The study protocol complied with the ethical guidelines of the 1964 Declaration of Helsinki and its later amendments. Before this observational cohort commenced, appropriate approval was obtained from the Institution Review Board and Ethics Committee of Chang Gung Medical Foundation, Taoyuan, Taiwan (permit numbers: 201601181A3 and 201701735A3C501). We used the KDIGO guideline to define the stages of CKD (18). The estimated glomerular filtration rate (eGFR) was calculated according to the Schwartz formula (19). CKD children with eGFR ≥90 ml/min/1.73 m² were categorized as stage G1; G2 denotes eGFR 60–89 ml/min/1.73 m²; G3 represents 30–59 ml/min/1.73 m²; and G4 stands for 15–29 ml/min/1.73 m². Inclusion criteria for CKD subjects were the following: children and adolescents aged 3–18 years with CKD stages G1–G4. CKD patients who were pregnant, already at eGFR <15 ml/min/1.73 m², or on dialysis, as well as those who had congenital heart disease or renal transplant, or were unable to receive CV assessment, were excluded from the study. All of the participants provided signed informed consent for study. The causes of CKD were divided into two categories, CAKUT or non-CAKUT, as described previously (20). CAKUT structural anomalies included renal agenesis, posterior urethral valves, duplex collecting system, multi-cystic kidney dysplasia, kidney dysplasia, kidney hypoplasia, horseshoe kidney, and ureter abnormalities (4).

This ancillary study was performed in a subgroup of 88 participants who received the analysis of plasma apoC-II and apoC-III levels and CV assessments simultaneously. Fasting blood samples were drawn and aliquoted, and spot urine samples were collected and stored in a −80°C freezer until analysis. Blood urea nitrogen (BUN), creatinine (Cr), uric acid, glucose, hemoglobin, hematocrit, sodium, potassium, calcium, phosphate, urine total protein-to-Cr ratio, total cholesterol, triglyceride, and low-density lipoprotein (LDL) were measured by the hospital central laboratory, as described previously (8). Plasma apoC-II concentrations were determined by an ELISA kit (Catalog No. EHAPOC2; Invitrogen, Thermo Fisher Scientific Inc., Carlsbad, CA, USA), and inter-assay coefficients of variations were <12%. Plasma apoC-III concentrations were analyzed by an ELISA kit (Catalog No. EHAPOC3; Invitrogen, Thermo Fisher Scientific Inc., Carlsbad, CA, USA), and inter-assay CVs were <12%.

Four groups of plasma samples (n = 3/group) from CAKUT and non-CAKUT patients, either with or without BP abnormalities, were studied using iTRAQ labeling and liquid chromatography–tandem mass spectrometry (LC-MS/MS). For iTRAQ labeling proteomic technology, pooling small-size samples into groups and evaluating pools rather than individuals are commonly applied to identify potential differential expressed biomarkers between groups (21–23). This ancillary study was performed in the subgroups of 12 and 88 participants, which were recruited from the same cohort study with power calculation to determine the effective sample size. Abnormal BP profile was defined by ABPM. ITRAQ labeling was performed according to previously reported methods (21). In brief, the plasma samples were subjected to high-abundance protein depletion with the Pierce Top 12 Abundant Protein Depletion Spin Columns (Thermo Fisher Scientific Inc., Waltham, MA, USA). Later, the plasma samples of three subjects were evenly pooled, resulting in four samples: CAKUT+HTN, CAKUT+N, non-CAKUT+HTN, and non-CAKUT+N. The four pooled plasma samples were then subjected to sample preparation with the iTRAQ Reagents Multiplex Kit (AB Sciex, Redwood City, CA, USA). After passing the standard QC check, the labeled samples were analyzed with LC/Q-Exactive Orbitrap MS (Thermo Fisher Scientific Inc., Waltham, MA, USA) for 24 h. The raw data were analyzed with Proteome Discoverer v2.4 (Thermo Fisher Scientific Inc., Waltham, MA, USA). Database search was performed using the MASCOT 2.5 database (Matrix Science Inc., Boston, MA, USA) for protein identification. As a result, the relative abundances of detected proteins were obtained. Proteins were considered to be differentially expressed if the difference was statistically significant (p < 0.05) and the fold change was >1.5 or <0.67.

Subjects were instructed to measure office BP in the usual clinical fashion. Children rested for 5 min prior to measurement. Three seated, non-dominant arm BP measurements were recorded at 1-min intervals. The ABPM was conducted using an Oscar II monitoring device (SunTech Medical, Morrisville, NC, USA) over a 24-h period for subjects aged 6–18 years. An experienced specialist nurse performed in-office test measurements to ensure comfort and adequate fitting (8). The device was programmed to measure BP at 20-min intervals from 7 a.m. to 10 p.m. and at 30-min intervals from 10 p.m. to 7 a.m. Waking and sleep periods were defined by self-report diaries, and activities that might influence BP measurements were recorded. An abnormal ABPM profile was determined by (1) daytime BP, nighttime BP, systolic BP (SBP), or diastolic BP (DBP) ≥95th percentile according to ABPM reference data (24); (2) daytime, nighttime, systolic, or DBP load ≥25%; and (3) a sleep decrease in BP load by <10% compared with average awake BP load. The AASI, an indirect arterial stiffness index, was calculated from one minus the slope of the DBP on SBP during ABPM (18). Echocardiographic examination was carried out by pediatric cardiologists using a Philips IE33 system machine (Philips, Bothell, WA, USA). The parasternal long-axis or short-axis views of the left ventricle were obtained by M-mode echocardiography to calculate LV mass. The LVMI was calculated by indexing LV mass to height³ (25). PWV, an index of arterial stiffness, was determined by echo-tracking methods using the ProSound α7 ultrasound (ProSound α7, e-TRACKING system; Aloka Co., Tokyo, Japan) (8).

Continuous variables are presented as medians and inter-quartile ranges (IQRs; 25th−75th percentile), while categorical variables are described as numbers and percentages. The Mann–Whitney U-test or chi-square test was used to test the differences in variables between two groups. Spearman's rank correlation coefficient was used to determine the associations between variables. A linear regression model was performed, followed by stepwise multivariable analyses for integrating relevant parameters to explain apoC-II, apoC-III, and CV risk markers. A p < 0.05 was considered statistically significant. All analyses were performed using Statistical Package for the Social Sciences (SPSS) software 14.0 (Chicago, IL, USA).

The plasma samples from patients with different causes of CKD (CAKUT vs. non-CAKUT) and patients with or without abnormal ABPM profiles were studied using iTRAQ labeling and LC-MS/MS. As a result, 502 proteins were identified with a <1% false discovery rate (FDR) and a minimum of two peptides (the overall iTRAQ data are available in Supplementary Table 1). Next, we analyzed the quantitative differences between the groups of patients. As shown in Table 1, the first set (HTN/N) of differentially expressed proteins consisted of those who had BP abnormalities, regardless of whether their underlying CKD was CAKUT or non-CAKUT, including apoC-II, platelet factor 4, and calmodulin, etc. A total of 20 proteins were identified: one-half was upregulated, and the other was downregulated. The second set (CAKUT-HTN/N) of differentially expressed proteins consisted only of those who differed significantly between CAKUT patients with abnormal and normal BPs. The third set of altered proteins comprised those who differed significantly between abnormal and normal BP in CVD patients with non-CAKUT. The abundance of apoC-II was 1.71-, 1.78-, and 1.65-fold higher in three different sets of patients (Table 1). Although it did not reach significance when comparing BP abnormalities in CAKUT patients, alterations in the abundance of apo-C-III between in the first (HTN/N) and third sets (non-CAKUT-HTN/N) were evident (Table 1). Among the proteins detected by proteomics, we selected apoC-II and apoC-III for subsequent studies.

The characteristics of the study participants are summarized in Table 2. In total, there were 88 children and adolescents with CKD stages G1–G4, including 56 G1 subjects (63.6%), 23 G2 subjects (26.1%), seven G3 subjects (8%), and two G4 subjects (2.3%). The majority of the participants (96.6%) recruited for the present study fell into the mild-to-moderate categories. Our study population had a median age of 13.6 years; 42% of our participants were male and 55.7% had CAKUT. Nearly 36% (32/88) of the CKD children and adolescents were diagnosed with HTN according to their office BP readings. CKD children and adolescents were divided into two groups based on two levels of renal function. The G1 group had an eGFR ≥90ml/min/1.73 m², and the G2–G4 group had an eGFR <90 ml/min/1.73 m². CKD G2–G4 patients were older, were predominantly female, and had higher systolic and diastolic office BPs, BUN, Cr, and uric acid; however, they also had lower eGFR and phosphate than had G1 patients.

As presented in Table 3, plasma apoC-II levels were higher in children with CKD G2–G4 than G1 patients. In the current study, there was no difference in apoC-III levels between the two groups. The results obtained from the present study demonstrated that lipid abnormalities were not common in children with early-stage CKD, with 13.4% (12/88) having LDL levels of >130 mg/dl, 18.2% (16/88) with total cholesterol levels of >200 mg/dl, and 14.8% (13/88) with triglyceride levels of >130 mg/dl. Although lipid abnormalities are associated with lower GFR and are common in children with advanced CKD (17), we observed no difference between the G2–G4 and G1 groups in our lipid profile findings. None of the participants received lipid-lowering medication.

Analyses that evaluated the existence of associations between plasma apoC-II and apoC-III levels and lipids, as evaluated by data pooled from all subjects, indicated that there were strong positive correlations between the plasma apoC-II level and total cholesterol (Figure 1A, r = 0.439, p < 0.001), triglyceride (Figure 1B, r = 0.445, p < 0.001), and LDL (Figure 1C, r = 0.52, p < 0.001). Additionally, apoC-III was positively correlated with total cholesterol (Figure 1D, r = 0.594, p < 0.001), triglyceride (Figure 1E, r = 0.567, p < 0.001), and LDL (Figure 1F, r = 0.591, p < 0.001). The association between plasma eGFR and apoC-II and apoC-III levels was calculated, and no significant correlation was found between eGFR, apoC-II (r = −0.236, p = 0.027), and apoC-III (r = −0.09, p = 0.402).

PWV and AASI are both arterial stiffness parameters used for CV risk assessment (9, 26). Our data showed that PWV and AASI were greater in patients with CKD stages G2–G4 than in G1 patients. LV mass was also higher in the G2–G4 group, while LVMI did not differ between the two groups (Table 4).

The frequency of HTN was higher on the basis of ABPM compared with office BP. Overall, 56.8% (50/88) of participants had at least one BP load abnormality on ABPM, including 12 subjects (13.6%) with 24-h BP or nighttime BP >95th percentile, 14 subjects (15.9%) with daytime BP >95th percentile, 32 subjects (36.4%) with a BP load ≥95th percentile, and 36 patients (40.9%) with a non-dipping nocturnal BP profile (Table 5).

Analysis of plasma apoC-II and apoC-III levels, stratified according to the ABPM profile, showed that there was a difference between the CAKUT and non-CAKUT groups. In Table 5, we compared the difference of apoC-II and apoC-III levels between abnormal and normal ABPM profiles in CAKUT and non-CAKUT patients, separately. It shows that plasma apoC-II level was significantly lower in CAKUT patients with a high BP load, non-night dipping, and an abnormal ABPM profile. Nevertheless, analysis of apoC-III levels in different elements on ABPM profiles did not reveal any significant difference between CKD children with abnormal and normal profiles. However, in patients with non-CAKUT, plasma apoC-II and apoC-III levels were significantly higher in CKD children with abnormal asleep BP than in normal patients. Additionally, in Table 6, we compared these levels of apoC-II and apoC-III between the CAKUT and non-CAKUT patient groups, and we found that apoC-III levels were higher in non-CAKUT patients with abnormal 24-h BP, non-night dipping, and an abnormal ABPM profile than in those with CAKUT.

Then, we performed multivariate linear regression analyses to specify the exact role of apoC-II and apoC-III in individual CV risk (Table 7). A multivariate linear regression model using the stepwise selection was applied for sex, age, CAKUT, and eGFR. In the best predictive model (r = 0.543, p < 0.001), plasma apoC-II level positively correlated with 24-h SBP (p = 0.004). Additionally, an association between 24-h DBP (p = 0.037), awake SBP (p = 0.006), asleep SBP (p = 0.011), awake DBP (p = 0.021), DBP load (p = 0.011), and LVMI (p = 0.033) was found in the adjusted regression model when controlling for age. ApoC-III level showed a positive correlation with 24-h hypertension (p = 0.003), awake HTN (p = 0.039), and asleep HTN (p < 0.001). Although Tables 5, 6 did not adjust CKD stage, Table 7 reveals apoC-II and apoC-III to have a positive correlation with CV risk markers in children with mild-to-moderate CKD, regardless of eGFR. These findings suggest a strong influence of apoC-II and apoC-III on surrogate markers of CV risk.