The futility of thrombophilia testing

PC/PC/AT

As mentioned, PC/PC/AT are rare in the general community (<0.5%) and will at best only be identified at a rate of approximately 5% in unselected VTE. Laboratory normal reference ranges (NRR) are most typically established using 95% confidence intervals, meaning that they will ‘confidently capture’ 95% of the normal population. However, what this also means is that 5% of the normal population will (by definition of the NRR process) fall outside these limits; thus, approximately 2% of normal individuals will be identified as falsely having a PC or PC or AT deficiency with each round of testing for each parameter. As these ‘false low values’ may identify different normal individuals for each of the three different tests evaluated, upwards of 6% of normal individuals may in total therefore be falsely identified as having a PC or PC or AT deficiency following testing of all three parameters. Thus, the risk of falsely being identified as PC or PC or AT deficient (approx. 6%) is similar to the chance of identifying a true deficiency in unselected VTE cases (approx. 5%), and >10× the chance of identifying a true deficiency in a normal population (<0.5%). A second consideration is the timing of the clinical request or blood sampling. Reductions in PC/PC/AT levels may occur just after a thrombosis due to ‘consumption’ or inflammatory events, and thus testing patients at this time can lead to detection of additional false deficiencies. Moreover, patients having suffered a thrombosis such as a VTE are subsequently placed on anticoagulant therapy, to both treat the thrombosis and prevent extension or reoccurrence of thrombosis, and potentially including unfractionated heparin, low molecular weight heparin, vitamin K antagonists such as warfarin, or perhaps one of the newer oral anticoagulants such rivaroxaban [4, 5]. Each of these anticoagulants will variably affect PC/PC/AT assays [6, 7], and samples taken from patients once commenced on these drugs have a very high chance of yielding a false deficiency. We also know from audits of clinical ordering practice that between 30% and 50% of test requests for PC/PC/AT assays occur while patients are on anticoagulant therapy [8, 9]. Indeed, one recent audit revealed that an alarming 80% of test cases determined to have low PC or PC or AT were likely to have derived from patients on anticoagulant therapy, and thus were potentially false-positive events [9].

FVL/PGA

As mentioned, FVL/PGA are not so rare in the general community (approx. 5% in whites). Compared to PC/PS/AT, there is much less chance of a ‘false-positive’ finding since genetic tests tend to provide more definitive yes/no answers. However, the relatively common presentation of FVL/PGA in the general population means that these can ‘easily’ be found in both thrombosis affected and unaffected individuals. Although FVL/PGA increase an individual’s risk of thrombosis, particularly when they coexist with other identifiable acquired risk factors, there is a clear danger that both clinicians and patients will over-interpret the importance of detection of FVL/PGA mutations. When clinicians become less selective of patients to investigate, the FVL/PGA identified becomes increasing irrelevant, with total irrelevance achieved when detection rates approach those of the normal population, since in essence this means that the normal population is being tested for these thrombophilia markers. Recent audits of clinical ordering and test practice have indeed confirmed this to be the case [9, 10] (Figure 1). Moreover, the same audits have revealed that current clinical requests are simply following other natural event trends, namely births by age for women, and age-related VTE rates for males; thus, the current clinical ordering patterns simply appear to be following pregnancy trends in women and VTE occurrence trends in males.

Figure 1

Decreasing rate of FVL heterozygote detection (as a percentage of FVL tests performed) during the past 17 years at the author’s institution (modified from [9, 10]).

Antiphospholipid antibodies

These can be identified by either clot-based assays that detect so-called lupus anticoagulants (LA) or solid phase assays that detect other antiphospholipid antibodies (aPL) [6, 11]. The anticoagulants mentioned previously, given as thrombosis therapy and affecting PC/PS/AT assays, also affect the clot-based assays used to identify LA, and can differentially produce both false-positive and false-negative results [6]. Although solid phase assays aPL assays are generally unaffected by these anticoagulants, they are instead biased by much higher inter-laboratory variation than LA testing [12, 13], thereby limiting their clinical utility. In summary, some laboratories will report an aPL negative sample as being aPL positive, some laboratories will report an aPL positive sample as being aPL negative, and the strength of an aPL positive sample will also be variably reported. Finally, evidence that clinicians are inappropriately ordering aPL testing is also available. As an example, despite its proven association with pregnancy morbidity/mortality, one recent audit identified that none of the 72 consecutive obstetric patients tested at one institution within a 6-month period was identified to have aPL [14]. The conclusion here is that the presence of aPL is either uncommon in the obstetric patient group (which we know to be untrue), or that the clinical reasoning for laboratory investigation of aPL by the obstetrics team require refinement, and specifically better patient selection.

In summary, the current evidence indicates that appropriate patient selection for thrombophilia investigation is simply not occurring, and instead clinicians are requesting such tests fairly unselectively. This has several adverse consequences (Table 1). First, it is costly and wasteful of health resources. Second, it is important to remember that identification of a ‘false positive’, be it PC, PC, or AT deficiency in an anticoagulated patient, or a FVL or PGA mutation in a 50-year-old woman without any additional risk factors, do not represent benign discoveries. Significant adverse effects will arise from these events, including the patient’s psychological distress at discovering that they have a PC/PC/AT deficiency or perhaps worse a ‘genetic mutation’ (such as FVL/PGA); there are also subsequent family issues when asymptomatic family members are also ‘discovered’ to have similar ‘genetic mutations’. Potentially worse is the risk that clinicians will over treat clinical conditions based on false positives for PC/PC/AT deficiency or the presence of FVL/PGA or aPL, e.g., by extending the duration of anticoagulant therapy for these ‘lifelong conditions’ and thus increased risk of eventual bleeding events [4, 15, 16]. Also important is that all the false diagnoses (e.g., PC/PC/AT deficiencies) will either remain in the patient’s clinical history files, or will need to be reversed by additional retesting. For example, once an individual is marked as being ‘PC/PC/AT deficient’, it becomes difficult to reverse this diagnosis, since the record of ‘deficiency’ is often retained forever within the patient’s clinical notes or within laboratory information systems.

Finally, in many cases, the utility of even a true-positive diagnosis remains limited. For example, these laboratory markers do not provide very high relative risk for thrombosis reoccurrence, so knowledge of their presence or absence provides only limited utility for advising on future risk. Moreover, discovery of true PC/PC deficiencies or FVL/PGA mutations do not in general alter clinical management, which in most cases involves short-term (3–6 months) anticoagulant therapy and avoidance of high thrombosis risk activities, which would be similar irrespective of the presence or absence of these markers.

In conclusion, thrombophilia testing is more likely to impact negatively on the health care of the people being tested, as well as the health care of their family members, than the positive impact that is promised by theoretical considerations [1, 2, 17], thereby leading the author to conclude on the general futility of thrombophilia testing.

Conflict of interest statement

Author’s conflict of interest disclosure: The author stated that there are no conflicts of interest regarding the publication of this article.

Research funding: None declared.

Employment or leadership: None declared.

Honorarium: None declared.