Oral Presentation Australasian Melanoma Conference 2026

Towards Safe Discontinuation of Immune Checkpoint Inhibitor Therapy: Integrating Digital Pathology and Gene Profiling in Melanoma (#89)

Ka Won Noh 1 , Yuri Tolkach 1 , Doris Helbig 2 , Vincenzo Mitchell Barroso 1 , Yannick Förster 3 , Max Schlaak 4 , Tilo Biedermann 3 , Reinhard Büttner 1 , Oana-Diana Persa 3
  1. Institute of Pathology, University Hospital of Cologne and Medical Faculty, Cologne, Germany
  2. Department of Dermatology, University Hospital Cologne and Medical Faculty, Cologne, Germany
  3. Department of Dermatology and Allergy, School of Medicine and Health, Technical University of Munich, Munich, Germany
  4. Department of Dermatology, Venereology and Allergology, Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany

Background:
Immune checkpoint inhibitors (ICIs) have transformed the management of advanced melanoma, with many patients achieving long-term remission even after treatment cessation. However, determining which patients can discontinue ICIs without relapse remains a clinical challenge and algorithms to guide clinical decision-making are lacking.

Methods:
This study included patients with advanced melanoma who discontinued ICI treatment in the absence of progression and had at least two years of follow-up after cessation. Time to progression (TTP) was defined as the interval between ICI discontinuation and disease recurrence. Pre-treatment hematoxylin and eosin (H&E)-stained whole-slide images (WSI) and clinical data were analyzed. Immune cell populations and tumor cells were identified using the CONIC challenge implementation of the StarDist algorithm trained on the Lizard dataset. Gene expression profiling was performed using the NanoString PanCancer IO 360 Panel.

Results:

Best overall response (BOR) was the only clinical parameter significantly affecting TTP. AI-based single-cell analysis revealed that tumor cell density (TCD) and the lymphocyte-to-plasma cell ratio (LYM/PC) were significant predictors of relapse after ICI cessation. Transcriptomic analysis identified TGFBR1, CRABP2, LTBP1, and NOD2 as genes associated with relapse, whereas CST2 and SOX11 were enriched in relapse-free patients. Multivariable Cox regression integrating BOR, TCD, LYM/PC, and the differentially expressed genes identified TCD, LYM/PC, and TGFBR1 as independent prognostic factors. A Multivariate Adaptive Regression Spline (MARS) model combining these predictors achieved a cross-validated accuracy of 84.6% in predicting relapse after therapy discontinuation.

Conclusion:
TGFBR1 expression, tumor cell density, and the lymphocyte-to-plasma cell ratio are promising biomarkers to select patients who may safely discontinue ICI therapy. The identified prognostic markers are fully explainable and easily implementable in routine practice and should be further evaluated in prospective studies.