P-87 Clinical score to predict recurrence in patients with stage II and III colon cancer

Abstract

Colorectal cancer is the 3º most common tumor worldwide. In patients with stage II and III colon cancer the prognosis is heterogeneous, and clinical and pathological characteristics, such as tumor budding, may help to further refine the recurrence risk. The aim of this study is to create a score to predict recurrence using clinical and pathological variables available in routine clinical practice and to select a subgroup of patients with excellent prognosis according to this score. We included all of the patients with pathologically confirmed diagnosis of stage II and III colon cancer at Hospital Universitario La Paz from October 2016 to September 2020. All statistical analyses were carried out using SPSS v.25. We performed a univariate and multivariate Cox regresión model for the endpoint Time to Recurrence (TTR). We built a prognostic score for recurrence assigning 1 point for each variable that remained P < 0.10 at the multivariate analysis. A total of 440 patients were included. 222 (50%) and 218 (50%) patients were diagnosed with stage II and III disease, and 48% were located in the right colon. After a median follow-up of 36 months (range, 0,1 to 56 months), 72 (16%) patients had a first tumor recurrence, and 80 (17%) patients died. Median TTR, and OS were not reached for the whole cohort. Univariate Cox regression analysis showed that T4, N2, R1, Stage III, bowel obstruction and perforation at diagnosis, lymphovascular and perineural invasion, high tumor budding, and deficient mismatch repair were significantly associated with TTR. Only T4 (hazard ratio (HR), 3.27 [95% confidence interval (CI): 1.52-7.00], p < 0.01), N2 (HR, 2.03 [95%CI, 0.99-4.16], p =0.05), R1(HR, 3.58 [95%CI, 1.77-7.21], p < 0.01) and high tumor budding (HR, 2.80 [95%CI, 1.56-5.03], p < 0.01) remained with a p value < 0.10 at the last step of the multivariate cox regression model. Based on these characteristics, patients were assigned from 0 to 4 points. A total of 135, 97, 52, 16, and 4 had 0,1,2,3, and 4 points, respectively. Freedom from recurrence at 24 months in patients with 0 to 4 points was 95%, 79%, 68%, 54% and 33% (p < 0.001). The area under the ROC curve for tumor recurrence at 24 months was 0.771 (95%CI, 0.65-0.85), p < 0.01. We compared patients with score = 0 (n =135; 44%) vs ≥ 1 (n = 169; 56%). Patients with score 0 had significantly longer median TTR (not reached (NR) in either group, p < 0.01), with a HR for disease recurrence of 0.13 (95%CI, 0.05-0.33), p < 0.01. 95%, and 72% of the patients were recurrence-free at 24 months in the score 0, and ≥1 groups, respectively. In this study, we built a simple score to accurately predict tumor recurrence based on T4, N2, R1 and high tumor budding. Patients with a score = 0, that comprises 44% of the cohort, had an excellent prognosis. The positive results of this score need to be confirmed in a validation cohort.