Open Life Sci. 2017; 12: 501–509

Research Article

Yunyi Liang, Yi Shu, Haizhao Luo, Riqiu Chen, Zhifu Zeng* Correlation between CDKAL1 rs10946398C>A single nucleotide polymorphism and mellitus susceptibility: A meta-analysis https://doi.org/10.1515/biol-2017-0059 showed people with AA or CA genotype had decreased Received September 19, 2017; accepted October 17, 2017 risk of developing T2DM compared to CC genotype (OR=0.83, P<0.05); For a homozygous genetic model (CC Abstract: Objective The purpose of this meta- vs AA), the OR was calculated through random effect analysis was to assess the correlation between CDKAL1 model for statistical heterogeneity among the included rs10946398C>A single nucleotide polymorphism (SNP) 13 studies. The combined OR was 1.33 which indicated and type 2 diabetes mellitus (T2DM) susceptibility by people with CC genotype had increased risk of developing pooling the open published studies. Method Electronic T2DM. Conclusion According to the present results, searching of PubMed, EMBASE, Medline, Cochrane, CDKAL1 rs10946398C>A single nucleotide polymorphism Biology Medicine disc (CBM) and China National had correlation with the susceptibility of type 2 diabetes Knowledge Infrastructure (CNKI) databases were mellitus. performed by two reviewers independently to collect the open published studies related to CDKAL1 rs10946398C>A Keywords: type 2 diabetes mellitus; CDKAL1 ; single nucleotide polymorphism and T2DM susceptibility. susceptibility; polymorphism; meta-analysis. The association between CDKAL1 rs10946398C>A single nucleotide polymorphism and T2DM susceptibility was expressed by odds ratio (OR) and the corresponding 95% confidence interval (95%CI). Results Thirteen studies with a total of 13,966 T2DM and 14,274 controls were 1 Introduction finally included for analysis in this meta-analysis. Of the Type 2 diabetes mellitus (T2DM) is a metabolic disorder included 13 publications, 2 studies were carried out in caused by insufficient insulin secretion or insulin Europe, 8 in Asia, 2 in Africa and 1 in Latin America. Being resistance [1,2]. Epidemiological studies have suggested significant statistical heterogeneity, the data was pooled that T2DM is a multifactor genetically heterogeneous through random effect model. In a dominant genetic disease that involves in the effects of multiple and model, there was significant correlation between CDKAL1 environmental factors [3,4]; thus, its pathogenesis is rs10946398C>A SNP and T2DM risk (OR=1.22, P<0.05). In a complex and the cause of this disease has not yet been dominant genetic model people with CC or CA genotype fully elucidated. CDKAL1 is a newly identified gene that had increased risk of developing T2DM; Because of is relevant to T2DM pathogenesis [5,6]. However, research statistical heterogeneity for the included studies in a on the relationship between CDKAL1 gene rs10946398 site recessive genetic model (AA+CA vs CC), the data was polymorphism and T2DM is inconclusive [7-9]. Wu et al. calculated by random effect method. The combined data [10] evaluated the correlation between common variants of CDKAL1 gene and type 2 diabetes. The authors found *Corresponding author: Zhifu Zeng, Department of Endocrinology, in Chinese Hans, the common variants of rs10946398C>A People’s Hospital, Lishui Province, 323000 PR China. single nucleotide polymorphism was associated with No.15 Dazhong Road, Liandu Lishui City Zhejiang Province, the susceptibility of type 2 diabetes mellitus. However, 323000 PR China, E-mail: [email protected] Cruz [9] didn’t find the correlations in all genetic models. Yunyi Liang, Yi Shu, Haizhao Luo, Department of Endocrinology, Nanhai People’s Hospital of Foshan City 528200 China In this study, a meta-analysis has been conducted to Riqiu Chen, Department of Endocrinology, Lishui People’s Hospital, comprehensively and quantitatively evaluate previous Lishui Zhejiang Province, 323000 PR China study results to obtain medical evidence on the relationship

Open Access. © 2017 Yunyi Liang et al., published by De Gruyter Open. This work is licensed under the Creative Commons Attribution- NonCommercial-NoDerivs 4.0 License. 502 Y. Liang, et al. between CDKAL1 gene rs10946398 polymorphism and (CC, CA and AA) for CDKAL1 rs10946398 from each T2DM susceptibility. included studies were also extracted. The data was cross examined by the two reviewers to make consensus. A third reviewer (Fuzhi Zeng) was consulted if disagreement was 2 Methods encountered.

2.1 Publication searching 2.4 Statistical analysis Electronic searching of PubMed, EMBASE Medline, Cochrane, China Biology Medicine disc (CBM) and China Stata 11.0 (for meta-analysis) was used to deal with the data National Knowledge Infrastructure (CNKI) databases was analysis. Firstly, the data extracted from each included performed by two reviewers (Yunyi Liang and Yi Shu) publication was tested for statistical heterogeneity independently to collect the open published studies related through chi-square test [11] and the inconsistency was to CDKAL1 rs10946398C>A SNP and T2DM susceptibility. calculated by I2 [12]. If significant statistical heterogeneity The electronic searching words were described as follows: existed among the included studies, the data was CDKAL1 gene, T2DM, susceptibility, rs10946398 and pooled by random effect method otherwise the data was genetic polymorphism. The studies searching process was calculated by fixed effect methods. The odds ratio (OR) done by two reviewers (Yunyi Liang and Yi Shu) and cross was used to demonstrate the correlation between DKAL1 checked. The references of included publications were rs10946398C>A single nucleotide polymorphism and also reviewed to find additional suitable studies. T2DM susceptibility. The publication bias was evaluated by Begg’s funnel plot and Egger’s line regression test. Two tail p≤0.05 was considered statistical significance. 2.2 Publication inclusion and exclusion criteria 3 Results The study inclusion criteria were: ① Open published studies about CDKAL1 rs10946398C>A SNP and T2DM 3.1 General characteristics of the included susceptibility with the language restriction of English thirteen studies and Chinese; ② The study was restrict to case-control or cohort study; ③ The study should provide the genotype After searching the electronic databases, 364 publications distribution; ④ The T2DM patients had confirmed were initially identified. After reading the title, abstract diagnosis according to WHO DM diagnostic criteria; and full text paper, 351 studies were did not meet the ⑤ Subjects in the control group did not have family inclusion criteria and were excluded. 13 studies [7-10, history of DM. The publication exclusion criteria were: ① 13-21] with a total of 13966 T2DM and 14274 controls were Duplicated publications; ② Case report or review study finally included for analysis in this meta-analysis, Figure type; ③ Studies published in other language; ④ The 1. Of the included 13 publications, 2 studies were carried original study did not provided genotype distribution or out in Europe, 8 in Asia, 2 in Africa and 1 in Latin America. enough data to calculate the OR. The detailed information for the included studies were showed in Table 1.

2.3 Data extraction 3.2 Genotype distribution Two reviewers (Haizhao Luo and Riqiu Chen) independently reviewed the whole paper to extract The median CC, CA and AA genotype distribution frequency the necessary information and data. The general for T2DM group were 0.22, 0.49 and 0.28. And for control characteristics of the included studies such as first and group, the median distribution frequency CC, CA and AA corresponding authors, the paper publication year, and genotype were 0.18, 0.48 and 0.34 respectively. There was the race of subjects were extracted and cross checked no statistical difference for the CC, CA and AA genotype by the two reviewers. The data of genotype distribution distribution between the two groups (P>0.05), Figure 2. Correlation between CDKAL1 rs10946398C>A single nucleotide polymorphism... 503

Records identified through Additional records identified database searching through other sources (n = 364 ) (n = 0 ) Identification

Records after duplicates removed (n =360) Screening Records screened (n = 360 )

Records excluded (n = 340)

Full-text articles assessed

Eligibility for eligibility (n = 20 )

Full-text articles excluded, with reasons (n = 7 )

Studies included in quantitative synthesis (meta-analysis)

Included (n = 13 )

Figure 1. Publication inclusion flow-chart of this meta-analysis

Table 1. General information for the included studies

Authors Year Region Race Case Control

CC CA AA CC CA AA

Grarup [13] 2007 Europe Caucasus 184 424 244 159 422 281 Scott [7] 2007 Europe Caucasus 171 549 440 150 539 486 Lewis [14] 2008 Africa African 353 442 138 357 513 185 Wu [10] 2008 Asia East Asian 106 212 106 319 922 666 Hu [8] 2009 Asia East Asian 360 912 578 306 866 613 Cruz [9] 2010 Latin America America 52 225 242 51 224 244 Han [15] 2010 Asia East Asian 227 498 273 177 484 330 Lin [16] 2010 Asia East Asian 310 757 463 195 666 567 Wen [17] 2010 Asia East Asian 226 574 365 175 542 420 Cooke [18] 2012 Africa African 968 1269 416 453 683 258 Chen [19] 2013 Asia East Asian 97 221 125 296 559 264 Al-Sinani [20] 2015 Asia Europa 401 460 131 140 126 28 Sun [21] 2015 Asia East Asian 108 231 108 99 294 215 504 Y. Liang, et al.

3.3 Statistical heterogeneity evaluation 3.4 Dominant genetic model (CC+CA vs AA)

The statistical heterogeneity for the included studies were For significant heterogeneity, the data was pooled evaluate by I2 test. In the dominant genetic model (CC+CA through random effect model. In dominant genetic vs AA), the I2=71.4% indicating significant statistical model, there was significant correlation between heterogeneity was existed. For the recessive genetic model CDKAL1 rs10946398C>A single nucleotide polymorphism (AA +CC vs CC), the statistical heterogeneity was significant (OR=1.22, P<0.05). In dominant genetic model people with I2=73.0%. Significant heterogeneity also found in with CC or CA genotype had increased risk of developing homozygous genetic model (CC vs AA) with the I2=79.7%. DM (Figure 3).

Figure 2. Scatter plot of genotype distribution in DM and control group (a: CC genotype distribution; b: CA genotype distribution; c: AA genotype distribution).

Study % ID OR (95% CI) Weight

Grarup (2007) 1.21 (0.98, 1.48) 7.94 Scott (2007) 1.15 (0.98, 1.36) 8.87 Lewis (2008) 1.23 (0.96, 1.56) 7.15 Wu (2008) 1.61 (1.27, 2.04) 7.18 Hu (2009) 1.15 (1.00, 1.32) 9.50 Cruz (2010) 1.02 (0.80, 1.30) 7.08 Han (2010) 1.33 (1.09, 1.61) 8.26 Lin (2010) 1.52 (1.30, 1.77) 9.19 Wen (2010) 1.28 (1.08, 1.53) 8.71 Cooke (2012) 1.22 (1.03, 1.45) 8.75 Chen (2013) 0.79 (0.61, 1.01) 6.97 Al-Sinani (2015) 0.69 (0.45, 1.06) 3.94 Sun (2015) 1.72 (1.31, 2.26) 6.46 Overall (I-squared = 71.4%, p = 0.000) 1.22 (1.09, 1.35) 100.00

NOTE: Weights are from random effects analysis

.443 1 2.26

Figure 3. Forest plot of OR with a random-effects model for evaluation of CDKAL1 rs10946398C>A single nucleotide polymorphism and type 2 diabetes mellitus susceptibility in dominant genetic model (CC+CA vs AA) Correlation between CDKAL1 rs10946398C>A single nucleotide polymorphism... 505

3.5 Recessive genetic model (AA+CA vs CC) test. For dominant genetic model (CC+CA vs AA). The Begg’s funnel plot was upper and lower symmetry (Figure Because of statistical heterogeneity across the studies in 6). Egger’s line regression test indicated no publications recessive genetic model (AA+CA vs CC), the data was calculated bias (P=0.34). In recessive (AA+CA vs CC), (Figure 7) and by random effect model. The combined data showed people homozygous (CC vs AA), (Figure 8) genetic model the with AA or CA genotype had decreased risk of developing Begg’s funnel plot was seemingly asymmetry. However, the T2DM compared to CC genotype (OR=0.83, P<0.05), Figure 4. Egger’s line regression test demonstrated no publication bias (P=0.72, P=0.36).

3.6 Homozygous genetic model (CC vs AA) 4 Discussion For homozygous genetic model (CC vs AA), the OR was Insulin resistance is the major pathogenesis of T2DM in calculated through random effect model because of humans [22,23]. As a multifactor genetically heterogeneous statistical heterogeneity among the included 13 studies. disease, the pathogenesis and prognosis of T2DM are The combined OR was 1.33 which indicated people with caused by a combination of inherited and acquired CC genotype had increased of developing T2DM, Figure 5. risk factors. Literature on the screening of candidate genes related to T2DM have shown that the variation in some genes related to insulin secretion and blood sugar 3.7 Publications bias metabolism is certainly correlated with the occurrence and development of T2DM [24] . The publication bias of this meta-analysis was evaluated A number of case-control studies have revealed by both Begg’s funnel plot and Egger’s line regression that single-nucleotide polymorphisms in multiple loci

Study % ID OR (95% CI) Weight

Grarup (2007) 0.82 (0.65, 1.04) 7.64 Scott (2007) 0.85 (0.67, 1.07) 7.67 Lewis (2008) 0.84 (0.70, 1.01) 8.71 Wu (2008) 0.60 (0.47, 0.77) 7.38 Hu (2009) 0.86 (0.72, 1.01) 9.01 Cruz (2010) 0.98 (0.65, 1.47) 4.78 Han (2010) 0.74 (0.59, 0.92) 7.99 Lin (2010) 0.62 (0.51, 0.76) 8.47 Wen (2010) 0.76 (0.61, 0.94) 8.05 Cooke (2012) 0.84 (0.73, 0.96) 9.61 Chen (2013) 1.28 (0.99, 1.67) 7.17 Al-Sinani (2015) 1.34 (1.03, 1.74) 7.16 Sun (2015) 0.61 (0.45, 0.83) 6.36 Overall (I-squared = 73.0%, p = 0.000) 0.83 (0.73, 0.93) 100.00

NOTE: Weights are from random effects analysis

.45 1 2.22

Figure 4. Forest plot of OR with a random-effects model for evaluation of CDKAL1 rs10946398C>A single nucleotide polymorphism and type 2 diabetes mellitus susceptibility in recessive genetic model (AA+CA vs CC) 506 Y. Liang, et al.

Study % ID OR (95% CI) Weight

Grarup (2007) 1.33 (1.01, 1.75) 7.87 Scott (2007) 1.26 (0.98, 1.62) 8.10 Lewis (2008) 1.33 (1.02, 1.73) 7.97 Wu (2008) 2.09 (1.55, 2.82) 7.53 Hu (2009) 1.25 (1.03, 1.51) 8.85 Cruz (2010) 1.03 (0.67, 1.57) 6.02 Han (2010) 1.55 (1.20, 2.00) 8.11 Lin (2010) 1.95 (1.57, 2.42) 8.54 Wen (2010) 1.49 (1.17, 1.89) 8.25 Cooke (2012) 1.33 (1.09, 1.60) 8.84 Chen (2013) 0.69 (0.51, 0.95) 7.37 Al-Sinani (2015) 0.61 (0.39, 0.96) 5.73 Sun (2015) 2.17 (1.52, 3.11) 6.82 Overall (I-squared = 79.7%, p = 0.000) 1.33 (1.13, 1.57) 100.00

NOTE: Weights are from random effects analysis

.322 1 3.11

Figure 5. Forest plot of OR with a random-effects model for evaluation of CDKAL1 rs10946398 C>A single nucleotide polymorphism and type 2 diab

Begg's funnel plot with pseudo 95% confidence limits

1

.5

logor

0

-.5 0 .1 .2 s.e. of: logor

Figure 6. The Begg’s funnel plot for evaluation publication bias in dominant genetic model (CC+CA vs AA) Correlation between CDKAL1 rs10946398C>A single nucleotide polymorphism... 507

Begg's funnel plot with pseudo 95% confidence limits

.5

0

logor

-.5

0 .05 .1 .15 .2 s.e. of: logor

Figure 7. The Begg’s funnel plot for evaluation publication bias in recessive genetic model (AA+CA vs CC)

Begg's funnel plot with pseudo 95% confidence limits

1

.5

logor

0

-.5 0 .1 .2 .3 s.e. of: logor

Figure 8. The Begg’s funnel plot for evaluation publication bias in Homozygous genetic model (CC vs AA) 508 Y. Liang, et al.

paternal and maternal lineages[J]. J Community Health, of CDKAL1 gene are related to T2DM pathogenesis[9, 13, 2009,34(6):553-85. 16]. The risk that the population with a specific genotype [4] Sheu WH, Song YM, Lee WJ, et al. Family aggregation and suffers from T2DM is significantly increased. However, maternal inheritance of Chinese type 2 diabetes mellitus in some studies have reported that the single-nucleotide Taiwan[J]. Zhonghua Yi Xue Za Zhi (Taipei), 1999,62(3):146-51. polymorphism of CDKAL1 genes is not associated with [5] Nikitin AG, Potapov VY, Brovkina OI, et al. Association of susceptibility to T2DM[9]. CDKAL1 gene is located on the polymorphic markers of genes FTO, KCNJ11, CDKAL1, SLC30A8, and CDKN2B with type 2 diabetes mellitus in the Russian short arm of 6, with a full length of 697948bp, population[J]. PeerJ, 2017,5:e3414. encoding a 65ku protein. The changes in CDKAL1 gene [6] Kommoju UJ, Samy SK, Maruda J, et al. Association of CDKAL1, expression can lead to the decrease in ATP and cause a CDKN2A/B & HHEX gene polymorphisms with type 2 diabetes mellitus in the population of Hyderabad, India[J]. Indian J Med KATP channel response induced by glucose to weaken. The activity of Ca2+ channel is then impaired, and first-phase Res, 2016,143(4):455-63. [7] Scott LJ, Mohlke KL, Bonnycastle LL, et al. A genome-wide insulin secretion is decreased. Studies have also shown association study of type 2 diabetes in Finns detects multiple that CDKAL1 can inhibit the activity of pancreatic β cells susceptibility variants[J]. Science, 2007,316(5829):1341-5. CDK5 under high glucose toxicity and prevent the decline [8] Hu C, Zhang R, Wang C, et al. PPARG, KCNJ11, CDKAL1, CDKN2A- in insulin gene caused by sugar toxicity[25]. Thus, insulin CDKN2B, IDE-KIF11-HHEX, IGF2BP2 and SLC30A8 are associated secretion is decreased, and the insulin secretion of people with type 2 diabetes in a Chinese population[J]. PLoS One, with rs10946398 dangerous allele after glucose load is 2009,4(10):e7643. [9] Cruz M, Valladares-Salgado A, Garcia-Mena J, et al. Candidate also decreased. CDKAL1 gene influencing the secretion gene association study conditioning on individual ancestry in function of pancreatic β cells is related to insufficient patients with type 2 diabetes and metabolic syndrome from insulin secretion but not to insulin resistance[26]. Mexico City[J]. Diabetes Metab Res Rev, 2010,26(4):261-70. In this study, meta-analysis is conducted for 13 [10] Wu Y, Li H, Loos RJ, et al. Common variants in CDKAL1, studies of CDKAL1 rs10946398C>A SNP from different CDKN2A/B, IGF2BP2, SLC30A8, and HHEX/IDE genes are associated with type 2 diabetes and impaired fasting glucose in regions. Results showed that population carrying C allele a Chinese Han population[J]. Diabetes, 2008,57(10):2834-42. has a significantly increasing risk of suffering from T2DM. [11] DerSimonian R, Laird N. Meta-analysis in clinical trials[J]. Thereafter, CDKAL1 rs10946398C>A single nucleotide Control Clin Trials, 1986,7(3):177-88. polymorphism had correlation with the susceptibility of [12] Higgins JP, Thompson SG, Deeks JJ, et al. Measuring type 2 diabetes mellitus. inconsistency in meta-analyses[J]. BMJ (Clinical research ed.), However, meta-analysis also has limitations. First, 2003,327(7414):557-60. [13] Grarup N, Rose CS, Andersson EA, et al. Studies of association statistical heterogeneity exists among the included of variants near the HHEX, CDKN2A/B, and IGF2BP2 genes with various studies. The presence of statistical heterogeneity type 2 diabetes and impaired insulin release in 10,705 Danish reduces the stability of conclusions. Second, meta- subjects: validation and extension of genome-wide association analysis is only conducted for one of CDKAL1 gene SNP studies[J]. Diabetes, 2007,56(12):3105-11. loci, but the interaction among polymorphic loci has not [14] Lewis JP, Palmer ND, Hicks PJ, et al. Association analysis in african americans of European-derived type 2 diabetes single been explored yet. nucleotide polymorphisms from whole-genome association studies[J]. Diabetes, 2008,57(8):2220-5. Ethical approval: The conducted research is not related [15] Han X, Luo Y, Ren Q, et al. Implication of genetic variants near to either human or animals use. SLC30A8, HHEX, CDKAL1, CDKN2A/B, IGF2BP2, FTO, TCF2, KCNQ1, and WFS1 in type 2 diabetes in a Chinese population[J]. Conflict of interest: Authors state no conflict of interest BMC Med Genet, 2010,11:81. [16] Lin Y, Li P, Cai L, et al. Association study of genetic variants in eight genes/loci with type 2 diabetes in a Han Chinese population[J]. BMC Med Genet, 2010,11:97. Reference [17] Wen J, Rönn T, Olsson A, et al. Investigation of type 2 diabetes risk alleles support CDKN2A/B, CDKAL1, and TCF7L2 as [1] Szendroedi J, Phielix E, Roden M. The role of mitochondria susceptibility genes in a Han Chinese cohort[J]. PLoS One, in insulin resistance and type 2 diabetes mellitus[J]. Nat Rev 2010,5(2):e9153. Endocrinol, 2011,8(2):92-103. [18] Cooke JN, Ng MC, Palmer ND, et al. Genetic risk assessment [2] Małecki MT. [--insulin resistance--type 2 diabetes of type 2 diabetes-associated polymorphisms in African mellitus][J]. Kardiol Pol, 2006,64(10 Suppl 6):S561-6. Americans[J]. Diabetes Care, 2012,35(2):287-92. [3] Imperato PJ, Imperato GH. The role of genealogy and clinical [19] Chen G, Xu Y, Lin Y, et al. Association study of genetic variants family histories in documenting possible inheritance patterns of 17 diabetes-related genes/loci and cardiovascular risk for diabetes mellitus in the pre-insulin era: part 2. Genealogic and diabetic nephropathy in the Chinese She population[J]. J evidence for type 2 diabetes mellitus in Josephine Imperato’s Diabetes, 2013,5(2):136-45. Correlation between CDKAL1 rs10946398C>A single nucleotide polymorphism... 509

[20] Al-Sinani S, Woodhouse N, Al-Mamari A, et al. Association of [25] Brambillasca S, Altkrueger A, Colombo SF, et al. CDK5 gene variants with susceptibility to type 2 diabetes among regulatory subunit-associated protein 1-like 1 (CDKAL1) is a Omanis[J]. World J Diabetes, 2015,6(2):358-66. tail-anchored protein in the endoplasmic reticulum (ER) of [21] Sun XF, Xiao XH, Zhang ZX, et al. Positive Association Between insulinoma cells[J]. J Biol Chem, 2012,287(50):41808-19. Type 2 Diabetes Risk Alleles Near CDKAL1 and Reduced [26] Lyssenko V, Jonsson A, Almgren P, et al. Clinical risk factors, Birthweight in Chinese Han Individuals[J]. Chin Med J (Engl), DNA variants, and the development of type 2 diabetes[J]. N Engl 2015,128(14):1873-8. J Med, 2008,359(21):2220-32. [22] Czech MP. Insulin action and resistance in obesity and type 2 diabetes[J]. Nat Med, 2017,23(7):804-814. [23] Gougeon R. Insulin resistance of protein metabolism in type 2 diabetes and impact on dietary needs: a review[J]. Can J Diabetes, 2013,37(2):115-20. [24] Barroso I, Luan J, Middelberg RP, et al. Candidate gene association study in type 2 diabetes indicates a role for genes involved in beta-cell function as well as insulin action[J]. PLoS Biol, 2003,1(1):E20.