Poster, Podium & Video Sessions
Presentation Authors: Philip Abbosh*, Philadelphia, PA, David Liu, Boston, MA, Woonyoung Choid, Houston, TX, Wafik El-Deiry, Philadelphia, PA, Jonathan Rosenberg, New York, NY, David McConkey, Baltimore, MD, Elizabeth Plimack, Philadelphia, PA, Eliezer Van Allen, Boston, MA
Introduction: The immune system is increasingly recognized as both a key player in cancer control and as druggable target. We hypothesized that the immune system impacts pathologic response in patients undergoing cisplatin-based neoadjuvant chemotherapy (NAC) for muscle-invasive bladder cancer though the immune system.
Methods: Whole exome sequencing (WES) was performed on tumor DNA from patients in two independent cohorts who underwent cisplatin-based NAC. The Memorial Sloan Kettering / Dana Farber Cancer Institute (n=50; 25 responders) and Philadelphia (n=48; 20 responders) cohorts were treated with gemcitabine / cisplatin or methotrexate / vinblastine / doxorubicin / cisplatin. Mutation analysis was performed using standard analytical pipelines. Macrohistocompatibility complex (MHC)-restricted neoantigens were identified with netMHCpan and PolySolver. High affinity neoantigens were defined to have ≤500 nM binding affinity (Kd). Pathologic response was defined as ≤ypTis cystectomy specimen. CIBERSORT was used to infer immune cell infiltrate based on DASL Illumina microarray expression profiles (n=41; 17 responders).
Results: Chemoresponders had >twice as many putative neoantigens as nonresponders (471 vs 207 neoantigens respectively; p=1x10-6, Wilcoxon). This relationship maintained significance when neoantigens were limited to Kd≤100 nM or ≤50 nM, and when chemoresponse was defined as ypT0-only, or ≤ypT1. ERCC2 loss-of-function mutations were 36% sensitive in identifying chemoresponders but 96% specific. Above-median neoantigen burden was 81% sensitive in identifying chemoresponders and 78% specific. CD8+ cells were enriched in responders (13.5% vs 8.4% of infiltrate; p<0.008, t-test) as were activated NK cells (15.5% vs 12.7%; p=0.05). In 28 samples where WES and CIBERSORT data was available, the neoantigen burden was correlated to the CD8+ infiltrate value (R2=0.33, Pearson).
Conclusions: Neoantigen burden and CD8+ infiltrate correlate strongly with chemoresponse. Neoantigen burden and CD8+ infiltrate are directly proportional. NAC may therefore exert known tumor cell autonomous effects and an extrinsic effect involving the immune system via neoantigens. Anti-neoantigen responses have been shown to impact immune checkpoint responses and a similar mechanism may mediate cytotoxic chemoresponse. Experiments are underway to directly and functionally characterize the effect of the immune system in chemoresponse.
Source Of Funding: PHA-AUA Foundation Research Scholar Award, Kidney Cancer Association Young Investigator Award
WSED-NCI #CA173453, CA176289, CA181419
JER- Starr Cancer Consortium
EVM-NCI #CA188615, Starr Cancer Consortium