Immunologic predictors of therapeutic response to neoadjuvant chemotherapy for pancreatic ductal adenocarcinoma (PDA) in SWOG S1505.

Abstract

419 Background: A role for the immune system in predicting treatment outcomes in PDA has not been well-studied prospectively, especially in the curative setting. Here, we investigated the capacity of two established chemotherapy regimens to trigger an immune response against PDA when administered as neoadjuvant therapy. Methods: We used available PDA surgical specimens from SWOG S1505 (a randomized phase II trial of perioperative chemotherapy with either mFOLFIRINOX [FFX] or gemcitabine/nab-paclitaxel [GA] for resectable PDA; ASCO 2020 abstr 4504) resected after 3 months of neoadjuvant chemotherapy, and matched untreated controls from the biorepository at the University of Utah. Multiplex immunohistochemistry (IHC) for CD8, CD68, CK19, FOXP3, PDL1, CD3, and Ki67 was performed; regions of interest (ROIs) were transcribed by a central pathologist. ROIs were digitally quantified using custom image analysis algorithms created using Visiopharm Integrator System software to detect and classify cells within superimposed grids for quantification; medians and coefficients of variation (CV) were calculated. Wilcoxon signed-rank test and Cox regression models were used to analyze associations between IHC cell counts and pathologic response (pR) and overall survival (OS), respectively. Results: IHC cell counts varied between treated (Rx) (n = 57; FFX = 34, GA = 23) and control (C) (n = 61) specimens for: CK19 percent area was 7.9 vs 15.6 (p < 0.001); total Ki67/mm2 was 211 vs 400 (p < 0.001); total CD3/mm2 was 376 vs 676 (p < 0.001); Foxp3/mm2 was 81.5 vs 152.8 (p < 0.001); CD8/mm2 was 313 vs 477 (p < 0.001); CD68/mm2 was 507 vs 741 (p = 0.015). PDL1 expression was undetectable in the majority of specimens, both Rx and C. FFX (vs GA) was associated with fewer CK19+Ki67+ cells (459 vs 1026, p = 0.017) but more CD68+ cells (33,241 vs 13,334, p = 0.007) and slightly more CD8+ cells (19,809 vs 14,344, p = 0.049). In all Rx patients, complete/major (n = 19) vs poor/no (n = 36) pR was associated with total CD3/mm2: 461 vs 308 (p = 0.019); the other parameters showed no notable differences. OS showed no remarkable associations with the tested parameters. Conclusions: We have demonstrated: 1) Decreased tumor cells and proliferating cells in Rx vs C samples, as expected. This decrease is more pronounced with FFX compared with GA; 2) Decreased total CD3+ T cells as well as regulatory Foxp3+ T cells in Rx vs C samples, which is unexpected; 3) A reduction in CD68+ myeloid cells in Rx vs C samples, which is expected but more pronounced with GA compared with FFX; 4) Rx samples did not show an increased expression of PDL1, compared with C; and 5) Improved pR was associated with increased T cell infiltrate, pointing toward a possible mechanism. Together, these data support the capacity of neoadjuvant chemotherapy to modulate the immune response to PDA. Therapeutic implications of such changes merit further investigation. Clinical trial information: NCT02562716.