Background
Chronic kidney disease (CKD), an independent risk factor for cardiovascular (CV) disease and CV mortality, is associated with dyslipidemia [
1‐
4]. Among other derangements, the circulating lipid profile seen in CKD includes elevated levels of lipoprotein(a) [Lp(a)], a modified low-density lipoprotein particle covalently linked to the highly polymorphic apolipoprotein(a) [apo(a)], a glycoprotein of genetically-variable protein length that varies widely among individuals [
5‐
8]. Dyslipidemia in CKD is also characterized by elevated levels of apolipoprotein C-III (apoC-III), an exchangeable apolipoprotein on lipoprotein particles that inhibits lipoprotein lipase, thereby reducing clearance of very-low-density-lipoproteins (VLDL) and the triglycerides (TGs) they carry [
2,
9‐
18]. Both Lp(a) and apoC-III levels have recently been identified as genetic causal risk factors for CV disease, [
5‐
7,
19‐
24] but their role in CKD development and progression remain unclear.
The purpose of this study was to probe whether baseline circulating Lp(a) and apoC-III levels are associated with the development of renal impairment in type 2 diabetes mellitus (T2DM) patients, a population already at increased risk for adverse CV and renal outcomes [
2,
9‐
18]. Although aberrant concentrations of other lipid fractions are associated with CKD, we chose to focus on Lp(a) and apoC-III due to their already proven causality in CV disease. Using an observational cohort of T2DM patients without baseline clinical CV disease or CKD, we examined the relationship between estimated glomerular filtration rate (eGFR) decline and baseline concentrations of these lipoproteins. We hypothesized that higher baseline plasma Lp(a) and apoC-III levels are each associated with eGFR decline.
Discussion
In this study of individuals with T2DM, we demonstrate that baseline plasma Lp(a) levels strongly associate with eGFR decline and that a significant although more modest association exists between baseline plasma apoC-III levels and eGFR decrease over time in a TG-dependent fashion. To our knowledge, this study is among the first to demonstrate an independent association between baseline circulating Lp(a) levels and eGFR decline in a population without pre-existing CKD.
Several prior studies have shown that plasma Lp(a) levels are elevated in CKD patients, [
7,
34‐
36] but the direction of causality in this trend has been unclear. Past studies have explored whether Lp(a) elevations in CKD are due to reduced renal clearance, pointing to differences in Lp(a) concentrations in the renal arteriovenous circulation [
37] as well as reduction in Lp(a) levels after renal transplant [
38]. Although the kidney may play a role in Lp(a) metabolism, the current study introduces the possibility that circulating Lp(a) may also play a role in renal impairment. Here we show that in diabetics without baseline CKD, higher baseline Lp(a) concentrations, including above the clinical used atherogenic cutpoint of 30 mg/dL, are strongly associated with decline in eGFR. The eGFR slope for subjects Lp(a) levels above that cutpoint is more than twice as steep as the slope for subjects in the lower Lp(a) group, indicating that higher baseline Lp(a) levels may accelerate the rate of decline in eGFR that would otherwise take place due to age and the natural course of diabetic disease. Because circulating Lp(a) levels do not vary much over time within healthy individuals, [
39] our results suggest that exposure to higher baseline Lp(a) concentrations that would remain unchanged over time may be a risk factor for eGFR decline. Within the PDHS cohort, the strong dose-dependent, temporal relationship between baseline plasma Lp(a) concentration and negative eGFR slope is consistent with a potential causal relationship between this lipoprotein risk factor and renal outcome. However, as has been performed for CVD outcomes, [
5,
6,
20,
40,
41] genetic data and Mendelian randomization studies will be important to definitively address Lp(a) causality in renal disease in humans.
Our findings add valuable insight to the conflicting results of prior studies examining the relationship between Lp(a) and GFR. In 2005, Song
et al. found in a small cohort of 81 individuals with proteinuric diabetic nephropathy that each 10 mg/dL increase in log-transformed Lp(a) concentration was associated with a 1.4 increased odds of serum creatinine doubling independent of albuminuria, hypertension, and glycemic control [
42]. Our results complement Song
et al.’s finding as we further demonstrate an independent inverse relationship between Lp(a) concentration and decline renal function in diabetics without baseline CKD, suggesting that Lp(a) is not only involved in progression of renal disease but also possibly the development of GFR impairment.
However, also in 2005, Uhlig
et al. found no association between Lp(a) and GFR after adjusting for age, sex, and race in the Modification of Diet in Renal Disease (MDRD) cohort of 804 participants [
35]. The MDRD study was cross-sectional in nature and focused on non-diabetic patients with pre-existing moderate CKD (iothalamate-determined GFR between 13 and 55 mL/min/1.73 m
2), a population with different baseline characteristics from the PDHS cohort of diabetic individuals without baseline CKD. Considering these differences in study population, the cross-sectional findings of Uhlig’s study are not directly comparable to our cohort. Furthermore, over half of the participants in the MDRD cohort had renal disease of polycystic or glomerular origin and thus had other competing pathophysiologic mechanisms driving their GFR status.
In a more recent prospective study utilizing the Chronic Renal Insufficiency Cohort (CRIC), Rahman
et al. also did not find an association between baseline plasma Lp(a) levels and their endpoints of 50 % decline in eGFR or progression to ESRD [
43]. They did not find a statistically significant association between Lp(a) and any degree of eGFR decline in their fully adjusted analysis, which included the covariates of age, sex, race, diabetes status, blood pressure, statin use, smoking, proteinuria, BMI, and alcohol use. The CRIC study may suggest that Lp(a) does not predict progression of CKD, but it does not negate our findings as its baseline population characteristics are quite different. Like the MDRD cohort, CRIC consists of a patient population (both diabetics and non-diabetics) with pre-existing moderate to severe CKD. Our results suggest that Lp(a) plays a role in the early development of CKD. After CKD has developed, other more dominant pathologies (such as hyperfiltration and fibrosis) may drive subsequent eGFR decline.
Adding to previous knowledge that, like Lp(a), apoC-III is elevated in CKD, [
2,
13,
15,
36,
44] here we report apoC-III’s association with eGFR decline in a TG-dependent manner. A prior small study did not find an association between apoC-III and eGFR slope, but again it utilized a very different study population consisting of 73 non-diabetic adults who had primary renal disease due to glomerular disease, polycystic kidney disease, and interstitial nephritis [
15]. Our findings demonstrate a statistically significant association between baseline apoC-III levels and eGFR decline that was attenuated only partly after adjusting for known risk factors of CKD progression including hemoglobin A1c, a confounder in both CKD progression and in the lipid metabolism pathway as hyperglycemia drives TG and VLDL production [
10]. As expected, the association between baseline apoC-III levels and eGFR slope was then strongly attenuated after additional adjustment for baseline TGs; this result did not change when a subset of subjects with HOMA-IR data underwent further adjustment for HOMA-IR. This result is consistent with other studies showing that TGs are a well-established intermediate of the apoC-III pathway in causing CVD, [
10,
13,
45,
46] as VLDL-associated apoC-III plays a role in TG lipolysis. Although insulin resistance, which contributes to increased VLDL secretion and decreased catabolism of TG rich lipoproteins, would be an expected confounder in the interplay between apoC-III and TGs in diabetics, adjustment for it did not change our results. Further work, including Mendelian randomization strategies, will be needed to investigate whether a direct relationship between VLDL-apoC-III and eGFR decline exists and the renal mechanisms of this effect, if present.
Conclusions
The current study has several strengths. First, the PDHS sub-sample with longitudinal eGFR data is similar to the overall PDHS cohort, is well-characterized and representative of the broader T2DM population, and thus is potentially generalizable among non-insulin-dependent diabetics. Second, the study utilizes a cohort without baseline CKD but with longitudinal follow-up, allowing for the study of potential development of early CKD. This complements other study cohorts that focus on individuals who already have pre-existing CKD. Third, our study has a unique focus of leveraging CV biomarkers measured in the PDHS cohort for the study of CKD risk as well as CVD risk.
Our study also has limitations worth consideration. Within the larger full PDHS cohort, the sample with longitudinal follow-up is considerably smaller, although the sample size was large enough to power the multivariable analysis employed in our study. We only included individuals in our study who continued to receive care at our institution and who had follow-up serum creatinine values. This could lead to selection bias, if individuals with higher levels of apoC-III or Lp(a) were more likely to seek medical care and, due to greater degrees of observation, more likely to be detected to have declining eGFR. Also, while genotypic data was available for a portion of the larger PDHS cohort, our final sample size precluded a robust Mendelian randomization study of the rare or low-frequency
LPA and
APOC3 genetic variants and renal outcomes. As studies on
LPA and
APOC3 variants have established a causal link between their respective lipoprotein biomarker and increased risk of CV disease, [
5,
6,
20,
22,
23] leveraging genetic data as an instrumental variable for CKD risk would more fully address reverse causation and exclude other confounders. In addition, prior studies have established an inverse correlation between the size of the apo(a) portion of Lp(a) and circulating Lp(a) concentrations; [
5,
19,
20] but because the Lp(a) assay used in this current study is not sensitive to apolipoprotein(a) isoform size, we are not able to establish whether the observed association between baseline Lp(a) and eGFR decline is isoform dependent. Our focus is on CKD within a T2DM setting so our findings cannot be extrapolated to non-T2DM settings. Other study limitations include the use of eGFR rather than measured GFR or cystatin-C and recruitment of patients from a single geographical area.
In summary, our data show that higher circulating baseline Lp(a) levels have a robust association with eGFR decline and that higher baseline apoC-III levels have a similar, but more modest relationship. These findings may have implications for whether Lp(a) and apoC-III are possible therapeutic targets for CKD prevention in the diabetic population. With emerging Lp(a)-lowering therapies and
APOC3 anti-sense oligonucleotides targeting apoC-III on the horizon, [
47,
48] further prospective studies, including Mendelian randomization, of Lp(a) and apoC-III in larger cohorts are needed to determine causality and assess the prospects for future clinical trials.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
JL contributed to the design of the analysis plan, performed the statistical analyses, contributed to the data interpretation, and drafted the manuscript. SAK contributed to the interpretation of the data and revision of the manuscript. MPR contributed to the conception and design of the manuscript, acquisition of the data, data interpretation, and manuscript revisions. KT contributed to acquisition of the data. FPW contributed to the design of the analysis plan, data interpretation, and manuscript revisions. All authors read and approved the final manuscript.