Clinical Research

Blood HER2 and Uromodulin as Causal Mediators of CKD

Jennifer Sjaarda, Hertzel C. Gerstein, Salim Yusuf, Darin Treleaven, Michael Walsh, Johannes F.E. Mann, Sibylle Hess and Guillaume Paré
JASN April 2018, 29 (4) 1326-1335; DOI: https://doi.org/10.1681/ASN.2017070812

Abstract

Many biomarkers have been epidemiologically linked with CKD; however, the possibility that such associations are due to reverse causation or confounding limits the utility of these biomarkers. To overcome this limitation, we used a Mendelian randomization (MR) approach to identify causal mediators of CKD. We performed MR by first identifying genetic determinants of 227 serum protein biomarkers assayed in 4147 participants of the Outcome Reduction with Initial Glargine Intervention (ORIGIN) trial who had early or prediabetes, and assessing the effects of these biomarkers on CKD in the CKD genetics consortium (n=117,165; 12,385 cases) using the inverse-variance weighted (fixed-effects) method. We then estimated the relationship between the serum concentration of each biomarker identified and incident CKD in ORIGIN participants. MR identified uromodulin (UMOD) and human EGF receptor 2 (HER2) as novel, causal mediators of CKD (UMOD: odds ratio [OR], 1.30 per SD; 95% confidence interval [95% CI], 1.25 to 1.35; P<5×10−20; HER2: OR, 1.30 per SD; 95% CI, 1.14 to 1.48; P=8.0×10−5). Consistent with these findings, blood HER2 concentration associated with CKD events in ORIGIN participants (OR, 1.07 per SD; 95% CI, 1.01 to 1.13; P=0.01). Additional exploratory MR analyses identified angiotensin-converting enzyme (ACE) as a regulator of HER2 levels (β=0.13 per SD; 95% CI, 0.08 to 0.16; P=2.5×10−7). This finding was corroborated by an inverse relationship between ACE inhibitor use and HER2 levels. Thus, UMOD and HER2 are independent causal mediators of CKD in humans, and serum HER2 levels are regulated in part by ACE. These biomarkers are potential therapeutic targets for CKD prevention.

CKD is a growing public health problem that increases the risk of cardiovascular disease, kidney failure, and other complications.1 Whereas glucose lowering, BP lowering, and therapies that target the renin-angiotensin-aldosterone system (RAAS)2 can slow progression of CKD, the mechanism(s) by which they work are not fully understood. Elucidation of these and other CKD-related mechanisms may identify novel therapies, and the identification of causal biomarkers for CKD represents one promising approach. Unfortunately, candidate biomarkers identified using traditional epidemiologic approaches may be confounded with, or caused by, other unmeasured biomarkers or mechanisms. Under a strict set of assumptions, Mendelian randomization (MR) can overcome these problems.3

MR is a powerful genetic methodology that is on the basis of the principle that genetic variants are inherited randomly and independently of other risk factors for disease. If the levels of a particular biomarker are affected by the presence of a genetic variant, and if that variant also affects the incidence of a disease, the variant may be causing the disease by modulating the levels of the risk factor. The random distribution of the genetic variant at birth minimizes the possibility of confounding or reverse causation as explanations for the link between the biomarker and disease, in the same way that the random allocation of a therapy in a randomized, controlled trial minimizes this possibility.4 These principles have been successful in identifying risk factors that are causal for CVD, and biomarkers identified using MR have been subsequently validated in randomized trials of therapeutic agents.5 Specifically, MR techniques have confirmed LDL cholesterol, IL-6 receptor, and lipoprotein(a) as causal biomarkers of coronary artery disease.6–8 Whereas there are fewer examples of MR in the field of nephrology, this approach has recently identified a causal effect of lower iron and ferritin levels on decreased kidney function and has ruled out a causal relationship between fetuin-A and mortality in patients on dialysis.9,10

MR methodology has traditionally been used to determine whether a candidate biomarker is causally related to a clinical outcome (clinicaltrials.gov; NCT 00069784). However, when combined with a large panel of biomarkers measured in a prospective study which accrued many clinical outcomes, it can also be used to identify new, unsuspected biomarkers that are likely to be causally related to the clinical outcome. We therefore sought to identify such CKD biomarkers by applying MR techniques to a comprehensive panel of 237 biomarkers covering cardiovascular, metabolic, and inflammatory processes within the recently completed Outcome Reduction with an Initial Glargine Intervention (ORIGIN) trial that was performed in people with type 2 diabetes or prediabetes.11

Results

Identification of CKD Biomarkers Using MR

Two hundred twenty-seven serum biomarkers were tested for an association with CKD using an MR approach. After removing biomarkers without any significant cis single nucleotide polymorphism (SNP) associations and with minor allele frequency<0.05 (according to CKD genetics consortium [CKDGen]), 197 biomarkers were retained for downstream analysis. After MR analysis, two biomarkers were found to be significantly associated with CKD after Bonferroni correction for multiple hypothesis testing (P<0.05/197), namely uromodulin (UMOD, also known as Tamm–Horsfall glycoprotein) and human EGF receptor 2 (HER2). As noted in Figure 1, the MR analysis suggested a deleterious effect of UMOD (odds ratio [OR], 1.30 per SD; 95% confidence interval [95% CI], 1.25 to 1.35; P<5×10−20, number of SNPs=17) and HER2 (OR, 1.30 per SD; 95% CI, 1.14 to 1.48; P=8.0×10−5, number of SNPs=5) on CKD. All SNPs used in the MR models had P<0.01 and FDR<0.05 for the SNP-biomarker associations (Supplemental Tables 1 and 2). Regional plots of SNP associations with serum UMOD and HER2 at the UMOD and ERBB2 loci, respectively, are depicted in Supplemental Figure 1. To assess for the presence of unmeasured horizontal pleiotropy, we utilized the MR–Egger12 method where the y intercept is allowed to float, rather than be fixed at zero. We found no evidence of pleiotropy as determined by the significance of the y intercept (P>0.05). As a sensitivity analysis, we use a leave-one-out strategy in which MR analyses were repeated, excluding one variant at a time, and consistent estimates were obtained for each SNP excluded (see Supplemental Tables 3 and 4 for full results).

Figure 1.
Figure 1.

UMOD and HER2 identified as novel markers of CKD using MR. Forest plots depict a summary of the MR results for (A) UMOD and (B) HER2 at the UMOD and ERBB2 locus, respectively. A single SNP MR was conducted for each independent SNP (pairwise R2<0.1). ORs were determined by the IVW method by regressing the effect estimates from the CKD association (from CKDGen) on the biomarker association (from ORIGIN). A two-tailed P-value was calculated using a z-test from 100,000 random simulations. IVW, inverse-variance weighted.

Association of UMOD and HER2 Concentration with CKD in ORIGIN

The MR-generated hypothesis that these biomarkers promoted CKD was then tested using the ORIGIN data. We therefore assessed the epidemiologic relationship of baseline UMOD and HER2 with incident CKD. We found that increased levels of blood UMOD were significantly associated with decreased risk of CKD, whereas increased levels of blood HER2 were associated with an increased risk of incident CKD in models adjusting for age, sex, and ethnicity (UMOD: OR, 0.83 per SD; 95% CI, 0.78 to 0.88; P<0.001 and HER2: OR, 1.07 per SD; 95% CI, 1.01 to 1.13; P=0.01). Consistent results were also observed in models fully adjusted for CKD risk factors and are provided in the Supplemental Material. We performed subgroup analyses to assess the consistency of the association of UMOD and HER2 concentration with CKD. No significant interaction across subgroups was observed after adjustment for multiple hypothesis testing (Figure 2).

Figure 2.
Figure 2.

No heterogeneity was observed among subgroups for the effect of serum UMOD and HER2 levels with risk of CKD. Models were adjusted (where appropriate) for age, sex and ethnicity. Subgroups were as follows: age (≤65, >65), sex (male, female), baseline diabetes status (yes/no), obese (BMI≥30, BMI<30), current smoker, hypertension diagnosis, and hypercholesterolemia (LDL-cholesterol>4.5 mmol/L, LDL-cholesterol≤4.5 mmol/L). BMI, body mass index.

Association of UMOD with Kidney Mass in Healthy Patients Undergoing Nephrectomy

We assessed for the possibility that reverse causation could play a role in the apparent discrepant epidemiologic association of UMOD with CKD (OR, 0.83; 95% CI, 0.78 to 0.88) and the evidence for UMOD as a CKD risk factor found in the MR (OR, 1.30; 95% CI, 1.25 to 1.35). Because UMOD is exclusively synthesized in the kidney, we hypothesized that UMOD is linked to a reduced kidney mass, and therefore lower UMOD expression, in patients with CKD. Therefore, the hypothesis that UMOD concentration is a marker of kidney mass, rather than CKD progression, was explored in ten healthy kidney donors. Briefly, participants had to meet clinical criteria for kidney donation; namely, normal BP, nonsmoker, eGFR>80 ml/min, and absence of major chronic disease. Mean age was 49.9 years and 30% were male (see Supplemental Material for further details). Indeed, UMOD blood levels were almost halved after uninephrectomy as compared with the presurgery period (μ=217.7 ng/ml, SD=75.6 versus μ=129.5, SD=39.1; P=7.6×10−5, paired samples t test) (see Figure 3). We also tested UMOD levels in urine (indexed to creatinine) and found a significant, positive correlation between blood and urine levels (R2=0.29, P=0.02).

Figure 3.
Figure 3.

Serum UMOD concentration decreased after nephrectomy in healthy donors. Panel A depicts concentration of blood UMOD concentration (ng/mL) in 10 otherwise healthy nephrectomy patients, before (left) and after (right) surgery. Notably, every patient showed a decrease in UMOD after nephrectomy, (paired t-test P=7.6x10−5). Panel B illustrates the box-plot of UMOD levels at the two time-points.

Identification of Regulators of UMOD and HER2 Using MR

To explore the mechanism by which UMOD and HER2 exert their effect on CKD, we tested for a causal effect of all other biomarkers on UMOD and HER2 levels using MR. Specifically, we investigated whether any of the other biomarkers play a causal role in the regulation of UMOD and HER2 through a similar MR analysis using cis SNPs for the biomarker under study (where possible) as instrumental variables. Significant cis SNPs (P<0.001) were identified for 169 biomarkers (decreased from 197 due to the more stringent threshold applied for IV selection to the one sample MR design) and were thus tested for their effect on UMOD and HER2. No significant biomarkers were found as regulators of UMOD (P<0.05/169). However, after adjusting for multiple hypothesis testing (P<0.05/169), angiotensin-converting enzyme (ACE) was identified as a causal regulator of serum HER2 levels (β=0.13 per SD; 95% CI, 0.08 to 0.16; P=2.5×10−7). We identified 16 independent cis ACE SNPs which were associated with ACE levels at P<0.001 to be used as instrumental variables in the MR analysis. We also tested for pleiotropic effects of UMOD and HER2 with other CV traits using MR, but found no significant associations after multiple hypothesis testing (data not shown).

After the identification of ACE as a causal mediator of HER2, we decided to further explore this relationship by investigating the effect of different classes of BP-lowering medications on HER2 and ACE concentration in ORIGIN. The following medications were assessed as dichotomous (yes/no) variables: ACE inhibitors or angiotensin II receptor blockers (ARBs), diuretics (grouped as one variable), aldosterone inhibitors, β-blockers, and calcium channel blockers. Specifically, a linear model was used, with HER2 or ACE concentration as the dependent variable, and use of medication (yes/no) as the independent variable (Table 1). The models were adjusted for age, sex, ethnicity, hypertension diagnosis, and prior renal disease; HER2 models were further adjusted for serum ACE levels. We identified a significant association between use of ACE inhibitors/ARBs and HER2 concentration, indicating lower levels of HER2 in patients using ACE inhibitors or ARBs (β=0.25 SD decrease with ACE inhibition; 95% CI, −0.30 to −0.20; P<5×10−16), consistent with our MR findings indicating ACE increases HER2 levels. Conversely, no other BP medication was associated with lower levels of ACE after adjusting for multiple hypothesis testing (P<0.05/5); in fact, diuretics showed a marginal increase in HER2 levels consistent with an activation of the renin-angiotensin system with diuretics. Additionally, ACE inhibitors/ARBs, diuretics, and β-blockers were associated with increased levels of ACE, consistent with the RAAS inhibitory feedback loop. Aldosterone blockers and calcium channel blockers showed no effect on ACE levels after adjusting accounting for test multiplicity (P<0.05/5). A summary of ACE/HER2 findings and their effect on CKD risk can be seen in Figure 4.

View this table:
Table 1.

Epidemiologic association of BP medications on HER2 serum levels

Figure 4.
Figure 4.

Proposed mechanism by which HER2 mediates the effects of ACE on CKD risk. Flow chart shows summary of ACE/HER2 results, red lines depict epidemiological associations, and blue lines depict MR associations. Black lines indicate previously known relationships.

Discussion

Using MR to screen a comprehensive panel of 227 blood biomarkers, we identified blood UMOD and HER2 as causal mediators of CKD. UMOD is a kidney-specific protein ubiquitously expressed by the epithelial cells of the thick ascending loop of Henle. Under normal physiologic conditions, UMOD is the most abundant protein found in urine. Despite much research after its discovery over 50 years ago, the role of UMOD in renal physiology remains unclear. Clinical and experimental studies implicated UMOD in several forms of bacteria clearance from the urinary tract and in inflammatory kidney disease, although these findings were inconsistent.13

With the advent of large genome-wide association studies, UMOD has emerged as an important locus in both CKD and hypertension.14–16 A meta-analysis by Olden et al.17 identified an SNP in the UMOD promoter to be strongly associated with UMOD urinary levels, confirming the role of variants at the UMOD locus in UMOD excretion. Furthermore, this same allele was previously identified to be associated with increased CKD risk.18 These results indicate a positive relationship between urinary UMOD and CKD, consistent with our MR findings in blood. Notably, Trudu et al.19 have demonstrated that UMOD overexpression in transgenic mice led to salt-sensitive hypertension and activation of the renal sodium cotransporter NKCC2; this is consistent with their data in humans where pharmacologic inhibition of NKCC2 was found to be more effective in lowering BP in patients homozygous for UMOD risk variants. These findings establish a link between UMOD, hypertension, and CKD. Together with our results, these data point to UMOD as a therapeutic target for lowering BP and preserving renal function. However, it should be noted that it is impossible to know from these results if UMOD is acting through urine, blood, or possibly an intracellular mechanism to exert its risk on CKD. In our epidemiologic analysis, however, increased UMOD levels were associated with a decreased risk of incident CKD, consistent with other studies.20–22 Possible explanations for the divergent direction of effect between the MR and epidemiologic association include confounding and reverse causation. Although adjustment for relevant risk factors did not alter conclusions, our analysis in healthy patients undergoing nephrectomy revealed almost halving of UMOD blood levels after uninephrectomy in healthy donors, consistent with previous studies.23 Although we cannot rule out other biologic and statistical explanations, including residual confounding, these results suggest the protective epidemiologic finding may be a result of reverse causation, reflective of loss of nephron mass in progressive kidney disease.

HER2 is a member of the human EGF receptor (EGFR) family which are key regulators of cellular proliferation. The EGFR family has been implicated in CKD previously because EGFR signaling is involved in renal physiology through nephrogenesis, tissue repair, and electrolyte balance. Numerous experimental studies have shown that pharmacologic and genetic blockade of the EGFR system inhibits renal deterioration and fibrosis in animal models of kidney damage.24 Additionally, in three independent cohorts of patients with CKD, low urine excretion of EGF predicted accelerated loss of kidney function.25 The authors suggested that low urine EGF excretion reflects reserved concentration in the tubules and represents a key factor in CKD progression. EGFR has also been shown to play a role in hypertensive and diabetic nephropathy.26,27 For instance, EGFR expression is increased in the kidneys of hypertensive rats. Similarly, administration of gefitinib, an EGFR–tyrosine kinase inhibitor, improves renal function in rats with hypertension-induced renal disease.28 Furthermore, significant reduction of diabetes-associated glomerular hypertrophy and renal enlargement has been seen upon blockade of EGFR signaling.29 These animal studies are consistent with our MR and with epidemiologic findings suggesting a causal effect of HER2 on CKD progression and development. Currently, HER2 inhibitors are used clinically in EGFR-mediated cancers.30,31 Our data suggest to explore those drugs in models of kidney disease, as others have suggested.32

We identified ACE as a positive regulator of HER2 levels, consistent with the observation of increased EGFR activity in hypertensive rats. Furthermore, we identified a lower concentration of HER2 in patients on ACE inhibitors or ARBs versus those not on these medications, consistent with the hypothesis that ACE does increase HER2 levels. Moreover, as medication control we found no evidence of other BP-lowering medications to have an effect on lowering HER2 levels. Together, these results implicate HER2 as a mediator by which ACE inhibitors and ARBs exert their protective effect on patients with CKD, beyond that of other classes of antihypertensive drugs.33,34 These results signify that blockade of RAAS through ACE inhibitors and ARBs does indeed reduce HER2, and also suggest that HER2 levels may be able to guide RAAS inhibition. Finally, as noted previously, HER2 inhibitors are commonly used to treat EGFR-mediated cancers. Therefore, given these findings, it is possible that inhibition of ACE would decrease cancer risk by decreasing HER2 levels. Indeed, a very large, cross-sectional cohort study, of nearly 300,000 individuals, identified a lower cancer risk in individuals on ACE inhibitors and ARBs as compared with those not on these medications.35 Although these results should be cautiously interpreted and may be confounded, they do suggest a novel application of ACE inhibitors which should be explored in future studies.

These findings are limited by the two-sample MR study design, where genetic estimates were obtained from independent populations. In such a design, weak instrumental variables lead to estimates which are biased toward the null hypothesis, which reduced the likelihood of type 1 error but decreased power.36 Thus not all of the causal biomarkers may have been identified. Additionally, the genetic variants studied may have other effects beyond their effect on the biomarker being studied (i.e., genetic pleiotropy). We mitigated this source of bias by limiting our investigation to variants at or near the gene coding for the biomarker of interest. Furthermore, associated loci were individually inspected for proximity to other potential genes and we did not identify any genes near the UMOD and ERBB2 loci that were plausible sources of pleiotropy.

Identification of CKD risk factors is instrumental to further our understanding of the disease, evaluate its risk, and guide treatment. Using MR, we have investigated a comprehensive panel of biomarkers for involvement in CKD. Our study presents the first MR analysis of blood UMOD and identified HER2 as a novel causal mediator of CKD, consistent with previous model systems and the known biologic role of these biomarkers. We also found compelling evidence to suggest HER2 as a mediator through which ACE inhibitors and ARBs protect again CKD progression and development. Increased serum UMOD and HER2 concentrations represent independent mechanisms leading to CKD, which can be assessed through a simple blood test. These findings pave the way for risk stratification and therapeutic interventions and provide important insights into the pathophysiology of CKD, and its relation to the current clinical practice of ACE inhibitors and ARBs in CKD treatment. Future research should be aimed at identifying the causal mechanisms and whether interventions targeted at reducing UMOD and HER2 levels can reduce CKD.

Concise Methods

Study Population—ORIGIN

The design and findings of the ORIGIN trial have been described in detail.37,38 Briefly 12,537 people with established cardiovascular risk factors who also had diabetes, impaired glucose tolerance, or impaired fasting glucose were studied. After random allocation to two therapies using a factorial design (basal insulin glargine versus standard care and omega 3 fatty acid supplements versus placebo), they were followed for a median of 6.2 years for cardiovascular events and other health outcomes. The ethics committee at each participating site approved the trial, and all participants provided written informed consent. As previously described,38 a subset of 8401 participants from the ORIGIN trial consented to further biologic analysis and were therefore included in the biomarker ORIGIN substudy. Biomarker levels were analyzed using the serum that was drawn at the beginning of the study (a detailed description of biomarker measurement and quality control is found in the Supplemental Material and a complete list of biomarkers analyzed is found in Supplemental Table 5).

A further subset of the 8401 participants in the biomarker substudy also consented to genetic analyses and were genotyped on Illumina’s HumanCore Exome chip, comprising 5078 ORIGIN individuals. Standard quality control measures were assessed. After quality control, the sample consisted of 4147 participants and 284,024 SNPs from two ethnic groups (European and Latin American). Imputation was then performed on the post-QC data to predict unobserved genotypes in the study population. Over 30 million SNPs were imputed, allowing for comprehensive coverage of known genetic variants (a detailed description of quality control and imputation procedures is in the Supplemental Material). Key clinical characteristics of the study populations are shown in Supplemental Table 6.

CKDGen Consortium Data

Genetic data on SNP associations with CKD (defined as eGFRcrea<60 ml min−1 per 1.73 m2) were obtained from the CKDGen database and downloaded from https://www.nhlbi.nih.gov/research/intramural/researchers/ckdgen. Specifically, we used the most recent meta-analysis (released in 2015) of 43 genome-wide association studies with up to 117,165 individuals for CKD (12,385 cases), of European descent.39

SNP Association with Biomarkers and CKD

The analysis was restricted to biomarkers directly encoded by a gene(s) on autosomal chromosomes (i.e., chromosomes 1–22). Thus, removal of five biomarkers because they were products of genes on the X chromosome, and five biomarkers because they were not a direct gene product (e.g., cortisol), left 227 biomarkers for analysis.

SNP selection was carried out in four steps. First, as noted in Supplemental Figure 2, we restricted our analysis for each of the 227 biomarkers to SNPs within 300 kb of the gene(s) encoding the corresponding protein or protein component, hereafter referred to as cis associations. This process identified 1,067,955 SNP/biomarker cis pairs (note that some SNPs are in cis with multiple biomarkers). Second, after removing SNPs not found in the CKDGen database and those with a minor allele frequency <0.05 according to CKDGen, we estimated the relationship between the remaining SNPs and their corresponding cis biomarker(s) in ORIGIN, by regressing each SNP against the concentration of its cis biomarker (with biomarker concentration as the dependent variable and SNP dosage as the independent variable). In other words, for each biomarker we only tested SNPs near the respective encoding gene(s). The regression models were first computed in each ethnic group separately, adjusting for age, sex, and the first five principal components, using SNPtest.40 The ethnic-specific models were then meta-analyzed across the two ethnicities using fixed effects models to minimize the risk of confounding caused by population stratification. Third, cis SNPs with a biomarker association of P<0.01 were selected. Finally, SNPs were pruned for linkage disequilibrium at a stringent threshold of r2<0.1 using the 1000 Genomes data (Europeans) to ensure associations retained for MR analysis were nonredundant. SNPs were selectively prioritized on the basis of the significance of the association with their biomarker. For each biomarker, the cis SNP with the most significant association with the biomarker was first retained and all SNPs in linkage (r2>0.1) with that SNP removed. This process was then repeated for any remaining SNPs. Thus, 1307 cis SNP/biomarker associations remained after pruning. SNP filtering was performed in R (version 3.0.1) and PLINK was used to calculate R2 statistics in 1000 genomes. A summary of the SNP and biomarker selection can be found in Supplemental Figure 3.

Identification of Blood Mediators of CKD Using MR

A two-sample MR was performed on the 197 biomarkers which had at least one significant cis SNP (P<0.01) and that were also found in the CKDGen data. Input variables for the MR analysis for each biomarker were (1) the β coefficients of the SNPs on their cis biomarker that were estimated using the regression models above, and (2) the β coefficients of the SNPs on CKD that were estimated from the CKDGen (Supplemental Figure 4). MR associations were performed using the inverse-variance weighted (IVW) method by regressing genetic effect estimates for CKD (dependent variable) on genetic effect estimates of biomarkers.41 To determine significance, a bootstrap method was used under the null hypothesis of no effect between CKD and biomarkers. Predicted effects on CKD were sampled from a normal distribution with mean and SDs as determined from CKDGen. A two-tailed P value was calculated using a z-test from 100,000 random simulations. In other words, CKD estimates for each SNP were sampled 100,000 times and regressed onto the corresponding β estimates from ORIGIN (this procedure is equivalent to the IVW fixed effect method). Biomarkers were deemed significant after adjusting for multiple testing hypothesis (P<0.05/197).

Association of Biomarker Levels with Incident CKD in ORIGIN

Once significant biomarkers were identified by the MR analysis, we tested whether biomarker levels showed a consistent association with incident CKD in 8197 ORIGIN participants with ethnicity information and biomarker levels. CKD was defined as the composite of either doubling of serum creatinine, worsening in albuminuria category, RRT, or death due to end stage renal failure. This was assessed using logistic models with incident CKD as the dependent variable and biomarker concentration as the independent variable of interest. Models were adjusted for age, sex, and ethnicity to remain consistent with MR model adjustment. We also tested models after further adjusting for prior type 2 diabetes, prior renal disease, BMI, current smoker, diagnosis of hypertension, baseline eGFR, and LDL. Subgroup analyses were performed to test for heterogeneity between groups using models adjusted for age, sex, and ethnicity (where appropriate).

Identification of Regulators of CKD Biomarkers Using MR

To gain further biologic insights regarding the novel CKD biomarkers, a second set of MR analyses were then performed to explore whether the levels of the novel CKD-causing biomarkers were determined by any of the other biomarkers. Specifically, we tested all biomarkers for an effect on both UMOD and HER2 levels (i.e., the novel biomarkers identified in the CKD MR), where the input variables for each MR were (1) the effect of SNPs on their cis biomarker as the independent variable (where possible), and (2) the effect of the same set of SNPs on the novel CKD biomarkers as the dependent variable. Because of the fact that both of these estimates were obtained from ORIGIN (i.e., one-sample), a more conservative significance threshold of P<0.001 (or F>10) was applied for the inclusion of SNPs into the MR model because weak instruments can bias results toward false-positives.42 Statistical analyses were performed using R (version 3.0.1), unless stated otherwise. A summary of the analysis plan can be seen in Figure 5.

Figure 5.
Figure 5.

Flow chart depicts a summary of the four analyses conducted and their respective sample sizes.

Disclosures

H.G. has received consulting fees from Sanofi, Novo Nordisk, Lilly, AstraZeneca, Boehringer Ingelheim, and GlaxoSmith-Kline and support for research or continuing education through his institution from Sanofi, Lilly, Takeda, Novo Nordisk, Boehringer, Ingelheim, and AstraZeneca. G.P. has received consulting fees from Sanofi, Bristol Myers Squibb, Lexicomp, and Amgen and support for research through his institution from Sanofi. S.H. is an employee of Sanofi. J.F.E.M. has received consulting fees from Novo Nordisk, AstraZeneca, Amgen, Braun, ACI, Fresenius, Celgene, Gambro, Abbvie, Roche, Sandoz, Lanthio, Sanifit, Relypsam, and ZS Pharma; and grants from the European Union, McMaster University Canada, Abbvie, Medice, Novo Nordisk, Roche, and Sandoz. J.S., D.T., and M.W. report no conflicts.

Acknowledgments

We are thankful to all of the participants having agreed to contribute to this project, and to the CKDGen for making their data available.

The ORIGIN trial and biomarker project were supported by Sanofi and the Canadian Institutes of Health Research (CIHR) (award 125794).

Sanofi was not involved in the design or conduct of the paper, but provided comments on the manuscript. CIHR had no role in the conduct of the study or the manuscript. Clinicaltrials.gov: NCT00069784.

J.S., H.G., and G.P. designed the study, planned the analyses, interpreted the results, and wrote the first draft of the report. J.S. performed the statistical and bioinformatics analyses. All authors contributed to the critical reading and revision of the manuscript.

Footnotes

References

  1. ↵
    1. James MT,
    2. Hemmelgarn BR,
    3. Tonelli M
    : Early recognition and prevention of chronic kidney disease. Lancet 375: 1296–1309, 2010pmid:20382326
    OpenUrlCrossRefPubMed
  2. ↵
    1. Shlipak MG,
    2. Day EC
    : Biomarkers for incident CKD: A new framework for interpreting the literature. Nat Rev Nephrol 9: 478–483, 2013pmid:23752888
    OpenUrlCrossRefPubMed
  3. ↵
    1. Sekula P,
    2. Del Greco M F,
    3. Pattaro C,
    4. Köttgen A
    : Mendelian randomization as an approach to assess causality using observational data. J Am Soc Nephrol 27: 3253–3265, 2016pmid:27486138
    OpenUrlAbstract/FREE Full Text
  4. ↵
    1. Thanassoulis G,
    2. O’Donnell CJ
    : Mendelian randomization: Nature’s randomized trial in the post-genome era. JAMA 301: 2386–2388, 2009pmid:19509388
    OpenUrlCrossRefPubMed
  5. ↵
    1. Ference BA,
    2. Majeed F,
    3. Penumetcha R,
    4. Flack JM,
    5. Brook RD
    : Effect of naturally random allocation to lower low-density lipoprotein cholesterol on the risk of coronary heart disease mediated by polymorphisms in NPC1L1, HMGCR, or both: A 2 × 2 factorial Mendelian randomization study. J Am Coll Cardiol 65: 1552–1561, 2015pmid:25770315
    OpenUrlFREE Full Text
  6. ↵
    1. Clarke R,
    2. Peden JF,
    3. Hopewell JC,
    4. Kyriakou T,
    5. Goel A,
    6. Heath SC, et al.; PROCARDIS Consortium
    : Genetic variants associated with Lp(a) lipoprotein level and coronary disease. N Engl J Med 361: 2518–2528, 2009pmid:20032323
    OpenUrlCrossRefPubMed
    1. Teslovich TM,
    2. Musunuru K,
    3. Smith AV,
    4. Edmondson AC,
    5. Stylianou IM,
    6. Koseki M, et al.
    : Biological, clinical and population relevance of 95 loci for blood lipids. Nature 466: 707–713, 2010pmid:20686565
    OpenUrlCrossRefPubMed
  8. ↵
    1. Sarwar N,
    2. Butterworth AS,
    3. Freitag DF,
    4. Gregson J,
    5. Willeit P,
    6. Gorman DN, et al.; IL6R Genetics Consortium Emerging Risk Factors Collaboration
    : Interleukin-6 receptor pathways in coronary heart disease: A collaborative meta-analysis of 82 studies. Lancet 379: 1205–1213, 2012pmid:22421339
    OpenUrlCrossRefPubMed
  9. ↵
    1. Del Greco M F,
    2. Foco L,
    3. Pichler I,
    4. Eller P,
    5. Eller K,
    6. Benyamin B, et al.; Genetics of Iron Status Consortium; CKDGen Consortium
    : Serum iron level and kidney function: A Mendelian randomization study. Nephrol Dial Transplant 32: 273–278, 2017pmid:28186534
    OpenUrlPubMed
  10. ↵
    1. Verduijn M,
    2. Prein RA,
    3. Stenvinkel P,
    4. Carrero JJ,
    5. le Cessie S,
    6. Witasp A, et al.
    : Is fetuin-A a mortality risk factor in dialysis patients or a mere risk marker? A Mendelian randomization approach. Nephrol Dial Transplant 26: 239–245, 2011pmid:20605840
    OpenUrlCrossRefPubMed
  11. ↵
    1. Gerstein HC,
    2. Yusuf S,
    3. Riddle MC,
    4. Ryden L,
    5. Bosch J; Origin Trial Investigators
    : Rationale, design, and baseline characteristics for a large international trial of cardiovascular disease prevention in people with dysglycemia: The ORIGIN Trial (Outcome Reduction with an Initial Glargine Intervention). Am Heart J 155: 26–32, e1–6, 2008
    OpenUrlPubMed
  12. ↵
    1. Bowden J,
    2. Davey Smith G,
    3. Burgess S
    : Mendelian randomization with invalid instruments: Effect estimation and bias detection through Egger regression. Int J Epidemiol 44: 512–525, 2015pmid:26050253
    OpenUrlCrossRefPubMed
  13. ↵
    1. Scolari F,
    2. Izzi C,
    3. Ghiggeri GM
    : Uromodulin: From monogenic to multifactorial diseases. Nephrol Dial Transplant 30: 1250–1256, 2015pmid:25228753
    OpenUrlCrossRefPubMed
  14. ↵
    1. Padmanabhan S,
    2. Melander O,
    3. Johnson T,
    4. Di Blasio AM,
    5. Lee WK,
    6. Gentilini D, et al.; Global BPgen Consortium
    : Genome-wide association study of blood pressure extremes identifies variant near UMOD associated with hypertension. PLoS Genet 6: e1001177, 2010pmid:21082022
    OpenUrlCrossRefPubMed
    1. Köttgen A,
    2. Hwang S-J,
    3. Larson MG,
    4. Van Eyk JE,
    5. Fu Q,
    6. Benjamin EJ, et al.
    : Uromodulin levels associate with a common UMOD variant and risk for incident CKD. J Am Soc Nephrol 21: 337–344, 2010pmid:19959715
    OpenUrlAbstract/FREE Full Text
  16. ↵
    1. Pattaro C,
    2. Köttgen A,
    3. Teumer A,
    4. Garnaas M,
    5. Böger CA,
    6. Fuchsberger C, et al.; CARDIoGRAM Consortium; ICBP Consortium; CARe Consortium; Wellcome Trust Case Control Consortium 2 (WTCCC2)
    : Genome-wide association and functional follow-up reveals new loci for kidney function. PLoS Genet 8: e1002584, 2012pmid:22479191
    OpenUrlCrossRefPubMed
  17. ↵
    1. Olden M,
    2. Corre T,
    3. Hayward C,
    4. Toniolo D,
    5. Ulivi S,
    6. Gasparini P, et al.
    : Common variants in UMOD associate with urinary uromodulin levels: A meta-analysis. J Am Soc Nephrol 25: 1869–1882, 2014pmid:24578125
    OpenUrlAbstract/FREE Full Text
  18. ↵
    1. Köttgen A,
    2. Glazer NL,
    3. Dehghan A,
    4. Hwang S-J,
    5. Katz R,
    6. Li M, et al.
    : Multiple loci associated with indices of renal function and chronic kidney disease. Nat Genet 41: 712–717, 2009pmid:19430482
    OpenUrlCrossRefPubMed
  19. ↵
    1. Trudu M,
    2. Janas S,
    3. Lanzani C,
    4. Debaix H,
    5. Schaeffer C,
    6. Ikehata M, et al.; SKIPOGH team
    : Common noncoding UMOD gene variants induce salt-sensitive hypertension and kidney damage by increasing uromodulin expression. Nat Med 19: 1655–1660, 2013pmid:24185693
    OpenUrlCrossRefPubMed
  20. ↵
    1. Steubl D,
    2. Block M,
    3. Herbst V,
    4. Nockher WA,
    5. Schlumberger W,
    6. Satanovskij R, et al.
    : Plasma uromodulin correlates with kidney function and identifies early stages in chronic kidney disease patients. Medicine (Baltimore) 95: e3011, 2016
    OpenUrlCrossRefPubMed
    1. Delgado GE,
    2. Kleber ME,
    3. Scharnagl H,
    4. Krämer BK,
    5. März W,
    6. Scherberich JE
    : Serum uromodulin and mortality risk in patients undergoing coronary angiography. J Am Soc Nephrol 28: 2201–2210, 2017pmid:28242751
    OpenUrlAbstract/FREE Full Text
  22. ↵
    1. Devuyst O,
    2. Olinger E,
    3. Rampoldi L
    : Uromodulin: From physiology to rare and complex kidney disorders. Nat Rev Nephrol 13: 525–544, 2017pmid:28781372
    OpenUrlCrossRefPubMed
  23. ↵
    1. Pruijm M,
    2. Ponte B,
    3. Ackermann D,
    4. Paccaud F,
    5. Guessous I,
    6. Ehret G, et al.
    : Associations of urinary uromodulin with clinical characteristics and markers of tubular function in the general population. Clin J Am Soc Nephrol 11: 70–80, 2016pmid:26683888
    OpenUrlAbstract/FREE Full Text
  24. ↵
    1. Tang J,
    2. Liu N,
    3. Zhuang S
    : Role of epidermal growth factor receptor in acute and chronic kidney injury. Kidney Int 83: 804–810, 2013pmid:23325080
    OpenUrlCrossRefPubMed
  25. ↵
    1. Ju W,
    2. Nair V,
    3. Smith S,
    4. Zhu L,
    5. Shedden K,
    6. Song PXK, et al.; ERCB, C-PROBE, NEPTUNE, and PKU-IgAN Consortium
    : Tissue transcriptome-driven identification of epidermal growth factor as a chronic kidney disease biomarker. Sci Transl Med 7: 316ra193, 2015pmid:26631632
    OpenUrlAbstract/FREE Full Text
  26. ↵
    1. Swaminathan N,
    2. Vincent M,
    3. Sassard J,
    4. Sambhi MP
    : Elevated epidermal growth factor receptor levels in hypertensive Lyon rat kidney and aorta. Clin Exp Pharmacol Physiol 23: 793–796, 1996pmid:8911715
    OpenUrlCrossRefPubMed
  27. ↵
    1. Benter IF,
    2. Canatan H,
    3. Benboubetra M,
    4. Yousif MHM,
    5. Akhtar S
    : Global upregulation of gene expression associated with renal dysfunction in DOCA-salt-induced hypertensive rats occurs via signaling cascades involving epidermal growth factor receptor: A microarray analysis. Vascul Pharmacol 51: 101–109, 2009pmid:19410658
    OpenUrlCrossRefPubMed
  28. ↵
    1. François H,
    2. Placier S,
    3. Flamant M,
    4. Tharaux P-L,
    5. Chansel D,
    6. Dussaule J-C, et al.
    : Prevention of renal vascular and glomerular fibrosis by epidermal growth factor receptor inhibition. FASEB J 18: 926–928, 2004pmid:15033924
    OpenUrlCrossRefPubMed
  29. ↵
    1. Wassef L,
    2. Kelly DJ,
    3. Gilbert RE
    : Epidermal growth factor receptor inhibition attenuates early kidney enlargement in experimental diabetes. Kidney Int 66: 1805–1814, 2004pmid:15496151
    OpenUrlCrossRefPubMed
  30. ↵
    1. Govindan R
    : A review of epidermal growth factor receptor/HER2 inhibitors in the treatment of patients with non-small-cell lung cancer. Clin Lung Cancer 11: 8–12, 2010pmid:20085861
    OpenUrlPubMed
  31. ↵
    1. Schroeder RL,
    2. Stevens CL,
    3. Sridhar J
    : Small molecule tyrosine kinase inhibitors of ErbB2/HER2/Neu in the treatment of aggressive breast cancer. Molecules 19: 15196–15212, 2014pmid:25251190
    OpenUrlCrossRefPubMed
  32. ↵
    1. Melenhorst WB,
    2. Mulder GM,
    3. Xi Q,
    4. Hoenderop JGJ,
    5. Kimura K,
    6. Eguchi S, et al.
    : Epidermal growth factor receptor signaling in the kidney: Key roles in physiology and disease. Hypertension 52: 987–993, 2008
    OpenUrlFREE Full Text
  33. ↵
    1. Sarafidis PA,
    2. Khosla N,
    3. Bakris GL
    : Antihypertensive therapy in the presence of proteinuria. Am J Kidney Dis 49: 12–26, 2007pmid:17185142
    OpenUrlCrossRefPubMed
  34. ↵
    1. Molnar MZ,
    2. Kalantar-Zadeh K,
    3. Lott EH,
    4. Lu JL,
    5. Malakauskas SM,
    6. Ma JZ, et al.
    : Angiotensin-converting enzyme inhibitor, angiotensin receptor blocker use, and mortality in patients with chronic kidney disease. J Am Coll Cardiol 63: 650–658, 2014pmid:24269363
    OpenUrlFREE Full Text
  35. ↵
    1. Chiang YY,
    2. Chen KB,
    3. Tsai TH,
    4. Tsai WC
    : Lowered cancer risk with ACE inhibitors/ARBs: A population-based cohort study. J Clin Hypertens (Greenwich) 16: 27–33, 2014pmid:24304931
    OpenUrlCrossRefPubMed
  36. ↵
    1. Inoue A,
    2. Solon G
    : Two-sample instrumental variables estimators. Rev Econ Stat 92: 557–561, 2010
    OpenUrlCrossRef
  37. ↵
    Origin Trial Investigators: Rationale, design, and baseline characteristics for a large international trial of cardiovascular disease prevention in people with dysglycemia: the ORIGIN Trial (Outcome Reduction with an Initial Glargine Intervention). Am Heart J 155: 26–32, 2008pmid:26518765
    OpenUrlPubMed
  38. ↵
    1. Bosch J,
    2. Gerstein HC,
    3. Dagenais GR,
    4. Díaz R,
    5. Dyal L,
    6. Jung H, et al.;
    ORIGIN Trial Investigators: n-3 fatty acids and cardiovascular outcomes in patients with dysglycemia. N Engl J Med 367: 309–318, 2012pmid:26518765
    OpenUrlCrossRefPubMed
  39. ↵
    1. Gerstein HC,
    2. Paré G,
    3. McQueen MJ,
    4. Haenel H,
    5. Lee SF,
    6. Pogue J, et al.; Outcome Reduction With Initial Glargine Intervention Trial Investigators
    : Identifying novel biomarkers for cardiovascular events or death in people with dysglycemia. Circulation 132: 2297–2304, 2015pmid:26518765
    OpenUrlAbstract/FREE Full Text
  40. ↵
    1. Pattaro C,
    2. Teumer A,
    3. Gorski M,
    4. Chu AY,
    5. Li M,
    6. Mijatovic V, et al.; ICBP Consortium; AGEN Consortium; CARDIOGRAM; CHARGe-Heart Failure Group; ECHOGen Consortium
    : Genetic associations at 53 loci highlight cell types and biological pathways relevant for kidney function. Nat Commun 7: 10023, 2016pmid:26831199
    OpenUrlCrossRefPubMed
  41. ↵
    1. Marchini J,
    2. Howie B
    : Genotype imputation for genome-wide association studies. Nat Rev Genet 11: 499–511, 2010pmid:20517342
    OpenUrlCrossRefPubMed
  42. ↵
    1. Burgess S,
    2. Butterworth A,
    3. Thompson SG
    : Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol 37: 658–665, 2013pmid:24114802
    OpenUrlCrossRefPubMed
    1. Burgess S,
    2. Thompson SG; CRP CHD Genetics Collaboration
    : Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol 40: 755–764, 2011pmid:21414999
    OpenUrlCrossRefPubMed
Back to top

In this issue

View Selected Citations (0)
Print
Download PDF
Email Article
Citation Tools
Request Permissions
Blood HER2 and Uromodulin as Causal Mediators of CKD
Jennifer Sjaarda, Hertzel C. Gerstein, Salim Yusuf, Darin Treleaven, Michael Walsh, Johannes F.E. Mann, Sibylle Hess, Guillaume Paré
JASN Apr 2018, 29 (4) 1326-1335; DOI: 10.1681/ASN.2017070812
del.icio.us logo Digg logo Reddit logo Twitter logo CiteULike logo Facebook logo Google logo Mendeley logo

Jump to section

Keywords