Association between Genetic Polymorphisms and Risk of Kidney Posttransplant Diabetes Mellitus: A Systematic Review and Meta-Analysis

Objectives The purpose of this study was to clarify the role of genetic factors on posttransplant diabetes mellitus (PTDM) risk. Methods Relevant publications were systematically retrieved from PubMed, EMBASE, and the Cochrane Library up to December 2020. Data from eligible case-control and cohort studies were extracted for qualitative and quantitative analyses. Odds ratios (ORs) and 95% confidence intervals (CIs) were used to estimate the association between gene polymorphisms and PTDM in the quantitative meta-analysis. Results A total of 43 eligible articles were identified, and 16 studies on 9 DNA variants from 8 genes were included in the meta-analysis. TCF7L2 rs7903146 was significantly associated with PTDM risk in 5 genetic models (OR (95% CI): allelic: 1.59 (1.17–2.16), P=0.003; dominant recessive: 1.62 (1.14, 2.31), P=0.007; recessive: 1.87 (1.18, 2.94), P=0.007; homozygote: 2.21 (1.23, 3.94), P=0.008; and heterozygote 1.50 (1.08, 2.10), P=0.017). KCNQ1 rs2237892 was significantly correlated with PTDM risk in 3 genetic models (allelic: 0.68 (0.58, 0.81), P < 0.001; dominant: 0.6 (049, 0.74), P < 0.001; and heterozygote: 0.61 (0.48, 0.76), P < 0.001). KCNJ11 rs5219 was significantly linked with PTDM in the recessive genetic model (1.59 (1.01, 2.50), P=0.047). No significant correlations of PTDM with TCF7L2 rs12255372, SLC30A8 rs13266634, PPARγ rs1801282, CDKN2A/B rs10811661, HHEX rs1111875, and IGF2BP2 rs4402960 polymorphisms were found. Conclusions The gene polymorphisms of TCF7L2 rs7903146, KCNQ1 rs2237892, and KCNJ11 rs5219 may predispose kidney transplant recipients to PTDM. Large sample size studies on diverse ethnic populations were warranted to confirm our findings.


Introduction
PTDM is a common serious complication after kidney transplantation, which is often associated with increased risk of graft failure, cardiovascular disease, and mortality [1]. Approximately 5.5% to 60.2% of kidney transplant patients develop PTDM in the first year after surgery [2]. A large retrospective study involving 11,659 kidney recipients from the United States Renal Data System (USRDS) demonstrated that the cumulative incidence of PTDM was 9.1%, 16%, and 24% at 3 months, 12 months, and 36 months, respectively [3]. Its etiopathogenesis is multifactorial, and transplantrelated risk factors for PTDM include immunosuppressants, ethnicity, age, sex, body mass index, genetic factors, hepatitis C and cytomegalovirus infections, and family history of diabetes [2]. Immunosuppressive drugs consisting of corticosteroids and calcineurin inhibitors are important risk factors of PTDM, contributing to the development of hyperglycemia and diabetes [4]. Tacrolimus (TAC) and cyclosporin (CsA) are two major calcineurin inhibitors required after transplantation to prevent acute or chronic graft rejections [1]. e mechanisms underlying the diabetogenic effect of immunosuppressive regimen include enhancing insulin resistance, reducing insulin secretion, and direct toxic effects on pancreatic β-cells [4]. It has also been suggested that glucocorticoid-induced hyperglycemia is partially reversible through avoidance or early withdrawal of the drugs [5].
More evidence suggests that genetic risk factors play a significant role in the development of PTDM. Many genes associated with diabetes mellitus (DM) have also been correlated with PTDM risk. Gene mutations such as single nucleotide polymorphisms (SNPs) are the most common type of genetic variation. SNPs of TCF7L2 rs7903146, TCF7L2 rs12255372, KCNQ1 rs2237892, KCNJ11 rs5219, SLC30A8 rs13266634, PPARc rs1801282, CDKN2A/B rs10811661, HHEX rs1111875, and IGF2BP2 rs4402960 have recently been detected and shown to affect PTDM occurrence. Among them, TCF7L2 rs7903146 had an established strong effect across different populations and is the most common susceptible gene for PTDM [6][7][8][9][10][11][12]. One previous meta-analysis assessed the potential association between TCF7L2 rs7903146 polymorphism and PTDM [13]. However, there was a lack of systematic review on the correlation between other genes polymorphisms and PTDM. e metaanalysis by Benson et al. evaluated the allelic distribution of 18 gene polymorphisms in PTDM development [14]. In this study, we included several updated articles and comprehensively examined the association of nine SNPs from eight genes including TCF7L2, KCNQ1, KCNJ11, SLC30A8, PPARc, CDKN2A/B, HHEX, and IGF2BP2 with PTDM risk in all allelic and genotype models. Moreover, we reviewed the literature on genetic SNP markers susceptible to PTDM, which might help predict the risk of PTDM and facilitate the early prevention of this disease.

Literature Search.
According to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline (see Supplementary Materials), we systematically searched PubMed, EMBASE, and the Cochrane Library for studies published up to December 2020.

Eligibility Criteria.
e inclusion criteria included (1) kidney transplant recipients diagnosed with new-onset diabetes after transplantation (NODAT) or PTDM according to ADA or WHO guideline, (2) original studies examining the relationship between the gene polymorphism and NODAT or PTDM in patients after kidney transplantation, (3) study type: cohort or case-control studies, and (4) language restricted to English.

Search Strategy.
When searching for possible eligible studies in the PubMed, EMBASE, and Cochrane Library databases, we used the mesh term of "kidney transplantation," "polymorphism, genetic," "posttransplant diabetes mellitus," and "new-onset diabetes mellitus after transplantation," as well as relevant keywords.

Data Extraction and Quality Assessment.
e selection and inclusion of studies were performed in two stages by two independent reviewers, which included the analysis of titles/ abstracts followed by the full texts. Disagreements were resolved by a third reviewer. Data retrieved from the eligible studies consisted of main demographical and clinical variables, including names of authors, publication year, study design, country, ethnicity, mean age, mean BMI, female percentage, genetic risk factors for PTDM, genotyping method and genotypes, diagnosis of PTDM, immunosuppressive therapy, time of PTDM diagnosis after transplantation, and age at transplant. We selected SNPs that showed significant associations with PTDM in allelic and/or genotype models from individual studies. e outcome was the evaluation of the impact of SNPs on the development of PTDM. Excel spreadsheet was used for the collection of extracted data. e methodological quality of included studies was evaluated by NOS. e base information was shown in Supplementary Table 1, and data used for all analyses were shown in Supplementary Table 2.
Cochran's Q statistic was used to assess the heterogeneity among studies (P < 0.10 indicated evidence of heterogeneity; https://doi.org/10.1136/bmj.327.7414.557). When significant heterogeneity (P < 0.10) was achieved, the randomeffects model was used to combine the effect sizes of the included studies; otherwise, the fixed-effects model was adopted [15]. In addition, sensitivity analyses were performed to identify the effects of individual studies on pooled results and test the reliability of the estimates. All statistical analyses were performed using the STATA SE 14.0 software (StataCorp, College Station, Texas, USA).

Study Selection and Characteristics of Included Studies.
A total of 173 relevant publications were identified through searching the databases and other resources. After initial screening, duplicated documents; conference abstracts; reviews; publications on unrelated diseases, transplants, and interventions; and articles without full text were removed. e remaining 62 publications were assessed carefully; then 19 articles were excluded due to insufficient data. Finally, 43 eligible studies were included for the qualitative analysis. Among them, the data from 16 studies were retrieved for the quantitative meta-analysis. e study screening flow chart was shown in Figure 1. e characteristics of the selected studies for qualitative analysis were summarized in Table 1, which covered a total of 2,849 PTDM patients and 9,816 non-PTDM patients after undergoing renal transplantation. e overall incidence of PTDM varied from 8% to 42% at 3 months after transplantation and from 17% to 46% at 12 months. ere were 40 retrospective or prospective cohort studies, and the rest were all retrospective case-control studies. Except for the study by Kao [16], most patients received a TAC-based treatment regimen, mainly combined with CsA, MMF, or steroid. Generally, the diagnosis of PTDM was in accordance with ADA or WHO guidelines. e mean age of patients at transplantation was 35.4 to 60 years old. e follow-up time after transplantation ranged from 1 to 36 months. e quantitative meta-analysis consisted of 16 studies involving 1,455 PTDM patients and 4,483 non-PTDM patients.

Quality Assessment.
e quality assessment of included studies using NOS was shown in Table 2, with the maximum of 9 points representing the least risk of bias. Overall, the methodological quality scores were 9 for 24 studies, 8 for 13 studies, 7 for 4 studies, and 6 for the other 2 studies, suggesting moderate to low risk of bias. e majority of the studies in the meta-analysis had a very low bias. Among them, 12 studies were assigned 9 points; 3 studies received 8 points; and only 1 study got 7 points.
CDKN2A/B rs10811661 polymorphism was also not shown to be related with PTDM risk in all five genetic models (OR (95% CI): allelic: 1. 10 Figure 4(b) and Table 3 Figure 4(c) and Table 3).
In addition, the overall analysis revealed that KCNJ11 rs5219 polymorphism was significantly associated with PTDM risk in the recessive genetic model (OR (95% CI):  Table 3).

Sensitivity Analysis.
For meta-analyses on the association of three gene polymorphisms including TCF7L2 rs7903146, SLC30A8 rs13266634, and PPARc rs1801282 with PTDM risk, the sensitivity analysis results showed that in all five genetic models, the reestimated ORs were all similar to the overall effects when excluding any individual study and assessing the remaining ones ( Supplementary  Figures 1-3).

Discussion
Genetic factors have been increasingly considered to play an important role in the pathogenesis of PTDM. is metaanalysis showed that gene polymorphisms of TCF7L2 rs7903146, KCNQ1 rs2237892, and KCNJ11 rs5219 contributed to PTDM occurrence and development. e genetic variations of TCF7L2 rs12255372, SLC30A8 rs13266634, PPARc rs1801282, CDKN2A/B rs10811661, HHEX rs1111875, and IGF2BP2 rs4402960 SNPs were not found to be associated with PTDM risk.
Previous studies indicated that these nine gene SNPs were associated with T2DM. Many genes associated with International Journal of Clinical Practice T2DM have also been associated with an increased risk of PTDM. T2DM and PTDM were thought to share certain common pathophysiological processes. Impaired insulin secretion and increased insulin resistance have been suggested as mechanisms underlying the development of PTDM. One of the most intensively studied genes was TCF7L2. TCF7L2, a key component of the Wnt signaling pathway, is involved in the regulation of pancreatic β-cell proliferation, differentiation, and insulin secretion [6,10]. Two common SNPs, rs7903146 and rs12255372, were located in TCF7L2 introns 3 and 4, respectively. TCF7L2 rs7903146 C/T emerged as the most common susceptible gene for T2DM in genome-wide association studies (GWAS) [2,51]. Its association with PTDM has been well demonstrated in Asian (Indian and Korean), White, and Caucasian populations [6][7][8][9][10][11][12]. e T allele mutation at TCF7L2 rs7903146 loci has been linked with impaired insulin secretion and hepatic insulin resistance. e results of the association between TCF7L2 rs12255372G/T and PTDM remained conflicting [6,11,12]. TCF7L2 rs7903146 and rs12255372 haplotype analyses did not reveal any significant association with PTDM [11].
KCNQ1 encodes a subunit of the voltage-gated K + channel. It is expressed in the pancreas and may help regulate the membrane potential of insulin-secreting cells and is involved in triggering and maintaining glucosestimulated insulin secretion [25,43]. Although this metaanalysis suggested the susceptibility of the most common KCNQ1 rs2237892 SNP to PTDM, opposite effects of KCNQ1 rs2237892 polymorphism have been discussed. Hwang et al. showed that KCNQ1 rs2237892C/T, located in intron 15, was significantly associated with decreased risk of PTDM in both allelic and genotype models, suggesting a protective effect on the development of PTDM [20]. Kang et al. reported that the T allele of KCNQ1 rs2237892 was correlated with a high risk of PTDM in an allele-specific manner [8]. e pooled analysis of KCNJ11 genes suggested its role in the pathogenesis of PTDM. ATP-sensitive potassium channel KCNJ11 plays an important role in the regulation of insulin secretion by pancreatic β cells, as well as glucose metabolism. KCNJ11 rs5219 glutamic acid to lysine amino acid substitution reduces potassium channels' sensitivity to ATP molecules, resulting in overactivity of the channel and subsequent inhibition of

Study
Representativeness of the exposed cohort Selection of the nonexposed cohort

Study
Representativeness of the exposed cohort Selection of the nonexposed cohort Yang (2011) Khan (2015) Alagbe (2017  e meta-analysis of the Asian Indian population showed no significant association of KCNJ11 rs5219 polymorphism with risk of T2DM [52]. However, other meta-analyses demonstrated a significant effect of KCNJ11 rs5219 in susceptibility to T2DM in East Asians, Caucasians, and North Africans [53]. Controversial results have been reported for the association of SLC30A8, PPARc, CDKN2A/B, HHEX, and IGF2BP gene polymorphisms with PTDM. In this overall analysis, these extensively evaluated genes were not found to contribute to the development of PTDM. SLC30A8 belongs to the zinc transporter family, which plays a major role in transporting zinc from the cytoplasm to intracellular vesicles for insulin maturation, storage, and secretion from β-cells [7,8,10,22,38,50]. e SLC30A8 rs13266634 arginine to tryptophan variant, associated with impaired β-cell function, has been proposed as important genetic markers of T2DM in Europeans and East Asians but not the African population [54,55]. PPARc gene belongs to the nuclear hormone receptor subfamily that controls the expression of genes involved in glucose and lipid homeostasis. e SNP rs1801282 (C/ G) is the most common variant located in exon-2 of PPARc, and the substitution of proline to alanine of PPARc reduces its transcriptional activity and insulin sensitivity [7,12,21,38,41]. One meta-analysis suggested that PPARc rs1801282 was significantly associated with T2DM under the heterozygote genetic model in Asian and Caucasian populations [56]. CDKN2A/B, which encodes two kinase inhibitors p16INK4a and p15INK4b, regulates pancreatic β-cell regeneration. e locus rs10811661 locates ∼100 kb upstream of CDKN2A/B gene-coding sequence, but the mechanism by which this SNP affects T2DM and PTDM susceptibility remains to be investigated [7,8,22,38]. HHEX gene encodes a   transcription factor involved in hepatic and pancreatic development via the Wnt signal pathway [7,8,22,38]. e SNP rs1111875 at the 3′-flanking region of the HHEX gene, which may decrease pancreatic beta-cell function, is reported to be associated with T2DM risk as lead SNP in Chinese Han and European populations [57]. e metaanalysis of IGF2BP2 rs4402960 suggested a significant association with T2DM in Asian populations [58]. e mRNA-binding protein IGF2BP2 is highly expressed in pancreatic islets and participates in a spectrum of the biological process including cellular metabolism.
McCaughan et al. examined in GWAS the association between PTDM and 26 gene SNPs in the White population [59].
is study had several limitations. First, the etiopathogenesis of PTDM was multifactorial. Immunosuppressive regimen, ethnicity, older age, sex, BMI, and other related clinical characteristics contributed significantly to the risk of PTDM. However, crude estimates of effect were often used to evaluate the association between genes polymorphisms and PTDM without adjustments for other confounding variables. Second, PTDM in kidney recipients occurred mainly during the first months. Additionally, there could be a reversible phenotype change from PTDM to non-PTDM. In this study, there was high heterogeneity regarding the observational follow-up time after renal transplantation, which varied from 3 to 12 months among the studies. ird, treatment modality varied greatly for different studies, which may substantially influence the overall incidence of PTDM. Fourth, certain minor allele frequencies (MAF) differed greatly in different races. e sample size in some studies might be too small to detect minor effects, and some study populations presented with various genetic backgrounds. Furthermore, for most studies, it is unclear whether there was preexisting impaired glucose tolerance, which may affect the estimated incidence of PTDM.
Our meta-analysis revealed a significant association between PTDM and gene polymorphisms of TCF7L2 rs7903146, KCNQ1 rs2237892, and KCNJ11 rs5219. Furthermore, we reviewed the literature on available gene SNPs that were susceptible to PTDM. e regulatory mechanism of relevant genes SNPs in the occurrence and development of PTDM was worthy of further exploration. SNPs showing association may serve as genetic markers for the prediction of the development of PTDM, combined with other risk factors of PTDM. Alternate medication of diabetogenic drugs may be considered for early prevention of PTDM based on risk assessment. Further large sample studies with diverse race populations are necessary to confirm our findings.

Data Availability
Since it is a meta-analysis, all data were extracted from public databases, and all data were available in Supplementary Materials.