A total of 1217 participants, 510 with normoglycemia, 318 with prediabetes, and 389 with T2DM, were identified from different cohorts of 4 participating institutions belonging to the same health care organization in the north-northeast region of Spain. After excluding those participants who were under lipid-lowering therapy, 929 participants were analyzed (463 with normoglycemia, 250 with prediabetes, and 216 with T2DM). Normoglycemic and prediabetic groups were selected from 3 previously published cross-sectional studies13–15 (2 university hospital cohorts and one primary care cohort in Spain). The study participants with T2DM were selected from the same 2 cross-sectional studies performed at the university hospitals from where participants with normoglycemia and prediabetes were recruited. T2DM participants were recruited from the outpatient clinic of 1 of the participating centers as well as from those identified by screening patients enrolled in the diabetic eye disease program.16 Some of them were new-onset T2DM patients recruited from the outpatient clinic of the department of endocrinology.14

The inclusion criteria for all 3 groups (normoglycemic, prediabetic, and T2DM) were as follows: absence of established impaired renal function (defined as an estimate glomerular filtration rate <60mL/min/1.73 m2), and absence of known heart disease (defined as any known peripheral artery disease, stroke, heart failure, coronary artery disease, including previous myocardial infarction, angina, previous coronary artery bypass surgery, or percutaneous coronary intervention). Exclusion criteria were active lipid-lowering treatment (defined as statin or fibrate drugs) and a diagnosis of type 1 diabetes mellitus (T1DM) or suspicion of having any other specific type of diabetes secondary to genetic defects, endocrinopathies, pancreatic exocrine dysfunction or chemically-induced diabetes.

We recorded participant characteristics, anthropometric parameters, and laboratory measurements (glucose, HbA1c, TGs, total/HDL/LDL-C, creatinine, estimated glomerular filtration rate, alanine transaminase (ALT), aspartate transaminase (AST), gamma-glutamyltransferase (GGT), insulin and homeostatic Model Assessment of Insulin Resistance [HOMA-IR]). To evaluate the presence of NAFLD, the Fatty Liver Index (FLI) was also calculated in all study participants. An FLI> 60 was considered suggestive of NAFLD.

This study was conducted in accordance with the Declaration of Helsinki and was approved by the local Ethics Committees. All participants signed a written informed consent form.

In the present study, the diagnosis of diabetes was made according to the criteria established by the American Diabetes Association. Prediabetes was also defined according to the American Diabetes Association criteria.17 We considered as prediabetic any participant who met 1 of the 2 following criteria: a) impaired fasting plasma glucose, defined as fasting plasma glucose between 100 and <126mg/dL (5.55-6.99 mmol/L), or b) HbA1c levels between 5.7% and <6.5% (39-48 mmol/mol). Participants in the normoglycemic group had fasting plasma glucose and HbA1c values below 100 mg/dL and 5.7%, respectively.

Participants’ weight, height and waist circumference were measured using standardized methods, and their blood pressure (mean of 2 measurements, 5minutes apart) was measured after 10minutes in a seated position using a blood pressure monitor (HEM-7001E, Omron, Barcelona, Spain). Body mass index (BMI) was calculated as weight (kg)/stature (m2).

Serum and spot urine samples were collected in the fasting state, and all serum and urine tests were performed using standard laboratory methods. LDL-C was estimated using the Friedewald formula, and estimate glomerular filtration rate was estimated using the Modification of Diet in Renal Disease-4 formula.18 HbA1c levels were determined using HPLC (Variant, Bio-Rad Laboratories SA, Spain), and its concentrations are expressed in the National Glycohemoglobin Standardization Program/Diabetes Control and Complications Trial units. Urine albumin was measured using an immunoturbidimetric method and a Roche/Hitachi Modular P analyzer (Roche Diagnostics, Spain). Insulin-resistance was assessed by homeostatic model assessment (HOMA)-IR.

Blood serum samples were shipped on dry ice to the Biosfer Teslab facilities (Reus, Spain) for advance lipoprotein testing by using the Liposcale test, an advanced lipoprotein test (CE) based on 2D NMR diffusion-ordered spectroscopy that enables exhaustive analysis of lipoprotein particles.19 We determined the lipid composition and the mean size (nm) for each particle class, and the particle concentration of 9 lipoprotein subclasses, namely large, medium, and small VLDL, LDL, and HDL-P. As previously described, the within-assay precision of the method for the determination of cholesterol and TG concentrations, and particle numbers for the LDL class and its small, medium, and large subclasses, is ≤ 5%. The interassay precision for the same parameters is ≤ 8%. Similarly, the within-assay and interassay precision for cholesterol and TG concentrations, and VLDL and HDL-P, are ≤ 6%. Finally, both within-assay and interassay precision for the mean particle size for each lipoprotein class is ≤ 1%.19

Data management and analyses were conducted using the free R statistical software version 3.6.0. Descriptive statistics are summarized with median, interquartile ranges [25th-75th], mean and standard deviation for continuous variables, or frequency and percentage for categorical variables. The Student t test or chi-square test was used to explore the differences between groups. P-values corresponding to pairwise comparisons were calculated by multiple testing with the Tukey method. Pearson correlation coefficients were computed between NMR-assessed advanced lipoprotein profiles and clinical and laboratory parameters in the normoglycemic control group. To assess the adjusted differences between groups from each lipoprotein subclass, we used multivariate regression models. The covariate variables included in the adjusted analysis were age, sex, and BMI. Furthermore, we applied multi testing correction according to the Bonferroni method to control the family-wise error rate prefixed to 0.05.

Participants were classified according to their serum concentration of LDL-P and HDL-P particles and lipid content, ie, normal and abnormal, according to previously published cutoffs in relation to CVD risk established in previous studies.20,21 Thus, the cutoffs for abnormal level of lipoprotein particle concentrations in the normoglycemic group were as follows: total LDL-P> 1300 nmol/L, and <24μmol/L for total HDL-P concentration; the LDL-C levels were considered elevated when they exceeded 130mg/dL, whereas HDL-C was considered low below 40mg/dL in men and 50mg/dL in women, respectively.

Serum lipoprotein particle concentrations were analyzed in 929 participants. From this cohort, 463 (49.8%) were participants with normoglycemia, 250 (26.9%) were participants with prediabetes, and 216 (23.3%) were participants with T2DM. A descriptive analysis of clinical and analytical variables by group is shown in table 1. Compared with the normoglycemic control group, participants with prediabetes and T2DM groups were older, had higher BMI and waist circumference, and a higher percentage had hypertension. They also had a higher fasting plasma glucose and insulin, HbA1c and a higher median HOMA-IR. Regarding the lipid profile, there were some differences in fasting cholesterol subclasses across the 3 groups, with lower serum HDL-C concentrations in participants with T2DM compared with normoglycemic participants and higher serum LDL-C in the prediabetic group than in the normoglycemic group.

Table 2 and table 3 describe the lipoprotein values according to the age range in both sexes. The analysis of the whole normoglycemic group showed a significant increase of several lipid parameters with increasing age with a P trend <.001 for total cholesterol, P <.01 for LDL-C, P <.001 for total LDL-P, P <.001 for large LDL-P, P <.001 for medium LDL-P and P <.001 for small LDL-P. HDL particles also increased with increasing age: P=.03 for total HDL-P and P <.001 for medium HDL-P. Interestingly, the increase in age was more marked in younger and middle-aged age ranges than between middle-aged and older age ranges.

Lipoprotein profile was strongly dependent on the participants’ sex. Compared with men, women had lower mean serum TG and LDL-C concentrations (82.5mg/dL vs 114mg/dL, P=.001; 110mg/dL vs 118mg/dL, P=.010, respectively). In line with serum TG, women showed concomitantly lower concentrations of VLDL-P and lipids, accompanied by elevated concentrations of HDL-P and lipids (P <.001 for all comparisons except for medium VLDL-P, which was P=.010). LDL characteristics also showed significant sex-related differences. Compared with men, total concentrations of LDL-P and small LDL-P were significantly lower in women (1256 [697-11 981] vs 1319 [572-12 179] nmol/L; P=.014; 638 [324-1047] nmol/L vs 735 [401-1171]; P <.001, respectively). Similarly, serum concentrations of non-HDL-P, including VLDL-P and LDL-P, were significantly lower in women than in men (P=.001). Women also showed lower cholesterol and TG content in VLDL particles and higher cholesterol and TG content in HDL particles (table 1 of the supplementary data).

Only 75 out of 462 normoglycemic participants (16%) showed abnormal values for both HDL-C (with values <40mg/dL in men and <50mg/dL in women) and LDL-C (with values> 130mg/dL), with the rest of the participants (84%) showing normal values for both parameters. Interestingly, among normoglycemic participants with normal LDL-C and HDL-C, nearly 50% showed higher than recommended LDL-P concentrations (50.4% with high levels of total LDL-P). In contrast, only 1.8% of normoglycemic participants showed lower than recommended HDL-P concentrations (table 4).

Figure 1 shows the correlation between different NMR-derived lipoprotein characteristics and different clinical variables. The most notable observations were that FLI, waist circumference and BMI were positively related to VLDL-related characteristics (total number, size and cholesterol and TG content), LDL-P small subclass and non-HDL-P. The same clinical variables were inversely related to total HDL-P, medium, and small subclasses, as well as their cholesterol content.

Figure 1.

Correlation between NMR-assessed advanced lipoprotein profile and clinical parameters in the normoglycemic control group. HbA1c: glycosylated hemoglobin; HDL, high-density lipoprotein; LDL, low-density lipoprotein; VLDL, very low-density lipoprotein.

(0.26MB).

The comparison between lipoprotein subclass sizes between the normoglycemic, prediabetic and T2DM groups are shown in table 5. Multivariate analysis (adjusted for age, gender, and BMI) showed that almost all the VLDL subclasses were significantly higher in participants with T2DM than in normoglycemic participants, with higher serum concentrations of large and small VLDL-P (P=.040 and P=.042, respectively). No significant differences in the LDL-P distribution were observed when comparing the 3 groups. Furthermore, we found a statistically significant reduction in LDL size (LDL-Z) in participants with T2DM compared with that of normoglycemic participants (P=.018). There was a trend toward a higher content of cholesterol in VLDL particles in participants with T2DM than in normoglycemic participants (P=.083). Regarding HDL-related characteristics, total HDL-P levels in the T2DM group were lower than in the normoglycemic group (P=.017), especially medium particles (P=.013). Median cholesterol content in HDL-P was concomitantly lower in participants with T2DM than in controls (P=.002).

Our findings should be interpreted within the context of some potential limitations. First of all, we could not distinguish participants with an impaired fasting glucose from those with impaired glucose tolerance (IGT) among the prediabetic group since study participants did not undergo an oral glucose tolerance test. This may be important because distinct lipoprotein and apolipoprotein changes in individuals with impaired fasting plasma glucose and IGT have been reported, with the latter being associated with lipoprotein changes similar to those previously described in insulin-resistance.33 The authors hypothesized that these differences could be the result of distinct pathophysiologic mechanisms in these distinctive states of glucose tolerance, which may be related to the site of insulin-resistance (skeletal muscle or hepatic). Another limitation is that all T2DM participants were recruited mainly from the outpatient clinic, and there may have been a selection bias toward including healthier patients with fewer complications than in the general T2DM population.