Cutaneous malignant melanoma: Incidence
For cutaneous melanoma, global data are available on incidence and mortality. The global burden of disease estimates for the year 2000 (available at www.who.int/evidence/bod) used incidence and mortality estimates from Globocan 2000 to calculate the burden of disease due to melanoma.
The assessment of the burden of disease due to UVR from melanoma was derived in the current work by applying the calculated population attributable fraction estimates to these data. Population attributable fraction. The fraction of disease in the population attributable to UVR exposure has been estimated at 96% in males and 92% in females in the USA, by comparison of white and black populations. Comparison of white populations in New South Wales, Australia, with ethnically similar populations in England and Wales gives a PAF of 89% (males) and 79% (females).
Examination of ecological and individual-level studies indicates little relationship of PAF to latitude (see Appendix 3). There is also little relationship between PAFs estimated from ecologic studies and those estimated from case-control studies. As discussed in section 2.3 above, this presumably reflects both a difficulty with measuring exposure and the difficulty in finding a truly non-exposed population as the control group in epidemiological studies. We therefore did not apply a PAF which varies with latitude, but used constant PAFs for upper and lower estimates of the burden of disease from CMM, that is caused by UVR. Estimation of disease burden.
There is generally an increase in incidence of melanoma with decreasing latitude. This has been shown within the Nordic countries, the USA and Australia. However, this relationship does not persist across non-homogeneous populations – mortality from melanoma is four to six times higher in Nordic countries than in the Mediterranean countries and there is an opposite relationship of melanoma incidence to latitude in Italy.
Since melanoma is likely to be related to intermittent high intensity sun exposure, particularly in fair-skinned individuals, those at greatest risk are likely to be fair skinned people from higher latitudes who intermittently are exposed to high intensity UVR on holidays. Langford used multilevel modeling to examine the relationship between melanoma mortality and UVB exposure in several countries. He found that the United Kingdom, Ireland, Belgium and the Netherlands generally showed a positive relationship, whereas France showed very little relationship, Italy showed a negative relationship.
Germany and Denmark, while having higher rates of melanoma mortality, did not show a positive relationship of UVB exposure with mortality. Few studies have been done in dark-skinned populations and these have been mainly descriptive. In these populations, the incidence of melanoma is very low and the behaviour of the disease is quite different – melanoma occurs at a later age and affects the plantar and palmar surfaces of the feet and hands. This is unlikely to be due to UVR exposure (lack of exposure to this site) and may represent a baseline of incidence of cutaneous melanoma.
WHO has estimated the burden of disease for the year 2000 from cutaneous malignant melanoma using incidence and mortality data derived from Globocan 2000. As noted in Appendix 3, case control studies indicate that the population attributable fraction is approximately 0.2. However, it seems likely that there is a great deal of error inherent in the exposure measurement in these individual-level epidemiological studies that may systematically bias the effect estimate towards the null. Thus, upper (0.9, derived from ecological data) and lower (0.5, based on a consensus of expert opinion) estimates for population attributable fraction were applied to the WHO melanoma GBD estimates.