A melanoma risk prediction model showed good discrimination between patients with and without melanoma, a study published in the British Journal of Dermatology reports. The prediction model incorporated several candidate predictors, among which whole-body nevi were the strongest risk factor.
The prediction model was developed using data abstracted from the Australian Melanoma Family Study, a population-based case-control study (461 cases; 329 controls). The model was then externally validated in the Leeds Melanoma Case-Control Study (960 cases; 513 controls). In both studies, patients with first-primary cutaneous melanoma were frequency matched with controls by age and sex. For all participants, clinical assessment of nevi was conducted, with separate counts for melanocytic nevi and dysplastic nevi. Other candidate predictors included demographic factors, solar lentigines, self-assessed pigmentation phenotype, sun exposure, family history of melanoma, and personal history of keratinocyte cancer. The model was developed using unconditional logistic regression with backward selection. Relative risks and odds per age- and sex-adjusted standard deviation were calculated to compare the predictive strengths of model variables. The ability of the model to discriminate between cases and controls was estimated using area under the curve.
The final model included the total number of nevi ≥2mm, solar lentigines on the upper back, hair color at age 18 years, and personal history of keratinocyte cancer. The area under the curve for the model was 0.79 (95% CI, 0.76-0.83) and 0.73 (95% CI, 0.70-0.75) in the Australian and Leeds studies, respectively. Nevi were the strongest risk factor, with an odds per age‐ and sex‐adjusted standard deviation of 3.51 (95% CI, 2.71-4.54) in the Australian study and 2.56 (95% CI, 2.23-2.95) in the Leeds study. The Hosmer-Lemeshow test value was 0.30 on internal validation and <0.001 on external validation; the poor calibration for the Leeds study was a result of the model underestimating risk at lower risk levels and overestimating risk at higher levels.
Although the risk prediction model had high calibration internally, poor calibration was observed on external validation, suggesting that measures of lifetime risk generated using the model may be less accurate for non-Australian populations. Additionally, the case-control study design introduced potential risk for selection and recall bias.
This melanoma risk prediction model which incorporated clinically-assessed risk factors had good discrimination, with ability to distinguish between individuals with and without melanoma across 2 populations with different ambient sun exposure.
The model “may be useful for offering tailored preventive interventions…in primary care and other clinical settings where dermatologic risk factors can be assessed,” investigators concluded.
Vuong K, Armstrong BK, Drummond M, et al. Development and external validation study of a melanoma risk prediction model incorporating clinically-assessed naevi and solar lentigines [published online August 4, 2019]. Br J Dermatol. doi:10.1111/bjd.18411