Abstract 538: Liquid biopsy approaches for determining pathologic response to neoadjuvant immunotherapy in esophageal cancer

We examined the utility of serial liquid biopsies to monitor clonal dynamics and predict pathologic response in patients with esophageal/gastroesophageal junction (E/GEJ) cancer undergoing treatment with neoadjuvant immunotherapy and concurrent chemoradiation (CA209-906; NCT03044613).

Methods: Using Targeted Error Correction sequencing (TEC-Seq), we performed high-depth next generation sequencing on 79 serial plasma samples from 16 patients with operable stage II/III E/GEJ cancer undergoing treatment with neoadjuvant nivolumab, followed by nivolumab plus chemo-radiation and surgery as part of the CA209-906 trial. Liquid biopsies were evaluated pre-treatment, after each of two cycles of neoadjuvant nivolumab, and after concurrent nivolumab and chemoradiation immediately prior to surgery, for an average of 4 time points per patient. Deep sequencing of matched leukocyte DNA and whole exome sequencing (WES) of pre-treatment tumors was performed. The origin of plasma variants was determined by comparison to tumor WES (tumor-derived) and leukocyte TEC-Seq (clonal hematopoiesis (CH)-derived or germline) and their clonal dynamics over time were evaluated. Resected tumors were reviewed by central pathology and percent residual tumor was estimated. Patients were followed for an average of 2 years post-surgery.

Results: Eight of 16 patients had detectable circulating tumor-derived DNA (ctDNA) at any time point. Additionally, 13 CH-derived mutations were detected in plasma of 8 patients. The number of CH-derived mutations was correlated with increasing patient age. Identification and removal of CH-derived mutations via comparison to matched leukocyte sequencing allowed for accurate assessment of kinetics of bona fide tumor-derived mutations in plasma. Tumor mutation burden as determined by tumor WES was not associated with pathological response (mean baseline mutations per exome 90 vs 81 for patients with residual disease >20% and <20% respectively). Detectable ctDNA at the last pre-surgery time point was found in 3 patients and was associated with residual tumor >20% (50% vs 23% with or without detectable ctDNA respectively). ctDNA clearance, defined as detectable ctDNA at one or more earlier time points that subsequently becomes undetectable before surgery, occurred in 5 patients and was associated with improved pathologic response (80% of patients with ctDNA clearance had residual tumor <=20% and no evidence of disease progression). Furthermore, of the three patients who did not have ctDNA clearance, two subsequently developed disease progression.

Conclusion: Comprehensive analyses of ctDNA from E/GEJ patients undergoing neoadjuvant immunotherapy with concurrent chemoradiation revealed ctDNA dynamics that were associated with pathologic response and disease recurrence. These approaches may be used to identify patients at high risk for disease progression.

Citation Format: Jenna VanLiere Canzoniero, Vincent Lam, Zineb Belcaid, Mara Lanis, Lamia Rhymee, Blair Landon, Archana Balan, Kristen Marrone, Patrick Forde, Benjamin Levy, David Ettinger, Heather Schneider, Hanika Rodavia, Richard Battafarano, Stephen Yang, Stephen Broderick, Jinny Ha, Blair Jobe, Ali Zaidi, Elizabeth Thompson, Julie Brahmer, Victor Velculescu, Ronan Kelly, Josephine Feliciano, Valsamo Anagnostou. Liquid biopsy approaches for determining pathologic response to neoadjuvant immunotherapy in esophageal cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 538.

Immunogenomic features of pathologic response to neoadjuvant immune checkpoint blockade in esophageal cancer

4042

Background: Improving immunotherapy efficacy remains an unmet need in esophagogastric cancer and a deeper understanding of tumor and immune system dynamics during therapy may tailor immuno-oncology approaches. Methods: We performed whole exome sequencing (WES) and bulk RNA sequencing (RNAseq) of 70 serial tumor samples from 23 patients with stage II/III esophageal/gastroesophageal junction (E/GEJ) cancer treated on a phase 1B clinical trial with neoadjuvant nivolumab with or without relatlimab (anti-LAG-3) and chemoradiation followed by surgery (NCT03044613; CA209-906). Pathologic response was measured by tumor regression at the time of resection. Median follow up was 23 months post-surgery. Serial tumor samples were collected prior to therapy, after 2 cycles of induction immune checkpoint blockade (ICB), and at the time of resection. Twenty-two baseline tumor/normal DNA pairs were analyzed by WES and 48 serial tumor samples were analyzed by RNAseq. WES data was analyzed to identify somatic mutations, generate tumor mutation burden (TMB) estimates and assess the fraction of expressed mutations in conjunction with RNAseq data. Immune cell subset composition was determined by RNAseq data deconvolution by CIBERSORT and gene set enrichment analyses were performed utilizing GSEA. B-cell density was inferred by immunoglobulin rearrangements detected by RNAseq. Results: Gene set enrichment expression analyses revealed an upregulation of effector pro-inflammatory cytokines after induction ICB. Interferon-gamma, interferon-alpha and TNF-alpha related genes were significantly upregulated after induction ICB compared to baseline (p < 0.0001). In contrast, significant downregulation of E2F targets (p = 0.002), G2M checkpoint genes (p = 0.005) and DNA damage repair genes (p = 0.004) was observed post ICB; enrichment analyses were independent of response to therapy and treatment arm. While TMB was not predictive of pathologic response (p = 0.22), patients with tumors harboring a higher number of expressed mutations were more likely to achieve a pathologic complete response (pCR; p = 0.026). RNAseq deconvolution analyses revealed a higher B-cell density post ICB induction in tumors with pCR (p = 0.018). Furthermore, an increased baseline content of intra-tumoral activated M1 macrophages differentiated tumors from patients achieving a pCR (p = 0.0034), which was further exemplified post induction ICB. Conclusions: Neoadjuvant immunotherapy induces an inflammatory immune response in the tumor microenvironment that is linked with tumor elimination and pathologic response. Our findings highlight the importance of nuanced multi-omics analyses to understand the wiring of response to immunotherapy and guide therapy for E/GEJ cancer.