What is the cause of Cushing's disease?

Exactly 60 years ago, Harvey Cushing published the first full description of his eponymous disease, associating the clinical features with a basophil adenoma of the pituitary, although a similar case had in fact been described by John Anderson in Glasgow in 1915 (Medvei, 1991). However, a putative hypothalamic cause for this disorder had previously been suggested by a Russian physician, Itsenko (quoted by Reschini & Catania, 1991), and the dispute as to possible aetiology has continued ever since. It is clear that the great majority of clinical symptoms and signs in Cushing’s disease are the consequence of the excess circulating corticosteroids, with the hypersecretion of POMC-related peptides rarely being of direct importance. There are additionally a constellation of biochemicaI neuroendocrine abnormalities (such as attenuated TSH responses to TRH, hypogonadotrophic hypogonadism and poor GH responses to a variety of stimuli) which demonstrate a derangement of neuroendocrine function and, in the absence of a major space occupying lesion, have been interpreted as evidence in favour of primary hypothalamic pathology (Krieger, 1983). However, it has become increasingly clear that most, if not all, of these abnormalities resolve with normalization of serum cortisol, although this may take a considerable period of time. The demonstration of the presence of a pituitary adenoma in the great majority of such patients is further direct evidence for the pituitary as a principle locus of the disease, as is the lack of recurrence seen in patients initially ‘cured’, that is, with undetectable serum cortisol. In contrast, most authorities find corticotroph hyperplasia to be extremely rare; occasional cases of cure by hypophysectomy without evidence of any abnormal histological tissue may be attributed to inadvertent loss of a miniscule tumour at operation. Indeed, the mean size of such tumours has been estimated to be approximately 5-6 mm in diameter. Finally, if hypothalamic overdrive by a corticotrophin releasing factor such as CRH-41 were important in the pathogenesis of Cushing’s disease, it is strange that severe depressive illness is so infrequently followed by this disorder. In fact, continuous infusion of CRH-41 with or without metyrapone leads to a resetting of the pituitary-adrenal axis which is uniformly elevated (Schulte et al., 1985; Ur et al., 1991), similar to the resetting seen in depression (Pfohl et al., 1985), and quite unlike the flattened circadian rhythm characteristic of Cushing’s disease. In Cushing’s disease CSF levels of CRH-41 are suppressed, as opposed to circulating levels which are unlikely to reflect hypothalamic activity. Nevertheless, there are discordant reports as to pituitary pathophysiology (Burke er al., 1990), and the possibility of Cushing’s disease being secondary to hypothalamic dysfunction continues to be entertained. In this issue, Stewart and colleagues from the Queen Elizabeth Hospital, Birmingham, have employed a battery of contemporary techniques (ACTH measurement by 2-site IRMA, rapid sampling and a sophisticated pulse analysis algorithm) to assess ACTH/ cortisol pulse parameters in Cushing’s disease. Their elegant analysis demonstrates an increase in ACTH pulse amplitude in this condition, compatible with a pituitary defect. In addition, there is also a consistent loss of the normal nocturnal deceleration in ACTH pulse frequency, from which they infer that the hypothalamic corticotroph releasing factors may be abnormally regulated. The data certainly confirm the absence of the circadian rhythm characteristic of Cushing’s disease, and neatly demonstrate pulse parameters in normal subjects. However, postulation of a pituitary defect in pulse amplitude and a hypothalamic abnormality in pulse generation sits uneasily on Ockham’s razor, and rests on the supposition that pituitary pulse frequency is driven by a hypothalamic pulse generator. In fact, pulsatility is a characteristic of a variety of biological systems: even intracellular calcium fluxes show rhythmic oscillations (Berridge, 1988). The secretion of several pituitary hormones such as prolactin certainly show endogenous rhythms (Samuels et al., 1991) that can be modulated by, but are not contingent upon, hypothalamic pulse generators. Thus, an alternative interpretation is that normal subjects show hypothalamic entrainment of an endogenous pituitary rhythm, and that this is absent in the autonomous adenoma in Cushing’s disease. It is possible that the late afternoon deceleration in ACTH pulse frequency seen in normal subjects is due to activation of a hypothalamic inhibitory factor (ANP?; Fink et al., 1991), but more probable that normal corticotrophs demonstrate a relatively slow endogenous rhythm which may be accelerated by CRH-41 or vasopressin, and which may be inherently faster in tumorous corticotrophs. The fundamental nature of the defect in Cushing’s disease requires further detailed studies, probably using molecular biological techniques. For example, Herman et al. (1990) have shown monoclonality in the great majority, if not all, of basophil adenomas. A more recent report by Schulte et al. (1 99 1) found monoclonality in, approximately, only twothirds of such tumours, but sampling problems in these small