Standardization of steroid tests and implications for the endocrine community

Two papers have evaluated a new cortisol immunoassay (IA) produced for Roche instruments (Roche Diagnostics, 9115 Hague Road, PO Box 50457, Indianapolis, IN 46250-0457) in the investigation of adrenal insufficiency^1,2; Roche II cortisol is a competitive IA based on a monoclonal antibody with improved specificity compared with the Roche Cortisol I assay that used polyclonal antibodies. The new assay performs favourably against the Institute for Reference Materials and measurements/IFCC-451 panel by isotope dilution gas chromatography/mass spectrometry, although the case is unproven for all types of samples sent to the clinical laboratory. Results for Cortisol II can be 20% or 30% lower than for Cortisol I assay which may make interpretation difficult for clinicians.

Using different approaches to data collection in adrenal function tests, including the Synacthen© (tetracosactide) test, the authors propose a cut-off for stimulated cortisol concentrations in plasma/serum of 374 or 350 nmol/L, respectively. The Roche Cortisol II kit insert suggests a cut-off of 420 nmol/L for adrenal insufficiency. These concentrations contrast with the 500 or 550 nmol/L values, often used in the literature for nearly 50 years and still in many laboratories, based on fluorometric assay and other assays with poor specificity. ³ Judging by the differences in baseline cortisol concentrations between the most commonly used IA, the differences in the peak concentration after ACTH stimulation are not unexpected. ⁴ It is astounding that after around 40 years of using IA for cortisol this remains a conundrum in endocrinology. When dissecting the likely issues it is, however, not surprising.

The analytical methodology has improved considerably over the years although IAs were never expected to enable accurate measurements of cortisol in all situations and still do not perform in harmony.^5,6 The assay of cortisol in samples after solvent/solid phase extraction and chromatography probably achieved the most accurate results until mass spectrometric (MS) testing became available. Those approaches until recently were not practical in routine hospital settings and were confined to research studies. Liquid chromatography (or better gas chromatography) coupled with tandem mass spectrometry is now more user friendly and is the method of choice for steroid analysis. The cost of MS methods needs be no more than using IA. Some journals threatened to reject papers where steroids had not been analysed by the best of analytical methods.

Clinicians and patients are now faced with having to understand new reference ranges in basal, stimulated and suppressed situations whatever method is used. These changes in results will have a huge impact on longitudinal and serial studies. There are known to be issues in IA with gender, matrix, cortisol-binding globulin concentrations, oral contraceptive use, pregnancy, dexamethasone suppression, treatment with synthetic glucocorticoids (prednisone, prednisolone). Hospital laboratories for many years have been dependent on suppliers of reagents and instruments. ⁷ The workloads for tests have grown exponentially and so automated systems became essential many years ago. This then led to the ability to provide complete panels of tests including many individual tests that were not necessary, but convenient, within the operation of the equipment. Inevitably, some compromises had to be made. Not all tests on one platform were accurate, so selection of equipment was often a best fit-for-purpose decision. Along with the high volumes of testing went lower reagent costs, so hospital management was pleased to fund the automated platforms now to be found in almost every hospital laboratory. An inevitable progression was the development of financially competitive contracts whereby a single manufacturer provides equipment covering a broad range of IAs to multidisciplinary laboratories. A recent development has been a move by instrument manufacturers to expand the company repertoire beyond diagnostics to include for example imaging equipment.

Huge sums of money are spent on quality assurance of assay performances yet external quality assurance (EQA) programmes have failed for decades to use a powerful evidence-base of the poor quality of assays to put pressure on assay suppliers to improve accuracy and specificity of tests. MS calibration of the EQA sample pools for steroids was promised a long time ago ⁸ but has failed to be introduced for all analytes to encourage manufacturers to make reagent kits that gave results close to the authentic targets rather than to overall or method means.

There are further issues with commercial IAs. Manufacturers do not describe the methodology in detail in order to protect their intellectual property (IP). So, for example the nature of the hapten and ligand are not stated even though there are a limited number of sites on steroids where a bridge to albumin or the label can be attached. The method for displacement of steroid from binding proteins may not be revealed to the user. Although all these steps will be familiar to users of IAs, the disclosure of more detailed analytical information by manufacturers would be helpful in investigating possible assay interference in the clinical laboratory. For the manufacturer, the greatest IP may lie in the polyclonal or monoclonal antibody used in the IA. Cross reactivity studies in most commercial IAs are merely a demonstration of where a supra-physiological level of a possible cross reactant does not give a significant result and has to assume response with concentration parallels the cortisol curve, which is unlikely. MS methods may turn out to be inaccurate in some (one hopes, few) clinical situations but is likely to arise because the need for rapid chromatographic steps does not allow separation of an interfering substance (of which for steroids there can be many isomers, epimers and metabolites) that give near common fragments in the mass spectrometer. In the method development, the reference point for the isotope dilution analysis must be evaluated in every step (especially in ion suppression and enhancement) as well as the analyte. The signal from the mass of a stable isotope labelled compound may be affected by different steroids co-eluting (dihydro- and tetrahydro-reduced metabolites are examples).

The endocrine community will to have to accept new clinically relevant cortisol cut-off concentrations for basal hormone measurement and the results following provocative or suppressive testing. Laboratories should now use MS methods and it is hoped that the cut-off concentrations will then remain stable because the specificity of the method and accuracy is at last assured.

There is an additional need for the timing of samples and doses of agent used in provocative tests to be standardized and harmonized. One particular example is the low dose (1 µg) synacthen test where the method for dilution of the Synacthen© (teracosactide) from the 250 µg ampoule must be clearly defined in the test protocol that must be adhered to in order to avoid losses by binding of the ACTH to surfaces to which it contacts. Furthermore, the peak of the cortisol response is earlier than 30 min ⁹ so bloods should be taken at 15, 20, 25 and 30 min to be sure the peak cortisol concentration is approached. This pattern was first described in 1996 yet the single 30-min result has been accepted in nearly all subsequent publications. There is a great need to standardise provocative tests in all the pituitary endocrine axes. Laboratories and clinicians need to work together to achieve harmony in endocrine tests for the benefit of the patient and to reduce wasted resources so that results are interpreted correctly and can be accurate and commutable between institutions.