A study of the effect of incidental drug exposures, body mass index and diabetes on survival from melanoma

Introduction: Drugs intended to treat incidental medical conditions could moderate host-tumour interaction and therefore melanoma survival. Method: Drug exposure data were collected from the 2184 newly diagnosed melanoma patients in the Leeds Melanoma Cohort (recruited 2000-2012) and their primary care physicians. An ever-never analysis and drug usage at diagnosis of melanoma (including 12 months prior) were chosen as the most applicable analysis methods (overall and sex stratified). The effects of exposure to different classes of drugs on MSS and overall survival (OS) were then assessed using Cox Proportional Hazards models whilst adjusting for confounding variables including diabetes and BMI, firstly in unadjusted models followed by adjustment for known predictors of MSS in a multivariate model. Results: For most drugs there were no statistically significant effects on MSS. The drugs that I ultimately chose to look at in detail were aspirin, simvastatin and metformin. Whilst adjusting for age and Breslow thickness, women who had ever taken aspirin were significantly less likely to die from their melanoma compared with those who never used the drug at any point in their lifetime with hazard ratios (HR) for MSS of 0.51 (95% CI: 0.30-0.87, P= 0.014) compared to men with an HR (MSS) 0.99 (95% CI: 0.71-1.37, P= 0.948). With both ever/never use of simvastatin and at diagnosis (including 12 months prior) analysis, when adjusting for age and Breslow thickness, men had a significantly reduced risk of death from melanoma with HRs of 0.54 (95% CI: 0.37-0.79, P=0.002) and 0.57 (95% CI: 0.38-0.85, P=0.006) respectively when compared to females who had HRs of 1.22 (95% CI: 0.82-1.83, P=0.327) and 1.21 (95% CI: 0.79-1.86, P = 0.379). Metformin usage was negatively associated with MSS in individuals with primaries on the trunk, which was used here as a surrogate marker for BRAF mutated tumours with an HR of 3.87 (95% CI 1.29-11.57, P= 0.02) for chest primaries. Conclusion: The associations seen in my thesis require further validation in larger international data sets as well as examination of biological models to assess if these represent real effects or whether confounding factors are responsible for these changes. I would propose that future studies looking at factors influencing melanoma survival should consider stratifying their findings by sex.