肝硬化食管胃底静脉曲张破裂出血患者再出血预测模型的建立

张群; 时克; 王宪波

doi:10.3969/j.issn.1001-5256.2022.11.011

肝硬化食管胃底静脉曲张破裂出血患者再出血预测模型的建立

DOI: 10.3969/j.issn.1001-5256.2022.11.011

首都医科大学附属北京地坛医院中西医结合中心，北京 100015

基金项目:

国家自然科学基金 (81774234);

北京市科学技术委员会资助 (Z191100006619033)

伦理学声明：本研究方案于2018年4月28日经由首都医科大学附属北京地坛医院伦理委员会审批，批号：京地伦科字[2018]第(017)-01号。

利益冲突声明：本研究不存在研究者、伦理委员会成员、受试者监护人以及与公开研究成果有关的利益冲突。

作者贡献声明：张群、时克参与数据收集，分析数据；张群负责论文撰写及修改；王宪波负责课题设计，拟定写作思路，指导撰写文章并最后定稿。

详细信息

通信作者:
王宪波，wangxianbo638@163.com

计量
- 文章访问数: 456
- HTML全文浏览量: 94
- PDF下载量: 87
- 被引次数: 0
出版历程
- 收稿日期: 2022-03-07
- 录用日期: 2022-04-19
- 出版日期: 2022-11-20

Developing the prediction model of esophagogastric variceal rebleeding in patients with liver cirrhosis based on artificial neural network

Center of Integrative Medicine, Beijing Ditan Hospital, Capital Medical University, Beijing 100015, China

Research funding:

National Natural Science Foundation of China (81774234);

Beijing Municipal Science & Technology Commission (Z191100006619033)

More Information

Corresponding author: WANG Xianbo, wangxianbo638@163.com(ORCID: 0000-0002-3593-5741)

摘要

摘要: 目的构建可以用来预测肝硬化食管胃底静脉曲张破裂出血患者1年内再出血风险的预测模型。方法选取2008年8月—2017年10月于首都医科大学附属北京地坛医院住院的肝硬化食管胃底静脉曲张破裂出血患者441例，并对其进行为期1年的随访观察。以再出血作为研究终点，根据1年内是否发生再出血事件将患者分为再出血组(249例)和未再出血组(192例)。计数资料采用Fisher精确检验或者χ²检验。计量资料符合正态分布两组间比较采用t检验，不符合正态分布的两组间比较采用Mann-Whitney U检验。采用Cox单因素和多因素回归分析影响食管胃底静脉曲张破裂出血患者1年内再出血的独立影响因素，并构建预测模型，利用ROC曲线下面积评估该模型的预测性能。结果有249例(56.5%)患者在1年内发生食管胃底静脉曲张破裂再出血事件，Cox多因素分析显示：INR(AHR=1.566，95%CI：1.023~2.398，P=0.039)、NLR(AHR=1.033，95%CI：1.009~1.058，P=0.006)是1年内再出血的独立危险因素；而CHB(AHR=0.769，95%CI：0.597~0.991，P=0.042)、Na(AHR=0.967，95%CI：0.936~0.999，P=0.044)、内镜(AHR=0.829，95%CI：0.743~0.926，P=0.001)及手术治疗(AHR=0.246，95%CI：0.120~0.504，P＜0.001)是保护因素。利用上述6个独立影响因素，成功构建了一个人工神经网络模型(https://wangxianbo.math.ink/PoRiEV-zq/index.html)，其预测1年内再出血的ROC曲线下面积为0.782(95%CI：0.740~0.825)，明显优于Cox回归模型的0.672(95%CI：0.622~0.722，P＜0.001)、Child-Pugh评分的0.557(95%CI：0.504~0.610，P＜0.001)和MELD评分的0.562(95%CI：0.509~0.616，P＜0.001)。结论人工神经网络模型具有良好的个体化预测性能，可以作为临床食管胃底静脉曲张破裂再出血的风险评估工具。
- 食管和胃静脉曲张 /
- 出血 /
- 神经网络(计算机)
Abstract: Objective To develop an artificial neural network model to predict the risk of rebleeding within one year in cirrhotic patients with esophagogastric variceal bleeding. Methods We retrospectively collected 441 cirrhotic patients with esophagogastric variceal bleeding hospitalized at Beijing Ditan Hospital, Capital Medical University, from August 2008 to October 2017. The enrolled patients were followed up for one year. According to the primary endpoint which was rebleeding within one year, patients were divided into rebleeding (249 cases) and non-rebleeding (192 cases) groups. Fisher exact test or chi-square test were used for comparison of categorical data. Comparison of continuous data with normal distribution between groups was performed using t test, while comparison of non-normally distributed data was performed by Mann-Whitney U test. Cox univariate and multivariate regression were used to identify independent factors affecting rebleeding within one year in cirrhotic patients with esophagogastric variceal bleeding, and then an artificial neural network prediction model was constructed using identified factors. The predictive performance of the model was evaluated using the area under the receiver operating characteristic curve (AUC). Results In total, 249 (56.5%) patients developed esophagogastric variceal rebleeding within one year. Cox multivariate regression showed INR (AHR=1.566, 95%CI: 1.023~2.398, P=0.039) and NLR (AHR=1.033, 95%CI: 1.009~1.058, P=0.006) were risk factors for 1-year rebleeding, while CHB (AHR=0.769, 95%CI: 0.597~0.991, P=0.042), Na (AHR=0.967, 95%CI: 0.936~0.999, P=0.044), endoscopic (AHR=0.829, 95%CI: 0.743~0.926, P=0.001) and surgical treatment (AHR=0.246, 95%CI: 0.120~0.504, P < 0.001) were protective factors. Using the above six independent influence factors, we successfully constructed an artificial neural network model (https://lixuan.me/annmodel/myg-v3/). The model's ability to predict 1-year rebleeding had an AUC of 0.782 (95%CI: 0.740-0.825), which was higher than 0.672 (95%CI: 0.622-0.722, P < 0.001) of Cox regression model, 0.557 (95%CI: 0.504-0.610, P < 0.001) of Child-Pugh and 0.562 (95%CI: 0.509-0.616, P < 0.001) of MELD scores. Conclusion The artificial neural network model has good individualized prediction performance and can be used as a risk assessment tool for esophagogastric variceal rebleeding.
- Esophageal and Gastric Varices /
- Hemorrhage /
- Neural Networks(Computer)

HTML全文

图 1 自变量对ANN模型构建的重要性排序

Figure 1. Independent variables rank the importance to ANN model construction

下载: 全尺寸图片幻灯片

图 2 ANN模型

Figure 2. ANN model

下载: 全尺寸图片幻灯片

图 3 ANN模型、回归模型、Child-Pugh评分和MELD评分的ROC曲线

Figure 3. ROC curve of ANN model, regression model, Child-Pugh scores and MELD scores

下载: 全尺寸图片幻灯片

表 1 肝硬化EVB患者基线特征

Table 1. Baseline characteristics of cirrhotic patients with EVB

指标	所有患者 (n=441)	未再出血组 (n=192)	再出血组 (n=249)	统计值	P值
年龄(岁)	52.46±10.51	52.43±10.81	52.48±10.28	t=0.446	0.505
男/女(例)	309/132	136/56	173/76	χ²=0.095	0.758
病因[例(%)]
CHB	251(56.9)	124(64.6)	127(51.0)	χ²=8.152	0.004
CHC	44(10.0)	15(7.8)	29(11.6)	χ²=1.774	0.183
ALD	122(27.7)	41(21.4)	81(32.5)	χ²=6.766	0.009
AIH	26(5.9)	14(7.3)	12(4.8)	χ²=1.194	0.274
细菌感染[例(%)]	245(55.6)	100(52.1)	145(58.2)	χ²=1.660	0.198
Child-Pugh分级[例(%)]				χ²=3.948	0.047
A/B(5~9)	377(85.5)	171(89.1)	206(82.7)
C(10~13)	60(13.6)	19(9.9)	41(16.5)
MELD评分	10.00(9.00~13.00)	10.00(9.00~12.00)	11.00(9.00~14.00)	Z=-2.238	0.025
ALT(U/L)	24.40(17.20~39.40)	24.35(17.23~39.93)	24.40(17.20~39.30)	Z=-0.273	0.785
AST(U/L)	30.20(21.65~49.55)	28.75(21.20~48.08)	31.50(22.15~49.75)	Z=-0.676	0.499
TBil(μmol/L)	19.50(12.75~31.25)	18.90(12.55~29.75)	20.30(12.85~33.05)	Z=-1.162	0.245
GGT(U/L)	8.50(5.50~13.85)	23.10(13.60~46.10)	29.10(15.65~69.70)	Z=-2.529	0.011
Alb(g/L)	29.97±5.78	30.17±5.76	29.82±5.80	t=0.006	0.937
Na(mmol/L)	139.0(136.7~141.2)	139.4(137.0~141.3)	138.5(136.2~141.2)	Z=-2.056	0.040
Cr(μmol/L)	65.30(52.65~77.50)	64.25(50.93~76.78)	66.00(54.00~78.55)	Z=-1.234	0.217
GLU(mmol/L)	8.43(6.72~11.27)	8.00(6.53~10.65)	8.83(6.91~11.63)	Z=-2.241	0.217
WBC(×10⁹/L)	4.36(2.92~6.39)	4.19(2.88~6.29)	4.50(2.92~6.50)	Z=-1.001	0.317
NLR	4.15(2.71~6.78)	3.85(2.62~6.09)	4.50(2.80~7.42)	Z=-2.402	0.016
Hb(g/L)	76.20(59.20~96.00)	76.20(58.55~97.75)	76.20(60.45~95.00)	Z=-0.159	0.874
PLT(×10⁹/L)	60.90(43.40~79.00)	59.50(43.25~79.98)	61.00(44.05~79.00)	Z=-0.055	0.956
PT(s)	15.30(14.05~17.20)	15.00(13.93~16.80)	15.40(14.15~17.45)	Z=-1.721	0.085
INR	1.29(1.18~1.45)	1.27(1.17~1.42)	1.31(1.19~1.49)	Z=-2.478	0.013
内镜治疗次数	1.0(1.0~2.0)	2.0(1.0~2.0)	1.0(1.0~3.0)	Z=-0.836	0.403
手术治疗[例(%)]	37(8.4)	29(15.1)	8(3.2)	χ²=19.944	＜0.001
TIPS[例(%)]	6(1.4)	4(2.1)	2(0.8)	χ²=1.324	0.411
β受体阻滞剂[例(%)]	24(5.4)	14(7.3)	10(4.0)	χ²=0.040	0.841
注：GLU，葡萄糖；NLR，中性粒细胞与淋巴细胞比值；PT，凝血酶原时间；INR，国际标准化比值；TIPS，经颈静脉肝内门体分流术。

下载: 导出CSV

表 2 肝硬化EVB患者再出血的Cox单因素和多因素分析

Table 2. Univariate and multivariate Cox analysis with esophagogastric variceal rebleeding

变量	单因素分析		多因素分析
变量	HR(95%CI)	P值	AHR(95%CI)	P值
年龄(岁)	1.001(0.989~1.013)	0.891
男/女	1.094(0.835~1.433)	0.514
CHB	0.686(0.535~0.880)	0.003	0.769(0.597~0.991)	0.042
CHC	1.330(0.903~1.960)	0.149
ALD	1.356(1.039~1.768)	0.025
AIH	0.732(0.410~1.308)	0.293
细菌感染	1.248(0.970~1.606)	0.084
ALT(U/L)	1.001(1.000~1.002)	0.040
AST(U/L)	1.000(1.000~1.001)	0.105
TBil(μmol/L)	1.004(1.001~1.006)	0.002
GGT(U/L)	1.001(1.000~1.003)	0.076
Alb(g/L)	0.989(0.968~1.011)	0.313
Na(mmol/L)	0.950(0.919~0.982)	0.003	0.967(0.936~0.999)	0.044
Cr(μmol/L)	1.004(1.000~1.007)	0.030
GLU(mmol/L)	1.023(0.999~1.046)	0.056
WBC(×10⁹/L)	1.035(1.012~1.057)	0.002
NLR	1.029(1.009~1.049)	0.004	1.033(1.009~1.058)	0.006
Hb(g/L)	1.000(0.996~1.005)	0.862
PLT(×10⁹/L)	1.001(0.998~1.004)	0.542
PT(s)	1.050(1.012~1.089)	0.010
INR	1.947(1.273~2.977)	0.002	1.566(1.023~2.398)	0.039
内镜治疗次数	0.863(0.775~0.961)	0.007	0.829(0.743~0.926)	0.001
手术治疗	0.261(0.129~0.527)	＜0.001	0.246(0.120~0.504)	＜0.001
TIPS	0.422(0.105~1.698)	0.225
β受体阻滞剂	0.998(0.582~1.711)	0.994

下载: 导出CSV

参考文献(19)

[1]	HOLSTER IL, TJWA ET, MOELKER A, et al. Covered transjugular intrahepatic portosystemic shunt versus endoscopic therapy+β-blocker for prevention of variceal rebleeding[J]. Hepatology, 2016, 63(2): 581-589. DOI: 10.1002/hep.28318.
[2]	Chinese Society of Hepatology, Chinese Medical Association; Chinese Society of Gastroenterology, Chinese Medical Association; Chinese Society of Endoscopy, Chinese Medical Association. Guidelines for the diagnosis and treatment of esophageal and gastric variceal bleeding in cirrhotic portal hypertension[J]. J Clin Hepatol, 2016, 32(2): 203-219. DOI: 10.3969/j.issn.1001-5256.2016.02.002. 中华医学会肝病学分会, 中华医学会消化病学分会, 中华医学会内镜学分会. 肝硬化门静脉高压食管胃静脉曲张出血的防治指南[J]. 临床肝胆病杂志, 2016, 32(2): 203-219. DOI: 10.3969/j.issn.1001-5256.2016.02.002.
[3]	VERDA D, PARODI S, FERRARI E, et al. Analyzing gene expression data for pediatric and adult cancer diagnosis using logic learning machine and standard supervised methods[J]. BMC Bioinformatics, 2019, 20(Suppl 9): 390. DOI: 10.1186/s12859-019-2953-8.
[4]	CHI S, TIAN Y, WANG F, et al. A novel lifelong machine learning-based method to eliminate calibration drift in clinical prediction models[J]. Artif Intell Med, 2022, 125: 102256. DOI: 10.1016/j.artmed.2022.102256.
[5]	GARCIA-TSAO G, SANYAL AJ, GRACE ND, et al. Prevention and management of gastroesophageal varices and variceal hemorrhage in cirrhosis[J]. Hepatology, 2007, 46(3): 922-938. DOI: 10.1002/hep.21907.
[6]	PUGH RN, MURRAY-LYON IM, DAWSON JL, et al. Transection of the oesophagus for bleeding oesophageal varices[J]. Br J Surg, 1973, 60(8): 646-649. DOI: 10.1002/bjs.1800600817.
[7]	FONG TV, HUNG FC, CHIU KW, et al. Model for end-stage liver disease (MELD) score for predicting late esophageal varices rebleeding in cirrhotic patients[J]. Hepatogastroenterology, 2008, 55(84): 1055-1058.
[8]	DING RH, HU YM, LI XG, et al. Current status of prevention and treatment of esophagogastric variceal bleeding in cirrhotic portal hypertension patients in Ningxia region: a multicenter study[J]. Chin J Dig Surg, 2021, 20(10): 1078-1084. DOI: 10.3760/cma.j.cn115610-20210928-00464. 丁荣华, 胡燕梅, 李小果, 等. 宁夏地区肝硬化门静脉高压食管胃底静脉曲张出血防治现状的多中心研究[J]. 中华消化外科杂志, 2021, 20(10): 1078-1084. DOI: 10.3760/cma.j.cn115610-20210928-00464.
[9]	DENG SM, ZHANG JX, QI Y, et al. Study on the application of endoscopic ligation in esophageal varices and the risk factors of postoperative rebleeding[J]. Chin J Med Offic, 2021, 49(6): 718-720. DOI: 10.16680/j.1671-3826.2021.06.43. 邓水苗, 张嘉星, 齐晔, 等. 内镜下套扎术在食管静脉曲张中应用及其术后再出血高危因素研究[J]. 临床军医杂志, 2021, 49(6): 718-720. DOI: 10.16680/j.1671-3826.2021.06.43.
[10]	HAO Y, TAN P, ZHAO YG, et al. Antibiotic prophylaxis in gastroesophageal variceal bleeding: A prospective study[J]. J Pract Hepatol, 2012, 15(1): 29-31. DOI: 10.3969/j.issn.1672-5069.2012.01.010. 郝勇, 谭萍, 赵亚刚, 等. 肝硬化食管胃底静脉曲张出血患者预防性抗菌治疗的前瞻性研究[J]. 实用肝脏病杂志, 2012, 15(1): 29-31. DOI: 10.3969/j.issn.1672-5069.2012.01.010.
[11]	YANG H, LIU YX, LI P, et al. Analysis of risk factors for early rebleeding from esophageal and gastric varices in patients with liver cirrhosis[J]. J Clin Hepatol, 2014, 30(6): 540-542. DOI: 10.3969/j.issn.1001-5256.2014.06.18. 杨花, 刘云霞, 李鹏, 等. 肝硬化患者食管胃底静脉曲张破裂早期再出血的危险因素分析[J]. 临床肝胆病杂志, 2014, 30(6): 540-542. DOI: 10.3969/j.issn.1001-5256.2014.06.18.
[12]	LIU X, HE L, HAN J, et al. Association of neutrophil-lymphocyte ratio and T lymphocytes with the pathogenesis and progression of HBV-associated primary liver cancer[J]. PLoS One, 2017, 12(2): e0170605. DOI: 10.1371/journal.pone.0170605.
[13]	PLEVRIS JN, DHARIWAL A, ELTON RA, et al. The platelet count as a predictor of variceal hemorrhage in primary biliary cirrhosis[J]. Am J Gastroenterol, 1995, 90(6): 959-961.
[14]	CHEN LH, HU NZ, WANG YL. The value of MESO (MELD/Na index) scoring system in predicting prognosis of patients with cirrhosis[J]. J Clin Hepatol, 2011, 27(10): 1044-1046, 1054. DOI: 10.3969/j.issn.1001-5256.2011.10.008. 陈丽红, 胡乃中, 王亚雷. 终末期肝病模型与血清钠比值在肝硬化患者预后判断中的价值[J]. 临床肝胆病杂志, 2011, 27(10): 1044-1046, 1054. DOI: 10.3969/j.issn.1001-5256.2011. 10.008.
[15]	HONG JB, LYU NH, WANG AJ, et al. Analysis of the risk factors for early rebleeding in oesophageal and gastric varices[J]. Chin J Dig, 2010, 30(11): 836-837. DOI: 10.3760/cma.j.issn.0254-1432.2010.11.013. 洪军波, 吕农华, 汪安江, 等. 食管胃静脉曲张早期再出血的危险因素分析[J]. 中华消化杂志, 2010, 30(11): 836-837. DOI: 10.3760/cma.j.issn.0254-1432.2010.11.013.
[16]	HUO TI, LIN HC, WU JC, et al. Proposal of a modified Child-Turcotte-Pugh scoring system and comparison with the model for end-stage liver disease for outcome prediction in patients with cirrhosis[J]. Liver Transpl, 2006, 12(1): 65-71. DOI: 10.1002/lt.20560.
[17]	KAMATH PS, WIESNER RH, MALINCHOC M, et al. A model to predict survival in patients with end-stage liver disease[J]. Hepatology, 2001, 33(2): 464-470. DOI: 10.1053/jhep.2001.22172.
[18]	HEUMAN DM, ABOU-ASSI SG, HABIB A, et al. Persistent ascites and low serum sodium identify patients with cirrhosis and low MELD scores who are at high risk for early death[J]. Hepatology, 2004, 40(4): 802-810. DOI: 10.1002/hep.20405.
[19]	ANGELI P, WONG F, WATSON H, et al. Hyponatremia in cirrhosis: Results of a patient population survey[J]. Hepatology, 2006, 44(6): 1535-1542. DOI: 10.1002/hep.21412.