Melanoma dormancy: Predictors of relapse in resected early-stage cutaneous melanoma.

Abstract

e21557 Background: Early-stage (IA, IB, IIA) cutaneous melanoma (CM) is treated with wide local excision (WLE). However, some patients (PTS) relapse (R), indicating preoperative micrometastasis. Identifying high-risk PTS may guide adjuvant therapy. No studies have compared coding mutations (MUT) in IA-IIA CM PTS with vs without R. We analyze clinical/pathological factors to support whole-exome sequencing (WES) to identify R-associated MUT. Methods: Retrospective study of 180 IA-IIA CM PTS diagnosed 2003–2022 in the UPMC Hillman Melanoma Program biospecimen registry (1,511 of 7,750 PTS). Of 1,511 PTS: 414 R, 808 non-R; others data incomplete. Inclusion required follow-up data, pathology reports, primary tumor specimen, WLE with negative margins, and negative sentinel lymph node (LN) biopsy. Controls had no R ≥3 years post-WLE with documented follow-up within 18 months. Cases (R) and controls (NR) were matched 1:1 by age ≥50, stage, and gender. Variables included: BT, Clark level, ulceration (ULC), MR, lymphovascular invasion (LVI), tumor-infiltrating lymphocytes (TIL), regression, age, gender, stage, histology. We assessed overall survival (OS), R-free survival (RFS), R rates, year (Y), and site(s). Logistic regression, Wilcoxon rank-sum, and t-tests were used for continuous variables; Fisher’s exact test for categorical; Cox regression for OS/RFS. Kaplan-Meier and log-rank tested survival. Results: R and NR stages: 19% IA, 38% IB, 43% IIA. 43% NR and 41% R were female. Median age: 61Y [R] vs. 59Y [NR]. Superficial spreading CM: 43% R vs 46% NR; nodular CM: 26% R vs 20% NR. Survival was 99% in NR vs. 49% in R (HR 51.2, p<0.001). R were 18.9% <1Y, 44.4% 1-3Y, 35.6% 3-10Y, 1 after 10Y. R: 34% local, 21% LN, and 44% distant (14% brain). Age was associated with lower risk of distant R (OR 0.96; p=0.04) and brain metastases (OR 0.95; p=0.03). Pathological features summarized in Table 1. MR ≥1/mm 2 and residual tumor in WLE were more prevalent among R (Table 1). MR ≥1/mm 2 associated with worse RFS in all intervals (HR 1.99, p=0.03), and <1Y (HR 1.99, p=0.03), and 1-3Y (HR 2.25, p=0.02). Worse RFS with residual tumor in WLE (HR 1.83, p=0.005) and LVI (HR 3.22, p=0.02). Stage IB showed worse OS (HR 3.33; p=0.01) vs IA, while IIA was worse than IA (HR 2.02; p=0.2). Brisk TIL was paradoxically adverse for OS (HR 2.95; p=0.003). BT worsened OS in IIA (HR 1.86; p=0.04) and all stages when adjusted for rising MR (HR 1.34, p=0.04) or MR≥2/mm 2 (HR 1.35, p=0.03). MR≥1/mm 2 did not affect OS. Conclusions: In this CM cohort, MR ≥1/mm 2 , LVI, tumor in WLE, and BT increase R risk. WES of primary tumors will define MUT tied to micrometastasis and refine risk stratification. Pathological features in our cohorts. NR R p BT (mm) Median [Min, Max] 1.33 [0.3, 8] 1.53[0.25, 4.12] 0.6 Clark ≥ 4 65.2% 68.4% 0.4 ULC 14% 16% 0.8 MR ≥ 1/mm 2 71% 86% 0.03 Brisk TIL 13% 13% 0.9 Tumor Regression 10% 16% 0.2 LVI 1.1% 4.4% 0.2 Tumor in WLE 21% 39% 0.01