Potential conflict of interest: Dr. Höblinger received grants from Roche and Novartis.
Supported by BONFOR (Bonner Forschungsförderprogramm) and HOMFOR Saarland University research program. This work was also financed by the Deutsche Forschungsgemeinschaft (DFG Collaboration Research Centre SFB/TRR 57 TP 01, LA 997/5-1 and LA997/6-1 to F.L.). 25(OH)-vitamin D measurements were supported by BioPersMed (COMET K-project 825329), which is funded by the Austrian Federal Ministry of Transport, Innovation and Technology (BMVIT) and the Austrian Federal Ministry of Economics and Labour/the Federal Ministry of Economy, Family and Youth (BMWA/BMWFJ) and the Styrian Business Promotion Agency (SFG).
Recently, genome-wide studies identified genetic variants that affect serum 25-hydroxyvitamin D levels in healthy populations (rs12785878, near dehydrocholesterol reductase, DHCR7; rs10741657, at CYP2R1; and rs7041, at vitamin D binding protein, GC). Because vitamin D deficiency is associated with advanced liver disease, we hypothesized that these variants are associated with 25(OH)-vitamin D levels and liver fibrosis. Overall, 712 Caucasian patients with chronic liver diseases were included. Liver fibrosis was assessed by transient elastography (TE) and/or histology. Serum levels of 25(OH)-vitamin D were correlated with TE and fibrosis stages. Genotypes were determined using TaqMan assays and tested for association with vitamin D and liver stiffness. Serum 25(OH)-vitamin D levels were inversely correlated with liver stiffness and histology (P < 0.001). Homozygous carriers of the rare DHCR7 allele or the common CYP2R1 allele presented with reduced 25(OH)-vitamin D levels (P < 0.05). The variant rs12785878 in the DHCR7 locus was associated with liver stiffness in both patients with TE <7.0 kPa and TE between 7.0 and 9.5 kPa. 25(OH)-vitamin D levels correlated with sunshine hours at the time of inclusion (P < 0.001). Conclusion: Common variation in 25(OH)-vitamin D metabolism is associated with liver stiffness in patients presenting with low to moderately increased elasticity. Although the susceptible DHCR7 genotype confers small risk, we speculate that the observed stiffness differences indicate a stronger influence of 25(OH)-vitamin D on initiation rather than progression of hepatic fibrosis. (HEPATOLOGY 2012;56:1883–1891)
Vitamin D is a key regulator of calcium homeostasis.1, 2 Recent studies have shown that it also modifies immune reactions, including T-cell functions that are critical for the host response to chronic hepatitis C virus (HCV) infection.3-5 Indeed, supplementation of vitamin D is associated with a favorable outcome in chronic inflammatory diseases such as tuberculosis, multiple sclerosis, and psoriasis.6, 7 In line with these results, large observational studies associated vitamin D deficiency with overall mortality in the general population.8-11 Data from the latest population-based surveys report that up to 90% of patients with chronic liver disease present with severe vitamin D deficiency12 as compared to 40%-45% in healthy Western populations.13 Moreover, individuals with chronic HCV infection and concomitant vitamin D deficiency have a reduced chance of viral clearance.3, 14, 15 Conversely, preliminary reports suggest a beneficial effect of vitamin D supplementation on treatment outcome in patients with chronic HCV infection under interferon and ribavirin treatment.14, 16
The metabolism of 25-hydroxy-vitamin D (25(OH)-vitamin D) is regulated by several environmental factors, in particular sunlight and diet. In addition, genetic factors may contribute up to 43% of the variation of serum 25(OH)-vitamin D concentrations as determined by studies in twin populations and large screening populations.17, 18 These results have been underpinned by two large genome-wide association studies (GWAS) demonstrating that serum concentrations of 25(OH)-vitamin D are linked to variants of genes controlling vitamin D synthesis and transport. However, the detected variants account for only 5% of the variation of vitamin D levels in the general population.19, 20
Here we aimed to clarify whether the reported single nucleotide polymorphisms (SNPs) near the DHCR7 locus (encoding 7-dehydrocholesterol reductase), the CYP2R1 gene (encoding cytochrome P450, family 2, subfamily R, polypeptide 1) and the vitamin D binding protein (also known as group-specific component [GC] globulin) are also associated with 25(OH)-vitamin D concentrations and liver fibrosis in patients with chronic liver diseases. All three genes control key steps in vitamin D metabolism and transport (Fig. 1): DHCR7 is a rate-limiting enzyme that removes the substrate 7-dehydrocholesterol from the 25(OH)-vitamin D synthesis pathway, which is otherwise transported to the skin where conversion in cholecalciferol is induced by ultraviolet light. Subsequently, CYP2R1 hydroxylates cholecalciferol to 25(OH)-vitamin D3, which is then transported to other hydroxylation sites by GC globulin.21 We now studied the common SNPs controlling vitamin D homeostasis in a large cohort of patients with chronic liver diseases of various etiologies, which we recently recruited for the first noninvasive elastography-based genetic study of liver fibrosis.22
GC, group-specific component; GWAS, genome-wide association study; HCV, hepatitis C virus; HWE, Hardy-Weinberg equilibrium; SNP, single nucleotide polymorphism; TE, transient elastography.
Patients and Methods
Patients (n = 712) were recruited in a project aiming to identify genetic susceptibility factors for liver fibrosis in patients with chronic liver diseases, irrespective of the underlying etiology, at University Hospital Bonn, Germany.22 Patients were included if they presented with known chronic liver disease such as chronic replicating viral hepatitis B or C with or without elevated serum aminotransferase activities, chronic alcohol abuse and elevated aminotransferase or γ-GT levels, previously diagnosed primary biliary cirrhosis or primary sclerosing cholangitis, known hereditary liver diseases (HFE-hemochromatosis, α1-antitrypsin deficiency, Wilson's disease) and with persisting (>6 months) elevation of liver enzymes.
Applying stringent criteria from genetic studies, only patients of Caucasian descent were included. The ethnicity of our patients was based on the places of birth of their grandparents. Informed consent was obtained from all patients prior to inclusion. The study protocol was approved by the Research Ethics Committee of the University of Bonn and the study was performed according to the Declaration of Helsinki.
Assessment of Liver Fibrosis by Transient Elastography and Liver Histology.
In all subjects, levels of liver fibrosis were determined by transient elastography (TE, Fibroscan) as described in the Supporting Methods, expressing hepatic stiffness in kPa. In a subgroup of patients (n = 213), liver histology had been performed as described in our previous study.22 Liver sections were stained with hematoxylin and eosin and Sirius red. Two pathologists, who were blinded to the elastography results, staged fibrosis according to Desmet and Scheuer, and significant interobserver agreement for fibrosis staging was indicated by Cohen's κ value (0.71, P < 0.0001).22
Measurement of 25(OH)-vitamin D Serum Levels, Clinical Chemical Assays, and Genotyping.
Vitamin D measurements, standard clinical chemical assays, and genotyping of the variants rs12800438 (DHCR7), rs10741657 (CYP2R1), and rs7041 (GC) are described in the Supporting Materials.
Statistical analysis was performed with SPSS 19.0 (IBM, Ehningen, Germany). For statistical significance, a two-sided P-value of 0.05 was used as threshold. TE results were analyzed for correlation with liver fibrosis stages obtained by histopathological assessment of liver biopsies. Allele and genotype frequencies of the three polymorphisms were tested for consistency with Hardy-Weinberg equilibrium (HWE) using exact tests. The 25(OH)-vitamin D levels were correlated to the genotypes of all SNPs and compared by nonparametric Kruskal-Wallis tests. Post-hoc pairwise comparison was conducted for comparisons of vitamin D levels between histological liver fibrosis stages or TE. Nonparametric tests (Spearman rank) were applied to assess correlations.
Linear regression analyses were applied to identify independent determinants of serum vitamin D levels. We tested whether the following parameters are associated with vitamin D levels: age, gender, body mass index (BMI), inclusion during “winter” months (October to March), liver stiffness, and the presence of any risk genotype.
Similarly, linear regression analyses were performed to identify independent determinants of liver stiffness. Here we tested whether the following factors are associated with stiffness: age, gender, alcohol consumption, and the presence of any risk or the presence of any risk genotype.
Potentially confounding factors from these univariate analyses with P ≤ 0.1 entered the respective multivariate linear regression analysis. In the stepwise multivariate analysis, likelihoods for inclusion and exclusion were set to P = 0.05 and 0.1, respectively. For categorical variables, dummy variables were entered in the linear regression analysis: gender (1 = women; 2 = men); inclusion time (1 = summer; 2 = winter); DHCR7 rs12785878 genotype (1 = TT/TG; 2 = GG); and CYP2R1 rs10741657 genotype (AA = 1; GG/GA = 2).
We recruited 712 individuals (median age 52 years) with chronic liver disease. Table 1 summarizes the clinical characteristics of the study cohort. All patients were of Caucasian ethnicity (see Methods). Male patients were overrepresented in our cohort (83.3%). The majority of patients tested anti-HCV positive (60.8%). Alcoholic and nonalcoholic fatty liver diseases were the second most common causes for chronic liver injury (22.1%). All other etiologies were rare (Table 1).
Table 1. Clinical Data of Patients
Abbreviations: AIH, autoimmune hepatitis; BMI, body mass index; HCV, hepatitis C virus; kPa, kilopascal; n, number; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; SD, standard deviation.
Age (median, range; years)
Male (n, %)
Etiology (n, %)
Nonalcoholic fatty liver disease
Autoimmune disease (AIH, PBC, PSC)
Alcohol (mean ± SD; g/d)
BMI (median, range)
Liver stiffness (median, range; kPa)
Histological fibrosis stage (n, %)
Liver histology was performed in 29.9% of all patients, and the interval from liver biopsy to inclusion varied from 4 weeks to 2 years. Liver stiffness measurements (kPa) and fibrosis stages were significantly correlated (r = 0.66; P < 0.001), in line with previous reports.26, 27
Distribution of Serum 25(OH)-vitamin D Concentrations.
The mean 25(OH)-vitamin D concentration was 27.7 ± 15.4 ng/mL. Figure 2 illustrates that the distribution of 25(OH)-vitamin D concentrations was skewed toward low levels. According to previous and recently published recommendations,12, 13, 28, 29 63.1% of patients presented with vitamin D deficiency, as defined by 25(OH)-vitamin D levels below 30 ng/mL (Fig. 2).
We detected a positive correlation of serum 25(OH)-vitamin D concentrations with serum albumin levels (r = 0.45) and reverse correlations with serum bilirubin levels (r = −0.25) as well as aspartate aminotransferase (AST) (r = −0.14) and γ-GT (r = −0.28) activities (all P < 0.05). Subgroup analysis showed that these associations were limited to patients with TE results ≥7 kPa (data not shown).
Association of 25(OH)-vitamin D Levels with Liver Stiffness and Stage of Liver Fibrosis.
Supporting Fig. 1A demonstrates that 25(OH)-vitamin D concentrations correlated negatively with TE results (r = −0.267; P < 0.001; Spearman rank). Although the correlation coefficient decreased markedly after exclusion of patients with high TE values (e.g., >20 kPa), the association remained highly significant (r = −0.109; P = 0.01) (Supporting Fig. 1B). In HCV patients, who represent the largest subgroup in the study cohort, we found a similar correlation with liver stiffness (r = −0.218; P < 0.001).
Supporting Fig. 1C displays that patients with liver stiffness ≥7 kPa presented with reduced 25(OH)-vitamin D levels as compared to patients with TE <7 kPa (28.3 versus 22.1 ng/mL; P < 0.001). Figure 3 shows that, similarly, lower 25(OH)-vitamin D levels were associated with advanced stages of liver fibrosis according to the Desmet and Scheuer staging system (n = 213, P < 0.001).
Association of Genetic Variants with 25(OH)-vitamin D Concentrations.
Table 2 summarizes the allele and genotype distributions for the entire cohort. Call rates for all variants in the study were >97%. No deviation from HWE was detected (P > 0.05 for all genotypes).
Table 2. Allele and Genotype Distribution of the Three Tested SNPs in the Total Cohort (n = 712)
Frequency in % (Number of patients)
Polymorphisms in the DHCR7 and the CYP2R1 loci were associated with serum 25(OH)-vitamin D levels. Figure 4A illustrates that homozygous carriers of the rare allele of variant rs12785878 (DHCR7) presented with lower median level 25(OH)-vitamin D levels as compared to carriers of the common allele (20.1 ng/mL [interquartile range, IQR, 16.8] versus 24.9 ng/mL [IQR 20.6], P = 0.03). In addition, as illustrated in Fig. 4B, carriers of the common allele of SNP rs10741657 (CYP2R1) also presented with lower 25(OH)-vitamin D levels as compared to homozygous carriers of the rare allele (23.9 ng/mL [IQR 20.0] versus 29.5 ng/mL [IQR 20.0], P = 0.045).
Overall, 42 (6.1%) of the patients for whom both genotypes were available carried compound “at risk” genotypes, i.e., were homozygous for the rare allele of rs12785878 (DHCR7) and carried at least one common allele of rs10741657 (CYP2R1). Median vitamin D levels in carriers of this at-risk combination were significantly (P = 0.04) lower than in patients with any other genotype combination (19.8 versus 24.9 ng/mL). In the cohort, 25(OH)-vitamin D concentrations were not associated with the GC globulin variant (Fig. 4C).
Subsequently, we tested the associations in patients with TE <7 kPa (n = 371) and TE ≥7 kPa (n = 341) separately. Noteworthy, in patients with TE <7 kPA the association remained significant (P = 0.008) for rs12785878 (DHCR7), and a trend (P = 0.08) was observed for rs10741657 (CYP2R1), whereas all SNPs failed to pass the significance threshold in patients with TE ≥7 kPa.
As a majority of patients in the study cohort suffered from chronic HCV infection, we explored this subgroup. In line with the results in the total cohort, we observed significantly decreased vitamin D concentrations in homozygous carriers of the rare allele SNP rs12785878 near DHCR7 (21.4 ng/mL [IQR 17.2] versus 27.3 g/mL [IQR 20.9], P = 0.02) However, in this subcohort vitamin D concentrations were not associated with the CYPR1 or GC polymorphisms.
To clarify whether the SNPs in the DHCR7 and the CYP2R1 loci are independent determinants of vitamin D levels, uni- and multivariate linear regression analysis was applied. Age, gender, BMI, inclusion during “winter” months (see Methods and Supporting Fig. 2), liver stiffness, and the two genotypes were tested by univariate analyses. Supporting Table 1 shows that gender, inclusion during “winter” months, stiffness, and the genotypes were significantly associated with vitamin D levels. These factors were tested in the multivariate analysis. Supporting Table 2 summarizes that the DHCR7 (R(B) = −5.59; 95% confidence interval [CI] = −9.63 to −1.56; P = 0.007) and CYP2R1 (R(B) = −3.50; 95% CI = −6.50 to −0.51; P = 0.02) genotypes were independently associated with vitamin D levels in addition to liver stiffness and time of inclusion.
Association of Gene Variants Controlling Vitamin D Homeostasis with Liver Stiffness.
In the overall cohort, none of the tested variants was associated with liver stiffness. Of note, when restricting the analysis to patients with no or slightly increased liver stiffness (i.e., with TE <7 kPa), we observed an association with DHCR7. Figure 5A demonstrates that the rare allele of the DHCR7 variant (rs12785878), which was associated with lower 25(OH)-vitamin D concentrations (Fig. 4A), was also associated with increased liver stiffness. Figure 5B shows that the association was not present in the group of patients with TE ≥7 kPa.
Of note, when we changed the cutoff up to 9.5 kPa the association remained significant for the lower stiffness group (i.e., <9.5 kPa) of the total cohort, but in patients with TE ≥9.5 kPa no association could be detected. To further clarify the association in patients with higher liver stiffness, we also analyzed patients with TE results between 7.0 and 9.5 kPa separately (n = 98). For this small group we observed P = 0.05 for the association of the rare DHCR7 allele with increased liver stiffness (7.9 kPa [IQR 1.2] versus 8.7 kPa [IQR 0.8]).
Because patients with TE results <7 kPa may have no fibrosis in liver biopsy, we analyzed the available histological data of these patients (n = 63) and found that the majority of our patients (81%), despite low TE (<7 kPa) values, presented with histological evidence for liver fibrosis (fibrosis stage ≥F1). From the patients with intermediate liver stiffness measurements between 7.0 and 9.5 kPa, only 19 liver biopsies were available, and 16 of 19 (84%) reported liver fibrosis ≥F1.
Although we detected an association with vitamin D levels, no association with liver stiffness was observed for the CYP2R1 variant (Supporting Fig. 3). In the HCV subcohort we observed a similar association of the rare DHCR7 allele (P = 0.005) with liver stiffness in patients with TE results <7 kPA and no association with the other variants. As a noninvasive fibrosis surrogate marker, Forns index could be calculated in 230 patients (median 8.2; range 2.9-18.2). The index was negatively correlated with vitamin D levels (r = −0.192; P < 0.001), but we detected no association with any of the three SNPs tested.
To analyze whether the DHCR7 risk genotype is independently associated with liver stiffness, uni- and multivariate linear regression analysis was applied. Age, gender, alcohol consumption, and the DHCR7 genotype were tested by univariate analyses in patients with TE levels up to 9.5 kPa (Supporting Table 3). Gender and genotype were entered in the multivariate analysis and Supporting Table 4 shows that the DHCR7 SNP was independently associated with liver stiffness (R(B) = 0.59; 95% CI = 0.05-1.14, P = 0.03), whereas we observed a trend for gender (R(B) = −0.32; 95% CI = −0.65 to 0.13; P = 0.06). Of note, restricting the linear regression analysis to patients with moderately increased liver stiffness (7.0-9.5 kPa) confirmed the independent association of the DHCR7 variant with liver stiffness (R(B) 0.58; 95% CI = 0.10-1.07; P = 0.02), whereas none of the other risk factors was independently associated with liver stiffness (all P > 0.05).
Our knowledge of the 25(OH)-vitamin D status in chronic liver disease is based on studies with small numbers of patients with predominant inclusion of patients with cholestatic liver disease or advanced cirrhosis.12, 30-32 Here we analyzed the 25(OH)-vitamin D status in a large cohort of Caucasian patients (n > 700) with chronic liver diseases of mixed etiology and fibrosis stages and relate our findings to genetic variants affecting 25(OH)-vitamin D homeostasis. The current analysis demonstrates that not only the progression of liver fibrosis and environmental factors but also recently identified common gene variants affect 25(OH)-vitamin D homeostasis and liver stiffness in patients with chronic liver diseases to a certain extent. Our findings indicate that this association is apparently confined to mild elevation of liver stiffness, and we conclude that in advanced liver disease environmental factors appear to be more critical for the control of 25(OH)-vitamin D levels.
Recently, serum 25(OH)-vitamin D levels have been related to common genetic variants.19, 20 The candidate genes identified by these studies represent key players in vitamin D synthesis and transport (Fig. 1). Our study is the first to confirm the association of the same variants (DHCR7 and CYP2R1) with 25(OH)-vitamin D levels in patients with chronic liver diseases. The previously described association with variation of the GC gene could not be confirmed. We note, however, that the initial GWAS19, 20 studied healthy individuals, and it cannot be ruled out that in patients with liver disease other genetic factors affect vitamin D and GC globulin levels. Furthermore, differences in gene/environment interactions in defined populations cannot be excluded.
It has been reported that vitamin D levels depend on a variety of factors such as BMI, sunshine exposure, gender, dietary and supplemental intake. However, we could confirm that the two associated SNPs are also independent determinants of serum vitamin D levels. Interestingly, the association was confined to patients with no or mildly increased liver stiffness (TE <7 kPa), whereas we did not detect any association between the SNP and 25(OH)-vitamin D levels in patients with increased liver stiffness (TE ≥7 kPa). This indicates that other factors, and advanced liver disease per se, are more likely to have a significant impact on serum 25(OH)-vitamin D levels in this setting.
With respect to the influence of the gene variants on fibrogenesis, we show that the SNP near the DHCR7 gene is associated with increased liver stiffness. Variations of the cutoff from 7.0 to 9.5 kPa uniformly lead to association in the lower stiffness group. However, including more patients with advanced fibrosis by further increasing the cutoff TE ≥9.5 kPa abolished the association, supporting our view that genetic predisposition to fibrosis by these variants is limited to early increase in liver stiffness. Our results are limited to a single cohort and by incomplete additional information (e.g., surrogate markers) on fibrosis phenotypes. Hence, it may be argued that patients with TE <7 kPa have no fibrosis at all. Therefore, we also performed a subgroup analysis of patients with intermediate liver stiffness values (between 7.0 and 9.5 kPa), which confirmed the association between DHCR7 and liver stiffness.
None of the studied SNPs has been directly related to functional changes and the variants might be in linkage disequilibrium with yet unknown functional mutations. Thus, the mechanism of the described association remains to be defined. Because serum levels of vitamin D were not correlated with aminotransferase activities in these patients, it is unlikely that the effect is due to increased hepatocellular injury. Furthermore, the inclusion of 25(OH)vitamin D levels in a linear regression model of TE revealed a significant contribution of this factor in the overall cohort of patients (data not shown), whereas in the subcohort of patients with no or mildly elevated TE levels, no significant effect could be observed. Therefore, the concept of low vitamin D levels as profibrogenic stimulus might be too simplistic, and indirect functional mechanisms linking the DHCR7 variant to liver stiffness (and fibrogenesis) could also be present. In this regard, dysregulation of vitamin D synthesis pathways might lead to altered signaling by way of the nuclear receptor VDR (a.k.a. NR1I1) on hepatic stellate cells.33-35 Alternatively, vitamin D and its metabolites could play a role as natural antiviral mediators36 or might ameliorate the progression of liver fibrosis by way of natural killer cells.37, 38
It has long been recognized that patients with chronic liver diseases suffer from vitamin D deficiency, especially in chronic cholestatic conditions such as primary biliary cirrhosis.39-41 Vitamin D deficiency in these patients has been attributed to impaired intestinal absorption and increased urinary excretion.40 It remains a matter of debate whether progressive vitamin D deficiency in advanced liver fibrosis is related to a net loss of hepatocytes and/or reduction in hydroxylation capacity for cholecalciferol.31 Here, we confirm that patients with advanced fibrosis display the lowest serum vitamin D levels as compared to patients with early fibrosis. However in contrast to previous reports, the proportion of patients with very severe vitamin D deficiency was only 65%, comparable to epidemiological studies in healthy Western populations,12, 13 and our cohort included less than 10% of patients with chronic cholestatic liver diseases but only 20% of patients with liver stiffness values corresponding to advanced fibrosis or cirrhosis.
One important environmental factor of vitamin D synthesis is sunlight.42, 43 Supplementation can overcome vitamin D deficiency, but sunlight exposure also appears to be effective, albeit to a lesser extent.43 Notably, the serum 25(OH)-vitamin D levels in our cohort were closely related to sunlight hours at the time of patient enrolment (Supporting Fig. 2). In contrast to the genetic associations, this was irrespective of the underlying degree of liver stiffness or fibrosis stage. This observation indicates that vitamin D concentrations in patients with chronic liver disease depend strongly on external factors such as vitamin D intake and/or sunlight exposure.
Although our patients presented with various chronic liver diseases, we predominantly included patients with chronic HCV infection. Increasing evidence suggests an important role of vitamin D and gate-keepers of vitamin D concentrations such GC globulin for the spontaneous and treatment associated course of patients with chronic HCV infection and HCV/HIV (human immunodeficiency virus) coinfection, with a more favorable outcome in individuals with sufficient vitamin D levels.14, 15, 44, 45 The exact mechanisms underlying these findings remain to be elucidated. Immune modulatory functions of vitamin D and vice versa control of vitamin D levels by immune cells have been shown to influence T-cell and monocyte functions, both important for control of HCV infection and fibrogenesis.46 Subgroup analysis of the patients with chronic hepatitis C infection did not reveal differences in serum vitamin D concentrations as compared to other etiologies of fibrosis in our cohort (data not shown). In these patients, we observed a consistent association with vitamin D levels and liver stiffness for the DHCR7 polymorphism but a lack of association with CYP2R1 and GC variants. The differences in associations may be due to etiology-specific association or decreased power of the subgroup analysis.
In summary, gene variation in vitamin D metabolism is associated with liver stiffness in patients with chronic liver disease. Although the absolute differences in stiffness are small, we speculate that the consistent association in patients with mild to moderate fibrosis due to various chronic liver diseases indicates an influence of vitamin D on fibrosis initiation rather than progression in general.
We thank Miriam Langhirt (Saarland University Hospital) for technical assistance. The authors also thank four anonymous reviewers for helpful and critical comments.