Association of UCP1 and UCP2 variants with diabetic retinopathy susceptibility in type-2 diabetes mellitus patients: a meta-analysis

Background Genetic association of uncoupling proteins (UCPs) variants with the susceptibility of diabetic retinopathy (DR) in diabetes mellitus (DM) patients has been reported but with controversy. Here we aimed to conduct a meta-analysis to confirm the association of different UCPs variants with DR. Methods Three databases (Medline Ovid, Embase Ovid and CENTRAL) were applied in the literature search. Five genetic models, including allelic, homozygous, heterozygous, dominant and recessive models, were evaluated. Odds ratios (OR) were estimated under the random or fixed-effects models. Subgroup analyses, publication bias and sensitivity analyses were also conducted. Results Eleven studies on 2 UCPs variants (UCP1 rs1800592 and UCP2 rs659366) were included. Our meta-analysis showed that UCP1 rs1800592 was not associated with DR in type-2 DM patients, and UCP2 rs659366 also showed no association with DR. In the subgroup analyses on the stage of DR, allele G of UCP1 rs1800592 significantly increased the susceptibility of proliferative diabetic retinopathy (PDR) in type-2 DM patients in the allelic (OR = 1.26, P = 0.03) and homozygous models (OR = 1.60, P = 0.04). Subgroup analysis on ethnicity did not found any significant association of rs1800592 and rs659366 with DR. Conclusion Our meta-analysis confirmed the association of UCP1 rs1800592 variant with PDR in patients with type-2 DM, suggesting its potential as a genetic marker for PDR prediction in population screening. Supplementary Information The online version contains supplementary material available at 10.1186/s12886-021-01838-1.

to the oxidative phosphorylation complexes and abolished adenosine triphosphate biosynthesis [8], and induces capillary endothelial cell apoptosis, which subsequently leads to the dysregulation of the angiogenesisrelated genes [9] and diabetic microvascular complications, including DR [10,11]. Ion channels physiologically play a role in signal transmission and visual processing, but also link to induced oxidative stress and significantly contribute to a wide spectrum of ocular diseases [12]. Uncoupling proteins (UCPs) belong to a group of proton carrier transporters (H + ) in the inner membrane of mitochondria [13]. UCPs are able to uncouple the oxidized substrates and dissipate the potential energy on the inner membrane as heat to reduce ROS overproduction from mitochondria [14][15][16][17]. The overproduced ROS could cause increases proton conductance by UCP1-3, leading to decease in superoxide radicals through the mitochondria respiratory chain reaction [18]. In human genome, there were five different UCPs, named UCP1 to 5, with various tissue distributions and functions [19]. Uncoupling protein 1 (UCP1) gene is located on chromosome 4q31.1 and found to be expressed in brown adipose tissue, endothelial cells and pericytes of retina [20]. UCP1 mainly plays a role in the maintenance of body temperature in a cold environment through non-shivering thermogenesis [17]. It has been shown that elevated of glucose levels upregulates UCP1 expression, protecting cells from glucose-induced ROS damage [21]. Uncoupling protein 2 (UCP2) and 3 (UCP3) genes are both located in the same cluster on chromosome 11q13.4. UCP2 is ubiquitously expressed across different tissues in the body, whereas UCP3 is mainly expressed in the skeletal muscle tissue [22]. In UCP2 knockout mice, ROS production increases in macrophages and pancreatic islets [23,24], whereas overexpression of UCP2 inhibits mitochondrial death pathway in cardiomyocytes [25], indicating that UCP2 could be involved in cell protection from ROS damage. UCP2 and UCP3, together with SLC25A27 (UCP4) and BMCP1 (UCP5), exert cytoprotective effects by reducing oxidative stress under certain conditions [22].
Since UCPs are involved in the pathophysiology of glucose-related ROS cell damage, it is reasonable to hypothesize that the UCPs variants could be related to the susceptibility of DR. Yet, inconsistent results have been reported on the association analysis of UCPs variants with the risk of DR [26][27][28]. Herein, we aimed to conduct a meta-analysis to clarify the association of different UCPs variants with the susceptibility of DR.

Study design
The protocol of this meta-analysis has been registered in the international prospective register of systematic reviews (PROSPERO protocol CRD42020173510; available at https://www.crd.york.ac.uk/prospero/).

Searching strategies and selection criteria
Three databases, including Medline Ovid, Embase Ovid and CENTRAL, were applied in the literature search for the potential studies. The eligible studies related to the susceptibility of DR and UCPs variants would be included in this meta-analysis. The following terms were used in this search: "diabetic retinopathy", "uncoupling protein", and "polymorphisms OR variants". The detail search strategies and results were shown in Supplementary document.
Literature language was not limited to English. For languages other than Chinese and English, Google Translate (http://translate.google.com/) was used to translate the full text. The bibliographies of the screened articles have been carefully browsed to identify the omitted relevant studies.
The inclusion criteria included: 1) studies on the analysis of the association of UCPs variants with DR; 2) the recruited participants were independent and unrelated to each other; 3) sufficient genotype data for the calculation of odd ratio (OR) with 95% confidence interval (C.I.); 4) participants diagnosed with diabetes without retinopathy (DWR) would be served as the control subjects for the Hardy-Weinberg equilibrium (HWE) analysis or the data provided should be able to calculate HWE of the control group; and 5) the type of diabetes in the participants was clearly provided, including type-1 and type-2 DM. The exclusion criteria included: 1) the genotype distributions of the control subjects did not follow HWE (P HWE < 0.05); and 2) the variants reported only by one study would not be included in this analysis.

Data extraction
Two researchers (X.L. and Z.J.) independently extracted and assessed the full-text reports for all potentially eligible studies. The included studies were evaluated by the Newcastle-Ottawa quality assessment scale (NOS). The extracted items include: first author, year of publication, region of study, ethnicity, number of cases and controls, diagnostic criteria, allele or genotype frequency, Hardy-Weinberg equilibrium (HWE) status, and genotyping method. If there was any disagreement regarding to the eligibility, scores of NOS and extracted items, the judicator (Z.W.) would make the final decision. If any full-text reports have been rejected, the reasons for the rejection would be given. The data extraction form included the following data: 1) the first author and the year of publication; 2) the country and the ethnicity of the studied subjects; 3) the methodology of genotyping; 4) the methodology of DR diagnosis; 5) P-value of HWE in the control group; 6) The genotypic count of each variant in the patient and control groups.
Non-proliferative diabetic retinopathy (NPDR) and proliferative diabetic retinopathy (PDR) are regarded as different stages of DR, which shows different pathology and pathophysiology. This meta-analysis evaluated three types of case groups: (1) DR, (2) only PDR, and (3) combined NPDR, PDR and DR.

Statistical analysis
A publicly available program (https://ihg.gsf.de/cgi-bin/ hw/hwa1.pl) was used to estimate the HWE of the included studies. HWE in the control subjects was evaluated by χ 2 test, and P < 0.05 was considered as deviation from HWE.
The association of UCPs variants with DR was evaluated by five genetic models, including the allelic (reference allele versus variant allele), homozygous (homozygous reference genotype versus homozygous variant genotype), heterozygous (homozygous reference genotype versus heterozygous genotype), dominant (homozygous reference genotype versus homozygous variant and heterozygous genotypes), and recessive models (homozygous reference and heterozygous genotypes versus homozygous variant genotype). Subgroup analyses were also conducted based on the stage of DR and the ethnicity.
Heterogeneity was examined by the Q statistic (significance defined as P < 0.1) and the I 2 statistic (significant inconsistency defined as I 2 > 50%) [29]. If heterogeneity test showed significance (P < 0.1 or I 2 > 50%), the random-effect model was selected to measure the pooled effect value (DerSimonian and Laird method) [30]; otherwise, the fixed-effect model was applied (Mantel-Haenszel method) [31]. The pooled odds ratio (OR) with 95% confidence intervals (C.I.) was calculated to measure the strength of association between the UCPs variants and DR, which was assessed by the Z test (significance defined as P < 0.05). Sensitivity analysis was used to measure the stability of the results by excluding one study at a time when there were more than two studies. Egger's test was used to quantitatively evaluate the potential publication bias. All statistical analysis was calculated by the STATA software (version 14.0; STATA Corporation, College Station, TX).

Studies characteristics
Forty studies were resulted and retrieved from the literature search in the 3 databases. After screening on the abstracts and full-text reports, 11 studies met with the  the allelic model. The quality of the included studies was evaluated by NOS, ranging from 5 to 8, and the overall quality was moderate. The flow chart of the study selection was presented in Fig. 1.
A total of 5 UCPs variants, UCP1 rs1800592, UCP2 rs659366, UCP2 rs660339 (p.A55V), UCP2 45-bp Ins/ Del and UCP2 rs1800849, were identified from the literature search (Table 1 and Supplementary Table 1). Yet, only rs1800592 and rs659366 were subjected to further analysis as they were reported in more than 3   (Table 3 and Fig. 2). Similarly, for UCP2 rs659366, the pooled effect analysis showed no statistically significant association with DR in type-2 DM patients for all five genetic models (Table 3 and Fig. 3).

Evaluation of publication bias and sensitivity analyses
The Egger's test, which was used to quantitatively measure the publication bias, showed no statistically significant publication bias (P > 0.05; Table 5). The results of the sensitivity analysis showed that the pooled OR lied within the 95% C.I. of the total pooled OR (Fig. 5).

Discussion
DR is one of the most common microvascular complications in DM patients. The hyperglycemia-induced ROS is considered as one of the initial and major pathways causing the damage to the endothelial cells. The UCPs are anion carrier proteins in mitochondrial inner membrane. UCPs function to reduce mitochondrial ROS, especially hyperglycemia-induced oxidative stress, and protect endothelial cells from oxidative stress by balancing the proton motive force across the mitochondrial inner membrane [22]. Therefore, UCPs could possibly participate in the development and pathogenesis of DR. This meta-analysis verified the association of the reported UCPs variants with the susceptibility of DR. Our results showed that UCP1 rs1800592 variant was not significantly associated with DR in type-2 DM patients in the pooled effects analysis (Table 3 and Fig. 2); yet, in the subgroup analysis, UCP1 rs1800592 was significantly associated with PDR in type-2 DM patients in the allelic and homozygous models (Table 4 and Fig. 4). The patients carrying allele G of UCP1 rs1800592 variant have 26% higher risk developing PDR than those carrying allele A. A previous study demonstrated that the carriers of rs1800592 GG genotype exhibited higher UCP1 gene expression than those with AA genotype in the retina samples [20]. This could suggest that higher UCP1 expression in retina by allele G of rs1800592 variant could be associated with higher susceptibility of PDR. Conversely, UCP1 expression was lower in carriers of GG genotype than those with AA genotype in intraperitoneal adipose cells, indicated the tissue-specific effect of rs1800592 on UCP1 expression activity [38]. Moreover, allele G of UCP1 rs1800592 also showed elevated expression of MnSOD2 gene, which is another major scavenger for mitochondrial ROS [20,39]. Our discovery was resulted from 2 reported studies, and our approach is similar to that from our previous studies [40][41][42]. Nevertheless, further studies with larger cohorts in different populations are needed to verify its association with PDR.
UCP2 is the most widely distributed uncoupling protein and most frequently studied in DM and DR, and itis associated with the increased oxidative stress and negatively regulates the insulin secretion [43,44]. Total 4 UCP2 variants, UCP2 rs659366, UCP2 rs660339 (p.A55V), UCP2 45-bp Ins/Del and UCP2 rs1800849, were reported in the association analysis with DR; however, only UCP2 rs659366 variant comprised enough studies for the meta-analysis, and other UCP2 variants have not been further analyzed in this study. UCP2 rs659366 has been reported to be associated with type-2 DM [45]; however, in this meta-analysis, we demonstrated that UCP2 rs659366 variant showed no pooled association with DR in the type-2 DM patients (Table 3 and Fig. 3). The elevation of UCP2 expression could be induced by high glucose treatment in epithelial cell of human vein, and the A allele of UCP2 rs659366 increases promoter activity as compared to the G allele, which can be exacerbated under hyperglycemic condition to exert a protective effect [46]. The negative association of UCP2 rs659366 variant with DR in this metaanalysis might indicated that UCP2 gene variation may   [28]. One report showed that the haplotype of 3 differentUCP2 variants [Ins (45-bp Ins/Del), A (rs659366) and Ala (rs660339)] is associated with the decreased UCP2 gene expression in human retina [47]. This could be an independent risk factor for PDR in both type-1 and 2 DM patients [28]. Additional association studies are necessary in order to confirm the association of all 4 UCP2 variants with DR in different ethnic groups. We conducted the subgroup analyses on ethnicity in this meta-analysis. There was no significant association in different ethnic group, which could be due to the limited and sample sizes after stratification. Thus, the ethnicity-specific effects of these variants need to be determined with larger sample sizes in additional cohort studies.
There are several limitations in this meta-analysis. First, the number of reported studies for each UCPs variant was still limited. Second, the lack of original clinical information would be difficult to adjust the relevant variables, such as duration of diabetes, medications and other chronic diseases.

Conclusions
In summary, our meta-analysis revealed no significant pooled association of UCP1 rs1800592 and UCP2 rs659366 with DR in DM patients; yet allele G of UCP1 rs1800592 variant could be associated with the increased risk of PDR in type-2 DM patients. Our results suggest that UCP1 rs1800592 variant could be clinically applied as a genetic marker for PDR prediction and risk analysis in DM clinics.
Additional file 4: Table S1. Characteristics of other UCPs variants. Table  S2. Analysis of the included studies by Newcastle-Ottawa quality assessment scale. Figure S1. Subgroup analysis of UCP1 rs1800592 by ethnicity. Figure S2. Subgroup analysis of UCP2 rs659366 by ethnicity. Figure S3. Subgroup analysis of UCP2 rs659366 by stage of DR. Figure S4. Sensitivity analyses of UCP1 rs1800592 in all genetic models. Figure S5.

Availability of data and materials
All the data supporting our findings is contained within the manuscript.
Ethics approval and consent to participate Not applicable.