探讨直肠癌患者术后近期并发症发生的相关危险因素并构建临床预测列线图模型。

采用回顾性研究方法，收集分析2015年1月至2018年9月在南京医科大学第二附属医院普外科行直肠癌根治术的患者临床病理资料及相关手术信息资料。共纳入204例患者，其中男性患者135例，女性患者69例；病理TNM分期0期11例，Ⅰ期42例，Ⅱ期62例，Ⅲ期89例。根据2009版改良Clavien-Dindo手术并发症分级标准对患者术后出现的并发症进行分级，将Ⅱ级以上的并发症定义为临床有意义并发症。分析患者一般基线特征、病理特征及手术相关信息特征与临床有意义并发症发生之间的关系，对临床变量采用单因素、多因素分析。运用R软件（R4.0.3）绘制列线图临床预测模型，采用ROC曲线及C-index验证和评价列线图模型。

204例患者中，共有39例（19.11%）患者出现临床有意义并发症，术后常见的并发症依次为吻合口漏（14例，9.03%；行Dixon术式和结肠肛管吻合术共155例）、肠梗阻（14例，6.86%）、肺部感染（13例，6.37%）、肠造口相关并发症（5例，5.95%；包括Miles术式、Hartmann术式预防性造口共84例）、术后出血（9例，4.41%）、腹腔感染/盆腔感染（6例，2.94%）、排尿困难/尿潴留/尿路感染（5例，2.45%）、切口感染（3例，1.47%）、静脉血栓（1例，0.49%）。单因素分析显示：合并基础疾病（χ²=6.677，P=0.010）、既往腹部手术史（χ²=5.260，P=0.022）、术前白蛋白＜40 g/L（χ²=9.495，P=0.002）、术前CEA增高（χ²=4.976，P=0.026）、肿瘤下缘距肛缘距离≤7 cm（χ²=6.683，P=0.010）、术中出血量＞100 mL（χ²=10.694，P=0.001）是直肠癌术后出现临床有意义并发症的相关危险因素。多因素分析结果提示：合并基础疾病（OR=2.770，95% CI：1.175~6.531，P=0.020）、既往腹部手术史（OR=2.538，95% CI：1.023~6.295，P=0.044）、肿瘤下缘距肛缘距离≤7 cm（OR=2.376，95% CI：1.077~5.239，P=0.032）、术中出血量＞100 mL（OR=3.154，95% CI：1.339~7.427，P=0.009）、术前白蛋白＜40 g/L（OR=3.403，95% CI：1.514~7.653，P=0.003）是直肠癌根治术后出现临床有意义并发症的独立危险因素。由此构建直肠癌术后有临床意义并发症列线图临床预测模型，合并基础疾病78分、既往腹部手术史72分、肿瘤下缘距肛缘距离≤7 cm为74分、术前白蛋白＜40 g/L为100分、术中出血量＞100 mL为80分，各因素积分总和即为总分，总分对应的概率即为该模型预测直肠癌根治术后有临床意义并发症发生率。区分度测试结果显示该模型AUC值为0.775（95% CI：0.694~0.855），重复抽样内部验证校正后C-index指数为0.747，校正曲线显示该列线图模型的预测结果与实际结果之间具备较好的一致性。

合并基础疾病、既往腹部手术史、肿瘤下缘距肛缘距离≤7 cm、术中出血量＞100 mL、术前白蛋白＜40 g/L是直肠癌根治术后出现临床有意义并发症的独立危险因素。充分的术前评估、营养支持和术中精细操作、减少出血量是降低直肠癌患者术后出现临床有意义并发症的有效措施。本研究构建的列线图预测模型对直肠癌根治术后出现临床有意义并发症的概率有较高的预测价值。

To investigate the risk factors of postoperative short-term complications in patients with rectal cancer and to establish a nomogram model to predict the postoperative short-term complications.

A retrospective study was conducted to collect and analyze the clinicopathological data and related surgical information of patients who underwent radical resection of rectal cancer in the Second Affiliated Hospital of Nanjing Medical University from January 2015 to September 2018. A total of 204 patients were collected, including 135 male patients and 69 female patients. Pathological TNM staging: 11 cases of stage 0, 42 cases of stage Ⅰ, 62 cases of stage Ⅱ and 89 cases of stage Ⅲ. The postoperative complications of grade Ⅱ and above according to the modified Clavien-Dindo classification of surgical complications (Version 2019) were defined as clinically significant complications. The relationship between the general baseline characteristics, pathological characteristics, operation information characteristics and the occurrence of clinically significant complications was analyzed and the clinical variables were analyzed by univariate analysis and multivariate analysis. The nomogram prediction model was established by R software (R4.0.3) and the nomogram model was verified and evaluated by ROC curve and C-index.

Of the 204 patients, 39 (19.11%) had clinically significant complications. The common postoperative complications were anastomotic leakage (14 cases, 9.03%; 155 cases of Dixon and coloanal anastomosis), intestinal obstruction (14 cases, 6.86%), pulmonary infection (13 cases, 6.37%), stoma related complications (5 cases, 5.95%; 84 cases of Miles operation, Hartmann operation or preventive stoma),postoperative bleeding (9 cases, 4.41%), abdominal infection/pelvic infection (6 cases, 2.94%), dysuria/urinary retention/urinary tract infection (5 cases, 2.45%), incision infection (3 cases, 1.47%), venous thrombosis (1 case, 0.49%). Univariate analysis showed that basic diseases (χ²=6.677, P=0.010), previous abdominal surgery history (χ²=5.260, P=0.022), preoperative albumin＜40 g/L (χ²=9.495, P=0.002), preoperative CEA increased (χ²=4.976, P=0.026), distance between the lower edge of the tumor and the anal edge ≤7 cm (χ²=6.683, P=0.010), intraoperative bleeding＞100 mL (χ²=10.694, P=0.001) were risk factors of clinically significant complications after radical resection of rectal cancer. Multivariate analysis showed that basic diseases (OR=2.770, 95% CI: 1.175~6.531, P=0.020), previous abdominal surgery history (OR=2.538, 95% CI: 1.023~6.295, P=0.044), distance between the lower edge of the tumor and the anal edge ≤7 cm (OR=2.376, 95% CI: 1.077~5.239, P=0.032), intraoperative blood loss＞100 mL (OR=3.154, 95% CI: 1.339~7.427, P=0.009), preoperative albumin＜40 g/L (OR=3.403, 95% CI: 1.514~7.653, P=0.003) were independent risk factors for clinically significant complications after radical resection of rectal cancer and the nomogram clinical prediction model was established. Basic diseases were for 78 points, previous abdominal surgery history was for 72 points, distance between the lower edge of the tumor and the anal edge ≤7 cm was for 74 points, preoperative albumin＜40 g/L was for 100 points and intraoperative blood loss＞100 mL was for 80 points. Adding all the points was the total score and the probability corresponding to the total score is that the model predicts the incidence of clinically significant complications after radical resection of rectal cancer. The discrimination test showed that the AUC value was 0.775 (95% CI: 0.694~0.855), and the c-index index was 0.747 after repeated sampling internal validation. The calibration curve showed a good consistency between the prediction results and the actual results.

Basic diseases, previous abdominal surgery history, distance between the lower edge of the tumor and the anal edge ≤7 cm, intraoperative blood loss＞100 mL, preoperative albumin＜40 g/L are the independent risk factors for clinical significant complications after radical resection of rectal cancer. Adequate preoperative evaluation, nutritional support, accurate surgical manipulation and reduction of blood loss are effective measures to reduce the incidence of clinically significant complications in patients with rectal cancer. The nomogram prediction model constructed in this study has high predictive value for the probability of clinically significant complications after radical resection of rectal cancer.