Association of Serum Paraoxonase/Arylesterase Activity With All-Cause Mortality in Maintenance Hemodialysis Patients

Abstract

Context

In end-stage renal disease (ESRD), serum high-density lipoprotein cholesterol (HDL-C) level is not an accurate predictor of mortality, partly because it does not necessarily correlate with indices of HDL function. Paraoxonase (PON) is a major enzyme constituent of HDL and a key component of HDL antioxidant activity. Apolipoprotein A-I (Apo A-1) is the core HDL structural protein that plays a major role in various aspects of HDL function.

Objective

We sought to examine PON activity and Apo A-I levels in patients with ESRD vs healthy controls.

Design and Setting

PON/arylesterase activity was measured in 499 patients with maintenance hemodialysis (MHD) and 24 healthy controls with similar distributions of age, sex, and race/ethnicity. Serum acrolein-modified Apo A-I was measured in 30 patients with MHD and 10 healthy controls.

Main Outcome Measures

Multilevel Cox models were used to assess associations among PON activity, Apo A-I, and HDL-C levels with 12-month all-cause mortality.

Results

PON activity was significantly lower in patients with MHD vs controls. Furthermore, acrolein-modified Apo A-I levels were higher in patients with MHD vs controls. In fully adjusted models, high PON activity was associated with lower 12-month mortality, whereas no difference of mortality risk was observed across HDL-C levels. The combination of high PON and low Apo A-I compared with low PON and low Apo A-I was associated with lower mortality risk.

Conclusions

In patients with MHD, PON activity had a stronger association with 12-month mortality than HDL-C. Future studies are needed to examine the role of these markers as potential diagnostic and therapeutic tools in ESRD.

End-stage renal disease (ESRD) is associated with a significantly higher risk of all-cause and cardiovascular (CV) mortality (1). Nearly one-half of all deaths in ESRD are attributed to CV disease (2). Although increasing serum concentrations of high-density lipoprotein (HDL) cholesterol (HDL-C) are associated with reduced risk of all-cause and CV mortality in the general population (3, 4), in certain subsets of patients, including those with ESRD being treated with hemodialysis, elevated HDL-C levels are paradoxically associated with worse outcomes (5–7).

It is well-known that HDL can play an important role in prevention of atherosclerosis and CV disease via mechanisms such as reverse cholesterol transport as well as antioxidant, antiapoptotic, vasoprotective, and anti-inflammatory properties (8–11). Apolipoprotein A-I (Apo A-I) is the major apolipoprotein component of HDL and also plays a key role in these functions (12). However, there is accumulating evidence that, under certain conditions, including those associated with inflammation such as ESRD, HDL becomes dysfunctional and loses these protective characteristics (13). Oxidant/chemical modification of Apo A-I or enrichment of HDL with proinflammatory proteins (such as serum amyloid A) can lead to this substantial impairment of HDL function and potentially result in a deleterious proinflammatory HDL, which can be associated with worse outcomes (14–16).

More recently, there has been increasing interest in determining whether measures of HDL function can be better predictors of HDL-related outcomes. However, determining which index of function to measure can be difficult (17–19), given the complexities involved in isolation of the HDL particle and the assays used to evaluate its function (17). Thus, serum measurements that are related to HDL function and that can be easily adopted in the laboratory setting would be of substantial value. Paraoxonase (PON) is a major antioxidant enzyme which is mostly associated with the HDL particle in serum and is thought to play an important role in HDL-mediated antioxidant activity (20, 21). Therefore, increasing serum PON activity can be associated with improved HDL function and several studies have found low PON activity to be a predictor of atherosclerosis and CV disease (22–24). In patients with ESRD, serum PON activity has been consistently shown to be decreased compared with healthy controls (25, 26), especially in those on long-term hemodialysis therapy (27). Furthermore, recent studies have found that low serum and HDL PON function is associated with significantly worse outcomes in patients with advanced (predialysis) chronic kidney disease (28, 29).

In light of these observations, we sought to determine the relationship between serum PON activity alone as well as in consideration of serum HDL-C and Apo A-I levels with ESRD-related mortality. Given the previous studies indicating increased oxidative modification of Apo A-I in patients with MHD, we also determined and compared the concentration of acrolein-modified Apo A-I in a subgroup of ESRD patients and healthy controls (30–32).

Materials and Methods

Study population

In this study, a cohort of healthy controls (n = 24) was recruited from the University of California (UC), Irvine Institute for Clinical and Translational Science. Eligible control subjects were at least 18 years old; not diagnosed with hypertension, diabetes, or other major CV comorbidities; and not taking medications. We then identified a subset of 500 patients with maintenance hemodialysis (MHD) enrolled in the prospective Malnutrition, Diet, and Racial Disparities in Chronic Kidney Disease (MADRAD) study (NCT01415570) between June 2014 and May 2017 with a similar distribution of age, sex, and race/ethnicity to that of the healthy control cohort. In brief, MADRAD is a prospective cohort study that examines the differences in dietary factors and nutritional status across racial/ethnic groups of patients with MHD recruited from large dialysis outpatient (LDO) facilities in the Los Angeles-Orange County, California, areas. A detailed description of the MADRAD study design and enrollment criteria have been previously reported (33). From the MADRAD MHD cohort, one patient was excluded on the basis of not having a serum PON measurement.

Clinical characteristics of patients with MHD

Baseline data on patient demographic and clinical characteristics and medication use were collected at study entry by MADRAD study coordinators. A combination of patient self-reports and International Classification of Diseases-9 codes from LDO records were used to identify the presence of diabetes as a preexisting condition. During regular hemodialysis sessions, body composition surrogates were measured by MADRAD study coordinators. Additional details about the collection of body anthropometric data have been previously described (33). Information on body mass index (BMI) (determined from postdialysis weight) was obtained from LDO electronic records and MADRAD study coordinators. We used the nearest body anthropometric data and BMI values collected within 90 days before or after the measurement of serum PON activity. Given that data on residual kidney function were unavailable, we used patient self-reported urine output and frequency (as measured by a validated questionnaire) closest to PON measurement as an estimate of residual kidney function. We defined dialysis vintage for MHD patients as the time interval between the dates of serum PON activity measurement and first hemodialysis treatment.

Medication ascertainment

Information on medication use was obtained from LDO records and entries by MADRAD study coordinators. Given limited data on medication use for MHD patients in our study cohort, we defined medication ever-use as having received a prescription within 1 year before or after the measurement of serum PON activity.

Serum samples and laboratory tests

Serum samples in MHD patients were collected predialysis during routine hemodialysis sessions, coinciding chronologically with blood tests performed at the LDOs, and frozen at −80°C until analysis. Laboratory tests obtained from patients with MHD at the LDOs were drawn using standardized techniques and measured using automated and standardized methods at a central laboratory in Deland, Florida, typically within 24 hours. Serum from healthy controls was purchased from Innovative Research (Novi, MI) and obtained via the assistance of UC Irvine Institute for Clinical and Translational Science. In all analyses, we used laboratory values measured closest to the time of serum PON activity measurement.

This study was approved by the institutional review committees of the Los Angeles Biomedical Research Institute at Harbor-UC Los Angeles (Torrance, CA) and the UC Irvine Medical Center, Orange, CA (MADRAD study institutional review board protocol #2012-9045).

Laboratory measures

Serum Apo A-I was checked using Apolipoprotein A-I Human SimpleStep ELISA Kit purchased from Abcam (Cambridge, MA). Serum PON activity was measured using an arylesterase/paraoxonase assay kit purchased from ZeptoMetrix, Inc. (Buffalo, NY). Serum concentrations of IL-6 were determined using ELISA assay kits from R&D systems (Minneapolis, MN) and Affymetrix ThermoFisher Scientific. All measurements were performed according to the manufacturer’s specifications and provided protocols.

Acrolein-modified Apo A-I

In additional analyses, we randomly selected 30 nonsmoking patients with MHD and 10 healthy controls from the study cohort. Only nonsmokers were selected (given that smoking can increase serum acrolein levels) via measurement of serum cotinine levels using a cotinine ELISA kit purchased from MyBioSource, Inc. (San Diego, CA) following the manufacturer’s protocol. Serum acrolein-modified Apo A-I adduct was measured with a sandwich ELISA assay using the ELISA kit purchased from Abcam, commercial antibody (GeneTex, Irvine, CA) against acrolein and acrolein-modified Apo A-I as a positive control (obtained from V.N.’s laboratory). Serum (100 μL) was inoculated in duplicate manner in the wells precoated with antihuman Apo A-I and incubated for 2.5 hours at room temperature with gentle shaking. After incubation, the wells were decanted and washed four times. A total of 100 μL acrolein antibody (GeneTex) was added to each well and incubated for 2 hours. After four washes, 100 μL antirabbit IgG horseradish peroxidase preadsorbed (Abcam, Cambridge, MA) was added to each well and incubated for 1 hour. After four washes, we added 3,3′,5,5′-tetramethylbenzidine ELISA substrate and stop solution (Abcam) and performed colorimetric detection at 450 nm in a microplate reader. Measurement was corrected by both negative controls and blanks.

Exposure and outcome ascertainment

The main exposure of interest was serum PON activity, which we categorized into quartiles (<44.9, 44.9 to <76.0, 76.0 to <104.4, and ≥104.4 kU/L). Other exposures of interest included HDL-C (n = 498) and Apo A-I (n = 493), which were divided into quartiles as follows: <29, 29 to <39, 39 to <50, and ≥50 mg/dL, and <79.7, 79.7 to <101.3, 101.3 to <126.3, and ≥126.3 mg/dL, respectively.

We also estimated rank scores of PON, HDL-C ,and Apo A-I separately based on their distribution in our cohort. We defined a low and high threshold for scores as <50th and ≥50th percentiles, respectively. We then created a 2 × 2 matrix of high and low categories for exposure groups based on PON with HDL-C or PON with Apo A-I levels.

The primary end point was 12-month all-cause mortality. Follow-up began at the date of measured serum PON activity to death, transplantation, loss-to-follow-up, end of study period (7 April 2018) or 12-month follow-up, whichever occurred first. Information on mortality and censored events were collected every 6 months by MADRAD study coordinators and reviewed by MADRAD study nephrologists (C.M.R. and K.K.-Z.).

Statistical analysis

Values are reported as mean (± SD) or median (interquartile range) for continuous variables, and as percentages for categorical variables. Parametric and nonparametric tests for trend were used, as appropriate. Baseline patient characteristics of demographics, comorbidities, and laboratory tests were compared between included and excluded patients. Using the Shapiro-Wilk test, we determined that PON activity was nonnormally distributed. We therefore used the nonparametric Wilcoxon-Mann-Whitney test to compare PON activity in MHD patients and healthy controls. Spearman’s rank correlations were used to assess the relationship between PON and laboratory, clinical, and body anthropometric data. Extreme outliers <0.5 or >99.5 percentiles were removed and replaced with lower or upper threshold values, respectively, for laboratory tests, time on hemodialysis, dialysis efficiency, BMI, and body anthropometric measurements.

Cox proportional hazards models were used to assess the association between exposure groups and 12-month all-cause mortality with hierarchical adjustment for covariates in the following three models: (i) model 1, unadjusted; (ii) model 2, adjusted for case-mix variables (age, sex, race, and ethnicity); and (iii) model 3, adjusted for covariates in model 2 plus diabetes and dialysis vintage.

Additionally, we conducted subgroup analyses examining the association of high (≥76) vs low (<76 kU/L, reference) PON and 12-month all-cause mortality in the fully adjusted model. We tested for potential effect modification by sex and race (white or nonwhite and black or nonblack) on the PON-mortality relationship by using the Wald test and including interaction terms for PON-gender and PON-race, respectively, in separate Cox models.

To identify other possible confounders on the PON quartile-mortality association, we also conducted sensitivity analyses of expanded models composed of covariates in model 3 plus either of the following: albumin, BMI, albumin and BMI, hypertension, myocardial infarction, other CV disease, cerebrovascular accident, heart failure, polycystic kidney disease, peripheral vascular disease, and current smoking status.

All patients had complete data on all covariates used in the primary analysis. In sensitivity analyses, we used imputation by mean for missing data (7% and 16% for BMI and albumin, respectively). A two-sided P < 0.05 was considered significant for all analyses in this study. All statistical analyses were conducted with SAS, version 9.4 (SAS Institute Inc., Cary, NC).

Results

Characteristics of patients with MHD

There were 499 patients included in our analytical cohort [comparisons between selected patients and 662 excluded patients are summarized in an online repository (34)]. Baseline characteristics of demographic, clinical, and medication use data for the 499 patients with MHD according to quartiles of PON levels are shown in Table 1. The mean (± SD) age of the cohort was 55 ± 15 years with 44% female, 34% black, and 49% Hispanic patients; 53% had diabetes. Baseline values of laboratory tests for the cohort are reported in Table 2. Patients in the highest PON quartile were more likely to have higher albumin and serum Apo A-I levels vs patients in the lowest PON quartile. Additionally, serum PON activity was significantly lower in patients with MHD than in healthy controls (mean ± SD, 77.2 ± 35.8 kU/L and 135.2±38.7 kU/L, respectively; P < 0.0001; Fig. 1).

Table 1.

Baseline Patient Characteristics of 499 Maintenance Hemodialysis Patients According to Serum PON Activity Quartiles

Variables	Total	Serum PON Activity, kU/L				P for Trend
		<44.9	44.9 to <76.0	76.0 to <104.4	≥104.4
N, %	499	122 (24)	126 (25)	128 (26)	123 (25)
Age, y	55 ± 15	56 ± 15	55 ± 16	54 ± 14	55 ± 13	0.45
Female, %	44	35	47	44	50	0.05
Race, %
Caucasian	56	53	44	63	63	0.01
Black	34	33	39	31	33	0.65
Asian	7	7	12	5	4	0.17
Other	3	7	6	0.78	0	0.0002
Hispanic ethnicity, %	49	47	39	55	55	0.04
BMI, kg/m2	27.8 ± 6.5	27.8 ± 6.8	27.1 ± 6.4	27.6 ± 5.9	28.9 ± 7.0	0.18
Current smoking status, %	11	3	8	17	16	0.0001
Vintage, %
12 mo	11	18	15	6	5	0.0001
12-<36 mo	27	34	29	27	19	0.006
36-<72 mo	29	20	21	33	41	<0.0001
≥72 mo	33	28	35	34	35	0.28
Comorbidities, %
Hypertension	28	25	25	30	32	0.14
Diabetes	53	50	52	55	54	0.41
Dyslipidemia	18	12	19	19	24	0.03
Cardiovascular disease	19	20	17	20	19	0.96
Myocardial infarction	7	7	6	10	7	0.72
Cerebrovascular incident	0.20	0	0	0	0.81	0.18
Heart failure	11	13	11	11	8	0.23
Polycystic kidney disease	0.40	0	0	0.78	0.81	0.20
Peripheral vascular disease	3	2	2	3	7	0.02
Medication use, %
Statin	23	24	27	23	20	0.34
ACEI/ARBs	38	37	34	44	37	0.56
Urine output, %	73	69	74	77	71	0.66
Urination frequency, %						0.66
More than once a day	66	64	66	69	64
Approximately once a day	25	27	26	24	23
Every 2 to 3 d	10	8	9	8	13

Values are reported as mean (±SD) for continuous variables and as percentages for categorical variables. Percentages may not add up to 100 as a result of rounding. Urine output (%) represents percentage of self-reported “yes” responses to the MADRAD study question, “Do you still urinate?”

Abbreviations: ACEI, angiotensin-converting-enzyme inhibitor; ARB, angiotensin II receptor blocker.

Table 2.

Baseline Characteristics of Laboratory Data in 499 Maintenance Hemodialysis Patients According to Serum PON Activity Quartiles

Variables	Total	Serum PON Activity (kU/L)				P for Trend
		<44.9	44.9 to <76.0	76.0 to <104.4	≥104.4
Albumin, g/dL	4.0 ± 0.3	3.9 ± 0.4	4.0 ± 0.3	4.0 ± 0.3	4.0 ± 0.3	0.01
Hemoglobin, g/dL	10.7 ± 1.1	10.7 ± 1.1	10.6 ± 0.9	10.6 ± 1.0	10.8 ± 1.2	0.50
Phosphorus, mg/dL	5.2 ± 1.5	5.1 ± 1.4	5.1 ± 1.6	5.5 ± 1.6	5.1 ± 1.3	0.70
HDL-C, mg/dL	41.3 ± 16.3	39.6 ± 13.2	43.8 ± 16.3	40.0 ± 16.9	41.6 ± 18.2	0.75
IL-6, pg/mL	2 (1, 4)	2 (1, 5)	3 (1, 5)	2 (1, 4)	2 (1, 4)	0.12
Apo A-I, mg/dL	105.7 ± 35.3	92.7 ± 27.9	110.5 ± 33.1	106.6 ± 39.0	112.9 ± 36.8	<0.0001

Values are reported as mean (±SD) or median (interquartile range) for continuous variables, where appropriate.

Figure 1.

Distribution of serum PON activity in controls and patients with MHD. Serum PON activity in healthy control subjects (n = 24) and patients with MHD (n = 499) are shown. P < 0.0001.

Spearman correlation coefficients between serum PON activity and laboratory data are presented in Table 3. Serum PON positively correlated with levels of albumin (rho = 0.13, P = 0.009), Apo A-I (rho = 0.20, P < 0.0001), and BMI (rho = 0.11, P = 0.02), but negatively correlated with IL-6 (rho = −0.13, P = 0.01) after adjustment for case-mix, diabetes, and dialysis vintage covariates.

Table 3.

Correlations of Serum PON Activity With Laboratory, Clinical, and Body Anthropometric Data in 499 Maintenance Hemodialysis Patients

Variables	Unadjusted		Adjusted
	Rho	P Value	Rho	P Value
Laboratory tests
Albumin, g/dL	0.12	0.01	0.13	0.009
Hemoglobin, g/dL	0.02	0.68	0.03	0.60
Phosphorus, mg/dL	0.04	0.46	0.01	0.76
HDL-C, mg/dL	−0.02	0.61	−0.03	0.45
IL-6, pg/mL	−0.11	0.05	−0.13	0.01
Apo A-I, mg/dL	0.20	<0.0001	0.20	<0.0001
Time on hemodialysis, min	−0.005	0.91	0.03	0.52
Kt/V	0.03	0.60	−0.07	0.15
BMI, kg/m2	0.10	0.03	0.11	0.02
Body anthropometric measurements
Biceps average, mm	0.12	0.03	0.09	0.11
Triceps average, mm	0.14	0.009	0.11	0.05
Midarm muscle circumference, cm	−0.04	0.44	0.04	0.46
Midarm circumference, cm	0.07	0.23	0.12	0.03
Near-IR body fat, %	0.11	0.04	0.06	0.28

Spearman correlation coefficients (Rho) shown for all variables. Correlation coefficients were adjusted for covariates in the following models: (i) unadjusted; and (ii) adjusted (age, sex, race, ethnicity, diabetes, and dialysis vintage).

Abbreviations: IR, infrared; Kt/V, dialysis efficiency.

Associations of PON activity, serum HDL-C, and Apo A-I with 12-month all-cause mortality

Among 499 patients, there were 61 deaths during a total 12-month follow-up of 459 patient-years, and the incidence rate of 12-month mortality was 13.3 (95% CI, 10.0 to 16.6) per 100 person-years.

Patients in the highest serum PON quartile (≥104.4 kU/L) had the lowest risk of 12-month all-cause mortality compared with patients in the first quartile (<44.9 kU/L, reference) across all levels of adjustment, with an adjusted hazard ratio of 0.26 (95% CI, 0.11 to 0.59) in model 3 [Fig. 2A (34)].

Figure 2.

Twelve-mo all-cause mortality and PON, HDL-C, and Apo A-I quartiles. Associations among (A) PON quartiles, (B) HDL-C quartiles, and (C) serum Apo A-I quartiles and 12-mo all-cause mortality with adjustments for covariates in the following models: (i) model 1 (unadjusted); (ii) model 2 (case-mix variables: age, sex, race, and ethnicity); and (iii) model 3 (model 2 covariates plus diabetes and dialysis vintage). For visual purposes, the plots have log-transformed y-axes. HR, hazard ratio.

In analyses evaluating the association between HDL-C quartiles and 12-month all-cause mortality, no important association was observed [Fig. 2B (34)].

The lowest serum Apo A-I quartile (<79.7 mg/dL) was associated with 3.3-, 4.3-, and 4.2-fold higher risks of 12-month all-cause mortality in models 1, 2, and 3, respectively, when compared with the third quartile (101.3 to <126.3 mg/dL, reference) [Fig. 2C (34)]. [Results of the Apo A-I mortality associations using quartile 1 (<79.7 mg/dL) as the reference category are presented elsewhere (34)].

In analyses using rank scores, compared with a reference of low PON/low HDL-C, we observed that high PON/low HDL-C was significantly associated with lower 12-month all-cause mortality risk across all adjustment models [adjusted hazard ratios; 95% CI, 0.26 (0.09 to 0.80), and 0.24 (0.08 to 0.74) for models 2 and 3, respectively; Fig. 3B and 3C (34)]. However, there were no differences in mortality risk for either high PON/high HDL-C or low PON/high HDL-C vs low PON/low HDL-C. Compared with low PON/low Apo A-I, all other PON with Apo A-I levels rank score combinations had lower 12-month all-cause mortality risk in all adjustment models [Fig. 3D and 3F (34)].

Figure 3.

Twelve-mo all-cause mortality with rank scores of PON and HDL-C, and PON and Apo A-I. Associations between (A-C) low or high PON and low or high HDL-C rank scores and (D-F) low or high PON and low or high Apo A-I with 12-mo all-cause mortality with adjustment for covariates in the following models: (A, D) model 1 (unadjusted); (B, E) model 2 (case-mix variables: age, sex, race and ethnicity); and (C, F) model 3 (model 2 covariates plus diabetes and hemodialysis vintage). *P < 0.05. Ref., reference.

In subgroup analyses, we analyzed the association between high PON (≥76 vs <76 kU/L) and 12-month mortality risk according to sex (male, female), and race (white, nonwhite, black, and nonblack). A significantly lower mortality risk was observed for high PON across all subgroups, except in black patients, who also had a lower mortality risk but did not reach statistical significance (34). Furthermore, we did not find a significant interaction between sex and PON (P_interaction = 0.68) or between race and PON (P_interaction = 0.36 for black and 0.63 for white) for the association with 12-month all-cause mortality.

As sensitivity analyses, adjustment for additional covariates in expanded models did not change the association between higher PON and lower 12-month mortality risk (34).

Serum concentration of acrolein-modified Apo A-I

Of note, in a subset of 10 healthy controls and 30 patients with MHD and negative serum cotinine tests, we found that the MHD patients had significantly higher (+60%) levels of acrolein-modified Apo A-I adduct compared with healthy controls (P = 0.047) (34).

Discussion

ESRD is associated with abnormal HDL metabolism and function, with evidence indicating that in subgroups of patients HDL can become proinflammatory in nature (35–37). Furthermore, it has been postulated that this HDL dysfunction may underlie the paradoxical lack of relationship observed between higher concentrations of serum HDL-C levels and mortality in patients with MHD (5). In this study, we found that although HDL-C concentration was not significantly associated with 12-month all-cause mortality risk, higher PON was associated with lower mortality. Compared with patients with low PON/low Apo A-I, all other combinations of PON and Apo A-I in the 2 × 2 analysis had a lower risk of death. Similarly, patients with high PON/low HDL-C also had a significantly lower risk of death vs the reference group of patients with low PON/low HDL-C. Last, we did not find an important correlation between serum PON activity and HDL-C concentrations, but did find a modest correlation between serum PON activity and serum Apo A-I levels. The latter findings highlight the important point that serum HDL-C levels are not a good marker for indices of HDL function.

Although decreased serum and HDL activity of PON has been extensively studied and reported in MHD patients, our study examines the role of serum PON activity in ESRD-related mortality (28, 38, 39). We did not find serum HDL-C concentrations to be associated with any change in outcomes in this small cohort, whereas previously we had found a U-shaped relationship between HDL-C and mortality (5). This is most likely from the small number of patients in this investigation compared with previous studies. In our study, however, high PON combined with either low or high Apo A-I or low HDL-C was associated with lower mortality risk, possibly indicating the potential for a strong role of PON in predicting mortality outcomes in ESRD patients.

A previous study reported that increased oxidized Apo A-I levels in hemodialysis patients is associated with worse CV outcomes (16). We also found that acrolein-modified/content of Apo A-I is significantly increased in patients with ESRD on MHD when compared with healthy controls. This is consistent with previous reports which demonstrated that serum protein acrolein adducts are increased in ESRD patients (40). Increased acrolein content/modification of Apo A-I impairs ATP-binding cassette transporter A1 mediated cholesterol efflux and HDL function (41, 42). It is also important to note that free serum acrolein levels have been found to be inversely correlated with serum PON activity in ESRD patients (43), and increased acrolein-modified Apo A-I in MHD patients may be contributing to reduced PON activity and impaired HDL antioxidant activity in this patient population. Therefore, our findings provide another potential mechanism by which Apo A-I may become dysfunctional in ESRD setting. Further studies investigating the association of acrolein-modified Apo A-I with mortality in ESRD patients are needed.

Several limitations need to be mentioned. First, this is an observational study, and future mechanistic studies are needed to confirm our hypotheses. In addition, the relatively small sample size of our cohort limits our ability to adjust for all possible confounders. Therefore, large validation studies are needed to further elucidate the relationship between serum PON activity, HDL-C, Apo A-I levels, and outcomes in ESRD. Moreover, in this study, we did not have access to genetic information and hence could not assess PON genetic variants that may play a role in the activity of this enzyme and its association with outcomes. We were also unable to consider high-sensitivity C-reactive protein as a covariate in the adjusted models because this inflammatory marker was not measured regularly in MHD patients. However, additional adjustment for albumin as a marker of nutritional status and inflammation did not alter observed associations. Future investigations will be needed to evaluate the role of acrolein-modified Apo A-I in reduced PON activity and the association of increased Apo A-I levels and mortality.

In conclusion, we found a stronger relationship between serum PON activity as measured by its arylesterase function with 12-month all-cause mortality, compared with exposure measurements of serum HDL-C and Apo A-I. Furthermore, lower serum Apo A-I levels were associated with a significantly higher risk of death at 12 months and acrolein content/modified Apo A-I was increased in patients on MHD. Further studies are needed to determine the potential utility of these markers as diagnostic and therapeutic tools in the ESRD population.

Glossary

Abbreviations:

Apo A-I

apolipoprotein A-I

BMI

body mass index

CV

cardiovascular

ESRD

end-stage renal disease

HDL

high-density lipoprotein

HDL-C

high-density lipoprotein cholesterol

LDO

large dialysis outpatient

MADRAD

Malnutrition, Diet, and Racial Disparities in Chronic Kidney Disease

MHD

maintenance hemodialysis

PON

paraoxonase

UC

University of California