Immune checkpoint inhibitors have revolutionized the treatment of metastatic melanoma, with a survival now reaching several years in some patients. The anti-PD-1 monoclonal antibodies nivolumab and pembrolizumab increase both the quality of life and overall survival of metastatic melanoma patients. They are currently the first-line treatment for metastatic melanoma. However, only 40% of patients respond to anti- PD-1 monotherapy. In the era of personalized medicine, it has become essential to identify those patients who will benefit from these treatments that are, by the way, responsible for rare but serious adverse events. We tested several immunohistochemical markers in order to discriminate between responders and nonresponders.
Among the 50 metastatic melanoma patients who received pembrolizumab from 2014 to 2016, we were able to analyze the tumor of 18 patients. Immunohistochemical stainings were performed on pretreatment metastatic tissue in order to analyze the interaction between tumor cells and the immune system. Staining for MELANA, TYR, and GP100 antigens was performed to identify tumor cells, and staining for CD3 and CD8 antigens was performed to identify T lymphocytes. We also performed stainings for antigens such as PD-L1, IDO1, HLA class I heavy chains, and β2-microglobulin. Two reviewers independently assessed the stainings. For PD-L1 status, we used the MEL-score. For T-cell infiltration, we evaluated the presence and location of CD3+ CD8+ T cells.
We found that only PD-L1 expression (p=0.043) and T-cell infiltration (p=0.025) as well as T-cell location at the periphery of the tumor nodules (p=0.025) were associated with a better response to pembrolizumab. PD-L1 expression and the presence of T lymphocytes were also associated with a longer survival. The median overall survival was 31.5 months for patients with T-cellrich metastases and high PD-L1 expression versus 4 months for those with non-T-cell-infiltrated tumors and 7 months for those with PD-L1-negative tumors. Furthermore, 50% of patients showed reduced or undetectable HLA class I expression by the tumor cells. They presented progressive or stable disease, while none presented partial or complete response.
Although significant correlations were observed between some markers and the response to anti-PD-1 therapy, no predictive factor could be identified in our small series. Some patients with non-T-cell-infiltrated and/or PD-L1-negative tumor showed a clinical benefit with pembrolizumab. Conducting a larger scale study would possibly allow proving that these two factors are predictive for the response to pembrolizumab. Recent and future technical progresses will enable to perform new studies aimed at finding robust biomarkers. The loss of HLA class I expression by tumor cells is probably a resistance mechanism to anti-PD-1 therapy, and new researches in this field will be helpful.