Establishment and verification of a nomogram model for individualized prediction of poor prognosis in patients with cirrhosis of esophagogastric variceal bleeding / 中国医师进修杂志

ABSTRACT

Objective:To establish a nomogram model for individualized prediction of poor prognosis in patients with cirrhosis of esophagogastric variceal bleeding (EGVB), and verify its efficacy, so as to provide a scientific basis for the prevention and treatment of EGVB.Methods:The clinical data of 389 patients with cirrhosis of EGVB from January 2010 to December 2018 in Hangzhou Hospital of Zhejiang Medical and Health Group were retrospectively analyzed. All patients were followed up for 3 years, including 232 cases with poor prognosis (poor prognosis group) and 157 cases with good prognosis (good prognosis group). The general clinical data and laboratory results were compared between 2 groups. Receiver operating characteristic (ROC) curve was used to analyze the optimal cut-off value of poor prognosis factors in patients with cirrhosis of EGVB; multivariate Logistic regression analysis was used to analyze independent risk factors of poor prognosis in patients with cirrhosis of EGVB. A nomogram model to predict poor prognosis in patients with cirrhosis of EGVB was established with R language software 4.0 "rms" package. Internal validation of the nomogram model was performed using correction curves, and the prediction efficiency of the nomogram model was evaluated using decision curves.Results:The age, ascites rate, liver surface roughness rate, end-stage liver disease model score (MELD score), Child-Turcotte-Pugh score (CTP score), alanine aminotransferase (ALT), aspartate transaminase (AST), international standard ratio (INR) and total bilirubin (TBIL) in poor prognosis group were significant higher than those in good prognosis group (62.48 ± 6.21) years old vs. (58.71 ± 5.93) years old, 51.29% (119/232) vs. 35.03% (55/157), 60.78% (141/232) vs. 42.03% (66/157), (13.89±1.93) scores vs. (11.32 ± 1.69) scores, (8.93 ± 0.77) scores vs. (7.46 ± 0.63) scores, (37.73 ± 5.21) U/L vs. (32.13 ± 5.03) U/L, (64.19 ± 11.31) U/L vs. (57.36 ± 10.29) U/L, 1.73 ± 0.41 vs. 1.61 ± 0.39 and (24.31 ± 2.63) μmol/L vs. (19.86 ± 2.17) μmol/L, the albumin, hemoglobin and serum sodium were significantly lower than those in good prognosis group (36.21 ± 4.51) g/L vs. (39.12 ± 4.96) g/L, (86.31 ± 8.27) g/L vs. (92.28 ± 9.67) g/L and (136.58 ± 18.24) mmol/L vs. (141.21 ± 19.26) mmol/L, and there were statistical differences ( P<0.01 or<0.05). ROC curve analysis results show that the optimal cut-off values of age, MELD score, CTP score, albumin, ALT, AST, hemoglobin, INR, TBIL and serum sodium for predicting poor prognosis in patients with cirrhosis of EGVB were 55 years old, 14.20 scores, 9.30 scores, 35 g/L, 38 U/L, 67 U/L, 88 g/L, 1.90 scores, 25 μmol/L and 135 mmol/L, respectively. Multivariate Logistic regression analysis results showed that age≥55 years old, ascites, MELD score ≥14.20 scores, CTP score ≥9.30 scores, albumin<35 g/L and INR≥1.90 were independent risk factors for poor prognosis in patients with cirrhosis of EGVB ( HR = 1.528, 1.439, 1.637, 1.795, 1.521 and 1.596; 95% CI 1.165 to 1.891, 1.088 to 1.790, 1.308 to 1.966, 1.385 to 2.205, 1.262 to 1.780 and 1.259 to 1.933; P<0.05 or<0.01). To construct a nomogram model that integrates independent risk factors for poor prognosis in patients with cirrhosis of EGVB, the predictive power of the model was good (C-index 0.839, 95% CI 0.781 to 0.948). The corrected curve of nomogram model to predict poor prognosis in patients with cirrhosis of EGVB was close to the ideal curve; when the high risk threshold>0.02, nomogram model provided a significant additional clinical net benefit to predict poor outcome in patients with cirrhosis of EGVB, which was higher than the individual risk factors. Conclusions:The nomogram model based on age, ascites, MELD score, CTP score, albumin, INR and other independent risk factors that affect the high risk of poor prognosis in patients with cirrhosis of EDVB has great clinical value in screening and identifying high risk of poor prognosis in patients with cirrhosis of EDVB.