Can childhood obesity influence later chronic kidney disease?

Discussion of study designs used in the paediatric age group

Problems of measuring obesity and kidney function in the paediatric age

The association has been measured most often either as the difference in mean eGFR, which involves creatinine and less often cystatin C concentrations or level of albuminuria or microalbuminuria in obese children compared to children with normal weight for age and sex. Cystatin C-based eGFR are believed to be less influenced than creatinine by changes in muscle mass [49]. In a meta-analysis of studies comparing creatinine- to cystatin C-based eGFR in adults and children, most of whom had some form of kidney disease, Roos et al. [50] concluded that cystatin C was more accurate.

The most commonly used measure for obesity is body mass index (BMI), with obesity and overweight classified according to World Health Organization (WHO) growth data http://www.who.int/childgrowth/en/. Other measures such as waist circumference and waist to height ratio have also been used. Associations have been measured using either simple or adjusted correlation coefficients, with the measure of obesity and either eGFR or level of microalbuminuria as continuous variables. These factors make the interpretation of the findings difficult. Differences in the findings may be explained further by differences in the range of BMI studied, age and sex of participants, age of onset of obesity and its duration, ethnicity, diet and other confounders as well as by differences in the choice of substrate and equations to estimate GFR and by differences in the conditions of measurements of urine albumin (e.g. diurnal effects, exercise, degree of hydration, timed versus spot samples, single or multiple urine sampling). In a large population-based study involving 4305 children aged 6.0 years, Miliku et al. [51] showed that BMI was not associated with creatinine-based eGFR and that BMI was negatively associated with cystatin C-based eGFR. The effect was relatively small (− 0.61 ml/min/1.73 m², 95% CI – 0.96 to − 0.26 per unit increase in BMI standard deviation score (SDS), p < 0.01). However, they showed that creatinine-based eGFR but not cystatin C-based GFR was influenced by fat mass (2.68 ml/min/1.73 m², 95% CI 2.14 to 3.21 per 1 SD increase in fat mass percentage, p < 0.001) and by lean mass percentage (− 2.74 ml/min/1.73 m², 95% CI – 3.27 to − 2.20 per unit increase in lean mass percentage, p < 0.001).

Which statistical association answers the question?

In the search for “independent” effects of obesity, many studies present measures of effect adjusted for risk factors which may be on the causal pathway between obesity and renal function, for example blood pressure, insulin resistance, abdominal obesity, HbA1c. For epidemiologists the question is: Is Obesity a risk factor for CKD? This means that what is of most importance from a public health perspective is the overall effect of obesity, either acting through direct nephrotoxic effects or acting indirectly on the kidneys by its effects in raising blood pressure or causing diabetes, with their ensuing effects on the kidneys. From this perspective, the overall effect of obesity needs to take into account age, sex, smoking status, socio-economic status and possibly ethnicity, as the key confounders. Therefore, many of these studies are over-adjusted.

Main findings of studies to date

Table 1 summarises the findings of studies on eGFR in relation to obesity. The cross-sectional population-based studies found a mix of positive, negative and no associations for eGFR with BMI, whilst for children from obese clinic populations compared to controls, there seems to be more of a consistent positive association between eGFR and BMI. The study by Pacifico et al. [25] showed that obese children were more likely than children with normal BMI to have a low eGFR, as well as being more likely than children with normal BMI to hyperfiltrate. These findings might possibly be explained by a changing effect of obesity on eGFR depending on the duration of obesity. This is an important factor which has not been measured in this study, or in any of the other listed studies, as most of these only assess overweight or obesity at a single time-point.

Most population-based studies on microalbuminuria have shown either no association with obesity or higher microalbuminuria in children with normal BMI (Table 2). The reason for this is not clear and some authors have suggested that the finding of microalbuminuria in non-obese children and adolescents might be confounded by orthostatic proteinuria. However, microalbuminuria has been found in obese children in studies based on obesity clinics, possibly linked to higher blood pressure.

Paediatric obesity and renal allograft function

The situation in which a recipient receives a new kidney is a unique model in which the influence of recipient-specific factors such as obesity on the transplant kidney function can be prospectively explored. There are a number of retrospective studies on the effect of paediatric obesity on subsequent renal allograft function during childhood. Obesity is common in children receiving a renal transplant and the first 4 months post-transplantation have been found to be a particularly sensitive period for weight gain [52]. This is thought to be at least in part mediated by steroid immunosuppressive therapy [53].

In a retrospective review of the records of 76 renal allograft recipients from 1994 to 2000, Mitsnefes et al. [54] found pre-transplant obesity to be associated with decreased allograft function after renal transplantation. Paediatric patients who were obese (BMI ≥ 95th percentile) both prior to transplant and 1 year post-transplant had lower mean directly measured GFR than both those who were non-obese prior to transplant but became obese by 1 year post-transplant (46.1 ± 15.0 mL/min/1.73 m² vs. 57.7 ± 24.5 mL/min/1.73 m², p < 0.05) and those who were non-obese prior to transplant and 1 year post-transplant (46.1 ± 15.0 mL/min/1.73 m² vs. 60.4 ± 21.5 mL/min/1.73 m², p < 0.01). The impact of paediatric obesity on long-term outcomes in paediatric patients with renal transplant was studied in an analysis of data from the North American Paediatric Renal Transplant Cooperative on 6658 paediatric age recipients. This showed that there had been an increase in the proportion of obese children from 8.2% in transplant recipients during the years 1987–1995 to 12.4% in children transplanted during the years 1996–2002 [55]. This study found no overall difference in patient mortality or graft survival between obese and non-obese children. However, in a post-hoc subgroup analysis it found a greater risk for death in obese children than non-obese children aged 6 to 12 years, and greater risk of living donor graft loss in obese than non-obese children aged 13–17 years. A study of 197 children from Brazil found decreased graft survival at 36 months post-transplant in children who had had an increase equal to or greater than 1 standard deviation in BMI between 1 and 6 months post-transplant [53]. In a more recent study of 750 renal transplant recipients in Australia, aged 2 to 18 years and with a median follow-up of 8.4 years, obesity was associated with a 60% greater risk of graft loss (hazard ratio (HR) 1.61, 95% CI 1.05–2.47) compared to children with normal BMI. There was no association between BMI and acute allograft rejection [56].

In summary, there are some data from the paediatric transplant population suggesting that obesity of the transplant recipient may be associated with accelerated graft loss.

Effects of bariatric surgery on renal function in paediatric age patients

Another way of looking at this question is to investigate the effect of weight reduction through bariatric surgery performed in children and adolescents on their subsequent kidney function. In a case report published in 2009, Fowler et al. [57] described the resolution of proteinuria resistant to medical treatment after bariatric surgery on a 15-year-old girl. This was associated with a reduction in her BMI from 56.8 to 32 kg/m² post-surgery. The impact of bariatric surgery on renal function was described in a cohort of 242 severely obese adolescents [58] with a mean age of 17 ± 1.6 years, 29% of whom were aged 13–15 years. Post bariatric surgery there was an average increase in cystatin C-based eGFR of 3.9 (95% CI 2.4–5.4) ml/min/1.73 m² per 10 kg/m² decrease in BMI (p < 0.0001). The majority of improvement in renal function occurred in those with a baseline eGFR of < 90 ml/min/1.73 m², whose mean (+/-SD) baseline eGFR was 76 ± 12 ml/min/1.73 m². In this group, eGFR improved by 34% to 102 ± 28 ml/min/1.73 m² at the 3-year follow-up evaluation (p < 0.0001). Amongst children with albuminuria pre-surgery, there was a fall in the geometric mean from a pre-surgery level of 74 mg/g (95% CI 45–121 mg/g) to 17 mg/g (95% CI 10–28 mg/g) at 3 years (p < 0.0001) post bariatric surgery. In 69% of children albuminuria normalised by the 2-year follow-up evaluation, and in 75% albuminuria resolved by 3-years post bariatric surgery. This study did not have a comparator group, there were no standardised selection criteria for surgery, and in addition, 24–29% of the 3 year follow-up data were missing. It is important to recognise that bariatric surgery is often followed by a change in diet, and if proteinuria was related to high blood pressure in those sensitive to salt, then it may not have been the surgery itself driving those changes in proteinuria over time. The evidence from studies in adults has already been mentioned [12, 13]. There have been no randomised controlled trials in adults and many of the studies are underpowered, and potential confounding by factors such as those mentioned remains a possibility.

What do epidemiological studies in the paediatric age show?

In summary, there is clear evidence that childhood obesity is associated with cardiovascular risk factors in adulthood, and that obesity in adulthood is associated with markers of incident CKD. The data are equivocal about whether an association between childhood obesity and renal function markers in childhood exists amongst children of the general population, whilst in clinic populations of referred obese children, there appears to be a subgroup of children with albuminuria and reduced eGFR. It may be that associations are difficult to detect depending on the timing of when the child was overweight and became obese, and how kidney function was measured (bearing in mind the limitations of creatinine-based equations). In contrast, the evidence from the paediatric renal transplantation population is very suggestive in that those who are most obese have accelerated graft loss. In addition, there are suggestive cross-sectional data on children aged between 5 and 19 years with eGFR of 30–90 mL/min/1.73 m² and a diagnosis of CKD participating in the prospective cohort “Chronic Kidney Disease in Children” (CKiD) [59]. Children were divided into non-glomerular (n = 513) and glomerular (n = 224) forms of CKD. Using univariate linear regression, in the non-glomerular group BMI was not related to either urinary protein or eGFR, but waist circumference (WC) was associated with an increase in urinary protein of 7.51% (95% CI 0.45–15.06, p < 0.05) and no association with eGFR; in the glomerular group, BMI and WC were related to eGFR (BMI: 6.81%, 95% CI 2.67–11.12%, p < 0.0; WC: by 4.34%, 95% CI 0.00–8.88%, p < 0.05), but neither BMI nor WC were related to urinary protein. The main aim of the study was a comparison between BMI and WC as measures of obesity in this group of patients with established CKD and the conclusion was that WC added limited information to BMI in this cohort. However, these cross-sectional data suggest that the association of obesity with renal markers amongst children with CKD may depend on the type of underlying renal damage.

Epidemiological studies of the association of childhood obesity with adulthood kidney function

There have been a number of studies of risk factors with onset in childhood for CKD in adult life, either using end-stage renal disease (ESRD) or laboratory measurements to confirm CKD.

Muntner et al. [60] reported on 15 cases of ESRD in the Bogalusa cohort. Seven of the cases of ESRD had an unknown aetiology, and amongst other features, they had a higher BMI recorded in childhood compared to their counterparts. This study was of limited scope given the weakness in the ascertainment of cases of ESRD and absence of detailed follow-up data from childhood to the time when ESRD occurred.

Vivante et al. used a sample of 1.2 million Israelis examined prior to their military conscription at age 17 years in the period 1967–1997, linked to the Israeli ESRD registry for the period 1980–2010 [61]. In an adjusted model, overweight (BMI between 85th and 95th centile) and obesity (BMI ≥ 95th centile) at age 17 years were both strongly associated with all-cause treated ESRD (HR 3.00, 95% CI: 2.50–3.60 and 6.89, 95% CI: 5.52–8.59, respectively). The authors could not adjust for confounders such as socioeconomic status or smoking.

A third, carefully designed matched case-control study [62] examined the incidence of ESRD up to the age of 40 years in relation to risk factors measured in adolescence in males only using data collected at conscription to the Swedish military service between 1970 and 1975 when most were aged 18 or 19 years of age. The National Patient Register was used to ascertain data on new cases of ESRD from 1985, at least 10 years after the conscription examinations, to reduce confounding by undiagnosed kidney disease in adolescence. There were 534 incident ESRD diagnoses and 5127 men were selected as controls, representing the most powerful study to date. Elevated BMI at conscription was associated with increased ESRD risk with an adjusted odds ratio (OR) of 3.53 (95% CI 2.04–6.11, p < 0.001) for BMI > 30 kg/m² compared to those with normal BMI (18.5–25 kg/m²). The odds for ESRD were also increased but to a lesser extent in those who had been overweight (BMI at conscription between 25 and 30 kg/m²), with an adjusted OR of 2.36 (1.79–3.13, p < 0.001) raising the possibility of a dose effect by BMI. The possible confounding effects of social status, and smoking, and the role of the subsequent BMI trajectory post conscription or current BMI were not examined.

In a nationwide population-based case-control study using incident cases of CKD as defined by laboratory test results [63], overweight (BMI > 25 kg/m²) at age 20 was associated with increased odds for CKD in middle-aged men and women, relative to BMI < 25 kg/m²: OR (adjusted for age, education, smoking, alcohol, and use of paracetamol and salicylates) = 3.1 (95% CI 2.1 to 4.8) in males, and 3.0 (95% CI 1.4 to 6.1) in females. The problem with this and other health record database studies using routine creatinine tests to examine this question is that people with BMI have more cardiovascular risk factors and therefore get tested earlier for presence/absence of kidney disease, which introduces information bias in favour of finding an association between BMI and incident CKD. As not everybody was tested the same way, no definitive conclusions can be drawn from this or similar routine health care studies.

To our knowledge the Medical Research Council National Survey of Health and Development (NSHD), a socially stratified sample of 5362 singleton children born in 1 week in March 1946 in England, Scotland and Wales, was the first study to systematically examine the effect of overweight throughout childhood and adulthood on later kidney function in a general population cohort [64, 65]. Repeated measurements of BMI between ages 2 and 20 years were used to derive childhood overweight latent classes (‘never’, ‘pre-pubertal only’, ‘pubertal onset’ or ‘always’), which were then related to measures of CKD at age 60–64 years: creatinine- or cystatin C-based eGFR < 60 ml/min/1.73 m² or urine albumin-creatinine ratio (uACR) ≥ 3.5 mg/mmol. Relative to being in the never overweight latent class, being in the pubertal onset or always childhood overweight latent classes was associated with CKD at age 60–64 years when considering cystatin C-based eGFR (OR 2.04, 95% CI 1.09 to 3.82). The associations with CKD at age 60–64 years defined by creatinine-based eGFR (OR 1.27, 95% CI 0.71 to 2.29) and uACR (OR 1.33, 95% CI 0.70 to 2.54) were less marked but in the same direction. Adjustment for social class, lifestyle and health factors (e.g. smoking) had little impact on the effect estimates. The low prevalence of CKD (resulting in low statistical power) and the lack of repeated measurements of creatinine and cystatin C, meaning that the CKD chronicity criterion was not met, were limitations of the study [66].

Overall, there is some suggestive evidence from the NSHD cohort study, but otherwise, studies may be affected by the co-occurrence of social deprivation and smoking.

Mechanisms

The mechanisms relating childhood obesity to later CKD are not well researched [67], which is perhaps unsurprising considering that a follow-up over several decades may be needed to detect associations.

From studies conducted in animals and adults, there is evidence that obesity can lead to inflammation, oxidative stress, abnormal lipid metabolism, activation of the renin-angiotensin-aldosterone system, and increased production of insulin and insulin resistance. These abnormalities could lead to haemodynamic abnormalities, lipotoxicity and inflammation in the kidneys constituting the main mechanisms for obesity-related renal damage [68]. It is thought that the most important mechanism might be haemodynamic abnormalities affecting the renal microvasculature leading to hyperfiltration, development of proteinuria and ultimately to glomerulosclerosis and a reduction in GFR [69, 70, 71]. Studies in young people include a study in young men (mean age 18.4 years), where hyperfiltration was associated with overweight, elevated blood pressure, and low high-density lipoprotein cholesterol with multivariate-adjusted OR of glomerular hyperfiltration of 6.6 (95% CI 3.8–11.6), 1.8 (95% CI 1.0–3.0) and 2.5 (95% CI 1.5–4.3), respectively [72]. Franchini et al. [27] found that the prevalence of microalbuminuria and hyperfiltration was similar between obese and youths with Type 1 diabetes and higher than their control peers (6.0 vs. 8.0 vs. 0%, p = 0.02; 15.9 vs. 15.9 vs. 4.3%, p = 0.03, respectively). Possible mechanisms mediating paediatric kidney injury are thought to include inappropriate activation of the renal-angiotensin-aldosterone system, the inflammatory role of adipokines secreted by fat tissue such as leptin and adiponectin, renal effects of dyslipidaemia and insulin resistance [73, 74, 75].

Children who are overweight are at a higher risk of developing high blood pressure and diabetes, though there is a lack of evidence for effects independent of adulthood overweight [8]. A further study using NHSD data investigated the mediatory roles of diabetes, hypertension and obesity in the relationship between low birth weight and later renal function [76], and found that the observed association between low birth weight and low eGFR was not confounded by socioeconomic position and was not explained by diabetes or hypertension. There was some evidence that it was stronger in study members who were overweight in adulthood. These findings thus again suggested that overweight/obesity may accelerate kidney function loss in the presence of other kidney problems.

Challenges and future research directions

Although there is a plethora of studies looking at childhood obesity and renal function, they do not provide enough evidence for a direct causal pathway between childhood obesity and adult CKD. The evidence that obesity is associated negatively with renal function in the paediatric age is not conclusive, and it may be instead that obesity may accelerate renal function loss in the context of other (subclinical) renal damage.

One deficiency in research to date has been an inability to distinguish whether the effects of childhood obesity on CKD are due to childhood obesity in and of itself (i.e. that childhood is a ‘critical period’ or a ‘sensitive period’ [77]), or whether the observed effect is merely due to childhood obesity being indicative of greater lifelong exposure to obesity pressure (i.e. an ‘accumulation of risk’ [77]). In addition, to date, authors have not fully explored the role of other factors that co-occur with obesity, such as a bad diet with high calories and salt, leading to both obesity and high blood pressure. The challenge of research investigating the question of whether childhood obesity causes kidney damage independently of adulthood obesity (‘sensitive period’) is the statistical necessity to have sufficient subjects who become normal weight in adulthood after being obese in childhood to make analysis tenable, and the fact that reasons for weight loss and the way weight loss occurs may itself impact on kidney health. There were very few subjects in the NSHD data who lost weight, which meant that data were not available to address this important question [65].

The number of years lived with adulthood obesity are known to be directly associated with the risk of all-cause and cause-specific mortality [78]. Since the obesity-related metabolic, oxidative and inflammatory damage to many tissues and organs may accumulate over time [79], kidney dysfunction resulting from lifelong obesity beginning in childhood would not be unexpected. To date, we have not seen evidence that CKD prevalence or ESRD incidence increase as a result of increasing prevalence of childhood obesity. However, based on the data from the NHSD, the lag time between childhood overweight and adult CKD may well be several decades [64, 65].

It is unclear whether childhood obesity causes pre-adulthood kidney damage. A further investigation of this question would require repeated follow up of very sensitive kidney function markers in a prospective birth cohort study. In general population cohorts, this necessitates large sample sizes. Such a study should carefully characterise nutrition and physical activity to better understand which of the pathways that lead to obesity may have most impact on kidney health.

The finding that obesity strengthens the association of low birth weight with later poorer kidney function is particularly worrying in the setting of low middle income countries which are undergoing a rapid rate of urbanisation with an associated rise in obesity [80]. Indeed, in the Indian Migrant study, the years lived in urban settings were associated with decreased eGFR amongst rural-urban migrants in age-adjusted analyses [81].

Conclusions

The association between adulthood obesity and kidney disease is well-documented, but the evidence for a similar effect of childhood obesity is patchy. Although there remains much research to be done in this area, in view of the overall impact of childhood obesity on health it is becoming apparent that efforts to prevent and treat obesity early in life could potentially have an impact on the progression, costs and comorbidities of kidney disease [75]. It is thought that reducing the amount of time that individuals live with obesity (“obesity-years”) by preventing early onset may be critical in minimising a variety of long-term hazards of obesity on health [79].