Proliferation indices of phosphohistone H3 and Ki67: strong prognostic markers in a consecutive cohort with stage I/II melanoma

Abstract

Cellular proliferation is correlated with the progression of melanoma. Accordingly, the proliferation index of H&E-stained thin melanomas was recently included in the staging system of the American Joint Committee on Cancer. Yet, the immunohistochemical markers of proliferation phosphohistone H3 and Ki67 may improve such indices. To accurately quantify these markers, they should be combined with a melanocytic marker, for example, MART1 in an immunohistochemical double stain; also enabling automated quantification by image analysis. The aim of the study was to compare the prognostic impact of phosphohistone H3/MART1, Ki67/MART1, and H&E stains in primary cutaneous melanoma, and to determine the difference between indices established in hot spots and the global tumor areas. The study included 153 consecutive stage I/II melanoma-patients. The follow-up time was 8–14 years for event-free melanoma. Recurrent disease occurred in 43 patients; 37 died of melanoma. Both events occurred in only three thin melanomas. Their paraffin-embedded tissue was stained for phosphohistone H3/MART1, Ki67/MART1, and with H&E. And proliferation indices were established in 1-mm² hot spots and in the global tumor areas. In multivariate Cox analyses, only hot spot indices of phosphohistone H3/MART1 and Ki67/MART1 were independent prognostic markers. Phosphohistone H3/MART1 tended to be better than Ki67/MART1 with adjusted hazard ratios of 3.66 (95% CI, 1.40–9.55; P=0.008) for progression-free survival and 3.42 (95% CI, 1.29–9.04; P=0.013) for melanoma-specific death. In all stains, prognostic performance was substantially improved by using hot spots instead of the global tumor areas. In conclusion, phosphohistone H3/MART1 and Ki67/MART1 were superior to H&E stains, and hot spots superior to the global tumor areas. Given the potential for automated analysis, these double stains seem to be robust alternatives to conventional mitotic detection by H&E in stage I/II melanomas in general. This was particularly true for thick melanomas whereas no specific analyses for thin melanomas only could be performed.