Association Between Vitamin D and Uric Acid in Adults: A Systematic Review and Meta-Analysis.

Association between vitamin D and uric acid is complex and might be bidirectional. Our study aimed to determine the bidirectional association between vitamin D and uric acid in adults. Using MEDLINE via PubMed and Scopus, we systematically searched for observational or interventional studies in adults, which assessed the association between serum vitamin D and serum uric acid, extracted the data, and conducted analysis by direct and network meta-analysis. The present review included 32 studies, of which 21 had vitamin D as outcome and 11 had uric acid as outcome. Meta-analysis showed a significant pooled beta coefficient of serum uric acid level on serum 25(OH)D level from 3 studies of 0.512 (95% confidence interval: 0.199, 0.825) and a significant pooled odds ratio between vitamin D deficiency and hyperuricemia of 1.496 (1.141, 1.963). The pooled mean difference of serum 25(OH)D between groups with hyperuricemia and normouricemia was non-significant at 0.138 (-0.430, 0.707) ng/ml, and the pooled mean difference of serum uric acid between categories of 25(OH)D were also non-significant at 0.072 (-0.153, 0.298) mg/dl between deficiency and normal, 0.038 (-0.216, 0.292) mg/dl between insufficiency and normal, and 0.034 (-0.216, 0.283) mg/dl between deficiency and insufficiency. In conclusion, increasing serum uric acid might be associated with increasing 25(OH)D level, while vitamin D deficiency is associated with hyperuricemia. These reverse relationships should be further evaluated in a longitudinal study. The Author(s). This is an open access article published by Thieme under the terms of the Creative Commons Attribution-NonDerivative-NonCommercial-License, permitting copying and reproduction so long as the original work is given appropriate credit. Contents may not be used for commercial purposes, or adapted, remixed, transformed or built upon. (https://creativecommons.org/licenses/by-nc-nd/4.0/).

Introduction

Vitamin D (VIT D) deficiency is still a major public health problem in the world 1 2 3 , with an estimation of over 1 billion people suffering from VIT D insufficiency or deficiency 4 . VIT D deficiency is not only a problem in areas with limited sunlight exposure such as the polar or temperate regions, but also in the tropics 3 . The prevalence of VIT D insufficiency was reported as 33.5–64.6% in Thailand 5 and 77.4% in Brazil 6 , similar to the prevalence of 65% in near-polar Finland 7 .

VIT D deficiency could cause bone-related diseases such as rickets and osteoporosis 8 and may increase the risk for cancer 9 , tuberculosis 10 , and several degenerative diseases such as type 2 diabetes (T2D) 11 , hypertension (HT) 12 , metabolic syndrome 13 , and cardiovascular diseases (CVDs) 14 . VIT D deficiency occurs if there is a decreased intake or synthesis, or increased metabolism or excretion of VIT D. Sources of VIT D are from food and the synthesis by the skin when exposed to sunlight’s ultraviolet B ray while it is minimally excreted in a healthy person 15 16 . Several factors have been reported to be associated with VIT D level, including age, obesity, several diseases such as chronic kidney disease (CKD), liver disease, tuberculosis, sarcoidosis, and malignancies 17 . In addition, genetic factors may also regulate VIT D, for example, the Fok-I polymorphism of VIT D receptor gene 18 and the cytochrome P450 24A1 19 and 2R1 20 21 22 genes.

Previous evidence 23 24 25 26 27 showed that VIT D deficiency might increase the risk of high serum uric acid (SUA). On the other hand, some studies 28 29 30 31 32 showed that high SUA (hyperuricemia) could decrease VIT D level and/or induce VIT D deficiency. Therefore, we conducted a systematic review and meta-analysis to assess whether VIT D deficiency is associated with high SUA or vice versa.

Materials and Methods

The study was conducted after registering with PROSPERO (CRD42018105283). It has no funding source. A literature search was performed in MEDLINE via PubMed and Scopus databases to identify relevant studies published through July 31st, 2018. The search strategy was provided in Supplementary Tables 1S and 2S .

Table 1 Characteristics of included studies.

Author, Year [Ref]	Design	Country	Outcome	Mean age (years)	Mean BMI (kg/m 2 )	Mean serum creatinine (mg/dl)	Mean HbA1c (mmol/mol)	Mean eGFR (ml/min/1.73 m 2 )	Male%	Obese%	Smoking%	T2D%	HT%	CVD%	CKD%
Bakirdöğen, 2018 44	Cross-sectional	Turkey	25(OH) 2 D	53	28.2	9.9			52.1			22.9	52.1	0	100
Chen, 2017 49	Cross-sectional	China	25(OH) 2 D	43.7	22.6	0.67	78.0		12.6
Faridi, 2017 30	Cross-sectional	US	25(OH) 2 D	60.4		0.9	39.5		54
Jalali, 2017 50	Cross-sectional	Iran	25(OH) 2 D	36.8					55.8
Pirro, 2017 62	Cross-sectional	Italy	Uric acid	66.5	26.0			80.56	0		25.6
Sipahi, 2017 31	Interventional	Turkey	25(OH) 2 D	53.5		0.8	52.9	92.7	36.6
Uzunova, 2017 56	Cross-sectional	Bulgaria	25(OH) 2 D	42.1					54.3	62.8
Wang, 2017 57	Cross-sectional	China	25(OH) 2 D	52.4	25.4		81.3		62.4		32.5	100	41.1
Bener, 2016 47	Case-control	Qatar	25(OH) 2 D	48.6	30.0	0.85	57.5		48.5
Nardin, 2016 53	Cross-sectional	Italy	25(OH) 2 D	69.1	28.5	1.04	54.5		70.5		19.1		85.2	25.9	25.9
Sakoh, 2016 63	Cross-sectional	Japan	Uric acid	68.9	23.6			29.29	64.4			41.5	83.4		100
Takir, 2016 26	Cross-sectional	Turkey	Uric acid	54.8		0.79		88.37	18.9
Xiao, 2016 58	Case-control	China	25(OH) 2 D	56.8					44.7		26.7
Zhao, 2016 27	Cross-sectional	China	Uric acid	54.1	25.1	0.85			100
Chan, 2015 48	Interventional	World	25(OH) 2 D	37.1					72
Chien, 2015 29	Cohort	Taiwan	25(OH) 2 D	60.2	23.3				54.9		29.3	15.6	35.9	14.5
Chin, 2015 60	Cross-sectional	Malaysia	Uric acid	42.9	24.3	0.93			100
Hernandez, 2015 61	Cross-sectional	Spain	Uric acid	65	29	1.06		76.99	100
Myrda, 2015 52	Cohort	Poland	25(OH) 2 D	53.4					85.6
Parveen, 2015 54	Cross-sectional	Pakistan	25(OH) 2 D	60.0		1.19			75
Veronese, 2015 65	Cohort	Italy	Uric acid	74.4	27.9			70.46	39.2		26.8	15.1		17.7
Barker, 2014 46	Cross-sectional	USA	25(OH) 2 D	48.6	32.9				44.6
Kim, 2014 51	Case-control	South Korea	1,25(OH)D	45.8	24.5				62.6
Lane, 2014 23	Case-control	USA	Uric acid	73.8	27.4	0.93		74.08	100		18.5	18.6
Barcelo, 2013 45	Cross-sectional	Spain	25(OH) 2 D	52.3	31.3	0.99			76.9	54.6		17.7	16.5
Peng, 2013 25	Cross-sectional	China	Uric acid		24.4				0		0.6	7.2	40.8
Shrestha, 2012 55	Cross-sectional	Nepal	25(OH) 2 D						73.9
Vallianou, 2012 32	Cross-sectional	Greek	25(OH) 2 D	45.9				105.01	39.4	57.8	34.7
Yilmaz, 2012 59	Cross-sectional	Turkey	25(OH) 2 D	56.5	31.3	0.94	70.4		33.9		32.2	0	59.6	21.1
Nabipour, 2011 24	Cross-sectional	Australia	Uric acid	76.9	27.8	1.08		68.7	100		62.7
Şengül, 2011 64	Cross-sectional	Turkey	Uric acid	46.6	25.7	3.13			42.9						100
Bonakdaran, 2009 28	Cross-sectional	Iran	25(OH) 2 D	55.3	27.3	0.9	65.0	102.8	44			100

Selection of studies

Studies were included if they were comparative studies, which assessed the association between serum VIT D and SUA in adults. They were excluded if they were published in untranslatable languages, multiple publications of the same original research, or had incomplete data after 3 attempts of contacting the authors.

Study factor and outcome of interest

The main exposure and outcome were VIT D and SUA level which were measured by any method according to the original studies. The SUA could be dealt with as a continuous variable or categorized using different cut-off points (i. e., 5.7, 6, or 6.6 mg/dl in females and 7 or 7.7 mg/dl in males). VIT D level could be measured as calcifediol [25(OH)D] and calcitriol [1,25(OH) 2 D] forms. The 25(OH)D level was categorized into deficiency, insufficiency, and normal if it was <20, 20–30, and >30 ng/ml, respectively.

Data extraction

Data extraction was done using a standardized data extraction form independently by 2 reviewers (RI and SB) and checked by senior reviewers (KT and AT). The data extraction form included characteristics of the article, study, and participants, exposure and outcome data of VIT D and SUA, and covariables, including mean age, body mass index (BMI), serum creatinine, serum glycated hemoglobin (HbA1c) levels, estimated glomerular filtration rate (eGFR), percentage of males, obesity, smoking, T2D, HT, CKD, and CVD. All disagreements were solved by discussion and reference to the original article with consent from a senior reviewer (KT).

Risk of bias assessment

Two reviewers (RI and SB) assessed the risk of bias for each study independently. For observational studies, we used the Newcastle Ottawa Scale for cohort 33 , adapted for cross-sectional 34 and case-control 33 studies, while the revised Cochrane risk of bias tool for randomized trials (RoB2) 35 and Risk Of Bias In Non-randomized Studies of Interventions tool (ROBINS-I) 36 were used for interventional studies. All disagreements were solved by discussion between the 2 reviewers (RI and SB) with consent from a senior reviewer (KT).

Statistical analysis

Direct meta-analysis

Direct meta-analysis was performed to pool the effect sizes, including beta coefficients (slopes) of SUA level on serum VIT D (25(OH)D) level, odds ratio (OR) between hyperuricemia and VIT D deficiency, and unstandardized mean difference (MD) of serum VIT D between the subjects with hyperuricemia and normal SUA. Heterogeneity was assessed and it was considered present if a Cochrane Q test p-value was <0.1 or Higgins I 2 ≥25% 37 . Effect sizes were pooled using DerSimonian and Laird method if they were heterogeneous, otherwise, the inverse-variance method was used 37 .

We explored the source of heterogeneity using Galbraith plot and sensitivity or subgroup analysis where appropriate. A potential source (e. g., T2D, CVD, HT, CKD, age, BMI, and obesity) was fitted individually in a meta-regression model. If the τ 2 was decreased by ≥50%, a subgroup analysis was performed accordingly 38 . Publication bias was assessed by funnel plot and Egger test 39 .

Network meta-analysis

Network meta-analysis using two-stage approach was applied to estimate MDs of SUA between groups with different VIT D status (i. e., normal, insufficiency, and deficiency). Initially, linear regression analysis was applied to estimate the MD along with the variance-covariance for each individual study using normal VIT D status as the reference group. Then, a multivariate random-effects meta-analysis with a consistency model was used to pool the MDs across the studies 40 41 .

Inconsistency was assessed using the design-by-treatment interaction model 42 . Publication bias was assessed using a comparison-adjusted funnel plot 43 .

All statistical analysis was performed using Stata version 15.1 SE by StataCorp (College Station, Texas, USA). A p-value <0.05 was the threshold for statistical significance, except for heterogeneity where p-value <0.1 was used.

Results

Study Selection

There were 243 and 804 articles identified from the MEDLINE and Scopus databases, respectively. Thirty-two studies met our eligibility criteria (see Supplementary Fig. 1S ). Characteristics of the studies are described in Table 1 . Most of them (23) were cross-sectional studies. Their participants were either from the general population or had specific diseases, with mean age ranging from 36.9 to 76.9 years.

Among 32 studies, 21 28 29 30 31 32 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 and 11 23 24 25 26 27 60 61 62 63 64 65 considered VIT D and SUA as the outcome, respectively. Within the 21 studies with VIT D outcome, 6 31 32 44 48 49 50 provided data of SUA and VIT D levels as continuous variables, 14 28 29 30 31 32 45 46 47 49 52 53 54 57 59 assessed association between SUA level as continuous data and categorical VIT D status, while 3 55 56 58 provided categorical data for both. For the 11 studies with SUA outcome, 1 64 had continuous data for both VIT D and SUA levels, 9 23 24 25 27 60 61 62 63 65 provided continuous data for VIT D level and categorical SUA status, and 2 25 26 provided categorical data for both.

The risk of bias assessment was performed and the results can be seen in Supplementary Table 3S-5S . For observational studies, 20 23 24 25 26 27 29 30 32 45 46 47 53 57 59 60 61 62 63 65 66 , 4 51 52 56 58 , and 6 28 44 50 54 55 64 were of low, moderate, and high risk of bias, respectively. The 2 interventional studies 31 48 included had low risk of bias.

Effects of SUA on VIT D outcome

Beta coefficient

Three studies 31 32 48 reported beta coefficient of SUA level on serum 25(OH)D level (see Fig. 1 and Supplementary Table 6S ) with the pooled coefficient (95% confidence interval) of 0.512 (0.199, 0.825; I 2 =8.1%). In other words, there was a statistically significant, positive correlation between SUA and VIT D levels in that each mg/dl increase in SUA level was associated with an increase in 25(OH)D by 0.512 ng/ml. The funnel plot was symmetrical, corresponding to the non-significant Egger test (coefficient=1.94, standard error=2.19, p-value=0.539).

Fig. 1

Meta-analysis of beta coefficients of SUA level on serum VIT D level: a forest plot and b funnel plot.

Mean difference in SUA

Thirteen studies assessed associations between SUA and VIT D. Among them, 9 studies 29 30 31 32 45 46 47 49 59 compared SUA levels between VIT D deficiency, insufficiency, and normal groups, whereas 4 28 53 54 57 reported SUA only between deficiency or insufficiency and normal groups. The most commonly used cut-offs for deficiency and insufficiency were <20 and <30 ng/ml, respectively (see Supplementary Table 7S ). A network meta-analysis applied to estimate MDs of SUA levels among 3 groups across the studies yielded the non-significant pooled MDs of 0.072 (−0.153, 0.298), 0.038 (−0.216, 0.292), and 0.034 (−0.216, 0.283) mg/dl for deficiency vs normal, insufficiency vs normal, and deficiency vs insufficiency, respectively; all of them were not significant (see Fig. 2 and Table 8S ). Comparison-adjusted funnel plot from the network meta-analysis was symmetrical, indicating no significant publication bias.

Fig. 2

Meta-analysis of mean differences in SUA levels between categories of VIT D status: forest plots for a deficiency vs. normal, b insufficiency vs. normal, ( c deficiency vs. insufficiency, and d comparison-adjusted funnel plot from network meta-analysis.

Exploring the source of heterogeneity by subgroup analysis in the first direct meta-analysis (VIT D deficiency vs. normal) based on percentage of T2D patients showed that studies with partial T2D patients 29 45 had a statistically significant pooled MD in SUA level of −0.379 mg/dl (−0.552, −0.205; I 2 =0.0%), compared to the non-significant pooled MD of −0.247 mg/dl (−0.874, 0.379; I 2 =74.9%) from studies with 100% T2D patients 28 57 (see Supplementary Tables 9S). From the second direct meta-analysis (insufficiency vs normal), subgroup analysis by BMI showed that the studies with mean BMI <30 kg/m 2 29 49 had a non-significant pooled MD of 0.087 mg/dl (−0.306, 0.132; I 2 =0.0%) and those with higher mean BMI 45 46 47 59 showed a non-significant pooled MD of 0.076 mg/dl (−0.240, 0.393; I 2 =58.8%) (see Table 10S ). No specific source of heterogeneity was found in the direct meta-analysis of deficiency vs insufficiency groups.

Effects of VIT D on SUA outcome

Odds ratio

Five studies 25 26 55 56 58 assessed association between VIT D deficiency and hyperuricemia (see Table 11S ). ORs were estimated and pooled across studies, yielding the statistically significant pooled OR of 1.496 (1.141, 1.963; I 2 =3.5%), indicating a significant association in which the odds for hyperuricemia of the group with VIT D deficiency was 1.496 times that of the group with normal VIT D. There is no evidence of publication bias by the funnel plot (see Fig. 3 ) and Egger test (coefficient=−0.188; standard error=0.593; p-value=0.772).

Fig. 3

Meta-analysis of odds ratios between VIT D deficiency and hyperuricemia: a forest plot and b funnel plot.

Mean difference of VIT D

Eight studies 23 24 25 27 60 61 62 65 compared 25(OH)D levels between hyperuricemia and normouricemia groups (see Table 12S ). The MDs were moderately heterogeneous (I 2 =30.0%) with the non-significant pooled MD of 0.138 ng/ml (−0.430, 0.707; see Fig. 4 ). The funnel plot and Egger test showed no evidence of publication bias (coefficient=−0.726; standard error=1.162; p-value=0.555).

Fig. 4

Meta-analysis of mean differences in serum VIT D levels between groups with hyperuricemia and normal SUA levels. a forest plot and b funnel plot.

Discussion

We performed a systematic review and meta-analysis on the association between serum VIT D level and SUA level and vice versa, pooling the effect sizes of association between VIT D and SUA. We found that each mg/dl increase in SUA corresponded to an increase in serum 25(OH)D by approximately 0.5 ng/ml. Additionally, the odds for hyperuricemia of VIT D deficiency was approximately 1.5 times that of normal VIT D level.

According to our findings, high SUA might increase 25(OH)D level. This may be explained by the fact that hyperuricemia can inhibit 1α-hydroxylase and prevent the conversion of the 25(OH)D form of VIT D into calcitriol [1,25(OH) 2 D] 25 66 , hence increasing the level of 25(OH)D but decreasing the level of calcitriol. Genetic backgrounds may also be involved in these metabolic pathways as shown by evidence from genome-wide association studies 67 68 69 70 71 . For instance, the rs4588 and rs2282679 polymorphisms of GC (VIT D binding protein) gene and rs10766197 polymorphism of cytochrome P450 2R1 gene were highly associated with VIT D level 72 73 , whereas the rs2231142 polymorphism of adenosine triphosphate-binding cassette subfamily G member 2 (ABCG2) gene was associated with increased SUA level 71 . A Mendelian randomization study 74 also showed a causal association between the rs2231142 polymorphism of ABCG2 gene and VIT D through SUA by reducing urate transporter in renal proximal tubules. In the same site, the increasing SUA is associated with an inhibition of CYP27B1 gene expression for the 1α-hydroxylase enzyme, which converts 25(OH)D to calcitriol 66 . Furthermore, increased level of SUA might also lead to nephropathy in diabetic patients or initiation and progression of renal disease in non-diabetic patients, manifested as microalbuminuria due to hyperuricemia 75 , which might reduce the circulating VIT D binding protein and decrease VIT D level. Conversely, treatment with allopurinol, a xanthine oxidase inhibitor, in diabetic patients can decrease the SUA level and also increase the serum vitamin D level 76 . Therefore, the direction of the relationship between SUA and VIT D levels is still controversial.

It should be noted also that the study with the biggest weight for the pooled beta coefficient was a study using worldwide data on patients with chronic hepatitis B 48 , which could, in turn, cause detrimental effects on serum VIT D level. However, this description might only explain a part of the bigger picture, since there are several other factors which could affect VIT D level such as diet, BMI, age, gender, geographical location, and season 21 .

Meanwhile, for the association between VIT D and SUA outcome, the results from pooling ORs 25 26 55 56 58 showed a significant association between VIT D deficiency and hyperuricemia. This pooled OR is inconsistent with the above-mentioned pooled beta coefficient of SUA on serum 25(OH)D level. The mechanism behind this might be the secondary hyperparathyroidism caused by VIT D deficiency 77 which affects the ABCG2 gene 78 and thus decreases excretion or increases reabsorption of uric acid in the kidney tubules 60 . This VIT D deficiency could be caused by limited intake, insufficient ultraviolet exposure from sunlight, or genetic factors. It should be noted that as genetic factors can affect both VIT D and SUA levels as previously mentioned, each with their own specific gene polymorphisms, the fact that these genetic factors are not evenly distributed across different populations might also explain the conflicting results for the associations between SUA and VIT D, being pooled from a number of different populations.

This study is the first systematic review on the bidirectional association between SUA and VIT D which includes a total of 32 studies within the analysis. However, there are also some limitations in the present review, for example, the 2 significant pooled effect sizes are from a limited number of studies, whereas the other meta-analyses which consist of higher number of studies produce heterogeneous results.

In summary, increasing SUA might increase 25(OH)D VIT D level, while VIT D deficiency is associated with hyperuricemia. These reverse relationships should be further evaluated in a longitudinal study.