Impact of metformin on the incidence of thrombotic events in ­older patients with polycythemia vera and essential thrombocythemia in the United States

Abstract

Abstract XM, NP co-senior authors Background Patients (pts) with Philadelphia chromosome negative myeloproliferative neoplasms (MPNs) and pre-existing cardiovascular risk factors, such as diabetes mellitus (DM), have a high risk for thrombosis (Lekovic et al. Eur J Haematol 2015). Therapeutic phlebotomies, antiplatelet, and cytoreductive therapies aim to mitigate this risk but fall short of eliminating it. Metformin (MET) was recently described to have protective effects against the development of MPNs (Oevlisen et al. Blood 2022) but clinical outcomes among MET users with MPNs are unknown. Methods We conducted a retrospective cohort study using the Surveillance, Epidemiology, and End Results-Medicare linked database to evaluate the impact of MET on thrombotic events in older pts with MPNs. Eligible pts were diagnosed with polycythemia vera (PV) or essential thrombocythemia (ET) between 2008-2019, aged between 66-99 years at diagnosis, continuously enrolled in Medicare Parts A, B and D from diagnosis to end of follow-up, and followed for ≥6 months after MPN diagnosis. MET users were defined as pts who received MET prescriptions within 6 months of MPN diagnosis and continued treatment for ≥180 days after diagnosis. All MET users had a diagnosis of DM. DM severity was classified using the DM Complications Severity Index (DCSI) as mild (0-2) or severe (>2) (Chang et al. Am J Manag Care 2012). Thrombotic events included both venous and arterial thromboses. Pts were defined as having experienced thrombosis after diagnosis if the occurrence of the first thrombotic event was recorded ≥30 days after PV/ET diagnosis to the end of follow-up. We followed pts from diagnosis through December 31, 2020, death, or change of insurance status—whichever occurred first. We used multivariable Fine–Gray proportional sub-distribution hazards regression models with death as the competing risk to evaluate the association between MET and thrombotic risk, adjusting for covariates including for DM severity. Results Among 7202 older pts with a diagnosis of PV/ET during the observation period, 5449 pts met eligibility criteria (PV, n=2435; ET, n=3014), of which 431 (7.9%) were MET users and 5018 (92.1%) were MET non-users. The majority of pts were female (62.7%), had ≥1 comorbid condition (61.3%), were classified as frail (23.6%), and had received influenza immunization (62.1%) (indicative of access to healthcare). MET users had a greater number of comorbidities and higher frailty scores than non-users (p<0.01). A total of 506 pts (9.3%) had a prior history of thrombosis with no difference between MET users and non-users (p=0.86). 1153 pts had a diagnosis of DM (431 MET users, 722 non-users), of which 846 (73.4%) had mild and 307 (26.6%) had severe DM. MET users received MET for a median of 1.92 (interquartile range [IQR]: 1.07-3.25) years, and had a median percentage of days covered of 87.9% (IQR: 66.5-97.4%) (higher number indicating treatment adherence). Overall, 1897 (34.8%) pts had a thrombotic event after diagnosis, including 150 (34.8%) MET users and 1747 (34.8%) non-users with or without a diagnosis of DM. The incidence rate of thrombosis was 110.3 cases per 1000 person-years among MET users and 165.3 among MET non-users (p=0.02). In multivariable analysis, MET use was associated with a reduced thrombotic risk (hazard ratio [HR]=0.80, 95% CI [confidence interval]: 0.66-0.97; p=0.02). When stratifying by DM severity, MET use was associated with a significantly lower thrombotic risk in pts with PV/ET and severe DM (HR=0.57, 95% CI: 0.38-0.86; p<0.01). No significant association was observed between MET use and thrombosis risk in pts with PV/ET and mild DM (HR=0.87, 95% CI: 0.68-1.11; p=0.25). The lower thrombotic risk associated with MET use was also observed within individual disease subgroups: PV (HR=0.43, 95% CI: 0.19-0.96; p=0.04) and ET (HR=0.55, 95% CI: 0.32-0.93; p=0.03). MET users without a prior history of thrombosis had a lower thrombotic risk than MET non-users with DM (HR=0.80, 95% CI: 0.65-0.99; p=0.04). There was no significant association between MET use and thrombotic risk (HR=1.05, 95% CI: 0.83-1.32; p= 0.69) among pts with DM and a prior history of thrombosis. Conclusions Our study suggests that the use of MET reduces the risk of thrombosis in older pts with PV/ET and DM, with the protection being more pronounced among pts with severe DM and without prior history of thrombotic events. Research supported by the ASH Minority Fellow Award.