Main

End-stage kidney disease seems to partly originate in early life (1,2). Smaller kidneys in early life with a reduced number of nephrons lead to glomerular hyperfiltration and sclerosis and predispose to kidney disease in adulthood (3,4). Also, risk factors for impaired kidney function track from childhood to adulthood (5,6). Early kidney growth and development is a complex developmental process and is influenced by many environmental factors (7,8). Various nutritional and environmental exposures have been suggested to affect early kidney development and may subsequently influence the risk of kidney disease in later life (9,10).

However, the associations between early life kidney characteristics and kidney disease in later life may also be explained by common genetic variants. These variants may affect early kidney development and function and thereby predispose individuals to kidney disease in later life. Several genes are known to be involved in kidney development (8). Rare mutations in these genes can cause severe anomalies, such as agenesis or dysgenesis of the kidney (7). These mutations do not explain variation in kidney development within normal range. However, large genome-wide association studies (GWAS) in adults have identified 30 common genetic variants associated with impaired kidney function (11,12,13). For this study, we hypothesized that these genetic variants also affect kidney growth and function in early childhood, and thereby partly explain the well-known associations of early life kidney measures with kidney disease in later life.

To test this hypothesis, we examined, in a population-based prospective cohort study among 4,119 children, the associations of genetic risk scores combining previously identified common genetic variants related to impaired kidney function in adults with kidney outcomes in school-age children.

Results

Participant Characteristics

The flowchart of study participants is given in the Figure 1 . Table 1 shows the characteristics of the participants. In the full group, the mean combined kidney volume and eGFRcreat were 120 cm3 (SD ± 23.3) and 119 ml/min/1.73 m2 (SD ± 16), respectively. Of all the children, 7.9% had microalbuminuria and 0.07% had an eGFR <60 ml/min/1.73 m2. The characteristics of the subgroup of European children only were similar to those of the full group (Supplementary Table S1 online).

Figure 1
figure 1

Flow chart of the study participants.

PowerPoint slide

Table 1 Subject characteristics (N = 4,119)

Individual Genetic Variants and Childhood Kidney Outcomes

Results for the associations of the individual single-nucleotide polymorphisms (SNPs) with childhood kidney outcomes are given in the Supplementary Material online. Of the 20 available SNPs known to be associated with eGFRcreat and the single SNP associated with microalbuminuria in adults, none were individually significantly associated with kidney volume, eGFRcreat, eGFRcystC, and microalbuminuria, in the full group (Supplementary Table S2 online). In both, the full group and the European ancestry subgroup, rs3925584, located upstream of MPPED2, was associated with childhood creatinine blood levels (P values 1.8 × 10−3 and 5.6 × 10−4, respectively). In addition, in the European ancestry subgroup, rs11078902 in CDK12 was associated with childhood creatinine blood levels and with eGFRcreat (P values 1.1 × 10−3 and 2.3 × 10−3, respectively) (Supplementary Tables S3–S5 online). The 20 adult kidney function SNPs together explained 1.3, 1.4, and 0.6% of the variance in combined kidney volume, eGFRcreat, and eGFRcystC in childhood, respectively.

Genetic Risk Score and Childhood Kidney Outcomes

Table 2 presents the associations of the weighted genetic risk score based on 20 SNPs known to be associated with impaired adult eGFRcreat with childhood kidney outcomes. The weighted risk score ranged from 13 to 33 with a mean of 22.8 (SD 2.8) and was significantly associated with childhood combined kidney volume (P value 3.0 × 10−3). For each additional average risk allele, combined kidney volume was −0.39 cm3 (95% confidence interval (CI): −0.64, −0.13) smaller. The difference in combined kidney volume between the lowest and highest risk categories (≤ 22 and ≥ 33 risk alleles, respectively) was 2.62 cm3 ( Figure 2a ), which corresponds with a 2.2% difference. A higher weighted genetic risk score was also associated with a lower eGFRcreat (P value 4.0 × 10−4). For each additional average risk allele, eGFRcreat was −0.38 ml/min/1.73m2 (95% CI: −0.60; −0.17) lower. The difference in mean eGFRcreat between the lowest and highest risk categories was 4.46 ml/min/1.73 m2 ( Figure 2b ), which corresponds with a 3.7% difference. Similarly, per additional average risk allele of the adult risk score, childhood eGFRcystC was −0.27 ml/min/1.73 m2 (95% CI: −0.47; −0.07) lower. The difference in mean eGFRcystC between the lowest and highest risk categories was 3.38 ml/min/1.73m2 ( Figure 2c ). We did not observe an association of the weighted risk score with the risk of childhood microalbuminuria (data not shown).

Table 2 Associations of the genetic risk score based on adult kidney function with childhood kidney outcomes (N = 4,119)a
Figure 2
figure 2

Effect of the weighted adult eGFRcreat genetic risk score on childhood kidney outcomes (N = 4,119). (ac) Effect of the weighted adult eGFRcreat genetic risk score on childhood kidney outcomes (N = 4,119). (a) Effect of adult eGFR creatinine-based genetic risk score on childhood combined kidney volume: The x-axis gives the categories of the risk score (overall sum of risk alleles, weighted by previously reported effect sizes, rescaled and rounded to the nearest integer). The left y-axis gives the number of individuals in each risk-score category, whereas the right y-axis gives the mean combined kidney volume. The histogram reflects the observed number of individuals in each risk-score category. The dots and regression line reflect the estimated mean combined kidney volume for each risk-score category. P value is based on the continuous risk score, as presented in Table 2. **P value 3.0 × 10−3. (b) Effect of adult eGFR creatinine-based genetic risk score on childhood eGFR creatinine-based: The x-axis gives the categories of the risk score (overall sum of risk alleles, weighted by previously reported effect sizes, rescaled and rounded to the nearest integer). The left y-axis gives the number of individuals in each risk-score category, whereas the right y-axis gives the mean eGFRcreat. The histogram reflects the observed number of individuals in each risk-score category. The dots and regression line reflect the estimated mean eGFRcreat for each risk-score category. P value is based on the continuous risk score, as presented in Table 2. **P value 4.0 × 10−4. (c) Effect of adult eGFR creatinine-based genetic risk score on childhood eGFR cystatin C-based: The x-axis gives the categories of the risk score (overall sum of risk alleles, weighted by previously reported effect sizes, rescaled and rounded to the nearest integer). The left y-axis gives the number of individuals in each risk-score category, whereas the right y-axis gives the mean eGFRcystC. The histogram reflects the observed number of individuals in each risk-score category. The dots and regression line reflect the estimated mean eGFRcystC for each risk-score category. P value is based on the continuous risk score, as presented in Table 2. **P value 8.0 × 10−3.

PowerPoint slide

Results from the unweighted risk score (mean 21.8, SD 2.8) were similar to those of the weighted score ( Table 2 ). Supplementary Table S6 online shows that higher genetic risk scores were also associated with higher childhood creatinine blood levels and cystatin C blood levels (P values < 0.05). Results in the subgroup of only European children tended to be similar to those in the full group, although the association of both risk scores with eGFRcystC lost its significance and the unweighted risk score lost its significance with combined kidney volume in this group (Supplementary Table S7 online). The effect of the genetic risk score on combined kidney volume did not change after adjusting our model additionally for child height (Supplementary Table S8 online).

The genetic risk scores including 27 SNPs, with 7 additional SNPs related to creatinine metabolism in adults, and the genetic risk score including 11 SNPs which may have a known function related to kidney development, revealed similar results as our main risk score. Results are shown in Supplementary Tables S9 and S10 online. In the Supplementary Table S11 online, the effect of the unweighted risk score which includes the additional 2 SNPs related to eGFRcystC is shown. The results did not change, but the effect estimates on eGFRcystC and cystatin C were larger.

Discussion

We observed in a large population-based prospective cohort study, that a higher genetic risk score combining previously identified common genetic variants related to lower kidney function in adults was also associated with a smaller combined kidney volume and lower eGFRcreat and eGFRcystC in childhood.

Previous GWAS have identified many common genetic variants related to impaired kidney function in adults (11,12,13). Of the 20 available SNPs known to be associated with adult eGFRcreat and the single SNP associated with adult microalbuminuria, only 2 were individually associated with childhood kidney outcomes. First, we observed an association of rs3925584, located upstream of MPPED2, with creatinine blood levels both in the full group and in the subgroup of European children only. In zebrafish, knockdown of MPPED2 caused abnormal podocyte anatomy (11). Rs3925584 has also been associated with magnesium levels in a large GWAS (14). Rs3925584 is also located close to DCDC5, which has been associated with bone mineral density (15). Second, rs11078902 in CDK12 was associated with eGFRcreat and creatinine blood levels in the subgroup of European children only (11). CDK12, cyclin-dependent kinase 12, regulates the expression of genes involved in DNA repair and is required for the maintenance of genomic stability. CDK inhibitors have been described to play a role in human glomerular disease (16). All these associations were directionally consistent with the results reported in previous GWAS among adults (11,13). None of the included common genetic variants were associated with childhood microalbuminuria. The results for the associations of individual SNPs with childhood kidney outcomes should be interpreted carefully. The original GWAS meta-analysis discovery studies in adults were performed in much larger samples than the current study and our negative results may have occurred due to a lack of power.

In the current study, the weighted genetic risk score related to impaired kidney function in adults was associated with smaller combined kidney volume and lower eGFR in childhood. For an average child in our population, the observed effect estimates correspond with a 0.3% smaller combined kidney volume, a 0.3% lower eGFRcreat, and a 0.3% lower eGFRcystC per each additional risk allele. We did not observe an association of the genetic risk score with microalbuminuria. We found similar results for the weighted and unweighted risk genetic scores, suggesting that the adult weights used in this score did not strongly influence the observed associations. Adding to the risk score 7 SNPs previously reported to play a role in creatinine production and secretion revealed similar results. Restricting the genetic risk score to the 11 SNPs with a likely or known function related to kidney development revealed similar results with slightly stronger effect estimates. This restricted genetic risk score may be more specific to our outcomes because of the young age range. Including in the unweighted risk score two additional SNPs known to be associated with eGFRcystC (13,17), showed similar associations, with stronger effect estimates on eGFRcystC and cystatin C, suggesting a strong influence of the two added SNPs on these two markers. Ideally, the weights in the genetic risk score would be child-specific, as these best reflect the relative importance of the variants in children. For this, we would need results from large-scale GWAS with independent replication samples on childhood kidney function outcomes. However, a GWAS focused on kidney function among healthy children is not available yet. A recent paper described a GWAS to identify common genetic variants influencing renal function in children with chronic kidney disease (CKD). Although no genome-wide significant associations were found, a number of potentially interesting subthreshold SNPs were identified (18). This underlines the need for large-scale GWAS efforts to discover the genetic background of kidney function in children, both with and without CKD.

We hypothesized that common genetic variants may partly explain the associations of early life kidney measures with kidney function in later life. The observed associations of the genetic risk scores related to kidney function in adults with kidney outcomes in childhood support this hypothesis. However, confirmation of our findings in other children cohorts and further functional studies are required to establish the causal genes and the mechanisms underlying the associations of these variants with kidney function in childhood and disease in adulthood. In the same cohort as the current study, we have reported associations of smaller kidneys with lower kidney function (19). To explore if the associations of the genetic risks score with kidney function outcomes were explained by kidney volume, we additionally adjusted the kidney function measures for combined kidney volume. However, the results did not change. Also, adding childhood height to the models associating the genetic risk score with combined kidney volume did not materially change the results. The observed effect estimates in the present study are small, but important from an etiological perspective. They provide further insights into pathways leading to changes in kidney function from the earliest phase of life. Whether common genetic variants related to kidney outcomes in childhood and disease in adulthood are clinically useful for identification of groups at risk needs further study.

Some methodological issues need to be considered. Major strengths of the current study are the detailed phenotypes and the relatively large number of subjects. Still, our sample size was too small to identify many of the associations of individual common genetic variants with childhood kidney outcomes, assuming similar effect sizes as in adults. It has been shown that kidney function and the prevalence of chronic kidney disease vary across ethnic groups in adults (20). Our study population was multi-ethnic, but a sensitivity analysis in European children only, as the largest ethnic subgroup, revealed similar results as in the full group. The other ethnic subgroups were too small to analyze individually. Of all children with genetic information, data on kidney outcomes was available in 72%. As compared to children with kidney follow-up measurements, those without these measurements had a lower birth weight (P value < 0.05) (data not shown). A selective loss to follow-up may have reduced variation in birth weight and early kidney size and therefore reduced the power to detect differences. We performed detailed measurements of childhood kidney outcomes. Kidney size correlates with the number of glomeruli and can be used as a measure of kidney development (21). In children the estimation of GFR is challenging. Blood creatinine is most commonly used to calculate eGFR. In addition to blood creatinine levels, we also calculated eGFR based on cystatin C levels using Zappitelli’s formula (22,23). It has been suggested that blood cystatin C levels might be a better biomarker to estimate GFR because the production rate is constant, it is freely filtered, and less dependent on child weight, height, and sex compared to creatinine (24,25). Finally, in our study, we have captured the vast majority of, but not all, known SNPs related to kidney function, as not all SNPs were available in the GWAS dataset.

Conclusion

Our results suggest that common genetic variants related to kidney function in adults could influence kidney structure and function from early childhood onwards. The previously observed associations of early life kidney measures with kidney disease in later life seem to be partly explained by common genetic variants.

Methods

Design and Study Population

This study was embedded in the Generation R Study, a multiethnic population-based prospective cohort study from fetal life onwards in Rotterdam, the Netherlands, which has been described in detail previously (26). The study has been approved by Medical Ethical Committee of Erasmus MC, University Medical Center Rotterdam. All children were born between April 2002 and January 2006 and form a largely prenatally enrolled birth cohort that is currently being followed until young adulthood. Written consent was obtained for every participant. A genome-wide association screen was available in 5,732 children (see below for details). The present analyses were performed among 4,119 singleton live births for whom we have detailed information on kidney outcomes at the median age of 6 y (95% range 5.7, 7.8). The flowchart of study participants is given in the Figure 1 .

Genetic Variants

DNA was isolated from cord blood samples. If DNA samples from cord blood were missing (in 6.3% of the participants), DNA was isolated from blood samples at follow-up measurements. Genome-wide association arrays were run using the Illumina 610 Quad and 660 platforms (27). A stringent process of quality control was applied. Individuals with low sample call rates or sex mismatches were excluded. Before imputation, SNPs were excluded in case of high levels of missing data (SNP call rate <98%), highly significant departures from Hardy-Weinberg equilibrium (P < 1 × 10−6), or low minor allele frequencies (< 1%) (26). MACH software was used to impute genotypes to the cosmopolitan panel of HapMap II (release 22) (28,29). Three SNPs (rs6431731, rs2453580, and rs881858) had imputation qualities between 0.68 and 0.9, indicating reasonable imputation. All other SNPs had imputation qualities above 0.9, indicating good imputation. Based on previous studies, we identified 30 SNPs robustly associated with adult kidney function: 29 for adult eGFRcreat and 1 for adult microalbuminuria (11,12,13). Of these 30 SNPs, information on 26 was available in the GWAS. Information on rs11078903, rs7805747, rs2279463, and rs7422339, all previously associated with adult eGFRcreat, was not available. However, rs3798156 was used as a perfect proxy (R2 = 1) for rs2279463 and rs11078902 was used as a perfect proxy for rs11078903. No perfect proxies were available in the GWAS dataset for the other two SNPs. As the focus of this study was to examine SNPs related to estimated glomerular filtration rate and not to creatinine metabolism, 20 out of these 28 SNPs were used for the genetic risk score analyses. The individual SNPs used are shown in Supplementary Table S2 online.

Childhood Kidney Outcomes

Kidney outcomes were assessed at a dedicated research center in the Erasmus MC—Sophia Children’s Hospital in Rotterdam by well-trained staff (30). We measured kidney volume with ultrasound, using an ATL-Philips HDI 5000 instrument (Seattle, WA), with a 2.0–5.0 MHz curved array transducer. We identified the left and right kidneys in the sagittal plane along its longitudinal axis. Also, we performed measurements of maximal bipolar kidney length, width, and depth. Kidney width and depth were measured at the level of the hilum. The cross-sectional area in which the kidney appeared symmetrically round at its maximum width was used. Kidney volume was calculated using the equation for a prolate ellipsoid: volume (cm3) = 0.523 × length (cm) × width (cm) × depth (cm) (31). Combined kidney volume was calculated by summing right and left kidney volume. We previously reported good intraobserver and interobserver correlation coefficients (32).

Nonfasting blood samples were drawn by antecubital venipuncture. Creatinine levels were measured with enzymatic methods and cystatin C levels with a particle-enhanced immunoturbidimetric assay (using Cobas 8000 analyzers, Roche, Almere, the Netherlands). Quality control samples demonstrated intra-assay and interassay coefficients of variation of 0.51% for creatinine and 1.65% for cystatin C, and 1.37% for creatinine and 1.13% for cystatin C, respectively. We estimated glomerular filtration rate (eGFR) according to the revised Schwartz 2009 formula: eGFRcreat = 36.5 × (height (cm)/serum creatinine (µmol/l)) (33). Additionally, we estimated the glomerular filtration rate using Zappitelli’s formula based on cystatin C levels: eGFRcystC = 75.94/[CysC1.17] (22).

Urine creatinine (mmol/l) and urine albumin (mg/l) levels were determined with a Beckman Coulter AU analyser, creatinine levels were measured with the Jaffe reaction. We also calculated the albumin-creatinine ratio. In line with clinical cut-offs, microalbuminuria was defined as an albumin-creatinine ratio >2.5 mg/mmol for boys and > 3.5 mg/mmol for girls (34).

Statistical Analysis

First, we performed multiple linear and logistic regression analyses, adjusted for sex, age at measurements and the first four principal components, to examine the associations of the 21 SNPs (20 SNPs for kidney function per se and 1 for microalbuminuria) individually with combined kidney volume, creatinine and cystatin C levels, eGFRcreat and eGFRcystC and the risk of microalbuminuria in childhood, assuming additive genetic effects. To adjust for multiple testing in the analyses of the individual SNPs, we applied Bonferroni correction for the number of SNPs tested. A P value of < 2.4 × 10−3 (0.05/21) was considered statistically significant. Second, for each outcome, we calculated the percentage of variance explained by all SNPs combined by deducting the unadjusted R2 of the model including only the covariates from the unadjusted R2 of the full model including all SNPs and the covariates, which were sex, age at measurements, and the first four principal components from our genome wide data. Third, we combined the 20 SNPs related to adult eGFRcreat in a genetic risk score that summed the number of eGFRcreat-decreasing alleles weighted by their previously reported effect sizes in adults. As for every SNP, the number of effect alleles can be 0, 1, or 2, the weighted risk score was rescaled to a score ranging from 0 to 40, the maximum number of effect alleles, and rounded to the nearest integer. Additionally, we computed an unweighted genetic risk score based on the 20 adult eGFRcreat SNPs to see whether the results were independent of the weight based on adults findings, by adding the number of risk alleles for the 20 SNPs. Linear and logistic regression analyses were performed to examine the association of these risk scores with combined kidney volume, creatinine and cystatin C levels, eGFRcreat, eGFRcystC, and the risk of microalbuminuria in childhood, adjusted for the same covariates as mentioned above. All analyses were performed first in the full group and subsequently repeated as sensitivity analyses in the subgroup of children of European ethnicity, the largest ethnic subgroup in our population. A child was classified as European if he/she was within four standard deviations from the HapMap CEU panel mean value for all first four principal components, based on the genetic data. To assess whether the associations were different by child sex we evaluated the statistical interaction by adding the product term of the child sex and individual SNPs to the models. However, no significant interactions were observed. We also aimed to explore whether child height affected the associations of the genetic risk score with combined kidney volume. We performed a sensitivity analysis to explore whether adding 7 SNPs related to creatinine metabolism (rs10774021, rs10794720, rs491567, rs6465825, rs9895661, rs2453533, rs3798156) to the genetic risk score would affect the results (11). In addition, we explored the association of a genetic risk score including only 11 of the 20 SNPs, which may have a role related to kidney developmental outcomes (nephrogenesis (rs881858, VEGFA; rs626277, DACH1); glomerular filtration barrier and podocyte function (rs11959928, DAB2; rs3925584, MPPED2; rs13538, NAT8), metabolic function of the kidney (rs1260326, GCKR; rs10109414, STC1), solute transport (rs12460876, SLC7A9; rs6420094, SLC34A1) and renal cell structure development (rs12917707, UMOD; rs17319721, SHROOM3)) (13,17,35,36). Furthermore, we explored the effect of 2 SNPs known to be associated with eGFRcystC, by adding them to the unweighted risk score. All analyses were performed using the Statistical Package for the Social Sciences version 21.0 for Windows (SPSS IBM, Chicago, IL).

Author Contributions

K.M., J.F.F., and V.W.V.J. conceptualized and designed the study, carried out the analyses, drafted the initial manuscript, and approved the final manuscript as submitted. K.M. and S.V. analyzed the data. S.V. critically reviewed and revised the manuscript, and approved the final manuscript as submitted. A.H. and O.H.F. contributed to the data collection, critically reviewed the manuscript, and approved the final manuscript as submitted. K.M., J.F.F., and V.W.V.J., had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data.

Statement of Financial Support

The general design of Generation R Study is made possible by financial support from the Erasmus MC, University Medical Center, Rotterdam, the Netherlands Organization for Health Research and Development (ZonMw), the Netherlands Organisation for Scientific Research (NWO), the Ministry of Health, Welfare and Sport and the Ministry of Youth and Families. K.M. has been financially supported through Erasmus Mundus Western Balkans (ERAWEB), a project funded by the European Commission. O.H.F. works in ErasmusAGE, a center for aging research across the life course funded by Nestlé Nutrition (Nestec Ltd.); Metagenics Inc.; and AXA. V.W.V.J. received an additional grant from the Netherlands Organization for Health Research and Development (VIDI 016. 136. 361) and a European Research Council Consolidator Grant (ERC-2014-CoG-648916). This research was partly funded by the European Commision’s Horizon 2020 Programme: EU H2020-PHC-2014-Grant no. 633595: DynaHEALTH—Understanding the dynamic determinants of glucose homeostasis and social capability to promote Healthy and active aging. The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclosure

All authors have no conflicts of interest to disclose.