Nathan Bahary

DEG Table for Pre vs Post (mRNAseq)

DEG Table for R vs NR (mRNAseq)

Supplementary Figures 1-5 and Supplementary Tables 1-2

Purpose:Determinants of treatment outcomes to chemotherapy-based regimens in metastatic pancreatic ductal adenocarcinoma (PDA) remain ill-defined. Our aim was to examine tissue-based correlates of treatment response and resistance using matched baseline and on-treatment biopsies collected from patients with PDA treated in the first-line metastatic setting.Experimental Design:Patients with treatment-naïve metastatic PDA were enrolled in a Phase II trial (NCT02077881) investigating gemcitabine plus nab-paclitaxel in combination with indoximod, an orally administered small-molecule inhibitor of the IDO pathway. Baseline and on-treatment biopsies (week 8) of metastatic lesions (88% liver) were collected from a cohort of responders (N = 8) and non-responders (N = 8) based on RECIST v1.1 and examined by multiplex IHC and mRNA sequencing.Results:Treatment altered the transcriptional profile of metastatic lesions with a decrease in tumor cell proliferation independent of treatment response. The antiproliferative response was seen in both basal and classical PDA subtypes. PDA subtype was not associated with survival outcomes; instead, genes involved in immune activation distinguished responders from non-responders. Tumor response was associated with an increase in CD3+ and CD8+ T-cell infiltrates into metastatic lesions. A composite of decreased tumor proliferation in response to treatment and increased CD8 T-cell infiltration in metastatic lesions identified responders and associated with a favorable survival outcome.Conclusions:Our findings suggest that inhibiting cancer cell proliferation alone in PDA is insufficient to produce tumor responses and support a role for tumor-extrinsic mechanisms, such as CD8+ T cells, which combine with the cancer cell proliferation index to define treatment outcomes.

Supplementary Table S7. Summary of Cox proportional hazards model for overall survival.

Supplementary Table S13. Summary of GLIPH2 TCR clustering analyses across patients analyzed.

Supplementary Table S6. Summary of significantly changing TCR clones across the study cohort

Supplementary Fig.S8 Overview of peripheral blood TCR dynamics in patients with immunotherapy-related pneumonitis. (A) Swimmer plot showing treatment duration for 17 patients with immune-related toxicity and paired baseline and on-treatment blood samples available for analysis. The time to identification of significant TCR clonotypic expansions and regressions in first on-treatment blood samples is shown alongside the time to clinical diagnosis of toxicity. Overall, TCR clonotype dynamics were identified on average 20 weeks prior to the clinical presentation of immune-related toxicity in these patients. (B-E) Logo plots showing global TCR clusters identified using GLIPH2 that were significantly expanded in on-therapy peripheral blood samples from patients who developed pneumonitis compared to individuals who developed other immune-related adverse events (irAEs) and individuals who did not develop immune-related toxicity. The size of each cluster, defined by the number of distinct CDR3 member sequences, is indicated alongside each panel. Enriched clusters with a minimum motif length of 5 are shown. (F-G) Boxplots showing the differential abundance of TCR clusters shown in (C-E) from baseline to on-treatment samples across patient groups.

Supplementary Table S12. Summary of global TCR clusters identified in at least 3 distinct patients using GLIPH2.

Supplementary Table S12. Summary of global TCR clusters identified in at least 3 distinct patients using GLIPH2.