Cancer mortality according to lipid-lowering drugs and lipoproteins in a general population

Abstract

Objective:The beneficial effect of lipid-lowering drugs (LLD) on cardiovascular risk is established, but long term safety data remain scarce. Our aim was to assess 10-year risk of cancer mortality according to blood lipoprotein levels and LLD exposure, in a general population.Methods:Our analysis was based on the Third French MONICA survey on cardiovascular risk factors (1994–1997). Participants were randomly recruited from the general population of three French areas and were aged 35–64 years. Subjects with a history of cancer at baseline were excluded from the analysis. Vital status and cause of mortality were obtained 10 years after inclusion.Results:There were 3262 participants and 177 deaths were recorded over the 10-year period (78 due to a cancer). The sample comprised 64% of normolipidaemic, 25% of untreated dyslipidaemic and 11% of dyslipidaemic subjects treated with LLD (4% statins, 6% fibrates and 1% other hypolipidaemic drugs). After adjustment for centre, age, gender, smoking, gamma-glutamyl transpeptidase and mean corpuscular volume, the hazard ratios (HR) for cancer mortality in subjects with non-HDL cholesterol <3.5 mmol/L (135 mg/dL) and in those with HDL cholesterol <0.90 mmol/L (35 mg/dL) were 2.74 (95% confidence interval: 1.66–4.52, p < 0.001) and 2.83 (1.62–4.96, p < 0.001), respectively. The adjusted HR for cancer mortality was 0.31 (0.11–0.86, p = 0.025) in people on LLD compared to untreated subjects.Conclusions:In the present study, we confirm the significant association between low cholesterol and cancer mortality without finding any harmful signal regarding cancer risk associated with the use of LLD. The main limitations are remaining baseline differences between treated and untreated subjects (due to the observational design but minimized by the use of extensive adjustments and propensity score methods), and the lack of re-assessment of LLD exposure and cholesterol levels during follow-up, possibly leading to a misclassification bias.