Clinical Care/Education/Nutrition/Psychosocial Research ORIGINAL ARTICLE

Ser1369Ala Variant in ABCC8 Is Associated With Antidiabetic Efficacy of Gliclazide in Chinese Type 2 Diabetic Patients

1 3 YAN FENG, MD, PHD XUEQI LI, MD he epidemic of in the 1 4 GUANGYUN MAO, MD, PHD LIRONG SUN, MD last decade in both developed and 1 5 XIAOWEI REN, MD JINQUI YANG, MD, PHD 1 6 developing countries has made it a OUXUN ING MD EIQING A MD T H X , W M , 1 7 major threat to global public health. At GENFU TANG, MD XIAOBIN WANG, MD, SCD 2 1 least 171 million people worldwide had QIANG LI, MD, PHD XIPING XU, MD, PHD diabetes in 2000, and this figure is likely to more than double by 2030 to reach 366 million (1). The majority of diabetes is OBJECTIVE — The purpose of this study was to investigate whether genetic variants could type 2 diabetes. Most of the recent rise in influence the antidiabetic efficacy of gliclazide in type 2 diabetic patients. diabetes prevalence is probably a result of lifestyle and dietary changes, but there is RESEARCH DESIGN AND METHODS — A total of 1,268 type 2 diabetic patients also clear evidence for genetic predisposi- whose diabetes was diagnosed within the past 5 years and who had no recent hypoglycemic tion to this complex disease. During the treatment were enrolled from 23 hospitals in China. All of the patients were treated with last decade, molecular genetic studies of gliclazide for 8 weeks. Fasting and oral glucose tolerance test 2-h plasma glucose, fasting , type 2 diabetes have shown significant and A1C were measured at baseline and after 8 weeks of treatment. We used two independent progress (2). Five genome-wide associa- cohorts to test the associations of 25 single nuclear polymorphisms in 11 candidate with the antidiabetic efficacy of gliclazide. A general linear regression model was used to test the tion studies have been published since association with adjustment for important covariates. February 2007, increasing the number of confirmed type 2 diabetes susceptibility loci from three (PPARG, KCNJ11, and RESULTS — After 8 weeks of gliclazide therapy, mean fasting plasma glucose (FPG) was TCF7L2) to 9 (with the addition of reduced from 11.1 mmol/l at baseline to 7.7 mmol/l. In cohort 1, we genotyped all 25 SNPs (n ϭ 661) and found that Ser1369Ala of the ABCC8 gene and rs5210 of the KCNJ11 gene were CDKAL1, CDKN2A/B, IGF2BP2, HHEX/ significantly associated with decreases in FPG (P ϭ 0.002). We further genotyped Ser1369Ala in IDE, FTO, and SLC30A8) (2). In addition, cohort 2 (n ϭ 607) and confirmed the association identified in cohort 1. In the pooled analysis, studies have lent support for the involve- compared with subjects with the Ser/Ser genotype, subjects with the Ala/Ala genotype had a ment of many other genes, such as ABCC8 7.7% greater decrease in FPG (P Ͻ 0.001), an 11.9% greater decrease in 2-h plasma glucose (P ϭ (3–6). In contrast, few studies have inves- 0.003), and a 3.5% greater decrease in A1C (P ϭ 0.06) after 8 weeks of treatment with gliclazide. tigated whether genetic variants may modulate the response to antidiabetic CONCLUSIONS — In two independent cohorts of Chinese type 2 diabetic patients, we agents in type 2 diabetic patients (7–9). found consistent evidence that the Ser1369Ala variant in the ABCC8 gene can influence the Such information can assist clinicians in antidiabetic efficacy of gliclazide. developing individualized treatment plans that will maximize therapeutic effi- Diabetes Care 31:1939–1944, 2008 cacy and minimize side effects. ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● Sulfonylurea is a widely used oral hy- From the 1Anhui Biomedical Institute, Anhui Medical University, Hefei, China; the 2Division of Endocri- poglycemic agent. Most type 2 diabetic nology, the Second Hospital, Harbin Medical University, Harbin, China; the 3Division of Endocrinology, patients respond well to this agent, but the First Hospital, Harbin Medical University, Harbin, China; 4Tianjin Medical University Metabolic variable efficacy is seen, and primary fail- 5 Hospital, Tianjin, China; the Division of Endocrinology, Tongren Hospital, Capital Medical University, ure to sulfonylurea treatment is seen in a Beijing, China; the 6Division of Endocrinology, the First Hefei People’s Hospital, Hefei, China; and the 7The Mary Ann and J. Milburn Smith Child Health Research Program, Children’s Memorial Hospital and small portion of patients. Secondary fail- Children’s Memorial Research Center, Chicago, Illinois. ure of sulfonylurea monotherapy devel- Corresponding author: Xiping Xu, [email protected]. ops in ϳ34% of patients at 5 years (10). Received 26 November 2007 and accepted 21 June 2008. To explore the underlying genetic factors Published ahead of print at http://care.diabetesjournals.org on 3 July 2008. DOI: 10.2337/dc07-2248. Y.F. and G.M. contributed equally to this study. that may explain individual variable re- Y.F. is currently affiliated with Bridgeport Hospital, Yale University School of Medicine, New Haven, Con- sponse to sulfonylurea, we conducted a necticut. hospital-based pharmacogenetic study. © 2008 by the American Diabetes Association. Readers may use this article as long as the work is properly Our goal was to examine whether type 2 cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons. diabetes candidate gene variants can in- org/licenses/by-nc-nd/3.0/ for details. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby fluence the antidiabetic efficacy of glicla- marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. zide, a commonly used sulfonylurea

DIABETES CARE, VOLUME 31, NUMBER 10, OCTOBER 2008 1939 Ser1369Ala variant in ABCC8 and gliclazide hypoglycemic agent, in Chinese type 2 di- Data collection and clinical our central laboratory. The insulin secre- abetic patients. laboratory methods tion (HOMA-B) and resistance index of At the first visit (screening), a standard homeostatic model assessment (HOMA- questionnaire was administered to collect IR) were calculated according to fasting RESEARCH DESIGN AND information on medical history and med- glucose and insulin level using the METHODS — We conducted a hospi- ication, diet, exercise, and lifestyle factors HOMA2 calculator (http://www.dtu.ox. tal-based pharmacogenetic study of gli- including smoking and alcohol drinking, ac.uk/homa). clazide in type 2 diabetic patients in household income, educational level, and China between December 2003 and Au- occupation. Height, weight, waist cir- Candidate gene and SNP selection gust 2005. Patients were recruited from cumference, and blood pressure were and genotyping method 23 hospitals located in Harbin of the Hei- measured using a standard protocol. We selected 11 type 2 diabetes candidate Ͼ longjiang province, Beijing, Tianjin, and Overnight ( 10 h) fasting blood samples genes on the basis of the published liter- Hefei and Anqing of the Anhui province were collected to determine FPG, insulin, ature (see Table 2). For each gene, one to following the same study protocol. The A1C, lipid profile, liver and renal func- six nonsynonymous or haplotype-tagging 23 hospitals selected were the major hos- tion, and routine blood cell counts. All SNPs, according to the HapMap data study patients underwent a 75-g oral glu- (http://www.hapmap.org/), were se- pitals in the study regions. These hospi- cose tolerance test (OGTT) unless FPG lected. In our central laboratory, we geno- tals were all government owned but were was Ն13.0 mmol/l, and blood samples typed all 25 selected SNPs for all of the run independently. were collected after2htodetermine subjects in cohort 1, which consisted of Type 2 diabetes was diagnosed ac- plasma glucose. Diabetes education was 661 patients from 12 participating hospi- cording to American Diabetes Association also provided to all potentially eligible tals located in northern China. We further criteria (11). To reduce the clinical heter- subjects during the screening visit. The genotyped one significant nonsynony- ogeneity of type 2 diabetes, this study was education included a brief introduction mous SNP (rs757110) identified in co- limited to Han Chinese subjects with on- on type 2 diabetes, with a particular focus hort 1 for all subjects in cohort 2, which set of type 2 diabetes after the age of 35 on diet and physical exercise. In addition, consisted of 607 patients from the re- years who also met all of the following a handbook with more detailed informa- maining 11 participating hospitals lo- criteria: 1) diabetes diagnosed within the tion was given to the subjects for them to cated in southern China. The reasons for past 5 years and no antidiabetes treatment read. the two-phase genotyping were twofold: within the past 2 months, 2)BMIϽ28 All study patients returned for fol- reduce the genotyping cost and minimize kg/m2, and 3) fasting plasma glucose low-up every 2 weeks. A follow-up ques- the multiple testing problem, which (FPG) between 7.8 and 15.0 mmol/l. We tionnaire was administered to monitor could inflate type I error. excluded patients with any acute or patients’ medications, diet, exercise, and DNA was extracted from leukocytes chronic diabetes complications, unstable side effects of gliclazide. Specifically, the in peripheral blood using standard tech- angina, myocardial infarction or heart research staff documented whether the niques. Genotyping was performed by failure, chronic gastrointestinal disease or subjects complied with the treatment and TaqMan genotyping assays that were de- abnormal liver function, renal insuffi- followed the instruction for diet (with a signed and manufactured by Applied Bio- ciency, or clinical problems potentially grade of very good, fair, and poor) and systems (Foster City, CA). causing hyperglycemia, including infec- exercise time. The side effects of gliclazide Statistical analysis. All data analyses tion, thyroid disease, or surgery. We also that we monitored in our study included were performed using SAS (version 8; excluded those taking other medications, hypoglycemia, abnormal liver function, SAS Institute, Cary, NC). The phenotype Ϯ such as corticosteroids or estrogen, as skin rash, or any other symptoms re- data are shown as mean SD and the well as cancer patients and pregnant or ported by patients. differences between groups were tested breast-feeding women. FPG (at the 29th day) or fasting fin- using t tests or one-way ANOVA. The as- Subjects were enrolled from the out- gerstick glucose (at the 13th and 43rd sociations between quantitative pheno- patient clinics of the participating hospi- days) were measured for each subject. At types (FPG, 2-h plasma glucose, or A1C the last visit (57th day), blood was drawn decrease [percent]) and genotypes were tals. After giving written informed at fasting and 2 h after the OGTT to repeat tested using linear regression models. The consent, patients started an 8-week treat- all of the tests that the patients had at their associations between binary phenotypes ment with gliclazide (Tianjin Huajin first visit. (response/nonresponse) and genotypes Pharmaceutical Company, Tianjin, Plasma glucose, serum lipids, and were tested using a logistic repression China). The initial dose of gliclazide was liver and renal function were measured at model. All analyses were performed with 40 mg twice daily. The patients continued the local hospitals using an automatic an- or without adjustment for age, sex, BMI, their initial dosage throughout the 8 alyzer (Hitachi 7020; Hitachi, Tokyo, Ja- total gliclazide dosage, and baseline weeks of treatment or had their dose in- pan, or a similar model). The A1C values HOMA-B and HOMA-IR. All P values creased to 80 mg twice daily if FPG was were determined with a high-perfor- were two tailed. Ն7.0 mmol/l after 2 weeks and increased mance liquid chromatography method at again by 40 mg (from 80 to 120 mg or four major participating hospitals using RESULTS — We enrolled a total of from 40 to 80 mg) twice daily if FPG was the unique standard procedure and the 1,464 patients; 196 patients were lost to Ն7.0 mmol/l after 4 weeks of treatment. same reagents. Serum insulin was mea- follow-up during the course of treatment. The study was approved by the institu- sured using an electrochemiluminescence The major reason for those lost to fol- tional review boards of Anhui Medical method on an Elecsys 2010 system low-up was inconvenience because they University and all participating hospitals. (Roche Diagnostics, Basel, Switzerland) at lived too far from the study hospitals. The

1940 DIABETES CARE, VOLUME 31, NUMBER 10, OCTOBER 2008 Feng and Associates subjects who completed the study and Table 1—Characteristics of patients in cohort 1 and cohort 2 those who were lost to follow-up were Ͼ similar (P 0.05) in major baseline de- Characteristics Cohort 1 Cohort 2 Pooled mographic and clinical characteristics, in- cluding age (50.1 Ϯ 8.4 vs. 49.2 Ϯ 9.4 n 661 607 1,268 years), sex (men 54 vs. 49%), age of diag- Age (years) 50.2 Ϯ 8.0 50.5 Ϯ 8.6 50.4 Ϯ 8.3 nosis of type 2 diabetes (49.4 Ϯ 8.4 vs. Men (%) 54.8 54.0 54.4 48.5 Ϯ 10.4 years), fasting glucose Age at diagnosis (years)* 49.0 Ϯ 7.8 49.1 Ϯ 8.5 49.1 Ϯ 8.1 (11.1 Ϯ 2.9 vs. 11.4 Ϯ 3.2 mmol/l), in- Duration of diabetes (months)* 14.2 Ϯ 19.9 14.6 Ϯ 22.3 14.4 Ϯ 20.9 sulin (5.6 Ϯ 2.2 vs. 5.3 Ϯ 2.5 ␮IU/ml), BMI (kg/m2) 24.9 Ϯ 2.9 24.3 Ϯ 3.4† 24.9 Ϯ 8.4 A1C (8.4 Ϯ 1.9 vs. 8.1 Ϯ 2.1%), and Waist circumference (cm) 88.7 Ϯ 9.0 86.7 Ϯ 9.5† 87.8 Ϯ 9.3 blood pressure (126.1 Ϯ 17.3/81.3 Ϯ Baseline 10.8 vs. 125.6 Ϯ 17.4/80.5 Ϯ 10.2 FPG(mmol/l) 10.8 Ϯ 2.6 11.4 Ϯ 3.2† 11.1 Ϯ 2.9 mmHg). Some difference was noted for 2-h plasma glucose (mmol/l)‡ 18.9 Ϯ 4.4 18.3 Ϯ 5.0† 18.6 Ϯ 4.7 total cholesterol (5.2 Ϯ 1.4 vs. 5.6 Ϯ 4.2 A1C (%)§ 8.5 Ϯ 1.8 8.3 Ϯ 2.1 8.1 Ϯ 1.9 mmol/l, P ϭ 0.01) and triglycerides Fasting insulin (␮IU/ml)ʈ¶ 6.7 Ϯ 2.0 5.0 Ϯ 2.2† 5.5 Ϯ 2.2 (2.1 Ϯ 1.7 vs. 2.8 Ϯ 4.1 mmol/l, P Ͻ HOMA-Bʈ¶ 20.1 Ϯ 2.0 14.9 Ϯ 2.5† 18.2 Ϯ 2.2 0.01). HOMA-IRʈ¶ 1.0 Ϯ 2.0 0.8 Ϯ 2.0† 0.9 Ϯ 2.0 We analyzed the data of 1,268 pa- Total cholesterol (mmol/l) 5.1 Ϯ 1.1 5.2 Ϯ 1.6 5.2 Ϯ 1.4 tients who completed the entire study HDL cholesterol (mmol/l) 1.3 Ϯ 0.4 1.5 Ϯ 3.2 1.4 Ϯ 2.2 procedures. Most patients followed the Triglyceride (mmol/l) 2.2 Ϯ 1.6 2.0 Ϯ 1.9 2.1 Ϯ 1.7 instructions for diet and exercise well. FPG at day 29 (mmol/l) 7.9 Ϯ 2.1 8.1 Ϯ 2.4 8.0 Ϯ 2.3 The percentage of patients with poor diet At day 57 control was consistently Ͻ6% through- FPG(mmol/l) 7.7 Ϯ 2.1 7.7 Ϯ 2.4 7.7 Ϯ 2.3** out the whole trial, 42 and 53% of pa- 2-h plasma glucose (mmol/l)†† 14.6 Ϯ 4.4 13.9 Ϯ 5.1† 14.3 Ϯ 4.7** tients had fair or good diet control, A1C (%)‡‡ 7.1 Ϯ 1.3 6.6 Ϯ 1.4† 6.9 Ϯ 1.3** respectively, at day 57. The mean weekly Fasting insulin (␮IU/ml)ʈ§§ 7.4 Ϯ 2.0 5.5 Ϯ 2.5† 6.7 Ϯ 2.2** exercise time was 8 h. In 245 and 111 HOMA-Bʈ§§ 44.7 Ϯ 1.8 33.1 Ϯ 2.5† 40.4 Ϯ 2.2** patients, respectively, the dose of glicla- HOMA-IRʈ§§ 1.1 Ϯ 2.0 0.8 Ϯ 2.0† 1.0 Ϯ 2.0 zide was increased at days 15 and 29 ac- Data are means Ϯ SD. *The sample sizes of cohort 1, cohort 2, and total are 383, 291, and 674, respectively. cording to the study protocol to achieve †P Ͻ 0.05 compared with cohort 1. ‡The sample sizes of cohort 1, cohort 2, and total are 534, 429, and 963, ʈ better glycemic control. The demo- respectively. §The sample sizes of cohort 1, cohort 2, and total are 378, 321, and 699, respectively. Log transformed before the analysis; geometric mean and anti-log SD are presented. ¶The sample sizes of cohort graphic and clinical characteristics of the 1, cohort 2, and total are 572, 465, and 1,037, respectively. **P Ͻ 0.05 compared with baseline, t test. ††The study patients are summarized in Table 1. sample sizes of cohort 1, cohort 2, and total are 646, 520, and 1,166, respectively. ‡‡The sample sizes of The mean age of the patients was 50.4 cohort 1, cohort 2, and total are 460, 249, and 709, respectively. §§The sample sizes of cohort 1, cohort 2, years and 54.4% of the patients were and total are 598, 304, and 902, respectively. male. After 8 weeks of gliclazide therapy, the patients’ mean FPG decreased from dosage of gliclazide during the 8 weeks of 33 of the ABCC8 gene (encoding the sul- 11.1 mmol/l at baseline to 7.7 mmol/l treatment, and baseline HOMA-B and fonylurea receptor), results in an amino (mean Ϯ SD decrease 3.4 Ϯ 2.8 mmol/l), HOMA-IR, on percent decrease in FPG acid substitution of Ser/Ala. SNP rs5210, and 43.7 and 63.0% patients lowered after 8 weeks of gliclazide therapy. We which is located in the 3Ј untranslated their FPG to Ͻ7.0 and 7.8 mmol/l, re- found that the last three factors were sig- region of KCNJ11, is located in the same spectively. A1C decreased from 8.1 Ϯ 1.9 nificantly associated with the antidiabetic as rs757110 and is only 10 to 6.9 Ϯ 1.3% (1.4 Ϯ 1.6%). Mean fasting efficacy of gliclazide (P Ͻ 0.001). Thus, kb apart. However, the two SNPs are not insulin level and HOMA-B increased by these factors, along with age and sex, were in significant linkage disequilibrium (LD) 22 and 122%, respectively, but the mean adjusted in the subsequent genetic asso- (DЈϭ0.08, R2 ϭ 0.006). HOMA-IR did not change significantly ciation analyses. The distribution of pa- We genotyped rs757110 for all of the (Table 1). The patients included in the tients with different levels of diet control study patients in the cohort 2 study and cohort 1 and 2 studies were similar with (good, fair, or poor) and average exercise confirmed the significant association regard to age and sex distribution. How- hours among three genotype groups of identified in cohort 1 (P ϭ 0.002, additive ever, patients in cohort 2 had lower BMI Ser1369Ala polymorphism were not sig- model). In the pooled analysis (combin- and a lower level of A1C, fasting insulin, nificantly different (data not shown). ing cohort 1 and cohort 2 patients), we HOMA-B, and HOMA-IR than those in The minor allele frequencies of the found that patients with Ser/Ala and Ala/ cohort 1 (Table 1). This may reflect pop- SNPs we genotyped in the cohort 1 study Ala genotypes had 2.8% (P ϭ 0.076) and ulation differences across geographic dif- ranged from 1 to 50% (Table 2). Geno- 7.7% (P Ͻ 0.001) greater decreases in ferences between cohort 1 patients, who types of all SNPs were in Hardy-Weinberg FPG, and 10.8% (P ϭ 0.001) and 11.9% were mainly from northern China, and equilibrium. We found that 2 of the 25 (P ϭ 0.003) greater decreases in OGTT cohort 2 patients, who were from south- SNPs, rs757110 and rs5210, were signif- 2-h plasma glucose, respectively, com- ern China. icantly associated with percent decrease pared with those with the Ser/Ser geno- We first explored the possible influ- of FPG, even after Bonferroni correction type (Table 3). The decrease in A1C after ence of demographic and clinical vari- for multiple testing (P ϭ 0.002) (Table 2). 8 weeks of treatment with gliclazide was ables, including age, sex, BMI, total SNP rs757110, which is located in exon 3.5% greater in patients with the Ala/Ala

DIABETES CARE, VOLUME 31, NUMBER 10, OCTOBER 2008 1941 Ser1369Ala variant in ABCC8 and gliclazide

Table 2—Associations of 25 candidate SNPs with percentage decrease in FPG after 8-week gliclazide treatment in type 2 diabetic patients of cohort 1

Amino Association with acid Minor allele FPG decrease Gene Encoded SNP Codon change frequency (P)* ABCC8 Sulfonylurea receptor (SUR1) rs757110 1,369 Ser3Ala 0.432 0.002 rs1799854 — — 0.416 0.498 rs2074312 — — 0.500 0.031 rs2237984 — — 0.369 0.616 rs2237981 — — 0.256 0.046 Kir6.2 (KCNJ11) ATP-sensitive potassium channel, rs5210 — — 0.480 0.002 Kir6.2 ENSA Endosulfine alpha rs7517 — — 0.222 0.085 PPARG Peroxisome proliferative activated rs2972164 — — 0.084 0.284 receptor ␥ rs10510412 — — 0.346 0.480 rs2959273 — — 0.412 0.332 CAPN10 Calpain 10 rs10933620 — — 0.342 0.462 rs3792267 — — 0.107 0.765 rs2975760 — — 0.106 0.958 TCF1 Hepatocyte nuclear factor 1␣ rs2464195 — — 0.498 0.490 rs1169300 — — 0.499 0.457 IRS1 Insulin receptor substrate 1 rs1801278 972 Gly3Arg 0.009 0.905 rs9653366 — — 0.189 0.065 rs10498210 — — 0.073 0.336 rs12052364 — — 0.172 0.039 GLP1R Glucagon-like peptide 1 receptor rs1042044 260 Phe3Leu 0.492 0.496 UCP2 Uncoupling protein 2 rs660339 55 Val3Ala 0.467 0.739 PPARGC1A Peroxisome proliferative activated rs8192678 482 Ser3Gly 0.425 0.596 receptor ␥, coactivator 1␣ rs3736265 612 Thr3Met 0.175 0.780 ␤ 3 ADRB2 2-Adrenergic receptor rs1042713 16 Arg Gly 0.422 0.386 rs1042714 27 Gln3Glu 0.099 0.897 *Linear regression under additive model with adjustment for age, sex, total gliclazide dose, baseline HOMA-B, and HOMA-IR. genotype than in those with the Ser/Ser therapy, the patients with the Ser/Ala and ciation between the Ser1369Ala polymor- genotype (P ϭ 0.06) (Table 3). The mean Ala/Ala genotypes had odds ratios of 1.4 phism and fasting insulin level, HOMA-B, gliclazide dosage requirements for Ser/ (95% CI 1.0–2.1, P ϭ 0.06) and 2.2 and HOMA-IR, either at baseline or af- Ser, Ser/Ala, and Ala/Ala genotypes were (1.4–3.6, P ϭ 0.001), respectively, for re- ter treatment, or the change in fasting 83.7, 77.6, and 78.4 mg/day, respec- sponse to gliclazide therapy compared insulin level or HOMA-B after gliclazide tively. When we define FPG at day 57 as with subjects with Ser/Ser genotype. treatment. Interestingly, however, Ͻ7.8mmol/l in response to gliclazide We did not find any significant asso- mean HOMA-IR decreased significantly

Table 3—Association of Ser1369Ala genotype with percentage decrease in FPG, 2-h plasma glucose, and A1C after 8 weeks of gliclazide treatment in type 2 diabetic patients (pooled sample of cohort 1 and cohort 2)

Regression, Outcome phenotype Genotype n Baseline Day 57 Decrease (%) ␤ (Se)* P FPG (mmol/l) Ser/Ser 363 11.1 Ϯ 2.9 7.9 Ϯ 2.4 26.1 Ϯ 20.2 — — Ser/Ala 562 11.0 Ϯ 2.9 7.6 Ϯ 2.0 27.9 Ϯ 18.9 2.8 (1.6) 0.076 Ala/Ala 224 11.5 Ϯ 3.3 7.6 Ϯ 2.5 31.6 Ϯ 19.8 7.7 (1.9) Ͻ0.001 2-h plasma glucose (mmol/l) Ser/Ser 269 18.9 Ϯ 4.7 15.2 Ϯ 5.9 22.3 Ϯ 22.8 — — Ser/Ala 404 18.4 Ϯ 4.7 14.0 Ϯ 4.1 23.3 Ϯ 23.4 10.8 (3.3) 0.001 Ala/Ala 157 18.8 Ϯ 4.4 13.9 Ϯ 4.4 27.6 Ϯ 20.3 11.9 (4.1) 0.003 A1C (%) Ser/Ser 151 8.4 Ϯ 1.9 7.0 Ϯ 1.5 14.2 Ϯ 17.6 — — Ser/Ala 251 8.3 Ϯ 2.0 6.8 Ϯ 1.2 15.8 Ϯ 15.3 1.9 (1.4) 0.195 Ala/Ala 106 8.7 Ϯ 2.0 7.0 Ϯ 1.4 17.4 Ϯ 13.5 3.5 (1.8) 0.060 Data are means Ϯ SD unless otherwise indicated. *Multiple line regression model, outcome variables were percent decrease in FPG, 2-h plasma glucose, and A1C, respectively. The analysis adjusted for age, gender, total gliclazide dose, and baseline HOMA-B and HOMA-IR. ␤ (Se), regression coefficient (SE) for genotype using Ser/Ser as reference.

1942 DIABETES CARE, VOLUME 31, NUMBER 10, OCTOBER 2008 Feng and Associates

(0.15 Ϯ 1.10, P ϭ 0.04) in patients with consistent associations were observed in with increased risk for secondary failure the Ala/Ala genotype, but not in patients the two different cohorts strengthens the of sulfonylurea treatment in type 2 dia- with the Ser/Ser or Ser/Ala genotypes genetic association and its generalizability betic patients. The E23K polymorphism (data not shown). across populations. is in high LD (DЈϭ0.98, R2 ϭ 0.87) with Of the 1,268 study subjects, 273 We did not examine the association of the Ser1369Ala of the ABCC8 gene (5), so (21.5%) met the criteria for metabolic a common variant of the KCNJ11 gene in we did not include the E23K polymor- syndrome. However, we did not find any cohort 2. It is possible that Ser1396Ala phism in our study. However, we did find association between the Ser1396Ala poly- SNP is not a causal variant. Instead, the that a common variant of KCNJ11, morphism and BMI or metabolic syn- association between Ser1396Ala and an- rs5210, was associated with gliclazide re- drome, and further adjustment for tidiabetic efficacy may be a result of other sponse in our cohort 1 study. metabolic syndrome in the analysis did functional mutations that are in high LD SUR1 is an important subunit of the ϩ not alter the association between the with Ser1396Ala. Future studies should ATP-sensitive K channel, which is a key Ser1396Ala polymorphism and antidia- screen for mutations within both the component in regulation of insulin secre- betic efficacy (data not shown). SUR1 gene and nearby genes, such as tion from pancreatic ␤-cell membranes. KCNJ11. The Ser1369Ala variant is located in the CONCLUSIONS — The sulfonylurea The association between A1C and the second nucleotide-binding fold, a func- receptor (SUR1), encoded by the ABCC8 Ser1369Ala polymorphism was only mar- tionally important region of the ABCC8 gene, is the internal molecular target for ginally significant. We speculate that sig- gene. It has not been found to be asso- gliclazide. Genetic variants of ABCC8, nificance would have been achieved if ciated with type 2 diabetes in either such as exon 16 Ϫ3t/c, exon 18 T759T, we had extended the trial for another 4 Caucasians (13) or Japanese (5,18). In- and most recently the missense mutation weeks, because hemoglobin turns over terestingly, the ABCC8 Ser1369Ala Y356C, have been reported to be associ- every 3 months. Another reason was that polymorphism was recently reported to ated with type 2 diabetes (3,5,12,13). missing data for A1C may have affected influence progression to diabetes (6). Al- Some missense mutations other than our statistical power to detect a significant though the Ser1369Ala is a missense Ser1369Ala in the ABCC8 gene have been association. polymorphism, its influence on SUR1 identified as causing Finally, we excluded patients with function still remains uncertain. Florez et (14). To our knowledge, this study is the longstanding diabetes (Ͼ5 years), who al. (6) found the Ala/Ala carriers had a first to report the association between a were more likely to have decreased ␤-cell significantly lower insulin index, i.e., in- genetic variant of the ABCC8 gene and function and to have complications of the sulin secretion function, compared with therapeutic efficacy of sulfonylurea in disease (which may affect the therapeutic Ser/Ser carriers in subjects with impaired type 2 diabetic patients. response to gliclazide). Caution is needed glucose tolerance. However, other studies The major strengths of this study in- in generalizing our findings to patients did not find any association between the clude large sample size, multiple well- with longstanding diabetes. Ser1369Ala variant and insulin secretion measured phenotypes, and a prospective Although our study may not have an in nondiabetic subjects (19,20). We also design. However, cautions are needed in immediate impact on clinical practice, it did not find any association between the interpreting our findings. First, we only may stimulate more investigations in this Ser1369Ala variant and fasting plasma in- studied the response to gliclazide for a area. Pharmacogenetics is an emerging sulin level or HOMA-B, an indicator for short period, i.e., the initial response to discipline investigating the influence of insulin secretion, either at baseline or af- monotherapy with a sulfonylurea. The as- genetic variants on drug response. It is an ter gliclazide treatment. The improve- sociation of Ser1369Ala with the long- important path toward personalized ment in insulin resistance (HOMA-IR) in term response to gliclazide remains medicine. So far, only a few pharmacoge- the Ser/Ser genotype group was probably unknown. Second, we excluded obese netic studies on diabetes have been re- due to good glucose control after glicla- patients. The InterASIA study, a recent ported. Patients with maturity-onset zide treatment. large-scale nationwide epidemiological diabetes of the young who had the hepa- In summary, we found that a com- study in China, showed that 18.7% of di- tocyte nuclear factor-1␣ mutation were mon variant in the ABCC8 gene, abetic subjects were obese (BMI Ն28 kg/ reported to be extremely sensitive to the Ser1369Ala, was significantly associated m2) (15). Thus, our study sample should hypoglycemic effects of sulfonylureas with the antidiabetic efficacy of gliclazide represent Ͼ80% of type 2 diabetic pa- (16). Recently, Shu et al. (9) took a mul- in nonobese type 2 diabetic patients in tients in China. tipronged approach, using cell-based ex- China. Patients with the Ala/Ala genotype Population admixture, which may periments, in vivo studies in mice, and in appeared to respond significantly better lead to false-positive association, is a ma- vitro human trials (healthy volunteers), to gliclazide than did patients with the jor concern in a genetic association study. and demonstrated that genetic variances Ser/Ser genotype. Although the difference Given the obvious difference between the of organic cation transporter 1 (OCT1) may have limited impact on clinical prac- northern and southern Han Chinese, we had a significant impact on response to tice, our results did demonstrate that ge- analyzed the two cohorts separately to metformin, a common antidiabetic agent netic variation can be a significant minimize potential confounding due to for type 2 diabetes. However, a small determinant of response to oral hypogly- population admixture. The differential study (n ϭ 24 responders and 9 nonre- cemic drugs. baseline characteristics of the two study sponders) did not confirm the association cohorts may affect antidiabetic efficacy of of OCT1 gene polymorphisms and re- the treatment; however, it would not be sponse to metformin in diabetic patients Acknowledgments— The study was sup- likely to influence the main purpose of (8). Sesti et al. (17) reported that the ported by Tianshili Pharmaceutical Company, this pharmacogenetic study. The fact that E23K variants in KCNJ11 are associated Tianjin, China.

DIABETES CARE, VOLUME 31, NUMBER 10, OCTOBER 2008 1943 Ser1369Ala variant in ABCC8 and gliclazide

We greatly appreciate the participation of to diabetes and response to interven- Aguilar-Bryan L, Vaxillaire M, Froguel P: the 23 hospitals in the study. We are grateful tions in the Diabetes Prevention Pro- Activating mutations in the ABCC8 gene to Dr. Rongh Qian for input and support of the gram. Diabetes 56:531–536, 2007 in neonatal diabetes mellitus. N Engl J Med study and Dr. Scott A. Venners for his editing 7. Sesti G, Hribal ML: Pharmacogenetics 355:456–466, 2006 of the manuscript. in type 2 diabetes: polymorphisms in 15. Hu D, Xie J, Fu P, Zhou J, Yu D, Whelton candidate genes affecting responses to PK, He J, Gu D: Central rather than overall antidiabetic oral treatment. Curr Phar- obesity is related to diabetes in the Chinese References macogenomics 4:69–78, 2006 population: the InterASIA study. Obesity 1. Wild S, Roglic G, Green A, Sicree R, King 8. Shikata E, Yamamoto R, Takane H, Shige- (Silver Spring) 15:2809–2816, 2007 H: Global prevalence of diabetes: esti- masa C, Ikeda T, Otsubo K, Ieiri I: Human 16. Pearson ER, Liddell WG, Shepherd M, mates for the year 2000 and projections organic cation transporter (OCT1 and Corrall RJ, Hattersley AT: Sensitivity to for 2030. Diabetes Care 27:1047–1053, OCT2) gene polymorphisms and thera- sulphonylureas in patients with hepato- 2004 peutic effects of metformin. J Hum Genet cyte nuclear factor-1␣ gene mutations: 2. Zeggini E: A new era for type 2 diabetes 52:117–122, 2007 evidence for pharmacogenetics in diabe- genetics, Diabet Med 24:1181–1186, 9. Shu Y, Sheardown SA, Brown C, Owen tes. Diabet Med 17:543–545, 2000 2007 RP, Zhang S, Castro RA, Ianculescu AG, 17. Sesti G, Laratta E, Cardellini M, An- 3. Tarasov AI, Nicolson T, Riveline JP, Yue L, Lo JC, Burchard EG, Brett CM, Gia- dreozzi F, Del Guerra S, Irace C, Gnasso Taneja TK, Baldwin SA, Baldwin JM, comini KM: Effect of genetic variation in A, Grupillo M, Lauro R, Hribal ML, Per- Charpentier G, Gautier JF, Froguel P, the organic cation transporter 1 (OCT1) ticone F, Marchetti P: The E23K variant Vaxillaire M, Rutter GA: A rare mutation on metformin action. J Clin Invest of KCNJ11 encoding the pancreatic ␤ Ј in ABCC8/SUR1 leading to altered KATP 117:1422–1431, 2007 -cell adenosine 5 -triphosphate-sensi- channel activity and ␤-cell glucose sens- 10. Kahn SE, Haffner SM, Heise MA, Herman tive potassium channel subunit Kir6.2 ing is associated with type 2 diabetes mel- WH, Holman RR, Jones NP, Kravitz BG, is associated with an increased risk of litus in adults. Diabetes 57:1595–1604, Lachin JM, O’Neill MC, Zinman B, Viberti secondary failure to sulfonylurea in pa- 2008 G; ADOPT Study Group: Glycemic dura- tients with type 2 diabetes. J Clin Endo- 4. Sakamoto Y, Inoue H, Keshavarz P, bility of rosiglitazone, metformin, or gly- crinol Metab 91:2334–2339, 2006 Miyawaki K, Yamaguchi Y, Moritani M, buride monotherapy. N Engl J Med 355: 18. Ohta Y, Tanizawa Y, Inoue H, Hosaka T, Kunika K, Nakamura N, Yoshikawa T, 2427–2443, 2006 Ueda K, Matsutani A, Repunte VP, Yasui N, Shiota H, Tanahashi T, Itakura 11. American Diabetes Association: Diagnosis Yamada M, Kurachi Y, Bryan J, Aguilar- M: SNPs in the KCNJ11-ABCC8 gene lo- and classification of diabetes mellitus. Di- Bryan L, Permutt MA, Oka Y: Identifica- cus are associated with type 2 diabetes abetes Care 29:S43–S48, 2006 tion and functional analysis of sulfonylurea and blood pressure levels in the Japa- 12. Hani EH, Clement K, Velho G, Vionnet N, receptor 1 variants in Japanese patients with nese population. J Hum Genet 52:781– Hager J, Philippi A, Dina C, Inoue H, Per- NIDDM. Diabetes 47:476–481, 1998 793, 2007 mutt MA, Basdevant A, North M, De- 19. Hansen T, Echwald SM, Hansen L, Moller 5. Yokoi N, Kanamori M, Horikawa Y, menais F, Guy-Grand B, Froguel P: AM, Almind K, Clausen JO, Urhammer SA, Takeda J, Sanke T, Furuta H, Nanjo K, Genetic studies of the sulfonylurea recep- Inoue H, Ferrer J, Bryan J, Aguilar-Bryan L, Mori H, Kasuga M, Hara K, Kadowaki T, tor gene locus in NIDDM and in morbid Permutt MA, Pedersen O: Decreased tolbu- Tanizawa Y, Oka Y, Iwami Y, Ohgawara obesity among French Caucasians. Diabe- tamide-stimulated insulin secretion in H, Yamada Y, Seino Y, Yano H, Cox NJ, tes 46:688–694, 1997 healthy subjects with sequence variants in Seino S: Association studies of variants in 13. Inoue H, Ferrer J, Welling CM, Elbein SC, the high-affinity sulfonylurea receptor gene. the genes involved in pancreatic ␤-cell Hoffman M, Mayorga R, Warren-Perry M, Diabetes 47:598–605, 1998 function in type 2 diabetes in Japanese Zhang Y, Millns H, Turner R, Province M, 20. Rissanen J, Markkanen A, Karkkainen P, subjects. Diabetes 55:2379–2386, 2006 Bryan J, Permutt MA, Aguilar-Bryan L: Se- Pihlajamaki J, Kekalainen P, Mykkanen L, 6. Florez JC, Jablonski KA, Kahn SE, quence variants in the sulfonylurea recep- Kuusisto J, Karhapaa P, Niskanen L, Franks PW, Dabelea D, Hamman RF, tor (SUR) gene are associated with Laakso M: Sulfonylurea receptor 1 gene Knowler WC, Nathan DM, Altshuler D: NIDDM in Caucasians. Diabetes 45:825– variants are associated with gestational di- Type 2 diabetes–associated missense 831, 1996 abetes and type 2 diabetes but not with polymorphisms KCNJ11 E23K and 14. Babenko AP, Polak M, Cave H, Busiah K, altered secretion of insulin. Diabetes Care ABCC8 A1369S influence progression Czernichow P, Scharfmann R, Bryan J, 23:70–73, 2000

1944 DIABETES CARE, VOLUME 31, NUMBER 10, OCTOBER 2008