Artifactual Hyperbilirubinemia Due to
Paraprotein Interference
Liron Pantanowitz, MD; Gary L. Horowitz, MD; Jan N. Upalakalin, MD; Bruce A. Beckwith, MD
● Context.—Paraprotein interference in automated chem-
istry is uncommon. We describe 2 patients with parapro-
teinemia and elevated total bilirubin levels measured er-
roneously using the Roche total bilirubin assay.
Objectives.—To explain the mechanism of this artifac-
tual hyperbilirubinemia and to determine its frequency in
patients with monoclonal or increased immunoglobulins.
Materials and Methods.—The assay was performed man-
ually using serum from 2 index patients and from control
patients (without M proteins). Total bilirubin was also de-
termined using another manufacturer’s assay. A prospective
study was then undertaken using serum from 100 consec-
utive patients with various monoclonal gammopathies and
from 13 patients with polyclonal hypergammaglobulinemia
and cryoglobulins. For all patients, serum immunoglobulin
(Ig) G, IgA, IgM, total and direct bilirubin, creatinine, and
a direct spectrophotometric assessment of icterus were
measured.
Results.—After the addition of assay reagents, a white
M onoclonal immunoglobulins (Igs) may interfere with
a wide variety of automated nephelometric, turbi-
dimetric, and immunologic assays.1 Their interference
with total bilirubin measurements has not been reported.
We identified 2 patients within a 10-month period who
had paraproteins and spuriously elevated total bilirubin
levels, when measured using the Roche total bilirubin as-
say on the Hitachi 917 automated analyzer (Roche Diag-
nostics, Indianapolis, Ind). The unexpected hyperbiliru-
binemia in these patients posed a perplexing problem, as
neither patient was jaundiced. We attempted to explain the
mechanism of this artifactual hyperbilirubinemia and also
to determine the frequency of this phenomenon in patients
with monoclonal or increased immunoglobulins.
REPORT OF CASES
Index Case 1
An 88-year-old hypertensive man with multiple myeloma was
admitted to our hospital because of a right lower lobe pneumo-
nia. He was taking metoprolol tartrate, lisinopril, and also furo-
semide for congestive cardiac failure. Initial laboratory tests re-
vealed normal values for complete blood count, serum electro-
Accepted for publication July 17, 2002.
From the Department of Pathology, Beth Israel Deaconess Medical
Center, Harvard Medical School, Boston, Mass.
Reprints: Department of Pathology, Beth Israel Deaconess Medical
Center, Harvard Medical School, 330 Brookline Ave, Boston, MA,
02215 (e-mail: lpantano@caregroup.harvard.edu).
Arch Pathol Lab Med—Vol 127, January 2003
precipitate formed in the reaction mixtures containing se-
rum from the index patients, but not in other samples. This
turbidity, rather than the expected color change to pink,
increased the absorbance and falsely elevated the total bil-
irubin value. Serum from both index patients was anicteric,
their direct bilirubin measurements were unaffected, and
total bilirubin measured using an alternate assay was nor-
mal. Among the 113 patients studied, no additional spuri-
ous total bilirubin values were detected.
Conclusion.—Paraprotein interference with the Roche
automated total bilirubin assay is caused by precipitate for-
mation. This interference is rare and probably idiosyncrat-
ic. Spurious hyperbilirubinemia from paraprotein interfer-
ence may cause clinical confusion. If artifactual elevation
of total bilirubin is suspected, the laboratory should ex-
amine the specimen for icterus (manually or by spectro-
photometry) or measure total bilirubin using a different
method.
(Arch Pathol Lab Med. 2003;127:55–59)
lytes, creatinine, and phosphate. Total bilirubin (refer to Table 2
for reference ranges) on admission, determined using the Roche
assay on the Hitachi 917 analyzer, was 9.0 mg/dL (153.9 mmol/
L), and the direct bilirubin was 0.1 mg/dL (1.7 mmol/L). The
total bilirubin was reported as 19.9 mg/dL (340.3 mmol/L) 4
hours later. These serum specimens were anicteric and showed
no evidence of lipemia or hemolysis. Clinically, the patient was
not jaundiced, and there was no supporting evidence for hemo-
lysis or liver disease. The IgG, IgA, and IgM levels were 6000
mg/dL (60 g/L), 12 mg/dL (0.12 g/L), and 6 mg/dL (0.06 g/
L), respectively. Serum protein electrophoresis and immunofixa-
tion electrophoresis revealed a monoclonal IgG-l M protein. Total
bilirubin levels measured using another assay were all less than
1.5 mg/dL (25.7 mmol/L). Four months later when he returned
for follow-up care, the patient’s total bilirubin levels using the
Roche assay remained erroneously elevated, up to 20.2 mg/dL
(345.4 mmol/L).
Index Case 2
A 56-year-old asthmatic woman was admitted to our hospital
for rituximab (anti-CD20 monoclonal antibody) therapy for pro-
gressive Waldenström macroglobulinemia. She was taking hor-
mone replacement therapy, iron supplements, fluticasone propi-
onate, and albuterol. Initial laboratory tests revealed a normal
complete blood count, serum electrolytes, and creatinine. Total
bilirubin measurements using the Roche assay on the Hitachi 917
instrument ranged between 1.4 mg/dL (24 mmol/L) and 4.9 mg/
dL (83.8 mmol/L), and the direct bilirubin level was 0.1 mg/dL
(1.7 mmol/L). Serum protein electrophoresis and immunofixation
electrophoresis revealed a monoclonal IgM-k. Immunoglobulin
levels were recorded as follows: IgG, 446 mg/dL (4.46 g/L); IgA,
Artifactual Hyperbilirubinemia—Pantanowitz et al 55
 Table 1. Repetitive Total Bilirubin Measurements for Index Case 1 (Patient With Myeloma) Using
the Hitachi 917 Analyzer
Blood Total Bilirubin, mg/dL∗
Specimen
No. Reading 1 Reading 2
1 19.9 57.8
2 9.0 84.0
3 68.4 9.2
∗ For SI units in mmol/L, multiply readings by a factor of 17.1.
Figure 1. Images of the Roche total bilirubin assay performed at mac-
roscale. Samples from control patients with total bilirubin levels of 0.2
mg/dL (A) and 21.3 mg/dL (B), and from the second index patient (C).
Figure 1, A shows neat serum and Figure 1, B shows the appearance
of the final reaction mixtures. Note that the final reaction mixture from
the patient with hyperbilirubinemia (B) was clear and pink, whereas
that of the index patient shows marked turbidity that obscures the back-
ground lines and is not pink. (To convert bilirubin values to mmol/L,
multiply by 17.1.)
,7 mg/dL (,0.07 g/L); and IgM, 4088 mg/dL (40.9 g/L). One
week after admission, the total bilirubin level for this patient was
reported as 10.6 mg/dL (181.3 mmol/L), and the direct bilirubin
remained at 0.1 mg/dL (1.7 mmol/L). These serum specimens
were anicteric and showed no evidence of lipemia or hemolysis.
The patient’s liver enzymes were normal. To investigate this dis-
crepancy, we measured the total bilirubin using a different assay
56 Arch Pathol Lab Med—Vol 127, January 2003
Icterus
Reading 3 Reading 4 (I) Index
4.9 52.5 0.0
13.8 78.3 0.0
71.3 6.8 0.0
and found it to be 0.3 mg/dL (5.13 mmol/L). Up until the point
when we contacted the patient’s clinicians, they had been con-
fused about the hyperbilirubinemia, because she was not clini-
cally jaundiced. Subsequent specimens continued to show erro-
neous total bilirubin values, as high as 16.5 mg/dL (282.2 mmol/
L).
METHODS
Interference Demonstration
Total bilirubin, direct bilirubin, IgG, IgA, IgM, and creatinine
levels as well as a direct photometric measurement for icterus (I
index) were determined on the Roche Hitachi 917 analyzer
(Roche Diagnostics, Indianapolis, Ind). The Roche serum total
bilirubin assay used is an endpoint chromogenic assay. The re-
action in this assay requires the sequential addition of 2 liquid
reagents (R1 and R2) to the patient’s serum sample within the
reaction cuvette. In this assay, after unconjugated bilirubin is sol-
ubilized by the addition of a ‘‘solubilizing agent,’’ all of the bil-
irubin is coupled with a diazonium ion in a strongly acid me-
dium (pH 1–2) at 378C to form azobilirubin.2 The color (pink)
intensity of the azobilirubin produced is proportional to the total
bilirubin concentration. The absorbance (at 546 nm, with a bich-
romatic correction at 600 nm) is determined spectrophotometri-
cally. The R1 working reagent contains 85 mmol/L of sodium
acetate buffer, 110 mmol/L of sulfamic acid, surfactant, and ‘‘sol-
ubilizer.’’ R2 contains 3 mmol/L of diazonium ion and 100
mmol/L of hydrochloric acid.
Direct photometric measurements for icterus, lipemia, and he-
molysis are performed on the Hitachi 917 by adding sample to
saline in a separate cuvette and taking a total of 6 absorbance
readings. For the I index, the wavelengths used are 480 nm and
505 nm, with corrections for the contributions from hemolysis
(570 nm and 600 nm) and lipemia (660 nm and 700 nm).3 The I
index is calibrated such that the units correspond to the bilirubin
concentration in mg/dL. Total bilirubin levels in samples from
the index patients were also determined using the Dade Behring
Dimension (Newark, Del) clinical chemistry system. In this assay,
bilirubin (unconjugated) in the sample is solubilized by dilution
in a mixture of caffeine, benzoate, acetate, and EDTA.4
Imprecision of the Roche total bilirubin assay was demonstrat-
ed by measuring the total bilirubin value several times on the
Hitachi 917, using different samples received for the first index
patient. Each result was then compared with the corresponding
I index for that specimen. To illustrate the phenomenon causing
the interference with the total bilirubin measurements, the Roche
assay was performed manually for both index patients. All vol-
umes were accordingly increased in magnitude, while still main-
taining the sample-reagent ratios specified by the manufacturer.
For the first patient’s sample, absorbance at a wavelength of 546
nm was measured after each step of the assay using a Gilford
Stasar III spectrophotometer (Oberlin, Ohio). The same measure-
ments were manually repeated on distilled water (blank) and
again on serum from a control patient (with no M protein),
whose total bilirubin level was 18.3 mg/dL (313 mmol/L). For
the second index patient’s serum and for 2 control patient sam-
ples, one with a normal total bilirubin level of 0.2 mg/dL (3.4
mmol/L) and the other with an elevated level of 23.1 mg/dL (395
mmol/L), the individual steps of the assay were recorded using
Artifactual Hyperbilirubinemia—Pantanowitz et al

digital photography. All incubations were timed as in the auto-
mated assay and performed in a 378C water bath. The absorbance
data for these samples from the total bilirubin assay on the Hi-
tachi 917 were retrieved. Attempts to redissolve the precipitate
from the macroscale reaction mixture of the second index patient
were unsuccessful, precluding further analyses of the precipitate.
Patient Selection and Analysis
We performed a prospective study using serum samples from
100 available, consecutive patients with electrophoretically doc-
umented monoclonal gammopathy. These samples were identi-
fied during the routine interpretation of serum electrophoresis in
our clinical laboratory during a 6-month period. An additional
11 individuals who presented during this period with polyclonal
hypergammaglobulinemia and 2 with cryoglobulins were also
studied. For all patients, serum IgG, IgA, IgM, total and direct
bilirubin, creatinine, and an I index were determined on the Hi-
tachi 917 instrument. Any discrepancies between the serum color,
I index, and total bilirubin measurements were noted. Creatinine
was measured because of previous reports of false measurements
of this analyte due to paraprotein interference with automated
methods.5–7
RESULTS
Assay Interference
Total bilirubin measurements, when repeated several
times for our first patient, revealed an imprecise and pos-
itive bias (Table 1). When the assay was performed man-
ually, a fine white flocculent precipitate, but no change in
color, was noted after both R1 and R2 reagents were add-
ed to serum for both the first and second (Figure 1) pa-
tients. No such precipitate was evident in either the control
or blank specimens. The precipitate remained stable dur-
ing warming (378C) and cooling (48C) of the specimen.
Only the icteric serum from control patients developed a
color (clear pink) change at the end of the assay. Inspec-
tion of the absorbance readings from the total bilirubin
assay on the Hitachi 917 revealed that the specimen con-
taining the paraprotein showed increased, but delayed,
absorbance when compared with the controls (Figure 2).
Population Analysis
The total and direct bilirubin, monoclonal immunoglob-
ulin, and serum creatinine measurements for all patients
Arch Pathol Lab Med—Vol 127, January 2003
Figure 2. Absorbance readings taken during
the total bilirubin assay performed on the
Roche Hitachi 917 using samples from con-
trol patients with total bilirubin levels of 0.2
mg/dL (A) and 21.3 mg/dL (B), and from the
second index patient (C). Note the delayed
rise in absorbance of sample C, which gave
a reading of 8.5 mg/dL. The vertical bars in-
dicate the initial and final reading times used
to calculate results. (To convert bilirubin val-
ues to mmol/L, multiply by 17.1.)
are shown in Table 2. No spurious total bilirubin or cre-
atinine measurements using Roche assays on the Hitachi
917 analyzer were detected in any of our patients with
monoclonal gammopathy, polyclonal hypergammaglobu-
linemia, or cryoglobulins.
COMMENT
We identified 2 patients whose serum paraproteinemia
resulted in a spuriously elevated total bilirubin measure-
ment using a Roche assay on the Hitachi 917 automated
analyzer. Our first patient with myeloma had a monoclo-
nal IgG-l, and our second patient with Waldenström mac-
roglobulinemia had a monoclonal IgM-k. It seems clear
from the clinical histories, serum color, direct bilirubin
levels, and I index measurements on the Hitachi 917, as
well as the total bilirubin values determined using the
Dade Behring Dimension instrument, that the Roche total
bilirubin levels were falsely elevated and that the most
likely cause of the interference was the paraprotein.
The absence of a solubilizing agent in the direct bili-
rubin assay strongly suggests that it is this agent that
caused the interfering precipitate. The turbidity of this
suspended precipitate, and not a color change, resulted in
increased light absorption when measured spectrophoto-
metrically, which artifactually raised the total bilirubin
measurements. It is interesting that even with the bich-
romatic correction used in this assay on the Hitachi 917
the untoward effect of the turbidity is not eliminated.
Along these lines, one can ascribe the marked imprecision
of the Roche total bilirubin values on these samples to the
fact that this precipitate likely formed at different times,
in varying amounts, and with different particle sizes on
each separate occasion. It is ironic that the human eye, but
not the Hitachi 917, can easily distinguish the pink color
of the genuine diazo reaction from the white turbidity of
the precipitate. In contrast, the Hitachi 917 can detect the
icterus (or lack thereof) of a serum sample more precisely
than the human eye. Unfortunately, it cannot use the dis-
crepancy between the I index and the measured total bil-
irubin to alert the operator to the problem under discus-
sion.
Monoclonal immunoglobulins have been shown to in-
Artifactual Hyperbilirubinemia—Pantanowitz et al 57
 Table 2. Serum Immunoglobulin, Bilirubin, and Creatinine Concentrations (Shown as Means With Ranges in
Parentheses) for Patients Studied. Nonmonoclonal Immunoglobulin Levels Are Not Shown, Except for Those Patients
With Other Conditions (Polyclonal Hypergammaglobulinemia and Cryoglobulins)
Total Direct Serum
No of IgG, IgA, IgM, Bilirubin Bilirubin, Icterus Creatinine,
Patients (mg/dL)∗ (mg/dL)∗ (mg/dL)∗ (mg/dL)∗ (mg/dL)∗ Index† mg/dL∗
Reference range ... 600–1500 60–380 50–250 0–1.5 0–0.3 0 0–1.3
Monoclonal IgG 62 2244 ... ... 0.3 0.1 0.5 1.1
(461–8784) (0.1–0.8) (0–0.3) (0–1) (0.3–5.3)
Monoclonal IgA 16 ... 1188 ... 0.3 0.1 0.7 1.4
(330–3944) (0.1–1.6) (0–0.3) (0–2) (0.5–4.9)
Monoclonal IgM 22 ... ... 1351 0.4 0.1 0.5 1.1
(183–5180) (0.2–1.2) (0–0.3) (0–1) (0.7–2.4)
Other conditions 13 2205 283 161 0.5 0.2 0.8 1.1
(1599–3751) (81–602) (32–342) (0.2–2.5) (0–1.8) (0–2) (0.7–3.4)
∗ For SI units, multiply IgG results by 0.01 for g/L, IgA by 10 for mg/L, IgM by 10 for mg/L, bilirubin by 17.1 for mmol/L, and creatinine by 88.4
for mmol/L.
† Refer to ‘‘Material and Methods’’ for discussion of icterus index.

terfere with various automated measurements. False mea-
surements due to paraprotein interference with automated
methods have been reported for creatinine,5–7 urea,8 phos-
phate,9–13 calcium,14 acetaminophen,15 serum iron,16 hemo-
globin,17–18 C-reactive protein, and certain microbiological
serological tests.1 In these reports, the interfering parapro-
teins were either monoclonal IgG or IgM, but never IgA
proteins. The method of interference documented in these
assays, as was demonstrated in this study, was attributed
to the precipitation of the paraprotein, usually in an acid
medium. The likelihood of this interference in most of
these reports was also unrelated to either the paraprotein
type or concentration. Only a single report in the Japanese
literature noted a human IgG-l–type M protein that in-
terfered with the automated determination of direct bili-
rubin.19 However, paraprotein interference has never be-
fore been reported in the literature for total bilirubin mea-
surements. Abnormal bilirubin binding to paraproteins
can occur in myeloma patients,20 but this reaction has not
resulted in falsely elevated total bilirubin levels. Interfer-
ence with the measurement of total bilirubin due to pro-
pranolol with certain automated diazo techniques has
been documented.21–23 Our 2 index patients had not been
on propranolol therapy, and none of the drugs being taken
by these patients, including rituximab, have been reported
to erroneously affect bilirubin values.
A customer bulletin received from Roche Diagnostics
informed our clinical laboratory that a positive bias may
indeed occur with their liquid total bilirubin assay in my-
eloma patient samples,24 but to date they have provided
no explanation or information regarding the frequency of
this phenomenon. Based on our series of patients with
monoclonal gammopathies, the incidence of paraproteins
interfering with the Roche total bilirubin assay is very low.
This interference is probably idiosyncratic and is not de-
pendent on either the paraprotein type or concentration.
The interference in our 2 index patients occurred with
both an IgG-l and IgM-k paraprotein, and failed to occur
in the other 100 patients studied, among whom a wide
variety of monoclonal immunoglobulins was exhibited.
Furthermore, concentrations as high as 8784.4 mg/dL
(87.8 g/L) for monoclonal and 3751 mg/dL (37.5 g/L) for
polyclonal immunoglobulins did not demonstrate this
phenomenon.
Whenever high total bilirubin levels are found in pa-
tients with plasma cell dyscrasias, Waldenström macro-
globulinemia, and possibly lymphomas associated with
abnormal immunoglobulin synthesis, in the absence of
signs and symptoms of jaundice, the possibility of a spu-
rious hyperbilirubinemia due to an interfering paraprotein
should be entertained. Awareness of this phenomenon
may help prevent unnecessary concern and expensive in-
vasive investigations. If artifactual elevation of total bili-
rubin is suspected, we recommend the following:
1. check the serum color by eye to see if it is truly ic-
teric;
2. obtain the spectrophotometric measurement using an
I index, if available on the autoanalyzer (such as the Hi-
tachi 917);
3. rerun the assay to demonstrate any imprecision be-
yond what is typically observed;
4. measure direct bilirubin on the same specimen, as
this assay does not require any solubilizing agent;
5. measure the total bilirubin using a different method;
and
6. correlate the total bilirubin result with the clinical
information, if available.
References
1. Yamada K, Yagihashi A, Ishii S, et al. Interference with nephelometric assay
of c-reactive protein and antistreptolysin-O by monoclonal IgM-k from a mye-
loma patient. Clin Chem. 1997;43:2435–2437.
2. Total bilirubin [package insert]. Indianapolis, Ind: Roche Diagnostics; 2001.
3. Reference Guide: Boehringer-Mannheim/Hitachi 917. Indianapolis, Ind:
Boehringer-Mannheim; 1998:245–256.
4. Total bilirubin Flex reagent cartridge [package insert]. Newark, Mass: Dade
Behring Inc; 1999.
5. Jaynes PK, Witte DL, Simpson CH, Feld RD. Interference by some mono-
clonal IgM’s in discrete serum creatinine analysis. Clin Chem. 1982;28:1580.
6. Datta P, Graham GA, Schoen I. Interference by IgG paraproteins in the Jaffe
method for creatinine determination. Am J Clin Pathol. 1986;85:463–468.
7. Hummel KM, Von Ahsen N, Kuhn RB, et al. Pseudohypercreatinemia due
to positive interference in enzymatic creatinine measurements caused by mono-
clonal IgM in patients with Waldenström’s macroglobulinemia. Nephron. 2000;
86:188–189.
8. Smith JD, Nobiletti J, Freed M, Malkus H, Donabedian R. Interference with
the Astra 8 and Synchron CX3 assays of urea nitrogen in serum by a high-Mr
inhibitor in a patient with multiple myeloma. Clin Chem. 1992;38:598–599.
9. Sonnenblick M, Eylath U, Brisk R, Eldad C, Hershko C. Paraprotein inter-
ference with colorimetry of phosphate in serum of some patients with multiple
myeloma. Clin Chem. 1986;32:1537–1539.
10. McCloskey EV, Galloway J, Morgan MA, Kanis JA. Pseudohyperphospha-
taemia in multiple myeloma. Br Med J. 1989;299:1381–1382.
11. Mandry JM, Posner MR, Tucci JR, Eil C. Hyperphosphatemia in multiple
myeloma due to phosphate-binding immunoglobulin. Cancer. 1991;68:1092–
1094.
12. Bowles SA, Tait RC, Jefferson SG, Gilleece MH, Haeney MR. Character-

58 Arch Pathol Lab Med—Vol 127, January 2003 Artifactual Hyperbilirubinemia—Pantanowitz et al

istics of monoclonal immunoglobulins that interfere with serum inorganic phos-
phate measurement. Ann Clin Biochem. 1994;31:249–254.
13. Cohen AM, Magazanik A, van-der Lijn E, Shaked P, Levinsky H. Pseudo-
hyperphosphataemia incidence in an automatic analyzer. Eur J Clin Chem Clin
Biochem. 1994;32:559–561.
14. John R, Oleesky D, Issa B, et al. Psuedohypercalcaemia in two patients
with IgM paraproteinaemia. Ann Clin Biochem. 1997;34:694–696.
15. Hullin DA. An IgM paraprotein causing a falsely low result in an enzymatic
assay for acetaminophen. Clin Chem. 1999;45:155–156.
16. Bakker AJ. Influence of monoclonal immunoglobulins in direct determi-
nations of iron in serum. Clin Chem. 1991;37:690–694.
17. Goodrick MJ, Boon RJ, Bishop RJD, Copplestone JA, Prentice AG. Inac-
curate haemoglobin estimation in Waldenström’s macroglobulinemia: unusual re-
action with monomeric IgM paraprotein. J Clin Pathol. 1993;46:1138–1139.
18. Roberts WL, Fontenot JD, Lehman CM. Overestimation of hemoglobin in
Arch Pathol Lab Med—Vol 127, January 2003
a patient with an IgA-k monoclonal gammopathy. Arch Pathol Lab Med. 2000;
124:616–618.
19. Aoki YJ, Kameko M, Fujita K. A case of IgG4-lambda type monoclonal
immunoglobulin that interfered with determinations for albumin, direct bilirubin
and iron in serum [in Japanese]. Rinsho Byori. 2001;49:686–689.
20. Gulian JM, Dalmasso C, Pesquie M, Harle JR, Charrel M. Abnormal bili-
rubin binding to human serum albumin in a patient with unusual myeloma im-
munoglobulin G. Eur J Clin Chem Clin Biochem. 1993;31:115–119.
21. Stone WJ, McKinney TD, Warnock LG. Spurious hyperbilirubinemia in ure-
mic patients on propranolol therapy. Clin Chem. 1979;25:1761–1765.
22. Al-Damluji S, Meek JH. Interference of a propranolol metabolite with se-
rum bilirubin estimation in chronic renal failure. Br Med J. 1980;280:1414.
23. Belsey R, Mueggler P, Swanson JR. Propranolol and spurious hyperbiliru-
binemia. JAMA. 1984;251:38.
24. Multiple Myeloma Interference in Total Bilirubin. Customer bulletin 00-
117. Indianapolis, Ind: Roche Diagnostics; June 30, 2000.
Artifactual Hyperbilirubinemia—Pantanowitz et al 59