The SNPs affecting warfarin dosing are not associated with any clinical features of PCa nor risk of PCa death. We found no association between prevalence of warfarin use and SNP mutations affecting warfarin dosing. Overall risk of death was non-statistically elevated among warfarin users compared to non-users. This is likely mainly driven by non-cancer deaths, as the risk increase was less pronounced, though still observable for risk of cancer death. Since warfarin metabolism does not associate with cancer death risk and also does not modify the risk between warfarin use and cancer death, the result suggests that warfarin or its anticoagulant effects do not play a direct role in PCa progression.
Metastatic cancer is known to increase the risk of venous thrombosis through changes in the coagulation cascade (30–32). Therefore, elevated cancer mortality in warfarin users observed in our study and also in previous studies is likely affected by reverse causation i.e. advanced cancer is the cause of the warfarin use in order to manage cancer-associated thrombosis. Clinically, treatment and prevention of thrombosis during chemotherapy for metastatic cancer is preferably done with LMWHs instead of warfarin (29). If unaccounted for, this could create bias leading to underestimation of cancer mortality among warfarin users and overestimation among LMWH users. We aimed to minimize this bias by keeping warfarin users in the user category even if the drug purchases ceased. In such analysis there was no protective association with warfarin use but rather only the observed increase in risk of death.
Warfarin metabolic genotype is associated with maintenance dose requirements, accounting for up to 40–45% of individual variability (33). Time in Therapeutic Range (TTR) should be a minimum of > 60% to achieve the benefit of warfarin use, prevention of thromboembolic events (34). Besides the genetic factors, INR fluctuations occur due to various reasons such as INR monitoring frequency, dosage adjustments, foods high in vitamin K, alcohol consumption, reduced oral intake due to non-adherence, malignancy, drug interactions and congestive heart failure. Every increase in warfarin dosage fluctuation by 10% is associated with an HR of 1.58 for bleeding or thromboembolism (35). Despite optimally adjusted anticoagulation a small proportion of patients develop venous thromboembolism, especially those with prior thromboembolism or extensive metastatic cancer (36).
The SNP carriers could supposedly be more susceptible to cancer-associated thrombosis despite warfarin use due to dose fluctuations. Nevertheless, our results do not support this hypothesis, as warfarin metabolism SNPs were not found to modify the risk associations between warfarin use and cancer outcomes.
The strength of this study is that we were able to combine genetic information on SNP carrier status and very detailed information on warfarin use through Finnish national registries and two population-based cohorts. Our data allowed examination of PCa risk and survival, overall cancer mortality and all-cause mortality. To our knowledge this is the first study to be able to examine these associations. We were able to analyze warfarin metabolism through the SNP carrier status. By connecting warfarin metabolism to warfarin user status we were able to draw conclusions of whether warfarin use affects PCa prognosis and outcome.
The limitation of our study is relatively small study population, causing low statistical power to study the risk associations among men using warfarin especially among combined carrier men. Also, number of PCa deaths was small, limiting our ability to estimate PCa-specific survival. The Finns are almost exclusively Caucasian and a genetic isolate (37–39), therefore generalizability of our results to other ethnicities is uncertain and need to be confirmed in other, more diverse study populations.