Association of Clinical Characteristics with the Proportion of the HDL-Cholesterol Subclasses HDL2b and HDL-3 Among Patients Undergoing Hemodialysis

Background: Altered high-density lipoprotein cholesterol (HDL-C) composition in patients with chronic kidney disease is common. However, reports on the distribution of HDL-C subclasses in patients undergoing hemodialysis (HD) are limited. Objective: We aimed to compare the two main HDL-C subclasses, HDL-2b and HDL-3, in two cohorts of HD patients and healthy individuals and examine their associations with clinical characteristics. Methods: A total of 164 prevalent HD patients and 71 healthy individuals in one hospital-facilitated outpatient clinic were enrolled from May 2019 to July 2019. The HDL-2b and HDL-3 proportions were measured and statistical analysis was performed. Results: The mean ages of HD patients and healthy individuals were 63 and 49.9 years, respectively. HD patients showed lower HDL-2b and HDL-3 proportions compared with those of healthy individuals (23.6% vs. 31.2%, P < 0.001; 31.7% vs. 33.6%, P = 0.137, respectively). The HDL-2b proportion was signicantly higher with a high-sensitivity C-reactive protein (hs-CRP) levels of <3 mg/L compared with hs-CRP ≥ 3mg/L in the HD cohort (P = 0.005). HDL-3 proportion was lower with a hs-CRP level of <3 mg/L compared with hs-CRP ≥ 3mg/L in the HD cohort (P = 0.022). Sex and diabetes did not inuence the HDL2b and HDL-3 proportions in the HD cohort. Conclusions: HD patients had lower HDL-2b and HDL-3 proportions than those of healthy individuals. The distribution of the HDL-2b and HDL-3 subclasses in HD patients is inuenced by proinammatory status, not by sex and diabetic status.


Introduction
Patients with chronic kidney disease (CKD) are at high risk for cardiovascular diseases (CVD) [1,2]. In particular, CVD confers to high overall mortality in the dialysis population [3,4]. High-density lipoprotein cholesterol (HDL-C) has been well recognized as an independent predictor of CVD in the general population [5,6]. Interestingly, several conditions are free from this association, for example, drug intervention to elevate the HDL-C levels, re ned HDL-C by genetic effect, and speci c higher HDL-C levels[7-9].
In the CKD population, dyslipidemia has been considered as a major risk factor for CVD. Screening for dyslipidemia in patients with CKD is supported by clinical guidelines [10][11][12]. Although decreased HDL-C levels and its impaired composition and function are generally documented in the CKD population, the association of HDL-C with CVD risk is still controversial [13][14][15][16][17].
HDL is composed of several subclass particles, varying in size, density, chemical composition, and physicochemical properities [18]. Generally, HDL can be divided into the following subclasses by gel electrophoresis coupled with immunoblotting: HDL-2a, HDL-2b (large size, cholesterol rich), HDL-3a, HDL-3b, HDL-3c, preβ 1 -HDL (small size), and preβ 2 -HDL [19]. In prior studies, HDL subclasses contribute to different effects on atherosclerosis. An increased large particle level, HDL-2b, exerts an antiatherogenic effect. In contrast, increased smaller particle levels, HDL-3 and preβ 1 -HDL, are positively associated with CVD[18,[20][21][22]. Due to complex technique and instrument requirement, the measurement of HDL subclasses is not widely applied in the past years. Recently, a micro uidic chip-based technique is developed. This fast, easy-to-operate technique is successfully applied to measure the HDL subclasses in the epidemiological studies [23,24].
In this study, we hypothesize that patients with hemodialysis (HD) have different proportions of the HDL subclasses compared with healthy individuals. Using the micro uidic chip-based technique, we measured the HDL-2b and HDL-3 subclass proportions in HD patients and correlated these with their demographic characteristics and proin ammatory status.

Study design and participants
Adult patients (> 18 years) who underwent maintenance HD thrice weekly for at least 3 months in the outpatient clinic in Kaohsiung Chang Gung Memorial Hospital in Taiwan from May 2019 to July 2019 were enrolled in this study. The exclusion criteria were as follows: ongoing treatment for malignancy, acute in ammatory diseases, hospitalization within 3 months, malnutrition de ned by serum albumin level < 3.5 g/dL, and pregnancy. Healthy controls were recruited voluntarily in the outpatient clinic by posted protocol noti cation. All blood samples from HD patients in the fasting status and in mid-week (Wednesday and Thursday) were obtained. Informative patient data, including demographic pro les and laboratory parameters, were also collected.

Analytic Parameters
All blood samples for biochemistry measurement were obtained using commercial kits and an autoanalyzer (Hitachi 7600 − 210, Hitachi Ltd., Tokyo, Japan). Albumin levels were measured using the bromocresol green method. Intact parathyroid hormone level was measured using a chemiluminescence immunoassay (Siemens Healthcare Diagnostics Inc., USA). The high-sensitivity C-reactive protein (hs-CRP) level was assayed using the immunoturbidimetric method (Spectra East Laboratories, Rockleigh, NJ, USA).
The plasma total cholesterol, triglyceride, and HDL-C levels were determined enzymatically on the Eroset Hitachi 7600 − 210 analyzer. The low-density lipoprotein cholesterol (LDL-C) levels were calculated using to the Friedewald formula, which provides reliable values up to a triglyceride level of 4.0 mmol/L. HDL-C subclass pro les were measured by electrophoresis of a micro uidic chip system. Brie y, serum samples, calibrator, and QC materials were diluted 1:50 in sample buffer in the presence of a mixture of lipophilic uorescent dyes and allowed to incubate for 5-15 min prior to loading on to chips. Separation was carried out in a micro uidic device (MICEP-30, Ardent BioMed). The entire procedure was performed in less than 1 h. The HDL-2b and HDL-3 subclasses were automatically calculated in line by a proprietary algorithm (Ardent BioMed LLC, Mt. View, California, USA).

Statistical analysis
The baseline demographic characteristics and laboratory measurements in HD patients and healthy controls are presented as frequency (percentage) and mean (standard deviation). The distribution difference was estimated using the independent two-sample t-test or chi-square test. The correlation between the HDL-2b and HDL-3 proportions and associated variables was estimated using the Pearson correlation test. A boxplot was used to illustrate the HDL-2b and HDL-3 subclass proportions in different subgroups, and the difference in these proportions between comparison groups was estimated using the independent two-sample t-test. All P values were two-sided, and P < 0.05 was considered statistically signi cant. All statistical analyses were performed using the R 3.6.3 software (R Core Team, 2020).

Characteristics of participants
Participants were screening by inclusion and exclusion criteria. Finally, a total of 164 participants with HD and 71 healthy controls were included in the analyses (Fig. 1). The mean ages of participants with HD and healthy controls were 63 and 49.9 years, respectively. The male-to-female percentages in the HD cohort and healthy controls were 48.8-51.2% and 31.0-69.0%, respectively. In the comparison of main laboratory parameters, HD patients showed a signi cantly lower HDL-2b subclass proportion compared with healthy controls (23.6% vs. 31.2%, P < 0.001) as well as lower total cholesterol (152 vs. 189 mg/dL, P < 0.001), HDL-C (44.5 vs. 59.8 mg/dL, P < 0.001), and LDL-C (88.1 vs. 112.1 mg/dL, P < 0.001) levels. The HDL-3 subclass proportion was not signi cantly different between the HD patients and healthy controls (31.7% vs. 33.6%, P = 0.137) ( Table 1).

Associations of the hs-CRP levels with the HDL-2b and HDL-3 subclass proportions
We strati ed the entire cohort with the cutoff hs-CRP level of 3 mg/L (normal range < 3 mg/L in the laboratory) and examined the associations with the HDL-2b and HDL-3 subclass proportions. The results showed that the HDL-2b (P = 0.031) and HDL-3 (P = 0.030) subclass proportions were signi cantly lower in the HD cohort under the hs-CRP level of < 3 mg/L compared with healthy controls. Similarly, the HDL-3 (P = 0.004) subclass proportion was signi cantly lower in the HD cohort under the hs-CRP level of ≥ 3 mg/L compared with healthy controls. However, this relationship was not noted in the HDL-2b(P = 0.274) subclass proportion (Fig. 3 − 1). We further examined the associations of the hs-CRP levels < 3 mg/L with the HDL-2b and HDL-3 subclass proportions in the HD cohort. The results showed that the HDL-2b (P = 0.005) subclass proportion was signi cantly lower under hs-CRP levels ≥ 3 mg/L compared with those with hs-CRP levels < 3 mg/L. In contrast, HDL-3 (P = 0.022) subclasses revealed an opposite trend (Fig. 3 − 2).

In uence of diabetes on the HDL-2b and HDL-3 subclass proportions in the HD cohort
Diabetic patients with HD did not demonstrate signi cant differences in the HDL-2b and HDL-3 subclass proportions compared with nondiabetic HD patients (Fig. 4).

Discussion
The key ndings in our study are that HD patients presented lower HDL-2b and HDL-3 subclass proportions compared with healthy controls. In HD patients, the HDL-2b and HDL-3 subclass proportions was in uenced by proin ammatory status, not by sex differentiation and diabetes. Our ndings indicate altered HDL composition in HD patients and potential utilization of the proportion of the HDL subclass analysis in clinical practice in these patients. We prefer to use proportions or ratio of HDL subclass analysis instead of a simple subclass measurement to generalize across the whole spectrum of the HDL levels [25,26]. Moreover, the micro uidic chip-based technique is easy to be applied in the clinical investigation and could forward to cause-relationship investigation. . In our study, we found that patients with HD showed lower HDL-2b and HDL-3 subclass proportions in both sexes compared with healthy individuals. We also examined the associations of HDL-C subclasses with various clinical parameters. However, there is only HDL-C showing strong association with HDL-2b and HDL-3 subclasses. Due to the lack of compared reports, above ndings need to be clari ed in the future.
CKD is a proin ammatory status. CRP is commonly used as a marker of systemic in ammation.
Recently, an elevated hs-CRP level has been linked to malnutrition, in ammation, and atherosclerosis syndrome and is considered to be a risk factor for morbidity and mortality in patients with CKD[37-39]. For mechanism investigation, CRP/oxLDL/β2GPI complex-aggravated atherosclerosis by increasing lipid uptake has been reported in a diabetic mouse study [40]. In our study, we found that decreased HDL-2b and HDL-3 subclass proportions in HD patients in either low or high hs-CRP levels compared with healthy individuals. Interestingly, the HDL-2b and HDL-3 subclass proportions were in uenced by the hs-CRP levels in HD patients. However, diabetic status did not in uence the HDL-2b and HDL-3 subclass proportions in HD patients. Accordingly, proin ammatory status could suppress the HDL-2b and increase HDL-3 subclass proportions in patients with HD. However, the relationship between the severity of proin ammatory status and diabetes and HDL subclass distribution still needs to be determined in a future investigation.
Our study has several limitations. First, this was a small-sized, single-center study on an Asian HD population. The results may not be extrapolated to other ethnic HD populations. Second, our study only measured the HDL subclass proportion in HD patients in one time point. The longitudinal effect of HD on HDL subclass distribution cannot be obtained in our study. Third, our study participants were relatively clinically stable; therefore, associations between proin ammatory status and HDL subclass distribution cannot be strati ed by wide-range hs-CRP levels. Despite the aforementioned limitations, the strengths of our study are that this is the rst clinical study to examine the proportions of the HDL subclasses HDL-2b and HDL-3 in HD patients and there is availability of cause-speci c data to ll knowledge gap about reasons for HDL subclass distribution in these patients. The clinical utility of HDL subclass distribution analysis would be further facilitated by spanning a full-range HDL-C population and determine the subclass distribution on adverse CVD event.
Conclusions HD patients have lower HDL-2b and HDL-3 subclass proportions compared with healthy individuals. The distribution of the HDL-2b and HDL-3 subclasses is in uenced by proin ammatory status, not by sex and diabetic status. Figure 1 Participants ow diagram.

Figure 2
Proportion of HDL-2b and HDL-3 subclasses among male and female study cohorts Proportion of HDL-2b and HDL-3 subclasses in patients with hemodialysis strati ed by an hs-CRP concentrations 3 mg/L. Proportion of HDL-2b and HDL-3 subclasses in diabetic and non-diabetic patients with hemodialysis.