See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/233891934

Interference in Coagulation Testing: Focus on

Spurious Hemolysis, Icterus, and Lipemia

Article in Seminars in Thrombosis and Hemostasis · December 2012

Impact Factor: 3.88 · DOI: 10.1055/s-0032-1328972 · Source: PubMed

CITATIONS READS

28 1,254

3 authors:

Giuseppe Lippi Mario Plebani

University of Verona University of Padova
1,376 PUBLICATIONS 17,499 CITATIONS 1,373 PUBLICATIONS 21,475 CITATIONS

SEE PROFILE SEE PROFILE

Emmanuel Favaloro
Westmead Hospital
506 PUBLICATIONS 9,224 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Giuseppe Lippi
letting you access and read them immediately. Retrieved on: 26 May 2016
258

Interference in Coagulation Testing: Focus on

Spurious Hemolysis, Icterus, and Lipemia
Giuseppe Lippi, MD 1 Mario Plebani, MD 2 Emmanuel J. Favaloro, PhD, FFSc (RCPA) 3

1 Dipartimento di Patologia e Medicina di Laboratorio, U.O. Address for correspondence Giuseppe Lippi, MD, Dipartimento di
Diagnostica Ematochimica, Azienda Ospedaliero-Universitaria di Patologia e Medicina di Laboratorio, U.O. Diagnostica Ematochimica,
Parma, Parma, Italy Azienda Ospedaliero-Universitaria di Parma, Strada Abbeveratoia 2/a,
2 Dipartimento di Medicina di Laboratorio, Azienda Ospedaliera- 43100, Parma, Italy (e-mail: glippi@ao.pr.it; ulippi@tin.it).
Università di Padova, Padova, Italy
3 Department of Haematology, Institute of Clinical Pathology and
Medical Research (ICPMR), Westmead Hospital, Westmead, Australia

Semin Thromb Hemost 2013;39:258–266.

Abstract The chance that errors might jeopardize the quality of testing is inherently present
throughout the total testing process, especially in the preanalytical phase. In the
coagulation laboratory, as well as in other areas of diagnostic testing, spurious

Downloaded by: Universita di Parma. Copyrighted material.

hemolysis, icteria, and lipemia in test samples represent by far the leading diagnostic
challenges. Interference in hemostasis testing due to spurious hemolysis is attributed to
both analytical and biologic elements, namely high absorbance of cell-free hemoglobin
at wavelengths used by optical instrumentation and release of both cytoplasmatic and
plasma membrane molecules (e.g., tissue factor, proteases, phospholipids, and ADP)
that can spuriously activate blood coagulation and platelets. The interference attribut-
able to hyperbilirubinemia is mostly due to spectral overlap, whereas that of hyper-
triglyceridemia mainly reflects elements of light scatter and volume displacement as
well as direct interference of lipid particles with hemostasis. In practical terms, spurious
hemolysis reflects a more generalized process of endothelial and blood cell damage, so
that test results on spuriously hemolyzed specimens should be systematically sup-
pressed. The bias attributable to hyperbilirubinemia is less significant using modern
coagulometers equipped with dedicated wavelengths (i.e., with readings at 650 nm or
above), so that test results in samples with a bilirubin concentration up to 20 mg/dL can
still be analytically reliable. The interference observed in lipemic samples is most evident
Keywords with readings using wavelengths lower than 500 nm and can hence be prevented with
► interference readings at 650 nm or above, and/or using higher dilutions of the test sample, or can be
► coagulation testing abated in high hypertriglyceridemic specimens (i.e., > 1,000 mg/dL) using high speed
► hemolysis microcentrifugation or lipid extraction with organic solvents such as fluorine-chlorinat-
► lipemia ed hydrocarbon, or lipid-clearing agents such as LipoClear (StatSpin Inc., Norwood, MA)
► errors and n-hexane.

The total testing process develops through a broad series of Throughout this complex series of processes, the chance of
operations and activities, wherein the analytical phase is only errors being made that may ultimately jeopardize the quality
the central portion, being respectively preceded and followed of testing, patient safety, or both is inherently high, especially
by the preanalytical and postanalytic phases, but all of which in the presence of a poor degree of standardization and when
contribute to complete the so-called “brain-to-brain loop.”1,2 a hierarchical quality management system is not thoughtfully

published online Issue Theme Quality in Hemostasis and Copyright © 2013 by Thieme Medical DOI http://dx.doi.org/
December 10, 2012 Thrombosis, Part II; Guest Editors, Publishers, Inc., 333 Seventh Avenue, 10.1055/s-0032-1328972.
Giuseppe Lippi, MD, Mario Plebani, MD, New York, NY 10001, USA. ISSN 0094-6176.
and Emmanuel J. Favaloro, PhD, Tel: +1(212) 584-4662.
FFSc (RCPA).
 Interference in Coagulation Testing Lippi et al. 259

applied.3–5 Several lines of historical as well as contemporary 2.4

evidence support the hypothesis that processes comprised 2.2
Bilirubin
Hemoglobin
within the preanalytical phase are the most vulnerable to 2.0 Turbidity
errors and thus most easily flaw-prone, due to inappropriate
1.8
standardization, poor adherence to best practice, and invol-
1.6
untary human mistakes.6–9 In the coagulation laboratory,

Absorbance
1.4
spurious hemolysis is by far the leading source of interference
1.2
in plasma samples (40%), prevailing over pre-analysis clotting
(29%), inappropriate filling of primary blood collection tubes 1.0

(28%), contamination from infusion route (2%), lipemia, and 0.8

hyperbilirubinemia (both approximately 1%),10 whereas li- 0.6

pemia (13%) and hyperbilirubinemia (11%) both precede 0.4
spurious hemolysis (4%) as prevailing nonconformances in 0.2
whole blood specimens.11 It is, however, noteworthy that the
0.0
inherent complexity of hemostasis and the sophistication of 300 350 400 450 500 550 600 650 700

analytical methods and instrumentations make the coagula- Wavelength (nm)

tion laboratory more vulnerable to preanalytical errors and Fig. 1 Absorbance of potentially interfering substances (i.e., cell-free
interferences than other areas of in vitro diagnostics.12 This is hemoglobin, hyperbilirubinemia and lipemia [turbidity]).
also clearly recognized in the most recent H21-A5 standard of
the Clinical and Laboratory Standards Institute (CLSI),13

Downloaded by: Universita di Parma. Copyrighted material.

which advises that samples that are hemolyzed, icteric, coagulant activity and exposure of the subendothelial surface,
lipemic, or that contain substances that potentially interfere which would ultimately compromise the quality of the speci-
with light transmission represent a serious challenge for men (►Fig. 2). In any of these cases, test results should be
hemostasis testing. suppressed and not reported to the requesting clinician. This
would be true for both optical and mechanical detection
systems because the biologic effects are significant, even if
Spurious Hemolysis
the analytical effects on mechanical systems are negligible.
Hemolyzed specimens are the leading source of nonconfor- The chance of producing hemolyzed samples spans
mance in several areas of diagnostic testing,14,15 including throughout the preanalytical phase and thereby comprises
clinical chemistry,16 immunochemistry,17 complete blood sample collection,23 handling,24 and transport as well as
cell counting,18 arterial blood gas analysis,19 and—last but preparation or processing of samples for testing25
not least—platelet function20,21 and coagulation testing.12,22 (►Table 1). The problem may be largely amplified in organi-
Once the likelihood of in vivo hemolysis (i.e., hemolytic zational settings where preanalytical processing modules are
anemia) has been ruled out, these samples pose the most physically attached to analytical platforms, and these “hide”
serious problem for the laboratory due to the potential for specimens from visual inspection, from the point of receipt in
both analytical and biologic interference. The former is the laboratory until the analyses have been completed.26
mostly attributable to the strong absorbance of cell-free Technically speaking, hemolyzed plasma can be classified
hemoglobin still present in the sample after centrifugation according to the concentration of cell-free hemoglobin, that
at those wavelengths that are conventionally used by the is, nonhemolyzed (
0.05 g/L, yellow hue), slightly hemo-
(optical) instrumentation for coagulation testing (►Fig. 1). lyzed (from 0.05 to
0.3 g/L, yellow to slightly pink hue),
Thus, high baseline absorbance values are detected by the mildly hemolyzed (0.3 to
0.6 g/L, pink to slightly red hue),
instrument, and test results will either be unreliable or not frankly hemolyzed (
0.6 to 2.0 g/L, slightly red hue), and
reported. The latter interference is often overlooked but is grossly hemolyzed ( 2.0 g/L, red to brown hue).27 In daily
probably just as important because it affects both optical and practice, the vast majority of hemolyzed specimens—approx-
mechanical instrumentation and can be attributed to release imating 95%—are slightly hemolyzed (i.e., cell-free hemoglo-
of both cytoplasmatic and plasma membrane molecules (e.g., bin
0.3 mg/L)28 and are most prevalently referred from the
tissue factor, proteases, phospholipids, and adenosine di- emergency department and intensive care units.29 Hemolysis
phosphate [ADP], among others), which can produce spurious is most frequently identified by the laboratory personnel
activation of blood coagulation and platelets, generation of— through visual scrutiny, once the sample has been centrifuged
seldom insignificant—blood clots, which may further bias test and when cell-free hemoglobin in serum or plasma is greater
results for premature activation of blood coagulation (i.e., than 0.3 g/L, which reflects a lysis of 2.0 to 2.3% of red blood
shortening of clotting times) or consumption of clotting cells in anticoagulated blood samples with total hemoglobin
factors (i.e., prolongation of clotting times), and hyperactiva- concentration comprised between 13.0 and 15.0 g/L. The
tion of platelets. Moreover, the presence of cell-free hemo- serum indices (which also include the hemolysis index) are
globin frequently reflects problems related to the collection now widespread in preanalytical modules and clinical chem-
of blood specimens (e.g., a cumbersome venipuncture), so istry platforms,30 whereas they have only recently been
that it can be considered a reliable index of concomitant implemented in coagulation instrumentation (e.g., the so-
endothelium injury with release of thromboplastin-like pro- called “HIL-System” on CS-2100i and CS-5100, Sysmex

Seminars in Thrombosis & Hemostasis Vol. 39 No. 3/2013

260 Interference in Coagulation Testing Lippi et al.

Fig. 2 Leading mechanisms causing hemolysis-related interference in coagulation testing.

Downloaded by: Universita di Parma. Copyrighted material.

Company, Kobe, Japan).31,32 A more consolidated feature in Technical Considerations on Hemolysis Studies
modern coagulometers is, however, the possibility to use One of the leading problems in finding consistency across the
multi-wavelength detection (e.g., simultaneous analysis at different studies it the vast array of techniques used to test
340, 405, 575, 660, and 800 nm), which would enable selection hemolysis interference. Basically, these include spiking plas-
of the appropriate absorbance for specific types of interfering ma with hemolysate preparations, freezing and thawing
substances such as lipemia and bilirubin other than hemoglo- whole anticoagulated blood, lysis of whole anticoagulated
bin, thus allowing practitioners to process those samples with samples by means of deionized water with or without
a modest degree of photometrical interference. detergents, mechanical lysis of whole anticoagulated blood

Table 1 Leading causes of sample hemolysis

1. Sample collection
a. Poor venous access
b. Inappropriate technique
c. Traumatic venipuncture
d. Use of syringes rather than vacuum system
e. Use of needle of smaller size (i.e., lower than 23 gauge)
f. Underfilling of primary blood tubes
g. Prolonged tourniquet placing
h. Fist clenching
2. Sample handling
a. No mixing or excessive shaking of collection tubes
3. Sample transportation
a. Excessive time before centrifugation
b. Transport at extremes of temperature (too low or too high)
c. Trauma and mechanical injury to the specimens
4. Sample preparation
a. Centrifugation at inappropriate force (i.e., too high), time (i.e., too long) and/or temperature (i.e., too low or too high)
b. Generation of a poor barrier separation or re-spun sample
c. Transfer of plasma containing contaminating cell fractions, followed by freezing and thawing before testing

Seminars in Thrombosis & Hemostasis Vol. 39 No. 3/2013

 Interference in Coagulation Testing Lippi et al. 261

by sonication, stirring with a metallic bar, and application of both “normal” and “abnormal” plasma samples, but the latter
the blade of a tissue homogenizer and aspiration through a specimens should also be adequately heterogeneous (e.g.,
fine blood collection needle (i.e., lower than 25 gauge).33 representative of clotting factor deficiencies as well as col-
Although each of these approaches carries its own advantages lected from patients undergoing therapy with various anti-
and limitations, the most reliable techniques seem to be those platelet43 or anticoagulant drugs,44 especially different
that more closely reproduce mechanical hemolysis of blood formulations of heparins,45 vitamin K antagonists,46 or the
during collection, which is incidentally the leading source of novel oral anticoagulants).47–49 A final issue is the test
spurious hemolysis throughout the preanalytical phase. As methodology, wherein the different combination of reagents
such, spiking samples with hemolysate or pure hemoglobin and instrumentations makes comparison across different
solution is considered unsuitable because this method would studies very difficult, so that a local evaluation of bias may
exclude the potential biologic interferences attributable to be advisable.
the lysis of leukocytes and platelets, which may be injured
together with the erythrocytes.18 The method of freezing Influence of Hemolysis on Hemostasis Testing
whole anticoagulated blood appears more suited, but a highly Only a few studies have been published so far on the influence
standardized technique is necessary because temperature of spurious hemolysis on coagulation and hemostasis testing
and duration of freezing are critical, and will otherwise and, unfortunately, these are barely comparable due to the
produce a heterogeneous, osmotically induced injury to use of different techniques for producing the hemolysate,
blood cells.34 The preferable approach is therefore that based different patient samples, heterogeneous criteria to assess the
on mechanical injury to whole anticoagulated blood, which bias, as well as different instrumentation and reagents. Thus,
would technically mimic the physical damage to the plasma direct comparison is essentially unfeasible and potentially

Downloaded by: Universita di Parma. Copyrighted material.

membrane of all types of blood cells that may occur during misleading; however, some general points can be usefully
drawing and handling of the specimen. Among “injury tech- identified.
niques,” the one that most closely reproduces the breakdown As regards routine coagulation testing, in 1998, Roß and
of cells due to traumatic blood collection is that originally Paar assessed the interference due to hemolysis using the
proposed by Dimeski35 and further modified by Lippi et al36; MDA 180 analyzer (BioMérieux, Marcy l’Etoile, France),50 and
this entails serial aspiration of whole anticoagulated blood by observed that results of APTT were relatively unbiased by
a 0.5 mL insulin syringe equipped with a very thin needle spiked hemoglobin, whereas PT exhibited significant devia-
(e.g., 30 gauge or lower). It is also noteworthy that the number tions (shortening) at concentrations of hemoglobin > 30
of aspirations is directly linked to the degree of damaged µmol/L, thrombin time (shortening) at concentrations of
blood cells and cell-free hemoglobin, so that an arbitrary scale hemoglobin > 210 µmol/L, fibrinogen (decrease) at concen-
of hemolysis can be easily and suitably reproduced. trations of hemoglobin > 60 µmol/L, and antithrombin (de-
The identification of reliable thresholds of bias is the crease) at concentrations of hemoglobin > 150 µmol/L. Laga
second critical issue in hemolysis as well as in other interfer- et al investigated pairs of hemolyzed (cell-free hemoglobin
ence studies. Different approaches can be used also in this concentration from 0.2 to 7.0 mg/L) and subsequently recol-
circumstance. Regardless of appreciation of an analytical bias lected nonhemolyzed (cell-free hemoglobin concentration
in unsuitable samples, which may still be technically mean- < 0.3 mg/L) patient samples for PT, APTT, and some clotting
ingful but otherwise clinically negligible in many occasions, factor assays.51 Additional nonhemolyzed specimens from
the use of thresholds derived from reliable biologic evidence healthy volunteers were also subjected to mechanical hemo-
is advisable (e.g., the use of the critical difference of the lysis by means of a tissue homogenizer. In hemolyzed patient
quality specification criteria based on biologic variabili- plasma samples, both PT and APTT measured on an MDA-II
ty).37,38 Whenever the percentage of absolute bias exceeds (BioMérieux) were significantly shortened (  3.0 and  2.8%,
the boundaries of clinical significance, there is evidence that respectively). As regards the clotting factor assays, significant
test results should be systematically suppressed. differences were observed for activated factor VII ( þ 32.6%),
Another aspect to be considered is the appropriate selec- factor V ( þ 11.9%), factor X ( þ 2.6%), and for prothrombin
tion of patient samples for these studies. Although this issue fragment 1.2 ( þ 125%), but not for factor XIIa and factor VIII.
would apply to any type of interference study as well as any Paradoxically, different results were obtained when assessing
type of laboratory parameter, the appropriate selection of the mechanically hemolyzed plasma specimens, wherein no bias
reference samples is more critical in the hemostasis labora- was observed for PT, whereas a prolongation (rather than a
tory, wherein specimens with abnormal values (exceeding shortening) was observed for APTT in hemolyzed specimens
either the lower or upper limit of the reference range) may (from þ 1.0 to þ 8.6% in samples with 0.9 to 3.3% hemolysis,
display a rather different behavior to samples yielding normal i.e., from approximately 1.8 to 6.8 g/L cell-free hemoglobin).
values. This is specifically due to the nature of analyses such as Interestingly, no correlation was found between the degree of
the prothrombin time (PT),39 activated partial thromboplas- hemolysis and the relative bias (either shortening or pro-
tin time (APTT),40 D-dimer,41 and even platelet function longation) of both PT and APTT in patient specimens.
testing,42 which are considered “global tests” that thereby We have also prepared aliquots by serial dilutions of
reflect multiple abnormalities of clotting factors or platelets homologous samples collected from 10 different health vol-
rather than a punctiform and circumstantiate defect. As such, unteers spiked with hemolysate produced by a single freez-
this type of investigation should not only be performed on ing-thawing cycle, to obtain a final percentage of lysis ranging

Seminars in Thrombosis & Hemostasis Vol. 39 No. 3/2013

262 Interference in Coagulation Testing Lippi et al.

from 0 to 9.1%, which correspond to a range of cell-free samples were dramatically prolonged with both agonists. In
hemoglobin between 0 and 17 g/L.52 These aliquots were aliquots containing cell-free hemoglobin concentrations as
then tested for PT, APTT, fibrinogen on a Behring Coagulation high as 12.0 g/L, the results were flagged as “flow obstruction”
System (Dade-Behring, Marburg, Germany), and D-dimer on in 67% of cases for CADP and in all samples for CEPI, whereas
Mini Vidas (BioMérieux). Although analytically significant the closure times produced with CADP in the remaining
interference was recorded at cell-free hemoglobin concen- samples were also dramatically prolonged. These results
trations of 0.9 g/L for PT (prolongation), 1.7 g/L for APTT clearly attest that the mechanical injury of blood, which
(shortening) and fibrinogen (decrease), and 5.0 g/L for D- causes considerable release of ADP (mostly from erythro-
dimer (increase), clinically significant variations were ob- cytes), prothrombotic material (mostly from leukocytes), and
served at 1.7 g/L for PT and APTT (prolongation and shorten- cellular debris, along with a potential decrease of platelet
ing, respectively) and 3.4 g/L for fibrinogen (decrease). The count in the specimen, might in fact lead to a various degree of
value of D-dimer exhibited a bias exceeding 10% only at a cell- hyperactivation of platelets that may variously interfere with
free hemoglobin concentration of 13.6 g/L or higher. Likewise testing, up to generation of flow obstruction in the PFA-100
clinical chemistry testing,53 and in agreement with the data test cartridges. It is thereby essential, in the presence of
of Laga et al,51 the sample-specific bias was predominantly unexpectedly prolonged results and/or flow obstruction, to
unpredictable, with coefficient of variations as high as 62%, include sample hemolysis as a potential cause of spurious test
which virtually prevented the use of any equation to correct results. Indeed, should blood be left over following testing by
test results for hemolysis degree, thus validating that the the PFA-100 and identification of a flow obstruction, the
decision to suppress test results on hemolyzed specimens is sample could be centrifuged and the plasma assessed for
not empirical. This is in agreement with the current CLSI evident hemolysis.

Downloaded by: Universita di Parma. Copyrighted material.

standard, recommending that samples with visible interfer- The effect of spurious hemolysis on canine kaolin-activat-
ing substances (including hemolysis) should not be proc- ed thromboelastography has also been recently investigated
essed for clotting assays due to potential activation of the by Bauer et al.56 Blood cells collected from healthy dogs were
coagulation cascade and interference with end point tested before and after freezing or mechanical injury by
measurements.13 repetitions of blood expulsion through a 23-gauge needle.
More recently, Tantanate et al also spiked hemolytic hemo- As compared with nonhemolyzed specimens, mechanical
globin to residual sodium-citrated patient plasma with normal injury caused a > 50% decrease of reaction time, whereas
PT, APTT, and fibrinogen values on Sysmex CS-2100i54 and freezing remarkably increased coagulation time by nearly
showed that all tests exhibited a clear trend toward increased three times and decreased the angle by more than 50%. The
values in hemoglobin-spiked specimens. In a subsequent maximum amplitude was also decreased by approximately
study, we assessed the influence of mechanical hemolysis of 20 and 47% in mechanically hemolyzed and frozen samples,
anticoagulated blood on two different D-dimer immunoassays, respectively. Finally, both types of hemolysis decreased mean
the former based on an immunoturbidimetric technique (He- platelet component concentration by about 20 and 26% in
mosIL D-dimer HS for ACL TOP, Instrumentation Laboratory, mechanically hemolyzed and frozen samples, respectively.
Bedford, MA) and the latter on a chemiluminescent immuno- These results are hence suggestive for activation of both
assay (HemosIL AcuStar D-dimer, Instrumentation Laborato- primary and secondary hemostasis in spuriously hemolyzed
ry).55 It is noteworthy that a rather similar trend toward samples, whereas the presence of hyporeactive platelets
reduced values was observed with both methods on four could not be excluded as attested by the reduced clot firm-
different plasma pools, thus suggesting that the interference ness. No other studies have addressed the issue of hemolysis
from hemolysis seems more biologic than analytical. In agree- interference in hemostasis testing to the best of our
ment with the previous investigation, the overall bias observed knowledge.
in samples with gross hemolysis (i.e., concentration of cell-free
hemoglobin > 11.5 g/L) was modest, always lower than 10%,
Hyperbilirubinemia
thus indicating that test results obtained in the vast majority of
hemolyzed specimens might still be reliable for this parameter Serum bilirubin, the final result of the catabolism of heme, is a
and thereby could be safely released to the clinicians. highly hydrophobic compound transported in blood linked
Platelet function tests may be also negatively biased for the with albumin, with < 0.01% circulating in unbound, free
presence of hemolysis and/or generalized cell injury involv- form. The main site of bilirubin metabolism is the liver, where
ing erythrocytes, platelets, and leukocytes. We have recently the molecule is dissociated from albumin and then conjugat-
assessed whole blood samples collected from healthy volun- ed with one or two molecules of glucuronic acid to form
teers on Platelet Function Analyzer-100 (PFA-100, Siemens bilirubin monoglucuronide and diglucuronide (i.e., conjugat-
Healthcare Diagnostics, Tarrytown, NY) before and after ed or direct bilirubin). Hyperbilirubinemia is commonly
mechanical hemolysis by aspiration through a very fine defined as the presence of excess bilirubin in the blood (i.e.,
needle.21 When compared with the nonhemolyzed aliquot, bilirubin concentration > 1.5 mg/dL), and can be classified as
44% of results were flagged as “flow obstruction” for both conjugated or unconjugated, according to the predominant
collagen and ADP (CADP) or collagen and epinephrine (CEPI) form.57 The interference in hemostasis testing due to the
in samples containing approximately 6.0 g/L of cell-free presence of hyperbilirubinemia is mostly attributable to
hemoglobin, whereas the closure times of the remaining spectral overlap (i.e., the compound has a high absorbance

Seminars in Thrombosis & Hemostasis Vol. 39 No. 3/2013

 Interference in Coagulation Testing Lippi et al. 263

between 400 and 520 nm, with absorbance peak at roughly varied. Basically, turbidity is typically caused by the presence
456 nm; ►Fig. 1), rather than for its biologic properties (e.g., of an excess of triglycerides—usually above 500 mg/dL (i.e.,
bilirubin can be rapidly oxidized to biliverdin and bilipur- 5.65 mmol/L)—which can be attributed to physiologic causes
purin and then reacts with all oxidants present in the (e.g., postprandial metabolism), para-physiologic causes (e.g.,
sample).58 As such, once the photometric interference has administration of intravenous lipids), as well as other meta-
been eliminated, no other sources of bias should be expected bolic disorders such as diabetes, chronic alcohol abuse, im-
in clotting assays. paired renal function, thyroid diseases, acute pancreatitis,
myeloma, primary biliary cirrhosis, systemic lupus erythe-
Influence of Hyperbilirubinemia on Hemostasis Testing matosus, and medication such as protease inhibitors, estro-
Although several studies have assessed the interference of gen, and steroids.64,65
hyperbilirubinemia on clinical chemistry parameters, with The use of “nontransparent turbid milky samples” (more
most of these recently reviewed by Dimeski,59 there is less simply “turbid samples”) for laboratory testing carries some
abundant information on this type of interference on routine technical, analytical and clinical problems. The interference
and specialized coagulation testing. The first study dealing mechanism seem basically attributable to light scatter, reflec-
with this topic was published in 1998 by Roß and Paar, using an tance, or absorption, thus causing measurement bias in pho-
optical test system, the MDA 180 analyzer.50 After spiking tometric methods, especially at shorter wavelengths (i.e.,
plasma samples with unconjugated bilirubin, the authors Rayleigh law; ►Fig. 1);58 volume displacement by lipids,
concluded that APTT results were virtually unaffected, where- which is particularly evident with using indirect ion selective
as PT (shortening) and thrombin time (prolongation) exhibited electrodes66; and direct interference of lipid particles with
significant deviations at concentrations of bilirubin > 14.6 mg/ hemostasis. Several modern coagulometers now use optical

Downloaded by: Universita di Parma. Copyrighted material.

dL (250 µmol/L); moreover, effects were also observed with detection of the clot along with optical assays for measuring
fibrinogen (increase) at concentrations of bilirubin > 21.9 mg/ the absorbance in the samples. It is thereby predictable that no
dL (375 µmol/L) and antithrombin (decrease) at concentrations reliable baseline readings will be obtained on highly lipemic
of bilirubin > 1.5 mg/dL (25 µmol/L). Quehenberger et al specimens, in which case test results will either be unreliable
compared results of icteric samples (bilirubin concentration or not reported by the instrument; in either case, test results
between 46 and 925 mmol/L, i.e., between 2.7 and 54.1 mg/dL) should be suppressed and not reported to the requesting
obtained on the optical instrument CA-6000 with those of an clinician. Understandably, analyzers using mechanical or elec-
STA analyzer (Diagnostica Stago, Cedex, France), which is tromechanical detection are much less influenced by this type
instead based on mechanical detection of the clot.60 At vari- of analytical interference.
ance with the results earlier obtained by Roß and Paar, the
correlations between the instruments were satisfactory, al- Influence of Lipemia on Hemostasis Testing
ways higher than 0.97 except for thrombin time (0.641). Roß and Paar first assessed the bias in turbid samples using
Another research group tested icteric plasma samples with the MDA 180 analyzer,50 observing that results of PT and
the 671 nm wavelength on ACL TOP (Instrumentation Labora- APTT were relatively unbiased by very high concentrations of
tory)61 and Sysmex CA-700062 and found that test results of all triglycerides ( > 965 mg/dL, i.e., 10.9 mmol/L), whereas
coagulation, chromogenic, and immunologic assays obtained thrombin time exhibited significant bias (prolongation) at
on these photo-optical clot detection–based instruments were concentrations of bilirubin > 965 mg/dL (10.9 mmol/L), fi-
highly comparable to those obtained using a mechanical clot brinogen (reduction), and antithrombin (increase) at concen-
detection–based analyzer (i.e., STA, Diagnostica Stago). Nearly trations of lipids > 487 mg/dL (5.5 mmol/L). In a subsequent
identical results were reported by Dorn-Beineke et al,63 who study, Arambarri et al simulated the turbidity due to lipemia
tested icteric samples (i.e., concentrations of total bilirubin by addition of increasing quantities of 20% intralipid to
from 7.4 to 41.5 mg/dL [127 to 710 µmol/L]) on Sysmex CA- patient plasma samples with normal or pathologic coagula-
6000 and Sysmex CA-7000 and failed to find any significant tion tests and assessed PT and APTT on an ACL-3000 analyzer
bias. Analogously, Tantanate et al spiked exogenous bilirubin to (Instrumentation Laboratory).67 In those specimens with
patient plasmas with normal PT, APTT, and fibrinogen values54 hypertriglyceridemia (i.e., triglycerides > 9.79 mmol/L, 866
and observed no interference using the 660 nm wavelength on mg/dL), no test data could be obtained. Nevertheless, test
Sysmex CS-2100i up to a final concentration of 14.07 mg/dL results were instead produced after treatment of plasma with
(241 µmol/L) of free bilirubin, and 14.55 mg/dL (249 µmol/L) of n-hexane, on average 6% lower for PT and 12% higher for
conjugated bilirubin. According to the data of these recent APTT, respectively, compared with control. Quehenberger et
studies, it therefore seems reasonable to conclude that hyper- al compared results of lipemic samples (triglyceride concen-
bilirubinemia does not represent a substantial analytical prob- tration ranging between 2.42 and 18.70 mmol/L, i.e., 214 to
lem for the modern coagulometers, provided that a dedicated 1,655 mg/dL) obtained on the optical instrument Sysmex CA-
wavelength is used for analysis of visually icteric samples. 6000 with those of a mechanical clot detection-based analyz-
er, the STA.60 Final correlations of 0.907, 0.988, 0.288, 0.961,
and 0.902 were reported for PT, APTT, thrombin time, fibrin-
Lipemia
ogen, and antithrombin, respectively, indicating a major
The potential origin of “nontransparent turbid milky sam- problem only for thrombin time testing. Dorn-Beineke et al
ples” (more simply “turbid samples” or “lipemic samples”) is assessed lipemic samples (i.e., triglyceride concentrations

Seminars in Thrombosis & Hemostasis Vol. 39 No. 3/2013

264 Interference in Coagulation Testing Lippi et al.

between 3.51 mmol/L and 18.5 mmol/L) on Sysmex CA-6000 ples with a bilirubin concentration lower than 20 mg/dL (342
and Sysmex CA-7000,63 describing a negative bias as well as µmol/L).
invalid results for PT in samples containing triglycerides The interference observed in lipemic samples is of both
> 13.5 mmol/L (1,195 mg/dL), and a positive bias as well as optical and biologic origin and develops from a different source
invalid results on Clauss fibrinogen in samples with trigly- but produces a similar negative influence on testing as that
cerides exceeding 3.51 mmol/L (317 mg/dL). The results of previously described for spurious hemolysis. Due to the most
APTT were substantially unbiased up to a final triglyceride evident interference with readings using wavelengths lower
concentration of 18.5 mmo/L (1,637 mg/dL). Bai et al further than 500 nm (►Fig. 2), the optical bias can, however, be easily
compared optical and mechanical clot detection instruments prevented using dedicated wavelengths (i.e., > 650 nm) and/
for routine coagulation testing in a large volume clinical or using higher dilutions of the sample. Naturally, sample
laboratory (i.e., Sysmex CA-1500 with readings at 660nm dilutions can only be applied to assays that already use
vs. Diagnostica Stago STA), and found a high correlation for PT, dilutions (e.g., fibrinogen, factor assays, etc) and not to global
APTT, and fibrinogen (all greater than 0.98) when assessing assays (e.g., PT, APTT). Because the presence of lipid particles
turbid samples.68 No clot was observed for APTT in only 2.4% can still bias the measurement for biologic interference, the
of turbid samples. In the study of Tantanate et al, who spiked CLSI currently recommends the removal of excess triglycerides
20% intralipid to patient plasmas with normal PT, APTT, and by ultracentrifugation for at least 30 minutes at a speed above
fibrinogen values,54 consistent trends toward increased val- 40,000  g.70 Nevertheless, this approach is time consuming,
ues of PT as well as toward reduction of APTT and fibrinogen requires dedicated instrumentation (i.e., the ultracentrifuge),
were observed with the Sysmex CS-2100i analyzer. In a recent and hence is incompatible with the routine daily practice of
publication, we have also assessed a vast array of coagulation most clinical laboratories. Moreover, ultracentrifugation may

Downloaded by: Universita di Parma. Copyrighted material.

tests including PT, APTT, fibrinogen, protein C, protein S, and cause precipitation of the large molecular proteins such as
antithrombin on ACL TOP in 17 healthy volunteers eating a fibrinogen or factor VIII/von Willebrand factor complex. An
standardized meal that contained carbohydrates, protein, and alternative approach entails high-speed microcentrifugation
lipids for a total caloric intake of 563 Kcal.69 Samples were (e.g., double centrifugation at > 20,000  g for 15 minutes) or
drawn before the meal, immediately after, and in the follow- lipid extraction by means of organic solvents such as fluorine
ing 1, 2, and 4 hours. The only clinically significant bias was chlorinated hydrocarbon or lipid-clearing agents such as
recorded for APTT, 2 hours after the ingestion of the meal. LipoClear (StatSpin Inc., Norwood, MA) and n-hexane.71

Conclusions
References
Regardless of analytical considerations, which, although im-
1 Plebani M, Lippi G. Closing the brain-to-brain loop in laboratory
portant, may not necessarily influence clinical practice and
testing. Clin Chem Lab Med 2011;49(7):1131–1133
patient safety, the type of interference encountered with 2 Plebani M, Favaloro EJ, Lippi G. Patient safety and quality in
spurious hemolysis, icterus, and lipemia is substantially laboratory and hemostasis testing: a renewed loop? Semin
different, and thus requires distinctive approaches. Thromb Hemost 2012;38(6):553–558
Spurious hemolysis typically reflects a more generalized 3 Lippi G. Governance of preanalytical variability: travelling the
right path to the bright side of the moon? Clin Chim Acta
process of endothelial and blood cell damage, so that any
2009;404(1):32–36
consideration about the potential optical interference at- 4 Lippi G, Plebani M, Simundic AM. Quality in laboratory diagnos-
tributable to the absorbance of cell-free hemoglobin is tics: from theory to practice. Biochem Med (Zagreb) 2010;20
ancillary to the presence of a critical biologic bias caused (2):126–130
by release of prothrombotic and proaggregant material 5 Lippi G, Simundic AM, Mattiuzzi C. Overview on patient safety in
from the injured cells. Test results of both routine and healthcare and laboratory diagnostics. Biochem Med (Zagreb)
2010;20(2):131–143
specialized hemostasis testing should therefore be system-
6 Lippi G, Guidi GC, Mattiuzzi C, Plebani M. Preanalytical variability:
atically suppressed in all samples with visible degree of the dark side of the moon in laboratory testing. Clin Chem Lab Med
hemolysis (i.e., cell-free hemoglobin > 0.3 to 0.6 g/L). It may 2006;44(4):358–365
also be advisable to routinely assess the hemolysis index in 7 Lippi G, Chance JJ, Church S, et al. Preanalytical quality improve-
these samples, clearly establish the degree of potential ment: from dream to reality. Clin Chem Lab Med 2011;49(7):
1113–1126
interference, and decide the most suitable actions. Natural-
8 Lippi G, Simundic AM. Total quality in laboratory diagnostics. It’s
ly, test results from patients with in vivo hemolysis should
time to think outside the box. Biochem Med (Zagreb) 2010;
instead be reported to avoid repeated collections with 20(1):5–8
similar outcome, albeit with a caveat detailing the limita- 9 Simundic AM, Lippi G. Preanalytical phase—a continuous chal-
tion of the test result. lenge for laboratory professionals. Biochem Med (Zagreb) 2012;22
The bias attributable to hyperbilirubinemia is almost (2):145–149
10 Salvagno GL, Lippi G, Bassi A, Poli G, Guidi GC. Prevalence and type
exclusively analytical, due to spectral overlap. There is prob-
of pre-analytical problems for inpatients samples in coagulation
ably insignificant clinical bias using modern coagulometer laboratory. J Eval Clin Pract 2008;14(2):351–353
equipped with dedicated wavelengths (i.e., with readings at 11 Salvagno GL, Lippi G, Gelati M, Guidi GC. Hemolysis, lipaemia and
650 nm or above) so that test results may be reliable and icterus in specimens for arterial blood gas analysis. Clin Biochem
hence released to requesting clinicians, particularly in sam- 2012;45(4-5):372–373

Seminars in Thrombosis & Hemostasis Vol. 39 No. 3/2013

 Interference in Coagulation Testing Lippi et al. 265

12 Favaloro EJ, Adcock Funk DM, Lippi G. Pre-analytical variables in 33 Lippi G. Interference studies: focus on blood cell lysates prepara-
coagulation testing associated with diagnostic errors in hemosta- tion and testing. Clin Lab 2012;58(3-4):351–355
sis. Lab Medicine 2012;43(2):1–10 34 Lippi G, Musa R, Aloe R, Mercadanti M, Pipitone S. Influence of
13 Adcock DM, Hoefner DM, Kottke-Marchant K, Marlar RA, Szamosi temperature and period of freezing on the generation of hemolysate
DI, Warunek DJ. Collection, Transport, and Processing of Blood and blood cell lysate. Clin Biochem 2011;44(14–15):1267–1269
Specimens for Testing Plasma-Based Coagulation Assays and 35 Dimeski G. Effects of hemolysis on the Roche ammonia method for
Molecular Hemostasis Assays; Approved Guideline—Fifth Edition. Hitachi analyzers. Clin Chem 2004;50(5):976–977
Wayne, PA: Clinical Laboratory Standards Institute; 2008. CLSI 36 Lippi G, Mercadanti M, Aloe R, Targher G. Erythrocyte mechanical
document H21–A5 fragility is increased in patients with type 2 diabetes. Eur J Intern
14 Lippi G, Blanckaert N, Bonini P, et al. Haemolysis: an overview of Med 2012;23(2):150–153
the leading cause of unsuitable specimens in clinical laboratories. 37 Lippi G, Banfi G, Buttarello M, et al. Recommendations for detec-
Clin Chem Lab Med 2008;46(6):764–772 tion and management of unsuitable samples in clinical laborato-
15 Simundic AM, Topic E, Nikolac N, Lippi G. Hemolysis detection and ries. Clin Chem Lab Med 2007;45(6):728–736
management of hemolysed specimens. Biochem Med (Zagreb) 38 Fraser CG, Lippi G, Plebani M. Reference change values may need
2010;20(2):154–159 some improvement but are invaluable tools in laboratory medi-
16 Lippi G, Salvagno GL, Montagnana M, Brocco G, Guidi GC. Influence cine. Clin Chem Lab Med 2012;50(5):963–964
of hemolysis on routine clinical chemistry testing. Clin Chem Lab 39 Favaloro EJ, Adcock DM. Standardization of the INR: how good is
Med 2006;44(3):311–316 your laboratory’s INR and can it be improved? Semin Thromb
17 Lippi G, Avanzini P, Dipalo M, Aloe R, Cervellin G. Influence of Hemost 2008;34(7):593–603
hemolysis on troponin testing: studies on Beckman Coulter UniCel 40 Lippi G, Favaloro EJ. Activated partial thromboplastin time: new
Dxl 800 Accu-TnI and overview of the literature. Clin Chem Lab tricks for an old dogma. Semin Thromb Hemost 2008;34(7):
Med 2011;49(12):2097–2100 604–611
18 Lippi G, Musa R, Avanzini P, Aloe R, Pipitone S, Sandei F. Influence of 41 Lippi G, Cervellin G, Franchini M, Favaloro EJ. Biochemical markers
in vitro hemolysis on hematological testing on Advia 2120. Int J Lab for the diagnosis of venous thromboembolism: the past, present

Downloaded by: Universita di Parma. Copyrighted material.

Hematol 2012;34(2):179–184 and future. J Thromb Thrombolysis 2010;30(4):459–471
19 Lippi G, Ippolito L, Fontana R. Prevalence of hemolytic specimens 42 Favaloro EJ, Lippi G, Franchini M. Contemporary platelet function
referred for arterial blood gas analysis. Clin Chem Lab Med testing. Clin Chem Lab Med 2010;48(5):579–598
2011;49(5):931–932 43 Linden MD, Tran H, Woods R, Tonkin A. High platelet reactivity and
20 Favaloro EJ. More on preanalytical variables affecting platelet antiplatelet therapy resistance. Semin Thromb Hemost 2012;38
function testing using light transmittance aggregometry. Clin (2):200–212
Chem Lab Med 2011;49(4):737–739 44 Tripodi A. Problems and solutions for testing hemostasis assays
21 Lippi G, Fontana R, Avanzini P, et al. Influence of mechanical while patients are on anticoagulants. Semin Thromb Hemost
trauma of blood and hemolysis on PFA-100 testing. Blood Coagul 2012;38(6):586–592
Fibrinolysis 2012;23(1):82–86 45 Walenga JM, Jackson CM, Kessler CM. Low molecular weight
22 Bonar R, Favaloro EJ, Adcock DM. Quality in coagulation heparins differ substantially: impact on developing biosimilar
and haemostasis testing. Biochem Med (Zagreb) 2010;20(2): drugs. Semin Thromb Hemost 2011;37(3):322–327
184–199 46 Lippi G, Franchini M, Favaloro EJ. Pharmacogenetics of vitamin K
23 Lippi G, Salvagno GL, Montagnana M, Lima-Oliveira G, Guidi GC, antagonists: useful or hype? Clin Chem Lab Med 2009;47(5):
Favaloro EJ. Quality standards for sample collection in coagulation 503–515
testing. Semin Thromb Hemost 2012;38(6):565–575 47 Favaloro EJ, Lippi G. Laboratory testing and/or monitoring of the
24 Lippi G, Plebani M. Primary blood tubes mixing: time for updated new oral anticoagulants/antithrombotics: for and against? Clin
recommendations. Clin Chem Lab Med 2012;50(4):599–600 Chem Lab Med 2011;49(5):755–757
25 Adcock Funk DM, Lippi G, Favaloro EJ. Quality standards for sample 48 Favaloro EJ, Lippi G, Koutts J. Laboratory testing of anticoagulants:
processing, transportation, and storage in hemostasis testing. the present and the future. Pathology 2011;43(7):682–692
Semin Thromb Hemost 2012;38(6):576–585 49 Favaloro EJ, Lippi G. The new oral anticoagulants and the future of
26 Lippi G, Plebani M. Continuous-flow automation and hemolysis haemostasis laboratory testing. Biochem Med (Zagreb) 2012;22
index: a crucial combination. J Lab Autom 2012 (e-pub ahead of (3):329–341
print). 10.1177/2211068212450014 50 Roß RS, Paar D. Analytisch und klinisch relevante Interferenzen in
27 Lippi G, Plebani M, Di Somma S, Cervellin G. Hemolyzed speci- der Gerinnungsanalytik am Beispiel des MDA 180. Laboratoriums-
mens: a major challenge for emergency departments and clinical medizin 1998;22:90–94
laboratories. Crit Rev Clin Lab Sci 2011;48(3):143–153 51 Laga AC, Cheves TA, Sweeney JD. The effect of specimen hemolysis
28 Carraro P, Servidio G, Plebani M. Hemolyzed specimens: a reason on coagulation test results. Am J Clin Pathol 2006;126(5):748–755
for rejection or a clinical challenge? Clin Chem 2000;46(2): 52 Lippi G, Montagnana M, Salvagno GL, Guidi GC. Interference of
306–307 blood cell lysis on routine coagulation testing. Arch Pathol Lab Med
29 Lippi G, Salvagno GL, Favaloro EJ, Guidi GC. Survey on the preva- 2006;130(2):181–184
lence of hemolytic specimens in an academic hospital according to 53 Lippi G, Avanzini P, Pavesi F, et al. Studies on in vitro hemolysis and
collection facility: opportunities for quality improvement. Clin utility of corrective formulas for reporting results on hemolyzed
Chem Lab Med 2009;47(5):616–618 specimens. Biochem Med (Zagreb) 2011;21(3):297–305
30 Plebani M, Lippi G. Hemolysis index: quality indicator or criterion 54 Tantanate C, Teyateeti M, Tientadakul P. Influence of plasma
for sample rejection? Clin Chem Lab Med 2009;47(8):899–902 interferences on screening coagulogram and performance evalua-
31 Martin-Toutain I, Jobic L, Mancic T, Ankri A. [Evaluation of the tion of the automated coagulation analyzer Sysmex® CS-2100i.
automated coagulation analyzer Sysmex® CS-2100i (Siemens)]. Siriraj Medical Journal 2011;63(5):151–156
Ann Biol Clin (Paris) 2011;69(6):699–704 55 Lippi G, Avanzini P, Zobbi V, Ippolito L. Influence of mechanical
32 Molenaar PJ, Leyte A. Pre-acquisition system assessment of the hemolysis of blood on two D-dimer immunoassays. Blood Coagul
Sysmex® Coagulation System CS-2100i and comparison with end- Fibrinolysis 2012;23(5):461–463
user verification; a model for the regional introduction of new 56 Bauer NB, Eralp O, Moritz A. Effect of hemolysis on canine kaolin-
analysers and methods. Clin Chem Lab Med 2011;49(9):1479– activated thromboelastography values and ADVIA 2120 platelet
1489 activation indices. Vet Clin Pathol 2010;39(2):180–189

Seminars in Thrombosis & Hemostasis Vol. 39 No. 3/2013

266 Interference in Coagulation Testing Lippi et al.

57 Franchini M, Targher G, Lippi G. Serum bilirubin levels and with ultracentrifugation. Biochem Med (Zagreb) 2011;21(1):
cardiovascular disease risk: a Janus Bifrons? Adv Clin Chem 86–92
2010;50:47–63 66 Lippi G, Aloe R. Hyponatremia and pseudohyponatremia: first, do
58 Grafmeyer D, Bondon M, Manchon M, Levillain P. The influence of no harm. Am J Med 2010;123(9):e17
bilirubin, haemolysis and turbidity on 20 analytical tests per- 67 Arambarri M, Oriol A, Sancho JM, Roncalés FJ, Galán A, Galimany R.
formed on automatic analysers. Results of an interlaboratory [Interference in blood coagulation tests on lipemic plasma.
study. Eur J Clin Chem Clin Biochem 1995;33(1):31–52 Correction using n-hexane clearing]. Sangre (Barc) 1998;43(1):
59 Dimeski G. Interference testing. Clin Biochem Rev 2008; 13–19
29(Suppl 1):S43–S48 68 Bai B, Christie DJ, Gorman RT, Wu JR. Comparison of optical and
60 Quehenberger P, Kapiotis S, Handler S, Ruzicka K, Speiser W. mechanical clot detection for routine coagulation testing in a large
Evaluation of the automated coagulation analyzer SYSMEX CA volume clinical laboratory. Blood Coagul Fibrinolysis 2008;19
6000. Thromb Res 1999;96(1):65–71 (6):569–576
61 Appert-Flory A, Fischer F, Jambou D, Toulon P. Evaluation and 69 Lippi G, Salvagno GL, Lima-Oliveira G, et al. Interference of a
performance characteristics of the automated coagulation analyz- regular standardized meal on routine coagulation testing. Poster
er ACL TOP. Thromb Res 2007;120(5):733–743 presented at: Second Annual Joint Meeting SIBioC –SIMeL; Octo-
62 Fischer F, Appert-Flory A, Jambou D, Toulon P. Evaluation of the ber 27–30, 2009; Naples, Italy
automated coagulation analyzer Sysmex CA-7000. Thromb Res 70 McEnroe RJ, Burritt MF, Powers DM, Rheinheimer DW. Interfer-
2006;117(6):721–729 ence Testing in Clinical Chemistry; Approved Guideline—Second
63 Dorn-Beineke A, Dempfle CE, Bertsch T, Wisser H. Evaluation of the Edition. Wayne, PA: Clinical Laboratory Standards Institute; 2005.
automated coagulation analyzer Sysmex CA-7000. Thromb Res CLSI document EP7–A2
2005;116(2):171–179 71 Vermeer HJ, Steen G, Naus AJ, Goevaerts B, Agricola PT, Schoen-
64 Kroll MH. Evaluating interference caused by lipemia. Clin Chem makers CH. Correction of patient results for Beckman Coulter LX-
2004;50(11):1968–1969 20 assays affected by interference due to hemoglobin, bilirubin
Dimeski G, Jones BW. Lipaemic samples: effective process for or lipids: a practical approach. Clin Chem Lab Med 2007;45(1):

Downloaded by: Universita di Parma. Copyrighted material.

65
lipid reduction using high speed centrifugation compared 114–119

Seminars in Thrombosis & Hemostasis Vol. 39 No. 3/2013