Do Age at Diagnosis, Tumour Thickness and Tumour Site Explain Sex Differences in Melanoma Survival? A Sequential Causal Mediation Analysis Using Cancer Registry Data

Abstract

Background: Previous research has shown that women diagnosed with melanoma have better survival than their male counterparts, but little is known about the underlying mechanisms by which sex affects melanoma survival. Aim: The aim of this study was to quantify the contribution of age at diagnosis, tumor thickness and tumor site to sex differences in 1-year and 5-year melanoma-specific survival. Methods: We conducted a population-based study using cancer registry data including 6,009 men and 5,232 women aged 15-70 years with first primary melanoma diagnosed between 2007 and 2015 in Victoria, Australia. We excluded cases notified via death certificate only. Deaths to the end of 2015 were identified through linkage to the Victorian and national death registries. We decomposed the total effect of sex on melanoma-specific survival into three possible pathways, examining the mediating role of age at diagnosis, tumor thickness and tumor site. The natural indirect effects through the mediators were assessed using weighted, sequential causal mediation analysis. Results: Compared with women, there were 60 (95% confidence interval (CI) 34 to 91) and 201 (95% CI 135 to 267) additional deaths per 10,000 in men within 1 year and 5 years following diagnosis, respectively. Within 1 year following diagnosis, the indirect effect through age at diagnosis was estimated to be 4 (95% CI -2 to 9) additional deaths per 10,000 in men compared with women, explaining 7% of the total effect of sex, while a considerable proportion of the total effect (79%) explained by melanoma thickness [47 (95% CI 32 to 66) additional deaths per 10,000 in men], excluding the potential influence of age at diagnosis on thickness. Similarly, within 5 years after diagnosis, about 13% of the total effect was estimated to be explained by age at diagnosis [26 (95% CI 15 to 38) additional deaths per 10,000 in men] and 86% by tumor thickness [174 (95% CI 142 to 207) additional death per 10,000 in men]. There was no indirect effect through tumor site within 1 year and 5 years postdiagnosis. Conclusion: The lower survival from melanoma of men relative to women appears to be explained primarily by tumor thickness rather than age at diagnosis. Our findings highlight the need for earlier detection of melanoma in men.