|
|
 Canadian
Medical Association Journal
Home |
|
Criteria for evaluating evidence that laboratory abnormalities are associated with the development of venous thromboembolism
See also: Abstract The identification of conditions associated with an increased risk of venous thromboembolism may indicate the need for aggressive prophylaxis during periods of high risk, prolonged anticoagulant therapy after an initial venous thromboembolic episode, the investigation of asymptomatic family members and the avoidance of oral contraceptives. Advances in laboratory medicine have led to the identification and assessment of many proteins responsible for normal hemostasis, and associations between abnormalities in a number of these proteins and venous thromboembolism have been reported. Without the ability to appraise this information critically, physicians may be unable to determine whether or how they should modify their clinical practice. Criteria for determining whether specific laboratory abnormalities have a relationship with venous thromboembolism are proposed here, and one example of the application of these guidelines is provided. Venous thromboembolism (VTE) can be subclassified as being associated with either a short-term (e.g., post hip surgery) or long-term (e.g., metastatic cancer) clinical risk factor, a laboratory abnormality (e.g., antithrombin [AT] deficiency), or there may be no associated factor (idiopathic). The development of "screening for thrombophilia" (an increased tendency to develop VTE) was driven by the observation that some young individuals with recurrent VTE and no other associated risk factors had a strong family history of VTE, suggesting an autosomal dominant genetic defect.1 Subsequently, AT was discovered, and an inherited deficiency was characterized that causes AT levels to be about one-half of the norm and leads to a predisposition to recurrent VTE.2,3,4,5,6 Other proteins responsible for normal hemostasis have since been identified. Assays that measure peptides or enzyme-inhibitor complexes reflecting thrombin generation or activity have become routinely available, as have "global tests" that reflect interactive cascades. Associations between abnormalities in the results of many of these tests and VTE have been reported. Laboratory parameters that are considered to be useful in diagnosing thrombophilia are summarized in Table 1. Many centres now routinely test young patients for thrombophilia after one episode of thrombosis or when there is a family history of thrombosis. Well-designed prospective studies show that the risk of recurrent VTE differs in the subgroups of patients described earlier.7,8,9,10,11 Patients with a short-term risk factor have a low risk of recurrence after 3 months of anticoagulants.8,9,10,11 Patients who develop idiopathic VTE and those with a persistent risk factor have a higher risk of recurrence,7,8,9,10,11 necessitating prolonged anticoagulation. Many experts recommend long-term anticoagulants when VTE is associated with a laboratory abnormality,12 although this approach has not been evaluated in clinical trials. Testing is also performed to identify individuals who might benefit from more aggressive prophylaxis in situations associated with an increased risk of VTE, for example, pregnant women with AT deficiency.12,13,14 The role of many of these laboratory abnormalities in predicting VTE is unclear, and an abnormal assay with an unproven link to VTE might mislead clinicians into inappropriate decision-making. In this review we propose criteria for determining whether laboratory abnormalities have a relationship with the development of VTE, and we apply the criteria to a recently discovered laboratory assay. Levels of evidence for assessing the significance of laboratory abnormalities in VTE Many studies suggest that the inappropriate ordering of tests is a significant problem in several clinical settings.15,16,17,18,19,20 When evaluating the usefulness of a laboratory assay in diagnosing thrombophilia, the credibility of the assay, the strength of the association between VTE and the laboratory abnormality, and the evidence for a causal relationship between the 2 must be considered. Elements relevant to assessing causation20 and information contained in previously published "User's Guides to the Medical Literature" that evaluate harm associated with a particular exposure21 can be used in this situation. The following criteria may prove helpful for evaluating laboratory assays reported to diagnose thrombophilia. 1. Is the laboratory assay measured in a credible fashion?
To assess an assay's credibility, 4 questions must be answered.
2. Is there supporting evidence from clinical or observational investigations that the abnormality is associated with VTE?
Since randomized trials are not available in this area, cross-sectional family studies, cohort studies or case–control studies, or all 3, constitute the usual evidence. Relevant criteria are listed below.
3. Have potential confounders been ruled out as causes for the observation?
Even if the laboratory measurements are valid, a comparison between groups for a given attribute can be biased if some extraneous factor that can affect that attribute independently is unequally distributed. Systematically biased measurements may either create or obscure associations.
Confounders may alter laboratory results. Normal values of components of the hemostatic system vary with age,23 sex and nutritional status. Heparin administration24 or disseminated intravascular coagulation,25 or both, have been associated with temporary reductions in AT levels. Similarly, coumarin derivatives cause low levels of proteins C and S.26,27,28 Liver disease affects many hemostatic factors. Therefore, when a laboratory abnormality occurs, the test should be repeated once the confounder(s) has (have) resolved.
Other confounders may independently alter the risk of developing VTE. Clinical studies of VTE should control for concomitant laboratory abnormalities independently associated with VTE as well as clinical confounders, such as recent trauma or surgery. 4. Is the magnitude of the abnormality sufficient to explain thrombophilia? A minimum clinically important alteration in levels of laboratory variables appears necessary to increase the risk of thrombosis in individuals with congenital thrombophilia. For example, symptomatic individuals with a congenital AT deficiency have levels that are approximately 50% of the norm. For most acquired abnormalities, the critical level at which the risk of VTE is increased is unknown. A statistically significant decrease in AT levels of 10% (from 100% to 90%) associated with oral contraceptive use29 has been advanced as an explanation for its association with VTE. The clinical relevance of such an alteration is dubious in view of the 50% reduction noted in symptomatic individuals with inherited AT deficiency, and because the AT levels are still well within the normal range (about 80% and above). On the other hand, patients with severe liver disease may have AT levels of 50%, however, because of concomitant reductions in procoagulants, they usually bleed rather than develop VTE (see earlier section on confounders). 5. Does the link between the laboratory abnormality and VTE make biologic sense?
The association should be plausible and consistent with current knowledge.
Summary The strength of the evidence available to support an assay's role in diagnosing thrombophilia should be described. The more criteria an assay meets, the more likely its validity. However, not all criteria are equally important. Biologic plausibility is less important than credible laboratory technique. Therefore, it is not sufficient to simply tally the number of criteria met in order to describe the strength of the evidence available to support an assay's diagnostic role. Fig. 1 shows an algorithm for assessing levels of evidence. At a minimum, we suggest that there should be supporting evidence that the abnormality is measured using credible laboratory techniques and is associated with VTE (level III evidence). If the magnitude of the abnormality is sufficient to explain thrombophilia and the elimination of potential laboratory and clinical confounders has been achieved, this strengthens the available evidence (level II evidence). Level I evidence implies that all the criteria for level II have been met and there is biologic plausibility for an association. Application of criteria Induced resistance to the activated protein C (APC) anticoagulant system, as diagnosed by the APC sensitivity ratio (APC-sr) assay,35 has been proposed by some to explain the stronger association between VTE and the third-generation oral contraceptives than between VTE and the second-generation oral contraceptives demonstrated in some,36,37,38 but not all,39,40,41 studies. For illustrative purposes, the validity of this conclusion will undergo an initial assessment by determining whether this assay meets the criteria for association with VTE. 1. Is the laboratory assay measured in a credible fashion? The APC-sr has been evaluated in young healthy volunteers, users and nonusers of oral contraceptives, individuals with a history of VTE and carriers of the Factor V Leiden mutation. Pregnant women and patients with intercurrent diseases have not been assessed. Other than in those initiating and discontinuing oral contraception, the variability of the APC-sr in a given individual over time has not been described. Although the assay has been independently and consistently reproduced in another laboratory,42 others have had difficulty replicating the methods (Marilyn Johnston, Hamilton Civic Hospitals Research Centre, Hamilton, Ont.; personal communication, 1999). 2. Is there supporting evidence from clinical or observational investigations that the abnormality is associated with VTE? The available evidence is conflicting. In a case–control study of 172 consecutive men with a first, objectively confirmed episode of deep vein thrombosis and 201 age- and sex-matched controls,43 the APC-sr predicted risk of venous thrombosis in the highest quartile of the ratio compared with the lowest quartile of the ratio, even after exclusion of Factor V Leiden carriers, (odds ratio [OR] 3.4, 95% confidence interval [CI] 1.1–10.8; p < 0.05).43 However, another case–control study involving 67 women with confirmed VTE and 290 age-matched controls found no association between VTE and APC-sr (OR 0.65, 95% CI 0.35–1.22).42 Exclusion of Factor V Leiden–positive women and oral contraceptive users did not change the results significantly. 3. Have potential confounders been ruled out as causes for the observation? Although the APC-sr predicted risk of thrombosis after exclusion of Factor V Leiden carriers in the first study described above,43 other abnormalities associated with abnormal APC-sr results (protein C or S deficiency) were not excluded. 4. Is the magnitude of the abnormality sufficient to explain thrombophilia? In women taking the third-generation oral contraceptive, the APC-sr results were similar to those seen in Factor V Leiden heterozygotes and significantly higher than those in subjects taking second-generation oral contraceptives;35 therefore, the magnitude of the abnormality could be sufficient to explain thrombophilia. However, as noted earlier, since this is an acquired abnormality, one cannot conclude that similar decrements to those observed in a congenital abnormality confer a similar magnitude of (or any) thrombophilia.
5. Does the link between the laboratory abnormality and VTE make biologic sense?
The association between APC-sr and thrombophilia remains controversial. On the one hand, there is biologic plausibility, and the test appears to be capable of distinguishing subjects who have Factor V Leiden from those who do not. On the other hand, a number of criteria have not been met by the APC-sr. Although the assay has been independently reproduced by one group, others have had difficulty with the assay (Marilyn Johnston, Hamilton Civic Hospitals Research Centre, Hamilton, Ont.; personal communication, 1999). Furthermore, given the conflicting clinical evidence regarding an association between the abnormality and VTE, there is insufficient evidence, using our algorithm, to support the diagnostic role of this assay. Until these issues have been addressed, we believe it is premature to conclude that this assay is useful in diagnosing thrombophilia. Conclusion The clinician will continue to be inundated with new laboratory tests that are said to predict VTE. Without critically appraising the available information, physicians may be unable to determine whether or how they should modify their clinical practice. We have proposed criteria to assist physicians in using information from studies to estimate the usefulness of laboratory tests in the diagnosis of venous thrombophilia.
Competing interests: The authors received an unrestricted educational grant from NV Organon to complete this paper. Dr. Rekers is an employee of NV Organon, a manufacturer of oral contraceptives. Dr. Sackett has received honoraria and research funding from a number of pharmaceutical firms. He has been a paid consultant on 2 occasions. His wife inherited and sold stock in a pharmaceutical company. Dr. Sackett received the Pharmaceutical Manufacturers' Association of Canada Medal of Honour (and cash) for "Contributions to Medical Science in Canada" for the decade 1984–1994.
Contributors: All authors contributed to the manuscript's conception. Dr. Bates was primarily responsible for reviewing the literature and drafting the manuscript. Dr. Ginsberg reviewed each draft of the article. Drs. Ginsberg, Straus, Rekers and Sackett critically reviewed the manuscript and were responsible for important revisions to the intellectual content.
Acknowledgements: The authors thank Drs. Harry Buller, Jack Hirsh and Clive Kearon for their critical review.
Dr. Bates is a recipient of a Heart and Stroke Foundation of Ontario Research Fellowship. Dr. Straus received support from the R. Samuel McLaughlin Foundation. Dr. Ginsberg is a recipient of a Career Investigator Award from the Heart and Stroke Foundation of Ontario. This work was supported by an unrestricted educational grant from NV Organon. From *the Department of Medicine, McMaster University, Hamilton, Ont.; †Hamilton Civic Hospitals Research Centre, Hamilton, Ont.; ‡the Department of Medicine, University of Toronto, Toronto, Ont.; §NV Organon, Oss, the Netherlands; and ¶the Trout Education and Research Centre, Trout Lake, Ont. This article has been peer reviewed. Reprint requests to: Dr. Shannon M. Bates, Thromboembolism Unit, HSC 3W15, McMaster Site, Hamilton Health Sciences Corporation, 1200 Main St. W, Hamilton ON L8V 1C3; fax 905 521-4997; batesm@fhs.mcmaster.ca References
Copyright 2000 Canadian Medical Association or its licensors |
