Abstract
Background and objectives In dialysis patients, the associations between apoprotein profile and all-cause or cardiovascular disease (CVD)-related mortality are not well known. We, therefore, investigated whether apoprotein levels are associated with these events.
Design, setting, participants, & measurements We undertook a prospective observational cohort study of prevalent hemodialysis patients aged ≥18 years (n=1081), who were followed for 4 years (2011–2014). Outcomes were all-cause and CVD-related mortality. Predictors used were baseline apoprotein levels, particularly the apoprotein B (apo B)/ apoprotein A-1 (apo A-1) ratio. A Cox regression analysis was used to calculate the hazard ratios (HRs) for mortality. Apo A-1, apo B, and apo B/ apo A-1 ratio were analyzed with adjustments in three models: model 1, basic adjustment for age and sex; model 2, basic adjustments plus dialysis conditions (dialysis vintage, mean predialysis systolic blood pressure, dry weight, and mean intradialytic weight gain); and model 3, model 2 plus metabolic and inflammatory conditions (basal kidney disease, serum albumin, C-reactive protein level, and statin use).
Results Of the 1081 patients included in the study, 203 deaths were recorded, 92 of which were related to CVD. The apo B/ apo A-1 ratio was significantly associated with all-cause and CVD-related mortality when analyzed by 1-SD increments or quartile IV versus I in all models. In model 3, HRs and 95% confidence intervals (95% CIs) for 1-SD increments of apo B/ apo A-1 ratio for all-cause mortality or CVD-related mortality were: HR, 1.16 (95% CI, 1.00 to 1.35), or HR, 1.38 (95% CI, 1.11 to 1.71), respectively, and for quartile IV versus I: HR, 1.65 (95% CI, 1.05 to 2.57), or HR, 2.56 (95% CI, 1.21 to 5.40), respectively. Apo A-1 was significantly associated with both mortalities in models 1 and 2. However, apo B was only significantly associated with CVD-related mortality in model 3.
Conclusions Apoprotein measurement, especially the apo B/ apo A-1 ratio, was significantly associated with all-cause and CVD-related mortality in prevalent dialysis patients.
- Apoprotein
- hemodialysis
- cardiovascular disease
- mortality
- blood pressure
- C-Reactive Protein
- Humans
- Kidney Diseases
- Prospective Studies
- Regression Analysis
Introduction
Lipid metabolism is reportedly different in dialysis patients than in patients without CKD. For example, within the extrinsic lipid metabolism pathway, uremia in dialysis patients suppresses lipoprotein lipase activity and hepatic chylomicron (CM) uptake, resulting in increased triglyceride-rich CM or CM remnants. Within the intrinsic pathway, uremia also suppresses lipoprotein lipase and hepatic lipase activity, resulting in relatively low levels of LDL and HDL (1,2).
Only a few studies have assessed the associations between lipid indexes and dialysis patient survival. Both low and oxidized HDL levels have been associated with cardiovascular disease (CVD)-related events (3,4); however, HDL is not always associated with CVD-related events or mortality in dialysis patients (5). Higher LDL levels have been associated with myocardial infarction (MI) and cerebral infarction (6). However, higher LDL or higher total cholesterol (TC) has also been associated with better survival (5,6). These paradoxical associations were sometimes explained by the coexistence of malnutrition and inflammation in dialysis patients (7). In some patients with lower TC or lower LDL, the promotion of atherosclerosis has been reported (8). Recently, Shoji et al. reported that both higher non-HDL and LDL were associated with atherosclerotic events; they also showed that non-HDL better predicted the incidence of MI (6). However, a cross-sectional study revealed that higher apo B, an essential structural component of VLDL, intermediate-density lipoprotein, and LDL, was associated with coronary artery disease prevalence (3). Among the major HDL constituents, lower apoprotein A-2 (apo A-2), but not apo A-1, was significantly associated with both CVD-related events and all-cause mortality (9).
Only a few studies have assessed the contribution of apoproteins to CVD-related events or all-cause mortality in dialysis patients (9,10). Therefore, in this study we investigated apoprotein profiles in addition to conventional lipid indexes in dialysis patients. We were particularly interested in the apo B/apo A-1 ratio, which represents the ratio of atherogenic and anti-atherogenic apoproteins, and has been reported to be associated with CVD-related events and mortality in non-CKD patients, (11–14) but not in dialysis patients.
Materials and Methods
Study participants
We enrolled 1379 prevalent chronic hemodialysis patients maintained on outpatient dialysis at 27 dialysis centers on December 31, 2010. Exclusion criteria included patients with <3 months’ hemodialysis vintage, patients <18 years old, pregnant women, hospitalized patients, and patients not wishing to participate; 287 patients were excluded for missing TC or HDL data. One patient whose triglyceride (TG) level was >400 mg/dL was also excluded because LDL was calculated using the Friedewald formula. Overall, 1081 patients (median age, 69 years; men, 57.7%; median dialysis vintage, 81 months; diabetes, 23.1%) were followed for 48 months. During the study period, 53 patients (4.9%) were transferred to other clinics or lost to follow-up.
Data collection
Clinical data, including age, sex, hemodialysis vintage, basal kidney disease, predialysis blood pressure, interdialytic body weight gain, dry weight, statin use, cause of death, and transfer to other clinics or hospitals, were collected from questionnaires recorded by attending physicians. Mean predialysis blood pressure and mean intradialytic weight gain were calculated from data from three consecutive dialysis sessions. The mean rate of intradialytic weight gain was calculated as the mean intradialytic weight gain divided by dry weight. Questionnaires were sent to physicians each January between January 2011 and January 2014, and returned the following January. All data were collected and stored in the Dialysis Division of the University of Miyazaki Hospital.
Cause of death was reported by attending physicians and confirmed with medical records, which were searched by Y.S. and T.T. if necessary. CVD-related mortality was defined as death from ischemic or hemorrhagic stroke, acute MI, causes related to congestive heart failure, sudden death, or aortic aneurysm rupture. Stroke was diagnosed using typical imaging and physical findings from examinations. Acute MI was diagnosed using typical electrocardiogram findings or elevation in myocardium-derived enzymes. Cardiac disease was determined based on a history of ischemic heart disease and/or congestive heart failure. Ischemic heart disease was defined as prior hospitalization or medication for angina pectoris and/or MI. Congestive heart failure was confirmed using electrocardiography, chest radiography or echocardiography together with symptoms of dyspnea or edema. Sudden death was judged as unexpected death occurring in the first hour following the start of symptoms or when the patient was found dead and had been seen alive 24 hours earlier.
This study was conducted in accordance with the principles contained in the Declaration of Helsinki and was approved by the University of Miyazaki Research Ethics Committee (No. 516). This was a non-invasive observational study and all data were anonymized. Verbal informed consent was obtained and recorded on patient medical charts by attending physicians, and a poster announcing the study and stating that all participants had the right to reject participation at the beginning of or half-way through the study was placed in a conspicuous location at each dialysis clinic or center.
Measurements
All serum was drawn before the first dialysis session (Monday or Tuesday) and sent to SRL Inc. (Tokyo, Japan) to assess TC, HDL, TG, apo A-1, and apo B. LDL was calculated using the Friedewald formula; accordingly, patients with TG >400 mg/dL were excluded from the study. Apo A-1 and apo B were measured using immunonephelometry. Non-HDL was calculated by subtracting HDL from TC. Patients were not asked to fast, as there were several dialysis sessions, including morning, afternoon, and evening. Serum albumin and C-reactive protein (CRP; not high-sensitivity) were measured at the same time of measuring lipid profiles. Blood pressure was measured by trained staff with patients in the supine position after at least 5 minutes rest. Body weight was measured before and after each hemodialysis session.
Statistical Analyses
Data are expressed as mean±SD, except for age, CRP level, dialysis vintage, apo A-1, apo B, and the apo B/apo A-1 ratio, which are expressed as median and 25th–75th percentile. Clinical parameters according to classification of apo B/apo A-1 ratio quartiles were compared using a one-way analysis of variance or the Kruskal–Wallis test, as appropriate. Categorical parameters were compared using the chi-squared test. To examine the correlation between lipid parameters, the Spearman rank correlation test was used.
Survival curves according to apoprotein quartile were analyzed using the Kaplan–Meier method, and statistical significance was evaluated with the log-rank test.
To compare which parameters of apoprotein profiles or conventional lipid indexes predicted future mortality in dialysis patients, we used multivariate Cox regression analyses to examine the independent associations between lipid parameter levels and mortality.
First, continuous variables of lipid parameters were analyzed by 1-SD increments. Second, each lipid parameter was divided into quartiles, and the statistical significance of the highest quartile (Q4) was compared with the lowest quartile (Q1), which was used as the reference. Using a multivariate Cox regression analysis, hazard ratios (HRs) and 95% confidence intervals (95% CIs) for all-cause and CVD-related mortality were determined independently after simultaneous adjustment in three models. Model 1 adjusted for basic confounders (age and sex). Model 2 included adjustments for the basic confounders plus adjustments for dialysis conditions (hemodialysis vintage, mean predialysis systolic blood pressure, dry weight, and mean intradialytic weight gain). Model 3 included model 2 confounders plus metabolic and inflammatory parameters (serum albumin, CRP, basal kidney disease, and statin use). A P value <0.05 was considered statistically significant.
Analyses were performed using SPSS version 20.0J software (IBM Corp., Armonk, NY).
Results
Participant Characteristics
Patients were divided into quartiles based on their apo B/apo A-1 ratio (Table 1). Age, serum albumin, CRP, and lipid profiles were significantly different. Older age, lower serum albumin levels, and higher CRP levels were prominent in the higher apo B/apo A-1 ratio quartiles. With regard to basal kidney diseases, diabetes mellitus was prominent in higher quartiles, whereas chronic glomerulonephritis was prominent in lower quartiles. Hypertensive nephrosclerosis did not show a consistent tendency in any quartile. Overall, basal kidney diseases were not statistically significant between the quartiles using the chi-squared test. With regard to lipid profiles, higher TC, TG, LDL, non-HDL, and apo B were more prominent in the higher quartiles, whereas higher HDL and apo A-1 were more prominent in the lower quartiles.
Population characteristics according to apo B/apo A-1 ratio quartiles
Lipid Parameter Correlations
Apo A-1 was significantly correlated with the apo B/apo A-1 ratio and HDL (r = −0.59 and 0.86, respectively), and apo B was significantly correlated with the apo B/apo A-1 ratio, LDL, non-HDL, and the LDL/HDL ratio (r = 0.74, 0.86, 0.93, and 0.64, respectively). We also found that CRP negatively correlated with apo A-1 and HDL (r = −0.32 and −0.32, respectively) but not with apo B (r = 0.04; Supplemental Table 1).
Cause of Death
During the 4-year follow-up, 203 deaths were recorded (Table 1); 92 of which were related to CVD. The most common cause of CVD-related mortality was congestive heart failure (n=49), followed by hemorrhagic stroke (n=13), sudden death (n=10), acute MI (n=9), and ischemic stroke (n=7). Other causes included infection (n=44) and malignancy (n=21). For infection-related mortality, pneumonia (n=25) and sepsis (n=17) were the main causes. The remaining deaths (n=46) were caused by unknown or other causes.
Survival Analyses
In the survival analyses, apo A-1, apo B, and the apo B/apo A-1 ratio were significantly related to all-cause and CVD-related mortality. Estimated survival curves by apo A-1 quartiles for all-cause and CVD-related mortality were significant (P<0.001 and P=0.001, respectively), but those for apo B were not (P=0.74 and P=0.27, respectively; Supplemental Figure 1). Survival curves by apo B/apo A-1 ratio quartiles for all-cause and CVD-related mortality were significant (P=0.001 and P=0.02, respectively; Figure 1).
All-cause and CVD-related mortality-free survival curves according to apo B/apo A-1 ratio quartiles. (A) All-cause mortality-free survival curves according to apo B/apo A-1 ratio quartiles (P=0.001). (B) CVD-related mortality-free survival curves according to apo B/apo A-1 ratio quartiles. (P=0.02). Apo A-1, apoprotein A-1; apo B, apoprotein B; CVD, cardiovascular disease.
Cox Analysis by Lipid Index
In a multivariate Cox analysis, the apo B/apo A-1 ratio was associated with all-cause and CVD-related mortality, even after adjustment in models 1–3 (Tables 2 and 3). In model 3, HRs and 95% CIs for 1-SD increments of the apo B/apo A-1 ratio for all-cause mortality and CVD-related mortality were: HR, 1.16 (95% CI, 1.00 to 1.35), and HR, 1.38 (95% CI, 1.11 to 1.71), respectively, and those for quartile IV versus quartile I were: HR, 1.65 (95% CI, 1.05 to 2.57), and HR, 2.56 (95% CI, 1.21 to 5.40), respectively. Apo A-1 was also associated with both all-cause and CVD-related mortality in models 1 and 2, but not in model 3. Apo B was only associated with CVD-related mortality in model 3.
The associations of apo A-1, apo B, and the apo B/apo A-1 ratio (per 1-SD increase) with all-cause or CVD-related mortality
Association of apo A-1, apo B, and the apo B/apo A-1 ratio (quartile IV versus quartile I) with all-cause or CVD-related mortality
With regard to conventional lipid parameters (Supplemental Tables 2 and 3), the LDL/HDL ratio showed a similar pattern to the apo B/apo A-1 ratio. However, the quartile analysis did not demonstrate statistical significance for all-cause mortality. HDL and apo A-1 were associated with all-cause mortality in models 1 and 2. Neither LDL nor non-HDL were associated with all-cause and CVD-related mortality.
Discussion
The main finding of our study is that the apo B/apo A-1 ratio is strongly associated with both all-cause and CVD-related mortality in prevalent chronic dialysis patients. This finding is consistent with results from non-CKD patients (11–14); however, our study is the first to investigate dialysis patients.
Apo B is a fundamental structural constituent of VLDL, intermediate-density lipoprotein, and LDL. As each particle of these lipoproteins contains one apo B molecule, measuring apo B can accurately estimate the number of these atherogenic particles (15). In contrast, the number of molecules of apo A-1, a major structural HDL constituent, varies among lipoprotein particles. However, our study, in agreement with a recent study (16), showed that apo A-1 levels correlate strongly with HDL levels. Accordingly, the apo B/apo A-1 ratio is a representative marker of the atherogenic/antiatherogenic molecule ratio, and the LDL/HDL ratio is thought to convey similar information. In this study, however, the LDL/HDL ratio was only associated with CVD-related mortality and not with all-cause mortality. Why the apo B/apo A-1 ratio was associated with all-cause mortality in addition to CVD-related mortality is not clear. We speculate that an advanced atherosclerotic state might affect not only CVD-related mortality, but also all-cause mortality.
The interaction between apo A-1 and apo B is an interesting issue. The apo A-1 versus apo B correlation was quite low in our dialysis patients. A direct interaction between apo A-1 and apo B is therefore unlikely. Apo A-1 was intermediately correlated with CRP and serum albumin, but apo B was not. It is possible that inflammation affected apo A-1 metabolism to some degree, as this phenomenon has been reported in the Chinese general population (1).
When the apo B/apo A-1 ratio was divided into quartiles, there was not a clear linear relationship between HRs for all-cause or CVD-related mortality (data not shown). We were unable to clarify why, but sex, mean predialysis systolic blood pressure, and basal kidney disease were not significantly different between quartiles. Some unknown factor may have affected the results.
Apo A-1 was significantly negatively associated with all-cause and CVD-related mortality. In the German Diabetes Dialysis study, apo A-1 did not significantly predict CVD-related events or all-cause mortality (9). However, there are some differences between that study and ours, including basal kidney disease. In our cohort, <30% had diabetes mellitus, whereas all of the German Diabetes Dialysis study cohort had diabetes mellitus. Furthermore, the mean dialysis vintage was quite different. In our cohort, dialysis vintage was around 100 months, whereas that in the German Diabetes Dialysis study was <10 months. These differences might account for the different results.
Apo B was significantly positively associated with CVD-related mortality in the fully adjusted model (model 3). That apo B values are representative of the number of atherogenic particles (15) is consistent with the present data.
Consistent with other studies (5), the HDL level was not a strong predictor for either all-cause or CVD-related mortality. Dysfunctional HDL has been reported in dialysis patients (17,18), perhaps explaining why HDL level per se was not considered a significant factor for mortality in dialysis patients.
Most studies from Western countries failed to demonstrate the significance of lipid parameters for mortality or heart attacks (5,9,19); however, some papers from Japan, including our cohort, did (6). We believe that the most prominent factors contributing to this difference are that the dialysis vintage is much shorter in Western countries than in Japan, diabetes prevalence is much higher in Western countries, and both regular and high-sensitive CRP levels are much higher in Western countries (5,6,9,19–21). The significance of lipid indexes for mortality or CVD events has been reported to be different based on the presence or absence of inflammation (7). In patients with higher CRP levels, higher lipid parameters reflect better survival; however, in patients with lower CRP levels, lower lipid parameters reflect better survival. Therefore, we speculate that inflammation status is key for understanding how lipids affect CVD-related as well as all-cause death. However, the mechanism for how lower lipid levels affect mortality in individuals with a higher inflammation status remains to be clarified.
Finally, previous studies have failed to demonstrate the usefulness of statins for preventing CVD-related events or mortality in dialysis patients (19,22); however, LDL was used as the marker in these studies. Therefore, an apoprotein-targeting study is needed to clarify whether lipid-lowering therapy brings significant benefits to dialysis patients.
Our study has some limitations. Our cohort consisted of prevalent but not incident chronic dialysis patients. Therefore, some patients with severe dyslipidemia might have died before study enrollment. Moreover, a major cause of CVD-related mortality is congestive heart failure, but we were unable to identify the specific underlying etiologies. Thus, there might have been some valvular heart disease in addition to ischemic heart disease; it is unclear how dyslipidemia might affect valvular heart diseases. Kt/V data were not obtained, so we could not adjust for them on the Cox analysis. Finally, we did not evaluate HDL function.
In summary, although controversies persist regarding whether lipids affect the prognosis of dialysis patients, our data revealed that the apoprotein profile, especially the apo B/apo A-1 ratio, was significantly associated not only with CVD-related mortality but also with all-cause mortality in prevalent dialysis patients.
Disclosures
None.
Acknowledgments
This work was supported by a grant for Clinical Research from University of Miyazaki Hospital.
Research idea and study design: Y.S. and S.F.; data collection: Y.S., T.T., H.N., Y.Y., T.I., A.F., and S.I.; data analysis: Y.S. and T.T.; supervision: S.F.
The authors appreciate the help of the following attending physicians for their participation in the study by collecting data and providing useful suggestions: Y. Yamamoto (Yamamoto Naika Clinic, Miyazaki, Japan), O. Wakisaka (Wakisaka HD Clinic, Miyazaki, Japan), T. Tanaka (Tanaka Takashi Naika, Mimata, Japan), H. Ebihara (Ebihara Naika, Miakonojo, Japan), M. Kuroki (Kuroki Naika Clinic, Takanabe, Japan), M. Yamashita (Yamashita Clinic, Kunitomi, Japan), J. Miyata (Miyata Naika Clinic, Nobeoka, Japan), K. Aso (Aso Naika, Miyazaki, Japan), H. Ochiai (Ochiai Naika, Miyazaki, Japan), S. Hisanaga (Koga General Hospital, Miyazaki, Japan), S. Morita (Morita Naika Clinic, Miyazaki, Japan), F. Iemura (Iemura Naika, Hyuga, Japan), T. Uchida (Uchida Clinic, Takanabe, Japan), N. Yokota (Yokota Naika, Miyazaki, Japan), F. Sawano (Sawano Naika Clinic, Miyazaki, Japan), M. Kawamura (Kawamura Naika Clinic, Sukumo, Japan), H. Washimine (Matsuo Naika Clinic, Sasebo, Japan), T. Ishihara (Fujimoto Central Hospital, Miyazaki, Japan), N. Ueno (Ueno Clinic, Saito, Japan), H. Kinoshita (Kinoshita Naika Clinic, Miyazaki, Japan), F. Matsuoka (Nobeoka Clinic, Nobeoka, Japan), K. Yamada (Miyazaki Konan Hospital, Miyazaki, Japan), K. Fukudome (Koga General Hospital, Miyazaki, Japan), H. Inagaki (Chiyoda Hospital, Hyuga, Japan), K. Hidaka (Ebihara Clinic, Miyazaki, Japan), M. Kuboyama (Morino Clinic, Miyazaki, Japan), A. Baba (Koga General Hospital, Miyazaki, Japan), S. Sonoda (Nojiri Central Hospital, Kobayashi, Japan), R. Nishizono (Chiyoda Hospital, Hyuga, Japan), and F. Ebihara (Ebihara Clinic, Miyazaki, Japan).
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
This article contains supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.09830915/-/DCSupplemental.
- Received September 16, 2015.
- Accepted January 13, 2016.
- Copyright © 2016 by the American Society of Nephrology
References
In this issue
Jump to section
More in this TOC Section
Original Articles
Epidemiology and Outcomes
Cited By...
- No citing articles found.