Cystatin C serum levels in healthy children are related to age, gender, and pubertal stage

This article is structured according to the STROBE Checklist (Strengthening the Reporting of Observational Studies in Epidemiology) [25]. As part of the Leipzig Research Centre for Civilization Diseases (LIFE), the population-based cohort study LIFE Child has started recruiting urban, primarily healthy infants, children, and adolescents in Leipzig (Germany) in 2011. This large population-based cohort has already been used to establish reference intervals for serum lipids [26], liver enzymes [27], and iron-related blood parameters [28] in children. The examinations take place in the LIFE Child study center and are carried out by trained medical staff using highly standardized procedures [29, 30]. LIFE Child pursues the Declaration of Helsinki [31] and has been approved by the Ethics Committee of the University of Leipzig (Reg. No. 264-10-19042010). It is registered under the NCT trial number 02550236. All data were appropriately anonymized to comply to the German data protection law. More information including the recruitment process and repetitive examinations can be found in Poulain et al. and Quante et al. [29, 30].

In this study, all participants of the LIFE Child cohort having valid CysC measurements taken between 2011 and 2017 (2926 participants) were included. Children with an age of 0–16 years can participate in the study and receive invitations for follow-up examinations until the age of 18 years. Furthermore, during the first year of life, there are visits at the age of 3, 6, and 12 months. Thus, participants provided data on one to six follow-up visits. We excluded all participants with renal anomalies, nephrolithiasis, or febrile urinary tract infections (118 participants). This information was obtained through computer-assisted personal interview and sonography diagnostic. Furthermore, we identified and excluded four remaining isolated extreme values of CysC (< 0.4 mg/l) as well as one participant with implausible anthropometric data. Thus, a total of 6217 observations of 1337 females and 1466 males (age 0–18 years) are included in this study (Fig. 1).

We examined the CysC levels depending on age and gender. Furthermore, we examined sCrea levels in order to analyze whether or not our data is comparable to sCrea cohorts of earlier studies. For CysC and sCrea, morning venous blood was drawn from each participant by venipuncture using serum monovettes (Sarstedt AG&Co, Nümbrecht, Germany). The analyses were performed by the Institute for Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics (ILM), University Hospital Leipzig. Serum samples were analyzed on an automated laboratory analyzer, Cobas8000 (Roche Diagnostics, Mannheim Germany), according to manufacturer’s protocol. sCrea was measured with an enzymatic assay (Roche Diagnostics). Measurement of serum CysC was performed using the turbidimetric immunoassay (PETIA) Tina-quant® Cystatin C (Roche Diagnostics). The primary measurement range is 0.4–8.0 mg/l. Traceability of the method was standardized against a Roche in-house reference preparation of recombinant human CysC. In April 2015, the Tina-quant® Cystatin C-assay was advanced to the second generation (n = 2070 during second versus n = 4401 observations during the first generation), now standardized against the international reference material ERM-DA471/IFCC [30]. Between October 2011 and April 2015, the variation coefficient of control level 1 varied between 1.3 and 6.4% (mean 3.0%), control level 2 varied between 0.9 and 4.5% (mean 2.0%). The primary measurement range of the Tina-quant® Cystatin C 2nd generation is 0.4–6.8 mg/l. Comparative measurement of 143 serum samples was performed between Tina-quant® Cystatin C and Tina-quant® Cystatin C 2nd generation. Using the MedCalc (MedCalc Software bvba, Belgium), a Passing-Bablok-regression [32] and Bland-Altman-plot [33] were calculated (Online Resources 2 and 3). The comparison showed a good conformity between the first and second generation of the immunoassay Tina-quant®. The mean bias accounts for 0.03 mg/l. The CysC reference values were not corrected for this clinically not relevant bias, which is also comparable to the usual batch effects.

Height, weight, BMI, and puberty status were taken into account as potential confounding variables. BMI was calculated using height and weight measured by instructed and qualified personnel applying standardized procedures and regularly calibrated devices (a stadiometer with a measurement accuracy of 0.1 cm and a Seca 701 scale with a measurement accuracy of 50 g). The puberty status was examined by means of Tanner stages and assessed by trained staff members [34, 35].

The percentiles were estimated applying generalized additive models for location, shape, and scale as implemented in the gamlss package combined with a resampling method using the ChildSDS packages as described by Vogel et al. [36, 37]. All statistical analyses and visualization were done using the R-Software (version 3.3.2) [38]. To examine the influence of anthropometric measures on CysC levels, we stratified the data into four age intervals of linear course identified through visual inspection and local non-parametric regression (infancy 0–2 years, childhood 2–11 years, and adolescence with 11–15 and 15–18 years; see also Fig. 2 created with ggplot) [39]. Linear modeling was favored due to better interpretability. Hierarchical linear regression analyses (backward deletion) were applied to determine the effects of the independent variables on CysC levels (lmer-function of the R-package lme4) [40]. To account for repetitive measurements in follow-up participants, the subject was added as random effect on the intercept. T tests were used to compare mean CysC and sCrea levels of boys and girls (Table 1).

The percentiles of sCrea levels for girls and boys are provided in the Online Resources 1 and 4. First, we evaluated the sCrea distribution to show that the LIFE Child cohort is comparable to other studies and, therefore, a representative sample of the population; sCrea levels rise continuously until the age of 12.5 years (ß-slope = 2.744 μmol/l/a = 0.031 mg/dl/a; a = period of 1 year of life) for both boys and girls. At that age, median sCrea levels are 53 μmol/l = 0.60 mg/dl in girls and 55 μmol/l = 0.62 mg/dl in boys. sCrea levels increase more rapidly in 12.5- to 18-year-old boys (ß = 5.905 μmol/l/a = 0.067 mg/dl/a). In contrast, the slope (ß) in 12.5- to 18-year-old girls is constant at around 2.8 μmol/l/a = 0.032 mg/dl/a. From the age of 13 years, boys exhibit significantly (p < 0.01) higher sCrea levels than girls.

The distribution of CysC levels in the LIFE Child cohort, the percentiles and the degree of freedom spread, skewness, location, and kurtosis parameters are shown in Fig. 3 and Table 3. Measurements were elevated for children between the age of 3 months and 18 years. The median CysC serum concentrations are highest in toddlers (males 1.06 mg/l, females 1.04 mg/l). They decrease during the first 2 years of life (ß = − 0.154 mg/l/a) to slightly but significantly lower levels (p < 0.001; males: 0.88 mg/l; females: 0.87 mg/l) and remain constant during childhood until the age of 11 years. The mean CysC values in girls and boys do not differ significantly at this age (p > 0.05). While the serum levels of female adolescents start to decrease at 11 years (ß = − 0.023 mg/l/a), those for male adolescents increase (ß = 0.028 mg/l/a). Thus, at the age of 13 years, CysC levels differ significantly between males and females (p < 0.001). After reaching 15 years of age and median levels of 0.97 mg/l in males and 0.84 mg/l in females, CysC levels of male participants drop again (ß = − 0.033 mg/l/a). In our study cohort, we found that CysC levels in males and females remained significantly different until the age of 18 (p < 0.05) with the highest and most significant difference at the age of 15 years (p < 0.001, mean CysC levels 0.97 mg/l in males and 0.84 mg/l in females).

Among all participants of the LIFE Child cohort, the scale remains constant as indicated by the sigma-value of 0.12–0.14 mg/l (Table 3).

To identify potential influential factors for the changes in CysC levels during infancy and adolescence, we correlated height, weight, BMI, the Tanner stage, and age with the CysC concentrations for boys and girls separately. For better interpretability, linear regression models were applied to four different intervals of linear course identified through visual inspection (infancy 0–2 years, childhood 2–11 years, adolescence with 11–15 and 15–18 years; Fig. 2 and Table 1). All effects are corrected for age (except age itself) and repetitive measurement in follow-up participants.

In a simple linear regression analysis of the participants aged 0–2 years, age, height, and weight were shown to be negatively associated with CysC serum concentrations (p < 0.001), whereas BMI does not show a significant effect. In hierarchical regression analyses, CysC levels were negatively correlated with height (ß = − 0.010 mg/l/cm, p < 0.001) as well as weight (ß = − 0.033 mg/l/kg, p < 0.001).

Using stepwise multiple regression models including age, height, weight, BMI, and Tanner stages CysC levels of male participants between the age of 11 and 14 years showed a strong dependency on puberty status (p < 0.001). Even after adjustment for age, CysC concentrations are significantly higher in pubertal boys (especially in Tanner stages three and four with ß ≈ 0.1 mg/l/Tanner stage, p < 0.001), compared to prepubertal boys. In females at the same age, pubertal stage is also the strongest predictor of CysC levels (p < 0.001), but in contrast to males negatively associated with CysC levels (ß = − 0.093 mg/l/Tanner stage, p < 0.01). We observe a peak of CysC concentrations in female adolescents during pubertal stage two that however is not significantly different from prepubertal CysC concentrations. Weight and BMI show no significant effect on CysC levels in adolescents. Height is another predictor of serum CysC concentrations in pubertal males (ß = 0.003 mg/l/cm, p < 0.001).