The complexities of predicting outcome in cirrhotic patients with acute variceal bleeding

Abstract

We read with interest the paper “Bedside risk-scoring model for predicting 6-week mortality in cirrhotic patients undergoing endoscopic band ligation for acute variceal bleeding” published in the May issue of Journal of Gastroenterology and Hepatology. The authors used Cox regression analysis to assess the relationship between clinical, biological, and endoscopic variables and 6-week mortality risk after EBL in a derivation cohort (n = 1373). The primary outcome was the predictive accuracy of the new model for the 6-week mortality in the validation cohort. The adequacy of the mortality risk-based stratification and the discriminative performance of the new model in comparison with the Child– Turcotte–Pugh (CTP) and the model for end-stage liver disease scores in the validation cohort (n = 200) was tested. Five objective variables (use of beta-blockers, hepatocellular carcinoma, CTP class C, hypovolemic shock at initial presentation, and history of hepatic encephalopathy [HE]) were scored to generate a 12-point risk-prediction model. The model stratified the 6-week mortality risk in patients as low (3.5%), intermediate (21.1%), and high (53.4%) (P < 0.001). Time-dependent area under the receiver operating characteristic curve for 6-week mortality showed that this model was a better prognostic indicator than the CTP class alone in the derivation (P < 0.001) and validation (P < 0.001) cohorts. The authors concluded that their simplified scoring model refines the prediction of 6-week mortality in high-risk cirrhotic patients, thereby aiding the targeting and individualization of treatment strategies for decreasing the mortality rate. The authors are to be congratulated on developing and validating a new bedside risk-scoring model for AVB. We, too, have grappled with the complexities of predicting outcome in cirrhotic patients with acute variceal bleeding in view of the inadequate and flawed existing scoring systems. We investigated the relative predictive performances of the original CTP, the model for end-stage liver disease (MELD) and a new four-category recalibrated CTP (rCTP) in 225 consecutive adult patients with bleeding esophageal varices secondary to alcoholic cirrhosis. The new rCTP classification with four recalibrated grades (A–D) had a significantly improved discriminatory capacity (AUROC 0.83 95% CI 0.77–0.89) compared to the original CTP (A–C) score (AUROC 0.75, 95% CI 0.71–0.81) and MELD (AUROC 0.71, 95% CI 0.62–0.80) to predict death (P = 0.004). It is not clear how the authors differentiated between a small peripheral HCC and a large central HCC with portal vein invasion as the clinical and portal hypertensive implications of each are vastly different but are scored similarly in their model. The authors also awarded 25% of the weighting (3 points) in their new score to CTP class C and to HE, both of which are known to fluctuate due to the influence of clinical events and are susceptible to differences in interpretation. In neither the original CTP nor in the modified CTP classification were the division of the grades or scores ever statistically validated. Over the years, intrinsic deficiencies in the CTP classification have become apparent. First, both the degree of ascites and presence of HE are subjective clinical assessments. Ultrasonography, widely available and increasingly used as a bedside test has led to more sensitive detection of ascites, but no current guidelines exist for accurate scoring when clinically undetected ascites is identified by ultrasonography. Likewise, subtle manifestations of HE are ideally assessed by psychometric testing or slowing of frequency on electroencephalography, both of which are superior to clinical evaluation and should be accounted for in calculating the CTP score. Secondly, both ascites andHE are influenced by diuretic therapy, albumin infusion, and lactulose administration, and it is not evident whether ascites and HE should be scored at their best, or worst, or independent of any specific therapy. In addition, the CTP classification does not distinguish between “mild” grade C and severe grade C with sufficient clinical discrimination. Furthermore, the continuous variables in the CTP classification are categorized with arbitrary cut-off points. Thus, a patient with a bilirubin of 55 μmol/L has a better prognosis than one with a bilirubin of 250 μmol/L, but both these patients in the CTP classification have the same score of severity for bilirubin concentration (“the ceiling effect”). The same applies to serum albumin in that the CTP classification does not differentiate between patients with an albumin of 17 g/L vs 25 g/L (“the floor effect”). Reweighting the above three variables, HCC, CTP class C, and HE, may assist the authors to further enhance discrimination and accuracy in their scoring model.