High Level of Fasting Plasma Proenkephalin-A Predicts Deterioration of Kidney Function and Incidence of CKD

Abstract

High levels of proenkephalin-A (pro-ENK) have been associated with decreased eGFR in an acute setting. Here, we examined whether pro-ENK levels predict CKD and decline of renal function in a prospective cohort of 2568 participants without CKD (eGFR>60 ml/min per 1.73 m2) at baseline. During a mean follow-up of 16.6 years, 31.7% of participants developed CKD. Participants with baseline pro-ENK levels in the highest tertile had significantly greater yearly mean decline of eGFR (Ptrend<0.001) and rise of cystatin C (Ptrend=0.01) and creatinine (Ptrend<0.001) levels. Furthermore, compared with participants in the lowest tertile, participants in the highest tertile of baseline pro-ENK concentration had increased CKD incidence (odds ratio, 1.51; 95% confidence interval, 1.18 to 1.94) when adjusted for multiple factors. Adding pro-ENK to a model of conventional risk factors in net reclassification improvement analysis resulted in reclassification of 14.14% of participants. Genome-wide association analysis in 4150 participants of the same cohort revealed the strongest association of pro-ENK levels with rs1012178 near the PENK gene, where the minor T-allele associated with a 0.057 pmol/L higher pro-ENK level per allele (P=4.67x10-21). Furthermore, the T-allele associated with a 19% increased risk of CKD per allele (P=0.03) and a significant decrease in the instrumental variable estimator for eGFR (P<0.01) in a Mendelian randomization analysis. In conclusion, circulating plasma pro-ENK level predicts incident CKD and may aid in identifying subjects in need of primary preventive regimens. Additionally, the Mendelian randomization analysis suggests a causal relationship between pro-ENK level and deterioration of kidney function over time.

Keywords: PENK; chronic kidney disease; creatinine; cystatin C; glomerular filtration rate; proenkephalin.

Copyright © 2016 by the American Society of Nephrology.

Figures

Figure 1.

High levels of pro-ENK associate with decreased longitudinal kidney function and risk for incident CKD. Changes per year in (A) eGFR, (B) creatinine, and (C) cystatin C through levels of fasting plasma pro-ENK in a linear model adjusted for age, (sex), and baseline levels. (D) Risk for CKD at follow-up re-examination in relation to baseline pro-ENK concentration in 2568 participants of MDCS-CC. The logistic regression model was adjusted for sex, age, eGFR, fasting glucose, SBP, antihypertensive medication, BMI at baseline, and follow-up time.

Figure 2.

Associations between pro-ENK and change in eGFR according to age in the MDCS. Yearly change in eGFR (ml/min per 1.73 m2) according to tertiles of pro-ENK among the individuals in the youngest, middle, and oldest age tertiles of MDCS-CC at baseline. Change in eGFR is shown as mean (SEM). P for trend from age, sex, and baseline adjusted general linear model through *tertiles of age (young/middle/old); or tertiles of log transformed pro-ENK in either **youngest (mean: 50.01; range, 46.09–53.08 years), ***middle (mean: 56.34; range, 53.09–59.89 years), or ****oldest (means: 63.16; 59.61–68.05 years) age tertile at MDCS-CC baseline. We observed a significant interaction between age and eGFR (Pinteraction<0.001) by introducing an interaction term for age (tertiles) × log transformed pro-ENK into the age, sex, and baseline eGFR adjusted general linear model. eGFR was calculated based on equation by Inker et al. 2012.

Figure 3.

Genome Wide Association Analysis for fasting plasma concentration of pro-ENK in the MDCS. (A) Q–Q and (B) Manhattan plot from genome-wide analysis in n=4150 participants from MDCS-CC. Red line indicates a P value of 5×10−8. Regional locus zoom plot of associations at 8q12.1 with fasting plasma pro-ENK concentration in n=4150 participants from MDCS-CC for (C) 850,658 directly genotyped SNP, and (D) additional imputation with 1000G reference panel up to 21,575,257 SNPs after QC. The purple diamond indicates the lead SNP in addition to all identified SNPs within different degrees of perfect LD (r2≥0.80% [red], 0.8–0.6 [orange], 0.6–0.4 [green], 0.4–0.2 [light blue], and ≤0.2, respectively) at this locus. Visual inspection of the Q–Q plot showed no indication for the presence of population stratification. Conducting a secondary GWA analysis in the same dataset with in total 21,575,257 SNPs imputed to the 1000G reference panel, the results remained similar. The imputed SNP rs2068321, in high LD with rs1012178 (D´=1.00, r2=0.951; in Utah residents with ancestry from northern and western Europe (CEU) using 1000G reference population SNAP version 2.2 Broad Institute) reached the strongest association (P=4.6×10−21) with an associated effect size comparable to that of rs2068321. LD, linkage disequilibrium.

Figure 3.

Genome Wide Association Analysis for fasting plasma concentration of pro-ENK in the MDCS. (A) Q–Q and (B) Manhattan plot from genome-wide analysis in n=4150 participants from MDCS-CC. Red line indicates a P value of 5×10−8. Regional locus zoom plot of associations at 8q12.1 with fasting plasma pro-ENK concentration in n=4150 participants from MDCS-CC for (C) 850,658 directly genotyped SNP, and (D) additional imputation with 1000G reference panel up to 21,575,257 SNPs after QC. The purple diamond indicates the lead SNP in addition to all identified SNPs within different degrees of perfect LD (r2≥0.80% [red], 0.8–0.6 [orange], 0.6–0.4 [green], 0.4–0.2 [light blue], and ≤0.2, respectively) at this locus. Visual inspection of the Q–Q plot showed no indication for the presence of population stratification. Conducting a secondary GWA analysis in the same dataset with in total 21,575,257 SNPs imputed to the 1000G reference panel, the results remained similar. The imputed SNP rs2068321, in high LD with rs1012178 (D´=1.00, r2=0.951; in Utah residents with ancestry from northern and western Europe (CEU) using 1000G reference population SNAP version 2.2 Broad Institute) reached the strongest association (P=4.6×10−21) with an associated effect size comparable to that of rs2068321. LD, linkage disequilibrium.

Figure 4.

Associations between rs1012178 genotype and longitudinal kidney function, and risk for incident CKD in the MDCS. (A) Risk for CKD at follow-up re-examination in relation to rs101278 genotypes in 2766 participants of MDCS-CC. The logistic regression model was adjusted for sex, age, eGFR, fasting glucose, SBP, antihypertensive medication, BMI at baseline, and follow-up time. Changes per year in (B) eGFR, (C) creatinine, and (D) cystatin C through rs1012178 genotypes in a linear model adjusted for age, (sex), and baseline levels.

Figure 5.

IV analysis for rs1012178 genotype and longitudinal kidney function, and risk for incident CKD in the MDCS. (A) IV analysis for rs1012178 genotype for risk of incident CKD in 2308 participants of MDCS-CC. The logistic regression model was adjusted for sex, age, eGFR, fasting glucose, SBP, antihypertensive medication, BMI at baseline, and follow-up time. IV analysis for rs1012178 genotype for changes in (B) eGFR, (C) creatinine, and (D) cystatin C through rs1012178 genotypes in a linear model adjusted for age, (sex), and baseline levels.