Main Session
Sep 28
SS 05 - Lung 1: NSCLC Locally Advanced and Oligometastatic

130 - High-Dimensional Immunophenotyping of Patients with Localized Non-Small Cell Lung Cancer (NSCLC) Reveals Distinct Immune Cell Subsets Predictive of Disease-Free Survival (DFS) in AFT-16 and LCMC3 Phase II Immunotherapy Clinical Trials

03:00pm - 03:10pm PT
Room 156/158

Presenter(s)

Nicholas Eustace, MD, PhD Headshot
Nicholas Eustace, MD, PhD - City of Hope National Medical Center, Duarte, CA

N. J. Eustace1, M. Seweryn2, M. Pietrzak3, G. Lozanski3, F. Oezkan4, T. Talabere3, A. Nicholas5, A. Johnson6, K. Schulz7, R. Helen8, D. E. Kozono9, J. Urbanic10, G. Nelson11, D. P. Carbone12, and T. M. Williams1; 1Department of Radiation Oncology, City of Hope National Medical Center, Duarte, CA, 2University of Lodz., Lodz, Poland, 3The Ohio State University, Columbus, OH, 4University Medicine Essen-Ruhrlandklinik, Essen, Germany, 5Genentech, South San Francisco, CA, 6Genentech, Inc., South San Francisco, CA, 7Genetech, South San Francisco, CA, 8Rush University Medical Center, Chicago, IL, 9Department of Radiation Oncology, Dana-Farber Cancer Institute/Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, 10Univesity of California San Diego, San Diego, CA, 11Mayo, Rochester, MN, 12The Ohio State University, Department of Internal Medicine, Division of Medical Oncology, Columbus, OH

Purpose/Objective(s): Immune checkpoint inhibitors (ICIs) have benefits in localized NSCLC, but predictive markers are needed to determine which patients benefit most. We analyzed peripheral blood mononuclear cells (PBMCs) using high-dimensional flow-cytometry-based-immunophenotyping (IMMUNOME) to identify immune cell subset surface marker combinations (SMCs) at screening associated with improved clinical outcomes in two prospective phase II trials: LCMC3 neoadjuvant +/- adjuvant atezolizumab with surgically resectable stage Ib-III NSCLC (NCT02927301), and AFT-16 neoadjuvant and adjuvant atezolizumab with inoperable stage III NSCLC receiving chemoradiation (NCT03102242).

Materials/Methods: From screening samples of 120 patients treated on LCMC3, we identified 1,139 SMCs above a median percentage of 0. The SMCs were first grouped into 5 clusters and then into clusters of 11-29 SMCs. A synthetic dataset was generated using a generalized adversarial network (GAN) approach to select the final set of SMCs using two criteria: (1) informativeness in a random forest model for RECIST response and survival in LCMC3 and (2) goodness of fit of the synthetic data to the actual distribution of each feature. The top 5 informative SMCs were chosen by random projection method from LCMC3 and validated on AFT-16 (39 patients).

Results: Using OS and RECIST response in LCMC3 as the training dataset, these 5 SMCs were associated with the best RECIST response in AFT-16, yielding an AUC of 0.812.

Cytotoxic T Cell (late activation): CD45+ CD3+ CD8+ CD274+ CD4- CD13- CD14- CD49a- CD63- CD107a/b-

NK: CD45+ CD56+ CD16+ CD3- CD94+ CD117- CD127- CD159a- CD161- CD314+

NK: CD45+ CD56- CD16+ CD3- CD94- CD117- CD127- CD159a- CD161- CD314-

Myeloid Dendritic Cell (MDC): CD45+ CD11b+ CD1c+ CD15+ CD16+ CD33+ HLADR+ LIN-

Classical Monocyte: CD45+ CD11b+ CD14+ CD15+ HLADR+ CD16- CD33+ CD66b- CD124-

Using OS in LCMC3 as the training dataset, these 5 SMCs were associated with DFS in AFT-16 yielding an AUC of 0.799.

NKT: CD45+ CD56+ CD3+ CD4+ CD16+ CD8- CD19- CD69- CD134- HLADR-

MDC: CD45+ CD1c+ CD33+ HLADR+ CD11b- CD15- CD16- LIN-

PMN MDSC: CD45+ CD11b+CD15+ CD33+ CD16- HLADR- LIN-

NK: CD45+ CD56+ CD3- CD107a/b+ NKp80+ CD16B- CD158a- CD158b- CD158e1- CD159c-

T Cell (Memory): CD45+ CD3+ CD127+ CD314+ CD16- CD56- CD94- CD117- CD159a- CD161-

Conclusion: IMMUNOME identified 5 SMCs present pre-therapy associated with RECIST response and DFS in patients with NSCLC receiving neoadjuvant +/-adjuvant atezolizumab, and all were independent of PD-L1 status. No SMCs were shared between groups; but both contained subsets of NK, myeloid dendritic, and T cells. IMMUNOME is a powerful tool for studying a patient's immune landscape and identifying SMCs associated with differential responses to ICIs. Additional validation of these novel SMCs as distinct immune cell subtypes is warranted as IMMUNOME may be useful for prognostication, prediction of ICI response, or selecting patients for treatment intensification or de-intensification.