Casual serum triglyceride as a predictor of premature type 2 diabetes mellitus: an 8-year cohort study of middle-aged Japanese workers

Abstract

Background

The utility of casual serum triglyceride (TG) as a predictor of type 2 diabetes mellitus (DM) is unclear, especially during the most productive years.

Methods

Participants were 3271 workers (913 men and 2358 women, age 20–57) without DM at baseline. They underwent consecutive annual medical check-ups for 8 years. The association between newly diagnosed DM and casual serum TG level was determined by classifying the participants into 4 groups according to casual serum TG level at baseline: below 50 mg/dL (group A), 50–100 mg/dL (group B), 100–150 mg/dL (group C), and ≥150 mg/dL (group D). The effects of casual serum TG level in combination with sex, obesity, or serum glucose level on newly diagnosed DM were also evaluated.

Results

A total of 222 newly diagnosed type 2 DM cases with a mean age of 50 years old were observed during the follow-up period, i.e., 10/406 in group A, 66/1534 in group B, 58/712 in group C, and 88/619 in group D. Compared with group A, the odds ratio (ORs) for newly diagnosed DM (after adjusting for DM-associated factors) was found to increase with casual serum TG level: 1.38 (group B), 1.79 (group C), and 2.36 (group D). Moreover, the OR for newly diagnosed DM was higher in participants with high casual serum TG levels who were also male (OR 2.46), obese (OR 4.18), or had a high serum glucose level (OR 6.96) than in the reference group.

Conclusions

Serum TG level ≥150 mg/dL when fasting or nonfasting is a significant predictor of type 2 diabetes in middle-aged Japanese workers.

Introduction

Insulin resistance is a predisposing factor for metabolic syndrome, leading to the development of type 2 diabetes and cardio- and cerebrovascular diseases over a long period [1–4]. Detecting individuals with insulin resistance and modifying their lifestyle accordingly from their early years are very important aims when attempting to prevent type 2 diabetes. Various methods of evaluating individual insulin resistance, such as the homeostatic model assessment of insulin resistance (HOMA-IR), have been developed [5–8]; however, these are expensive and frustratingly complicated. Therefore, a simple alternative method of screening for insulin resistance or monitoring it in medical check-ups is desired [9].

High levels of fasting serum triglycerides (TGs) have been reported to be a marker of insulin resistance [10, 11] and a predictor of type 2 diabetes [12–19]. Ectopic fat deposition in visceral adipose tissues is considered to play an important role in the pathogenesis of insulin resistance [20–23]: fatty acid synthesized in visceral adipose tissues is delivered to the liver and promotes the synthesis of very low density lipoprotein (VLDL), which induces hypertriglyceridemia, leading to insulin resistance [20, 24, 25]. Thus, the association between the serum level of TG and the development of type 2 diabetes has been investigated.

Compared with fasting serum TG levels, less attention has been paid to non-fasting serum TG levels, because fasting overnight is recommended before attempting to obtain a lipid profile [26]. A few studies have, however, demonstrated that non-fasting serum TG levels predict the development of type 2 diabetes [13, 27]. Also, in some settings it is unrealistic to require all participants to fast, especially those who visit in the afternoon. Furthermore, in Japan, general screening for CVD risk factors is performed under non-fasting conditions to improve participation rates.

We therefore focused on assessing whether the casual serum TG level, regardless of the content of the meal consumed before the lipid profile is obtained and the time since that last meal, could be used to predict the development of premature diabetes mellitus (DM) in a cohort study of relatively young Japanese workers.

Methods

The cohort consisted of 6045 individuals who underwent medical examination in 2004. These individuals were aged 20–57 years and worked in a retail company in Japan. Among them, 5833 did not have DM at baseline, and 9 of those were pregnant women who were therefore excluded from the study. Of the remaining 5824 subjects, 3271 (2358 females and 913 males; mean age 45.4 years) who underwent a medical examination every year for 8 years were eligible for analysis in the present study. This study was approved by the ethics committee of the Shiga University of Medical Science (24-18).

All of the medical examinations performed included a questionnaire, physical examination, and laboratory tests. Among the results of the annual examinations, those for age, sex, medical history, alcohol intake, smoking history, blood pressure, height, weight, casual serum glucose, total cholesterol (TC), high density lipoprotein cholesterol (HDLC), and TG were used in the present study. Body mass index (BMI) was calculated from height and weight. Obesity was defined as a BMI level of 25.0 kg/m² or more. Optimal blood pressure was defined as systolic blood pressure (SBP) <120 mmHg and diastolic blood pressure (DBP) <80 mmHg [28].

Following the classification of the Committee of the Japan Diabetes Society on the Diagnostic Criteria of Diabetes Mellitus, the participant was considered to have DM if the fasting glucose level was ≥126 mg/dL (7.0 mmol/L); the casual glucose level was ≥200 mg/dL (11.1 mmol/L); or they used antidiabetic agents. Participants with a fasting serum glucose level of <110 mg/dL (6.1 mmol/L) or a casual serum glucose level of <140 mg/dL (7.8 mmol/L) were classified as “normal type”. Those who did not have DM and were not normal type were classified as “borderline type”. The first development of DM during follow-up was used as the outcome variable. The date of the examination when a new case of DM was identified was considered the date of the first development of DM. The new case of DM was also defined according to the classification of the Committee of the Japan Diabetes Society on the Diagnostic Criteria of Diabetes Mellitus, or based on the use of antidiabetic agents.

All subjects were divided into 4 groups according to the level of serum TG at baseline (in 2004): <50 mg/dL (group A), 50–100 mg/dL (group B), 100–150 mg/dL (group C), and ≥150 mg/dL (group D). The risk characteristics of each group at the baseline survey were described as means and standard deviations (SDs) for continuous variables and proportions for categorical variables. The odds ratio (OR) of each group for the development of DM during the 8-year monitoring period in comparison with group A was calculated using a multivariate logistic regression model after adjusting for age, sex, blood pressure, casual serum glucose level, total cholesterol, HDL cholesterol, smoking, and alcohol intake. Furthermore, the association between casual serum TG level and the development of DM was evaluated via combinations of casual serum TG level and sex, BMI, and casual serum glucose level, using 150 mg/dL of serum TG as a cutoff point [26, 29]. All statistical analyses were performed with SPSS version 22.0 for Windows (SPSS Inc.). All significance tests were two-tailed, and p < 0.05 was considered to indicate significance in all analyses.

Results

Groups A, B, C, and D included 406, 1534, 712, and 619 subjects, respectively (total: 3271 subjects). The baseline characteristics of these 4 groups are presented in Table 1. There was no significant difference in mean age between group A (44.4 ± 6.7 years) and group D (45.0 ± 6.8 years); the male to female ratio was higher in group D than in group A; and the number of typical risk factors for DM in group D was greater than those in the other groups. During the observation period, 222 subjects were newly diagnosed with DM (137 upon fasting and 85 postprandially): 2.5 % (n = 10) in group A, 4.3 % (n = 66) in group B, 8.1 % (n = 58) in group C, and 14.2 % (n = 88) in group D. The mean age at which DM developed was 49.8 ± 7.4 years, and the mean time taken to develop DM after the baseline survey was 4.5 ± 2.1 years. There was no significant difference in the OR for developing DM between the anterior half and the posterior half of the observation period.

Table 1.

Baseline characteristics of the participants in groups A–D (i.e., participants were categorized according to their levels of casual serum triglyceride)

	TG < 50	100 > TG ≥ 50	150 > TG ≥ 100	TG ≥ 150	Total
Subjects	406	1535	712	620	3271
Number of males	32 (7.9 %)	280 (18.3 %)	244 (34.3 %)	357 (57.7 %)	913 (27.9 %)
Current smokers	91 (22.4 %)	415 (27.0 %)	223 (31.3 %)	251 (40.5 %)	980 (29.9 %)
Alcohol intake	254 (62.6 %)	899 (58.6 %)	424 (59.6 %)	413 (66.7 %)	1990 (60.8 %)
Age (years)	44.4 ± 6.7	45.6 ± 6.4*	45.9 ± 6.4*	45.0 ± 6.8	45.4 ± 6.5
SBP (mmHg)	120.8 ± 17.9	122.4 ± 15.3	126.0 ± 16.0*	130.9 ± 15.7*	124.6 ± 16.2
DBP (mmHg)	72.4 ± 12.0	73.5 ± 11.2	76.6 ± 11.2*	79.5 ± 11.2*	75.2 ± 11.6
Glucose (mg/dL)	93.5 ± 12.7	95.2 ± 14.7	99.1 ± 18.7*	101.4 ± 18.3*	97.0 ± 16.4
T-chol (mg/dL)	195.6 ± 31.1	204.7 ± 31.8*	213.5 ± 34.2*	222.4 ± 36.8*	208.8 ± 34.3
HDL chol (mg/dL)	79.2 ± 15.6	71.9 ± 15.5*	62.1 ± 15.0*	51.4 ± 12.2*	66.8 ± 17.4
BMI	21.3 ± 3.3	21.9 ± 3.5*	23.5 ± 3.8*	24.9 ± 4.0*	22.8 ± 3.9
Sampling performed while fasting	123 (30.3 %)	326 (21.3 %)	115 (16.2 %)	60 (9.7 %)	624 (19.1 %)

TG casual serum triglyceride level

* p < 0.05 when mean value was compared with that of the TG <50 group using Dunnett’s C test

The OR for the development of DM increased as the level of casual serum TG at baseline increased; in comparison with group A, the OR after adjusting for age and sex (model 1) was 1.64 (0.83–3.23) in group B, 2.92 (1.46–5.85) in group C, and 4.81 (2.41–9.59) in group D (Table 2). A similar tendency was observed after adjusting for age, sex, blood pressure (<120/80 or ≥120/80 mmHg), casual serum glucose (<140 or ≥140 mg/dL), BMI (<25 or ≥25), currently smoking, and currently drinking alcohol (model 2), and after adjusting for all of these variables plus TC (<220 or ≥220 mg/dL) and HDLC (≥40 or <40 mg/dL) too (model 3).

Table 2.

Association of serum triglyceride level with the development of diabetes mellitus

	TG < 50	100 > TG ≥ 50	150 > TG ≥ 100	TG ≥ 150
Development of DM	10	66	58	88
Odds ratio (95 % CI)
Model1		1.64 (0.83–3.23)	2.92 (1.46–5.85)	4.81 (2.41–9.59)
Model2		1.42 (0.71–2.81)	1.95 (0.96–3.95)	2.85 (1.41–5.77)
Model3		1.38 (0.69–2.74)	1.82 (0.89–3.69)	2.36 (1.15–4.84)

Model 1: adjusted for age and sex

Model 2: as for model 1 but adjusted for blood pressure (<120/80, ≥120/80 mmHg), casual serum glucose (<140, ≥140 mg/dL), BMI (<25 or ≥25), currently smoking, and currently drinking alcohol (more often than sometimes, or not) too

Model 3: as for model 2 but adjusted for total cholesterol (<220, ≥220 mg/dL) and HDL-cholesterol (>40, ≤40 mg/dL) too

Table 3 shows the multivariate adjusted ORs for the development of DM for different combinations of casual serum TG level and sex, BMI, or casual serum glucose level. For all three combinations, the ORs for developing DM were higher than for the corresponding reference group (female/lower TG group, lower BMI/lower TG group, and normal casual serum glucose level/lower TG group); a higher casual serum TG level was strongly associated with a higher OR for the development of DM. Even in relatively low-risk groups such as those with female participants, lower BMIs, or normal casual serum glucose levels, a high casual serum TG level was shown to be a risk for the development of DM (fully adjusted ORs: 1.55, 1.84 and 1.57, respectively).

Table 3.

Associations of serum triglyceride level in combination with sex, BMI, or casual serum glucose level with the development of diabetes mellitus

	Female and TG <150	Female and TG ≥150	Male and TG <150	Male and TG ≥150
Development of DM	88/2096	27/262	46/556	61/357
Model 1a		2.58 (1.64–4.06)	2.14 (1.46–3.14)	4.87 (3.40–6.95)
Model 2c		1.78 (1.11–2.86)	1.66 (1.09–2.52)	2.95 (1.97–4.44)
Model 3		1.55 (0.96–2.52)	1.64 (1.07–2.50)	2.46 (1.60–3.78)
	BMI <25 and TG <150	BMI <25 and TG ≥150	BMI ≥25 and TG <150	BMI ≥25 and TG ≥150
Development of DM	73/2150	33/348	61/502	55/271
Model 1b		2.36 (1.51–3.68)	3.84 (2.68–5.50)	5.73 (3.84–8.55)
Model 2d		2.15 (1.37–3.39)	3.40 (2.34–4.94)	5.15 (3.40–7.79)
Model 3		1.84 (1.16–2.95)	3.23 (2.22–4.71)	4.18 (2.70–6.48)
	Serum glucose <140 and TG <150	Serum glucose <140 and TG ≥150	Serum glucose ≥140 and TG <150	Serum glucose ≥140 and TG ≥ 150
Development of DM	117/2584	78/594	17/68	10/25
Model 1b		2.52 (1.82–3.48)	6.20 (3.44–11.15)	10.81 (4.67–24.99)
Model 2e		1.84 (1.31–2.57)	6.04 (3.27–11.15)	7.74 (3.24–18.52)
Model 3		1.57 (1.10–2.24)	6.32 (3.41–11.73)	6.96 (2.90–16.67)

^a Model 1: adjusted for age; model 1b: adjusted for age and sex

^c,d,e Model 2: as for model 1^a,b but also adjusted for blood pressure (<120/80, ≥120/80 mmHg), casual serum glucose (<140, ≥140 mg/dL) (but not in ^e), BMI (<25, ≥25) (but not in ^d), currently smoking, and currently drinking alcohol (more often than sometimes, or not)

Model 3: as for model 2^c,d,e but also adjusted for total cholesterol (<220, ≥220 mg/dL) and HDL-cholesterol (>40, ≤40 mg/dL)

Discussion

This is the first cohort study to clarify the utility of casual serum TG level as a predictor of premature DM, for which the mean age of incidence is about 50 years old. The present study demonstrated that higher casual serum TG level was a risk for DM development in a healthy and relatively young Asian population. In combination with sex, BMI, and casual serum glucose level, higher casual serum TG level was significantly associated with the development of DM, even after adjusting for the effects of other DM-associated factors.

Although the change in serum TG level after normal food intake is reported to be slight [30–32], the non-fasting state should not be ignored because periods of fasting during the day tend to be very short in most individuals, considering current lifestyles (eating 3 times a day, sometimes with snacks between meals) [33]. In fact, postprandial serum TG levels are considered pro-atherogenic [34, 35], and are strongly associated with future cardiovascular diseases [31, 35–39]. The mechanism of this phenomenon is probably explained by the finding that non-fasting TG levels are associated with remnants from chylomicrons and very low-density lipoproteins [36], and these postprandial triglyceride-rich lipoproteins penetrate the arterial endothelium [35, 40].

The effect of postprandial serum TG level on the development of DM is unclear, although a few cohort studies have suggested that non-fasting serum TG level is a useful predictor of the development of DM [13, 27]. Lipotoxicity—excess lipid accumulation (including TG) in non-adipose tissues—is thought to play an important role in the onset of insulin resistance and β-cell damage [41–43]. As the net whole-body importation of fatty acids occurs in the postprandial state, tissue fatty acid overexposure has been linked to this physiological state [44]. Therefore, postprandial elevation of the serum TG level is a potential candidate for the cause of impaired insulin secretion and insulin impairment [45]. However, the precise mechanism of lipotoxicity is unclear. Further studies are required to determine the effect of postprandial serum TG levels on insulin impairment or the development of DM.

A few cohort studies have reported that higher serum TG level is a risk factor for the development of DM in Japanese [15, 27]. Kametani et al. followed up 7222 (3916 females and 3306 males) normoglycemic subjects who were divided into 2 groups based on the fasting serum TG level at baseline: normotriglyceridemic (TG <150 mg/dL, age; 52.3 ± 10.5 years) and hypertriglyceridemic (TG ≥150 mg/dL, age; 53.3 ± 10.3 years) [15]. During the follow-up period (average: 4.5 years), 114 (1.57 %) subjects developed DM. However, the subjects in that study were slightly older and had a shorter follow-up period on average than ours. Recently, Fujihara et al. followed up 127,176 non-diabetic subjects (87,980 females and 39,196 males), who were divided into fasting and non-fasting groups based on the number of hours since the previous meal [27]. Among these participants, 8867 (4855 females and 4012 males, 6.97 %) developed DM during a mean follow-up of 5.5 years. They found that the fasting and non-fasting TG levels in males and the non-fasting TG level in females were predictive of future DM among participants with normal BMI (18.5–24.9). However, the mean age of the participants was around 60 years, and serum glucose level was not included in their multivariate adjusted models.

The present study has some advantages. First, it demonstrates that serum TG level is a predictor of DM development during the early years in Asians; the mean age of the participants was 45.4 ± 6.5 years and the time taken to develop DM was 4.5 ± 2.1 years from the baseline examination. Before the present work, in spite of the significance of ethnicity [10, 46–48], the importance of serum TG level in the development of DM in Asians was unclear because only a few cohort studies had been conducted in Asians, and the mean age of the population in each of those studies was relatively high (in their late 40s [18], 50s [15, 27], or 60s [10]). Second, in the present study, the outcome was obtained by evaluating those who underwent consecutive annual medical check-ups including laboratory tests for 8 years. Third, even in subjects with a low serum non-fasting glucose level (<140 mg/dL), a high serum TG level (≥150 mg/dL) independently predicted the development of DM in the present study. Some cohort studies of healthy populations in which blood glucose level was included in the multivariate analysis have shown an association of serum TG level with the development of DM [14, 15, 18, 49]. However, it is very important to evaluate the effect of serum TG level in low serum glucose subjects, because a high normal blood glucose level is known to be associated with the development of DM [14, 49, 50].

The present study also has some limitations. First, this study disregarded any information about the duration of fasting and the content of the last meal, so it was not designed to clarify the effects of these factors. However, this information is not always obtained correctly in a medical check-up; therefore, the present study proposed even the non-fasting serum TG level can be effectively used as a predictor for the development of DM. Second, our research was not conducted to compare the effects of fasting and non-fasting on the development of DM. This study therefore cannot answer the question of whether non-fasting serum TG level is a better predictor of DM than fasting serum TG level. Third, there is a possibility that subjects with impaired glucose tolerance (IGT) were included in the <140 mg/dL category. However, the number of subjects with a fasting serum glucose level ≥110 mg/dL was 32, and similar results were observed when these subjects were excluded. Finally, the participants in the present study were relatively young, but might not be young enough to confirm the effect of serum TG level on the premature development of type 2 DM. A large cohort with a narrow age distribution in young subjects is ideal; however, gathering young subjects is usually a more difficult task than gathering seniors. Utilizing casual blood sampling may facilitate further studies.

Conclusion

The present study demonstrated that a serum TG level of ≥150 mg/dL when fasting or non-fasting is a significant predictor of type 2 diabetes in middle-aged Japanese workers. Monitoring the casual serum TG level, which has frequently been overlooked so far, should contribute to the early detection of individuals at high risk for DM and should indicate suitable diets for young and healthy subjects.

Sources of funding

This work was supported by a Grant-in-Aid for Scientific Research (C25460799 and C24500814) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.

Conflict of interest

None.

Human rights statement and informed consent

This article does not contain any studies with human or animal subjects performed by any of the authors.