Prediction of survival in patients with thin melanoma: results from a multi-institution study

Cutaneous melanoma incidence is increasing. Most new cases are thin (≤ 1 mm) with favorable prognoses, but survival is nonetheless variable. Our aim was to investigate new prognostic factors and construct a nomogram for predicting survival in individual patients.

Data from 2,243 patients with thin melanoma were retrieved from prospectively maintained databases at six centers. Kaplan-Meier survival and crude cumulative incidences of recurrence were estimated, and competing risks were taken into account. Multivariable Cox regression was used to investigate survival predictors.

Median follow-up was 124 months (interquartile range, 106 to 157 months); 12-year overall survival was 85.3% (95% CI, 83.4% to 87.2%). Median times to local, regional, and distant recurrence were 79, 78, and 107 months, respectively. Relapse was significantly related to age, Breslow thickness, mitotic rate (MR), ulceration, lymphovascular invasion (LVI), and regression; incidence was lower and subgroup differences were less marked for distant metastasis than for regional relapse. The worst prognosis categories were age older than 60 years, Breslow thickness more than 0.75 mm, MR ≥ 1, presence of ulceration, presence of LVI, and regression ≥ 50%. Breslow thickness more than 0.75 mm, MR ≥ 1, presence of ulceration, and LVI (all P = .001) were significantly associated with sentinel node positivity. Age, MR, ulceration, LVI, regression, and sentinel node status were independent predictors of survival and were used to construct a nomogram to predict 12-year overall survival. The nomogram was well calibrated and had good discriminative ability (adjusted Harrell C statistic, 0.88).

Our findings suggest including LVI and regression as new prognostic factors in the melanoma staging system. The nomogram appears useful for risk stratification in clinical management and for recruiting patients to clinical trials.