Saenger, Yvonne M.

Importance Tumor-infiltrating lymphocytes (TILs) are a provocative biomarker in melanoma, influencing diagnosis, prognosis, and immunotherapy outcomes; however, traditional pathologist-read TIL assessment on hematoxylin and eosin–stained slides is prone to interobserver variability, leading to inconsistent clinical decisions. Therefore, development of newer TIL scoring approaches that produce more reliable and consistent readouts is important. Objective To evaluate the analytical and clinical validity of a machine learning algorithm for TIL quantification in melanoma compared with traditional pathologist-read methods. Design, Setting, and Participants This multioperator, global, multi-institutional prognostic study compared TIL scoring reproducibility between traditional pathologist-read methods and an artificial intelligence (AI)-driven approach. The study was conducted using retrospective cohorts of patients with melanoma between January 2022 and June 2023 across 45 institutions, with tissue evaluated by participants from academic, clinical, and research institutions. Participants were selected to ensure diverse expertise and professional backgrounds. Main Outcomes and Measures Intraclass correlation coefficient (ICC) values were calculated for the manual and AI-assisted arms using log-transformed data. Kendall W values were calculated for Clark scores (brisk = 3, nonbrisk = 2, and sparse = 1). Reliabilities of ICC and W values were classified as moderate (0.40-0.60), good (0.61-0.80), or excellent (>0.80). AI TIL measurements were dichotomized using the 16.6 and median cutoffs. Univariable and multivariable Cox regression analyses assessed the prognostic value of TIL scores adjusted for clinicopathologic variables. Results There were 111 patients with melanoma in the independent testing cohort (median [range] age at diagnosis, 61.0 [25.0-87.0] years; 56 [50.5%] male) who contributed melanoma whole tissue sections. A total of 98 participants evaluated TILs on 60 hematoxylin and eosin–stained melanoma tissue sections. All 40 participants in the manual arm were pathologists, while the AI-assisted arm included 11 pathologists and 47 nonpathologists (scientists). The AI algorithm demonstrated superior reproducibility, with ICCs higher than 0.90 for all machine learning TIL variables, significantly outperforming manual assessments (ICC, 0.61 for AI-derived stromal TILs vs Kendall W , 0.44 for manual Clark TIL scoring). AI-based TIL scores showed prognostic associations with patient outcomes (n = 111) using the median cutoff approach with a hazard ratio (HR) of 0.45 (95% CI, 0.26-0.80; P = .005), and using the cutoff of 16.6, with an HR of 0.56 (95% CI, 0.32-0.98; P = .04). Conclusions and Relevance In this prognostic study of TIL quantification in melanoma, the AI algorithm demonstrated superior reproducibility and prognostic associations compared with traditional methods. Although the retrospective nature of the cohorts limits demonstration of clinical utility, the publicly available dataset and open-source AI tool offer a foundation for future validation and integration into melanoma management.

3553 Background: Immuno-STATs are modular fusion proteins designed for the selective activation of tumor antigen specific CD8+ T cells. CUE-102, the second Immuno-STAT in clinical trials, is composed of a human leukocyte antigen (HLA) complex, HLA-A*0201, a peptide epitope derived from the Wilms Tumor 1 (WT1) protein, and 4 molecules of reduced affinity human interleukin-2 (IL-2). In preclinical studies CUE-102 binds, activates and expands WT1-specific CD8+ T cells with the potential to enhance anti-tumor immunity in patients with WT1+ cancers. Here we present the results of the dose-finding portion (Part A) of the CUE-102-01 study. Methods: CUE-102-01 is a 2 part phase 1 study. HLA-A*02:01 positive patients with WT1+ advanced colorectal (CRC), gastric (GA) and gastroesophageal junction (GEJ), pancreatic (PDAC) or ovarian (OvCa) tumors refractory to standard therapies were eligible for participation. WT1 status is determined by IHC via central testing. CUE-102 monotherapy was administered intravenously every three weeks in doses ranging from 1 mg/kg to 8 mg/kg following 3+3 design rules with a Bayesian Logistic Regression Model overlay. Dose levels that exhibit an immune or tumor response may be expanded to further characterize activity and toxicity as allowed by safety rules. Results: Dose escalation was successfully completed in 28 patients with CRC (17), PDAC (6), GA/GEJ (3) and OvCa (2). No DLTs were observed. Three dose levels (2, 4 and 8 mg/kg) exhibited evidence of immune activity and were expanded up to 9 patients each for further evaluation. CUE-102 was well-tolerated with > 96% of treatment-related adverse events (TRAEs) grade ≤2. The most common TRAEs, were fatigue, infusion reactions, nausea and vomiting. There were no cases of CRS Grade ≥ 2. Preliminary PK data demonstrate dose dependent increases in exposure across the dose range assessed. Preliminary analysis of patient samples demonstrates selective expansion of WT1-specific CD8+ T cells. Stable disease of ≥ 6 weeks (range 6-36 weeks), as determined by RECIST 1.1, has been observed in 10 of 23 patients (DCR 43.5%). Conclusions: CUE-102 is a novel T cell engager that demonstrates acceptable tolerability, favorable PK, and supportive preliminary PD readouts. No DLTs or drug-related SAEs have been observed in doses up to 8 mg/kg as of the data cut-off. Adverse events have been manageable and consistent with the CUE-102 mechanism of action and underlying disease. Early signs of disease stabilization and anti-tumor activity in late-stage patients are encouraging. Part B is a dose expansion that further evaluates safety, PK/PD, recommended phase 2 dose (RP2D), antitumor efficacy and is currently enrolling patients. Clinical trial information: NCT05360680 .