Martin H. Kroll, Christopher R.

McCudden
Endogenous Interferences in Clinical Laboratory Tests
Patient Safety

Edited by
Oswald Sonntag and Mario Plebani

Volume 5
Martin H. Kroll, Christopher R. McCudden

Endogenous
Interferences in Clinical
Laboratory Tests

Icteric, Lipemic and Turbid Samples

Authors
Prof. Martin H. Kroll, MD Christopher R. McCudden, MD
Quest Diagnostics University of Ottawa
3 Giralda Farms Faculty of Medicine
Madison, NJ 07940 Department of Pathology & Laboratory Medicine
Park Ottawa, Ontario
USA Canada
E-mail: martinkroll500@gmail.com E-mail: cmccudde@uottawa.ca

The book has 28 figures and 19 tables.

ISBN 978-3-11-026620-7
e-ISBN 978-3-11-026622-1

Library of Congress Cataloging-in-Publication Data

A CIP catalog record for this book has been applied for at the Library of Congress.

Bibliographic information published by the Deutsche Nationalbibliothek

The Deutsche Nationalbibliothek lists this publicatioin in the Deutsche Nationalbibliografie;
detailed bibliographic data are available in the Internet at http://dnb.dnb.de.

© 2013 Walter de Gruyter GmbH, Berlin/Boston. The publisher, together with the authors and
editors, has taken great pains to ensure that all information presented in this work (programs, ap-
plications, amounts, dosages, etc.) reflects the standard of knowledge and the time of publication.
Despite careful manuscript preparation and proof correction, errors can nevertheless occur. Authors,
editors and publisher disclaim all respondibility and for any errors or omissions or liability for the
results obtained from use of the information, or parts thereof, contained in this work.
The citation of registered names, trade names, trade marks, etc. in this work dies not imoly, even
in the absence of a specific statement, that such names are exempt from laws and regulations
preotecting trade marks etc. and therefore free for general use.

Typesetting: PTP-Berlin Protago-TEX-Production GmbH, Berlin

Printing and Binding: Hubert & Co. GmbH & Co. KG, Göttingen
Cover image: Comstock/Getty Images

♾ Printed on acid-free paper

Printed in Germany
www.degruyter.com
To my wife Ellen and children Allison, Lauren and Jonathan.
Martin H. Kroll

To my wife Liesje and children Katie and Sam.

Christopher R. McCudden
Preface
Medicine has evolved to a new level, where not only is it expected that physicians
diagnose and treat patients efficaciously, but also that all patients are protected from
harm. Protecting patients from harm is part of patient safety and implies that the
processes used in taking care of patients are free from error. Medical care depends on
obtaining useful information from laboratory tests. Biochemical tests provide a great
deal of information at relatively low cost and usually with rapid turnaround times.
The achievement of low cost and rapid turnaround times depends, to a large extent,
on the use of automation. The dependence on automation subsequently results in a
diminution of individualized attention to each individual sample. To protect patient
safety, laboratories need to establish detection systems to identify situations that
could lead to biased results and rules to correct for the biased problems.
A bias occurs when the result obtained during an assay deviates from the true
value of the analyte in question. A systematic bias occurs when there is an inher-
ent problem in the measurement technique, as can occur with calibration errors and
reagent deterioration. All samples are affected by a systematic bias. Interferences
cause a non-systematic bias. Here, the bias occurs only for the individual sample. It
is important to identify common features that occur for interferences, and to identify
ways of not only identifying the interferences, but also of quantifying their impact.
For biochemical tests, especially those using serum or plasma as a matrix, a
high concentration of bilirubin and turbidity can affect biochemical tests. The most
common cause of turbidity is lipemia. The intent of this book is to provide a founda-
tion for those running laboratories to identify, quantify and correct for the presence
of hyperbilirubinemia and lipemia (turbidity). Because most laboratories will need to
perform these processes in an automated fashion, the people working in the labora-
tory will need to design the appropriate procedures and to manage them.
To establish the necessary foundation to effectively design processes to manage
the interferences caused by bilirubin and lipemia (turbidity), this book contains
several different perspectives. The early chapters of the book provide information
on the physical and chemical mechanisms involved in interferences. There is consid-
erable emphasis on the interaction of bilirubin and lipemic particles with light, the
most common form of energy used to detect clinical biochemical species. Additional
chapters provide an emphasis on the clinical conditions where one might expect to
encounter high concentrations of bilirubin or lipemia. The latter half of the book dis-
cusses means of detecting bilirubin or lipemia, as well as means to quantify their
presence, to allow for appropriate reporting of results. Finally, the last chapter dis-
cusses means of characterizing and quantifying interferences in complex reactions,
viii       Preface

as frequently occurs with bilirubin, where the analyte may interact with the analyte or
species directly related to the concentration of the analyte. The intent of the book is to
provide the laboratorian with sufficient background to deal with these interferences
and protect patient safety.

November, 2012 Martin H. Kroll, MD

Contents
Preface   vii

1 Accuracy Goals for Laboratory Tests   1

1.1 Accuracy and Precision   1
1.1.1 Definition   1
1.1.2 Imprecision as a Form of Error   2
1.2 Types of Error   2
1.2.1 Bias   2
1.2.2 Impact of Bias   4
1.3 Interference as a Type of Bias   6
1.4 References   8

2 Nature of Interferences   11

2.1 Definition   11
2.2 Nature of Interferences   11
2.3 Instrumentation   12
2.4 The Chemistry of the Absorbance of Light   15
2.5 References   20

3 The Nature of Icteric Interference   21

3.1 Source Information on Bilirubin Interference   21
3.2 Allen Correction as a Source of Bilirubin Interference   21
3.3 Bilirubin Interference with Oximetry   22
3.3.1 Co-oximetry Interference   24
3.3.2 Pulse Oximetry   25
3.3.3 Cerebral Oximetry   26
3.3.4 Interference with Methemoglobin   27
3.4 Chemical Reactions as a Cause of Bilirubin Interference   28
3.4.1 Bilirubin Reaction with Creatinine Methods   29
3.4.2 Bilirubin Reactions with Peroxidase Methods   31
3.5 References   32

4 The Nature of Lipemic and Turbidity Interferences   35

4.1 Types of Interferences   35
4.2 Lipemia Causes Turbidity   36
4.3 Lipemia Interference Mechanisms   37
4.3.1 Light Scattering   37
4.3.2 Lipoprotein Particles   40
4.3.3 Intralipid® and Lipemia Simulation   42
4.3.4 Empirical Studies in Lipemia Turbidity   43
x       Contents

4.4 Lipoprotein Particles and Lipemia   44

4.5 References   45

5 Measurement of Interference   47

5.1 A Typical Commercial Study   47
5.2 Guidelines for Interference Studies   48
5.3 Bilirubin   49
5.4 Intralipid®   50
5.5 Procedure to Make Five Concentrations   52
5.6 Interference Criteria   52
5.7 Data Analysis   54
5.8 References   60

6 Origin of Icteric Samples   63

6.1 The Origin of Bilirubin   63
6.2 Bilirubin Toxicity   65
6.3 Transport of Bilirubin in the Blood   65
6.4 Uptake of Bilirubin by the Liver   66
6.5 Clinical Aspects of Bilirubin   66
6.6 Neonatal Jaundice   67
6.7 Cholestasis   69
6.8 Hepatitis   70
6.9 Alcoholic Liver Disease   70
6.10 Hemolysis   71
6.11 Drug Induced Hyperbilirubinemia   71
6.12 Summary   72
6.13 References   72

7 Impact of Icterus   75

7.1 Introduction   75
7.2 Estimated Impacts Based on Interference Studies   75
7.3 Differential Interference with Different Bilirubin Isoforms   77
7.4 Non-spectrophotometric Icterus Interference   79
7.5 Resolving Icterus Interference   80
7.6 Summary   81
7.7 References   81

8 Origin of Lipemia and Turbidity   83

8.1 Lipoprotein Pathways   83
8.2 Classification of Hypertriglyceridemia   85
8.2.1 Frederickson Classification of Dyslipidemias   85
8.2.2 Obesity, Metabolic Syndrome and Diabetes   87
 Contents       xi

8.2.3 Alcohol   88

8.2.4 Nonalcoholic Fatty-liver Disorder   89
8.2.5 Medications   89
8.2.6 HIV Infection   89
8.2.7 Renal Disease   90
8.3 References   91

9 Impact of Lipemia/Turbidity   93

9.1 Introduction   93
9.2 Estimated Impacts Based on Interference Studies   95
9.2.1 Interference by Light Scattering   95
9.2.2 Interference by Volume Displacement   96
9.2.3 Interference by Lipid Partitioning   99
9.3 Summary   99
9.4 References   99

10 Endogenous Interferences in Clinical Laboratory Tests: Icteric, Lipemic and

Turbid Samples   101
10.1 Interference Indices   101
10.2 Generating Interference Indices   101
10.2.1 Preparation of Standards   102
10.2.2 Data Collection and Deconvolution of Non-Target Interferences   103
10.2.2.1 Subtraction Using Selected Wavelengths   104
10.2.2.2 Index Calculation Using Derivative Spectrometry   105
10.2.3 Establishing Indices and Defining Ranges   107
10.3 Limitations   110
10.4 Summary   110
10.5 References   111

11 Reporting of Results   113

11.1 Introduction   113
11.2 Procedures for Handling Samples with Interference Within the
Laboratory   113
11.3 Reporting of Results in Icteric and Turbid Samples   115
11.4 Autoverification and Reporting Algorithms   116
11.5 Practical Issues: Education and Implementation   117
11.6 References   118

12 Analyte-dependent Interference   119

12.1 Complex Interferences   119
12.1.1 Model for Analyte-dependent Interference   120
12.1.2 Examples of Analyte-Dependent Interference   121
xii       Contents

12.2 Statistical Testing for Significance   129

12.3 Failure to Design the Interference Study   133
12.4 Advantages of Using Multiple Regression Analysis   133
12.5 Concluding Remarks   135
12.6 References   137

Index   139
1 Accuracy Goals for Laboratory Tests
It is often said that laboratory tests account for 70 % of the objective information used
to diagnose and monitor patients. Even though it is true that a good history and physi-
cal examination provide a significant amount of information, physicians and clini-
cians, as well as nurses and other healthcare professionals, depend on laboratory
test results to provide a final diagnosis, determine the degree of illness (the disease
spectrum) and to monitor patients.

1.1 Accuracy and Precision

1.1.1 Definition

Accuracy of laboratory tests plays a vital role in health care, stipulating the quality
and assuring patient safety [1]. Typical process steps that infringe on the quality of
laboratory results and thus patient safety include patient misidentification, failure of
reagents, mismanagement, and failure to communicate [2]. The accuracy of labora-
tory tests is critically important for achieving and maintaining quality in delivering
good medical care. When the accuracy of laboratory tests is breached, the patient’s
safety is put at risk. Therefore, safe medical practice places a significant responsibil-
ity on the laboratory to maintain a high accuracy of test results. High accuracy of test
results depends on good laboratory practice and includes such processes as Quality
Control and Quality Assurance.
Accuracy is a generalized term. In the vernacular it may refer to how good the
quality of the test result is from an analytical perspective. Theoretically, one judges the
quality of the result arising from the laboratory by comparing it to a perfect method,
i.e., a method without defect, for which one has obtained a perfect specimen and the
reproducibility is perfect. The term Reproducibility is an ISO term [3] and refers to the
closeness of the agreement between the results of measurements of the same meas-
urand (analyte) carried out under controlled conditions of measurement. Essentially,
the term Reproducibility refers to the precision of the measurement made for a par-
ticular analyte. The laboratory easily determines the precision of an analyte by deter-
mining values for control materials. On a day to day basis, the results obtained for any
particular analyte for any control material will tend to a mean or average value. The
typical scatter around this value will demonstrate a normal (Gaussian) distribution,
and thus have a definable standard deviation (SD). Because results for any particular
analyte may take on any value across the reportable range of the analyte, a standard
deviation determined at a particular value for the given quality control material may
not be directly applicable. To extend the precision measurements over the report-
able range, one can use the ratio of the standard deviation to the mean of the quality
control value and express it as a percentage. This ratio is called the coefficient of
2       1 Accuracy Goals for Laboratory Tests

variation and for most tests in Chemistry it ranges between 1 % to 10 %, depending
on the analyte being measured and the magnitude of the value in the quality control
material. The coefficient of variation provides a measure of the precision. Ideally, the
clinician would like the precision, as measured by the coefficient of variation, to be
as low as possible.

1.1.2 Imprecision as a Form of Error

Another way to think of precision is that it represents the closeness of agreement

between independent measurements to each other. Of course, in the laboratory, in
order to put structure into the analytical process, the laboratory develops rules to stip-
ulate the conditions for performing the assay. Clinicians assume that all the values
for laboratory tests that they receive have an extremely high precision. They presume
that if they took a specimen and had the laboratory run that sample today, then if
they gave the laboratory the same specimen tomorrow, they would receive exactly the
same result.
The laboratory has to conduct itself with the knowledge that most clinicians are
not expecting that there are going to be errors in results. For this reason, laboratories,
and the people who manage them, spend a lot of time and effort in controlling the
processes to minimize the errors generated by running laboratory tests. Precision, or
in actuality, imprecision represents a non-systematic error. A non-systematic error is
not part of the designed process of deriving a value from the collection and analysis
of the specimen. Even though imprecision can be measured for the process, random
error causes the deviations from the central value (central tendency). Random errors,
though characteristic of the process, occur independently of one another. Even
though the measure of a random error allows one to predict how the population of
specimens will behave, one cannot predict for each individual specimen exactly what
will happen. In order to be able to predict exactly what will happen to each individual
specimen, one needs to examine the systematic errors.

1.2 Types of Error

1.2.1 Bias

Systematic errors are inherent in the process. Systematic errors are part of the process
of measurement, that is, they are the result of the way the sample and reagent are
mixed, the amplification of the detection system, and most importantly, how values
are assigned to the readings generated in the sensing process. How values are
assigned to the readings generated by the sensing process relates to the calibration
of the method. The calibration of the method can be biased if the standards used for
 1.2 Types of error       3

calibrating the method are not properly assigned. Most methods in the clinical labo-
ratory use calibrators instead of standards. Standards contain purified analyte dis-
solved in pure water or solvent of determined composition. Calibrators contain puri-
fied analyte or measured analyte dissolved in the matrix of the naturally occurring
constituents comprising the environment of the samples used for testing. The matrix
often is serum, plasma, or urine. Any of these matrices contains all sorts of unidenti-
fied and unspecified materials, typically protein, lipids, and organics. Typically the
laboratories making the calibrators will control the concentration of the electrolytes
and some of the organics. What makes a matrix material different from a standard
is the analyte of interest plus other analytes are bound or complexed with naturally
occurring constituents. The naturally occurring constituents may alter the way the
analytical method interacts with the analyte of interest, altering the signal from the
sensor.
Testing and assigning values in the laboratory are separated into three phases:
the pre-analytic, analytic and post-analytic phase. The pre-analytic phase includes
preparing the patient to obtain the specimen, collecting the specimen into an appro-
priate container (often with an anti-coagulant for blood), labeling and transporting
the specimen to the laboratory and processing of the specimen to present it to the
analyzer. The post-analytical phase includes communicating the value for the test
result to the clinician. The analytical phase includes physically introducing the speci-
men into a reaction vessel, chemically or biologically reacting the specimen with
other materials, physical interaction with some form of energy to produce a signal,
and translation of that signal into a number or value that can be communicated to
the clinician.
In the analytical phase, calibrators do not always translate the signal into exactly
the same set of values that a purified standard would. The mistranslation results in
a systematic error. Systematic errors can be separated into two types of error, based
on how they relate to the underlying true concentration. If the error, for example for
creatinine, were high or low and did not depend on the value for creatinine over the
entire range of results, then the error is constant. To illustrate the constant error, take
a value of 115 μmol/L of creatinine. If there is a constant error or bias of 27 μmol/L,
then the reported value would be 88  μmol/L instead of 115  μmol/L. Further, if the
true value of creatinine were 71 μmol/L, then the reported value would be 44 μmol/L;
and if the true value of creatinine were 398 μmol/L, then the reported value would be
371 μmol/L. The deviation from the true value would always be the same. What differs
in the error for each of these examples is the percentage of error that occurs. For the
115 μmol/L the percentage error is a negative 23 %, for the 71 μmol/L, the percentage
error is a negative 37 % and for the 398 μmol/L of creatinine, the percentage error is
a negative 7 %. The impact of a constant bias decreases with an increasing true value
of the analyte. More important is the effect that the error has on the interpretation of
the laboratory result. If the bias is negative and the true value falls within the refer-
ence interval and values below the reference interval have no clinical impact, then
4       1 Accuracy Goals for Laboratory Tests

the negative bias itself has no clinical impact. For a true value that exceeds the upper
limit of the reference interval, if the negative bias causes the reported value to fall
within the reference interval, then the interpretation would indicate that the patient
does not have the condition implied by abnormal values. Thus, if the upper limit for
creatinine in the reference interval were 106 μmol/L and the true value of the analyte
was 115 μmol/L, a constant bias of −27 μmol/L would cause the reported value to be
88 μmol/L, which falls within the reference interval. The reported result would indi-
cate that there is not a condition of renal dysfunction or impairment, which is classi-
fied as a false negative. At a creatinine concentration of 398 μmol/L, the clinician is
already aware that the patient has renal dysfunction. If the physician receives a result
of 371 μmol/L instead of 398 μmol/L, it would not change the assessment by the phy-
sician, because the interpretation of the test is that the patient has renal dysfunction
and the interpretation of the test is unchanged by the creatinine result. These exam-
ples are typical of those used for the purpose of making a diagnosis.

1.2.2 Impact of Bias

In addition to making a diagnosis, clinicians use laboratory tests to monitor the

disease or condition that the patient is experiencing. Here the situation is different,
because the clinician has already made a diagnosis for the patient’s disease or condi-
tion. The clinician is interested in whether the patient is getting better or worse, how
well the therapy is working or predicting the course of the disease and giving a prog-
nosis. The clinician may be observing the patient to follow the natural course of the
disease, waiting until the patient crosses a particular threshold of disease severity or
demonstrates enough change in their condition to indicate a time to institute therapy.
If the clinician is waiting for the values reported from the laboratory to indicate that
the patient has crossed into a more severe degree of their disease, then a constant
bias may disturb the proper conclusion. If the constant bias is negative, and the clini-
cian is waiting for the laboratory values to exceed a reference interval limit, then the
patient’s condition will exceed the limit before the reported laboratory values do. In
such a case, the clinician may not institute therapy soon enough and may inadvert-
ently postpone therapy. If the constant bias is positive, and the clinician is waiting
for the laboratory values to exceed the reference interval limit, then the reported
laboratory values will exceed the limit before the patient’s condition truly does, and
the clinician may institute therapy too early, potentially exposing the patient to risk
from the therapy. If the institution of therapy is not warranted, because it is a false
positive, then in addition to exposing the patient to the risk of therapy, the clinician
may cause valuable resources to be expended when they are not needed. In a cost-
conscious world, expending resources when they are not required results in a waste
of resources, which potentially can risk the safety of the entire patient population,
because abuse of resources may prevent the use of resources for another patient.
 1.2 Types of error       5

Clinicians often monitor patients observing changes in results. A constant error or bias
may have minimal impact here, because if the reported value was initially 115 μmol/L
with a true value of 141 μmol/L, when the next value is reported as 97 μmol/L with a
true value of 124 μmol/L, the net change in value from before to after is 18 μmol/L for
both the reported and true values. Thus, in observing the absolute change over time,
a constant bias has no effect.
The situation is different for proportional bias. With a proportional bias, the
degree of bias depends on the true concentration. For a positive proportional bias, the
degree of bias increases with increasing concentration of the analyte, while for a neg-
ative proportional bias, the degree of bias decreases with increasing concentration.
For a proportional bias of 10 % and creatinine, at 71 μmol/L true value, the reported
value would be 78 μmol/L. At a creatinine concentration of 106 μmol/L, the reported
value would be 117 μmol/L; while at a creatinine concentration of 398 μmol/L, the
reported value would be 438 μmol/L, and so on. The proportional bias demonstrates
a constant percentage of error over all the values of the reportable range. The per-
centage bias can be positive or negative. Typically the proportional bias is reported
as a slope. A positive bias of 10 % would have a slope of 1.1, while a negative bias of
10 % would have a slope 0.9. The proportional bias can cause the same problems with
diagnosis as does the constant bias: false negative results and false positive results.
Proportional bias shows a greater impact with monitoring of patients than the
constant bias does. Monitoring of patients entails comparing laboratory results from
one time to the next. If there is no change in the patient’s condition, then one would
not expect the laboratory values to change and there would be no problem. If there
is a change in the patient’s condition, one would expect the laboratory results to
change. In monitoring a patient for renal function, as their renal function worsens,
one would expect their creatinine and urea values to increase. If there was a nega-
tive proportional bias, their values for creatinine and urea would not rise as quickly
as their condition. If there was a positive proportional bias, their values for creati-
nine and urea would rise quicker than the actual condition. For example, if pharmacy
needed to adjust the dosage of a drug based on the patient’s renal clearance of that
drug, then if the reported creatinine value was 20 % higher than the true concentra-
tion, the calculated dosage would be too low and the patient would not receive a
sufficient amount of drug; likewise, if the reported creatinine value was 20 % lower
than the true value, the patient would be overdosed on the drug and run the risk of
becoming drug toxic. Acyclovir, amikacin, ceftazidime, ciprofloxacin, digoxin, gen-
tamicin, lithium, ofloxacin, piperacillin, tobramycin, and vancomycin are just some
of the medications that require adjustment of dosage based on the creatinine and
creatinine clearance values [4].
Even though there may be pure cases of constant bias by itself, or proportional
bias by itself, most biases are mixed. In mixed bias, both constant and proportional
biases have an effect on the reported results. Frequently, the constant and propor-
tional biases run in opposite directions, i.e., if the constant bias is positive, the pro-
6       1 Accuracy Goals for Laboratory Tests

portional bias will be negative, or if the constant bias is negative, then the propor-
tional bias will be positive. The net effect of the constant and proportional biases
running in opposite directions is that there will be some central value where the net
bias is zero. If the biases balance out one another in this manner, then it means that
the true value of the analyte dictates whether there is a positive or a negative bias.
For example, if there is a positive constant bias balanced by a negative proportional
bias, then when the true value of the analyte is less than the central value with zero
bias, the net bias is positive, but when the true value of the analyte is greater than
the central value with zero bias, the net bias is negative. Usually in calibration issues
these biases are small. Fortunately, one can characterize these biases quite well. The
laboratory expends considerable energy and resources, such as proficiency testing
surveys and comparisons with other laboratories, in characterizing and minimizing
the systematic error or biases.
Constant and proportional biases represent systematic errors in analysis, while
imprecision represents non-systematic error. In both cases the error can be well-
characterized and is predictable. Often, the degree of error can be controlled. Both
types of errors depend on the mechanics of the instrumentation, the selection of the
reagents, and the quality of the calibration. Interferences in Laboratory Medicine rep-
resent another type of error. The error due to interference is not systematic, in that it
does not depend on the mechanics of the analyzer nor on the quality of calibration.
It does not apply to all specimens, but is specimen-specific. It does depend on the
choice of sensor and reaction. Interference error, even though it is non-systematic, is
not measurable in the same way that imprecision error is. Most of the time, it is epi-
sodic. Because it is episodic, laboratories need to develop ways to detect its presence
and report results in a suitable manner and reduce its impact on laboratory results [5].

1.3 Interference as a Type of Bias

Interferences depend on the specimen, the method and the type of reaction involved.
Owen and Keevil examined the interference cause by bilirubin with the Jaffe (alka-
line picrate) and enzymatic methods for creatinine [6]. They varied the concentration
of bilirubin and compared the measurements for both reaction methods with liquid
chromatography tandem mass spectrometry (LC-MS/MS) as the reference method.
Samples with low concentrations of creatinine showed more than a 10  % reduc-
tion in creatinine at bilirubin concentrations greater than 220 μmol/L as measured
by the Jaffe method. For samples measured by the enzymatic method, more than a
10 % reduction in creatinine was observed for bilirubin concentrations greater than
200  μmol/L. The greatest percentage of reduction occurred for specimens with the
lowest concentration of creatinine.
A 20  % reduction in the reported value for creatinine can affect the quality of
care and patient safety. The glomerular filtration rate is estimated by two different
 1.3 Interference as a Type of Bias       7

equations, the Cockroft-Gault equation and the Modification of Diet in Renal Disease
(MDRD) study [7]. The Cockroft-Gault equation is used to modify drug dosage, while
the MDRD equation is used to assess risk in patients and identify patients for further
workup for diminishing renal function.
Examination of the use of a test in varying disease states provides a way to assess
the full impact of an interference. Clinicians are interested in assessing glomerular
function. Many diseases of the kidney or injuries to the nephron result in decreased
glomerular function. In physiologic studies the glomerular filtration rate provides
an assessment of the number of sufficiency of the glomerular and nephrons in the
kidney. In physiologic studies, the glomerular filtration rate is assessed by injecting
and measuring inulin, which provides the inulin clearance.
The inulin clearance is not a practical way to assess renal function. Instead, clini-
cians use the creatinine clearance. Creatinine is an endogenous compound, the break-
down product of creatine produced by skeletal muscle. Its main advantage is that
there is no requirement to infuse it into the patient. Creatinine is freely filtered by the
glomerulus, but the renal tubules also secrete a small quantity of creatinine into the
forming urine [8]. In normal kidneys, the tubular secretion accounts for 10 % of the
total secretion, but as renal disease progresses, the percentage of the total creatinine
excretion contributed by the tubules increases. Though imperfect as a marker, the use
of creatinine to assess renal function is a standard practice. In the past, many clini-
cians used the creatinine clearance to assess renal function. Calculating a creatinine
clearance requires the collection of a 24-hour urine sample to measure the creatinine
in it. The collection must be done at the same time that the serum or plasma sample
is obtained for creatinine. The clearance is calculated by dividing the urine creatinine
production rate by the serum or plasma creatinine concentration [8]. Because there is
a discrepancy in the orders of magnitude of the creatinine concentration in urine and
serum (plasma) and there is some degree of error in the collection of the 24-hour urine
specimen, there is considerable error in calculating a creatinine clearance. Today,
most clinicians depend on the Cockroft-Gault equation and the Modification of Diet
in Renal Disease (MDRD) formula to assess renal function.
Renal disease can be acute or chronic. Any patient who develops renal disease,
whether acute or chronic, requires a further workup to establish the cause, especially
if they are young. Other tests, more expensive than creatinine or with some degree of
discomfort and risk are required, such as renal biopsy. Invasive tests, such as renal
biopsy, put the patient at risk, and are extremely costly.
Primary glomerular diseases include glomerulosclerosis, acute postinfectious
glomerulonephritis, membranoproliferative glomerulonephritis, IgA nephropathy,
and chronic glomerulonephritis [9]. Systemic diseases also affect the kidneys and
include lupus nephritis, diabetic nephropathy, amyloidosis, Goodpasture syndrome,
Wegener’s granulomatosis, Henoch-Schönlein purpura, bacterial endocarditis, and
thrombotic micrangiopathy [9]. Diseases that affect the tubules ultimately decrease
glomerular filtration and increase creatinine concentrations, and include such dis-
8       1 Accuracy Goals for Laboratory Tests

eases as acute pyelonephritis, chronic pyelonephritis, and acute tubular necrosis [9].
Some of these diseases represent medical emergencies, such as any acute glomerulo-
nephritis, acute pyelonephritis, and acute tubular necrosis. These medical emergen-
cies require quick action, and a reported creatinine result that falls below the speci-
fied limit slows the response for appropriate treatment and puts the patient at risk.
The blood vessels may be involved in renal disease and include stenosis or occlusion
of the renal artery secondary to atherosclerosis, malignant hypertension with result-
ing nephrosclerosis and benign nephrosclerosis [9]. Essential hypertension causes
hyperplastic arteriolosclerosis in the kidneys. Essential hypertension is one of the
major causes of decreasing renal function. The other major cause of renal dysfunc-
tion is diabetes mellitus. It causes diabetic macrovascular disease and hyaline arte-
riolosclerosis and is related to the duration of the disease and the level of the blood
pressure [9].
Dehydration and hypovolemia cause both blood urea nitrogen and creatinine to
rise, usually urea rises faster than creatinine. Dehydration and hypovolemia occur
in numerous diseases such as diabetes mellitus, acid base disorders, shock, hemor-
rhage, infection, trauma and surgical emergencies. In dehydration and hypovolemia,
blood urea nitrogen rises fast than creatinine. Normally, the urea to creatinine ratio
is around 10 (urea and creatinine in units of mg/dL (conversation ratio mg/dL into
μmol/L = 40)). During dehydration and hypovolemia the urea/creatinine ratio rises to
above 20 [10]. If the value reported for creatinine is falsely depressed because of inter-
ference, then this ratio will rise even higher. Clinicians use it in emergency situations
and acute care to decide whether to give more fluids. A falsely elevated value for this
ratio would prompt clinicians to give fluid inappropriately, and, perhaps, not search
for other causes for the azotemia and increased creatinine values.
Interferences represent a non-systematic type of error that is episodic and dif-
ficult to predict. Interferences impact patient safety by creating false positive or false
negative results when used for diagnosis, or by falsely increasing or decreasing the
difference between two results separated by time used in monitoring. Because it is
difficult to predict the exact errors caused by interferences, especially for bilirubin,
lipemia and turbidity, one needs to be aware of the nature of interferences, ways to
measure them, the clinical situations that give rise to elevated bilirubin, lipemia and
turbidity and ways to detect their presence.

1.4 References
[1] Stankovic AK. The laboratory is a key partner in assuring patient safety. Clin Lab Med USA
2004,24,1023–1035.
[2] Valenstein P. Quality Management in Clinical Laboratories. Northfield IL College of American
Pathologists, 2005.
[3] VIM93 ISO, International Vocabulary of Basic and General Terms in Metrology. Geneva,
Switzerland International Organization for Standardization, 1993.
 1.4 References       9

[4] Dipiro JT, Talbert RL, Yee GC, Matzke GR, Wells BG, Posey LM. Pharmacotherapy: A Pathophys-
iologic Approach. 7th ed. New York, NY, USA, McGraw Hill, 2008.
[5] Sonntag O. Analytical interferences and analytical quality. Clin Chim Acta The Netherlands,
2009,404,37–40.
[6] Owen LJ, Keevil BR. Does bilirubin cause interference in Roche creatinine method? Clin Chem
USA 2007,53,370–371.
[7] Kaplan LA, Pesce AJ. Clinical Chemistry: Theory, Analysis, Correlation. St. Louis, MO, USA,
Mosby, 2010.
[8] Noe DA, Rock RC. Laboratory Medicine: The Selection and Interpretation of Clinical Laboratory
Studies. Baltimore, MD, USA, Williams & Wilkins, 1994.
[9] Kumar V, Abbas AK, Fausto N, Mitchell RN. Robbins Basic Pathology. 8th ed. Philadelphia, PA,
USA, Saunders, 2007.
[10] McPherson RA, Pincus MR. Henry’s Clinical Diagnosis and Management by Laboratory Methods.
21st ed. Philadelphia, PA, USA, Saunders, 2007.
2 Nature of Interferences

2.1 Definition

There are two commonly used definitions of interference. The first definition claims
that an analytic interference occurs when there is a component in the sample that
causes an error in measurement of the analyte in the analyzer but by itself does not
produce a signal [1]. This definition presents a problem because it fails to include any
effect that the interferent may have in the absence of the analyte. It is better to define
interference as any agent present in the sample, which causes the result of a measure-
ment process to deviate (demonstrate a bias) from the true value or value that would
have been obtained in the absence of the agent [2]. Defining the term interference
in this broad fashion would allow the definition to cover such agents as bilirubin,
lipemia and turbidity. The definition for interference can include the effect of an agent
in the detection or determination of concentration or activity of an analyte occurring
at any step along the path to providing a result [3]. For a more detailed discussion of
the concepts of interference in analytical systems, see the review by Büttner [4]. For
the purposes of this book, we define interference as the effect of a substance present
in the sample that alters the correct value of the determination of the result, quantita-
tively or qualitatively, for the analyte under consideration [2].
The interfering agent is a substance that alters the determination of the correct
value. That substance may be exogenous or endogenous. An exogenous substance is
one that is not a naturally occurring chemical or agent, e.g., an antibiotic. An endog-
enous substance is one that occurs naturally, but is in a higher concentration than
normally encountered. Hemolysis, lipemia, bilirubinemia, and paraproteinemia are
examples of endogenous substances.

2.2 Nature of Interferences

Bilirubin and lipemia generate interferences by three basic mechanisms. Bilirubin

and lipemia may alter the way light is absorbed or scattered. Also, bilirubin may react
with reagents or with the analyte itself. Hyperlipidemia may alter the balance between
aqueous and non-aqueous phases. Melvin Glick et al., in their book, Interferographs,
presented many examples of interferences caused by bilirubin and lipemia [5]. This
book is arranged by analyzers and demonstrates the effect of interference for bili-
rubin, and for that matter, lipemia as well, over increasing concentrations on many
different types of analyzers that were in use at that time. Today, it remains a valuable
resource, because the book clearly demonstrates the ubiquitiousness of interference
over many different tests and varying methodologies.
12       2 Nature of Interferences

Awareness of bilirubin interference was heightened in the 1990s and manufacturers

responded by improving their methods. Improvement of methods occurred through
better selection of wavelengths, addition of reagents that inhibited the interference
effects, and changes in the way that analysis occurred in the analyzer. Despite these
changes for improvement, for which the manufacturers can be thanked, knowledge
of the methods affected by bilirubin is still helpful today, because the mechanisms
behind the interferences have not been elucidated in all cases and the evolution of
newer analyzers and the search for cheaper reagents may return one to the method-
ologies used in the past. Observation of the different methods and the noted inter-
ferences serves as a resource to evaluate methods used in the present and in the
future.

2.3 Instrumentation

The most commonly used instrumentation for analysis in clinical chemistry is the
spectrophotometer. The purpose of the spectrophotometer is to sense the absorbance
of light by a substance. It represents an extension of the human eye. Also, it con-
verts the amount of light absorbed into a voltage change, which can be quantified and
translated into a concentration.
The basic design of a spectrophotometer consists of a source of radiant energy,
a dispersive device, a vessel to hold the analyte in a solvent and for the reaction to
occur, a photodetector and finally, a readout device [6]. The most common sources of
radiant energy consist of a lamp. Typically these lamps emit electromagnetic radia-
tion from 150–10,000 nm. The spectrophotometer contains a dispersive device that
separates the different wavelengths of light into discrete bandwidths of radiation [6].
Examples of dispersive devices are absorption filters, interference filters, prisms and
diffraction gratings. The dispersive device may be placed either before or behind the
sample. Finally, the incident radiation (light) that passes through the sample will hit
a detector. Commonly used detectors are photodiodes or photomultiplier tubes.
The radiation or light hitting the sample may be absorbed by molecules with
that ability. The light absorbed by each of these molecules depends on the chemical
structure of that particular molecule and the wavelength of the radiation. In the light
spectrum, on the left-hand side are the higher energies, shorter wavelengths associ-
ated with ultraviolet light, while on the right-hand side are the lower energies, longer
wavelengths associated with red light. When electromagnetic radiation encounters
a molecule, it may be absorbed, scattered or completely pass by the molecule. In
a spectrophotometer, when the wave of light passes by the molecule, that process
is known as transmission. Spectrophotometers are designed to measure the light
passing through a cuvette that contains the analyte of interest as well as a cuvette
in which the analyte of interest is omitted. The cuvette without the analyte is known
as the blank. The photometer provides a readout of the amount of light transmitted
 2.3 Instrumentation       13

through the blank cuvette and the test cuvette, the one containing the analyte, for
which the amount of light from both of these sources can be compared. The amount
of light transmitted through the cuvette, i.e., the light that bypasses the molecules in
the sample, is known as the transmittance [6].
The transmittance represents a comparison between the amount of light trans-
mitted through the test cuvette and the amount of light transmitted through the blank
cuvette. A comparison is required because light may be absorbed or scattered by the
material composing the cuvette itself, even if it is glass, quartz or plastic, and by
the solvent within the cuvette. In the clinical laboratory, the most commonly used
solvent consists of water, but may include other chemicals as well, typically such
chemicals as salts, protein, surfactants, etc. By matching the cuvettes, the amount
of light absorbed by the cuvette and solvent by both the blank and the sample can
be essentially handled and accounted for. The amount of light transmitted through
the blank is referred to as Io, while the amount of light transmitted through the test
sample is I [6]. The transmittance for the test sample is defined as T = I/Io [6]. The
spectrophotometer will translate the amount of light transmitted, i.e., the amount of
light impinging on the detectors, into a voltage difference. Ultimately this voltage dif-
ference is sent to a readout, which can quantify the amount of voltage difference for
the test sample and the blank. For our purposes here, the quantity can be considered
to be arbitrary. If the amount of transmitted light had a value of 500 for the blank and
400 for the test sample, then the transmittance would be 0.9.
The amount of light transmitted through the test sample, where our analyte is
located, depends on the concentration of the analyte; however, it does so in an expo-
nential manner. Therefore, the relationship between the concentration of analyte
and transmittance is not linear. To make it easier to understand, the logarithm of the
reciprocal of the transmittance is used to define absorbance, A [6].

   
1 I
A = log = − log (2.1)
T I0

The power of defining the absorbance, A, in this way comes from the Lambert-Beer
Law, usually known as simply Beer’s law (Fig. 2.1) [6].

1 4
0.8 3
transmission

absorbance

0.6
2
0.4
0.2 1
0 0
0 10 20 30 0 10 20 30
concentration concentration

Fig. 2.1: The relationship between percent transmittance or absorbance and concentration.
14       2 Nature of Interferences

According to Beer’s law the absorbance, A, is related to the capacity of a molecule to

absorb light, a, the pathlength of the cuvette, b, and the concentration of the analyte,
C, in a linear fashion.

A = abC (2.2)

One knows the pathlength of the cuvette. Also, one can determine the capacity of the
molecule to absorb light, known as the specific absorptivity or specific absorptivity
coefficient. Then, one can directly relate the absorbance to the concentration of the
molecule being identified. A transmittance of 1.0 yields an absorbance of 0, that of 0.5
yields an absorbance of 0.3, that of 0.1 an absorbance of 1.0, that of 0.01 an absorb-
ance of 2.0, and that of 0.001 an absorbance of 3.0. Absorbance provides an easy way
to measure the molecule of interest. Spectrophotometers are limited in the range of
concentrations that they can measure because of error of detection. There is consider-
able error of detection when the concentration of the molecule of interest is very low
or very high. For most practical purposes, the relative error is considered high when
the absorbance, A, falls below 0.1 or above 2.4 [6].
Methods using spectrophotometers to measure analytes are designed to relate a
set of standards or calibrators that cover absorbances from 0.1–2.0. The standards or
calibrators represent known entities. The reaction is run and the absorbance for each
concentration is determined, yielding, in most cases, a linear relationship, known
as a standard curve. The slope of the line determined by this standard curve directly
relates the absorbance to a concentration. When the unknown test sample is ana-
lyzed, the absorbance is compared to this standard curve, from which one can read
the concentration for the unknown. In non-automated methods, one typically reads
the concentration from a graph. In automated analyzers, one can use the inverse of
the standard curve. If the standard curve is represented as
 
A = slope · analyte + constant, (2.3)

then the concentration of the analyte, [analyte], can be found from the equation

  A − constant
analyte = . (2.4)
slope

Spectrophotometry works because molecules often demonstrate a particular absorb-

ance spectrum that is characteristic of their structure, or can be made characteristic of
their structure. The simplest methods simply measure the absorbance of the test mol-
ecule in the appropriately diluted sample. If the intent is to measure the test molecule,
such as hemoglobin, without dilution, then the test apparatus can be modified to
control the pathlength, in order to bring the absorbance within the appropriate range.
Simply measuring the absorbance of the test molecule requires that the test molecule
 2.4 The Chemistry of the Absorbance of Light       15

absorbs a sufficient quantity of light within a spectrum of wavelengths that can be

considered to be fairly specific for that molecule. If the molecule of interest does not
absorb a sufficient quantity of light within the range of the spectrophotometer, or
the specificity is not sufficient, then the molecule can be reacted with reagents that
will produce another molecule that can meet these criteria. The challenge in Clini-
cal Chemistry and much of Laboratory Medicine is to design methods that allow the
determination of a molecule of interest in a sea of other molecules that have similar
physical and chemical characteristics.

2.4 The Chemistry of the Absorbance of Light

The spectral range that one may utilize to determine molecules of interest is some-
what limited. The infrared region of the spectrum begins at 700 nm and extends to
5,000 nm. Molecules absorb light in the infrared region as the result of vibrational
motions of atoms as well as bending, rotating and twisting [6]. Infrared spectroscopy
is useful in identifying fairly pure solutions of a molecule, because the absorbance is
specific for functional groups making up the molecule. One needs to examine absorb-
ance versus a fairly wide spectrum and relate the absorbance peaks to known func-
tional groups. Sometimes the absorbance spectrum may give way to a ‘fingerprint’
pattern compared to known compounds.
At the other end of the spectrum, with wavelengths below 400 nm, begins the
ultraviolet region. The human eye cannot detect ultraviolet light. There are many
molecules that absorb in this region; however, the use of the ultraviolet region is
limited because of absorbance by materials used to hold the sample, such as glass,
and by solvents. Most organic solvents absorb significant amounts of light between
100 and 300  nm [6]. For example, acetone absorbs light at 340  nm and ethanol at
210 nm. Water, the most common solvent for Laboratory Medicine purposes, absorbs
at 191  nm  [6]. For these reasons, most wavelengths usable in determinations fall
between 340 and 700 nm.
Most molecules of interest do not absorb enough light between the ranges of
340–700 nm without some chemical reaction. Usually, as in the case of creatinine, the
molecule of interest will shift the absorption band of a reagent, allowing for quanti-
fication of the analyte. Bilirubin is unique in that it absorbs light without the need of
reaction. Lipemia is unique in that it is the major source of turbidity, a form of light
scattering.
As mentioned before, light, a form of electromagnetic radiation can interact with
matter. If the light is scattered by the matter, no change in the matter occurs. If the
matter absorbs the light, then there is an alteration of the molecule in some fashion.
Radiation with short wavelengths, such as X-rays, when they interact with matter can
break chemical bonds. Often, this interaction is destructive. At long wavelengths,
16       2 Nature of Interferences

160000
carotene
extinction coefficient
140000
120000
100000
80000
60000
40000
20000
0
-20000
250 300 350 400 450 500 550 600
wavelength (nm)

60000
bilirubin
50000
extinction coefficient

40000
30000
20000
10000
0
-10000
250 300 350 400 450 500 550 600
wavelength (nm)

Fig. 2.2: Absorbance spectra for carotene and bilirubin.

carotene

M P M P V M M V

O N N N N O

bilirubin

Fig. 2.3: Polyene structure of carotene and bilirubin.

 2.4 The Chemistry of the Absorbance of Light       17

such as infrared light, nearly all the chemical bonds can absorb the light and the
energy is given off in some form of heat, because it represents vibrational energy.
In regards to analytical methodology and spectrophotometry, absorption of visible
and near ultraviolet radiation is the most interesting and useful form of absorption.
Molecules absorbing light in this region undergo electron excitation [7]. If the surface
of a bulky material is very smooth, it may reflect light, for example, a mirror. If the
surface is rough, it will reflect light diffusely. The latter effect is the one we experience
in everyday life. Diffuse reflection demonstrates the color of the object [7]. Scattered
light, as occurs with a powder, does not demonstrate a color, because there is no
interaction with the matter itself [7].
For absorption in analytical chemistry, the most important sources of color for
objects are transition metals and transitions between molecular orbitals [7]. The
absorption by transition metals is an important facet for the chemistry of minerals
and inorganic chemistry and has been used to detect and quantify transition metals
in clinical chemistry. They are notable for the absorption of a specific wavelength of
light, with the transition of an electron in a d-orbital to a specific higher energy level,
which is known as a ligand field effect [7]. The resulting absorption band often is very
sharp.
Far more important for the purposes of clinical chemistry are absorption and
color in organic molecules. Absorption by organic molecules can be described by
the molecular orbital theory [7]. Much of the knowledge concerning color in organic
molecules is derived from the study of organic dyes. In dyes, it has been noted that
conjugated double bonds, as occur in carotene, give rise to a particular color (Fig. 2.2
and Fig. 2.3).
Dyes possess several resonance structures [7]. In molecular orbital theory, there
are several types of orbitals involved, including bonding or antibonding orbitals of
sigma (σ or σ*) or pi (π or π*) types, and in addition, n-type nonbonding orbitals [7].
In the lowest energy state of the molecule, all of the bonding and nonbonding orbit-
als are fully occupied, while the antibonding orbitals are empty [7]. When the organic
molecule absorbs light, the energy of the light now becomes part of the molecule,
often described as a transition of orbitals from the highest occupied molecular orbital
(HOMO) to the lowest unoccupied molecular orbital (LUMO), i.e., an electron transfers
from a nonbonding or bonding orbital to an antibonding orbital, usually n → π* and
π → π* [7]. In formaldehyde the π → π* produces a strong absorption band at 185 nm
and the n → π* produces a strong absorption band at 290 nm [7]. Neither of these two
bands produce a color in the visible region, but addition of another double bonded
carbon to formaldehyde produces acrolein (Fig. 2.4), and shifts the n → π* absorption
from 290–330 nm, much closer to the visible region in the spectrum [7]. Addition of
another double bonded carbon to acrolein (Fig. 2.4) produces a molecule with yellow
absorbance.
Continued addition of double bonds in a polyene fashion produces dyes, such as
carotene (Fig. 2.2 and Fig. 2.3). Absorbance of this molecule is predominated by the
18       2 Nature of Interferences

O O
C H2C C
H H H
formaldehyde acrolein

Fig. 2.4: Structure of formaldehyde and acrolein.

π → π* transition [7]. In organic molecules, the p-orbitals are unshielded from electro-
static interactions; the electrostatic interactions from other electronic orbitals have
their effect on the absorption bands, broadening them out, with the result that the
absorption band for organic molecules are typically very broad. Also, in organic mol-
ecules, the change in energy levels between the LUMO and the HOMO states depend
on the length of the π system in the molecule and the bond length; molecules like
bilirubin are rather plastic and there can be small variations in the length of the π
system, causing a dispersed population of molecules absorbing light [8].
As the number of double bonds increases in a dye, so does the wavelength of
light absorbed by the dye [8]. If the wavelength of light absorbed is 420  nm, then
the color of the light absorbed is blue, and the solution appears yellow; if the wave-
length is 540 nm, the color of the light absorbed is green and the color of the solu-
tion is red; if the wavelength of light absorbed is 640 nm, then the color of the light
absorbed is orange and the color of the solution is blue; and if the wavelength of the
light absorbed is 740 nm, then the color of the light absorbed is red and the color of
the solution is bluish green [8]. As the number of π electrons increases in a polyene,
so does the wavelength, with 16–30 p-electrons placing its absorbance in the visible
region between 400 and 550 nm [8]. Beta-carotene is a typical example, with a strong
orange absorption (Fig. 2.2 and Fig. 2.3). Bilirubin also has a polyene structure with
nitrogens in the place of some of the carbons (Fig. 2.3). Carotene has 11 double bonds
while bilirubin has ten. In both carotene and bilirubin, the double bonds are laid
out in adjacent fashion without ring structure formation. Benzene shows aromatic
resonance because it has three double bonds arranged in hexagonal ring structure.
Neither carotene nor bilirubin demonstrate aromatic resonance. Because bilirubin
does not demonstrate aromatic resonance as in benzene, it acts as a strong chromo-
phore in the visible region of light (Fig. 2.2). The absorbance spectra for carotene and
bilirubin are almost identical in the visible region. Bilirubin is not quite as strong as
carotene in its molar absorptivity, but both molecules represent important natural
dyes.
Polyenes do not need to exist only in a straight chain, but may also occur in cyclic
form, and they can represent strong absorbers of light if they do not present with
benzenoid conjugation and aromatic resonance [7]. The porphyrins are strong rep-
resentatives of this class and include chlorophyll and heme. Of course, bilirubin is
the breakdown product of heme. Bilirubin has a strong potential for causing interfer-
ence because its absorption band covers a broad area of the spectrum. In the past, if
 2.4 The Chemistry of the Absorbance of Light       19

one directly measured absorbance without a chemical reaction, bilirubin has a strong
potential for causing interference, because its absorbance might be measured as well.
This interfering potential is an important problem for co-oximetry, where hemoglobin
in its various forms is measured without chemical alteration.
One of the most common ways to minimize interferences with methods is using
a sample blank. In a manual spectrophotometer, that process is rather easy, because
one would use a sample diluted with buffer as the blank in the analyzer. Automated
equipment used to present a challenge, because automated analyzers used a single
blank channel for multiple samples. More modern automated analyzers solved this
problem by taking multiple readings. Thus, an initial reading is taken after the addi-
tion of the sample and the buffer or other diluents, prior to adding the reagents neces-
sary to start the reaction. This blank reading for the sample is held in the analyzer’s
computer and subtracted from the final readout answer. The multiple readings
approach works well, but may be confounded if the blank reading is so high that the
final total absorbance is very high. If the final total absorbance reading is very high,
it may exceed the photometric error allowed for the assay and result in a flag or code,
indicating that there is a problem with the sample [9].
An alternate approach to taking a blank reading is to subtract out a baseline of
the interferent. This approach is commonly used to deal with the effects of hemo-
globin present in the sample and is mentioned here because its use has the potential
to cause problems with bilirubin. This method is commonly referred to as the Allen
correction. In the classic Allen correction, in addition to the wavelength chosen for
the primary chromogen (usually chosen close to its maximum value of the absorption
curve after subtracting out the absorbance for the reagent), two other wavelengths are
chosen, one to the left of the chromogen wavelength and the other to the right of the
chromogen wavelength (Fig. 2.2) [9]. This method works if the spectrum of the inter-
ferent differs from that of the chromogen in the reaction. It assumes that the spectrum
of the interferent is approximately the same on both the left-hand side (lhs) and the
right-hand side (rhs), and that the chromogen has minimal absorbance at these two
wavelengths. The formula for calculating the Allen correction is

(Alhs + Arhs )
Corrected Achromogen = Achromogen − . (2.5)
2

In many modern analyzers, a modified version of the Allen correction is used, where
only one wavelength is used to adjust for the presence of hemoglobin. Use of only
one wavelength may incur effects from other potential interferents, such as bilirubin.
20       2 Nature of Interferences

2.5 References
[1] IFCC, IFCC provisional recommendation on quality control in clinical chemistry. J Clin Chem Clin
Biochem Germany 1976,14,270.
[2] Kroll MH, Elin RJ. Interference with clinical laboratory analysis. Clin Chem USA 1994,40,
1996–2005.
[3] chemistry; proposed guidelines. NCCLS Document EP7-P. Villanova, PA, USA, National
Committee for Clinical Laboratory Standards, 1986.
[4] Büttner J. Unspecificity and interference in analytical systems: concepts and theoretical
aspects. DG Klin Chem Mitteilungen Germany 1991,22,3–11.
[5] Glick MR, Ryder KW, Glick SJ. Interferographs User’s Guide to Interferences in Clinical Chemistry
Instruments. 2nd ed. Indianapolis, IN, USA, Science Enterprises, Inc, 1991.
[6] Willard HH. Merritt Jr LL, Dean JA, Settle Jr FA. Instrumental Methods of Analysis. 6th ed.
Belmont, CA, USA, Wadsworth Publishing Company, 1981.
[7] Nassau K. The Physics and Chemistry of Color: The Fifteen Causes of Color. New York, NY, USA,
John Wiley & Sons, 1983.
[8] Kuhn H, Forsterling H-D. Principles of Physical Chemistry: Understanding Molecules, Molecular
Assemblies, Supramolecular Machines. Chichester, UK, John Wiley & Sons, Ltd, 2000.
[9] Kaplan LA, Pesce AJ. Clinical Chemistry:Theory Analysis Correlation. 5th ed. St. Louis, MO, USA,
Mosby Elsevier, 2010.
3 The Nature of Icteric Interference

3.1 Source Information on Bilirubin Interference

The most comprehensive and informative source of information on bilirubin inter-

ference comes from the book, Interferographs, by Melvin Glick, Kenneth Ryder and
Starla Glick [1]. As might be expected, the degree of interference varies by manufac-
turer and by method. Some analyzers demonstrate next to no interference for their
methods, while other analyzers demonstrate many methods with an interference due
to bilirubin. Some of the more common methods affected by a bilirubin interference
are creatinine, uric acid, and alkaline phosphatase. Often the creatinine, cholesterol
and uric acid interferences are negative, that is, yield lower results, while the bicarbo-
nate results are positive, that is yielding higher results than expected. [1].
The College of American Pathologists (CAP), through its Instrumentation
Resource Committee, offers a survey that provides materials for testing the impact
of interfering substances with common chemistry tests. The survey materials include
samples with specified amounts of either hemoglobin or bilirubin. The survey is
designed for verifying manufacturers’ interference specifications and investigating
discrepant results caused by interfering substances. The survey is designed to cover
more than 20 common analytes.
The survey provides information for the individual laboratory, showing the
effects of hemoglobin and bilirubin interference for their own analyzers. In addition,
the survey provides information at the instrument or group level. The survey allows
a laboratory to evaluate the quality of their detection system for hyperbilirubinemia.
The CAP survey has demonstrated bilirubin can cause interference with many
common chemistry tests, including ALT, albumin, alkaline phosphatase, calcium,
creatinine, sodium, glucose, lipase, total protein, creatine kinase, magnesium, phos-
phorus, urea and uric acid. Because bilirubin interference varies by peer group, plat-
form, test methodology and reagent formulation, each laboratory should evaluate the
methods used in their laboratory. The College of American Pathologists Interfering
Substance Survey provides a convenient process to obtain samples containing biliru-
bin and analysis of the results.

3.2 Allen Correction as a Source of Bilirubin Interference

Bilirubin might also cause an interference if the Allen correction is used to handle
hemoglobin absorbance interference. This approach is called the bichromatic correc-
tion and uses a primary wavelength for following the analyte in the reaction and a
secondary wavelength to subtract contributions for other substances, usually hemo-
globin. Use of the bichromatic correction works well in most cases, but it can present
22       3 The Nature of Icteric Interference

a problem if the choice of the secondary wavelength shows absorbance with biliru-
bin. As an example, consider the problem with using two wavelengths for measure-
ment, 400 nm for the primary wavelength and 450 for the secondary wavelength.
Bilirubin’s extinction coefficient is 22,910 at 400 nm and 55,000 at 450 nm. If the
reaction is measured at 400 nm and the absorbance at 450 nm is subtracted from the
total absorbance, there will be an overestimation of the amount of absorbance due to
bilirubin, in this case because the ratio of bilirubin’s absorbance at 450 nm compared
with 400 nm is 2.4–1. The effect is small for very low concentrations of bilirubin, but
as the concentration of bilirubin increases it may have a much greater effect. Such
an occurrence can explain why sometimes the interference from bilirubin may be
negative.

3.3 Bilirubin Interference with Oximetry

Bilirubin may cause an interference in the measurement of the different forms of

hemoglobin by absorbing light at the same wavelengths as that of hemoglobin. The
active portion of the hemoglobin molecule is the heme unit, which is a porphyrin
ring. It contains one atom of iron, which may be in the ferrous (Fe2+) or ferric (Fe3+)
oxidation state. In order to bind oxygen (O2), the iron must be in the ferrous state. On
binding oxygen, there is a tendency for the iron to become oxidized from the ferrous
to the ferric state, which is known as methemoglobin. Normal individuals reduce
the ferric form of iron back to the ferrous form by means of the enzyme NADH-cyto-
chrome-b5 reductase [3]. The blood of normal persons contains approximately 1.5 %
methemoglobin.
Hemoglobin provides a way for the blood to capture oxygen in the lungs and
transport it back to the tissues. Lung diseases, both acute and chronic, demonstrate
decreased concentrations of oxyhemoglobin. Determination of oxyhemoglobin and
percent saturation of hemoglobin by oxygen represent key measurements in the care
of many acutely ill patients and that information is critical in reducing morbidity and
mortality. The percent saturation of hemoglobin by oxygen can be determined by both
pulse oximetry and co-oximetry. Pulse oximetry is an in vivo, non-invasive technique
where the oxy and deoxy forms of hemoglobin are measured through the skin. Co-
oximetry measures oxy and deoxy forms of hemoglobin, but requires a whole blood
sample.
One can distinguish between the different forms of hemoglobin by using oxime-
try. There are two types of analyzers for oximetry, the pulse form and the co-oximeter.
Both of these types of devices work on the principle of using multiple wavelengths
of light to measure the different forms. Both oxyhemoglobin and deoxyhemoglobin
absorb light at the same wavelengths, but the amount of light absorbed varies
depending on the wavelength, which is expressed as the molar absorptivity or molar
extinction coefficient.
 3.3 Bilirubin Interference with Oximetry       23

One could measure the total amount of hemoglobin at 431 nm, which has an extinc-
tion coefficient of 528,600 (extinction coefficients are in units of per Mole/L for a
pathlength of 1 cm). The other forms of hemoglobin absorb light at wavelengths very
close to 431 nm with nearly similar extinction coefficients (Fig. 3.1) [3].

600000
extinction coefficient

500000 hemoglobin absorbance spectra

400000
oxyhemoglobin
300000 deoxyhemoglobin
200000
100000
0
300 400 500 600 700
wavelength (nm)

60000
extinction coefficient

50000 bilirubin
40000
30000
20000
10000
0
-10000
300 350 400 450 500 550 600 650 700
wavelength (nm)

Fig. 3.1: Absorbance spectra for oxyhemoglobin, deoxyhemoglobin and bilirubin.

The extinction coefficients are too high, near 431 nm, to be very useful; however, one
can measure the hemoglobin molecule absorption at slightly above 500 nm to dis-
criminate among the varying hemoglobin forms. At a wavelength of 555 nm, deoxyhe-
moglobin has an extinction coefficient of 54,520, oxyhemoglobin an extinction coeffi-
cient of 36,815, carboxyhemoglobin and methemoglobin have extinction coefficients
that are distinct from these other species. The absorption maximum in the 500–700 nm
region for oxyhemoglobin is 578 nm, while 621 is the absorption maximum for meth-
emoglobin, with carboxyhemoglobin showing absorbance maxima at 541 nm and 577
nm [4]. By measuring the peaks for deoxyhemoglobin, oxyhemoglobin, carboxyhe-
moglobin and methemoglobin at three other wavelengths, one can distinguish among
the varying forms, because the peaks occur at different wavelengths and the extinc-
tion coefficient are different.
24       3 The Nature of Icteric Interference

3.3.1 Co-oximetry Interference

By measuring the absorbance at four carefully chosen wavelengths, one can discern
among the four species of hemoglobin. In some methods, one measures the total
hemoglobin as cyanohemoglobin, so a fifth wavelength is added. To find the concen-
trations of the various species one solves the set of simultaneous equations given by

C1 "1 1 + C2 "2 1 + C3 "3 1 + C4 "4 1 + C5 "5 1 = A1 ,

C1 "1 2 + C2 "2 2 + C3 "3 2 + C4 "4 2 + C5 "5 2 = A2 ,
C1 "1 3 + C2 "2 3 + C3 "3 3 + C4 "4 3 + C5 "5 3 = A3 ,
C1 "1 4 + C2 "2 4 + C3 "3 4 + C4 "4 4 + C5 "5 4 = A4 ,
C1 "1 5 + C2 "2 5 + C3 "3 5 + C4 "4 5 + C5 "5 5 = A5 , (3.2)

where C represents the concentration of each hemoglobin species, ε represents the

extinction coefficient for that species at that particular wavelength, λ, and Aλ rep-
resents the absorbance at that particular wavelength. The equations represent five
equations and five unknowns, and can be directly solved for the concentration of
each species. The problem is that bilirubin can also absorb light in the region of the
selected wavelengths and thus masquerade as one of the hemoglobin species.
An example of this type of interference occurs in oximetry where it has been noted
that pulse oximetry is often not affected by bilirubin, but co-oximetry is [5]. Beall and
Moorthy reported a case that is interesting. A patient was treated for nodular scleros-
ing Hodgkin’s lymphoma with several rounds of chemotherapy and whole body irra-
diation followed by autologous bone marrow transplantation. He developed hepatic
venous occlusive disease, which resulted in hepatic failure with bilirubin concentra-
tions ranging between 633–770 μmol/L and respiratory failure requiring intubation
and mechanical ventilator support. The patient was monitored with continuous pulse
oximetry and blood gas analysis. The blood gas analysis included determination of
hemoglobin oxygen saturation by co-oximetry.
The staff were able to maintain the patient’s arterial pO2 in a range of 92–
133  mmHg, which should be sufficient for normal oxygenation of hemoglobin;
however, the results from the co-oximeter of the blood gas unit were lower than
expected, with hemoglobin saturations of 88–93 % (IL 282 Co-oximeter from Instru-
mentation Laboratory). These results for the co-oximeter do not correlate well with
the arterial pO2, obtained on the same specimen as used for the co-oximetry deter-
mination of hemoglobin saturation. Further evidence helped to discern this discrep-
ancy. The staff recorded values for the hemoglobin saturation from pulse oximetry
using two different devices, the Nellcor N100c and the Ohmeda Biox 3700® [5]. The
results from pulse oximetry gave results of hemoglobin saturation of 98–99 %, which
is within the normal range and consistent with the arterial pO2 measurements.
Additional information from the co-oximeter showed a slight increase in the frac-
tion of hemoglobin represented as carboxyhemoglobin, in this case the values ranged
 3.3 Bilirubin Interference with Oximetry       25

from 2.4–2.9 % and increased fractions measured as methemoglobin, in this case the
values ranged from 3.2–11.9  % [5]. Differences besides the specimen, transcutane-
ous nonvasive for pulse oximetry and an arterial blood sample for co-oximetry are
required to explain these differences.
The basis for measurements of oxygen saturation depends on the differences in
extinction coefficients at various wavelengths for deoxyhemoglobin and oxyhemo-
globin. Oxyhemoglobin absorbs less light in the longer wave, red region, near 660
nm, than does deoxyhemoglobin [6]. Because deoxyhemoglobin absorbs a consider-
able amount of light in the 660 nm range, associated with red color, its color does
not appear as red as oxyhemoglobin. The color of the substance is determined by the
wavelengths that the light reflects. If a substance, such as oxyhemoglobin, absorbs
more light in the blue, yellow and green regions of the spectrum, it reflects more red
light and therefore looks red. Carboxyhemoglobin, the form of hemoglobin which has
combined with carbon monoxide, CO, instead of oxygen, absorbs even less light in the
red region, and thereby appears even brighter red than oxyhemoglobin.

3.3.2 Pulse Oximetry

A pulse oximeter works by synchronizing the absorbances with the arterial pulse [6].
In the skin, the capillaries dilate in response to the pulse and synchronizing the meas-
urements with the pulse gives the most reliable results. In pulse oximetry, absorb-
ances are measured at two absorbances, one at 660 nm, at which the absorbance of
oxyhemoglobin is less than that of deoxyhemoglobin, and at a much longer wave-
length. The secondary measurement of the absorbance is typically done at a wave-
length between 815 and 940 nm. In this region of the spectrum, the absorbance of the
oxyhemoglobin is slightly greater than that of deoxyhemoglobin [6]. Pulse oximeters
are normalized to account for variation in skin absorbance from person to person.
By examining a ratio of the absorption of light in the red region (R) and the infra-
red region (IR), pulse oximetry can account for variations from person to person. The
ratio is calculated as

R
Ratio (660 : 815 − 940) = . (3.3)
IR

Manufacturers calibrate the pulse oximeters empirically by observing the ratio from a
group of normal volunteers and comparing them with results obtained by a co-oxime-
ter [6]. Subjects are asked to breathe hypoxic gas mixtures so that a calibration curve
may be generated. Around an oxyhemoglobin saturation of 85 %, the absorbance at
the red and infrared wavelengths measures are about the same, the ratio of these two
absorbances would be 1.0. Using a standard or calibration curve provides for a robust
relationship because there is a large difference in the values for the ratio; at 99  %
26       3 The Nature of Icteric Interference

oxygenation, the ratio is around 0.4, while at 50 % it is around 2.0 [6]. Bilirubin does
not absorb much light at either of the two wavelengths chosen for measurement with
pulse oximeters; therefore, bilirubin is fairly unlikely to cause an interference with
this methodology.
Pulse oximeters do have limitations though. Because pulse oximeters measure
light at only two wavelengths, they can only distinguish two different species of
hemoglobin. At 660 nm, methemoglobin absorbs light in a fashion similar to oxy-
hemoglobin [6]. Carboxyhemoglobin absorbs similarly to oxyhemoglobin as well.
Caution must be used in depending on pulse oximetry alone, because it cannot
provide information about these two important species of hemoglobin. Frequently,
acutely ill patients have their oxygen saturation determined by both methods.
The IL 282 Co-oximeter (Instrumentation Laboratories) used in the case measured
light at four wavelengths, 535, 585, 594 and 626 nm. Even though bilirubin exhibits
its absorption peak at 450 nm, the bilirubin absorption band extends its tail into the
535–585 nm range, adding its absorption to that measured for hemoglobin. Based on
the set of equations used to calculate deoxyhemoglobin, oxyhemoglobin, carboxyhe-
moglobin and methemoglobin, the presence of bilirubin in the specimen will add to
the apparent absorbance of carboxyhemoglobin and methemoglobin, which is why
the carboxyhemoglobin and methemoglobin were overestimated in the co-oximeter.
In the co-oximeter, the percent hemoglobin saturation by oxygen is determined by
dividing the calculation fraction of oxyhemoglobin by the total hemoglobin. Total
hemoglobin is determined from the fractions for deoxyhemoglobin, carboxyhemo-
globin and methemoglobin. Because the calculations for the co-oximeter are per-
formed differently than for the pulse oximeter, bilirubin may cause an interference
with co-oximeters not seen with pulse oximeters. Many co-oximeters today use an
increased number of wavelengths, so that they can calculate the contribution to
absorbance from bilirubin and subtract it from the total, thus avoiding bilirubin inter-
ference.
Pulse oximetry is utilized by transmitting light through skin, usually a finger.
New techniques in oximetry provide a noninvasive method assessing cerebral oxygen
saturation using dual-wavelength near-infrared spectrophotometry (NIRS) [7]. NIRS
has been used for monitoring cerebral oxygenation during carotid endarterectomy,
acute heart failure and orthotopic liver transplantation [7]. NIRS works by measuring
absorbance in the cerebral tissue at 733 and 809 nm [7]. The wavelength at 733 nm
provides a measure of deoxygenated hemoglobin, while the wavelength at 809 nm
provides a sum of deoxygenated and oxyhemoglobin.

3.3.3 Cerebral Oximetry

Cerebral oximetry employs infrared light which can penetrate the tissues. The selec-
tion of the two wavelengths, one near 730 nm and the other near 810 nm allows for
 3.3 Bilirubin Interference with Oximetry       27

maximal tissue penetration, with the light being scattered back from the skin, the
skull, and up to 15 mm of cerebral tissue [8]. The operation of orthotopic liver trans-
plantation is divided into four phases: dissection, anhepatic, reperfusion and end.
Patients requiring liver transplantation are likely to be jaundiced. During the reperfu-
sion phase a rise in the cerebral oxygen saturation is expected; however, in patients
with elevated bilirubin this increase in the cerebral oxygen saturation is blunted [7].
Examination of the hemoglobin oxygen saturation showed no effect of bilirubin
in arterial or venous samples; however, examination of the cerebral oxygen satura-
tion demonstrated a negative interference with increasing concentrations of blood
bilirubin [7]. It is suspected that rather than bilirubin deposited in the cutis, it is
biliverdin, the oxidative product of bilirubin, deposited in the cutis that is interfer-
ing with the cerebral oxygen saturation measurement [7]. The absorption of light by
biliverdin changes depending on the orientation that the molecule takes: in nonpolar
solvents it takes on a ring form and absorbs light in the ultraviolet region, around 350
nm, but in polar solvents or attached to proteins, it takes on a straight chain form with
four connected pyrrole groups and shifts its light absorbance to the visible region [9].

3.3.4 Interference with Methemoglobin

Falsely elevated values for methemoglobin, caused by bilirubin when using co-oxi-
metry, can cause erroneous diagnosis of methemoglobinemia. Methemoglobinemia is
defined when the hemoglobin in the red blood cells possess greater than 1 % hemo-
globin, and even though a small amount of methemoglobin is normal and does not
pose a health risk, elevated fractions of methemoglobin decrease the oxygen carry-
ing and delivery capacity of the blood and may pose a risk, resulting in a functional
anemia [10]. Methemoglobin is formed when the iron in normal hemoglobin is oxi-
dized from the ferrous to the ferric form. When this occurs the skin become cyanotic
or blue in appearance; neurologic and cardiac symptoms begin when the fraction
increases above 15 % because of hypoxia and death occurs for fractions of 70 % or
above [10].
The normal method for reduction of ferric iron to ferrous iron involves the
adenine dinucleotide (NADH)-dependent reduction, called the diaphorase pathway,
the enzyme cytochrome b5 reductase playing the major role in the pathway [10]. The
diaphorase pathway reduces 95–99 % of the methemoglobin, while another enzyme
system, the nicotinamide adenine dinucleotide phosphate (NADPH)-dependent
methemoglobin reduction accounts for the rest; this enzyme uses glutathione and
glucose-6-phosphate dehydrogenase (G6PD) to reduce methemoglobin to hemo-
globin [10]. Hereditary methemoglobinemia is a rare condition and is most commonly
found among Native American tribes such as the Navajo, Athabascan Alaskans and
the Yakutsk people of Siberia; it involves a deficiency in cytochrome b5 reductase [10].
Most cases of methemoglobinemia are acquired through exposure to drugs or toxins,
28       3 The Nature of Icteric Interference

such as benzocaine, with infants, and especially premature infants being the most
susceptible [10].
Acquired methemoglobinemia requires treatment and is caused by several recog-
nized agents, such as nitrites, nitrates, chlorates, and dapsone [10]. There are several
tests used in the diagnosis and management of methemoglobinemia, including pulse
oximetry, co-oximetry, potassium cyanide test, as well as complete blood count. Co-
oximetry is the best way to quantify the fraction of methemoglobin, but bilirubin can
cause falsely elevated results, unless the co-oximeter has additional wavelengths that
can detect the presence of bilirubin. Once the diagnosis is suspected, a search for the
cause prompts the ordering of additional tests such as a complete blood count, hemo-
globin M by electrophoresis, mass spectrometry and DNA sequencing.
Hemoglobin Barts is a variant hemoglobin where the alpha (α) chains are absent
and the gamma (γ) chains form tetramers and it is associated with hydrops fetalis,
occurs in the cord blood of healthy neonates and neonates with hemoglobin S,
a-thalassemia and other hemoglobinopathies [11]. High-performance liquid chroma-
tography (HPLC) is the most common method used to identify hemoglobin variants
and there are several reports that the presence of hemoglobin can produce a peak in
the HPLC procedure that is mistaken for hemoglobin Barts [11, 12]. Hemoglobin Barts
elutes within the first 0.5 minutes after injection and a peak identified as bilirubin
appeared in 0.2 minutes from the time of injection [11]. In a series of 8,000 hemo-
globin chromatograms, 90 chromatograms were identified with an unusually large
peak in the first 0.2 minutes after injection, and of these 90 patients, 86 had hemo-
globin SS, one had hemoglobin AS and three had hemoglobin AA. The height of
these peaks ranged from 3 % to more than 46.5 %. When these peaks were compared
against the concentration of bilirubin, a positive interference was revealed with a
correlation coefficient of 0.87 [11]. When serum samples with more than 513 μmol/L
of bilirubin were taken and analyzed by chromatography using the HPLC analyzer,
the chromatograms showed a single peak at 0.2 minutes of injection [11]. This study
clearly elucidated that bilirubin could cause an interference with the HPLC method
for hemoglobin variants and showed several useful techniques to prove the inter-
ference.

3.4 Chemical Reactions as a Cause of Bilirubin Interference

The previous mechanisms of interference caused by bilirubin described above have all
depended on the physical chemical properties of bilirubin or its breakdown product,
biliverdin. The mechanism dealt with light absorption or elution in chromatography.
In addition to its physical chemical properties, bilirubin may cause interference by
means of its chemical reaction with reagents involved in the methods for determina-
tion of analytes.
 3.4 Chemical Reactions as a Cause of Bilirubin Interference       29

3.4.1 Bilirubin Reaction with Creatinine Methods

One well-studied interference is bilirubin’s effect on creatinine determinations. Cre-

atinine, a breakdown product of creatine, is an important analyte for the assess-
ment of renal function. It is widely used and one of the most frequently ordered tests
among clinical chemistry test panels. Creatinine measurement, as determined by the
Jaffe reaction (alkaline picrate), suffers from interferences from many different com-
pounds, especially those with strong ketone groups or cepha rings [13].
Before the kinetic method for the determination of creatinine was instituted, the
protein of serum was separated from the filtrate and the creatinine was determined as
an endpoint procedure. The effects of bilirubin on creatinine determinations by the
kinetic method have always been mixed, with some methods not demonstrating an
interference, while others showing a strong negative interference [14].
Bilirubin is a very reactive compound with a relatively strong absorbance near
the wavelengths used in the Jaffe reaction [14]. In the kinetic method, an absorbance
reading is taken immediately after mixing of the sample with picric acid and NaOH,
followed by at least one more absorbance reading taken sometime later in the reac-
tion. The advantage of the kinetic method is that the first absorbance reading can act
as a blank for the reaction and one does not need to wait until an endpoint is reached
to calculate the creatinine concentration. If bilirubin interfered simply by its absorb-
ance alone, the first kinetic reading would subtract out this absorbance; further, an
early absorbance reading would not cause a negative interference, thus there must
be some reaction occurring with bilirubin that would give rise to a negative interfer-
ence [14].
In the alkaline picrate method for the determination of creatinine, NaOH mixed
with picric acid makes a strongly alkaline solution. The actual chromogen in the reac-
tion is the picrate, which when it reacts with creatinine, shifts the picrate absorbance
band approximately 10 nm, which is sufficient to measure an absorbance change
[15]. The concentration of creatinine is determined by the change in absorbance near
500 nm and this change in absorbance is a positive number. In an alkaline medium,
bilirubin can be oxidized to form biliverdin, which demonstrates a major decrease
in the change in absorbance as measured at 510 nm (biliverdin absorbs more light
at 630 nm than does bilirubin, which absorbs more at 510 nm) [14]. In the initial part
of the reaction, bilirubin absorbs light near 500 nm, the wavelength for the assay for
creatinine. As the reaction proceeds in time, creatinine reacts with picrate and the
absorbance for that species increases, but bilirubin is converted to biliverdin and that
absorbance decreases. The net change in absorbance decreases and a depression in
the value for creatinine is reported.
The effect of bilirubin depends on its form. Bilirubin can be either conjugated or
unconjugated. The conjugated bilirubin has glucuronide sugars attached to it, which
make it soluble in water and it is typically measured as direct bilirubin. Unconjugated
bilirubin is not soluble in water and is attached to albumin in the plasma. Unconju-
30       3 The Nature of Icteric Interference

gated bilirubin is typically measured as the difference between total bilirubin and
direct bilirubin. Unconjugated bilirubin is liable to oxidation to biliverdin in alkaline
solutions and gives rise to a negative interference for creatinine. Conjugated bilirubin
is less likely to oxidize to biliverdin, and thereby less likely to cause a negative inter-
ference [16]. The choice of method also has an effect. If one were to dialyze or ultrafil-
trate the sample first, and perform the reaction on the filtrate (as done with some older
methods), then there would not be any unconjugated bilirubin present in the filtrate
and therefore no negative interference [17, 18]. The concentration of NaOH, temperature
and timing kinetics, i.e., the time between the first and last absorbance readings will
affect the bilirubin interference as well, and is a source for the varied results seen with
this analyte; a greater decrease is seen with increasing concentrations of NaOH [16].
One would think that a simple solution to the bilirubin interference with the Jaffe
method for creatinine would be to replace the Jaffe method with an enzymatic method
for the determination; however, the enzymatic method for creatinine has demon-
strated negative interference for creatinine [19].
There are three major reactions for the determination of creatinine utilizing an
enzymatic reaction [4]. The first two methods use creatininase.

Creatinine + H2 O → Creatine
Creatine + ATP → Creatine phosphate + ADP
ADP + Phosphoenolpyruvate → Pyruvate + ATP
Pyruvate + NADH + H + → Lactate + NAD + (3.4)

This method has not been widely accepted.

The second method to use creatininase produces hydrogen peroxidase.
Its reactions are as follows:

Creatinine + H2 O → Creatine
Creatine + H2 O → Sarcosine + Urea
Sarcosine + O2 + H2 O → Formaldehyde + Glycine + H2 O2
Phenolderivative + 4-Aminoantipyrine + H2 O2 → H2 O + Colored product (3.5)

This method is fairly popular as is the next one, based on the use of creatinine imi-
nohydrolase:

Creatinine + H2 O → N -Methylhydantoin
N -Methylhydantoin + ATP + H2 O → N -Carbamoylsarcosine + ADP + Phosphate
N -Carbamoylsarcosine + H2 O →NH3 + CO2 + Sarcosine
Sarcosine + O2 + H2 O → Formaldehyde + Glycine + H2 O2
Aniline dye + 4-Aminoantipyrine + H2 O2 → H2 O + Colored product (3.6)
 3.4 Chemical Reactions as a Cause of Bilirubin Interference       31

These two methods using hydrogen peroxide and peroxidase as the detection system
experience a negative bilirubin interference [20]. The magnitude depends on both
the concentrations of creatinine and of bilirubin. The absorbance spectrum of biliru-
bin, with a peak near 460 nm, overlaps the absorbance band of the colored product
(Trinder chromophore) [20]. In addition, it has been speculated that bilirubin may
react with one of the peroxidase reaction intermediates, decreasing the concentration
of chromophore produced and its net absorbance [20].

3.4.2 Bilirubin Reactions with Peroxidase Methods

Other evidence suggests that bilirubin may cause an interference with peroxidase
methods by five possible mechanisms: as a substrate for peroxidase; absorbance at
the same wavelengths as the product; intercepting and removing an oxidase interme-
diate; reacting with a peroxidase intermediate, such as hydrogen peroxide; bleach-
ing of the final reaction color [21]. Bilirubin does not appear to have an high enough
Km to displace creatinine from peroxidase. The Trinder chromophore does appear to
be stable in the presence of bilirubin. Bilirubin does absorb at 505 nm, the absorp-
tion band for the Trinder chromophore, but in this case, it could potentially provide a
positive interference, but even more important, the coupled enzyme reaction follows
a kinetic reaction and proper blanking of the reaction would eliminate this interfer-
ence. It is most likely that bilirubin reacts with the hydrogen peroxide. Hydrogen
peroxide is a strong oxidizer and bilirubin is readily oxidized to biliverdin, bilirubin
acting as a strong anti-oxidant. Bilirubin could then remove hydrogen peroxide from
the detection system. Control of the bilirubin interference often uses bilirubin oxidase
or potassium ferricyanide.
Other common chemistry methods utilize the Trinder chromophore, or a hydro-
gen peroxide intermediate, in their detection system. Lipase acts on triglycerides to
produce fatty acids and glycerol; in turn, glycerol oxidase acts on glycerol to produce
dihydroxyacetone and hydrogen peroxide [22]. Uricase acts on uric acid to produce
allantoin and hydrogen peroxide [4]. Cholesterol oxidase acts on cholesterol to
produce hydrogen peroxide as well [23]. These three methods share the production
of hydrogen peroxide as an intermediate species with detection of hydrogen peroxide
through peroxidase-coupled assays as a common step with the enzymatic determina-
tion of creatinine [23]. Bilirubin interferes with these methods by means of the same
mechanisms as described for creatinine. The sensitivity of commercial methods utiliz-
ing peroxidase-coupled assays varies depending on the choice of wavelength, substi-
tution of an aniline rather than phenol derivative, the presence of ferrocyanide in the
reagent mix, and other reaction conditions [23]. Some assays are highly sensitive to
bilirubin, the interference becoming significant at concentrations below 43 μmol/L,
ranging up to high concentrations of bilirubin, e.g., 684 μmol/L [23]. The interfer-
ences are mixed as well, some showing negative interference and others showing
32       3 The Nature of Icteric Interference

positive interference, again suggesting that more than one mechanism of interference
may be operating at a time [23].
Not only endogenous substances, but also exogenous analytes suffer from biliru-
bin interference. One method for measuring salicylate is to use the Trinder method, in
which ferric ions (Fe3+) react with salicylate to form a purple color which can be meas-
ured at 540 nm. There is a positive interference related to an increase in absorbance at
the chosen wavelength with bilirubin and is caused by the direct reaction of the ferric
ions with bilirubin [24]. Use of dual wavelength blanking at 700, 750 and 800 nm did
not have an effect on the interference [24].

3.5 References
[1] Glick MR, Ryder KW, Glick SJ. Interferographs User’s Guide to Interferences in Clinical Chemistry
Instruments. 2nd ed. Indianapolis, IN, USA, Science Enterprises, Inc, 1991.
[2] Kaplan LA, Pesce AJ. Clinical Chemistry: Theory, Analysis, Correlation. 5th ed. St. Louis, MO,
USA, Mosby Elsevier, 2010.
[3] McPherson RA, Pincus MR. Henry’s Clinical Diagnosis and Management by Laboratory Methods.
21st ed. Philadelphia, PA, USA, Saunders Elsevier, 2007.
[4] Burtis CA, Ashwood ER. Tietz Textbook of Clinical Chemistry. 3rd ed. Philadelphia, PA, USA, W.B.
Saunders Company, 1999.
[5] Beall SN, Moorthy SS. Jaundice, oximetry, and spurious hemoglobin desaturation. Anesth Analg
USA 1989,58,806–807.
[6] Mendelson Y. Pulse oximetry: theory and applications for noninvasive monitoring. Clin Chem
USA 1992,38,1601–1607.
[7] Madsen PL, Skak C, Rasmussen A, Secher NH. Interference of cerebral near-infrared oximetry in
patients with icterus. Anesth Analg USA 2000,90,489–493.
[8] Samra SK, Dorje P, Zelenock GB, Stanley JC. Cerebral oximetry in patients undergoing carotid
endarterectomy under general anesthesia. Stroke 1996,27,49–55.
[9] New Scientist. Sep. 10, 1987, p. 35.
[10] Denshaw-Burke M. Methemoglobinemia: Emedicine, Medscape article 204178.
[11] Howanitz PJ, Kozarshi TB, Howanitz JH, Chauhan YS. Spurious hemoglobin Barts caused by
bilirubin. Am J Clin Pathol USA 2006,125,608–614.
[12] Kar R, Sharma CB. Bilirubin peak can be mistaken as Hb Bart’s or Hb H on high-performance
liquid chromatography. Hemoglobin UK 2011,35,171–174.
[13] Kroll MH, Elin RJ. Mechanism of cefoxitin and cephalothin interference with the Jaffe method for
creatinine. Clin Chem USA 1983,29,2044–2048.
[14] Watkins RE, Feldkamp CS, Thibert RJ, Zak B. Interesting interference in a direct serum creatinine
reaction. Microchemical J USA 1976,21,370–384.
[15] Kroll MH. Some observations on the reaction mechanism of cefoxitin and cephalothin with
picrate. Microchem J USA 1990,42,241–249.
[16] Knapp ML, Hadid O. Investigations into negative interference by jaundice plasma in kinetic Jaffe
methods for plasma creatinine determination. Ann Clin Biochem UK 1987,24,85–97.
[17] Soldin SJ, Henderson L, Hill JG. The effect of bilirubin and ketones on reaction rate methods for
the measurement of creatinine. Clin Biochem USA 1978,11,82–86.
 3.5 References       33

[18] da Fonseca-Wollheim F, Heinze K-G, Lomsky K, Schreiner H. Serum ultrafiltration for the
elimination of endogenous interfering substances in creatinine determination. J Clin Chem Clin
Biochem Germany 1988,26,523–525.
[19] Schoenmakers CH, Kuller T, Lindemans J, Blijenberg BG. Automated enzymatic methods for
creatinine measurement with special attention to bilirubin interference. Eur J Clin Chem Clin
Biochem Germany 1993,31,861–863.
[20] Crocker H, Shephard MDS, White GH. Evaluation of an enzymatic method for determining
creatinine in plasma. J Clin Pathol UK 1988,41,576–581.
[21] Witte DL, Brown LF, Feld RD. Effects of bilirubin on detection of hydrogen peroxide by use of
peroxidase. Clin Chem USA 1978,24,1778–1782.
[22] Klotzsch SG, McNamara JR. Triglyceride measurements. A review of methods and interferences.
Clin Chem USA 1990,36,1605–1613.
[23] Spain MA, Wu AHB. Bilirubin interference with determination of uric acid, cholesterol, and
triglycerides in commercial peroxidase-coupled assays, and the effect of ferrocyanide. Clin
Chem USA 1986,32,518–521.
[24] Broughton A, Marenah C, Lawson N. Bilirubin interference with a salicylate assay performed on
an Olympus analyser. Ann Clin Biochem UK 2000,37,408–410.
4 The Nature of Lipemic and Turbidity Interferences
The interferences from lipemia are fundamentally different from interferences associ-
ated with icterus (bilirubin) and hemolysis. Assays reported to be affected by lipemia
cover a wide range of different analytes, including cholesterol, hemoglobin, total
protein, albumin, lactate dehydrogenase, amylase, electrolytes, and bilirubin [1]. In
lipemia, chylomicrons and VLDL particles scatter light, producing cloudiness or tur-
bidity similar to that seen in milk. Lipemia may interfere in any assay that uses the
transmission of light as part of the detection scheme [1].

4.1 Types of Interferences

Lipemia causes three types of errors or interferences: electrolyte exclusion, partition-

ing and light scattering. Specimens that are lipemic have an increased concentration
of lipoproteins in the blood, but they form microemulsions, which exclude water, the
aqueous solvent. For a specific volume of blood, normally about 7 % is occupied by
the lipid fraction. In lipemia, the lipid fractions occupy a greater percentage of the
volume, and the aqueous phase a lesser percentage. Small ions, both positive and
negative, are polar and excluded from the lipoproteins, and even though the concen-
tration of the ions in the polar fraction is undisturbed, for a given volume of specimen
aspirated into an analyzer, the total number of ions is decreased compared with a
normal sample, because the total number of ions is equal to the concentration of the
ions in the polar fraction times the volume of the polar fraction [2]. If one separates the
polar fraction from the total specimen, as occurs using an ion-selective electrode in
sampling without dilution, then the decreased polar fraction would not have an effect
on accuracy, but when the aspirated sample is diluted, as occurs in flame photometry,
atomic absorption, and indirect potentiometry (the electrode method for determining
electrolytes on most automated clinical chemistry analyzers), then the presence of the
lipoproteins interferes with the volume and the reported results are decreased (nega-
tive interference) [2]. Chylomicrons and VLDL particles are most prone to this type of
error, because they occupy greater volumes than LDL or HDL particles [2].
Nonpolar substances, such as steroid hormones, lipophilic drugs such as dilan-
tin, and phenobarbital may partition themselves into the nonpolar fraction of plasma
which includes the chylomicrons and VLDL particles. The movement of these lipo-
philic analytes into chylomicrons and VLDL particles may cause an interference if an
extraction step is utilized [2]. Significant errors have been reported with radioimmu-
noassay methods for testosterone and progesterone by this mechanism [2]. Postpran-
dial lipemia has been shown to be particularly prone to causing this type of interfer-
ence [2].
36       4 The Nature of Lipemic and Turbidity Interferences

4.2 Lipemia Causes Turbidity

Lipemia is the most common form of turbidity in serum and plasma samples, and it
may appear in whole blood samples as well, even though it may be very difficult to
ascertain that turbidity is present because it is obscured by the red cells in the speci-
men. Lipemia usually presents as a milky-appearing sample. The source of the lipemia
may be endogenous, as occurs following a meal (postprandial chylomicronemia) and
disease, or exogenous, as occurs from providing lipid nutrients with total parenteral
nutrition and the use of such agents as Intralipid®. The turbidity of lipemic samples
interferes with measurements of absorbance. The turbidity of lipemic samples results
from light scattering. Light scattering itself has many characteristics that it shares
with absorbance, but actually results from a different type of physical chemical phe-
nomenon.
There are several factors that play a part in giving rise to turbidity from lipemic
samples, most of them relate to the size of the lipemic particles. Light scattering
behaves in a slightly different manner from the absorbance observed for bilirubin or
other endogenous or exogenous chemicals or substances found in the blood. To get
an idea of what the turbidity from lipemia looks like, one needs to observe the rela-
tionship between absorbance and wavelength.

1.4
1.2
lipemic absorbance
1.0
0.8
absorbance

0.6
0.4
0.2
0
300 400 500 600 700
wavelength (nm)

Fig. 4.1: Apparent absorbance caused by lipemic specimens.

Fig. 4.1 shows the average absorbance for 11 lipemic plasma samples. These samples
were grossly lipemic and diluted 1:20 before being placed in a cuvette in a Beckman
Spectrophotomer. Several things are noteworthy concerning the relationship between
absorbance and wavelength. First, absorbance occurs across all wavelengths. Lipemic
turbidity thus differs from absorbance by a chromophore, because the typical chromo-
phore demonstrates one or more clear bands of absorbance. A chromophore’s bands
of absorbance are clearly distinguishable from the wavelengths surrounding them.
Clearly distinguishable bands of absorbance are readily apparent by looking at the
bands of absorbance found for bilirubin or hemoglobin (see Fig. 3.1). Because of the
apparent absorbance over all the wavelengths of the visible spectrum, the lipemic
 4.3 Lipemia Interference Mechanisms       37

turbidity will not show a definite color, but instead will appear white in color and
milky in consistency. The lipemic particles cause light scattering (and thus turbidity)
whose nature depends less on the actual chemical substances but more on the size
and shape of the particles. The second thing to note is that there is a distinctive bump
in the absorbance spectrum between 400 and 420 nm. This bump is not caused by the
lipemia of the samples, but rather represents the absorbance by bilirubin present in
the serum samples.
The third thing to recognize looking at the apparent absorbance by lipemia is that
the absorbance is relatively low at long wavelengths, around 700 nm, slowly increases
in an almost linear fashion until about 500 nm, then increases more rapidly between
500 and 320 nm in a curvilinear manner. In the ultraviolet range, with wavelengths
shorter than 320 nm, the apparent absorbance markedly increases and eventually, as
one decreases the wavelength, runs off the absorbance scale.
Turbidity from lipemia interferes with chemical methods by mimicking absorb-
ance of the chromophore. The degree of interference varies widely and is depend-
ent on the wavelength used, the amount of dilution of the sample, the methodology
used and whether or not blanking is effective [3]. Obviously, methods that measure
absorbance with shorter wavelengths will be more affected than those with longer
wavelengths, especially if they do not have an effective means to blank the sample.

4.3 Lipemia Interference Mechanisms

To evaluate the susceptibility of methods to interferences from icterus or hemolysis,

it is appropriate to prepare samples with added bilirubin or hemoglobin, respec-
tively [3]. By contrast, the lack of readily available, standardized materials to produce
lipemic samples complicates the evaluation of lipemia. Glick et al. added Intralipid®,
a synthetically produced emulsion for intravenous administration, to serum to simu-
late lipemic samples [3]. Bornhorst et al. showed that samples with added Intralipid®
do not perfectly mimic lipemic samples [17]. Thus, native lipemic patient samples
have falsely low results for ceruloplasmin, prealbumin, and transferrin measured
by immunoturbidimetry, whereas simulated lipemic samples prepared by adding
Intralipid® do not.

4.3.1 Light Scattering

To understand this discrepancy and appreciate its potential occurrence with all light-
based methodologies in the clinical laboratory, one must review the features of light
scattering pertinent to clinical laboratory instrumentation and the physical chemical
differences between naturally lipemic and Intralipid®-supplemented samples.
38       4 The Nature of Lipemic and Turbidity Interferences

When electromagnetic radiation in the form of light interacts with matter, such as
lipid particles, a dipole moment is induced in the particles [4]. The magnitude of the
dipole moment is proportional to the strength of the electric field and the polariz-
ability of the particles [4]. The particles do not need to reflect the light; instead, the
phenomenon arises from the difference in the refractive index between the particles
(solute) and the solvent [5]. The polarizability is related to the refractive index, which
for most lipid particles is around 1.3 [4].
In spectrophotometry, Beer’s law,

A = abC (4.1)

(ϵ being the molar absorptivity, b the pathlength of the cuvette, and c the concentra-
tion), relates the absorbance (A) to the concentration.
In turn, absorbance is the negative log measure of the transmittance, or
A = log(Io/I), where Io is the intensity of the incident light and I is the intensity of
the light as it leaves the cuvette and strikes the detector. From the perspective of the
center of the cuvette, the angle toward the direction of the light source is 0 degrees,
and the angle toward the direction of the detector is 180 degrees. In absorbance spec-
trophotometry, one assumes that the decrease in intensity of the light striking the
detector is attributable to light being absorbed by the sample.
In the presence of light scattering, the light scatters in all directions, but the inten-
sity varies according to the angle (the angle between the line of observation and the
X-axis) and the expression + cos2 (θ) gives the relative intensity [4]. One can measure
the diminution of the incident beam of light caused by light scattering, Io − I, where
I is the light scattered. By analogy with absorption spectroscopy, turbidity is defined
as log(Io/I) [4]. As long as the intensity of light scattering remains relatively low, the
turbidity relates to the concentration of particles in a linear fashion and can be used
analytically in instruments that measure absorbance. Light scattering interferes with
absorbance spectrophotometric methods by diminishing the light intensity by this
mechanism.
Light scattering interferes with nephelometric and turbidimetric methods by
mimicking the analyte-reagent product, analogous to the protean interferences of
bilirubin in spectrophotometric methods [6]. Empirical observations have shown that
the intensity of light scattering is affected by the number of particles suspended in
solution, the size of the particles, the dependence of the refractive index on particle
concentration, and although it is not readily apparent, the wavelength of the light
[7]. The Rayleigh ratio, Rθ, represents the relative ratio of the scattered light, taking
the angle of scatter and the distance of the observer from the scattering particles into
consideration, and it is directly proportional to I/Io.
The relative intensity of the scattered light becomes proportional to the particle
molecular weight (M), particle concentration (c), the refractive index (n), the refrac-
 4.3 Lipemia Interference Mechanisms       39

 
∂n
tive increment , and the inverse of the wavelength of the light (λ) raised to the
∂c
fourth power:
 
2 ∂n
R = Kn −4 Mc, (4.2)
∂c

where K is a constant [7]. The relative intensity is dependent on the molecular weight
and the concentration of the number of particles. When the number of particles, as
given by the concentration, is low, then there will be a low intensity of scattering; that
is why most serum and plasma samples do not appear turbid.
For molecules that are small compared to the wavelength of the incident light,
each molecule acts as a dipole oscillating with the frequency of the incident light and
the molecule will emit light of the wavelength in all directions [8]. The amplitude of
the oscillating dipole depends on the difference in the refractive index of the solution
(n̂) relative to that of the surrounding solvent (n0 
̂ ) and the intensity of the scattered
light becomes:
 2
2 (n̂ − n̂0 ) · n̂ mM
Iscattered = Iincident · 4 ⁄ m/ , (4.3)
V Nd 2 4

where d is the distance between the sample and the observer, m is the dissolved
mass in the volume V, M is the molar mass [8] and N is Avagadro’s number, some-
times referred to as Avagadro’s constant. Note that there is a strong dependency on
the wavelength of the incident light and that the amount of light scattered increases
dramatically as one moves to shorter wavelengths. This relationship holds only for
particles that are much smaller than the wavelength of the incident light [8]. Thus,
it would hold for VLDL particles, but not for chylomicrons. Chylomicrons, which are
much larger particles, follow Mie scattering [5]. The scattering becomes more intense
and much more prominent in the forward direction, and the intensity of the scattered
light becomes less dependent on the wavelength [5]. A population of particles of the
same size scatters light in colors, but for a mixed population, the scattered light will
appear white [5].
The amount of light scattering depends on the size of the particle and the wave-
length. Plasma proteins, as long as they stay in solution, are not known for causing
turbidity or light scattering because they are too small to scatter light in the visible
region of the spectrum. Haptoglobin, IgM and alpha-macroglobulin represent some
of the largest circulating proteins found in the serum or plasma and have molecular
weights of 1,000, 900, and 800 kD, respectively. One can estimate the radius by using
the Stokes-Einstein radius, which is equal to

SE radius = 0.74M 0.333 Ångströms [9]. (4.4)

40       4 The Nature of Lipemic and Turbidity Interferences

By this formula, a protein with a molecular weight of 1,000 kD would have a radius of
147 Ångströms, or 14.7 nm, which is too small to cause light scattering in the visible
region or near ultraviolet region.

4.3.2 Lipoprotein Particles

Even though lipoproteins can be quite large, the constituents that make up the lipo-
proteins are small when compared with the large proteins that circulate in blood. The
Apo A proteins have molecular weights of 44 kD or less, the protein Apo B-100 has a
molecular weight of 512 kD, and the protein Apo B-48 has a molecular weight of 241
kD. The molecular weights of Apo C and Apo E are less than 34 kD. The constituents
of the lipid pool of lipoproteins show even lower molecular weights. Triglycerides run
about 0.8 kD. Cholesterol shows a molecular weight of 386 D, while cholesteryl ester
about 1,000 D or 1.0 kD. Thus, the individual components of lipoprotein particles are
never big enough to scatter light in the visible or near ultraviolet region (Apo B-100 is
only about 11 nm in diameter and cholesteryl esters only about 1.4 nm in diameter).
Particles need to be at least 50 nm in diameter to effectively scatter light.
Chylomicrons effectively scatter light, causing turbidity. The particles compos-
ing VLDL also effectively scatter light, as shown using a Spectra Physics Model 125A
laser, with an incident wavelength of 633 nm, clearly in the visible region of the light
spectrum [10].
Chylomicrons and VLDL (as well as the other lipoproteins) represent conglom-
erates of many smaller chemicals. The purpose of lipoproteins is to transport lipids
in the blood. The major lipids transported are cholesterol, cholesteryl esters and tri-
glycerides. Cholesterol has a very low solubility in water. Cholesteryl esters basically
are completely nonpolar and exhibit extremely limited solubility in water. Triglyc-
erides are composed of glycerol and fatty acids (see Fig. 4.2). Glycerol itself is fairly
soluble in water because it consists of three alcohol groups on a three-hydrocarbon
backbone. The carboxylate group on fatty acids makes that end of the molecule polar
and thereby soluble in water, but the long hydrocarbon chain is strongly nonpolar
and insoluble in water. The polar carboxylate group and the long hydrocarbon chain
structure that makes up fatty acids make these molecules amphiphilic.
The lipids are transported in blood (an aqueous phase) in the form of micelles
(see Fig. 4.3). The micelles are aggregates of polar, amphiphilic and nonpolar sub-
stances. The polar and amphiphilic substances contact the water in the aqueous
phase as surface active materials. The apolipoproteins are polar, while the phospho-
lipids and fatty acids are amphiphilic. They form spherical micelles as a result of the
thermodynamics of the interactions between the solvent and the hydrophobic (polar)
portions of the molecule [11].
Because a strictly polar solvent cannot dissolve a strictly or strongly nonpo-
lar substance, the solvent and substance separate. In the presence of amphiphilic
 4.3 Lipemia Interference Mechanisms       41

polar
O
O P O N
O

O O
C O C O

nonpolar

Fig. 4.2: Phosphatidylcholine.

H H
O H2O

H2O
H2O

H2O
H2O

H2O

Fig. 4.3: Micelles formed by phospholipids.

species (one end polar and the other end nonpolar), the polar ends of the species will
direct themselves towards the water and the nonpolar ends will direct themselves
towards one another. Thermodynamically, the collection of nonpolar and amphiphil-
ics species will attain the lowest entropy and coalesce into a sphere with the nonpolar
material directed to the inside. Without the formation of micelles, the nonpolar phase
would separate out of solution. The phospholipids and apolipoproteins act as sur-
factants and reduce the surface and interfacial energies for the polar ends of the free
fatty acids; the nonpolar cholesterol and cholesterol esters remain on the interior of
the micelles protected from the polar environment of the water [11]. Physiologically,
it is dangerous for free fatty acids to circulate freely in the blood or plasma because
42       4 The Nature of Lipemic and Turbidity Interferences

when at physiologic concentrations they interact with Ca2+ and Mg2+ and will precipi-
tate as fatty acid soaps in the tissues [11]. The precipitation of fatty acids into fatty
soaps is well recognized pathologically when it occurs in acute pancreatitis. Much
of the free fatty acids found in the circulation in blood are actually non-covalently
bound to albumin. The formation and solubility of the micelles depends on tempera-
ture, being more likely to form at higher temperatures. The dependence on tempera-
ture explains why placing a lipemic sample in the refrigerator will separate out the
chylomicrons onto the surface of the specimen.
VLDL circulates in the plasma in three size classes: small (27–35 nm), intermedi-
ate (35–60 nm), and large (60–200 nm) [12]. VLDL particles actually represent swollen
micelles or microemulsions, defined as having a range between 10–100 nm in diam-
eter [11]. Only the intermediate and large VLDL play a major role in light scattering. In
normal subjects, the VLDL1 particles are typically 55 nm in diameter, but in Type III,
they range in size from 58–74 nm, and in Type IV, they range in size from 60–66 nm
[10]. The normal triglyceride concentration falls below 1.1 mmol/L, but in dyslipidem-
ias the triglycerides concentration increases, ranging between 2.3–4.6  mmol/L for
Type III dyslipidemia and 3.4–5.7 mmol/L for Type IV dyslipidemia [10]. All plasma
samples contain a small concentration of large VLDL, but the number of VLDL par-
ticles increases in insulin resistance and diabetes, and can give rise to lipemia [12].
Chylomicrons represent a heterogeneous group of particles ranging in size from
70–1,000 nm and varying greatly in size distribution and number among individuals
[13]. Chylomicrons would be considered to be colloidal dispersions and range even up
to emulsions or suspensions [11]. Microemulsions with minimum diameters between
100 and 200 nm are quite turbid and border on being opaque [11]. Because VLDL and
chylomicron particles vary greatly in size and triglycerides content, one might expect
that a direct measure of triglycerides would not show good correlation with light
scattering. Sonntag and Glick first reported that triglycerides concentration poorly
correlates with lipemic index (light scattering determined as the difference between
absorbance at 660 and 700 nm) [14]. The effects of these two classes of lipoproteins
complicate the analysis further, because their diameters range from 50–1,000 nm,
thus producing Mie light scattering in addition to Rayleigh light scattering. Because
one specimen from a patient with lipemia may not be the same as specimens from
other patients with lipemia, the use of one type of sample may not be sufficient to
truly evaluate a method for the effects of turbidity caused by lipemic samples. It is
wiser to evaluate the effect with multiple lipemic samples.

4.3.3 Intralipid® and Lipemia Simulation

Intralipid® is different from VLDL and chylomicrons. It is a sterile, nonpyrogenic fat

emulsion for intravenous infusion containing, per liter, 200 mL of soybean oil, 12
mL of egg yolk phospholipids, and 22 mL of glycerin with the balance made up by
 4.3 Lipemia Interference Mechanisms       43

water [15]. The particles range in size from 200–600 nm with a mean of 345 nm [16].
Thus, Intralipid® completely misses the range of values for large VLDL and misses
the lower and upper ranges for chylomicrons. Furthermore, the refractive index of
Intralipid® is near 1.47 and differs from those of lipoproteins, which is closer to 1.3
[16]. Lipemia differs from icterus and hemolysis, in that one cannot obtain a simple
chemical substance, like bilirubin or hemoglobin, respectively, that can mimic many
of the physical chemistry properties of the interfering substance.
If one changes the physical chemical properties of the particles in solution, such
as adding a reagent that improves or decreases the solubility of the particles in the
solvent, such as the effect of a surfactant on lipid particles, then the refractive index
and the refractive increment change. Such changes might explain why interferences
were noted for lipemic samples with some, but not all of the immunoturbidimetric
methods in the study reported by Bornhorst et al. [17]. The wavelength of the light and
the molecular weight of the particle remain as important variables in explaining the
observed differences between Intralipid® and lipemic samples observed by Bornhorst
et al. [17]. Rayleigh scattering, as described above, applies for particles smaller than
the wavelength of the incident light [7]. The intensity of the scattered light relative to
the incident light intensity follows λ−4; thus, the solution of particles will scatter light
with an intensity more than ninefold greater for violet (400 nm) than for red (700 nm)
light [5]. Molecular weight enters as the third major variable. The molecular weight
relates to the density times one half of the particle diameter cubed. Thus, as particles
increase in size, their ability to scatter light greatly increases as well. Furthermore,
particles whose diameter is near that of the wavelength of the incident light present a
distortion of the Rayleigh scattering [7]. The distorted configuration becomes notice-
able when the particle (for a spheroid) diameter exceeds one fourth of the wavelength
[7], which for visible light (400 nm) starts at 100 nm. Furthermore, as the diameter of
the particle approaches the wavelength, Rayleigh light scattering loses importance
and Mie scattering predominates [18]. Particles as small as 50 nm in diameter are
large enough to generate turbidity interference in the visible region of the spectrum
(around 400 nm), as determined by back scatter [8].
In Mie scattering, back scattering of light exceeds forward scattering as parti-
cle diameters approach the wavelength in size, with angle-dependent separation of
colors (rainbow effect) [18]. Heterogeneous mixtures appear white instead of colored
because the various particle diameters scatter the light at various angles [5].

4.3.4 Empirical Studies in Lipemia Turbidity

In our own study of lipemic samples (see Fig. 4.1) we regressed the pseudo-absorb-
ance as measured in the spectrophotometer against the wavelength. The result was a
power law relationship, expressed as
44       4 The Nature of Lipemic and Turbidity Interferences

A = 3 · 107 −3.03 (4.5)

with a correlation coefficient of 0.99. This power law disagrees with the one for Ray-
leigh light scattering, and suggests that light scattering by lipemic samples more
closely follows Mie scattering than it does Rayleigh light scattering, and also that one
needs to rely more on empirical data for evaluating the effect of turbidity from lipemia
on interference. Still, it fits the data extremely well, implying that the relationship
between the apparent absorbance due to turbidity and the wavelength can be clearly
correlated between wavelength in a relative fashion, and that determination of the
pseudo-absorbance at one wavelength can be used to predict the pseudo-absorbance
at another wavelength.

4.4 Lipoprotein Particles and Lipemia

Chylomicrons and VLDL particles cause turbidity and interference. One must exercise
care in interpreting the results of interference studies that use samples with added
Intralipid® (or other synthetic emulsions) to simulate lipemia. Such samples may
not behave the same as native lipemic specimens. Interference may occur for native
samples but not for Intralipid®; interference may occur for Intralipid® samples but
not for native samples; both lipemic specimens and Intralipid® may show interfer-
ences, but the interferences could differ in magnitude, direction (positive as opposed
to negative) or wavelength. Accurate evaluation of lipemic interference is important
to prevent the reporting of erroneous values, and investigators should be encouraged
to use native lipid samples covering a wide range of VLDL and chylomicron concen-
trations in these studies. In using lipemic samples to evaluate methodologies for
interference, one can create a quantitative scale by measuring triglyceride content or
the apparent absorbance at a given wavelength, e.g., 400 or 660 nm.
The heterogeneous nature of lipemia creates difficulties in simulating samples.
Further, there is an imperfect relationship between the degree of turbidity and the
concentration of triglycerides [14, 17]. Several approaches have been taken to mini-
mize the effects of lipemia on analytical methods. Proper blanking can be effective,
either a true sample blank or a kinetic method can be quite effective, but assumes
that the reagents do not have an effect on the lipids [2]. In glycerol-blanked methods
for triglycerides, the sample is frequently lipemic and thus turbid, but the reaction
to measure the endogenous concentration of glycerol may actually clear the sample
to some extent, thus causing a negative interference. Some laboratories use clearing
solutions to rid the sample of turbidity, but this type of approach needs to be care-
fully studied in the individual laboratory because the clearing solution may cause
its own interference with the analytical method. A common approach to lipemic
samples taken by many laboratories is to separate the lipemic phase from the rest
 4.5 References       45

of the sample by ultracentrifugation, often using a table top model centrifuge with a
special centrifuge tube [2].

4.5 References
[1] Artiss JD, Zak B. Problems with measurements caused by high concentrations of serum solids.
CRC Critical Reviews. Clin Lab Sci USA 1987,25,19–41.
[2] Creer MH, Ladenson J. Analytical errors due to lipemia. Lab Med USA 1983,14,351–355.
[3] Glick MR, Ryder KW. Analytical systems ranked by freedom from interferences. Clin Chem USA
1987,33,1453–1458.
[4] Tanford C. Physical chemistry of macromolecules. New York, NY, USA, Wiley and Sons, 1961.
[5] Nassau K. The Physics and Chemistry of Color: The Fifteen Causes of Color. New York, NY, USA,
Wiley and Sons, 1983.
[6] Kroll MH, Elin RJ. Interference with clinical laboratory analysis. Clin Chem USA
1994,40,1996–2005.
[7] Cantor CR, Schimmel PR. Biophysical chemistry. San Francisco, CA, USA, W.H. Freeman & Co.,
1980.
[8] Kuhn H, Forsterling H-D. Principles of Physical Chemistry: Understanding Molecules, Molecular
Assemblies, Supramolecular Machines. Chichester, UK, Wiley and Sons, Ltd 2000.
[9] Venturoli D, Rippe B. Ficoll and dextran vs globular protein probes for testing glomerular
permselectivity: effects of molecular size, shape, charge, and deformability. AJP Renal USA
2005,288,605–613.
[10] Packard CJ, Shephard J, Joerns S, Gotto Jr AM, Taunton OD. Very low density and low density
lipoprotein subfractions in type III and type IV hyperlipoproteinemia: Chemical and physical
properties. Biochim Biophys Acta (BBA)/Lipids Lipid Metabolism Netherlands 1979,
572,269–282.
[11] Myers D. Surfactant Science and Technology. New York, NY, USA VCH Publishers, Inc, 1988.
[12] Garvey WT, Kwon S, Zheng D, Shaughnessy S, Wallace P, Hutto A, et al. Effects of insulin
resistance and type 2 diabetes on lipoprotein subclass particle size and concentration
determined by nuclear magnetic resonance. Diabetes USA 2003,52,453–62.
[13] Park Y, Grellner WJ, Harris WS, Miles JM. A new method for the study of chylomicron kinetics in
vivo. Am J Physio l Endocrinol Metab USA 2000,279,E1258–63.
[14] Sonntag O, Glick MR. Serum-Index und Interferogramm: Ein neuer Weg zur Prüfung und
Darstellung von Interferenzen durch Serumchromogene. Lab Med Germany 1989,13,77–82.
[15] Baxter Healthcare. Intralipid [Package insert]. Deerfield, IL, USA Fresenius Kabi AB, Baxter
Healthcare Corporation, 2000.
[16] Wabel C. Influence of lecithin on structure and stability of parenteral fat emulsions: Doctoral
thesis. Nuremberg, Germany, Naturwissenschaftliche Fakultäten der Friedrich-Alexander-
Universität Erlangen, 1998.
[17] Bornhorst JA, Roberts RF, Roberts WL. Assay-specific differences in lipemic interference in
native and Intralipid-supplemented samples. Clin Chem USA 2004,50,2197–201.
[18] Weiner I, Rust M, Donnelly TD. Particle size determination: an undergraduate lab in Mie
scattering. Am J Phys USA 2001,69,129–36.
5 Measurement of Interference
The manufacturer of the analytical method or reagent systems used to report results
in the clinical laboratory must be able to make claims about the independence of their
methods concerning interference or be able to predict when the presence of bilirubin
or lipemia may begin to affect their methods. Likewise, the laboratory needs to know
the potential for bilirubinemia or lipemia to cause an interference with the methods
they use to report results.

5.1 A Typical Commercial Study

Manufacturers will provide laboratories with information on the studies that they
have performed to assess their methods for interference. What follows is the way
that Beckman Instruments reported on their evaluation for the Synchron EL-ISE
System  [1]. First, they established a serum pool by combining and freezing patient
serum samples. Before use, they thawed the pool and filtered it. They added lithium
to bring it up to the desired concentration. Then the pool was aliquoted into volumes
suitable for one-day use and the aliquots refrozen.

Interferent Na (mmol/L) K (mmol/L) Cl (mmol/L) CO2 (mmol/L) Ca (mmol/L)

Bilirubin (μmol/L)

26 0.3 −0.1 −0.4 0.5 0.00

51 0.5 0.1 −0.5 0.7 0.01
77 1.0 0.1 −0.1 0.8 0.02
103 0.9 0.2 0.3 0.8 0.02
154 0.5 0.1 −0.1 0.9 0.03
205 −0.2 0.1 0.1 0.6 0.03
257 1.6 0.1 0.1 1.6 0.04
308 2.0 −0.2 0.6 0.5 0.02
410 1.3 0.2 0.3 1.1 0.04
513 0.8 0.2 0.1 −0.5 −0.01

Intralipid® (g/L)

1 0.4 0.0 0.1 0.4 −0.01

2 1.2 0.0 0.1 0.5 −0.01
3 0.6 0.0 0.2 0.3 −0.01
4 0.8 0.0 0.1 0.7 −0.01
6 0.2 0.0 −0.1 0.5 0.00
8 1.8 0.1 0.4 0.7 0.00
10 1.0 0.0 1.1 0.2 −0.01
Within run SD 2.0 0.2 4.0 2.0 0.10

Tab. 5.1: Beckman bilirubin and Intralipid® delta values.

48       5 Measurement of Interference

To prepare the bilirubin samples they added 53 mg of bovine bilirubin into 10 mL of

0.1 N KOH, with a resulting concentration of 10,773 μmol/L. They protected the solu-
tion from light with aluminum foil. They then made further dilutions, from 5–100 %.
To prepare samples for lipemia testing they diluted Intralipid® (KabiVitrum, Inc.)
with deionized water to make spiking solutions at concentrations of 0–100 %, in steps
of 10 %. The Intralipid® concentration ranged from 0–210 g/L in steps of 21 g/L. Then
they added 263 μL of the Intralipid® spike to 5.0 mL of the serum pool, resulting in
Intralipid® concentrations ranging from 0–10 g/L in increments of 1 g/L.
Next they determined the analytes sodium, potassium, chloride, carbon dioxide
and calcium with the analyzer in duplicate. They defined the degree of interference
as Delta (Δ) = [Test Analyte Concentration] − [Blank Analyte Concentration]. They
defined interference if the Delta exceeds twice the within-run standard deviation for
the methodology. The results for the calculated Delta are listed in Tab. 5.1. Based on
their criteria for evaluating interference they are justified in making the claim that
their method is free of interferences for bilirubin and lipemia. Their study raises some
important questions. What are the appropriate materials necessary to test for interfer-
ence, how should interference be defined and how should the results be interpreted?

5.2 Guidelines for Interference Studies

Answers to these questions can be found in the NCCLS guideline, “Interference

Testing in Clinical Chemistry; Approved Guideline” [2]. The document provides back-
ground information, guidance and experimental procedures for investigating, iden-
tifying, and characterizing the effects of interfering substances on clinical chemistry
test results.
Manufacturers and laboratories need a way to evaluate potentially interfering
substances with laboratory tests. Because assays may vary from analyzer to analyzer,
interfering substances may demonstrate varying behaviors, and the use of laboratory
tests may depend on the clinical setting. The guidance document outlines approaches
to take in the assessment of interfering agents rather than strict protocols. Ultimately,
the users of these guidelines need to apply judgment based on a thorough under-
standing of the strengths and weakness of any given approach.
Inaccuracy or total analytical error in methods consists of the imprecision of the
method, the bias of the method and bias introduced by the sample or specimen uti-
lized in testing for the given analyte [2]. Interference from endogenous substances
implies a bias independent of the population or type of matrix holding the analyte,
but dependent on a substance present endogenously for that particular individual. To
include the effect of the interfering substance with the reported result for that particu-
lar patient could result in a misinterpretation of the patient’s diagnosis or progress in
the course of disease or treatment and thus pose a threat to the patient’s safety.
 5.3 Bilirubin       49

The effect of an endogenous interferent must be made relative to the normal amount
of that interferent occurring in the patient’s serum, plasma, blood, etc. All samples
contain some bilirubin as well as chylomicrons and VLDL particles. The evaluation of
the interference requires then that the base material represents the amounts of these
substances found typically in healthy subjects. Because endogenous interferents are
common and in some sense uncontrollable, all methods need to be tested for them [2].
Icterus (bilirubin) and lipemia are two of these endogenous interferences, hemolysis
being the third.
In establishing the amount of interferent to add, one should assume the worst
and conduct the evaluations with the highest concentrations of interferent possible
[2]. The purpose of the evaluation is to demonstrate that there is no interference or if
an interference is present, at what concentration of interferent is it still safe to report
results or hold results because the accuracy of the analyte result is not a given. Accord-
ing to the guideline, two concentrations of the analyte should be evaluated [2]. Some
analytes normally occur at relatively low concentrations and a negative interference
may not be detectable at low concentrations of the analyte. The guideline provides
suggestions of concentrations of analytes to use in its Appendix B.
For evaluating the effects of bilirubin and lipemia, one should establish a base
pool from normal sera or plasma. Next, one needs to choose an appropriate material
to mimic the endogenous effects of bilirubin or lipemia.
Glick et al. have established clear protocols for material for testing for bilirubin
and lipemia, especially as referenced in their book Interferographs [3].

5.3 Bilirubin

Bilirubin solutions can be prepared from purified bilirubin. Purified bilirubin is not
soluble in water, so it is necessary to dissolve it with dimethyl sulfoxide (DMSO). One
should first create a base serum in a solution of 0.25 mL of DMSO, 0.5 mL of sodium
carbonate solution, 0.1 M and 0.5 mL of 0.1 molar HCL [3]. Next, with constant stirring
add the base solution, drop by drop, into 24 mL of the serum pool [3].
Dissolving the bilirubin may take some time, one should make certain that the
bilirubin is completely dissolved before going onto the next step. (If one tries to add
bilirubin to the serum pool first, it will adhere to the surface of the container and not
go into solution.) First protect the bilirubin from light and measure out 6.0 g of the
bilirubin powder, then add 0.1 mL of DMSO and wait until it dissolves; then add 0.2
mL of the 0.1 molar sodium carbonate solution and mix immediately, then quickly
add 9.5 mL of the already prepared serum and stir well; finally add 0.2 mL of the
0.2 molar HCL solution. This solution has a bilirubin concentration of 1,026 μmol/L.
Transfer 4.0 mL of the main bilirubin solution to the next dilution test tube, adding
4.0 mL of the base solution to create a 513 μmol/L solution; continue these steps to
produce concentrations of 257, 125, 65, and 32 μmol/L of added bilirubin [3].
50       5 Measurement of Interference

5.4 Intralipid®

Intralipid®  20  % (a 20  % Intravenous Fat Emulsion) is a sterile, non-pyrogenic fat

emulsion prepared for intravenous administration as a source of calories and essen-
tial fatty acids. It is made up of 20  % Soybean Oil, 1.2  % Egg Yolk Phospholipids,
2.25  % Glycerin, and Water for Injection. In addition, sodium hydroxide has been
added to adjust the pH so that the final product pH is 8 pH range is 6–8.9.
The soybean oil is a refined natural product consisting of a mixture of neutral
triglycerides of predominantly unsaturated fatty acids. The major component fatty
acids are linoleic (44−62 %), oleic (19−30 %), palmitic (7−14 %), linolenic (4−11 %) and
stearic (1.4−5.5 %) (Fig. 5.1) [4].

O O OCR1
OH R2CO
OCR3
linoleic acid O
triglycerides
O
O
OH
O OCR1
O CH3
oleic acid R2CO
O P O CH2 CH2 N CH3
O O O CH3
OH phosphatidylcholine

linolenic acid O
O OCR1
O
O R2CO
O P O CH2 CH2 NH3
OH
O
stearic acid phosphotidylethanolamine

Fig. 5.1: Structures of fatty acids, triglycerides and phospholipids.

Purified egg phosphatides are a mixture of naturally occurring phospholipids which

are isolated from the egg yolk. These phospholipids have the following general struc-
ture of an amine group bonded to a glycerol backbone through a phosphate (Fig. 5.1).
In addition, two fatty acids are bonded to the backbone of the remaining alcohol
groups of the glycerol. These fatty acids may be saturated or unsaturated. The fatty
acid groups make the phospholipids nonpolar on their end of the molecule, while the
amine groups make the molecular polar.
Intralipid®  20  % has an osmolality of approximately 350 mOsmol/kg water
(which represents 260 mOsmol/liter of emulsion) and contains emulsified fat parti-
cles of approximately 500 nm size. Intralipid® 20 % is indicated as a source of calo-
 5.4 Intralipid®       51

ries and essential fatty acids for patients requiring parenteral nutrition for extended
periods of time (usually for more than 5 days) and as a source of essential fatty acids.
The prime destabilizers of emulsions are excessive acidity (low pH) and inappropri-
ate electrolyte content. Careful consideration should be given to additions of divalent
cations (Ca2+ and Mg2+) which have been shown to cause emulsion instability. Amino
acid solutions exert a buffering effect protecting the emulsion.
The admixture should be inspected carefully for “breaking or oiling out” of the
emulsion. “Breaking or oiling out” is described as the separation of the emulsion and
can be visibly identified by a yellowish streaking or the accumulation of yellowish
droplets in the admixed emulsion. The admixture should also be examined for par-
ticulates.
Intralipid® is not a perfect simulating agent for lipemia because it is derived from
soybeans and its emulsions do not necessarily have the same structure of VLDL par-
ticles and chylomicrons. Intralipid® is also sensitive to low pH and it may “oil out”
of the emulsion, further changing its light scattering properties; however, its use is
superior to the measurement of triglycerides as a way to quantify the effect of light
scattering caused by lipemia. Even though chylomicrons are the major carrying par-
ticle for triglycerides, triglycerides can be increased in particles that carry VLDL. The
light scattering effects of VLDL particles and chylomicrons have more to do with the

Degree of Interferent* Absorbance Dimension Max Architect ci8200 Vitros

Turbidity (670nm) Creatinine (μmol/L) Creatinine (μmol/L) Creatinine (μmol/L)

Baseline 0.42 74 78 84
+ 0.64 79 79 83
1+ 0.89 80 79 83
2+ 1.32 84 82 83
3+ 1.50 83 81 83
4+ 1.77 85 82 86
Baseline 0.42 140 138 142
+ 0.64 139 137 137
1+ 0.89 139 138 138
2+ 1.32 141 139 137
3+ 1.50 140 140 139
4+ 1.77 145 141 140
Baseline 0.42 285 280 262
+ 0.64 286 276 267
1+ 0.89 291 276 268
2+ 1.32 292 275 269
3+ 1.50 289 277 263
4+ 1.77 297 278 260

* Starting significant negative interference >11.4 % decreasing from baseline level

Tab. 5.2: Effect of lipemia for three methods for creatinine.

52       5 Measurement of Interference

size of the particles, rather than their number and the total concentration of triglyc-
erides. Glick et al. have recommended creating a set of dilutions of Intralipid® that
follow the following sequence: 0.3, 0.6, 1.25, 5.0, and 20.0 g/L [3]. Dilutions are usually
stable for several days.
Another approach to making materials to test for lipemic interference is to pool
several lipemic specimens. Specimens that show marked turbidity can be pooled and
then serially diluted to create varying concentrations of chylomicrons and VLDL par-
ticles. To quantify the dilutions one can measure the absorbance at a standardized
wavelength, 660 nm, or measure the triglycerides content (Tab. 5.2) [5]. Further, one
can compare the serially diluted samples against a visual assessment of the amount
of turbidity.

5.5 Procedure to Make Five Concentrations

The NCCLS guideline presents an easy way to make five concentrations. Begin by
establishing a serum or plasma pool, this is considered the low pool and is desig-
nated L. For bilirubin, it will contain a small amount of bilirubin if the pool was made
by combining serum or plasma from disease free persons. For lipemia, if the samples
were obtained from disease free persons and they were fasting, the pool will contain a
minimum of lipemia and turbidity. One could also make an analyte pool with elevated
analytes if one used serum or plasma from diseased persons, but without apparent
hyperbilirubinemia or lipemia.
Create a high pool in serum or plasma, as mentioned above and designate it as H.
Mix equal volumes of the low pool and high pool, to make a mid pool. Then mix equal
volumes of the mid pool and the low pool to make a 25 % pool. Likewise, mix equal
volumes of the mid pool with the high pool to make a 75 % pool.

5.6 Interference Criteria

One should establish the criteria for declaring that a potential interfering agent has
actually caused an interference. Based on how an error might be interpreted medi-
cally, often a 10 % deviation is considered appropriate. Actually, the percentage error
depends on the analyte in question and how that analyte is used. Manufacturers often
use the 10 % deviation or a combination of common denominators, or, as seen above,
+/− 2 standard deviations of within run precision. Individual laboratories must assess
whether the above error criteria are sufficient.
The argument for using a limit based on the within run precision is that the error
would not normally be detected by the clinician and is not greater than normally
expected based on the natural errors built into the analytical system. Arguments
against using the within run precision as a limit are that the precision utilized may
 5.6 Interference Criteria       53

not be tight enough or too tight. The precision utilized might be the one in the manu-
facturer’s claim for the method, which may be larger than the actual precision, result-
ing in a false negative result. The precision for the method may be extremely tight,
resulting in a false positive result.
Beyond precision, using the claimed accuracy limits for the test, based on bias,
may be preferred. If the biases are extremely small, use of biological variation may be
the most appropriate. Biological variation limits vary depending on the analyte.
The rationale for using biological variation for determining an analytical goal
stems from the concepts of total error and total allowable error [6]. The total error
(TE) of an analytical method is composed of the random error (RE) and the systematic
error (SE) known as bias for the method. One should allow for 95 % of the allowable
error so the formula for the total error is

TE = 1.96 CVA + SE , (5.1)

where CVA is the coefficient of variation for the particular analytical method [7]. The
total error for the analytical system represents what the system, as construed, can do
or how well it can perform. In reality, for goals, one wants what is best to serve the
clinical purpose for the analytical tests. Given that, one should establish the allow-
able error and the allowable error should fit the clinical need for information. The
total error should be less than the allowable error [7].
To establish an acceptable allowable error, one should minimize the systematic
error, which should be kept to near zero, and use a precision goal that should be
less than or equal to one-half of the within-subject biological variation (CVi) [7–9].
When the precision goal is set at one-half of the within-subject biological variation,
the maximum contribution to error of the imprecision is only 11.8 % that of the bio-
logical variation, which would be 11.2 % for the 95 % center of the distribution of the
imprecision.
One can calculate the allowable error using the one-half within-subject biological
variation rule as follows [7]:

1
EA < 1.96 CVi for P ≤ 0.05, or EA < CVi . (5.2)
2

The presence of an interferent (I) adds to the systematic error, so that

TE = 1.96 CVA + SE + I with TE < EA < CViI , (5.3)

and thus

1.96 CVA + SE + I < CVi , (5.4)

and solving for the interference term yields,

54       5 Measurement of Interference

I < CVi − (1.96 CVA + SE ). (5.5)

The implication of using this criterion is that one needs to establish the precision and
bias in one’s method, then subtract it from the CV for the within biological variation
to determine the limits for an interference. For example with creatinine, say that the
systematic error is 1 % and the CV of the random error is 1.5 %. The within-person
biological variation for diseases such as type 1 diabetes mellitus, chronic renal failure
and impaired renal function is around 6 % [10]. The percentage error limit for creati-
nine in this instance would be I = 6% − (1.96 ⋅ 1.5% + 1%) = 2%. For triglycerides, with
a 3 % CV for random error and 3 % systematic error and a within-individual biological
variation CV of 18 %, the allowable error due to interference would be 9 %. The use of
biological variation and the within-individual biological variation coefficient of vari-
ation offer a rational approach to addressing the concepts of developing the appro-
priate limits to use in studies. At present such a manner of determining the limits of
acceptability for interferences has not been universally accepted as a standard for
evaluating interferences, and it is typically not present in the literature obtainable
from manufacturers, but it does offer a reasonable starting point for arriving at an
appropriate limit to use. Ultimately, the laboratory director should use his or her judg-
ment to set the criteria for determining the limit in deviation to establish the presence
of interference.

5.7 Data Analysis

The NCCLS (now CLSI) guideline [2] describes using a difference test to screen poten-
tial interferents, achieved by observing the difference between the test average and
a control average. One could evaluate the results by using a Student’s t-test. Using a
screen may be an effective way to look for interference when it comes to drug interfer-
ence, but as explained in the preceding chapter, bilirubin and lipemic interferences
may demonstrate complex behavior that could be missed using a single concentra-
tion of the interferent. It is much better to examine these types of interferences using
at least five concentrations of the interferent. This approach also fits in with the char-
acterization in the NCCLS guideline.
In the Interferographs Glick would plot the points as a deviation from the origi-
nal value. He determined the original point by dividing the result for the bilirubin or
Intralipid® value by the original results and multiplying by 100. Creating a graph in
this manner allows one to see the effects of bilirubin for a particular analyzer for all
the methods at once.
In Tab. 5.3 are results for a study of the effect of bilirubin on three methods for
creatinine [5]. The first two methods are based on the Jaffe reaction while the last
method (the Vitros) is based on the sarcosine oxidase enzymatic method (peroxidase
as the indicator system). Fig. 5.2 shows a plot in the manner of an Interferograph [3].
 5.7 Data Analysis       55

Degree Icteric Interferent* Bilirubin Dimension Max Architect ci8200 Vitros

(μmol/L) Creatinine (μmol/L) Creatinine (μmol/L) Creatinine (μmol/L)

Baseline 0 87 82 88
+ 85.5 84 70 a 92
1+ 171.0 76 61 97
2+ 343.0 65 a 42 104 a
3+ 514.0 58 24 109
4+ 686.0 44 14 108
Baseline 0 145 137 145
+ 85.5 142 127 154
1+ 171.0 136 114 a 164 a
2+ 343.0 124 a 94 177
3+ 514.0 114 75 180
4+ 686.0 99 61 184
Baseline 0 284 272 278
+ 85.5 287 259 302
1+ 171.0 280 249 328 a
2+ 343.0 279 232 a 352
3+ 514.0 272 207 364
4+ 686.0 247 a 191 369

* Bilirubin values given in SI units: to convert to conventional units (mg/dL), multiply by 0.058.
a
Starting significant negative interference >11.4 % decreasing from baseline level

Tab. 5.3: Effect of bilirubin for three methods for creatinine.

bilirubin interference
140

120
Dimension Max
100 · (final/original result)

100 Architect ci8200

Vitros 350
80

60

40

20

0
0 200 400 600 800
bilirubin ( mmol/L)

Fig. 5.2: Relative amount of interference caused by bilirubin for creatinine.

By plotting the y-axis scale in terms of the percentage related to the original result,
all three methods start at the same point when examining them at the initial sample,
the one without the added interferent. If one were using a 10 % deviation from the
56       5 Measurement of Interference

original value as the limit for detecting interference, then it is fairly easy to see that all
three methods exceed either 110 % or 90 %, thus indicating that bilirubin interferes
with the determination of creatinine. Further, it is readily apparent that the Dimen-
sion Max and Architect ci8200 methods demonstrate a negative interference and that
the Vitros 350 demonstrates a positive interference. One thing that the Interferograph
plot does not show well is at what concentrations of bilirubin the interfered value of
creatinine crosses the limit.
Fig. 5.3 shows a plot of the values for creatinine versus bilirubin concentration
in a manner similar to that suggested by the NCCLS guideline [2]. Again, it is easy to
see that the Jaffe reaction based methods experience negative interference and the
enzymatic based method experiences positive interference with bilirubin. The plot
also demonstrates that the enzymatic method recovers slightly higher values than
the Jaffe based methods, when one looks at the point on the y-axis near a bilirubin
concentration of 0. From this type of plot it is a bit more difficult to see where to draw
a cutoff for the limit of interference acceptability.

bilirubin interference
120

100
Dimension Max
Architect ci8200
80
creatinine ( mmol/L)

Vitros 350
60

40

20

0
0 200 400 600 800
bilirubin ( mmol/L)

Fig. 5.3: Absolute amount of bilirubin interference with creatinine.

The next step in the guideline is to perform regression analysis [2]. Regression analy-
sis probably provides the best and most quantitative information concerning inter-
ferences. Results for standard linear regression analysis for these three methods are
shown in Tab. 5.4. For the Dimension analyzer, note that the r2 (R sq in Tab. 5.4) is
0.99. As long as r2 is larger than 0.95, then interpretation of the regression analysis
is straightforward. The slope is −0.06 and the standard error of the slope is 0.003. To
assess if there is a linearly dependent interference in relation to bilirubin, one needs
to assess the significance of the slope. One can assess the significance of the slope by
performing a t-test, comparing the slope to zero. In this case,
 5.7 Data Analysis       57

|−0.06 − 0|
t= = 20. (5.6)
0.003

For a t = 20, with 4 degrees of freedom and a two-tailed test, the probability of the
null hypothesis is <0.0001, so the slope is significant. Examining the slope is the best
test for determining whether or not a method is affected by a potential interferent. If
the slope is not significant, then one can say with confidence that no interference is
occurring. Of course, in order to make this statement one must examine the curvature
of the analyte-interferent relationship. If it is straight, then one can make this claim.
If it is curvilinear, then one must apply a more rigorous regression analysis.

Dimension        

  Slope −0.062 87.63 Intercept

  SE slope 0.003 1.14 SE int
  R sq 0.991 1.75 SE reg
  F 437.968 4.00 df
  SS reg 1.347.691 12.31 SS res

Architect        

  Slope −0.101 78.98 Intercept

  SE slope 0.006 2.13 SE int
  R sq 0.988 3.28 SE reg
  F 328.810 4.00 df
  SS reg 3.529.892 42.94 SS res

Vitros        

  Slope 0.031 90.32 Intercept

  SE slope 0.005 2.00 SE int
  R sq 0.900 3.08 SE reg
  F 35.840 4.00 df
  SS reg 339.448 37.88 SS res

Analyte: creatinine
Interferent: bilirubin

Tab. 5.4: Regression analysis of bilirubin interference with creatinine.

From the slope and intercept one can calculate at which value of the interferent sig-
nificant interference begins. Say, for example, that one’s criterion for a clinically
significant interference is +/− 10  %. The goal is to find the bilirubin concentration
that forces the creatinine concentration to exceed by 10 % the unfettered value or fall
below by 10  % the unfettered value. For creatinine with bilirubin interference one
could state this criterion as

[creatinine ] · (1 + 0.1) = slope · [bilirubin ] + intercept (5.7)

58       5 Measurement of Interference

and

[creatinine ] · (1 − 0.1t ) = slope · [bilirubin ] + intercept (5.8)

which yields the two formulas:

1.1 · [creatinine ] = slope · [bilirubin ] + intercept (5.9)

and

0.9 · [creatinine ] = slope · [bilirubin ] + intercept . (5.10)

Re-arranging, the upper limit becomes

1.1 · [creatinine ] − intercept

[bilirubin ]upper = (5.11)
slope

and

0.9 · [creatinine ] − intercept

[bilirubin ]lower = . (5.12)
slope

Now the [creatinine] in each of these formulas should be equal to the intercept, thus
the formulas are simplified to

1.1 · intercept − intercept 0.1 · [intercept ]

[bilirubin ]upper = = (5.13)
slope slope

and

0.9 · intercept − intercept −0.1 · [intercept ]

[bilirubin ]lower = = . (5.14)
slope slope

Thus, the upper limit for bilirubin is −146 μmol/L and the lower limit is 146 μmol/L;
of course, because the slope is negative, only the lower limit is relevant, because in
order to exceed the unfettered value of creatinine by 10 % would require a negative
concentration of bilirubin. If, however, the slope was positive, then only the upper
limit becomes relevant. By using regression analysis, one can determine the limits
rather exactly.
One question that arises is how to discern with two different methods that both
demonstrate interference which one shows the worst interference. This can be accom-
 5.7 Data Analysis       59

plished by comparing the slopes. The one with the larger slope will have the worse
interference.
One needs to answer the question, in this case, as to whether the two slopes are
different. That question is answered with a t-test.
The formula for the t-test is:

slope − slope

− 0.06 + 0.1

1 2
t = =  (5.15)
se 2slope 1 + se 2slope 2 0.0032 + 0.00552

which is equal to 0.04/0.0067=6, with 8 degrees of freedom, the probability for the
null hypothesis is <0.01 and therefore the two slopes are different; thus, the interfer-
ence by bilirubin for the Architect analyzer is worse than that for the Dimension.
One goes through exactly the same type of analysis for a positive interference. In
the case of the Vitros, the slope is a positive 0.03. The t-test analysis for the slope gives
a value of 6 for t, which has a probability of 0.004. Because p < 0.05, one rejects the
null hypothesis, and accepts that the slope is statistically significant.
Evaluation of the graph and the regression analysis for the Vitros demonstrates
that the interference relationship does not show a straight line. To properly evalu-
ate this relationship, one should use a polynomial regression. Using the graphic
Trend Analysis in EXCEL, the following regression formula was found, [creatinine]=
−0.0005 [bilirubin]2+0.058 [bilirubin] + 87.3, with a r2 of 0.99. The fact that the correla-
tion coefficient is now greater than 0.99 and that the intercept is extremely close to
the value of 88 found for the sample with zero bilirubin added, indicate that this is
a very good fit for the data. The presence of a polynomial fit for interference analysis
strongly suggests that a more complicated mechanism is involved than simple absorb-

turbidity with Dimension creatinine

160
140
120
creatinine ( mmol/L)

100
80
60
creatinine1
40
creatinine2
20
0
0 0.5 1.0 1.5 2.0
turbidity (absorbance at 670 nm)

Fig. 5.4: Turbidity interference with the creatinine method.

60       5 Measurement of Interference

ance changes, that there might be a depletion of reagents, or that more than one type
of process is occurring.
Results for lipemia from the creatinine study are shown in Tab. 5.3 [5]. In this
study, instead of using Intralipid®, the researcher used lipemic samples and meas-
ured the native apparent absorbance at 670 nm. If one looks at the values for the
Dimension for values of creatinine near 74 and 140 μmol/L one can see that it is dif-
ficult to discern whether a true interference is occurring or not (Fig. 5.4). Regression
analysis reveals that the slope for low values of creatinine (Creatinine1 in Fig. 5.4) is
7.3 with a standard error of 1.3. The t-test for this slope is 5.5, which, with 4 degrees of
freedom, has a probability of 0.005; which means that one can reject the null hypoth-
esis and accept that there is an interference from lipemia (turbidity) with this method.
For the intermediate values of creatinine (Creatinine2 in Fig. 5.4), the slope is 3.1 with
a standard error of 1.5. The t-test for this slope is 2.1, which, with 4 degrees of freedom,
has a probability of 0.1, which is >0.05; which means that one cannot reject the null
hypothesis. Failure to be able to reject the null hypothesis suggests that the slope is
not statistically different from zero and can be used to substantiate and claim that
there is no interference from lipemia (turbidity) for this method and this concentra-
tion of creatinine.
The proper use of regression analysis makes the analysis of interference more
precise, quantitative, and easier to compare. It can provide a more accurate limit for
reporting of results.

5.8 References
[1] Brzezicki J, Muirhead C, Worthington L, Schmitz J, Dewberry T, Haden B. Potential interfereing
substances on Beckman’s Synchron EL-ISE System. Brea, CA, USA, Beckman Instruments, Inc.
[2] NCCLS. Interference Testing in Clinical Chemistry; Approved Guideline. NCCLS document EP7-A,
NCCLS USA 2002. Vol. 22 No. 27, 19087–1898.
[3] Glick MR, Ryder KW, Glick SJ. Interferographs: User’s Guide to Interferences in Clinical
Chemistry Instruments. 2nd Edition. Indianapolis, IN, USA, Science Enterprises, Inc, 1991.
[4] Baxter Healthcare. Intralipid [Package insert]. Deerfield, IL, USA, Fresenius Kabi AB, Baxter
Healthcare Corporation, 2007.
[5] Srisawasdi P, Chaichanajarernkul U, Teerakanjana N, Vanavanan S, Kroll MH. Exogenous
interferences with Jaffe creatinine assays: addition of sodium dodecyl sulfate to reagent
eliminates bilirubin and total protein interference with Jaffe methods. J Clin Lab Anal USA
2010,24,123−133.
[6] Westgard JO, Carey RN, Wold S. Criteria for judging precision and accuracy in method
development and evaluation. Clin Chem USA 1974,20,825–833.
[7] Fuentes-Arderiu X, Fraser CG. Analytical goals for interference. Ann Clin Biochem UK
1991,28,393–395.
[8] Fraser CG. Desirable performance standards for clinical chemistry tests. Adv Clin Chem USA
1983,23,299–229.
[9] Fraser CG. The application of theoretical goals based on biological variation in proficiency
testing. Arch Pathol Lab Med USA 1988,112,404–415.
 5.8 References       61

[10] Ricos C, Iglesias N, Garcia-Lario J-V, Simon M, Cava F, Hernandez A, et al. Within-subject
biological variation in disease: collated data and clinical consequences. Ann Clin Biochem UK
2007,44,343–352.
6 Origin of Icteric Samples
Hyperbilirubinemia occurs in 20 % of specimens from hospitalized patients [1]. Even
though personal examination of samples can provide a way to demonstrate that they
are icteric, many processes today have become automated and samples are not per-
sonally observed prior to being placed into the analyzer for analysis or even after
determinations have been made. Automated processes to establish the presence
of hyperbilirubinemia using interference indices provide a suitable way to detect
samples liable to interference, but such detection systems are not perfect. Another
means to detect samples liable to interference is to recognize the conditions that lead
to hyperbilirubinemia. Knowledge of these conditions can lead to alterations in the
way that Laboratory Information Systems or Electronic Medical Records interact with
the laboratory staff. Medical Directors and Laboratory Professionals should have a
broad and deep knowledge base on the biochemistry, physiology and pathology that
can give rise to samples with hyperbilirubinemia.

6.1 The Origin of Bilirubin

Bilirubin originates from the biochemistry of hemoglobin. Hemoglobin contained in

erythrocytes carries the oxygen to the tissues, releases it to the tissues at a reasonable
partial pressure of oxygen, carries back carbon oxide to the lungs and picks up more
oxygen. In the adult, on average, 30 trillion erythrocytes, or 9 grams of hemoglobin,
circulate in the bloodstream [2]. The erythrocytes have no nuclei in humans and they
only survive on average 120 days in circulation. They are destroyed in the spleen at a
rate of 1⋅1010 per hour, or 0.3 grams of hemoglobin, per hour [2]. The amount of hemo-
globin circulating is equivalent to 14 mmol. There are four heme molecules for each
mole of hemoglobin, which means that there are 56 mmol of heme present in the cir-
culation at any given time. Based on the number of erythrocytes destroyed, 1.9 mmol
of heme are released per hour, which is equivalent to 47 mmol per day.
Bone marrow continually produces new erythrocytes, while senescent erythro-
cytes are destroyed by macrophages in the spleen and in some cases the liver. The
macrophages separate the hemoglobin into the globin portion and the heme. The iron
from the heme is extracted and recirculated, while the rest of the heme is converted to
alpha-hydroxyhemin by the microsomal heme oxygenase, NADPH and oxygen, which
effectively oxidizes the iron to the ferric state and converts the hydrogen on the alpha
carbon to an alcohol group [3]. The alpha-hydroxyhemin is acted upon by oxygen,
which oxidizes the molecule, creating two aldehyde groups where the alpha carbon
was located with the release of a carbon monoxide and the iron atom; this conver-
sion creates an open ring structure [3]. Under the influence of biliverdin reductase
with NADH acting as the electron donor, the biliverdin is reduced to bilirubin, which
differs from biliverdin in that all the four nitrogens are bonded to a hydrogen (bili-
64       6 Origin of Icteric Samples

verdin has only three nitrogens bonded to hydrogen) and it contains only 10 double
bonds, instead of 11 (Fig. 6.1) [3]. These four molecules contain conjugated double
bonds without resonance, giving them the ability to absorb light in the visible region.

OH
H
V C M V M

M V M V
N N N N
heme oxygenase
HC Fe 2 CH HC Fe 2 CH
N N NADPH N N
O2 M M
M M

P C P P C P
H H

O2
V OO M V OO M

M V M V
NH HN NH HN
CO + Fe + HC CH HC CH
biliverdin reductase
NH N NH HN
NADH
M M M M
H
P C P P C P
H H

Fig. 6.1: Formation of bilirubin from heme.

Unconjugated bilirubin has it main absorption band at 450 nm, clearly in the visible
region, with an extinction coefficient of 48,000–63,000 m2/mol [3]. In alkaline solu-
tions, as occur in the Jaffe reaction for creatinine, the main absorption bands shifts
10–30 nm to shorter wavelengths (called a hypsochromic shift) and another absorp-
tion band appears around 280–300 nm [3]. In alkaline solutions, when bilirubin is
fairly dilute, its main absorption band is 440 nm, but if the concentration of the bili-
rubin is increased, it develops a new shoulder in the absorption band around 520 nm,
because bilirubin begins to dimerize [3]. When bilirubin concentration exceeds its
solubility in a solution of neutral pH, its extinction coefficient decreases and another
shoulder is seen around 490 nm [3]. At this point as the bilirubin concentration
increases above it’s solubility, bilirubin self-aggregates into particles sufficient in size
to scatter light [3]. Bilirubin is light sensitive, and upon exposure, will lose color and
form maleimides and propentdyopent adducts as well as a small amount of biliverdin
[3].
The liver readily excretes biliverdin in amphibians, birds and fish, but in humans
most of the biliverdin is converted to bilirubin. Bilirubin itself is less polar than biliver-
 6.3 Transport of Bilirubin in the Blood       65

din and thus more readily crosses placental membranes [3]. Bilirubin does serve a role
as an anti-oxidant in humans. Bilirubin is unstable in aqueous solutions and forms
colloids or surface films, an observation readily available to anyone who has tried to
dissolve bilirubin in water without a solubilizing agent [3]. Bilirubin, which contains
two aldehyde groups, would be expected to be soluble in most polar solvents, but it
loses this solubility when it forms strong hydrogen bonds. When the hydrogen bonds
in bilirubin are broken, bilirubin shows good solubility in polar solvents [3]. Bilirubin
does not dissolve well into fat or nonpolar lipids, but it does adhere to the phospho-
lipids found in fats, allowing it to stain lipids and lipophilic materials [3].
It takes three days for heme to be converted into bilirubin [3]. Approximately 80 %
of the bilirubin is derived from the destruction of circulating erythrocytes, the other
20 % is derived from ineffective erythropoiesis, myoglobin and porphyrin structures.
The percentage derived from ineffective erythropoiesis increases in certain condi-
tions, such as congenital dyserythropoietic anemias, megaloblastic anemias, iron
deficiency anemia, and lead poisoning [3].

6.2 Bilirubin Toxicity

Bilirubin itself is a toxic molecule. It inhibits RNA synthesis, protein synthesis, car-
bohydrate metabolism in the brain and uncouples oxidative phosphorylation and
ATPase activity in brain mitochondria [3]. Bilirubin can act in vitro and inhibit hydro-
lytic and dehydrogenase enzymes [3]. For newborns, toxicity occurs at 340 μmol/L
and above [3].
Kernicterus occurs in young infants exposed to high concentrations of unconju-
gated bilirubin. It also may occur in adults. In infants the signs and symptoms of ker-
nicterus are hypotonia, hyporeflexia, athetoid movements, and reflex opisthotonos,
progressing to lethargy, atonia, and death [3]. If the infant survives the episode of
kernicterus, he or she may suffer from hearing loss, cranial nerve palsy, athetosis and
mental retardation [3]. At autopsy, staining of the hippocampus, basal ganglia and
cerebellar nuclei is frequently observed, while in chronic cases there is necrosis of the
neurons and glia accompanied by gliosis [3].

6.3 Transport of Bilirubin in the Blood

Bilirubin is insoluble in water, but is transported in the blood by binding to albumin [2].

Two types of complexes emerge between albumin and bilirubin. First, bilirubin binds
as a dianion, in a reaction that is fast and reversible; second, at pH below 7.4 albumin
aggregates with bilirubin [3]. Hydrogen peroxide rapidly destroys unbound bilirubin,
but bound bilirubin is more resistant [3]. Certain agents compete with the binding
sites on albumin for bilirubin, and the effects of these agents must be considered.
66       6 Origin of Icteric Samples

Sulfonamides, anti-inflammatory drugs, and contrast media displace bilirubin from

albumin and may lead to kernicterus in infants with elevated bilirubin [3].

6.4 Uptake of Bilirubin by the Liver

On encountering hepatocytes, the bilirubin leaves the albumin and enters the hepato-
cyte, bound to cytoplasmic proteins. The transfer of bilirubin from the blood to the
hepatocyte occurs rapidly, 90 % of injected radiolabeled bilirubin leaving the blood
in 15 minutes [3]. Bilirubin, along with other organic anions, is stored in the liver
bound to ligandin [3]. The enzyme glucuronyl transferase conjugates one or two glu-
curonic acids to bilirubin to form a monoglucuronide or diglucuronide bilirubin [2].
The glucuronide bilirubin is now water soluble, some enters the blood where it is
bound by albumin, while the majority of the bilirubin glucuronide is secreted into the
bile duct where it enters the intestine and is eventually excreted [2]. The rate limiting
step for excretion of bilirubin into the bile is canalicular transport, not glucuronide
conjugation, because of studies in rats where unconjugated bilirubin was injected
intravenously and concentrations of conjugated bilirubin increased [3]. Most of the
conjugated bilirubin is excreted; however, urobilinogen is reabsorbed by the intes-
tine, where it circulates in blood, until it is finally excreted in the urine [2]. Intesti-
nal bacteria degrade bilirubin into urobilinogen and related products [3]. Normally,
unconjugated bilirubin is cleared quickly from the blood, with a half-life of 18 minutes
in humans [4]. A third form of bilirubin may also circulate. This form of bilirubin
has a half-life the same as albumin and has been demonstrated in humans and rats.
In rats the half-life for this albumin is 2 days, the same as for that of albumin [4].
In humans, the half-life of albumin is 19 days. Unconjugated bilirubin is noncova-
lently bound to albumin, however, this third form of bilirubin is covalently bound to
albumin, which explains its longer half-life and represents a persistent hyperbiliru-
binemia in some patients. In addition, covalently attached bilirubin does interfere
with dye-binding assays for albumin by competing for dye [4]. In addition, unless the
protein is removed, covalently attached albumin has the potential to interfere with
other absorbance based assays.

6.5 Clinical Aspects of Bilirubin

Jaundice is detectable when the bilirubin concentration exceeds (34 μmol/L) [2].
Clinically, information about patients occurs based on demographic factors, such
as pediatric or adult populations or the presence or absence of varying types of dis-
eases, thus, the classification of jaundice is distributed among different types of cat-
egories of disease, such as neonatal, cholestatic, infectious, hepatitis etc.
 6.6 Neonatal Jaundice       67

1. Excess production as in hemolytic anemia

2. Decreased uptake of bilirubin into hepatocytes
3. Disturbed intracellular protein binding or conjugation
4. Disturbed secretion of conjugated bilirubin into the bile canaliculi
5. Intrahepatic or extrahepatic bile duct obstruction

Tab. 6.1: Classification of jaundice.

6.6 Neonatal Jaundice

In the neonate bilirubin is elevated and is basically unconjugated; in a study of 4,000

consecutive infants, 16 % had a serum bilirubin concentration of 171 μmol/L or higher,
while 5 % had concentrations exceeding 257 μmol/L [5]. In normal infants, the biliru-
bin concentration rises rapidly after birth for the first day or two, peaking at a concen-
tration of 86–103 μmol/L at the third day, and eventually reaches normal concentra-
tions by a week to ten days [6]. The rise of bilirubin in infants results from increased
production of bilirubin, from such entities as increased absorbance of bilirubin by
the intestines because infants lack the same bacteria as adults that degrade bilirubin
in the intestinal lumen, and hemolytic disease of the newborn, as expressed by Rh
incompatibility, and an immature liver which cannot handle the bilirubin load  [3].
Hepatic bilirubin uptake capacity is reduced during an infant’s first day, slowing
down the ability of the liver to take in the unconjugated bilirubin, and the conjuga-
tion process in the neonatal liver is slower than in adults [3]. Like adults, the excre-
tion of bilirubin into the biliary canal is a rate limiting step, so as the conjugation of
bilirubin in the infant hepatocyte improves, the bilirubin load increases as well and
the entire system backs up, increasing the concentration of conjugated bilirubin in
the blood [3].
The most common cause of hyperbilirubinemia occurs from overproduction. In
this case the bilirubin is unconjugated. Unconjugated bilirubin can be measured clin-
ically by measuring the total bilirubin and the direct bilirubin. Direct bilirubin cor-
relates with conjugated bilirubin, while total bilirubin correlates with the sum of the
conjugated and the unconjugated bilirubin. Therefore, if one wanted to determine the
concentration of unconjugated bilirubin, one could do that by subtracting the direct
bilirubin from the total bilirubin.
With increased production of bilirubin, the concentration usually does not go
higher than 51–68 μmol/L [3]. The most common causes of overproduction result from
hemolytic anemia, as seen in sickle-cell anemia, hereditary spherocytosis, and toxic
drug reactions [3]. The destruction of red cells produces so much free heme, that the
excretory system of the liver is overwhelmed.
Three diseases typify the inherited causes of unconjugated hyperbilirubine-
mia found in infants (Tab. 6.2). These diseases are Crigler-Najjar types I and II and
the Gilbert syndrome [3]. Crigler-Najjar type I is caused by an abnormal UGT1 gene
68       6 Origin of Icteric Samples

(uridine diphosphoglucuronate glucuronosyltransferase, UGT) which may result

from isoforms or deletion of nucleotides [3]. Because there is an absence of any UGT
enzymatic activity, all the bilirubin remains unconjugated. These patients typically
have concentrations of total bilirubin around 342  μmol/L, all of it total bilirubin,
and none of it direct bilirubin [3]. These patients develop kernicterus and bilirubin
encephalopathy. The most effective treatment is plasmapheresis, but orthotopic liver
transplant is the only long term solution [3].

Feature Crigler-Najjar I Crigler-Najjar II Gilbert Syndrome

Bilirubin, μmol/L 342–855 <342 <51
Bile Pale Increased Increased
monoglucuronide monoglucuronide
Glucuronosyltranferase Absent Markedly reduced Reduced
activity
Inheritance Autosomal recessive Autosomal recessive Autosomal dominant
Prevalence Rare Uncommon Common (<5 % of the
population)
Prognosis Kernicterus Usually benign Benign

Tab. 6.2: Neonatal unconjugated hyperbilirubinemia (modified from a table in [3]).

Crigler-Najjar type II, also known as the Arias syndrome, is similar to type I except
that it is less severe [3]. Half of patients develop icterus by one year of life, but for
many patients diagnosis is made later in life, sometimes even as late as 30 years of
age [3]. In Crigler-Najjar type II the concentration of bilirubin usually ranges between
137–308 μmol/L and most of the bilirubin is unconjugated. The bilirubin in bile that
is conjugated is monoconjugated instead of diconjugated [3]. There are no apparent
morphological signs, with a normal appearing liver and lack of kernicterus [3]. Studies
have shown that there is a genetic lesion in the variable region of UGR1, resulting in
one amino acid substitution.
Gilbert syndrome is characterized by a mild to moderate unconjugated hyperbili-
rubinemia; it is also called constitutional hepatic dysfunction or familial nonhemo-
lytic jaundice [3]. Gilbert syndrome typically presents in young adults, with total and
unconjugated bilirubin <51 μmol/L that fluctuates over time and may increase during
another illness [3]. The syndrome represents a group of heterogeneous disorders,
some of which have an anion uptake defect [3]. In some subjects, there is a genetic
defect in the promoter region of the UGT1A1 gene, where instead of a TATAA box with
six repeats, there are two extra bases of thymidine-adenine [7]. Additional factors,
though, are required for the development of hyperbilirubinemia [7]. The syndrome
also demonstrates hyperbilirubinemia as the result of fasting.
Two inherited diseases are associated with predominantly conjugated bilirubin,
the Dubin-Johnson Syndrome and the Rotor Syndrome. Dubin-Johnson Syndrome
presents with a mild hyperbilirubinemia, typically having concentrations up to
 6.7 Cholestasis       69

342 μmol/L and the evidence of jaundice during physical examination; the degree of
elevation of bilirubin is exacerbated by intercurrent illness [3]. The disease is rarely
detected before puberty, although it has occasionally been reported in neonates; often
in women the disease does not become apparent until a woman receives oral contra-
ceptives or becomes pregnant [3]. In most patients the bilirubin averages between
34–86 μmol/L and more than 50 % of the bilirubin is conjugated (direct bilirubin) [3].
Rotor Syndrome is typified by hyperbilirubinemia that is predominantly conju-
gated with impaired organic acid secretion [3]. Rotor Syndrome follows an autoso-
mal recessive pattern of inheritance. Benign Recurrent Intrahepatic Cholestasis is
characterized by recurrent attacks of pruritus and jaundice with marked elevation of
alkaline phosphatase and bile acids. The liver shows features of cholestasis, such as
accumulation of bile pigment inside hepatocytes.
Some infants develop jaundice because of the ingestion of breast milk, with
serum concentrations of bilirubin reaching as high as 171 μmol/L [7]. Breast milk jaun-
dice may be caused by a missense mutation of UGT1A1 similar to that found in Gilbert
syndrome, triggered by breast milk components, such as pregnane diol or nonesteri-
fied fatty acids [7].

6.7 Cholestasis

Cholestasis is caused by intrahepatic or extrahepatic obstruction and leads to hyper-

bilirubinemia (jaundice), pruritus, intestinal malabsorption, fat soluble vitamin
deficiencies, along with the elevation of alkaline phosphatase and gamma-glutamyl
transpeptidase (GGT) [8]. Stones or tumors may block the bile duct, causing obstruc-
tion and cholestasis. Stones caught in the bile duct, associated with gall bladder
stones, are common causes of jaundice. Carcinoma at the head of the pancreas may
impinge on the bile duct, leading to obstruction and jaundice. Carcinoma at the head
of the pancreas and focal chronic pancreatitis are common causes of painless jaun-
dice [9].
Cholestasis beginning in infancy with severe pruritus and inherited in families
may be caused by a defect in the ATP8B1 gene with impaired bile secretion, resulting
in Progressive Familial Intrahepatic Cholestasis 1 (PFIC) [8]. PFIC 2 is caused by muta-
tions in the canalicular bile salt export pump. These patients suffer from pruritus and
growth failure, leading eventually to cirrhosis at an early age [8]. In PFIC 3 there is an
absence of secreted phosphatidylcholine in the bile and damage of the biliary tree
epithelia, the latter of which release GGT into the circulation [8].
70       6 Origin of Icteric Samples

6.8 Hepatitis

There are five known viruses that cause infectious hepatitis, labeled Hepatitis A, B,
C, D, and E.
Hepatitis A and E are contracted through the fecal-oral route, with hepatitis A
usually transmitted through contaminated food or water. Hepatitis B is transmitted
through blood contamination or parenterally as well as through sexual contact. Hepa-
titis C is transmitted parenterally and is common in intranasal cocaine drug abuse.
Hepatitis D is transmitted parenterally. Hepatitis A accounts for 25 % of cases of acute
hepatitis worldwide and thirty to fifty thousand cases a year in the United States
[8]. One-third of the world’s population has been infected with hepatitis B and 400
million have a chronic infection [8]. All of these causes of hepatitis cause acute hepa-
titis, while most cases of chronic hepatitis are caused by hepatitis C, with fewer cases
caused by hepatitis B and uncommonly hepatitis D [8]. Hepatitis A takes 2–4 weeks
for symptoms to appear; hepatitis B, 1–4 months; and hepatitis C, 7–8 weeks [8]. All
the forms of hepatitis are typified by the development of jaundice, accompanied by
elevations in AST and ALT, as the first signs and symptoms. Approximately 0.5 % of
cases of hepatitis B result in a fulminant form with massive hepatic necrosis, which
can lead to death [8].
Many patients infected with hepatitis C go on to a chronic state of infection. These
patients may go on to permanent liver damage and eventually demonstrate jaundice
[8]. Many patients with HIV infections are also infected with hepatitis B (10  %) or
hepatitis C (30 %) [8]. Such patients eventually develop jaundice and lipemia.
Bacteria can cause ascending cholangitis. Malaria, schistosomiasis, strongly-
loidiasis, cryptosporidiosis, leishmaniasis, echinococcosis, and liver flukes all can
cause hepatic infections [8]. Amoeba may cause abscesses. These infections may or
may not be associated with jaundice. Exposure to viral infections may give rise to an
autoimmune hepatitis. Autoimmune disorders such as celiac disease, systemic lupus
erythematosus, rheumatoid arthritis, thyroiditis, Sjögren syndrome and ulcerative
colitis can affect the liver [8].

6.9 Alcoholic Liver Disease

While infectious hepatitis infections, mainly hepatitis C, cause liver destruction and
eventually cirrhosis, chronic ethanol ingestion causes considerable morbidity and
mortality throughout the world as well. Alcoholic hepatitis is an acute disorder and is
associated with elevated alkaline phosphatase and hyperbilirubinemia [8]. Alcoholic
cirrhosis is typified by progressive hepatics failure of which hyperbilirubinemia plays
a significant part [8]. Cirrhosis is a fatal disease and the bilirubin concentrations con-
tinue to rise as the disease progresses. In alcoholic cirrhosis, the bilirubin concentra-
 6.11 Drug Induced Hyperbilirubinemia       71

tion rise as high as 342–513 μmol/L. In many laboratories, the bilirubin concentration
in those patients exceeds the reportable range.

6.10 Hemolysis

Hemolysis is a common cause of overproduction of bilirubin and is typified by uncon-

jugated bilirubin [10]. The bone marrow can increase its production of red cells by at
most 6–8 fold. Hemolytic anemias are classified into hereditary and acquired. Exam-
ples of hereditary hemolytic anemias are those related to erythrocyte membrane
abnormalities, hemoglobinopathies and enzymopathies [11]. Examples of acquired
hemolytic anemias are immune and nonimmune; nonimmune examples include
microangiopathy, physical agents, chemical agents, infectious agents, plasma lipid
abnormalities, hypersplenism and paroxysmal nocturnal hemoglobinuria [11].
Dyserythropoiesis, which is an inefficient erythropoiesis and where immature red
cells fail to reach maturity, is a fairly uncommon cause of hyperbilirubinemia [10].
Causes of dyserythropoiesis are thallasemia, vitamin B12 deficiency, folate deficiency,
aplastic anemia, and myelopdysplasia [10].

6.11 Drug Induced Hyperbilirubinemia

Drug-induced hepatic injury is the most frequent reason for withdrawal from the
market of an approved drug [12]. Drug-induced hepatic injury accounts for 50 % of the
cases of acute liver failure in the United States. The liver is the most important organ
for metabolizing almost all foreign substances; the liver transforms the drugs through
oxidative pathways involving the cytochrome P-450 enzyme system [12]. The liver
makes these metabolites water soluble by conjugating a glucuronide, a sulfate, or a
glutathione; then the drug metabolite is exported into the bile or plasma [12]. Most
frequent hepatotoxic drug reactions cause moderate to severe hepatocyte injury and
clinically resembles viral hepatitis, characterized by malaise, jaundice and elevated
liver enzymes [12]. Cholestatic symptoms are most likely to raise bilirubin concentra-
tions and they are also associated with a rise in alkaline phosphatase [12].
Different drugs demonstrate different diagnostic patterns. The liver injury is con-
sidered hepatocellular if the ALT rises three times above normal as the initial sign.
Drugs that cause this type of injury are acetaminophen, allopurinol, amiodarone, iso-
niazid, ketoconazole, NSAIDs, statins and valproic acid as examples [13].
Acetaminophen demonstrates a dose-related toxicity to the bile secretory com-
ponents, and accounts for 39 % of cases of acute liver failure [12]. Acetaminophen’s
toxicity is exacerbated by chronic alcohol abuse, phenytoin, isoniazid or starvation
[12]. Other drugs that demonstrate a dose-dependency for toxicity are amiodarone,
72       6 Origin of Icteric Samples

bromfenac, cocaine, phencyclidine, cyclophosphamide, cyclosporine, methotrexate,

niacin, and oral contraceptives [12].
Drugs that cause a cholestatic pattern of injury raise the alkaline phosphatase the
most, with a two-fold elevation of total bilirubin above the upper limit of the reference
interval [13]. Examples of this type of injury are amoxicillin-clavulanic acid, anabolic
steroids, clopidogrel, oral contraceptives, phenothiazines and tricyclics [13]. Some
drugs show a mixed pattern with elevation of both ALT and alkaline phosphatase,
examples are captopril, carbamazepine, clindamycin, phenobarbital, phenytoin,
sulfonamides, trazodone, and verapamil [13]. In all of these types of injuries, when
severe, jaundice and hyperbilirubinemia occur.

6.12 Summary

There are many causes of hyperbilirubinemia in patients, ranging from genetic to infec-
tious to cholestatic and cirrhosis. There are a wide range of diseases that are associated
with fairly low concentrations of bilirubin, still abnormal, from 26–86 μmol/L [14].
Levine and Klatskin studied 366 patients with unconjugated hyperbilirubinemia to
ascertain the cause [14]. In their study, they found that 33 patients had cholecystitis
(acute in ten and chronic in 23), four had pancreatitis (two had chronic and two had
acute), 13 were chronic alcoholics, three had cirrhosis, 86 (by far the largest group)
had cardiovascular disease, with or without heart failure, nine had compensated
hemolysis, eight had hemolytic anemia, 38 had acute or chronic infections, 23 had
cancer, five had diabetes, and four had hyperthyroidism [14] (In about 140 they did
not report a cause.). Thirty-three of the 366 patients had no evidence of any associated
disease [14]. The causes of hyperbilirubinemia are extremely varied and the incidence
of hyperbilirubinemia in hospitalized patients is extensive. In a more recent study,
55 % of the cases of jaundice were caused by a hepatic cause and 45 % were due to
extrahepatic causes [15]. In that study, 21  % of cases were caused by decompensa-
tion of a pre-existing liver disease, 17 % were caused by alcoholic hepatitis, 9 % were
caused by an acute viral infection (hepatitis B, hepatitis C, hepatitis A, Epstein-Barr
virus, and HIV), 0.3  % were autoimmune, and 3.9  % were drug-induced (acetami-
nophen, 3.3 %, and HAART, 0.4 %) [15]. Of the non-hepatic causes, 22 % were caused
by sepsis/abnormal hemodynamic state, 14 % by gallstones, 2.5 % by hemolysis and
6 % by malignancy [15]. The incidence of hyperbilirubinemia ranges from five to nine
to 20 %, depending on the patient population [16].

6.13 References
[1] Glick MR, Ryder KW, Glick SJ, Woods JR, Unreliable visual estimation of the incidence and
amount of turbidity, hemolysis, and icterus in serum from hospitalized patients. Clin Chem USA
1989,35,837–839.
 6.13 References       73

[2] Ganong WF. Review of Medical Physiology. 22nd ed. New York, NY, USA, Lange Medical
Publishers, 2005.
[3] Chowdhury JR, Wolkoff AW, Chowdhury NR, Arias IM. Hereditary Jaundice and Disorders of
Bilirubin Metabolism. In: Scriver CR, Beaudet AL, Sly WS, Valle D.,ed. The Metabolic and
Molecular Bases of Inherited Disease. 7th ed. New York, NY, USA, McGraw-Hill, Inc, 1995,2161–
2208.
[4] Reed RG, Davidson LK, Burrington CM, Peters Jr T. Non-resolving jaundice: bilirubin covalently
attached to serum albumin circulates with the same metabolic half-life as albumin. Clin Chem
USA 1988,10,1992–1994.
[5] Hardy JB, Peeples MO. Serum bilirubin levels in new born infants: Distributions and
associations with neurological abnormalities during the first year of life. Johns Hopkins Med J
Canada 1971,128,265–272.
[6] Gartner LM, Lee K, Vaisman S, Lane D, Sarafu I. Development of bilirubin transport and
metabolism in the newborn Rhesus monkey. J Pediatr USA 1977,90,513–531.
[7] Fabris L, Cadamuro M, Okolicsanyi L. The patient presenting with isolated hyperbilirubinemia.
Dig Liver Dis Netherlands 2009,41,375–381.
[8] Kumar V, Abbas AK, Fausto N, Aster JC, Robbins and Cotran Pathologic Basis of Disease. 8th ed.
Philadelphia, PA, USA, Saunders, 2010.
[9] Patlas M, Deitel W, Taylor B, Gallinger S, Wilson SR. Focal chronic pancreatitis mimicking
pancreatic head carcinoma: are there suggestive features on ultrasound? Can Assoc Radiol J
Canada 2007,58,15–21.
[10] Fevery J. Bilirubin in clinical practice: a review. Liver Int USA 2008,28,592–605.
[11] McKenna RW, Keffer JH. The Handbook of Clinical Pathology. 2nd ed. Chicago, IL, USA, American
Society of Clinical Pathologists, 2000.
[12] Lee WM. Drug-induced hepatotoxicity. New Eng J Med USA, 2003,349,474–485.
[13] Navarro VJ, Senior JR. Drug-related hepatotoxicity. New Eng J Med USA 2006,354,731–739.
[14] Levine RA, Klatskin G. Unconjugated hyperbilirubinemia in the absence of overt hemolysis. Am J
Med USA 1964,36(4),541–552,
[15] Vuppalanchi R, Liangpunsakul S, Chalasani N. New onset jaundice. Am J Gastroenerol USA
2007,102,558–562.
[16] Glick MR, Ryder KW, Glick SJ. Interferographs. User’s Guide to Interferences in Clinical
Chemistry Instruments. 2nd ed. Indianapolis, IN, USA, Evaluation Sciences, Inc, 1991.
7 Impact of Icterus

7.1 Introduction

Icterus refers to the yellow color of serum or plasma specimens with high levels of
bilirubin. Hyperbilirubinemia occurs frequently in critically ill patients and is com-
monly encountered in the clinical laboratory. It is widely known that bilirubin can
interfere with many different laboratory tests. As described in Chapter 6, bilirubin
interfere occurs via two different mechanisms, chemical and spectral. Spectral
interference occurs because bilirubin has two absorption peaks at a wavelength of
400 and 540 nm. Any test that relies on absorbance measurements at or near these
wavelengths may be subject to interference. In addition, bilirubin may also inter-
fere through chemical mechanisms due to its interaction with hydrogen peroxide
(H2O2) [1–4]. It is also possible for both of these mechanisms to occur simultaneously.
Regardless of the mechanisms, high levels of bilirubin in patient samples can cause
erroneous results.
It is important to recognize that bilirubin is heterogeneous. It can occur in mono-
glucuronide-conjugated, diglucuronide-conjugated, and unconjugated forms. In the
case of neonates, it is also possible to encounter photobilirubin or bilirubin photoiso-
mers, which are structurally different than conjugated and unconjugated forms and
may not be detected the same as other forms [5]. The problem with this heterogeneity
is that most interference studies, which are used to quantify the extent of interfer-
ence, do not determine the effect of each type of bilirubin on a test. As described in
Chapter 5, studies typically rely on unconjugated ditaurobilirubin. This bovine biliru-
bin has different chemical and structural properties than human bilirubin (Fig. 7.1).
Not surprisingly, these structural differences sometimes translate to differential inter-
ference with laboratory tests. This chapter describes the reported effects of icterus of
the accuracy of patient results, providing insight into study design limitations and
general principles that can be applied in the laboratory.

7.2 Estimated Impacts Based on Interference Studies

Icterus has been recognized as a clinically significant interferent for several decades.
One widely reported example of icterus interference is with the kinetic Jaffe (alka-
line picrate) method for creatinine determination. In one study, it was reported that
patient samples with icterus >300 μmol/L had a negative bias of 10–50 % for creati-
nine [2]. The authors concluded that this bias would be clinically significant in a few
patients with high levels of icterus and could obfuscate detection of renal damage.
An important side note of this study was that the authors determined the effect of
icterus on creatinine by comparing the Jaffe method to HPLC, which is unaffected
76       7 Impact of Icterus

HO OH
O O S
O S O

O
NH
NH
O

ditaurobilirubin
NH HN

N N
O H H O

HO O O OH

H3C CH3

N N
H H
unconjugated bilirubin H H
N N

H3C OO

H3C CH2
CH2

H H OH HO
H OH
HO H H H
O HO
H OH
HO O O H
OH O
H H
O O O O

H3C CH3

N N
H H
bilirubin diglucoronide H H
N N

H3C OO

H3C CH2
CH2

Fig. 7.1: Structural differences between ditaurobilirubin, human unconjugated bilirubin ((4Z,15Z)-Bili-
rubin IXa), human conjugated bilirubin (bilirubin diglucuronide). Ditaurobilirubin is frequently used
for interference studies despite its physico-chemical differences from that encountered clinically.
Molecular weights, refractivity, the three pKas, and pI are different between these structures.
 7.3 Differential Interference with Different Bilirubin Isoforms       77

by icterus. This approach is ideal for determining interference as it avoids issues

with spiking non-physiological isoforms or concentrations of bilirubin. Because this
approach requires an unaffected reference method, it is difficult to achieve in cases
where robust reference methods or icteric samples are not readily available.
A recent publication that did use an appropriate reference method approach
to detect interference (liquid chromatography-isotope dilution mass spectrometry;
LC-IDMS) supports that icterus remains a significant problem with some creatinine
methods [6]. As many as 95 % of icteric samples (bilirubin 137–560 μmol/L) showed
significant (>10 % or >8.8 μmol/L) bias for some methods. Of the 7 different methods
tested (4 enzymatic and 3 Jaffe), 5 showed negative bias, 1 was unaffected, and 1
showed positive bias. This well-designed study emphasizes the need to understand
how interference affects individual assays. It can be dangerous to make assumptions
from literature regarding the effect of icterus on tests from different vendors. For
example, it has previously been suggested that enzymatic creatinine methods were
largely immune to icterus interference. However, several reports support that not
only is the enzymatic method subject to interference, but sometimes it may be more
affected than the Jaffe method [7–6]. In another study, the majority of samples ana-
lyzed by the enzymatic creatinine method showed significant negative bias (≥10 %) in
spiked and patient samples at bilirubin concentrations of >400 μmol/L. While even
lower bilirubin concentrations (>200 μmol/L) affected creatinine at low concentra-
tions, the Jaffe method was unaffected by total bilirubin concentrations >700 μmol/L
[8]. Collectively, the evidence supports that test users must determine the effects of
icterus on their own assay.

7.3 Differential Interference with Different Bilirubin Isoforms

The difference between the effects of different forms of bilirubin has been highlighted
in several published studies. Some methods to detect phosphorus rely on formation
of ammonium molybdate complexes at a wavelength of 340 nm. It has been observed
that there was significantly lower (0.13–0.32 mmol/L) phosphorus in icteric patient
samples as compared with a method using a reduced complex measured at 680 nm
[9]. Attempts to replicate this interference with spiking studies showed that addition
of unconjugated bilirubin to healthy patient samples did not affect the measured
phosphorus concentration. In contrast, the addition of conjugated bilirubin effec-
tively mimicked the icteric samples. While dual wavelength measurements have since
all but eliminated icterus interference with phosphorus methods, the study does
highlight how different forms of bilirubin have differential effects on test results.
If structural differences with conjugated and unconjugated bilirubin seem over-
whelming, things become truly complex when it comes to photoisomers. There are
five isomers of bilirubin, which include 4Z,15Z-bilirubin, 4E,15Z-bilirubin, 4E,15E-bil-
irubin, as well as Z- and E-lumirubin (Fig. 7.2). The most abundant or ‘native’ form is
78       7 Impact of Icterus

O O O O
HO OH HO OH
4E, 15E-bilirubin 4Z, 15E-bilirubin

H3C CH3 H3C CH3

H NH HN H NH HN H
O N N O N O
CH3

H2C CH3
CH3 HN
CH3
CH2 CH2 CH2
O

HO O O OH
O O
HO OH 4Z, 15E-bilirubin
4E, 15Z-bilirubin

H3C CH3
H3C CH3
N N
H NH HN H H
O N
H H
H2C N N
H2C
NH H3C OO
CH3 H3C
H3C CH2
O CH2

O O
HO HO HO HO
OH OH
E-lumirubin Z-lumirubin

H3C H3C
CH3 CH3
NH HN H NH HN
N O
H3C H3C
HN
N CH2 N
H3C O O CH3
O

H2C

Fig. 7.2: Bilirubin photoisomers. Isomers are more polar than 4Z, 15Z-bilirubin IXa, making photo-
therapy effective. These physico-chemical differences equate to differential interference with some
tests.
 7.4 Non-spectrophotometric Icterus Interference       79

4Z,15Z-bilirubin. Upon exposure to light, 4Z,15Z-bilirubin is rapidly converted to the

other forms. These photoisomers are more polar and therefore more water soluble
than 4Z,15Z-bilirubin. These chemical changes are the basis for phototherapy where
increased water solubility facilitates clearance of bilirubin. It is reported that 25–30 %
of the total bilirubin occurs as photoisomers in jaundiced neonates treated with
phototherapy [10–12]. However, photoisomerization also occurs in icteric serum or
plasma samples exposed to ambient light during collection and processing. Pho-
toisomerization is of particular concern in small sample volumes, such as capillary
blood collections from neonates. If these samples are not protected from light, then
their high relative surface area will serve to increase relative light exposure and there-
fore the prevalence of photoisomers. Bilirubin photoisomers effectively interfere with
bilirubin measurements because of their differential detection by different methods
[13]. In the presence of photoisomers, calculation of unconjugated bilirubin in neo-
nates may be overestimated by spectrophotometric measurements using dry slides.
In contrast, measurements using diazo-based methods are reportedly unaffected as
they do not react with photoisomers. There is growing evidence supporting the meas-
urement of ‘free bilirubin’ for detection of kernicterus using peroxidase methods. Free
bilirubin refers to the non-protein bound fraction analogous to free thyroxine or free
phenytoin, where the biologically active portion is only a small fraction of the total
and therefore a more appropriate target analyte for measurement. The problem with
‘free bilirubin’ is that peroxidase methods react differentially with bilirubin 4Z,15Z
and the more polar photoisomers, such that measurement may be a challenge in neo-
nates with photoisomers. In addition, bilirubin photoisomers have different spectral
properties such that their detection is lower than 4Z, 15Z-bilirubin. The issue of pho-
toisomers is likely to remain a concern for interference for the foreseeable future.
With the aforementioned differences in bilirubin effects in mind, laboratorians
should recognize that most manufacturers use only one form of bilirubin for interfer-
ence studies. While this is largely consistent with laboratory practice guidelines (e.g.
CLSI EP07), it is important to consider the limitations of this practice during method
validation studies and when handling discrepant results. It is good laboratory prac-
tice to assess the effect of interferences on tests experimentally as part of method
validation. The challenge in this case is acquiring appropriate materials or accessing
unaffected reference methods.

7.4 Non-spectrophotometric Icterus Interference

Icterus does not simply interfere by absorbing light. The structural nature of bilirubin
also makes it possible to cause chemical interference. Several studies have identified
icterus interference with peroxidase-based methods [3, 4, 14]. It was recognized as
early as the 1940s that peroxidase interacted with bilirubin [15]. Some years later,
it was reported that bilirubin caused negative bias in peroxidase-based methods,
80       7 Impact of Icterus

including uric acid, cholesterol, and triglycerides [4]. Further studies also revealed
clinically significant negative bias (≥10 % bias) with as little as 43 μmol/L bilirubin [3].
It was later recognized that analyte-dependent and independent mechanisms were
involved in bilirubin interference and that the effect consistently resulted in negative
bias [14]; this modeling study also identified bilirubin isoform differences in interfer-
ence as described in the previous paragraph. Still without resolution of the problem,
another report demonstrated interference with uric acid and lactate [16]. Uric acid
was variably affected depending on the type of bilirubin added; conjugated ditauro-
bilirubin caused negative bias at a bilirubin concentration of 30 μmol/L and unconju-
gated bilirubin caused positive or negative bias. The variable effect was attributed to
concentration-dependent positive spectral bias in combination with negative chemi-
cal bias. Lactate detection by another peroxidase-based assay was affected variably
depending on the type of bilirubin added. Ditaurobilirubin caused negative bias at a
concentration of ≥60 μmol/L while unconjugated bilirubin caused significant nega-
tive bias at a concentration of >250 μmol/L. For both uric acid and lactate, patient
specimens behaved differently than spiked samples, with a bias threshold in between
the ditaurobilirubin and unconjugated bilirubin. These differences between bilirubin
isoforms as well as the sample and concentration-dependent make it challenging to
generalize the effect of bilirubin in an individual patient. Ultimately, it is generally
accepted that bilirubin can serve as a substrate for peroxidase, such that tests relying
on this principle will have reduced production of H2O2, and thus typically show nega-
tive bias with high concentrations of bilirubin.

7.5 Resolving Icterus Interference

Endogenous interferences, such as icterus, would seem to offer few options for labs
when it comes to affected assays. Clearly sample redraws will not eliminate icterus
interference, as they might in samples with phlebotomy-induced hemolysis. However,
there are a few mechanisms by which results may be salvaged in icteric samples.
Simple dilution can reduce the concentration of icterus for tests which are only mildly
affected by bilirubin. Repeated dilutions are used to reduce the bilirubin concentra-
tion to below the threshold of unacceptable bias. This practice can only be used when
the analyte of interest remains within the analytical measurement range. Thus, in
patients with low concentrations of analyte there may be only a few dilutions avail-
able. As with any additional analytical step, these practices need to be thoroughly
evaluated. Dilution protocols are usually part of a typical method validation, but
special consideration may be required for samples that are already in the measuring
interval with respect to acceptable imprecision and the maximum allowable dilution
to remove an interferent. Another option that has been proposed to remove bilirubin
interference is peroxidase treatment [17]. Considering the preceding section, this may
seem ironic, but peroxidase enzyme does indeed use bilirubin as a substrate such that
 7.7 References       81

it can oxidize bilirubin and eliminate interference. In a study of icterus interference

with the Jaffe creatinine assay, samples were subjected to pretreatment with caffeine
and sodium benzoate to displace bilirubin from proteins followed by the addition
of peroxidase. Remarkably, the addition of bilirubin up to 435 μmol/L did not affect
the accuracy of creatinine measurements. Analytical precision was also largely unaf-
fected by the pretreatment steps. Although such pretreatments are perhaps impracti-
cal for routine use, they may be useful for precious samples. This approach may also
be useful in the context of assay development, where reagents could be added as part
of the analytical assay to remove interferences.

7.6 Summary

Icterus interference remains a clear and present danger for patient care. A wide
number of tests are affected by icterus such that laboratories need to remain vigilant
in detecting and where possible eliminating interference. Targeted dilution protocols
and creative use of peroxidase treatment may serve as useful tools for laboratories
to salvage irreplaceable specimens while maintaining an acceptable level of analyti-
cal accuracy. However, additional treatments are considered home-brew assays that
require thorough validation to avoid introducing different errors. Interference studies
are the basis for estimation of icterus interference and determining the thresholds
reporting or rejecting samples. While useful, interference studies seldom replicate
true physiological icterus, limiting the ability to generalize data yielded from these
experiments to individual patients. Ideally, interference studies would compare
patient samples with icterus between the method of interest and an unaffected refer-
ence method. Given that this is not practical or even possible for some tests, it remains
the laboratory’s responsibility to use caution when reporting and interpreting results
in patients with icterus. As with many other aspects of the laboratory, education of
test users is likely to benefit patient care and reduce the risks of error due to icterus.

7.7 References
[1] Weber JA, van Zanten AP. Interferences in current methods for measurements of creatinine. Clin
Chem USA 1991,37,5,695–700.
[2] Guy JM, Legg EF. Bilirubin interference in determinations of creatinine with the Hitachi 737
analyzer. Clin Chem USA 1990,36,10,1851–2.
[3] Spain MA, Wu AH. Bilirubin interference with determination of uric acid, cholesterol, and trigly-
cerides in commercial peroxidase-coupled assays, and the effect of ferrocyanide. Clin Chem
USA 1986,32,3,518–21.
[4] Witte DL, Brown LF, Feld RD. Effects of bilirubin on detection of hydrogen peroxide by use of
peroxidase. Clin Chem USA 1978,24,10,1778–82.
[5] McDonagh AF, Vreman HJ, Wong RJ, Stevenson DK. Photoisomers: obfuscating factors in clinical
peroxidase measurements of unbound bilirubin? Pediatrics USA 2009,123,1,67–76.
82       7 Impact of Icterus

[6] Greenberg N, Roberts WL, Bachmann LM, Wright EC, Dalton RN, Zakowski JJ, et al. Specificity
characteristics of 7 commercial creatinine measurement procedures by enzymatic and Jaffe
method principles. Clin Chem USA 2012,58,2,391–401.
[7] Dimeski G, McWhinney B, Jones B, Mason R, Carter A. Extent of bilirubin interference with
Beckman creatinine methods. Ann Clin Biochem UK 2008,45,Pt 1,91–2.
[8] Owen LJ, Keevil BG. Does bilirubin cause interference in Roche creatinine methods? Clin Chem
USA 2007,53,2,370–1.
[9] Alvarez F, Whalen K, Scott MG. Conjugated, but not unconjugated, bilirubin negatively interferes
in Hitachi 747 assay of inorganic phosphorus. Clin Chem USA 1993,39,11 Pt 1,2345–6.
[10] Itoh S, Kusaka T, Imai T, Isobe K, Onishi S. Effects of bilirubin and its photoisomers on direct
bilirubin measurement using bilirubin oxidase. Ann Clin Biochem UK 2000,37 ,Pt 4,452–6.
[11] McDonagh AF. Photolysis and photoisomerization of bilirubin in serum specimens exposed to
room lighting. Clin Chim Acta Netherlands 2008,393,2,130, author reply 131.
[12] Mreihil K, McDonagh AF, Nakstad B, Hansen TWR. Early isomerization of bilirubin in
phototherapy of neonatal jaundice. Pediatr Res USA 2010,67,6,656–9.
[13] Gulian JM, Dalmasso C, Millet V, Unal D, Charrel M. Influence of photoisomers in bilirubin
determinations on Kodak Ektachem and Hitachi analysers in neonatal specimens study of the
contribution of structural and configurational isomers. Eur J Clin Chem Clin Biochem Germany
1995,33,8,503–12.
[14] Eng CD, Delgado R, Kroll MH. Complex analyte-dependent and analyte-independent
interferences with conjugated bilirubin in the enzymatic phenol-aminophenazone peroxidase
(PAP) method for creatinine determination. Eur J Clin Chem Clin Biochem Germany
1993,31,12,839–50.
[15] Sumner JB, Nymon M. The oxidation of bilirubin by peroxidase. Science USA
1945,102,2643,209.
[16] Beyne P, Lettéron P, Hervé C, Roullet JB, Delacoux E. Bilirubin interference with determination
of creatinine, lactate, phosphorus, and uric acid on Beckman Synchron CX7. Clin Chem USA
1996,42,6 Pt 1,988–90.
[17] Rajs G, Mayer M. Oxidation markedly reduces bilirubin interference in the Jaffé creatinine assay.
Clin Chem USA 1992,38,12,2411–3.
8 Origin of Lipemia and Turbidity
Evidence of lipemia occurs in approximately 3  % of specimens from hospitalized
patients [1]. Even though personal examination of samples can provide a way to
determine if samples demonstrate lipemia or turbidity, many processes today have
become automated and samples are not personally observed prior to being placed
into the analyzer for analysis or even after determinations have been made. Auto-
mated processes to establish the presence of lipemia or turbidity using interference
indices provide a suitable way to detect samples liable to interference, but such detec-
tion systems are not perfect. Another means to detect samples liable to interference
is to recognize the conditions that lead to lipemia and turbidity. Knowledge of these
conditions can lead to alterations in the way that Laboratory Information Systems or
Electronic Medical Records interact with the laboratory staff. Medical Directors and
Laboratory Professionals should have a broad and deep knowledge base on the bio-
chemistry, physiology and pathology that can give rise to samples with lipemia and
turbidity. VLDL particles and chylomicrons are strongly associated with the triglycer-
ides concentration, though elevated triglycerides per se is not synonymous with the
turbidity seen with lipemia; however, identifying triglyceridemia increases the prob-
ability of identifying those subjects and patients who demonstrate lipemia.

8.1 Lipoprotein Pathways

Triglycerides from dietary fat sources (exogenous) are circulated as chylomicrons,

while those synthesized by the liver are circulated as VLDL particles (Fig. 8.1). Tri-
glycerides are the major store of lipids for both plants and animals. Once ingested
lipids are fairly insoluble, until they reach the duodenum, where they interact with
bile salts, which emulsify them [2]. Next, pancreatic lipase and colipase hydrolyze the
fatty acids from position 1 and position 3 of the triglycerides, resulting in fatty acids
and 2-monoacylglycerol [2]. Pancreatic esterases and phospholipase A2 hydrolyze
cholesteryl esters and phospholipids, respectively. The fatty acids, 2-monoacylglyc-
erols, cholesterol, and lysophospholipids combine with bile salts to form micelles,
which interact with the microvilli on the intestinal epithelial cells, here the various
species of lipid are absorbed, but the bile salts are left behind [2]. Fatty acids that are
short and medium in length (4–12 carbons) are more soluble than the long chain fatty
acids and they are absorbed directly into the intestinal lumen and pass directly into
the portal vein [2].
Once inside the cells the fatty acids and 2-monoacylglycerols are re-synthesized
back into triglycerides. The triglycerides coalesce with Apo B-48, Apo A-1, Apo A-2,
Apo C, Apo E, cholesterol, and phospholipids to form nascent chylomicrons [2]. The
nascent chylomicrons leave the intestinal cells and enter the lymphatic system and
84       8 Origin of Lipemia and Turbidity

apoprotein

phospholipid
monolayer
triglycerides
esterified cholesterol

free cholesterol

Fig. 8.1: Structure of lipoprotein particles.

liver LPL lipoprotein lipase

VLDL LPL
liver
apo B, HDL
E receptor LPL

HPL
HPL HPL IDL

oxidation LDL LDL LDL

dense light
LPL: lipoprotein lipase
HPL: hepatic lipase

Fig. 8.2: Metabolism of VLDL.

eventually enter the bloodstream through the lymphatic duct, which pours the chyle
into the left subclavian vein, which can give rise to postprandial lipemia.
HDL lipoproteins transfer Apo E and Apo C-II to the nascent chylomicrons and
they mature. The Apo C-II in the chylomicrons activates lipoprotein lipase which
hydrolyzes the triglycerides and provides fatty acids for cells [2]. As adipocytes
increase their concentration of fatty acids, they release leptin, which feed back to the
hypothalamus to decrease the appetite [2]. The portion of the chylomicrons remain-
ing after the action of lipoprotein lipase is known as a chylomicron remnant; hepato-
cytes bind and take in the remnants and degrade them in the lysosomes [2]. Chylomi-
crons measure between 75–1,200 nm in diameter, which truly covers the absorbance
range of visible light; chylomicron remnants measure between 30–80 nm in diameter,
which is close enough to have some light scattering capabilities.
The liver forms another particle known as very low density lipoprotein, VLDL,
which is composed of Apo B-100, 80–95 % triglycerides, and a small amount of cho-
lesterol and phospholipids [2]. Lipoprotein lipase again hydrolyzes the triglycerides
 8.2 Classification of Hypertriglyceridemia       85

in the VLDL particles. As the VLDL particles become smaller they are eventually
converted into intermediate density lipoproteins (IDL), which in normal persons is
usually not detectable, and the IDL particles are eventually converted into low density
lipoproteins (LDL), which are taken up by peripheral cells through the LDL receptor
(Fig. 8.2) [2]. VLDL particles range in diameter from 30–80 nm, sufficient to cause the
scattering of visible light.
Insulin acts on adipocytes to increase the secretion of lipoprotein lipase. The
action of lipoprotein lipase on chylomicrons and VLDL results in the hydrolysis of
triglycerides to form fatty acids and glycerol [2]. Those fatty acids are stored in adi-
pocytes as triglycerides, the glycerol circulates to the liver, where it is converted to
triglycerides and packaged into VLDL [2].

8.2 Classification of Hypertriglyceridemia

Dyslipidemia are a common phenomenon, worldwide. In Brazil, 1 out 4 (24 %) adults

demonstrated a dyslipidemia in one study, with rates, depending on the city, ranging
from 21 % to 53 % [3]. 11 % of adults demonstrated borderline triglycerides (1.7–2.3
mmol/L), 15 %, high triglycerides (2.3–5.7 mmol/L) and 1.8 %, very high triglycerides
(equal or greater than 5.7 mmol/L) [3]. The high prevalence of hypertriglyceridemia
means that laboratories will encounter a significant number of samples with lipemia
and turbidity. Hypertriglyceridemia is associated with increased risk of cardiovascu-
lar disease and pancreatitis. Typically hypertriglyceridemia is classified by whether
it has a genetic component, which is termed primary and presented in the section
on the Frederickson Classification, or secondary in nature. Hypertriglyceridemia by
other causes is termed secondary and includes obesity, metabolic syndrome, diabe-
tes, alcohol, renal disease, pregnancy, nonalcoholic fatty-liver disorder, autoimmune
disorders, and medication [4].

8.2.1 Frederickson Classification of Dyslipidemias

Familial chylomicronemia (Frederickson Class Type I) and primary mixed hyperlipi-

demia (Frederickson Class Type V) demonstrate chylomicrons in the circulation even
after 12–14 hours of fasting [4]. In addition, these familial forms demonstrate other
signs of dyslipidemia, including eruptive xanthomata, lipemia retinalis, hepatospeno-
megaly, irritability and recurrent epigastric pain [4]. The samples from these patients
are noteworthy in that they will demonstrate a creamy layer on top of a blood sample
when left in the refrigerator overnight. Fasting triglycerides exceed 20 mmol/L [4].
Familial chylomicronemia is caused by a deficiency in apolipoprotein C-II (Apo
C-II), or the presence of inhibitors to or deficiencies in lipoprotein lipase [5, 6]. Famil-
ial chylomicronemia initially presents in childhood and shows more severe lipopro-
86       8 Origin of Lipemia and Turbidity

tein lipase deficiency, while primary mixed hyperlipidemia presents in adulthood and
shows less severe lipoprotein lipase deficiency [4].
Familial hypercholesterolemia (Frederickson Class Type II) shows an elevation
of LDL and cholesterol, it is a common disorder [7]. Familial combined hyperlipopro-
teinemia (Frederickson Class Type IIB) has a prevalence of 2–5 % [4]. Genetically it
is inherited as an autosomal dominant with variable penetrance; it has been asso-
ciated with the USF1 gene. HDL-cholesterol is low, but VLDL and LDL particles are
increased [4].
Familial chylomicronemia occurs rarely, in about one out of a million persons [4].
Dysbetalipoproteinemia (Frederickson Class Type III) shows chylomicron remnants
and VLDL remnants, it is fairly rare, but the chylomicron remnants may give rise to
turbidity [7]. Primary mixed hyperlipidemia is more common (one out of a thousand
persons) [4].
On the other hand, familial hypertriglyceridemia (Frederickson Class Type IV) is
much more common with a population prevalence of 5–10 % [4]. It is typified by an
increase in VLDL particles, but not an increase in chylomicrons. As of yet its molecu-
lar genetics has not been specified, and it is believed to be polygenic and may depend
on secondary factors. Patients with this disorder present with moderately elevated
triglycerides, from 3–10 mmol/L, and with low concentrations of HDL-cholesterol [4].
Patients with this disorder have an increased risk for cardiovascular disease, obesity,
insulin resistance, diabetes, hypertension and hyperuricemia.
Familial dysbetalipoproteinemia (Frederickson Class Type III) is rarer than these
latter two disorders with a prevalence of about one per 20,000 persons [4]. Biochemi-
cally, familial dysbetalipoproteinemia is typified by the presence of β-VLDL, which is a
triglycerides-rich remnant and known as intermediate-density lipoprotein (IDL) with
equimolar concentrations of cholesterol and triglycerides [4]. Genetically the disease
is associated with homozygous inheritan for a binding-defective APOE E2 isoform,
but requires accompanying factors for expression, including obesity, type 2 diabetes

Phenotype Predominant Lipoprotein Disease

I Chylomicrons Familial LPL deficiency;

Familial apo C-II deficiency
IIa LDL Familial hypercholesterolemia
IIb VLDL and LDL Apolipoprotein E deficiency
Familial combined hyperlipidemia
III IDL Familial dysbetalipoproteinemia
IV VLDL Familial combined hyperlipidemia;
Familial hypertriglyceridemias
V Chylomicrons and VLDL Familial hypertriglyceridemia;
Familial apo C-II deficiency

Tab. 8.1: Classification of lipid disorders (modified from Mais[8]).

 8.2 Classification of Hypertriglyceridemia       87

or hypothyroidism [4]. Patients present with tuberous or tuberoeruptive xanthomata

on the extensor surfaces of their extremities, planar or palmar crease xanthoma and
increased risk of cardiovascular disease. Diagnosis depends on finding an increased
VLDL-cholesterol:triglycerides ratio, E2/E2 homozygosity, and decreased LDL [4].
Hyperlipoproteinemia (Frederickson Class Type V) demonstrates increases in
both chylomicrons and VLDL. It is thought to have a genetic component, but other
diseases, such as diabetes, alcoholism, obesity and renal disease, clearly have an
effect [7].

8.2.2 Obesity, Metabolic Syndrome and Diabetes

Obesity has been associated with hypertriglyceridemia if the additional fat deposits
are mainly abdominal. The triad of hyperinsulinemia, increased Apo B and small
dense, LDL particles increases the risk of cardiovascular disease by 20 times [4].
Approximately 25 % of adults in the United States suffer from the metabolic syn-
drome, a syndrome typified by visceral obesity, insulin resistance, dyslipidemia,
hypertension and a pro-inflammatory/thrombotic state [9]. The World Health Organi-
zation defines the metabolic syndrome as insulin resistance, obesity (BMI >30 kg/m2
or waist to hip circumference ratio >0.90 in men and >0.85 in women, dyslipidemia
(triglycerides >1.7 mmol/L, HDL-C <0.9 mmol/L in men, and <1.0 mmol/L in women),
blood pressure > or = to 140/90 mmHg, and microalbuminuria exceeding 20 μg/
min [9]. The National Cholesterol Education guidelines agree with the World Health
Organization in most points, differing in defining obesity as a waist circumference
greater than 102 cm for men and 88 cm for women, not including microalbuminuria,
and defining hypertension as greater than or equal to 130/85 mmHg [9]. The European
Group for the Study of Insulin Resistance agrees with the World Health Organization
on the major defining points, but differ in defining obesity as a waist circumference
exceeding 94 cm in men and 80 cm in women, triglycerides exceeding 2.0 mmol/L,
not including microalbuminuria, and defining decreased HDL-cholesterol as values
falling below 1.0 mmol/L for both men and women [9]. High plasma triglycerides are
an important feature in all three definitions.
Subjects with central obesity have elevated VLDL-Apo-B secretion compared with
non-obese persons and this is associated with the delayed clearance of IDL (inter-
mediate density lipoprotein), LDL and chylomicron remnant particles [9]. Visceral
obesity increases plasma VLDL concentrations by providing an increase in the lipid
availability from the liver and through a mechanism of insulin resistance, altering
the kinetics of VLDL synthesis and catabolism in the peripheral tissues. Visceral
adipose tissue is lipolytically active and presents an increase of free fatty acids to the
portal vein and ultimately the liver, which in turn stimulates the synthesis of Apo B
by increasing the synthesis of cholesteryl esters and triglycerides. The result of this
activity is an increase in production of VLDL1 particles which are the larger particles
88       8 Origin of Lipemia and Turbidity

and most likely to cause turbidity in serum and plasma samples. A similar increase in
the VLDL1 particles has been noted for diabetes as well [9].
The principle ways that insulin resistance increases the secretion of VLDL par-
ticles is by increasing the amount of free fatty acids reaching the liver, interfering
with insulin inhibitory action on the secretion of Apo B, decreasing the post-trans-
lational degration of Apo B, increasing the expression of microsomal triglycerides
transfer protein, increasing the de novo lipogenensis through the sterol regulatory
element-binding protein-1c (hyperinsulinemia increases expression), and decreas-
ing the expression of peroxisome proliferator-activated receptors [9]. In addition,
insulin resistance acts on peripheral tissue (skeletal muscle and adipose tissue) and
decreases the activity of lipoprotein lipase (LPL) which in turn impairs the catabolism
of triglycerides rich lipoproteins. Insulin resistance decreases the clearance of exog-
enous lipoproteins, thus increasing the competition of chylomicrons and VLDL for
LPL [9]. The decreased activity of LPL decreases the clearance of chylomicrons from
the circulation. Insulin resistance increases the production of Apo B-100 which can
give rise to the overproduction of VLDL [9].
Mice with the ob/ob genotype on a background of low density lipoprotein recep-
tor (LDLR) deficiency (−/−) exhibited elevations in triglycerides and total cholesterol
concentrations; with the average elevation of triglycerides to 11.6  mmol/L [10]. An
increase in the Apo-B-containing broad-beta remnant lipoprotein fraction of lipopro-
teins was the cause of the elevation of the triglycerides and cholesterol, and the hyper-
triglyceridemia was associated with leptin deficiency in these mice [10]. A decrease
in leptin causes insulin resistance and is considered an important mechanism in the
development of diabetes and is probably associated with the development of hyper-
triglyceridemia [11].

8.2.3 Alcohol

Ethanol is known to impair lipolysis and can result in an increase in VLDL with or
without an increase in chylomicrons [4]. Zieve’s syndrome is the presence of icterus,
lipemia and hemolysis occurring after heavy ethanol ingestion; once ethanol inges-
tion is stopped it resolves in about a week [12].
Alcohol stimulates lipolysis in fatty tissues increasing the supply of fatty acids
for the liver [13]. The effect of alcohol in the production of hypertriglyceridemia is
affected by other conditions. The highest triglycerides are found in patients with
obesity and diabetes [13]. Extremely high values for triglycerides, greater than 15.9
mmol/L, are associated with the development of pancreatitis, another cause of hyper-
triglyceridemia [13]. Alcohol induces the synthesis of triglycerides and the production
of VLDL in the liver [14]. Alcohol may act through several different mechanisms, such
as stimulating catecholamines, which exert a lipolytic effect on adipose tissue, inhibi-
tion of lipoprotein lipase and the development of insulin resistance [13].
 8.2 Classification of Hypertriglyceridemia       89

8.2.4 Nonalcoholic Fatty-liver Disorder

Nonalcoholic fatty-liver disorder elevates lipids with elevated triglycerides but

decreased HDL-cholesterol [4]. It is thought that one-third of North Americans dem-
onstrate this disorder. It is believed to be related to oxidative stress, cytokines, and
proinflammatory mediators.

8.2.5 Medications

Medications such as corticosteroids, oral estrogens, estradiol, tamoxifen, androgens,

testosterone, progenstins, glucocorticoids, cyclosporine, tacrolimus, isotretinoin,
noncardioselective β-blockers, thiazides, bile-acid-binding resins, cyclophospha-
mide and psychotropic drugs such as phenothiazines and second-generation antipsy-
chotics elevate triglycerides [4, 14]. Patients taking antiviral drugs for HIV infections
also experience an increase in triglycerides.

8.2.6 HIV Infection

Patients treated for HIV infection often develop hypertriglyceridemia. Treatment with
Highly Active Antiretroviral Therapy (HAART) has turned HIV infection from a ter-
minal illness to a chronic infection. Hypertriglyceridemia along with low HDL-cho-
lesterol and hypercholesterolemia are common in patients with HIV infection even
before therapy is begun [15]. An Italian study found that 37 % of patients treated with
saqunavir hard gel, indinavir or ritonavir developed hypertriglyceridemia [16]. In a
study in West Africa, 17  % of patients presenting with HIV infection had hypertri-
glyceridemia; and after introduction of HAART that percentage rose to 54 % [15]. The
hypertriglyceridemia often presents with marked turbidity and even a creamy layer.
The introduction of omega-3 fatty acids can significantly reduce the hypertriglyceri-
demia [17].
HIV infected patients treated with protease inhibitors (PIs) and nucleoside-
reverse transcriptase inhibitors (NRTIs) develop fat redistribution or lipodystrophy,
hyperlipidemia, insulin resistance and hyperglycemia [18]. Calza et al. found in a
cohort of patients (n = 212) receiving PIs for HIV infection that 38 % of them devel-
oped hypertriglyceridemia, especially those patients receiving ritonavir. The degree
of hyperlipidemia is both dose and duration related, with onset usually within 3–12
months of beginning therapy [18]. As a possible explanation for the development of
hypertriglyceridemia during treatment with PIs, it has been noted that cytoplasmic
retinoic acid-binding protein type 1 (CRABP-1) and low density lipoprotein-receptor-
related protein (LRP) show similar structures to that of the catalytic region of HIV-1
protease [18]. HIV-1 protease shows 58 % homology with the amino acid sequence of
90       8 Origin of Lipemia and Turbidity

the C-terminal region of CRABP-1 and 63 % with LRP [18]. CRABP-1 binds retinoic acid
and presents it to cytochrome P450 (CYP)3A enzymes that convert the retinoic acid
to cis-9 retinoic acid. The latter binds to adipocyte nuclei heterodimer RXR-PPARγ,
which inhibits adipocyte apoptosis, while the PIs bind to the same site and inhibit the
formation of cis-9 retinoic acid. The inhibition of the formation of cis-9 retinoic acid
allows more adipocytic apoptosis, with subsequent release of lipid into the blood-
stream and decreased lipid storage [18].

8.2.7 Renal Disease

The hypertriglyceridemia of renal disease occurs frequently and is accompanied by

increased concentrations of Apo B, Apo C and Apo E with reduced concentrations
of Apo A-1 and Apo A-2 [19]. It is thought to be caused by increased synthesis and
decreased renal clearance of lipoproteins [19]. In nephrotic syndrome, the increase
in triglycerides is thought to arise from decreased catabolism of VLDL [20]. Kronen-
berg has shown that in nephrotic syndrome the apolipoproteins change and that this
change may explain the increase in VLDL. The ratio of apolipoproteins C-III to C-II
increases; Apo C-III is an inhibitor of lipoprotein lipase while Apo C-II is an activator
of this enzyme, a situation that could explain the decreased catabolism of VLDL [21].
In children, chronic renal disease is associated with dyslipidemia. In a study by
Saland et al., they found that 126 out of 391 children with chronic renal disease demon-
strated hypertriglyceridemia, also with increased non-HDL-cholesterol and reduced
HDL-cholesterol; children with a GFR < 30 ml/min were much more likely to demon-
strate the dyslipidemia [22]. In most patients, chronic renal disease is associated with
hypertriglyceridemia and with increased VLDL and chylomicrons. The increase in
VLDL and chylomicrons is thought to be the result of a decreased activity of lipopro-
tein lipase [23]. Lipoprotein lipase adheres to capillary endothelium and catalyzes the
hydrolysis of the triglycerides found in VLDL and chylomicrons. In studies in patients
with chronic renal disease and severe hypertriglyceridemia, the amount of lipoprotein
lipase mRNA, derived from skeletal muscle, myocardium or adipose tissue, was mark-
edly reduced compared with controls [23]. In addition, in these patients, the amount
of mRNA of glycosylphosphatidylinositol-anchored binding protein 1 (GPIHBP1), an
endolium-based molecule that plays a critical role in anchoring the lipoprotein lipase
to the endothelium and binding chylomicrons, was reduced compared with control
subjects [23]. The decreased quantities of mRNA found for the enzyme and anchoring
molecules indicate that the patients with chronic renal disease suffer from deficien-
cies in the enzyme and the anchoring protein; together these deficiencies decrease
the clearance of the triglycerides found in VLDL and chylomicrons, contributing to
lipemia [23].
Many patients with severe renal disease eventually receive a renal transplant. Even
after renal transplant, these patients still suffer from increased lipoprotein concentra-
 8.3 References       91

tions and hypertriglyceridemia [24]. Post-transplant renal patients, for example, show
increased triglycerides with or without statin treatment, with an average triglycerides
measurement of 2.0 mmol/L and a range of 0.5–3.0 mmol/L, with patients receiving
statin drugs showing even higher triglycerides than those post-transplant patients not
receiving statins [25]. For patients with renal failure and post-transplant, such find-
ings are important because cardiovascular risk is the leading cause of death in these
patients, and it is important to the laboratory because they have a high likelihood
of having lipemic samples. These patients’ lipid abnormalities progress, possibly
because of immunosuppressive therapy [26].

8.3 References
[1] Glick MR, Ryder KW, Glick SJ, Woods JR. Unreliable visual estimation of the incidence and
amount of turbidity, hemolysis, and icterus in serum from hospitalized patients. Clin Chem USA
1989,35,837–839.
[2] Lieberman M, Marks AD. Mark’s Basic Medical Biochemistry: A Clinical Approach. 3rd ed.
Baltimore, MD, USA, Walters Kluwer/Lippincott/Williams & Wilkins, 2009.
[3] De Souza LJ, Filho JTDS, de Souza TF, Reis AFF, Neto CG, Bastos DA, Cortes VA, Chalita FEB,
Teixeira CL. Prevalence of dyslipidemia and risk factors in Campos dos Goytacazes, in the
Brazilian state of Rio de Janeiro. Arq Bras Cardiol Brazil 2003,81,257–264.
[4] Yuan G, Al-Shali KZ, Hegele RA. Hypertriglyceridemia: its etiology, effects and treatment. CMAJ.
Canada 2007,176,1113–1120.
[5] Brunzell JD, Miller NE, Alaupovic P. Familial chylomicronemia due to circulating inhibitor of
lipoprotein lipase activity. J Lipid Res USA 1983,24,12–19.
[6] Fojo SS, Brewer HB. Hypertriglyceridemia due to genetic defects in lipoprotein lipase and
apoliprotein C-II. J Intern Med UK 1992,231,669–677.
[7] Meisenberg G, Simmons WH. Principles of Medical Biochemistry. 2nd ed. Philadelphia, PA,
USA, Mosby, 2006.
[8] Mais DD. Quick Compendium of Clinical Pathology. 2nd ed. Chicago, IL, USA, American Society
for Clinical Pathology Press, 2009.
[9] Chan DC, Barrett PH, Watts GF. Recent studies of lipoprotein kinetics in the metabolic syndrome
and related disorders. Curr Opin Lipidol UK 2006,17,28–36.
[10] Hasty AH, Shimano H, Osuga J, Namatame I, Takahashi A, Yahagi N, et al. Severe hypercholes-
terolemia, hypertriglyceridemia, and atherosclerosis in mice lacking both leptin and the low
density lipoprotein receptor. J Biol Chem USA 2001,276,37402–37408.
[11] Kumar V, Abbas AK, Faust N, Aster JC. Robbins and Cotran Pathologic Basis of Disease. 8th ed.
Philadelphia, PA, USA, Saunders, 2010.
[12] Zieve L. Hemolytic anemia in liver disease. Medicine USA 1966,45,497–505.
[13] Bessembinders K, Wielders J, van de Qiel A. Severe hypertriglyceridemia influences by Alcohol
(SHIBA). Alcohol and Alcoholism UK 2011,46,113–116.
[14] Brunzell JD. Hypertriglyceridemia. N Engl J Med USA 2007,357,1009–1017.
[15] Salami AK, Akande AA, Olokoba AB. Serum lipids and glucose abnormalities in HIV/AIDS
patients on antiretroviral therapies. West Afr J Med Nigeria 2009,28,10–15.
[16] Manfredi R, Chiodo F. Disorders of lipid metabolism in patients with HIV disease treated with
antiretroviral agents: frequency, relationship with administered drugs, and role of hypoli-
pidaemic therapy with bezafibrate. J Infect UK 2001,42,181–188.
92       8 Origin of Lipemia and Turbidity

[17] Oliveira JM, Rondo PH. Omega-3 fatty acids and hypertrigllyceridemia in HIV-infected
subjects on antiretroviral therapy: systematic review and meta-analysis. HIV Clin Trials USA
2011,12,268–274.
[18] Calza L, Manfredi R, Chiodo F. Dyslipidaemia associated with antiretroviral therapy in
HIV-infected patients. J Antimicrobial Chemotherapy UK 2004,53,10–14.
[19] Attman PO, Alapovic P. Pathogenesis of hyperlipidemicain the nephritic syndrome. Am J Nephrol
Switzerland 1990,10,69–75.
[20] De Sain-van der Velden MG, Kaysen FA, Barrett HA, Stellaard F, Gadellaa MM, Voorbij HA,
Reijngoud DJ, Rabelink TJ. Increased VLDL in nephritic patients results from a decreased
catabolism while increased LDL results from increased synthesis. Kidney Int USA
1998,53,994–1001.
[21] Kronenberg F. Dyslipidemia and nephritic syndrome: recent advances. J Ren Nutr USA
2005,15,195–203.
[22] Saland JM, Pierce CB, Mitsnefes MM, Floynn JT, Goebel J, Kupferman JC, Warady BA.
Dyslipidemia in children with chronic kidney disease. Kidney Int USA 2010,78,1154–1163.
[23] Vaziri ND, Yuan J, Ni A, Nicholas SB, Norris KC. Lipoprotein lipase deficiency in chronic kidney
disease is accompanied by down-regulation of endothelial GPIHBP1 expression. Clin Exp
Nephrol Japan 2011, Oct 19 [epub ahead of print] PMID 22009636.
[24] Razeghi E, Shafipour M, Ashraf H, Pourmand G. Lipid disturbances before and after renal
transplant. Exp Clin Transplant Turkey 2011,9,230–5.
[25] Kimak E, Halabis M, Baranowicz-Gaszczyk I. Relationships between serum lipid, lipoprotein,
triglyceride-rich lipoprotein, and high-density lipoprotein particle concentrations in post-renal
transplant patients. J Zhejiang Univ-Sci B China 2010,11,249–257.
[26] Kasiske BL, Chakkera HA, Roel J. Explained and unexplained ischemic heart disease risk after
renal transplantation. J Am Soc Nephrol USA 2000,11,1735–1743.
9 Impact of Lipemia/Turbidity

9.1 Introduction

It is widely recognized that turbidity can affect the accuracy of laboratory test results.
If inaccurate results are reported by the laboratory and acted upon by care providers,
there is a risk of harm to patients. This chapter is focused on the reported effects of
turbidity on laboratory results, and some of the challenges in predicting how turbidity
may affect the accuracy of a given result. One of the key concepts to appreciate when
considering the impact of turbidity on laboratory results is that turbidity is caused
by a heterogeneous set of compounds. For example, most interference studies rely
on Intralipid®, a nutritional supplement, as a surrogate for lipemia. Patient samples
that do not have the same lipid composition as this supplement (i.e. all patients
with endogenous lipemia) may or may not display similar interference profiles [1].
A lipemic specimen is typically a result of excess VLDL and chylomicrons, but may
be a combination of the two or include other lipids. Though lipemic specimens are
indeed turbid (refract light and appear partially or completely opaque), the term tur-
bidity also includes samples that have high concentrations of particulate matter, such
as erythrocyte debris, platelets, leukocytes, and fibrin clots, that are not removed
through routine centrifugation (1,500⋅g). Because of the heterogeneity of compounds
that cause turbidity, it can be challenging to estimate the effect of turbidity in an indi-
vidual patient.
Turbidity interferes with laboratory tests through multiple mechanisms includ-
ing volume displacement, solvent partitioning, and light scattering. Light scatter-
ing causes a decrease in transmittance, which translates to higher or lower reported
analyte concentrations depending on the nature of the method. For most common
nephelometric and turbidimetric tests, turbidity causes a positive bias [1], where light
transmittance is inversely proportional to the concentration of the analyte. However,
there may be a negative bias with competitive tests, such as immunoassays for small
molecules or drugs, where transmittance is directly proportional to the analyte con-
centration. The extent of the interference is dependent on the size, concentration,
and type of particle; variable particle size translates to variable spectral interference
making it difficult to predict how a given patient sample will behave [2]. Very high
lipid concentrations that make the sample appear opaque may make it impossible to
get any value for a test result. While tests that rely on nephelometry and turbidimetry
are obvious subjects of interference, any method that uses light for detection may be
affected. Thus, some immunoassays are also affected by turbidity interference due to
light scattering.
Lipemia can also interfere with lab tests by displacing volume. The classic
example of this phenomenon is pseudohyponatremia [3, 4]. In pseudohyponatremia,
high concentrations of large lipid particles displace volume effectively increasing the
94       9 Impact of Lipemia/Turbidity

apparent total volume of the sample. When a portion of the sample is pipetted and
mixed with other reagents, as it occurs in indirect ion-selective electrode (ISE) meas-
urement of sodium, the apparent concentration of the analyte is decreased. This is
because sodium (and other electrolytes) are found only in the water portion of the
sample and the dilution is based on the total volume (Fig. 9.1). Pseudohyponatremia
remains a significant issue given that there are many laboratories using indirect
ISE methods. Volume displacement can also occur with other electrolytes such as
potassium and chloride. While this is a problem with indirect ion-selective electrode
methods, it can be avoided with direct ISE; direct ISE methods do not rely on dilu-
tions but rather the activity of the ion in the sample. While pseudohyponatremia is
relatively well recognized, it is important to be aware that volume displacement can
affect the measurement of other analytes.

lipemic specimen non-lipemic specimen

dilution for measurememt

falsely low
electrolyte
concentration

lipid phase accurate

electrolyte
diluent concentration

aqueous phase

Fig. 9.1: Mechanism of pseudohyponatremia in lipemic specimen.

A third mechanism of interference in turbidity due to lipemia is by partitioning of

lipophilic compounds. It is recognized that lipophilic compounds may preferentially
partition into the lipid fraction. This can occur not only with the analyte of interest,
but also with lipophilic reagents or metabolites. Thus, both the target analyte and
the reagent used for measurement of the target may be affected by partitioning. Par-
titioning is also dependent on lipid composition; compounds partition differentially
 9.2 Estimated Impacts Based on Interference Studies       95

depending on the lipid composition, making it difficult to predict the extent or even
the direction of interference for dissimilar matrices.

9.2 Estimated Impacts Based on Interference Studies

9.2.1 Interference by Light Scattering

Most of the existing data on turbidity interference is derived from the use of non-
endogenous lipid emulsion. It is the industry standard to use the soybean-derived
lipid emulsion Intralipid® to estimate lipemia/turbidity in interference studies [5].
Thus, most manufacturers’ package inserts effectively indicate how Intralipid® affects
a given test. This is not entirely without merit, as there are a number of patients who
receive this as nutritional supplement. However, when the cause of turbidity is not
Intralipid®, this information may be misleading. Direct comparison of Intralipid® and
endogenous lipemia has shown significant differences between the extent of bias [1].
In one study, pooled lipemic samples were compared against Intralipid® spiked into
the same samples. Each pool was assayed for triglycerides, α-1 antitrypsin, cerulo-
plasmin, haptoglobin, transthyretin, and albumin on a turbidimetric chemistry ana-
lyzer. While Intralipid® had no significant effect on any of the measured analytes,
endogenous lipids showed substantial negative bias for transthyretin (up to −52 %),
ceruloplasmin (up to −38 %). It was thought that the mechanism of interference was
reduction of the absorbance due to interaction between the endogenous lipids and the
reagents in the tests. However, the complex composition of endogenous lipids makes
it possible that other mechanisms, such as lipid partitioning or volume displacement
(see Section 9.2.2 below), were also involved. The same study also compared detection
of interference using the ‘L-index’. The detection of interference was largely similar
between Intralipid® and endogenous lipids as was correlation with triglycerides.
However, a few individual samples showed marked differences between triglycerides
and the ‘L-index’, supporting abnormally triglycerides-rich or triglycerides-poor lipo-
proteins. The implications are that turbidity detection indices may not always corre-
late with triglycerides. The more significant message was that Intralipid® cannot be
used to reliably predict the effect of interference for individual patient results.
Other studies have also identified differential interference between lipid emul-
sion and endogenous lipemia [6]. Endogenous lipemia has been shown to affect cre-
atinine measurement using the Jaffe method. The Jaffe method is notoriously sensi-
tive to interferences in general and so it is not surprising that turbidity due to lipemia
interferes with this method. What is surprising, is that lipemic interference can vary
tremendously in different patients. In this study [6], the effect of lipemia on creatinine
values ranged from none to −100 % (absurd values were found with some samples,
e.g. −354 μmol/L). As noted in the aforementioned study [1], lipid emulsions and tri-
glycerides concentration were not predictive of the extent of interference. Another
96       9 Impact of Lipemia/Turbidity

important finding was that the age of the lipids affected the extent of the interference;
overnight storage eliminated the interference. This means that even studies that use
endogenous lipids may not identify clinically significant bias if the materials used are
not identical to those encountered clinically. Because of most of these factors, it can
be conclude that some analytical errors due to hyperlipidemia may go unrecognized
even despite the best attempts of the laboratory to detect turbidity and avoid report-
ing unacceptably biased results. While there is no easy solution to these problems, it
is worthwhile being aware of these potential pitfalls, such that they may be addressed
on a case by case basis when consulted regarding results that are inconsistent with
clinical findings.

9.2.2 Interference by Volume Displacement

The second most common type of interference due to lipemia is volume displace-
ment. The clinical importance of turbidity interference is highlighted in numerous
case reports of pseudohyponatremia [7–10]. For clarity, pseudohyponatremia in this
chapter refers to factitiously low sodium due to hyperlipidemia (as opposed to low
sodium due to cellular shifts in a hyperosmolar state). Pseudohyponatremia effec-
tively amounts to short sampling (not pipetting enough volume), as a high concen-
tration of lipids displaces water. Electrolyte measurements that depend on pipetting
aqueous volume accurately are subject to error in the presence of high lipid (or protein)
concentrations. In a classic pseudohyponatremia scenario [10], a patient with hyper-
lipidemia, for example in pancreatitis, has a measurement of sodium by indirect ISE.
The high lipid concentration results in falsely low sodium values and the patient may
be erroneously treated with hypertonic saline. Because of the erroneous lab value,
the treatment results in iatrogenic hypernatremia and hyperosmolarity putting the
patient at risk for cerebral dysfunction. While this phenomenon is well recognized
in the literature, there are frequently cautionary reports published throughout the
medical literature when the phenomenon goes unrecognized. Pseudohyponatremia
due to turbidity is serious enough that even mild hyponatremia can prompt unneces-
sary treatment and cause significant morbidity and mortality [7].
Besides pseudohyponatremia, it is reported that hyperlipidemia can affect other
commonly measured electrolytes, such as potassium, and chloride among others
(Tab. 9.1) [11]. In one study, sodium and chloride were decreased by ~1 mmol/L and
potassium was decreased by 0.04 mmol/L for every 10 mmol/L increase in total lipid
concentration [12]. Because of the apparent linear relationship they observed, the
authors derived correction formula to correct for lipemia. While the findings of this
study do highlight the effect of lipemia on the accuracy of electrolytes, it does bring
to light an important concern. The practice of correcting laboratory results with inter-
ference is in general strongly discouraged [13]. The heterogeneous lipid composition,
 9.2 Estimated Impacts Based on Interference Studies       97

variable mechanisms of interference, and biologically variability between patients

make it risky to try and correct results for a single patient.

Analyte Direction of Effecta Interferent Tested

α-1 antitrypsin increase/no effect Intralipid®, patient specimes

Albumin increase Intralipid®, patient specimes
ALT increase Intralipid®
AST increase Intralipid®
Calcium increase Intralipid®, patient specimes
Ceruloplasmin decrease Intralipid®, patient specimes
Chloride decrease Intralipid®
Conjugated Bilirubin decrease Intralipid®
Creatinine (Jaffe Method) no effect/decrease Intralipid®, patient specimes
CRP decrease Intralipid®
GGT decrease Intralipid®
HDL-C decrease Intralipid®
IgG increase Intralipid®
IgM decrease Intralipid®
Magnesium increase Intralipid®
Phenytoin decrease Intralipid®
Potassium decrease Intralipid®
Progesterone decrease Intralipid®
Salicylate decrease Intralipid®
Sodium decrease Intralipid®
Testosterone decrease Intralipid®
Theophylline decrease Intralipid®
Total Bilirubin decrease Intralipid®
Transferrin decrease Intralipid®, patient specimes
Transthyretin decrease Intralipid®, patient specimes
Uric Acid decrease Intralipid®
Vancomycin decrease Intralipid®
a
Extent and direction of interference are instrument dependent; list is not exhaustive

Tab. 9.1: Partial list of analytes affected by turbidity.

An approach to both identification and elimination of turbidity interference due

to lipemia is ultracentrifugation. Ultracentrifugation refers to applying centrifugal
force in excess of 30,000⋅g to fractionate the sample; high-speed centrifugation may
also be used where samples are spun repeatedly in microcentrifuges, which achieve
~ 10,000–15,000⋅g. In principle, the sample is fractionated by density. Low density
particles, such as lipids, rise to the top and heavy particles, such as cells, fibrinogen,
or cellular debris sediment to the bottom. In one study, which used ultracentrifugation
to identify analytical interference due to lipemia, clinically significant bias was found
for creatinine, phosphorus, and liver function tests (ALT, GGT) [14]. Lipemic samples
from 110 patients were ultracentrifuged at 40,000⋅g for 18 hrs and the results before
98       9 Impact of Lipemia/Turbidity

and after centrifugation were compared. Predictably, triglycerides and cholesterol

showed the largest changes of 21 % and 7.5 % respectively; lipid particles are of course
rich in these substances. What was not expected, however, was the high number of
statistically significant differences found for other analytes, including phosphorus
(5 %), creatinine (4 %), GGT (3 %), urea (2 %), iron (2 %), ALP (2 %), calcium (1.6 %)
among others. Using error goals based on biological variation, the authors concluded
that the effect of lipemia on phosphorus, creatinine, total protein and calcium was
clinically significant. While the 18 hrs centrifugation times are unlikely to be repli-
cated in clinical practice, it does show that lipemia affects a wide array of analytes.
Ultra-centrifugation is therefore a viable option for determining the effect of turbidity
on analytical tests experimentally.
Unfortunately, ultracentrifugation is not a cure-all for lipemic specimens. Some
analytes are themselves affected by the ultracentrifugation process, which can lead to
inaccurate results. One widely-cited example of this is with coagulation testing (PT,
APTT) [15], where ultracentrifugation causes sedimentation of fibrinogen or factor
VIII when bound to von Willebrand factor. Loss of these key complexes results in
no detectable clot with some non-mechanical methods. Thus, the solution (ultracen-
trifugation) can be worse than the problem (lipemia) if one is not careful. The effect of
lipemia on coagulation tests is itself method dependent, such that ultracentrifugation
is not needed when using some methods. Newer analyzers with photo-optical clot
detection have been shown to be unaffected by lipemia [16]. Because of this method-
dependence, laboratorians need to be aware of how lipemia affects their own in-house
methodology. When significant interference is found, it is essential to experimentally
confirm that the process of removing interference (e.g. ultracentrifugation) does not
cause another analytical error. When properly validated, ultracentrifugation serves as
a useful tool for clinical laboratories to handle lipemic specimens.
Another issue for laboratorians to be aware of is that turbidity indices do not
detect all lipid interferences that cause volume displacement. It is well known that
factitiously low electrolytes may be found in patients with lipoprotein X. Lipoprotein
X is an abnormal lipoprotein that may appear in patients with cholestasis. However,
samples with high concentrations of lipoprotein X may remain clear and thus unde-
tectable by turbidity indices. In one case study [17], the inability to immediately iden-
tify this interference delayed appropriate treatment of a patient with severe obstruc-
tive cholestasis secondary to pancreatic cancer; sodium was reported as 20 mmol/L
lower than the correct value and potassium was reported up to 1 mmol/L lower than
the accurate value. The solution to volume displacement due to lipoprotein X is actu-
ally the same as turbidity due to lipemia, where direct ISEs and ultracentrifugation
ameliorate the problem. Astute clinical awareness of apparent discrepancies between
laboratory results and clinical findings remains essential for detection of all types of
interference.
 9.3 Summary       99

9.2.3 Interference by Lipid Partitioning

Another mechanism by which lipemia can interfere with test results is lipid partition-
ing. Lipid partitioning in lipemic samples has been observed to affect measurement
of both steroids and drugs. Several lipophilic therapeutic drugs have been reported to
be affected by lipemia. This phenomenon was recognized several decades ago where
the nonpolar therapeutic drugs phenytoin and phenobarbital were observed to parti-
tion with lipids causing errors [18]. In a recent comprehensive analysis of interfer-
ence in an array of analytes, lipemia was reported to decrease phenytoin, salicylate,
theophylline, and vancomycin [19]. Interestingly, in that report the authors described
differences between the manufacturer-derived cutoffs and those found in the study,
despite the fact that both used the same interferent (Intralipid®). It is possible that
sample-matrix or lot-to-lot differences in either the manufacturer’s reagent or the
widely-used Intralipid® may be the underlying cause of the differences. With respect
to steroid immunoassays, samples spiked with Intralipid® showed a negative bias of
>10  % for progesterone and testosterone [11]. While the mechanism was not eluci-
dated, it is hypothesized that the bias was due to lipid partitioning. At the least it can
be concluded that patients who have lipemia due to nutritional supplementation may
have falsely decreased steroid results.

9.3 Summary

Estimation of the effects of turbidity is a challenge. The majority of reports in the

literature and most manufacturers use Intralipid® as a surrogate for lipemia despite
its well-known limitations. The heterogeneous nature of lipemia makes replication
of this phenomenon difficult for routine interference experiments. With reports of
matrix instability and the potential for differences between reagent lots, it is likely
that estimating the interference of turbidity will remain a challenge for some time
to come. Challenges aside, it is clear that the accuracy of results is grossly affected
by turbidity for some analytes on some platforms in some patients. Awareness of the
potential for interference is a good first step in identifying these errors. Patient care is
likely to be improved by making more people aware of the phenomenon, particularly
those making clinical-decisions. The ongoing patient risk of unnecessary treatment
of pseudohyponatremia underscores the importance of clinical awareness of uncom-
mon analytical interferences such as turbidity.

9.4 References
[1] Bornhorst JA, Roberts RF, Roberts WL. Assay-specific differences in lipemic interference in
native and intralipid-supplemented samples. Clin Chem USA 2004,50,11,2197–201.
100       9 Impact of Lipemia/Turbidity

[2] Park Y, Grellner WJ, Harris WS, Miles JM. A new method for the study of chylomicron kinetics in
vivo. Am J Physiol Endocrinol Metab USA 2000,279,6,E1258–63.
[3] Aw TC, Kiechle FL. Pseudohyponatremia. Am J Emerg Med USA 1985,3,3,236–9.
[4] Fortgens P, Pillay TS. Pseudohyponatremia revisited: a modern-day pitfall. Arch Pathol Lab Med
USA 2011,135,4,516–9.
[5] Glick MR, Ryder KW, Jackson SA. Graphical comparisons of interferences in clinical chemistry
instrumentation. Clin Chem USA 1986,32,3,470–5.
[6] Hortin GL, Goolsby K. Lipemia interference with a rate-blanked creatinine method. Clin Chem
USA 1997,43,2,408–10.
[7] Lippi G, Aloe R. Hyponatremia and pseudohyponatremia: first, do no harm. Am J Med USA
2010,123,9,e17.
[8] Weisberg LS. Pseudohyponatremia: a reappraisal. Am J Med USA 1989,86,3,315–8.
[9] Coakley JC, Vervaart PP, McKay MR. Factitious hyponatremia in a patient with cholestatic
jaundice following bone marrow transplantation. Pathology UK 1986,18,1,158–9.
[10] Howard JM, Reed J. Pseudohyponatremia in acute hyperlipemic pancreatitis. A potential pitfall
in therapy. Arch Surg USA 1985,120,9,1053–5.
[11] Steen G, Klerk A, Laan K, Eppens EF. Evaluation of the interference due to haemoglobin,
bilirubin and lipids on Immulite 2500 assays: a practical approach. Ann Clin Biochem UK
2011,48,Pt 2,170–5.
[12] Dimeski G, Mollee P, Carter A. Effects of hyperlipidemia on plasma sodium, potassium, and
chloride measurements by an indirect ion-selective electrode measuring system. Clin Chem
USA 2006,52,1,155–6.
[13] Vermeer HJ, Steen G, Naus AJM, Goevaerts B, Agricola PT, Schoenmakers CHH. Correction of
patient results for Beckman Coulter LX-20 assays affected by interference due to hemoglobin,
bilirubin or lipids: a practical approach. Clin Chem Lab Med Germany 2007,45,1,114–9.
[14] Calmarza P, Cordero J. Lipemia interferences in routine clinical biochemical tests. Biochem Med
(Zagreb) Croatia 2011,21,2,160–6.
[15] Arambarri M, Oriol A, Sancho JM, Roncalés FJ, Galán A, Galimany R. Interference in blood
coagulation tests on lipemic plasma. Correction using n-hexane clearing. Sangre (Barc) Spain
1998,43,1,13–9.
[16] Appert-Flory A, Fischer F, Jambou D, Toulon P. Evaluation and performance characteristics of the
automated coagulation analyzer ACL TOP. Thromb Res USA 2007,120,5,733–43.
[17] Sivakumar T, Chaidarun S, Lee HK, Cervinski M, Comi R. Multiple lipoprotein and electrolyte
laboratory artifacts caused by lipoprotein X in obstructive biliary cholestasis secondary to
pancreatic cancer. J Clin Lipidol USA 2011,5,4,324–8.
[18] Creer MH, Ladenson J. Analytical errors due to lipemia. Lab Med USA 1983,14,351–355.
[19] Ji JZ, Meng QH. Evaluation of the interference of hemoglobin, bilirubin, and lipids on Roche
Cobas 6000 assays. Clin Chim Acta Netherlands 2011,412,17–18,1550–3.
10 Endogenous Interferences in Clinical Laboratory
Tests: Icteric, Lipemic and Turbid Samples

10.1 Interference Indices

There are two main methods to detect icterus and turbidity: direct measurement and
visual inspection. Visual inspection has been used in the laboratory for decades once
it was established that hemolysis, icterus, and turbidity affected the accuracy of labo-
ratory results. In fact, this practice remains commonplace today. However, it has been
experimentally proven that visual inspection is inconsistent and inaccurate [1]. In this
widely-cited example, individuals visually inspecting samples varied in their estima-
tions of icterus and turbidity by an order of magnitude; estimates of icterus ranged
from <17–342 μmol/L bilirubin and estimates of turbidity ranged from 1.1–11.4 mmol/L
triglycerides (using a +1, +2, +3, +4 scale). Such inaccuracies are clearly undesirable
for routine use in the clinically laboratory. To ameliorate this problem, manufacturers
have developed automated mechanisms to detect interference in patient samples to
augment or replace visual determinations. Currently, most large chemistry platforms
and some smaller specialized platforms have modules to detect hemolysis, icterus,
and turbidity with automated technology. Automated detection offers numerous
advantages over visual detection, including speed, accuracy, and consistency. Auto-
mated detection of icterus and turbidity (as well as hemolysis) yields shorter turna-
round times, higher precision than visual inspection, is semi-quantitative, and ame-
nable to autoverification. These workflow advantages are discussed in the context of
result reporting (see Chapter 11). Because of these advantages, where possible, it is
recommended that laboratories use automated indices. The focus of this chapter is
defining how indices are derived from absorbance measurements.

10.2 Generating Interference Indices

Existing instruments vary widely in their methodology to detect interferences, but all
essentially rely on the same underlying spectrophotometric principles. Instrument
manufacturers (or laboratories if they are so inclined) establish indices using spectro-
photometric absorbance readings at multiple key wavelengths to detect icterus and
turbidity. The principles of icterus and turbidity detection are based on the absorb-
ance spectra of these compounds (discussed in Chapter 2). All manufacturers use
two or more wavelengths to detect icterus and one or more wavelengths to detect tur-
bidity. Typically, peak absorbance wavelengths of 454 nm for bilirubin and 700 nm
for turbidity are used; additional wavelengths are used to account for the overlap
between absorbance spectra of hemoglobin. Analogous to detection of different forms
of hemoglobin (oxyhemoglobin, deoxyhemoglobin, HbF, methemoglobin, and car-
102       10 Endogenous Interferences in Clinical Laboratory Tests

boxyhemoglobin) in a blood gas analyzer, the contribution of hemoglobin, bilirubin,

and turbidity can be deconvoluted based on subtracting known absorbance at differ-
ent wavelengths using extinction coefficients. Calibration curves are generated using
known amounts of icterus and turbidity using a linear regression to derive indices.
The steps of interference index development are shown in Fig. 10.1.

prepare standards – hemolysate

– bilirubin
– turbidity

record absorbance deconvolution/subtraction

for standards using of absorbance from
multiple wavelengths other interferences

calibration curve regression

– absorbance vs.
concentration

define and establish

‘units’ or ‘bins’
for indices

determine
analyte-specfic
interferences (chapter 5)

Fig. 10.1: Flow chart of the steps to generate interference indices.

10.2.1 Preparation of Standards

Indices are established from calibration curves using samples with known amounts
of interference. Calibration samples are made as described in Chapter 5, by adding
known amounts of ditaurobilirubin and Intralipid®. A bilirubin stock solution of
1,026 μmol/L diluted with buffer (saline typically) is used to make a series of samples
spanning from 0–1,026 μmol/L. It is recommended to use 6 or more concentrations for
the calibration curve (e.g. 0, 43, 120, 257, 513, 1,026 μmol/L). Consideration should be
given to the highest concentration of bilirubin used for the calibration curves. Labora-
tories at tertiary care centers are likely to receive highly icteric samples, such that the
indices should span this range; community hospital labs may use a smaller range, but
manufacturers need to consider all extremes.
Turbidity is simulated by adding Intralipid® to buffer to span a range of 0–10 g/L
with at least 7 concentrations (0, 0.25, 0.50, 1.25, 2.50, 5.00, and 10 g/L) as described
 10.2 Generating Interference Indices       103

in Chapter 5. The materials used for interference test calibration must all be prepared
fresh, as it is known that bilirubin and Intralipid® breakdown over time [2] and yield
different results. Common practice is to prepare bilirubin stock solutions in a dark-
room and protect them from light through the entire process.
Hemolysate should also be prepared for interference index studies to subtract
the contribution of hemoglobin to the total absorbance. Hemolysate is prepared by
‘osmotic shock’ according to CLSI guideline EP7-P as previously described [3]. Hep-
arinized whole blood is collected from healthy volunteers, centrifuged at 1,500⋅g
for 10 min, and the plasma fraction is discarded. The red cells are washed with two
volumes of normal saline and re-centrifuged again discarding the supernatant. Cells
are lysed by freezing the red cells at –20°C for 8–16 hours in a single volume of ddH2O.
The hemolysate is warmed to room temperature and cellular debris is removed by
centrifugation. The hemolysate is then brought to the desired final concentration with
ddH2O. As above, at least 7 concentrations of hemolysate should be made to span the
expected range of hemolysis (e.g. 0, 0.25, 0.50, 1.25, 2.50, 5.00, and 10.0 g/L) hemo-
globin) encountered in a clinical laboratory. Again the purpose of the hemolysate is to
account for the absorbance spectrum overlap with bilirubin.

10.2.2 Data Collection and Deconvolution of Non-Target Interferences

Each set of calibration solutions is measured using the spectrometric system to be

employed recording absorbance readings for key wavelengths. This process should
be done with several replicates using multiple different instruments to allow for inter-
instrument spectrophotometer variability and lot differences between the calibration
materials. Because of the spectrophotometer differences and according to Beer’s Law
(A = abc), where path length affects absorbance, interference indices are not transfer-
able between different types of instrument.
An important part of developing the index calibration is to account for absorb-
ance of other interferences at the selected measurement wavelength(s). Detection
of turbidity is simplified because hemoglobin and bilirubin do not absorb at wave-
lengths above 700 nm. However, for calibration of the icterus index, absorbance
readings should be taken at at least 2 wavelengths because hemoglobin (hemolysis)
and turbidity (lipemia or particulate matter) both absorb light across the absorbance
spectrum of bilirubin (340–500 nm). There are two main approaches to handling
the absorbance of other interferences across the light spectrum. One is to account
for the contribution of hemolysis and turbidity by measuring the absorbance of the
standards at selected wavelengths and subtracting these from the total absorbance at
the wavelength(s) used to measure icterus. Alternatively, indices may be calculated
directly from the measurement of absorbance across a continuous spectrum of wave-
lengths using derivative spectrometry.
104       10 Endogenous Interferences in Clinical Laboratory Tests

10.2.2.1 Subtraction Using Selected Wavelengths

The selected wavelength subtraction approach relies on measurement of absorb-
ance of the 3 sets of standards (hemolysate, bilirubin and Intralipid®) at 2–3 dif-
ferent selected wavelengths. If the icterus index method relies on the peak absorb-
ance of bilirubin at 454 nm, then additional wavelengths could include 550 nm for
hemolysate and 700 nm for turbidity. The absorbance of hemolysate at 550 nm is
plotted against the absorbance of hemolysate at 454 nm and a polynomial regression
is applied (Fig. 10.2). In this example, a 2nd order polynomial regression is used; H0,
H1, H2 are constants determined by the regression and A550 or A454 is the absorbance
of the hemolysate at the specified wavelength.

2
HemolysateA454 = H0 + H1 · A550 + H2 · A550 (10.1)

0.8
hemolysate absorbance

0.6

0.4
(at 454 nm)

0.2

0
0 0.5 1.0 1.5 2.0
hemolysate absorbance (at 550 nm)

0.35
0.30
0.25
turbidity absorbance

0.20
0.15
(at 454 nm)

0.10
0.05
0
0 0.5 1.0 1.5 2.0
turbidity absorbance (at 700 nm)

Fig. 10.2: Hemolysate and turbidity absorbances at peak wavelengths plotted against absorbance
at the wavelength used for determination of icterus (454 nm). Polynomial regressions are used to
calculate corrections factors.
 10.2 Generating Interference Indices       105

Similarly, the absorbance of Intralipid® at 700 nm is plotted against the Intralipid®

absorbance at 454 nm again applying a polynomial regression (Fig. 10.2), where T0, T1,
T2 are constants determined by the regression.

2
TurbidityA454 = T0 + T1 · A700 + T2 · A700 (10.2)

For detection of icterus at 454 nm, the sum of HemolysateA454+TurbidityA454 is equal

to the total absorbance contributions of hemolysate and turbidity at 454 nm. This
sum is then subtracted from the absorbance at 454 nm, yielding the concentration of
icterus alone. Thus, by measuring samples at these 3 wavelengths (454, 550, 700 nm)
the interference index for icterus is corrected for contributions of hemolysis or tur-
bidity when present. In the absence of these other interferences, the values of these
correction factors are negligible. Note that Equations 1 and 2 are both 2nd order poly-
nomials, which were determined to be the best fit for the example data. The best fit for
regression data is determined empirically with each experimental dataset and may
range from simple 1st order polynomials (y = a0 + a1x) to more complex 3rd order (y = a0
+ a1x + a2x2 + a3x3) equations. These equations are simply a tool to calculate and sub-
tract the contribution of a given interferent to the absorbance at another wavelength.

10.2.2.2 Index Calculation Using Derivative Spectrometry

Another option to handle the absorbance contributions of different interferences
across the light spectrum is to use derivative spectrometry. Derivative spectrometry
refers to transforming the absorbance readings over a range of wavelengths with a 1st
derivative. The basis for this approach is that substances with overlapping absorb-
ance spectra, such as hemoglobin and bilirubin, can be differentiated by using the
derivatives in combination with the extinction coefficients. Thus, measurement of
interferences using derivative spectrometry effectively replaces the need for subtrac-
tion. Fig. 10.3 shows the absorbance spectra and 1st derivatives for hemolysate, biliru-
bin, and turbidity.
With this approach, the absorbance of each standard is recorded across the rel-
evant range of wavelengths. For hemolysis and icterus this range is ~400–800 nm.
Again, because hemoglobin and bilirubin do not absorb at >700 nm, turbidity meas-
urement is somewhat simplified (discussed below). For icterus absorbance readings,
the extinction coefficient (ε) can be calculated according to Beer’s Law for each wave-
length where the concentration of interference added is known and the path length
(l)=1 using the following equations.

Extinction coefficient at 1st Wavelength:

"1 = A1 /c1 (10.3)

106       10 Endogenous Interferences in Clinical Laboratory Tests

bilirubin bilirubin 1st derivative

2.0 0.03
absorbance
1.5 0.02
0.01
1.0 0

dA/dl
0.5 앥0.01
앥0.02
0
앥0.03
앥0.5 앥0.04
300 400 500 600 700 800 300 400 500 600 700 800
wavelength (nm) wavelength (nm)

hemoglobin hemoglobin 1st derivative

2.0 0.10
0.08
1.5
absorbance

0.06
1.0 0.04

dA/dl
0.5 0.02
0
0
앥0.02
앥0.5 앥0.04
300 400 500 600 700 800 300 400 500 600 700 800
wavelength (nm) wavelength (nm)

turbidity turbidity 1st derivative

2.5 0
2.0 앥0.01
absorbance

1.5
앥0.02
dA/dl

1.0
0.5 앥0.03
0 앥0.04
300 400 500 600 700 800 300 400 500 600 700 800
wavelength (nm) wavelength (nm)

Fig. 10.3: Hemoglobin, bilirubin, and turbidity absorbance curves and first derivatives.

Extinction coefficient at 2nd Wavelength:

"2 = A2 /c1 (10.4)

Extinction coefficient at nth Wavelength:

"n = An /c1 . (10.5)

Extinction coefficients are thus calculated for interferences at each measured wave-
length. With the extinction coefficients computed, the concentration of each interfer-
ent can be calculated using the vector dot product of the derivative of the absorbance
 10.2 Generating Interference Indices       107

and the extinction coefficient. The equation to calculate the vector dot product for
icterus is as follows:


dA
Icterus = · ["]Icterus (10.6)
d

The vector dot product is summed for all measured wavelengths as follows:

800
 
dA
Icterus · ["]Icterus i (10.7)
i =400
d i

The use of the derivative effectively accounts for the absorbance of other interfer-
ences, eliminating the need for subtraction. Currently, few manufacturers employ this
creative approach.

10.2.3 Establishing Indices and Defining Ranges

For methods employing selected wavelengths, the corrected absorbance readings

are used to establish calibration curves. The absorbance of the standards is plotted
against the interference concentration and a linear regression is applied. The regres-
sion equation can be used to estimate the concentration of a given absorbance reading.
With this information, the indices can now be defined. This involves grouping con-
centrations into ‘bins’ representing a range of interference concentrations as shown
in Fig. 10.4. A significant consideration when establishing indices is the number of
‘bins’ into which the values should be split. For selected wavelength measurements,
this may range from 1–5 to 1–20 depending on the manufacturer. Using the continuous
wavelength approach, the index is the sum of the vector dot product, which is roughly
calibrated to the actual concentration of the interferent; for example icterus indices
from 1–60 correlate with bilirubin concentrations of ~17–1,026 μmol/L).
The range of indices varies between manufacturers by several orders of magni-
tude. The smallest range of indices for turbidity is 1–5, whereas the largest is 1–800.
Icterus indices also vary greatly from 1–5 for the smallest range and 1–60 for the
widest. Currently there are no guidelines or published data supporting the strengths
and weaknesses of either of these approaches, so this remains the subject of opinion.
It is argued that the fewer the number of indices, the easier it is for laboratory staff
and clinical providers to understand them. This is largely based on the concept the
staff are used to visual detection systems using +1, +2, +3, +4 to provide a qualita-
tive measure of an interference (similar to how urine dipsticks are still used). In
favor of wider indices is the concept that more ‘bins’ provide a wider spectrum to
fine-tune rejection thresholds. With this approach, the likelihood of falsely rejecting
a specimen with an acceptable level of interference is reduced. Consider an example
108       10 Endogenous Interferences in Clinical Laboratory Tests

1.5

1.0
absorbance
(at 454 nm)

0.5

0
0 171 342 513 684 855 1026
bilirubin (mmol/L)

1.5

1.0
absorbance
(at 454 nm)

0.5

0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
small icterus bins

1.5

1.0
absorbance
(at 454 nm)

0.5

0
0 1 2 3 4 5
large icterus bins

Fig. 10.4: Calibration curve regression between absorbance and bilirubin concentration (top).
Calibration curve regression between absorbance and small size bins (middle). Calibration curve
regression between absorbance and large size bins (bottom).

where there are 5 ‘bins’ for icterus, ranging from 0–171, 188–342, 359–513, 530–684,
and 701–855  μmol/L of bilirubin (Tab. 10.1). With few bins, cutoffs can only be set
at 171 μmol/L increments, such that specimens with acceptable icterus may still be
rejected. If interference studies show that icterus above a threshold of 257 μmol/L is
unacceptable, then samples with bilirubin concentrations of 274, 290, 308, 325, and
342 μmol/L would all be rejected as they cannot be differentiated by the large bins
(concentrations between 188–342 μmol/L are all in the same bin). With a continuous
index (e.g. 1–60), the exact cutoff can be selected to maximize the number of report-
able results. Continuous indices do require a certain amount of precision if the goal
is to set tight thresholds. Using the example of a 257 μmol/L cutoff, if the index is
measured with 10 % coefficient of variation, samples with bilirubin concentrations
 10.2 Generating Interference Indices       109

ranging from 233–280 μmol/L may fall on either side of the cutoff (based on a para-
metric 95 % confidence interval). A rational goal is to set the number of bins wide
enough that samples with mild interference are not unnecessarily rejected, but not so
wide that values fluctuate by a large range of bins due to imprecision alone.

Bilirubin (μmol/L) Wide Bin Icterus Index Small Bin Icterus Index Continuous Icterus Index

17 1 1 1

34 2

51 3

68 2 4

85 5

102 6

119 3 7

136 8

153 9

170 4 10

187 2 11

204 12

221 5 13

238 14

255 15

272 6 16

289 17

306 18

323 7 19

340 20

357 3 21

374 8 22

391 23

408 24

Concentrations from 425–1,003 μmol/L are not shown.

1,020 5 20 60

Tab. 10.1: Example of icterus indices with different bin sizes.

110       10 Endogenous Interferences in Clinical Laboratory Tests

10.3 Limitations

While automated indices are a significant improvement over visual detection, they
are not perfect. There are instances where automated indices do not detect colorimet-
ric interferences, such as methemalbumin, drugs/dyes, and biliverdin. Even if instru-
ments were equipped with detection systems for these compounds, there is almost no
data indicating how and if such compounds interfere as they are rarely encountered
and are often found in patients with numerous co-morbidities. In addition, automated
detection systems for interferences are themselves subject to interferences (interfer-
ence index interferences!). Either of the subtraction approaches above do not account
for compounds other than hemolysis, icterus, and turbidity; theoretically derivative
spectrometry should be more resistant to other interferences than a two wavelength
process, but this method is by no means immune to interferences.
In addition, it can be inferred from the sections on index calibration that if the
compounds used to derive the extinction coefficients for the calibration differ from
those in patient samples then there is the possibility of unexpected results. This
emphasizes the need to use calibration materials that are similar to those that will be
found in patients.
One final consideration is quality control of the indices themselves. Few manufac-
turers offer QC material and there are no guidelines for if, how, and when the indices
should be assessed for accuracy and imprecision. Good laboratory practice suggests
that some assessment of the index stability and accuracy be considered. However, in
true practice, many labs find it difficult to perform even basic interference studies, let
alone inter-instrument comparability and proficiency testing for each type of interfer-
ence. At the end of the day, interference indices have a number of limitations, but
offer the opportunity to improve the quality of results reported.

10.4 Summary

Interference indices are determined using absorbance readings of standards prepared

for each type of interference (hemolysate, icterus, turbidity). Different methods exist
to account for interferences that have overlapping absorbance spectrum, each with its
own advantages with respect to accuracy and simplicity. The selection of the number
of ‘bins’ for each index requires careful consideration of imprecision, accuracy, and
the ability to accurately report results. Indices are only as accurate as the standards
from which they are generated, such that the calibration materials must be selected
carefully. As always, it is essential to keep focused on the clinical goal of interfer-
ence indices: improve the accuracy of patient results. Index calibration, bin sizes, and
quality practices should all be driven by this underlying goal.
 10.5 References       111

10.5 References
[1] Glick MR, Ryder KW, Glick SJ, Woods JR. Unreliable visual estimation of the incidence and
amount of turbidity, hemolysis, and icterus in serum from hospitalized patients. Clin Chem USA
1989,35,5,837–9.
[2] Bornhorst JA, Roberts RF, Roberts WL. Assay-specific differences in lipemic interference in
native and intralipid-supplemented samples. Clin Chem USA 2004,50,11,2197–201.
[3] Snyder JA, Rogers MW, King MS, Phillips JC, Chapman JF, Hammett-Stabler CA. The impact of
hemolysis on Ortho-Clinical Diagnostic’s ECi and Roche’s elecsys immunoassay systems. Clin
Chim Acta Netherlands 2004,348,1–2,181–7.
11 Reporting of Results

11.1 Introduction

Before implementing interference indices for icterus and turbidity (and hemolysis),
laboratorians need to decide what to do with results that exceed their self-defined
allowable limits (discussed in Chapter 5). As described in Chapter 5, the Laboratory
Director must decide what is or is not an acceptable level of bias due to endogenous
interference. Handling results that exceed thresholds can be complicated because
there is no universally unacceptable level of bias due to interference. In particular,
there will be cases where results may be clinically useful even in the presence of an
interference at a level that exceeds the defined allowable limits. Likewise, there will
be samples for which results should not be reported as they may be clinically mislead-
ing and could put the patient at risk of harm. If too much time is spent identifying
and reporting results with interferences, the overall turnaround times may increase,
again putting patients at risk. Thus, there are many subtleties to consider when it
comes to reporting results with detectable interferences. Ultimately, clinical judgment
is required, both for development of the result reporting process and assessment of
individual cases where results exceed defined thresholds.

11.2 Procedures for Handling Samples with Interference Within

the Laboratory

Even with exceptions and subtleties, there remains a need for a general approach to
handling samples with interferences. The Laboratory Director cannot examine every
result, so the bulk of samples need to be handled according to a detailed procedure
developed using available interference data. Procedures should include available
data for the extent and direction of interferences for each reportable analyte, how and
when to report or not report results, whether comments should be appended to results
with interferences, and clear instructions on how to communicate results verbally
when necessary. It is also useful to have figures or tables with interferograms (see
Chapter 5), as well as manufacturer and the laboratory data on the effects of interfer-
ences either as quick reference documents or embedded in procedures (e.g. Tab. 11.1).
The first step in developing an interference reporting procedure is to establish
the protocol for handling samples with interference within the laboratory prior to
reporting. Some laboratories choose to repeat results, search for alternate samples, or
perform additional processing to attempt to remove interference. In many cases, an
interference flag prompts visual inspection of the sample. Visual inspection remains
a useful practice to detect unexpected or uncommon sample problems, particularly
when considering the heterogeneous nature of icterus and turbidity interferences,
114       11 Reporting of Results

Test Hemolysis index Effect Index Effect Index Effect

ALB 4 increase 10 – 10 –
ALKP 4 decrease 5 increase 10 –
ALT 4 decrease 10 – 10 –
AST 1 increase 10 – 10 –
Ca 10 – 10 – 10 –
Cl− 10 – 10 – 10 –
CREA 10 – 10 – 10 –
GGT 3 increase 2 increase 10 –
GLU 3 increase 5 increase 10 –
K+ 1 increase 10 – 10 –
Mg 8 increase 10 – 10 –
Na+ 10 – 10 – 10 –
PHOS 2 increase 10 – 10 –
TBIL 2 increase NA – 10 –
UREA 10 – 10 – 10 –
URIC 10 – 5 decrease 10 –

Tab. 11.1: An example of an interference table for a chemistry procedure. Interference indices are
scaled from 1–10; 10 indicates a high concentration of interference; the numbers indicate the cutoff
value for the interferent.

which often occur in combination with multiple co-morbidities. Laboratory proce-

dures often include steps to phone the ordering physician to request a recollection
and determine if there was a preanalytical error associated with the sample collec-
tion. In the case of turbidity, preanalytical errors include a non-fasting patient with
postprandial lipemia or a draw from a line containing a nutritional supplement. The
laboratory must develop specific procedures for deciding when and if additional pro-
cesses should be used to remove an interference. As described in Chapter  9, ultra-
centrifugation or high-speed centrifugation may be used to remove turbidity. When
it comes to icterus, there are not many options for re-collection, as production of
bilirubin is endogenous. However, some analytes are amenable to dilution, such that
icterus interference may be reduced in concentration to permit accurate measurement
of the analyte in question. Either of these approaches requires experimental valida-
tion and their own procedures. When results are derived from additional processing,
such as dilution or ultracentrifugation, it is often informative to add a comment or
footnote to reports. Reports can indicate the presence of the interference and how
the sample was treated to remove the interference. In some cases, the presence of an
interference by itself is clinically informative. Turbidity is often associated with very
high levels of triglycerides, which put the patient at risk for pancreatitis. Turbidity is
also found in patients with dyslipidemia, which puts patients at high risk for cardio-
vascular disease (see Chapter 8). Likewise, the presence of icterus may alert the physi-
cian to hepatic injury, biliary obstruction, or hemolytic anemia (Chapter 6).
 11.3 Reporting of Results in Icteric and Turbid Samples       115

11.3 Reporting of Results in Icteric and Turbid Samples

There are several different options for handling icteric and lipemic results: not report-
ing the result, reporting the result with a comment indicating the presence of an inter-
ferent, reporting the result with a comment indicating the presence of an interferent
and the direction of bias, reporting the results without a comment, or performing
additional processing and analysis. Each of these many options has its own set of
advantages and disadvantages that are discussed below.
The most stringent, and arguably the most straightforward approach to handling
samples with icterus or turbidity levels that exceed self-defined limits is to not report
the results. One could argue that this makes sense in the context of allowable error
limits, where values that have an unacceptably high degree of error increase the risk of
patient harm. In this commonly used approach, results are provided for analytes that
do not exceed interference thresholds or limits, while those that do not are ‘cancelled’
and not reported due to icterus or turbidity. A sample report is shown in Tab. 11.2.
Footnotes are appended to each result printout and are also linked to the result in
the electronic medical record. These could also be linked to a website with additional
descriptions of interference and include contact information for physicians to ask for
more information; this approach is useful for laboratories performing reference ser-
vices, where low volume test users may be unfamiliar with interference reports.

Test Result Units Reference Interval

Total Bilirubin 814 μmol/L 3–17

Direct Bilirubin 673 μmol/L 2–9
ALT No Result* U/L 17–63
AST No Result* U/L 15–37
Sodium 135 mmol/L 135–145
Potassium 4.0 mmol/L 3.5–5.0
Chloride 97 mmol/L 97–105

* No result due to bilirubin interference.

Tab. 11.2: Sample laboratory report: no result for affected analytes.

Some laboratories may choose to report results with interference using an additional
comment describing the nature, direction, and extent of the bias (Tab. 11.3). As shown
in Tab. 11.3, creatinine is reported with a comment specifying that bilirubin causes
negative bias. In this example, the eGFR is not reported because it is calculated from
the creatinine result which is known to be affected. This highlights the need for con-
sideration of calculated values, which require decisions as to which parameters will
be reported and which will not in the presence of interference. Calculated parameters
may magnify error, particularly when they involve multiplication or exponents. Each
116       11 Reporting of Results

Test Result Units Reference Interval

Total Bilirubin 286 μmol/L 3–17

Direct Bilirubin 192 μmol/L 2–9
Creatinine 180† μmol/L 50–100
eGFR No Result‡ mL/min/1.73 m2 >60
Sodium 142 mmol/L 135–145
Potassium 3.5 mmol/L 3.5–5.0
Chloride 101 mmol/L 97–105

† Specimen is icteric. Large elevations in bilirubin may cause a negative interference with the creati-
nine assay. Interpret results with caution.
‡ No result due to bilirubin interference.

Tab. 11.3: Sample laboratory report: reporting results and comments for affected analytes.

reportable result, measured or otherwise requires careful consideration of the clinical

impact of a biased result due to interference.

11.4 Autoverification and Reporting Algorithms

Any or all of the above options can be applied to groups of tests or individual analytes.
While it is important to detect interferences, laboratories also need to consider prac-
tical issues, such as turnaround times and workflow. Accordingly, it is essential to
incorporate interference detection and reporting practices into routine workflow. One
effective method for maintaining workflow is to integrate interference detection and
reporting into autoverification procedures. With autoverification, results that exceed
a defined set of limits are subject to additional testing, review, or analysis. The bulk
test volume is passed through the system without human intervention. A benefit of
autoverification is that laboratory staff can focus efforts on problem samples and com-
municating issues to physicians and nurses at the bedside. Laboratories may choose
to automatically report results with comments on samples with interferences as part
of the bulk test volume. Alternatively, labs may define specific interference criteria to
prompt intervention by the medical technologist. For example, a sample with a low
level of interference may be processed automatically, whereas a sample with multiple
interference flags may prompt visual inspection, additional processing, and alternate
communication or reporting.
Attempts have been made to develop algorithms to facilitate detection of inter-
ference while maintaining workflow efficiency. A useful algorithm for analysis of
interference testing in the laboratory has been reported [1]. In this algorithm, consid-
eration was given to detection, communication, comment reporting, validation, and
sample rejection. Interference below a certain threshold (e.g. lipemia <6 for sodium
using an interference scale of 1–10) was automatically reported with a comment indi-
cating the type and relative concentration of the interferent. Above this threshold
 11.5 Practical Issues: Education and Implementation       117

(lipemia ≥6 for sodium), the laboratory information system required the technolo-
gist to actively investigate the sample and perform ultracentrifugation. Use of this
algorithm increased documented detection of icteric and lipemic samples by several
orders of magnitude (10–100x) as compared with the manual reporting and visual
inspection. Collectively, the data support that automated detection, reporting, and
integration with the LIS enable more accurate results, and a wealth of useful data to
understand the process. An example of interference data that can be derived from
the LIS is shown in Fig. 11.1. The figure illustrates that the intensive care unit has the
highest number of samples with interference. This is consistent the clinical expecta-
tions, where the ICU is the location with the most patients with extreme laboratory
values. While perhaps not surprising, it is useful to understand the sources of inter-
ference to make improvements and align workflow. LIS data can be further broken
down by phlebotomist, nurse, and patient, enabling laboratories to visualize trends
and identify strengths and weaknesses in the preanalytical process. Use of computer
algorithms facilitates long-term monitoring of processes and process improvement.
It allows for changes to be made and accurately measured to determine the effect of
changes. For example, if an improvement is made to a preanalytical process, then
data collected using the laboratory algorithm will enable determination of the impact
of the change. Hospitals or regions with an unexpected high frequency of interfer-
ences may be identified and addressed.

11.5 Practical Issues: Education and Implementation

When developing a reporting procedure using comments, it is essential to consider

the end user who ordered the test and will receive the result and comments. If the

50

40
number of samples

30
with interference

20

10

0
ICU 5NWE 5NES 5EST 6WST B5 7EST SICU NICU EMCU

location

Fig. 11.1: Example of using LIS data to identify the source of samples with interference; # of samples
with interference by location.
118       11 Reporting of Results

information provided in the report is not meaningful to the user then it serves little
purpose. Test users may be unfamiliar with the significance of an ‘Icterus’ flag or
‘T+++’ indicating high levels of turbidity. It is useful to provide a concise explanation
of the intent of interference flags as footnotes to the results, links with the results, and
in laboratory handbooks or websites. These are frequently educational opportunities
to engage clinical stakeholders and explain how the process of detecting interference
is intended to benefit patients. The laboratory also benefits from the perspective of
test users who may help refine cutpoints where results may or may not be flagged.
For example, a physician may be able to determine the trend of a result (is creatinine
going up or down) in the presence of a consistent concentration of interferent, such
that policies need to permit this type of clinical consultation. Likewise, physicians
may be shocked to learn that a small amount of interference may be the difference
between a normal and significantly abnormal troponin result. Good communication
is the key to effective utilization and reporting of interference indices.
Before implementation of indices, it is also useful to model the types and
frequency of different interferences and results using existing data where possible to
avoid unintended consequences. For example, if a threshold for a common analyte
is stringent and the laboratory chooses not to report that result, there may be a large
number of re-collections. Consider high frequency test orders at in outpatients clinics.
A high volume of re-collections is likely to strain the relationship between the labora-
tory and clinic while putting patients at unnecessary risk; phlebotomy is not without
risk and having patients who travel long distances to the hospital return for a single
result can be a significant personal relations and cost issue. Selection of reporting
procedures is a balancing act between different types of risk. Knowledge of the test
utility, biological variation, and analytical performance of each analyte is extremely
valuable when making decisions on reporting results with icterus and turbidity inter-
ferences.
One final consideration is implementation of interference indices. Implementa-
tion of interference indices requires engagement with key personnel responsible for
the Laboratory Information System (LIS) and the Electronic Medical Record (EMR) to
ensure the appropriate information is effectively communicated to the clinical care
team. Unfortunately, the laboratory is sometimes not at the forefront of controlling
changes to the EMR or even the LIS, such that efforts must be carefully explained to
the leadership responsible for making these changes. As with all other areas of the
laboratory, getting buy-in from all the relevant stakeholders is essential for effective
change implementation.

11.6 References
[1] Vermeer HJ, Thomassen E, de Jonge N. Automated processing of serum indices used for
interference detection by the laboratory information system. Clin Chem USA 2005,51,1,244–7.
12 Analyte-dependent Interference
Chapter 5 discussed at length the measurement of interference. Even though that
discussion leads to ways to determine appropriate limits based on the presence of
simple interferences, the discussion in Chapters 3 and 4 proposed that for bilirubin
and lipemia complex interactions may occur. One cannot assume that interferences
occur by increases or decreases in absorbance or pseudo-absorbance only and that
no interaction occurs with the analyte itself or with intermediates within the reaction.
The problem becomes more apparent when one observes that the slope of interference
changes, depending on the concentration of the analyte. This problem is especially
apparent when interferences do not follow simple linear trajectories or they demon-
strate positive interference at one concentration of analyte and negative interference
at a different analyte concentration.

12.1 Complex Interferences

Interferences become complex when they involve the analyte or species involved with
the detection system of the analyte. Most models of interferences assume that the
interferent acts independently of the analyte or other reactions occurring because of
the presence of the analyte. As an example, the interference caused by cepha antibi-
otics with the Jaffe method for creatinine occurs by the independent reaction of the
cepha antibiotics with alkaline picrate and absorbance at the same wavelength and
kinetics as the reaction of creatinine with alkaline picrate [1]. Interferences of this
type may be considered simple. Standard linear regression analysis may be adequate
for the evaluation of simple interferences, but more powerful regression algorithms
are necessary for the evaluation of interferences that occur by complex mechanisms.
One type of complex mechanism is one easily observed from a graph of the recov-
ered analyte concentrations, the output from the analyzer, versus the concentration
or other increasing unit of the interferent. The type of complex mechanism that is
easily observed is when the line describing the relationship is not straight. In terms of
incidence, this type of observation occurs frequently. Simple, straight-line standard
regression analysis is inadequate to do justice in first summarizing the relationship
in the interference or provide adequate information about the mechanism involved in
the interference. Whenever the relationship between the output by the analyzer and
the concentration of the analyte is not a straight line, i.e., nonlinear, the evaluation
needs to include a polynomial regression.
Classically one thinks of an interferent reacting directly with the reagent to
produce a new chromophore that absorbs at the same wavelength as the chromo-
phore produced by the reagent and the analyte; or the interferent reacts with the
reagent to produce a species that decreases absorbance at the test wavelength. In
some cases, the interferent may react with the analyte or it may react with an interme-
120       12 Analyte-dependent Interference

diary species on the path to produce the chromophore and thus decrease absorbance
in the reaction [2].
If one uses the standard linear regression model, one limits the ability to detect an
analyte-dependent interference, because the model for a standard linear regression
model assumes that there is only one independent variable [2]. In order to evaluate
interferences where analyte-dependence may be a factor, one must use a model that
includes analyte-dependence; otherwise, the study design and regression analysis
will be inadequate. Regression analysis works when the proper model is employed. To
accept one model over another, one must compare the models using the same form of
the regression analysis. To assess for interferences with complex reactions and inter-
actions, as bilirubin represents, one should use a model that assumes the possibility
of both analyte-independent and analyte-dependent variables [2].

12.1.1 Model for Analyte-dependent Interference

The model for analyte-dependence uses multiple regression analysis with three inde-
pendent variables. The independent variables are the concentration of analyte, the
concentration of interferent, and a mixed term, created by multiplying the concentra-
tion of analyte by the concentration of interferent. The multiple regression formula is

F ([analyte], [interferent])
= b1 [analyte] + b2 [interferent] + b3 [analyte][interferent] + b0 (12.1)

where F([analyte],[interferent]) is a function of both the analyte and the interferent,

the b’s represent coefficients for the equations, the terms in brackets represent the
concentration of the constituent, and b0 is the coefficient for the intercept [2]. The
coefficient b1 tells how well the method recovers the analyte, usually this coefficient
should be equal to 1.0. The coefficient b2 captures the degree to which the interferent
by itself acts like the analyte in producing a positive interference, if the value for the
coefficient is greater than zero, or detracts from the recovery of the analyte, acting
independently of the analyte, if the coefficient is negative. Using regression analysis
provides a statistical means to assess if this type of interference truly exists. If the
analyte-independent interference truly exists, then the coefficient will have a non-
zero value. Using A for analyte and I for interferent, the above formula can be more
compactly expressed as

F ([A ], [I ]) = b1 [A ] + b2 [I ] + b3 [A ][I ] + b0, (12.2)

 12.1 Complex Interferences       121

12.1.2 Examples of Analyte-Dependent Interference

Examples of analyte-dependent interference will make it easier to understand how

to design the studies and interpret the results. Crocker et al. evaluated an enzymatic
method for determining creatinine using the 4-aminophenazone/peroxidase detec-
tion system with hydrogen peroxide as an intermediate [3]. In short, creatininase
converts creatinine to creatine, creatinase converts creatine to sarcosine, sarcosine
oxidase converts sarcosine to glycine and hydrogen peroxide, and peroxidase con-
verts hydrogen peroxide, phenol derivative and 4-aminophenazone to a red benzo-
quinone imine dye.
In the study, Crocker et al. examined potential interference by glucose, acetoac-
etate, bilirubin and cefoxitin and compared those interferences with the kinetic Jaffe
method (alkaline picrate) [3]. They designed the study with four concentrations of
bilirubin (the zero value was whatever was already in the sample pool) and three
different concentrations of creatinine, one low, one moderate and one high. The con-
centrations of bilirubin used and the three concentrations of creatinine in the pools
are shown in Tab. 12.1. The bilirubin concentration as well as each concentration of
creatinine is given its own table. In this way, the three concentrations of creatinine

Bilirubin (μmol/L) Creatinine1 (μmol/L) Creatinine2 (μmol/L) Creatinine3 (μmol/L)

0 87 332 572
100 76 291 509
200 60 253 456
300 47 218 396

Tab. 12.1: Enzymatic creatinine bilirubin interference data.

Bilirubin (μmol/L) Creatinine (μmol/L) BiliCreat (μmol/L) Result (μmol/L)

0 87 0 87
100 87 87 76
200 87 174 60
300 87 261 47
0 332 0 332
100 332 332 291
200 332 664 253
300 332 996 218
0 572 0 572
100 572 572 509
200 572 1,144 456
300 572 1,716 396

Tab. 12.2: Independent and dependent variables for the bilirubin-creatinine interactive term.
122       12 Analyte-dependent Interference

are referred to as Creatinine1, Creatinine2 and Creatinine3. If one wanted to perform

multiple regression analysis using a spreadsheet program, one would enter the data
in this form on the spreadsheet. Tab. 12.2 shows the same data as Table 12.1, except
additional columns have been entered in addition to that of bilirubin, one for the cre-
atinine concentration, and another for the bilirubin-creatinine interaction (BiliCreat).
Fig. 12.1 shows plots of the recovered values for creatinine from the analytical
method plotted against the bilirubin concentration.

enzymatic creatinine and bilirubin

600
500
creatinine (mmol/L)

400
300
200 creatinine1
creatinine2 creatinine3
100
0
0 50 100 150 200 250 300
bilirubin (mmol/L)

jaffe creatinine and bilirubin

600
500
creatinine (mmol/L)

400
300
200 creatinine1
creatinine2 creatinine3
100
0
0 50 100 150 200 250 300
bilirubin (mmol/L)

Fig. 12.1: Interference with creatinine by bilirubin showing variation in slopes.

The upper graph shows three curves, one for each creatinine concentration in the
base pool for the enzymatic method. The lower graph shows three curves, again
for the three creatinine concentrations in the base pool, but for the Jaffe creatinine
method. Each curve is negative. For the Jaffe creatinine method, the curves are almost
all parallel, but for the enzymatic creatinine method, the curves are not parallel,
which means that they have different slopes. For the interference to be independent
of the concentration of creatinine (analyte-independent) the slopes of the creatinine
values vs the bilirubin concentration should all be the same [4].
Tab. 12.3 shows linear regression analyses of the creatinine vs bilirubin curves
at each individual concentration of the creatinine pool. The configuration of these
 12.1 Complex Interferences       123

results are the result of the output from the spreadsheet program EXCEL, using the
LINEST function. The r-squared values (square of the correlation coefficient) are at
least 0.99, meaning that the regressions were adequately run and the data fit well
with the model. The intercept for Creatinine 1 is 87.9 which is close to that of 87 in the
base pool, for Creatinine 2 it is 330.5, which is close to 332, as found in the base pool,
and for Creatinine 3, it is 570.4, which is close to 572, as found in the base pool. Given
these results, it is acceptable to go on with this analysis. The slope for each base pool
of creatinine is negative. Furthermore, the standard errors of the slopes (SE slope) for
each of the creatinine pools is small, especially compared with the magnitude of the

Creatinine 1 vs Bilirubin      

Slope −0.136 87.9 Intercept

SE slope 0.006 1.2 SE intercept
r-squared 0.995 1.4 SE regression
F 440 2 degrees of freedom
SSregression 925 4.2 SSresiduals

Creatinine 2 vs Bilirubin      

Slope −0.38 330.5 Intercept

SE slope 0.009 1.8 SE intercept
r-squared 0.999 2.1 SE regression
F 1,604 2 degrees of freedom
SSregression 7,220 9 SSresiduals

Creatinine 3 vs Bilirubin      

Slope −0.581 570.4 Intercept

SE slope 0.013 2.4 SE intercept
r-squared 0.999 2.9 SE regression
F 2021.323 2 degrees of freedom
SSregression 16,878 16.7 SSresiduals

Slope vs Creatinine      

Slope −0.00092 −0.062 Intercept

SE slope 4.6E-05 0.018 SE intercept
r-squared 0.998 0.016 SE regression
F 403 1 degrees of freedom
SSregression 0.099 0.00025 SSresiduals

Creatinine vs Slope      

Slope −1,087 −67.1 Intercept

SE slope 54 22.1 SE intercept
r-squared 0.998 17.1 SE regression
F 403 1 degrees of freedom
SSregression 117,325 291 SSresiduals

Tab. 12.3: Linear regression analysis of enzymatic creatinine method with bilirubin.
124       12 Analyte-dependent Interference

slope. Presented in the section of Statistical Testing for Significance will be a discus-
sion on how to use a t-test to test for the significance of the slope. As one looks at these
slopes, it becomes apparent that the absolute magnitude of the slopes increases as
the concentration of creatinine increases.
One can plot the slopes, as obtained with the first set of regressions, against the
creatinine concentrations in the base pools. The results of such plots are shown in
Fig. 12.2 for both the enzymatic and the Jaffe methods.

enzymatic slope vs creatinine

0
0.1 y 0.0009x 0.0625
R2 0.9975
0.2
0.3
0.4
slope

0.5
0.6
0.7
0 100 200 300 400 500 600 700
creatinine (mmol/L)

jaffe creatinine slope vs creatinine

0
y 0.0002x 0.0316
0.05 R2 0.9678

0.10
slope

0.15

0.20
0 100 200 300 400 500 600
creatinine (mmol/L)

Fig. 12.2: Regression of slope vs creatinine for bilirubin interference.

In both cases, the slopes are negative, but the scale for the kinetic method is much
greater in magnitude than for the Jaffe method. Regressing the slopes vs the creati-
nine concentrations is a form of multiple regression. The statistical results of this
regression are shown in Tab. 12.3 for the kinetic method, under the heading of slope
vs creatinine. The slope for this regression is −0.00092 and is much greater than the
standard error of the slope. A t-test yields a value of 20 which is statistically signifi-
cant because the value for P is less than 0.05. One could also plot the creatinine con-
centrations of each pool vs the slopes and obtain a slope for this regression. The result
is a slope of −1,087. These two regressions are inverses of each other, as shown by
taking the reciprocal of −1,087, which is identical to −0.00092. The implication of the
statistical significance of the slope vs creatinine regression means that the magnitude
 12.1 Complex Interferences       125

of the interference changes not only with the variation in the bilirubin concentration,
but also with the variation in the creatinine concentration. In other words, the inter-
ference is analyte-dependent.
One problem with regressing the slopes vs the creatinine concentration proce-
dure to perform the multiple regression analysis is that one loses degrees of freedom.
In this case, there is only one degree of freedom. The more degrees of freedom one
has, the greater the statistical power of the test.
To circumvent the problem of loss of degrees of freedom, one can perform the
multiple regression analysis using all the data at once. One needs to use a special sta-
tistical program for that, or one can set it up as a function in a spreadsheet program.
Tab. 12.4 shows the results of a fully-executed multiple regression program
output. Tab. 12.2 shows how the data must be arranged to use a fully executed mul-
tiple regression program. The first three columns, labeled Bilirubin, Creatinine, and
BiliCreat, represent the independent variables. The fourth column, labeled Result,
represents the creatinine result reported by the analyzer at that particular bilirubin
and creatinine concentration. The column labeled BiliCreat is a combination of the
values in the Bilirubin and the Creatinine column. The BiliCreat column is now a new
independent variable for the model of the regression. The creation of the BiliCreat
column represents the keystone in developing a model for assessing analyte-depend-
ent interference.

  BiliCreat Creatinine Bilirubin Intercept

Slope −0.092 0.995 −0.062 0.975

SE slope 0.004 0.008 0.016 3.067
r-squared 0.9998 SE regression 3.3 
F 11595 DF 8 
SS regression 370,673 SS residuals 85 

Tab. 12.4: Multiple regression analysis of interference slope vs creatinine for enzymatic method.

The BiliCreat column was created by multiplying the concentration of bilirubin by

the concentration of Creatinine. If the creation of this new independent variable had
been made by adding the two columns together, or by subtracting one column from
another, then the newly created variable would not be independent from the other
two. Because the new variable was created by a process (multiplication), it is linearly
independent of the other two. Further, because the two variables used to create the
third variable are still included in the regression process, the results gleaned from the
newly created variable, BiliCreat, will not reflect a dependence on Creatinine alone or
Bilirubin alone.
The results from the multiple regression analysis based on the newly created
BiliCreat variable are shown in Tab. 12.4. The first thing to note is that the degrees of
freedom are 8, which provides much more statistical power. The top row shows the
126       12 Analyte-dependent Interference

independent variable being analyzed. The first column shows results for the Biliru-
bin-Creatinine interaction, the second column for Creatinine, the third column for
Bilirubin and the fourth column for the intercept. The value for the intercept is 0.975,
but its standard error is 3.067. The value for t is 0.3, and p>0.05, so it is not statisti-
cally significant, which means that the regression should make sense. The value for
r-squared is 0.998, which is very high and also a good indication that the regression
analysis was carried out in an adequate manner. The slope for Creatinine is 0.995,
which is nearly 1.0, which makes sense and the slope of Bilirubin is −0.062, with a t
value of 3.8, with p <0.05, which is significant; this information implies that there is
an interference caused by bilirubin which is independent of the creatinine. The slope
for the BiliCreat variable is −0.092, which is identical to that obtained by the regres-
sion of the slope vs creatinine. The value for t for this slope is 23, which is statistically
significant. The significance of the slope for this variable is that it means that there is
an interference caused by bilirubin that also depends on creatinine. One can write the
formula that gives the result based on the regression study as

Result = 0.995[creatinine ]− 0.062[bilirubin ]− 0.092[bilirubin ][creatinine ]+ 0.975.

(12.3)

One can use this formula to predict a result based on the concentrations of creatinine
and bilirubin. One can use the formula to find out how much interference there would
be at any given concentration of bilirubin for any given concentration of creatinine.
Tab. 12.5 shows the table of the creatinine results and the effect of bilirubin as in
Tab. 12.1, but this time for the Jaffe method for determining creatinine. Tab. 12.6 shows
the regression results for this method. Again there is negative interference, but it is
less than for the enzymatic method. Fig. 12.1 shows the graph of these results. The
slope vs creatinine is only 0.00022 and the t-test result is only 5.5, which is not sig-
nificant. Tab. 12.7 shows the three independent variable results for the Jaffe reaction
method. One difference here in the calculations is that the column created by multi-
plying the concentrations of bilirubin by those of creatinine, was then multiplied by
100 because the numbers were so small. The slope and the SE of the slope should be
multiplied by 0.01 to make them equivalent to the values in Tab. 12.6. The multiplica-
tion by 100 does not alter the results of the t-test. The value for t is 3.14, but with 8
degrees of freedom the critical value for t is 2.3, which is less than 3.1, so p <0.05, and
the coefficient for analyte-dependent interference is statistically significant. Using a
direct multiple regression calculation is more powerful than applying a simple linear
regression multiple times.
 12.1 Complex Interferences       127

Bilirubin (μmol/L) Creatinine 1 (μmol/L) Creatinine 2 (μmol/L) Creatinine 3 (μmol/L)

0 81 317 551
100 64 297 535
200 68 290 526
300 65 282 505

Tab. 12.5: Jaffe reaction method creatinine bilirubin interference data.

Creatinine 1 vs Bilirubin      

Slope −0.044 76.1 Intercept

SE slope 0.030 5.6 SE intercept
r-squared 0.523 6.6 SE regression
F 2 2 degrees of freedom
SSregression 97 88.2 SSresiduals

Creatinine 2 vs Bilirubin      

Slope −0.112 313.3 Intercept

SE slope 0.021 4.0 SE intercept
r-squared 0.932 4.8 SE regression
F 27 2 degrees of freedom
SSregression 627.2 45.8 SSresiduals

Creatinine 3 vs Bilirubin      

Slope −0.147 551.3 Intercept

SE slope 0.016 2.9 SE intercept
r-squared 0.978 3.5 SE regression
F 88.926 2 degrees of freedom
SSregression 1,080 24.3 SSresiduals

Slope vs Creatinine      

Slope −0.00022 −0.032 Intercept

SE slope 0.00004 0.015 SE intercept
r-squared 0.968 0.013 SE regression
F 30 1 degrees of freedom
SSregression 0.005 0.00018 SSresiduals

Creatinine vs Slope      

Slope −4414 −129.5 Intercept

SE slope 805 88.3 SE intercept
r-squared 0.968 59.6 SE regression
F 30 1 degrees of freedom
SSregression 106,893 3,558 SSresiduals

Tab. 12.6: Linear regression analysis of Jaffe creatinine with bilirubin.

128       12 Analyte-dependent Interference

  BiliCreat Creatinine Bilirubin Intercept

Slope −0.022 1.011 −0.032 −6.264

SE slope 0.007 0.013 0.025 4.735
r-squared 0.9995 SE regression  5.1
F 5,476 DF 8 
SS regression 424,518 SS residuals  207

Tab. 12.7: Multiple regression analysis of bilirubin interference vs creatinine.

Fig. 12.3 shows another enzymatic assay for creatinine, this time using the Vitros, and
its interference by bilirubin. It is plotted in a fashion similar to that of Fig. 12.1.

Jaffe creatinine interference by bilirubin

300
creatinine (mmol/L)

250
200
creatinine1
150 creatinine2 creatinine3
100
50
0
0 200 400 600 800
bilirubin (mmol/L)

Vitros interference by bilirubin

400
350
300
creatinine (mmol/L)

creatinine1
250 creatinine2 creatinine3
200
150
100
50
0
0 200 400 600 800
bilirubin (mmol/L)

Fig. 12.3: Difference in plot appearance for bilirubin interference for Jaffe creatinine and enzymatic
creatinine methods; the interference with the enzymatic (Vitros) method is nonlinear.

Not only are the curves formed by each of the varying concentrations of creatinine
not parallel, suggesting analyte-dependent interference, but they are not straight.
In fact, these curves need to be fitted with a second-order polynomial regression to
adequately fit the data. As an example, at the high concentration of creatinine, the
regression that fits the data is given by
 12.2 Statistical Testing for Significance       129

Result = −0.0002[Bilirubin ]2 + 0.299[Bilirubin ] + 279. (12.4)

Notice that the first term for Bilirubin is squared. These results indicate that the inter-
ference mechanisms are complicated. If one is not aware of the nonlinear nature of
the process one might miss important details.

12.2 Statistical Testing for Significance

To test for statistical significance one can use a t-test using the following formula

|b1 − 0|
t = , (12.5)
SE b1

where SE represents the standard error of the coefficient (the slope in standard linear
regression). In this case the subscript ‘1’ indicates that the coefficient and the stand-
ard error are for the first term in the formula, which is because it serves as the coef-
ficient for the concentration of the analyte, and indicates the recovery of the signal for
the analytical method. Usually the b1 coefficient will have a value very close to 1.00.
The standard error of the coefficient is calculated when one performs the multiple
regression analysis. To test for bias in the recovery of the method, one would calculate
assuming that the coefficient should be equal to 1.00. One would calculate t as

|b1 − 1|
t= , (12.6)
SE b1

if the value for t, based on the number of standard deviations, is low enough not to
reject the null hypothesis, e.g., t = 1.5 for 8 degrees of freedom, then there is no bias
for the method due to some sort of calibration error. On the other hand, if t = 3.0 for
8 degrees of freedom, then, for a two-tailed test (a two-tailed test implies that the
values can be higher or lower), then p <0.05 and one would reject the null hypothesis.
The null hypothesis in this case is that the coefficient is not different from 1.00. If the
coefficient were less than 1.00, e.g., 0.9, one would say that there was a negative bias.
If the coefficient were greater than 1.00, e.g., 1.1, one would say that there is a positive
bias.
The b2 coefficient represents the quantification of the effect that a potential inter-
ferent has on the determination process as captured by the signal and that signal
is transformed into a value based on the calibration of the method. If the potential
interferent has no effect, which means that it does not interfere with the method, then
the value of the coefficient would be 0.0, the null hypothesis in this case. If the value
of the coefficient is different from 0.0, the alternate hypothesis, then that means that
130       12 Analyte-dependent Interference

the potential interferent does interfere with assay. To test the statistical significance
of the coefficient, one uses the t-test

|b2 − 0|
t = , (12.7)
SE b2

where the assumed value for no interference is zero. If the result for the t-test yields
a p >0.05, then one can accept the null hypothesis and claim that there is no interfer-
ence; however, if the result for the t-test yields a p <0.05, then one would reject the
null hypothesis and accept the alternate hypothesis, that there is an interference from
this substance. If the coefficient is positive, then that means that there is a positive
interference. If the coefficient is negative, then that means that there is a negative
interference. If one stopped here, then one would be analyzing the data using the
standard linear regression for the model F=b1[A]+b2[I]+b0, and one assumes that the
interference model is simple (Fig. 12.4).

analyte reagent chromophore

bilirubin reagent chromophore

Fig. 12.4: Diagram for analyte-independent interference.

The b2 coefficient would encompass any interference, whether dependent or inde-

pendent, but it would be prone to errors. Evaluation of the third coefficient allows one
to ascertain any contribution by analyte-dependent interference and clearly sepa-
rates the contribution caused by analyte-independent interference which is indicated
by the second coefficient.
The b3 coefficient represents the quantification of the contribution to the signal
that a potential interferent makes when it interacts directly with the analyte or with
an intermediary in the analytical reaction or even with the final product. If the poten-
tial interferent has no effect, which means that it does not interact in the reaction
and interfere with the method, then the value of the coefficient would be 0.0, the null
hypothesis. If the value of the coefficient is different from 0.0, the alternate hypoth-
esis, then that means that the potential interferent does interact with the analyte, an
intermediary or final product and interfere with assay. To test the statistical signifi-
cance of the coefficient, one uses the t-test

|b3 − 0|
t= , (12.8)
SE b3
 12.2 Statistical Testing for Significance       131

where the assumed value for no interaction and interference is zero. If the result for
the t-test yields a p >0.05, then one can accept the null hypothesis and claim that there
is no interaction leading to interference; however, if the result for the t-test yields a
p <0.05, then one would reject the null hypothesis and accept the alternate hypoth-
esis, that there is an interaction that leads to interference for this substance. If the
coefficient is positive, then that means that there is a positive interference. If the coef-
ficient is negative, then that means that there is a negative interference (Fig. 12.5).

analyte reagent intermediary

intermediary chromophore
bilirubin

Fig. 12.5: Diagram for analyte-dependent interference.

In some cases, the interference caused by an interferent, when plotted on a graph,

does not show a straight line. Interferences caused by analyte-interferent interactions
are more likely than simple analyte-independent interferences to demonstrate devia-
tion from a straight line. Deviation from a straight line represents a nonlinear effect.
Interactions between two reacting species are prone to nonlinearities. A require-
ment for methods to be linear is that the concentrations of the reagents used in the
determination of the analyte have to be much greater than the concentration of the
analyte. Keeping the concentration of the reagents greater than the concentration of
the analyte means that as the reaction proceeds, the apparent concentration of the
reagents does not appear to decrease. Usually, the ratio of the concentration of the
reagents compared with the concentration of the analyte is at least 20–1. In enzymatic
reactions, the substrate’s concentration must be kept high, otherwise in specimens
with extremely high concentrations of the analyte substrate depletion can occur.
When the interferent reacts with an intermediary in the reaction, the concentration
of the intermediary is very close to that of the analyte. The reaction can be described
using an equilibrium constant, K:

[interferent − intermediary − product ]

K = (12.9)
[interferent ][intermediary ]

but as the reaction proceeds, the amount of interferent and intermediary decrease as
the product is consumed. Using P as the product, and ITM as the intermediary, with
subscripts of 0 to indicate the initial concentration,
132       12 Analyte-dependent Interference

[P ]
K = (12.10)
([I ]0 − [P ])([ITM ]0 − [P ])

and the concentration of product appears in both the numerator and the denomina-
tor in a nonlinear fashion. With increasing concentrations of interferent the amount
of product decreases. Often this pattern follows a pattern that looks like substrate
exhaustion. For the simpler formula for K where the interferent is not being greatly
consumed,

[P ]
K = (12.11)
([I ])([ITM ]0 − [P ])

and the equation for the production of the product becomes

K [I ] [ITM ]0
[P ] = , (12.12)
1 + K [I ]

which is a hyperbolic curve and clearly nonlinear (see the lower graph in Fig. 12.3).
Nonlinear relationships can be successfully captured in the multiple regression anal-
ysis using polynomial regressions [4].
For example, in peroxidase based reactions, hydrogen peroxide is produced and
aminoantipyrine reacts with a phenol derivative to produce a chromophore. If bili-
rubin reacts directly with hydrogen peroxide or with an intermediary, the apparent
analytes (creatinine, uric acid, triglycerides, cholesterol, etc.) would appear to be
decreased. The concentration of bilirubin needs to be only one-tenth of the concentra-
tion of analyte to produce this nonlinearity. As examples, the normal concentration of
bilirubin has a limit at about 20 μmol/L. At ten times this concentration, the concen-
tration of bilirubin would be 200 μmol/L. A mildly elevated concentration of triglyc-
erides is 2.26 mmol/L, which has a ratio of about 11 and could affect the total amount
of the interference. A mildly elevated uric acid has a concentration of 500  μmol/L
and with a bilirubin of 200 μmol/L, that ratio is 2.5 to 1. And for creatinine, a mildly
elevated concentration would be 150 μmol/L and with a bilirubin concentration of
200 μmol/L, the ratio of creatinine to bilirubin is 0.75 to 1, enough to readily reduce
the amount of apparent creatinine, but also to produce a major nonlinearity.
The complexity of the nonlinear interactions between conjugated bilirubin and
creatinine has been documented in detail. Using an enzymatic method that employs
creatininase to form creatinine, creatinase to form sarcosine, sarcosine oxidase to
form glycine and hydrogen peroxide and finally hydrogen peroxide with a phenol
derivative, 4-aminophenazone and peroxidase (PAP) to form a red benzoquinon-
eimine dye, Eng et al. showed that one needed to use polynomial regressions to find
the best fit of the data for the interference [5]. Stepwise regression analysis showed
that the best fitting of the data occurred when using models that included terms such
 12.4 Advantages of Using Multiple Regression Analysis       133

as [creatinine]2[bilirubin] or even higher terms, and that it did not matter whether one
used an end-point or a kinetic method with the PAP reagents. The nonlinearities in
the data set were so large that fitting the data to a linear model produced large errors.
Also, it suggests that there may be multiple interactions occurring between bilirubin
and the intermediaries [5].

12.3 Failure to Design the Interference Study

One should design the study to evaluate for interference with at least three different
concentrations of bilirubin and at least three different concentrations for the analyte.
PAP methods are particularly vulnerable to the effect of bilirubin. Designing the
studies with more than three concentrations of analyte improves the accuracy of the
study and provides improved confidence in the fit of the data. Failure to design the
study the proper way means that one will miss important information. Depending on
the directions of the interference, gross mistakes can happen, as when one mecha-
nism demonstrates a positive interference and another mechanism a negative one.

12.4 Advantages of Using Multiple Regression Analysis

Use of multiple regression analysis allows one to evaluate more than one potential
interfering substance at a time. If properly entered into the data stream for independ-
ent variables the multiple regression analysis clearly separates one variable from
another. Besides interactions among analytes and even other interfering agents, one
can account for nonlinearities in the interference by using polynomial regression.
Polynomial regressions can account for many types of nonlinearities because the non-
linear relationships can be successfully approximated using a Taylor expansion [4].
Hyperbilirubinemia may present with more than one interfering agent. Aoki et
al. reported on bilirubin interference with a urate oxidase based methodology utiliz-
ing peroxidase [6]. They noted the presence of a negative interference for conjugated
bilirubin, unconjugated bilirubin and ditaurobilirubin (a chemically synthesized con-
jugated bilirubin), with the magnitude of the slopes of interference for conjugated
bilirubin and ditaurobilirubin being almost identical. The magnitude of the slope for
the unconjugated bilirubin was significantly less than that seen for the conjugated
bilirubin. Further, upon addition of albumin to the solutions of conjugated biliru-
bin, the magnitude of the interference decreased, suggesting that bilirubin bound
to albumin causes less negative interference than free bilirubin. Covalently bound
bilirubin (delta bilirubin) caused very little negative interference as well. The study
points out that the magnitude of interference will be different based on the types of
bilirubin present, which would be summed together in a Bilirubin Index measure on
an automated analyzer or in the Total Bilirubin measurement.
134       12 Analyte-dependent Interference

Zieve’s syndrome, due to the chronic effect of ethanol on the liver and excessive binge
drinking, results in a transient hyperlipidemia, hyperbilirubinemia and hemolytic
jaundice [7]. As an example, in a report of a case of a man who presented with abdom-
inal pain and had an enlarged liver along with hyperlipidemia, hyperbilirubinemia
and increased red cell fragility, the latter of which is likely to present with hemolysis
in vivo or in vitro [8]. On admission the patient’s bilirubin was 171 μmol/L, the serum
milky in appearance, with a triglycerides concentration of 5.0 mmol/L [8]. It took five
days for the milky serum to clear and the bilirubin to fall back to a near normal con-
centration. The mechanism for the hyperlipidemia in Zieve’s syndrome is believed to
be due to an episode of massive mobilization of fat to or from fatty acids, occurring on
the background of a fatty or cirrhotic liver after ingestion of large amounts of ethanol
[7]. Zieve’s syndrome is an example of how easy it is for more than one interferent to
present itself in a patient’s specimen.
Multiple regression provides a way to evaluate the impact of multiple mecha-
nisms for interference. Bilirubin can introduce a spectral interference based on its
ability to absorb light in the visible region. In addition, bilirubin can react chemically.
It serves as an anti-oxidant. Bilirubin has the ability to easily convert back to its parent
molecule, biliverdin, through an oxidation step. Many analytes are detected using a
peroxidase-based detection system, resulting in a chromophore. It has been noted
that the spectral properties of bilirubin can cause a positive interference with the
detection of uric acid using a peroxidase-based method, and a negative interference,
chemically, using this same method [6]. Multiple regression analysis separates these
two mechanisms. Two mechanisms for interference, with different directions for their
interference (one positive and the other negative) complicate the matter of evaluating
the presence of an interference by bilirubin. Based on the concentration of bilirubin
and the concentration of the analyte, uric acid in this case, along with the coefficients
of the interferences can lead to a situation where no interference would be detected.
Such a situation has been demonstrated for the interfering effect of hemoglobin on
the determination of total bilirubin [2].
The cancellation of the interferences when one is negative and the other posi-
tive presents a good reason for performing interference studies at more than one con-
centration of the analyte. To see how a negative interference can cancel out a posi-
tive interference, return to the basic formula for the measurement of the information
reaching the detector for the assay:

F (A , I ) = b1 [A] + b2 [I ] + b3 [A][r ] + b0 , (12.13)

where F represents the information reaching the detector (the absorbance, for example
in a uric method), [A] represents the concentration of the analyte, [I] represents the
concentration of the interferent, in this case, bilirubin, b1 represents the coefficient
relating the response of the method to the analyte concentration, b2 represents the
response of the system to the concentration of the interferent acting independently,
 12.5 Concluding Remarks       135

b3 represents the response of the system to the interaction of the interferent with the
analyte or intermediaries in the reaction started by the analyte, and b0 represents
the intercept. Multiple regression analysis helps separate out the effect of the analyte
response from the response resulting from the interferent; it is represented by the
derivative obtained by holding the concentration of analyte constant, which is

F (I ) = b2 + b3 [A]. (12.14)

If the coefficients are opposite in sign, they can cancel each other out when the con-
centration of the analyte is equal to the negative of the ratio of the independent coef-
ficient to the dependent coefficient:

b2
[A ] = − . (12.15)
b3

This situation may not happen frequently, but in evaluating methods for the presence
or absence of interferences by bilirubin, and even lipemia, it should be considered.
The use of multiple regression analysis can clearly separate interferences into
analyte-dependent vs analyte-independent, as well as positive or negative. Knowl-
edge of the reagents systems and particulars of the testing methodology and instru-
mentation coupled with the above information allows one to piece together probable
mechanisms for the observed interferences. With the knowledge that one obtains
concerning interference mechanisms, one can construct better means to detect the
presence of an interfering agent and reconfigure testing methodologies to minimize
their influence.

12.5 Concluding Remarks

Letellier has designated four types of interference [9].

1. The interferent interacts with the analyte. An example is inhibition of alkaline
phosphatase activity by theophylline.
2. The interferent interacts with a species that is important for the analyte’s activ-
ity. An example is D-penicillamine chelating magnesium or zinc, both of which
serve as enzyme activators.
3. The interferent absorbs light at the same wavelength as the chromophore for the
chemical reaction or the analyte itself. Examples of this type of interference are
the interference caused by methotrexate, which absorbs light at 340 nm, and its
interference with the determination of phosphorus and triglycerides; bilirubin’s
absorbance at a wavelength similar to that of methemoglobin and hemoglobin
[10]; and light scattering by chylomicron and VLDL particles which similates
absorbance at many wavelengths.
136       12 Analyte-dependent Interference

4. The interferent reacts with the reagents or a constituent in the reagent mix that
converts it to absorb light at the same wavelength as the final chromophore
resulting from the reaction of the analyte with the reagent system. Examples of
this type of interference are reaction of certain drugs with phosphotungstate to
form a product similar to that formed for uric acid; the reaction of cefoxitin with
alkaline picrate, the reagent for the determination of creatinine [1]; and the reac-
tion of bilirubin with the reagents used for the determination of creatinine by
the Jaffe reaction and the peroxidase-based detection system [11–13].
5. To the types of interferences mentioned above, one should add some more,
especially focusing on negative interference types. The reagent can interact with
the interferent and modify it so that it loses absorbance at the wavelength for
measuring the production of the chromophore in the chemical reaction, or, if the
analyte itself is the chromophore, its wavelength; this loss of absorbance can be
either end-point or kinetic.
6. The interferent can interact with the intermediary or final product of a chemical
reaction mechanism, either enhancing absorbance or decreasing it. An example
of this type of interference occurs in the measurement of triglycerides that are
glycerol blanked and the specimen has turbidity as the result of increased chy-
lomicrons (lipemia); the enzymatic removal of glycerol may alter the properties
of the chylomicrons, causing some clearance of the turbidity and decrease in the
baseline absorbance.

Types 3, 4 and 5 represent analyte-independent interference. Types 1, 2 and 6 repre-

sent analyte-dependent interference.
The main goal of an interference study is to ascertain the presence or absence of
interference. The design of the study should depend on the knowledge of possible
mechanisms, the entertainment of the appropriate models for interactions of the sup-
posed interferent with the reaction methodology, and the concentrations expected
for the reportable range of the analyte and expected concentrations of the interfer-
ent. Once one has obtained results from a properly designed and executed study,
one can determine, statistically, whether or not a potential interferent does cause an
interference. Once the determination has been made that a certain potential inter-
ferent does cause an interference for a given method, then one needs to proceed to
determine the potential impact that interferent will have on patient specimens, the
quality of care and ultimately, the patient’s safety. The practice of medicine is dif-
ficult enough without the generation, presentation and perpetuation of misinforma-
tion. Misinformation can have an extensive negative impact on the quality of patient
care because the information will be perpetuated in the patient’s medical record and
history, causing increased anxiety on the part of the patient and the clinical care
givers, pursuit of additional studies to continually check on the misinformation and
the pursuance of additional studies to follow up the information. For example, in a
patient where there was a falsely elevated methemoglobin caused by an icteric speci-
 12.6 References       137

men, the question as to whether the patient has a medical problem that might lead to
methemoglobinemia will always be present. At any hospital admission or experience
with a major illness, or even on annual physical examinations, additional testing will
be done to rule out the presence of methemoglobinemia. Additional testing would
probably include blood co-oximetry and genetic testing.
Once the potential impact has been determined, the laboratory will need to
establish protocols for communicating the presence of an interfering substance with
laboratory staff and the clinical team taking care of the patient. The protocols need
to include how the result for the analyte will be reported, e.g., no result or a result
with a number but including the flag to alert the clinical team. The laboratory needs
to establish a way to detect the presence of potential interfering substances. In the
case of bilirubin and lipemia (turbidity), the opportunities to detect their presence
are many. The opportunities to detect the presence of hyperbilirubinemia or lipemia
include direct detection using an interference index, integration of other laboratory
data, e.g., from the direct analytical determination of total and direct bilirubin and
the determination of triglycerides, and from information gleaned from the medical
record, including past presence of these conditions or a list of medical diseases and
medications born by the patient. Improved detection of hyperbilirubinemia and
lipemia depend on the sophistication of the electronic medical record and the labora-
tory information system.
Study design and statistical analysis of interference should include the analyte-
dependent interference model and multiple regression analysis. The analyte-depend-
ent interference model encompasses the ability to detect both analyte-dependent and
analyte-independent interference as well as more complex interferences. Important
in the design of studies that examine for the presence of analyte-dependent interfer-
ence is the production of test specimens that vary the concentration of the analyte, in
at least three different concentrations, over a range (at least three) of interferent con-
centrations for bilirubin or Intralipid® or absorbance in the case of lipemic samples.
Multiple regression analysis extends the ability of the statistical analysis to separate
and quantify analyte-dependent from analyte-independent interference. The quanti-
fication of the interference is important to establish the proper protocols for commu-
nicating the presence of interferences.

12.6 References
[1] Kroll MH, Hagengruber C, Elin RJ. Reaction of picrate with creatinine and cepha antibiotics. Clin
Chem USA 1984,30,1664–1666.
[2] Kroll MH, Ruddel M, Blank DW, Elin RJ. A model for assessing interference. Clin Chem USA
1987,33,1121–1123.
[3] Crocker H, Shephard MDS, White GH. Evaluation of an enzymatic method for determining
creatinine in plasma. J Clin Pathol UK 1988,41,576–581.
138       12 Analyte-dependent Interference

[4] Kroll MH, Chesler R. Rationale for using multiple regression analysis with complex
interferences. Eur J Clin Chem Clin Biochem Germany 1992,30,415–424.
[5] Eng CD, Delgado R, Kroll MH. Complex analyte-dependent and analyte-independent
interferences with conjugated bilirubin in the enzymatic phenol-aminophenazone peroxidase
(PAP) method for creatinine determination. Eur J Clin Chem Clin Biochem Germany
1993,31,839–850.
[6] Aoki Y, Ihara H, Nakamura H, Aoki T, Yoshida M. Effects of serum bilirubin on determination of
uric acid by the uricase-peroxidase coupled reaction. Clin Chem USA 1992,38,1350–1352.
[7] Zieve L. Jaundice, hyperlipemia and hemolytic anemia: a heretofore unrecognized syndrome
associated with alcoholic fatty liver and cirrhosis. Ann Intern Med USA 1958,48,471–496.
[8] Gitlin N. Zieve’s syndrome and porphyrinuria in an alcoholic. Br Med J UK 1969,1,96–98.
[9] Letellier G. How does drug interference relate to quality assurance? In: Rand R, Eilers R,
Lawson N, Braughton A, ed. Quality assurance in health care: a critical appraisal of clinical
chemistry. Washington, DC, USA, AACC,1980,135–168.
[10] Sharma A, Artiss JD, Strandbergh DR, Zak B. The turbid specimen as an analytical medium:
hemoglobin determination as a model. Clin Chim Acta Netherlands 1985,147,7–14.
[11] Srisawasdi P, Chaichanajarernkul U, Teerakanjana N, Vanavanan S, Kroll MH. Exogenous
interferences with Jaffe creatinine assays: addition of sodium dodecyl sulfate to reagent
eliminates bilirubin and total protein interference with Jaffe Methods. J Clin Lab Anal USA
2010,24,123–133.
[12] Spain MA, Wu AHB. Bilirubin interference with determination of uric acid, cholesterol, and
triglycerides in commercial peroxidase-coupled assays, and the effect of ferrocyanide. Clin
Chem USA 1986,32,518–521.
[13] Witte DL, Brown LF, Feld RD. Effects of bilirubin on detection of hydrogen peroxide by use of
peroxidase. Clin Chem USA 1978,24,1778–1782.
Index
absorb 134 – glucuronic acid 66
– light 11, 17 – glucuronide 66
absorbance 12, 13, 38 – hemolysis 71
absorption 17 – interference with methemoglobin 27
accuracy 1 – oxidized to biliverdin 29
acetaminophen 71 – toxicity 65
acrolein 17 – unconjugated 64
active antiretroviral therapy – water solubility 49
– hypertriglyceridemia 89 biliverdin 27, 28, 63, 134
adipocytes 84 – reductase 63
albumin biochemistry 83
– bilirubin bound 29 biological variation 53
alcohol 88 blank 19
alcoholism
– elevated lipids 87 calibrators 3, 14
Allen correction 19, 21 captopril 72
alternate hypothesis 129 carbamazepine 72
amoxicillin-clavulanic acid 72 carboxyhemoglobin 24
amphiphilic 40 carboxylate group 40
anabolic steroids 72 carcinoma at the head of the pancreas 69
analyte 1, 3, 5, 6 carotene 17
analyte-dependence 120 change in absorbance 29
analyte-dependent interference 120, 121, 125, chemical interference 79
126, 128, 136, 137 cholestasis 69
analytic 3 cholesterol 31, 40
anti-oxidant 134 cholesteryl esters 40
apolipoproteins chromogen 29
– polar 40 chromophore 18, 119, 132, 134
Arias syndrome 68 chylomicron remnant 84
aromatic resonance 18 chylomicrons 35, 39, 44, 83
ascending cholangitis 70 clindamycin 72
atomic absorption 35 clopidogrel 72
autoimmune disorders 70 Cockroft-Gault equation 7
automated processes 63 coefficient of variation 2
autosomal dominant 86 colloidal dispersions 42
color 17
Beer’s law 13, 38 complex 119
benzoquinone imine dye 121, 132 complexity 132
bias 3, 48 conjugated bilirubin 77, 132
– mixed 5 conjugated double bonds 17
– negative 3 co-oximetry 22
– proportional 5 creatininase 30, 121
bichromatic correction 21 creatinine 3, 8, 15, 29, 54, 77, 98, 121
bilirubin 6, 11, 15, 18, 26, 48, 49, 132, 133 – clearance 7
– anti-oxidant 64 – enzymatic method 30, 122
– crosses placental membranes 64 – Jaffe method 6
140       Index

degrees of freedom 129 heme 18, 22

delta bilirubin 133 – origin of bilirubin 63
design hemoglobin 21
– study 136 – Barts 28
diabetes – saturation 22
– elevated lipids 87 hemolysis 71
diagnosis 48 hemolytic anemias 71
dilution 80 hepatitis
dimethyl sulfoxide (DMSO) 49 – alcoholic 70
disease – infectious 70
– progress 48 HIV infection
ditaurobilirubin 133 – lipemia 89
drug-induced hepatic injury accounts 71 hydrogen peroxide 31, 75, 121
dysbetalipoproteinemia 86 hyperbilirubinemia 21, 63, 133, 134
dyslipidemia 85 – alcoholic cirrhosis 70
– renal disease in children 90 – alcoholic hepatitis 70
– types 42 – causes 72
– Crigler-Najjar types I and II 67
effects – Dubin-Johnson syndrome 68
– interfering substances 48 – Gilbert syndrome 67
electrolye exclusion 35 – intrahepatic or extrahepatic obstruction 69
electromagnetic radiation 12 – Rotor syndrome 68
emulsion 50 hyperinsulinemia 87
endogenous 11 hyperlipidemia 11, 134
error 2 hyperlipoproteinemia 87
– constant 3 hypertension 87
ethanol 88 hypertriglyceridemia 85
evaluate 48 – after renal transplant 91
extinction coefficient – HIV and active antiretroviral therapy 89
– bilirubin 22
– deoxyhemoglobin 25 icteric samples 63
– oxyhemoglobin 25 icterus 75
– Zieve‘s syndrome 88
familial imprecision 48
– chylomicronemia 85 inaccuracy 48
– combined hyperlipoproteinemia 86 independent variable 120
– dysbetalipoproteinemia 86 indirect potentometry 35
– hypercholesterolemia 86 infrared region 15
– hypertriglyceridemia 86 insulin resistance 87, 88
fatty acids 40, 50, 83 interactions 133
ferric iron interference 6, 8, 11, 44, 48, 119, 130, 133
– reduction to ferrous iron 27 – bilirubin 21, 22, 24, 26, 27, 28, 29, 30, 31,
flame photometry 35 32, 33
formaldehyde 17 – hyperbilirubinemia 47
Frederickson classification 85 – lipemia 35, 47
free bilirubin 79 interferent 11
– endogenous 49
Gilbert syndrome 68 interferographs 11, 21, 49, 54
guidelines 48 interpretation 3
 Index       141

Intralipid® 37, 42, 44, 48, 50, 93, 95 microalbuminuria 87

– light scattering properties 51 microemulsions 35, 42
– oil out 51 Mie
ion-selective electrode 35 – light scattering 42
iron – scattering 39
– oxidized 27 misinformation 136
misinterpretation 48
Jaffe reaction model 130, 132, 136
– creatinine 29 Modification of Diet in Renal Disease 7
jaundice – MDRD 7
– classification 67 molecular
– neonatal 67 – orbitals 17
– orbital theory 17
kernicterus 65, 68 – weight 39, 43
kinetic method 124 monitor 4
multiple
laboratory tests 1, 2, 4, 48 – interactions 133
lactate 80 – regression 120, 124, 134
Lambert-Beer law 13 – regression analysis 129, 134
leptons 84
light 11 NCCLS guideline 48, 54
– scattering 35, 36, 38, 84 negative
linear model 133 – interference 56
linear regression 126 nephrotic syndrome
lipemia 8, 11, 15, 35, 36, 83 – increase in triglycerides 90
– interferes 37 NIRS 26
– Zieve’s syndrome 88 nonalcoholic fatty-liver disorder 89
lipodystrophy 89 nonlinear 129, 131, 132
lipolysis nonpolar 40, 50
– effect of alcohol 88 – substances 35
lipophilic drugs 35 null hypothesis 59
lipoproteins 35 – reject 60
lipoprotein lipase 85, 90
lipoprotein X 98 obesity 87
liver 84 oral contraceptives 72
lymphatic duct 84 orbitals 17
organic molecules 17
matrix 3 oximetry
mechanisms 12, 134 – cerebral 26
metabolic syndrome 87 – pulse 22
methemoglobin 22, 25, 27 oxyhemoglobin
– bilirubin interference 27 – decreased 22
methemoglobinemia
– drug exposure 27 pancreatitis 85
– hereditary 27 partitioning 35, 94
methods pathology 83
– nephelometry 38 patient safety 8
– turbidimetry 38 percentage error 52
micelles 40, 42 peroxidase 30, 121, 132, 133, 134
142       Index

persistant hyperbilirubinemia 66 reproducibility 1

phenobarbital 72 resonance 17
phenothiazines 72
phenytoin 72 salicylate 32
phospholipids 40, 50 scatter
phosphorus 77, 98 – light 11, 15
photobilirubin 75 scattered light 17
photoisomers 77 signal 3, 11, 129
physiology 83 skin
picrate – cyanotic 27
– creatinine 29 slope 124
– Jaffe reaction 29 – of interference 59
plot specific absorptivity 14
– of interference 56 spectral interference 134
polar 50 spectrophotometer 12
– molecule 40 spectrophotometry 14, 17, 38
polyene 17 – near infrared 26
polynomial regressions 132 spectrum 25
porphyrin 18, 22 standard deviation 1
positive statistically different 60
– interference 56 statistical testing for significance 129
post-analytic 3 Student’s t-test 54
postprandial study design 120
– chylomicronemia 36 sulfonamides 72
– lipemia 84
pre-analytic 3 thrombotic state 87
precision 1, 2 total error 53
primary mixed hyperlipidemia 85, 86 transmission 12
progressive familial intrahepatic cholestasis 69 transmittance 13
pro-inflammatory 87 trazodone 72
protocol 137 tricyclics 72
pseudohyponatremia 93, 94, 96 triglycerides 31, 40, 83, 132
pulse oximetry 25 – elevated by medications 89
– hydrolysis of 85
quality 1 Trinder chromophore 31
quantification 130 t-test 59, 129, 130
turbidity 8, 11, 15, 35, 44, 93
ratio
– absorbance 25 ultraviolet
Rayleigh – light 12
– light scattering 42 – region 15
– ratio 38 unconjugated bilirubin 30, 71, 77, 133
reflect uric acid 31, 80
– light 17
regression Verapamil 72
– analysis 56, 119 visible
– second-order polynomial 128 – light 84
– stepwise 132 – region 134
renal disease 7
 Index       143

VLDL 42 Zieve’s syndrome 134

– particles 35, 39, 44, 83, 88 – cause of icterus and lipemia 88

waist circumference 87
wavelength 12, 36
– light 18