Background

Chronic kidney disease (CKD) contributes a substantial proportion of disease burden with an estimated global prevalence of 9.1% in 2017 [1]. CKD is an independent risk factor for cardiovascular disease [2] as well as endstage kidney disease. The key to prevention of CKD and of its complications is understanding causation.

Low birthweight is associated with increased risk of subsequent adult CKD [3,4,5] and is also associated with adult chronic diseases associated with CKD such as hypertension and Type 2 diabetes [6,7,8] suggesting foetal kidney development may be a key factor. The third trimester of pregnancy is when foetal kidney development occurs and intra-uterine growth retardation (IUGR) during this period leads to reduction in nephron number [9, 10] and lower kidney size at birth [11,12,13]. As nephron number is fixed at birth [9], compensatory glomerular hypertrophy, hyperfiltration and hypertension, could lead to further reduction in nephrons, and susceptibility to kidney damage [14, 15].

IUGR has been found to be associated with albuminuria in infancy [16, 17], with similar estimated Glomerular Filtration Rate (eGFR) at age 2 despite smaller kidney size, suggesting hyperfiltration [17]. Kidney size tracks through into early childhood [18] and lower foetal kidney size was found to be independently associated with both reduced eGFR and kidney volume at age 6 years in the Generation R cohort in the Netherlands [19].

We aimed to investigate the relationship between foetal kidney volume and kidney function and blood pressure in childhood using the Born in Bradford (BiB) birth cohort. This is a prospective longitudinal multi-ethnic birth cohort study that aims to examine the impact of environmental, psychological and genetic factors on health and wellbeing in a deprived population. Foetal kidney volume was measured at 34 weeks gestation in a sub sample, and follow-up undertaken in childhood (age 7–11 years).

Methods

Baseline recruitment in pregnancy

Women were recruited to the BiB cohort study while attending for their glucose tolerance test (OGTT), offered to all pregnant women registered at Bradford Royal Infirmary at 24–28 weeks of gestation.

BiB children aged 7–11 years and their families were followed up using a multi-method approach between 2017 and 2020 (BiB Growing Up Study) [20]. Detailed parent and child questionnaires, BP, anthropometry and blood samples were collected. Written informed consent was collected for the follow-up and for continued routine data linkage.

Renal ultrasound sub-study recruitment in pregnancy

A renal sub-study was nested within the full BiB cohort at baseline. Women who were attending for the OGTT at 26–28 weeks of gestation, had consented for the main BiB study and completed the baseline questionnaire, were invited to undertake a further foetal ultrasound scan (USS) at 34 weeks for standard anthropometrics and foetal renal dimension measurement. Data on renal ultrasound was available for 1802 women, details of recruitment and ultrasonography were published previously [21]. Our analyses were restricted to those who reached 37 weeks of gestation.

Follow-up population

Recruitment for the renal US sub-study follow-up commenced in January 2018 and ended in March 2020. Figure 1 shows the pattern of recruitment.

Fig. 1
figure 1

CONSORT flow chart of follow-up recruitment of the BiB renal ultrasound sub-study in pregnancy

Full size image

Data collection was undertaken by trained staff in schools, clinics or at the participant’s home using standard operating procedures. Child participants’ height and weight were measured without shoes and in light clothing. BP measurements were taken at the brachial artery preferably on the left arm twice using an Omron electronic monitor 705-CPII. Skinfold measurements were taken in between the first and second BP measurement. Non-fasting blood samples were taken. First morning urine samples were taken at home and transported to the laboratory at Bradford Royal Infirmary laboratory by a pre-arranged taxi service, by a parent or a member of the team.

Laboratory methods

Blood samples were analysed for creatinine, urea and cystatin C. Creatinine and urea were analysed on Beckman AU650. The creatinine calibrator is traceable to the Isotope Dilution Mass Spectrometry (IDMS) reference method used by the National Institute of Standards and Technology Standard Reference Material 967. The coefficient of variation (CV) for serum creatinine was 5.7%. Cystatin C was measured using particle enhanced immunoturbidimetric assay on Roche/Hitachi cobas c systems. The assay is standardised against the ERM-DA471/IFCC certified reference material for cystatin C. CVs were 1.6–2.6.

Urine samples were analysed for protein, albumin, creatinine and retinol binding protein (RBP). The urine protein method was based on Pyrogallol red, urine albumin using a turbidometric assay and urine creatinine was an enzymatic assay, all measured on Beckman AU680. The limit of detection for urine albumin was 3 mg/L. CVs were 2.6% for creatinine, 8% for albumin and 7.5% for protein. RBP was measured using the immunonephelometry method on a Siemens Atellica630 Neph Nephelometer. The reference range is < 15 mg/L, the limit of detection is 3 mg/L and CV was 5.3% [22].

Renal and BP outcomes

Outcomes included eGFR, urinary albumin to creatinine ratio (ACR), urinary protein to creatinine ratio (PCR), RBP and systolic and diastolic blood pressure. Each outcome was considered separately and all (except RBP) were assessed as continuous variables.

eGFR was calculated using four published equations as follows:

  • Schwartz creatinine only = 41.3*(height (m)/ Scr) [23].

  • 2012 CKiD Schwartz combined serum creatinine and cystatin C = 39.8 x [(height (m2)/Scr(mg/dl)]0.456 x [1.8/cystatin C (mg/L)]0.418 x [30/blood urea nitrogen (mg/dl)]0.079 x [1.076male] x [height (m)/1.4]0.179 [23].

  • Zappittelli cystatin C only = 75.94/[cystatin C^1.17] [24].

  • Filler cystatin C only log(eGFR) = 1.962 + (1.123*log(1/cystatin C)) [25].

Child height required to calculate eGFR using both Schwartz equations was measured at the same visit as the blood sample in 60.0% (n = 358), the height measurement was standardised to time of sample using centile charts if not measured at the same visit [22].

Urine albumin was only detectable in 211/416, 50.7% of the sample [22].

Exposure

Renal volume was derived from the volume of an ellipsoid using the formula: length × width × depth × 0.523 [26].

Covariates

Maternal

body mass index (BMI), age, parity, smoking, alcohol consumption, measures of socio-economic position (maternal highest educational attainment, housing tenure [buying/own house and renting or other related], and employment status), marital status and gestational diabetes (all recorded at first antenatal booking or during pregnancy).

Child

Birth weight, gender and gestational age at birth; age and measured weight and height at follow-up. Measured weight and height were used to calculate BMI and body surface area (BSA). BMI was converted to age- and sex- adjusted z-scores according to the UK 1990 growth reference charts. BSA was calculated using the Du Bois formula [27] (BSA = weight (kg)0.425 x height (cm)0.725 × 0.007184). Blood pressure was not identified as a covariate to adjust for from the directed acyclic graph (DAG) but as the direction of this relationship is not known, we included blood pressure as a covariate [22].

Statistical analysis

All analysis was carried out in Stata [28]. The selection of covariates into the multivariable models were guided by a DAG (Fig. 2) constructed using DAGitty [29]. The assumptions underlying the associations in the DAG were decided based on existing knowledge of the literature and topic area and through detailed discussion within the team.

Fig. 2
figure 2

Directed acyclic graph illustrating the relationship between ethnicity and childhood kidney function

Full size image

Basic descriptive statistics are presented for the overall sample and stratified by tertiles of foetal renal volume. Linear regression was used to examine the association between foetal renal volume and the outcome measures, except for the ACR detectable binary outcome where logistic regression was used with the ACR result outcome conditional on a positive ACR detectable outcome. A statistical significance level of 0.01 with 95% CI was used due to the multiple comparisons performed.

Covariates were added sequentially to form multivariable regression models in stages by grouping explanatory factors:

Model 1: Ethnicity and child age at kidney function measurement (ethnicity);

Model 2: Additionally adjusted for maternal age, maternal education, housing tenure and employment status (socio-demographic);

Model 3: Additionally adjusted for maternal BMI, maternal height, smoking and alcohol consumption in pregnancy, and gestational diabetes (maternal/pregnancy);

Model 4: Further adjusted for birthweight, gestational age at birth, BSA (kidney outcomes) or BMI (BP outcomes), gender, systolic and diastolic blood pressure (kidney outcomes) (fully adjusted).

Sensitivity analysis to explore the impact of differences in timing of height measurement and blood sampling for the Schwartz eGFR formulae was undertaken. We compared estimates using standardised height (accounting for the difference in timing between sampling and measurement), height as measured (without accounting for difference in timing), and restricting the sample to those with sampling and measurement at the same time. Separate sensitivity analysis replacing BSA with BMI in eGFR models was also carried out. Complete case analysis was carried out due to the low percentage of missing data on model covariates (n < = 2.5%).

Results

Of the 1802 children whose mothers had a renal ultrasound in pregnancy, 630 children provided a blood (n = 447) or urine (n = 433) sample for renal analysis, of which 250 children provided both a blood and urine sample (Fig. 1). The response rate was higher in South Asians and in boys but there was no difference in birthweight or renal volume between those that responded and those that did not (Table 1).

Table 1 Selected descriptive variables from the baseline data to assess difference in sample characteristics at follow-up by response*
Full size table

52.7% of the sample with follow-up blood or urine sampling (n = 630) were South Asian and 44.6% had further and higher maternal education (Table 2). Mean maternal age was 28.2 years (standard deviation (SD) 5.6) and mean maternal BMI at antenatal booking was 26.7 kg/m2 (SD 5.9). Mothers of babies in the lowest tertile of renal volume were more likely to be of South Asian origin; to be married, and to own their own home, less likely to smoke or drink during pregnancy, and were younger, had lower BMI, and higher parity.

Table 2 Demographic and clinical details of mothers and children for the full sample and stratified by tertiles of fetal renal volume
Full size table

The prevalence of CKD at 7–11 years was rare (11 (2.6%) with ACR ≥ 2.5 mg/mmol in boys/3.5 mg/mmol in girls and none with eGFR < 60 ml/min/1.73m2). RBP results for the whole sample were within the normal range.

Males were more likely to be in the highest tertile for foetal kidney volume. Foetal kidney volume was inversely associated with cystatin C and positively with eGFR on all measures. There were no obvious patterns in urinary markers or blood pressure. The correlations between renal volume and the four measures of eGFR are shown in Fig. 3.

Fig. 3
figure 3

Scatter plots of eGFR and renal volume

Full size image

Kidney volume was positively associated in unadjusted and adjusted models with eGFR calculated using Schwartz creatinine only formula (fully adjusted 0.80 ml/min diff per unit increase in volume, 95% CI 0.06 to 1.53), Schwartz combined (0.64 ml/min diff per unit increase in volume, 95% CI 0.25 to 1.02), Zappitelli (0.79, 95% CI 0.38 to 1.20) and Filler (2.84, 95% CI 1.40 to 4.28) (Table 3, Supplementary Table 1 for complete results of adjusted models). These findings were consistent using cystatin C as an outcome which was negatively associated with renal volume. A higher renal volume was associated with a higher probability of having a detectable ACR value but was not associated with having a higher ACR result. There were no differences in PCR, RBP or blood pressure.

Table 3 Summary table of differences in kidney function, kidney damage and blood pressure by foetal kidney volume: univariable and multivariable*
Full size table

Adjusting for child BMI instead of BSA resulted in similar estimates for eGFR (Supplementary Table 2). There was little difference in eGFR calculated using measured height regardless of timing between height measurement and blood sampling or on restricting to the subset with blood sampling and height measurement taken at the same time (Supplementary Table 3).

Discussion

Prevention of adult CKD partly rests on better understanding of the early origins and childhood determinants of kidney function. Using a population based multi-ethnic birth cohort, we have shown here that foetal kidney volume is positively associated with eGFR using cystatin C only, and cystatin C and creatinine combined formulae and with having a detectable ACR result but not with other urinary measures or blood pressure.

Strengths and limitations

Our study strengths include a multi-ethnic population from the same city with diverse socio-economic background with detailed ultrasonography with high reliability in pregnancy for renal measures, and first morning urine sample taken to measure albuminuria and proteinuria. Foetal kidney volume was used as a proxy for nephron number [30,31,32]. This is the first renal follow-up of this cohort and a key strength is the possibility of following up this cohort as the children grow. We measured both creatinine and cystatin C and calculated eGFR using four different formulae, all of which have been validated in paediatric populations. eGFR calculated using the Schwartz creatinine only formula was higher than that using the Zappitelli cystatin C as previously shown in other studies [33, 34]. It has been previously suggested that cystatin C formulae may be more sensitive than creatinine formulae for evaluating kidney function. This is because cystatin C is freely filtered by glomeruli, fully catabolised by renal tubules and not excreted by non-renal routes [35]. There was no correction for ethnicity in any of the eGFR formulae.

There were limitations. Uptake to the follow-up component was lower than expected and whilst there was no obvious major response bias this limited our statistical power. The low uptake highlights the challenge of undertaking follow-up, especially with invasive testing, in a population with high levels of socio-economic disadvantage and ethnic diversity. Differences in timing between anthropometric measurements and blood sampling in some children may have affected the creatinine based eGFR measurement. We used growth centile charts to take this into account, and no difference in study findings was found in sensitivity analysis. Our assessments of kidney function and blood pressure were based on a single measure and any variation is likely to reduce the strength of any associations but this approach was the same as in another study [36]. Unlike the Generation R cohort [30], we had no valid measures of kidney volume in childhood as this was not practically feasible. Our data are restricted to full term infants as we excluded preterm infants.

These findings add to our knowledge of the biology of kidney development by examining the influence of foetal kidney size through childhood. There are limited data on the childhood associations of kidney volume, predominantly from studies using the Generation R Dutch birth cohort. Third trimester foetal kidney volume was associated with eGFR (Schwartz creatinine based) and cystatin C at 6 years of age but there was no association with albuminuria or blood pressure in the Generation R cohort (n = 870) [37]. The effect sizes were small and without clinical consequences at school age. Foetal kidney size on ultrasound was correlated with serum creatinine at age 6 in a single centre study of 1748 children in Italy [36].

The Generation R cohort were able to extend the investigations by including ultrasound in childhood at age 6 and hence childhood kidney volume. Foetal kidney volume was positively associated with childhood kidney volume, and both were associated with eGFR and inversely with Cystatin C but not blood pressure [30, 37]. In Generation R, lower foetal growth was associated with lower childhood kidney volume and function independently of childhood growth [30]. The relationship of kidney volume with microalbuminuria is less clear. In the Generation R cohort, childhood kidney volume was positively associated with microalbuminuria and in our study foetal kidney volume was associated with the presence though not level of microalbuminuria. The significance is unclear and requires further follow-up .

Foetal kidney volume reflects nephron number which are fixed at birth. Nephron number is an important determinant of kidney function through childhood. Limited nephron number may lead to hyperfiltration and later impaired function [5]. Impaired foetal growth influences birth weight and foetal kidney size and this has persistent effects on kidney development and function and may affect adult CKD risk. Additionally, renal tissue relative to body weight may impact on eGFR. This may be part of the explanation for ethnic differences in adult CKD, as we have previously shown ethnic differences in foetal kidney volume [21].

Further longitudinal follow-up is required in the Born in Bradford cohort to investigate the relationship between foetal kidney volume, childhood measures of kidney function, blood pressure and metabolic risk with adolescent kidney function and size as children go through puberty and in later life, to see how any effects track. This would inform population strategies and risk stratification in utero/childhood to prevent adult CKD through optimising foetal and postnatal growth and kidney development. This study suggests that acting early may have advantages in reducing late-onset adult disease presentation for higher risk groups.