Usefulness of salivary cortisol as a marker of secondary adrenal insufficiency in paediatric patients

Background: The main cause of adrenal insuffi ciency (AI) in paediatric patients is prolonged treatment with corticosteroids. Determination of plasma cortisol (PC) during ACTH test is the most used adrenal function indicator in clinical practice. However, determination of salivary cortisol (SC), a simple test especially useful in children in order to avoid invasive procedures, can be used as an alternative technique for the diagnosis of adrenal disease.


Introduction
The activation of the hypothalamic-pituitary-adrenal axis in response to critical illness and the resulting release of cortisol from the adrenal cortex are essential to stress adaptation. Adrenal insuf iciency (AI) is described as the inability of adrenal glands to produce an appropriate hormonal secretion not only under stress but also in basal situation. Therefore, a low baseline plasma cortisol (PC) (< 5 μg/dL) and a poor cortisol response to stimulation with exogenous adrenocorticotropic hormone (peak < 18 μg/dL) are some of the de ining criteria of this condition [1,2]. It is well known that the main cause of AI in paediatric patients is prolonged treatment with exogenous corticosteroids, which is an iatrogenic cause derived from the increasing complexity of paediatric pathologies and the increased use of prolonged high-dose corticosteroid therapy.
In clinical practice, adrenal function is usually assessed by the total PC (determined by low-dose ACTH test). This implies the placement of a vascular access which is often a traumatic experience for children.
PC includes protein-bound fraction and serum-free cortisol. The latter constitutes the biologically active form of the hormone and is responsible for glucocorticoid activity on peripheral organs. Most of the circulating cortisol is bound to plasma proteins (over 90%), such as cortisol-binding globulin (CBG) and albumin, whereas only about 10% of circulating cortisol is free. Hence, the measurement of plasma-free cortisol level has been considered more representative of adrenal function (especially in critically ill adults and children) [1,2], because some conditions, such as hypoalbuminaemia or hypoproteinaemia (frequent in critically ill patients or in patients with cirrhosis), may lead to misinterpretation of adrenal function with an overestimation of the prevalence of AI. But the direct measurement of free PC is a laboratorydependent and time-consuming procedure that is not available for routine use. Salivary cortisol (SC) is one of the several indirect methods available to determine free PC [3], as SC levels accurately re lect free PC [4] even in cases of hypoalbuminaemia or CBG abnormality [1,5]. For this reason, in the last years, this technique (SC) has been introduced as a non-invasive tool in the diagnosis of adrenal cortical disorders, for its simplicity and applicability in the paediatric population. However, few studies to date have evaluated the usefulness of SC as a diagnostic method in children with AI. No interactions between exogenous corticoids and SC have been described [6].
The aim of the present study was to assess the usefulness of determining salivary cortisol levels as a diagnostic tool in children with suspected secondary iatrogenic AI.

Material and methods
The study was carried out at the outpatient Paediatric Endocrine Department of Sabadell Hospital. We designed a prospective 2-year study, from January 2014 to January 2016, in paediatric patients (aged between 2 and 18 years). Inclusion criterion was: children with suspected AI secondary to prolonged corticosteroid therapy (over 15 days), received as a treatment for various underlying pathologies. These patients were referred to the Paediatric Endocrinology Unit from any paediatric section of our hospital. Written informed consent was obtained from all patients' parents and the investigation was conducted according to the Declaration of Helsinki. The study was approved by the Ethics Committee of our Institution (reference code CEIC 2014510).
Samples of twenty children (4 male), median age 10.93 years ± 3.69 SD years were evaluated. All the patients had received long-term steroid treatment for various disorders (mainly rheumatic diseases, such as systemic lupus erythematosus, juvenile dermatomyositis, and others). After informed consent was obtained, clinical data (age, gender, ethnicity, anthropometric data, Cushing features and main pathology) were recorded. In a second visit, plasma and salivary cortisol levels after ACTH test was performed at 8:00 -9:00 am, after fasting for at least 8 hours, 1 hour after waking up [6], 24-48 hours after steroid withdrawal, always in the outpatient day hospital and performed by same nurse [7] to avoid variability. A peripheral vascular catheter was placed and blood sample for PC baseline levels was collected. After careful mouth cleansing, a speci ic device was used to collect a saliva sample (Sarstedt Cortisol Salivette ® Device). According to the supplier's instructions, the sponge is placed directly into the mouth and patients gently chew and roll it around for 2-3 minutes before spitting it back into the tube and the sample can be collected. After collecting baseline plasma and saliva samples, ACTH stimulation test was performed using a dose of 1μg synthetic adrenocorticotropic hormone, administered by intramuscular injection (Synacthen ® ). Blood samples for plasma cortisol were collected at 30, 60 and 90 minutes; saliva samples were collected simultaneously. These saliva and plasma samples were subjected to electrochemiluminescence (E-170, Roche Diagnostics). Detection limit: 0.018 μg/dL. Coef icient of variation (CV): 6.1% intraday at concentration of 0.17μg/dL; 11.5% interday at concentration of 0.292 μg/dL. Data are expressed as mean ± SD for quantitative variables and as percentages for categorical variables unless otherwise indicated. Pearson's linear regression analysis was used to analyze the correlation between plasma and saliva cortisol, while the ROC curve was used to determine cut-off for SC. A p value < 0.05 was considered as signi icant. All analyses were performed with SPSS 21.0 version (SPSS, Chicago, USA).

Results
A total of 230 samples were analyzed (118 of plasma cortisol and 112 of salivary cortisol), of 30 studies (consisting of four blood samples and four saliva samples) applied in 20 patients (Table 1). Studies with a baseline PC or a peak of PC in low-dose ACTH stimulation test higher than 18 μg/ dL were considered normal (without AI). According to this classi ication, 8 studies were normal and AI was identi ied in 22.
The cut-off for peak salivary cortisol (during ACTH stimulation test) that allowed us to differentiate the two populations (healthy children versus patients with AI) with the ROC analysis, taking into account the maximum values of SC of each study was 0.61 μg/dL ( Figure 2) with a sensitivity and speci icity of 93.94% and 66.67%, respectively.

Discussion
Adrenal insuf iciency is a life-threatening condition requiring prompt diagnosis, and currently, the main cause in the paediatric population is iatrogenic, result of prolonged treatments with high-dose exogenous corticosteroids. Laboratory evaluation of patients with suspected AI begins with the measurement of total baseline PC, but this is not suf icient to accurately diagnose this condition. For this reason, the most widely used adrenal function indicator in clinical practice is the determination of total PC during ACTH stimulation. Yet PC is not the best tool because 80% -90% of circulating cortisol is bound to CBG and measurement of plasma cortisol is compromised by conditions altering CBG levels [8,9]. Although the exact mechanism of cortisol secretion in saliva is unclear, salivary cortisol re lects circulating free cortisol levels better than total plasma cortisol (SC is not in luenced by CBG levels, only free plasma cortisol passes into saliva). In cirrhotic patients, free cortisol seems to be more strongly correlated with salivary cortisol than with total plasma cortisol (Spearman coef icient 0.91 vs. 0.76, respectively, p < 0.001) [10]. Consequently, the measurement of salivary cortisol levels can be used as an alternative technique in the diagnosis of adrenal disease. Saliva samples can easily be obtained in a non-invasive way while avoiding stress-related cortisol release that may occur during blood sampling. Furthermore, it is an accurate tool to test adrenocortical function under basal and stimulated conditions, especially in infants and young children, and offers a viable alternative to test plasma cortisol levels.
We included children over 2 years old, because under this age the amount of saliva produced seems to be more variable  and unpredictable [11]. We chose a saliva testing device because it is considered the best method to obtain salivary cortisol [12,13]. Intramuscular ACTH administration was used to evaluate the correlation between PC and SC in order to avoid intravascular administration in the future, if SC was optimal.
Some studies have shown that plasma cortisol passes and concentrates in the saliva in approximately 1 to 5 minutes (reaching a plateau in 6-12 minutes), after IV injection of cortisol or after ACTH stimulation test [8,14]. All these adult's studies showed a correlation between PC and SC levels (r = 0.83-0.932), but a good correlation during low-dose ACTH test between these two parameters was only found in one paediatric study [8]. In critically ill children admitted in the Paediatric Intensive Care Unit, there was a signi icant positive correlation between total plasma cortisol and salivary cortisol concentrations at baseline (r = 0.67; p < 0.0001) and following the ACTH test (r = 0.41; p < 0.02) [1]. However, in athletes, SC did not correlate with PC, but this difference may be due to the different methods used and time of collection [15]. In children with hyperactivity attention de icit disorder, SC was higher than in controls, which is a good stress indicator [16]. SC was not useful in preterm infants probably due to small sample size [17].
Our study showed a highly positive correlation between salivary and total plasma cortisol levels in children with adrenal insuf iciency caused by long-term steroid use, which is in agreement with previously published studies [1,8]. Furthermore, this positive correlation has been shown in adults with Cushing's syndrome and end-stage renal disease [1,18,19]. These indings reveal that the pharmacodynamics of cortisol in the paediatric population may not be different than in adults, and SC becomes a tool that could be applied to a range of disease settings to monitor adrenocortical function, and is also feasible in critically ill patients, especially in children.
Several published studies have attempted to identify cutoff levels for SC in the diagnosis of Cushing syndrome, in both adults and the paediatric population [8]. In Cushing patients, there is a good concordance with free urine cortisol and SC can be offered as an alternative better screening study [20,21]. However, data regarding the value of SC in the diagnosis of AI in children are scarce [2,22,23]. In the current study, a SC peak during low-dose ACTH test higher than 0.61 μg/dL identi ied patient without AI, in agreement with other published studies.
This study has several limitations: the sample size can be considered limited and the samples were collected during different seasons. Additional studies including patients with adrenal insuf iciency are needed to con irm these indings (especially to assess the diagnostic value of the suggested cutoff levels) in a larger cohort and to determine the validity of measuring salivary cortisol in other adrenal diseases.

Conclusion
Salivary cortisol is a non-invasive, easily applicable test.
Cortisol levels in saliva and plasma are highly correlated and a cut-off for peak SC during low-dose ACTH stimulation test of 0.61 μg/dL is useful to rule out adrenal insuf iciency. It could replace plasma cortisol as a screening tool for the diagnosis of secondary AI.

Declaration of interest
The authors declare that there is no con lict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding
This project was supported by a grant from Foundation Parc Taulí (CEIC 2014510).