基于LASSO回归的慢性乙型肝炎肝纤维化无创诊断模型的构建及验证
DOI: 10.3969/j.issn.1001-5256.2022.08.014
Establishment and validation of a noninvasive diagnostic model for chronic hepatitis B liver fibrosis based on LASSO regression
-
摘要:
目的 利用血清学指标建立基于LASSO回归的慢性乙型肝炎(CHB)肝纤维化的无创诊断模型,并评估该模型对CHB肝纤维化的诊断价值。 方法 纳入2019年9月—2021年9月南京医科大学附属常州第二人民医院诊断为CHB的240例患者为研究对象,根据肝穿刺活检和病理结果将其分为显著纤维化组175例(F2~4期)和无显著纤维化组65例(F0~1期)。比较两组患者性别、年龄、血生化指标和二维剪切波弹性成像测量肝脏硬度值(LSM),根据LASSO回归和多因素logistic回归分析筛选肝纤维化的危险因素,建立列线图模型并绘制受试者工作特征曲线(ROC曲线)、Calibration曲线和Decision曲线进行验证。符合正态分布的计量资料多组间比较采用单因素方差分析,两两比较采用LSD-t检验;不符合正态分布的计量资料多组间比较采用Kruskal-Wallis H检验;计数资料多组间比较用χ2检验。 结果 F3、F4期与F2、F0~1期患者年龄、ALT、AST、ALP、GGT、TBil、PLT、Ⅲ型前胶原、Ⅳ型胶原、透明质酸和LSM比较,差异均有统计学意义(P值均<0.05)。采用LASSO回归筛选出5个重要变量,logistic回归分析显示,透明质酸(OR=1.432)、Ⅳ型胶原(OR=1.243)、Ⅲ型前胶原(OR=1.146)和LSM(OR=1.656)是肝纤维化的独立危险因素,而PLT(OR=0.567)是保护因素(P值均<0.05)。F3和F4期患者列线图模型评分、LSM、APRI指数、King评分、Forns指数和FIB-4指数显著高于F2和F0~1期患者,差异均有统计学意义(P值均<0.05)。ROC曲线分析列线图模型的预测价值,其AUC为0.876,显著高于LSM、APRI、King评分、Forns指数和FIB-4,差异均有统计学意义(P值均<0.05)。Calibration曲线和Decision曲线显示列线图模型的一致性和获益性尚可。 结论 利用血清学指标透明质酸、Ⅳ型胶原、Ⅲ型前胶原、PLT和LSM,基于LASSO回归建立无创列线图模型作为临床诊断CHB肝纤维化的量化工具,具有较高的诊断效能,值得推广应用。 Abstract:Objective To establish a noninvasive diagnostic model for chronic hepatitis B (CHB) liver fibrosis based on LASSO regression using serological parameters, and to investigate the value of this model in the diagnosis of CHB liver fibrosis. Methods A total of 240 patients who were diagnosed with CHB in Changzhou Second People's Hospital, Nanjing Medical University, from September 2019 to September 2021 were enrolled as subjects, and according to the results of liver biopsy and pathology, they were divided into significant liver fibrosis (stage F2-F4) group with 175 patients and non-significant liver fibrosis (stage F0-F1) group with 65 patients. The two groups were compared in terms of sex, age, blood biochemical parameters, and liver stiffness measurement (LSM) measured by two-dimensional shear wave elastography, and LASSO regression and the multivariate logistic regression analysis were used screen out the risk factors for liver fibrosis. A nomogram model was established and then verified by the receiver operating characteristic (ROC) curve, calibration curve, and decision curve. A one-way analysis of variance was used for comparison of normally distributed continuous data between multiple groups, and the least significant difference t-test was used for further comparison between two groups; the Kruskal-Wallis H test was used for comparison of non-normally distributed continuous data between groups; the chi-square test was used for comparison of categorical data between groups. Results There were significant differences between the patients with stage F3/F4 liver fibrosis and those with stage F2/F0-F1 liver fibrosis in age, alanine aminotransferase, aspartate aminotransferase, alkaline phosphatase, gamma-glutamyl transpeptidase, total bilirubin, platelet count, procollagen type Ⅲ, type Ⅳ collagen, hyaluronic acid, and LSM (all P < 0.05). Five important variables were screened out by LASSO regression, and the logistic regression analysis showed that hyaluronic acid (odds ratio [OR]=1.432, P < 0.05), type Ⅳ collagen (OR=1.243, P < 0.05), procollagen type Ⅲ(OR=1.146, P < 0.05), and LSM (OR=1.656, P < 0.05) were the independent risk factors for liver fibrosis, while platelet count (OR=0.567, P < 0.05) was a protective factor. Compared with the patients with stage F2/F0-F1 liver fibrosis, the patients with stage F3/F4 liver fibrosis had significantly higher score of the nomogram model, LSM, aspartate aminotransferase-to-platelet ratio index (APRI), King score, Forns index, and fibrosis-4 (FIB-4) index (all P < 0.05). The ROC curve was used to analyze the predictive value of the nomogram model, and the results showed an area under the ROC curve of 0.876, which was significantly higher than that of LSM, APRI, King score, Forns index, and FIB-4 (all P < 0.05). Calibration curve and decision curve showed that the nomogram model had acceptable consistency and benefit. Conclusion The noninvasive nomogram model based on LASSO regression is established by using serum parameters including hyaluronic acid, type Ⅳ collagen, procollagen type Ⅲ, platelet count, and LSM, and as a quantitative tool for the clinical diagnosis of CHB liver fibrosis, it has a high diagnostic efficiency and thus holds promise for clinical application. -
Key words:
- Liver Cirrhosis /
- Hepatitis B, Chronic /
- Diagnosis /
- Models, Statistical
-
表 1 CHB肝纤维化不同分期患者的一般资料比较
Table 1. Comparison of baseline data in patients with chronic hepatitis B-related liver fibrosis at different stages
项目 F0~1期(n=65) F2期(n=70) F3期(n=65) F4期(n=40) 统计值 P值 男/女(例) 35/30 37/33 36/29 22/18 χ2=0.031 0.860 年龄(岁) 56.3±7.8 56.6±7.7 59.6±9.2 61.2±9.6 F=4.956 0.013 抗病毒治疗[例(%)] 19(29.2) 23(32.9) 20(30.8) 15(37.5) χ2=0.852 0.837 HBV DNA(×103/mL) 3.5±1.6 3.7±1.8 3.6±1.51)2) 3.9±2.01)2) F=1.124 0.169 ALT(U/L) 32.6(20.1~65.2) 34.2(21.2~68.9) 45.3(26.3~75.3)1)2) 46.6(27.3~79.8)1)2) H=15.238 0.001 AST(U/L) 30.0(18.5~62.3) 32.3(19.6~64.5) 42.6(24.4~72.9)1)2) 44.5(26.3~75.8)1)2) H=17.501 <0.001 GGT(U/L) 52.3(30.2~70.2) 53.6(32.3~72.3) 65.9(42.3~75.9)1)2) 67.8(44.2~76.9)1)2) H=11.384 0.007 Alb(g/L) 42.3(32.6~49.9) 42.0(32.1~50.1) 41.0(31.2~49.2) 41.3(31.3~52.6) H=3.214 0.402 Glo(g/L) 28.9(22.3~35.6) 28.5(21.6~34.8) 27.4(21.0~34.9) 27.5(21.2~35.0) H=1.987 0.741 ALP(U/L) 77.5(51.6~92.5) 79.2(53.3~93.6) 92.3(72.3~112.4)1)2) 94.5(73.6~115.5)1)2) H=9.573 0.017 TBil(μmol/L) 10.2(8.2~15.5) 10.3(8.3~15.6) 15.6(11.1~18.9)1)2) 16.2(11.3~20.1)1)2) H=8.465 0.028 DBil(μmol/L) 6.2(4.6~8.9) 6.3(4.5~9.0) 6.7(4.8~9.7) 6.6(4.5~10.2) H=2.993 0.452 PLT(×109/L) 175.9(132.0~301.2) 172.2(130.2~298.6) 105.6(92.3~164.5)1)2) 100.2(90.1~156.8)1)2) H=18.247 <0.001 总胆固醇(mmol/L) 5.0(4.0~5.7) 5.1(4.1~5.5) 5.3(4.2~5.9) 5.4(4.0~5.8) H=2.342 0.620 低密度脂蛋白(mmol/L) 3.0(2.3~4.0) 3.1(2.3~4.2) 3.2(2.5~3.9) 3.1(2.3~3.8) H=3.175 0.409 Ⅲ型前胶原(ng/mL) 5.8(4.2~8.3) 6.0(4.3~8.5) 9.2(6.0~12.3)1)2) 9.4(6.3~15.2)1)2) H=16.519 <0.001 Ⅳ型胶原(ng/mL) 42.5(28.9~62.3) 44.2(29.3~63.5) 59.8(43.6~70.5)1)2) 62.4(44.5~73.6)1)2) H=22.356 <0.001 透明质酸(ng/mL) 100.5(89.6~133.4) 102.3(86.5~136.5) 123.5(105.1~156.4)1)2) 130.2(112.3~165.9)1)2) H=19.574 <0.001 LSM(kPa) 10.0(8.2~15.6) 10.3(8.3~16.0) 14.5(11.0~18.9)1)2) 15.2(11.3~20.1)1)2) H=14.268 0.002 注:与F0~1期比较,1)P<0.05;与F2期比较,2)P<0.05。 表 2 CHB肝纤维化的多因素logistic回归分析
Table 2. Multivariate logistic regression analysis of HBV liver fibrosis
因素 β值 Wald P值 OR 95%CI 透明质酸 0.801 9.629 <0.001 1.432 1.212~1.895 Ⅳ型胶原 0.619 7.052 0.001 1.243 1.100~1.759 Ⅲ型前胶原 0.524 6.659 0.001 1.146 1.056~1.423 LSM 1.002 13.206 <0.001 1.656 1.325~2.001 PLT -0.346 5.624 0.007 0.567 0.232~0.865 表 3 肝纤维化不同分期患者评分系统的比较
Table 3. Comparisons of different grading systems in patients with different stages of liver fibrosis
指标 F0~1期(n=65) F2期(n=70) F3期(n=65) F4期(n=40) H值 P值 列线图模型评分 0.58(0.33~0.86) 0.69(0.42~0.94) 1.02(0.68~1.43)1)2) 1.56(1.12~1.89)1)2) 17.352 <0.001 LSM 10.0(8.2~15.6) 11.9(8.5~17.2) 14.0(10.0~17.8)1)2) 16.3(12.3~18.9)1)2) 9.517 0.017 APRI 0.29(0.13~0.37) 0.33(0.21~0.40) 0.59(0.32~0.72)1)2) 0.67(0.39~0.83)1)2) 11.241 0.007 King评分 7.05(5.26~8.23) 7.98(5.98~9.04) 12.23(6.59~15.54)1)2) 13.64(7.89~17.26)1)2) 15.431 <0.001 Forns指数 7.23(5.64~8.24) 8.01(5.77~9.00) 9.65(7.26~13.25)1)2) 10.52(8.01~15.59)1)2) 11.587 0.006 FIB-4指数 1.23(1.00~1.67) 1.65(1.16~2.01) 2.48(1.56~3.12)1)2) 2.79(1.79~3.55)1)2) 9.368 0.018 注:与F0~1期比较,1)P<0.05;与F2期比较,2)P<0.05。 表 4 列线图模型与各评分系统预测CHB肝纤维化的ROC曲线分析
Table 4. ROC curve of predicting liver fibrosis in HBV patients by nomogram model and different grading systems
指标 AUC 95%CI P值 敏感度(%) 特异度(%) 列线图模型 0.876 0.806~0.923 <0.001 82.3 86.9 LSM 0.800 0.823~0.856 0.002 72.0 76.3 APRI 0.745 0.701~0.824 0.005 66.5 70.4 King评分 0.698 0.612~0.752 0.008 61.3 64.9 Forns指数 0.644 0.602~0.723 0.010 60.2 62.2 FIB-4 0.601 0.564~0.710 0.015 63.9 60.7 -
[1] ALTAMIRANO-BARRERA A, BARRANCO-FRAGOSO B, MÉNDEZ-SÁNCHEZ N. Management strategies for liver fibrosis[J]. Ann Hepatol, 2017, 16(1): 48-56. DOI: 10.5604/16652681.1226814. [2] KHAN S, SAXENA R. Regression of hepatic fibrosis and evolution of cirrhosis: a concise review[J]. Adv Anat Pathol, 2021, 28(6): 408-414. DOI: 10.1097/PAP.0000000000000312. [3] LEMMER A, VANWAGNER LB, GANGER D. Assessment of advanced liver fibrosis and the risk for hepatic decompensation in patients with congestive hepatopathy[J]. Hepatology, 2018, 68(4): 1633-1641. DOI: 10.1002/hep.30048. [4] AGBIM U, ASRANI SK. Non-invasive assessment of liver fibrosis and prognosis: an update on serum and elastography markers[J]. Expert Rev Gastroenterol Hepatol, 2019, 13(4): 361-374. DOI: 10.1080/17474124.2019.1579641. [5] LOOMBA R, ADAMS LA. Advances in non-invasive assessment of hepatic fibrosis[J]. Gut, 2020, 69(7): 1343-1352. DOI: 10.1136/gutjnl-2018-317593. [6] CHENG CH, CHU CY, CHEN HL, et al. Subgroup analysis of the predictive ability of aspartate aminotransferase to platelet ratio index (APRI) and fibrosis-4 (FIB-4) for assessing hepatic fibrosis among patients with chronic hepatitis C[J]. J Microbiol Immunol Infect, 2020, 53(4): 542-549. DOI: 10.1016/j.jmii.2019.09.002. [7] CUI AL, WANG JB, XU LL, et al. Evaluation of two-dimensional shear wave elastography and transient elastography in the diagnosis of hepatic fibrosis in patients with chronic hepatitis B[J]. Clin Ultras Med, 2018, 20(12): 819-822. DOI: 10.3969/j.issn.1008-6978.2018.12.008.崔艾琳, 王佳冰, 徐莉力, 等. 二维剪切波弹性成像与瞬时弹性成像对慢性乙型肝炎患者肝纤维化诊断效能的探讨[J]. 临床超声医学杂志, 2018, 20(12): 819-822. DOI: 10.3969/j.issn.1008-6978.2018.12.008. [8] JIANG W, HUANG S, TENG H, et al. Diagnostic accuracy of point shear wave elastography and transient elastography for staging hepatic fibrosis in patients with non-alcoholic fatty liver disease: a meta-analysis[J]. BMJ Open, 2018, 8(8): e021787. DOI: 10.1136/bmjopen-2018-021787. [9] Chinese Society of Hepatology, Chinese Medical Association; Chinese Society of Gastroenterology, Chinese Medical Association; Chinese Society of Infectious Diseases, Chinese Medical Association. Consensus on diagnosis and treatment of liver fibrosis (2019)[J] J Clin Hepatol, 2019, 35(10): 2163-2172. DOI: 10.3969/j.issn.1001-5256.2019.10.007.中华医学会肝病学分会, 中华医学会消化病学分会, 中华医学会感染病学分会. 肝纤维化诊断及治疗共识(2019年)[J]. 临床肝胆病杂志, 2019, 35(10): 2163-2172. DOI: 10.3969/j.issn.1001-5256.2019.10.007. [10] ABDEL-HAMEED EA, ROUSTER SD, KOTTILIL S, et al. The enhanced liver fibrosis index predicts hepatic fibrosis superior to FIB4 and APRI in HⅣ/HCV infected patients[J]. Clin Infect Dis, 2021, 73(3): 450-459. DOI: 10.1093/cid/ciaa646. [11] MEDHIOUB M, BEN SALAH W, KHSIBA A, et al. Performance of FIB4 and APRI scores for the prediction of fibrosis in patients with chronic hepatitis B virus infection[J]. Tunis Med, 2020, 98(12): 998-1004. [12] CHANG ML, YANG SS. Metabolic signature of hepatic fibrosis: from individual pathways to systems biology[J]. Cells, 2019, 8(11). DOI: 10.3390/cells8111423. [13] IQBAL U, DENNIS BB, LI AA, et al. Use of anti-platelet agents in the prevention of hepatic fibrosis in patients at risk for chronic liver disease: a systematic review and meta-analysis[J]. Hepatol Int, 2019, 13(1): 84-90. DOI: 10.1007/s12072-018-9918-2. [14] CABALLERÍA L, TORÁN P, CABALLERÍA J. Markers of hepatic fibrosis[J]. Med Clin (Barc), 2018, 150(8): 310-316. DOI: 10.1016/j.medcli.2017.08.009. [15] UEDA J, YOSHIDA H, MAMADA Y, et al. Evaluation of the impact of preoperative values of hyaluronic acid and type iv collagen on the outcome of patients with hepatocellular carcinoma after hepatectomy[J]. J Nippon Med Sch, 2018, 85(4): 221-227. DOI: 10.1272/jnms.JNMS.2018_85-34. [16] SULYOK M, FERENCI T, MAKARA M, et al. Hepatic fibrosis and factors associated with liver stiffness in HⅣ mono-infected individuals[J]. Peer J, 2017, 5: e2867. DOI: 10.7717/peerj.2867. [17] ZHANG R, CHEN J, JIANG Y, et al. Prognostic nomogram for hepatocellular carcinoma with fibrosis of varying degrees: a retrospective cohort study[J]. Ann Transl Med, 2020, 8(21): 1429. DOI: 10.21037/atm-20-3267. [18] XU DX, ZHANG F. Expression levels of coagulation indicators in patients with different stages of hepatitis B virus infection[J]. Chin J Med Offic, 2020, 48(11): 1349-1350, 1353. DOI: 10.16680/j.1671-3826.2020.11.32.许德翔, 张飞. 不同阶段乙型肝炎病毒感染患者凝血功能指标表达水平分析[J]. 临床军医杂志, 2020, 48(11): 1349-1350, 1353. DOI: 10.16680/j.1671-3826.2020.11.32. [19] CAMPOS-MURGUÍA A, RUIZ-MARGÁIN A, GONZÁLEZ-REGUEIRO JA, et al. Clinical assessment and management of liver fibrosis in non-alcoholic fatty liver disease[J]. World J Gastroenterol, 2020, 26(39): 5919-5943. DOI: 10.3748/wjg.v26.i39.5919. [20] BERARDI G, MORISE Z, SPOSITO C, et al. Development of a nomogram to predict outcome after liver resection for hepatocellular carcinoma in Child-Pugh B cirrhosis[J]. J Hepatol, 2020, 72(1): 75-84. DOI: 10.1016/j.jhep.2019.08.032. [21] XIONG J, LIN DN, WENG WZ, et al. Assessment of hepatitis B virus associated liver fibrosis by FIB-4 combined with serum PCⅢ, ⅣC[J]. Int J Virol, 2020, 27(5): 403-406. DOI: 10.3760/cma.j.issn.1673-4092.2020.05.012.熊静, 林登娜, 翁伟镇, 等. FIB-4联合血清PCⅢ、ⅣC评估乙型肝炎病毒相关肝纤维化[J]. 国际病毒学杂志, 2020, 27(5): 403-406. DOI: 10.3760/cma.j.issn.1673-4092.2020.05.012. [22] XU W, CHENG Y, TU B. Construction and validation of a nomogram for predicting the risk of portal vein thrombosis after splenectomy in patients with hepatitis B cirrhosis[J]. J South Med Univ, 2020, 40(9): 1265-1272. DOI: 10.12122/j.issn.1673-4254.2020.09.07.徐伟, 程瑶, 涂兵. 乙型肝炎肝硬化患者行脾切除术后门静脉血栓形成的列线图预测模型的建立与验证[J]. 南方医科大学学报, 2020, 40(9): 1265-1272. DOI: 10.12122/j.issn.1673-4254.2020.09.07.