New reference intervals for endocrinological biomarkers in pediatric patients: what can we learn from the LIFE child study?

Introduction

Clinically relevant reference intervals are a mandatory precondition for the interpretation of endocrine biomarker measurements. The absolute level of the biomarker is determined by the basal stage and functional capacity of the hormone-producing cells. Additionally, various confounders may regulate the final circulating level of each hormone. In pediatric patients, age and sex are the most important covariates because both determine growth and maturity during childhood. At the onset of adolescence, puberty-induced sex hormones can be observed and may stimulate the hormone levels of the related endocrine axis (e.g., the bone turnover marker). Although pubertal development can be determined by the individual Tanner stage, the role of puberty as an endocrine confounder for reference values remains largely ambiguous. This is mainly caused by the issue that patients, physicians or study assistants frequently refuse the examination of pubertal features of adolescents.

Recently, overweight or obesity has been found to be prevalent in every sixth child and tends to persist into later life [1]. An increase in the amount of adipose tissue may have effects on growth and maturity during childhood, but also on the pubertal development of children and adolescents. This can be observed, for instance, as increasing levels of the sex hormones testosterone and estradiol from Tanner Stages 1–5 or by the increase of dehydroepiandrosterone sulfate over the adrenarche. However, it remains unclear whether or not the secretion of hormones from another axis (e.g., bone markers or thyroid hormones) is under the control of BMI and puberty.

The LIFE Child study (clinical trial number NCT02550236) is a population-based longitudinal cohort study conducted at the Research Center for Civilization Diseases in Leipzig. It aims to monitor healthy child development from birth to adulthood and to understand the etiology of lifestyle diseases such as obesity [2]. The study consists of three interrelated cohorts; the birth cohort, the health cohort, and the obesity cohort (Figure 1). Subjects were invited to participate at 3, 6 and 12 months of age, followed by annual invitations from 1 to 18 years of age. Due to the large numbers of healthy infants, children and adolescents included in the study, the use of their laboratory and clinical data to establish reference values for endocrine biomarkers was a temptation for each clinical chemist. Moreover, the LIFE study investigated various confounders in detail, such as puberty, weight status, medication, and others.

Figure 1:

Study cohorts in the LIFE Child study and their interrelations. W, weeks of gestation; M, age of children in month; Y, age of children in years. With permission from ref. [2].

Thereby we were able to assess whether confounders such as puberty, body mass index (BMI), and the frequently neglected use of contraceptives have effects on age- and gender-specific reference intervals for the thyroid parameters thyroid-stimulating hormone (TSH), free triiodothyronine (FT3), free thyroxine (FT4), and calcitonin (Ct); the bone markers osteocalcin, procollagen type 1 N-propeptides (PINP) and carboxy-terminal cross-linking telopeptide of type 1 collagen (CTX-I), the calciotropic hormones 25-hydroxyvitamin D (25(OH)D) and parathyroid hormone (PTH); as well as for the steroids cortisol, progesterone, 17-hydroxyprogesterone, androstenedione, testosterone, estradiol, dehydroepiandrosterone sulfate, and aldosterone. Accordingly, this article presents a brief overview from the current set of four published articles on endocrine biomarkers from the LIFE Child study [3], [4], [5], [6].

Materials and methods

Results

The gender-specific age-dependent reference values derived from our studies are presented for the thyroid parameters in Figure 2; for Ct in Figure 3, for the bone marker and calcium homeostasis parameters in Figure 4 and for steroids in Figure 5.

Figure 2:

Reference percentiles for TSH, FT3, and FT4 by age and sex. The 2.5th (dashed line), 50th (solid line), and 97.5th (dashed line) percentiles for males and females are shown for TSH (A), FT3 (B), and FT4 (C). The intensity of the shading of the hormone level data points for the reference cohort indicates the relative hormone SDS value. SDS, standard deviation score. With permission from ref. [3].

Figure 3:

Serum Ct measurements of healthy boys (A) and girls (B) depending on age.

A solid line indicates the 2.5th, 50th, and 97.5th percentiles. The 99th percentile was shown as a dashed line. Re-sampling levels were marked by the shaded areas around the percentiles. With permission from ref. [4].

Figure 4:

Percentiles for osteocalcin (OC), PINP, CTX-I, 25(OH)D, and PTH by age and gender.

Boys’ and girls’ percentiles are shown (2.5th, 50th, 97.5th) for the OC (A), PINP (B), 459 CTX-I (C), 25(OH)D (D), and PTH (E) levels in the reference cohort (transparent dots), and the distribution of levels for obese children (BMI-SDS >1.88) are shown as black dots. With permission from ref. [5].

Figure 5:

Steroid hormone levels change throughout the lifespan in males and females. Blue: males; red: females without oral contraceptives (oC); black: females with oC. With permission from ref. [6].

Moreover, details on the effects of age, gender, puberty, BMI-SDS and oC on the reference values are summarized in Table 1. The most important results are described below:

1. TSH and FT3 demonstrated a decrease in hormone levels before puberty, whereas the age-dependent changes in FT4 were rather negligible (Figure 2). Gender-dependent small variations were observed for a restricted age range in TSH and FT3. During puberty, TSH and FT3 were lowest during P5 and P3, respectively. FT4 was increased at P3 or P4. TSH-SDS and FT3-SDS demonstrated a significant positive and FT4-SDS a significant negative correlation with BMI-SDS, leading to distinctly different thyroid values in obese subjects [3].

2. Ct showed an age-dependent decrease, especially in infants (Figure 3). Puberty or high BMI-SDS in children older than six years were associated with slightly reduced values [13].

3. Osteocalcin, PINP and CTX-I showed a distinct age- and gender-dependence with peak levels at 10–11 years (girls, Tanner 3) and 13 years (boys, Tanner 3–4) (Figure 4). For 25(OH)D and PTH, we found no distinct age-dependency. The maximum of 25(OH)D in infants is likely to be induced by exogenous supplementation. During puberty, the values of osteocalcin, P1NP and CTX-I reached their maximum later in boys than in girls. A faint peak was also detected in P3 or P4 for PTH and there was a respective nadir for 25(OH)D in P4. Interestingly, the mean SDS levels in our obese cohort differed significantly (p<0.001) from the values in the reference cohort: Osteocalcin by −0.44, PINP by −0.27, CTX-I by −0.33, 25(OH)D by −0.43, PTH by +0.44 [5].

4. Excluding cortisol and aldosterone, all other steroid hormones showed at least slightly increased serum values during mini-puberty in the age interval of around 0.5–1 year (Figure 5; Ref. [6]). High aldosterone levels during the same period revealed physiological hyperaldosteronism due to renal tubular immaturity and mineralocorticoid insensitivity. A similar trend could be seen during the actual puberty, whereas cortisol appeared to demonstrate slightly increasing values. We also found the well-known difference between boys and girls with higher levels of male testosterone and female estradiol and progesterone, at least with puberty onset. Interestingly, distinctly increased cortisol and to a lower extent testosterone levels were measured when adolescent girls used oC. Unfortunately, this effect was not investigated in the non-steroid hormone studies.

Discussion

Adult reference intervals for hormones are frequently presented by the manufacturer’s package insert as applicable for routine diagnostics of endocrinological diseases in infants, children or adolescents. However, if adult reference intervals are applied in pediatrics, the number of diagnostic errors will be distinctly higher due to differences in the physiology of both groups, especially the hormone secretion. Our study found that covariates such as age and gender as well as confounders such as puberty, BMI, or contraceptive use were responsible for significant variations in the reference intervals of all 17 investigated hormone biomarkers during infancy, childhood and adolescence. Although these variations are important for the clinical interpretation for most of the measures, single parameters (e.g., FT4) demonstrated only minor covariate- and confounder-related changes that are largely negligible for clinical practice.

We observed age-related changes in reference intervals between birth and puberty. The physiological post-partum maturation was related to an adjusted secretion of hormones, especially in bone-, growth- and energy-related hormone axes. Moreover, minipuberty [14] is characterized by an increase in sexual hormones with huge inter-individual variations and has to be considered while interpreting diagnostic or therapeutic findings.

During puberty, gender is the major influencing factor, especially as the individual stages of puberty are delayed in boys compared to girls. The stages themselves represent the different puberty-related secretion capacity of sex hormones and are, therefore, the most important confounder of hormones during adolescence. Accordingly, we found respective changes in hormone levels for 15 of our 17 parameters, although some of them were rather small. Moreover, other hormones, such as the biomarkers of the growth-hormone insulin-like growth factor (IGF) axis, IGF-I [15] and IGF binding protein-3 [16], showed a corresponding peak in their serum levels during puberty.

Interestingly, we were able to demonstrate an effect of BMI on thyroid function, bone markers and the calciotropic hormones 25(OH)D and PTH. The relationship between BMI and TSH was recently suggested by an Israeli group by a data mining approach [17]. Accordingly, in obese children, conspicuous findings regarding these two axes have to be interpreted with caution. This is also true for cortisol and testosterone in girls taking oC: the drug-induced increased binding proteins for cortisol and testosterone led to dramatically increased values for both steroids.

The outstanding unique feature of our study is the huge sample of primarily healthy subjects whose ages covered the entire interval of infancy (excluding the neonatal period), childhood and adolescence. All of them were examined at a single study center, following the same highly standardized methods carried out by professionally trained staff. In the literature, such so-called “direct” procedures for establishing reference intervals have been very rare due to the large effort and costs. To the best of our best knowledge, there was only one analogous approach known as the “Canadian Laboratory Initiative on Pediatric Reference Intervals” (CALIPER) [18], which recruited subjects from across Canada, but was not primarily focused on endocrine parameters. Due to the large number of included subjects, LIFE Child and CALIPER were able to detect statistically significant moderate or even small confounding effects on the reference intervals. Moreover, there was no need to integrate samples or data from different clinical studies obtained from different centers located in different international regions, an approach that frequently fails in quality control or in collecting detailed, reliable and consistent data from all patients.

Recently, “indirect” procedures have been applied more and more often to establish pediatric reference intervals (e.g. data mining) in lieu of extensive and laborious “direct” reference interval studies. However, such “indirect” procedures need routine findings from huge numbers of patients released from the laboratory archive. Thus, this procedure is usually applied for parameters from classical clinical chemistry and blood cell analysis [19, 20]. Publications of validated hormone reference intervals based on data mining procedures are largely restricted to thyroid parameters [21]. More research is needed to develop a respective approach covering the entire spectrum of hormones for diagnostics in pediatric endocrinology.

The reference intervals for the thyroid parameters, the bone markers, and parameters of calcium homeostasis were based on the measurement using an ECLIA method; for steroid measurement, we used the liquid chromatography tandem mass spectrometry (LC-MS/MS). The latter method was developed to improve the comparability of results from different analytical methods and to overcome potential interference from immunoassays [8]. However, it is still infrequently used in routine labs so far.

Strictly speaking, our reference intervals are valid only for a Middle-European, Caucasian pediatric population. However, qualitatively, the confounder-dependent course of our biomarkers should be considered and interpreted along with the patient’s clinical appearance, independently on the used analytical method. Moreover, our reference intervals can be adapted for immunoassays of different manufacturers or for different LC-MS/MS approaches after mathematical transformation and respective validation according to CLSI guideline C28-A3 [22].

We conclude that age- and gender-specific reference intervals are important for the correct interpretation of hormone measurements in infants, children and adolescents. In addition, confounding factors such as puberty, BMI and the use of contraceptives should be considered when hormones are used for the diagnosis and treatment monitoring of endocrine diseases.