Abstract 2576: Radiomic features of intrathoracic lesions can predict the mixed response in non-small cell lung cancer treated with immunotherapy

Abstract Background: Although immunotherapy has emerged as a promising treatment for lung cancer, the decision to continue immunotherapy becomes challenging when mixed response (MR) occurs. This study is aimed to evaluate whether radiomic features from pre-treatment intrathoracic images can predict MR. Methods: 127 consecutive patients with NSCLC who received first line systemic treatment containing immunotherapy were included in this study. Chest CT scans were obtained at baseline and 1-3 times from 11 different CT scanners and harmonization method was applied to minimize scanner-related bias. The durable response of intrathoracic lesions (n=266) at least 24 weeks was evaluated based on RECIST 1. 1 and irRECIST. MR was defined as the simultaneous presence of at least one lesion that increased and decreased in a single patient during immunotherapy and two different criteria of MR for the change of tumor lesions were applied: Definition-A (MR-A), Δ baseline - first follow-up ≥ 5 mm; Definition-B (MR-B), any change of Δ baseline - first follow-up. All intrathoracic lesions were identified and segmented by four different physicians. Radiomic features, including morphological, intensity, GLCM, GLRLM, GLSZM, and NGTDM, were extracted using LIFEX software (IMIV/CEA, Orsay, France). Confusion matrix was used to assess tumor responses, and area under the curve (AUC) for intrathoracic lesions was calculated. Results: Tumor responses based on irRECIST were CR (n=1, 0. 8%), PR (n=33, 26. 0%), SD (n=55, 43. 3%), and PD (n=39, 30. 7%). There were 10 cases (7. 9%) of MR-A and 25 cases (19. 7%) of MR-B. While the median progression-free survival (PFS) and overall survival (OS) in MR-A group were 22. 9 (range, 0. 5-50. 6) and 23. 5 (range, 1. 1-42. 2) months, those in MR-B group were 2. 0 (range, 0. 5-50. 6) and 3. 3 (range, 0. 7-49. 0) months respectively. While the overall sensitivity (SN) and specificity (SP) for MR-A in intrathoracic lesions were 0. 55 and 0. 87, those for MR-B were 0. 52 and 0. 71, respectively. The AUC for MR-A and MR-B were 0. 71 and 0. 62 respectively, and the AUC for durable response was 0. 56. In the SD group, median PFS and OS were 3. 3 (1. 4-18. 9) and 6. 1 (4. 8-26. 2) in the MR-A group and 9. 7 (0. 7-18. 9) and 23. 9 (0. 7-26. 2) respectively in the MR-B group. Median PFS and OS were 18. 2 (0. 5-65. 2) and 23. 9 (0. 9-55. 3) respectively in the non-MR-A group and 18. 2 (0. 5-65. 2) and 23. 9 (0. 9-55. 3) respectively in the non-MR-B group. In the PD group, median PFS and OS were 23. 5 (0. 5-41. 3) and 24. 7 (1. 1-42. 2) respectively in the MR-A group and 5. 1 (0. 5-41. 3) and 11. 5 (1. 1-49. 0) respectively in the MR-B group. Median PFS and OS were 20. 3 (0. 7-71. 4) and 37. 1 (1. 0-72. 9) respectively in the non-MR-A group and 20. 3 (0. 7-71. 4) and 37. 1 (1. 0-72. 9) respectively in the non-MR-B group. Conclusion: The radiomic features in NSCLC patients who received immunotherapy may help predict MR in NSCLC patients treated with immunotherapy. Citation Format: Monica Yadav, Jeeyeon Lee, Peter Haseok Kim, Maria J. Chuchuca, Taegyu Um, Liam Il-Young Chung, Yury S. Velichko, Young Kwang Chae. Radiomic features of intrathoracic lesions can predict the mixed response in non-small cell lung cancer treated with immunotherapy abstract. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts) ; 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84 (6Suppl): Abstract nr 2576.