Impaired Vitamin D Metabolism in Hospitalized COVID-19 Patients

There is increasing data regarding the association between vitamin D and COVID-19. This study aimed to reveal the alterations of vitamin D metabolism in the setting of COVID-19. We examined 119 adult COVID-19 inpatients and 44 apparently healthy adult individuals with similar serum 25OH-D3 levels as a reference group. The assessment included serum biochemical parameters (total calcium, albumin, phosphorus, creatinine), parathyroid hormone (PTH), vitamin D-binding protein (DBP), vitamin D metabolites (25OH-D3, 25OH-D2, 1,25(OH)2D3, 3-epi-25OH-D3, 24,25(OH)2D3 and D3) and free 25OH-D. COVID-19 patients had in general very low vitamin D levels (median 25OH-D3 equals 10.8 ng/mL), accompanied by an increased production of the active vitamin D metabolite (1,25(OH)2D3), estimated as higher 1,25(OH)2D3 serum levels (61 [44; 81] vs. 40 [35; 50] pg/mL, p < 0.001) and lower 25OH-D3/1,25(OH)2D3 ratio (175 [112; 260] vs. 272 [200; 433], p < 0.001) which is presumably aimed at preventing hypocalcemia. Patients with COVID-19 also had elevated DBP (450 [386; 515] vs. 392 [311; 433] mg/L, p < 0.001) and low free 25OH-D levels (<LoB vs. 3.9 [3.2; 4.4] pg/mL, p < 0.001). Follow-up assessment of the COVID-19 inpatients showed recovery of the observed changes. Overall, hospitalized patients with an acute course of COVID-19 have not only very low levels of 25OH-D but also profound abnormalities in the metabolism of vitamin D regardless of the clinical course of the disease. These alterations might exacerbate existing vitamin D deficiency and its negative impact.


Introduction
COVID-19 pandemic remains a major worldwide public health threat. COVID-19 is an infectious disease with predominant lung involvement caused by a highly infectious novel coronavirus known as severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) [1]. Vitamin D has emerged as a factor that may be involved in susceptibility to the disease and is considered a potential therapeutic option.
The three main steps in vitamin D metabolism, 25-hydroxylation, 1α-hydroxylation, and 24-hydroxylation, are all performed by cytochrome P450 mixed-function oxidases (CYPs) (Figure 1). To obtain biological activity, vitamin D undergoes two subsequent hydroxylation reactions-first to 25-hydroxyvitamin D (25OH-D), then to the active form 1,25dihydroxyvitamin D (1,25(OH) 2 D). CYP2R1 is the most important 25-hydroxylase; CYP27B1 is the key 1-hydroxylase. Both 25OH-D and 1,25(OH) 2 D are catabolized by CYP24A1. Isomerization of the C-3 hydroxy group from the α to β orientation by 3-epimerase does not restrict the action of CYP27B1 or CYP24A1 but might reduce biologic effects [2]. Airway epithelia, alveolar macrophages, and dendritic cells all express 1α-hydroxylase and therefore have the potential to locally synthesize the active form of vitamin D (1,25(OH)2D) from its precursor (25OH-D) [3]. Although not demonstrated for coronaviruses such as SARS-CoV-2, for other viruses and other respiratory pathogens, activation of the innate immunity leading to increased local production of 1,25(OH)2D has been shown to enhance viral neutralization and clearance while modulating the subsequent proinflammatory response [4].
The effect of vitamin D on acquired immunity is considered predominantly in the context of its ability to influence the proliferation and phenotype of T cells [5]. In vitro 1,25(OH)2D modifies the cytokine profile of T-lymphocyte into an overall anti-inflammatory humoral immune response. Vitamin D may also stimulate the generation of the regulatory T cells, which are considered crucial for the induction of immune tolerance. These effects are suggestive of 1,25(OH)2D significance in preventing the cytokine storm associated with the severe respiratory disease caused by viral infections.
There is an increasing scientific interest in the possible association between vitamin D status and COVID-19 infection. According to the available evidence to date, most of the studies indicate a significant relationship between low 25OH-D levels and risk of SARS-CoV-2 infection [6][7][8][9][10][11][12][13], COVID-19 severity, and mortality [10,11,[13][14][15], having considerable heterogeneity in methodological and statistical approach. The available evidence regarding the effectiveness of vitamin D in the treatment of COVID-19 comes predominantly from observational data and lacks thorough verification in clinical trials as the results of available meta-analyses are conflicting [16][17][18][19][20][21][22][23][24]. Vitamin D supplementation might be associated with improved clinical outcomes in COVID-19 patients, such as transfer to intensive care unit [16,19,20] and mortality rates [25]. Meta-analyses which included only randomized clinical trials and quasi-experimental studies showed uncertain or no evidence for the effectiveness of vitamin D supplementation in the treatment of COVID-19 [21,22]. However, vitamin D supplementation was significantly associated with reduced intensive care unit admission and mortality when administered after the diagnosis of COVID-19, according to the subgroup analysis [23]. One systematic review focused on patients aged 60 years or over demonstrated better primary clinical outcomes of COVID- Airway epithelia, alveolar macrophages, and dendritic cells all express 1α-hydroxylase and therefore have the potential to locally synthesize the active form of vitamin D (1,25(OH) 2 D) from its precursor (25OH-D) [3]. Although not demonstrated for coronaviruses such as SARS-CoV-2, for other viruses and other respiratory pathogens, activation of the innate immunity leading to increased local production of 1,25(OH) 2 D has been shown to enhance viral neutralization and clearance while modulating the subsequent proinflammatory response [4].
The effect of vitamin D on acquired immunity is considered predominantly in the context of its ability to influence the proliferation and phenotype of T cells [5]. In vitro 1,25(OH) 2 D modifies the cytokine profile of T-lymphocyte into an overall anti-inflammatory humoral immune response. Vitamin D may also stimulate the generation of the regulatory T cells, which are considered crucial for the induction of immune tolerance. These effects are suggestive of 1,25(OH) 2 D significance in preventing the cytokine storm associated with the severe respiratory disease caused by viral infections.
There is an increasing scientific interest in the possible association between vitamin D status and COVID-19 infection. According to the available evidence to date, most of the studies indicate a significant relationship between low 25OH-D levels and risk of SARS-CoV-2 infection [6][7][8][9][10][11][12][13], COVID-19 severity, and mortality [10,11,[13][14][15], having considerable heterogeneity in methodological and statistical approach. The available evidence regarding the effectiveness of vitamin D in the treatment of COVID-19 comes predominantly from observational data and lacks thorough verification in clinical trials as the results of available meta-analyses are conflicting [16][17][18][19][20][21][22][23][24]. Vitamin D supplementation might be associated with improved clinical outcomes in COVID-19 patients, such as transfer to intensive care unit [16,19,20] and mortality rates [25]. Meta-analyses which included only randomized clinical trials and quasi-experimental studies showed uncertain or no evidence for the effectiveness of vitamin D supplementation in the treatment of COVID-19 [21,22]. However, vitamin D supplementation was significantly associated with reduced intensive care unit admission and mortality when administered after the diagnosis of COVID-19, according to the subgroup analysis [23]. One systematic review focused on patients aged 60 years or over demonstrated better primary clinical outcomes of COVID-19 in patients with vitamin D supplementation [24]. In a hypothesis-generating study by Giannini et al., comorbidity burden significantly modified the effect of vitamin D treatment with amplification of positive effect on clinical outcomes [26].
Overall, issues regarding the appropriate dose, duration, and mode of administration of vitamin D in COVID-19, as well as identifying the patient group with the greatest possible benefit, need further research. A comprehensive assessment of vitamin D metabolism during the course of COVID-19 could provide additional information for planning further clinical trials. Simultaneous measurement of 25OH-D and 24,25(OH) 2 D with the calculation of 25OH-D/24,25(OH) 2 D ratio has recently emerged as a valuable new diagnostic tool in the differential diagnosis of hypercalcemia particularly for screening of 24-hydroxylase deficiency [27][28][29][30][31]. Assessment of vitamin D metabolite profile and calculation of metabolites ratios is also a promising direction for better characterization of vitamin D status [29,[32][33][34]; moreover, 25OH-D/24,25(OH) 2 D ratio has recently been shown to be associated with important clinical outcomes [35,36]. Tang et al. showed that the production of serum 1,25(OH) 2 D is favored over 24,25(OH) 2 D in vitamin D-insufficient young, healthy adults as the availability of vitamin D precursors in circulation diminishes [37]. A single retrospective study showed no association between parameters of vitamin D catabolism and fatal outcomes or the need for respiratory support in COVID-19 patients [38]. However, to the authors' knowledge, there are no data regarding the activity of 1α-hydroxylase and production of active vitamin D metabolite in COVID-19, which might be of particular interest since promising clinical data was obtained with the administration of active vitamin D metabolites in COVID-19 patients [39].
This study aimed to reveal alterations in vitamin D metabolism in the setting of the COVID-19 course.

Results
COVID-19 group was presented by patients of various ages with a median age of 61 years, equally men and women (p > 0.05) ( Table 1). Most of the patients with COVID-19 were overweight or obese; 17 patients (14%) had diabetes mellitus. A detailed description of the condition on admittance and the course of inpatient treatment is presented in Table 2; most patients were characterized by a moderately severe course of infection. A total of 34 patients (29%) were characterized by a severe course of the disease, including 21 patients (18%) who were prescribed immunobiological treatment, 11 patients (9%) admitted or transferred to the intensive care unit, 10 patients (8%) with a fatal outcome.  The patients from the COVID-19 group had lower total calcium levels (p < 0.001) and serum albumin levels (p < 0.001), while levels of albumin-adjusted calcium were similar to the reference group (p = 0.06) ( Table 3). Secondary hyperparathyroidism was revealed in 15 patients (13%) with COVID-19 and in 7 individuals (16%) from the reference group (p = 0.61), one COVID-19 patient had primary hyperparathyroidism. PTH levels were equal between the groups (p = 0.65). We also observed high creatinine levels in COVID-19 patients (p < 0.001), indicative of the decrease in kidney function and lower phosphorus levels (p = 0.04). The detailed data on vitamin D metabolites, free 25OH-D, and DBP measurement in the COVID-19 group and the reference group with similar 25OH-D 3 levels (p = 0.88) are presented in Table 4. Only 3 patients (3%) with COVID-19 had sufficient vitamin D levels, according to the Endocrine Society and the Russian Association of Endocrinologists guidelines (≥30 ng/mL [40,41]). COVID-19 patients had not only higher serum levels of 1,25(OH) 2 D 3 (p < 0.001), but also lower 25OH-D 3 /1,25(OH) 2 D 3 ratios (p < 0.001) implying increased production of the active metabolite (1,25(OH) 2 D 3 ). They also had higher 25OH-D 3 /24,25(OH) 2 D 3 ratios (p = 0.001). The rest of the studied metabolites did not differ significantly from the reference group. The levels of 25OH-D 2 did not exceed 0.5 ng/mL in all examined individuals and detectable levels of vitamin D 3 were observed only in 10 patients (8%) from the COVID-19 group (maximum 19.7 ng/mL) and in 7 individuals (16%) from the reference group (maximum 22.6 ng/mL). In addition, the COVID-19 group had significantly higher DBP levels (p < 0.001) and very low (mostly lower than the limit of blank) levels of free 25OH-D (p < 0.001) against the reference group. Follow-up samples with median follow-up period 11 [9; 14] days were available in 62 patients (52%) from the COVID-19 group, of which 20 patients (32%) were characterized by severe course of the disease, defined as prescription of immunobiological treatment (n = 12), admission or transfer to the intensive care unit (n = 7) or fatal outcome (n = 5). We observed clinically insignificant change in 25OH-D 3 levels (9. Patients with a severe course of COVID-19 showed no significant deviations in the studied laboratory parameters versus the main body of patients with a moderately severe course of the disease, except for a lower baseline total calcium (p = 0.04) ( Table 5).  Patients with a severe course of COVID-19 showed no significant deviations in the studied laboratory parameters versus the main body of patients with a moderately severe course of the disease, except for a lower baseline total calcium (p = 0.04) ( Table 5).  1 The first value corresponds to the baseline examination; the second value corresponds to the follow-up examination. 2 Ranges are specified according to kit manufacturers' recommendations. 3 Range is given for a total of 25OH-D according to the clinical guidelines [40,41]; the 25OH-D 2 fraction is negligible for the purposes of this study. 4 Ranges are given according to the literature data [42,43].

Discussion
First of all, our study further supported the prevalence of very low vitamin D levels in patients with COVID-19. The median level of 25OH-D 3 in COVID-19 patients was only 10.8 ng/mL, which corresponds to a pronounced vitamin D deficiency in a major part of the examined patients [41] and depicts worse vitamin D status than epidemiological data from a similar season [44]. Similar results have been demonstrated in a number of previous studies from various countries [45], and the amount of evidence continues to increase [46]. The obtained results raise awareness of vitamin D supplementation in COVID-19; however, further clinical trials are required to produce detailed guidelines. As the effect of vitamin D supplementation in COVID-19 was shown to be modified by several factors, including the timing of therapy initiation [23], age [24], and comorbidity burden [26], we believe that prospective trials should be aimed at assessing the effectiveness of vitamin D therapy in vitamin D supplementation-naive patients and prioritize older and comorbid groups.
Looking deeper than the traditional 25OH-D 3 levels, the first major finding was increased levels of active vitamin D metabolite (1,25(OH) 2 D 3 ) in COVID-19 patients compared to the reference group, with a recovery observed alongside the course of the disease. The majority of COVID-19 patients were of older age, had higher creatinine levels and BMI compared to the reference group, and the association with a decrease -instead of an increase -of circulating active vitamin D metabolite was shown for these conditions [47][48][49]; hence one cannot explain the observed difference from this standpoint. The observed higher levels of 1,25(OH) 2 D 3 cannot be fully explained by the lower activity of 24-hydroxylase due to the vitamin D deficiency either, since the production of 24,25(OH) 2 D 3 appeared to be stable during the follow-up, unlike the levels of 1,25(OH) 2 D 3 . We can hypothesize that an increase in extrarenal 1α-hydroxylase activity might be a contributing factor and hereby question the generally accepted speculation that in the setting of vitamin D defi- ciency, lesser quantities of 25OH-D are available for the synthesis of 1,25(OH) 2 D, which might cause impairment of the innate immune response [50,51]. Given that, the localized intracrine mechanism is now considered a cornerstone of the interaction between vitamin D and the immune system [52]. Apart from that, previously reported studies on patients with sepsis showed a decrease in circulating 1,25(OH) 2 D 3 levels [53] and the correlation of low 1,25(OH) 2 D 3 levels with poor survival [54], so the observed findings may be specific to COVID-19. The proposed mechanism of vitamin D dysregulation among COVID-19 patients is illustrated in Figure 3. The second significant finding was an increase in DBP levels and a concord crease in free 25OH-D in patients with COVID-19 in the acute phase of the diseas is partially consistent with recent literature data [55] and indicates the presence o phase protein properties in DBP. This feature has been described previously in acute conditions, in particular, sepsis [56] and minor traumas [57].
Hypocalcemia in critically ill patients is universally known [58], and a high lence of low serum calcium levels are also described in hospitalized patients with 19 [59]. The observed increase in circulating active vitamin D metabolite could b aimed at preventing hypocalcemia in COVID-19 patients with low total and free levels. Recent data on impaired compensatory PTH response in hypocalcemic with COVID-19, obtained by other research groups [60,61], support this hypoth may be due to negative feedback from 1,25(OH)2D3, although in our cohort, PT were similar to the reference group.
Our study had several major limitations. First of all, individuals from the r group were matched by the level of 25OH-D3, which allowed us to separate the underlying vitamin D deficiency [32,37], but significantly differed in age, sex ra BMI, and therefore it is not possible to separate the contribution of these factor The second significant finding was an increase in DBP levels and a concordant decrease in free 25OH-D in patients with COVID-19 in the acute phase of the disease, which is partially consistent with recent literature data [55] and indicates the presence of acutephase protein properties in DBP. This feature has been described previously in various acute conditions, in particular, sepsis [56] and minor traumas [57].
Hypocalcemia in critically ill patients is universally known [58], and a high prevalence of low serum calcium levels are also described in hospitalized patients with COVID-19 [59]. The observed increase in circulating active vitamin D metabolite could be partly aimed at preventing hypocalcemia in COVID-19 patients with low total and free 25OH-D levels. Recent data on impaired compensatory PTH response in hypocalcemic patients with COVID-19, obtained by other research groups [60,61], support this hypothesis and may be due to negative feedback from 1,25(OH) 2 D 3 , although in our cohort, PTH levels were similar to the reference group.
Our study had several major limitations. First of all, individuals from the reference group were matched by the level of 25OH-D 3 , which allowed us to separate the effect of underlying vitamin D deficiency [32,37], but significantly differed in age, sex ratio, and BMI, and therefore it is not possible to separate the contribution of these factors within the framework of this work, although the authors consider such an impact as unlikely, taking into account the above-discussed data on the influence of these factors [47][48][49]. Also, the reference group was much smaller than the patients' group, which requires a very cautious interpretation of the data obtained. Recruiting a matched control group was complicated due to the pandemic restrictions, especially for the elderly and comorbid individuals. Secondly, our study only allows us to make an assumption about the altered activity of vitamin D metabolism enzymes as the surrogate parameters were investigated and direct activity was not actually measured, as well as additional factors involved in the regulation of vitamin D metabolism (for instance, fibroblast growth factor-23) were lacking. This seems particularly important given that knowledge of the enzymes involved in vitamin D metabolism has increased in recent years. For example, other hydroxylases besides CYP2R1 (such as CYP3A4, CYP27A1, CYP2C11, and CYP2J2/3) have been shown to have a 25-hydroxylase activity that might affect 25OH-D 3 levels [2], while CYP2C11 and CYP3A4 demonstrated 24-hydroxylase activity [62,63]; however, current literature data suggest that CYP24A1 remains the only established 24-hydroxylase involved in vitamin D metabolism [64]. The dietary habits of the patients were not taken into account, so that the low phosphorus dietary intake could have some contribution to the increase in 1,25(OH) 2 D 3 production in COVID-19 patients [47]. Finally, the reduced amount of followup data attenuates the results of this part of the study. However, the study had a number of strengths: the substantial number of patients included, the wide range of parameters assessed for vitamin D metabolism and the high-quality method validated by an external control scheme used for measurement of vitamin D metabolites.

Study Population and Design
This was a retrospective cohort study based on a single-center experience. The main cohort included 119 adult patients hospitalized at the COVID-19 patient care hospital based in Moscow, Russia, for inpatient treatment during the period from 5 May to 4 June 2020. Diagnosis and inpatient treatment of COVID-19 patients was performed according to temporary federal guidelines [65]. COVID-19 was diagnosed with at least one of the following features in the presence of clinical findings suggestive of COVID-19: positive PCR nasal swab, positive PCR throat swab, a characteristic pattern of changes on chest CT. The severe course of the disease was defined within the framework of the current study as at least one of the following features: prescription of immunobiological treatment, either admission or transfer to the intensive care unit, fatal outcome. Recruitment of patients was carried out throughout the entire period of operation of the COVID-19 patient care hospital in a continuous manner, except for patients with insufficient residual serum for the planned laboratory measurements.
To exclude the direct impact of vitamin D deficiency on vitamin D metabolism, a reference group with similar to the COVID-19 patients' serum 25OH-D 3 levels were recruited. The reference group was comprised of 44 otherwise healthy adult individuals from the hospital staff who did not receive vitamin D supplementation for at least 3 months preceding the participation in the study.
Anthropometric and clinical data were collected from the patients' records. The study was approved by the Ethics Committee of the Endocrinology Research Centre, Moscow, Russia, on 30 April 2020 (abstract of record No. 6). All patients signed informed consent to participate in the study. The assessment included serum vitamin D metabolites (25OH-D3, 25OH-D2, 1,25(OH)2D3, 3-epi-25OH-D3, 24,25(OH)2D3 and D3), biochemical parameters (total calcium, albumin, phosphorus, creatinine), parathyroid hormone (PTH), vitamin D-binding protein (DBP) and free 25OH-D. Follow-up reassessment of COVID-19 patients further in the course of inpatient treatment included vitamin D metabolites, DBP, and free 25OH-D ( Figure 4). Anthropometric and clinical data were collected from the patients' records.

Laboratory Measurements
After the completion of all routine laboratory testing, residual serum was stored at −80 • C until batched analysis, avoiding repeated freeze-thaw cycles. Biochemical parameters of blood serum were evaluated using the ARCHITECT c8000 analyzer (Abbott, Chicago, IL, USA) according to the standard methods using reagents from the same manufacturer. PTH levels were assessed by the electrochemiluminescence immunoassay (ELECSYS, Roche, Basel, Switzerland). Serum DBP and free 25OH-D levels were measured by enzyme-linked immunosorbent assay (ELISA) using commercial kits (Immundiagnostik AG, Bensheim, Germany, and DIAsource, ImmunoAssays S.A., Louvain-la-Neuve, Belgium respectively). The assay used for DBP levels assessment (Immundiagnostik AG, Bensheim, Germany) has a 3.3% intra-and inter-assay coefficient of variation (CV), the limit of blank is equal 0.154 mg/L, and the limit of detection is equal 0.944 mg/L. The assay used for free 25(OH)D levels assessment (DIAsource, ImmunoAssays S.A., Belgium) has 4.9% intraassay CV and 5.9 inter-assay CV; the limit of blank is equal to 1.5 pg/mL and the limit of detection is equal 2.4 pg/mL. The limit of blank is defined as the highest measurement result that is likely to be observed (with a stated probability) for a blank sample, the limit of detection is defined as the lowest amount of analyte in a sample which can be detected but not necessarily quantitated as an exact value.
The serum levels of vitamin D metabolites (25OH-D 3 , 25OH-D 2 , 1,25(OH) 2 D 3 , 3-epi-25OH-D 3 , 24,25(OH) 2 D 3, and D 3 ) were determined by ultra-high performance liquid chromatography in combination with tandem mass spectrometry (UPLC-MS/MS) using an in-house developed method, described earlier [66]. With this technique, the laboratory participates in the DEQAS quality assurance program (lab code 2388) and the results fall within the target range for the analysis of 25OH-D and 1,25(OH) 2 D metabolites in human serum (Supporting Information, Figures S1 and S2).

Statistical Analysis
Statistical analysis was performed using Statistica version 10.0 (StatSoft, Tulsa, OK, USA). All data were analyzed with non-parametric statistics and expressed as median [interquartile range], absolute values (percentages), or ratios. Mann-Whitney U-test and Fisher's exact two-tailed test were used for comparisons between the two groups. Wilcoxon test was performed to evaluate changes in indices during follow-up. A p-value of less than 0.05 was considered statistically significant.

Conclusions
Taking into account all the limitations, this is the first study to our knowledge aimed at the comprehensive evaluation of vitamin D metabolism in COVID-19. Our data suggest that hospitalized patients with an acute course of COVID-19 have profound abnormalities in the metabolism of vitamin D apart from very low levels of 25OH-D (in particular, an increase in circulating active vitamin D metabolite and an increase in DBP levels) regardless of the clinical course of the disease. These alterations might likely exacerbate existing vitamin D deficiency and its negative impact.