Model for End-Stage Liver Disease Score Predicts Outcome in Cirrhotic Patients During Pregnancy

BACKGROUND & AIMS: Pregnancy is rare among patients with cirrhosis, and data about complications and outcomes are sparse. We evaluated the utility of prognostic models of severity of cirrhosis in determining outcomes in pregnant women with cirrhosis. METHODS: We evaluated all cirrhotic patients who self-reported pregnancy at our center and correlated prognostic scores at the time of conception with outcomes. RESULTS: Sixty-two pregnancies occurred in 29 women. The median model for end-stage liver disease (MELD) score at conception was 7 (range, 6-17), the median MELD sodium score was 9 (range, 6-17), the median United Kingdom end-stage liver disease (UKELD) score was 44 (range, 36-53), and the median Child-Pugh score was 5 (range, 5-8). The live birth rate was 58%; the median gestational age was 36 weeks. Higher MELD (P = .01), MELD sodium (P = .01), UKELD (P = .01), and Child-Pugh (P = .03) scores were associated with gestation <37 weeks. Maternal complications (ascites, encephalopathy, or variceal hemorrhage) occurred in 10% of patients and were associated with higher MELD (P = .01) and UKELD (P = .02) scores. Receiver operator curve analysis demonstrated that a MELD score >= 10 predicted, with 83% sensitivity and 83% specificity, which patients were likely to have significant, liver-related complications (area under curve, 0.8); a UKELD score >= 47 had 83% sensitivity and 79% specificity (area under curve, 0.8). No patient who had a MELD score <= 6 or a UKELD score <= 42 developed any significant hepatologic complications. CONCLUSIONS: MELD and UKELD scores at the time of conception can be used to predict specific clinical outcomes in pregnant women with cirrhosis.