Best Practice & Research Clinical Endocrinology & Metabolism 27 (2013) 795–801
Contents lists available at ScienceDirect
Best Practice & Research Clinical Endocrinology & Metabolism journal homepage: www.elsevier.com/locate/beem
5
Determination of cortisol in serum, saliva and urine Ursula Turpeinen, Ph.D., Chemist *, Esa Hämäläinen, M.D., Head of the Women's Clinic Laboratory 1 HUSLAB, Laboratory of Women’s Clinic, Haartmaninkatu 2, 00290 Helsinki, Finland
Keywords: serum cortisol urinary cortisol HPLC LC–MS/MS
Cortisol is quantitatively the major glucocorticoid product of the adrenal cortex. The main reason to measure cortisol is to diagnose human diseases characterised by deficiency of adrenal steroid excretion in Addison’s disease or overproduction in Cushing’s syndrome (CS). In both cases a sensitive, accurate and reproducible assay of cortisol is required. Several methods have been described for the quantitative measurement of cortisol in both serum and urine. The most widely used methods in routine clinical laboratories are immunoassays (IA) and enzyme immunoassays (EIA), luminescence and fluorescence assays, which are available in numerous commercial kits and on automated platforms. However, there remains a number of problems in the so-called direct immunoassays if extraction and prepurification are not carried out before the assay. Recently, more specific chromatographic methods have been introduced, such as high pressure liquid chromatographic (HPLC) or liquid chromatography tandem mass spectrometric assays (LC–MS/MS). The high specificity especially of LCMS/MS facilitates reliable measurement of cortisol both in plasma, urine and saliva samples. Ó 2013 Elsevier Ltd. All rights reserved.
Physiological and clinical background Cortisol is the main glucocorticoid hormone produced by the adrenal cortex. Its synthesis is stimulated by the hypothalamic corticotropin releasing hormone (CRH) and adrenocorticotropin * Corresponding author. Tel.: þ358 9 4717 2845; Fax: þ358 9 4717 4806. E-mail addresses: ursula.turpeinen@hus.fi (U. Turpeinen), esa.hamalainen@hus.fi (E. Hämäläinen). 1 Tel.: þ358 9 4717 74944. 1521-690X/$ – see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.beem.2013.10.008
796
U. Turpeinen, E. Hämäläinen / Best Practice & Research Clinical Endocrinology & Metabolism 27 (2013) 795–801
(ACTH) from the hypophysis and regulated by cortisol itself through a feedback mechanism. Cortisol has large circadian variation, which follows the sleep-wake rhythm with high levels in the morning and the lowest ones at midnight [1]. Cortisol is transported bound to the specific carrier protein, cortisol binding globulin (CBG) and to albumin. The biologically active free fraction comprises only 2–5% of the total hormone concentration. Most of cortisol is excreted into urine as tetrahydrocortisol metabolites conjugated glucuronides and sulphates, and only about 3% occurs as the native hormone. Urinary excretion correlates with cortisol production and moderately with serum cortisol concentration. The main clinical use of cortisol determinations is the assessment of overproduction in CS or hypocortisolism in Addison’s disease. Furthermore, high serum levels are also found in stress responses, psychiatric diseases, obesity, diabetes, alcoholism and pregnancy, which may cause diagnostic problems in patients with CS. Low levels of cortisol are seen in patients with rare adrenal enzyme defects and after long-lasting stress. For diagnostic purposes the following analyses are used: Total and free cortisol in serum, overnight or 24-h excretion of urinary cortisol, mid-night serum or salivary cortisol. In addition, serum measurement is often combined with dexametasone suppression or ACTH-stimulation test [2]. Sample collection A validated sampling protocol for serum and urine is important since incorrect sampling may significantly influence the results. Because of the circadian variation of cortisol secretion, the deviation from a fixed daily sampling procedure may increase variation of the results. In some cases the clinical interpretation may not be possible without exact knowledge of the collection time of the sample. A urinary cortisol sample represents the production of cortisol over a fixed time period and urine excretion may therefore better reflect cortisol production than a single serum sample. Repeated 24-h urine collections may, however, be difficult in clinical practice. Accurate collection of urine needs good cooperation with the patient. Therefore, in many laboratories, the mean of two to three 24-h urine collections are used for reliable estimation of the average daily cortisol excretion. Serum cortisol Serum total cortisol has previously been measured by colorimetric, spectroscopic or chromatographic methods. Presently, immunoassays with radioactive or non-isotopic labels are mainly used. However, mass spectrometric methods are becoming more popular in routine laboratories. According to data from the UK National External Quality Assurance Scheme (UKNEQAS, www.birminghamquality. org.uk), which extensively monitors the reliability of currently used cortisol methods, most of the 264 laboratories participating in the survey (October 2012) used immunological methods. Only three used a mass spectrometric method. Different immunoassays show variable bias from the all laboratory mean value. The laboratories that use either gas chromatography mass spectrometry (GC–MS) or LC–MS (liquid chromatography mass spectrometry) report lower values than the all laboratory mean. Immunoassays can give satisfactory results but the specificity is often a problem especially in socalled direct assays. In these, serum is directly added to the reagents without any prepurification and this can cause both matrix interference and cross-reactions. Antisera raised against cortisolprotein antigens conjugates are rarely totally specific for cortisol. Chemically similar steroids such as prednisolone and 11-deoxycortisol often cross-react in cortisol assays. Another problem in direct assays is that endogenous plasma proteins compete with the antibody for the same antigen. This problem is usually suppressed, but not completely eliminated, by addition of a blocking agent. The structure of the blocking agent used in commercial kits is not always given by the manufacturer and it is not clear how well it prevents the problem. Better specificity requires more complex methods, e.g., prepurification of the sample by chromatography and using sophisticated detection methods such as tandem mass spectrometry. One of the first reports of a better method for serum cortisol determination used dichloromethane extraction, reversed phase liquid chromatography and UV detection [3]. Cortisol concentrations down to 10 nmol/ L could be detected in 1 mL of serum. Later Hariharan et al. developed a similar assay but with a better detection limit of 0.8 nmol/L by using solid phase extraction (SPE) and a minibore reversed phase column [4]. McBride et al. [5] also used SPE followed by reversed phase chromatography and UV
U. Turpeinen, E. Hämäläinen / Best Practice & Research Clinical Endocrinology & Metabolism 27 (2013) 795–801
797
detection. Their lowest detectable concentration was about 27 nmol/L and, in addition to cortisol, the method could also be used to determine prednisone and prednisolone. The technically demanding sample preparation step in chromatographic methods is time consuming and alternative sample preparation procedures have been developed. Turnell et al. used a fully automated technique combining dialysis and trace enrichment. The dialysis removes macromolecules and the trace enrichment concentrates the solutes in the dialysate. This permits loading of untreated serum samples directly onto an HPLC autosampler [6]. Specific immunological assessment of steroid hormones in biological fluids requires effective purification before the assay. Automatic high pressure liquid chromatography (HPLC) can be used as a purification step after sample extraction and before immunoassay [7]. The most widely used methods for serum cortisol assay are based on immunoassay (IA), which can be easily automated. More than 30 different methods are available as manual kits and on automated platforms. The label in these assays can be a radioactive isotope (RIA), an enzyme (EIA), a fluorophore (FIA) or a luminescent label (LIA). The between-assay agreements of these methods is usually unsatisfactory. When compared with tandem mass spectrometry, many immunoassays show both over- and underestimations of true cortisol concentrations [8], which may lead to erroneous clinical decisions. In urine assays, high results may be attributed to cross-reactions with steroid metabolites, while precursor steroids and cortisol binding proteins tend to interfere in serum assays. Since IA methods are not satisfactory for cortisol measurements it is worth considering chromatographic methods with specific detection such as LC–MS. In the past GC–MS were fairly widely used but because it requires derivatization it is time-consuming making it incompatible with the high throughput required in routine clinical laboratories. Two almost identical isotope dilution LC–MS and LC–MS/MS methods have been developed using positive ESI (electrospray ionisation) and MRM (multiple reaction monitoring) by Jung et al. [9] and Tai and Welch [10]. In both methods the results were validated using certified reference materials (CRM) 192 and 193 [11]. The authors suggest that their methods qualify as candidate reference methods for serum cortisol based on accuracy, precision and lack of interference. Sensitivity of LC–MS methodology depends on ionisation efficiency and in order to improve it, it might be necessary to consider alternative ions or to use negative ESI [12]. Reduction of background noise can usually be achieved by using MRM or by derivatizing the analyte prior to LC. Derivatization adds a step to the procedure thus lengthening total analysis time. Prior to LC– MS/MS, cortisol is usually extracted from plasma into ethyl acetate or dichlormethane after addition of the deuterated internal standard. The solvent is then removed by evaporation and the steroids reconstituted in the mobile phase before injection. The lower limit of quantification achieved by LC– MS/MS is perfectly satisfactory for measurement of cortisol concentrations in plasma. Serum free cortisol Analysis of total cortisol in serum is adequate for most clinical purposes but seriously ill patients often have low plasma protein concentrations. This lowers the fraction of protein bound cortisol and thus total cortisol levels. In these patients, free cortisol is therefore a more reliable indicator of adrenal function and therefore it reflects disease activity better than total serum cortisol [13]. However, analysis of free cortisol using ultrafiltration, equilibration dialysis or gel filtration is complicated making it unsuitable for routine use. These methods are hampered by poor precision, they are laborious and time-consuming and difficult to standardise. Furthermore a very sensitive assay is needed because of the low concentrations in serum. Quantitation of free cortisol, based on the measurement of both total cortisol and CBG, have problems because of varying affinity of cortisol to CBG [14]. Thus the routine use of free serum cortisol assays is very limited. Salivary cortisol Salivary cortisol is an ultrafiltrate of plasma cortisol and reflects the levels of biologically active, nonprotein bound cortisol in serum. It follows the circadian variation of serum cortisol, with the highest levels in the morning and lowest at midnight. Late-night salivary cortisol is commonly used as a screening test for Cushing’s syndrome (CS). For diagnosis of hypercortisolism, an increased
798
U. Turpeinen, E. Hämäläinen / Best Practice & Research Clinical Endocrinology & Metabolism 27 (2013) 795–801
concentration of late-night salivary cortisol is highly sensitive and specific, i.e., >90%. It is thus comparable with other screening tests for CS such as the 1 mg dexamethasone suppression test and 24 h urinary free cortisol [15]. Recent studies have shown that salivary cortisol may be useful in combination with the ACTH stimulation test to detect adrenal insufficiency, especially in patients with low CBG concentrations [16]. Salivary cortisol is also used to monitor glucocorticoid treatment of patients with congenital adrenal hyperplasia and may be useful for monitoring of hydrocortisone replacement therapy. An advantage of salivary cortisol testing is the easy, non-invasive sample collection. By using the Salivette polyester swab device, which does not adsorb steroids, patients can collect saliva samples at home or in hospital wards under stress-free conditions. Salivary samples can be transferred to the laboratory during the following day. Salivary cortisol is stable at room temperature for 1–2 days and at refrigerator temperature for a week. However, eating, smoking and brushing of the teeth should be avoided 2 h before collection of saliva and the mouth should be rinsed with water 10–15 min before sampling. Salivary assays may be impossible in patients with oral diseases such as bleeding gingivitis or in conditions with decreased salivary excretion, such as Sjögren’s syndrome. Pregnant women need their own reference ranges for saliva, due to high total and free cortisol levels. Several techniques have been used to measure salivary cortisol. The most commonly used ones have been IAs, including in-house RIAs and commercial cortisol assays, modified to improve sensitivity. Recently, LC–MS/MS has provided promising results. Because the salivary cortisol concentration is less than one tenth of that in serum and decreases to the low nanomolar range during late-night, a sensitive cortisol assay is needed. However, the variation in sensitivity is method dependent and large. The detection limits of LC–MS/MS have reported to be 0.07–0.11 nmol/L [17,18]. which is clearly better than that of immunoassays (i.e., 0.4–5.8 nmol/L) [19– 22]. Furthermore, other steroids tend to crossreact in immunoassays, especially 6beta-OH-cortisol, 21deoxycortisol and 11-deoxycortisol [23]. Thus, the agreement between salivary cortisol concentrations measured by LC–MS/MS and immunoassays is poor [17,18,24]. LC–MS/MS gives significantly lower reference ranges and cut-off limits than immunoassays. Baid et al. compared their LC–MS/MS method and RIA and showed that the cut-off level for salivary cortisol with LC–MS/MS was 2.8 nmol/l, which was about half of that obtained by RIA (4.7 nmol/L) [17]. Similar observations have been reported in other studies using other LC–MS/MS methods [18,24–26]. The upper limit for late-night salivary cortisol using LC–MS/MS has varied between 2.7 and 3.0 nmol/l, whereas that determined by immunoassays has been reported to be 3.6–15.2 nmol/L [23]. Salivary cortisol is a useful tool for detection of hypercortisolism and fluctuation of cortisol production. The determination should preferably be done by LC–MS/MS due to its a superior specificity and sensitivity. 0ther advantages of modern LC–MS/MS methods are rapid turnaround time when compared to RIAs with extraction steps, the possibility for pre-analytical automation and lack of interference by drugs and antibodies, which are frequent problems in immunoassays. Urinary cortisol The concentration of cortisol in urine reflects unbound and biologically active cortisol in plasma. Measurement of urinary free cortisol is clinically important in the diagnosis of CS. Because of the diurnal rhythm of cortisol excretion, 24-h urine is usually collected. In comparison with the more widely used serum determinations, 24-h urinary cortisol is not affected by diurnal variation or the confounding effect of cortisol binding proteins in serum. There are several methods for measurement of urinary free cortisol (UFC) concentrations. Routine determinations have been performed by competitive protein-binding assay [27] RIA [28] and HPLC [29–31] and recently by LC–MS/MS [32–34]. RIA has been used to determine cortisol in urine either directly or after extraction. While the specificity of antisera employed in most immunoassays is acceptable for measurement of serum cortisol, urine contains more crossreacting substances that interfere in immunoassays [35]. Some of the compounds that cause interference are known but most of them remain unidentified- [35,36]. The general standard of immunoassays for urinary cortisol is unsatisfactory [37,38]. Owing to cross-reacting substances, urinary free cortisol tends to be overestimated by immunoassays and the results obtained are often two- to threefold those obtained by HPLC [30,31]. Despite the superiority of HPLC methods, immunological methods are still
U. Turpeinen, E. Hämäläinen / Best Practice & Research Clinical Endocrinology & Metabolism 27 (2013) 795–801
799
mainly used [38]. Improved performance of direct immunoassays can often be achieved by solvent extraction prior to assay, usually with dichloromethane or ethyl acetate. Solid-phase extraction using disposable cartridges is now widely used for this purpose. However, in order to achieve satisfactory specificity by immunoassays, some form of chromatographic purification is required [29]. Several coeluting steroids or drugs may interfere with HPLC and UV detection leading to difficulties in identification and quantitation. In order to reduce analytical interference, improve accuracy and shorten the analysis time, it is worth considering LC–MS/MS, which provides positive identification of each compound detected making it superior to immunoassays. LC–MS/MS also solves the problem of long analysis times encountered in HPLC methods while improving sensitivity and specificity. The results obtained by LC–MS/MS and HPLC show strong correlation, especially at concentrations above 20 nmol/l [32,39]. Only a few direct cortisol immunoassays have been properly evaluated for urine determinations. Ching et al. compared urine cortisol results obtained by a direct immunoassays and an HPLC method [40]. The upper reference limit of the immunoassay was
Contents lists available at ScienceDirect
Best Practice & Research Clinical Endocrinology & Metabolism journal homepage: www.elsevier.com/locate/beem
5
Determination of cortisol in serum, saliva and urine Ursula Turpeinen, Ph.D., Chemist *, Esa Hämäläinen, M.D., Head of the Women's Clinic Laboratory 1 HUSLAB, Laboratory of Women’s Clinic, Haartmaninkatu 2, 00290 Helsinki, Finland
Keywords: serum cortisol urinary cortisol HPLC LC–MS/MS
Cortisol is quantitatively the major glucocorticoid product of the adrenal cortex. The main reason to measure cortisol is to diagnose human diseases characterised by deficiency of adrenal steroid excretion in Addison’s disease or overproduction in Cushing’s syndrome (CS). In both cases a sensitive, accurate and reproducible assay of cortisol is required. Several methods have been described for the quantitative measurement of cortisol in both serum and urine. The most widely used methods in routine clinical laboratories are immunoassays (IA) and enzyme immunoassays (EIA), luminescence and fluorescence assays, which are available in numerous commercial kits and on automated platforms. However, there remains a number of problems in the so-called direct immunoassays if extraction and prepurification are not carried out before the assay. Recently, more specific chromatographic methods have been introduced, such as high pressure liquid chromatographic (HPLC) or liquid chromatography tandem mass spectrometric assays (LC–MS/MS). The high specificity especially of LCMS/MS facilitates reliable measurement of cortisol both in plasma, urine and saliva samples. Ó 2013 Elsevier Ltd. All rights reserved.
Physiological and clinical background Cortisol is the main glucocorticoid hormone produced by the adrenal cortex. Its synthesis is stimulated by the hypothalamic corticotropin releasing hormone (CRH) and adrenocorticotropin * Corresponding author. Tel.: þ358 9 4717 2845; Fax: þ358 9 4717 4806. E-mail addresses: ursula.turpeinen@hus.fi (U. Turpeinen), esa.hamalainen@hus.fi (E. Hämäläinen). 1 Tel.: þ358 9 4717 74944. 1521-690X/$ – see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.beem.2013.10.008
796
U. Turpeinen, E. Hämäläinen / Best Practice & Research Clinical Endocrinology & Metabolism 27 (2013) 795–801
(ACTH) from the hypophysis and regulated by cortisol itself through a feedback mechanism. Cortisol has large circadian variation, which follows the sleep-wake rhythm with high levels in the morning and the lowest ones at midnight [1]. Cortisol is transported bound to the specific carrier protein, cortisol binding globulin (CBG) and to albumin. The biologically active free fraction comprises only 2–5% of the total hormone concentration. Most of cortisol is excreted into urine as tetrahydrocortisol metabolites conjugated glucuronides and sulphates, and only about 3% occurs as the native hormone. Urinary excretion correlates with cortisol production and moderately with serum cortisol concentration. The main clinical use of cortisol determinations is the assessment of overproduction in CS or hypocortisolism in Addison’s disease. Furthermore, high serum levels are also found in stress responses, psychiatric diseases, obesity, diabetes, alcoholism and pregnancy, which may cause diagnostic problems in patients with CS. Low levels of cortisol are seen in patients with rare adrenal enzyme defects and after long-lasting stress. For diagnostic purposes the following analyses are used: Total and free cortisol in serum, overnight or 24-h excretion of urinary cortisol, mid-night serum or salivary cortisol. In addition, serum measurement is often combined with dexametasone suppression or ACTH-stimulation test [2]. Sample collection A validated sampling protocol for serum and urine is important since incorrect sampling may significantly influence the results. Because of the circadian variation of cortisol secretion, the deviation from a fixed daily sampling procedure may increase variation of the results. In some cases the clinical interpretation may not be possible without exact knowledge of the collection time of the sample. A urinary cortisol sample represents the production of cortisol over a fixed time period and urine excretion may therefore better reflect cortisol production than a single serum sample. Repeated 24-h urine collections may, however, be difficult in clinical practice. Accurate collection of urine needs good cooperation with the patient. Therefore, in many laboratories, the mean of two to three 24-h urine collections are used for reliable estimation of the average daily cortisol excretion. Serum cortisol Serum total cortisol has previously been measured by colorimetric, spectroscopic or chromatographic methods. Presently, immunoassays with radioactive or non-isotopic labels are mainly used. However, mass spectrometric methods are becoming more popular in routine laboratories. According to data from the UK National External Quality Assurance Scheme (UKNEQAS, www.birminghamquality. org.uk), which extensively monitors the reliability of currently used cortisol methods, most of the 264 laboratories participating in the survey (October 2012) used immunological methods. Only three used a mass spectrometric method. Different immunoassays show variable bias from the all laboratory mean value. The laboratories that use either gas chromatography mass spectrometry (GC–MS) or LC–MS (liquid chromatography mass spectrometry) report lower values than the all laboratory mean. Immunoassays can give satisfactory results but the specificity is often a problem especially in socalled direct assays. In these, serum is directly added to the reagents without any prepurification and this can cause both matrix interference and cross-reactions. Antisera raised against cortisolprotein antigens conjugates are rarely totally specific for cortisol. Chemically similar steroids such as prednisolone and 11-deoxycortisol often cross-react in cortisol assays. Another problem in direct assays is that endogenous plasma proteins compete with the antibody for the same antigen. This problem is usually suppressed, but not completely eliminated, by addition of a blocking agent. The structure of the blocking agent used in commercial kits is not always given by the manufacturer and it is not clear how well it prevents the problem. Better specificity requires more complex methods, e.g., prepurification of the sample by chromatography and using sophisticated detection methods such as tandem mass spectrometry. One of the first reports of a better method for serum cortisol determination used dichloromethane extraction, reversed phase liquid chromatography and UV detection [3]. Cortisol concentrations down to 10 nmol/ L could be detected in 1 mL of serum. Later Hariharan et al. developed a similar assay but with a better detection limit of 0.8 nmol/L by using solid phase extraction (SPE) and a minibore reversed phase column [4]. McBride et al. [5] also used SPE followed by reversed phase chromatography and UV
U. Turpeinen, E. Hämäläinen / Best Practice & Research Clinical Endocrinology & Metabolism 27 (2013) 795–801
797
detection. Their lowest detectable concentration was about 27 nmol/L and, in addition to cortisol, the method could also be used to determine prednisone and prednisolone. The technically demanding sample preparation step in chromatographic methods is time consuming and alternative sample preparation procedures have been developed. Turnell et al. used a fully automated technique combining dialysis and trace enrichment. The dialysis removes macromolecules and the trace enrichment concentrates the solutes in the dialysate. This permits loading of untreated serum samples directly onto an HPLC autosampler [6]. Specific immunological assessment of steroid hormones in biological fluids requires effective purification before the assay. Automatic high pressure liquid chromatography (HPLC) can be used as a purification step after sample extraction and before immunoassay [7]. The most widely used methods for serum cortisol assay are based on immunoassay (IA), which can be easily automated. More than 30 different methods are available as manual kits and on automated platforms. The label in these assays can be a radioactive isotope (RIA), an enzyme (EIA), a fluorophore (FIA) or a luminescent label (LIA). The between-assay agreements of these methods is usually unsatisfactory. When compared with tandem mass spectrometry, many immunoassays show both over- and underestimations of true cortisol concentrations [8], which may lead to erroneous clinical decisions. In urine assays, high results may be attributed to cross-reactions with steroid metabolites, while precursor steroids and cortisol binding proteins tend to interfere in serum assays. Since IA methods are not satisfactory for cortisol measurements it is worth considering chromatographic methods with specific detection such as LC–MS. In the past GC–MS were fairly widely used but because it requires derivatization it is time-consuming making it incompatible with the high throughput required in routine clinical laboratories. Two almost identical isotope dilution LC–MS and LC–MS/MS methods have been developed using positive ESI (electrospray ionisation) and MRM (multiple reaction monitoring) by Jung et al. [9] and Tai and Welch [10]. In both methods the results were validated using certified reference materials (CRM) 192 and 193 [11]. The authors suggest that their methods qualify as candidate reference methods for serum cortisol based on accuracy, precision and lack of interference. Sensitivity of LC–MS methodology depends on ionisation efficiency and in order to improve it, it might be necessary to consider alternative ions or to use negative ESI [12]. Reduction of background noise can usually be achieved by using MRM or by derivatizing the analyte prior to LC. Derivatization adds a step to the procedure thus lengthening total analysis time. Prior to LC– MS/MS, cortisol is usually extracted from plasma into ethyl acetate or dichlormethane after addition of the deuterated internal standard. The solvent is then removed by evaporation and the steroids reconstituted in the mobile phase before injection. The lower limit of quantification achieved by LC– MS/MS is perfectly satisfactory for measurement of cortisol concentrations in plasma. Serum free cortisol Analysis of total cortisol in serum is adequate for most clinical purposes but seriously ill patients often have low plasma protein concentrations. This lowers the fraction of protein bound cortisol and thus total cortisol levels. In these patients, free cortisol is therefore a more reliable indicator of adrenal function and therefore it reflects disease activity better than total serum cortisol [13]. However, analysis of free cortisol using ultrafiltration, equilibration dialysis or gel filtration is complicated making it unsuitable for routine use. These methods are hampered by poor precision, they are laborious and time-consuming and difficult to standardise. Furthermore a very sensitive assay is needed because of the low concentrations in serum. Quantitation of free cortisol, based on the measurement of both total cortisol and CBG, have problems because of varying affinity of cortisol to CBG [14]. Thus the routine use of free serum cortisol assays is very limited. Salivary cortisol Salivary cortisol is an ultrafiltrate of plasma cortisol and reflects the levels of biologically active, nonprotein bound cortisol in serum. It follows the circadian variation of serum cortisol, with the highest levels in the morning and lowest at midnight. Late-night salivary cortisol is commonly used as a screening test for Cushing’s syndrome (CS). For diagnosis of hypercortisolism, an increased
798
U. Turpeinen, E. Hämäläinen / Best Practice & Research Clinical Endocrinology & Metabolism 27 (2013) 795–801
concentration of late-night salivary cortisol is highly sensitive and specific, i.e., >90%. It is thus comparable with other screening tests for CS such as the 1 mg dexamethasone suppression test and 24 h urinary free cortisol [15]. Recent studies have shown that salivary cortisol may be useful in combination with the ACTH stimulation test to detect adrenal insufficiency, especially in patients with low CBG concentrations [16]. Salivary cortisol is also used to monitor glucocorticoid treatment of patients with congenital adrenal hyperplasia and may be useful for monitoring of hydrocortisone replacement therapy. An advantage of salivary cortisol testing is the easy, non-invasive sample collection. By using the Salivette polyester swab device, which does not adsorb steroids, patients can collect saliva samples at home or in hospital wards under stress-free conditions. Salivary samples can be transferred to the laboratory during the following day. Salivary cortisol is stable at room temperature for 1–2 days and at refrigerator temperature for a week. However, eating, smoking and brushing of the teeth should be avoided 2 h before collection of saliva and the mouth should be rinsed with water 10–15 min before sampling. Salivary assays may be impossible in patients with oral diseases such as bleeding gingivitis or in conditions with decreased salivary excretion, such as Sjögren’s syndrome. Pregnant women need their own reference ranges for saliva, due to high total and free cortisol levels. Several techniques have been used to measure salivary cortisol. The most commonly used ones have been IAs, including in-house RIAs and commercial cortisol assays, modified to improve sensitivity. Recently, LC–MS/MS has provided promising results. Because the salivary cortisol concentration is less than one tenth of that in serum and decreases to the low nanomolar range during late-night, a sensitive cortisol assay is needed. However, the variation in sensitivity is method dependent and large. The detection limits of LC–MS/MS have reported to be 0.07–0.11 nmol/L [17,18]. which is clearly better than that of immunoassays (i.e., 0.4–5.8 nmol/L) [19– 22]. Furthermore, other steroids tend to crossreact in immunoassays, especially 6beta-OH-cortisol, 21deoxycortisol and 11-deoxycortisol [23]. Thus, the agreement between salivary cortisol concentrations measured by LC–MS/MS and immunoassays is poor [17,18,24]. LC–MS/MS gives significantly lower reference ranges and cut-off limits than immunoassays. Baid et al. compared their LC–MS/MS method and RIA and showed that the cut-off level for salivary cortisol with LC–MS/MS was 2.8 nmol/l, which was about half of that obtained by RIA (4.7 nmol/L) [17]. Similar observations have been reported in other studies using other LC–MS/MS methods [18,24–26]. The upper limit for late-night salivary cortisol using LC–MS/MS has varied between 2.7 and 3.0 nmol/l, whereas that determined by immunoassays has been reported to be 3.6–15.2 nmol/L [23]. Salivary cortisol is a useful tool for detection of hypercortisolism and fluctuation of cortisol production. The determination should preferably be done by LC–MS/MS due to its a superior specificity and sensitivity. 0ther advantages of modern LC–MS/MS methods are rapid turnaround time when compared to RIAs with extraction steps, the possibility for pre-analytical automation and lack of interference by drugs and antibodies, which are frequent problems in immunoassays. Urinary cortisol The concentration of cortisol in urine reflects unbound and biologically active cortisol in plasma. Measurement of urinary free cortisol is clinically important in the diagnosis of CS. Because of the diurnal rhythm of cortisol excretion, 24-h urine is usually collected. In comparison with the more widely used serum determinations, 24-h urinary cortisol is not affected by diurnal variation or the confounding effect of cortisol binding proteins in serum. There are several methods for measurement of urinary free cortisol (UFC) concentrations. Routine determinations have been performed by competitive protein-binding assay [27] RIA [28] and HPLC [29–31] and recently by LC–MS/MS [32–34]. RIA has been used to determine cortisol in urine either directly or after extraction. While the specificity of antisera employed in most immunoassays is acceptable for measurement of serum cortisol, urine contains more crossreacting substances that interfere in immunoassays [35]. Some of the compounds that cause interference are known but most of them remain unidentified- [35,36]. The general standard of immunoassays for urinary cortisol is unsatisfactory [37,38]. Owing to cross-reacting substances, urinary free cortisol tends to be overestimated by immunoassays and the results obtained are often two- to threefold those obtained by HPLC [30,31]. Despite the superiority of HPLC methods, immunological methods are still
U. Turpeinen, E. Hämäläinen / Best Practice & Research Clinical Endocrinology & Metabolism 27 (2013) 795–801
799
mainly used [38]. Improved performance of direct immunoassays can often be achieved by solvent extraction prior to assay, usually with dichloromethane or ethyl acetate. Solid-phase extraction using disposable cartridges is now widely used for this purpose. However, in order to achieve satisfactory specificity by immunoassays, some form of chromatographic purification is required [29]. Several coeluting steroids or drugs may interfere with HPLC and UV detection leading to difficulties in identification and quantitation. In order to reduce analytical interference, improve accuracy and shorten the analysis time, it is worth considering LC–MS/MS, which provides positive identification of each compound detected making it superior to immunoassays. LC–MS/MS also solves the problem of long analysis times encountered in HPLC methods while improving sensitivity and specificity. The results obtained by LC–MS/MS and HPLC show strong correlation, especially at concentrations above 20 nmol/l [32,39]. Only a few direct cortisol immunoassays have been properly evaluated for urine determinations. Ching et al. compared urine cortisol results obtained by a direct immunoassays and an HPLC method [40]. The upper reference limit of the immunoassay was
