Background and Objectives: The prognostic significance of regression in predicting melanoma recurrences is unknown. We present a large multicenter study correlating regression with recurrence. Methods: The Sentinel Lymph Node Working Group database was queried from 1993 to 2018 for cases with regression data. Clinicopathologic factors were correlated with overall and first-site of recurrence and with recurrence-free survival (RFS). Results: There were 4790 patients and the median follow-up was 39.6 months. Regression and recurrences were seen in 1081 (22.6%) and 773 (16.1%) cases, respectively. First-site locoregional and distant recurrences were seen in 412 (8.6%) and 352 (7.3%) patients, respectively. Regression was seen in 15.8% and 24.7% of all cases with and without recurrences (p < 0.0001), respectively, while regression was seen in 14.3% and 17.9% of first-site locoregional and distant recurrent cases, respectively, compared with 23.3% and 22.9% of patients with regression and without first-site locoregional and distant recurrences, respectively (p = 0.29). On multivariable analysis, after controlling for age, gender, thickness, ulceration, lymphovascular invasion, and sentinel lymph node status, regression significantly predicted improved RFS (p = 0.004) and fewer first-site regional recurrences (p = 0.017). Conclusion: Our data suggest that regression is a favorable prognostic marker in melanoma and predicts significantly better RFS and decreased first-site regional recurrences.
ASJC Scopus subject areas