What is Subclinical Cushing's Syndrome?
Cortisol has multiple target organs as illustrated in clinical practice by the spectrum of symptoms associated with chronic cortisol excess as observed in long-standing, overt Cushing's syndrome. The focus of this article is not on overt Cushing's, but rather on "subclinical" Cushing's syndrome (SCS). This could be considered an ambiguity, as why would a physician look for a silent entity that has no detectable impact on a patient's health? Today, SCS refers to autonomous cortisol production that is insufficient to generate the typical, clinically recognizable overt syndrome. However, based on studies published in the medical literature, SCS could be involved in the pathogenesis (development) of non-specific pathologies such as obesity, diabetes, and hypertension.
What populations are most likely to have subclinical Cushing's syndrome?
One might look for SCS in two main patient populations with the expectation that making this diagnosis would lead to a beneficial therapeutic intervention: 1) patients with an incidentally discovered tumor that may inappropriately secrete cortisol (adrenal gland) or ACTH (pituitary gland). Usually these lesions are called incidentalomas. 2) patients that display non-specific pathologies that may be secondary to hypercortisolism such as obesity, diabetes, hypertension, osteoporosis etc..
Subclinical" Cushing's syndrome and adrenal incidentalomas
The first report (Baierwaltes, et al.) of what we may call subclinical cortisol secreting adenomas (SCSA) described 2 patients with no clinical evidence of overt Cushing's that harboured a unilateral adrenal adenoma that showed exclusive uptake during noriodocholesterol scintigraphy (the other adrenal gland was no longer visible, implying decreased function). This suggested that the adrenal tumor produced a slight excess of cortisol that decreased the HPA activity through negative feedback. In vivo demonstration of excessive cortisol production by the tumor responsible for this scintigraphic pattern, came from: 1) catheterization studies that showed a larger cortisol output in the adrenal vein from the tumor side with suppression of the other adrenal gland, and 2) the report of scintigraphic reappearance of the other adrenal gland after ACTH infusion or several months after removal of the tumor.
How frequent are SCSA among adrenal incidentalomas ?
To date the true prevalence is unknown since different criteria have been adopted for the definition of SCSA. For example, in a study that we published in 2001, 41% of incidentalomas matched the historical scintigraphic criteria of unilateral uptake. Using a 60 nmol/L (2.16 ug/dL) threshold for impaired cortisol suppression after low doses of dexamethasone Tsagarakis et al identified 31% as being SCSA. The prevalence decreases to 12% in a German study that used a higher cortisol threshold and a modified Dexamethasone Test. The prevalence falls to around 10% in studies by Italian groups who required at least two abnormal routine tests of the HPA axis with a necessary condition of impaired dex suppression. These studies clearly indicate that no single alteration allows the diagnosis of SCSA.
Are current tests adequate to diagnose SCSA ?
The current biological tests used to diagnose overt Cushing's may be inadapted to patients with subtle hypercortisolism. For example, 24h UFC, has a very low sensitivity in this situation and is elevated in less than 15% of cases. Using a cut off of 140 nmol/L (5ug/dL), 75% of cases suppress with overnight 1 mg dex test. Another frequent problem is the interpretation of HPA axis tests when there are not in agreement with each other. Thus the biological testing for SCS seems to be more complex than that of overt Cushing's. Indeed there is a spectrum of biological activity between tumors that ranges from the non-functioning adenoma to the evident cortisol secreting tumor. Since cortisol output and autonomy is continuously distributed between these two extremes, it is probably pertinent, from a biological point of view, to consider the diagnosis of SCSA as, possible, probable or definitive according to the number and type of biochemical abnormality.
Variable or cyclic hormonogenesis is a classical feature of overt Cushing's that complicates biological diagnosis and could be an even larger problem in diagnosis of mild hypercortisolism. In a recently published Italian prospective follow-up study of adrenal incidentalomas, the authors pointed out that 28 alterations of the HPA axis observed in 23 patients at initial diagnosis were not confirmed during follow up and that only 4 alterations in 3 patients were permanent. We have similar study results that confirm these findings: biological variations of the HPA axis occur not only between patients but also frequently within patients with incidentalomas. 51 consecutive patients from 4 cities in the west of France, with unilateral adenomas characterized by CT and iodocholesterol scintigraphy, were enrolled and followed at yearly intervals for a mean duration of 4 years. In this series, 35% have possible SCSA at baseline and subtle hypercortisolism is confirmed at each follow-up evaluation. 18% of patients classified initially has having non-secreting adenomas display an evolution towards the disease that is confirmed in subsequent evaluations, while up to 20% of patients fluctuate over time between the two conditions (having SCSA or having a non-secreting adenoma). Thus, one take-home lesson could be "take great care in the interpretation of a single set of biological investigations and sample twice before diagnosing biochemical evolution".
Clearly there is no easy way to diagnose SCSA but we need a simple and reliable test useful for screening purposes in order to select individuals for complementary investigations. Ideally this test will have a high negative predictive value when normal. In 2003 the NIH consensus panel on the management of Adrenal Incidentalomas recommended to look for impaired suppression to 1 mg Dex for this purpose. However, there was no real consensus about the definition of adequate cortisol suppression. Some experts recommended the use of the traditional 140 nmol/L (5ug/dL) threshold for post dex cortisol while others proposed further testing of individuals with serum cortisol values between 50 and 140 nmol/L (1.8-5ug/dL) to increase the detection of subclinical hypercortisolism. Several arguments are in favour of using a low threshold. Indeed, many groups have shown that using modern immunoassays, cortisol levels are low following a 1 mg overnight dex suppression test. In our study of a cohort of 70 hospitalized obese subjects, post dex cortisol levels were below the limit of detection of the assay in more than 50% of cases and below 50 nmol/L (1.8ug/dL) in 94% of cases. Also, in a previous study, we found excellent correlation between the 1 mg overnight dex test and presumably more robust 4mg IV test. Thus, assuming that patients are not in acute stressful conditions and do not take drugs known to interfere with dex metabolism, we recommend the use of the 50 nmol/L threshold to screen for SCSA in order to improve the sensitivity of the screening (e.g. to identify all putative cases). Since this threshold might generate up to 15 % of false positives, further testing of patients with a cortisol measurement greater that 50nmol/L is mandatory.
Why should subtle hypercortisolism be diagnosed?
Is there evidence that these biochemical abnormalities significantly affect patient's health? What has been most studied to date are the metabolic consequences of subtle hypercortisolism. Numerous studies have shown that cortisol increases hepatic gluconeogenesis, opposes the action of insulin resulting in insulin resistance, promotes hypertension and abdominal fat distribution. Altogether these actions result in the full spectrum of metabolic X syndrome in overt Cushing's and its resultant increase in cardiovascular mortality and morbidity. A number of studies have shown a remarkably high prevalence of obesity, hypertension and impaired glucose metabolism among patients with adrenal incidentalomas. Similar findings were observed in studies dealing more specifically with SCSA. Interestingly, the prevalence of diabetes and hypertension was greater in patients with SCSA than in patients with putatively non-secreting incidentalomas. A study published by Terzolo et al compared non-obese patients with SCSA to patients with non-secreting tumors with equivalent BMI (body mass index). SCSA was associated with increased glycemia 2h after a Glucose Tolerance Test, increased triglyceride levels and a reduced index of insulin sensitivity. Although there was probably important variability among patients, this study suggests that subtle hypercortisolism is specifically associated with an adverse metabolic profile. Such findings were confirmed in another study that also documented the impact of biochemical abnormalities on the vascular system. Compared to controls matched for age and BMI, SCSA patients had accelerated atherosclerosis and frequent impaired cardiac function.
The ultimate justification in looking for subclinical Cushing's will be the demonstration that removal of the tumor reverses end-organ complications. While we are far from that point, a few preliminary studies, involving a limited number of patients, suggest that adrenalectomy reduces the cardiovascular risk. It is also interesting to note that parameters such as insulin sensitivity and blood pressure were also improved after surgery in some patients diagnosed as having non-secreting tumors. Such findings illustrate the lack of close correlation at the individual level between biochemical tests of subtle hypercortisolism and end-organ complications.
Subclinical Cushing's syndrome and pituitary incidentalomas
Very little data is available for the frequency of autonomous ACTH hypersecretion among pituitary incidentalomas. In tests performed at autopsy, the prevalence of ACTH immunostaining among pituitary adenomas is at most 8%. As presented later in this article, Cushing's disease is found in approximately 1.5% of patients with metabolic syndrome X. Thus, to date it is probably wise to screen for subclinical Cushing's in this situation when patients present with features of metabolic syndrome.
In conclusion of the first part of this article, what we clearly need from now on is: 1) to identify variables that correlate with end-organ complications and that may be usable at the individual level. Due to the huge variability in the tissular impact of hypercortisolism, this might be a holy grail! 2) to demonstrate using control studies the short and long-term benefit of surgery in SCSA, however the design of such studies is complicated. 3) To set strategies of monitoring for patients who are not operated on. In light of present knowledge, it is probably wise not only to focus on tumor size and endocrine biology but also to keep an eye on metabolic parameters and the cardiovascular system of patients.
Looking for subclinical Cushing's in other patient populations
To date, the most studied patient population evaluated for SCS, is the diabetic population. Several studies have reported that a significant portion of type 2 diabetics, actually have SCS. Several years ago, we conducted a prospective study in 200 overweight type-2 diabetic patients consecutively referred to our unit for poor metabolic control. None of the patients had overt CS after careful clinical examination. A first screening step was performed with the 1-mg overnight dex test using a low threshold for cortisol suppression to maximize the sensitivity of the procedure. Fifty-two patients had impaired cortisol suppression. Among these, 47 had further evaluation of the HPA axis. Thirty patients (i.e. 15% of the whole series) had normal investigations and were considered as false positives of the 1-mg DST. On the contrary, 17 displayed at least one additional abnormal test. Finally 14 of these underwent a third step of imaging investigations that allowed identification of pituitary dependent Cushing's disease in three patients, a unilateral adrenal tumor showing prevalent or unilateral uptake at iodocholesterol scintigraphy in 8 patients, while no tumor was found in 3 patients. Overall and considering only patients with obvious tumors, we found a 5.5% prevalence of subclinical Cushings.
Given the presumed rarity of Cushing's, we were surprised by these findings and wondered if this was an epidemic limited to the Bordeaux area of France. However, a recent Italian publication (Chiodini, et al.) replicated and improved our study and the higher than anticipated prevalence is not limited to France. The Italian authors studied not only diabetics patients (P) but also control subjects (C) matched for age and BMI (thus most of the controls were obese but did not have type 2 diabetes). Using the same screening strategy as our study, 17% of P had on two occasions impaired 1 mg dex suppression. During the second step of detailed biological investigations, 35 P (12%) and 4 C (2%) had additional markers of impaired HPA axis function and underwent imaging studies. Finally, an adrenal mass was found in 21 P and 3 C, a pituitary microadenoma in 4 P and 1 C, a pulmonary mass in 2 P. Again, no tumor was found in 3 patients. Overall, an impressive prevalence of SCS was found in this study; 9% of P and 2% of C. Since a minority of patients underwent surgery, definitive diagnosis of cortisol or ACTH secretion by tumors is lacking so that the true prevalence of SCC remains to be more precisely determined. However, these studies clearly suggest that type 2 diabetics probably represent a high-risk patient population for SCS. Also, given that the BMI matched controls showed an increased prevalence of 2% compared to the generally accepted incidence of overt Cushing's, perhaps appropriate testing of the obese population should be considered.
The last important point I want to mention is the serious lack of controlled interventional studies to prove the causality of SCS on impaired glucose metabolism e.g. the reversal of metabolic abnormalities after cure of SCS.
In conclusion, we definitely need studies in putative high-risk patients for SCS such as women with PCOS and young patients with osteoporosis. An Italian study presented at the recent Endo 2006 meeting presented data that indicated an elevated prevalence of subclinical Cushing's in a study among osteoporotic patients. Finally interventional studies are urgently needed to confirm the suspected causal links between diabetes and the adrenal tumors found in diabetic patients. Such studies may provide a rationale for a systematic screening of SCS in selected groups of patients.
Editor's Note: Dr. Antoine Tabarin is head of the Endocrinology Department at University Hospital, Bordeaux, France and teaches endocrinology at Bordeaux 2 University. Dr Tabarin is also adjunct associate professor in the department of Neuropharmacology at Scripps Research Institute (La Jolla, CA). Dr. Tabarin has treated Cushing's patients for over 20 years and has published extensively in the area. Dr. Tabarin can be reached at firstname.lastname@example.org