SO-36 The prognostic impact of mismatch repair phenotype in colon cancer is highly stage-dependent

Mismatch repair deficiency (dMMR) is associated with an improved prognosis in stage II colon cancer, as reflected in current treatment guidelines. However, for patients with colon cancer stage IV, dMMR is a marker of poor prognosis in many studies. The reason for this stage-dependent function of MMR phenotype is not known. In this study, we assess whether the prognostic impact of dMMR differs between stage II and stage III colon cancer, adjusting for possible confounders including density of tumor-infiltrating CD3+ and CD8+ lymphocytes (TILs), expression of caudal-type homeobox transcription factor 2 (CDX2), and administration of adjuvant chemotherapy. The study included 544 patients with surgically treated colon cancer stage II and III with thorough clinical and follow-up data. Immunohistochemistry for CD3+ and CD8+ T lymphocytes, CDX2 and MMR proteins MLH1, MSH2, MSH 6 and PMS2 was performed on Tissue Micro Arrays. The study included colon cancer patients from two different cohorts, including a cohort of patients randomized to surgery only or to receive adjuvant fluorouracil-based chemotherapy after surgery, allowing us to also study the predictive value of MMR status on the effect of adjuvant chemotherapy. In stage III colon cancer patients, dMMR status conveyed a significantly worse mean cancer-specific survival (CSS) compared to patients with pMMR (38.1 months; 95% CI 28.4-47.9 vs. 48,1 months; 95% CI 45.1-51.1; p=0.021). CSS in stage II colon cancer for patients with dMMR was 58.0 (95% CI 55.9-60.0) versus 56.3 (95% CI 54.8-57.9) months for patients with pMMR. The tendency towards improved survival for patients with dMMR was not statistically significant (p=0.147). A stage-dependent prognostic impact of dMMR status was demonstrated in our multivariate cox regression models where a significant interaction between MMR phenotype was found (p=0.001). dMMR was a marker of poor prognosis in stage III colon cancer (HR 2.55; 95% CI 1.13-5.71; p=0.023) but a marker of improved prognosis in stage II (HR 0.25; 95% CI 0.071-0.89; p=0.032). Our multivariate model adjusted for the effects of tumor-infiltrating CD3+ and CD8+ lymphocytes (TILs), CDX2 expression, adjuvant chemotherapy, differentiation, different hospitals, age, and sex. CDX2 expression and density of tumor-infiltrating lymphocytes did not impact the prognosis of dMMR patients. For stage II colon cancer stage there was a tendency towards a worsened CSS for patients with dMMR receiving adjuvant chemotherapy compared to patients going through surgery only (54,7; 95% CI 48.6-60.7 vs. 59.0; 95% CI 57.1-60.8; p=0.058). In stage III, adjuvant chemotherapy did not affect the prognosis of dMMR patients. Our study demonstrates the stage-dependent prognostic value of MMR phenotype in colon cancer stage II and III. As shown in other studies, dMMR phenotype is associated with an improved prognosis in stage II colon cancer. Our study demonstrated that dMMR status is a poor prognostic marker in colon cancers with lymphatic spread (stage III), in compliance with results previously seen in dMMR patients with hematological spread (stage IV). This stage-dependent difference is not explained by variations in CDX2 expression, density of TILs or administration of adjuvant chemotherapy.