Altered HDL composition is associated with risk for complications in type 2 diabetes mellitus in South Asian descendants: a cross-sectional, case-control study on lipoprotein subclass profiling

doi:10.21203/rs.3.rs-2345386/v1

Download PDF

Research Article

Altered HDL composition is associated with risk for complications in type 2 diabetes mellitus in South Asian descendants: a cross-sectional, case-control study on lipoprotein subclass profiling

https://doi.org/10.21203/rs.3.rs-2345386/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 17 May, 2023

Read the published version in Diabetes, Obesity and Metabolism →

Version 1

posted

You are reading this latest preprint version

Background: Composition of high-density lipoproteins (HDL) is emerging as an important determinant in the development of microvascular complications in type 2 diabetes mellitus (T2DM). Dutch South Asian (DSA) individuals with T2DM display an increased risk of microvascular complications compared to Dutch white Caucasian (DwC) individuals with T2DM. In this study, we aimed to investigate whether changes in HDL composition associate with increased microvascular risk in this ethnic group and lead to new lipoprotein biomarkers.

Methods: Using ¹H nuclear magnetic resonance (NMR) spectroscopy and Bruker IVDr Lipoprotein Subclass Analysis (B.I.LISA^TM) software, plasma lipoprotein changes were determined in 51 healthy individuals (30 DwC, 21 DSA) and 92 individuals with T2DM (45 DwC, 47 DSA) in a cross-sectional, case-control study.

Results: Using multinomial logistic and linear regression analyses we identified specifically reduced ApoA2 and HDL-4 plasma concentrations in DSA individuals with T2DM, revealing unique HDL compositional changes, associated with higher waist circumference, waist-to-hip ratio, HbA1c, and glucose levels, that correlated with microvascular complications.

Conclusion: While in both ethnic groups HDL composition differed between controls and T2DM, the lipid changes in DSA with T2DM were of more clinical relevance with higher odds of having diabetes-related pan-microvascular complications such as retinopathy and neuropathy. These typical changes in HDL could be used as ethnicity-specific T2DM biomarkers.

High-density lipoprotein composition

NMR

Dutch South Asians

type 2 diabetes mellitus

microvascular complications

The worldwide rising prevalence of type 2 diabetes mellitus (T2DM) is one of the major challenges to public health in the 21st century [1]. Being overweight and obese are the most prominent risk factors for T2DM. However, South Asians (SA) have a significantly higher risk of developing T2DM compared to white Caucasians (wC) even with a body mass index (BMI) of around 23, which strongly suggests that different pathophysiological pathways in this particular ethnic group may lead to T2DM progression [2]. Additionally, compared to other ethnic groups, SAs tend to develop T2DM 5 to 10 years earlier [2], and more strikingly, SA patients with T2DM are more prone to develop microvascular complications such as diabetic retinopathy, with a faster progression and greater disease severity compared to West European patients with T2DM [3–5].

Previous studies showed that since the reduction of in particular low-density lipoproteins (LDL) by statins was associated with insulin resistance and increased risk of hyperglycaemic complications in T2DM, high-density lipoproteins (HDL) may be a better biomarker in diabetes progression [6–8]. High-density lipoproteins are composed of various lipids and proteins including apolipoproteins, enzymes, and lipid transfer factors, and exhibit large heterogeneity in size, density, and composition. The various HDL fractions represent different functionalities that mediate cholesterol efflux, anti-oxidation, anti-inflammation, and anti-thrombotic processes [9, 10]. Low plasma HDL cholesterol (HDL-C) levels, for instance, were found associated with an increased risk of T2DM and cardiovascular disease [11–13], while also the importance of HDL functionality to cardiovascular disease incidence and prognosis of heart failure was observed [14–16], one of the main complications among T2DM patients. Emerging evidence further indicated that HDL particle quality has a causative impact on diabetes [17–20].

Although promising, the majority of these studies were performed in Western populations, and very little is known about HDL functional changes in individuals with T2DM from other ethnic groups. Given that SA represent a unique high-risk population with different pathophysiological pathways to T2DM progression, dysfunction of HDL might be an important aspect overlooked in this population. Therefore, we hypothesized that specific changes in HDL composition in T2DM Dutch SA (DSA) related to their increased vulnerability to diabetes-related microvascular complications. In the present study, we aimed to investigate whether differences in plasma HDL composition in T2DM DSA in relation to Dutch wC (DwC) with T2DM and their respective non-diabetic controls are associated with diabetes-related microvascular complications to provide possible new biomarkers. For this, we used ¹H nuclear magnetic resonance (NMR) spectroscopy and the validated Bruker IVDr Lipoprotein Subclass Analysis (B.I.LISA™) software [21–26] to measure plasma HDL subclass and lipid concentrations for HDL composition determination in DSA individuals with T2DM and free of T2DM, compared with DwC without or with T2DM.

Study population. For the present cross-sectional study, baseline samples were used from the MAGNetic resonance Assessment of VICTOza efficacy in the Regression of cardiovascular dysfunction in type 2 diAbetes mellitus (MAGNA VICTORIA) study from two previous randomized controlled trials (RCT, ClinicalTrials.gov NCT01761318 [27] and NCT02660047 [28], respectively), together with age and gender matched healthy controls from both ethnic groups (16). The details of both trials can be found elsewhere [19–20]. In the first trial, 50 DwC with T2DM were recruited while 51 DSA with T2DM were recruited in the second trial. T1DM patients were excluded in both trials. One DwC-T2DM and four DSA-T2DM patients withdrew from the RCTs. Healthy controls were recruited by advertisement in Leiden University Medical Center (Leiden, The Netherlands) and local newspapers, as mentioned elsewhere [29]. Additionally, 51 healthy non-diabetic control participants (30 DwC-C and 21 DSA-C) were prospectively enrolled with a similar age and sex distribution to the corresponding T2DM patients. Ethnicity was based on the self-identified and self-reported biological parents' and ancestors' origins. Participants with complete informed consent were included. The study conformed to the revised Declaration of Helsinki and ethical approval was obtained from the Institutional Review Board (Leiden University Medical Center, Leiden, The Netherlands).

For the present study, we also excluded participants without plasma samples.

Sample preparation for ¹H nuclear magnetic resonance (NMR) spectroscopy. Sample preparation was performed consistently with the requirements of the Bruker B.I.LISA lipoprotein analysis protocol. The EDTA plasma samples were thawed at room temperature and immediately homogenized by inverting the tubes 10 times. Next, 200µL of plasma was manually transferred to a Ritter 96 deep-well plate. A Gilson 215 liquid handler robot was used to mix 120µL of plasma with 120µL, 75 mM disodium phosphate buffer in H₂O/D₂O (80/20) with a pH of 7.4 containing 6.15 mM NaN₃ and 4.64 mM sodium 3-[trimethylsilyl] d4-propionate (Cambridge Isotope Laboratories). Using a modified second Gilson 215 liquid handler, 190µL of each sample was transferred into 3 mm Bruker SampleJet NMR tubes. Subsequently, the tubes were sealed by POM ball insertion and transferred to the SampleJet autosampler where they were kept at 6°C while queued for acquisition.

NMR spectroscopy measurement and processing. All proton nuclear magnetic resonance (¹H-NMR) experiments were acquired on a 600 MHz Bruker Avance Neo spectrometer (Bruker Corporation, Billerica, USA) equipped with a 5-mm triple resonance inverse (TCI) cryogenic probe head with a Z-gradient system and automatic tuning and matching.

The NMR spectra were acquired following the Bruker B.I.Methods protocol. Before the measurements, a standard 3 mm sample of 99.8% methanol-d4 (Bruker) was used for temperature calibration [30]. A standard 3 mm QuantRefC sample (Bruker) was measured as the quantification reference and for quality control. All experiments were recorded at 310 K. The duration of the π/2 pulses was automatically calibrated for each sample using a homonuclear-gated nutation experiment on the locked and shimmed samples after automatic tuning and matching of the probe head [31]. For water suppression, presaturation of the water resonance with an effective field of γB1 = 25 Hz was applied during the relaxation delay and the mixing time of the NOESY1D experiment [32].

The NOESY1D experiment was recorded using the first increment of a NOESY pulse sequence with a relaxation delay of 4 s and a mixing time of 10 ms [33]. After applying 4 dummy scans, 32 scans of 98,304 points covering a sweep width of 17,857 Hz were recorded.

The lipoprotein values were extracted from the NOESY1D plasma spectra by submitting the data to the commercial Bruker IVDr Lipoprotein Subclass Analysis (B.I.LISA) platform [21–26]. This approach extracts information about lipoproteins and lipoprotein subfractions in plasma. In the current study, we focused on the HDL particles; dissected into concentration, composition, and four subclasses (sorted according to increasing density and decreasing size), and accompanying lipids within the lipoprotein subclasses including total cholesterol, free cholesterol, phospholipids, and triglycerides. The calculated esterified cholesterol was not included in the present study.

Isolation of primary human umbilical vein endothelial cells (HUVEC). Primary HUVECs were isolated from human umbilical cords obtained at the Leiden University Medical Center after written informed consent was obtained and the umbilical cord was collected and processed anonymously. The umbilical cord was rinsed with PBS to remove any remaining blood. To detach the endothelial cells from the umbilical cord, trypsin/EDTA (BE17-161E, Lonza, B-4800 Verviers, Belgium) was used. After 20 minutes of incubation at 37°C, the trypsin was inactivated with 20% FCS in PBS, and the entire solution in the umbilical cord was collected. The umbilical vein was flushed with PBS one more time to ensure that all detached cells were collected before centrifugation at 1200 rpm for 7 minutes. Cells were then cultured in 1% gelatin pre-coated flasks in endothelial growth medium (EGM2 medium, C-22011 supplemented with SupplementMix, Promocell, Heidelberg, Germany) with 1% antibiotics (penicillin/streptomycin, 15070063, Gibco, Paisley, UK). We cultured HUVECs from two different donors (one boy and one girl). We pooled them together and froze them in several vials for future experiments once they were confluent.

HDL function isolation. Leftover plasma samples from ¹H NMR measurements were used to isolate HDL based on the protocol from a previous study by Mulder et al. [34]. First, a 1:2 mixture of 36% polyethylene glycol (PEG 6000, 442714K, VWR International, Lutterworth, UK) in 10 mM HEPES (pH = 8, H3375, Sigma Aldrich, the Netherlands) and plasma samples was prepared. Following that, samples were incubated for 30 minutes on ice before being centrifuged for 30 minutes at 2200 g. Apo-B-containing lipoproteins were then precipitated, and the HDL-containing supernatant was collected, stored on ice, and used the following day for HDL functionality tests.

HDL anti-thrombotic capacity. Primary HUVECs were used to test HDL's anti-thrombotic properties. In the 96-well plate, HUVECs (passage 3) were seeded at a density of 4×10⁵ cells/well. The following day, HUVECs were pre-incubated for 30 minutes with 2% apoB-depleted plasma or an equal volume of precipitation reagent in HEPES as a control. Tumor necrosis factor α (TNF-α, H8916, Sigma Aldrich, the Netherlands) was then added at a concentration of 10ng/mL. After another 5 hours of incubation, the cell surface was washed once with HBSS, no calcium, and no magnesium (14170120, Gibco™, Paisley, UK). Each well received 50µL of normal pooled plasma before being placed in the Fluorometer for a 10-minute incubation at 37°C. The formation of thrombin was started by mixing 10µL of the fluorogenic substrate with calcium (TS50.00 FluCa-kit; Thrombinoscope BV, Maastricht, the Netherlands). The final reaction volume was 60µL. Thrombin formation was measured every 20 seconds for 60 minutes and calibrated using Thrombinoscope software (Additional file 1: Fig. S4a). To assess HDL anti-thrombotic capacity, endogenous thrombin potential (ETP) was used. To limit potential variation due to different plate conditions, HDL anti-thrombotic capacity measurements were performed at the same time using the same batch of pooled HUVECs and reagents. For each individual, measurements were taken in three technical replicates. To avoid batch effects, each plate contains samples from four groups.

Statistical analyses. Differences in baseline characteristics were tested between 4 groups, namely DwC healthy controls (DwC-C), DwC-T2DM, DSA healthy controls (DSA-C), and DSA-T2DM. Categorical variables were expressed as total number (percentage, %) and differences between groups were tested with the Chi-square test or Fisher's exact test. Normally distributed continuous variables were expressed as mean (standard deviation, SD), and differences were evaluated by post hoc tests of unpaired One-way ANOVA test; skewed continuous variables were presented as median (25–75 percentile) and differences were assessed using the Kruskal-Wallis test.

Based on ¹H NMR spectroscopy, a total of 112 lipoprotein main fractions and subclass features were extracted. To confirm the reliability of the NMR measurements, correlation analyses were performed between clinical routine measurements and corresponding parameters from NMR. We first extracted concentrations of all lipoprotein subclasses and performed a Pearson’s correlation between clinical routine measurements (total triglycerides, total cholesterol, LDL-cholesterol [LDL-C], and HDL-C) and corresponding parameters from NMR.

Four groups (DSA-T2DM, DSA-C, DwC-T2DM, and DwC-C) were considered as four outcomes, and concentrations of HDL-related features were scaled (z-normalization, i.e., with mean = 0 and SD = 1) to identify lipoproteins with different concentrations, and multinomial logistic regression analysis (MLR) was used. We set a specific reference for each comparison: between DSA-T2DM and DSA-C, DSA-C was the reference; between DwC-T2DM and DwC-C, DwC-C was the reference; between DSA-T2DM and DwC-T2DM, DwC-T2DM was the reference; between DSA-C and DwC-C, DwC-C was the reference. Age (continuous variable), sex (dichotomous variable), and current smoking status (dichotomous variable) were adjusted for the complete model. We considered a p-value < 0.05 as significant. The results were expressed as regression coefficient (β), and odds ratios (ORs) with a 95% confidence interval (CI) to evaluate the association between concentrations of HDL-related features and T2DM. OR was determined per 1 SD increase in lipoprotein concentration could have a higher/lower odds of having T2DM. We followed the STROBE guidelines to report our findings.

Next, based on HDL-related features, supervised dimension reduction analysis called sparse partial least squares discriminant analysis (sPLS-DA) was performed by using the “mixOmics” package in R [35]. sPLS-DA enables the selection of the most predictive or discriminative features in the data to classify the samples. Thus, by using this analysis, we aimed to rank and validate the features according to their contribution between DSA-T2DM and DSA-C, DwC-T2DM and DwC-C, DSA-T2DM and DwC-T2DM, and DSA-C and DwC-C. We then compared the rank of features determined by sPLS-DA and those generated from multinomial logistic regression by using Spearman’s correlation to further validate the findings. The contribution of loading features was calculated based on two components for each sPLS-DA model.

Multiple linear regression analyses were used to examine if there was an association between concentrations of differential HDL composition (exposures) and laboratory markers including fasting levels of gamma-glutamyl transferase (GGT), haemoglobin A1c (HbA1c), and glucose and anthropometrics markers including visceral adipose tissue (VAT), waist circumference, and waist-to-hip ratio (outcomes). Age (continuous variable), sex (dichotomous variable), and current smoking status (dichotomous variable) were adjusted for the complete model. We calculated differences in the mean levels of GGT, HbA1c, glucose, VAT, waist circumference, and waist-to-hip ratio, with 95% confidence intervals, associated with 1 SD concentration of differential HDL composition.

We further examined the associations between HDL composition and pan-microvascular-related complications, which were defined as composite diabetes-related complications including diabetic nephropathy, diabetic neuropathy, and diabetic retinopathy. A T2DM participant with one or more of these complications was considered as having a pan-microvascular-related complication. This analysis was performed separately for DSA and DwC populations, and the Mann-Whitney-Wilcoxon test was used to assess the statistical differences between the cases (e.g., with pan-microvascular-related complications) and controls (e.g., without pan-microvascular-related complications) for each ethnic group for all HDL features. Statistical analyses were performed in R (version 4.1.0) and GraphPad Prism version 8 (Graphpad Inc., La Jolla, CA, USA).

Clinical characteristics. After consecutive exclusion of 1 T1DM individual from the MAGNA VICTORIA study, 5 individuals who withdraw from the RCT, and 3 individuals with missing plasma samples, 92 individuals with T2DM were included. None of the healthy controls were excluded (Figure 1).

In total, 143 subjects were included in the present study: 47 DSA-T2DM subjects (19 men/28 women), 21 DSA-C (9 men/15 women), 45 DwC-T2DM subjects (25 men/20 women), and 30 DwC-C (16 men/14 women), (Table 1).

Compared to DSA-C, systolic blood pressure, BSA, BMI, waist circumference, waist-to-hip ratio, fasting glucose, HbA1c, triglyceride levels, and VAT were higher in DSA-T2DM; while total cholesterol, HDL-C, and LDL-C were lower (Table 1). In the DwC-T2DM group, BSA, BMI, waist circumference, waist-to-hip ratio, total cholesterol, LDL-C, total body fat, and SAT were higher compared to DwC-C, together with a higher proportion of current smokers. Comparing individuals with T2DM between the ethnic groups, DSA subjects had a longer diabetes duration, a higher incidence of vascular-related complications (e.g., retinopathy and macrovascular problems), and a higher albumin/creatine ratio compared to DwC subjects.

Comparison between NMR-based results vs clinical chemistry approach. NMR-based lipoprotein subclass measurements are shown in Figure 2a. To verify the quality of the NMR measurements we compared: total triglycerides, total cholesterol, HDL-C, and LDL-C to the clinical chemistry measurements revealing a high correlation in the total cohort (R = 0.81-0.99, p-value < 2e-16) as well as any single group (Supplemental Figure S1).

Distinct HDL composition between DSA-T2DM and DwC-T2DM. To identify differences in HDL composition between the various groups, 32 HDL-related features were tested (Supplemental Table S1) using multinomial logistic regression analyses. In total, 14 lipoprotein subfraction features were different (5 higher and 9 lower) in DSA-T2DM compared to DSA-C (Figure 2b and Supplemental Table S2); 21 lipoprotein subfraction features were different (4 higher and 17 lower) in DwC-T2DM compared to DwC-C (Figure 2c and Supplemental Table S3); 14 lipoprotein subfraction features (3 higher and 11 lower) were different in DSA-T2DM compared to DwC-T2DM (Supplemental Figure S2a and Supplemental Table S4); 20 lipoprotein subfraction features were lower in DSA-C than DwC-C (Supplemental Figure S2d and Supplemental Table S5). We found that 7 lipoprotein subfractions were specifically different in DSA, 13 lipoprotein subfraction features that were unique to DwC, and 8 lipoprotein subfractions that were common to both ethnicities (Figures 2c and f).

Multinomial logistic regression analysis was used to evaluate the association between HDL composition and the odds of having T2DM. The common trend in both ethnic groups showed that higher free cholesterol content in all HDL subfractions was associated with lower odds of having T2DM, while higher triglyceride content in total HDL, HDL-2, and HDL-3 was associated with higher odds of having T2DM (Figures 2d and g). Specifically, in DSA with T2DM we found a distinct HDL composition phenotype when compared to DwC with T2DM. These findings revealed that higher ApoA1-, Apo-A2-, and cholesterol content in the smallest and dense HDL (i.e., HDL-4), and higher total ApoA2 were associated with lower odds of having T2DM in DSA, whereas higher triglyceride content in the largest HDL particles (i.e., HDL-1) was associated with higher odds of having T2DM (Figure 2d). In the total DwC population, higher triglyceride content in HDL-4 was associated with higher odds of having T2DM while higher ApoA1 and total lipid content in the larger HDL fractions, especially in HDL-1, were associated with lower odds of having T2DM (Figure 2g).

When we compared T2DM subjects between DSA and DwC, notable changes were observed that in all the small and dense HDL subfractions, total ApoA1 and ApoA2, and phospholipid content in total HDL (HDPL) were higher and while in DwC-T2DM, large HDL subfractions such as ApoA1, cholesterol, and phospholipid content in HDL-1 (H1A1, H1CH, and H1PL) were lower in DSA with T2DM compared to those in DwC (Figure 2i). Considering the higher prevalence of T2DM in DSA, we also compared HDL composition among healthy individuals of both ethnicities and found that the majority of HDL composition except for small and dense HDL subfractions was lower in DSAs, and within small and dense HDL, only H4PL was lower in DSA-C (Supplemental Figures S2a and b). Our findings revealed ethnic differences in HDL composition between DSA and DwC, as well as changes in HDL composition in T2DM between two ethnicities that were not detectable by routine lipid or laboratory assessments.

HDL composition in plasma can differentiate patients with T2DM from healthy controls in two ethnic groups. Using HDL-related features, sPLS-DA, a supervised machine learning method combining variable selection and classification in a one-step procedure was performed. When comparing the T2DM between two ethnic groups, a significant separation was observed (Figure 3a).

Ranking HDL-related feature measurements by discriminating capability, we found all HDL-4 subfractions and total ApoA2 had the greatest contribution (Figure 3b). Notably, when comparing the rank generated from MLR and sPLS-DA, they were highly correlated (ρ = 0.9, p-value = 6.2e-8, Figure 3c), which further validated our previous findings. Meanwhile, similar but less profound findings were found in healthy individuals between the two ethnic groups (Supplemental Figures S2c-e). In addition, a clear separation between T2DM and healthy control was observed in both ethnicities, and the top 5 ranked features in DSA were free cholesterol content in total HDL and subclasses (HDL-2, HDL-3, and HDL-4) and ApoA2 in HDL-4 (HDFC, H2FC, H3FC, H4FC, and H4A2); those in DwC were free cholesterol content in total HDL and subclasses (HDL-1, HDL-2, and HDL-4) and ApoA1 in HDL1 (HDFC, H1FC, H2FC, H3FC, and H1A1), (Supplemental Figures S3c-d). The rank generated from MLR and sPLS-DA were highly correlated in both ethnic groups (ρ = 0.94-0.98, p-value < 2e-16, Figures 3e-f). These results suggested that HDL composition was different in T2DM and had ethnic specificity, especially ApoA2 and HDL-4 subfractions.

HDL anti-thrombotic capacity reduction in T2DM subjects in both ethnicities. Anti-thrombotic capacity was measured in 142 people to evaluate HDL functionality (47 DSA-T2DM, 21 DSA-C, 45 DwC-T2DM, and 29 DwC-C). Based on the thrombin generation curve, diabetic subjects had a higher curve than non-diabetic subjects in both ethnic groups (Supplemental Figures S4b and d). Furthermore, the endogenous thrombin potential (ETP) was significantly higher in T2DM patients versus healthy controls in both ethnicities (Supplemental Figures S4c and e), revealing that HDL functionality in anti-thrombin formation was impaired both in DSA and DwC with T2DM.

Associations between differential HDL composition and clinic outcomes. We further investigated the associations between differential HDL composition (ApoA2 and HDL-4 subtractions) and clinic outcomes (laboratory and anthropometric markers). In DSA-T2DM, with 1 SD increase of ApoA2 and cholesterol content in HDL-4 (H4A2 and H4CH), we found decreased levels of VAT. With 1 SD increase of ApoA1, ApoA2, phospholipids, free cholesterol, and cholesterol content in HDL-4 (H4A1, H4PL, H4FC, and H4CH), we observed lower waist circumference and waist-to-hip ratio; while with 1 SD increase of triglyceride content in HDL-4 (H4TG), we observed higher waist-to-hip ratio. 1 SD increase of H4PL and H4FC, was associated with lower fasting glucose and HbA1c; and per SD increase in H4CH was negatively associated with HbA1c. Interestingly, no association was observed between differential HDL composition in DSA-T2DM and GGT (Supplemental Figure S5a).

In DwC-T2DM, per 1 SD increase of total ApoA2 and H4A2, a negative association was observed with waist circumference and waist-to-hip ratio. 1 SD increase of H4FC was associated with lower fasting glucose. Per SD increase of H4TG was positively associated with GGT. No associations were found with levels of VAT and HbA1c. These data suggest that changes in HDL composition were clinically relevant, especially in DSA-T2DM which could partly reflect long-term dysregulated blood-glucose control.

Associations between differential HDL composition and pan-microvascular-related complications. Glycaemic control is associated with multiple diabetes-related complications, especially microvascular problems; therefore, we then investigated the associations between differential HDL composition (ApoA2 and HDL-4 subtractions) and microvascular complications. Interestingly, there was no association between HDL composition and pan-microvascular-related complications in white Caucasian subjects with T2DM, while, in DSA-T2DM, we observed total ApoA2, H4A1, H4FC and H4PL were lower in patients with pan-microvascular-related complications (Figure 4a).

When we compare single complications, TPA2, H4A1, H4A2, and H4FC were significantly lower in DSA-T2DM with retinopathy, while H4FC was higher in DwC-T2DM with retinopathy (Figure 4b). In DSA-T2DM, TPA2, H4A1, H4A2, H4CH, H4FC, and H4PL were significantly lower in diabetic neuropathy while none of these HDL compositions were associated with diabetic neuropathy in DwC-T2DM (Figure 4c). None of the differential HDL composition was changed between with and without diabetic nephropathy (Figure 4d). These data suggest that HDL composition changes, especially TPA2 and HDL-4 subfractions except for H4TG, in DSA were associated with diabetes-related microvascular complications such as neuropathy and retinopathy.

In the present study, we specifically observed that lower total ApoA2 and HDL-4 subclass concentrations in DSA individuals with T2DM were associated with higher odds of having diabetes-related pan-microvascular complications such as retinopathy and neuropathy. This change in HDL composition was not observed in DwC, in which typically ApoA1 levels changed upon T2DM, nor in either control group. When comparing individuals with T2DM of both ethnic groups with their controls, triglyceride content was found higher in most HDL subclasses and corresponding higher odds ratios for T2DM. Interestingly, contribution in DSA is specifically in the HDL-1 fraction while in DwC this is in the HDL-4 fraction. These results indicate that changes in HDL composition might be used as an ethnicity-specific biomarker for DSA with T2DM and in addition, may provide a mechanism underlying the increased risk of microvascular complications.

Previous studies did not account for the HDL composition, particularly the lipid content, and the majority of these studies were conducted on white ethnic groups [36, 37]. In our study, we included both DSA and DwC populations and generated HDL composition including 32 features, providing a higher resolution of HDL lipoprotein fraction than HDL-C alone. Sparse partial least squares discriminant analysis (sPLS-DA) based on HDL composition could not only distinguish healthy controls from T2DM in both ethnic groups but also show the distinction between T2DM in these two ethnic groups. This suggested that this technique may be clinically meaningful beyond the measurement of HDL-C, especially in different ethnicities.

We observed ethnicity-specific associations between HDL composition and T2DM and found a loss of large HDL subfractions and higher triglycerides in the smallest HDL in DwC with T2DM, consistent with the finding of the PREVEND study that HDL size is inversely associated with T2DM risk [36]. Interestingly, in DSA with T2DM, we observed an opposite trend, in that the smallest HDL subfractions were specifically lower and triglycerides were higher in the largest HDL. While in our study, T2DM patients of both ethnicities received similar medical care, we still discovered that total ApoA1 and ApoA2, as well as all HDL-4 subfractions, were significantly lower in DSA with T2DM than DwC with T2DM. Small and dense HDL particles play a vital role in multiple functions, such as cellular cholesterol efflux, antioxidative, antithrombotic, anti-inflammatory, and antiapoptotic capacities [38, 39]. These observations supported our findings in DSA and we hypothesized that in the DSA with T2DM group, impaired HDL function might be explained by the loss of functional small HDL subfractions; while in DwC with T2DM, increased triglyceride content in small HDL might lead to HDL dysfunction. Our in-house HDL functional assay revealed that T2DM patients in both ethnic groups had significantly reduced anti-thrombotic capacity due to impaired HDL function, which is also consistent with earlier findings showing that T2DM patients had impaired HDL function in terms of its capacity to suppress TNF-induced vascular cell adhesion molecule-1 expression in endothelial cells in vitro [19]. A previous study by Bakker et al. reported that in obese DSA, without T2DM, only the ability of HDL to prevent LDL oxidation was reduced in overweight DSA when compared to obese DwC (without T2DM) [25]. Following these findings, we discovered that obese DSA with T2DM had a specific change in HDL composition, indicating T2DM as a potential additional risk factor for cardiovascular disease.

HDL-4 subfractions in DSA with T2DM had the most clinical relevance, but not in DwC with T2DM. There were notable associations between HDL-4 subfractions with both waist circumference and waist-to-hip ratio, which were reported in HELIUS study as critical predictors for T2DM regardless of ethnic background [40]. Additional HDL-4 associations with glycemic control parameters, which were risk factors for diabetes progression, were discovered [41-43], and we did find differences in HDL-4 fraction concentrations between patients with and without pan-microvascular complications, particularly neuropathy and retinopathy. In contrast, there is no link between HDL compositional changes and diabetes-related complications in DwC with T2DM subjects, which could indicate that changes in HDL composition occurred as a consequence of hyperglycemia rather than as a factor associated with diabetes-related complications.

The strength of our study is that we measured detailed HDL compositions concerning two ethnic groups of diabetic and non-diabetic individuals. Our findings further confirmed impaired HDL function in T2DM, linking it with diabetes-related complications; meanwhile, we revealed an ethnic distinction in differences in HDL composition. Of note, there are still several limitations to our study. First, our study is a cross-sectional study, which precludes statements on causality. Second, sample sizes are relatively small, which might limit generalization potential and preclude stratification analyses. Third, statin use could affect HDL concentration, composition, and function [44, 45], suggesting that differences in HDL composition between healthy controls and T2DM may not be solely due to diabetes status. Fourth, due to limited plasma availability, we only performed an in-house HDL anti-thrombotic capacity assay, without performing cholesterol efflux capacity, anti-oxidation capacity, or anti-inflammatory capacity measurements. Fifth, the biological response of HUVECs is variable, therefore, our in-house assay is not comparable to clinical chemistry determinations.

In conclusion, Dutch T2DM patients of both white Caucasian and South Asian descent exhibited altered HDL composition when compared to healthy individuals, but revealed a distinct phenotype. In Dutch South Asian subjects, lower ApoA2 and HDL-4 were associated with higher odds of having T2DM as well as diabetes-related complications such as retinopathy and neuropathy, suggesting that they could be used as ethnicity-specific biomarkers for T2DM patients.

ApoA1: apolipoprotein A1; ApoA2: apolipoprotein A2; ACE: angiotensin-converting enzyme; B.I.LISA: Bruker IVDr Lipoprotein Subclass Analysis; BMI body mass index; BSA: body surface area; CI: confidence interval; DSA: Dutch South Asian; DwC: Dutch white Caucasian; ETP: endogenous thrombin potential; GGT: gamma-glutamyl transferase; HbA1c: hemoglobin A1c; HDL: high-density lipoprotein; HDL-C: high-density lipoprotein cholesterol; HUVEC: human umbilical vein endothelial cell; LDL: low-density lipoprotein; MAGNA VICTORIA: MAGNetic resonance Assessment of VICTOza efficacy in the Regression of cardiovascular dysfunction In type 2 diAbetes mellitus; MLR: multinomial logistic regression analysis; NMR: Nuclear magnetic resonance; OR: odds ratio; RCT: randomized controlled trial; SA: South Asian; SAT: subcutaneous adipose tissue; SD: standard deviation; T1DM: type 1 diabetes mellitus; T2DM: type 2 diabetes mellitus; TBF: total body fat percentage; TNF-α: tumor necrosis factor α; VAT: visceral adipose tissue; WC: white Caucasian.

Acknowledgments

We thank all the patients who participated in the MAGNA VICTORIA study and healthy volunteers. We express our gratitude to the researchers, nurses, and physicians that helped organize this study. We would also like to thank Wendy M.P.J. Sol for helping with HDL isolation.

Authors' contributions

LY, MAG, and BMvdB contributed to the study concept, design, and analysis; LY analysed and interpreted data, and critically revised the manuscript; RL-G interpreted data and revised the manuscript; AV performed NMR measurements; HJvE and MBB collected data for the MAGNA VICTORIA studies, IMJ supervised the MAGNA VICTORIA studies, with HJL serving as study director; PCNR and MAG interpreted NMR data; and LY, TJR, and BMvdB drafted the manuscript. The final manuscript was read, commented on, and approved by all authors.

Funding

The work was supported by ZonMW (The Enabling Technologies Hotels program, grant: 435005003, 2019) and by the China Scholarship Council grant to Lushun Yuan (CSC no. 201806270262).

Availability of data and materials

All data and methods supporting the findings of this study are available from the corresponding author upon reasonable request.

The authors declare that they have no relevant financial interests or personal relationships.

Ethics approval and consent to participate

The T2DM participants were from baseline samples of two previous randomized controlled trials (ClinicalTrials. gov NCT01761318 and NCT02660047). The study protocol was approved by the Institutional Review Board (Leiden University Medical Center, Leiden, The Netherlands), and all participants provided written informed consent.

Consent for publication

The manuscript was approved by all authors for publication.

Unnikrishnan R, Pradeepa R, Joshi SR, Mohan V. Type 2 Diabetes: Demystifying the Global Epidemic. Diabetes. 2017;66(6):1432–42.
Sattar N, Gill JM. Type 2 diabetes in migrant south Asians: mechanisms, mitigation, and management. Lancet Diabetes Endocrinol. 2015;3(12):1004–16.
van Niel J, Geelhoed-Duijvestijn P, Numans ME, Kharagjitsing AV, Vos RC. Type 2 diabetes in South Asians compared to Europeans: Higher risk and earlier development of major cardiovascular events irrespective of the presence and degree of retinopathy. Results from The HinDu The Hague Diabetes Study. Endocrinol Diabetes Metab. 2021;4(3):e00242.
Pardhan S, Gilchrist J, Mahomed I. Impact of age and duration on sight-threatening retinopathy in South Asians and Caucasians attending a diabetic clinic. Eye (Lond). 2004;18(3):233–40.
Raymond NT, Varadhan L, Reynold DR, Bush K, Sankaranarayanan S, Bellary S, Barnett AH, Kumar S, O'Hare JP, Group UKADSRS. Higher prevalence of retinopathy in diabetic patients of South Asian ethnicity compared with white Europeans in the community: a cross-sectional study. Diabetes Care. 2009;32(3):410–5.
Mansi IA, Chansard M, Lingvay I, Zhang S, Halm EA, Alvarez CA. Association of Statin Therapy Initiation With Diabetes Progression: A Retrospective Matched-Cohort Study. JAMA Intern Med. 2021;181(12):1562–74.
Liew SM, Lee PY, Hanafi NS, Ng CJ, Wong SS, Chia YC, Lai PS, Zaidi NF, Khoo EM. Statins use is associated with poorer glycaemic control in a cohort of hypertensive patients with diabetes and without diabetes. Diabetol Metab Syndr. 2014;6:53.
Sukhija R, Prayaga S, Marashdeh M, Bursac Z, Kakar P, Bansal D, Sachdeva R, Kesan SH, Mehta JL. Effect of statins on fasting plasma glucose in diabetic and nondiabetic patients. J Investig Med. 2009;57(3):495–9.
Mineo C, Shaul PW. Novel biological functions of high-density lipoprotein cholesterol. Circ Res. 2012;111(8):1079–90.
Jialal I, Jialal G, Adams-Huet B. The platelet to high density lipoprotein -cholesterol ratio is a valid biomarker of nascent metabolic syndrome. Diabetes Metab Res Rev. 2021;37(6):e3403.
Schmidt MI, Duncan BB, Bang H, Pankow JS, Ballantyne CM, Golden SH, Folsom AR, Chambless LE. Atherosclerosis Risk in Communities I: Identifying individuals at high risk for diabetes: The Atherosclerosis Risk in Communities study. Diabetes Care. 2005;28(8):2013–8.
Wilson PW, Meigs JB, Sullivan L, Fox CS, Nathan DM, D'Agostino RB. Sr.: Prediction of incident diabetes mellitus in middle-aged adults: the Framingham Offspring Study. Arch Intern Med. 2007;167(10):1068–74.
Sacks FM, Hermans MP, Fioretto P, Valensi P, Davis T, Horton E, Wanner C, Al-Rubeaan K, Aronson R, Barzon I, et al. Association between plasma triglycerides and high-density lipoprotein cholesterol and microvascular kidney disease and retinopathy in type 2 diabetes mellitus: a global case-control study in 13 countries. Circulation. 2014;129(9):999–1008.
Jia C, Anderson JLC, Gruppen EG, Lei Y, Bakker SJL, Dullaart RPF, Tietge UJF. High-Density Lipoprotein Anti-Inflammatory Capacity and Incident Cardiovascular Events. Circulation. 2021;143(20):1935–45.
Ajala ON, Demler OV, Liu Y, Farukhi Z, Adelman SJ, Collins HL, Ridker PM, Rader DJ, Glynn RJ, Mora S. Anti-Inflammatory HDL Function, Incident Cardiovascular Events, and Mortality: A Secondary Analysis of the JUPITER Randomized Clinical Trial. J Am Heart Assoc. 2020;9(17):e016507.
Emmens JE, Jia C, Ng LL, van Veldhuisen DJ, Dickstein K, Anker SD, Lang CC, Filippatos G, Cleland JGF, Metra M, et al. Impaired High-Density Lipoprotein Function in Patients With Heart Failure. J Am Heart Assoc. 2021;10(9):e019123.
Constantinou C, Karavia EA, Xepapadaki E, Petropoulou PI, Papakosta E, Karavyraki M, Zvintzou E, Theodoropoulos V, Filou S, Hatziri A, et al. Advances in high-density lipoprotein physiology: surprises, overturns, and promises. Am J Physiol Endocrinol Metab. 2016;310(1):E1–14.
Xepapadaki E, Nikdima I, Sagiadinou EC, Zvintzou E, Kypreos KE. HDL and type 2 diabetes: the chicken or the egg? Diabetologia 2021, 64(9):1917–1926.
Ebtehaj S, Gruppen EG, Parvizi M, Tietge UJF, Dullaart RPF. The anti-inflammatory function of HDL is impaired in type 2 diabetes: role of hyperglycemia, paraoxonase-1 and low grade inflammation. Cardiovasc Diabetol. 2017;16(1):132.
Heier M, Borja MS, Brunborg C, Seljeflot I, Margeirsdottir HD, Hanssen KF, Dahl-Jorgensen K, Oda MN. Reduced HDL function in children and young adults with type 1 diabetes. Cardiovasc Diabetol. 2017;16(1):85.
Lounila J, Ala-Korpela M, Jokisaari J, Savolainen MJ, Kesaniemi YA. Effects of orientational order and particle size on the NMR line positions of lipoproteins. Phys Rev Lett. 1994;72(25):4049–52.
Jeyarajah EJ, Cromwell WC, Otvos JD. Lipoprotein particle analysis by nuclear magnetic resonance spectroscopy. Clin Lab Med. 2006;26(4):847–70.
Jiang H, Peng J, Zhou ZY, Duan Y, Chen W, Cai B, Yang H, Zhang W. Establishing (1)H nuclear magnetic resonance based metabonomics fingerprinting profile for spinal cord injury: a pilot study. Chin Med J (Engl). 2010;123(17):2315–9.
Straat ME, Martinez-Tellez B, Nahon KJ, Janssen LGM, Verhoeven A, van der Zee L, Mulder MT, Kooijman S, Boon MR, van Lennep JER, et al. Comprehensive (apo)lipoprotein profiling in patients with genetic hypertriglyceridemia using LC-MS and NMR spectroscopy. J Clin Lipidol. 2022;16(4):472–82.
Bakker LE, Boon MR, Annema W, Dikkers A, van Eyk HJ, Verhoeven A, Mayboroda OA, Jukema JW, Havekes LM, Meinders AE, et al. HDL functionality in South Asians as compared to white Caucasians. Nutr metabolism Cardiovasc diseases: NMCD. 2016;26(8):697–705.
Nodeland M, Klevjer M, Saether J, Giskeodegard G, Bathen TF, Wisloff U, Bye A. Atherogenic lipidomics profile in healthy individuals with low cardiorespiratory fitness: The HUNT3 fitness study. Atherosclerosis. 2022;343:51–7.
Bizino MB, Jazet IM, Westenberg JJM, van Eyk HJ, Paiman EHM, Smit JWA, Lamb HJ. Effect of liraglutide on cardiac function in patients with type 2 diabetes mellitus: randomized placebo-controlled trial. Cardiovasc Diabetol. 2019;18(1):55.
van Eyk HJ, Paiman EHM, Bizino MB, de Heer P, Geelhoed-Duijvestijn PH, Kharagjitsingh AV, Smit JWA, Lamb HJ, Rensen PCN, Jazet IM. A double-blind, placebo-controlled, randomised trial to assess the effect of liraglutide on ectopic fat accumulation in South Asian type 2 diabetes patients. Cardiovasc Diabetol. 2019;18(1):87.
Paiman EHM, van Eyk HJ, Bizino MB, Dekkers IA, de Heer P, Smit JWA, Jazet IM, Lamb HJ. Phenotyping diabetic cardiomyopathy in Europeans and South Asians. Cardiovasc Diabetol. 2019;18(1):133.
Findeisen M, Brand T, Berger S. A 1H-NMR thermometer suitable for cryoprobes. Magn Reson Chem. 2007;45(2):175–8.
Wu PS, Otting G. Rapid pulse length determination in high-resolution NMR. J Magn Reson. 2005;176(1):115–9.
Price WS. Water signal suppression in NMR spectroscopy. Annu Rep Nmr Spectro. 1999;38:289–354.
Kumar A, Ernst RR, Wuthrich K. A two-dimensional nuclear Overhauser enhancement (2D NOE) experiment for the elucidation of complete proton-proton cross-relaxation networks in biological macromolecules. Biochem Biophys Res Commun. 1980;95(1):1–6.
Mulder DJ, de Boer JF, Graaff R, de Vries R, Annema W, Lefrandt JD, Smit AJ, Tietge UJ, Dullaart RP. Skin autofluorescence is inversely related to HDL anti-oxidative capacity in type 2 diabetes mellitus. Atherosclerosis. 2011;218(1):102–6.
Rohart F, Gautier B, Singh A, Le Cao KA. mixOmics: An R package for 'omics feature selection and multiple data integration. PLoS Comput Biol. 2017;13(11):e1005752.
Sokooti S, Flores-Guerrero JL, Heerspink HJL, Garcia E, Connelly MA, Bakker SJL, Dullaart RPF. Lipoprotein particle sizes and incident type 2 diabetes: the PREVEND cohort study. Diabetologia. 2022;65(2):402–5.
Sokooti S, Flores-Guerrero JL, Kieneker LM, Heerspink HJL, Connelly MA, Bakker SJL, Dullaart RPF. HDL Particle Subspecies and Their Association With Incident Type 2 Diabetes: The PREVEND Study. J Clin Endocrinol Metab. 2021;106(6):1761–72.
Camont L, Lhomme M, Rached F, Le Goff W, Negre-Salvayre A, Salvayre R, Calzada C, Lagarde M, Chapman MJ, Kontush A. Small, dense high-density lipoprotein-3 particles are enriched in negatively charged phospholipids: relevance to cellular cholesterol efflux, antioxidative, antithrombotic, anti-inflammatory, and antiapoptotic functionalities. Arterioscler Thromb Vasc Biol. 2013;33(12):2715–23.
He Y, Ronsein GE, Tang C, Jarvik GP, Davidson WS, Kothari V, Song HD, Segrest JP, Bornfeldt KE, Heinecke JW. Diabetes Impairs Cellular Cholesterol Efflux From ABCA1 to Small HDL Particles. Circ Res. 2020;127(9):1198–210.
Zethof M, Mosterd CM, Collard D, Galenkamp H, Agyemang C, Nieuwdorp M, van Raalte DH, van den Born BH. Differences in Body Composition Convey a Similar Risk of Type 2 Diabetes Among Different Ethnic Groups With Disparate Cardiometabolic Risk-The HELIUS Study. Diabetes Care. 2021;44(7):1692–8.
Stuhldreher WL, Becker DJ, Drash AL, Ellis D, Kuller LH, Wolfson SK, Orchard TJ. The association of waist/hip ratio with diabetes complications in an adult IDDM population. J Clin Epidemiol. 1994;47(5):447–56.
Li W, Gong X, Wang W, Xiong K, Meng J, Li Y, Wang L, Liang X, Jin L, Huang W. Association of different kinds of obesity with diabetic retinopathy in patients with type 2 diabetes. BMJ Open. 2022;12(5):e056332.
Marcovecchio ML, Lucantoni M, Chiarelli F. Role of chronic and acute hyperglycemia in the development of diabetes complications. Diabetes Technol Ther. 2011;13(3):389–94.
Orsoni A, Therond P, Tan R, Giral P, Robillard P, Kontush A, Meikle PJ, Chapman MJ. Statin action enriches HDL3 in polyunsaturated phospholipids and plasmalogens and reduces LDL-derived phospholipid hydroperoxides in atherogenic mixed dyslipidemia. J Lipid Res. 2016;57(11):2073–87.
Pirillo A, Catapano AL. Pitavastatin and HDL: Effects on plasma levels and function(s). Atheroscler Suppl. 2017;27:e1–9.

Table 1 is available in the Supplementary Files section

No competing interests reported.

Download PDF

Journal Publication

published 17 May, 2023

Read the published version in Diabetes, Obesity and Metabolism →

Version 1

posted

You are reading this latest preprint version

Altered HDL composition is associated with risk for complications in type 2 diabetes mellitus in South Asian descendants: a cross-sectional, case-control study on lipoprotein subclass profiling

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Material And Methods

Results

Discussion

Conclusions

Abbreviations

Declarations

References

Table

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1