Association between 19 medication use and risk of common cancers: A cross-sectional and Mendelian randomisation study

Abstract

Background: Previous studies have yielded inconsistent results concerning drug use and the risk of cancers. We conducted a large-scale cross-sectional study and a two-sample Mendelian randomisation (MR) study to reveal the causal effect between the use of 19 medications and the risk of four common cancers (breast, lung, colorectal, and prostate).

Methods: We obtained information on medication use and cancer diagnosis from National Health and Nutrition Examination Survey participants. After propensity score matching, we conducted survey-weighted multivariate logistic regression and restricted cubic spline analysis to assess the observed correlation between medication use and cancer while adjusting for multiple covariates. We also performed MR analysis to investigate causality based on summary data from genome-wide association studies on medication use and cancers. We performed sensitivity analyses, replication analysis, genetic correlation analysis, and reverse MR analysis to improve the reliability of MR findings.

Results: We found that the use of agents acting on the renin-angiotensin system was associated with reduced risk of prostate cancer (odds ratio (OR) = 0.42; 95% confidence interval (CI) = 0.27-0.63, P < 0.001), and there was a nonlinear association of 'decrease-to-increase-to-decrease' (P < 0.0001). The random-effects inverse variance weighted (IVW) model-based primary MR analysis (OR = 0.94, 95% CI = 0.91-0.97, P = 0.0007) and replication MR analysis (OR = 0.90, 95% CI = 0.85-0.96, P = 0.0006) both provided robust evidence of the causality of genetic liability for the use of agents acting on the renin-angiotensin system on a decreased risk of prostate cancer.

Conclusions: Our study provides robust evidence that the use of drugs acting on the renin-angiotensin system can reduce prostate cancer risk. Given the high prevalence of prostate cancer, these findings have important implications for drug selection and prostate cancer prevention in patients with cardiovascular disease.

Figure 1

Outline of the study design. SNPs – single nucleotide polymorphisms, GWAS – genome-wide association studies

Figure 2

Survey-weighted multivariate logistic regression on the association of 19 medication use with cancers in the NHANES study. All models were adjusted for age, sex, BMI, race, educational attainment, family income, drinking status, and smoking status. Only the findings with significant association (P < 0.05) are shown in the figure. The colour of each square is related to the significance of the association, and the legend on the right is used as a reference. The number above the box is the odds ratio, and the P-value is displayed in parentheses below. BMI – body mass index, NHANES – the National Health and Nutrition

Figure 3

A primary Mendelian randomisation study identified a significant causal relationship between medication use and cancers. CI – confidence interval, MR method – Mendelian randomisation method, N SNPs – number of SNPs, OR – odds ratio

Figure 4

Scatterplot for the significant Mendelian randomisation association (P < 0.05) between medication use and cancers. Panel A. Antithrombotic agents and breast cancer. Panel B. Antithrombotic agents and lung cancer. Panel C. Antihypertensives and lung cancer. Panel D. Thyroid preparations and lung cancer. Panel E. Anti-inflammatory and antirheumatic products, non-steroids and colorectal cancer. Panel F. Agents acting on the renin-angiotensin system and prostate cancer. Horizontal and vertical lines represented standard errors for medication use and cancer, respectively. The slope of each line corresponds to the estimated MR effect from different methods. SNP – single nucleotide polymorphism

Figure 5

Meta-analysis of significant association (IVW derived P < 0.05) identified in primary MR analysis. CI – confidence interval, OR – odds ratio

Figure 6

Causal evidence between agents acting on the renin-angiotensin system and prostate cancer was established by the cross-sectional study and MR analysis. Panel A. Analysis of the shape of the nonlinear association between agents acting on the renin-angiotensin system and prostate cancer using restricted cubic spline based on NHANES database. The solid blue line is the estimated odds ratio, and the shaded blue area is the 95% confidence interval. Panels B and (C). Two-sample Mendelian randomisation reveals causal evidence for agents acting on the renin-angiotensin system and prostate cancer. The forest plots illustrate the standardised odds ratio (OR) (95% confidence interval (CI)) for each two-sample MR method. PRACTICAL – Prostate Cancer Association Group to Investigate Cancer Associated Alterations in the Genome.