Advance Publication by J-STAGE Circulation Journal Official Journal of the Japanese Circulation Society http://www.j-circ.or.jp

Bilirubin Is Negatively Associated With A1c Independently of Other Cardiovascular Risk Factors in Apparently Healthy Japanese Men and Women Eiji Oda, MD; Ryu Kawai, MD

Background: total (TB) is a potent antioxidant and may be a negative risk factor of cardiovas- cular disease. In non-diabetic adults, hemoglobin A1c (HbA1c), but not fasting plasma glucose, is an independent risk factor of .

Methods and Results: Linear regression using TB as a dependent variable and cardiovascular risk factors, in- cluding HbA1c, as independent variables, linear regression using HbA1c as a dependent variable and other cardio- vascular risk factors, including TB, as independent variables, and logistic regression using the highest decile (≥5.4%) of HbA1c as a dependent variable and TB and other cardiovascular risk factors as independent variables were performed for 893 apparently healthy male smokers, 1,607 male nonsmokers, and 1,680 women. The standardized regression coefficient of HbA1c for TB and that of TB for HbA1c was –0.12 (P=0.007) and –0.06 (P=0.02), respec- tively, in the smokers, –0.20 (P<0.0001) and –0.07 (P<0.0001), respectively, in the nonsmokers, and –0.21 (P< 0.0001) and –0.14 (P<0.0001), respectively, in the women. The odds ratio of 1 SD increment in TB for HbA1c ≥5.4% was not significant in the smokers, 0.67 (P=0.002) in the nonsmokers, and 0.55 (P<0.0001) in the women.

Conclusions: Bilirubin was negatively associated with HbA1c independently of other cardiovascular risk factors in apparently healthy Japanese men and women. The association was weak in male smokers.

Key Words: Antioxidant; Bilirubin; Cardiovascular disease; ; Hemoglobin A1c

ilirubin is a potent antioxidant in the human body.1 for CVD. In nondiabetic adults, however, fasting plasma Increased serum levels of total bilirubin (TB) have glucose (FPG) was not significantly associated with risk of B been reported to be associated with reduced risk for CVD and of death from any cause, adjusting for hemoglobin cardiovascular disease (CVD) in some epidemiological studies A1c (HbA1c). In contrast, HbA1c was significantly associated and TB is suggested to be an independent risk factor for with risk of CVD and of death from any cause, adjusting for CVD.2 Healthy subjects with lower TB levels have been found FPG.10 HbA1c levels appear to increase with age11 and will to have significant endothelial dysfunction and increased be influenced by any condition that changes red cell turnover, intima – media thickness (IMT) of the carotid artery, which such as iron deficiency, , chronic , are predictors for CVD.3 TB has been suggested to be a novel major blood loss, or blood transfusions.12 Other than age and risk marker of CVD in middle-aged men.4 Increased TB is hemoglobin , some cardiovascular risk factors associated with reduced peripheral artery disease prevalence.5 may have an independent association with HbA1c. Thus, we TB may have some protective function against stroke risk.6 hypothesized that the stronger association of HbA1c, than of Low TB is associated with coronary artery calcification, which FPG, with CVD, in nondiabetic adults may at least partly be is a predictor of coronary artery events in Japanese people.7 mediated by the association between HbA1c and bilirubin in Gilbert syndrome, a genetic variant characterized by moder- nondiabetic subjects. The aim of the present study was there- ate hyperbilirubinemia, was reported to be associated with fore to examine the cross-sectional association between TB a reduced risk for CVD.8 In subjects with Gilbert syndrome, and HbA1c in apparently healthy Japanese men and women elevated TB levels are associated with lower levels of who had no history of CVD and who were not on antidiabetic, advanced glycation endproducts that promote oxidant stress antihypertensive, or antihyperlipidemic . and inflammation.9 Thus, TB may be a negative risk factor

Received July 4, 2010; revised manuscript received August 13, 2010; accepted September 3, 2010; released online December 2, 2010 Time for primary review: 28 days Medical Check-up Center, Tachikawa Medical Center, Nagaoka, Japan Mailing address: Eiji Oda, MD, Medical Check-up Center, Tachikawa Medical Center, 2-2-16 Nagachou, Nagaoka 940-0053, Japan. E-mail: [email protected] ISSN-1346-9843 doi: 10.1253/circj.CJ-10-0645 All rights are reserved to the Japanese Circulation Society. For permissions, please e-mail: [email protected] Advance Publication by J-STAGE ODA E et al.

Table 1. Baseline Data Smoking men Non-smoking men Women (n=893) (n=1,607) (n=1,680) Age (years) 48.2±8.7 49.5±9.9* 48.5±9.5 BMI (kg/m2) 23.1±2.9 23.0±2.7 21.4±3.0 WC (cm) 83.9±8.0 83.4±7.8 77.9±8.3 Systolic BP (mmHg) 118.2±15.4 120.7±17.2* 109.8±15.4 Diastolic BP (mmHg) 75.6±10.3 77.1±11.0* 68.5±9.9 TB (mg/dl) 0.76±0.30 0.93±0.38* 0.73±0.28

HbA1c (%) 5.02±0.51 4.99±0.41 4.96±0.31 FPG (mg/dl) 94.6±15.9 94.1±11.4 88.1±8.5 (mg/dl) 112 [80, 162] 95 [69, 137]# 68 [52, 92] HDL-C (mg/dl) 55.0±13.9 59.2±14.5* 68.0±14.6 LDL-C (mg/dl) 122.4±30.8 122.8±29.0 119.4±30.9 hs-CRP (mg/L) 0.35 [0.20, 0.78] 0.28 [0.15, 0.56]# 0.21 [0.11, 0.46] FPG >– 100 mg/dl 21.1 20.2 7.2 FPG >– 126 mg/dl 2.7 1.8 0.5 HbA1c >– 5.2% 24.1 21.8 22.9 HbA1c >– 5.4% 11.4 9.4 9.1 Current smoker 100 0 6.6 Data are mean ± SD, median [25, 75 percentiles], or %. *P<0.01 compared with smoking men, #P<0.01 (log-transformed) compared with smoking men. BMI, body mass index; WC, waist circumference; BP, blood pressure; TB, total bilirubin; HbA1c, hemoglobin A1c; FPG, fasting plasma glucose; HDL-C, high-density lipoprotein ; LDL-C, low-density lipoprotein cholesterol; hs-CRP, high-sensitivity C-reactive protein.

factant method using Choletest-LDL (Sekisui Medical, Tokyo, Methods Japan). The measurement limit of hs-CRP was 0.02 mg/L, Subjects and any results under this measurement limit were taken as Between April 2008 and March 2010, 3,401 men and 2,095 being 0.01 mg/L. Average systolic blood pressure (SBP) and women visited Medical Check-up Center, Tachikawa Medical diastolic blood pressure (DBP) were calculated from 2 mea- Center, for general health screening. Among them 3,375 men surements done while the subjects were seated after a 5-min and 2,069 women signed informed consent. Excluding sub- rest. Body weight was measured with the subjects wearing jects with a history of CVD or who were taking antidiabetic, light clothes provided by the Center and the weight of the antihypertensive, or antihyperlipidemic medication from those clothing was subtracted from the measured body weight. who gave consent, the resulting 2,500 men and 1,680 women Waist circumference (WC) was measured at the level of the comprised the subjects in the present study, which is part of umbilicus with the subject standing. Body mass index (BMI) the study on obesity-related cardiovascular risk factors at was calculated as weight in kilograms divided by the square Tachikawa Medical Center. Data collected at the first visit of height in meters. were used for subjects who visited 2 or more times. The pro- tocol for the study was approved by the ethics committee at Statistical Analysis Tachikawa Medical Center and written informed consent Pearson correlation coefficients between age, BMI, SBP, was given by each subject. DBP, FPG, , HDL-C, LDL-C, hs-CRP, TB, and HbA1c, and both TB and HbA1c were calculated separately in Measurements male smokers and nonsmokers, and women. Linear regres- After an overnight fast, blood samples were obtained to mea- sions with stepwise exclusion of independent variables with sure levels of HbA1c, FPG, triglycerides, high-density lipo- P>0.05, using TB as a dependent variable and age, BMI, SBP, protein cholesterol (HDL-C), low-density lipoprotein choles- FPG, triglycerides, HDL-C, LDL-C, hs-CRP, and HbA1c as terol (LDL-C), high-sensitivity C-reactive protein (hs-CRP), initial independent variables; linear regressions with stepwise and TB. Chemistry was assessed at BML Nagaoka (Nagaoka, exclusion of independent variables with P>0.05, using HbA1c Japan) using routine laboratory methods except for hs-CRP, as a dependent variable and age, BMI, SBP, FPG, triglycer- which was measured at BML General Laboratory (Tokyo, ides, HDL-C, LDL-C, hs-CRP, and TB as initial independent Japan) with nephelometry using N-latex CRP-2 (Siemens variables; and logistic regressions with stepwise exclusion of Healthcare Japan, Tokyo, Japan). HbA1c was measured with independent variables with P>0.05, using the highest decile a latex aggregation immunoassay using Determiner HbA1c (≥5.4%) of HbA1c as a dependent variable and age, BMI, SBP, (Kyowa Medex, Tokyo, Japan). This assay is not affected by FPG, triglycerides, HDL-C, LDL-C, hs-CRP, TB, and smok- the presence of bilirubin when conjugated bilirubin concen- ing status as initial independent variables, were performed trations are <40 mg/dl. In Japan, HbA1c levels are expressed separately in the men, and women. Triglyceride and hs-CRP in Japan Diabetes Society (JDS) units, and HbA1c levels in were transformed to logarithms before calculations because National Glycohemoglobin Standardization Program (NGSP) of skewed distributions. WC and DBP were excluded from units in the following equation: HbA1c NGSPunits% = 1.02 × the initial independent variables in multivariate analysis to HbA1c JDSunits% + 0.3. LDL-C was measured with a direct sur- avoid multicollinearity. Continuous variables were divided Advance Publication by J-STAGE Bilirubin, A1c, and CVD

Table 2. Pearson Correlation Coefficients Smoking men Non-smoking men Women

HbA1c TB HbA1c TB HbA1c TB Age 0.18 –0.07 0.25 –0.04** 0.38 –0.04** BMI 0.19 –0.08 0.18 –0.11 0.20 –0.09 Systolic BP 0.11 –0.03** 0.13 –0.04** 0.17 –0.02** Diastolic BP 0.08 0.001** 0.12 –0.03** 0.13 0.01** FPG 0.83 –0.02** 0.73 –0.02** 0.55 –0.03** Log triglycerides 0.19 –0.11 0.18 –0.10 0.21 –0.12 HDL-C –0.12 0.16 –0.10 0.13 –0.13 0.17 LDL-C 0.16 –0.11 0.17 –0.04** 0.27 –0.02** Log hs-CRP 0.15 –0.11 0.08 –0.11 0.11 –0.12 TB –0.08 –0.10 –0.17

HbA1c –0.08 –0.10 –0.17 **P>0.05. Abbreviations see in Table 1.

Table 3. Multivariate Linear Regression Using TB as a Table 4. Multivariate Linear Regression Using HbA1c as a Dependent Variable Dependent Variable Independent variables† SRC 95%CI P value Independent variables† SRC 95%CI P value Smoking men Smoking men

HbA1c –0.12 –0.20, –0.03 0.007 FPG 0.77 0.74, 0.81 <0.0001 FPG 0.08 0.00, 0.16 0.047 Age 0.09 0.05, 0.14 <0.0001 HDL-C 0.12 0.07, 0.18 <0.0001 TB –0.06 –0.11, –0.01 0.02 Constant 2.36 1.72, 3.00 <0.0001 LDL-C 0.07 0.03, 0.11 0.001 Non-smoking men HDL-C –0.07 –0.11, –0.02 0.004

HbA1c –0.20 –0.28, –0.12 <0.0001 Systolic BP –0.07 –0.12, –0.03 0.002 FPG 0.15 0.07, 0.24 0.0006 Constant 5.77 5.28, 6.26 <0.0001 HDL-C 0.11 0.05, 0.16 <0.0001 Non-smoking men Log hs-CRP –0.09 –0.15, –0.04 0.0008 FPG 0.76 0.72, 0.80 <0.0001 Constant 3.17 2.49, 3.85 <0.0001 Age 0.09 0.06, 0.12 <0.0001 Women TB –0.07 –0.10, –0.04 <0.0001

HbA1c –0.21 –0.26, –0.15 <0.0001 LDL-C 0.09 0.06, 0.12 <0.0001 FPG 0.10 0.0,5, 0.16 0.0003 Systolic BP –0.06 –0.09, –0.03 <0.0001 HDL-C 0.15 0.10, 0.19 <0.0001 Constant 5.49 5.17, 5.82 <0.0001 Log hs-CRP –0.09 –0.14, –0.05 0.0001 Women LDL-C 0.06 0.01, 0.11 0.01 FPG 0.48 0.45, 0.52 <0.0001 Constant 3.83 3.03, 4.64 <0.0001 Age 0.25 0.21, 0.29 <0.0001 †Initial independent variables with P>0.05 were excluded. TB –0.14 –0.17, –0.1 <0.0001 CI, confidence interval; SRC, standardized regression coefficient. LDL-C 0.10 0.06, 0.14 <0.0001 Other abbreviations see in Table 1. HDL-C –0.05 –0.09, –0.01 0.007 Systolic BP –0.05 –0.09, –0.01 0.02 by 1 SD before multivariate analysis to standardize each value. Constant 10.20 9.71, 10.68 <0.0001 Statistical analyses were performed using Dr SPSS2 (SPSS †Initial independent variables with P>0.05 were excluded. Japan, Tokyo, Japan). Abbreviations see in Tables 1,3.

Results HDL-C were independently positively associated with TB in Baseline data are presented in Table 1. TB was significantly the men and women. The hs-CRP level was independently lower and hs-CRP was significantly higher in the smokers negatively associated with TB in the male nonsmokers and than in nonsmokers among the men, and HbA1c was not sig- the women, but not in the male smokers. The multivariate nificantly different between smokers and nonsmokers. The linear regressions using HbA1c as a dependent variable are Pearson correlation coefficients between age, BMI, SBP, given in Table 4. FPG, age, TB, SBP, and LDL-C were inde- DBP, FPG, triglycerides, HDL-C, LDL-C, hs-CRP, TB, and pendently associated with HbA1c in the men, and the women. HbA1c, and both TB and HbA1c are given in Table 2. The The multivariate logistic regressions using the highest decile Pearson correlation coefficients between TB and HbA1c in the (≥5.4%) of HbA1c as a dependent variable are listed in men and women were all significant. The multivariate linear Table 5. FPG, age, and TB were independently associated regressions using TB as a dependent variable are listed in with the highest decile (≥5.4%) of HbA1c in the nonsmokers Table 3. HbA1c was independently negatively, and FPG and and the women, but TB was not significantly associated with Advance Publication by J-STAGE ODA E et al.

carotid artery, which are predictors for CVD.3 Troughton et al Table 5. Multivariate Logistic Regression Using HbA1c >– 5.4% as a Dependent Variable suggested that TB is a novel risk marker of CVD in middle- Independent variables† OR 95%CI P value aged men, with a U-shaped relationship observed between TB and CVD risk in the Prospective Epidemiological Study Smoking men of Myocardial Infarction.4 Increased TB level is associated FPG 7.05 4.75, 10.47 <0.0001 with reduced peripheral artery disease prevalence in a study Age 1.69 1.21, 2.35 0.002 from the National Health and Nutrition Examination Survey BMI 1.51 1.13, 2.01 0.005 1999–2004.5 Kimm et al suggested that TB might have some HDL-C 0.63 0.44, 0.89 0.01 protective function against stroke risk in Korean men.6 Low Constant 5.7×10–10 <0.0001 TB level is associated with coronary artery calcification in Non-smoking men Japanese subjects.7 Gilbert syndrome, a genetic variant char- FPG 8.34 5.96, 11.67 <0.0001 acterized by moderate hyperbilirubinemia attributable to Age 1.80 1.46, 2.23 <0.0001 reduced hepatic expression of the UDP-glucuronosyltrans- TB 0.67 0.53, 0.86 0.002 ferase, which conjugates bilirubin, has been associated with a reduced risk for CVD.8 In patients with cyanotic con- Log triglycerides 1.27 1.02, 1.59 0.03 genital heart disease, carotid IMT, and hence atherosclerosis Systolic BP 0.74 0.60, 0.92 0.008 disease risk, was decreased, which might be partly because of Constant 1.3×10–9 <0.0001 higher TB levels.16 Yesilova et al suggested that the chronic Women hyperbilirubinemia in Gilbert syndrome leads to a lag-phase FPG 3.10 2.52, 3.83 <0.0001 prolongation in LDL oxidation and a decrease in LDL oxi- Age 2.15 1.73, 2.67 <0.0001 dation.17 In subjects with Gilbert syndrome, elevated TB TB 0.55 0.42, 0.73 <0.0001 levels are associated with lower levels of advanced glycation Log triglycerides 1.32 1.09, 1.61 0.005 endproducts.9 Recent data suggested a direct relationship Constant 1.8×10–9 <0.0001 between urinary excretion of biopyrrins, a novel group of bilirubin oxidative , and severity of oxidative †Initial independent variables with P>0.05 were excluded. OR, odds ratio for 1 SD increment. Other abbreviations see in stress. Yamaguchi et al investigated the antioxidative effect Table 1. of bilirubin in a rat model of ischemia – reperfusion and demonstrated that the hepatic mRNA level of oxygen- ase-1 (HO-1), the rate-limiting of bilirubin biosyn- the highest decile of HbA1c in the smokers. thesis, and the amount of biopyrrins in increased after ischemia – reperfusion injury.18 They suggested that bilirubin acts as a physiological antioxidant in vivo in ischemia – Discussion reperfusion and that bilirubin biosynthesis is evoked by oxi- In the present cross-sectional study, TB was independently dative stress.18 Vítek et al demonstrated an inverse relation- negatively associated with HbA1c in multivariate linear re- ship between serum bilirubin level and urinary excretion of gressions in both smoking and nonsmoking men and in biopyrrins and suggested antioxidative effects of elevated women, and was also independently negatively associated serum bilirubin levels in Gilbert syndrome.19 Yamamoto et al with the highest decile (≥5.4%) of HbA1c in multivariate studied the dynamics of oxidative stress after myocardial logistic regressions in the nonsmoking men and the women. ischemia reperfusion in rat models using urinary excretion The results were essentially the same when we excluded inde- of biopyrrins and their generation in various organs as oxida- pendent variables with P>0.1 instead of P>0.05 in regression tive stress markers and suggested that urinary excretion of analyses (Tables S1–S3). The results were also similar when biopyrrins can serve as a useful marker of systemic oxidative we examined the associations of TB with the highest quartile stress after myocardial ischemia – reperfusion.20 Ueyama et of HbA1c instead of the highest decile of HbA1c (Table S4). al reported that cardiac and aortic macrophages upregulate We did not analyze smoking status in women separately be- HO-1 in response to emotional stress and catecholamine surges cause they comprise only 6.6% of current smokers in Japan. as an endogenous defense mechanism against oxidative stress In men, the association between TB and HbA1c was weaker in rat models of takotsubo cardiomyopathy.21 Kunii et al dem- in smokers than in nonsmokers. This phenomenon may have onstrated that serum bilirubin and biopyrrins were elevated resulted from the decrease in TB levels (Table 1) because in patients with acute myocardial infarction, and that bio- of the pro-oxidant, pro-inflammatory (hs-CRP, Table 1), pyrrins and HO-1 expression were increased in the infarcted and pro-insulin-resistant13,14 (triglycerides, HDL-C, Table 1) myocardium and that plasma and urinary biopyrrin elevation effects of smoking. was associated with mortality and morbidity.22 These data The catabolism of heme by generates suggest the involvement of bilirubin/biopyrrin pathway in , free iron, and , which is rapidly the of CVD. Inoguchi et al reported a lower converted to bilirubin by ubiquitously expressed biliverdin prevalence of CVD in patients with diabetes and Gilbert reductase. Although oxidation of LDL-C is critical for athero- syndrome, and suggested that sustained hyperbilirubinemia sclerotic vascular changes,15 bilirubin suppresses the oxida- inhibits oxidative stress and prevents the development of tion of lipid more than α-tocopherol.1 Bilirubin efficiently CVD in diabetic patients.23 scavenges a wide range of physiological oxidants by electron HbA1c is a reliable marker of chronic glycemia and corre- donation. In this process it is often re-converted to biliverdin, lates well with the risk of long-term diabetes complications,24 but quickly regenerates bilirubin, thereby and is the test of choice for the chronic management of dia- greatly boosting its antioxidant potential. Schwertner et al betes and is recommended for its diagnosis.12 In the Athero- suggested that serum TB is an inverse and independent risk sclerosis Risk in Communities (ARIC) study of non-diabetic factor for CVD.2 Healthy subjects with lower TB levels have adults, HbA1c was similarly associated with a risk of diabetes significant endothelial dysfunction and increased IMT of as FPG and was more strongly associated with risks of CVD Advance Publication by J-STAGE Bilirubin, A1c, and CVD and death from any cause than FPG.10 FPG was not signifi- 2. Schwertner HA, Jackson WG, Tolan G. Association of low serum cantly associated with risks of CVD and death from any cause concentration of bilirubin with increased risk of coronary artery disease. Clin Chem 1994; 40: 18 – 23. adjusting for HbA1c, whereas HbA1c was significantly asso- 3. Erdogan D, Gullu H, Yildirim E, Tok D, Kirbas I, Ciftci O, et al. ciated with risks of CVD and death from any cause adjusting Low serum bilirubin levels are independently and inversely related for FPG.10 HbA1c levels appear to increase with age11 and in to impaired flow-mediated vasodilation and increased carotid intima- iron deficiency.25,26 Other than age and iron deficiency, some media thickness in both men and women. Atherosclerosis 2006; cardiovascular risk factors may be independently associated 184: 431 – 437. 4. Troughton JA, Woodside JV, Young IS, Arveiler D, Amouyel P, with HbA1c. Ohnaka et al studied the association of TB with Ferrières J, et al. Bilirubin and coronary heart disease risk in hs-CRP, HbA1c, and the prevalence of diabetes in middle- the Prospective Epidemiological Study of Myocardial Infarction aged and elderly Japanese men and women.27 In their study, (PRIME). Eur J Cardiovasc Prev Rehabil 2007; 14: 79 – 84. geometric means of hs-CRP and HbA1c were progressively 5. Perlstein TS, Pande RL, Beckman JA, Creager MA. Serum total bilirubin level and prevalent lower-extremity peripheral arterial lower with increasing levels of TB, and an inverse associa- disease: National Health and Nutrition Examination Survey tion between TB and HbA1c adjusted for age, BMI, smoking (NHANES) 1999 to 2004. Arterioscler Thromb Vasc Biol 2008; 28: and other lifestyle-related confounding factors, and liver func- 166 – 172. tion tests was slightly attenuated after further adjustment for 6. Kimm H, Yun JE, Jo J, Jee SH. Low serum bilirubin level as an 27 independent predictor of stroke incidence: A prospective study in hs-CRP, but still remained significant. They suggested that Korean men and women. Stroke 2009; 40: 3422 – 3427. TB probably confers protection against the development of 7. Tanaka M, Fukui M, Tomiyasu K, Akabame S, Nakano K, diabetes.27 In the present study among apparently healthy Hasegawa G, et al. Low serum bilirubin concentration is associated Japanese men and women, TB was independently negatively with coronary artery calcification (CAC). Atherosclerosis 2009; 206: 287 – 291. associated with HbA1c in multivariate linear regressions 8. Vítek L, Jirsa M, Brodanová M, Kalab M, Marecek Z, Danzig V, adjusted for age, BMI, hs-CRP, and other cardiovascular et al. Gilbert syndrome and ischemic heart disease: A protective risk factors in male smokers and nonsmokers, and was also effect of elevated bilirubin levels. Atherosclerosis 2002; 160: 449 – independently negatively associated with the highest decile 456. 9. Kalousová M, Novotny L, Zima T, Braun M, Vítek L. Decreased ( 5.4%) of HbA1c in the nonsmokers. Thus, TB, a potent ≥ levels of advanced glycation end-products in patients with Gilbert antioxidant in vivo, was negatively associated with HbA1c syndrome. Cell Mol Biol 2005; 51: 387 – 392. adjusting for cardiovascular risk factors and so the stronger 10. Selvin E, Steffes MW, Zhu H, Matsushita K, Wagenknecht L, association of HbA1c, than of FPG, with CVD in nondiabetic Pankow J, et al. , diabetes, and cardiovascular subjects may at least partly be mediated by the inverse asso- risk in nondiabetic adults. N Engl J Med 2010; 362: 800 – 811. 11. Pani LN, Korenda L, Meigs JB, Driver C, Chamany S, Fox CS, et ciation between HbA1c and TB. al. Effects of aging on A1C levels in individuals without diabetes. Diabetes Care 2008; 31: 1991 – 1996. Study Limitations 12. The International Expert Committee. International Expert Commit- The present study was a cross-sectional study and the sub- tee report on the role of the A1C assay in the diagnosis of diabetes. Diabetes Care 2009; 32:1327 – 1334. jects were not a general population but visitors to a medical 13. Facchini FS, Hollenbeck CB, Jeppesen J, Chen YD, Reaven GM. check-up center that is in a central city of a rural region in Insulin resistance and cigarette smoking. Lancet 1992; 339: 1128 – Japan. Studies in larger populations and longitudinal studies 1130. are warranted to test the present conclusions. 14. Farin HM, Abbasi F, Kim SH, Lamendola C, McLaughlin T, Reaven GM. The relationship between insulin resistance and dys- lipidaemia in cigarette smokers. Diabetes Obes Metab 2007; 9: Conclusions 65 – 69. 15. Steinberg D, Parthasarathy S, Carew TE, Khoo JC, Witztum JL. Bilirubin was negatively associated with HbA1c independent- Beyond cholesterol: Modifications of low-density lipoprotein that ly of FPG, age, and other cardiovascular risk factors in appar- increase its atherogenicity. N Engl J Med 1989; 320: 915 – 924. 16. Duffels MGJ, Mulder KM, Trip MD, de Groot E, Gort J, van Dijk ently healthy Japanese men and women who had no history APJ. Atherosclerosis in patients with cyanotic congenital heart dis- of CVD and who were not taking antidiabetic, antihyperten- ease. Circ J 2010; 74: 1436 – 1441. sive, or antihyperlipidemic medication. The association was 17. Yesilova Z, Serdar M, Ercin CN, Gunay A, Kilciler G, Hasimi A, weak in male smokers. It suggests that the stronger associa- et al. Decreased oxidation susceptibility of plasma low density lipoproteins in patients with Gilbert’s syndrome. J Gastroenterol tion of HbA1c, than FPG, with CVD in nondiabetic subjects Hepatol 2008; 23: 1556 – 1560. may at least partly be mediated by the inverse association 18. Yamaguchi T, Terakado M, Horio F, Aoki K, Tanaka M, Nakajima between HbA1c and TB. H. Role of bilirubin as an antioxidant in an ischemia-reperfusion of rat liver and induction of heme oxygenase. Biochem Biophys Res Commun 1996; 223: 129 – 135. Acknowledgments 19. Vítek L, Kráslová I, Muchová L, Novotný L, Yamaguchi T. Uri- The authors thank all subjects who participated in the study, the para- nary excretion of oxidative metabolites of bilirubin in subjects with medical staff at Tachikawa Medical Center who assisted with the study, Gilbert syndrome. J Gastroenterol Hepatol 2007; 22: 841 – 845. and Dr Shinpei Yoshii and Dr Masaaki Okabe at Tachikawa Medical 20. Yamamoto M, Maeda H, Hirose N, Yamamoto M, Nakagawa A, Center, Professor Yoshifusa Aizawa at Niigata University Graduate Radhakrishnan G, et al. Biphasic elevation of bilirubin oxidation School of Medical and Dental Sciences, and Professor Kenichi Watanabe during myocardial ischemia reperfusion. Circ J 2008; 72: 1520 – at Niigata University of Pharmacy and Applied Life Sciences for their 1527. assistance to set up the study. 21. Ueyama T, Kawabe T, Hano T, Tsuruo Y, Ueda K, Ichinose M, et al. Upregulation of heme oxygenase-1 in an animal model of Takotsubo cardiomyopathy. Circ J 2009; 73: 1141 – 1146. Disclosures 22. Kunii H, Ishikawa K, Yamaguchi T, Komatsu N, Ichihara T, The authors received no grant and have no conflict of interest to dis- Maruyama Y. Bilirubin and its oxidative biopyrrins in close. patients with acute myocardial infarction. Fukushima J Med Sci 2009; 55: 39 – 51. 23. Inoguchi T, Sasaki S, Kobayashi K, Takayanagi R, Yamada T. References Relationship between Gilbert syndrome and prevalence of vascular 1. Stocker R, Yamamoto Y, McDonagh AF, Glazer AN, Ames BN. complications in patients with diabetes. JAMA 2007; 298: 1398 – Bilirubin is an antioxidant of possible physiological importance. 1400. Science 1987; 235: 1043 – 1046. 24. The International Expert Committee on the Diagnosis and Classi- Advance Publication by J-STAGE ODA E et al.

fication of Diabetes Mellitus. Report of the International Expert Res Clin Pract 2010; 88: 103 – 110. Committee on the diagnosis and classification of diabetes mellitus. Diabetes Care 1997; 20: 1183 – 1197. 25. Hashimoto K, Osugi T, Noguchi S, Morimoto Y, Wasada K, Imai Supplementary files S, et al. A1C but not serum glycated is elevated because Table S1. Multivariate Linear Regression Using TB as a Dependent of iron deficiency in late in diabetic women. Diabetes Variable Care 2010; 33: 509 – 511. Table S2. Multivariate Linear Regression Using HbA1c as a Depen- 26. Kim C, Bullard KM, Herman WH, Beckles GL. Association be- dent Variable tween iron deficiency and A1C levels among adults without dia- Table S3. Multivariate Logistic Regression Using HbA1c ≥5.4% as a betes in the National Health and Nutrition Examination Survey, Dependent Variable 1999–2006. Diabetes Care 2010; 33: 780 – 785. 27. Ohnaka K, Kono S, Inoguchi T, Yin G, Morita M, Adachi M, et al. Table S4. Multivariate Logistic Regression Using HbA1c ≥5.2% as a Inverse associations of serum bilirubin with high sensitivity C-reac- Dependent Variable tive protein, glycated hemoglobin, and prevalence of type 2 diabetes Please find supplementary file(s); in middle-aged and elderly Japanese men and women. Diabetes http://dx.doi.org/10.1253/circj.CJ-10-0645