Abstract
Background and objectives Hemostatic factors have been associated with kidney function decline, and evidence suggests stronger effects among African Americans. The presence of APOL1 renal risk variants, common in African Americans, might partly underlie this risk difference.
Design, setting, participants, & measurements A total of 13,337 participants in the Atherosclerosis Risk in Communities study were followed from 1987–1989 until 2010. Participants were categorized into three groups by ancestry and APOL1 risk status: European Americans (n=10,206), African Americans with zero or one APOL1 risk allele (n=2,733), and African Americans with two APOL1 risk alleles (n=398). ESRD events were ascertained through linkage to the US Renal Data System. Cox regression was used to estimate the risk for ESRD associated with hemostatic factors (factor VIIc, factor VIIIc, fibrinogen, von Willebrand factor, protein C, and antithrombin III).
Results There were 232 cases of ESRD over 21.5 years (European Americans, 119; African Americans with zero or one APOL1 risk allele, 94; African Americans with two APOL1 risk alleles, 19). In unadjusted and adjusted analysis of the overall population, higher levels of all hemostatic factors, except antithrombin III, were significantly associated with ESRD (all P<0.05). Factor VIIc had the strongest association (per one interquartile range; adjusted hazard ratio, 1.46; 95% confidence interval, 1.21 to 1.76). With the exception of fibrinogen, the risk associated with each hemostatic factor was stronger in African Americans with two APOL1 risk alleles compared with the other two groups. Statistically significant interactions were seen for factor VIIIc and protein C (interaction between those with two APOL1 risk allele and the other two groups: P<0.02 for factor VIIIc and <0.04 for protein C).
Conclusions Higher levels of factor VIIc, VIIIc, fibrinogen, von Willebrand factor, and protein C were associated with ESRD risk, with a significantly stronger association of factor VIIIc and protein C in African Americans with two APOL1 risk alleles.
Introduction
ESRD is an important public health burden, affecting over half a million people (disproportionately African American) and costing an estimated $40 billion annually in public and private funds (1). Epidemiologic studies have linked higher levels of hemostatic factors in the coagulation cascade to faster kidney function decline, with stronger associations in African Americans than European Americans (2,3). The pathophysiology underlying the association of hemostatic factors to kidney function decline is unclear, although triggers of hemostatic activation, including vascular injury, endothelial dysfunction, and inflammation, have been proposed as potential mechanisms (4,5).
APOL1 genetic variants common in African Americans are known to be associated with progressive kidney disease (6–9). The APOL1 high-risk genotype (two copies of the G1 or G2 alleles) has a population frequency of approximately 13% in African Americans and confers an approximately 2-fold higher risk for kidney function decline in cohort studies (8,9). Although the role of APOL1 in relation to kidney function is largely unknown, environmental factors probably affect APOL1-associated renal risk (10). For example, HIV infection (and particularly an unsuppressed viral load) in individuals with the APOL1 high-risk genotype appears to increase the susceptibility for progressive kidney disease (11,12). Other events that activate the innate immune response might similarly increase APOL1-associated susceptibility. Inflammatory cytokines induce the expression of the APOL1 encoded protein, apolipoprotein L1 (apoL1) in macrophages, and endothelial and epithelial cells (13–15). Increased expression of the high-risk variants has been associated with organ damage in a transgenic mice study and reduced cell survival in in vitro studies (14–16).
Using the Atherosclerosis Risk in Communities (ARIC) study, a community-based prospective cohort of both African Americans and European Americans, we investigated the association between levels of hemostatic factors (factor VIIc, factor VIIIc, fibrinogen, von Willebrand factor [vWF], protein C, and antithrombin III) and incident ESRD. Given the previously demonstrated interaction of ancestry and hemostatic factors with kidney function decline (2,3), and that APOL1 expression and hemostatic activation may share risk factors, we hypothesized that the risk of ESRD would be higher in participants with higher levels of hemostatic factors, with a stronger association in those with two APOL1 risk alleles.
Materials and Methods
Study Population
The ARIC Study is a community-based prospective observational study of 15,792 individuals age 45–64 years at the baseline visit from 1987 to 1989. Details of the ARIC cohort have been published elsewhere (19). For this study, we excluded individuals missing eGFR or with eGFR<15 ml/min per 1.73 m2 (n=172), those who were not fasting at the time of phlebotomy (n=552), those receiving anticoagulant or hemostatic medications (warfarin, heparin, dipyridamole, aminocaproic acid, or pentoxifylline; n=293), those missing hemostatic factor values (n=420), African Americans missing APOL1 genotype (n=343), those with self-reported race other than black or white (n=43), and those with missing data on covariates in the final model (n=633). Our final analyzed sample included 13,337 participants. For the purpose of this study, participants were followed until death, ESRD, or December 31, 2010, whichever came first. ESRD events were ascertained through linkage with the US Renal Data System.
Hemostatic Factor Measurements
The design of the hemostasis component of the ARIC study and the methods for the measurement of hemostatic factors have been described previously (18–20). Briefly, six hemostatic factors were measured in the overall population (factor VIIc, factor VIIIc, fibrinogen, vWF, protein C, and antithrombin III) from blood drawn from the antecubital vein after an 8-hour fast. Samples were processed according to a standardized protocol. Factor VII and VIII activity (factor VIIc and factor VIIIc, respectively) were measured by the coagulation test; fibrinogen by the thrombin-time titration method; vWF and protein C antigen by ELISA; and antithrombin III by a chromogenic substrate for thrombin. The reliability coefficients for repeated measures from a subsample of participants over 1–2 weeks were 0.78 for factor VIIc, 0.86 for factor VIIIc, 0.72 for fibrinogen, 0.68 for vWF, 0.56 for protein C, and 0.42 for antithrombin III (21).
Genotyping and APOL1 Risk Group Definition
Taqman assays were used for direct genotyping of the APOL1 risk variants in African Americans. The G1 risk variant consists of two missense mutations (rs73885319 [S342G] and rs60910145 [I384M]) that are in almost total linkage disequilibrium on the same haplotype, and the G2 risk variant (rs71785313) is a 6–base-pair deletion (22). The G1 and G2 variants are in high linkage disequilibrium on different haplotypes. The G1/G2 haplotypes were inferred using PLINK (23). All inferred haplotypes had a posterior probability of 1. Because both G1 and G2 variants are rare (minor allele frequency <1%) in European Americans (24), we assumed all European Americans had zero or one copy of the risk allele. Previous studies have shown that the G1 and G2 variants confer risks for ESRD in a recessive manner (11,25); therefore, we defined APOL1 high-risk status as having two risk alleles (G1/G1, G1/G2, or G2/G2). The participants were categorized into three groups by ancestry and the number of APOL1 risk alleles: European Americans, African Americans with zero or one APOL1 risk allele (APOL1 0/1 risk allele), and African Americans with two APOL1 risk alleles (APOL1 2 risk alleles).
Other Measurements
Prevalent diabetes mellitus was defined as having a fasting glucose level ≥126 mg/dl, nonfasting glucose level ≥200 mg/dl, self-reported diabetes medication use, or physician diagnosis of diabetes. BP measures were calculated as the average of the last two measures of three seated BP measures performed by certified technicians using a random-zero sphygmomanometer after the participant rested for 5 minutes. Medication use was determined on the basis of the inspection of medication bottles. Prevalent coronary heart disease was defined as a self-reported physician diagnosis or an electrocardiogram obtained during the baseline visit with signs of a previous myocardial infarction. Enzymatic methods were used to measure total plasma cholesterol and triglyceride levels. LDL cholesterol was calculated using the Friedewald equation (26) (excluding those with incalculable LDL cholesterol levels because of triglyceride values >400 mg/dl). Smoking status was based on self-report. eGFR was calculated using the CKD-Epidemiology Collaboration equation (27) with calibrated and standardized serum creatinine (28).
Statistical Analyses
Baseline characteristics of participants by ancestry-APOL1 risk status (European American, African-American APOL1 0/1 risk allele, and African-American APOL1 2 risk alleles) were compared using t tests for nonskewed continuous variables, Wilcoxon tests for skewed continuous variable, and chi-squared tests for categorical variables.
For the analysis of the association between hemostatic factors and ESRD, the hemostatic variables were log-transformed (because of right skewness) and standardized to interquartile range (IQR), such that risk estimates represent an increase in one IQR. We estimated unadjusted and adjusted hazard ratios (HRs) and 95% confidence intervals (95% CIs) using Cox regression. The covariates in the adjusted model were determined a priori on the basis of a literature review of CKD progression factors (29,30), medication or hematologic factors that may influence hemostatic factor levels, and their availability in the ARIC study. The covariates included ancestry-center-APOL1 risk status, sex, and year of hemostatic marker measurement, as well as baseline age, systolic BP, diabetes status, hypertension medication use, prevalent coronary heart disease, smoking status, log-transformed triglycerides, HDL cholesterol, log-transformed LDL cholesterol, eGFR, serum albumin, use of salicylates, hematocrit, log-transformed white blood cell count, and log-transformed platelet count.
We assessed the proportional hazard assumption by testing the time interaction term of the hemostatic factors for ESRD. The time interaction terms of factor VIIIc, fibrinogen, and vWF were significant, demonstrating slightly weaker association over time (P<0.05). Therefore, for these three hemostatic factors, Cox regression evaluated their average association during the follow-up period. We assessed the assumption of linearity for the continuous association of hemostatic factors with ESRD in two ways: first by adding a square term of the log-transformed hemostatic factor in the unadjusted and adjusted analysis, and second using piecewise linear splines with knots at each tertile of the hemostatic factor. Neither the square terms nor the slope differences between splines were significant (P>0.05). Therefore, we retained the model using the log-transformed hemostatic factors. For each ancestry-APOL1 risk group, we estimated the risk for ESRD per one IQR of hemostatic factor levels using an interaction term between log-transformed hemostatic factor levels and ancestry-APOL1 risk status.
For the two hemostatic factors (factor VIIIc and protein C) found to have significantly stronger association in African Americans with two APOL1 risk alleles, we evaluated their association with mortality. The square terms of the log-transformed hemostatic factors were significant, indicating violation of the linear assumption. Therefore, we used linear splines with knots at each quintile for this analysis. Baseline characteristics were analyzed using R software. Other analyses were conducted using Stata/SE 13.1 (Stata Corp., College Station, TX).
Results
Study Population Characteristics
Over a median follow-up of 20.5 years, 232 ESRD events occurred in 13,337 individuals. Characteristics of the study populations are presented by European American, African-American APOL1 0/1 risk allele group, and APOL1 2 risk alleles group in Table 1. Compared with European Americans, both African-American APOL1 groups had a higher prevalence of diabetes (7.9% in European Americans, 15.7% in the African-American APOL1 0/1 risk allele group, and 15.3% in the APOL1 2 risk alleles group) and higher proportions of individuals receiving hypertension medication (24.2% in European Americans, 41.9% in the African-American APOL1 0/1 risk allele group, and 43.0% in the APOL1 2 risk alleles group). Overall, the baseline characteristics of the two African-American APOL1 groups were similar, except for factor VIIc levels (115% in the APOL1 0/1 risk allele group versus 110% in the APOL1 2 risk alleles group; P=0.05). The correlation between hemostatic factors varied from 0.01 between vWF and antithrombin III to 0.73 between vWF and factor VIIIc (Supplemental Table 1).
Baseline characteristics of the participants by ancestry-APOL1 status
Association between Hemostatic Factors and Incident ESRD
In unadjusted and adjusted analyses of the overall population, higher levels of five hemostatic factors (factor VIIc, factor VIIIc, fibrinogen, vWF, and protein C) were associated with higher risk of incident ESRD (P<0.05) (Table 2). In adjusted analyses of the overall population, higher levels of factor VIIc had the strongest association with incident ESRD (per one IQR higher: HR, 1.46; 95% CI, 1.21 to 1.76; P<0.001). Supplemental Figure 1, A–F, presents the Kaplan–Meier estimates of the proportion free of ESRD by tertiles of the hemostatic factors.
Unadjusted and adjusted hazard ratios for ESRD in the overall population
In the adjusted analysis by subgroup of ancestry-APOL1 status, factor VIIc was associated with incident ESRD in all three ancestry-APOL1 risk groups (per one IQR higher: European American, HR, 1.32 [95% CI, 1.02 to 1.71]; APOL1 0/1 risk allele, HR, 1.46 [95% CI, 1.12 to 1.89]; APOL1 2 risk alleles, HR, 1.97 [95% CI, 1.29 to 3.02]; for interaction between European Americans and APOL1 2 risk alleles, P=0.11; for interaction between APOL1 0/1 risk alleles and 2 risk alleles, P=0.22) (Table 3). In contrast, higher levels of factor VIIIc and protein C were associated with significantly higher risk for ESRD in those with two APOL1 risk alleles (factor VIIIc: European American, HR, 1.07 [95% CI, 0.83 to 1.38]; APOL1 0/1 risk allele, HR, 1.19 [95% CI, 0.92 to 1.53]; APOL1 2 risk alleles, HR, 2.48 [95% CI, 1.39 to 4.41]; P for interaction <0.02; protein C: European American, HR, 0.99 [95% CI, 0.79 to 1.25]; APOL1 0/1 risk allele, HR, 1.31 [95% CI, 1.03 to 1.67]; APOL1 2 risk alleles, HR, 2.61 [95% CI, 1.43 to 4.76]; P for interaction <0.04). The Kaplan–Meier estimates of the proportion free of ESRD by tertile 3 versus tertiles 1 and 2 of factor VIIIc and protein C in each ancestry-APOL1 risk group are presented in Supplemental Figure 2, A and B. Antithrombin III had no significant association with incident ESRD.
Adjusted hazard ratios for ESRD for each ancestry-APOL1 risk group
With respect to mortality, we observed a total of 4632 events (European Americans, 3290; African Americans: APOL1 0/1 risk allele, 1178; APOL1 2 risk alleles, 164). Higher factor VIIIc activity was associated with higher mortality risk, with a steeper gradient above the median value (per IQR change in the lowest quintile, adjusted HR, 1.05 [95% CI: , 0.86 to 1.24]; per IQR change in the highest quintile, adjusted HR, 1.49 [95% CI, 1.33 to 1.67]) (Supplemental Figure 3). Protein C had a slight U-shaped association with mortality. In the first quintile of protein C levels, higher protein C levels were associated with lower risk for death (per IQR change, adjusted HR, 0.78; 95% CI, 0.70 to 0.87) (Supplemental Figure 4). In the highest quintile of protein C levels, higher protein C levels were associated with higher risk for death (per IQR change, adjusted HR, 1.15; 95% CI, 0.99 to 1.33). The interaction between hemostatic factor and ancestry-APOL1 risk status for mortality was not statistically significant for either factor VIIIc or protein C (P>0.1).
Discussion
This study of 13,337 persons from a population-based cohort with >20 years of follow-up demonstrates that higher factor VII activity (factor VIIc), factor VIII activity (factor VIIIc), fibrinogen, vWF, and protein C were associated with higher risk of developing ESRD, independent of baseline kidney function and other traditional risk factors. We also demonstrate a significant interaction of factor VIIIc and protein C with APOL1 risk status for incident ESRD, with stronger associations observed in African Americans with two APOL1 risk alleles. This might suggest that hemostatic activation (or its triggers) acts synergistically with the APOL1 risk alleles to increase susceptibility for kidney disease.
The present study expands on previous studies of hemostatic factors and kidney disease risk. In the Multi-Ethnic Study of Atherosclerosis, higher factor VIII activity was significantly associated with rapid decline in eGFR, and higher levels of fibrinogen had a weak, nonsignificant association in the same direction (5). In the Cardiovascular Health Study, higher factor VII levels were significantly associated with annual increase in serum creatinine, with stronger association in blacks than in nonblacks (3). A prior investigation in the ARIC study demonstrated that higher levels of factor VIIIc, fibrinogen, and vWF were significantly associated with increased risk of incident CKD in both European Americans and African Americans, while the association of factor VIIc was significant only in African Americans with the same direction of association in European Americans (2). Although differences exist in the hemostatic factors tested and the statistical significance of associations among studies, the qualitative associations between higher hemostatic factor levels and kidney function decline were consistent across studies. The present study evaluated ESRD, arguably the most clinically meaningful kidney outcome, and found significant associations with factor VIIc, factor VIIIc, fibrinogen, vWF, and protein C in a population-based cohort, with stronger associations of factor VIIIc and protein C in participants with two APOL1 risk alleles.
Multiple factors that increase APOL1-related renal susceptibility have been proposed or reported (10). The interaction of factor VIIIc activity and protein C levels with APOL1 high-risk status may suggest an additional synergistic mechanism of action. Although speculative, APOL1 expression and higher levels of factor VIIIc and protein C might be linked through vascular injury or infection. Both factor VIII and protein C are activated in response to vascular injury (31,32), with factor VIII serving as a procoagulant and protein C serving as an anticoagulant (33). Vascular injury can trigger the production of inflammatory cytokines (34), which can induce the expression of APOL1 (35). Increased expression of the APOL1 G1 or G2 risk variants causes organ damage in transgenic mice (16) and reduces cell survival (14,15). On the other hand, a recent study of HIV-infected individuals found that total plasma apoL1 levels had little cross-sectional correlation with inflammation biomarker levels or CKD status (36). Additional research is needed to prospectively evaluate the link between circulating apoL1 levels and kidney function decline.
The association between higher levels of protein C and ESRD risk is somewhat surprising because previous studies have shown an association of higher levels of protein C with lower risk of venous thromboembolism, atrial fibrillation, and ischemic stroke (37–39). We found not only an association of protein C with ESRD but also a slight U-shaped association between protein C and mortality. Higher protein C levels have been associated with prevalent hypertension or diabetes (40–42). Some have hypothesized that a reactive anticoagulatory response to hemostatic activity could increase protein C levels (42). We did note positive correlations between the anticoagulant protein C and the other procoagulant hemostatic factors, particularly factor VIIc. In addition, protein C levels may not perfectly correlate with activated protein C levels, a more direct measure of anticoagulant activity. Indeed, patients with diabetes and those undergoing hemodialysis may exhibit normal or higher protein C levels but lower activated protein C levels or thrombomodulin-induced anticoagulant activity (43,44).
Strengths of this study include a large community-based cohort of both African Americans and European Americans, a long follow-up period spanning >20 years, and careful measurement of hemostatic markers not obtained routinely in clinical care. Although the results from the present study are robust and biologically plausible, a few limitations should be noted. First, although many known risk factors of ESRD and hematologic variables were available to include as covariates, measures of albuminuria, an important marker of kidney damage, were not available at baseline. We cannot exclude the possibility that the observed interactions between hemostatic factor and APOL1 status may be driven by albuminuria. Regarding hemostasis, the available hemostatic factors in this study provide only a partial view of the hemostasis process. The association of the hemostatic factors may represent the role of other factors in the hemostasis system in kidney function decline. Inflammatory biomarkers were not available at baseline to evaluate the relation between hemostatic activation and inflammation. Hemostatic activation could be a consequence of inflammation (45). Finally, the hemostatic factors were measured in blood and not in kidney tissues. Our results provide insight but not direct evidence on the pathogenesis of ESRD.
In conclusion, this study shows that higher levels of factor VII activity, factor VIII activity, fibrinogen, vWF, and protein C are associated with the development of ESRD in a middle-aged European American and African-American cohort, with higher levels of factor VIII activity and protein C having significantly stronger associations in African Americans with two APOL1 risk alleles. These results suggest hemostatic activation or its triggers may work synergistically to increase APOL1-associated renal risk. Further investigation into the pathways of apoL1 expression and hemostatic activation may help to unravel the APOL1-associated susceptibility for ESRD.
Disclosures
None.
Acknowledgments
We would like to dedicate this manuscript to our friend, colleague, and mentor, Dr. Kao, who was instrumental in discovering the genetic association of kidney disease in the APOL1 region. She has inspired many to study genetic risk in kidney disease and work toward improving the lives of persons with kidney disease.
The authors thank the staff and participants of the Atherosclerosis Risk in Communities study for their important contributions.
Some of the data reported here have been supplied by the US Renal Data System. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the United States government.
A.T. is supported by a National Institute of Diabetes and Digestive and Kidney Diseases Renal Disease Epidemiology Training Program (T32-DK007732). This work is partly supported by a National Institutes of Health/National Heart, Lung, and Blood Institute Cardiovascular Epidemiology Training Grant (T32-HL007024) and the National Kidney Foundation of Maryland Mini-grant. The Atherosclerosis Risk in Communities Study is carried out as a collaborative study supported by National Heart, Lung, and Blood Institute contracts (HHSN268201100005C, HHSN268201100006C, HHSN268201100007C, HHSN268201100008C, HHSN268201100009C, HHSN268201100010C, HHSN268201100011C, and HHSN268201100012C).
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.08340814/-/DCSupplemental.
See related editorial, “Hemostatic Factors, APOL1, and ESRD Risk: Another Piece of the Puzzle?,” on pages 723–725.
- Received August 22, 2014.
- Accepted January 13, 2015.
- Copyright © 2015 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...
- HDL in CKD--The Devil Is in the Detail
- Race, APOL1 Risk, and eGFR Decline in the General Population
- Using Genetic Technologies To Reduce, Rather Than Widen, Health Disparities
- Examination of Potential Modifiers of the Association of APOL1 Alleles with CKD Progression
- Hemostatic Factors, APOL1, and ESRD Risk: Another Piece of the Puzzle?