Vitamin A levels reflect disease severity and portal hypertension in patients with cirrhosis

Fat-soluble vitamins A (retinol), D (cholecalciferol), E (tocopherols and tocotrienols), and K (phylloquinone and menaquinones) are lipophilic molecules with distinct physiological properties. While dietary intake is an important (or even exclusive) source for all vitamins, the lipophilicity of fat-soluble vitamins requires mediation of bile acids (BA) for intestinal uptake [1].

Dietary uptake is the exclusive source of vitamin A (VitA) in humans [2]. Notably, 60–95% of VitA is stored in the liver of healthy individuals, while only minor fractions are located in extrahepatic tissues [3]. Importantly, activated hepatic stellate cells (HSCs) are the main driver of fibrogenesis upon liver injury by producing extracellular matrix proteins. During their activation process, HSCs lose their lipid droplets containing retinyl esters [4, 5]. In patients with non-alcoholic fatty liver disease (NAFLD) undergoing bariatric surgery, reduced levels of serum and hepatic retinol, as well as retinoic acid (RA) have been observed, being inversely correlated with grade of hepatic steatosis and severity of non-alcoholic steatohepatitis (NASH) [6]. In other etiologies of (advanced) chronic liver disease, VitA deficiency was reported in a considerable percentage of patients and associated with higher fibrosis stages or cirrhosis [7‐9].

Vitamin D (VitD) is mainly derived from the endogenous synthesis in the skin followed by two critical steps of hydroxylation in the liver and the kidneys [10]. Importantly, a previous study on patients with cirrhosis displayed a link between severe vitamin D deficiency (VitD_Def) and increased levels of inflammatory biomarkers as well as the risk of hepatic decompensation [11]. Similarly, disease severity and mortality were associated with VitD_Def [12, 13].

Vitamin E (VitE) functions as an antioxidant by scavenging peroxyl radicals and regulating the oxidation process of polyunsaturated fatty acids, and is widely reported as a modulator and promoter of immunity [14]. However, while the direct effect of VitE on immunity-/oxidative stress-related pathways may be overestimated, prior studies have demonstrated an association between disease severity and oxidative stress in NAFLD patients [15].

This study aimed to investigate the prevalence of vitamin A, D, and E deficiencies in prospectively recruited patients with advanced chronic liver disease (ACLD) and to determine their association with severity of liver disease and portal hypertension (PHT). Furthermore, we assessed the previously unreported relation to serum levels of bile acids and fibrosis markers to address pathophysiological concepts of vitamin A homeostasis.

This study is the first to simultaneously describe VitA/D/E status in a large prospective cohort of 234 ACLD patients with characterized severity of PHT. We found a VitA_Def in 73% of our patients and its prevalence increased with CTP stages and was associated with dACLD. Similarly, 57% had VitD_Def, which also displayed a distinct association with disease severity. Conversely, we found that only 3% had low VitE serum levels and did not observe an association of VitE levels with liver dysfunction or PHT severity. Of note, our study provides novel data by investigating the relationship of these fat-soluble vitamins with serum BA levels as well as ELF score (both available in about 95% of patients), which is a well-established and an increasingly used non-invasive surrogate parameter for fibrosis.

The liver represents the key regulator of VitA homeostasis, its storage in HSCs, intestinal absorption by BA production and its systemic distribution by the synthesis of retinol-binding protein 4 (RBP4) and transthyretin [18]. As mentioned above, prior studies on patients with NAFLD reported reduced VitA levels both in the serum and the liver [6], being inversely correlated with the grade of hepatic steatosis and severity of NASH [19]. It is reasonable to hypothesize that during the progression of fibrosis, HSCs lose their essential ability to store VitA, subsequently leading to progressive VitA_Def as fibrosis accumulates in the liver. Ultimately, this process might be reflected by reduced VitA serum levels. This assumption is supported by data from the literature [7‐9], as well as the moderate/strong correlation of VitA levels with markers of hepatic dysfunction and, importantly, ELF score, in our study. However, it has to be acknowledged that VitA serum levels may not mirror hepatic VitA content [20].

Furthermore, VitA_Def may also be due to inadequate intake or intestinal malabsorption. In this regard, it seems very interesting that elevated bile acids were independently associated with VitA_Def. It might be argued that serum levels might not reflect the availability of bile acids in the intestines and that impaired bile acid homeostasis and biliary excretion may rather reflect impaired liver function in cirrhosis [21]. Intriguingly, retinoic acid is required for activation of retinoid X receptor-alpha (RXR-α) as a heterodimeric partner of the farnesoid X receptor (FXR) and other key nuclear receptors required for transactivation of genes maintaining proper bile acid homeostasis and excretion [22]. Importantly, elevated serum BA levels may also be a reflection of portosystemic shunting, and thereby, the degree of PHT [23, 24].

Dietary intake of VitA was not systematically assessed in our patients, which represents an important limitation of this study. Interestingly, a previous study on patients with cirrhosis reported that adequate VitA intake did not prevent VitA deficiency [25]. Similarly, another study on patients with (A)CLD applied the relative-dose-response method to assess hepatic VitA storages after oral VitA challenge and reported that neither duration nor dosage of VitA supplementation was independently associated with an effective test response. In contrast, low RBP was the only independent parameter associated with failed test response [26]. Since RBP is synthesized in the liver, it might be hypothesized that VitA serum levels are closely associated with its synthetic capacity and that low serum VitA levels are simply a consequence of impaired hepatic function. Several findings of our study support this hypothesis: First, CTP score – reflecting important aspects of hepatic dysfunction was independently associated with VitA_Def. Second, we observed a moderate/strong correlation of VitA with all laboratory parameters reflecting hepatic synthesis capacity, including vitamin K-dependent and independent coagulation parameters. For example, AT-III and Protein C activity were reduced by nearly 50% in the lowest vs. highest VitA quintile. Third, the independent association of CTP score with VitD (25-OH-vitamin D) deficiency further underlines the role of hepatic dysfunction towards decreased vitamin serum levels as this metabolic step in the course of VitD synthesis is localized in the liver [10].

Finally, we observed that patients with low VitA levels were significantly younger than patients with high VitA levels, which was also confirmed on multivariate regression analysis. At this point, we can only speculate that patients undergoing HVPG measurement (i.e., had developed ACLD) at a younger age had a more progressive course of the disease, which might be related to pronounced VitA depletion.

Correction of VitA levels in ACLD, however, is challenging: First, simple supplementation might not lead to effective amelioration of VitA serum levels, depending on the severity of liver disease [25, 26]. Second, it remains unclear whether VitA_Def (as assessed by serum VitA levels) really impacts on downstream signaling relevant for modulation of liver disease. Lastly, overdosing of VitA may cause hepatotoxicity which may be of particular concern in patients with ACLD [27]. However, the strong link between VitA_Def and fibrosis may indicate that treatment strategies promoting fibrosis regression will also ameliorate VitA homeostasis and vice versa.

VitD_Def was highly common in our study cohort, being present in 57% of patients and additional 26% with VitD insufficiency, according to commonly used cut-offs for 25-OH-vitamin D levels [10]. Of note, comparisons to other studies regarding the prevalence of VitD_Def are difficult to draw, as there are considerable differences in cut-offs and patient characteristics, however, other studies have demonstrated similar results [11, 12]. As indicated previously, a recent meta-analysis confirmed the association between VitD deficiency and mortality in patients with cirrhosis [13].

In our study, VitD levels were clearly associated with disease stage and severity of PHT. As mentioned above, only severity of hepatic dysfunction (i.e. CTP score) remained an independent risk factor for VitD_Def, which seems particularly interesting in regard to the observation that elevated BA levels were independently associated with VitA_Def but not with VitD_Def. While VitA supply is exclusively dependent on dietary uptake (mediated by BA [1]), VitD is mainly provided by endogenous synthesis/metabolization [10]. Therefore, the identification of hepatic (dys-)function – but not elevated BA levels – as the main determinant for 25-OH-vitamin D deficiency seems consistent with physiological concepts of VitD metabolism.

Importantly, a previous RCT by Pilz et al. demonstrated that VitD supplementation effectively increases VitD serum levels in patients with cirrhosis [28], while another study reported rapid declines of VitD levels after cessation of supplementation [29]. Furthermore, the efficacy of VitD supplementation towards osteopenia/osteoporosis remains unclear, as parathyroid hormone levels were not affected by VitD supplementation in the study by Pilz et al [28] and since several studies displayed no benefits of VitD supplementation on bone density/osteoporosis in patients with liver disease [30‐32]. Similarly, a systematic review on VitD supplementation in chronic liver disease suggested limited efficacy of VitD supplementation on mortality, but importantly no effects on morbidity and health-related quality of life [33]. Nevertheless, in the absence of significant toxicity related to VitD supplementation in patients with ACLD, VitD supplementation seems safe, and thus, it may be argued that ACLD patients with VitD deficiency should still receive VitD supplements in case of low VitD levels.

Interestingly, our study clearly refutes that VitE_Def is a regular or relevant condition in ACLD. First of all, only 7 (3%) patients fulfilled laboratory criteria for VitE_Def. This observation is particularly relevant in regard to previous studies reporting lower levels of VitE in patients with NASH as compared to healthy subjects [34]. Of note, the study by Erhardt et al. [34] reported mean VitE levels of 22.4 µmol/L and 26.8 µmol/L in NASH patients and controls, respectively, which was quite similar to our study (median 23.6 µmol/L). Therefore, we can only speculate whether “significantly” lower VitE serum levels – however, still within normal range between healthy controls and patients have an impact on disease or may even be classified as VitE_Def based on their potential implications on ACLD. Nevertheless, RCT displayed a significantly higher rate of histological improvement upon VitE treatment as compared to placebo in NASH patients [35], which was also suggested by a meta-analysis on VitE supplementation in NASH [36]. Similarly, a recent publication reported a remarkable reduction of risk for death or liver transplantation in compensated NASH patients receiving VitE supplements (adjusted HR 0.30) [37]. However, next to the retrospective and non-randomized study design, adherence to VitE intake in the treatment group was not assessed [37]. In any case, we did not observe any differences in VitE levels across CTP or HVPG strata or any meaningful correlation with parameters of hepatic (dys-)function. Of note, our study only included 25 (11%) NASH patients that exclusively displayed VitE levels above the cut-off for VitE_Def.

Our study has several limitations: First, we could not provide data on hepatic VitA content, as well as a profile of different VitA metabolites in serum. Second, specific binding proteins for these vitamins were not measured, which prevents us from reporting their relation to vitamin serum levels. Third, we did not perform a systematic assessment of dietary vitamin intake, which would include both regular food intake as well as over-the-counter vitamin supplements, as well as current adherence towards and prior duration of prescription vitamin supplement intake. Fourth, the limited duration of potential follow-up in our study cohort prevents us from reporting follow-up events such as mortality and (further) hepatic decompensation in these patients at this point, which would have allowed to assess the prognostic value of VitA/D/E deficiencies. While these limitations must be acknowledged, we provide novel and important data on the strong association between vitamin deficiencies and ACLD severity as well as the severity of PHT in a large series of prospectively recruited patients. Furthermore, this study displays a close link between VitA levels and ELF score, as well as an independent association between elevated BA levels and VitA_Def.

In summary, our study demonstrates a considerable prevalence of VitA and VitD deficiency in patients with ACLD, while VitE deficiency was uncommon. VitA and VitD deficiencies are closely linked to hepatic dysfunction, which may be explained by advanced fibrosis and/or synthetic dysfunction impacting on the uptake, storage, and metabolism of these vitamins. The prognostic value of VitA as well as its role in liver disease progression should be assessed in future studies.