A rare cause of Cushing's syndrome demonstrates analytical discrepancies between the Roche E170 and Bayer Centaur testosterone assays

Abstract

Abstract

The initial biochemical investigations of a female patient with suspected Cushing's syndrome revealed abnormal endocrine results, including a marked elevation of serum testosterone. Overnight and low-dose dexamethasone suppression tests confirmed the diagnosis of Cushing's syndrome. Imaging investigations revealed an appearance compatible with adrenocortical carcinoma with metastases in the lungs and liver. This tumour is a rare cause of Cushing's syndrome. Two different automated testosterone immunoassays were used during the investigation of this patient, and analytical discrepancies in the patient's testosterone results were found. The two assays used, as well as potential causes of the difference in results will be discussed.

Case history

A 40-year-old female patient with diabetes mellitus was admitted to hospital feeling weak and lethargic with erratic glycaemic control and intermittent blurred vision. The patient had a five-year history of hypertension and had been diagnosed with type 2 diabetes five months previously. On examination, the patient was noted to have Cushingoid features and she was referred to an endocrinologist for further investigation. Due to her suboptimal glycaemic and hypertensive control since diagnosis she was also referred to the local diabetes centre. The patient was on a range of medication including gliclazide (160 mg b.i.d. [twice daily]), metformin (1 g b.i.d.), valsartan (80 mg b.i.d.), atenolol (50 mg o.d. [once daily]), Natrilix MR (1.5 mg o.d.) and amlodipine (5 mg o.d.).

On further examination the patient was found to have a plethoric round face, telangiectasia, excess hair growth in a beard distribution and over the back, central adiposity with peripheral wasting of muscle, purple striae over the abdomen and a buffalo hump. She had a body mass index of 29 and an elevated blood pressure of 150/99. An abdominal examination revealed a 3 cm hepatomegaly. The patient had a 2–3-year history of redness over the cheeks, and over the previous six months she had experienced increasing lethargy and difficulty getting up from a chair. The patient's husband noted that she had a short temper, poor attention, mood swings and increasing weight for the previous 12 months.

Serum urea and electrolytes were normal with liver function tests showing slightly elevated alanine transferase (ALT) (48 U/L; reference range 0–31 U/L) and an elevated γ-glutamyltransferase (GGT) (191 U/L; reference range <64 U/L). Endocrine investigations revealed an elevated serum cortisol (895 nmol/L) on a mid-day sample and a markedly elevated testosterone (18.8 nmol/L; reference range 0–3 nmol/L). The high serum testosterone result with a low sex hormone binding globulin (SHBG) (10 nmol/L, reference range 20–120 nmol/L) gave rise to an abnormal testosterone:SHBG ratio (1.88; reference range 0–0.05). Thyroid function tests, prolactin, oestradiol and gonadotrophins were normal. An overnight (1 mg) dexamethasone suppression test was performed, which failed to suppress the serum cortisol (09:00 h cortisol 863 nmol/L). A two-day low-dose dexamethasone suppression test was subsequently performed, which also failed to suppress the serum cortisol (09:00 h cortisol 839 nmol/L).

A computerized tomography (CT) scan of the abdomen, pelvis and chest revealed a large 8 × 6 cm irregular mass arising from the left adrenal gland, along with focal mass lesions throughout both lobes of the liver and both lungs. The rounded pulmonary lesions were of varying sizes ranging from a few mm to over 1 cm in size with an increased number at the bases of the lungs, indicative of metastases. A working diagnosis of adrenocortical carcinoma with liver and lung metastases was made, which was confirmed following an ultrasound guided biopsy of the liver lesions. Appearance of the liver tissue was in keeping with metastatic adrenocortical carcinoma and the immunohistochemical analysis confirmed the diagnosis. Tumour cells were inhibin positive and were negative for Cam 5.2, AE1/AE3 (for renal cell carcinoma) and hepatocyte markers.

Additional biochemical investigations revealed 24 h urinary metadrenaline and normetadrenaline within the reference ranges. Serum adrenocorticotrophic hormone (ACTH) was suppressed (<5 ng/L, reference range for 09:00 h sample 7–51 ng/L), whilst urine free cortisol (1163 nmol/24 h, reference range < 300 nmol/24 h) and serum dehydroepiandrosterone sulphate (DHEA-S) (> 27 µmol/L, reference range 1.7–9.2 µmol/L) were elevated.

Due to the large size of the adrenal tumour and the extent of the metastases, surgery was not a curative option, and medical therapies were commenced. The patient was treated with metyrapone (1 g q.d.s. [four times daily]) and dexamethasone (0.5 mg b.i.d.) as a block and replace regimen, in an attempt to relieve the symptoms attributed to cortisol excess. Two months following diagnosis the patient developed leg weakness and pain which resulted in difficulty mobilizing. A bone and magnetic resonance imaging (MRI) scan revealed metastases to the lumbar spine, pelvis and femoral heads. A subsequent fracture to the left neck of femur was treated with a total hip replacement four months after initial diagnosis. The patient is currently receiving palliative care in the form of analgesia and radiotherapy for relief of bone pain.

Adrenocortical carcinoma

Adrenocortical carcinoma is a rare malignancy (incidence 1–2 per 1 million population), which presents with adrenal steroid hormone excess in 60% of cases. ¹ Androgen-secreting adrenal tumours in women induce hirsutism, virilization and oligomenorrhoea, while oestrogen-secreting tumours in men lead to gynaecomastia and testicular atrophy. Presentation exhibits a bimodal age distribution with a peak in childhood and a second higher peak in the fourth and fifth decade.

Tumour staging is dependent on the size of the primary tumour and whether there is local invasion or distant metastases. Therapeutic options include surgery, radiation and medical therapy which are tailored according to the severity of disease. At diagnosis, as in this case, most patients have severe disease, ² which leads to the poor prognosis associated with this form of cancer. Diagnosis of stage IV metastatic disease leads to a median survival of less than 12 months. ¹

Analytical discrepancies in testosterone results

During the investigation of this patient, serum testosterone was analysed locally using a Roche Diagnostics Modular E170 unit. The testosterone concentrations measured on this patient ranged between 15 and 20 nmol/L. Female testosterone results greater than 3.5 nmol/L are routinely sent to another laboratory (University Hospital of Wales, Cardiff), where they are analysed on the Bayer Centaur prior to and following ether extraction. There was a large difference in testosterone results obtained by the Roche E170 (17 nmol/L) and the Bayer Centaur analyser (8.2 nmol/L prior to and 7.3 nmol/L following ether extraction, respectively). The similarity in the results before and after ether extraction using the Bayer assay implied that there was significant interference with measurement of testosterone in this serum sample using the Roche assay. It is well recognized that ether extraction does not remove all interfering substances, ³ and so the Bayer testosterone result itself is not necessarily ‘accurate’. Studies have previously demonstrated that many direct immunoassays including the Roche Elecsys 2010 assay, have a positive bias compared with the reference method for testosterone, i.e. isotope-dilution gas chromatography mass spectrometry (ID-GCMS). ⁴

The Roche and Bayer assays are both competitive immunoassays using electrochemiluminescence and direct chemiluminescent technology with comparable measuring ranges of 0.07–52.0 nmol/L and 0.35–52.1 nmol/L, respectively. However, there are marked differences in both the assay architecture and the antibodies used in the two assays. First, the Roche assay employs a biotinylated monoclonal antitestosterone antibody, whereas the Bayer assay uses a polyclonal antitestosterone antibody. Secondly, the assays differ in the timing of the release of testosterone from its binding proteins. The Roche assay involves patient serum, the labelled monoclonal antitestosterone antibody and the competing labelled testosterone being combined in the first step. The labelled testosterone contains the releasing agent which causes displacement of the patient testosterone from its binding proteins. The Bayer assay initially involves the addition of a releasing agent to the patient serum, before any other components of the immunoassay. Generally, monoclonal antibodies are regarded as more specific but in this case it was the Roche monoclonal assay that was susceptible to interference. It is speculated that the different assay architecture and possibly the different affinities and avidities of the antibodies used in the two assays contribute to the Roche assay being more susceptible to interference.

The differences in the stated cross-reactivities of the two assays are provided in Table 1. The most prominent difference in cross-reactivities is that for 5-α-dihydrotestosterone, where cross reactivity for the Bayer assay is approximately three-fold greater than that for the Roche assay. However, the relative serum concentrations of testosterone and 5-α-dihydrotestosterone would make this cross-reactivity clinically insignificant.

Table 1

Stated analytical specificities for the Roche and Bayer assays

Compound	Cross-reactivity with testosterone assay (%)
	Roche E170 assay	Bayer Centaur assay
Androstenedione	0.91%	0.94%
Danazol	Not detected	<0.1%
Dehydroepiandrosterone sulphate	0.01%	<0.1%
Oestradiol	Not detected	0.02%
5-α-dihydrotestosterone	1.89%	5.4%

Despite the stated low cross-reactivities of the Roche E170 testosterone assay with other steroid hormones, the fact that some of these steroid hormones are normally found in the serum in markedly higher concentrations than testosterone may contribute to analytical interference. The stated cross-reactivity of the Roche testosterone assay with DHEA-S is low at 0.01%. However, 1000 nmol/L of DHEA-S equates to 0.1 nmol/L of ‘pseudo’ testosterone (Roche correspondence), and therefore 10,000 nmol/L of DHEA-S (i.e. 10 µmol/L) equates to 1 nmol/L of ‘pseudo’ testosterone. The upper limit of the quoted female DHEA-S reference range is 9.2 µmol/L, which means that it may normally contribute up to 0.9 nmol/L of ‘pseudo’ testosterone. In this case report the patient had an initial DHEA-S concentration of greater than 27 µmol/L (the upper limit of the analytical range). The exact concentration was not elucidated by dilution. If DHEA-S was solely responsible for the interference in the Roche assay, the 10 nmol/L difference in measured testosterone between the Roche and Bayer assays could have resulted from of a DHEA-S concentration in the region of 100 µmol/L. The Bayer testosterone results prior to and following ether extraction indicate that the high concentrations of DHEA-S in this patient's serum appeared not to have caused a problem in the Bayer assay.

This degree of cross-reactivity in the Roche assay could potentially be clinically misleading. DHEA-S within the reference range in a female patient's serum could contribute ‘pseudo’ testosterone and shift the patient's testosterone result above the reference range. This could potentially lead to further laboratory investigations, including ether extractions, or clinical investigations where malignancy may be wrongly suspected. In this case, the falsely elevated testosterone results did not lead to any harm to the patient. Both the Bayer and Roche testosterone results were in the region where malignancy would be suspected in a female patient. The cross-reactivity of DHEA-S in the Roche testosterone assay makes this assay unsuitable for investigation of patients with suspected adrenal carcinomas, where the DHEA-S is usually elevated. ⁵

The exact cause of interference in the Roche assay in this case report has not been experimentally proven, and it is only postulated that DHEA-S is the cause of the interference. Interference from endogenous steroid hormones as well as exogenous sources is a well-recognized problem associated with automated direct immunoassays for testosterone. Recently, it has been reported that significant cross-reactivity with DHEA-S was the cause of interference in the Abbott Architect assay. ⁶ In addition, the Roche Elecsys, Roche E170 Modular and Beckman Access/DxI assays have been shown to be susceptible to intereference to a similar extent, whilst minimal effects were observed for the Bayer Advia Centaur. ⁷ Consistent with this finding, the discrepant testosterone results reported in this case were higher when assayed using the Roche Modular E170 assay. Mass spectrometric methods such as ID-GCMS are now considered reference methods for testosterone measurement, and simpler liquid chromatography–tandem mass spectrometry (LC–MS/MS) methods are becoming routine practice in some clinical laboratories. ⁶ The improved specificity of these methods will lead to more reliable and meaningful testosterone results on female patients. This case illustrates that potentially misleading serum testosterone results can be reported.