This analysis was approved by the Ethics Committee of the Charité—Universitätsmedizin Berlin, Germany, and all patients gave informed consent for DNA isolation before genetic testing. We analyzed medical records of 144 patients with ischemic stroke and of 62 patients with either MI (n = 44) or PAOD (n = 18). These patients had been referred to the Institute of Transfusion Medicine, Charité—Universitätsmedizin Berlin, for thrombophilia testing. The following information was assessed from medical records: patients’ age, sex, cardiovascular risk factors, and history of VTE.

All patients in this study were referred for testing for hereditary thrombophilia, which is part of routine care performed at the Institute of Transfusion Medicine, Charité—Universitätsmedizin Berlin, Germany. Heterozygous and homozygous variants of the following gene polymorphisms were distinguished from wild type: plasminogen activator inhibitor type I (PAI-1) [−675 insertion/deletion (5G/4G)], tissue plasminogen activator (PLAT) (intron h deletion/insertion), F5 (1691G>A) (Leiden) and HR2 haplotype (4070A>G), F2 (20210G>A), FGB (455G>A), TFPI (536C>T), thrombomodulin (THBD) (127G>A), FVII (F7) (R353Q), FXIII (F13A) (Val34Leu), methylenetetrahydrofolate reductase (MTHFR) (677C>T), angiotensin converting enzyme (ACE) (intron 16 insertion/deletion), and glycoprotein Ia/IIa (ITGA2) (807C>T).

Genomic DNA was extracted from whole blood using GenoPrepTM Cartridges B and the GenoMTM-6 system (GenoVision, Vienna, Austria). DNA samples were analyzed by an amplification refractory mutation system using allele-specific primer pairs as previously described (11). All primers were synthesized by TIB Molbiol (Berlin, Germany). The polymerase chain reaction was carried out in a thermocycler (GeneAmp PCR System 9700, Applied Biosystems, Darmstadt, Germany) using the following temperature profiles: denaturation (96°C for 2 min), followed by 10 cycles (96°C for 15 s, 65°C for 60 s) and 20 cycles (96°C for 10 s, 61°C for 50 s, and 72°C for 30 s). Amplification products were separated using electrophoresis and made visible using electrophoresis and UV transillumination.

To assess the overall burden of thrombophilia per patient, a prothrombotic score was calculated dividing the number of risk-increasing polymorphisms (heterozygote or homozygote) minus the number of risk-lowering polymorphisms (heterozygote or homozygote) by the number of all 13 polymorphisms tested. Variant alleles of F7 and F13A were considered to be protective. This prothrombotic score ranges theoretically between 0% (none of tested variants are prothrombotic) and 100% (all of the tested variants are prothrombotic), with the remark that only a subset of values can be observed as a maximum of 13 markers have been tested in these patient groups. The score was categorized in tertiles based on the given distribution in the stroke cohort. Overall, 19 (10.2%) patients had to be excluded from this analysis because of incomplete data (defined as less than 12 of 13 hemostatic polymorphisms tested). The excluded patients comprised of 14 stroke patients, 3 MI patients, and 2 POAD patients.

For categorical data, absolute and relative frequencies were computed. In case of continuous variables with nearly normal distribution, the mean, SD, minimal and maximal values, otherwise median, quartiles, minimal and maximal values were determined. The Fisher’s exact test was used to compare proportions for dichotomous outcomes between independent groups. Distribution of variant alleles was explored comparing carrier vs. non-carrier status by Fisher’s exact test. A direct comparison between the MI/PAOD group and the stroke group with regard to the prothrombotic score was performed using binary logistic regression, yielding odds ratios (OR) and corresponding confidence intervals (CI). To do this, we a priori categorized the risk score in tertiles, with the middle category as reference category. Nominal scaled outcomes were analyzed using the Mann–Whitney test.

Baseline characteristics of stroke patients as well as patients with MI or PAOD are depicted in Table 1. Stroke patients were significantly younger and more often female than MI or PAOD patients. Compared to patients with MI or PAOD, stroke patients suffered less often from diabetes, arterial hypertension, hyperlipoproteinaemia, or coronary arterial disease.

In a tertile analysis, a low prothrombotic score (0–31.1%) was found in 22.3% (n = 29) of stroke patients and 36.8% (n = 21) of MI/PAOD [OR 2.66 (95% CI 1.14–6.21)] (Table 2), indicating that more MI/PAOD patients have a lower prothrombotic propensity then stroke patients. This difference was also observed if patients with a prior VTE were excluded from the analysis [OR 2.82 (95% CI 1.17–6.82)].

Comparing the distribution of individual variant alleles in stroke patients (referred to the thrombophilia clinic) to MI/PAOD patients (referred to the same thrombophilia clinic) there was no statistically significant difference across groups (Table 3). In more detail, protective variant F13A Val34Leu was numerically more frequent in stroke patients than in MI/PAOD patients [44.8 vs. 33.1%; OR 1.58 (95% CI 0.85–2.94)], whereas F7 R353Q was equally distributed [22.8 vs. 20.9%, OR 1.02 (95% CI 0.49–2.13)]. F5 1691G>A was found in 18.3% of stroke patients and 14.5% of MI/PAOD patients [OR 1.32 (95% CI 0.58–3.01)], and F2 20210G>A was present in 3.5% of stroke patients and 9.7% of MI/PAOD patients [OR 0.34 (95% CI 0.10–1.17)]. TFPI 536C>T as well as THBD 127G>A was found in one stroke patient (0.7%) and one MI/PAOD patient (1.6%). These results show that the prevalence of single polymorphisms was too low to draw strong conclusions about these individuals.

Some limitations have to be taken into account in the interpretation of these results. First, the sample size is limited, as we included only patients referred to a tertiary referral center for genetic testing on the presence of thrombophilia. Thus, a potential referral bias has to be taken into account. This is a major problem of similar studies, which use population or healthy controls. However, we believe that a potential referral bias is minimized in the present analysis, as the reason for referral is assumed to be similar for both groups. Furthermore, the lower number of patients with MI or PAOD limits the power of our analysis. Moreover, combining MI and PAOD patients for analysis might have introduced bias as one of the two combined groups might have a different distribution. However, this bias would dilute any true association, rendering the reported odds ratio of 2.7 an underestimation of the real association. Moreover, stroke is a heterogeneous disease, and a prothrombotic state may play a stronger role in (young) stroke patients without or with just a few established cardiovascular risk factors. Of note, stroke patients referred for thrombophilia testing do not represent the overall cohort of stroke patients. As follow-up data were not available, we cannot answer the question whether risk assessment by routine genetic testing can reduce recurrence of arterial thrombosis. Finally, by reporting a wide variety of gene polymorphisms, not all polymorphisms were analyzed in every single patient with stroke, MI, or PAOD, respectively.