Clinical Biochemistry 48 (2015) 911–914

Contents lists available at ScienceDirect

Clinical Biochemistry

journal homepage: www.elsevier.com/locate/clinbiochem

Short Communication

Body fluid matrix evaluation on a Roche cobas 8000 system☆

William E. Owen a, Mindy L. Thatcher b, Karolyn J. Crabtree b, Ryan W. Greer b, Frederick G. Strathmann a,c,
Joely A. Straseski a,c, Jonathan R. Genzen a,c,⁎
a
ARUP Institute for Clinical and Experimental Pathology, 500 Chipeta Way, Salt Lake City, UT 84108, USA
b
ARUP Laboratories, 500 Chipeta Way, Salt Lake City, UT 84108, USA
c
Department of Pathology, 15 North Medical Drive East, Salt Lake City, UT 84112, USA

a r t i c l e i n f o a b s t r a c t

Article history: Objectives: Chemical analysis of body fluids is commonly requested by physicians. Because most commercial
Received 10 April 2015 FDA-cleared clinical laboratory assays are not validated by diagnostic manufacturers for “non-serum” and “non-
Received in revised form 13 May 2015 plasma” specimens, laboratories may need to complete additional validation studies to comply with regulatory
Accepted 14 May 2015 requirements regarding body fluid testing. The objective of this report is to perform recovery studies to evaluate
Available online 22 May 2015
potential body fluid matrix interferences for commonly requested chemistry analytes.
Design and Methods: Using an IRB-approved protocol, previously collected clinical body fluid specimens
Keywords:
Body fluid
(biliary/hepatic, cerebrospinal, dialysate, drain, pancreatic, pericardial, peritoneal, pleural, synovial, and vitreous)
Method validation were de-identified and frozen (−20 °C) until experiments were performed. Recovery studies (spiking with high
Matrix effect concentration serum, control, and/or calibrator) were conducted using 10% spiking solution by volume; n = 5
Recovery study specimens per analyte/body fluid investigated. Specimens were tested on a Roche cobas 8000 system (c502,
Spiking experiment c702, e602, and ISE modules).
Assay interference Results: In all 80 analyte/body fluid combinations investigated (including amylase, total bilirubin, urea nitro-
gen, carbohydrate antigen 19-9, carcinoembryonic antigen, cholesterol, chloride, creatinine, glucose, potassium,
lactate dehydrogenase, lipase, rheumatoid factor, sodium, total protein, triglycerides, and uric acid), the average
percent recovery was within predefined acceptable limits (less than ±10% from the calculated ideal recovery).
Conclusions: The present study provides evidence against the presence of any systematic matrix interference
in the analyte/body fluid combinations investigated on the Roche cobas 8000 system. Such findings support the
utility of ongoing body fluid validation initiatives conducted to maintain compliance with regulatory
requirements.
© 2015 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.

Introduction chemistry assays, however, are not considered “FDA-approved” when

Chemical analysis of body fluids can provide clinically useful infor-
mation [1–4]. As such, it is an important service that clinicians frequent-
ly request. Few diagnostic companies routinely evaluate body fluids
(other than serum, plasma, or urine) as part of their assay development
process leading to submission to the Food and Drug Administration
(FDA), thus assay performance specifications for most body fluids are
rarely available from manufacturers. Exceptions include some common-
ly ordered body fluid tests (e.g. CSF glucose and total protein on high-
volume chemistry analyzers [5,6]) as well as select analytes on individ-
ual analyzers (e.g. pleural fluid pH on Radiometer ABL800 FLEX and
Siemens RAPIDPoint blood gas instruments [7,8]). Most clinical
☆ Conflict of Interest: The authors have no relevant conflicts of interest to disclose.
⁎ Corresponding author at: ARUP Laboratories, Department of Pathology, University of
Utah, 500 Chipeta Way, Mail Code 115, Salt Lake City, UT 84108, USA. Fax: +1 801 584
5207.
E-mail address: jonathan.genzen@path.utah.edu (J.R. Genzen).
using body fluids not included in the manufacturer’s package inserts.
This places the onus on individual clinical laboratories to verify assay
performance characteristics in many body fluids when they are offered
for clinical testing.
The Clinical and Laboratory Standards Institute (CLSI) C49-A Analysis
of Body Fluids in Clinical Chemistry; Approved Guideline describes a num-
ber of validation and analytical considerations when testing alternative
sample types [9]. Of note, a “primary concern is the validity of the body
fluid matrix for the measurement system” [9]. Additionally, a newly re-
vised checklist item (COM.40620) in the College of American Patholo-
gists (CAP) Laboratory Accreditation Program has more fully described
their requirements for body fluid validations and reporting by CAP-
accredited laboratories [10]. This checklist item specifically notes that
“method performance specifications for blood specimens may be used
for body fluids if the laboratory can reasonably exclude the existence
of matrix interferences affecting the latter either by reference in the pro-
cedure manual to published literature or by evaluation for interferences
due to matrix effects by performing an appropriate study…” [10]. Useful

http://dx.doi.org/10.1016/j.clinbiochem.2015.05.012
0009-9120/© 2015 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.
912 W.E. Owen et al. / Clinical Biochemistry 48 (2015) 911–914

guides describing several important considerations in body fluid valida-
tion studies are available online [11] and in print [4].
A shared concept described in each of the resources above is the im-
portance of evaluating body fluids for matrix effects. Recovery experi-
ments are one way to identify whether a matrix effect causes a
constant and/or systematic error [12]. In such studies, it is recommend-
ed that the supplementing (“spiking”) material make up no more than
10% of the final volume [11,12]. Ideally, the spiking material is a high-
concentration specimen of the same type used in the vendor’s FDA-
cleared validation (e.g. serum or plasma), although other materials
(standards, calibrators, or controls) may also be used [9].
In the present experiments, recovery studies were conducted across
10 different fluid types to look for potential matrix effects. These studies
were conducted on a Roche cobas 8000 chemistry system (c502, c702,
e602, ISE) for analytes and body fluids that have been described in text-
books or guidelines [1,9,13], identified in literature review [2,14,15],
and/or requested more frequently in current and/or prior laboratory
settings.
For each analyte/body fluid combination (Table 1), 5 specimens of a
given body fluid type were tested. Body fluids were chosen preferential-
ly based on having a range of different baseline analyte concentrations.
By using some body fluids in more than one analyte matrix evaluation
(as shown in Supplementary Table 2), the number of body fluids used
(n = 185) was less than the number that would have been needed if
each body fluid had been used only once (predicted n = 500).
For most recovery experiments, each body fluid specimen was
spiked with a high-concentration serum specimen in the ratio of
90% body fluid + 10% spiking solution (by volume) and then mixed
thoroughly (see Supplementary Table 2). For CSF total protein studies,
a 95% fluid/5% serum spiking solution ratio was used to keep final re-
sults within the assay analytical measurement range (AMR). Calibrator
and/or control material was used for spiking when a high-concentration
serum specimen was not available. The “expected final concentration”
was calculated as:

 
expected final concentration ¼ ½X fluid  0:9 þ ½X spike  0:1

Materials and methods
Analytes tested included amylase (AMY), total bilirubin (Bili T),
blood urea nitrogen (BUN), carbohydrate antigen 19-9 (CA 19-9),
carcinoembryonic antigen (CEA), cholesterol (CHOL), chloride (Cl−),
creatinine (Cr), glucose (GLU), potassium (K+), lactate dehydrogenase
(LDH), lipase (LIP), rheumatoid factor (RF), sodium (Na+), total protein
(TP), triglycerides (TRIG), and uric acid (URIC); all assays from Roche
Diagnostics (Indianapolis, IN) with specific cobas 8000 modules as
listed in Table 1 and Supplementary Table 1. All assays were conducted
using serum/plasma configurations, with the exception of CSF (total
protein), which used a separate Roche Total Protein Urine/CSF kit (Sup-
plementary Table 1).
According to an IRB-approved protocol, clinical body fluid specimens
(biliary/hepatic, cerebrospinal, dialysate, drain, pancreatic, pericardial,
peritoneal, pleural, synovial, and vitreous) were obtained from clinical
storage after assay-specific discard intervals: AMY, 14 days − 20 °C;
Bili T, 14 days − 20 °C; BUN, 14 days 4 °C; CA 19-9, 90 days − 20 °C;
CEA, 90 days − 20 °C; CHOL, 14 days − 20 °C; Cl−, 14 days − 20 °C;
Cr, 14 days 4 °C; GLUC, 14 days − 20 °C; K+, 14 days − 20 °C; LDH,
14 days 4 °C; LIP, 14 days − 20 °C; Na+, 14 days − 20 °C; RF, 14 days
− 20 °C; TP, 14 days − 20 °C; TRIG, 14 days − 20 °C; and URIC,
14 days − 20 °C. Retrieved specimens were de-identified and stored
for approximately 0–4 additional months (−20 °C) until experiments
were conducted. Evaluation of freezing on matrix integrity was not
conducted.
where [X]fluid is the baseline analyte concentration in body fluid and
[X]spike is the baseline analyte concentration in serum or other spiking
material. Each spiked fluid specimen was then tested in duplicate and
an average final concentration [X]spiked average was determined. As indi-
cated in Supplementary Table 2, there were two body fluid specimens
that did not have sufficient volume to be run in duplicate and were
therefore analyzed with single measurements. The “percent recovery”
was calculated as:
 
½X spiked
percent recovery ¼
average
 100
expected final concentration
Data analysis was performed in Excel 2010 (Microsoft Inc, Redmond,
WA) and is presented as mean ± standard deviation (SD), unless other-
wise indicated. The threshold for acceptable performance was defined
as an average difference from 100% recovery that was less than ±10%.
Results were plotted in SigmaPlot 11 (Systat Software Inc, San Jose, CA).
Results
All analytes tested met threshold criteria for acceptable average per-
cent recovery (less than ± 10% from expected 100% recovery) in the re-
spective fluids tested (Table 1). Only a small number of analytes
demonstrated a deviation from ideal average percent recovery of great-
er than ±5%: Bili T in pleural fluid (−7.3%), RF in pleural fluid (−7.3%)
and synovial fluid (−6.0%), LIP in pericardial fluid (7.9%), and GLUC in

Table 1
Percent recovery.

Analyte Instrument Biliary and CSF Dialysate Drain Fluid Pancreatic Pericardial Peritoneal/Ascites Pleural Synovial Vitreous
Hepatic Fluid Fluid Fluid Fluid Fluid Fluid

AMY c502 – – – 101.3 ± 5.8 104.7 ± 2.7 – 101.7 ± 1.2 102.6 ± 2.4 – –
Bili T c502 99.1 ± 1.4 – – 97.3 ± 3.2 – – 101.1 ± 1.0 92.7 ± 2.9 – –
BUN c502 – – – – – – – – – 101.0 ± 1.4
CA 19-9 e602 101.6 ± 2.5 103.1 ± 1.1 – – 104.0 ± 8.0 – 101.1 ± 6.5 96.2 ± 1.8 – –
CEA e602 – 101.1 ± 2.3 – – 107.1 ± 7.7 – 102.6 ± 5.7 100.2 ± 3.6 – –
CHOL c502 – – – 103.9 ± 1.9 – 102.6 ± 2.0 100.4 ± 6.1 102.6 ± 1.8 – –
Cl− ISE – 100.2 ± 0.5 – 99.9 ± 0.5 100.2 ± 0.5 99.3 ± 0.9 99.7 ± 0.4 99.5 ± 0.6 – 99.5 ± 0.6
Cr c502 – – – – – – – – – 104.4 ± 1.4
GLUC c502 – 100.4 ± 0.5 100 ± 0.6 – – 101.7 ± 1.1 102.9 ± 1.2 100.5 ± 0.9 103.1 ± 1.0 91.3 ± 3.9
K+ ISE – 99.7 ± 0.9 – 100.3 ± 0.4 100.1 ± 2.0 100.4 ± 0.5 100.2 ± 0.4 100.1 ± 0.3 – 100.4 ± 0.2
LDH c502 – 104.7 ± 1.9 – – – 101.3 ± 1.8 101.7 ± 1.4 104.0 ± 4.0 101.3 ± 1.0 –
LIP c502 102.3 ± 2.3 – – 101.6 ± 4.6 100.9 ± 1.1 107.9 ± 4.7 102.6 ± 4.0 102.9 ± 4.2 109.1 ± 6.2 –
Na+ ISE – 100.2 ± 0.2 – 100.3 ± 0.3 100.7 ± 1.2 100.0 ± 0.2 100.0 ± 0.4 100.1 ± 0.4 – 100.4 ± 0.3
RF c702 – 104.1 ± 1.5 – – – 99.5 ± 3.6 – 92.7 ± 7.7 94.0 ± 4.3 –
TP c502 – 99.8 ± 1.4 – – – 100.3 ± 1.1 104.3 ± 3.5 101.2 ± 0.7 100.9 ± 1.2 –
TRIG c502 – – – 101.1 ± 2.9 – 101.1 ± 3.4 96.5 ± 1.0 101.2 ± 1.8 – –
URIC c502 – – – 101.9 ± 1.2 – – 101.3 ± 1.9 100.3 ± 0.5 102.5 ± 1.7 –
 W.E. Owen et al. / Clinical Biochemistry 48 (2015) 911–914 913

vitreous fluid (−8.7%). Additionally, the SD for average percent recov-
ery was less than 7% for all analytes tested, emphasizing that no obvious
outliers were observed for any analyte/body fluid combinations. As an
example, Fig. 1 shows the individual specimen percent recovery results
for all analytes tested in pleural fluid, a specimen type commonly re-
quested for chemical analysis. While minor variability may be observed
between individual points—for example, slight under-recovery at the
low end for RF (Fig. 1L)—an overall consistent recovery near the 100%
goal is observed across analytes.
Conclusions
The present experiments demonstrate an excellent percent recovery
in spiking studies for all analytes tested in their respective fluids. These
data serve as one source of evidence that there is no consistent, assay-
specific “matrix effect” for these analytes in the body fluids tested on
this set of instrumentation (Roche cobas c502, c702, e602, ISE). The
clinical utility of these and other body fluid assays are described in
more detail elsewhere [1,2,4,9], but include differentiation of transu-
dates from exudates (PROT, LDH, GLUC), identification of chylous effu-
sions (CHOL, TRIG), pancreatic fluid collections (AMY, LIP), urine
contamination (Cr), biliary leak (TBIL), gout (URIC), and to provide sup-
portive evidence of malignancy (CA 19-9, CEA).
These experiments do not address all components of a complete
body fluid validation, which commonly include accuracy, precision,
sensitivity, AMR, and reference intervals. Additionally, we do not have
specific information on the source location of drain fluids used in this
analysis. Other possible interferences (such as pH) were also not inves-
tigated [14].
As matrix effects may be assay- and/or instrument-specific, the
present findings should not be considered generalizable to other plat-
forms without additional verification studies. These experiments do,

Fig. 1. Pleural fluid – percent recovery. Example showing all data for one body fluid type (pleural fluid). Percent recovery of all spiked pleural fluid specimens for A, AMY; B, Bili T; C, CA 19-
9; D, CEA; E, CHOL; F, Cl−; G, GLU; H, K+; I, LDH; J, LIP; K, Na+; L, RF; M, TP; N, TRIG; O, URIC.
914 W.E. Owen et al. / Clinical Biochemistry 48 (2015) 911–914

however, provide evidence for “reasonably exclude[ing] the existence design support to MLT and KJC during their student projects. This re-
of matrix interferences” when relying on method performance specifi- search was supported by the ARUP Institute for Clinical and Experimen-
cations derived from blood for potential body fluid testing on these tal Pathology.
instruments [10]. The present recovery experiments also suggest that
pursing such studies on other platforms may be productive in both
verifying assay performance in body fluids, as well as assisting in References
compliance with existing regulatory standards.
[1] Hussong J, Kjeldsberg C. Kjeldsberg's Body Fluid Analysis. Chicago, IL: American So-
Supplementary data to this article can be found online at http://dx. ciety for Clinical Pathology; 2015.
doi.org/10.1016/j.clinbiochem.2015.05.012. [2] Block DR, Algeciras-Schimnich A. Body fluid analysis: clinical utility and applicability
of published studies to guide interpretation of today's laboratory testing in serous
fluids. Crit Rev Clin Lab Sci 2013;50:107–24.
Abbreviations [3] Kopcinovic LM, Culej J. Pleural, peritoneal and pericardial effusions—A biochemical
AMR analytical measurement range approach. Biochem Med (Zagreb) 2014;24:123–37.
AMY amylase [4] Block DR, Franke DDH. Quick Guide to Body Fluid Testing. Washington, DC: AACC;
2015.
Bili T total bilirubin [5] Glucose Hexokinase (GLUC3) Package Insert. 2012-04 V8. Indianapolis, IN: Roche
BUN blood urea nitrogen Diagnostics; 2012.
CA 19-9 carbohydrate antigen 19-9 [6] Total Protein Urine/CSF (TPUC3) Package Insert. 2010-01 V7. Indianapolis, IN: Roche
Diagnostics; 2012.
CAP College of American Pathologists
[7] 510(k) Premarket Notification Submission (ABL835 FLEZ Analyzer with Pleaural pH)
CEA carcinoembryonic antigen and Determination of Substantial Equivalence. 2011. K110416. http://www.
CHOL cholesterol accessdata.fda.gov/cdrh_docs/pdf11/K110416.pdf. [Accessed 5/8/2015].
Cl− chloride [8] 510(k) Premarket Notification Submission (RAPIDPoint®& 400/405/500 Systems)
and Determination of Substantial Equivalence. 2011. K122539. http://www.
CSLI Clinical Laboratory Standards Institute accessdata.fda.gov/cdrh_docs/pdf12/K122539.pdf. [Accessed 5/8/2015].
Cr creatinine [9] Analysis of Body Fluids in Clinical Chemistry. Approved Guideline. C49-A. , vol. 27,
FDA Food and Drug Administration No 14Clinical and Laboratory Standards Institute (CLSI); April 2007 [www.clsi.org.
Accessed 3/26/2015].
GLUC glucose [10] College of American Pathologists. Laboratory Accreditation Program. COM.40620
IRB Institutional Review Board Body Fluid Validation. www.cap.org. [Accessed 11/11/2014].
K+ potassium [11] Block DR. Practical Guide to the Analytical Validation of Body Fluid Chemistry
Testing. Mayo Medical Laboratories Communique; March 2013 (http://www.
LDH lactate dehydrogenase mayomedicallaboratories.com/articles/communique/2013/03-bodyfluid-testing/
LIP lipase index.html. Accessed 3/13/2015).
Na+ sodium [12] Westgard JO. Basic Method Validation. Madison, WI: Westgard QC; 2008.
[13] Henry's Clinical Diagnosis and Management by Laboratory Methods. 22nd ed.
RF rheumatoid factor Philadelphia, MA: Elsevier Saunders; 2011.
TP total protein [14] Lin MJ, Hoke C, Dlott R, Lorey TS, Greene DN. Performance specifications of common
TRIG triglycerides chemistry analytes on the AU series of chemistry analyzers for miscellaneous body
fluids. Clin Chim Acta 2013;426:121–6.
URIC uric acid
[15] Schumann GB. Clinical utility of body fluid analyses: An overview. Clin Lab Sci 1994;
7:28–31.
Acknowledgements

The authors would like to thank the Department of Medical Labora-

tory Sciences at Weber State University for providing experimental