Clinical, dermatoscopic, histological and molecular predictive factors of distant melanoma metastasis: A systematic review and meta-analysis

Abstract

Background

Melanoma metastasis to distant sites is associated with diminished survival rates and poor prognosis. Except of Breslow thickness and ulceration that are currently used in melanoma staging, the investigation of additional clinicopathological, dermatoscopic and molecular factors that could predict tumors with aggressive biologic behavior is of paramount importance.

Methods

A literature search was conducted in PubMed, Scopus, Cochrane databases and gray literature until November 2023. Observational studies (including cohorts and case-control studies) were included and clinical and histopathological factors of primary cutaneous melanomas, along with dermatoscopic and molecular predictors of distant metastasis (DM) and distant metastasis-free survival (DMFS) were assessed. Random – effect models were preferred, the results were presented as Hazard Ratios (HRs) with 95 %Confidence Intervals (CIs) and the I² index quantified heterogeneity. Subgroup analysis according to AJCC stage and sensitivity analysis were also conducted.

Results

One hundred forty-three and 101 studies were included in the qualitive and quantitative synthesis, respectively. Regarding clinical factors, males, compared to females, and head and neck location, compared to trunk, demonstrated higher risk for DM [n=36, HR 1.49, 95%CI 1.36 – 1.63, I² 33% and n=21, HR 1.24, 95 %CI 1.01 – 1.52, I² 62 %]. Both factors had similar effects on DMFS. Breslow thickness and ulceration were significant predictors or DM. Additional factors that posed an increased risk for DM were nodular (n=15, HR 2.51, 95 %CI 1.83 – 3.43, I² 56 %) and lentigo maligna subtypes (n=12, HR 1.87, 95 %CI 1.27 – 2.75, I² 0 %), compared to superficial spreading subtype, lymphovascular invasion (n=9, HR 2.05, 95 %CI 1.18 – 3.58, I² 78 %), SLN positivity and BRAF+ mutational status. In contrast, regression was a negative predictor of DM (n=15, HR 0.59, 95 %CI 0.44 – 0.79, I² 68 %). Two studies focused on dermatoscopic factors and found that low pigmentation and the presence of blue-white veil might predict DM development. The results of subgroup analysis for stage I-II patients were essentially similar and sensitivity analysis did not reveal significant alterations, despite the moderate or high heterogeneity in some categories.

Conclusions

Clinical and histological characteristics of the tumor along with dermatoscopic features and molecular parameters hold significant prognostic information and could be incorporated into models to predict melanomas with high metastatic potential.