[Multivariate analysis and construction and validation of a nomogram model from data of 1610 patients with non-tumor-related anastomotic stenosis after rectal cancer surgery]
K M Qiu,W Jian,J X Zheng,M Y Feng,X M Liu,D S Lu,J Yan
DOI: https://doi.org/10.3760/cma.j.cn441530-20230926-00112
2024-06-25
Abstract:Objective: To assess the risk factors affecting development of non-tumor- related anastomotic stenosis after rectal cancer and to construct a nomogram prediction model. Methods: This was a retrospective study of data of patients who had undergone excision with one-stage intestinal anastomosis for rectal cancer between January 2003 and September 2018 in Nanfang Hospital of Southern Medical University. The exclusion criteria were as follows: (1) pathological examination of the operative specimen revealed residual tumor on the incision margin of the anastomosis; (2) pathological examination of postoperative colonoscopy specimens revealed tumor recurrence at the anastomotic stenosis, or postoperative imaging evaluation and tumor marker monitoring indicated tumor recurrence; (3) follow-up time <3 months; and (4) simultaneous multiple primary cancers. Univariate analysis using the χ2 or Fisher's exact test was performed to assess the study patients' baseline characteristics and variables such as tumor-related factors and surgical approach (P<0.05). Multivariate analysis using binary logistic regression was then performed to identify independent risk factors for development of non-tumor-related anastomotic stenosis after rectal cancer. Finally, a nomogram model for predicting non-tumor-related anastomotic stenosis after rectal cancer surgery was constructed using R software. The reliability and accuracy of this prediction model was evaluated using internal validation and calculation of the area under the curve of the model's receiver characteristic curve (ROC). Results: The study cohort comprised 1,610 patients, including 1,008 men and 602 women of median age 59 (50, 67) years and median body mass index 22.4 (20.2, 24.5) kg/m². Non-tumor-related anastomotic stenosis developed in 121 (7.5%) of these patients. The incidence of non-tumor-related anastomotic stenosis in patients who had undergone neoadjuvant chemotherapy, neoadjuvant radiotherapy, and surgery alone was 11.2% (10/89), 26.4% (47/178), and 4.8% (64/1,343), respectively. Neoadjuvant treatment (neoadjuvant chemotherapy: OR=2.455, 95%CI: 1.148-5.253, P=0.021; neoadjuvant chemoradiotherapy, OR=3.882, 95%CI: 2.425-6.216, P<0.001), anastomotic leakage (OR=7.960, 95%CI: 4.550-13.926, P<0.001), open laparotomy (OR=3.412, 95%CI: 1.772-6.571, P<0.001), and tumor location (distance of tumor from the anal verge 5-10 cm: OR=2.381, 95%CI:1.227-4.691, P<0.001; distance of tumor from the anal verge <5 cm: OR=5.985,95% CI: 3.039-11.787, P<0.001) were identified as independent risk factors for non-tumor-related anastomotic stenosis. Thereafter, a nomogram prediction model incorporating the four identified risk factors for development of anastomotic stenosis after rectal cancer was developed. The area under the curve of the model ROC was 0.815 (0.773-0.857, P<0.001), and the C-index of the predictive model was 0.815, indicating that the model's calibration curve fitted well with the ideal curve. Conclusion: Non-tumor-related anastomotic stenosis after rectal cancer surgery is significantly associated with neoadjuvant treatment, anastomotic leakage, surgical procedure, and tumor location. A nomogram based on these four factors demonstrated good discrimination and calibration, and would therefore be useful for screening individuals at risk of anastomotic stenosis after rectal cancer surgery.