Assessment of pathologic response in the colorectal cancer cohort of the IMHOTEP phase II trial of neoadjuvant pembrolizumab in dMMR/MSI tumors.

241 Background: Neoadjuvant treatment with immune check point inhibitors has shown promising results in localized deficient mismatch repair/microsatellite instability (dMMR/MSI) colorectal cancer (CRC), notably in terms of pathologic response. However, pathologic response has rarely been precisely assessed on both primary tumor and its corresponding lymph nodes (LN) as an end-point. We performed this analysis on the CRC samples selected from the IMHOTEP trial, a multicenter, single-arm study evaluating peri-operative Pembrolizumab in 4 different cohorts of patients with localized resectable dMMR/MSI tumors. Methods: We analyzed pathologic response on the 56 surgical specimens available from the CRC cohort. Pathologic response was assessed on primary tumor according to the percentage of Residual Viable Tumor (RVT) score, encompassing 4 groups : 0% = pathologic Complete Response (pCR) ; ≤10% = Major Pathologic Response (MPR) ; >10% < 50% = partial Pathologic Response (pPR) ; > 50% = lack of response (noPR). Pathologic response was assessed on lymph node (LN) according to 4 groups : negative LN without sign of sterilization (yN0 reg-), negative LN with sterilization (yN0 reg+) ; metastatic LN without tumor regression (yN+ reg-) ; metastatic LN with tumor regression (yN+ reg+).The type of regression, including colloid, fibrotic and necrotic features was reported, as well as the presence of tertiary lymphoid structures (TLS). pCR and MPR corresponded to pathologic responder patients and pPR and noPR to pathologic non responder patients. Results: In thisCRC cohort pCR, MPR, pPR and noPR were observed in n= 33 (58.9%), 7 (12.5%), 5 (8.9%) and 11 (19.6 %) cases and yN0 reg-, yN0 reg+, yN+ reg-, yN+ reg+ in 41 (73.2%) , 7 (12.5%), 7 (12.5%), and 1 (1.8 % ) cases. There was a significant association between pathologic responders (pCR + MPR) and negative LN status encompassing sterilized LN (p=0.04). Pathologic responders were more frequently associated with classical (Lieberkuhnian) adenocarcinoma than with other CRC special subtypes (mucinous, signet ring-cell, poorly cohesive subtypes)(p<0.01). Colloid response was the most frequent pattern of regression, observed in 42 (75%) cases, and predominant (>80% of pathologic response areas) in 14 (25%) cases. TLS were seen in 46 (82%) cases, without any significant association with the RVT score. However TLS were observed in all CRC cases with MPR. Conclusions: These preliminary results show pembrolizumab efficacy as a neoadjuvant monotherapy in terms of pathologic response on both primary CRC and corresponding LN, highlighting the need for a new and more accurate pathologic score. It also illustrates the potential impact of the histological subtypes. Complementary data will be available after updated pathologic reviewing of additional specimens from CRC patients and will also be correlated with outcome. Clinical trial information: NCT04795661 .