Reply:Regarding terlipressin-related patient outcomes in hepatorenal syndrome-acute kidney injury

We read with interest the letter by Luo and Yu,1 and we thank them for the insightful comments on our investigation of point-of-care ultrasound in patients with hepatorenal syndrome-acute kidney injury.2 The authors commented on the predictive models used for the prediction of terlipressin nonresponse and mortality. Our hypothesis is that cirrhotic cardiomyopathy, indicated by systolic function and/or diastolic dysfunction,3 confers a higher risk for poor response to treatment with terlipressin. Therefore, the cardiac variables that reflect cirrhotic cardiomyopathy were of main interest in our study. Since the cardiac function is affected by age, gender, presence of diabetes, and possibly liver disease etiology, we created adjustments for these variables in addition to liver disease severity, ie, Model for End-Stage Liver Disease-Sodium (MELDNa) in the main manuscript. Regarding terlipressin nonresponse, there were 38 patients who did not respond to terlipressin. As suggested by the authors, we reanalyzed the predictive model for terlipressin nonresponse using only the 3 pertinent cardiac function variables that differentiated responders and nonresponders [ie, systolic function, cardiac index at 72 h (delta CI72h), and diastolic function (E/e, e' velocity)]. Subsequently, an adjusted model was created, correcting only for MELDNa. Similarly, for the prediction of mortality, there were 23 deaths. We performed an analysis with 3 variables (delta CI72h, E/e' and e' velocity) that reflect the cardiac dysfunction in our study and then adjusted for baseline MELDNa. Table 1A indicates the findings of the new analyses, as recommended by Luo and Yu. The new analyses did not change the results, and they further corroborated our initial findings predicting terlipressin nonresponse with E/e' ratio (indicative of higher filling pressure), e' velocity, (indicative of impaired relaxation), and cardiac index at 72 hours (corresponding to impaired myocardial contractility).4 TABLE 1 - Cox proportional hazards analysis for (A) Predictors of terlipressin nonresponse (B) Predictors of mortality based on cardiac variables (E/e', e' velocity) and delta cardiac index at 72 hours Univariable analysis Multivariable analysis Multivariable analysis with adjustment for MELDNa HR (HR) (95% CI) p HR (95% CI) p Adjusted HR (95% CI) p POC-echo parameters (A) Predictors of terlipressin nonresponse E/e' (septal) 1.231 (1.111–1.363) <0.001 1.191 (1.057–1.342) 0.004 1.114 (1.005–1.243) 0.029 e' (septal) 0.530 (0.382–0.735) <0.001 0.647 (0.441–0.950) 0.026 0.698 (0.489–0.986) 0.04 Delta Cardiac Index 72h 1.096 (1.002–1.128) 0.012 1.221 (1.192–2.256) 0.007 1.286 (1.184–2.297) 0.017 POC-echo parameters (B) Predictors of Mortality HR (95% CI) p HR (95% CI) p Adjusted HR (95% CI) p E/e' (septal) 1.182 (1.042–1.342) 0.01 1.157 (1.053–1.672) 0.032 1.208 (1.021–1.429) 0.027 e' (septal) 0.951 (0.672–0.846) 0.047 0.832 (0.658–0.975) 0.04 0.812 (0.788–0.896) 0.035 Delta Cardiac Index 72h 1.108 (1.100–1.218) 0.044 1.212 (1.008–1.455) 0.03 1.204 (1.003–1.314) 0.04 Abbreviations: MELDNa, Model for End-Stage Liver Disease-Sodium; POC-Echo, point-of-care echocardiography. The same 3 parameters (delta CI72h, E/e' and e' velocity) also remained predictors of mortality when adjusted for baseline MELDNa (Table 1B). The focus of this study is that cardiovascular effect of terlipressin, which includes peripheral vasoconstriction and possible myocardial depression (reduced cardiac output or contractility), interacts with baseline cardiac function and consequently influences the treatment efficacy and adverse outcomes in patients with cirrhosis and hepatorenal syndrome-acute kidney injury who are treated with terlipressin. We do agree, however, that to establish a predictive model for all the variables that represent the heart, the liver, and the kidney in patients with cirrhosis and hepatorenal syndrome-acute kidney injury, a large study with pooled data from multiple sites representing diverse patient populations is required.