Association between high-density lipoprotein cholesterol and the risk of incident diabetes in the prediabetic and the normoglycemic Japanese men: A population-base longitudinal cohort study

doi:10.21203/rs.3.rs-4800115/v1

Background

While many studies indicate a negative correlation between high-density lipoprotein cholesterol (HDL-C) and the occurrence of diabetes, there are still some inconsistent findings. The contentious relationship between the two may be partially due to the undistingushement between the pre-diabetic and the normoglycemic participants in the previous studies, which may confound the association. This study aimed to investigate the relationship between the baseline HDL-C and incident type 2 diabetes mellitus (T2DM) in a Japanese cohort with normoglycemia or with prediabetes, respectively.

Method

In total, 10120 men (6791 with normoglycemia and 3329 with prediabetes) were enrolled from the NAGALA cohort from Jan 5th, 2004 to Dec 26th, 2015. Cox proportional hazards models were conducted to explore the association between baseline HDL-C levels and incident T2DM. A two-piecewise linear regression model was performed to evaluate the threshold effect of the baseline HDL-C concentration on T2DM incidence by using a smoothing function.

Results

During the median 5.95-year follow-up duration for participants with normoglycemia and 4.33-year follow-up period for prediabetes, 88 participantes with normoglycemia and 494 participantes with prediabetes developed T2DM. In the crude model and partly adjusted model, the risk of T2DM decreased significantly in both normoglycemia and prediabetes with increment in baseline HDL-C concentration. Howerver, the associations became nonsignificant after fully adjusting for possible confounders. Interestingly, in prediabetes, an L-shaped relationship between baseline HDL-C and risk of incident T2DM with a threshold HDL-C concentration of 32.4mg/dl was determined: the T2DM risk sharply decreased by 62% with the each 10mg/dl increment in HDL-C levels (HR = 0.377, 95%CI = 0.191–0.743) and the decline reaches a near plateau when the HDL-C concentration is higher than 32.4 mg/dl (HR = 0.986, 95%CI = 0.895–1.085).

Conclusions

Among a Japanese male population, an L-shape relationship between baseline HDL-C concentration and the risk of incident T2DM was explored in prediabetes, while no significant association was detected in men with normoglycemia.

Health sciences/Diseases

Health sciences/Endocrinology

Prediabetes

Euglycemia

HDL

Incident diabetes

Over the past four decades, the prevalence of type 2 diabete mellitus (T2DM) increased unprecedentedly and has become a major global health threat^{1, 2}. Most individuals will pass through a phase of prediabetes before progressing to full-blown T2DM³. Prediabetes, which is at high risk of future T2DM, is genenerally refered to a condition with blood glucose concntrations higher than normal, but lower than diabetes thresholds^{4, 5}. According to an ADA expert panel, around 70% individuals with prediabetes will eventually develop diabetes⁴. Notably, prediabetes is a reversible stage^{3, 6, 7}. If proper interventions are taken during this critical state, the transition from prediabetes to T2DM can be reduced^{3, 7}. The progression from prediabetes to T2DM is usually mild with further deterioration, whereas reversing back to normoglycaemia needs improvement in risk factors⁸. Thus, knowledge about the reliable risk factors of developing T2DM in people with prediabetes is important for screening for effective diabetes preventive intervention.

Many proposed risk factors for diabetes take account of metabolic syndrome components, including low levels of high-density lipoprotein cholesterol (HDL-C)⁹. Indeed, lower HDL-C concentrations confer higher risk to future T2DM in various ethnic populations^10–14. For exmple, an inverse association between serum HDL-C concentration and risk of T2DM was reported among middle-aged and elderly Chinese¹⁵ as well as inhabitants living in the city of Groningen in the Netherlands¹³. However, there are still some inconsistent findings. For example, a 4-year retrospective, longitudinal study performed in Koreans, suggested that HDL-C was not associated with incident T2DM in fully adusted model¹⁶. In the previous studies, the inclusion criteria regarding the presence of prediabetes in the participants were not clearly defined^{10, 11, 13, 17, 18}, thus, those studies may have inevitably consisted of people with prediabetes⁴, and the undistingushement between the pre-diabetic and the normoglycemic participants may have confounded the association, at least partially contributing to the contentious relationship between HDL-C and incident diabetes. Therefore, those results may not be well applied to pure prediabetic or normoglycemic populations, respectively. The association between HDL-C and incident diabetes in the subgroups with prediadiabetes or normoglycemia remain unkown. This study, thus, aimed to investigate the relationship between the baseline HDL-C and incident T2DM in a Japanese cohort with prediadiabetes or normoglycemia.

Design and participants

Data were extracted from the NAGALA cohort (NAfld in the Gifu Area, Longitudinal Analysis)¹⁹. Briefly, the NAGALA program was a population-based longitudinal cohort study performed at Murakami Memorial Hospital (Gifu, Japan) from May 1st, 1994 to Dec 31st, 2016. The NAGALA program aimed to detect chronic diseases and their risk factors, contributing to public health promotion ¹⁹. Approximately, 60% individuals in the program received one to two medical exams every year¹⁹. Male individuals participating in the program from Jan 5th, 2004 to Dec 26th, 2015 were extracted, and participants with at least one follow-up visit between 27 September 2004 and 27 December 2016 were included in the present study. Individuals with diabetes, medication usage, and missing data of HDL-C, HbA1C, as well as body weight at baseline were excluded. Finally, 10120 men (6791 nonprediabetes and 3329 prediabetes) were included (Fig. 1). This study was approved by the ethics committee of Murakami Memorial Hospital.

Figure 1. Flowchart of the participant enrollment process

Data collection and measurement

Our previous study has described data collection and measurement in detail¹⁹. Briefly, a standardized self-administered questionnaire was used to obtain the medical history, alcohol habits, smoking status, recreational and physical activity, and family history of diabetes^{19, 20}. The mean ethanol consumption per week was assessed by the amount and the type of alcohol use weekly in the past month. Based on the weekly ethanol consumption, the participants were categorized into the following four groups: no or minimal alcohol consumption, < 40 g/week; light, 40–140 g/week; moderate, 140–280 g/week; or heavy, > 280 g/week. Smoking status were also classified into three groups: never, ex or current. Non-smokers were referred to participants who never smoked, ex-smokers referred to those who had ever smoked but quitted until baseline, and current-smokers were defined as individuals who smoked at baseline visit. Regular exercisers were identified as individuals who regularly participated in any type of sports over once a week²⁰. Positive family history of T2DM was referred to a person with a father and/or mother diagnosed with diabetes. Fatty liver was evaluated by abdominal ultrasonography¹⁹. The criteria for fatty liver diagnosis included hepatorenal echo contrast, liver brightness, deep attenuation, and vascular blurring¹⁹. Additionally, hypertriglyceridemia was defined as triglyceride ≥ 150mg/dl²¹.

Definition of Prediabetes

Prediabetes was referred to participants with impaired fasting glucose (FPG ≥ 5.6), with 5.7%≤ HbAlc < 6.5%, either or both ^{4, 22, 23}. Participants with HbA1c < 5.7% and fasting plasma glucose < 5.6mmol/l were considered to have neither diabetes nor prediabetes^{4, 22, 23} (referred to here as “normoglycemia” or “nonprediabetes”).

Exposure

The exposure in this study was fasting HDL-C concentration of participants at baseline.

Primary outcomes

Incident T2DM was defined as FPG ≥ 7mmol/L or HbA1c ≥ 6.5% according to the diagnostic criteria of ADA or self-reported ²⁴.

Statistical analyses

Baseline participants’ characteristics were presented by categories of prediabetes and nonprediabetes (nomorglycemia). Continuous variables are presented as mean (S.D.) or as median༈Q1-Q3༉, while categorical data are described as percentage. Data normality was tested by Kolmogorov-Smirnov. For normally distributed data, statistical differences between the two groups were assessed by Student’s t test, while for non-normal distributed variables, the Manne-Whitney tests were used (Table 1). Chi-square tests were applied to eveluate statistical differences between categorical variables (Table 1). The potential effect of age, BMI, waist circumference, body weight, ALT, AST, GGT, HDL-cholesterol, triglyceride, HbA1c, systolic blood pressure, diastolic blood pressure, and FPG were screened by univariable logistic regression analysis (Table 2). To examine the association between baseline HDL-C concentration and incident T2DM, Cox proportional hazards models were performed with or without adjustment for covariates (table 3). A two-piecewise linear regression model was applied to explore the threshold effect of the log HDL-C on new-onset diabetes by using a smoothing function (Fig. 2, Table 4). A log likelihood ratio test was performed to compare the online linear regression model with a two-piecewise linear model. When P value < 0.05 (two-tailed), results were considered statistically significant. All statistical analyses were conducted by the statistical packages R (The R Foundation; http://www.r-project. org; version 3.6.1) and EmpowerStats²⁵.

Baseline characteristics

Data from 10120 men (3329 prediabetes and 6791 nonprediabetes) at baseline were analyzed. Values for age, BMI, waist circumference, body weight, ALT, AST levels, GGT, HbA1c, FPG, systolic blood pressure, diastolic blood pressure, as well as the proportions of fatty liver，ex-smoker, moderate and heavy alcohol consumption, hypertriglyceridemia were significantly higher in men with prediabetes (all P values <0.05, table 1). In contrast, HDL-C concentration and percentages of regular exerciser, nonsmoker, current-smoker, family history of T2DM, and minimal to light alcohol consumer were significantly lower in prediabetes compared to those in nonprediabetes (nomorglycemia) (all P values <0.05, table1). During the median 5.95-year follow-up duration for participants with nomorglycemia and 4.33-year follow-up period for individuals with prediabetes, 582 men (88 nonprediabetes and 494 prediabetes) developed T2DM.

Table 1. Baseline Characteristics of the participants with nomorglycemia or with prediabetes
	With normoglycemia	With prediabetes	P-value
N	6791	3329
Age(years)	43.56 (8.82)	46.88 (9.16)	<0.001
BMI (kg/m2)	22.74 (2.88)	24.06 (3.08)	<0.001
Waist circumferences (cm)	79.72 (7.73)	83.33 (7.93)	<0.001
Body weight (Kg）	66.51 (9.63)	69.92 (10.39)	<0.001
ALT (u/l)	23.05 (13.37)	28.15 (17.97)	<0.001
AST (u/l)	18.00 (15.00-22.00)	20.00 (16.00-25.00)	<0.001
GGT (u/l)	19.00 (14.00-28.00)	24.00 (17.00-37.00)	<0.001
HDL-C (mg/dl)	51.21 (13.57)	49.68 (13.75)	<0.001
HbA1C (%)	5.08 (0.28)	5.41 (0.36)	<0.001
FPG (mmol/l)	5.15 (0.28)	5.90 (0.34)	<0.001
DBP (mmHg)	74.08 (9.79)	78.28 (10.12)	<0.001
SBP (mmHg)	117.62 (13.86)	123.61 (14.68)	<0.001
Fatty liver (%)	21.84	42.23	<0.001
Regular exerciser (%)	19.69	16.88	0.003
Smoking states
nonsmoker (%)	33.21	30.37	0.04
ex-smoker (%)	28.38	35.6	<0.001
current-smoker (%)	38.07	33.73	<0.001
family history of T2DM (%)	2.90	2.31	<0.001
Alcohol consumption
no or minimal (%)	52.91	46.98	<0.001
light (%)	23.78	7.2328	0.577
moderate (%)	12.75	14.90	0.003
heavy (%)	9.75	13.82	<0.001
Hypertriglyceridemia (%)	13.46	24.12	<0.001
Continuous variables are presented as mean ± S.D. or as median (Q1-Q3). Categorical data are presented as frequencies (percentages). BMI: Body mass index; FPG: Fasting plasma glucose; SBP: Systolic blood pressure; DBP: Diastolic blood pressure.

Univariate analysis of the association between baseline characteristics and incident T2DM

On univariate regression analysis, age, BMI, waist circumference, body weight, AST, ALT, GGT, triglyceride, HbA1c, systolic blood pressure, diastolic blood pressure, and FPG were positively associated with risk of incident T2DM in men either with prediabetes or normoglycemia. Conversely, HDL-cholesterol presented a protective effect in both populations (all P<0.05, Table 2).

Table 2. Univariate associations between baseline variates and incident type 2 diabetes.
	With normoglycemia		With prediabetes
	HR (95%CI)	P value	HR (95%CI)	P value
Age(years)	1.05 (1.02, 1.07)	0.0001	1.03 (1.02, 1.04)	<0.0001
BMI (kg/m2)	1.20 (1.14, 1.26)	<0.0001	1.12 (1.09, 1.15)	<0.0001
Waist circumferences (cm)	1.07 (1.05, 1.10)	<0.0001	1.05 (1.04, 1.06)	<0.0001
Body weight (Kg）	1.04 (1.03, 1.06)	<0.0001	1.03 (1.02, 1.04)	<0.0001
ALT (u/l)	1.02 (1.02, 1.03)	<0.0001	1.02 (1.01, 1.03)	<0.0001
AST (u/l)	1.01 (1.01, 1.02)	<0.0001	1.02 (1.01, 1.03)	<0.0001
GGT (u/l)	1.01 (1.00, 1.01)	<0.0001	1.00 (1.00, 1.01)	<0.0001
HDL mg/dl	0.96 (0.94, 0.98)	<0.0001	0.98 (0.97, 0.99)	<0.0001
TG mg/dl	1.01 (1.00, 1.01)	<0.0001	1.00 (1.00, 1.00)	<0.0001
HbA1C (%)	30.78 (13.52, 70.09)	<0.0001	14.74 (11.32, 19.20)	<0.0001
SBP mmHg	1.03 (1.01, 1.04)	0.0001	1.01 (1.00, 1.01)	0.0058
DBP mmHg	1.04 (1.02, 1.06)	<0.0001	1.01 (1.01, 1.02)	0.0011
FPG (mmol/l)	7.06 (2.89, 17.26)	<0.0001	10.12 (8.06, 12.70)	<0.0001
CI: confidence interval; BMI: Body mass index; FPG: Fasting plasma glucose; SBP: Systolic blood pressure; DBP: Diastolic blood pressure.

Multivariate analysis of the association between baseline HDL-C concentration and incident T2DM

The independent effects of the baseline HDL-C levels on newly-onset T2DM was evaluated by multivariate Cox proportional hazard model (Table 3). In the crude model, the risk of T2DM decreased by 33% in nonprediabetes and 19% in prediabetes with per 10 mg/dl increment in baseline HDL-C concentration (nonprediabetes: hazard ratio [HR]=0.67, 95% confidence interval [CI]=0.54-0.82, P < 0.001; prediabetes: HR=0.81, 95% CI=0.74-0.87, P < 0.001, table 3). The association remained significant in nonprediabetes (HR=0.77, 95% CI=0.63-0.95, P=0.016, table 3) and prediabetes (HR=0.87, 95% CI=0.80-0.95, P=0.001, table 3) after adjusted for age, BMI, waist circumference and body weight in adjusted model I. However, further adjustment for ALT, AST, GGT, HbA1c, systolic blood pressure, diastolic blood pressure, FPG, fatty liver, regular exercise, smoking status, triglyceride, alcohol consumption, and family histoy of T2DM, the association between HDL-C and incident diabetes became non-significant in both prediabetes and nonprediabetes (Table 3).

The individuals were then divided into tertiles according to the baseline HDL-C concentration. The fully adjusted model showed that, with the lowest tertile as a reference, the HRs were 0.60 (95%CI=0.349-1.035) for Tertile 2 and 0.72 (95%CI=0.375-1.369) for Tertile 3 in nonprediabetes (nomorglycemia). Similarly, in prediabetes, the HRs were 0.89 (95%CI=0.72-1.11) for the second tertile and 0.94 (95%CI=0.72-1.23) for the third tertile with the first tertile as a reference. The risk of incident diabetes in individuals with different tertile of HDL-C levels were not changed in a dose-dependent manner, suggesting the existence of a nonlinear relationship between HDL-C levels and incident T2DM.

Table 3. Association between baseline HDL-C and incident diabetes
	incident diabetes
	Crude model		Adjusted model I		Adjusted model II
	HR (95%CI)	P value	HR (95%CI)	P value	HR (95%CI)	P value
With nomorglycemia
HDL-C per 10mg/dl increment (continuous)	0.67 (0.54, 0.82)	<0.0001	0.77 (0.63, 0.95)	0.016	0.84 (0.67, 1.06)	0.1347
HDL-C tertiles
T1(≤44.0mg/dl)	1		1		1
T2(44.1-54.9mg/dl)	0.42 (0.249, 0.697）	0.00087	0.51 (0.301, 0.855)	0.011	0.60 (0.349, 1.035)	0.06632
T3(≥55mg/dl)	0.38 (0.22, 0.67)	0.0008	0.58 (0.32, 1.05)	0.072	0.72(0.375, 1.369)	0.3135
With prediabetes
HDL-C per 10mg/dl increment (continuous)	0.81 (0.74, 0.87)	<0.0001	0.87 (0.80, 0.95)	0.001	0.95 (0.86, 1.04)	0.2582
HDL-C tertiles
T1(≤42.4mg/dl)	1		1		1
T2(42.5-53.0mg/dl)	0.69 (0.56, 0.85)	0.0004	0.76 (0.62, 0.93)	0.009	0.89 (0.72, 1.11)	0.2947
T3(≥53.1mg/dl)	0.58 (0.46, 0.72)	<0.0001	0.72 (0.57, 0.91)	0.005	0.94 (0.72, 1.23)	0.6373
Crude model adjusted for none.
Adjusted model I adjusted for age, BMI (body mass index, kg/m2), waist circumference (cm); body weight (kg).
Adjusted model II adjusted for model I plus ALT, AST, GGT, HbA1c, systolic blood pressure, diastolic blood pressure, and fasting plasma glucose, as continuous variables; fatty liver (positive, negative or not recorded), regular exercise (yes, no, or not recorded), smoking status (never, past, current or not recorded), triglyceride (<150mg/dl or ≥150mg/dl), alcohol consumption (no or minimal, light, moderate, heavy or not recorded), and family histoy of T2DM (positive, negative, or not recorded). CI denotes confidence interval.

Two-piecewise linear regression model analysis using a smoothing function

As the above multivariate Cox analysis suggesting nonlinear associations between HDL-C and incident diabetes, a smoothing function was used to further explore their relationship. Interestingly, in prediabetes, an L-shaped relationship between baseline HDL-C and risk of incident T2DM with a threshold HDL-C concentration of 32.4mg/dl was illustrated: the T2DM risk sharply decreased by 62% with the each 10mg/dl increment in HDL-C levels (HR=0.377, 95%CI=0.191-0.743) and the decline reached a near plateau when the HDL-C concentration is higher than 32.4 mg/dl (Figure 2 and table 4). However, in participants with normoglycemia (nonprediabetes), there was no significant threshold effect of HDL-C levels on risk of incident T2DM was found (Figure 2 and table 4).

Table 4. Threshold effect analysis of HDL-C on incident diabetes using two-piecewise linear regression.
	Adjusted model
	HR	95%CI	P value
With nomorglycemia (nonprediabetes) (Per 10mg/dl HDL-C increment)
HDL-C<47.6mg/dl	0.78	(0.51, 1.18)	0.2418
HDL-C>47.6mg/dl	0.89	(0.62, 1.28)	0.5322
P for log likelihood ratio test	0.678

With prediabetes (Per 10mg/dl HDL-C increment)
HDL-C<32.4mg/dl	0.375	(0.191, 0.738)	0.0045
HDL-C>32.4mg/dl	0.985	(0.895, 1.084)	0.7569
P for log likelihood ratio test	0.013
Adjusted for age, BMI (body mass index, kg/m2), waist circumference, body weight, ALT, AST, GGT, HbA1c, systolic blood pressure, diastolic blood pressure, and fasting plasma glucose, as continuous variables; fatty liver (positive, negative or not recorded); regular exercise (Yes, No, or not recorded); smoking status (never, past, current or not recorded); triglyceride (<150mg/dl or ≥150mg/dl); alcohol consumption (no or minimal, light, moderate, heavy, or not recorded), and family histoy of T2DM (positive, negative, or not recorded). CI denotes confidence interval.

Figure 2. The relationship between baseline HDL-C and incident diabetes in participants with nondiabetes (nomorglyciemia) (A) and with prediabetes (B). Adjusted for age, BMI (body mass index, kg/m2), waist circumference, body weight, ALT, AST, GGT, HbA1c, systolic blood pressure, diastolic blood pressure, and fasting plasma glucose, as continuous variables; fatty liver (positive, negative or not recorded); regular exercise (Yes, No, or not recorded); smoking status (never, past, current or not recorded); triglyceride (<150mg/dl or ≥150mg/dl); alcohol consumption (no or minimal, light, moderate, heavy, or not recorded), and family histoy of T2DM (positive, negative, or not recorded).

To the best of our knowledge, this is the first population study to demonstrate an L-shape relationship between HDL-C concentration and incident T2DM among male participants with prediabetes. A turning point of HDL-C at 32.4 mg/dl using the threshold effect analysis after adjustment for potential confounders was revealed (Fig. 2, Table 4).

Most previous studies suggested that the HDL-C was negatively linked with incident diabetes^{10, 11, 13, 17}. For example, the PREVEND study conducted in the city of Groningen in the Netherlands suggested that serum HDL-C concentration was inversely and independently associated with risk of T2DM¹³. Similarly, a higher risk of new-onset diabetes was revealed among middle-aged and elderly Chinese with low HDL-C levels^{15, 17}. Besides, higher HDL-C decreased the risk of incident T2DM in a case-cohort study²⁶. However, the negative correlation between HDL-C and diabetes is still a matter of debate. For example, a retrospective, longitudinal study performed in Koreans, indicated that HDL-C was not linked with incident T2DM in the fully adjusted model¹⁶;and Changchun Cao et.al. found that HDL-C was only negatively correlated with the risk of diabetes when it was lower than 1.72 mmol/L ²⁷, while Christiane L et.al. showed that genetically reduced HDL-C does not relate to increased risk of type 2 diabetes in a Mendelian Randomization Study²⁸. In the previous studies, as the inclusion criteria regarding the presence of prediabetes in the participants were not clearly defined^{10, 11, 13, 17, 18}, and not only people with normoglycemia but also with baseline prediabetes may have been inevitably included, thus, the associations between HDL-C and incident diabetes could have been confounded by the undistingushement between the baseline prediabete and the normoglycemia in participants.

In this study with clearly defined prediabetes and normoglycemia according to ADA criteria^{4, 22, 29, 30}, we found that, in the participants with prediabetes, a baseline HDL-C level below 32.4 mg/dl is negatively correlated with the risk of incident diabetes according to the two-piecewise linear regression model (Table 4 and Figure.2), and with each 10mg/dl increment in HDL-C, the risk of T2DM decreased by 62% (HR = 0.377, 95%CI = 0.191–0.743, Table 4), well in line with the previous studies conducted in non-diabetic individuals^{10, 11, 13, 17}; however, when the HDL-C concentration was above 32.4 mg/dl, the decline of the risk reached a near plateau (HR = 0.986, 95%CI = 0.895–1.085, Table 4 and Figure.2). Consisent with our results, an “L” shape relationship between HDL-C and risk of incident diabetes may also exist in a previous cohort study enrolling participants from Isfahan and adjoining areas¹⁸, in which, Janghorbani M et al found that, compared to the lowest HDL-C quartile, the risk of incident T2DM was 17% lower in the 2nd quartile and 15% lower in 3rd HDL-C quartile, but not lower in the 4th quartile in the adjusted models¹⁸, suggesting that the protective effect of elevating HDL-C on incident diabetes reachs a plateau within the 4th HDL-C quartile and indicative of an “L” shape association. However, smooth curve fitting was not explored in the aforementioned study. Interstingly, in participants with normoglycemia (nonprediabetes), concentration of HDL-C was inversely associated with incident T2DM with or without adjustments for age, BMI, waist circumference, and body weight. However, in the fully-adjusted model, the inverse association between baseline HDL-C and incident diabetes became non-significant (Table 3). These results were in paralle with a 4-year retrospective, longitudinal study performed in nondiabetic Koreans, suggesting that HDL-C was not associated with incident T2DM other than crude model¹⁶, as well as the Mendelian Randomization Study by Christiane L et.al. showing that genetically reduced HDL-C does not relate to increased risk of type 2 diabetes ²⁸, whereas discordant with the significant protective effects of HDL-C on T2DM in the previous studies^{10, 11, 13, 17}. With an L-shape association between HDL-C and incident T2DM in particiapnts with prediabetes, we speculated that, prediabetes, a high-risk state for developing diabetes⁴, not clearly defined in the previous reports, may have partially confound the results and lead to inconsistent interpretation on the relationships between HDL-C and incident diabetes in the previous reports. Indeed, a two-piecewise linear regression model was used to explore the correlation between baseline HDL-C and incident type 2 diabetes in men without prediabetes, and no L-shape association was found (Fig. 2).

This study has significant clinical implications. Firstly, low levels of HDL-C may increase the risk of incident T2DM, and the underlying mechanism may be mainly related to the regulation of β-cell insulin secretion and function by HDL-C³¹. HDL-C attenuates inflammatory actions, stimulates proliferation and migration in endothelial cells, governing vascular health³². As vascular consequence, HDL-C promotes β-cell insulin secretion, reduces β-cell apoptosis³². Low HDL-C contributes to sustain the abnormalities in insulin secretion³³, whereas Interventions that elevate plasma HDL-C enhancing pancreatic β-cell function³⁴. Additionally, HDLs have also been found to stimulate glucose uptake into skeletal muscle, adipose tissue and liver^{32, 35}. These experimental findings together with the negative association between HDL-C levels and the risk of diabetes development have generated the notion that appropriate HDL-C concentration must be maintained in humans to diminish the risks of incident diabetes, especially in people with baseline prediabetes. Secondly, the plateau of the L-shape association in the current studies suggested that therapeutic raising HDL-C beyond the turning point may not reduce risk of diabetes in prediabetes as expected. Morover, the Copenhagen Heart Studies suggested that not only low HDL-C levels but also elevated HDL-C (≥ 135 mg/dL and men > 97 mg/dL) consistently increased HR for all-cause and cause-specific mortality³⁶. Similarly, Madsen CM also suggested that extreme high levels of HDL-C were related paradoxically to high mortality in two prospective cohort studies³⁷. Taking the L-shape association in our study and the previous detreimental findings of inappropriately high levels of HDL-C into consideration, HDL-C should be maintained at an optimal level, either very high or very low HDL-C concentration may be detrimental in prediabetes. Additionally, correlations between HDL-C concentrations and β-cell function differed in participants within various degree of glucose tolerance³⁸, indicating that the effects of HDL-C on incident diabetes might be interpretated seperately according to glucose tolerant state, which at least partially explains our findings that an L-shape relationship between HDL-C and incident T2DM was observed in particiapnts with prediabetes, but not with normoglycemia.

There were several strengths in this study. Firstly, the participants with prediabetes or with normoglycemia (nonprediabetes) were clealy defined and analyzed separately to avoid confounding the results by the significant difference in baseline blood glucose levels (prediabetes). Secondly, the dose-response associations were further explored by two-piecewise linear regression model analysis using a smoothing function. However, several limitations should be also taken into account. Firstly, the present study is conducted only in the male Japanese population, and the relationship between baseline HDL-C and incident diabetes was uncertain in other ethnicities or in females. Secondly, oral glucose tolerance test was not performed and the baseline prediabetes and occurrences of incident type 2 diabetes (outcome) might have been partially underestimated in this study.

An L-shape relationship between baseline HDL-C concentration and the risk of incident T2DM was explored in prediabetes, while no significant association was detected in nonprediabetes (normoglycemia) among a Japanese male population.

Ethics approval and consent to participate

This study was approved by the ethics Committee of Murakami Memorial Hospital. We confirm that all research was performed in accordance with relevant regulations, and informed consent was obtained from all participants or their legal guardians.

Consent for publication

Not applicable.

Availability of data and materials

Data are available upon reasonable request from Masahide Hamaguchi.

Masahide Hamaguchi, MD, PhD

Department of Endocrinology and Metabolism, Kyoto prefectural University of Medicine, Graduate School of Medical Science

Address：465 Kajii-cho, Kawaramachi-Hirokoji, Kamigyo-ku, Kyoto 602-8566， Japan

Fax: +81-75-252-3721, Tel：+81-75-251-5505

E-mail: hama@koto. kpu-m.ac.jp.

0000-0002-8651-4445

Funding

This study was supported in part by the National Natural Science Foundation of China (grant number: 82001651), Guangxi Science and Technology Base and Talent Project (GuikeAA21220002), the 111 Project (D17011) and Advanced Innovation Teams and Xinghu Scholars Program of Guangxi Medical University.

Author contribution

Xiuping Xuan contributed to design the study and wrote the manuscript. Lijuan Kong helped to design the study and wrote the manuscript, also contributed to the data analyses. Qian Hu, Lan Zhou, Hai Zhu, Jixiang Liao, Jie Zhang, Song Huang and Songqing He contributed to analyze the data and revised the manuscript. Takuro O, Yoshitaka H, Akihiro O, Takao K and Michiaki F helped to collect the data and revised the article. Masahide H had full access to all the data in this study, provided the details, helped to revised the article. Xuemei Xie originated and designed this study, took responsibility of the accuracy of the data analysis, revised the article critically, and was the guarantor of this work. All authors were involved in the writing of the manuscript and approved the manuscript's final version.

Acknowledgements

We would like to thank all of the participants and the staff of the medical health checkup center at Murakami Memorial Hospital.

Chatterjee S, Khunti K, Davies MJ. Type 2 diabetes. Lancet (London, England). Jun 3 2017;389(10085):2239–2251. doi:10.1016/s0140-6736(17)30058-2
Zheng Y, Ley SH, Hu FB. Global aetiology and epidemiology of type 2 diabetes mellitus and its complications. Nature reviews Endocrinology. Feb 2018;14(2):88–98. doi:10.1038/nrendo.2017.151
Khetan AK, Rajagopalan S. Prediabetes. The Canadian journal of cardiology. May 2018;34(5):615–623. doi:10.1016/j.cjca.2017.12.030
Tabak AG, Herder C, Rathmann W, Brunner EJ, Kivimaki M. Prediabetes: a high-risk state for diabetes development. Lancet. Jun 16 2012;379(9833):2279–90. doi:10.1016/S0140-6736(12)60283-9
Beulens J, Rutters F, Rydén L, et al. Risk and management of pre-diabetes. European journal of preventive cardiology. Dec 2019;26(2_suppl):47–54. doi:10.1177/2047487319880041
Khan RMM, Chua ZJY, Tan JC, Yang Y, Liao Z, Zhao Y. From Pre-Diabetes to Diabetes: Diagnosis, Treatments and Translational Research. Medicina (Kaunas, Lithuania). Aug 29 2019;55(9)doi:10.3390/medicina55090546
Clark B, Boghani S, Grullon C, Batista M. The Impact of a Worksite-Based Diabetes Prevention Intervention: A Pilot Study. Population health management. Jun 2017;20(3):233–238. doi:10.1089/pop.2016.0055
Goldberg RB, Temprosa M, Haffner S, et al. Effect of progression from impaired glucose tolerance to diabetes on cardiovascular risk factors and its amelioration by lifestyle and metformin intervention: the Diabetes Prevention Program randomized trial by the Diabetes Prevention Program Research Group. Diabetes care. Apr 2009;32(4):726–32. doi:10.2337/dc08-0494
Abbasi A, Corpeleijn E, Gansevoort RT, et al. Role of HDL cholesterol and estimates of HDL particle composition in future development of type 2 diabetes in the general population: the PREVEND study. The Journal of clinical endocrinology and metabolism. Aug 2013;98(8):E1352-9. doi:10.1210/jc.2013-1680
Fiorentino TV, Succurro E, Marini MA, et al. HDL cholesterol is an independent predictor of β-cell function decline and incident type 2 diabetes: A longitudinal study. Diabetes/metabolism research and reviews. May 2020;36(4):e3289. doi:10.1002/dmrr.3289
Lee SH, Kim HS, Park YM, et al. HDL-Cholesterol, Its Variability, and the Risk of Diabetes: A Nationwide Population-Based Study. The Journal of clinical endocrinology and metabolism. Nov 1 2019;104(11):5633–5641. doi:10.1210/jc.2019-01080
Yahya R, Jainandunsing S, Rashid M, et al. HDL associates with insulin resistance and beta-cell dysfunction in South Asian families at risk of type 2 diabetes. Journal of diabetes and its complications. Oct 2021;35(10):107993. doi:10.1016/j.jdiacomp.2021.107993
Kunutsor SK, Kieneker LM, Bakker SJL, James RW, Dullaart RPF. Incident type 2 diabetes is associated with HDL, but not with its anti-oxidant constituent - paraoxonase-1: The prospective cohort PREVEND study. Metabolism: clinical and experimental. Aug 2017;73:43–51. doi:10.1016/j.metabol.2017.05.004
Hwang YC, Hayashi T, Fujimoto WY, et al. Differential Association Between HDL Subclasses and the Development of Type 2 Diabetes in a Prospective Study of Japanese Americans. Diabetes care. Nov 2015;38(11):2100–5. doi:10.2337/dc15-0625
Peng J, Zhao F, Yang X, et al. Association between dyslipidemia and risk of type 2 diabetes mellitus in middle-aged and older Chinese adults: a secondary analysis of a nationwide cohort. BMJ open. May 25 2021;11(5):e042821. doi:10.1136/bmjopen-2020-042821
Seo MH, Bae JC, Park SE, et al. Association of lipid and lipoprotein profiles with future development of type 2 diabetes in nondiabetic Korean subjects: a 4-year retrospective, longitudinal study. The Journal of clinical endocrinology and metabolism. Dec 2011;96(12):E2050-4. doi:10.1210/jc.2011-1857
Cao X, Tang Z, Zhang J, et al. Association between high-density lipoprotein cholesterol and type 2 diabetes mellitus among Chinese: the Beijing longitudinal study of aging. Lipids in health and disease. Jul 17 2021;20(1):71. doi:10.1186/s12944-021-01499-5
Janghorbani M, Amini M, Aminorroaya A. Low Levels of High-Density Lipoprotein Cholesterol Do Not Predict the Incidence of Type 2 Diabetes in an Iranian High-Risk Population: The Isfahan Diabetes Prevention Study. The review of diabetic studies: RDS. Summer-Fall 2016;13(2–3):187–196. doi:10.1900/rds.2016.13.187
Okamura T, Hashimoto Y, Hamaguchi M, Obora A, Kojima T, Fukui M. Ectopic fat obesity presents the greatest risk for incident type 2 diabetes: a population-based longitudinal study. International journal of obesity (2005). Jan 2019;43(1):139–148. doi:10.1038/s41366-018-0076-3
Okamura T, Hashimoto Y, Hamaguchi M, Obora A, Kojima T, Fukui M. Creatinine-to-bodyweight ratio is a predictor of incident non-alcoholic fatty liver disease: A population-based longitudinal study. Hepatology research: the official journal of the Japan Society of Hepatology. Jan 2020;50(1):57–66. doi:10.1111/hepr.13429
Kinoshita M, Yokote K, Arai H, et al. Japan Atherosclerosis Society (JAS) Guidelines for Prevention of Atherosclerotic Cardiovascular Diseases 2017. Journal of atherosclerosis and thrombosis. Sep 1 2018;25(9):846–984. doi:10.5551/jat.GL2017
Rooney MR, Rawlings AM, Pankow JS, et al. Risk of Progression to Diabetes Among Older Adults With Prediabetes. JAMA Intern Med. Apr 1 2021;181(4):511–519. doi:10.1001/jamainternmed.2020.8774
Akerblom A, Wojdyla D, Steg PG, et al. Prevalence and relevance of abnormal glucose metabolism in acute coronary syndromes: insights from the PLATelet inhibition and patient Outcomes (PLATO) trial. Journal of thrombosis and thrombolysis. Nov 2019;48(4):563–569. doi:10.1007/s11239-019-01938-2
Standards of medical care in diabetes–2011. Diabetes care. Jan 2011;34 Suppl 1(Suppl 1):S11-61. doi:10.2337/dc11-S011
EmpowerStats. http://www.empowerstats.com, X&Y Solutions, Inc., Boston, MA.
Bragg F, Kartsonaki C, Guo Y, et al. Circulating Metabolites and the Development of Type 2 Diabetes in Chinese Adults. Diabetes care. Feb 1 2022;45(2):477–480. doi:10.2337/dc21-1415
Cao C, Hu H, Zheng X, Zhang X, Wang Y, He Y. Non-linear relationship between high-density lipoprotein cholesterol and incident diabetes mellitus: a secondary retrospective analysis based on a Japanese cohort study. BMC Endocr Disord. Jun 18 2022;22(1):163. doi:10.1186/s12902-022-01074-8
Haase CL, Tybjaerg-Hansen A, Nordestgaard BG, Frikke-Schmidt R. HDL Cholesterol and Risk of Type 2 Diabetes: A Mendelian Randomization Study. Diabetes. Sep 2015;64(9):3328–33. doi:10.2337/db14-1603
Akerblom A, Wojdyla D, Steg PG, et al. Prevalence and relevance of abnormal glucose metabolism in acute coronary syndromes: insights from the PLATelet inhibition and patient Outcomes (PLATO) trial. J Thromb Thrombolysis. Nov 2019;48(4):563–569. doi:10.1007/s11239-019-01938-2
ElSayed NA, Aleppo G, Aroda VR, et al. Introduction and Methodology: Standards of Care in Diabetes-2023. Diabetes Care. Jan 1 2023;46(Suppl 1):S1-S4. doi:10.2337/dc23-Sint
Xepapadaki E, Nikdima I, Sagiadinou EC, Zvintzou E, Kypreos KE. HDL and type 2 diabetes: the chicken or the egg? Diabetologia. Sep 2021;64(9):1917–1926. doi:10.1007/s00125-021-05509-0
Mineo C, Shaul PW. Novel biological functions of high-density lipoprotein cholesterol. Circulation research. Sep 28 2012;111(8):1079–90. doi:10.1161/circresaha.111.258673
Natali A, Baldi S, Bonnet F, et al. Plasma HDL-cholesterol and triglycerides, but not LDL-cholesterol, are associated with insulin secretion in non-diabetic subjects. Metabolism: clinical and experimental. Apr 2017;69:33–42. doi:10.1016/j.metabol.2017.01.001
Rye KA, Barter PJ, Cochran BJ. Apolipoprotein A-I interactions with insulin secretion and production. Current opinion in lipidology. Feb 2016;27(1):8–13. doi:10.1097/mol.0000000000000253
von Eckardstein A, Widmann C. High-density lipoprotein, beta cells, and diabetes Cardiovascular research. Aug 1 2014;103(3):384 – 94. doi:10.1093/cvr/cvu143
Rodriguez A. High HDL-Cholesterol Paradox: SCARB1-LAG3-HDL Axis. Current atherosclerosis reports. Jan 5 2021;23(1):5. doi:10.1007/s11883-020-00902-3
[32]. Madsen CM, Varbo A, Nordestgaard BG. Extreme high high-density lipoprotein cholesterol is paradoxically associated with high mortality in men and women: two prospective cohort studies. European heart journal. Aug 21 2017;38(32):2478–2486. doi:10.1093/eurheartj/ehx163
Bardini G, Dicembrini I, Rotella CM, Giannini S. Correlation between HDL cholesterol levels and beta-cell function in subjects with various degree of glucose tolerance. Acta diabetologica. Apr 2013;50(2):277–81. doi:10.1007/s00592-011-0339-0

No competing interests reported.

Association between high-density lipoprotein cholesterol and the risk of incident diabetes in the prediabetic and the normoglycemic Japanese men: A population-base longitudinal cohort study

Status:

Version 1

Abstract

Background

Method

Results

Conclusions

Figures

Introduction

Methods

Design and participants

Data collection and measurement

Definition of Prediabetes

Exposure

Primary outcomes

Statistical analyses

Results

Discussion

Conclusions

Declarations

References

Additional Declarations

Status:

Version 1