Relation of Factor V Leiden Genotype to Risk for Acute Deep Venous Thrombosis after Joint Replacement Surgery

Methods

Study Samples

Patients undergoing total hip or knee replacement at Strong Memorial Hospital, Rochester, New York; The Genesee Hospital, Rochester, New York; McMaster University Medical Center, Hamilton, Ontario; or the Hamilton Civic Hospitals, Hamilton, Ontario, were recruited for one of six prospective studies [22-27]. Patients were excluded if they were younger than 18 years of age, were allergic to contrast media, were receiving aspirin or long-term anticoagulation therapy, or had an underlying bleeding disorder or recent bleeding from the gastrointestinal or urinary tract. Women with childbearing potential were included if the result of a pregnancy test was negative. Additional exclusion criteria (serum creatinine level > 1.7 mg/dL [150.28 µmol/L]; recent eye, ear, or central nervous system surgery; known hypersensitivity to heparin; severe hypertension; or weight < 90 lbs [40.5 kg]) were used in one study [27]. If a patient underwent two separate surgical procedures, only the first was used for analysis. All studies were approved by the appropriate institutional review boards, and all patients gave informed consent for enrollment in the individual studies.

Treatment Regimens

Anticoagulant regimens used in the studies are shown in Table 1. For our analysis, the regimens were grouped into one of three categories: warfarin (grouped by different dosing schedules), heparin (unfractionated or low-molecular-weight), and external compression (pneumatic compression or graduated compression stockings). Prophylaxis was continued until venography was performed.

Venography

Bilateral ascending venography [28] was performed between the fifth and ninth day after surgery in more than 90% of patients. Venograms were interpreted by radiologists who were unaware of the patients' treatment assignment or factor V genotype. A constant intraluminal filling defect identified in two different projections was the criterion for a positive venogram. Thrombi were classified as involving proximal (iliac, femoral, or popliteal) or distal (anterior tibial, posterior tibial, or peroneal) deep veins. Patients with clots involving superficial (superficial, communicating, or muscular) leg veins were not included in the group considered to have deep venous thrombosis.

Clinically evident bleeding complications were identified. In three studies, objective measures of bleeding (surgical blood loss and maximum hematocrit decrease) were determined [22, 23, 27].

Plasma Samples

We used 825 plasma samples collected from study patients within 2 weeks of surgery to determine factor V genotype. This number accounted for 47% of all patients in the six studies conducted between 1988 and 1994 and represented all samples available from these patients at the time that we began our retrospective analysis. Plasma samples were prepared by single or double centrifugation of citrated blood at 1500 g for 10 to 15 minutes, and platelet-poor plasma was then stored at −70°C.

Factor V Genotyping

We extracted DNA from 200 µL of thawed plasma by using the QIAamp Blood Kit (Qiagen, Chatsworth, California) according to the manufacturer's instructions. Preliminary studies indicated that successful amplification of DNA from plasma required nested polymerase chain reaction (PCR) to achieve the necessary sensitivity. The first round of PCR was performed as described elsewhere [12]. The second round of PCR (using 0.1 µL of PCR product amplified under the same conditions as in the first PCR) was primed with oligonucleotides that corresponded to factor V sequences internal to the first-round primers so that misprimed sequences would not be amplified in the nested reaction. The nested primers (sense, 5′TACTACAGTGACGTGGACATCA3′ and antisense, 5′TTTAGCCAGGAGACCTAACATG3′) yielded an expected product of 177 base pairs. Products of PCR were tested for purity and expected size by 4% NuSieve (FMC Bioproducts, Rockland, Maine) agarose gel electrophoresis with ethidium bromide staining. Contamination of PCR products was monitored by simultaneous amplification of a negative control that contained all reagents except template DNA.

The final PCR product was digested with MnlI endonuclease, as described elsewhere [12]. MnlI digested the 177-base pair PCR product of the nested primer reaction into three fragments (94 base pairs, 46 base pairs, and 37 base pairs) in normal persons and two fragments (131 base pairs and 46 base pairs) in persons who are homozygous for the factor V Leiden mutation. Heterozygotes for the mutation showed a combination of all three bands. Seven samples showed a weak variant band after amplification and digestion. In all cases, repeated PCR amplification and digestion showed an expected homozygote or heterozygote pattern. The nested PCR assay done by using plasma DNA has been validated by comparison with standard PCR done by using leukocyte DNA in selected samples (Ryan DH, Nuccie BL, Maize B, Arvan DA, Francis CW. Amplification and genotyping of coagulation factor V genomic sequence from stored frozen plasma. In preparation).

Statistical Analysis

Before doing sample genotyping, we performed a power analysis in which we assumed a prevalence of 5% for the factor V Leiden mutation and a 20% overall expected rate of thrombosis. The primary hypothesis was that factor V Leiden acts in concert with surgical risk factors to increase the risk for deep venous thrombosis to greater than 50% in affected patients undergoing joint replacement surgery and receiving antithrombotic prophylaxis. The number of patients needed to assure an 80% power of identifying a difference between a 20% and an 80% thrombosis rate in the two patient groups at a 5% significance level was 130. If the thrombosis rate in the population with the factor V Leiden mutation was 60% or 50% instead of 80%, the same power for identifying a significant difference could be achieved with 340 and 640 total patients, respectively [29]. We made comparisons by using a Fisher exact test and a likelihood ratio chi-square test for binary variables and a two-tailed Student t-test for continuous variables. Multivariate data analysis was done by using stepwise logistic regression. We calculated univariate relative risks and CIs according to standard methods [30] and considered results statistically significant if the 95% CI did not overlap with 1.0.

Role of Funding Sources

The funding sources for this work had no role in the gathering, analysis, or interpretation of data or in the decision to submit the manuscript for publication.

Results

The clinical characteristics of patients with and without the factor V Leiden allele are shown in Table 2. The prevalence of the mutation in the total combined patient population was 3.9% (32 of 825 patients). Only 1 patient was homozygous for the mutation. The proportions of patients with the factor V Leiden mutation in the University of Rochester and McMaster University studies were similar (3.3% [12 of 365 patients] and 4.3% [20 of 460 patients], respectively; P > 0.2).

In the combined patient population, the presence of a factor V Leiden allele did not significantly increase the risk for acute deep venous thrombosis (Table 3). The incidence of venographically determined deep venous thrombosis was 31% (95% CI, 15% to 47%) in patients with the mutation and 26% (CI, 22% to 29%) in patients without the mutation. The relative risk for deep venous thrombosis in patients with a factor V Leiden allele compared with patients who were homozygous for the normal allele was 1.2 (CI, 0.6 to 2.9). The 1 patient who was homozygous for the factor V Leiden allele had distal deep venous thrombosis. Only 4.7% of all patients with deep venous thrombosis had a factor V Leiden allele. The incidence of deep venous thrombosis was not significantly correlated with the presence of a factor V Leiden allele in separate analyses of proximal or distal deep venous thrombosis or analysis of subgroups of patients receiving different types of anticoagulants (Table 3). In contrast, history of venous thrombosis, although not associated with the factor V Leiden mutation (Table 2), was strongly correlated with the occurrence of deep venous thrombosis. Patients with a history of venous thrombosis had a 50% (20 of 40 patients) incidence of deep venous thrombosis; the corresponding incidence in patients without a history of venous thrombosis was 26.7% (117 of 438; P = 0.002).

The incidence of deep venous thrombosis was increased in patients having hip replacement who had a factor V Leiden allele (Table 3), but this result was of borderline statistical significance. No significant difference was seen in the larger group of patients undergoing knee replacement surgery. Logistic regression analysis to assess the independent contribution of surgical site, anticoagulant prophylaxis, medical center, and factor V genotype to the occurrence of deep venous thrombosis did not show that factor V genotype contributed significantly to the risk for deep venous thrombosis.

A small subset of patients (n = 39) had unilateral venography performed on the limb on which surgery was done. Analysis of the larger group of patients who had bilateral venography indicated that the limb that was not operated on was the sole site of thrombosis in only 20% of cases in which at least one side showed deep venous thrombosis. Because 12 of the 39 patients who had unilateral venography had deep venous thrombosis in the limb on which surgery was done, these data suggest that an additional 3 patients may have had deep venous thrombosis in the limb that was not operated on that was missed because unilateral venography was done. Reanalysis of the data after modeling inclusion of 3 additional patients who had deep venous thrombosis or after eliminating all patients who had unilateral venography did not significantly alter the relation between factor V Leiden status and deep venous thrombosis in the combined group or subgroups of patients.

In the overall group of patients, no significant correlation was seen between the presence or absence of a factor V Leiden allele and the incidence of severe hemorrhage (1 of 30 patients [3%] and 15 of 638 patients [2%], respectively; P > 0.2) or minor bleeding (2 of 30 patients [7%] and 35 of 638 patients [5%]; P > 0.2). Clinical bleeding events were not recorded in one study [26]. Because only 3 patients with a factor V Leiden allele had a clinically identified hemorrhagic complication, we could not evaluate the relation of factor V genotype to hemorrhagic risk in subgroups of patients. Quantitative data gathered in three studies [22, 23, 27] support the lack of association between factor V genotype and bleeding tendency. The results show that in patients who underwent hip surgery, presence or absence of a factor V Leiden allele was not correlated with mean surgical blood loss (864 ± 350 mL in 8 patients with the allele compared with 940 ± 602 mL in 255 patients without the allele; P > 0.2) or maximum hematocrit decrease (8.3 ± 3.2 mL in 8 patients with the allele compared with 9.6 ± 4.1 mL in 253 patients without the allele; P > 0.2). Similarly, in patients who underwent knee surgery, presence or absence of a factor V Leiden allele was not correlated with mean surgical blood loss (250 ± 194 mL in 5 patients with the allele compared with 263 ± 302 mL in 94 patients without the allele; P > 0.2) or maximum hematocrit decrease (11 ± 4.3 mL in 5 patients with the allele compared with 9.7 ± 3.2 mL in 94 patients without the allele; P > 0.2).

Discussion

We sought to determine whether the risk for venous thrombosis is significantly increased in patients with the factor V Leiden mutation who undergo major hip or knee surgery and receive antithrombotic prophylaxis. If risk is increased in these patients, routine preoperative screening for the presence of the factor V Leiden mutation and intensification of prophylaxis for affected patients may be indicated. Despite an association of the factor V Leiden mutation with venous thromboembolism in other patient populations, our data show that factor V Leiden genotype is not a significant risk factor for venographically determined deep venous thrombosis in patients who undergo hip or knee replacement and receive reasonable antithrombotic prophylaxis. The upper limit of the 95% CI for the observed rate of venous thrombosis in patients with the mutation is 47%. Therefore, our initial hypothesis that the rate of thrombosis in the overall group of patients is 50% to 80% is unlikely to be correct. The possibility that the factor V Leiden mutation may be associated with a markedly increased thrombotic risk in more narrowly defined categories of patients receiving antithrombotic prophylaxis was excluded in our secondary analyses of some subgroups (patients who had knee surgery or received heparin prophylaxis) but not others (patients who had hip surgery or received warfarin prophylaxis), largely on the basis of numbers of at-risk patients in each subgroup.

The overall incidence of venographically detectable deep venous thrombosis in our patient population was similar to that in previous reports: 27% in patients undergoing hip replacement surgery who received warfarin prophylaxis (compared with 24% in a recent meta-analysis [18]) and 15% in patients who received heparin (compared with 16% to 24% [18]). Moreover, the 43% incidence in patients undergoing knee replacement surgery with warfarin prophylaxis is similar to the reported range of 37% to 52% [31, 32], and the 22% incidence that we found in heparin-treated patients is similar to the reported range of 19% to 37% [31-34].

Few studies have examined the thrombotic risk presented by the factor V Leiden mutation in an acute postoperative setting. When venous thromboembolic events secondary to neoplasia or various surgical procedures were separately analyzed in a recent study based on samples from the Physician's Health Study [10], the excess risk associated with the factor V Leiden mutation was not statistically significant. Our study was designed for sensitive detection of whether a genetic risk factor can amplify a coexisting surgical risk: We included a larger number of patients limited to a specific type of surgery known to be associated with high thrombotic risk even with standard antithrombotic prophylaxis. Furthermore, because of the prospective design of the studies from which our patient samples were derived, all patients underwent venography, the most sensitive diagnostic end point for thrombosis. Most venographically detectable deep venous thromboses that occur after hip or knee replacement surgery are not clinically evident (for instance, venography was performed before the scheduled time because of clinical suspicion of deep venous thrombosis in only 2.5% of patients in two studies [22, 23]). We reasoned that if the presence of the factor V Leiden mutation amplified the thrombotic risk associated with surgery, this strategy would provide the best opportunity for detection of this interaction.

The lack of an enhanced postoperative thrombotic risk caused by the factor V Leiden mutation contrasts with the high rate of clinically evident thrombosis (33%) reported in patients with deficiency of ATIII, protein C, or protein S in lower-risk surgical settings [35]. It is therefore possible that a patient with the factor V Leiden mutation may be at lower risk for thrombosis than a patient with other hereditary causes of thrombophilia. However, the different outcome in the previous study may reflect selection of patients who had already had a thrombotic episode; in our study, healthy patients were screened for thrombotic risk factors. Recent evidence suggests that apparent differences in thrombotic tendency between patients with the factor V Leiden mutation and patients with protein C deficiency may be caused by patient selection factors [36].

Antithrombotic prophylaxis may have obscured the excess risk associated with the factor V Leiden mutation in major hip or knee surgery. Even with standard antithrombotic prophylaxis, however, venographically detectable thrombosis in these patients is still common. The important clinical issue is whether the factor V Leiden mutation amplifies this residual risk, because antithrombotic prophylaxis is accepted as standard practice for these patients. Of interest, our data linking a history of venous thrombosis with risk for postoperative thrombosis suggest that in contrast to the factor V Leiden mutation, other risk factors (genetic or acquired) may produce a thrombotic tendency that is not obscured by antithrombotic prophylaxis in this postoperative setting.

Our results are consistent with those of studies that show that in patients primarily in nonsurgical settings, the thrombotic risk attributable to the factor V Leiden mutation is less evident for secondary thrombosis [10, 37] than for recurrent idiopathic thrombosis [11]. The dependence of factor V Leiden-associated thrombotic risk on the underlying clinical setting is underscored by studies that show a lack of association of the factor V Leiden mutation with arterial thrombosis [10, 38], pulmonary embolism [39], or thrombosis accompanied by a lupus anticoagulant [40] in patients who did not have surgery. The temporal relation between the factor V Leiden mutation and thrombotic risk that we considered differs from that in studies in which the thrombotic risk of the variant factor V gene was assessed over several years of follow-up [11]. The lack of predictive value of the factor V Leiden genotype for thrombosis in patients having major hip or knee surgery may be related to the rapid time course in which deep venous thromboses evolve in these patients. The effects of the genetic risk factor may only become evident with time as the patient is repeatedly exposed to relatively infrequent procoagulant stimuli that interact with the factor V Leiden phenotype, resulting in venous thromboembolism.

Although it has been suggested that the factor V Leiden genotype may persist in the population by minimizing severe bleeding complications [6], this genetic factor does not ameliorate the bleeding diathesis of patients with severe hemophilia A [41, 42]. Our finding that a factor V Leiden allele is not associated with diminished bleeding after hip or knee replacement surgery is consistent with the conclusion that the factor V Leiden mutation has little, if any, effect on clinically evident bleeding.

Several considerations support the validity of combining the patient groups for evaluation. The inclusion criteria for patients in each of the six studies were similar, and all studies used venography (the gold standard [43, 44]) as the diagnostic end point for venous thrombosis. Venography was performed approximately 1 week after surgery because most deep venous thromboses after orthopedic surgery occur in the immediate postoperative period [45]. We did not address the risk for late thrombotic events occurring after discharge [46]. The main rationale for grouping these studies was the known high risk for venous thrombosis after either hip or knee replacement. We assessed the comparability of individual studies by subgroup analysis that compared patients in different studies who had similar prophylaxis and type of surgery. Three surgery-treatment combinations were included in more than one study and contained enough patients (n > 25) for analysis of the incidence of deep venous thrombosis. In each case, the incidence of deep venous thrombosis did not differ among comparable studies of patients whose factor V Leiden status was evaluated as part of the combined analysis. Although few patients had venography performed only on the limb that was operated on, unilateral thrombosis occurring only in the leg that was not operated on is uncommon [47]. Modeling with the inclusion of additional thrombotic events in the patient group that underwent unilateral venography or removing all patients who had unilateral venography from the analysis did not affect our conclusions.

We conclude that heterozygosity for the factor V Leiden mutation does not increase the risk for post-operative venous thrombosis after high-risk surgery in patients who receive antithrombotic prophylaxis. Our data do not justify routine screening of patients for the factor V Leiden mutation before orthopedic surgery for the purpose of antithrombotic prophylaxis intensification in patients with the mutation.

Drs. Ginsberg and Crowther: McMaster University Medical Center, 1200 Main Street West, HSC 3W11, Hamilton, Ontario L8N 3Z5, Canada.

Dr. Francis: Box 710, University of Rochester Medical Center, 601 Elmwood Avenue, Rochester, NY 14642.