Ki67 as a prognostic biomarker for patients with vertical growth phase (VGP) melanomas

Abstract

9043 Background: In VGP melanomas proliferation is reflected in dermal mitotic figures (“mitogenic” VGP) and/or tumor cell nests larger than any epidermal nest. An alternative to mitotic rate (MR) to characterize cell proliferation is the expression of Ki67 protein. Since Ki67 is expressed in all phases of the cell cycle except G0, it is potentially a more robust biomarker for proliferation and prognosis than mitoses. Methods: To test the hypothesis that Ki67 would replace MR as a prognostic factor, we did a retrospective cohort study of 432 patients with Stage I/II primary VGP melanomas who had at least 10 years of follow up. Tissue sections were stained using the monoclonal antibody MIB-1 to Ki67 and the % of positive melanoma cells were evaluated by two readers. ROC curves for Ki67 and MR were computed. Predicted probabilities (PP) of 10-year melanoma-specific death were computed from 3 multivariate logistic regression models, one for each biomarker (Models 1 and 2) and one with both (Model 3), controlling for established melanoma prognostic factors (thickness, gender, anatomic site, ulceration, regression and tumor infiltrating lymphocytes), and compared. Cross-validation was used to assess differences between using Ki67 and using MR including the differences in PP, Brier scores and the misclassification rates. A decision curve analysis was done to assess the clinical net benefit of the two. Results: The areas under the ROC curve (AUCs) for Ki67 and MR, both continuous factors, were 0.69 and 0.79, respectively. In the multivariate analysis, Ki67 expression was significant in Model 1 (OR=1.03, 95% CI: 1.01–1.05), mitotic rate was not significant in Model 2 (1.05, 0.99–1.1), and only Ki67 was significant in Model 3 (1.03, 1.01–1.05). The AUCs for the three models were 0.84, 0.84, and 0.85, respectively. Based on cross-validation, there was no difference between the two biomarkers in PP, Brier scores, or misclassification rates. The decision cost analysis demonstrated the same net benefit for the two. Conclusions: A prospective study needs to be conducted to confirm that Ki67 and MR are equivalent. No significant financial relationships to disclose.