Influence of melanoma type on incidence and downstream implications of cutaneous immune-related adverse events in the setting of immune checkpoint inhibitor therapy.

BACKGROUND: Emerging evidence suggests that cutaneous immune-related adverse events (cirAEs) are associated with a survival benefit in the setting of advanced melanoma treated with immune checkpoint inhibitor (ICI) therapy. Previous studies have not examined the role of melanoma subtypes on cirAE development and downstream therapeutic outcomes. OBJECTIVE: Examine the impact of melanoma subtypes on cirAE onset and survival among ICI recipients. METHODS: Retrospective multi-institutional cohort study. Multivariate time-series regressions were utilized to assess relationships between melanoma subtype, cirAE development, and survival. RESULTS: Among 747 ICI recipients, 236 (31.6%) patients developed a cirAE. Patients with acral melanoma were less likely to develop a cirAE (hazard ratio [HR] = 0.41, P = .016) compared to patients with nonacral cutaneous melanoma. Across all melanoma subtypes, cirAEs were associated with reduced mortality (HR = 0.76, P = .042). Patients with acral (HR = 2.04, P = .005), mucosal (HR = 2.30, P < .001), and uveal (HR = 4.09, P < .001) primaries exhibited the worst survival. LIMITATIONS: Retrospective cohort study. CONCLUSION: This is the first study to demonstrate differences in cirAE development among melanoma subtypes. The presence of cirAEs was associated with better survival. Further, the lower incidence of cirAEs may be a marker of immunotherapy response, which is reflected in the association between acral melanoma and mortality.

Pre-Existing Inflammatory Disease Predicts Cutaneous Immunotherapy Toxicity Development: A Multi-Institutional Cohort Study.

Background: Relationships between pre-existing inflammatory diseases (pIDs) and cutaneous immune-related adverse events (cirAEs) have not been well-studied. This study is to investigate associations between pIDs and cirAEs among immune-checkpoint inhibitor (ICI) recipients at the Mass General Brigham healthcare system. Methods: Electronic health records were reviewed to ascertain cirAE status. Patients' pID status was determined using International Classification of Diseases (ICD) codes. Cox proportional hazard, logistic regression, and linear regression models were performed. Results: Among 3607 ICI recipients, 1354 had pIDs, and 672 developed cirAEs. After covariate adjustments, patients with cutaneous pIDs (HR:1.56, p<0.001) or both cutaneous and non-cutaneous pIDs (HR:1.76, p<0.001) had increased cirAE risk in contrast to patients with non-cutaneous pIDs alone (HR:1.01, p=0.9). In adjusted ordinal logistic regression modeling, cutaneous pIDs (OR:1.55, p<0.0001) and the presence of both cutaneous pIDs and non-cutaneous pIDs (OR:1.71, p=0.002) were associated with increased cirAE severity. The time to cirAE onset was different between the cutaneous pID group and the non-cutaneous pID group (Mean: 98 vs. 146 days, p=0.021; Beta: -0.11, p=0.033). Conclusions: ICI recipients with cutaneous pIDs should have increased clinical monitoring due to their increased risk of cirAE development, severity, and earlier onset.

Prediction of early-stage melanoma recurrence using clinical and histopathologic features.

Prognostic analysis for early-stage (stage I/II) melanomas is of paramount importance for customized surveillance and treatment plans. Since immune checkpoint inhibitors have recently been approved for stage IIB and IIC melanomas, prognostic tools to identify patients at high risk of recurrence have become even more critical. This study aims to assess the effectiveness of machine-learning algorithms in predicting melanoma recurrence using clinical and histopathologic features from Electronic Health Records (EHRs). We collected 1720 early-stage melanomas: 1172 from the Mass General Brigham healthcare system (MGB) and 548 from the Dana-Farber Cancer Institute (DFCI). We extracted 36 clinicopathologic features and used them to predict the recurrence risk with supervised machine-learning algorithms. Models were evaluated internally and externally: (1) five-fold cross-validation of the MGB cohort; (2) the MGB cohort for training and the DFCI cohort for testing independently. In the internal and external validations, respectively, we achieved a recurrence classification performance of AUC: 0.845 and 0.812, and a time-to-event prediction performance of time-dependent AUC: 0.853 and 0.820. Breslow tumor thickness and mitotic rate were identified as the most predictive features. Our results suggest that machine-learning algorithms can extract predictive signals from clinicopathologic features for early-stage melanoma recurrence prediction, which will enable the identification of patients that may benefit from adjuvant immunotherapy.