建立预测直肠癌腹会阴联合切除术后会阴切口并发症危险因素的列线图模型

Establishment of a nomogram predicting risk factors of postoperative perineal wound complications after abdominoperineal resection for rectal cancer

目的探究直肠癌腹会阴联合切除术(APR)后会阴切口并发症发生的危险因素,并建立预测会阴切口并发症发生的列线图模型。方法采用病例对照研究方法,回顾性收集2010年1月至2016年12月于南京医科大学附属第一医院行APR手术、临床资料完整的213例结直肠癌患者病历资料。根据2019版改良Clavien-Dindo手术并发症分级标准,对APR术后会阴切口并发症进行分级,并将Ⅱ级及以上并发症定义为"临床有意义的并发症"。分析性别、年龄、手术方式、手术入路、会阴修复方式、引流管放置位置、引流管引出皮肤位置、手术时间、术中出血量、术前放化疗、术中局部灌注化疗、肿瘤分型、病理分级、肿瘤T分期、肿瘤TNM分期等22个与会阴切口并发症可能有关的因素,采用χ^2检验对所有变量进行并发症发生风险的单因素分析,将单因素分析中P<0.05的变量进一步纳入到多因素logistic回归分析,筛选出独立危险因素,引入R软件(R 3.3.2),运用rms软件包,构建列线图预测模型。计算C-index指数(范围为0.5~1.0,越接近1.0说明模型预测结果与实际一致性越好)来评价模型的区分度,采用Bootstrap法重复抽样进行内部验证,并选取2017年1—12月期间、在南京医科大学附属第一医院行APR手术的42例结直肠癌患者进行模型的外部验证,分别计算校正后的C-index指数。通过比较预测风险和实际风险的C-index指数校准差值,确定模型符合度。结果213例结直肠癌患者中男性131例,女性82例,年龄(59.6±11.6)岁。术后会阴切口并发症的发生率为20.2%(43/213),Clavien-DindoⅡ级及以上并发症27例。单因素分析显示,美国东部肿瘤协作组(ECOG)评分、术前白蛋白、引流管引出皮肤位置、术中出血量、术前放化疗与发生术后会阴切口发生并发症有关(均P<0.05)。多因素分析结果显示,术前白蛋白水平≤38 g/L(OR=105.261,95%CI:7.781~1 423.998,P<0.001)、引流管从会阴引出皮肤(OR=11.493,95%CI:1.379~95.767,P=0.024)、术中出血量>110 ml(OR=6.476,95%CI:1.505~27.863,P=0.012)和术前放化疗(OR=7.479,95%CI:1.887~29.640,P=0.004)是术后发生有临床意义的会阴切口并发症的独立危险因素。由此构建列线图模型,术前白蛋白水平<38 g/L为100分,术前接受过放化疗为52.5分,术中出血量>110 ml为28.5分,引流管从腹壁引出为17.5分。将所有积分相加即为总分,总分所对应并发症发生率即为模型预测的发生率。该模型C-index指数为0.863,经过内部验证后,C-index指数下降了0.005。外部验证显示C-index指数为0.841。结论术前营养水平、引流管引出皮肤位置、术中出血量和术前放化疗情况会影响直肠癌APR术后会阴伤口并发症的发生,本研究构建的列线图模型有助于预测术后有临床意义并发症的发生概率。

Objective To investigate the risk factors of perineal incision complications after abdominoperineal resection (APR) for rectal cancer, and to establish a nomogram model to predict the complications of perineal incision. Methods A case-control study was conducted to retrospectively collect the medical records of 213 patients with colorectal cancer who underwent APR at the First Affiliated Hospital of Nanjing Medical University from January 2010 to December 2016. The complications of perineal incision after APR were classified according to the modified Clavien-Dindo classification of surgical complications (Version 2019), and the complications of grade II and above were defined as "clinically significant complications".Twenty-two factors related to complication of perineal incision, such as gender, age, surgical procedure, surgical approach, perineal repair, placement of drainage tube, skin position of drainage tube, operation time, intraoperative blood loss, preoperative radiotherapy and chemotherapy, intraoperative local perfusion chemotherapy, tumor classification, pathological grade, tumor T stage, tumor TNM stage and so on, were analyzed by chi-square test for univariate risk factor of complication in all variables, and variables with P<0.2 in univariate analysis were further included in multivariate analysis. Logistic regression analysis was used to screen out independent risk factors. R software (R 3.3.2) was introduced. The rms software package was used to construct a nomogram prediction model. The C-index was calculated (higher meaning better consistency with actual risk) to evaluate the discriminant degree of the model. The Bootstrap method was used to repeat the sampling for internal verification. A total of 42 patients with colorectal cancer who underwent APR from January 2017 to December 2017 at the First Affiliated Hospital of Nanjing Medical University were externally validated, and the corrected C-index was calculated. The model conformity was determined by comparing the C-index calibration difference between the predicted and actual risks. Results Of the 213 patients with colorectal cancer, 131 were male and 82 were female, with mean age of (59.6±11.6) years. The incidence of postoperative perineal incision complications was 20.2%(43/213), including 27 cases of Clavien-Dindo II and above complications. Univariate analysis showed that the Eastern Cancer Cooperative Group (ECOG) score, preoperative albumin, skin position of drainage tube, intraoperative blood loss, preoperative radiotherapy and chemotherapy were associated with complications of postoperative perineal incision (All P<0.05). Multivariate analysis showed that preoperative albumin levels ≤38 g/L (OR=105.261, 95% CI: 7.781 to 1423.998, P<0.001), perinead drainage (OR=11.493, 95% CI: 1.379 to 95.767, P=0.024), intraoperative blood loss >110 ml (OR=6.476, 95% CI: 1.505 to 27.863, P=0.012) and preoperative radiotherapy and chemotherapy (OR=7.479, 95% CI: 1.887 to 29.640, P=0.004) were postoperative clinically significant independent risk factors for perineal incision complications. The nomogram model was established. Preoperative albumin level <38 g/L was for 100 points, the preoperative chemoradiotherapy was for 52.5 points, the intraoperative blood loss >110 ml was for 28.5 points, and the perineal drainage was for 17.5 points. Adding all the points was the total score, and the complication rate corresponding to the total score was the predicted rate of the model. The model had a C-index of 0.863. After internal verification, the C-index dropped by 0.005. External verification showed a C-index of 0.841. Conclusions Preoperative nutritional status, skin position of drainage tube, intraoperative blood loss and preoperative radiotherapy and chemotherapy may affect the occurrence of perineal wound complications after APR for rectal cancer. The nomogram model constructed in this study is helpful for predicting the probability of clinically significant complications after APR.