基于肿瘤负荷评分、血清学指标对肝癌微血管侵犯术前预测模型的建立.pdf

基金项目:国家重点研发计划(2022YFC2407304)第一作者简介:朱明强,硕士,研究方向:肝癌及肝移植的临床与基础研究。E-mail:ZMQ通讯作者:丁佑铭,博士,主任医师,教授,研究方向:肝癌及肝移植的临床与基础研究。E-mail:doi:10.3969/j.issn.1006-5709.2023.11.010基于肿瘤负荷评分、血清学指标对肝癌微血管侵犯术前预测模型的建立朱明强,杨大帅,裴俊鹏,熊祥云,苏洋,丁佑铭武汉大学人民医院肝胆外科,湖北 武汉 430060【摘要】目的探讨肿瘤负荷评分(tumor burden score,TBS)联合中性粒细胞与淋巴细胞比值(neutrophil-to-lymphocyte ratio,NLR)、白蛋白与碱性磷酸酶比值(albumin-to-alkaline phosphatase ratio,AAPR)等血清学指标对肝癌微血管侵犯(microvascular inva-sion,MVI)的术前预测模型的建立。方法回顾性收集 2020 年 1 月至 2022 年 8 月于武汉大学人民医院行肝切除术的 200 例原发性肝癌患者的临床资料,根据术后病理检测是否存在 MVI,将患者分为 MVI 组(n=74)和非 MVI 组(n=126)。采用 Logistic 回归分析筛选肝癌患者发生 MVI 的影响因素,并建立肝癌患者术前是否发生 MVI 的列线图风险预测模型。将所有患者按 7 3 随机分为训练集(n=140)和验证集(n=60),采用 Bootstrap 法对模型进行内部验证,应用模型校准曲线和 ROC 曲线来评价列线图模型的校准度和区分度。结果200 例患者中 74 例(37.0%)发生 MVI。Logistic 多因素分析显示,TBS、AAPR、术前 NLR、AFP、CONUT 评分是肝癌患者发生 MVI 的独立危险因素(P 均0.05)。基于上述 5 个独立危险因素构建的列线图,训练集的 C 指数为 0.794,ROC 曲线下面积为 0.791;验证集的 C 指数为 0.756,ROC 曲线下面积为 0.762(P 均0.05)。列线图模型校准曲线显示预测值与实际观测值基本一致,表明列线图模型预测的准确度较好。ROC 曲线显示此列线图模型用于预测术前 MVI 有较好的区分度。结论基于上述 5 个独立危险因素建立的个体化列线图风险预测模型对肝癌 MVI 的术前预测效能良好。【关键词】肝癌;微血管侵犯;肿瘤负荷评分;血清学指标;危险因素;列线图中图分类号:R735.5文献标识码:A文章编号:1006-5709(2023)11-1249-05收稿日期:2022-11-01Establishment of preoperative prediction model for microvascular invasion of liver cancer based on tumor burden score and serological indicatorsZHU Mingqiang,YANG Dashuai,PEI Junpeng,XIONG Xiangyun,SU Yang,DING YoumingDepartment of Hepatobiliary Surgery,Renmin Hospital of Wuhan University,Wuhan 430060,China【Abstract】Objective To investigate the establishment of a preoperative prediction model for microvascular invasion(MVI)of liver cancer by tumor burden score(TBS)combined with serological indicators such as neutrophil-to-lymphocyte ratio(NLR)and albumin-to-alkaline phosphatase ratio(AAPR).Methods The clinical data of 200 patients with pri-mary liver cancer who underwent hepatectomy in Renmin Hospital of Wuhan University,from Jan.2020 to Aug.2022 were retrospectively collected,and all patients were divided into MVI group(n=74)and non-MVI group(n=126)according to the postoperative pathological detection of MVI.Logistic regression analysis were used to screen the risk fac-tors of MVI in liver cancer patients,and a nomogram risk prediction model for whether MVI occurred before surgery in liver cancer patients was established.All patients were randomly divided into training cohort(n=140)and validation cohort(n=60)according to 73,the model was internally verified by Bootstrap method,and the calibration curve of the model and the ROC curve were used to evaluate the calibration and discrimination of the nomogram model.Results MVI occurred in 74 cases(37.0%)of the 200 patients.The results of Logistic multivariate analysis showed that TBS,AAPR,preoperative NLR,AFP and CONUT scores were independent risk factors for MVI in patients with liver cancer(P0.05).Based on the above five independent risk factors,the C index of the training cohort was 0.794,the area under the ROC curve was 0.791,the C index of the verification cohort was 0.756,the area under the ROC curve was 0.762(P0.05).The calibration curve of the nomogram model showed that the predicted value was basically consistent with the actual observed value,indicating that the prediction accuracy of the nomogram model was good.The ROC curve showed that this nomogram model was well differentiated for predicting preoperative MVI.ConclusionThe individual-9421胃肠病学和肝病学杂志2023 年 11 月第 32 卷第 11 期Chin J Gastroenterol Hepatol,Nov 2023,Vol.32,No.11ized nomogram risk prediction model based on the above five independent risk factors has good efficacy in predicting MVI of liver cancer.【Key words】Liver cancer;Microvascular invasion;Tumor burden score;Serological indicators;Risk factors;Nomogram肝癌是目前全球最常见的恶性肿瘤之一1。由于其起病隐匿、进展迅速,确诊时大多已为中晚期,预后较差2。目前,HCC 首选治疗方法仍是肝切除术和肝移植术,但术后复发率高,长期生存率仍不令人满意3。微血管侵犯(microvascular invasion,MVI)主要指在显微镜下于内皮细胞衬覆的血管腔内见到的癌细胞巢团,以癌旁门静脉分支为主4。大多数早期复发的肝癌患者病理提示 MVI 阳性,MVI 是肝癌患者肝切除或肝移植术后复发的关键预后因素之一5-6。目前,MVI 仅通过术后病理诊断,因此,本研究通过分析肝癌 MVI 的风险因素并建立术前预测模型可为肝癌患者的精准评估及治疗提供进一步的临床参考7。1资料与方法1.1一般资料回顾性收集 2020 年 1 月至 2022 年 8月武汉大学人民医院肝胆外科行肝癌肝切除术患者的临床资料。纳入标准:(1)病理诊断为肝癌且临床资料完整。(2)肝功能 Child 分级为 A 级或 B 级且无远处转移。(3)首次行肝切除术患者。排除标准:(1)手术切缘阳性,病理诊断非肝癌患者;(2)术前肝癌相关治疗史(放疗、化疗、经动脉化疗栓塞和免疫抑制治等);(3)肝外转移或合并急慢性感染、自身免疫性疾病等;MVI 的诊断标准:在显微镜下于内皮细胞衬覆的血管腔内见到的癌细胞巢团,主要分布在微中央静脉、门静脉分支或包膜外小静脉。1.2临床资料收集(1)收集资料:年龄、性别、糖尿病、高血压、乙肝、肝硬化、肿瘤大小、肿瘤数目、肿瘤负荷评分(tumor burclen score,TBS)、Edmonson 分级、术前AFP、控 制 营 养 状 态(controlling nutritional status,CONUT)评 分、ALBI 评 分、ALT、AST、TBIL、ALB、PT、PLT、Scr、总胆固醇、乳酸脱氢酶、ALP、中性粒细胞与淋巴细胞比值(neutrophil-to-lymphocyte ratio,NLR)、白蛋白与碱性磷酸酶比值(albumin-to-alkaline phosphatase ratio,AAPR);谷氨酰转肽酶与 PLT 比值(-glutamyc-transpeptidase/platelet ratio,GPR)。(2)计算公式:TBS=肿瘤数目2+肿瘤最大直径2。AAPR=ALB(g/L)/ALP(U/L);GPR=谷氨酰转肽酶(U/L)/PLT(mg/L);CONUT 评分标准为:血清 ALB 分为35.0 g/L,30.035.0 g/L,25.030.0 g/L 和25.0 g/L 共 4 个等级,分别为 0 分、2 分、4 分和 6 分。血清总胆固醇分为180 mg/dl、140 180 mg/dl、100 140 mg/dl、100 mg/dl共 4 个等级,分别计 0 分、1 分、2 分和 3 分。外周血淋巴细胞计数分为 4 个水平,1.60109 L-1、(1.20 1.60)109 L-1、(0.801.20)109 L-1及0.80109 L-1,分别为 0 分、1 分、2 分和 3 分。3 项参数的得分相加即为总 CONUT 评分。1.3统计学处理采用 SPSS 26.0 统计软件进行分析。符合正态分布的连续性变量以 xs 表示,两两比较采 用 t 检 验;不 符 合 正 态 分 布 的 连 续 性 变 量 用M(P25,P75)表示,两组间比较使用 Mann-Whitney U 检验。计数资料用例数(%)表示,组间比较采用2检验或 Fisher 确切概率法。对有显著性差异的单因素进行Logistic 回归多因素分析。根据回归分析结果运用 R软件建立个体化列线图预测模型。将 200 例患者按7 3 随机分为训练集和验证集,用 Bootstrap 法(抽样1 000 次)对该模型进行内部验证,用 C 指数、ROC 曲线及校准曲线评价模型的预测效能,P0.05 为差异有统计学意义。2结果2.1肝癌患者发生 MVI 的情况共纳入 200 例肝癌肝切除 术 患 者,男 159 例,女 41 例,年 龄(59.76 10.91)岁(22 79 岁)。其 中 74 例(37.0%)发生 MVI。2.2单因素分析比较 MVI 组和非 MVI 组的临床资料,筛选肝癌患者发生 MVI 的影响因素。两者在肿瘤大小、TBS 评分、乳酸脱氢酶、ALP、AAPR、GPR、术前NLR、AFP、CONUT 评 分 方 面 差 异 有 统 计 学 意 义(P 均0.05,见表 1)。2.3多因素分析以肝切除术后病理检测是否发生MVI 作为因变量,将单因素分析中的影响因素作为自变量进行非条件二元 Logistic 回归分析。分析结果显示,TBS 评分、AAPR、术前 NLR、AFP、CONUT 评分这 5个变量是肝癌患者发生 MVI 的独立影响因素(P 均0.05,见表 2)。2.4列线图模型的建立与验证基于独立影响因素构建个体化列线图风险预测模型(见图 1),并绘制校准曲线和 ROC 曲线。训练集和验证集的校准曲线显示预测值与实际观测值基本一致,表明列线图模型预测的准确度较好(见图 2 3)。DCA 曲线表明该预测模型具有较好的临床效用(见图 4 5)。训练集和验证集的 ROC 曲线下面积分别为 0.791、0.762,表明模型的区分度较好(见图 67)。0521胃肠病学和肝病学杂志2023 年 11 月第 32 卷第 11 期Chin J Gastroenterol Hepatol,Nov 2023,Vol.32,No.11表 1肝癌患者发生 MVI 的单因素分析Tab 1Univariate analysis of MVI in patients with liver cancer变量MVI 组(n=74)无 MVI 组(n=126)2/t/Z 值P 值年龄M(P25,P75),岁58(54,67)63(53,69)-1.4310.153性别(男/女)62/1297/291.3230.250高血压(是/否)30/4441/851.3030.254糖尿病(是/否)12/6219/1070.0460.830乙肝(是/否)48/2684/420.0670.795肝硬化(是/否)36/3862/640.0060.939肿瘤数目(单发/多发)46/2893/332.9840.084肿瘤大小(5 cm/5 cm)22/5266/609.7080.002TBS 评分M(P25,P75),分6.32(4.37,9.22)4.80(3.16,6.32)-4.0020.001Child-Pugh 评分(A/B)56/18108/183.1830.074Edmonson 分级(级/级)23/5129/971.5760.209AFP(200 ng/ml)/(200 ng/ml)32/14225/10112.5290.001CONUT 评分(4 分/4 分)42/3235/9116.5350.001ALBI 评分M(P25,P75),分-2.54(-2.81,-2.21)-2.61(-2.83,-2.41)-1.4650.143ALTM(P25,P75),U/L29.0(17,42.5)25.0(17.75,39.25)-1.3070.191ASTM(P25,P75),U/L30.0(24,43.5)28.5(21.75,40.25)-1.1700.242T-BilM(P25,P75),mol/L15.2(11.65,25.2)16.3(11.7,20.43)-0.3420.733ALBM(P25,P75),g/L39.0(36.2,42.2)40.0(37.4,43.0)-1.4050.160PTM(P25,P75),s11.6(10.9,12.5)11.6(11.0,12.2)-0.4900.624血小板M(P25,P75),mg/L177.0(127,222)159.5(116.8,205.3)-1.0650.287血肌酐M(P25,P75),mol/L65.0(55,74.5)65.5(56,75)-0.1510.880总胆固醇M(P25,P75),mmol/L4.0(3.33,4.79)3.94(3.42,4.42)-0.8000.424乳酸脱氢酶M(P25,P75),U/L216.0(194,273)201.5(177,247)-2.3320.020ALPM(P25,P75),U/L90.0(77.0,152.5)82.5(65.8,107.0)-2.7840.005术前 NLRM(P25,P75)3.01(2.50,4.08)2.40(1.86,3.16)-4.7950.001AAPRM(P25,P75)0.41(0.24,0.53)0.50(0.35,0.61)-3.1810.001GPRM(P25,P75)0.37(0.21,1.15)0.27(0.15,0.62)-2.1350.033表 2肝癌患者发生 MVI 的多因素分析Tab 2Multivariate analysis of MVI in patients with liver cancer变量 值标准误差Wald/2值OR 值 95%CIP 值TBS 评分0.1290.0555.3981.1371.0201.2680.020AAPR-2.0470.9244.9120.1290.0210.7890.027术前 NLR0.2560.1204.5831.2921.0221.6340.032AFP200 ng/ml0.9950.3677.3362.7051.3165.5570.007CONUT4 分0.9210.3496.9792.5111.2684.9730.008图 1肝癌患者术前发生 MVI 风险的列线图预测模型Fig 1Nomogram prediction model of preoperative risk of MVI in patients with liver cancer1521胃肠病学和肝病学杂志2023 年 11 月第 32 卷第 11 期Chin J Gastroenterol Hepatol,Nov 2023,Vol.32,No.11图 2训练集校准曲线;图 3验证集校准曲线;图 4训练集 DCA 曲线;图 5验证集 DCA 曲线;图 6训练集 ROC 曲线;图 7验证集 ROC 曲线Fig 2The calibration plot for the training cohort;Fig 3The calibration plot for the validation cohort;Fig 4The DCA curve for the training cohort;Fig 5The DCA curve for the validation cohort;Fig 6The ROC curve for the training cohort;Fig 7The ROC curve for the validation cohort3讨论肝切除术是早期肝癌主要的根治性治疗手段。MVI 是肝癌发生转移、术后复发的重要预后指标8-9。MVI 可用于指导是否进行新辅助放化疗、扩大手术切缘和超出米兰标准的肝移植,因此,术前预测 MVI 在手术治疗中具有重要意义10。目前,有关肝癌 MVI的术前风险预测模型已有一些探索11-13。本研究结合了形态学、营养状态、炎症指标及血清学指标建立了个性化列线图风险预测模型,有利于更好地评估肝癌患者的预后。TBS 是通过肿瘤形态学(即肿瘤最大直径和数目)计算得出,在预测结直肠癌肝转移患者和肝癌患者中具有较好的效能14-15。本研究发现,TBS 评分是肝癌患者发生 MVI 的独立危险因素。肿瘤直径和数量增加是瘤灶侵袭的表现,侵袭性越强,恶性程度越高,导致 MVI发生率高。CONUT 评分可以反映患者免疫营养状况,与肝癌、胃癌等恶性肿瘤以及肝移植术后的不良预后密切相关16-18。有研究发现,CONUT4 分是胰腺癌患者预后的独立危险因素19。本研究发现 CONUT4 分是肝癌患者发生 MVI 的有效预后指标。AAPR 是 ALB 与 ALP 的比值,作为一种新型预后指标广泛应用于多种疾病模型中20-21。既往研究发现,AAPR 是可切除肝癌患者重要危险因素,能更好地预测肿瘤患者的生存率22-23。本研究亦发现,MVI 组术前 AAPR 水平降低,AAPR 是肝癌患者发生 MVI 的独立危险因素。AAPR 反映肿瘤的高负荷和高异质性,是机体炎症、营养状况更全面的反馈24。有大量研究发现 NLR 是肝癌以及肝移植患者预后的独立危险因素25-27。本研究发现,肝癌患者 MVI 组 NLR 升高,NLR 是肝癌术前 MVI 的独立危险因素。这可能与其促肿瘤炎症反应有关,导致血管内皮功能紊乱,促进癌细胞侵袭28。有研究指出,APF 增高是肝癌患者发生 MVI 的独立危险因素29-30,这也与本研究结果相一致。但目前预测 MVI 的 AFP 临界值尚未统一,需要进一步的研究来验证31-32。本研究所建立的列线图模型具有可视化、直观化以及研究指标获取方便等优点使其易于在临床推广。同时,本研究也存在一定的局限性,该预测模型属于单中心研究,缺乏多中心资料进行外部验证。其次,结合影像学检查以及整合其他预后相关生物标志物或许能进一步提升该模型的准确性。综上所述,TBS、CONUT、AAPR、NLR、AFP 是预测术前肝癌 MVI 的独立危险因素。以此为基础建立个体化列线图风险预测模型预测效能良好,可直观地预测 MVI 的发生风险,筛查高风险患者,并对临床方案的制定提供一定的指导。参考文献1 Villanueva A.Hepatocellular carcinoma J.N Engl J Med,2019,2521胃肠病学和肝病学杂志2023 年 11 月第 32 卷第 11 期Chin J Gastroenterol Hepatol,Nov 2023,Vol.32,No.11380(15):1450-1462.DOI:10.1056/NEJMra1713263.2 Yang JD,Hainaut P,Gores GJ,et al.A global view of hepatocellular carcinoma:trends,risk,prevention and management J.Nat Rev Gastroenterol Hepatol,2019,16(10):589-604.DOI:10.1038/s41575-019-0186-y.3 Llovet JM,Kelley RK,Villanueva A,et al.Hepatocellular carcinoma J.Nat Rev Dis Primers,2021,7(1):6.DOI:10.1038/s41572-020-00240-3.4 Rodrguez-Perlvarez M,Luong TV,Andreana L,et al.A systematic review of microvascular invasion in hepatocellular carcinoma:diagnostic and prognostic variability J.Ann Surg Oncol,2013,20(1):325-339.DOI:10.1245/s10434-012-2513-1.5 Roayaie S,Blume IN,Thung SN,et al.A system of classifying micro-vascular invasion to predict outcome after resection in patients with hep-atocellular carcinoma J.Gastroenterology,2009,137(3):850-855.DOI:10.1053/j.gastro.2009.06.003.6 Pommergaard HC,Rostved AA,Adam R,et al.Vascular invasion and survival after liver transplantation for hepatocellular carcinoma:a study from the European Liver Transplant Registry J.HPB(Oxford),2018,20(8):768-775.DOI:10.1016/j.hpb.2018.03.002.7 Yang D,Zhu M,Xiong X,et al.Clinical features and prognostic fac-tors in patients with microvascular infiltration of hepatocellular carcino-ma:development and validation of a nomogram and risk stratification based on the SEER database J.Front Oncol,2022,12:987603.DOI:10.3389/fonc.2022.987603.8 Xu L,Li L,Wang P,et al.Novel prognostic nomograms for hepatocel-lular carcinoma patients with microvascular invasion:experience from a single center J.Gut Liver,2019,13(6):669-682.DOI:10.5009/gnl18489.9 Chong HH,Yang L,Sheng RF,et al.Multi-scale and multi-paramet-ric radiomics of gadoxetate disodium-enhanced MRI predicts microvas-cular invasion and outcome in patients with solitary hepatocellular carci-noma 5 cm J.Eur Radiol,2021,31(7):4824-4838.DOI:10.1007/s00330-020-07601-2.10 谷宇飞,刘豪,冯亮杰,等.术前-谷氨酰转肽酶/血小板比值、AST/血小板比 值 对 肝 细 胞 癌 患 者 并 发 微 血 管 浸 润 的 预 测 价值J.中华肝胆外科杂志,2021,27(6):454-456.DOI:10.3760/113884-20200722-00389.Gu YF,Liu H,Feng LJ,et al.Value of preoperative gamma-glu-tamyl transpeptidase to platelet ratio and AST to platelet ratio for pre-diction of microvascular invasion in patients with hepatocellular carci-noma J.Chinese Journal of Hepatobiliary Surgery,2021,27(6):454-456.DOI:10.3760/cm 113884-20200722-00389.11 Yoh T,Seo S,Ogiso S,et al.Quantitative assessment of microvascu-lar invasion in hepatocellular carcinoma using preoperative serological and imaging markers J.HPB(Oxford),2021,23(7):1039-1045.DOI:10.1016/j.hpb.2020.10.024.12 Wang L,Jin YX,Ji YZ,et al.Development and validation of a predic-tion model for microvascular invasion in hepatocellular carcinoma J.World J Gastroenterol,2020,26(14):1647-1659.DOI:10.3748/wjg.v26.i14.1647.13 Mao S,Yu X,Yang Y,et al.Preoperative nomogram for microvascu-lar invasion prediction based on clinical database in hepatocellular carcinoma J.Sci Rep,2021,11(1):13999.DOI:10.1038/s41598-021-93528-7.14 Sasaki K,Morioka D,Conci S,et al.The tumor burden score:a new Metro-ticket prognostic tool for colorectal liver metastases based on tumor size and number of tumors J.Ann Surg,2018,267(1):132-141.DOI:10.1097/SLA.0000000000002064.15 Wang J,Chen Z,Wang L,et al.A new model based inflammatory index and tumor burden score(TBS)to predict the recurrence of hep-atocellular carcinoma(HCC)after liver resection J.Sci Rep,2022,12(1):8670.DOI:10.1038/s41598-022-12518-5.16 Harimoto N,Yoshizumi T,Sakata K,et al.Prognostic significance of preoperative controlling nutritional status(CONUT)score in patients undergoing hepatic resection for hepatocellular carcinoma J.World J Surg,2017,41(11):2805-2812.DOI:10.1007/s00268-017-4097-1.17 Kuroda D,Sawayama H,Kurashige J,et al.Controlling Nutritional Status(CONUT)score is a prognostic marker for gastric cancer pa-tients after curative resection J.Gastric Cancer,2018,21(2):204-212.DOI:10.1007/s10120-017-0744-3.18 Pravisani R,Mocchegiani F,Isola M,et al.Postoperative trends and prognostic values of inflammatory and nutritional biomarkers after liver transplantation for hepatocellular carcinoma J.Cancers(Basel),2021,13(3):513.DOI:10.3390/cancers13030513.19 Kato Y,Yamada S,Suenaga M,et al.Impact of the controlling nutri-tional status score on the prognosis after curative resection of pancreat-ic ductal adenocarcinoma J.Pancreas,2018,47(7):823-829.DOI:10.1097/MPA.0000000000001105.20 Zhang F,Lu SX,Hu KS,et al.Albumin-to-alkaline phosphatase ra-tio as a predictor of tumor recurrence and prognosis in patients with early-stage hepatocellular carcinoma undergoing radiofrequency abla-tion as initial therapy J.Int J Hyperthermia,2021,38(1):1-10.DOI:10.1080/02656736.2020.1850885.21 Sandfeld-Paulsen B,Aggerholm-Pedersen N,Winther-Larsen A.Pre-treatment albumin-to-alkaline phosphatase ratio is a prognostic marker in lung cancer patients:a registry-based study of 7077 lung cancer pa-tients J.Cancers(Basel),2021,13(23):6133.DOI:10.3390/cancers13236133.22 Chan AW,Chan SL,Mo FK,et al.Albumin-to-alkaline phosphatase ratio:a novel prognostic index for hepatocellular carcinoma J.Dis Markers,2015,2015:564057.DOI:10.1155/2015/564057.23 Li Q,Lyu Z,Wang L,et al.Albumin-to-alkaline phosphatase ratio associates with good prognosis of hepatitis B virus-positive HCC pa-tients J.Onco Targets Ther,2020,13:2377-2384.DOI:10.2147/OTT.S242034.24 Tian G,Li G,Guan L,et al.Pretreatment albumin-to-alkaline phos-phatase ratio as a prognostic indicator in solid cancers:A meta-ana