Background: Geographic variation in melanoma incidence is not fully explained by total ultraviolet (UV) dose. Ecological evidence from Nordic populations has implicated the UV-A:UV-B ratio as a potential driver of latitudinal variation in melanoma incidence, but this has not been examined in a southern hemisphere, high-UV setting. Australia’s broad latitudinal range and historically low sunbed use provide an informative context to assess whether the UV-A:UV-B ratio explains geographic variation in melanoma incidence independently of total UV irradiance.
Methods: We obtained modelled, age-adjusted incidence ratios for invasive melanoma at Statistical Area Level 2 (SA2) from the Australian Cancer Atlas 2.0 for 2008-2015. Satellite-derived daily surface UV-A and UV-B dose estimates from NASA’s GOME-2 instruments were extracted for SA2 geographic centroids over the same period. For each SA2-year, we calculated mean daily UV dose measures and UV-A:UV-B ratios. We used linear regression models to examine the spatial variance in melanoma incidence rates explained by UV exposure variables, for persons overall and by sex, in both univariable and multivariable analyses. We conducted sensitivity analyses to assess the stability of associations between annual melanoma incidence and seasonal UV (warmer months: October-March; cooler months: April-September).
Results: The mean daily UV-A:UV-B ratio was the strongest single predictor of geographic variation in melanoma incidence, accounting for 20.5% of the variance in persons, substantially exceeding that explained by mean daily total UV (7.3%), UV-A (7.2%), or UV-B dose (9.7%) individually. The association was inverse, with areas having higher UV-A:UV-B ratios associated with lower melanoma incidence. After mutual adjustment, both the UV-A:UV-B ratio and total UV dose remained independent predictors of melanoma incidence, jointly explaining 24.5% of the variance. This explained variance was consistently higher in men than in women. Seasonal analyses showed greater variance explained using the UV measurements from cooler months than the measurements from warmer months.
Conclusions: The spectral composition of ambient solar UV radiation, characterised by the UV-A:UV-B ratio, is associated with geographic variation in melanoma incidence across Australia, independently of total UV. Although ecological in nature, these findings provide some support for a relatively greater role for UV-B than UV-A in population-level melanoma risk.