MELD-albumin score is strongly associated with all-cause mortality in patients with acute myocardial infarction complicated by cardiogenic shock

Abstract

Abstract Background Kidney and liver injury are frequent complications in patients with acute myocardial infarction complicated by cardiogenic shock (AMICS) because of hypoperfusion and venous congestion. Model of End-Stage Liver Disease replacing INR with albumin (MELD-Albumin) score quantifies liver and kidney function and has been associated with mortality in acute heart failure. Higher MELD-albumin score indicates particularly more severe hepatic dysfunction. The predictive value of MELD-albumin score has yet to be assessed in AMICS. Methods This is a retrospective, multicenter observational cohort study of 1,716 patients admitted with AMICS at two tertiary facilities in Denmark between 2010 and 2017. MELD-albumin was calculated for each patient individually from 0–72-hour peak s-albumin, s-creatinine and s-bilirubin values in patients alive beyond day 2. Patients were stratified into groups according to score (Low: 1-10, Intermediate: 11-25, High: >25). Using national health registries, we performed follow-up until a maximum of 5 years (median 3.4 years). The primary outcome was all-cause mortality. Results Of 1716 patients in the cohort, 1284 (75%) were alive beyond admission day 2. Albumin, creatinine, and bilirubin measurements were available in 764 patients (60%). The median MELD-albumin score was 17. In the comparison across high vs. intermediate vs. low score groups, the high score group revealed a significantly higher prevalence of hypertension (61 vs. 49 vs. 39%, p=0.038), diabetes mellitus (33 vs. 18 vs. 13%, p=0.003), and history of acute myocardial infarction (24 vs. 13 vs. 17%, p=0.026). Fewer had suffered out-of-hospital cardiac arrest (34 vs. 52 vs. 48%, p=0.014) and revascularization attempts (82 vs. 93 vs. 92%, p=0.007) in the high score group. No significant differences were observed in the initial hemodynamic pressures, lactate levels, left ventricular ejection fraction (LVEF), number of diseased vessels and culprit lesions between the three groups. Patients in the high score group had a 30-day mortality rate of 58%, which was higher compared to 40% in the medium- and 17% in the low score group (Figure 1, p<0.001, using the log-rank test). These findings were still significant when excluding s-creatinine from the calculations. In a Cox Regression analysis, a MELD-albumin score of >25 was associated with a hazard ratio (HR) of 4.34 for 5-year mortality (p < 0.001), whereas a score of 11-25 was associated with a HR of 2.52 (p < 0.001). When adjusting for age, diabetes, hypertension, dyslipidemia, peripheral arterial disease, STEMI, left ventricular ejection fraction and baseline lactate, scores of >25 and 11-25 were still associated with similar 4- to 2.5-fold increases in 5-year mortality risks. Conclusions This study revealed a strong association between MELD-albumin score and all-cause mortality among patients with AMICS, supporting kidney and liver function inclusion in future risk stratification models.