High Body Mass Index is an Important Risk Factor for the Development of Type 2 Diabetes

Abstract

OBJECTIVE

The aim of this study is to establish a causal relationship between excess body weight and the onset of diabetes in a retrospective cohort study.

METHODS

This 10-year observational cohort study investigated 969 men and 585 women (23 to 80 years of age), who underwent voluntary complete medical check-ups and an annual 75-g oral glucose tolerance test (75g-OGTT). Participants with fasting plasma glucose ≥ 126 mg/dl, 2-h glucose level in a 75g-OGTT ≥ 200 mg/dl and/or received medical treatment for type 2 diabetes during the previous year were counted as new-onset diabetics. We assessed the independent contribution of increased BMI to the risk of developing type 2 diabetes with Cox proportional hazard model.

RESULT

During the follow-up period, we diagnosed 86 men and 49 women with new-onset type 2 diabetes. In the Cox proportional hazards model, the risk of diabetes mellitus increased with increasing BMI, even after adjusting for age, sex, blood pressure, metabolic profiles, and insulin resistance. In the final model, setting BMI less than 25 as a reference group, the Hazard ratios for diabetes mellitus was 3.12 for those with a BMI of 25–27.4 and increased to 3.80 for participants with a BMI of 27.5 or higher.

CONCLUSION

Overweight/obesity (high BMI) is an independent and dose-dependent risk factor for type 2 diabetes in overweight Japanese patients. Our results confirmed the usefulness of BMI as a classic parameter, and the importance of lifestyle modification and better management among people with overweight/obesity for prevention of type 2 diabetes mellitus

INTRODUCTION

In many developed countries, including Japan, the prevalence of type 2 diabetes has increased dramatically in recent years ^1–4. One of the major health concerns believed to be associated with this phenomenon is the rapid increase in obesity, considered to be caused by unhealthy eating habits and lifestyle patterns ^{5– 6}.

A number of studies have reported a strong relationship between obesity/overweight and the onset of type 2 diabetes ^7–10. However, there are many coexisting conditions, such as hypertension, dyslipidemia, or insulin resistance, that may affect this relationship^{7, 11}. In addition, in many previous studies, the onset of diabetes was based on the initiation of treatment with a hypoglycemic agent or elevated fasting plasma glucose (FPG) levels. These diagnostic methods are problematic because diabetes is often asymptomatic in early stages and can only be diagnosed by a glucose tolerance test.

In this study, we used data obtained from annual voluntary complete medical check-ups, including a 75g-OGTT. In addition, we either adjusted the data or excluded participants from the final analysis to eliminate confounders such as hypertension, existing diabetes/borderline glucose tolerance, dyslipidemia, hyperuricemia and insulin resistance. This facilitated the assessment of the effect of obesityand overweight, per se, on the risk of new-onset type 2 diabetes mellitus.

METHODS

This was a retrospective cohort study of patients with new onset of type 2 diabetes. We collected data from 2,605 individuals who had voluntarily completed medical check-ups, including an annual 75g-OGTT from 1994 to 1996 in two adjacent hospitals in Fukushima prefecture, Japan. These Participants had undergone 75g-OGTT after overnight fasting. Of those, we excluded 286 men and 100 women with borderline glucose tolerance (75g-OGTT FPG of 110–125 mg/dl and/or 2-hr plasma glucose level of 140–199 mg/dl), 65 men and 14 women with diabetes (FPG ≥ 126 mg/dl and/or 75g-OGTT 2-hr plasma glucose level ≥ 200 mg/dl, or on hypoglycemic agents)¹². We also excluded 166 men and 53 women with hypertension (systolic blood pressure (SBP) ≥140 mmHg and/or diastolic blood pressure (DBP) ≥90 mmHg), or on antihypertensive agents because hypertension is known to be an important risk factor for diabetes ^13–15. Of the remaining 1,921 individuals, we analyzed only those 1,554 individuals (969 men and 585 women) with medical check-up data available for 10 consecutive years.

We defined the onset of type 2 diabetes as FPG ≥ 126 mg/dl, 2-h measured value for 75g-OGTT ≥ 200 mg/dl and/or initiation of treatment for type 2 diabetes within the previous year. Body height and weight were measured in the standing position. Body mass index (BMI) was calculated based on body weight (kg) divided by height squared (m²). Both SBP and DBP were obtained as the mean of two measurements on the upper arm after the subject had been seated for at least 5 minutes. Total cholesterol (TC) and triglyceride (TG) levels were measured using the enzymatic method. High-density lipoprotein cholesterol (HDL-C) was measured by the enzymatic method. Serum uric acid level (UA) was measured on a standard autoanalyzer by the uricase-peroxidase method. Plasma glucose levels were measured by the glucose dehydrogenase method. Immunoreactive insulin (IRI) was measured by enzyme immunoassay. The homeostasis model assessment-insulin resistance (HOMA-IR) index was obtained, using the formula: HOMA-IR = [FPG (mg/dl) × IRI (IU/ml)]/405^16–17.

First, we assessed the clinical characteristics of male and female participants at baseline and estimated the differences between the two genders using the Mann-Whitney method, because variables did not have a normal distribution. Second, we analyzed the baseline characteristics of the diabetic and non-diabetic groups. We compared the groups based on age, sex, BMI, SBP, DBP, TC, HDL-C, TG, UA, FPG, 2-hour PG, IRI, and HOMA-IR at baseline. Cut-off values for this study were: age ≥45 years; BMI ≥ 25, SBP > 130 mmHg and/or DBP > 85 mmHg; TC ≥ 220 mg/d1; TG ≥ 150 mg/dl; HDL-C < 40 mg/dl; UA ≥ 6.0 mg/dl; FPG ≥ 110 mg/dl; 2-hour PG ≥ 120 mg/dl and HOMA-IR ≥ 2.0^16–17, according to criteria established by the NCEP-ATPIII¹⁸, International Diabetes Federation 2005 criteria of the metabolic syndrome¹⁹ and the American Diabetes Association²⁰. Then, we compared those who developed diabetes and those who did not to estimate the odds ratio of each risk factor for the onset of type 2 diabetes, using a univariate logistic regression analysis. Third, we categorized the participants into three groups based on BMI: normal weight (BMI < 25 kg/m²), moderately overweight (BMI 25–27.4 kg/m²), and markedly overweight/obese (BMI ≥ 27.5 kg/m²) ¹⁸. As for category of overweight/ obesity, there were a few subjects with severe obesity such as BMI 30 kg/m²or over, and we decided BMI 27.5 kg/m² as cut off of mid between 25 and 30 by reference to prevalence of BMI. For each group, we obtained a diabetes-free survival curve using the Kaplan-Meier method and examined significant differences using the log-rank test. Only values of P < 0.05 were considered statistically significant. Finally, we applied the Cox proportional hazards model to assess the independent contribution of BMI to the risk of developing type 2 diabetes. To obtain hazard ratios, we adjusted for age, sex, SBP, DBP, TC, TG, HDL-C and HOMA-IR; Model 1(age and sex), Model 2 (age, sex, SBP and DBP), Model 3(age, sex, SBP, DPB, TC, HDLC, TG and UA), Model 4 (age, sex, SBP, DPB, TC, HDLC, TG, UA, FPG, 2h-PG and HOMA-IR). All statistical analyses were performed using SPSS software (version 16.0, SPSS Inc., IL, USA).

Our survey was conducted in compliance with the Ethical Guidelines for Epidemiological Studies established by the Japanese government²¹, and the work was performed in accordance with the Declaration of Helsinki of 1975 (revised in 2000)²². Fukushima Welfare Federation of Agricultural Cooperatives Council reviewed and approved the research protocol.

RESULTS

Table 1 shows baseline clinical characteristics of 969 men and 585 women. The median age of men was 50 years old (minimum 23, maximum 79), and 51 years (30, 80) for women. The BMI of participants ranged from 16.6 to 32.5 kg/m² (median: 22.9 kg/m²) in men, and from 15.8 to 37.8 kg/m² (median: 23.3 kg/m²) in women. The proportions of high BMI (≥25) were 28.5 % in menand 24.5 % in women. The median SBP was 121 mm Hg (89, 138) and that of DBP was 78 mm Hg (20, 89) in men, and 120 mm Hg (81, 139) and 72 mm Hg (45, 89) in women, respectively.

Table 1.

Characteristics at baseline survey (N=1264)

	Median (minimum, maximum) N (%)
	Men (N=969)	Women (N=585)	P valuea)
Age (years)	50 (23, 79)	51 (30, 80)	*
Body mass index (BMI)	22.9 (16.6, 32.5)	23.3 (15.8, 37.8)
(<25)	690 (71.2)	442 (75.6)
(25.0–27.4)	206 (21.3)	104 (17.8)
(≥27.5)	73 (7.5)	39 (6.7)
Hypertension-related factors
Systolic blood pressure (mmHg)	121 (89, 138)	120 (81, 139)
Diastolic blood pressure (mmHg)	78 (20, 89)	72 (45, 89)	**
Lipid profiles
Total cholesterol (TC) (mg/dl)	192 (98, 319)	195 (91, 356)	*
Triglyceride (TG) (mg/dl)	95 (24, 942)	77 (21, 690)	**
High density lipoprotein cholesterol (HDLC) (mg/dl)	51 (20, 117)	57 (24,132 )	**
Uric acid (UA) (mg/dl)	5.6 (1.9, 11.1)	4.5 (0.7, 8.8)	**
Fasting plasma glucose concentration (FPG) (mg/dl)	96 (73, 109)	93 (71, 109)	**
Two hour post load (2h) plasma glucose concentration (mg/dl)	113 (27, 139)	112 (33, 139)
Homeostasis model assessment as an index of insulin resistance (HOMA-IR)	1.63 (0.15, 8.68 )	1.66 (0.56, 4.37)	**

^a)* P<0.05, ** P<0.01, Mann-Whitney method or Chi square test with P value <0.05 was considered statistically significant.

Additional variables were as follows: TC [median 192 mg/dl (minimum 98, maximum 319) in men,195 mg/dl (91, 356) in women], TG [95 mg/dl (24, 942) in men,77 mg/dl (21, 690) in women], HDL-C [51 mg/dl (20, 117) in men, 57 mg/dl (24, 132) in women], UA [5.6 mg/dl (1.9, 11.1) in men, 4.5 mg/dl (0.7, 8.8) in women], FPG [96 mg/dl(73, 109) in men, 93 mg/dl (71, 109) in women] and 2h-PG [ 113 mg/dl (27, 139) in men, 112 mg/dl (33, 139) in women] and HOMA-IR [1.63 (0.15, 8.68) in men, 1.66 (0.56, 4.37) in women].

During the 10-year observation period, 135 subjects (86 men, and 49 women) developed type 2 diabetes. Table 2 shows factors associated with developing type 2 diabetes after ten-years of follow-up. The analysis indicated that overweight/obesity (BMI ≥ 25.0 kg/m²) is a strong risk factor [Odds ratio (OR) = 4.98, 95% confidence interval (CI) = 3.32–7.47 for BMI 25.0–27.4, and OR = 7.72, 95% CI = 4.64–12.87 for BMI ≥ 27.5]. In addition, elevated TG (≥ 150 mg/dl) (OR = 4.44, 95% CI = 3.07–6.44), UA (≥ 6.0mg/dl) (OR = 2.09, 95% CI = 1.46–3.00), FPG (100–109 mg/dl) (OR = 2.02, 95% CI = 1.41–2.91), 2h-PG (120–139 mg/dl) (OR = 1.83, 95% CI = 1.28–2.60) and HOMA-IR (≥2.0) (OR = 2.95, 95% CI = 2.05–4.25), and decreased HDL-C (< 40 mg/dl) (OR = 2.56, 95% CI = 1.72–3.81) were significantly associated with increased risk of new-onset type 2 diabetes.

Table 2.

Odds ratios of the factors associated with onset of Diabetes Mellitus

		N (%)		Odds ratio	95% confidence interval	P valueb)
		Subjects with DMa) (N=135)	Subjects without DM (N=1504)
Age (years)	≥45	101 (74.8)	1020 (71.9)	1.16	0.78–1.74
Male sex		86 (63.7)	883 (62.2)	1.07	0.74–1.54
Body mass index (BMI)	<25.0	49 (36.3)	1083 (76.3)		-
	25.0–27.4	57 (42.2)	253 (17.8)	4.98	3.32–7.47	**
	≥27.5	29 (21.5)	83 (5.8)	7.72	4.64–12.87	**
Systolic blood pressure (mmHg)	≥130	48 (35.6)	443 (31.2)	1.22	0.84–1.76
Diastolic blood pressure (mmHg)	≥85	14 (10.4)	141 (9.9)	1.05	0.59–1.87
Total cholesterol (TC) (mg/dl)	≥220	37 (27.4)	287 (20.2)	1.49	0.99–2.22
Triglyceride (TG) (mg/dl)	≥150	59 (43.7)	211 (14.9)	4.44	3.07–6.44	**
High density lipoprotein cholesterol (HDLC) (mg/dl)	<40	40 (29.6)	200 (14.1)	2.56	1.72–3.81	**
Uric acid (UA) (mg/dl)	≥6.0	59 (44.4)	391 (27.6)	2.09	1.46–3.00	**
Fasting plasma glucose concentration (FPG) (mg/dl)	100–109	55 (40.7)	360 (25.4)	2.02	1.41–2.91	**
Two hour post load plasma glucose concentration (mg/dl)	120–139	68 (50.4)	507 (35.7)	1.83	1.28–2.60	**
Homeostasis model assessment as an index of insulin resistance (HOMA-IR)	≥2.0	61 (45.9)	314 (22.3)	2.95	2.05–4.25	**

^a)Diabetes Mellitus,

^b)* P<0.05, ** P<0.01

A Kaplan-Meier survival curve for 10 years showed the cumulative probability of developing type 2 diabetes to be higher in the participants who were moderately (BMI 25.0–27.4 kg/m²) or markedly overweight/obese (BMI ≥ 27.5 kg/m²) (Figure 1).

Figure 1.

Figure 1 shows the cumulative probability of remaining free of diabetes, for subjects classified as moderately overweight (BMI 25.0–27.4 kg/m²), markedly overweight/obese (BMI ≥27.5 kg/m²) or normal weight (BMI <25 kg/m²). The probability of developing diabetes was significantly higher in participants who were overweight or obese compared to those of normal weight (P<0.01). P values were determined using the log-rank test to account for differences among the groups.

We used Cox proportional hazards model to determine the association between BMI and the risk of developing type 2 diabetes with the three groups stratified by BMI (Table 3). In Model 1, the Hazard ratios (HRs) adjusted only for age and sex were 4.53 (P<0.01) in those with a BMI of 25.0 – 27.4, and 6.61 (P<0.01) in those with a BMI ≥ 27.5, compared to those with a BMI < 25.0. After adjusting other associated factors, we also observed the positive relationship between increased BMI and new onset of type 2 diabetes; Model 2 [HR = 4.40 (P<0.01)in those with a BMI of 25.0 – 27.4, HR = 6.41 in those with a BMI ≥ 27.5 (P<0.01)], Model 3 [HR = 3.35 (P<0.01) in BMI 25.0 – 27.4, HR = 4.17 (P<0.01) in BMI ≥ 27.5], Model 4 [HR= 3.12 (P<0.01) in BMI 25.0 – 27.4, HR = 3.80 (P<0.01) in BMI ≥ 27.5].

Table 3.

Cox’s proportional hazards model of the incidence of type 2 Diabetes Mellitus over 10-years according to Body Mass Index

Body Mass Index (BMI)	Hazard Ratio (95% confidence Interval)
	Model 1a)			Model 2b)			Model 3c)			Model 4d)
<25.0 (N=1193)	1.00			1.00	-		1.00	-		1.00	-
25.0–27.4 (n=341)	4.53	3.09–6.64	**	4.40	2.99–6.46	**	3.35	2.25–4.99	**	3.12	2.09–4.66	**
≥27.5 (n=118)	6.61	4.17–10.46	**	6.41	4.04–10.18	**	4.17	2.59–6.71	**	3.80	2.36–6.12	**

^a)Adjusted with age and sex,

^b)Adjusted with age, sex, SBP and DBP,

^c)Adjusted with age, sex, SBP, DPB, TC, HDLC, TG and UA,

^d)Adjusted with age, sex, SBP, DPB, TC, HDLC, TG, UA, FPG, 2h-PG and HOMA-IR

^**p<0.01

Discussion

Our current data indicate that in Japanese increased BMI is an independent and dose-dependent risk factor for type 2 diabetes even for those who are only moderately overweight. Oguma et al.,⁹ in a prospective cohort study of 20,187 individuals, showed that BMI is strongly associated with increased risk of diabetes with a positive dose-response relationship in multivariate analyses after adjusting for physical activity, smoking, hypertension, and a family history of diabetes. In Japan, Nagaya et al.,⁸ also performed a follow-up study of 16,829 men and 8,370 women who were apparently healthy at baseline. That study demonstrated that the multivariate-adjusted hazard ratio for an increased BMI of 1 kg/m² was about 25 percent⁸. The present study is unique in that we followed 1,652 individuals for an entire decade during which the onset of diabetes was determined annually using the 75g-OGTT. Wannamethee et al.,¹⁰ reported that a strong positive relationship between BMI and the onset of diabetes remained significant after adjusting for SBP and for TC in a 20-year prospective study of British men. The current study adds another variable to that report; we found that the risk of type 2 diabetes increased with increased BMI after adjusting for hyperlipidemia and insulin resistance. Meigs et al.,⁷ compared the age-sex-adjusted multivariate relative risk for the incidence of type 2 diabetes with or without insulin resistance in Caucasian subjects. Even among participants with insulin-sensitivity, severe obesity remained a significant risk factor for the development of type 2 diabetes. These results showed that increased BMI, even in the moderately overweight, is a strong risk factor for diabetes independent of insulin resistance in Japanese.

The DECODE study indicated that 31% of people with diabetes who are hyperglycemic based on 2 hr plasma glucose levels have normal FPG concentrations²³. In order to accurately evaluate glucose tolerance, we selected only those participants with complete medical check-up data with results from an annual 75g-OGTT. Our analysis indicates that the prevalence of type 2 diabetes was much higher throughout the observation period in moderately overweight, and markedly overweight or obese participants than in those of normal weight after excluding those subjects whose 2 hr were 140 mg/dl or greater. Thus, our results may indicate that 75 g-OGTT is useful for obesity/overweight subjects with even normal glucose tolerance for early detection or prevention of type 2 diabetes.

Hypertension is an important risk factor for diabetes in the present study. In a prospective study of 12,550 hypertensive adults, it has been reported that the incidence of diabetes is 29.1 per 1,000 person-years in hypertensive subjects, and only 12.0 per 1,000 person-years in normotensive subjects, with a relative risk of 2.43 (95% CI= 2.16–2.73)¹¹. Some investigators have described the mechanism underlying the association between diabetes and hypertension. It has been suggested that insulin resistance leads to hypertension. Many hypertensive subjects take angiotensin II type I receptor blockers (ARB) that have positive effects on insulin resistance²⁴. In contrast, diabetes is more likely to develop among participants taking a beta-adrenergic blocker than those not on this medication ¹¹. We examined in normotensive subjects, the effect of blood pressure on the incidence of diabetes using the Cox proportional hazards model. Our data indicate that the hazard ratio for the onset of diabetes was not significantly different when we adjusted for other factors associated with the onset of diabetes. With the conceptualization of metabolic syndrome, interest in BMI has declined. This study confirmed that BMI is an independent risk factor for developing type 2 diabetes, although we could not compare directly BMI and waist circumference. In Japan, the proportion of obese subjects is increasing in all age groups of men. It is a significant problem that parallels the increase in the incidence of type 2 diabetes cases. The Ministry of Health, Labour and Welfare aims to decrease the number of diabetic patients by launching anti-obesity measures in their planned “Kenko Nippon 21”. However, the effects are not sufficient. This study provides valuable data which should prompt effective measures.

Our study has several limitations, including a selection bias; the study was conducted in two adjacent hospitals in Fukushima prefecture, Japan, and the participants were limited to those who were taken medical check-up. It is possible that participants could have had a greater awareness about their health condition and healthy lifestyles. In addition, important pieces of information, including patient background information, such as family history, alcohol use, smoking habits and/or the level of education, were lacking. A large scale multicenter study with detailed information is needed.

Conclusion

Increased BMI was an independent and dose-dependent risk factor for type 2 diabetes among overweight Japanese study participants. Our results suggested the usefulness of BMI which seems to be a classic parameter, and the importance of lifestyle modification and better management among people with overweight/obesity for prevention of future type 2 diabetes mellitus.