In this study, we have demonstrated that high immunohistochemical expression of MCM3 is an independent predictor of an increased risk of recurrence and death from melanoma in a large, prospective, population-based cohort study. These findings are in line with previous results from gene expression analysis and immunohistochemical validation in retrospective melanoma cohorts, where MCM3, together with MCM4 and MCM6, was found to correlate with poor outcome, although only MCM4 and MCM6 remained independent predictors of survival in multivariable analysis . While we are not aware of any other validatory studies on the prognostic value of MCM3 expression in melanoma, a recent study failed to confirm the prognostic value of MCM4 in a consecutive cohort of nodular melanoma (n = 220) . Given the uniformity regarding histological subtype in that study, the results may however not be applicable to the general population. In our study, MCM3 expression was strongly associated with nearly all unfavourable clinicopathological characteristics and, yet, remained an independent predictor of a reduced DFS and MSS, adjusted for all other factors. As denoted and discussed previously, the proportion of thin melanomas in this cohort is higher than the expected , despite the higher age of participants in the MDCS (all > 40 years at study entry) and the fact that older melanoma patients more frequently present with advanced disease . Nevertheless, as current clinical guidelines in Sweden recommend sentinel node biopsy only in melanomas > 1 mm, there is a great unmet need for identification of prognostic biomarkers in thin melanomas (≤ 1 mm) , not least since this category seems to make up for most of the increasing melanoma incidence . Therefore, despite the possibility of a selection bias and the small number of events, the comparatively large proportion of thin melanomas in the MDCS may offer some advantage in biomarker studies. The study by Winnepenninckx et al. was limited to melanomas in the vertical growth phase and/or with a thickness of > 1 mm  and, notably, in our study, MCM3 expression was found to be prognostic also in the category of thin melanomas. Thus, the potential value of MCM3 expression as a prognostic biomarker in thin melanomas merits further investigation in larger patient cohorts.
A cautionary remark should also be made regarding methodological aspects on the use of the TMA technique for biomarker studies in malignant melanoma, mainly concerning the technical difficulty in sampling of small tumours. As previously pointed out, the distribution of clinicopathological characteristics was similar in tumours included in the TMA cohort (n = 226) and tumours not suitable for TMA construction (n = 29) in the full cohort, with exception for histological subtype, with no tumours being denoted as nodular in the category not suitable or available for TMA construction, but an equal distribution of the other subtypes . Another limitation to the TMA technique is that it might not accurately reflect the expression of heterogenously expressed markers. Therefore, assessment of MCM3 expression was also performed in a subset of full-face sections, which resulted in a kappa-value corresponding to the best degree of concordance .
The here observed inverse association of MCM3 and RBM3, both regarding their tumour-specific expression and impact on survival, is in line with previous in vivo and in vitro observations in ovarian cancer . Notably, MCM3 expression was an independent predictor of both DFS and MSS while RBM3 only remained an independent predictor of CSS and OS . Longitudinal analysis did not reveal an altered expression of MCM3 in metastatic compared to primary melanoma, in contrast to RBM3, that was found to be downregulated in metastatic melanoma in two independent studies, including the present cohort [12, 15]. Moreover, RBM3 expression was not found to be prognostic in thin melanomas, but an independent prognostic factor in melanomas thicker than 1 mm . Expression of MCM3 but not RBM3 correlated significantly with Ki67 expression, which is in line with its role as a marker of proliferation [28, 29]. In contrast to MCM3, however, Ki67 expression was not prognostic in the full cohort in our previous study, only in male melanoma . Speculatively, these findings indicate that loss of RBM3 may be associated with a switch towards a more invasive and/or metastatic rather than proliferative phenotype, while up-regulated MCM3 expression may either be associated with increased proliferation or functions beyond DNA licensing, i.e. cell migration and invasion, as demonstrated in medulloblastoma cells .
In a translational context, while MCM3 expression appears to be a stronger prognostic biomarker than RBM3, not least in thin melanomas, the inverse correlation between tumour-specific expression of MCM3 and RBM3, which is in line with previous observations in ovarian cancer , may give some directions towards further functional studies of the role and interaction of these proteins in melanoma progression and metastasis. While both proteins have been demonstrated to be up-regulated in several premalignant conditions and cancer forms compared to their corresponding normal tissues [13, 17, 28–31], it is becoming increasingly evident that their oncogenic activities in human tumours influence clinical outcome in opposite ways. One explanation for this may be that elevated expression of RBM3 attenuates DNA damage response, in which MCM proteins play an important role [17, 32], thus preventing the selection of clones with an increased capability for invasion and metastatic spread [33, 34]. Along this line, it would also be of interest to study the expression of MCM3 and RBM3 in benign naevi, dysplastic naevi and melanoma in situ in order to assess their potential role as markers of genetic abnormalities and high-risk lesions, for which e.g. FISH testing has been identified as a valuable diagnostic tool . Ladstein et al. compared MCM4 expression in benign naevi and melanoma and found significantly higher MCM4 positivity in melanoma compared to benign naevi .