A High Factor II/Factor X Functional Ratio Is Not a Useful Predictor of the FII G20210A Gene Mutation in Thromboembolic Patients Undergoing Oral Anticoagulant Treatment

A G→A transition at nucleotide 20210 in the prothrombin gene has recently been associated with venous thromboembolism in a Dutch population (1). The prevalence of this genetic variation in Western countries is 5–15% among thrombotic patients and 1–5% in healthy controls (2)(3)(4)(5). The main pathogenic mechanism appears to be the increase of plasma prothrombin (FII) because many carriers of the FII20210A mutation have hyperprothrombinemia by functional assays. However, increased FII is not specific for this mutation (1)(3)(6). Even some carriers do not exhibit hyperprothrombinemia because of the variability in vitamin K metabolism or hepatic function. A functional, rapid, low-cost assay, preferably not influenced by the oral anticoagulant (OA) (7) required by many patients, would be desirable for screening purposes. Because factor X (FX) has a half-life close to that of FII, we used FX activity to control for changes in concentrations of vitamin K-dependent factors. The ratio of FII activity to FX activity (FII/FX) was used to screen for subjects with high FII activity during OA therapy. We studied 123 outpatients objectively diagnosed with venous thromboembolism (59 men and 64 women; mean age, 63 years; range, 17–87 years) to identify a reliable screening test for this genetic anomaly. All were receiving OA (120 were receiving acenocumarol, and 3 were receiving warfarin). Informed consent was required. The mean international normalized ratio (INR) was 2.43 (0.76) with 66% of the patients in the therapeutic range (INR, 2.0–3.5) at the time of sampling. The INR range was 1.3–2.0 in 26% of the patients, and none had an INR >4.5. The subjects were classified as carriers and non-carriers after a standard PCR assay for this prothrombin mutation (1). For the standard functional clotting tests (FII:C/FX:C), we avoided the large oscillations in the INR (especially during the first month of anticoagulant therapy). We used deficient plasmas and recombinant thromboplastin (Innovin®) from Dade-Behring® on a Sysmex® CA-6000® coagulometer. We calculated the imprecision for plasmas with an equivalent range of activity (<500 units/L). The intraassay CVs were 4.0% and 2.5%, and the interassay CVs were 4.6% and 3.0% for the FII and FX activities, respectively. During OA therapy, FII:C remains higher than FX:C (8). We observed activities of 320 (120) vs 160 (90) units/L, respectively (P <0.0001). The FII20210A allele was identified in 17 patients (13.8%; 16 heterozygous and 1 homozygous) and ruled out in 106 (86.2%). No statistical difference in age, gender, or INR was found between the two groups. FII:C was 340 units/L [95% confidence interval (CI), 290–400 units/L] among the 20210A carriers and 310 units/L (95% CI, 290–340 units/L) among the non-carriers (P = 0.36). During OA therapy, the FII:C seems useless as a diagnostic tool. FX:C was not significantly lower among the carriers [140 (95% CI, 110–170) units/L vs 170 (95% CI, 150–180) units/L; P = 0.31], but the ratio was higher [2.54 (95% CI, 2.26–2.82)] than in non-carriers [2.01 (95% CI, 1.92–2.10); P <0.0001]. The sensitivity was 0.82, and the specificity was 0.58 for a cutoff of 2.1. A good negative predictive value of 0.95 corresponded to an unacceptable positive predictive value (0.25; see Table 1 ). The 2.40 cutoff showed the best likelihood ratio of a positive result (3.1) but correctly classified only 77% of the patients. The nonparametric estimation for the area under the ROC curve was 0.73 (9). In summary, although a non-anticoagulated population should be evaluated to obtain definitive conclusions, many thromboembolic patients are referred to the outpatient clinic for investigation while being treated with OA, and many recurrent cases may require lifelong treatment. During stable OA therapy, FX activity could be a good marker of FII oscillations. In these subjects, the FII:C/FX:C ratio was significantly higher when the mutation FII20210A existed (2.54 vs 2.01; P <0.0001), but disappointing results were obtained when we attempted to use it for a diagnostic purpose. Operational parameters, predictive values, and ROC curve for several cutoff points of the FII/FX ratio. LR+, positive likelihood ratio; LR−, negative likelihood ratio; PPV, positive predictive value; NPV, negative predictive value. Operational parameters, predictive values, and ROC curve for several cutoff points of the FII/FX ratio. LR+, positive likelihood ratio; LR−, negative likelihood ratio; PPV, positive predictive value; NPV, negative predictive value. We thank Ana Arias, Emilia Adán, and Celia Rios for technical support.