Introduction

Despite recent advances in the field of chronic kidney disease (CKD), the prevalence of pediatric stage 5 chronic kidney disease (CKD 5) in the USA has been increasing [1]. CKD-mineral and bone disorder (MBD), a complication of CKD [2, 3], manifests by increases in circulating parathyroid hormone (PTH), phosphorus, alkaline phosphatase (ALP), fibroblast growth factor 23 (FGF23), and other biomarkers and leads to impaired bone health and vascular calcifications. In the chronic kidney disease in children (CKiD) study cohort, 67% of participants had CKD-MBD [4]. CKD-MBD increases the risk of fractures [5] and is associated with cardiovascular disease [6, 7]. In children, CKD-MBD contributes to the impairment of statural growth [8, 9]. In the North American Pediatric Renal Trials and Collaborative Studies (NAPRTCS), 36.9% of children had short stature at enrollment [10].

Chronic low-grade inflammation is another typical feature of CKD [11]. Specifically, it has been shown that tumor necrosis factor-alpha (TNF-α) and IL-6 are inversely and independently associated with GFR in CKD [12]. The specific role of individual cytokines in CKD remains incompletely understood. We have recently performed cytokine profiling to identify cytokines associated with anemia in children with CKD [13]. Only IL-6, but not the other cytokines that we screened, was associated with anemia. Interestingly, in our subsequent experimental study, IL-6 deletion, while improving anemia, failed to improve bone health, growth, and nutritional status of juvenile mice with CKD [13]. Thus, understanding of the role of individual cytokines in CKD pathophysiology is important and may inform novel therapeutic targets.

Chronic non-renal inflammatory disorders of childhood, such as juvenile idiopathic arthritis and inflammatory bowel disease, frequently present with bone disease [14,15,16,17] and poor statural growth [18, 19]. The use of TNF-α inhibitors in children with these conditions improves their bone health and growth [20, 21]. While it remains unclear if inflammation contributes to the development of CKD-MBD in children with CKD, emerging literature suggests that this is highly plausible. Indeed, serum phosphorus is independently associated with inflammation in adults with CKD [22]. Inflammation stimulates production of FGF23 [23]. In a population-based cohort, TNF-α was the only cytokine independently and positively associated with FGF23 [24]. Based on bone biopsy data, TNF-α was shown to inversely correlate with bone volume in adults with CKD [25]. TNF-α suppresses bone formation, particularly by inhibiting osteoblastic differentiation of mesenchymal stem cells in the bone [26]. Thus, based on the recent literature, investigation of the role of inflammation and specifically TNF-α in pediatric CKD-MBD and growth impairment is warranted.

The aim of the present study was to evaluate a panel of cytokines to detect potential inflammatory mediators implicated in CKD-MBD and in growth impairment of children with CKD.

Methods

Patients

This was a cross-sectional study. Children were enrolled from the pediatric nephrology clinics at Weill Cornell Medical Center from June 2016 to June 2019. The study was approved by the Institutional Review Board of Weill Cornell Medicine. Parental informed consent was obtained, as well as assent for children ≥ 7 years of age. Eligibility criteria included the following: pre-dialysis CKD stage II and above using the KDIGO definition [27] and no acute illnesses at the time of blood collection. Glomerular filtration rate (GFR) was calculated using the bedside Schwartz formula [28]. Age-sex-specific height z-scores were calculated using the 2000 Centers for Disease Control and Prevention standard growth charts for US children [29].

Biomarkers

Serum was separated immediately upon blood collection and stored at – 80 °C until analyzed. To delineate the role of individual cytokines in the development of CKD-MBD, we measured serum levels of 9 cytokines previously reported to be altered in CKD (IL-1β, IL-4, IL-6, IL-8, IL-10, IL-12, IL-13, TNF-α, and interferon-γ). Cytokines were measured using V-PLEX kit, an array-based multiplex electrochemiluminescent assay, from Meso Scale Discovery (Rockville, MD) on the MESO QuickPlex SQ120 Analyzer following the manufacturer’s instruction. The intra-assay coefficient of variation (CV) for these cytokines ranged from 2.4 to 6.4% and inter-assay CVs from 5.0 to 10.2%. The detection limits of IL-1β, IL-4, IL-6, IL-8, IL-10, IL-12, IL-13, TNF-α, and interferon-γ were 0.10, 0.016, 0.10, 0.07, 0.077, 0.10, 0.50, 0.20, and 0.40 pg/mL, respectively. Serum bone-specific alkaline phosphatase (BAP) was measured by an enzyme-linked immunosorbent assay (ELISA) according to the manufacturer’s manual (Quidel/Metra Bio Systems, San Diego, CA). The intra-assay and inter-assay CVs were < 6.5% and < 10.0%, respectively. The sensitivity was 0.7 μg/L. Serum procollagen type 1 N-terminal propeptide (P1NP) was measured by a radioimmunoassay following the manufacturer’s instruction (Orion Diagnostica, Oulunsalo, Finland). The intra-assay and inter-assay CVs were < 10.2% and < 9.8%, respectively. The sensitivity was 2.0 μg/L. Serum phosphorus, PTH, and ALP were measured by the New York Presbyterian Hospital clinical laboratory as part of the standard of care.

Power considerations

Mendoza et al. reported a direct correlation between TNF-α and FGF23 with a coefficient of correlation r = 0.4 in adult patients with CKD from the Chronic Renal Insufficiency Cohort (CRIC) study [23]. Assuming a correlation with similar strength exists between TNF-α and PTH in children with CKD, a priori power analysis indicated that a total sample size of 47 patients will be required to determine the significance of correlation coefficient with type I error rate α = 0.05 and type II error rate β = 0.2 [30].

Statistical analysis

The χ2 test was applied to detect differences between categorical variables. Normality of continuous variables was checked using Shapiro-Wilk W test. To examine the relationship between TNF-α and clinical characteristics of CKD in children, we stratified our cohort by TNF-α tertiles. ANOVA was used to compare normally distributed continuous variables among TNF-α tertiles and Kruskal-Wallis test for non-normally distributed variables. The correlations were assessed using Spearman’s rank test. Multiple linear regression was used for multivariate analyses. To assess the relationship between TNF-α and CKD-MBD, and between TNF-α and growth, we conducted a series of stepwise linear regression analyses where TNF-α was the exposure. PTH, ALP, serum phosphorus, BAP, P1NP, and height z-score were the outcomes for the respective individual analyses. We first performed unadjusted analyses (model 1), followed by adjustment for age and sex (model 2), GFR (model 3), and active vitamin D analog use (model 4). P values < 0.05 were considered statistically significant. STATA 12.0 (Stata Corp, College Station, TX) was used for conducting statistical analyses.

Results

Cohort characteristics

The cohort was composed of 63 children with median GFR 42.0 [30.11–53.41] mL/min/1.73 m2 (Table 1). Most children had stage III CKD. The majority had non-glomerular CKD etiology, predominantly congenital anomalies of the kidneys and urinary tract. Approximately half of the cohort had microalbuminuria. Although most children had normal body mass index (BMI), median height z-score was − 1.01 [− 2.1–0.04], and 21% of children had a short stature defined as height < 3rd percentile.

Table 1 Clinical characteristics of the study cohort
Full size table

Median PTH was 98 [54–295] ng/L, and one-third of the children (33.3%) were receiving active vitamin D analogs. Nutritional vitamin D was prescribed to 34.9% of children, and median 25(OH)D level was 27.9 [19.9–32.0] ng/mL. Serum phosphorus was normal in the majority of children. Very few children required phosphate binders in this cohort. Alkali therapy was used in 14 children (22.2%) and recombinant human growth hormone (rhGH) in 11 children (17.4%). Children receiving rhGH at the time of blood sample collection for this study did not have statistically significant differences from the rest of the cohort, with the exception of sex and height z-score (Table S1).

Cytokine profiling identified an association between TNF-α and PTH

In unadjusted analysis, TNF-α and IL-6 levels were significantly associated with PTH (Table S2). The rest of the cytokines that we measured (IL-1β, IL-4, IL-8, IL-10, IL-12, IL-13, and interferon-γ) were not associated with PTH. Given the known inverse relationship between GFR and inflammation in CKD [16], we adjusted our analysis for GFR. In this adjusted analysis, TNF-α remained significantly associated with PTH, while the association between IL-6 and PTH was no longer significant (Fig. 1).

Fig. 1
figure 1

Cytokine profiling and PTH in children with CKD. This figure depicts the associations between cytokines and parathyroid hormone (PTH) in children with chronic kidney disease (CKD). Each bar represents a separate regression analysis of the PTH on respective cytokine, adjusted for glomerular filtration rate; t-statistic values corresponding to the significance levels (P) for the overall models are graphed. Negative values correspond to inverse correlations. n = 63. IFN, interferon; IL, interleukin; TNFα, tumor necrosis factor alpha

Full size image

Clinical correlates of TNF-α in children with CKD

Significant differences between children from different TNF-α tertiles were found in PTH, serum phosphorus, ALP, BAP, P1NP, and height z-score (Fig. 2). There were no differences in age, sex, race/ethnicity, CKD etiology, GFR, 25(OH)D, and systolic and diastolic blood pressure between children from different TNF-α tertiles (Table 2). The percentages of children receiving calcitriol, nutritional Vitamin D, phosphate binders, and growth hormone were not significantly different between TNF-α tertiles.

Fig. 2
figure 2

Associations between TNF-α and biomarkers of chronic kidney disease-mineral and bone disorder and growth in children with chronic kidney disease: a parathyroid hormone (PTH), b alkaline phosphatase (ALP), c phosphorus, d bone alkaline phosphatase (BAP), and e procollagen type 1 N-terminal propeptide (P1NP) each positively correlate with TNF-α. f Height z-score inversely correlates with TNF-α. n = 63. Gray area surrounding black regression lines represents 95% confidence interval

Full size image
Table 2 Characteristics of bone health and growth stratified by TNF-α tertiles
Full size table

Association between TNF-α, parameters of CKD-BMD, and growth

In multivariate analyses, PTH, ALP, serum phosphorus, and height z-score were significantly associated with TNF-α in all models (Table 3). BAP was significantly associated with TNF-α in models 1–3 and showed a trend toward significance (p = 0.052) in model 4. Similarly, P1NP was significantly associated with TNF-α in models 1 and 2 and showed a trend toward significance in models 3 (p = 0.052) and 4 (p = 0.06). Association between TNF-α and markers of CKD-MBD also remained significant after adjustment for rhGH use (Table S3). Thus, our multivariate analyses indicated that the relationship between TNF-α and CKD-BMD parameters, as well between TNF-α and height in children, was independent of age, sex, kidney function, and use of active vitamin D analogs.

Table 3 Multivariate analysis of the relationship between serum TNF-α, parameters of CKD-MBD, and height
Full size table

Discussion

The pathophysiology of CKD-MBD and interrelated growth impairment in children with CKD remains incompletely understood. Specifically, the role of inflammation in the development of CKD-MBD and growth failure in pediatric CKD has not been elucidated. In this study, we identified a positive independent association between TNF-α and biomarkers of CKD-MBD, and an inverse association between TNF-α and height z-score in a cohort of children with CKD.

Cytokine profiling in our cohort identified TNF-α as the only cytokine associated with PTH after adjustment for GFR, among the 9 cytokines that we measured. Egli-Spichtig et al. recently conducted a similar cytokine profiling in an adult cohort and reported that TNF-α was the only cytokine associated with FGF23 [24]. IL-6, IL-1β, and IFN-y were not associated with FGF23 in the study of Egli-Spichtig et al. and were not associated with PTH in our study. Interestingly, Glosse et al. demonstrated that TNF-α mediated the stimulatory effect of high-fat diet on FGF23 formation in mice [31]. However, the role of TNF-α in the development of CKD-MBD is likely not exclusive, and other cytokines may also play a role. Indeed, McKnight et al. recently reported that IL-1β stimulated production of FGF23 at the onset of CKD in juvenile mice [32]. Wallquist et al. found that IL-12 and RANTES correlated with FGF23 in adult patients with pre-dialysis CKD [33]. Nevertheless, among several cytokines that were implicated in CKD-MBD, TNF-α has emerged as a promising target, which was consistent with our findings.

The link between inflammation and mineral metabolism has been previously reported in adult patients with CKD [22, 34]. Specifically, the association between TNF-α and serum phosphorus has been reported in adults with CKD [35]. However, this association has not been previously reported in children with CKD. In our pediatric study, we identified significant independent associations between TNF-α and several biomarkers of CKD-MBD, including serum phosphorus, PTH, total and bone-specific alkaline phosphatase, and P1NP. Vitamin D was reported to act as an inhibitor of inflammation in CKD [36,37,38]. However, we did not find an association between TNF-α and serum 25(OH) vitamin D in our cohort.

While the pathways linking TNF-α to CKD-MBD remain to be fully elucidated, there are several possible mechanisms by which TNF-α could be aggravating CKD-MBD. TNF-α inhibits the expression of RUNX2, a major transcription factor associated with osteoblast differentiation [39]. At the same time, TNF-α is a potent stimulator of osteoclastogenesis [40]. Furthermore, inflammation appears to stimulate production of FGF23 in CKD, both directly through upregulating FGF23 production and cleavage by osteocytes, as well as indirectly through activation of hepcidin and induction of hypoferremia, among other potential mechanisms [41]. Further studies are needed to delineate the mechanisms underlying the effects of TNF-α in CKD-MBD, particularly elucidating the role of soluble receptors for TNF-α.

The possible role of inflammation in growth impairment of children with CKD has been acknowledged in the literature [42]. The growth-inhibiting effect of inflammation has been traditionally explained by a suppressive action of inflammation on the nutritional status. TNF-α exhibits well recognized anorexigenic effects, which gave TNF-α its synonymous name, cachectin. Indeed, anorexia was observed following acute administration of TNF-α to mice [43]. Administration of recombinant TNF-α increased energy expenditure in patients [44]. In a meta-analysis, serum levels of TNF-α were elevated in patients with anorexia nervosa compared to controls [45]. In patients with CKD 5, TNF-α and its genetic variants were the major contributors to the development of malnutrition inflammation syndrome [46]. Inflammation may affect growth also by disrupting the growth hormone (GH), insulin-like growth factor 1 (IGF1) signaling. In patients with chronic liver disease, TNF-α appeared to be a major factor responsible for acquired GH resistance [47]. Children with sickle cell disease who had higher TNF-α suffered from more severe growth impairment [48]. Despite this evidence, the direct association between inflammation and growth delay in children with CKD has not been established to date. In our study, we found an inverse association between TNF-α and height z-score in children with CKD, which was independent of GFR. These data provide previously missing direct evidence for the role of inflammation in growth impairment of children with CKD. The suppressive effect of TNF-α on growth could be mediated by its effects on bone health, nutritional status, and GH-IGF1 signaling.

Whereas experience with anti-cytokine therapies in children with CKD has been limited [11], they have been widely and successfully used in other inflammatory conditions. Several TNF-α inhibitors are FDA-approved for various indications, including monoclonal antibodies such as infliximab and adalimumab, and receptor fusion protein (etanercept). In several chronic inflammatory disorders of childhood, studies have established protective effects of TNF-α inhibitors against bone disease and growth impairment. In children with inflammatory bowel disease, anti TNF-α therapy improved trabecular bone mineral density and cortical structure [49], as well as statural growth, which was not solely due to a steroid sparing effect [50]. TNF-α inhibition ameliorated bone loss [51] and restored impaired growth velocity in children with juvenile idiopathic arthritis [52].

The strengths of this study included unbiased cytokine profiling and comprehensive assessment of both traditional and novel biomarkers of CKD-MBD in a well-defined cohort of children with CKD. Our study had limitations, including single center cohort and relatively small sample size, as well as not assessing FGF23. We had only 11 children who were receiving rhGH in the cohort. Establishing causality was beyond the scope of our study. The relationship between inflammation and CKD-MBD may be bi-directional. Indeed, FGF23 directly stimulates production of pro-inflammatory cytokines by the liver in CKD [53]. Dietary phosphate loading increased serum TNF-α in uremic rats [54]. After parathyroidectomy, bone expression of TNF-α significantly decreases in hemodialysis patients [55]. Multicenter and prospective clinical studies, as well as additional basic science work, would be warranted to further elucidate the role of TNF-α in pediatric CKD.

In conclusion, this study demonstrates an independent positive association between circulating TNF-α and several biomarkers of CKD-MBD in children with CKD. Furthermore, we found an inverse independent association between TNF-α and height z-score in our cohort. These data call for attention to the role of TNF-α in bone health and growth in children with CKD and warrant studies that would examine the potential of TNF-α as a therapeutic target in CKD.