In this study, we identified four clinical indicators to identify patients with advanced or metastatic melanoma who may benefit from ICB monotherapy. These include baseline LDH level, the extent of disease, lymphopenia, and irAE. Among the favorable group who had normal baseline LDH level and no visceral/CNS metastasis, we observed very prolonged OS (40.4 months) even if all patients had advanced or metastatic melanoma. Although patients were classified into the unfavorable group, subgroups of patients who may have longer OS can be sorted early during the treatment course using other clinical indicators, such as lymphopenia and irAE. Baseline LDH level, the extent of disease, and lymphopenia were independent prognostic factors for both OS and PFS.
As response to immunotherapy can be achieved for only a subset of patients, there is a crucial need to identify biomarkers to predict the efficacy of anti-CTLA-4 or anti-PD-1 treatment or identify a specific subset of patients who may benefit from immunotherapy. Researchers have investigated many potential biomarkers for immunotherapy in advanced or metastatic melanoma patients using novel technologies, such as next-generation sequencing, T-cell receptor profiling, and mass cytometry . Among them, tumor mutations and neoantigen load as well as the expression of immune-related genes in tumor tissue and/or the presence of CD8 + T-cell infiltrates showed significant correlations to response in large genetic and transcriptional analyses of tumor tissue . The response prediction would be more powerful if these high-technology novel biomarkers are combined with clinical indicators which can be obtained using non-invasive and easily accessible techniques. The suggested clinical biomarkers in the current study can be determined through a simple blood test or routine clinical exam and are cost-effective and quickly identified. Thus, we surmise that utilizing these easily accessible biomarkers will be of clinical use to guide proper patient selection for immunotherapy.
In 2009, LDH was shown to be an independent predictor of survival in melanoma and was therefore added to the AJCC guidelines . Accelerated metabolism in cancer cells requires increased glycolysis that produces elevated levels of LDH as a byproduct, which is therefore a robust proxy to assess tumor burden . In the context of immunotherapy, baseline LDH level and visceral/CNS metastasis were both independent prognostic factors in this study. This finding is consistent with previous reports. A retrospective data from the Netherlands and the United Kingdom also suggested that patients with metastatic melanoma whose baseline serum LDH was greater than twice the upper limit were unlikely to benefit from ipilimumab treatment . A retrospective large cohort of advanced melanoma patients also showed that elevated LDH and the presence of liver metastasis predict poor response to anti-PD-1 therapy .
We also revealed that patients who developed lymphopenia within 3 months after ICB initiation are associated with poor OS. Lymphocytes are important mediators of ICB mechanism. Given that circulating lymphocytes are the cells that eventually infiltrate tumors, their depletion might contribute to suboptimal treatment outcomes after immunotherapy. Similar to our results, a retrospective analysis with melanoma patients treated with ipilimumab also showed that increases in absolute lymphocyte counts at 2–8 weeks and circulating CD4 + and CD8 + T cells at 8–14 weeks were associated with positive clinical outcomes .
In this study, 7% and 11% patients developed immunotherapy-related hypothyroidism and vitiligo, respectively, of which 5% and 6% developed within 6 months, respectively. Although irAE can present at any time, including after cessation of ICB therapy, we confined the biomarker to the irAE occurring within 6 months. Because it is difficult to make a clinical decision based on delayed irAE, we suppose that irAE occurring at least within 6 months would have clinical significance as a biomarker. A subgroup of patients who developed immune-related hyporthyroidism or vitiligo in the unfavorable group had longer OS than other subgroups of patients (median, 43.6 vs. 13.1 months; p = .008), although PFS was not significantly longer (median, 8.6 vs. 2.8 months; p = .199). The findings are supported by previous studies. A systematic review showed that patients who developed vitiligo were associated with two to four times less risk of disease progression or death, respectively, compared to those without . A prospective observational study of 67 melanoma patients showed that an objective response to treatment was associated with a higher incidence of vitiligo and all 17 patients who developed vitiligo were alive at the time of analysis, which means that patients with vitiligo had durable response. A retrospective study with 174 patients who received ICB for metastatic or advanced cancers showed a significantly longer PFS (median, 66 vs. 27 weeks) and OS (median, 156 vs. 59 weeks) in the thyroid dysfunction group than in the euthyroid group . As this group had an extremely long survival even after disease progression, we named the group “late responders.” The subsequent treatments after ICB monotherapy had a long-term response (Supplementary Table 1). We did not elucidate whether the long-term OS was attributed to the irAE itself or the good treatment response. If the former is correct, irAE would be a prognostic factor, and if the latter is correct, irAE would be a predictive factor. Further study regarding this issue is warranted.
Using a simple decision tree, we suggested an optimal way to apply these factors in the clinical decision making in a time dependent manner. Nosrati et al  developed a clinical scoring system to predict response to anti-PD-1 monotherapy in patients with advanced melanoma. The variables used in the scoring system were baseline clinical factors, such as sex, age, previous ipilimumab treatment, elevated LDH, and liver metastasis. Our model is practical in that it is not only simple but also includes variables of multiple time points, such as baseline, 3 months, and 6 months. This model has strength in that it can be applied before ICB monotherapy and during early ICB monotherapy.
Our study is retrospective in nature and therefore is limited by patient heterogeneity associated with the design, including the presence of uncontrolled confounding factors, variations in ICB therapy cycles, variable radiation doses/sites, and under-reporting of toxicity. In addition, information regarding date of progression on ICB was not uniformly assessed because it would have been on a prospective clinical trial. Thus, we focused mainly on OS given the atypical patterns of response that can be observed after ICB therapy. Moreover, the sample size was relatively small. However, we enrolled a homogenous cohort of patients with metastatic melanoma who received ICB therapy at a single institution. Furthermore, irAE was significant in the univariate analysis for OS, but it lost its significance in the subsequent multivariate analysis, which may be due to the small number of groups.
In conclusion, we suggested a clinical predictive model using easily accessible clinical indicators to predict treatment outcomes in patients with advanced or metastatic melanoma who received ICB monotherapy. These indicators are baseline LDH level, visceral/CNS metastasis, lymphopenia within 3 months, and hypothyroidism or vitiligo within 6 months. Although these indicators have been reported as prognostic factors separately in previous reports, we showed that all were independent prognostic factors in a patient cohort, which add to the growing body of literature. The identification of a clinical predictive model is critical due to the following reasons. Firstly, it allows patients who are unlikely to benefit from anti-PD-1 therapy to be spared from unnecessary risk of toxicity and to rationally select a combination that will better fit them. Secondly, it can spare those who are likely to respond to PD-1 monotherapy from unnecessary toxicities from combination immunotherapy approach. This model could potentially have a role in the therapeutic decision-making and proper patient selection regarding immunotherapy. Its validation in future studies is warranted.