ABSTRACT
BACKGROUND: Adjuvant treatment of stage II-IV melanoma with PD-1-based immune checkpoint inhibitors (ICI) has improved relapse-free survival (RFS) and has therefore become a standard-of-care treatment option. Approximately 25%-30% of patients still recur within 1 year. Predictive biomarkers reflecting real-world data are desired. The predictive relevance of tumour tissue PD-L1 expression in the adjuvant setting remains inconclusive. OBJECTIVES: This retrospective, observational study was conducted to evaluate the value of PD-L1 expression scores in different tumour tissue locations in predicting response towards adjuvant immunotherapeutic treatment. METHODS: Tumour tissue taken prior to anti-PD-1 adjuvant ICI in 243 stage II-IV melanoma patients was collected at University Skin Cancer Center Hamburg. PD-L1 expression was evaluated on immune cells (ICS), tumour cells (TPS) and combined (CPS). Scores were determined by independent pathological physician quantification and correlated with therapy outcome at different cut-off (CO) levels (relapse-free survival, RFS) for different tumour tissue locations (primary tumour, metastases). RESULTS: A total of 104 patients were eligible for analysis. Positivity of ICS, TPS and CPS showed no predictive RFS outcome association at different CO levels when analysed irrespective of tissue origin. In primary tumours, ICS at CO 1% showed a significantly improved RFS upon positivity (HR 0.22). In contrast, positivity to TPS (CO 1%) correlated significantly and independently with improved RFS when evaluated in metastatic tumour tissue specimens (HR 0.37). CONCLUSIONS: PD-L1 tumour tissue expression may serve as a predictive biomarker for adjuvant ICI treatment response stratification in melanoma, but caution should be spent on the origin of tumour tissue analysed. The cell-type relevant for the predictive value of PD-L1 expression is tissue-specific with immune cells being important in primary tumours while tumour cells are key in metastases. The present results should be validated in a multicentre cohort.
