- Email: [email protected]
A critical evaluation of salivary testosterone as a method for the assessment of serum testosterone
Steroids 86 (2014) 5–9
Contents lists available at ScienceDirect
Steroids journal homepage: www.elsevier.com/locate/steroids
A critical evaluation of salivary testosterone as a method for the assessment of serum testosterone q Tom Fiers a,⇑, Joris Delanghe a, Guy T’Sjoen b, Eva Van Caenegem b, Katrien Wierckx b, Jean-Marc Kaufman a,b a b
Ghent University Hospital, Clinical Chemistry, Belgium Ghent University Hospital, Endocrinology, Belgium
a r t i c l e
i n f o
Article history: Received 7 February 2014 Received in revised form 11 April 2014 Accepted 15 April 2014 Available online 30 April 2014 Keywords: Saliva Testosterone LC–MS/MS Equilibrium dialysis
a b s t r a c t Although salivary testosterone (T) is often used in clinical studies accuracy is mostly questionable. State of the art data for men is sparse and for women absent. Our objective was to perform a critical evaluation of salivary T (Sal-T) as a method for indirect assessment of serum T using state of the art methods. Saliva was collected via ‘Salivette’ and ‘passive drooling’ methods. Sal-T and free T in serum after equilibrium dialysis were measured by LC-MS/MS Results: Evaluation of Sal-T results versus free T by equilibrium dialysis (ED-T) for men gave: ‘Salivette’ Sal-T = 0.05 + 0.88x ED-T, r = 0.43; ‘passive drooling’ Sal-T = 0.17 + 0.91x ED-T r = 0.71. In women, correlation was comparable but values are higher than free T: ‘passive drooling’ Sal-T = 0.12 + 2.32x ED-T, r = 0.70. The higher than expected T values in saliva, appear to be explained by T binding to salivary proteins. Iso-electric focusing of saliva proteins, followed by fractionation and LC–MS/MS assay of T showed marked testosterone peaks at pH 5.3 and 8.4, providing evidence for T binding in saliva to proteins such as albumin and proline rich protein (PRP). Conclusions: Passive drooling is the collection method of choice for testosterone in saliva. Sal-T is not directly comparable to serum free T due to T binding to saliva proteins, which substantially affects the low Sal-T in women but not the higher Sal-T in healthy adult men. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction Testosterone (T) is one of the key hormones to assess gonadal status in males, females and children. In serum about 97–98% of T is protein-bound, i.e. for 40–60% in men and 60–70% in women bound with high affinity to sex hormone-binding globulin (SHBG) and the remaining for the major part bound with much lower affinity to albumin. Only the free T fraction of about 2–3% and probably also at least part of T loosely bound to albumin is available for biological action; free T together with albumin-bound T is often referred to as ‘bioavailable T’. For most indications (e.g. assessment of pubertal disorders, evaluation of androgen excess in females, diagnosis of clear-cut hypoandrogenism) an accurate measurement of total serum testosterone, preferably by a well standardised LC–MS/MS or GC–MS/MS method suffices [1–3]. However, total T is
q The SEXPERT study was supported by a Belgian federal Grant (government agency for Innovation by Science and Technology, IWT, www.iwt.be). ⇑ Corresponding author. Address: UZ Gent, Hormonologie 2P8, De Pintelaan 185, 9000 Gent, Belgium. Tel.: +32 93324565 (Office), mobile: +32 485412943. E-mail address: tom.fi[email protected] (T. Fiers).
http://dx.doi.org/10.1016/j.steroids.2014.04.013 0039-128X/Ó 2014 Elsevier Inc. All rights reserved.
not an optimal parameter to assess more subtle forms of hypo- and hyper-androgenism. In particular, situations with increased SHBG levels (e.g. estrogen treatment, hyperthyroidism, ageing in men) or decreased SHBG (e.g. obesity, glucocorticoid treatment) result in alterations of total T not necessarily reflected in the biologically active free T fraction that is controlled through negative feedback regulation of gonadotropins at the hypothalamo-pituitary compartment of the gonadal axis. In these situations free T will be a more reliable reflection of androgen exposure. Direct immunoassays to measure free T have proven to be inaccurate, gives results not independent of SHBG concentrations and are badly standardised [4–6], but unfortunately are still often used. In order to accurately measure serum free T, ultrafiltration isotope dilution mass spectrometry (UF-IDMS) or equilibrium dialysis (ED) isotope dilution mass spectrometry (ED-IDMS) has to be used [7]. However these are difficult and labour-intensive to perform correctly and are therefore scarcely used. Although UF is more convenient and has been used by some authors with good results [8], it is less robust than ED which Thienpont et al. used for subsequent candidate reference procedures [9,10]. Even if equilibrium dialysis is used, accuracy and precision differ widely across methods (e.g.
6
T. Fiers et al. / Steroids 86 (2014) 5–9
because of pH or temperature issues; because a dilution is used which influences the equilibrium [8,11–13]; or an indirect measurement is performed of radioactive tracer with subsequent percentage calculation of total T, increasing total error [10,13]; because total T was not measured by a LC–MS/MS method standardised against the gold standard; or a combination of the above. In the past, a number of publications have hinted at the possibility of salivary T (Sal-T) being an equally performing marker as serum free T. [14–17]. Sal-T reflects the T which diffuses through the salivary glands and was reported to be quite well correlated with serum free T [18,19]. The stress-free and not invasive nature of saliva sampling would seem to make it a perfect candidate for large scale population studies, studies with children or selected clinical usage (e.g. behavioural studies, repeated sampling, . . .) [17,20,21]. However, expected Sal-T concentrations are low in men and very low in women and children, so that Sal-T assessment is not without pitfalls, which may explain the sometimes mixed results to date. Blood contamination (e.g. teeth brushing) must be avoided prior to collection, saliva stimulation and the collection device itself may influence results and samples should be frozen at 80 °C if long term storage is needed prior to analysis [22–25]. Unfortunately, to date most studies have been performed using adapted immunoassays to measure the low Sal-T concentrations. These assays suffer from lack of specificity and from matrix related problems, similar to the long known pitfalls of serum total T assays, and may thus bias results [26,27]. Also, LC–MS/MS analysis of Sal-T has mostly been limited to samples from men, given the high sensitivity needed to measure Sal-T in women. To the best of our knowledge no critical evaluations of salivary testosterone have been described using the combination a state of the art methods i.e. Sal-T and serum total T by LC–MS/MS traceable to a primary gold standard and – free T using undiluted serum equilibrium dialysis with subsequent direct LC–MS/MS measurement of free T (ED-T). In addition we evaluated the impact of two commonly used collection devices on Sal-T. 2. Materials and methods Morning blood and saliva samples were collected in 50 male and 50 female healthy volunteers. All participants signed an informed consent and the study was approved by the ethical committee of the University Hospital of Ghent, Belgium. Serum was kept at 80 °C until batch analysis. SHBG assay was performed on E170 Modular, albumin on Cobas 8000 (Roche Diagnostics). For saliva polypropylene tubes for passive drooling and Salivette cotton rolls (Sarstedt, Nümbrecht, Germany) were used. Salivette rolls were chewed for 1 min and then replaced into the centrifugation tubes. After centrifugation at 2000g for 5 min, 250 ll supernatant from passive drooling and Salivette was collected and kept at 80 °C until LC–MS/MS analysis. Serum equilibrium dialysis was performed using Fast MicroEquilibrium dialyzer cartridges and regenerated cellulose 25 KD membranes (Harvard Apparatus; Holliston, USA) 500 ll serum (men) or 1000 ll (women) was dialysed at 37 °C for 24 h at pH 7.28 using protein free buffer prepared according to Yue et al. [28]. Serum free testosterone was also calculated (CalcFT), according to Vermeulen [11]. Samples of 100 ll serum with 400 ll physiologic solution added was used for assay of total testosterone. A liquid–liquid extraction (LLE) was used for all serum, saliva and dialysate samples: 20 ll d3-testosterone internal standard was added before extraction with 2 mL of diethylether; after mixing for 3 min samples were frozen and decanted with subsequent drying of the collected supernatant; the dried supernatant was then reconstituted in a final solution of 125 ll methanol of which 100 ll was injected for liquid chromatography. Testosterone was acquired from Sigma–Aldrich
(Saint Louis, USA), Testosterone d3 from CDN Isotopes (Quebec, Canada). All standards and internal standards were dissolved in methanol. Methanol, water and acetonitrile were LC–MS grade from BioSolve BV (Varkenswaard, The Netherlands). For measurement of serum T, Sal-T and T in dialysate following equilibrium dialysis of serum, i.e. free T(ED-T) by LC–MS/MS an AB Sciex 5500 triple-quadrupole mass spectrometer (AB Sciex; Toronto, Canada) was used, coupled with an APCI probe on the Turbo-V source. The liquid chromatography system consisted of a Shimadzu system using a C8 security guard column (5 lm, 4 2 mm) and a C8 Luna analytical column (3 lm, 50 3 mm) (Phenomenex; Torrance, USA). Measurements were performed by the tandem mass spectrometer running in multiple reaction monitoring (MRM) mode by using transitions m/z 289/109/97 for T and 292/109/97 for d3-T. It was validated for T against the IDMS candidate reference method [7] using a common serum panel. The method allows equally for determination of serum or salivary cortisol on 363/121/97 and androstenedione on 287/109/97. A declustering potential (DP) of 100 V and a collision energy (CE) of 32 eV was used for all the analytes. Data processing was performed through MultiQuant version 2.0.2. For analysis on 100 ll serum, inter-assay CV for testosterone is 6.5% at 3 ng/dL (104 pmol/L) (n = 30) with an LOQ of 1 ng/dL (35 pmol/L). For saliva and equilibrium dialysis where more sample volume is used inter-assay CV is 8.2% at 0.23 ng/dL (8 pmol/L) with an LOQ of 0.07 ng/dL (2.4 pmol/ L). To evaluate the presence of matrix specific effects, experiments were performed using sets of native and spiked eluent, serum, dialysate and saliva for 5 different samples. Compared to eluent, recoveries in serum (96–104%), dialysate (98–102%) and saliva (97– 105%) matrices proved there is no discernable matrix effect in the analysis. Saliva proteins were separated with a Bio-Rad MicroRotofor™ cell which uses patented RotoforÒ technology based on iso-electric focusing. It enables fractionation of proteins in their native state by isoelectric point (pI). An ampholyte solution (Bio-LyteÒ) with a pH gradient ranging from 3 to 10 was obtained from Bio-Rad Laboratories (CA, USA); T concentration was measured in each fraction with the assay as described above. 3. Results 3.1. Impact of collection device To investigate the potential impact of salivary collection device on T determination, ‘Salivette’ and ‘passive drooling’ Sal-T were compared to serum equilibrium dialysis (ED-T) for men. Better concordance with ED-T is observed for ‘passive drooling’ compared to ‘Salivette’ Sal-T: passive drooling Sal-T = 0.17 + 0.91x ED-T, r = 0.71; Salivette Sal-T = 0.05 + 0.88x ED-T, r = 0.43. A similarly important difference between Salivette Sal-T (r = 0.23) and passive drooling Sal-T (r = 0.71) was observed for samples of women (Supplementary data). Therefore, subsequent salivary analysis was only carried out on passive drooling samples (Fig. 1b and c) and all results reported below are for Sal-T as measured in samples obtained by passive drooling. Sal-T correlates somewhat better with total serum T for men (r = 0.68) than for women (r = 0.52). 3.2. Comparisons of salivary T with serum free T concentrations As results on free T presented in the literature are often a calculated estimation of serum free T (CalcFT) [10,29] CalcFT was also evaluated. Sal-T correlates moderately well to CalcFT: r = 0.68 for both men and women. Comparison between CalcFT and ED-T for men yielded: CalcFT = 0.09 + 1.25x ED-T; r = 0.85. Notable is the concentration dependent overestimation of CalcFT by 25% in
T. Fiers et al. / Steroids 86 (2014) 5–9
7
Fig. 1. Clockwise from left, Passing and Bablock regression: CalcFT vs equilibrium dialysis (ED-T) (female) (a), Salivette Sal-T vs ED-T (male) (b), passive drooling Sal-T vs ED-T (male) (c), passive drooling Sal-T vs ED-T (female) (d).
combination with a good linear correlation in comparison with EDT. For women, an agreement of CalcFT = 0.05 + 1.0x ED-T with r = 0.85 was found, thus a fixed bias of about 15% of an average female value (Fig. 1a). For the remainder of the analyses only EDT was considered as the ‘gold standard’ comparator. Sal-T correlates equally well to ED-T for men and women (r = 0.71), however in contrast to findings for samples from men, Passing and Bablock regression of female Sal-T vs ED-T reveals surprisingly higher than expected Sal-T values compared to ED-T: Sal-T = 0.12 + 2.32x ED-T for women; Sal-T = 0.17 + 0.91x ED-T for men (Fig. 1c and d). To elucidate this pronounced difference in men and women analytical errors were first ruled out: dialysis with different serum and dialysate volumes (increasing dialysate volume from 200 ll to 1 mL repeated for 0.2, 0.5 and 1 mL of serum) yielded comparable results (data not shown). Time to equilibrium was experimentally determined to be ±8 h for serum for men and ±3 h for serum from women, hereby validating the safety of the implemented 24 h equilibrium dialysis time and which is comparable to reported cortisol equilibrium dialysis findings by Kirchhoff et al. [30]. Determinations of duplicate extractions for serum free T dialysis showed excellent concordance (r = 0.94). As partial protein binding of T to saliva proteins was the most plausible hypothesis to explain the difference between serum free testosterone and the higher saliva testosterone, protein analysis of a subset of saliva samples was performed and yielded a median of 42 mg/dL, p25–p75: 35–53 mg/dL (i.e. about 1% of serum values). To determine whether T was actually binding to these saliva proteins, isoelectric focusing of salivary proteins in pH range 3–10 and subsequent fractionation and LC– MS/MS analysis for testosterone was performed. The results presented in Fig. 2 provide evidence of broad T binding with marked T peaks at pH 5.3 and 8.4.
Fig. 2. Isoelectric focusing and fractionating according to isoelectric point (pI) of saliva proteins. Amount of testosterone in each saliva fraction in pg/mL.
4. Discussion We present an assessment using state-of-the-art methodology of T measurements in saliva. Most studies focusing on or using measurements of T in saliva have traditionally relied on immunoassays, or have been limited to Sal-T in men given the high sensitivity needed to measure Sal-T in women. In the preparation for an epidemiological study we evaluated a commonly used saliva testosterone kit (Salimetrix; Carlsbad, USA) versus LC–MS/MS, which showed higher values for the immunoassay versus LC–MS/MS
8
T. Fiers et al. / Steroids 86 (2014) 5–9
Fig. 3. Bland and Altman difference plots of male (a) and female (b) passive drooling saliva (PD-Sal-T) vs equilibrium dialysis (ED-T).
(data not shown). More importantly, a very large spread exceeded functional sensitivity criteria (20%) at all levels confirming existing criticism on salivary T immunoassays [26]. We therefore proceeded to further assess the more laborious, but much more accurate LC–MS/MS analysis of Sal-T by comparing saliva levels to serum total T and serum free T by equilibrium dialysis. We evaluated the potential impact of saliva collection device on Sal-T. In contrast to the findings for cortisol (r = 0.94) or androstenedione (r = 0.84) which were determined on the same samples and for which results concurred well between the ‘Salivette’ collection method and the ‘passive drooling’ collection method, we demonstrated that only the ‘passive drooling’ sample collection is acceptable for Sal-T analysis. As for the assumption by many authors that saliva can be used as a surrogate for serum free T, we proved that for samples from healthy adult males collected by passive drooling, Sal-T experimentally seems a reasonably valid alternative to ED-T given proper pre-analytical care and use of reliable analytical methodology. Granger et al. [23] previously noted a bad correlation between female salivary testosterone and free serum testosterone. However they employed a direct immunoassay which was unable to correctly measure serum free T. We demonstrated that there actually is a reasonably good correlation between Sal-T and serum free T in women. However, for Sal-T in women there is a substantial positive bias of Sal-T compared to serum free T (Fig. 3). In women Sal-T cannot be considered a surrogate marker of serum free T. To explain this observation, we postulated that Sal-T is in fact not a measure of serum free T as is often stated, but rather a separate complex matrix with its own testosterone protein-binding properties. Given the low total protein content of saliva and conse-
quently only low expected T-binding, this effect would still be susceptible to significantly affect the low Sal-T in women or children, but not be apparent when measuring the much higher Sal-T in men (explaining the different observed slopes in men and women between Sal-T and ED-T). The concentration of saliva proteins we measured could indeed theoretically explain the higher than expected values in female saliva samples, if binding to a significant fraction of these proteins is occurring [11]. As the constitution of proteins in saliva is very different from that in serum, with a much smaller role for albumin, it was further unclear whether serum-like binding really occurs. The isoelectric focusing of saliva and subsequent T determination in the fractions provided evidence for this hypothesis. In addition to the expected binding to salivary albumin which is rather low in abundance, the major testosterone peaks (pH 5.3 and 8.4) correspond exactly to the migration patterns of proline rich protein (PRP) and glycosylated PRP (expressed predominantly in the parotid gland) as experimentally determined by Beeley et al. in two dimensional saliva electrophoresis [31]. This delicate mixed saliva protein balance and possible differences between individuals underscore the need for proper collection of unstimulated mixed saliva and analysis using LC–MS/MS. Although much progress has been made in these experiments as to the true analytical value of salivary testosterone and potential pitfalls, whether Sal-T for women could have a clinical applicability such as for detection of hyperandrogenism was not assessed in this study. Additional research is needed to determine the impact of nonstandard protein binding patterns or elevated female serum T on saliva T levels, which may be important factors to consider in future clinical saliva studies. In conclusion, we show that Sal-T by LC–MS/MS for salivary samples collected using the ‘passive drooling’ technique is not identical to serum free T. Sal-T correlates fairly well with serum free T, but due to binding of T to salivary proteins, in women values are higher than expected and cannot be considered a surrogate for serum free T. Moreover, this effect of protein binding can be expected to introduce additional variability of Sal-T among women. Given the prevalent Sal-T concentrations in healthy adult males the protein binding effect is negligible although formally also present. Disclosure statement The authors have nothing to disclose. Acknowledgements We thank Brigitte Bernaert, Martine De Bock and Eric Vandersypt for their invaluable experimental work without which this study would not have been possible. Alexis Dewaele, Maya Caen, Els Elaut, Katrien Vermeire, Sabine Hellemans, Lies Hendrickx, Wouter Pinxten, Katrien Symons, Ellen Van Houdenhove and Joke Vandamme for recruiting and collection of samples. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.steroids.2014. 04.013. References [1] Thienpont LM, Van Uytfanghe K, Blincko S, Ramsay CS, Xie H, Doss RC, et al. State-of-the-art of serum testosterone measurement by isotope dilution– liquid chromatography–tandem mass spectrometry. Clin Chem 2008;54(8):1290–7. [2] Rauh M. Steroid measurement with LC–MS/MS. Application examples in pediatrics. J Steroid Biochem Mol Biol 2010;121(3–5):520–7.
T. Fiers et al. / Steroids 86 (2014) 5–9 [3] Honour JW. Steroid assays in paediatric endocrinology. J Clin Res Pediatr Endocrinol 2010;2(1):1–16. [4] Rosner W. An extraordinarily inaccurate assay for free testosterone is still with us. J Clin Endocrinol Metab 2001;86:2903. [5] Fritz KS, McKean AJ, Nelson JC, et al. Analog-based free testosterone test results linked to total testosterone concentrations, not free testosterone concentrations. Clin Chem 2008;54:512–6. [6] Swerdloff RS, Wang C. Free testosterone measurement by the analog displacement direct assay: old concerns and new evidence. Clin Chem 2008;54(3):458–60. [7] Van Uytfanghe K, Stöckl D, Kaufman JM, Fiers T, De Leenheer A, Thienpont LM. Validation of 5 routine assays for serum free testosterone with a candidate reference measurement procedure based on ultrafiltration and isotope dilution–gas chromatography–mass spectrometry. Clin Biochem 2005;38(3):253–61. [8] Chen Y, Yazdanpanah M, Wang XY, Hoffman BR, Diamandis EP, Wong PY. Direct measurement of serum free testosterone by ultrafiltration followed by liquid chromatography tandem mass spectrometry. Clin Biochem 2010;43(4– 5):490–6. [9] Thienpont LM, Van Uytfanghe K, Beastall G, Faix JD, Ieiri T, Miller WG, et al. Report of the IFCC Working Group for Standardization of Thyroid Function Tests; part 2: free thyroxine and free triiodothyronine. Clin Chem 2010;56(6):912–20. [10] Hackbarth JS, Hoyne JB, Grebe SK, Singh RJ. Accuracy of calculated free testosterone differs between equations and depends on gender and SHBG concentration. Steroids 2011;76(1–2):48–55. [11] Vermeulen A, Verdonck L, Kaufman JM. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 1999;84:3666–72. [12] Rosner W. Errors in the measurement of plasma free testosterone. J Clin Endocrinol Metab 1997;82:2014–5. [13] Szydlarska D, Grzesiuk W, Kondracka A, Bartoszewicz Z, Bar-Andziak E. Measuring salivary androgens as a useful tool in the diagnosis of polycystic ovary syndrome. Endokrynol Pol 2012;63(3):183–90. [14] Morley JE, Perry III HM, Patrick P, Dollbaum CM, Kells JM. Validation of salivary testosterone as a screening test for male hypogonadism. Aging Male 2006;9:165–9. [15] Goncharov N, Katsya G, Dobracheva A, Nizhnik A, Kolesnikova G, Herbst V, et al. Diagnostic significance of free salivary testosterone measurement using a direct luminescence immunoassay in healthy men and in patients with disorders of androgenic status. Aging Male 2006;9:111–22. [16] Arregger AL, Contreras LN, Tumilasci OR, Aquilano DR, Cardoso EM. Salivary testosterone: a reliable approach to the diagnosis of male hypogonadism. Clin Endocrinol (Oxf) 2007;67(5):656–62.
9
[17] Cardoso EM, Contreras LN, Tumilasci EG, Elbert A, Aguirre EC, Aquilano DR, et al. Salivary testosterone for the diagnosis of androgen deficiency in endstage renal disease. Nephrol Dial Transplant 2011;26(2):677–83. [18] Vittek J, L’Hommedieu DG, Gordon GG, Rappaport SC, Southren AL. Direct radioimmunoassay (RIA) of salivary testosterone: correlation with free and total serum testosterone. Life Sci 1985;37:711–6. [19] Wang C, Plymate S, Nieschlag E, Alvin Paulsen C. Salivary testosterone in men: further evidence of a direct correlation with free serum testosterone. J Clin Endocrinol Metab 1981;53:1021–4. [20] Gatti R, De Palo EF. An update: salivary hormones and physical exercise. Scand J Med Sci Sports 2011;21(2):157–69. [21] Bui HN, Schagen SE, Klink DT, Delemarre-van de Waal HA, Blankenstein MA, Heijboer AC. Salivary testosterone in female-to-male transgender adolescents during treatment with intra-muscular injectable testosterone esters. Steroids 2013;78(1):91–5. [22] Gröschl M. Current status of salivary hormone analysis. Clin Chem 2008;54(11):1759–69. [23] Granger DA, Shirtcliff EA, Booth A, Kivlighan KT, Schwartz EB. The ‘‘trouble’’ with salivary testosterone. Psychoneuroendocrinology 2004;29:1229–40. [24] Schultheiss OC. Effects of sugarless chewing gum as a stimulant on progesterone, cortisol, and testosterone concentrations assessed in saliva. Int J Psychophysiol 2013;87(1):111–4. [25] Celec P, Ostatníková D. Saliva collection devices affect sex steroid concentrations. Clin Chim Acta 2012;413(19–20):1625–8. [26] Mitchell JS, Lowe TE. Matrix effects on an antigen immobilized format for competitive enzyme immunoassay of salivary testosterone. J Immunol Methods 2009;349(1–2):61–6. [27] Wang C, Catlin DH, Demers LM, Starcevic B, Swerdloff RS. Measurement of total serum testosterone in adult men: comparison of current laboratory methods versus liquid chromatography–tandem mass spectrometry. J Clin Endocrinol Metab 2004;89(2):534–43. [28] Yue B, Rockwood AL, Sandrock T, La’ulu SL, Kushnir MM, Meikle AW. Free thyroid hormones in serum by direct equilibrium dialysis and online solidphase extraction—liquid chromatography/tandem mass spectrometry. Clin Chem 2008;54:642–51. [29] Keevil B, Macdonald P, Macdowall W, Lee D, Wu F, NATSAL Team. Salivary testosterone measurement by liquid chromatography tandem mass spectrometry in adult males and females. Ann Clin Biochem 2014;51(3):368–78. [30] Kirchhoff F, Briegel J, Vogeser M. Quantification of free serum cortisol based on equilibrium dialysis and isotope dilution–liquid chromatography–tandem mass spectrometry. Clin Biochem 2011;44(10–11):894–9. [31] Beeley JA, Khoo KS, Lamey PJ. Two-dimensional electrophoresis of human parotid salivary proteins from normal and connective tissue disorder subjects using immobilised pH gradients. Electrophoresis 1991;12(7–8):493–9.
Contents lists available at ScienceDirect
Steroids journal homepage: www.elsevier.com/locate/steroids
A critical evaluation of salivary testosterone as a method for the assessment of serum testosterone q Tom Fiers a,⇑, Joris Delanghe a, Guy T’Sjoen b, Eva Van Caenegem b, Katrien Wierckx b, Jean-Marc Kaufman a,b a b
Ghent University Hospital, Clinical Chemistry, Belgium Ghent University Hospital, Endocrinology, Belgium
a r t i c l e
i n f o
Article history: Received 7 February 2014 Received in revised form 11 April 2014 Accepted 15 April 2014 Available online 30 April 2014 Keywords: Saliva Testosterone LC–MS/MS Equilibrium dialysis
a b s t r a c t Although salivary testosterone (T) is often used in clinical studies accuracy is mostly questionable. State of the art data for men is sparse and for women absent. Our objective was to perform a critical evaluation of salivary T (Sal-T) as a method for indirect assessment of serum T using state of the art methods. Saliva was collected via ‘Salivette’ and ‘passive drooling’ methods. Sal-T and free T in serum after equilibrium dialysis were measured by LC-MS/MS Results: Evaluation of Sal-T results versus free T by equilibrium dialysis (ED-T) for men gave: ‘Salivette’ Sal-T = 0.05 + 0.88x ED-T, r = 0.43; ‘passive drooling’ Sal-T = 0.17 + 0.91x ED-T r = 0.71. In women, correlation was comparable but values are higher than free T: ‘passive drooling’ Sal-T = 0.12 + 2.32x ED-T, r = 0.70. The higher than expected T values in saliva, appear to be explained by T binding to salivary proteins. Iso-electric focusing of saliva proteins, followed by fractionation and LC–MS/MS assay of T showed marked testosterone peaks at pH 5.3 and 8.4, providing evidence for T binding in saliva to proteins such as albumin and proline rich protein (PRP). Conclusions: Passive drooling is the collection method of choice for testosterone in saliva. Sal-T is not directly comparable to serum free T due to T binding to saliva proteins, which substantially affects the low Sal-T in women but not the higher Sal-T in healthy adult men. Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction Testosterone (T) is one of the key hormones to assess gonadal status in males, females and children. In serum about 97–98% of T is protein-bound, i.e. for 40–60% in men and 60–70% in women bound with high affinity to sex hormone-binding globulin (SHBG) and the remaining for the major part bound with much lower affinity to albumin. Only the free T fraction of about 2–3% and probably also at least part of T loosely bound to albumin is available for biological action; free T together with albumin-bound T is often referred to as ‘bioavailable T’. For most indications (e.g. assessment of pubertal disorders, evaluation of androgen excess in females, diagnosis of clear-cut hypoandrogenism) an accurate measurement of total serum testosterone, preferably by a well standardised LC–MS/MS or GC–MS/MS method suffices [1–3]. However, total T is
q The SEXPERT study was supported by a Belgian federal Grant (government agency for Innovation by Science and Technology, IWT, www.iwt.be). ⇑ Corresponding author. Address: UZ Gent, Hormonologie 2P8, De Pintelaan 185, 9000 Gent, Belgium. Tel.: +32 93324565 (Office), mobile: +32 485412943. E-mail address: tom.fi[email protected] (T. Fiers).
http://dx.doi.org/10.1016/j.steroids.2014.04.013 0039-128X/Ó 2014 Elsevier Inc. All rights reserved.
not an optimal parameter to assess more subtle forms of hypo- and hyper-androgenism. In particular, situations with increased SHBG levels (e.g. estrogen treatment, hyperthyroidism, ageing in men) or decreased SHBG (e.g. obesity, glucocorticoid treatment) result in alterations of total T not necessarily reflected in the biologically active free T fraction that is controlled through negative feedback regulation of gonadotropins at the hypothalamo-pituitary compartment of the gonadal axis. In these situations free T will be a more reliable reflection of androgen exposure. Direct immunoassays to measure free T have proven to be inaccurate, gives results not independent of SHBG concentrations and are badly standardised [4–6], but unfortunately are still often used. In order to accurately measure serum free T, ultrafiltration isotope dilution mass spectrometry (UF-IDMS) or equilibrium dialysis (ED) isotope dilution mass spectrometry (ED-IDMS) has to be used [7]. However these are difficult and labour-intensive to perform correctly and are therefore scarcely used. Although UF is more convenient and has been used by some authors with good results [8], it is less robust than ED which Thienpont et al. used for subsequent candidate reference procedures [9,10]. Even if equilibrium dialysis is used, accuracy and precision differ widely across methods (e.g.
6
T. Fiers et al. / Steroids 86 (2014) 5–9
because of pH or temperature issues; because a dilution is used which influences the equilibrium [8,11–13]; or an indirect measurement is performed of radioactive tracer with subsequent percentage calculation of total T, increasing total error [10,13]; because total T was not measured by a LC–MS/MS method standardised against the gold standard; or a combination of the above. In the past, a number of publications have hinted at the possibility of salivary T (Sal-T) being an equally performing marker as serum free T. [14–17]. Sal-T reflects the T which diffuses through the salivary glands and was reported to be quite well correlated with serum free T [18,19]. The stress-free and not invasive nature of saliva sampling would seem to make it a perfect candidate for large scale population studies, studies with children or selected clinical usage (e.g. behavioural studies, repeated sampling, . . .) [17,20,21]. However, expected Sal-T concentrations are low in men and very low in women and children, so that Sal-T assessment is not without pitfalls, which may explain the sometimes mixed results to date. Blood contamination (e.g. teeth brushing) must be avoided prior to collection, saliva stimulation and the collection device itself may influence results and samples should be frozen at 80 °C if long term storage is needed prior to analysis [22–25]. Unfortunately, to date most studies have been performed using adapted immunoassays to measure the low Sal-T concentrations. These assays suffer from lack of specificity and from matrix related problems, similar to the long known pitfalls of serum total T assays, and may thus bias results [26,27]. Also, LC–MS/MS analysis of Sal-T has mostly been limited to samples from men, given the high sensitivity needed to measure Sal-T in women. To the best of our knowledge no critical evaluations of salivary testosterone have been described using the combination a state of the art methods i.e. Sal-T and serum total T by LC–MS/MS traceable to a primary gold standard and – free T using undiluted serum equilibrium dialysis with subsequent direct LC–MS/MS measurement of free T (ED-T). In addition we evaluated the impact of two commonly used collection devices on Sal-T. 2. Materials and methods Morning blood and saliva samples were collected in 50 male and 50 female healthy volunteers. All participants signed an informed consent and the study was approved by the ethical committee of the University Hospital of Ghent, Belgium. Serum was kept at 80 °C until batch analysis. SHBG assay was performed on E170 Modular, albumin on Cobas 8000 (Roche Diagnostics). For saliva polypropylene tubes for passive drooling and Salivette cotton rolls (Sarstedt, Nümbrecht, Germany) were used. Salivette rolls were chewed for 1 min and then replaced into the centrifugation tubes. After centrifugation at 2000g for 5 min, 250 ll supernatant from passive drooling and Salivette was collected and kept at 80 °C until LC–MS/MS analysis. Serum equilibrium dialysis was performed using Fast MicroEquilibrium dialyzer cartridges and regenerated cellulose 25 KD membranes (Harvard Apparatus; Holliston, USA) 500 ll serum (men) or 1000 ll (women) was dialysed at 37 °C for 24 h at pH 7.28 using protein free buffer prepared according to Yue et al. [28]. Serum free testosterone was also calculated (CalcFT), according to Vermeulen [11]. Samples of 100 ll serum with 400 ll physiologic solution added was used for assay of total testosterone. A liquid–liquid extraction (LLE) was used for all serum, saliva and dialysate samples: 20 ll d3-testosterone internal standard was added before extraction with 2 mL of diethylether; after mixing for 3 min samples were frozen and decanted with subsequent drying of the collected supernatant; the dried supernatant was then reconstituted in a final solution of 125 ll methanol of which 100 ll was injected for liquid chromatography. Testosterone was acquired from Sigma–Aldrich
(Saint Louis, USA), Testosterone d3 from CDN Isotopes (Quebec, Canada). All standards and internal standards were dissolved in methanol. Methanol, water and acetonitrile were LC–MS grade from BioSolve BV (Varkenswaard, The Netherlands). For measurement of serum T, Sal-T and T in dialysate following equilibrium dialysis of serum, i.e. free T(ED-T) by LC–MS/MS an AB Sciex 5500 triple-quadrupole mass spectrometer (AB Sciex; Toronto, Canada) was used, coupled with an APCI probe on the Turbo-V source. The liquid chromatography system consisted of a Shimadzu system using a C8 security guard column (5 lm, 4 2 mm) and a C8 Luna analytical column (3 lm, 50 3 mm) (Phenomenex; Torrance, USA). Measurements were performed by the tandem mass spectrometer running in multiple reaction monitoring (MRM) mode by using transitions m/z 289/109/97 for T and 292/109/97 for d3-T. It was validated for T against the IDMS candidate reference method [7] using a common serum panel. The method allows equally for determination of serum or salivary cortisol on 363/121/97 and androstenedione on 287/109/97. A declustering potential (DP) of 100 V and a collision energy (CE) of 32 eV was used for all the analytes. Data processing was performed through MultiQuant version 2.0.2. For analysis on 100 ll serum, inter-assay CV for testosterone is 6.5% at 3 ng/dL (104 pmol/L) (n = 30) with an LOQ of 1 ng/dL (35 pmol/L). For saliva and equilibrium dialysis where more sample volume is used inter-assay CV is 8.2% at 0.23 ng/dL (8 pmol/L) with an LOQ of 0.07 ng/dL (2.4 pmol/ L). To evaluate the presence of matrix specific effects, experiments were performed using sets of native and spiked eluent, serum, dialysate and saliva for 5 different samples. Compared to eluent, recoveries in serum (96–104%), dialysate (98–102%) and saliva (97– 105%) matrices proved there is no discernable matrix effect in the analysis. Saliva proteins were separated with a Bio-Rad MicroRotofor™ cell which uses patented RotoforÒ technology based on iso-electric focusing. It enables fractionation of proteins in their native state by isoelectric point (pI). An ampholyte solution (Bio-LyteÒ) with a pH gradient ranging from 3 to 10 was obtained from Bio-Rad Laboratories (CA, USA); T concentration was measured in each fraction with the assay as described above. 3. Results 3.1. Impact of collection device To investigate the potential impact of salivary collection device on T determination, ‘Salivette’ and ‘passive drooling’ Sal-T were compared to serum equilibrium dialysis (ED-T) for men. Better concordance with ED-T is observed for ‘passive drooling’ compared to ‘Salivette’ Sal-T: passive drooling Sal-T = 0.17 + 0.91x ED-T, r = 0.71; Salivette Sal-T = 0.05 + 0.88x ED-T, r = 0.43. A similarly important difference between Salivette Sal-T (r = 0.23) and passive drooling Sal-T (r = 0.71) was observed for samples of women (Supplementary data). Therefore, subsequent salivary analysis was only carried out on passive drooling samples (Fig. 1b and c) and all results reported below are for Sal-T as measured in samples obtained by passive drooling. Sal-T correlates somewhat better with total serum T for men (r = 0.68) than for women (r = 0.52). 3.2. Comparisons of salivary T with serum free T concentrations As results on free T presented in the literature are often a calculated estimation of serum free T (CalcFT) [10,29] CalcFT was also evaluated. Sal-T correlates moderately well to CalcFT: r = 0.68 for both men and women. Comparison between CalcFT and ED-T for men yielded: CalcFT = 0.09 + 1.25x ED-T; r = 0.85. Notable is the concentration dependent overestimation of CalcFT by 25% in
T. Fiers et al. / Steroids 86 (2014) 5–9
7
Fig. 1. Clockwise from left, Passing and Bablock regression: CalcFT vs equilibrium dialysis (ED-T) (female) (a), Salivette Sal-T vs ED-T (male) (b), passive drooling Sal-T vs ED-T (male) (c), passive drooling Sal-T vs ED-T (female) (d).
combination with a good linear correlation in comparison with EDT. For women, an agreement of CalcFT = 0.05 + 1.0x ED-T with r = 0.85 was found, thus a fixed bias of about 15% of an average female value (Fig. 1a). For the remainder of the analyses only EDT was considered as the ‘gold standard’ comparator. Sal-T correlates equally well to ED-T for men and women (r = 0.71), however in contrast to findings for samples from men, Passing and Bablock regression of female Sal-T vs ED-T reveals surprisingly higher than expected Sal-T values compared to ED-T: Sal-T = 0.12 + 2.32x ED-T for women; Sal-T = 0.17 + 0.91x ED-T for men (Fig. 1c and d). To elucidate this pronounced difference in men and women analytical errors were first ruled out: dialysis with different serum and dialysate volumes (increasing dialysate volume from 200 ll to 1 mL repeated for 0.2, 0.5 and 1 mL of serum) yielded comparable results (data not shown). Time to equilibrium was experimentally determined to be ±8 h for serum for men and ±3 h for serum from women, hereby validating the safety of the implemented 24 h equilibrium dialysis time and which is comparable to reported cortisol equilibrium dialysis findings by Kirchhoff et al. [30]. Determinations of duplicate extractions for serum free T dialysis showed excellent concordance (r = 0.94). As partial protein binding of T to saliva proteins was the most plausible hypothesis to explain the difference between serum free testosterone and the higher saliva testosterone, protein analysis of a subset of saliva samples was performed and yielded a median of 42 mg/dL, p25–p75: 35–53 mg/dL (i.e. about 1% of serum values). To determine whether T was actually binding to these saliva proteins, isoelectric focusing of salivary proteins in pH range 3–10 and subsequent fractionation and LC– MS/MS analysis for testosterone was performed. The results presented in Fig. 2 provide evidence of broad T binding with marked T peaks at pH 5.3 and 8.4.
Fig. 2. Isoelectric focusing and fractionating according to isoelectric point (pI) of saliva proteins. Amount of testosterone in each saliva fraction in pg/mL.
4. Discussion We present an assessment using state-of-the-art methodology of T measurements in saliva. Most studies focusing on or using measurements of T in saliva have traditionally relied on immunoassays, or have been limited to Sal-T in men given the high sensitivity needed to measure Sal-T in women. In the preparation for an epidemiological study we evaluated a commonly used saliva testosterone kit (Salimetrix; Carlsbad, USA) versus LC–MS/MS, which showed higher values for the immunoassay versus LC–MS/MS
8
T. Fiers et al. / Steroids 86 (2014) 5–9
Fig. 3. Bland and Altman difference plots of male (a) and female (b) passive drooling saliva (PD-Sal-T) vs equilibrium dialysis (ED-T).
(data not shown). More importantly, a very large spread exceeded functional sensitivity criteria (20%) at all levels confirming existing criticism on salivary T immunoassays [26]. We therefore proceeded to further assess the more laborious, but much more accurate LC–MS/MS analysis of Sal-T by comparing saliva levels to serum total T and serum free T by equilibrium dialysis. We evaluated the potential impact of saliva collection device on Sal-T. In contrast to the findings for cortisol (r = 0.94) or androstenedione (r = 0.84) which were determined on the same samples and for which results concurred well between the ‘Salivette’ collection method and the ‘passive drooling’ collection method, we demonstrated that only the ‘passive drooling’ sample collection is acceptable for Sal-T analysis. As for the assumption by many authors that saliva can be used as a surrogate for serum free T, we proved that for samples from healthy adult males collected by passive drooling, Sal-T experimentally seems a reasonably valid alternative to ED-T given proper pre-analytical care and use of reliable analytical methodology. Granger et al. [23] previously noted a bad correlation between female salivary testosterone and free serum testosterone. However they employed a direct immunoassay which was unable to correctly measure serum free T. We demonstrated that there actually is a reasonably good correlation between Sal-T and serum free T in women. However, for Sal-T in women there is a substantial positive bias of Sal-T compared to serum free T (Fig. 3). In women Sal-T cannot be considered a surrogate marker of serum free T. To explain this observation, we postulated that Sal-T is in fact not a measure of serum free T as is often stated, but rather a separate complex matrix with its own testosterone protein-binding properties. Given the low total protein content of saliva and conse-
quently only low expected T-binding, this effect would still be susceptible to significantly affect the low Sal-T in women or children, but not be apparent when measuring the much higher Sal-T in men (explaining the different observed slopes in men and women between Sal-T and ED-T). The concentration of saliva proteins we measured could indeed theoretically explain the higher than expected values in female saliva samples, if binding to a significant fraction of these proteins is occurring [11]. As the constitution of proteins in saliva is very different from that in serum, with a much smaller role for albumin, it was further unclear whether serum-like binding really occurs. The isoelectric focusing of saliva and subsequent T determination in the fractions provided evidence for this hypothesis. In addition to the expected binding to salivary albumin which is rather low in abundance, the major testosterone peaks (pH 5.3 and 8.4) correspond exactly to the migration patterns of proline rich protein (PRP) and glycosylated PRP (expressed predominantly in the parotid gland) as experimentally determined by Beeley et al. in two dimensional saliva electrophoresis [31]. This delicate mixed saliva protein balance and possible differences between individuals underscore the need for proper collection of unstimulated mixed saliva and analysis using LC–MS/MS. Although much progress has been made in these experiments as to the true analytical value of salivary testosterone and potential pitfalls, whether Sal-T for women could have a clinical applicability such as for detection of hyperandrogenism was not assessed in this study. Additional research is needed to determine the impact of nonstandard protein binding patterns or elevated female serum T on saliva T levels, which may be important factors to consider in future clinical saliva studies. In conclusion, we show that Sal-T by LC–MS/MS for salivary samples collected using the ‘passive drooling’ technique is not identical to serum free T. Sal-T correlates fairly well with serum free T, but due to binding of T to salivary proteins, in women values are higher than expected and cannot be considered a surrogate for serum free T. Moreover, this effect of protein binding can be expected to introduce additional variability of Sal-T among women. Given the prevalent Sal-T concentrations in healthy adult males the protein binding effect is negligible although formally also present. Disclosure statement The authors have nothing to disclose. Acknowledgements We thank Brigitte Bernaert, Martine De Bock and Eric Vandersypt for their invaluable experimental work without which this study would not have been possible. Alexis Dewaele, Maya Caen, Els Elaut, Katrien Vermeire, Sabine Hellemans, Lies Hendrickx, Wouter Pinxten, Katrien Symons, Ellen Van Houdenhove and Joke Vandamme for recruiting and collection of samples. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.steroids.2014. 04.013. References [1] Thienpont LM, Van Uytfanghe K, Blincko S, Ramsay CS, Xie H, Doss RC, et al. State-of-the-art of serum testosterone measurement by isotope dilution– liquid chromatography–tandem mass spectrometry. Clin Chem 2008;54(8):1290–7. [2] Rauh M. Steroid measurement with LC–MS/MS. Application examples in pediatrics. J Steroid Biochem Mol Biol 2010;121(3–5):520–7.
T. Fiers et al. / Steroids 86 (2014) 5–9 [3] Honour JW. Steroid assays in paediatric endocrinology. J Clin Res Pediatr Endocrinol 2010;2(1):1–16. [4] Rosner W. An extraordinarily inaccurate assay for free testosterone is still with us. J Clin Endocrinol Metab 2001;86:2903. [5] Fritz KS, McKean AJ, Nelson JC, et al. Analog-based free testosterone test results linked to total testosterone concentrations, not free testosterone concentrations. Clin Chem 2008;54:512–6. [6] Swerdloff RS, Wang C. Free testosterone measurement by the analog displacement direct assay: old concerns and new evidence. Clin Chem 2008;54(3):458–60. [7] Van Uytfanghe K, Stöckl D, Kaufman JM, Fiers T, De Leenheer A, Thienpont LM. Validation of 5 routine assays for serum free testosterone with a candidate reference measurement procedure based on ultrafiltration and isotope dilution–gas chromatography–mass spectrometry. Clin Biochem 2005;38(3):253–61. [8] Chen Y, Yazdanpanah M, Wang XY, Hoffman BR, Diamandis EP, Wong PY. Direct measurement of serum free testosterone by ultrafiltration followed by liquid chromatography tandem mass spectrometry. Clin Biochem 2010;43(4– 5):490–6. [9] Thienpont LM, Van Uytfanghe K, Beastall G, Faix JD, Ieiri T, Miller WG, et al. Report of the IFCC Working Group for Standardization of Thyroid Function Tests; part 2: free thyroxine and free triiodothyronine. Clin Chem 2010;56(6):912–20. [10] Hackbarth JS, Hoyne JB, Grebe SK, Singh RJ. Accuracy of calculated free testosterone differs between equations and depends on gender and SHBG concentration. Steroids 2011;76(1–2):48–55. [11] Vermeulen A, Verdonck L, Kaufman JM. A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 1999;84:3666–72. [12] Rosner W. Errors in the measurement of plasma free testosterone. J Clin Endocrinol Metab 1997;82:2014–5. [13] Szydlarska D, Grzesiuk W, Kondracka A, Bartoszewicz Z, Bar-Andziak E. Measuring salivary androgens as a useful tool in the diagnosis of polycystic ovary syndrome. Endokrynol Pol 2012;63(3):183–90. [14] Morley JE, Perry III HM, Patrick P, Dollbaum CM, Kells JM. Validation of salivary testosterone as a screening test for male hypogonadism. Aging Male 2006;9:165–9. [15] Goncharov N, Katsya G, Dobracheva A, Nizhnik A, Kolesnikova G, Herbst V, et al. Diagnostic significance of free salivary testosterone measurement using a direct luminescence immunoassay in healthy men and in patients with disorders of androgenic status. Aging Male 2006;9:111–22. [16] Arregger AL, Contreras LN, Tumilasci OR, Aquilano DR, Cardoso EM. Salivary testosterone: a reliable approach to the diagnosis of male hypogonadism. Clin Endocrinol (Oxf) 2007;67(5):656–62.
9
[17] Cardoso EM, Contreras LN, Tumilasci EG, Elbert A, Aguirre EC, Aquilano DR, et al. Salivary testosterone for the diagnosis of androgen deficiency in endstage renal disease. Nephrol Dial Transplant 2011;26(2):677–83. [18] Vittek J, L’Hommedieu DG, Gordon GG, Rappaport SC, Southren AL. Direct radioimmunoassay (RIA) of salivary testosterone: correlation with free and total serum testosterone. Life Sci 1985;37:711–6. [19] Wang C, Plymate S, Nieschlag E, Alvin Paulsen C. Salivary testosterone in men: further evidence of a direct correlation with free serum testosterone. J Clin Endocrinol Metab 1981;53:1021–4. [20] Gatti R, De Palo EF. An update: salivary hormones and physical exercise. Scand J Med Sci Sports 2011;21(2):157–69. [21] Bui HN, Schagen SE, Klink DT, Delemarre-van de Waal HA, Blankenstein MA, Heijboer AC. Salivary testosterone in female-to-male transgender adolescents during treatment with intra-muscular injectable testosterone esters. Steroids 2013;78(1):91–5. [22] Gröschl M. Current status of salivary hormone analysis. Clin Chem 2008;54(11):1759–69. [23] Granger DA, Shirtcliff EA, Booth A, Kivlighan KT, Schwartz EB. The ‘‘trouble’’ with salivary testosterone. Psychoneuroendocrinology 2004;29:1229–40. [24] Schultheiss OC. Effects of sugarless chewing gum as a stimulant on progesterone, cortisol, and testosterone concentrations assessed in saliva. Int J Psychophysiol 2013;87(1):111–4. [25] Celec P, Ostatníková D. Saliva collection devices affect sex steroid concentrations. Clin Chim Acta 2012;413(19–20):1625–8. [26] Mitchell JS, Lowe TE. Matrix effects on an antigen immobilized format for competitive enzyme immunoassay of salivary testosterone. J Immunol Methods 2009;349(1–2):61–6. [27] Wang C, Catlin DH, Demers LM, Starcevic B, Swerdloff RS. Measurement of total serum testosterone in adult men: comparison of current laboratory methods versus liquid chromatography–tandem mass spectrometry. J Clin Endocrinol Metab 2004;89(2):534–43. [28] Yue B, Rockwood AL, Sandrock T, La’ulu SL, Kushnir MM, Meikle AW. Free thyroid hormones in serum by direct equilibrium dialysis and online solidphase extraction—liquid chromatography/tandem mass spectrometry. Clin Chem 2008;54:642–51. [29] Keevil B, Macdonald P, Macdowall W, Lee D, Wu F, NATSAL Team. Salivary testosterone measurement by liquid chromatography tandem mass spectrometry in adult males and females. Ann Clin Biochem 2014;51(3):368–78. [30] Kirchhoff F, Briegel J, Vogeser M. Quantification of free serum cortisol based on equilibrium dialysis and isotope dilution–liquid chromatography–tandem mass spectrometry. Clin Biochem 2011;44(10–11):894–9. [31] Beeley JA, Khoo KS, Lamey PJ. Two-dimensional electrophoresis of human parotid salivary proteins from normal and connective tissue disorder subjects using immobilised pH gradients. Electrophoresis 1991;12(7–8):493–9.