Steroid Analysis in Saliva: An overview

Abstract

The first report of steroid analysis in saliva was more than thirty years ago. Since that time its popularity has increased due to the attractiveness of non-invasive, repeated and simple stress-free sampling. It has proved a popular sampling fluid for psychobiology, sports medicine, pharmacology and paediatric studies as well as in the area of complementary medicine. In the diagnostic laboratory, salivary progesterone and oestradiol have been used for assessing ovarian function and 17α-OH progesterone for the diagnosis of congenital adrenal hyperplasia (CAH). Salivary cortisol is used for investigating adrenal function and recently there has been considerable interest in the use of bedtime salivary cortisol levels as a screening test for Cushing’s disease. However, there are several caveats on the use of saliva including collection techniques, the variable matrix of saliva, sensitivity, steroid stability, the presence of binding proteins and reference range anomalies. This brief review will attempt to address these issues and provide a balanced approach to steroid analysis in saliva.

Introduction

Following early reports on salivary steroids there has been a plethora of literature including a number of reviews.^1–5 For this reason, this overview will concentrate more on recent developments although a few introductory comments are useful. As a biological medium saliva is a variable and complex fluid and is mostly produced by three pairs of salivary glands (parotid, submandibular and sublingual) with a small contribution from the buccal glands which line the mouth. In addition, saliva contains variable amounts of gingival crevicular fluid, which leaks from the tooth-gum margin, as well as plasma exudates, or blood, from oral abrasions or lesions. Hormones can enter saliva by a variety of mechanisms but for the neutral steroids the most common route is rapid diffusion through the acinar cells and as such their concentration is independent of the rate of saliva flow.⁶ For charged steroids, like DHEAS, the mode of entry is by diffusion between the tight junctions of the acinar cells and its concentration is inversely related to saliva flow rate.⁵ Saliva pH also alters with saliva flow rate and hence affects the partitioning of charged steroids. Steroids can also enter saliva from blood or plasma via oral abrasions or directly from foodstuffs by contamination with exogenous steroids.⁶

Advantages and Disadvantages

The prime advantage of saliva is that it offers non-invasive, stress-free and real-time repeated sampling where blood collection is either undesirable or difficult. It is well suited for paediatric, time-shift and psychobiological studies. In addition, no special training or equipment is needed and subjects can conveniently collect samples themselves, if required. Salivary steroid levels can reflect the circulating level of free steroid rather than total circulating levels, which are confounded by the presence of circulating high affinity binding proteins.⁵ There are, however, disadvantages in the use of saliva. It can be a difficult matrix to deal with experimentally and may require physical or chemical disruption. Freeze/thaw cycles and centrifugation are often used to break up mucins and dithiothreitol treatment can facilitate filtration.⁷ The analysis of steroids in saliva can present analytical problems since they are present at far lower levels than in circulation. Existing assays may need to be adapted to improve sensitivity, although kits are available for measuring some steroids in saliva. Some use manual methods or automated platforms, designed for plasma, and care is required to address standardisation issues as well as the differing matrices of plasma and saliva.^8–10 The possibility of blood contamination and its likely interference can be quantified but the presence of both sex hormonebinding globulin and corticosteroid-binding globulin in uncontaminated saliva casts doubt on the reliability of salivary steroids to accurately reflect circulating free steroid levels.^11–13 The presence of 11β-hydroxysteroid dehydrogenase type 2 and 17-hydroxysteroid dehydrogenase in salivary glands also complicates the relationship between salivary and plasma free steroids.^14,15 During saliva sampling there is also the possibility of oral contamination by exogenously administered steroids. Another point, often overlooked, is that during the course of clinical investigation the physician is likely to investigate other analytes, besides salivary steroids, and more often than not a blood sample is likely to be taken.

Saliva Collection, Storage and Pretreatment

There are several saliva collection techniques and devices available. Flavoured beverage crystals and lemon juice have been used to stimulate saliva flow. Flavoured crystals may cause a small increase in measured cortisol values whereas lemon juice has been reported to compromise the extraction of cortisol from saliva.^16,17 To maintain consistency their use is probably best avoided. Various absorbent devices have been used to facilitate saliva collection. They are rated acceptable and easy to use by participants.¹⁸ These devices may be applicable for research with children and the elderly. However, there are reports that salivary cortisol values are reduced by more than 50% when saliva is not retrieved immediately from cotton buds.¹⁹ Conversely, others have reported no effect for cortisol and artificially elevated results for testosterone, DHEA, progesterone and oestradiol using cotton absorbent materials.²⁰ There have also been several reports on the stability of steroids in saliva. Storing saliva samples at room temperature is not desirable as bacterial activity has been reported to be associated with changes in testosterone but not cortisol or DHEA.²¹ Cortisol, 17α-OH progesterone and progesterone are reported to be stable after five days at 4° C in intact and centrifuged saliva and unaffected by two freeze /thaw cycles.²² Similarly others have shown that cortisol is stable in centrifuged saliva for even longer periods at 5° C and up to three months at −20° C with notable losses occurring at room temperature.²³ Pretreatment of saliva may affect the concentrations of some steroid hormones when measured following extraction and chromatography. Sonication resulted in significantly higher saliva values than supernatant values for progesterone, cortisone, 17α-OH progesterone, testosterone and oestradiol. No differences were observed for cortisol and androstenedione.²⁴ Comparisons were benchmarked against levels measured following extraction and chromatography of the saliva samples. Taken together, the collection of unstimulated saliva into small plain tubes and storage at −20° C would appear to avoid any possible pitfalls. If long term storage is anticipated then storage at −80° C is desirable. If special collection devices are to be used then ideally the same device should be used throughout a particular study, including the derivation of reference ranges. Furthermore, analysis of intact saliva, preferably following solvent extraction, is sound analytical practice.

Salivary Androgens

Salivary androstenedione correlates well with free plasma concentrations (r = 0.92) and has been used in paediatric studies along with DHEA and testosterone.^25,26 However, in the clinical setting, it has been deemed necessary that salivary androstenedione determination, along with progesterone and 17α-OH progesterone determinations, are performed following extraction and chromatography. This requirement is due to the presence of unidentified cross-reacting material in unpurified extracts.^27,28 Salivary DHEA levels, following organic extraction, are reportedly lower in adult depression but salivary DHEAS analysis is probably not as useful as its concentration is dependent on salivary flow rate.²⁹ This suggests an alternative transport mechanism into saliva. In addition, the high concentration of DHEAS in plasma could contaminate saliva levels via gingival fluid or oral abrasions.⁵ There are numerous reports of testosterone analysis in saliva and in psychobiological studies it appears to correlate with psychological parameters.^30–32 Salivary testosterone has also been used for monitoring i.v. replacement therapy with testosterone buciclate.³³ However, in view of the current controversies surrounding testosterone measurement by immunoassay at low concentrations in plasma, it would seem that the even lower testosterone concentrations in saliva would likely present an analytical challenge, although it is suggested that its reliability is acceptable for research purposes.^34,35 There are, however, real concerns regarding storage and interference.³⁶ In males, salivary testosterone declines with age and reflects testicular production although neither salivary testosterone nor plasma testosterone appears clearly superior to the other as a measure of bioavailability.^37,38 Extraction-based assays for salivary testosterone may mitigate some problems and improve sensitivity as well as display acceptable correlation between either total or free testosterone in male plasma and saliva levels (r = 0.71 and 0.67, respectively). However, the correlation between plasma free testosterone and saliva testosterone in females is poor (r = 0.37).³⁶ This is in agreement with our laboratory findings using extraction-based immunoassays for testosterone. The sex-related correlation between calculated free testosterone in plasma and salivary testosterone is shown in Figure 1.³⁹ We would conclude that mathematically derived indicators of bioavailable androgen status, using plasma total testosterone and sex hormone-binding globulin, are preferable as they overcome the limitations imposed by saliva.

Figure 1.

Correlation of calculated plasma free testosterone with salivary testosterone for females (a) and males (b). Testosterone analysis was by ELISA. From Broadbent JL. NZ J Med Lab Science 2002;56:85–9, with kind permission of the NZIMLS, copyright holders.

Salivary Progestins and Oestrogens

CAH has been diagnosed by the immunoassay of 17α-OH progesterone in both plasma and saliva and an excellent correlation reported.^40,41 An LC/MS method for saliva steroids has also been reported to effectively diagnose CAH in children.⁴² Latterly, commercial immunoassays for 17α-OH progesterone have been adapted and used to establish reference ranges in children. However, the use of oral collection devices for newborns induced stress responses, jeopardising both 17α-OH progesterone and cortisol determinations.⁴³ A recent study examined the use of three different saliva collection systems, plain tube and either cotton or polyester Salivette®. Here, all saliva samples were centrifuged prior to assay, using a modified commercial assay, and no differences detected.⁴⁴ However, a shortcoming was the absence of comparisons with intact extracted saliva.

Progesterone is metabolically stable in saliva and has been used in a variety of applications.⁴⁵ It has been used as an index of ovulation but there was a high degree of discrepant classifications between the follicular and luteal phases.⁴⁶ Similar findings were reported recently where salivary progesterone in the mid-luteal phase yielded a sensitivity and specificity of 78% and 77%, respectively.⁴⁷ While suited for research and long-term clinical observation salivary progesterone determination may have a limited application for individual diagnosis.⁴⁸ The parallel changes of salivary progesterone and 17α-OH progesterone over the menstrual cycle were suggested as an additional parameter to assess ovarian function.⁴⁹ Similarly, paired determinations of progesterone and oestradiol in saliva may be useful for fertility monitoring although wide variations in oestradiol were noted.⁵⁰ Probably, the most reliable way of assessing corpus luteal function is via plasma progesterone and if noninvasive testing is required the best option is the determination of urinary pregnanediol excretion indexed to creatinine.^46,51 Salivary progesterone determinations were used to monitor the absorption of progesterone following topical application of progesterone cream to pre- and postmenopausal women. Paradoxically, salivary progesterone levels were elevated whereas plasma and urinary progesterone and its metabolites remained low suggesting minimal benefit to target tissues using this alternative replacement therapy.^52,53

Salivary Glucocorticoids

There is little information on 11-deoxycortisol in saliva, presumably due to its very low level, although it can be detected following metyrapone.⁵⁴ On the other hand, salivary cortisol determinations are more promising. Since the earliest reports, there is evidence that saliva levels reflect unbound concentrations in plasma.⁵⁵ Direct measurement in saliva may be comparable to extraction but extraction has the advantage of allowing the analysis of low volume saliva samples. This can avoid the use of stimulants and the loss of data due to insufficient sample volume.⁵⁶ Salivary cortisol determinations have proved popular in psychobiology, stress and sports medicine studies.^57–61 Their use is based on the assumption that salivary cortisol is a reasonable reflection of hypothalamicpituitary- adrenal (HPA) axis function. Indeed, in the diagnostic setting, salivary cortisol levels parallel those in plasma following ACTH and CRH stimulation, and following exercise induced-stress.^62,63 However, the correlation of salivary cortisol levels with total plasma cortisol is confounded by the presence of corticosteroid-binding globulin in plasma which is largely saturated up to 500–600 nmol/L of cortisol.⁶² Salivary cortisol correlates better with measured plasma free cortisol than total plasma cortisol. However, it appears to be subject-specific as considerably variability is found between individuals for daily paired samples (Figure 2). Salivary cortisol determinations were used as markers of metabolic disturbances in obese and diabetic patients and used to investigate changes in glucocorticoid control of the HPA axis following oral prednisolone.^64–67 Conversely, salivary cortisol measurement is not a useful tool in determining dose adequacy in subjects on oral glucocorticoid replacement therapy.^68,69 Blood spots or serum is preferable, due to contamination of saliva by oral hydrocortisone. The diurnal variation of plasma cortisol is reflected by similar changes in salivary cortisol and hence timed salivary cortisols have been used in a diagnostic setting. Early morning salivary cortisols are useful as a screening tool for adrenal suppression and salivary cortisol is helpful to investigate the control of diurnal cortisol secretion following exposure to darkness and light.^70–72

Figure 2.

Correlation of salivary cortisol and plasma free cortisol in paired samples from two normal individuals (a and b). Salivary cortisol was measured by ELISA and plasma free cortisol by ligand binding/ultrafiltration.

Recently, there has been considerable interest in the use of night time salivary cortisols for the initial screening for Cushing’s syndrome. However, despite the optimism, an element of caution is required. Reported cut-off values differ considerably. Papanicolaou et al. determined a normal night time (2330–2400h) saliva cut-off of 15 nmol/L with values above this threshold suggestive of Cushing’s.⁷³ They reported 100% specificity and 91% sensitivity for saliva, similar to a late night serum cortisol measurement cut-off value exceeding 240 nmol/L. Both were reportedly clearly superior to urinary free cortisol excretion, contrasting with other recent reports. Yaneva et al.⁷⁴ determined a lower saliva cut-off value of 5.7 nmol/L, with 100% sensitivity and 96% specificity which effectively separated Cushing’s from obese subjects.⁷⁵ The corresponding excretion of urinary free cortisol determined a cut-off value of 248 nmol/day, with similar sensitivity and specificity to saliva. Viardot et al. reported a similar night time saliva cut-off value of 6.1 nmol/L and view salivary cortisol as a reliable first line alternative to urinary free cortisol, the urinary free cortisol:creatinine ratio, and the 1mg overnight dexamethasone suppression test.⁷⁵ However, their cut-off value for urinary free cortisol was considerably higher at 504 nmol/day compared to the 248 nmol/day reported by the Yaneva group.⁷⁴ Both groups used extraction based assays for urinary free cortisol which, as an aside, highlights the importance of well characterised methods and reagents for the determination of urinary free cortisol.

Late night salivary cortisol measurement is nevertheless very promising for the diagnosis of Cushing’s although elevation above threshold values can occur in the elderly, diabetics and women in late pregnancy.^75,76 The variations in reported late night salivary cortisol cut-off values could result from the relatively small numbers in each of the study control groups, which may have varying degrees of obesity, non-adrenal disorders and pseudo-Cushing states, which themselves may be influenced by periodic hypercortisolism.⁷³ Methodological and standardisation issues are also likely contributors to differences in reported cut-off values but a major factor is probably differing specificities of cortisol antibodies towards cortisone. The salivary gland has abundant 11 β-hydroxysteroid dehydrogenase type 2 activity and as a consequence, saliva, unlike plasma, has up to three times the level of cortisone compared to cortisol.^14,77 Depending on the relative cross-reactivity of cortisol antibodies towards cortisone, there could be quite different values of salivary cortisol measured by different immunoassays. Conversely, differences in plasma would be expected to be minimal as cortisone levels are normally only 10% of circulating cortisol levels.⁷⁷ It is therefore desirable that laboratories establish their own method-specific reference ranges before using salivary cortisol for diagnostic purposes.

Conclusion

Salivary steroid testing has a recognised place in research and diagnostic medicine although its limitations must be acknowledged. It is clearly not desirable for androgen assays as well as assays to assess ovarian function and the monitoring of absorption of steroids from transdermal creams. Caution must be exercised for these applications. The use of salivary cortisol for measuring endogenous cortisol is the most encouraging. It can be successfully applied to research studies, adrenal stimulation tests, investigating diurnal variation as well as night time samples as a screening test for Cushing’s syndrome. However, there is a need to establish methodspecific reference ranges and sample stimulation, collection and storage should remain consistent across study groups with a preference for the collection of whole unstimulated saliva, if possible.