Plasma Retinol Concentration Is Mainly Driven by Transthyretin in Hemodialysis Patients

Background Micronutrients deficiencies in hemodialysis patients are due to low dietary intakes and intradialytic losses for hydrophilic micronutrients. Conversely, lipophilic non-dialyzable compounds might accumulate due to a lack of elimination through renal metabolism or dialysis. Other compounds have complex metabolism: their concentration is not explained by these phenomenons. Study design Monocentric observational longitudinal study Subjects 123 hemodialysis patients Main outcome measure Plasma concentration of lipophilic micronutrients: retinol and its two cotransporters transthyretin and retinol binding protein 4, tocopherol, and carotenoids (αand β-carotene, βcryptoxanthin, lycopene, lutein, zeaxanthin) and all factors associated with one-year mortality Results Within the 123 patients of the study, median age [IQR] was 77.5[69.5–84.5] years and 58.5% were male. Median retinol plasma concentration was 4.07[2.65-5.51]μmol/L, and 91.9% of patient had high plasma retinol concentrations. In monovariate analysis, retinol levels were inversely correlated with mortality (HR=0.57[0.45-0.72]; p<0.001). This effect remained significant after adjustment with several parameters. Nevertheless, the correlation between retinol and mortality disappeared as soon as transthyretin was added in the statistical model, suggesting an effect of transthyretin as confusing bias. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16


Introduction
Protein energy wasting syndrome affects 30 to 60% of hemodialysis patients (HD) and contributes to high morbidity and mortality rates in this specific population [1]. Yet, in daily practices, very little concern is made about micronutrients (i.e. vitamins and trace elements) which are implicated in many metabolic functions including regulation of oxidative stress, or modulation of the immune system [2].
Micronutrients deficiencies in HD patients are due to low dietary intakes, but also from intradialytic losses for small hydrophilic micronutrients, and conversely micronutrients accumulation, to a lack of elimination through renal metabolism or dialysis techniques for some lipophilic or non-dialyzable compounds [2,3]. Nevertheless, some compounds have more complex metabolism and their plasma concentration is not only explained by these simple phenomenons.
In HD patients, plasma retinol levels have been described to be inversely related to mortality rate in two observational prospective studies [5,6]. However, pathophysiological explanations for the patients' benefit of survival in case of high concentrations of plasma retinol are not clear. In another study, high tocopherol levels were associated with a better survival, but this effect disappeared with adjustment with other parameters [7]. To our best knowledge, no 41 Plasma retinol in hemodialysis patients study has assessed the link between carotenoids plasma concentration and survival in HD patients.
In this study, we aimed 1/ to assess plasma concentrations of lipophilic micronutrients: retinol (and its physiological partners: retinol binding protein 4 (RBP4) and transthyretin), tocopherol, and carotenoids; and 2/ to analyze if these plasma concentrations are predictive of subsequent mortality in HD patients.

Subjects and study design
We conducted a retrospective longitudinal study on all patients from our HD center from July 2014 to July 2015. All patients with data available regarding micronutrients were included, except for pregnant women or patients aged <18 years. Written information was provided to all patients, and all gave consent for their personal data to be used for research purposes.
According to French law, it is neither necessary nor possible to obtain approval from an

Clinical, biological, and hemodialysis parameters
The following data were collected from the patients' medical files: age, gender, diabetes mellitus, nephropathy, height, post-dialysis weight, vascular access. Evaluation of daily urine output was based on oral questioning of the patients and was therefore semi-quantitative: ≥500 mL/d or <500 mL/d. Body mass index (BMI) was calculated as post-dialysis weight after the mid-week session in kilograms divided by squared height in meters. Survival was assessed for all patients at one year.
Dialysis parameters were recorded at the mid-week session, and biological analyses were all performed at the start of this hemodialysis session. Dialysis dose was estimated by a singlepool Kt/V (spKt/V), as recommended by Daugirdas et al. [8]. The ESRD adapted Charlson Biological data Albumin, transthyretin, predialysis creatinine and urea as well as C-reactive protein which are performed in routine practice in our hemodialysis center were recorded from patient's medical files. Transthyretin was dosed using an immunoturbidimetry assay (ADVIA® 1800 Clinical Chemistry System, Siemens, France). Albumin was dosed in serum using bromocresol green assay (ADVIA® 1800 Clinical Chemistry System, Siemens, France).
RBP4 quantification was performed using dedicated ELISA kits (Quantikine Elisa, R&D systems, France) according to the manufacturer's instructions.

Statistical analyses
Kaplan Meier tests were performed to assess association between transthyretin and mortality, as well as retinol and mortality. Univariate linear models were used to assess association between retinol and transthyretin, retinol and RBP4, and predialysis serum creatinine and RBP4. Cox models were used to determine factors associated with mortality. In a first step, variables with a statistical p-value of <0.10 in the univariate analysis were considered eligible for inclusion in the multivariate analysis. In a second step, using a descending stepwise   Figure 1). This effect remained significant after adjustment with several parameters (

Discussion
In this study, we confirmed that hemodialysis patients display high plasma retinol and RBP4concentrations, and that the highest concentrations of retinol were correlated with an improved survival in monovariate analysis. A new insight of our study was that this correlation disappears when retinol concentration was adjusted with transthyretin concentration, which suggests that high retinol plasma concentration is only a surrogate marker of higher concentration of transthyretin, which reflects a good nutritional status of hemodialysis patients.
To understand this new insight, one must be familiar with retinol physiology. Dietary sources of vitamin A include its provitamin, beta-carotene which is found in vegetable sources, and its active forms -retinol, retinal and retinoic acid-which are found in animal sources. Retinol storage is mainly in the liver stellar cells as retinyl esters. In the plasma, retinol which is a lipophilic compound is linked to a transport complex consisting of RBP4 and transthyretin -also known as prealbumin -forming a complex characterized by an equimolecular ratio of 1:1:1 [11].
RBP4 is a single polypeptide chain protein, synthesized in the liver, which binds and transports retinol to the target organs. It is a strong signal for stellar cells to remove retinol into the plasma and therefore, retinol plasma concentration is highly regulated by RBP4 [12].
During renal failure, RBP4, which has a renal degradation, accumulates. The high concentration of retinol is mainly due to RBP4 accumulation which is a signal for liver cells to remove retinol in the plasma. In our study, median retinol plasma concentration was 4 times higher than normal values, which was in accordance with previous literature reporting retinol concentrations to be 3 to 4 times higher in HD patients than in healthy subjects [13][14][15].
RBP4 is a 21 kDa molecule and is thus considered as a middle size (>500Da) protein bound uremic toxin [16,17]. In hemodialysis patients, convective therapy using online In HD patients, transthyretin concentration is representative of the nutritional status, exhibiting significant relationships with energy and protein intake as well as with fat stores and lean body mass. In addition to its relationship with nutritional status, transthyretin is involved in the inflammatory response and its serum concentration is negatively correlated with inflammatory markers. Serum transthyretin concentrations lower than 300 mg/l were associated with an increased risk of morbidity and mortality independently from serum albumin [11].
In malnourished HD patients, the serum transthyretin/RBP4 ratio is thus highly modified because of a high RBP4 concentration and low transthyretin concentration [13]. We show that in this specific population, retinol is better correlated with transthyretin (r²=0.72; p<0.001) than RBP4 (r²=0.26; p<0.001). Thus, in patients with a low transthyretin and a high RBP4 level, transthyretin, which is necessary to remove retinol from its storage in liver cells, could be the limiting factor for this removal [19]. Another explanation could be that transthyretin which reflects the daily dietary intakes is correlated with vitamin A intakes [3], but this latter is not likely because in plasma, retinol concentration is high. vitamin A stores which are mainly located in the liver. Interestingly, vitamin A accumulation hepatotoxicity might occur with a normal retinol plasma level [20] and in patients with hepatitis C virus, liver fibrosis has been associated with retinol plasma levels increase but lower liver storage of vitamin A [21]. In this latter study, no correlation was found between hepatic tissue retinyl palmitate-a retinyl ester-and plasma retinol concentration. Thus, to assess retinol whole body stores, a liver biopsy analysis is required [21].
The link between retinol concentration and mortality reflects only the relationship between transthyretin and mortality. In two previous studies, retinol plasma concentration was correlated with death, even after adjustment with many parameters including nutritional status and RBP4, but these studies did not analyze transthyretin levels which is a key element in retinol physiology [5,6]. The link between retinol plasma concentration and death is not likely because retinol plasma concentration couldincorrectly reflect the vitamin A status.
Moreover, these two studies hypothesize a role of low retinol in a high infection rate in HD patients, but one could argue an opposite relationship in which infection induces inflammation and reduces transthyretin concentration and thus retinol concentration [6].
Still, the lack of effect of high retinol concentrations on mortality should not be interpreted as if this substance has no effect. Roehrs et al. reported a study in which higher retinol concentrations were associated with higher superoxide dismutase and catalase activities, but these activities did not prevent lipid peroxidation, hypothesizing a pro-oxidant role of high retinol concentrations [15].
Regarding vitamin E, our findings are very similar to what has been reported in literature [7]: plasma concentration of tocopherol and tocopherol/TG are correlated with poor outcomes, but this effect does not resist to adjustment with other parameters and thus, plasma tocopherol concentration is only a surrogate marker of the measured outcome, i.e. death. Identically, plasma carotenoids levels were not independently associated with mortality.

Plasma retinol in hemodialysis patients
We agree this study suffers some limitations: retrospective analysis, small size and old age of the studied population, monocentric design, lack of follow-up. Our results require confirmation in an large independent cohort of patients. Nevertheless, the strength of the correlation between plasma retinol and transthyretin concentrations is striking.
In conclusion, our study reports that in HD patients with low transthyretin, retinol plasma concentration could might not represent vitamin A levels, but likely the nutritional status of the patients. Its correlation with mortality is could be more linked to the nutritional status rather than a direct biological effect of retinol.