基于经肛多点全层穿刺活检构建局部进展期直肠癌新辅助治疗后病理完全缓解预测模型

Construction of a model based on multipoint full-layer puncture biopsy for predicting pathological complete response after neoadjuvant therapy for locally advanced rectal cancer

目的探讨经肛多点全层穿刺活检(TMFP)病理结果在预测局部进展期直肠癌(LARC)患者新辅助放化疗(NCRT)后病理完全缓解(pCR)的价值,建立预测模型为LARC的治疗提供临床指导。方法本研究为多中心前瞻性队列研究,收集首都医科大学附属北京朝阳医院(50例)、首都医科大学附属北京友谊医院(41例)、山东大学齐鲁医院(16例)和武汉大学中南医院(3例)4家医院2020年4月至2023年3月间的110例LARC患者资料,患者均在完成标准NCRT治疗后接受TMFP。本研究的观察指标包括:(1)患者临床病理特征;(2)临床完全缓解(cCR)和TMFP对LARC患者NCRT后pCR的判断效力;(3)纳入不同医院、性别、年龄、临床T分期和N分期、肿瘤下缘距肛缘距离、基线血清癌胚抗原(CEA)和糖类抗原(CA)19-9水平、化疗方案、免疫抑制剂的使用与否、放疗剂量、放疗后等待手术时间、手术方式、放化疗后临床T分期和N分期、放化疗后CEA和CA19-9水平、临床完全缓解(cCR)、TMFP病理结果、穿刺方式(体内或体外)、穿刺针数、穿刺时间等因素,通过单因素分析及多因素logistic回归分析筛选出LARC患者NCRT后pCR的影响因素;(4)基于多因素分析结果构建预测模型,利用受试者工作特征(ROC)曲线、校准曲线及临床决策(DCA)曲线分析评估所建立模型的性能。pCR定义为直肠癌NCRT后手术切除标本(包括清扫淋巴结)中显微镜下肿瘤细胞完全消失,即ypT0+N0。cCR参照中国直肠癌新辅助治疗后等待观察数据库研究协作组(CWWD)标准,需满足内镜检查无溃疡和结节;直肠指诊阴性;直肠核磁T2及DWI序列无肿瘤信号、血清CEA水平正常以及盆腔CT/MRI未见复发征象等条件。结果110例患者中,45例(40.9%)达到pCR,NCRT治疗联合免疫检查点抑制剂34例(30.9%)。穿刺前评估为cCR病例为38例(34.5%),其中体内穿刺43例(39.1%),穿刺后无肠瘘、阴道损伤、前列腺损伤和骶前出血等并发症,仅1例(2.3%)少量便血,经肛门填塞压迫止血后症状缓解。cCR判断pCR的灵敏度为57.8%(26/45),特异度为81.5%(53/65),准确性为71.8%(79/110),阳性预测值为68.4%(26/38),阴性预测值73.6%(53/72)。而TMFP病理判断pCR的灵敏度为100.0%(45/45),特异度为66.2%(43/65),准确性为80.0%(88/110),阳性预测值为67.2%(45/67),阴性预测值为100.0%(43/43)。本研究TMFP对pCR的灵敏度(100.0%比57.8%,χ^(2)=24.09,P<0.001)高于cCR,差异有统计学意义;但两者对pCR的准确性比较,差异无统计学意义(80.0%比71.8%,χ^(2)=2.01,P=0.156)。单因素和多因素logistic回归分析结果显示:肿瘤下缘距肛缘距离≥4 cm(OR=7.84,95%CI:1.48~41.45,P=0.015)、非cCR(OR=4.81,95%CI:1.39~16.69,P=0.013)、TMFP病理结果(OR=114.29,95%CI:11.07~1180.28,P<0.001)是LARC患者NCRT后pCR预测的危险因素,而体内穿刺(OR=0.02,95%CI:0.05~0.77,P=0.020)是LARC患者NCRT后pCR预测的保护因素。基于前述4个因素所建立预测模型的ROC曲线下面积(AUC)为0.934(95%CI:0.892~0.977),提示模型具有较好区分度。校准曲线相对接近理想45°参考线,表明模型的预测值与实际值吻合良好。DCA曲线显示该模型具有较好的临床净获益。结论纳入TMFP病理结果的预测模型在预测局部进展期直肠癌NCRT后pCR方面有较好的准确率,可能为临床个体化精准治疗提供决策依据。

Objective To investigate the value of transanal multipoint full-layer puncture biopsy(TMFP)in predicting pathological complete response(pCR)after neoadjuvant radiotherapy and chemotherapy(nCRT)in patients with locally advanced rectal cancer(LARC)and to establish a predictive model for providing clinical guidance regarding the treatment of LARC.Methods In this multicenter,prospective,cohort study,we collected data on 11o LARC patients from four hospitals between April 2020 and March 2023:Beijing Chaoyang Hospital of Capital Medical University(50 patients),Beijing Friendship Hospital of Capital Medical University(41 patients),Qilu Hospital of Shandong University(16 patients),and Zhongnan Hospital of Wuhan University(three patients).The patients had all received TMFP after completing standard nCRT.The variables studied included(1)clinicopathological characteristics;(2)clinical complete remission(cCR)and efficacy of TMFP in determining pCR after NCRT in LARC patients;and(3)hospital attended,sex,age,clinical T-and N-stages,distance between the lower margin of the tumor and the anal verge,baseline and post-radiotherapy serum carcinoembryonic antigen(CEA)and carbohydrate antigen(CA)19-9 concentrations,chemotherapy regimen,use of immunosuppressants with or without radiotherapy,radiation therapy dosage,interval between surgery and radiotherapy,surgical procedure,clinical T/N stage after radiotherapy,cCR,pathological results of TMFP,puncture method(endoscopic or percutaneous),and number and timing of punctures.Single-factor and multifactorial logistic regression analysis were used to determine the factors affecting pCR after NCRT in LARC patients.A prediction model was constructed based on the results of multivariat analysis and the performance of this model evaluated by analyzing subject work characteristics(RoC),calibration,and clinical decision-making(DCA)curves.pCR was defined as complete absence of tumor cells on microscopic examination of the surgical specimens of rectal cancer(including lymph node dissection)after NCRT,that is,ypTO+NO.cCR was defined according to the Chinese Neoadjuvant Rectal Cancer Waiting Watch Database Study Collaborative Group criteria after treatment,which specify an absence of ulceration and nodules on endoscopy;negative rectal palpation;no tumor signals on rectal MRI T2 and DWI sequences;normal serum CEA concentrations,and no evidence of recurrence on pelvic computed tomography/magnetic resonance imaging.Results Of the 110 patients,45(40.9%)achieved pCR after nCRT,which was combined with immune checkpoint inhibitors in 34(30.9%).cCR was diagnosed before puncture in 38(34.5%)patients,43(39.1%)of the punctures being endoscopic.There were no complications of puncture such as enterocutaneous fistulae,vaginal injury,prostatic injury,or presacral bleeding.Only one(2.3%)patient had a small amount of blood in the stools,which was relieved by anal pressure.cCR had a sensitivity of 57.8%(26/45)for determining pCR,specificity of 81.5%(53/65),accuracy of 71.8%(79/110),positive predictive value 68.4%(26/38),and negative predictive value of 73.6%(53/72).In contrast,the sensitivity of TMFP pathology in determining pCR was 100%(45/45),specificity 66.2%(43/65),accuracy 80.0%(88/110),positive predictive value 67.2%(45/67),and negative predictive value 100.0%(43/43).In this study,the sensitivity of TMFP for pCR(100.0%vs.57.8%,χ^(2)=24.09,P<0.001)was significantly higher than that for cCR.However,the accuracy of pCR did not differ significantly(80.0%vs.71.8%,x=2.01,P=0.156).Univariate and multivariate logistic regression analyses showed that a≥4 cm distance between the lower edge of the tumor and the anal verge(OR=7.84,95%CI:1.48-41.45,P=0.015),non-cCR(OR=4.81,95%CI:1.39-16.69,P=0.013),and pathological diagnosis by TMFP(OR=114.29,the 95%CI:11.07-1180.28,P<0.001)were risk factors for pCR after NCRT in LARC patients.Additionally,endoscopic puncture(OR=0.02,95%CI:0.05-0.77,P=0.020)was a protective factor for pCR after NCRT in LARC patients.The area under the ROC curve of the established prediction model was 0.934(95%CI:0.892-0.977),suggesting that the model has good discrimination.The calibration curve was relatively close to the ideal 45°reference line,indicating that the predicted values of the model were in good agreement with the actual values.A decision-making curve showed that the model had a good net clinical benefit.Conclusion Our predictive model,which incorporates TMFP,has considerable accuracy in predicting pCR after nCRT in patients with locally advanced rectal cancer.This may provide a basis for more precisely selecting individualized therapy.