背景:术中转换为开放手术是微创远端胰腺切除术(MIDP)期间的不良事件,与术后不良结局相关。这项研究的目的是开发一种能够预测接受MIDP的患者转化的模型。
方法:共有352名接受MIPD的患者被纳入本回顾性分析,并随机分配到训练和验证队列。通过文献综述确定了与开放式转换相关的潜在风险因素,并相应地收集了我们队列中这些因素的数据.在训练组中,进行多因素logistic回归分析,调整混杂因素的影响,以确定模型构建的独立危险因素.使用接收器工作特性曲线对构建的模型进行了评估,决策曲线分析(DCA),和校准曲线。
结果:经过广泛的文献综述,总共确定了十种术前危险因素,包括性,BMI,白蛋白,吸烟者,病变的大小,靠近主要血管的肿瘤,胰腺切除类型,手术方法,MIDP经验,还有恶性肿瘤的嫌疑.多变量分析表明,性别,靠近主要血管的肿瘤,怀疑是恶性肿瘤,胰腺切除术的类型(胰腺次全切除术或左胰腺切除术),和MIDP经验仍然是MIDP期间转换为开放手术的重要预测因素。与现有模型相比,构建的模型提供了更高的判别能力(曲线下面积,培训队列:0.921vs.0.757,P<0.001;验证队列:0.834vs.0.716,P=0.018)。DCA和校准曲线揭示了列线图的临床实用性以及预测值和观察值之间的良好一致性。
结论:本研究中开发的基于证据的预测模型在预测MIDP转化方面优于以前的模型。该模型可以促进围绕手术方法选择的决策过程,并促进患者对MIDP转化风险的咨询。
BACKGROUND: Intraoperative conversion to open surgery is an adverse event during minimally invasive distal pancreatectomy (MIDP), associated with poor postoperative outcomes. The aim of this study was to develop a model capable of predicting conversion in patients undergoing MIDP.
METHODS: A total of 352 patients who underwent MIPD were included in this retrospective analysis and randomly assigned to training and validation cohorts. Potential risk factors related to open conversion were identified through a literature review, and data on these factors in our cohort was collected accordingly. In the training cohort, multivariate logistic regression analysis was performed to adjust the impact of confounding factors to identify independent risk factors for model building. The constructed model was evaluated using the receiver operating characteristics curve, decision curve analysis (DCA), and calibration curves.
RESULTS: Following an extensive literature review, a total of ten preoperative risk factors were identified, including sex, BMI, albumin, smoker, size of lesion, tumor close to major vessels, type of pancreatic resection, surgical approach, MIDP experience, and suspicion of malignancy. Multivariate analysis revealed that sex, tumor close to major vessels, suspicion of malignancy, type of pancreatic resection (subtotal pancreatectomy or left pancreatectomy), and MIDP experience persisted as significant predictors for conversion to open surgery during MIDP. The constructed model offered superior discrimination ability compared to the existing model (area under the curve, training cohort: 0.921 vs. 0.757, P < 0.001; validation cohort: 0.834 vs. 0.716, P = 0.018). The DCA and the calibration curves revealed the clinical usefulness of the nomogram and a good consistency between the predicted and observed values.
CONCLUSIONS: The evidence-based prediction model developed in this study outperformed the previous model in predicting conversions of MIDP. This model could contribute to decision-making processes surrounding the selection of surgical approaches and facilitate patient counseling on the conversion risk of MIDP.