Nocturnal Salivary Cortisol Is an Accurate Non-Invasive Test to Assess Endogenous Hypercortisolism in Children with Obesity and a Clinical Phenotype Suspicious for Cushing’s Syndrome

The increase in the prevalence of childhood obesity is associated with more children being referred for investigation of endocrine disorders that cause or contribute to obesity [1, 2]. Cushing’s syndrome (CS) is a rare disorder in children and adolescents characterized by central obesity and clinical signs and symptoms of chronic cortisol hypersecretion [1]. Most cases of simple obesity can be distinguished from CS on clinical grounds. These children tend to be tall with normal or increased growth velocity for their age, whereas children with CS tend to be short with growth failure [1]. Therefore, only a minority of children require screening for CS after a thorough clinical assessment. A progressive rise in body mass index during infancy, associated with growth retardation, is the classic presentation of hypercortisolism states in children [1]. In older children, the clinical picture can also include acanthosis nigricans, hirsutism, striae, and metabolic disorders such as glucose intolerance and dyslipidaemia [1, 3], some of them commonly observed in exogenous obesity. As a consequence, CS constitutes one of the most challenging diagnostic assessments for paediatric endocrinologists that must therefore be excluded in the obese by accurate first-line tests to avoid false-positive results and further biochemical investigations [4]. The circadian rhythm of cortisol is established after the first year of life and remains stable thereafter [5, 6]. It is characterized by higher serum cortisol concentration in the morning that gradually declines, reaching the lowest level by 11:00 p.m. The loss of the nadir of cortisol at nigh constitutes the biochemical hallmark in hypercortisolism assessment regardless of its cause. An elevated serum mid-night cortisol concentration is a highly specific and sensitive (>90%) indicator of CS in children referred for investigation [1, 2]. However, this test requires inpatient admission and in some cases the cut-off values used are derived from only a small number of patients that not always represent the clinical challenging problem [7]. Moreover, the cut-off points of most studies are highly influenced by the method used [8]. Salivary cortisol reflects the biologically free cortisol fraction and is a non-invasive first-line test to exclude CS [1, 9‒11]. Despite this, no consensus cut-off values have been suggested for children with obesity and clinical signs suggestive of hypercortisolism [2, 10, 12]. In addition, some divergences in cortisol concentration have been observed in childhood obesity, probably reflecting not only differences in preanalytical and analytical issues, but also differences in the clinical characteristics of the study population [3, 13‒18].

In the present study, we aimed to evaluate the diagnostic efficiency of salivary cortisol at 11 p.m. and the decrease of salivary cortisol at night in children with obesity and clinical overlapping signs of CS.

We conducted a prospective case-control study with retrospective analysis following the Standards for Reporting of Diagnostic Accuracy (STARD) [19]. We included children aged 2–18 years with obesity (body mass index score of standard deviation [BMI-SDS ≥2.0]) and increasing BMI-SDS for at least 1 year prior to enrolment. Argentinian population reference data were used for height and weight SDS [20]. BMI was calculated according to Cole et al. [21]. Children attended the endocrinology unit between March 2014 and March 2022 and had clinical and/or biochemical follow-up for at least 2 years. Genetic syndromes and iatrogenic hypercortisolism were exclusion criteria. The clinical features of the study group were listed according to the Endocrine Society Clinical Practice Guidelines for CS and categorized as follows: group A, features particularly suggestive of CS in still growing children such as decreasing height percentile and increasing weight; group B, features that are unusual for age and suggestive of CS such as moon face and striae, and group C, features with overlap between CS and other disorders and are hence of less discriminatory value, such as hypertension, headache, insulin resistance, acanthosis nigricans, ovarian dysfunction in girls, hypertrichosis/hirsutism and dyslipidaemia [2]. Severe obesity was defined as BMI ≥+3.0 SD. Dyslipidaemia was defined as total cholesterol, low-density lipoprotein cholesterol or triglycerides above the 95th percentile of the control population and/or high-density lipoprotein cholesterol below the 10th percentile [22]. The Homeostatic Model Assessment for insulin resistance (HOMA-IR) was calculated. A cut-off of >2.5 for HOMA-IR was used to define insulin resistance according to own local reference [23]. Parents and children signed informed consent and assent as appropriate.

Only patients with clinical features from groups A and B were the control group to calculate the diagnostic accuracy of the main outcome measures. Therefore, the control group consisted of children with BMI-SDS ≥2.0 with features strongly indicative or suggestive of paediatric CS. In this cohort, hypercortisolism was rule out by at least two 24-h urinary free cortisol (24-h UFC) <90 μg/24 h and an exhaustive clinical assessment for at least 2 years (non-CS with clinical signs of group A and B). In none of these children did the clinical picture progress or the Cushing’s phenotype accumulate, reducing the likelihood that the syndrome was present [2]. The case group were patients with CS confirmed by 24-h UFC ≥90 μg/24 h, measurement of the cortisol filtered into the urine between 22.00 and 23.00 h, expressed as ng/mg creatinine (UFC 22–23) ≥28 ng/mg creatinine, elevated plasma ACTH (>80 pg/mL), lack of suppression in the dexamethasone test (>1.8 μg/dL), pituitary tumour in CNS magnetic resonance imaging or lateralization by catheterization of the inferior petrosal sinus [2]. Two patients with adrenal tumour included abdominal TAC. All children collected saliva samples in the morning between 7 and 8:00 a.m. (mSC) and at 11 p.m. (nSC) on the same day according to own laboratory preanalytical instructions [24]. Briefly, the parents received written instructions for saliva collection. Children needed to be rested before sampling (>5 min), saliva accumulated in the floor of the mouth was spitted into a conic graduated test tube every 60 s for no longer than 3 min (minimum volume required 0.8 mL). The mSC and nSC results were used to calculate the percentage decrease of cortisol at night (%D) as follows: %D = [(mSC-nSC)/mSC*100]. For the assessment of the circadian rhythm of cortisol, we evaluated the following variables: absolute value of nSC and the %D. They were the main outcome measures of the study.

Saliva samples were stored at −20°C until processing, thawed at room temperature, centrifuged at 3,500 rpm for 15 min in a refrigerated centrifuge (4°C), and the supernatant was carefully isolated to avoid pellet contamination. Cortisol was measured in saliva by electrochemiluminescence immunoassay (ECLIA) in a Roche Cobas e411 autoanalyser. This immunoassay has a lower limit of quantification of 2.0 nmol/L, intra-assay coefficient of variation of 3.5%, and 2.5% for a mean concentration of 11.0 nmol/L and 29.2 nmol/L (Roche Salivary PreciControl, Lot No. 266657), respectively. Inter-assay CV <4.0%.

The sample size was estimated at 9 patients with confirmed CS, to achieve a sensitivity of 100% and specificity of 60% for the index test with a statistical power of 80% and α-error <5%. The main outcomes of the study were established by receiver operating characteristic curve (sensitivity, specificity, positive predictive value [PPV], negative predictive value [NPV] and their corresponding 95% confidence intervals). The distribution of the variables was assessed using the Shapiro-Wilks test; salivary cortisol was log-transformed to achieve a normal distribution. Paired t test was used to assess differences between morning and night SC within each child. Unpaired t test or one way ANOVA were used to assess differences between groups. Fisher’s exact test was used to investigate differences in proportions of categorical variables. The level of significance was set at p < 0.05. Data were analysed using GraphPad Prism version 8.00 for Windows (GraphPad Software, San Diego, CA, USA; www.graphpad.com).

Figure 1 shows the flow diagram of clinical phenotype categorisation of children with obesity. One hundred and twelve children met inclusion criteria. Thirty-seven children were excluded due to incomplete clinical and/or biochemical follow-up. The study group consisted of 75 patients with obesity (non-CS group) who were classified according to their clinical presentation. Fifty-three out of 75 children had only less specific features overlapping with CS (group C) and they were excluded from the study. The control group consisted in 18 females and 4 males that presented features strongly indicative (group A) and/or suggestive of paediatric CS (group B): growth delay in 8 (8/22, 36.4%), moon face in 10 (45.5%) and striae in 9 (40.9%). The case group consisted of 10 children with confirmed CS. Eight children out of 10 had Cushing’s disease (4 boys and 4 girls), one boy had an adrenal tumour and another boy had an ectopic ACTH-secreting carcinoid tumour. Clinical and biochemical characteristics of obese non-CS groups and children with confirmed CS are shown in Table 1. Except for one girl in the CS group, all children were pubertal. BMI-SDS, severe obesity, body surface area, proportion of overlapping less discriminating features of CS and glucose metabolism parameters were similar in the 3 groups of children. In CS, height-SDS was significantly lower than in the non-CS groups, while headache and hypertension were more frequently observed.

Concentrations of mSC were similar between the obese non-CS groups. Night SC was also similar, with a significant decrease from the morning levels in both non-CS groups (Table 1; Fig. 2). When considered as a single group, 57/75 (76%) non-CS children had undetectable nSC (<2.0 nmol/L), with a 97.5th percentile of 4.5 nmol/L. Children with CS had higher mSC and nSC concentrations than non-CS children (p < 0.0001; Table 1; Fig. 2). They also had higher basal serum cortisol, plasma ACTH and 24-h UFC than obese non-CS children with (A) and (B) clinical signs (Table 1). As a group, CS did not decrease nSC compared to mSC (p = 0.59). This was also the case in the subgroup of 6 children in whom a no significant reduction was observed (p = 0.28, Fig. 2). Children with CS had a lower percentage decrease of salivary cortisol at night compared to non-CS groups (p < 0.0001, Table 1).

Children with obesity and the best discriminating signs of CS ([A] and/or [B] clinical signs; control group, n = 22) and children with CS (case group, n = 10) were included to assess of the diagnostic accuracy of nSC and the (%D). A cut-off concentration for nSC of ≥8 nmol/L had a sensitivity, specificity, PPV and NPV of 100% for the diagnosis of CS (Table 2). The cut-off obtained for %D was 50%. No child with CS had a %D ≥50%, but 6/22 children in the control group had a %D below the cut-off, resulting in a lower overall diagnostic accuracy of 81% compared to nSC (Table 2).

CS is one of the most challenging diagnoses for paediatric endocrinologists. This study evaluated the diagnostic accuracy of salivary cortisol as a first-line non-invasive test to assess hypercortisolism in obese children with overlapping clinical features most commonly seen in CS. The main result of the study is that salivary cortisol at 11 p.m. ≥8.0 nmol/L is an accurate biomarker to confirm CS with a sensitivity and specificity within 100%. This cut-off was obtained in a paediatric population following strict clinical inclusion and exclusion criteria. To the best of our knowledge, this assessment has not previously been performed in children using the current automated ECLIA method. The nSC was also useful in showing that children with exogenous obesity preserved the circadian rhythm of cortisol.

Less healthy eating and sleep behaviours as well as sedentary lifestyle may induce weight gain in children and changes in their biological rhythms. In the diagnostic work-out of patients with signs of CS, the loss of the nadir of cortisol at night constitutes the biochemical hallmark assessment regardless of the hypercortisolism cause [1, 2]. Urine collection over 24 h can be difficult in paediatric practice. In addition, obese children may have elevated UFC due to the inference of cortisol metabolites, limiting its usefulness [25].

Most of the salivary cortisol studies by the ECLIA method have been conducted in the adult population. To our knowledge, at least one study was performed in children with obesity [14]. However, it did not evaluate diagnostic the accuracy of nSC as screening test for CS. Following strict inclusion criteria for children with obesity and overlapping clinical features of CS and 10 age-matched children with confirmed CS, we found that nSC ≥8.0 nmol/L accurately confirmed the pathology. We also showed that the circadian rhythm of cortisol was preserved in obese non-CS cohorts, most of whom had either undetectable nSC or levels below the 97.5th percentile (4.5 nmol/L). Other cut-off values indicative of CS have been reported using reference liquid chromatography-tandem mass spectrometry in adults (nSC >4 nmol/L) and in children (nSC >2.4 nmol/L) [2, 7, 26], the latter obtained with very few index cases of Cushing’s disease and a cohort of children with obesity from an outpatient clinic [7].

Morning SC levels overlapped between the 3 groups of children; therefore, our results do not support the inclusion of morning salivary cortisol as an isolated test to rule out hypercortisolism in obese children. Furthermore, the percentage decrease in nocturnal salivary cortisol did not improve the diagnostic accuracy of nSC due to lower specificity. In 6 children with CS, nSC fluctuated toward a lower level, but this did not reach statistical significance. Similar findings were observed by Martinelli et al. [12] using the RIA method, and in some case reports of spontaneous plasma cortisol secretion prolife in adults with CS [27, 28].

The strengths of the present study are the sample size of children with CS. The careful selection of the control group (BMI-SDS ≥2.0, clinical suspicion of CS and biochemical follow-up of at least 2 years) allowed us to exclude possible cyclic CS. We also note some potential limitations. First, nocturnal salivary cortisol was measured in a single day, whereas two tests are recommended to improve diagnostic accuracy [2]. Therefore, the single measurement of nocturnal salivary cortisol should be viewed as a screening test, rather than a diagnostic test. The inclusion of a second salivary cortisol measurement increases the cost of the initial assessment and may not be necessary as other authors have found no significant diurnal variation between at least 3 saliva samples taken 4–8 days apart [29]. We would like to point out a grey area of uncertainty for nSC between 4.5 and 8 nmol/L, as none of our cohorts had values in this range. In these cases, we recommend further investigation of hypercortisolism with other biomarkers [1, 2] and/or repeat nSC with extreme care in preanalytical sampling to avoid false positives. Finally, we measured saliva cortisol by using an immunoassay rather than the gold standard liquid chromatography-tandem mass spectrometry for steroid quantification [26, 30]. As this technology is not always available in most endocrine laboratories, we consider that nSC can be widely used as a feasible, non-invasive first-line test in the diagnostic algorithm of children with obesity and overlapping clinical features of CS.

Salivary cortisol at 11 p.m. is an accurate and non-invasive first-line test to assess endogenous hypercortisolism in children with obesity and clinical suspicion of CS. In patients with nSC ≥8.0 nmol/L, other dynamic biochemical assessments and imaging studies should be indicated to further confirm the aetiology.

The authors would like to thank Dra. Solange Rosembrock and Ms. Mónica Campos from the Centro de Investigaciones Endocrinológicas “Dr. César Bergadá” (CEDIE) CONICET-FEI-División de Endocrinología, Hospital de Niños Ricardo Gutiérrez for their technical assistance; Dra. Liliana Franchioni de Muñoz, Dra. Liliana Karina Silvano, Dra. Gabriela Sobrero, Dra. Mirta Beatriz Miras and Dra. Malvina Signorino from Hospital de Niños de la Santísima Trinidad de Córdoba, and Dr. Gonzalo Perez from Hospital Materno Infantil “Dr. Héctor Quintana” de Jujuy.

The present study protocol was reviewed and approved by the Research Ethics Committee and the Teaching and Research Committee of Hospital de Niños Dr. Ricardo Gutiérrez of Buenos Aires, in accordance with the Declaration of Helsinki; Approval No. CEI 19.18. Parents and children signed informed consent and assent as appropriate.

The authors have no conflicts of interest to declare.

The study was supported by grants of Consejo de Investigación en Salud, Ministerio de Salud del Gobierno de la Ciudad de Buenos Aires, Argentina to M.G.B.

María Gabriela Ballerini, Analía Freire, María Eugenia Rodríguez, Luciana Brenzoni, Luciana Daga, Laura Cecilia Castro, Ana Carolina Arias Cau, Graciela Testa, Melina Gil, Débora Braslavsky, Ana Vieites, Ana Keselman, Ignacio Bergadá, Andrea Arcari, and María Gabriela Ropelato constitute contributors to the present paper that fulfil the ICMJE Criteria for Authorship stated at http://www.icmje.org/recommendations/browse/roles-and-responsibilities/defining-the-roleof-authors-and-contributors.html. This included a substantial contributions to the conception or design of the work, the acquisition, analysis, and interpretation of data for the work (M.G.B., A.V.F., A.A., and M.G.R.), patient enrolment (A.A., L.B., C.C., A.C.A.C., G.T., M.G., D.B., A.V., A.K., and I.B.), methodological expertise (M.E.R. and L.D.), and the critically review for important intellectual content (M.G.B. and M.G.R.). All authors gave the final approval of the version to be published and in Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

M.G.B., A.A., and M.G.R. are members of Carrera de Investigador en Salud, Gobierno de la Ciudad Autónoma de Buenos Aires.

All data generated or analysed during this study are included in this article. Further enquiries can be directed to the corresponding author María Gabriela Ballerini.