Keywords
coagulation testing - interferences - case-based learning
Schlüsselwörter
Gerinnungstests - Interferenzen - fallbasiertes Lernen
Introduction
Assays in the coagulation laboratory are affected by numerous variables. When they
have impact on the test result, incorrect results may be reported to the physician,
leading (or, not leading) to consequences for further diagnostic and therapeutic decisions.
In this review, I would like to propose a classification of these interferences into
three main groups: (1) biological interferences—interferences due to an actual impairment of the patient's coagulation
system (congenital or acquired); (2) physical interferences—interferences due to pre-analytical errors; and (3) chemical interferences—interferences due to the presence of drugs (mainly anticoagulants)
in the test tube.
Interferences are often more easily recognized by reviewing coagulation tests in concert
with the clinical information. On the one hand, however, the laboratory often does
not have the clinical information required (or, even worse, no clinical information
at all). On the other hand, the patient's physicians may be unaware of the relationship
between a confounding variable and its interference with a specific laboratory test.
It is therefore recommended that appropriate quality systems covering the whole process
of a diagnostic test in coagulation—from taking the patient's blood to reporting the
results—should be established to minimize the effect of such influences. Educating
medical staff involved in this process, in the end, should be a cornerstone in any
quality system. Case-based discussions of (near) miss events are one approach to generate
more attention to these issues. Seven cases of (near) miss events are reported in
this review, and the reader is encouraged to share some or all of them with the medical
staff in his institution.
Biological Interferences
A specific coagulation assay or a group of assays that have been requested from the
laboratory may be affected by an actual impairment of the patient's coagulation system.
This includes congenital impairments (such as a mutation in a relevant gene encoding
a clotting factor) and acquired impairments (such as those mediated by inhibitors
directed against clotting factors). Often, it is difficult to identify a specific
interaction that has impact on a test result. The fewer tests requested, the more
difficult it is to detect individual incorrect results. The more different the requested
analyses are methodologically (e.g., based on prothrombin time [PT] and activated
partial thromboplastin time [APTT], antigen and clotting assays from the same sample),
the more likely it is to find inconclusive outliers before a report leaves the laboratory
as medically validated.
Interference of the FV R506Q (Leiden) Mutation
Observation
An external laboratory performed a thrombophilia screen and the patient was referred
to us with two distinct diagnoses: a homozygous factor V R506Q (Leiden) mutation and
protein S deficiency. We confirmed activated protein C (APC) resistance using a functional
assay but found free protein S antigen (and later total protein S antigen) within
the normal range.
Pitfall
Does this patient have protein S deficiency?[1]
[2] Protein S is a cofactor of APC and together, they are responsible for inactivating
the cofactors FVa and FVIIIa. A protein S activity assay measures the ability of protein
S (in the patient's sample) to prolong a one-stage clotting time (PT or APTT) in the
presence of a standardized amount of APC. The more protein S available in the sample,
the longer the APTT. If the patient's FV is mutated and carries the Leiden variant,
it acquires APC resistance. APC resistance shortens the APTT. Accordingly, the protein
S activity will be underestimated because in the presence of the same amount of protein
S, the APTT is shorter with APC resistance than with no APC resistance.
Remember that the same biological principle is used to identify APC resistance[3]: in an APTT assay, the total clotting time is measured in the patient's plasma in
the absence and in the presence of (externally added, high amounts of) APC, and a
ratio “APTT with APC added/APTT without APC added” is reported. The ratio is usually
<2 if the patient has APC resistance: his APTT is not adequately prolonged because
his FV is not sufficiently inhibited. Free protein S antigen is the assay recommended
by the ISTH when screening for protein S deficiency.[2]
Interference of Antiphospholipid Antibodies
Observation
A preoperative screen in a 6-year-old boy prior to tonsillectomy revealed normal test
results for PT and fibrinogen, but prolonged APTT (92 seconds). Subsequent analysis
of single factors revealed normal FXII (82%) and FXI (72%), but severely low FVIII
(<1%) and FIX (<1%) activity. The pathologist contacted the clinical team. The child
was apparently healthy, did not have any signs or symptoms of bleeding, and his family
history was unremarkable. The procedure was postponed for additional diagnostic workup.
Pitfall
A combined diagnosis of severe hemophilia A and severe hemophilia B in an apparently
healthy 6-year-old boy seemed unlikely. A plasma mixing study was performed the same
day and failed to normalize the APTT, indicating the presence of an inhibitor. Further
analysis in a Nijmegen-Bethesda assay gave 3 BU anti-FVIII/mL and 5 BU anti-FIX/mL.
Again, development of acquired hemophilia A and B in an apparently healthy child seemed
unlikely. The presence of a lupus anticoagulant (LA) was suspected to have caused
the alterations. An integrated screen and confirm assay employing normal and elevated
levels of phospholipids in a diluted Russell Viper venom time (DRVVT) revealed a ratio
of 1.6, confirming the presence of a (weak) LA in the sample. Immunological assays
for antibodies of the IgG and IgM type against beta-2-glycoprotein I and cardiolipin,
as recommended by the ISTH, were negative. With these results in mind, the first question
to address is whether this excludes the presence of inhibitors. In rare case, LA and
a specific inhibitor against a clotting factor may be present together.[4] Chromogenic factor assays appear to not suffer from LA interference to the same
extent as clotting assays, most likely because of the high dilution of the patient's
plasma.[5] Indeed, chromogenic FVIII and F IX were near-normal in this boy (66 and 70%, respectively).
Another way to exclude the presence of inhibitors in the presence of LA would be antigenic
enzyme-linked immunosorbent assays (unavailable in our laboratory). A preliminary
diagnosis of “transient LA in childhood” was made and the boy was cleared for surgery,
which was uneventful without any unexpected bleeding. Recovery was quick and complete.
The second question to address is whether the detection of LA is of any relevance
to the child. In a recent study in adults, isolated LA was highly associated with
thrombosis.[6] In children, a positive LA can be detected in approximately 2% of apparently healthy
children, usually discovered during preoperative assessment of coagulation.[7] These LA are usually transient and clinically insignificant, and are most likely
induced by infections or vaccinations. Of course, most of these LA do not show such
dramatic impact on clotting factor assays as in they did in this case. Our patient
was reevaluated in the coagulation clinic 3 months after surgery, and normal values
were obtained for APTT, and APTT-based FVIII and FIX activity. LA was no longer detectable
by DRVVT.
Interference of Fibrinogen Variants
Observation
A 22-year-old female patient in gestational week 41 + 3 was transferred from a birth
center to the hospital because of obstetrical arrest. After initial assessment, the
decision was made to deliver the child by caesarean section. An emergency coagulation
panel gave disturbed measurements for PT, APTT, and Clauss fibrinogen, and no values
were reported. Reflex testing of fibrinogen in an antigen assay showed 0.81 g/L (reference
interval: 1.9–4.3 g/L). Platelet function analysis was normal, and von Willebrand
antigen and activity were normal with normal activity/antigen ratio. The laboratory
and the obstetric team called the pathologist to seek advice.
Pitfall
Most routine coagulation assays use the patient's own fibrinogen to produce either
a change in light transmission (photo-optical method) or a mechanically detectable
change in the sample (mechanical method). This holds true for the “group tests” (PT,
APTT, and TT) and all clotting-based factor assays, such as Clauss fibrinogen, a diluted
variant of the TT. Antigen assays, such as the von Willebrand assays or the fibrinogen
antigen assay, are not based on clotting but on ligand binding to specific targets,
immobilized on latex particles or similar. These assays, and also the chromogenic
assays which use an artificial target antigen that is sensitive to FXa- or FIIa-mediated
cleavage and will give rise to color after being cleaved, are therefore independent
from the patient's own fibrinogen. Interpretation of the aforementioned case was fast
and easy because of the reflex testing of fibrinogen, which, for a pregnant woman
at term, was unusually low. She initially received 3 g of fibrinogen concentrate and
delivery was uneventful. Congenital fibrinogen disorders are classified according
to the fibrinogen levels.[8] Congenital dysfibrinogenemia should be suspected in the case of a decreased Clauss
fibrinogen and normal antigenic levels, a ratio of ≤0.7 is usually applied.[9]
[10] The term “hypodysfibrinogenemia” was coined to indicate that a low Clauss/antigen
ratio is observed together with reduced antigenic levels of fibrinogen. Hypofibrinogenemia,
in contrast, is defined by similarly decreased levels of both analytes. Our patient
apparently suffers from hypodysfibrinogenemia. A ratio could not be calculated, but
her results indicated nonfunctional fibrinogen (Clauss) plus reduced amounts of fibrinogen
(antigenic assay). A recent study indicated that a Clauss/antigen ratio of 0.7 would
have a sensitivity of 86%.[11] Additional tests are often necessary to distinguish between hypo- and hypodysfibrinogenemia,
and genetic testing may help confirm the diagnosis.
(Hypo-)dysfibrinogenemia is a rare disorder. What is seen more often in daily routine
are patients with acquired hypofibrinogenemia due to hepatic impairment, massive transfusion,
systemic thrombolytic therapy, and acute leukemias. Physicians should be aware that
the lower fibrinogen levels are, the more they affect clot-based assays. Severe hypofibrinogenemia
(<0.7–0.5 g/L) will result in failure of the assay to clot. In any no-clot sample,
the laboratory should also make sure that citrate plasma and not serum was investigated.
This mistake is unlikely to occur as long as the primary tube is used for analysis,
but may be observed when secondary tubes (aliquots) are received in the laboratory.[12] Be aware that analyzing serum may also give false high FVII and FIX activities.
This might be overlooked if these analytes are requested as single tests.
Physical Interferences
A specific coagulation assay or a group of assays that have been requested from the
laboratory may be affected by physical impact before it reaches the laboratory. This
includes, for example, length of tourniquet application to the patient's arm, increased
shear stress due to thin needles, inappropriate mixing of the sample, refrigeration,
and long or inappropriate transport to the laboratory. Inside the laboratory, adequate
centrifugation, timely separation and analysis, and appropriate freezing and defreezing
procedures are required. All steps prior to the analysis are usually covered under
the term “preexamination” or “preanalytical” phase, and this is where most errors
occur ([Table 1]). These errors may lead to platelet activation, consumption of coagulation factors,
or in vitro degradation of proteins, finally resulting in incorrect test results and,
in the worst case, inappropriate patient treatment. Therefore, the preanalytical phase
requires special attention, even though some or even all steps may occur outside the
control of the laboratory performing the coagulation tests. For a comprehensive overview,
see the article by Magnette et al.[13]
Table 1
Preanalytical errors
|
Error
|
Impact
|
Incorrect result
|
|
Refrigeration of citrated whole blood prior to centrifugation
|
Platelet activation
Loss of Willebrand factor
Loss of FVIII
|
Low von Willebrand
Low FVIII
|
|
Long transportation time of citrated whole blood
|
Protein degradation
|
Low FVIII, low FV, low protein S; prolonged PT, APTT
|
|
Shear forces, e.g., in pneumatic tube systems, heavy breaking in centrifuges
|
Thrombin generation
Platelet/microparticle contamination
|
Low APTT
Negative LA
|
Abbreviations: APTT, activated partial thromboplastin time; LA, lupus anticoagulant;
PT, prothrombin time.
Sample Handling in the Laboratory
Observation
A 30-year-old female patient attended the coagulation clinic after two episodes of
peripheral vein thrombosis and a transient ischemic attack during the last 2 years.
FVIII was ordered from the laboratory as part of the thrombophilia panel.[14] The FVIII clotting assay result was reported as 42%. All other coagulation parameters
including PT, APTT, TT, D-dimer, chromogenic protein C, free protein S antigen, LA,
antithrombin activity (both Xa-based and IIa-based), and APC resistance were found
to be normal.
Pitfall
Two important differential diagnoses in a female patient with reduced FVIII activity
are von Willebrand's disease and carrier status of hemophilia A. Further laboratory
testing was required and the patient returned to the coagulation clinic for additional
blood sampling. Her family history was taken again with a focus on bleeding disorders,
but no bleeders were identified. Using the bleeding assessment tool (BAT) from the
ISTH,[15] her ISTH-BAT score was calculated with 3 points (cut-off for women of 6 points).
Assessing women with carrier status, more than 90% have a positive family history;
and more than 60% have an abnormal bleeding score.[16] Assessing women with von Willebrand's disease, a positive ISTH-BAT has a sensitivity
of 70 to 100%, but low specificity.[15]
[17] Overall, additional clinical data seemed not very conclusive for either of the two
diagnoses. Repeat FVIII clotting test and chromogenic assay as well as von Willebrand
antigen and activity were ordered from the laboratory, together with the group tests
and von Willebrand multimers. Same-day results were as follows: PT of 100% (Quick),
APTT of 27 seconds, FVIII (clotting) of 88%, FVIII (chromogenic) of 92%, von Willebrand
antigen of 92%, von Willebrand activity of 88%, and activity/antigen ratio of 0.9.
Multimers were reported as normal with complete triplet structure some weeks later.
The discrepancy in FVIII prompted the pathologist to check all FVIII measurements
from the specific day when this patient was reported to have 42%. He revealed that
all requested assays were run the day the patient had attended the clinic except for
FVIII, for which the sample was frozen. The frozen sample was thawed and analyzed
2 days later, together with 20 other samples. Out of the 21 samples analyzed for FVIII
clotting that day, 14 (66%) were below the reference range (70–150). Internal quality
controls were within their range and did not show any trends. After discussing the
issue with the laboratory, it was finally realized that, against the standard operating
procedure, frozen plasma samples were thawed on the benchtop for a period of time
that could no longer be reconstructed. Frozen samples should always be thawed rapidly
(within 5–10 min) in an appropriate device at 37 °C until completely thawed,[18] and close monitoring during thawing is recommended. FVIII, but also FV and protein
S, is very sensitive to protein degradation and loss of activity is marked. This also
holds true if a sample is refrigerated prior to centrifugation or if the transport
time between phlebotomy and the laboratory is too long.
Chemical Interferences
A specific coagulation assay or a group of assays that have been requested from the
laboratory may be affected by chemical impact before it reaches the laboratory. This
includes the presence of an anticoagulant or a relevant active substance in the patient's
blood (such as vitamin K antagonists or direct oral anticoagulants [DOACs]), contaminating
anticoagulants present in lines (such as heparins in central venous catheters), or
inappropriate amounts of anticoagulants in the test tube, e.g., if the tube is underfilled.
Wrong Anticoagulant in the Collection Tube
Observation
FV activity test was requested as the only test in an outpatient and was very low
(<5%). Because homozygous FV deficiency is a very rare disorder,[19] the pathologist contacted the clinical team.
Pitfall
It was revealed that the patient had a family history and a personal history of thromboembolism
and that FV activity was ordered erroneously instead of a genetic test for FV R506Q
(Leiden). The initial analysis was performed from a secondary tube that was labelled
with the patient's details by the external laboratory which referred the sample to
us. Analyzing the plasma in this tube revealed gross hyperkalemia, in accordance with
the presence of K-EDTA and not Na-citrate as anticoagulant.[20] Potassium has strong effects especially on FV and FVIII activities[20] and should always be suspected when one (or both) of these two analytes are unexpectedly
low. Of note, K-EDTA may also affect the Bethesda assay,[21] and if both FV and FVIII inhibitors are detected, which would be very uncommon,
the presence of K-EDTA should also be suspected before reports are released by the
laboratory. Furthermore, EDTA is known to have high calcium-chelating capacity, which
interferes with routine recalcification, which has effects on all clotting assays.
Apparently, wrong anticoagulants may affect the test result.
Correct, but Too Strong Concentration of Anticoagulant in the Collection Tube
Observation
We received a preoperative inpatient sample for routine coagulation testing which
revealed a PT of 30% (Quick) and an APTT of greater than 120 seconds. Because both
results are above the alarm limits for coagulation assays in first-time patients,
the technician phoned the ward. The nurses reported that the patient was clinically
fine, was not taking any drugs, and was waiting for surgical shoulder repair after
a sport trauma. Under-filling of the tube was suspected, and a new sample was requested.
Unexpectedly, results did match the previous findings, and the laboratory contacted
the pathologist. A quick review of the additional laboratory results revealed unremarkable
results for biochemistry but a hematocrit of 65 L/L (reference range, 37–45 L/L).
Pitfall
If the concentration of the anticoagulant in the test tube is too strong, all clotting
times are affected. One frequently seen problem in the coagulation laboratory is under-filling
of the primary tube. Blood tubes for coagulation testing should be filled not less
than 90% of their nominal filling volume.[22] Filling usually requires an optical check at sample reception, since automated analyzers
do not perform this check properly. If the sample is under-filled (meaning that not
enough whole blood was added during the blood draw), a higher amount of Na-citrate
than required will be present in the patient's blood. Following centrifugation, the
patient's plasma will contain more citrate than necessary. When the effect of Na-citrate
is then reversed in the analyzer by adding a standard amount of calcium, higher amounts
of citrate will bind more calcium than calculated. Accordingly, the sample is not
fully recalcified and clotting times are prolonged. Incorrect filling may account
for up to 60% of sample rejections in the coagulation laboratory.[23] Of note, although coagulation tubes filling accuracy was recently reported to be
within ± 10% for the manufacturers tested, the overall bias was found to be highly
variable among manufacturers and lots[24] and lot-specific under-filling cannot be excluded. Also do not forget that vacuum-based
blood collection tubes lose their capability of complete filling over time and should
not be used after their expiry date.[25] All these potential interferences were excluded in our patient. One important additional
reason for the constellation seen in this patient would be high amounts of heparin
in the sample. This sometimes occurs if the coagulation sample is taken from existing
lines which either had been flushed with heparin or which are used actively for IV
heparin therapy.
Notably, the patient had a very high hematocrit and was later diagnosed with polycythemia
vera. When the cells are separated from her plasma, a relatively small amount of plasma
will be anticoagulated in a relatively high amount of Na-citrate. Comparable to what
happens in an under-filled sample, too much citrate remains after recalcification
and clotting times will be prolonged, starting somewhere at a patient's hematocrit
of 56 to 60 L/L.[26] The amount of citrate in the tube should therefore be adjusted according to the
individual hematocrit using the following formula: volume of citrate solution (3.2%) = volume
* (100-hematocrit)/(640-hematocrit).
Anticoagulant in the Patient's Plasma
Observation
A patient with recurrent thromboembolic events taking 20 mg rivaroxaban OD after segmental
pulmonary embolism 4 months before presented to the coagulation clinic. A partial
thrombophilia screening had been performed from a colleague in private practice, indicating
absence of protein C, protein S, and antithrombin deficiency, and absence of FV R506Q
(Leiden) and prothrombin G20210A mutations. Despite the fact that functional coagulation
tests were already performed, a regular thrombophilia panel adopted for patients under
direct Xa inhibitors was ordered. Laboratory results were normal for chromogenic protein
C activity, and free protein S antigen. The antithrombin results were as follows:
antithrombin FXa-based, 126%; antithrombin FIIa-based, 105%; and antithrombin FXa-based
after DOAC adsorption, 58%.
Pitfall
In this patient on rivaroxaban for pulmonary embolism, the effect of the DOAC on functional
coagulation assays was overlooked. DOACs include both direct thrombin inhibitors (dabigatran)
and direct factor Xa inhibitors (apixaban, edoxaban, rivaroxaban). These medications
may cause assay interference by falsely increasing or decreasing measured values,
depending on the analyte, the drug, and the drug's concentration (which largely depends
on the interval between the last drug intake and the time of blood sampling). For
an overview, see [Table 2]. Dabigatran and rivaroxaban significantly prolong the PT and APTT in a concentration-
and reagent-dependent manner; rivaroxaban has a higher impact on PT and dabigatran
on APTT. The presence of dabigatran can lead to falsely reduced Clauss fibrinogen
when measured with low thrombin concentration reagents. Antithrombin is overestimated
in FXa-based assays in the presence of rivaroxaban, and in FIIa-based assays in the
presence of dabigatran. Instrument-related differences can be found for all parameters.[27] Apixaban, in contrast, has only mild effects on coagulation tests.[28] PT and APTT are barely influenced. At 225 ng apixaban/mL, PT and APTT are approximately
1.15 times prolonged, with some reagents being more sensitive than others.[29] However, antithrombin activity measured in FXa-based assays is also overestimated
in the presence of apixaban. In comparison with the other two oral Xa inhibitors,
the in vitro effects of edoxaban are more similar to rivaroxaban than apixaban.[30] Apparently, our patient has an antithrombin deficiency that was not detected because
an anti-Xa-based antithrombin assay was performed despite the fact that she was on
rivaroxaban.
Table 2
Impact of DOAC on coagulation assays
|
Impact
|
Assay
|
Clinical risk
|
|
False increase
|
Antithrombin activity[a]
Clot-based protein C activity
Clot-based protein S activity
|
Deficiency of protein C, protein S, or antithrombin may be overlooked
|
|
APC resistance[b]
|
FV Leiden mutation may be overlooked erroneously
|
|
LA screen/LA ratio
|
The presence of LA may be suspected erroneously because of DOAC effect
|
|
False decrease
|
Single factors in clotting assays (PT or APTT based)
|
Misdiagnosis of single factor deficiency; mixing studies may show (unusual) inhibitor
effects that shall not be mistaken as indicative for the presence of an inhibitor
|
|
No change
|
Clauss fibrinogen[c]
Fibrinogen antigen
D-dimer
Chromogenic protein C activity
Free protein S antigen
Total protein S antigen
von Willebrand antigen and activity
|
–
|
Abbreviations: APC, activated protein C; APTT, activated partial thromboplastin time;
DOAC, direct oral anticoagulant; LA, lupus anticoagulant; PT, prothrombin time.
a Apixaban, edoxaban, and rivaroxaban have impact on FXa-based antithrombin assays,
and dabigatran on FIIa-based antithrombin assays.
b Assay-dependent observation.
c In high concentration, dabigatran may lower Clauss fibrinogen in some assays.
Physicians can choose from different strategies to minimize DOAC interference in coagulation
testing. The simplest way is avoiding sensitive coagulation tests in patients taking
DOACs ([Table 2]). If a sensitive test is required under DOAC therapy, the drug may be paused for
2 (or 3) days prior to collecting the blood sample. This interruption may put the
patient at increased risk of (re-)thrombosis, and transitioning to low-molecular-weight
heparin (LMWH) could be an option. A much more simple and effective way is the use
of adsorbing agents that can neutralize DOACs in the patient sample. Several studies
have demonstrated that DOAC removal by adsorbing agents is efficient in almost all
patient samples.[31] Some centers which receive large amounts of plasma samples potentially containing
DOAC, such as from regional heart centers, have adopted strategies to double test
sensitive parameters in the absence and presence of an adsorber. As seen in our patient,
the “true” FXa-based antithrombin activity is 58%, but was increased to 126% in the
presence of rivaroxaban. One might argue that the alternative option in this patient
would have been to rely on a FIIa-based assay which is unaffected by rivaroxaban.
Unfortunately, this assay gave normal antithrombin activity. Later, genetic testing
of this patient revealed the presence of a SERPINC1 c.[236G > A];[=] p.(Arg79His) mutation, also known as antithrombin Rouen I, which
is undetectable in FIIa-based assays. The lesson to be learned here is to exclude
false normal measurement of the important coagulation inhibitors: clot-based protein
C and protein S activity, and antithrombin, because of DOAC. The best strategy is
using adsorbers. We also believe that young patients with unexplained thrombosis should
also be assessed by at least two functionally different antithrombin assays.
This case can also be used to address another aspect—the question how to identify
DOACs in a sample. All DOACs can be measured “specifically” in anti-Xa or anti-IIa
assays calibrated with the specific drug. This might be helpful in critical clinical
situations, such as deciding the right treatment strategy in a patient with fresh
stroke taking DOAC for atrial fibrillation: lysis or no lysis?[32] Of course, “specific” anti-Xa assays are only inasmuch specific as they are calibrated
against the drug (e.g., rivaroxaban). Analyzing the rivaroxaban concentration in a
patient's plasma in the presence of additional drugs that interfere with the assay
system, such as LMWH, does not reflect the true rivaroxaban concentration.[33] Finally, it should be kept in mind that PT and APTT normal values cannot exclude
DOAC anticoagulant activity in a patient sample, and PT or APTT prolongation is not
always associated with DOAC anticoagulant effect when compared to specific tests.[34]
Conclusion
There are multiple interfering variables that can affect results in the coagulation
laboratory. Training of medical staff is relevant to reach an appropriate level of
awareness that biological, physical, and chemical interferences may disturb an assay
or a group of assays. Whenever a coagulation result is unexpected or seems highly
unlikely, close contact between the pathologist and the clinical team is the best
way for identifying the root cause.
