ORIGINAL ARTICLE Vitamin D deficiency is associated with mortality in patients with advanced liver cirrhosis € nhage* Caroline S. Stokes*, Marcin Krawczyk*, Christoph Reichel†, Frank Lammert* and Frank Gru € ckenau, *Department of Medicine II, Saarland University Medical Center, Homburg, Germany, †Rehabilitation Center Bad Bru € ckenau, Germany Clinic Hartwald, German Pension Insurance, Federal Office, Bad Bru
ABSTRACT Background Chronic liver disease is the fifth most common cause of mortality in Europe. Recently, vitamin D deficiency has been associated with an increased risk of mortality in the general population. As patients with advanced liver disease frequently exhibit vitamin D deficiency, we assessed for a possible association of vitamin D deficiency with survival in a cohort of patients with advanced liver disease. Methods Sixty-five patients with liver cirrhosis (median age, 58 years; range, 19–76 years; 66% male; ChildPugh stage C, 46%) were included in our prospective single-centre survival study. Serum 25-hydroxyvitamin D concentrations were measured by chemiluminescence immunoassay. The optimal cut-off was determined using receiver operating characteristic (ROC) and Kaplan-Meier analysis. Chi-square statistics and multivariate binary logistic regression analysis were also conducted. Results Median serum vitamin D levels were 82 ng/mL (range 12 years in patients with compensated cirrhosis in contrast to ~ 2 years in those with decompensated cirrhosis .
The model for end-stage liver disease (MELD) score has become widely accepted for predicting survival in patients with cirrhosis. It has been validated in a large variety and number of patients, nonetheless approximately 15–20% of patients remain, for which the MELD cannot accurately predict survival . Therefore, despite the increased incidence of mortality in advanced cirrhosis, patients with comparable risk profiles have variable outcomes . Thus, other yet-to-be-identified factors might mediate risk of death in these patients. Evidence is emerging in support of detrimental effects associated with vitamin D deficiency, which may contribute to the progression of various diseases and ultimately, mortality . An association between vitamin D and mortality was recently illustrated in a large observation study, where the US National Health and Nutrition Examination Survey (NHANES) inversely correlated serum vitamin D levels with all-cause mortality . Moreover, a meta-analysis of 14 prospective cohort studies
ª 2013 Stichting European Society for Clinical Investigation Journal Foundation. Published by John Wiley & Sons Ltd
VITAMIN D AND MORTALITY IN LIVER CIRRHOSIS
in the general population reported a nonlinear decrease in mortality with increasing vitamin D concentrations . With regard to liver diseases, such an observation has been illustrated in early-stage cirrhosis, in which serum 25-hydroxyvitamin D was inversely associated with degree of liver dysfunction and emerged as a possible predictor of mortality . Vitamin D deficiency (defined as 6 ng/mL-censored = ≤ 6 ng/mL-censored
(n = 7). Of note, death from infectious complications (i.e. sepsis) occurred more frequently in patients with serum vitamin D levels ≤6 ng/mL (n = 5) as compared to patients with serum vitamin D levels >6 ng/mL (n = 2). However, as shown in Table 2, no significant differences were detected in the overall cohort with regard to survival when comparing age, sex and aetiology of liver disease (all P > 005). Body mass index (BMI) differed significantly between nonsurvivors and survivors (2371 vs. 2663 kg/m2; P = 0005); however, this may have been influenced, at least in part, by the presence of ascites. Therefore, we did not include BMI in the univariate and multivariate analyses. There was a nonsignificant difference in median vitamin D levels between nonsurvivors and survivors (568 vs. 1008 ng/ mL), and no difference between alcoholic and nonalcoholic patients (853 vs. 804 ng/mL). Moreover, there was no significant difference (v2 = 298; P = 0084) between survivors and nonsurvivors in terms of number of patients sampled during summer and autumn months compared with spring and winter months which are synonymous with lower sun exposure. Also, signs of decompensated cirrhosis (presence of ascites, hepatic encephalopathy and/or variceal bleeding) did not differ between survivors and nonsurvivors. On the other hand, there was a significant difference in survival when comparing the number of patients with serum vitamin D levels above and below the detection threshold (see Methods) for vitamin D assessment (v2 = 514; P = 0024). Specifically, 40% of patients above the detection limit died compared with 73% of those below this threshold. When comparing general characteristics between patients with vitamin D levels ≤6 and >6 ng/mL, no significant differences were detected in age, sex, BMI, aetiology of liver disease, season of blood sampling, MELD score and biochemical parameters. The only significant difference (v2 = 825; P = 0004) was the proportion of nonsurvivors to survivors during follow-up (16/5 vs. 15/25 in the group of patients with vitamin D levels ≤6 and >6 ng/mL, respectively).
Regression analysis 0·2
Time (days) Vitamin D > 6 ng/mL
n = 42
Vitamin D ≤ 6 ng/mL
n = 23
Figure 2 Kaplan–Meier survival analysis confirms low vitamin D levels (≤6 ng/mL) as a significant (P = 0012) predictor of survival for 65 patients with liver cirrhosis.
Univariate analysis confirmed serum vitamin D and MELD scores as being significantly associated with survival; age, sex and aetiology of liver disease were nonsignificant factors (Table 3). This finding was maintained in the multivariate model, where only vitamin D serum concentrations (odds ratio [OR] = 63; 95% confidence interval [CI], 12–312; P = 0012) and MELD scores (OR = 14; 95% CI 12–17; P < 0001) were independent predictors of death.
Discussion Our results firstly confirm a high incidence of vitamin D deficiency in patients with cirrhosis (86% in the current
European Journal of Clinical Investigation Vol 44
C. S. STOKES ET AL.
Table 2 Patient characteristics based on survival status after 24 months Nonsurvivors N (males/females)
BMI (kg/m )
Median age, years (range)
Cryptogenic liver cirrhosis
Viral Hepatitis (HBV/HCV)
A (5–6 points)
B (7–9 points)
Primary liver disease (n) Alcoholic liver disease
Child-Pugh stage (n)
C (10–15 points) MELD score
Medium serum 25-hydroxyvitamin D (ng/mL) (range)