Cortisol (total plasma cortisol, corticosteroid binding globuline, salivary cortisol)

Authors: Marijke Bremmer, Geeske Peeters

LASA file name(s):
LASAE880 / LASAE881 / LASAE882 (salivary)

Contact: Natasja van Schoor

Cortisol is a hormone produced in the cortex of the adrenal gland. It plays a key role in the bodily stress response. The function of cortisol is to facilitate adaptations of the body to changes in demands / stressors from the outside (physical trauma, infections) or inside (physical disease, psychological stress, psychological trauma). The secretion of cortisol is tightly regulated within the hypothalamus-pituitary-adrenal system (HPA-axis). Cortisol secretion by the adrenal gland shows a diurnal variation, with high levels around awakening rapidly declining to low levels in the evening. Most research in stress-related diseases has been focused on different aspects of HPA-axis regulation. With ageing functioning of the HPA-axis changes, resulting in both an increased sensitivity to stressors and a flattening of the diurnal variation. Deregulation of HPA-axis functioning has been reported in chronic stress, as a long-term consequence of severe psychological trauma and in major depression. Moreover, deregulation of the HPA-axis has been reported in many age-associated diseases, such as diabetes, Alzheimer, cardiovascular diseases and osteoporosis. Cortisol levels can be determined in urine, saliva, blood, and liquor.

Determination of total plasma cortisol and corticosteroid binding globuline (LASA-C)
Respondents were invited to a health care center near their homes where blood samples were collected in the morning. Participants were allowed to take tea and toast before, but no dairy products. The blood samples were centrifuged and serum was stored at –70 °C until processing in 2002/2003. The serum levels of Cortisol were determined using a commercially available competitive immunoassay ACS:Centauer developed by Bayer Diagnostcs in The Netherlands. Corticosteroid binding globuline (CBG) levels were determined using a radio immunoassay from Medgenix Diagnostics in Belgium. Results were expressed as nmol/L (cortisol) and mg/L (CBG). The lower limit for accurate detection of Cortisol was 30 nmol/L and the inter-assay coefficients of variation (CV) were 6% at 150 nmol/L and 8% at 1000 nmol/L and the intra-assay CV of cortisol was 3% at 700 nmol/L. The lower limit for accurate detection of CBG concentrations was 11 mg/L and the inter-assay CV were 8% at 30 mg/L and 5% at 110 mg/L. The intra-assay CV of CBG was 8% at 30 mg/L and 3% at 110 mg/L. In none of the samples the concentrations of Cortisol or CBG fell below lower detection limits.

Results LASA-C
The distribution of total cortisol and CBG in the LASA-C sample

total cortisol

Total Cortisol
· n=1276
· Cortisol is normally distributed
· Mean: 498,6 ± 172,9 nmol/L
· The upper 1% are 11 respondents with cortisol levels above 1000 nmol/L.
· No outliers.


· n=1273
· CBG is not normally distributed (skewness >1.0).
· The upper 1% are 11 respondents with CBG levels over 68 mg/L.
· No outliers

The levels of total cortisol and CBG were correlated (Spearman r = 0.123; p<.001).

Handling of the data on total plasma cortisol and CBG
All respondents were excluded using oral corticosteroids (n=26), sex hormones as in OAC or HRT (n=13) or oncolytica which consist of active sex-hormones (n=8) or mineralocorticoids (n= 11) as with these medication it is known that they heavily interfere with endogenous cortisol production (corticosteroids), CBG levels (sex-steroids) and mineralocorticoid receptor functioning (aldosteron).

The Free Cortisol Index (FCI)
For an adequate interpretation of cortisol levels the CBG concentration is needed. Approximately 90% of naturally occurring glucocorticoids are bound to a single site on the corticosteroid-binding globulin (CBG), another 8% is bound to albumine. When bound to CBG or albumine, cortisol is unable to bind to its own receptor and therefore is biologically inactive. Under stressful conditions, when there is a higher demand for glucocorticoids, CBG plasma concentrations decrease. CBG varies between and within individuals as it is affected by estrogens, acute illness and excessive exercise. For an adequate interpretation of active cortisol concentrations the free cortisol fraction is necessary. Two recent articles can be found describing a laboratory test that can directly determine the concentration of free cortisol in plasma, but most researchers estimate the free cortisol index from the concentration of total cortisol and CBG.

How to assess the Free Cortisol Index
Three different methods can be found; the first is the most frequent applied method:

  1. Free Cortisol Index
    (LeRoux et al. ) serum total cortisol (nmol/L) / CBG (mg/L)

  2. Free Cortisol Index
    (Beishuizen, Thijs, Vermes 2001) cortisol (umol/L)/ CBG (mg/mL) x100

  3. Modified Free Cortisol Index
    (Qureshi, Endocine abstracts 2002) serum total cortisol 2 / CBG

Essentially, these methods do no differ among each other; they all will generate the same ranking, only the range between values (or errors) will differ. In our study the first method has been used (according to LeRoux) and continuous levels of Albumin have been included in the multivariate models.

Choice of cut-off
Depending on the research question at hand, you can either use continuous levels of total cortisol (or the FCI) or categorize the data.

Algorithm-based Free Cortisol
Although the Free Cortisol Index is often used in the literature to estimate free cortisol levels, it is not the most precise method. Cortisol binds to both CBG and albumin. The rate to which cortisol binds to CBG or albumin depends on the concentrations of cortisol, CBG and albumin in the serum. The FCI does not correct for this phenomenon. An algorithm based on the law of mass action can be used to calculate free cortisol. For this algorithm two binding proteins (CBG and albumin) and one hormone (cortisol) are assumed and dissociation constants (Ka) are entered as described by Lentjes et al, 1999: albumin=43 gr/l, CBG in nmol/l, KaCBG=0.43*10E8 and Kaalb=0.30*10E4.5 This cortisol measure is skewed: mean=28.47 [IQR 0.01-56.53]. Although the relationship between the cortisol/CBG ratio (Free cortisol index) and the algorithm based free cortisol concentration was non linear (as expected), the coefficient of correlation for the relationship between the two appeared to be high: r >0.95 (Spearman and Pearson).
In our analysis both the Free Cortisol Index and algorithm-based free cortisol were used and the results are highly similar. In the end, seemingly it does not make a difference which measure is used, though theoretically, algorithm-based free cortisol is preferred.

Regional differences in cortisol levels
As can be seen in the two figures above, the total and free cortisol levels in the Amsterdam sample are significantly lower as compared with the Zwolle and Oss samples. These regional differences were not found in evening salivary cortisol. Also, the differences could not be explained by time point of measurement or age/sex differences. To test whether the regional differences could be explained by the lab methods used, per region, the cortisol levels of 21 respondents were reassessed using different methods. In 2002 the levels were determined using Centaur. In 2006, the levels were measured using Centaur (updated assays), DPC and Architect. To compare the different methods in 2006, the same samples were used, at the same time and using the same tubes (not the same tubes as in 2002). The figure below shows that, regardless which method was used, the regional differences remain apparent: Centaur 2002 p<0.001, Centaur 2006 p=0.018, DPC 2006 p=0.09, Architect 2006 p<0.001.

It was concluded that the regional differences could not be explained by differences in laboratory techniques. The only remaining hypothesis is that the regional differences may be explained by stress: participants in Zwolle and Oss might have more stress than the participants in Amsterdam. Since no measure for stress is available, this hypothesis cannot be tested in LASA.

Salivary cortisol

Around awakening, cortisol levels reach their peak and late in the evening the lowest values are found. Salivary cortisol levels are an accurate reflection of circulating free cortisol levels. By measuring both waking cortisol and evening cortisol, information on both absolute concentrations and the diurnal variation in cortisol secretion is given. The diurnal variation is regulated by the nucleus suprachiasmaticus in the brain, which is an important focus in depression and Alzheimer research.
Disruption of the circadian rhythm has been considered a characteristic for Cushing Syndrome1,2. The normal range for cortisol in the morning is rather broad, and concentrations overlap with those in patients with Cushing’s syndrome3. In the evening, both high serum (>50nmol/l) and salivary (>3.6 nmol/l) cortisol concentrations are related to Cushing’s syndrome3,4.

In LASA salivary cortisol measures were determined within 30 minutes after awakening and in the evening (approximately 23.00 h). Participants were asked to rinse their mouth and wait for ten minutes before starting to chew the cotton ball. They had to prevent bleeding of the gum previous to and during the collection of the saliva. The cotton balls were chewed on for approximately 1.5 minutes and then put into a tube. The samples were kept in the refrigerator until mailing or collecting by the visiting nurse.
Salivary cortisol levels are an accurate reflection of circulating free cortisol levels. Salivary cortisol (nmol/L) was determined using radio immunoassay coated tubes (Spectria Orion Diagnostics, Finland). The detection limit was 1.5 nmol/L. The intra- and inter-assay coefficients of variation (CV) were less than 19% (see Table).
From the samples, 4 awakening levels and 92 evening levels were below the detection limit. Since evening levels are quite low, a fixed value (for instance 1.5 nmol/L) would have influenced the distribution too much (regression to the mean). Hence, these missing values with randomized values between 0 and 1.5 were imputed in LASAE882. Another possibility is to categorize the original data. This was done in LASAE881.

Handling of the data
The files have been cleaned. It is advisable to exclude respondents using sex-steroids or glucocorticoids from analysis (and optionally mineral corticoids, however, the impact of these medications on cortisol level is very small). Cortisol levels around waking and the evening cortisol levels can be analyzed independently of each other. Also, the difference between waking and evening can be informative of the diurnal variation.


  1. Findling JW, Doppman JL. 1994 Biochemical and radiological diagnosis of Cushing’s syndrome. Endocrinol Metab Clin North Am. 23:511–537.
  2. Newell-Price J, Trainer P, Perry L, Wass J, Grossman A, Besser M. 1995 A single sleeping midnight cortisol has 100% sensitivity for the diagnosis of Cushing’s syndrome. Clin Endocrinol (Oxf). 43:545–550.
  3. Raff H, Raff FL, Findling JW. 1998 Late-night salivary cortisol as a screening test for Cushing’s syndrome. J Clin Endorcrin Metab. 83:2681-2686.
  4. Jones MT, Gillham B. 1988 Factors involved in the regulation of adrenocorticotropic hormone/b-lipotropic hormone. Physiol Rev. 68:743– 818.
  5. Lentjes, E.G. & Romijn, F.H. (1999) Temperature-dependent cortisol distribution among the blood compartments in man. J Clin Endocrinol Metab, 84, 682-687.
  6. Peeters G.M., van Schoor N.M., Visser M., Knol D.L., Eekhoff E.M., de Ronde W., Lips P. Relationship between cortisol and physical performance in older persons. Clinical Endocrinol. (Oxf) 2007; 67:398-406.