Decreased plasma kallikrein activity is associated with reduced kidney function in individuals with type 1 diabetes

Aims/hypothesis Plasma kallikrein is the central mediator of the plasma kallikrein–kinin system, which is involved both in vascular control and thrombin formation cascades. The plasma kallikrein–kinin system has also been considered protective in pathological conditions, but the impact of plasma kallikreins on diabetic nephropathy remains unknown. The objective of this cross-sectional study was to explore the association of plasma kallikrein with diabetic nephropathy. Methods We measured plasma kallikrein activity in 295 individuals with type 1 diabetes at various stages of diabetic nephropathy, and we tested the genetic association between the plasma kallikrein–kinin system and kidney function in 4400 individuals with type 1 diabetes. Results Plasma kallikrein activity was associated with diabetes duration (p < 0.001) and eGFR (p < 0.001), and plasma kallikrein activity was lower with more advanced diabetic nephropathy, being lowest in individuals on dialysis. The minor alleles of the KNG1 rs5030062 and rs710446 variants, which have previously been associated with increased plasma pre-kallikrein and/or factor XI (FXI) protein levels, were associated with higher eGFR (rs5030062 β = 0.03, p = 0.01; rs710446 β = 0.03, p = 0.005) in the FinnDiane cohort of 4400 individuals with type 1 diabetes. Conclusions/interpretation Plasma kallikrein activity and genetic variants known to increase the plasma kallikrein level are associated with higher eGFR in individuals with type 1 diabetes, suggesting that plasma kallikrein might have a protective effect in diabetic nephropathy. Electronic supplementary material The online version of this article (10.1007/s00125-020-05144-1) contains peer-reviewed but unedited supplementary material, which is available to authorised users.

Decreased plasma kallikrein activity is associated with reduced kidney function in individuals with type 1 diabetes Methods Participants. All study participants were Finnish adult individuals with type 1 diabetes. The diagnosis of type 1 diabetes for each participant was made by his/her attending physician at the time of diabetes onset according to national evidence-based clinical practice guidelines. We further refined the diagnosis of type 1 diabetes by requiring an onset of the disease before the age of 40 years, and permanent insulin treatment initiated within one year of diagnosis. During the FinnDiane study visit, participants underwent a clinical examination and completed standardized questionnaires regarding health and medical history in collaboration with the attending physician/nurse. Anthropometric measurements (waist circumference, weight, height and hip circumference) and blood pressure measurements were performed by a trained nurse.
Blood pressure was measured two times with a two-minute interval in the sitting position after a ten-minute rest and the mean values of these two measurements were used in the analyses.
We calculated the mean arterial pressure as 1/3 systolic blood pressure + 2/3 diastolic blood pressure in mmHg.
Renal status was assessed by the albumin excretion rate (AER) in two out of three consecutive timed overnight or 24h urine sample collections or by the albumin-to-creatinine ratio (ACR) in morning spot urine samples. The following criteria were used: normoalbuminuria AER<20 µg/min or <30 mg/24h or ACR <2.5 mg/mmol (for men) and <3.5 mg/mmol (for women); microalbuminuria AER≥20 and <200 µg/min or ≥30 and <300 mg/24h or ACR≥2.5 and <25 mg/mmol (for men) and ≥3.5 and <35 mg/mmol (for women); macroalbuminuria AER≥200 µg/min or≥300 mg/24h or ACR≥25 (for men) and ≥35 mg/mmol (for women), and end-stage renal disease (ESRD). The ESRD group comprised of patients having received a kidney transplant or undergoing dialysis. DN was defined as microalbuminuria, macroalbuminuria or ESRD. The eGFR was calculated using the CKD-EPI equation [1]. The main study cohort of 295 individuals was divided into four groups based on AER or ACR: normoalbuminuria (n=165), microalbuminuria (n=41), macroalbuminuria (n=37) and ESRD (n=52, from which 36 individuals had received kidney transplantation, and 16 individuals were on dialysis). It can be concluded, that dextran sulfate activation in plasma does not reflect the total available plasma kallikrein activity; instead, it reflected the underlying plasma kallikrein potential after induction by a similar activator. As this study intended to evaluate the differences of the physiologically inhibited plasma kallikrein in the context of diabetic nephropathy, measuring the protease activity under inhibitory conditions should resemble physiologic conditions more than quantifying the total protein.
Factor XI (FXI) assay. In parallel with the plasma kallikrein assay, the same samples were prepared for the FXI assay according to the manufacturer's protocol, and FXI activity was measured following the microtiter method (COA0090 CoaChrom Factor XI, Coachrom Diagnostica, Maria Enzersdorf, Austria). Absorbance values were recorded at 405 nm at 37˚C, with 1-minute intervals for 60 min. The inter-assay CV was <10%. We expressed the results as baseline value subtracted from the end-point value and converted to units/mL against normal plasma standard.
Genotyping and quality control. Genotyping was performed in three batches and quality control as well as genotype imputation was performed as previously described [2]. Batch one ) and was not included in the current study, resulting in 5,161 participants with available genome-wide genotyping data. We further excluded those that did not meet the type 1 diabetes criteria in this study (diabetes onset age <40 years and insulin treatment within 1 year; n=283) and patients with missing data for age, diabetes duration and sex (n= 292) as well as RAAS blocker therapy (n=66) resulting in 4,520 individuals. Of these 4,520 individuals, we had data on eGFR for n=4400 (2.7% missing), diabetic nephropathy for n=4349 (3.8% missing), systolic blood pressure for n= 4414 (2.3% missing), diastolic blood pressure for n=4410 (2.4% missing), and for 24h urine sodium concentration for n=2491 (44.9% missing). The 10 first principal components were calculated with the EIGENSTRAT, version 3.0.

SNP selection.
To select SNPs affecting the KKS system (ESM Fig. 1), we used the curated literature-based GWAS catalogue (https://www.ebi.ac.uk/gwas/) and searched for genomewide significant genetic variants for FXII, plasma kallikrein and FXI using the search words 'plasma kallikrein', 'Factor XI' and 'Factor XII'. We then selected SNPs associations performed in studies with n≥500 individuals. The search resulted in two SNPs for plasma kallikrein (KLBK1 rs1511802, KNG1 rs5030062), and five SNPs for factor XI (KNG1 rs710446, F11 rs4253417, F11 rs2289252 and KNG1 rs5030062, ESM Table 1). There were no results for factor XII. However, F12 rs1801020 has previously been established to alter factor XII protein levels, therefore this SNP was additionally included in the analyses. We extracted genotypes from the FinnDiane GWAS for the selected SNPs. We converted imputed genotypes to most likely genotypes using a 90% threshold for the genotype posterior probability. Deviations from the Hardy-Weinberg Equilibrium (HWE) were assessed using the exact test statistics in PLINK1.9. [3]. None of the SNPs deviated significantly from the HWE (ESM Table 2).

Statistical analyses.
Five individuals (1.7%) with measured kallikrein and FXI activity had missing data for some variables and were thus excluded from the analysis. Genetic associations were tested using an additive model for the SNP and the associations were adjusted for age, sex, diabetes duration, renin-angiotensin-aldosterone system (RAAS) blocker therapy and the two first principal components calculated based on the genome-wide genotyping data to account for population substructure in Finland [2]. Genetic association and epistasis analyses were done in PLINK, and other statistical analyses were done in R [3]. For genetic comparisons, Bonferroni correction was applied.
We calculated the power of the genetic analyses using QUANTO [4] and the power of the other analyses using R (WebPower package). Power analyses were performed as post hoc analyses for the observed effect sizes in the study. In the genetic analyses, our study had 73% power to detect a significant association between KNG1 rs5030063 and eGFR with the observed effect size of β=0.02 and 74 % power to detect a significant association between KNG1 rs710446 and eGFR with the observed effect size of β=0.03. For the genetic analyses with 24h urine sodium concentration, we had 54 % power to detect an association between F12 rs1801020 and 24h urine sodium concentration with the observed effect size of β=2.15. In the non-genetic association analyses performed in the 295 individuals with measured plasma kallikrein activity and factor XI activity, we had 99% power to detect an association between plasma kallikrein activity and eGFR with the observed effect size of β=0.24 (Cohen effect estimate f2= 0.08) and 27% power to detect a significant association study between factor XI activity and eGFR with the observed effect size of β=0.36 (Cohen effect estimate f2=0.007). In the analyses with plasma kallikrein activity and blood pressure (β=0.001), we had 6% power to detect such subtle changes (Cohen estimate f2=0.0004).  Table 3.

Impact of age, diabetes duration and medication on plasma kallikrein activity. Cross-talk
between the KKS and the RAAS has been discussed [5][6][7][8], suggesting that KKS has a counterbalancing role to RAAS. Furthermore, plasma kallikrein and F12 rs1801020 are involved in the RAAS in healthy individuals [9]. By measuring plasma kallikrein activity, one can get some information about the functionality of this biologically inhibited protease under pathological conditions, and it is possible, that the RAAS-blocker therapy adds to this biological inhibition, preventing plasma kallikrein from functioning correctly. p=0.016). When data were stratified by DN group, a similar trend remained, but the statistical significance was lost, likely due to the low number of therapy free individuals (data not shown). In unadjusted logistic regression model, plasma kallikrein activity was associated with RAAS-blocker therapy (β=-0.32 [95% CI -0.61 --0.02], p=0.03).
Finally, we observed, that plasma kallikrein activity, was negatively correlated with systolic blood pressure (r=-0.13, p=0.03) and the association remained after adjustment for age, sex, SNPs that were associated with kallikrein and FXI activity were rs5030062 and rs710446 near the gene KNG1 (Table 1). These two SNPs were highly correlated (r 2 =0.91) and thus represented the same signal at that locus. Therefore, we focused only on the rs710446 in further analyses (ESM Fig. 2a-b). Altogether, KNG1 rs710446 and F12 rs1801020 accounted for 25% of the variance (r 2 =0.25) in plasma kallikrein activity. There were no significant SNP-SNP interactions for either plasma kallikrein or FXI activity.
We selected the three SNPs with confirmed associations with plasma kallikrein activity in our primary study cohort of 295 individuals and tested their association with DN, systolic and diastolic blood pressure. Additionally, we tested the SNP association with 24h urine sodium concentration (N=2675). The Bonferroni corrected statistical significance threshold for the three SNPs, was p<0.016.
None of the tested SNPs were associated with systolic blood pressure or DN in linear and logistic regression models, respectively, adjusted for age, sex, diabetes duration, RAAS-blocker therapy and for the principal components. The minor allele of F12 rs1801020 was associated with 24h urine sodium concentration (β=2.15, p=0.03). Table 1

Figures
ESM Fig. 1 The activation of the kallikrein-kinin system.