Use of Deep Learning to Evaluate Tumor Microenvironmental Features for Prediction of Colon Cancer Recurrence.

Deep learning may detect biologically important signals embedded in tumor morphologic features that confer distinct prognoses. Tumor morphologic features were quantified to enhance patient risk stratification within DNA mismatch repair (MMR) groups using deep learning. Using a quantitative segmentation algorithm (QuantCRC) that identifies 15 distinct morphologic features, we analyzed 402 resected stage III colon carcinomas [191 deficient (d)-MMR; 189 proficient (p)-MMR] from participants in a phase III trial of FOLFOX-based adjuvant chemotherapy. Results were validated in an independent cohort (176 d-MMR; 1,094 p-MMR). Association of morphologic features with clinicopathologic variables, MMR, KRAS, BRAFV600E, and time-to-recurrence (TTR) was determined. Multivariable Cox proportional hazards models were developed to predict TTR. Tumor morphologic features differed significantly by MMR status. Cancers with p-MMR had more immature desmoplastic stroma. Tumors with d-MMR had increased inflammatory stroma, epithelial tumor-infiltrating lymphocytes (TIL), high-grade histology, mucin, and signet ring cells. Stromal subtype did not differ by BRAFV600E or KRAS status. In p-MMR tumors, multivariable analysis identified tumor-stroma ratio (TSR) as the strongest feature associated with TTR [HRadj 2.02; 95% confidence interval (CI), 1.14-3.57; P = 0.018; 3-year recurrence: 40.2% vs. 20.4%; Q1 vs. Q2-4]. Among d-MMR tumors, extent of inflammatory stroma (continuous HRadj 0.98; 95% CI, 0.96-0.99; P = 0.028; 3-year recurrence: 13.3% vs. 33.4%, Q4 vs. Q1) and N stage were the most robust prognostically. Association of TSR with TTR was independently validated. In conclusion, QuantCRC can quantify morphologic differences within MMR groups in routine tumor sections to determine their relative contributions to patient prognosis, and may elucidate relevant pathophysiologic mechanisms driving prognosis. SIGNIFICANCE: A deep learning algorithm can quantify tumor morphologic features that may reflect underlying mechanisms driving prognosis within MMR groups. TSR was the most robust morphologic feature associated with TTR in p-MMR colon cancers. Extent of inflammatory stroma and N stage were the strongest prognostic features in d-MMR tumors. TIL density was not independently prognostic in either MMR group.

Efficacy of Difluoromethylornithine and Aspirin for Treatment of Adenomas and Aberrant Crypt Foci in Patients with Prior Advanced Colorectal Neoplasms.

Difluoromethylornithine (DFMO), an inhibitor of polyamine synthesis, was shown to act synergistically with a NSAID for chemoprevention of colorectal neoplasia. We determined the efficacy and safety of DFMO plus aspirin for prevention of colorectal adenomas and regression of rectal aberrant crypt foci (ACF) in patients with prior advanced adenomas or cancer. A double-blinded, placebo-controlled trial was performed in 104 subjects (age 46-83) randomized (1:1) to receive daily DFMO (500 mg orally) plus aspirin (325 mg) or matched placebos for one year. All polyps were removed at baseline. Adenoma number (primary endpoint) and rectal ACF (index cluster and total) were evaluated at a one year colonoscopy. ACF were identified by chromoendoscopy. Toxicity was monitored, including audiometry. Eighty-seven subjects were evaluable for adenomas or ACF modulation (n = 62). At one year of treatment, adenomas were detected in 16 (38.1%) subjects in the DFMO plus aspirin arm (n = 42) versus 18 (40.9%) in the placebo arm (n = 44; P = 0.790); advanced adenomas were similar (n = 3/arm). DFMO plus aspirin was associated with a statistically significant reduction in the median number of rectal ACF compared with placebo (P = 0.036). Total rectal ACF burden was also reduced in the treatment versus the placebo arm relative to baseline (74% vs. 45%, P = 0.020). No increase in adverse events, including ototoxicity, was observed in the treatment versus placebo arms. While adenoma recurrence was not significantly reduced by one year of DFMO plus aspirin, the drug combination significantly reduced rectal ACF number consistent with a chemopreventive effect.