Neoadjuvant immunotherapy (neoIT) in melanoma: A prediction tool for pathological response and recurrence, and the role of adjuvant (Adj) therapy (Tx) to improve recurrence-free survival (RFS).

9579 Background: NeoIT with anti-PD-1 (PD1) showed better event-free survival than adjuvant (adj)IT and is standard of care for pts with stage IIIB–D melanoma. Many pts achieve a Major Pathological Response with neoIT (MPR; complete [pCR] or near-complete [near-pCR] pathological response) and may not require additional Tx; pts with non-MPR (partial [pPR] or no [pNR] pathological response) are at higher risk of recurrence, and may benefit from adj Tx. We sought to: 1) predict pts with non-MPR and at higher risk of recurrence with PD1-based neoIT, and 2) study the role of adjuvant (adj) Tx in pts with pNR. Methods: Pts with stage IIIB–D melanoma treated with PD1-based neoIT were included. Demographics, disease characteristics, baseline blood parameters, radiological response, and clinical outcomes were examined. A multivariate penalized logistic regression model was developed to predict non-MPR and recurrence. Results: Of 184 pts, 30 (16%) had PD1 alone, 107 (58%) had PD1 + anti-CTLA-4 (PD1+IPI), and 47 (26%) had PD1 + novel IT agents. Of the 179 (97%) who underwent surgery, 114 (64%) had MPR (95 [53%] with pCR and 19 [11%] with near-pCR) and 65 (36%) had non-MPR (20 [11%] with pPR and 45 [25%] with pNR). Five pts did not receive surgery (due to clinical progression in 4). The combination of pre-neoIT clinical and pathological features predicted non-MPR pts with an AUC of 0.73. The addition of radiological response (% change from baseline in tumor size: adjusted odds ratio [AOR] 18.94, p=0.0001) to clinical features (pre-neoIT neutrophil [AOR 1.29, p=0.0476] and lymphocyte count [AOR 1.64, p=0.0467], largest lymph node long axis [AOR 1.02, p=0.1056], hemoglobin level [AOR 0.99, p=0.6728], and primary tumor staging [T4 vs. T1; AOR 1.83, p=0.0457]) improved the predictive accuracy (AUC = 0.86). For pts with ≥6 months FU, 2% (2/98) of pts with MPR and 34% (20/58) of pts with non-MPR (20% [4/20] with pPRs and 42% [16/38] with pNR) recurred. Amongst non-MPR pts, the combination of mitotic rate in the primary melanoma (>1 vs. ≤1 mitosis/mm 2 ; AOR 8.23, p<0.0001) and % change from baseline in tumor size (AOR 13.59, p=0.0005) predicted recurrence with an AUC of 0.81. The median % change from baseline in tumor size was +12% in recurrent vs -11% in non-recurrent pts (p=0.0047). Pts with >30% reduction in tumor size did not recur. Of the 38 pts with pNR, the recurrence rate was 64% (7/11) with no adj Tx, 27% (3/11) with adj targeted therapy (TT), and 38% (6/16) with adj PD1. Conclusions: Clinical, pathological, and radiological features can accurately predict MPR or non-MPR to neoIT, as well as recurrence, facilitating response-directed Tx. Acknowledging the small numbers of patients, these early data suggests that Adj TT or PD1 reduces recurrence and should be discussed with pts with pNR. De-escalation of surgery (± adj Tx) should be investigated in MPR pts.