Abstract
Background: Thrombosis patients are often evaluated for deficiency in protein C (PC, PROC), but the phenotype of PC deficiency remains poorly defined. Older studies are small and have relied on family-based or blood donor cohorts that are plagued by inherent residual biases and confounders. This work has generated widely discrepant effect size estimates for the venous thromboembolism (VTE) risk associated with PC deficiency and have not elucidated the role of PC deficiency in arterial thrombosis. Moreover, the relative prevalences and clinical impacts of quantitative (Type I) and qualitative (Type II) PC deficiency has never been studied at scale.
Methods: We analyzed paired genomic and phenotypic data from the UK Biobank (UKB, N=454,209) and NIH All of Us (AoU, N=204,006) biorepositories. Rare (minor allele frequency ≤0.1%) germline coding variants in PROC were each assigned an AlphaMissense score (AMS) between 0.0 and 1.0, with higher scores reflecting greater predicted likelihood of damage to protein function. AlphaMissense has designated AMS ≥0.564 as the cutoff for predicted deleterious variants, high-confidence loss-of-function variants (nonsense, essential splice site, or frameshift mutations) are assigned an AMS of 1.0. We interrogated the relationship between PROC variant carrier status and risk of thrombosis using Firth's logistic regression modeling followed by trans-cohort random effects meta-analysis. Additionally, circulating protein C antigen levels were measured in 44,567 UKB participants (Olink® Explore 3072).
Results: We identified 406 carriers of 101 unique coding variants (AMS ≥0.564) in PROC (99.9% heterozygous). Protein C antigen levels were significantly lower in variant carriers compared to non-carriers (P<0.0001). Variant carrier status was significantly associated with VTE at both AMS ≥0.564 (OR=4.86, 95% CI: 3.71-6.36, P<0.0001) and AMS=1.0 (OR=11.24, 95% CI: 5.60-22.56, P<0.0001). Increasing AMS cutoffs corresponded to higher VTE risk, reflecting a stronger biological impact from variants more likely to impair protein C function. Adjustment for VTE polygenic risk score and factor V Leiden and prothrombin gene mutation G20210A carrier status did not significantly change our findings, indicating that rare germline variants in PROC operate independently of common inherited risk factors for VTE. PROC variant carriers did not experience significantly increased risk for any of the arterial thrombosis phenotypes evaluated (myocardial infarction, non-cardioembolic ischemic stroke, and peripheral artery disease). UKB participants with plasma protein C levels <60% had increased VTE risk (OR=1.976, 95% CI: 1.278-2.922, p=0.003), but this effect was no longer significant when variant carriers were excluded from the analysis (OR=1.573, 95% CI: 0.929-2.491, p=0.089). We next assessed deleterious PROC variants based on their functional mechanism as reported in the literature. Prior consensus holds that Type I inherited PC deficiency is far more common than Type II (roughly 85% Type I vs. 15% Type II), but this conclusion has never been tested at the level of germline genetics or on a large scale in the general population. Across both biorepository datasets, we identified 1,705 carriers of a previously described PROC variant. Unexpectedly, we found that reported type II variants were collectively far more prevalent (60.5%, 1,031/1,705) than type I (39.5%, 674/1,705). Despite their differing mechanisms, variants of each type appeared to confer increased VTE risk when grouped and evaluated together: Type I (OR=3.49, 95% CI: 2.76-4.45, P<0.0001) and Type II (OR=2.38, 95% CI: 1.92-2.94, P<0.0001). This finding suggests that the majority of Type II carriers express a low-penetrance phenotype that may go undetected and/or that annotation for a substantial number of previously reported Type II variants is inaccurate.
Conclusions: Genetically-defined PC deficiency is significantly associated with increased risk of venous but not arterial thrombosis. Carriers of high-confidence loss-of-function PROC variants (AMS=1.0) constitute a separate, higher-risk category. Challenging prior dogma, we found that Type II PROC variants are far more common than Type I variants in the general population. Spot plasma PC levels in the absence of a deleterious PROC variant weakly predict VTE risk, suggesting that genetic testing is of value in at least some individuals.
