No abstract

BackgroundClinical response to MAPK inhibitors in metastatic melanoma patients is heterogeneous for reasons still needing to be elucidated. As the patient immune activity contributes to treatment clinical benefit, the pre-existing level of immunity at tumor site may provide biomarkers of disease outcome to therapy. Here we investigated whether assessing the density and spatial tissue distribution of key immune cells in the tumor microenvironment could identify patients predisposed to respond to MAPK inhibitors.MethodsPretreatment tumor biopsies from a total of 213 patients (158 for the training set and 55 for the validation set) treated with BRAF or BRAF/MEK inhibitors within the Italian Melanoma Intergroup were stained with selected immune markers (CD8, CD163, β-catenin, PD-L1, PD-L2). Results, obtained by blinded immunohistochemical scoring and digital image analysis, were correlated with clinical response and outcome by multivariate logistic models on response to treatment and clinical outcome, adjusted for American Joint Committee on Cancer stage, performance status, lactate dehydrogenase and treatment received.ResultsPatients with high intratumoral, but not peritumoral, CD8+ T cells and concomitantly low CD163+ myeloid cells displayed higher probability of response (OR 9.91, 95% CI 2.23–44.0, p = 0.003) and longer overall survival (HR 0.34, 95% CI 0.16–0.72, p = 0.005) compared to those with intratumoral low CD8+ T cells and high CD163+ myeloid cells. The latter phenotype was instead associated with a shorter progression free survival (p = 0.010). In contrast, PD-L1 and PD-L2 did not correlate with clinical outcome while tumor β-catenin overexpression showed association with lower probability of response (OR 0.48, 95% CI 0.21–1.06, p = 0.068).ConclusionsAnalysis of the spatially constrained distribution of CD8+ and CD163+ cells, representative of the opposite circuits of antitumor vs protumor immunity, respectively, may assist in identifying melanoma patients with improved response and better outcome upon treatment with MAPK inhibitors. These data underline the role of endogenous immune microenvironment in predisposing metastatic melanoma patients to benefit from therapies targeting driver-oncogenic pathways.

Our nomogram could be useful for follow-up personalization in clinical practice, and for patient risk stratification while conducting clinical trials or analyzing their results.

The contribution of recently established or candidate susceptibility genes to melanoma missing heritability has yet to be determined. Multigene panel testing could increase diagnostic yield and better define the role of candidate genes. We characterized 273 CDKN2A/ARF and CDK4-negative probands through a custom-designed targeted gene panel that included CDKN2A/ARF, CDK4, ACD, BAP1, MITF, POT1, TERF2IP, ATM, and PALB2. Co-segregation, loss of heterozygosity (LOH)/protein expression analysis, and splicing characterization were performed to improve variant classification. We identified 16 (5.9%) pathogenic and likely pathogenic variants in established high/medium penetrance cutaneous melanoma susceptibility genes (BAP1, POT1, ACD, MITF, and TERF2IP), including two novel variants in BAP1 and 4 in POT1. We also found four deleterious and five likely deleterious variants in ATM (3.3%). Thus, including potentially deleterious variants in ATM increased the diagnostic yield to about 9%. Inclusion of rare variants of uncertain significance would increase the overall detection yield to 14%. At least 10% of melanoma missing heritability may be explained through panel testing in our population. To our knowledge, this is the highest frequency of putative ATM deleterious variants reported in melanoma families, suggesting a possible role in melanoma susceptibility, which needs further investigation.