Performance of urinary vitamin D-binding protein in diabetic kidney disease: a meta-analysis

Abstract

Objective

To systematically evaluate the correlation between urinary vitamin D-binding protein (VDBP) and diabetic kidney disease and to evaluate the relationship between urinary VDBP and the albumin-to-creatinine ratio (ACR), renal function indicators [estimate glomerular filtration rate (eGFR), creatinine (CR), blood urea nitrogen (BUN)] and glycaemic control indices [glycated hemoglobin (HbA1c), fasting plasma glucose (FPG)].

Methods

We searched the CNKI, Wanfang, VIP, CBM, PubMed, Cochrane Library, Embase and Web of Science databases up to May 31, 2023, for relevant literature. RevMan 5.3 software was used for the meta-analysis.

Results

Ultimately, 9 articles were included. Due to heterogeneity in the pooled results, the random-effects model was chosen. Meta-analysis results showed that the urinary VDBP concentrations in the normal albuminuria diabetes group were significantly higher than those in the healthy control group [SMD 1.52, 95% CI (0.84, 2.19), p < 0.00001]. The urinary VDBP concentrations in the microalbuminuria diabetes group were significantly higher than those in the normal albuminuria diabetes group [SMD 1.81, 95% CI (1.40, 2.21), p < 0.00001]. The urinary VDBP concentrations in the macroalbuminuria diabetes group were also significantly higher than those in the microalbuminuria diabetes group [SMD 1.51, 95% CI (1.05, 1.96), p < 0.00001]. In addition, urinary VDBP was positively correlated with the ACR, CR, BUN and HbA1c [Summary r = 0.73, 95% CI (0.54, 0.85), p < 0.0001; Summary r = 0.38, 95% CI (0.10, 0.61), p = 0.009; Summary r = 0.37, 95% CI (0.16, 0.55), p = 0.0008; Summary r = 0.40, 95% CI (0.13, 0.62), p = 0.005, respectively] and tended to be negatively correlated with the eGFR [Summary r = −0.64, 95% CI (−0.92, 0.10), p = 0.08] but was not significantly correlated with the FPG [Summary r = 0.16, 95% CI (−0.03, 0.33), p = 0.10]. Sensitivity analysis showed that our pooled results are robust.

Conclusion

Urinary VDBP may be used as a novel biomarker for the early diagnosis of DKD and can be used to assess the severity of DKD.

1. Introduction

Diabetic kidney disease (DKD) is a serious microvascular complication of diabetes and is the leading cause of chronic kidney disease (CKD) and end-stage renal disease worldwide due to its insidious onset and low rate of early presentation [1]. Therefore, early detection of DKD and early intervention are crucial. At present, the noninvasive diagnostic method for diabetic kidney disease is the measurement of urinary microalbumin. However, previous studies have reported that some diabetic patients still develop chronic kidney disease when their urinary microalbumin levels are within the normal range [2]. Therefore, biomarkers with higher diagnostic value are urgently needed to increase early diagnosis and treatment of diabetic kidney disease and delay progression to CKD.

Vitamin D-binding protein is also called Gc globulin, with a molecular weight of 51,000–58,000. It was first isolated from plasma α2-globulin by Hirschfeld in 1959. VDBP is mainly synthesized by the liver, and the kidneys can also secrete a small amount of VDBP, which is filtered in the glomerulus and reabsorbed in the proximal tubules. In recent years, urinary VDBP, as an important renal tubular early damage marker in DKD, has received widespread attention, and a number of studies [3–11] have found that urinary VDBP is abnormally elevated in early DKD and can thus be used as a biomarker for the early diagnosis of DKD. However, one study found no association between VDBP and DKD [12]. Therefore, the current meta-analysis aimed to examine the role of VDBP in DKD to provide evidence-based findings for clinical practice.

2. Materials and methods

2.1. Literature search

We searched the CNKI, Wanfang Data Service Platform (Wanfang), Weipu Database (VIP), China Biomedical Database (CBM), PubMed, Embase, Cochrane Library, and Web of Science databases up to 31 May 2023. The search terms included ‘diabetic nephropathy’, ‘diabetic kidney disease’, ‘nephropathy, Diabetic’, ‘kidney disease, diabetic’, ‘diabetic glomerulosclerosis’, ‘Vitamin D-binding protein’, ‘VDBP’, ‘Vitamin D-Binding Globulin’, ‘25-Hydroxyvitamin D-Binding Protein’, ‘Gc Globulin’, etc. All subject words and free words were searched. Additionally, the Chinese Clinical Trials Registry and the US Clinical Trials Registry were manually searched for unpublished but completed literature.

2.2. Inclusion and exclusion criteria

Inclusion Criteria: (1) Participants: patients with type-1 or type-2 diabetes diagnosed according to the WHO diagnostic criteria [13]. (2) Study group: Diabetic patients were divided into normal albuminuria group, microalbuminuria group and macroalbuminuria group according to the ACR [14], with or without a healthy control group. (3) Outcomes: Comparison of urinary VDBP concentration between groups or correlation analysis of urinary VDBP with the ACR, eGFR, CR, BUN, HbA1c, and FPG. (4) Study designs: cross-sectional studies, case–control studies, or cohort studies.

Exclusion criteria: (1) duplicate literature; (2) reviews, conference abstracts, case reports, animal studies, etc.; (3) studies without full text; (4) studies that did not report VDBP data.

2.3. Selection process

Two researchers (C.HS. and N.LH.) independently reviewed the retrieved literature, first to eliminate duplicate articles. Then, in pairs, the researchers independently screened titles and abstracts to exclude obviously irrelevant literature and screened out which articles needed to be read in full. Finally, two investigators independently screened full-text articles for inclusion. In case of disagreement, consensus was reached in consultation with the third investigator (W.XY.) strictly based on the inclusion/exclusion criteria.

2.4. Data extraction and quality evaluation

We designed a data extraction form using Microsoft Excel, which two investigators (C.HS. and N.LH.) independently used to extract data from eligible studies. Any differences or disagreements were resolved through consultation with a third investigator (W.XY.). The two investigators independently reviewed all articles and extracted the following data: first author, time of publication, country, type of study design, age, sex, urinary VDBP detection method, urinary VDBP concentration (mean ± standard deviation) in each group, and the correlation coefficient of urinary VDBP with the ACR, eGFR, CR, BUN, HbA1c, and FPG. The Newcastle–Ottawa scale (NOS) was applied to evaluate the quality of the included studies [15]. The specific OS scoring criteria are as follows:

1. The study subjects were evaluated by 4 items (4 points): whether the case was appropriate (1 point); whether the case was representative (1 point); whether the control was selected from the same population as the case (1 point); and whether the control group had no history of the target disease (1 point).

2. Intergroup comparability was evaluated as follows (2 points): whether comparability of cases and controls was considered in the design and statistical analysis; whether the study controlled for the most important confounding factors, such as age and sex (1 point); and any other confounding factors, such as blood pressure, diabetes complications, BMI, hemoglobin, uric acid, etc. (1 point)

3. The results were assessed by 3 items (3 points): whether the results were blinded (1 point); whether the same method was used to determine the results of the case and control groups (1 point); and whether the nonresponse rate was the same (1 point).

In the NOS system, each study could be assigned a score of 0–9 points, and scores ≥7 suggest that studies are high-quality.

2.5. Statistical analysis

RevMan5.3 software was used for statistical analysis. The analysis index was a continuous variable; if the units and measurement methods were the same, the mean difference (MD) was used. Otherwise, the standardized mean difference (SMD) was used. Interstudy heterogeneity was assessed using the Cochran-Q test and the I² statistic. When there was statistically significant heterogeneity (I² > 50%, p < 0.05), subgroup analysis or sensitivity analysis was considered to identify the sources of heterogeneity. Subgroup analysis was performed based on the study population and whether urinary VDBP was correlated with creatinine. Sensitivity analyses were performed by excluding individual studies, excluding type-1 diabetes studies, and excluding low-quality studies. We estimated the E-value to evaluate potential unmeasured confounders and the certainty of evidence, and according to the method reported by VanderWeele et al. [16], specific measurement methods are provided in the Supplementary Material. If there was still heterogeneity, the data were pooled using a random-effects model; otherwise, a fixed-effects model was used for the pooled analysis. All effect sizes were expressed with 95% CIs. When p < 0.05, the difference was considered statistically significant. When the number of studies included in meta-analyses was ≥ 10, funnel plots were used to assess publication bias.

The correlation coefficients extracted in this study have two kinds of Pearson correlation coefficients and Spearman correlation coefficients, which cannot be directly combined. We converted the Spearman correlation coefficient to the Pearson correlation coefficient before meta-analysis according to the formula [r = 2sin(r_s π/6)] proposed by Rupinski et al. [17]. Since the correlation coefficient (r) does not follow a normal distribution, directly merging the correlation coefficients from each study would lead to serious result bias. Therefore, the ‘Fisher Z transform’ was performed first. The correlation coefficient was transformed using a normal Z distribution, and then, the SE was calculated. Then, Fisher’s Z and SE were combined and analyzed by RevMan software to calculate the combined Fisher’s Z value. Finally, the inverse Fisher transform was performed to obtain the summary r value. The Fisher Z transformed value, SE and inverse Fisher transformed value were converted in Excel according to the following formula reported by Borenstein et al. [18]. Finally, the summary r was used to evaluate the strength of the correlation: ≥ 0.8 high correlation, 0.3–0.8 moderate correlation, and ≤0.3 indicated low correlation.

$$\begin{array}{c}\text{Fisher}\prime \mathrm{s} z=0.5\times In\frac{1+r}{1-r}\\ Vz=1/\left(n-3\right)\\ S_E=\sqrt{Vz}\\ \text{Summary}\mathit{r} \text{=}\left(e^{2z}-1\right)/\left(e^{2z}+1\right)\end{array}$$

3. Results

3.1. Literature search results

A total of 656 relevant studies were retrieved from the databases. After removing 85 duplicate studies, 571 studies underwent title and abstract screening. In this process, 531 irrelevant studies, 8 reviews, 6 animal tests, and 9 articles that did not meet the inclusion criteria were excluded. Then, further screening was carried out by carefully reading the full texts, and 8 studies were excluded. Ultimately, 9 articles [3–11] were included in the meta-analysis, as shown in Figure 1.

Figure 1.

Flow chart of the screening process.

3.2. Basic characteristics and quality assessment of studies

The nine included studies involved 1179 samples. Five studies were conducted in China, two studies were conducted in Egypt, and the other studies were conducted in Saudi Arabia and the United States. Seven studies included people with type-2 diabetes, one study included people with type-1 diabetes, and one included people with type-1 diabetes and people with type-2 diabetes. The quality scores of the included studies ranged from 6 to 8. The basic characteristics of the included studies are detailed in Table 1.

Table 1.

Basic characteristics of the literature.

Study	Area	Study design	DM type	Age, years	N, M/F	Determination of VDBPVDBP	NOS
Thrailkill [3], 2011	States	cross-sectional	1	NC:24.3 ± 7.6 NO:21.1 ± 7.8 MI:19.4 ± 7.0	170, 78/92	ELISA	8
Tian [4], 2014	China	cross-sectional	1 and 2	NC:60.89 ± 13.92 NO:64.63 ± 11.76 MI:64.49 ± 11.78 MA:64.80 ± 12.98	150, 92/58	ELISA	7
Wang [5], 2015	China	cross-sectional	2	NC:60.1 ± 7.6 NO:63.2 ± 9.6 MI:64.00 ± 10.1 MA:63.8 ± 8.8	86, 43/43	ELISA	8
Shoukry [6], 2015	Egypt	cross-sectional	2	NC:51.6 ± 5.1 NO:52.3 ± 5.6 MI:53.5 ± 7.3 MA:54.6 ± 6.9	100, 66/34	ELISA	8
Yuan [7], 2015	China	cross-sectional	2	NC:60.1 ± 9.6 DM:59.2 ± 10.6	208, 104/104	ELISA	6
Khodeir [8], 2016	Egypt	cross-sectional	2	NC:NR NO:NR MI;NR MA;NR	60	ELISA	7
Su [9], 2018	China	cross-sectional	2	NC:54.7 ± 8.5 NO:54.2 ± 8.6 MI:53.6 ± 7.9 MA:55.1 ± 9.2	110, 64/46	ELISA	7
Fawzy [10], 2018	Saudi Arabia	cross-sectional	2	NC:47.6 ± 13.6 NO:50.6 ± 7.8 MI:50.4 ± 12.0 MA:45.9 ± 6.5	160, 26/134	ELISA	7
Huang [11], 2023	China	cross-sectional	2	NC:50.87 ± 12.30 NO:55.26 ± 11.76 MI:54.29 ± 11.56 MA:56.33 ± 10.27	135, 93/42	ELISA	7

DM: diabetes mellitus; N: number of participants; M: male; F: female; NOS: Newcastle–Ottawa Scale; ELISA: enzyme-linked immunosorbent assay; MA: macroalbuminuria group in DM patients; MI: microalbuminuria group of DM patients; NO: normal albuminuria group of DM patients; NR: not reported; NC: normal control.

3.3.1. Urinary VDBP in the normal albuminuria group versus the healthy control group

Seven studies [4–10] reported urinary VDBP values in the normal albuminuria group (215 patients) and healthy control group (272 patients). The SMD was calculated due to the different units of VDBP. There was statistical heterogeneity between studies [I² = 90%, 95% CI (34%, 96%), p < 0.1], and thus, the random-effects model was used. The results showed that the urinary VDBP concentrations in the normal albuminuria group were significantly higher than those in the healthy control group [SMD 1.52, 95% CI (0.84, 2.19), p < 0.00001] (Figure 2A). We performed sensitivity analyses by excluding individual studies, studies of type-1 diabetes [4], and low-quality studies [7], and no clear change in heterogeneity was observed. According to the previous method used to calculate the E-value, the E-value (point estimate) was 9.954, and the E-value (CI) was 3.865, thus suggesting that the results of the pooled analysis were robust. Of these seven studies, urinary VDBP was measured by creatinine correction in four studies [4,6,7,10], and three did not use creatinine correction [5,8,9]. Therefore, we performed a subgroup analysis based on whether VDBP was correlated with creatinine. The results showed that heterogeneity in the creatinine correction group remained high, while heterogeneity in the non-creatinine correction group decreased to 0 (Figure 2B), indicating that urinary VDBP measurement methods may be a cause of heterogeneity. In addition, four of the seven studies were conducted in China [4,5,7,9], two were conducted in Egypt [6,8], and one was conducted in Saudi Arabia [10]. We performed subgroup analyses based on study location, and no significant change in heterogeneity was observed in China and Egypt (Figure 2C).

Figure 2.

Meta-analysis forest plots of urinary VDBP in the normal albuminuria group versus the healthy control group.

3.3.2. Urinary VDBP in the microalbuminuria versus normal albuminuria groups

Seven studies [4–10] reported urinary VDBP values in the microalbuminuria group (224 patients) and normal albuminuria group (215 patients). The SMD was calculated due to the different units of VDBP. There was statistical heterogeneity between the studies [I² = 67%, 95% CI (0%, 87%), p < 0.1], and thus, the random-effects model was used. The results showed that urinary VDBP concentrations in the microalbuminuria group were significantly higher than those in the normal albumin group [SMD 1.81, 95% CI (1.40, 2.21), p < 0.00001] (Figure 3A). We performed sensitivity analysis by excluding individual studies, studies of type-1 diabetes [4], and low-quality studies [7], and no clear change in heterogeneity was observed. According to the previous method used to calculate the E-value, the E-value (point estimate) was 14.744, and the E-value (CI) was 8.505, thus suggesting that the results of the pooled analysis were robust. Since four studies [4,6,7,10] measured VDBP by creatinine correction and three studies [5,8,9] did not, subgroup analysis based on whether VDBP was corrected with creatinine was performed. The results showed that heterogeneity in the creatinine correction group remained high, while heterogeneity in the non-creatinine correction group decreased to 20% (Figure 3B), indicating that the urinary VDBP measurement method was likely a cause of heterogeneity. In addition, among the seven studies, four studies were conducted in China [4,5,7,9], two were conducted in Egypt [6,8], and one was conducted in Saudi Arabia [10]. Therefore, we performed subgroup analyses based on study populations. The heterogeneity in Chinese populations decreased to 45%, while heterogeneity was significantly increased in Egypt, thus indicating that the study population was one of the sources of heterogeneity (Figure 3C).

Figure 3.

Meta-analysis forest plots of urinary VDBP in the microalbuminuria versus normal albuminuria groups.

3.3.3. Urinary VDBP in the macroalbuminuria versus the microalbuminuria group

Six studies [4–6,8–10] reported urinary VDBP values in the microalbuminuria group (163 patients) and the macroalbuminuria group (168 patients). Due to the different units of VDBP, the SMD was calculated. There was significant heterogeneity between studies [I² = 69%, 95% CI (0%, 88%), p < 0.1], and thus, the random-effects model was used. The results showed that the urinary VDBP concentrations in the macroalbuminuria group were significantly higher than those in the microalbuminuria group [SMD 1.51, 95% CI (1.05, 1.96), p < 0.00001] (Figure 4A). We performed sensitivity analysis by excluding individual studies and studies of type-1 diabetes [4], and no clear change in heterogeneity was observed. According to the previous method used to calculate the E-value, the E-value (point estimate) was 9.820, and the E-value (CI) was 5.227, thus suggesting that the results of the pooled analysis were robust. Since three studies [4,6,10] measured VDBP by creatinine correction and three studies [5,8,9] did not, subgroup analysis based on whether VDBP was corrected with creatinine was performed. The results showed that heterogeneity in the creatinine correction group remained high, while heterogeneity in the non-creatinine correction group decreased to 0% (Figure 4B), indicating that urinary VDBP measurement methods may be a cause of heterogeneity. In addition, three studies were conducted in China [4,5,9], two were conducted in Egypt [6,8], and one was conducted in Saudi Arabia [10]. Therefore, we performed subgroup analyses based on study populations, and the results showed that the heterogeneity in Chinese populations decreased to 41%, while the heterogeneity increased in the Egyptian population. Therefore, the study population was one of the sources of heterogeneity (Figure 4C).

Figure 4.

Meta-analysis forest plots of urinary VDBP in the macroalbuminuria versus the microalbuminuria group.

3.3.4. Correlation of VDBP with the ACR

Eight studies [3,4,6–11] evaluated the association of urinary VDBP with the ACR, involving a total of 773 patients with DM, two of which [3,11] reported the Spearman correlation coefficient between urinary VDBP and the ACR, which was converted to the Pearson correlation coefficient according to the previous formula. There was statistical heterogeneity between studies [I² = 95%, 95% CI (79%, 98%), p < 0.1], so the random-effects model was used. As shown in Figure 5A, there was a positive correlation between urinary VDBP and the ACR [Summary Fisher’s Z = 0.94, 95% CI (0.60, 1.27), p < 0.0001]. The inverse Fisher transform yielded a summary r value of 0.73 [95% CI (0.54, 0.85)], indicating a moderate correlation.

Figure 5.

Forest plots of correlation analysis between VDBP values and different clinical index of ACR (A), eGFR (B), CR (C), BUN (D), HbA1c (E), FPG (F).

3.3.5. Correlation of VDBP with the eGFR

Five studies [5,6,8,10,11] evaluated the association between urinary VDBP and the eGFR, involving a total of 456 patients with DM, one of which [11] reported Spearman correlation coefficients, which were converted to Pearson correlation coefficients. There was statistical heterogeneity between studies [I² = 99%, 95% CI (89%, 99.6%), p < 0.1]; thus, the random-effects model was used. As shown in Figure 5B, there tended to be a negative correlation between urinary VDBP and the eGFR [Summary Fisher’s Z = −0.76, 95% CI (−1.62, 0.10), p = 0.08]. Inverse Fisher transform yielded a summary r value of −0.64 [95% CI (-0.92, 0.10)], indicating a moderate correlation.

3.3.6. Correlation of VDBP with CR

Four studies [5,6,8,10] evaluated the association of urinary VDBP with creatinine in 321 patients with DM. There was statistical heterogeneity between studies [I² = 86%, 95% CI (0%, 96%), p < 0.1]; thus, the random-effects model was used. As shown in Figure 5C, there was a positive correlation between urinary VDBP and creatinine [Summary Fisher’s Z = 0.40, 95% CI (0.10, 0.70), p = 0.009]. The inverse Fisher transform yielded a summary r value of 0.38 [95% CI (0.10, 0.61)], indicating a moderate correlation.

3.3.7. Correlation of VDBP with BUN

Four studies [5,6,8,10] evaluated the association between urinary VDBP and blood urea nitrogen in a total of 321 patients with DM. There was statistical heterogeneity between studies [I² = 75%, 95% CI (0%, 92%), p < 0.1]; thus, the random-effects model was used. As shown in Figure 5D, there was a positive correlation between urinary VDBP and urea nitrogen [Summary Fisher’s Z = 0.39, 95% CI (0.16, 0.62), p = 0.0008]. The inverse Fisher transform yielded a summary r value of 0.37 [95% CI (0.16, 0.55)], indicating a moderate correlation.

3.3.8. Correlation of VDBP with HbA1c

Five studies [3,5,6,8,10] evaluated the association of urinary VDBP with HbA1c, involving a total of 436 patients with DM, one of which [3] had the Spearman correlation coefficient, which was converted to the Pearson correlation coefficient. There was statistical heterogeneity between studies [I² = 89%, 95% CI (3%, 96%), p < 0.1]; thus, the random-effects model was used. As shown in Figure 5E, there was a positive correlation between urinary VDBP and HbA1c [Summary Fisher’s Z = 0.42, 95% CI (0.13, 0.72), p < 0.00001]. The inverse Fisher transform yielded a summary r value of 0.40 [95% CI (0.13, 0.62)], indicating a moderate correlation.

3.3.9. Correlation of VDBP with FPG

Four studies [5,6,8,10] evaluated the association of urinary VDBP with FPG in a total of 321 patients with DM. There was statistical heterogeneity between studies [I² = 64%, 95% CI (0%, 89%), p < 0.1]; thus, the random-effects model was used. As shown in Figure 5F, there was no significant association between urinary VDBP and FPG [Summary Fisher’s Z = 0.16, 95% CI (−0.03, 0.35), p = 0.10]. The inverse Fisher transform yielded a summary r value of 0.16 [95% CI (−0.03, 0.33)].

4. Discussion

VDBP belongs to the bound albumin family and is mainly synthesized by the liver. Its main function is to bind and transport vitamin D and its metabolites. VDBP can significantly extend the half-life of 25-hydroxyvitamin D through the binding reaction, which is essential for maintaining plasma vitamin D concentration and regulating its bioavailability. In addition, VDBP also binds to and removes actin, thereby reducing systemic vascular occlusion and cell damage. In recent years, VDBP has been widely studied with respect to its role in immune defence and inflammatory response. It has been shown to play a role in immune defense by enhancing the action of complement-related peptides and promoting neutrophil aggregation. In addition, VDBP can be converted to a potent macrophage activating factor (MAF) [19] under the deglycosylation of some enzymes, including sialidase in T lymphocytes and galactosidase in B lymphocytes, which has a potent chemotaxis effect on macrophages by stimulating macrophage inflammatory responses, participating in host defenses, and causing inflammatory damage. Macrophages infiltrate the glomeruli and tubules of DKD, which is an important histological feature of DKD progression [20] and is also the central link of its pathogenesis. This feature may be related to the pathogenesis of the effect of VDBP on DKD. In addition, Mirković et al. found that urinary VDBP is a biomarker of tubulointerstitial damage; it exists independently of proteinuria, and it occurs in the early stages of the inflammatory response and tubulointerstitial fibrosis [21]. Tubulointerstitial damage is an important pathogenesis of diabetic nephropathy. Furthermore, excessive urinary VDBP excretion predicts intrarenal and extrarenal vitamin D deficiency [22], which has been shown to be independently associated with the development and severity of DKD [23].

This study may be the first meta-analysis of the association between urinary VDBP and diabetic nephropathy. We found that urinary VDBP was significantly higher in the normal albuminuriauria group than in the healthy control group. Urinary VDBP was significantly higher in the microalbuminuria group than in the normal albuminuria group. Urinary VDBP was significantly higher in the macroalbuminuria group than in the microalbuminuria group. This suggests that urinary VDBP may be a novel marker for the early prediction or diagnosis of diabetic nephropathy. The specific mechanism of increased urinary VDBP excretion is not well understood. In normal kidneys, VDBP is reabsorbed as a 25-(OH)-vitamin D3/VDBP complex by megalin-mediated endocytosis and catabolized by proximal tubular epithelial cells, thereby reducing their urinary excretion levels [24]. On the one hand, studies have shown that the excretion of proximal tubular megalin/cubilin protein in the urine is increased in patients with DKD [25–26]; the megalin/cubilin protein receptor is a polyligand endocytic receptor expressed by the brush-like boundary of proximal tubular cells and is involved in the reuptake of low-molecular-weight proteins (such as albumin and VDBP) filtered into the glomerular filtrate. In addition, it has been demonstrated that the main factors involved in the development of DN glomerulosclerosis and interstitial fibrosis (e.g., TGF-β and angiotensin II) may negatively regulate receptor-mediated endocytosis [27–28], participating in enhanced uVDBP excretion. The loss of macroprotein receptors, as well as the negative regulation of their endocytosis, leads to a decrease in the reabsorption of uVDBP from the renal tubules, which leads to an increase in its excretion. On the other hand, considering that increased urinary VDBP excretion is associated with renal interstitial damage, which is closely related to the development of DKD, independent of urinary protein, this hypothesis can be supported by the results of Mirkovic et al. [21] and Chaykovska et al. [29]. They found that uVDBP increased with the severity of renal damage in a rat model of proteinuria nephropathy, which was not associated with proteinuria. In the former study, they had a partial response to intensive renal protective therapy, while in the latter study, there was a partial response to contrast-induced nephropathy. It has been hypothesized that tubulointerstitial fibrosis-damaged tubular epithelial cells may no longer be able to process VDBP, resulting in a significant loss of VDBP in urine [29]. As mentioned earlier in this study, Sharif et al. [12] found no correlation between VDBP and diabetic nephropathy. After carefully reading their study, we found that they grouped patients based on the eGFR values, and the sample size was small; therefore, their study could not be included in the comparison of urinary VDBP among the three UACR groups. In this study, heterogeneity was high in all three groups, and we performed subgroup analysis and sensitivity analysis. The results showed that heterogeneity was reduced to varying degrees in both the non-creatinine correction subgroup and the Chinese subgroup within the three groups, indicating that VDBP measurement methods and population may be one of the sources of heterogeneity. Differences in urinary VDBP between study populations may indicate genetic differences. One study [30] found that the most common haplotype among White individuals was Gc1s/Gc1s and that among Chinese individuals was Gc1f/Gc1f. After sensitivity analysis, the results of all three groups were found to be stable, and the calculated E-values were large, indicating that a large unmeasured confounding effect was required to explain away our results, which indicated that the pooled results were more reliable.

This study also found that urinary VDBP was positively correlated with the ACR, CR, and BUN but was not correlated with the eGFR. Based on previous results, urinary VDBP is gradually elevated with an increase in urinary microalbumin, reflecting the severity of renal impairment. In general, DKD is characterized by persistent albuminuria or a progressive decline in the eGFR; the ACR has long been known to be closely related to DKD kidney damage, and creatinine and BUN are common markers of kidney damage, which is consistent with the results of this study. However, there tended to be a negative correlation between urinary VDBP and the eGFR in this study, and sensitivity analysis showed that after the removal [6] of the study by Shoutry et al. the results indicated a significant negative correlation between urinary VDBP and the eGFR [summary r = −0.30, 95% CI (–0.52, −0.05), p = 0.02]. In this study, the eGFR was calculated based on SCr; thus, the absolute value of the correlation coefficients between urinary VDBP and SCr and between urinary VDBP and the eGFR were similar, as shown in several other studies. However, after carefully reading the study by Shoutry et al. we found that urinary VDBP had a correlation coefficient of 0.44 with SCr, while the correlation coefficient with the eGFR was −0.998, which exhibited significant differences. Therefore, we hypothesized that measurement errors or statistical errors may have contributed to significant heterogeneity.

This study also found that urinary VDBP was positively correlated with glycated hemoglobin and not with FBG. Vitamin D bioavailability decreases as urinary VDBP increases, and low vitamin D levels are strongly associated with poor diabetes control [31]. Vitamin D exerts normal insulin secretion by acting on vitamin D3 receptors present on pancreatic islet β cells. Elevated urinary VDBP may promote hyperinsulinaemia and insulin resistance by regulating the active vitamin D content of pancreatic β cells, thereby causing poor glycaemic control [32]. This is consistent with the positive correlation between urinary VDBP and glycosylated hemoglobin. However, no association with fasting plasma glucose was found in this study, which may be related to the fact that fasting blood glucose reflects immediate blood glucose and is susceptible to factors such as diet, exercise, and emotional state.

This study includes the most recent clinical trials and may be the first meta-analysis to investigate the role of urinary VDBP in diabetic nephropathy. Compared with previous individual studies, the results of our meta-analysis further confirmed the correlation between urinary VDBP and DKD and confirmed the correlations between urinary VDBP and renal function-related indicators and glycaemic control indicators, providing an alternative biomarker for early clinical diagnosis of DKD and assessment of DKD severity.

However, this study also had some limitations. First, the heterogeneity of this study was generally high, and although subgroup analysis and sensitivity analysis found that some of the heterogeneity was derived from the measurement method of urinary VBDP and different populations, there were other unknown sources of heterogeneity. The primary literature populations included in this study were Chinese, Egyptian, Arab, and American populations, but none of them mentioned specific racial/ethnic proportions, which may have had an impact on the results. Moreover, most of the primary literature did not mention or did not control for confounding factors affecting DKD; only 1 [10] controlled for diabetic retinopathy, [3–5, 11,] controlled for blood glucose levels, and none mentioned hemoglobin, nutritional status, or socioeconomic factors, which may have a large impact on the reliability of the pooled results. Second, due to the small number of included studies, publication bias could not be assessed, implying that statistical deficiencies exist. This was a small cross-sectional study, most of which was conducted in patients with type-2 diabetes. To further explore the role of urinary VDBP in diabetic nephropathy, prospective, large-sample, high-quality studies with people with type-1 diabetes and people with type-2 diabetes are needed. Finally, our review is not registered online and the review protocol is not prepared, which may indicate problems with transparency.

In conclusion, urinary VDBP can be used as a novel marker for the early diagnosis of DKD and can be used to assess the severity of DKD.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The search strategy, the method of E-value measurement and sensitivity analysis results are available in the Supplementary Material.