We read with interest the comments on our recent article1 made by Dr. Dogru and colleagues and by Dr. Wang and colleagues.
Regarding the issues raised by Dr. Dogru, a 75-g oral glucose tolerance test (OGTT) was not performed in our nondiabetic patients, so we are unable to evaluate the proportion of subjects with impaired glucose tolerance (2-hour plasma glucose of 140-199 mg/dL during OGTT). However, this test, even if more sensitive and modestly more specific than fasting plasma glucose (FPG) for diagnosing diabetes or a prediabetic status, is poorly reproducible and rarely performed in routine practice.2 As a result, FPG, because of its ease of use, acceptability to patients, and lower cost, is the preferred diagnostic test, and the use of OGTT is restricted to the evaluation of patients with impaired fasting glucose (IFG; FPG of 100-125 mg/dL) or those for whom diabetes is still suspected despite a normal FPG.2 According to FPG, we found that 17% of our patients (18/104) had IFG, another expression of a prediabetic status. As suggested by Dr. Dogru and colleagues, we have compared patients with and without varices according to glucose abnormalities. We have observed, as reported in our article,1 that the prevalence of diabetes was higher in patients with esophageal varices (EV; 32%, 20/63) than in subjects without EV (17%, 7/41), even if this was not statistically significant (P = 0.10). Similarly, patients with EV more frequently had glucose abnormalities (IFG and diabetes; 49%, 31/63) than their counterparts without EV (31%, 14/41), although this was not significant (P = 0.12). In contrast, in the 77 nondiabetics, the prevalence of IFG was comparable in subjects with EV (25%, 11/43) and in those without EV (21%, 7/34; P = 0.60). All evidence therefore suggests a possible additional effect of diabetes as a risk factor for EV, underlining the central role of homeostasis model assessment/insulin resistance (HOMA-IR) as a predictor of EV. In particular, we confirmed that HOMA-IR is an independent predictor of EV not only in nondiabetics, as reported in our article, but also in the 59 patients without IFG or frank diabetes. In this subgroup, the platelet count/spleen diameter ratio (odds ratio, 0.998; 95% confidence interval, 0.996-0.999; P = 0.009) and HOMA-IR (odds ratio, 1.463; 95% confidence interval, 1.128-2.195; P = 0.04) remained the only variables independently associated with the presence of EV by multivariate analysis. Finally, in the setting of 45 patients with glucose abnormalities (27 with diabetes and 18 with IFG), we found that HOMA values were higher in those with EV compared to their counterparts without EV (5.46 ± 3.71 versus 4.30 ± 3.21), even if this difference was not statistically significant (P = 0.36), probably because of the small sample size.
We excluded from our sample cohort patients with type 2 diabetes on insulin therapy because in this setting HOMA-IR is not a reliable marker of IR. We included instead diabetics on oral hypoglycemic agents (metformin, thiazolidenediones, alpha-glucosidase inhibitors, and repaglinides). Independently of the class of agents used, we are confident in fact that HOMA-IR is able to reflect a condition of IR. However, considering that these treatments could easily influence measures of insulin sensitivity, we suggest considering a diabetes diagnosis per se, independently of HOMA-IR values, as a risk factor for EV presence.
We stress again that in patients with hepatitis C virus cirrhosis, HOMA represents a simple, easy-to-get, and low-cost test useful for assessing the likelihood of the presence of EV, regardless of the glucose tolerance status. In patients being treated with oral antidiabetics, HOMA could be less accurate in evaluating IR. In this setting of patients, the diagnosis of glucose abnormalities per se could define a profile at higher risk of the presence of EV.
In response to Dr. Wang and colleagues, we evaluated, as reported in the article,1 not only the discriminating ability in predicting EV of the platelet count/spleen diameter ratio and HOMA separately but also the whole ability of the model. We made a receiver operating characteristic curve of the whole model, showing the high ability of the rule to discriminate between patients with or without EV (area under the curve, 0.800). We have clearly stressed that the main limitation of our study lies in its cross-sectional nature and its inability to dissect the temporal relation between IR and the presence of EV. Therefore, we agree with Dr. Wang and colleagues that further studies are needed to evaluate in cirrhosis of different stages and etiologies the usefulness of IR and changes in IR over time in predicting not only the presence of EV but also their development and the changes in hepatic venous pressure gradient levels. Recently, Li and colleagues3 validated, in a large training set of 349 Chinese patients with cirrhosis of mixed etiology, the platelet count/spleen diameter ratio and HOMA as independent predictors of the presence of EV in Child A, B, and C class cirrhosis and then confirmed these results in another independent test set.