Predictive score for perineural invasion in colorectal cancer: A SEER-based case-control analysis.

[BACKGROUND] Although perineural invasion, a negative prognosticator in colorectal cancer, is typically detected after surgical resection, preoperative prediction may be useful. We used a large population-based registry to develop a predictive model of perineural invasion in colorectal cancer, including clinical and pathologic features of tumors.

[METHODS] Retrospective case-control analysis of patients with colorectal adenocarcinoma from the Surveillance, Epidemiology, and End Results Research Data database. The study involved a case-control analysis of predictive factors of perineural invasion in colorectal cancer. The main outcome measures were perineural invasion, overall survival, and cancer-specific survival.

[RESULTS] The study included 223,468 patients (52% male, mean age: 66.4 years). Perineural invasion was detected in 13.1%. Independent predictors of perineural invasion were age (odds ratio [OR], 0.99; P < .001), male sex (OR, 1.05; P = .009), Black race (OR, 1.19; P < .001), American Indian race (OR, 0.69; P = .001), left colon cancer (OR, 1.29; P < .001), rectal cancer (OR, 1.32; P < .001), moderately differentiated adenocarcinomas (OR, 1.41; P < .001), poorly differentiated adenocarcinomas (OR, 2.28; P < .001), undifferentiated carcinomas (OR, 2.13; P < .001), mucinous adenocarcinomas (OR, 0.64; P < .001), N1 stage (OR, 2.28; P < .001), N2 stage (OR, 3.83; P < .001), and elevated carcinoembryonic antigen levels (OR, 1.45;P < .001). These predictors were integrated into a risk prediction score, with an excellent negative predictive value of 93%. Perineural invasion was associated with lower 5-year overall survival (54 vs. 76.4%; P < .001) and cancer-specific survival rates (56.6 vs 81.8%; P < .001).

[CONCLUSIONS] Age, sex, race, tumor location, histology and grade, lymph node involvement, and carcinoembryonic antigen levels were independently associated with perineural invasion. Perineural invasion was an independent predictor of liver and lung metastases and worse overall survival and cancer-specific survival. The Cleveland Clinic Florida-Perineural Invasion Prediction score had an excellent negative predictive value in excluding patients without perineural invasion. Pending those trials, these findings may be considered in multidisciplinary team management conferences and in shared decision-making.