Reply:

Abstract

In their letter, Dr. Li and colleagues add further evidence to the fact that insulin resistance (IR) may be used as a surrogate noninvasive marker to predict the presence of esophageal varices (EVs). They first validated, in a large training set of 349 Chinese patients with cirrhosis of mixed etiology and stage, homeostasis model assessment (HOMA) and platelet count/spleen diameter ratio as independent predictors of EV presence in patients with Child class A, B, and C cirrhosis, and then confirmed these results in another independent test set. Results comparable to ours1 were obtained in the subgroup of patients with hepatitis C virus (HCV)-derived Child A cirrhosis. Dr. Li and colleagues found a positive relation between HOMA and liver function, which is to be expected given the reduced liver function in patients with Child B and C cirrhosis. In our study,1 to exclude confounding factors for HOMA evaluation such as hepatic failure and “hepatogenous diabetes”,2 we included only patients with Child A cirrhosis. Li et al. remark that in their cohort the 95% confidence interval of the two predictors are somewhat different compared from those of our study. These differences could be related not only to ethnic/geographical variations, as they suggest, but also to the differences in prevalence of EVs in the study population, to the etiology of cirrhosis, and to the different metabolic and anthropometrical features of the patients. In the letter, no information is given about anthropometric parameters and prevalence of diabetes in the study population. This is of major relevance, considering that HOMA is not a reliable index of IR in diabetic patients on insulin therapy. Furthermore, they do not report the etiology of cirrhosis in their patients. In our study,1 we included only HCV-infected patients because much clinical and experimental evidence suggests that HCV by itself is able to interfere with insulin signaling.3 Furthermore, in the setting of chronic HCV infection, IR has been associated with progression of fibrosis.4 Some of the clinical expression of IR, namely obesity and diabetes, correlate with an increased risk of hepatocellular carcinoma.5 It would be interesting to know whether, in the Chinese population, IR is associated to presence of EVs not only in cirrhosis derived from HCV, but also from nonalcoholic fatty liver disease cirrhosis, where IR is the pathogenetic key of the disease,6 and in patients with hepatitis B virus (HBV), considering recent experimental data suggesting the potential ability of HBV to interfere with nuclear receptors involved in liver lipid homeostasis.7 In conclusion, it must be stressed that in any population of patients, prediction rules—although valuable in predicting the average probability of specific events in a group of patients—are much less accurate in predicting the outcome in individual patients. Although confident that HOMA and platelet count/spleen ratio are useful to predict EV, we feel that to reach a degree of precision useful for the clinician, these predictors should probably be combined with other noninvasive screening tools, such as transient fibroelastometry.8 Salvatore Petta*, Calogero Cammà* , Antonio Craxì*, * Cattedra ed Unità Operativa di Gastroenterologia, University of Palermo, Italy, Istituto di Biomedicina e di Immunologia Molecolare, Consiglio Nazionale delle Ricerche, Palermo, Italy.