Meta-analysis: Vitamin D and Non-alcoholic Fatty Liver Disease

M. Eliades; E. Spyrou; N. Agrawal; M. Lazo; F. L. Brancati; J. J. Potter; A. A. Koteish; J. M. Clark; E. Guallar; R. Hernaez


Aliment Pharmacol Ther. 2013;38(3):246-254. 

In This Article


Our search identified 805 unique references of which 789 did not meet our inclusion criteria, resulting in 17 papers included for the systematic review. Reasons for exclusions are shown in Figure 1.

Figure 1.

Flow diagram of the study.

Of the 17 studies included,[18–23,25–35] six originated in North America, four from Asia and the rest in Europe or Israel (Table S3). Six studies were conducted in general population settings, five in out-patient settings, one included both inpatients and out-patients, one only inpatients and two studies did not report the clinical setting.[25,34] The mean age of the participants was similar amongst the studies with the exception of the study by Katz et al.,[29] which was conducted in adolescents. Not unexpectedly, those subjects with NAFLD had higher body mass index (BMI) and homeostasis model assessment-insulin resistance (HOMA-IR) indexes compared with subjects without NAFLD. Half of the studies reported participant ethnicity of which four studies were of Korean or Chinese population.

NAFLD was ascertained by liver biopsy in four studies,[19,25,32,35] by imaging (ultrasound or CT) in ten studies[18,20–23,26–28,33,34] and by elevated ALT levels in three studies after excluding other causes of abnormal liver enzymes.[29–31] The cut-off level of 25(OH)D for vitamin D deficiency also varied across studies, ranging from 12 to 30 ng/mL; however, most of the studies that reported dichotomous outcomes used the value of 20 ng/mL as the cut-off level. In four studies, the analysis was conducted based on quartiles of vitamin D levels. Three studies did not report the cut-off level of vitamin D deficiency.[19,25,30]

Overall, most of the studies showed higher prevalence of vitamin D deficiency in NAFLD subjects compared with controls.[19,22,23,26,28,33–35] In two studies,[18,29] the association of vitamin D and NAFLD was not significant after adjustment for other covariates; and, two other studies[30,31] showed an inverse association between vitamin D levels and presumed NAFLD (as evident by elevated ALT levels). Using histopathological results, two reports[25,32] showed that low levels of vitamin D were associated with increased likelihood of fibrosis and inflammation and therefore progression to NASH.


Of the 17 studies included in the systematic review, 11 provided data to conduct the pooled analyses: 9 were used for the continuous vitamin D levels[19,20,22,26,28,29,31,33,35] (Table 1 and Figure 2) and 9 for the dichotomous exposure (vitamin D deficiency),[18,20,22,23,26,28,29,31,35] Figure 3. The reasons for exclusion were the presence of only NAFLD subjects in the study population in two studies;[25,32] the evaluation of metabolic syndrome as the study outcome;[30] the lack of detailed data on vitamin D levels,[18,21,23] the publication of duplicate data from another study from the same investigators[34] and the restriction to hypertensive NAFLD participants that could introduce heterogeneity in the analyses.[27] The meta-analysis for the continuous levels of vitamin D by NAFLD status included 12 794 participants (4855 NAFLD cases and 7939 controls). On average, NAFLD patients had 0.36 ng/mL lower levels of 25(OH)D levels compared to controls (SMD: 0.36 ng/mL, 95% CI: 0.32,0.40 ng/mL) (I 2 99%, P < 0.01). Western NAFLD participants had lower levels of vitamin D compared with their Eastern NAFLD counterparts (Figure 2).

Figure 2.

Meta-analysis of studies reporting 25(OH)D levels in NAFLD vs. controls, standardised mean difference with 95% confidence interval, chronologically ordered.

Figure 3.

Meta-analysis of studies reporting dichotomous outcomes of 25(OH)D levels in NAFLD vs. controls and estimated ORs with 95% confidence interval, chronologically ordered.

For the presence of vitamin D deficiency, nine studies totalling 5202 NAFLD participants and 8520 controls were included. NAFLD participants were 1.26 times more likely to have vitamin D deficiency (OR 1.26, 95% CI: 1.17, 1.35) compared to their controls. After stratifying for ethnic background, Western NAFLD participants were more likely to be vitamin D deficient as compared with their Eastern counterparts (Figure 3).

The exclusion of any individual studies did not markedly affect the overall measure of the association. In addition, different methods of ascertainment on the quantitative levels of vitamin D have not shown changes in the inferences. More specifically, in the dichotomous analysis (presence or absence of vitamin D deficiency), even after using different models of adjustment – from none in Barchetta and Foster;[18,26] metabolic syndrome, insulin resistance and serum triglycerides in Liangpunsakul;[31] age, gender, race and poverty status in Katz[29] and finally, age, sex, BMI, calcium, creatinine, HOMA-IR and ATP III in Targher[35] – none of the positive associations between NAFLD and low vitamin D deficiency disappeared. We found no evidence of publication bias as evidence by Egger's test P-value (P = 0.32)