TITLE PAGE - European Journal of Endocrinology

Page 1 of 33 Accepted Preprint first posted on 16 March 2009 as Manuscript EJE-09-0046 TITLE PAGE THE PREVALENCE AND CHARACTERISTIC FEATURES OF CYCLICITY AND VARIABILITY IN CUSHING’S DISEASE Krystallenia I. Alexandraki1,2, Gregory A. Kaltsas1,2, Andrea M. Isidori1, Scott A. Akker1, William M. Drake1, Shern L. Chew1, John P. Monson1, G. Michael Besser1 and Ashley B. Grossman1. 1 Department of Endocrinology, St Bartholomew's Hospital, London, United Kingdom and 2Division of Endocrinology, Department of Pathophysiology, Laiko General Hospital, School of Medicine, National & Kapodistrian University of Athens, Athens, Greece. Correspondence to: Prof. Ashley Grossman, Dept. of Endocrinology, St. Bartholomew’s Hospital, London EC1A 7BE, UK, Tel: +44-207-6018343, Fax: +44207-6018505, Email: a.b.grossman@qmul.ac.uk Running Title: Cyclicity in Cushing’s disease Word count: 2910 1 Copyright © 2009 European Society of Endocrinology. Page 2 of 33 ABSTRACT Objective: Cyclical Cushing’s syndrome may render the diagnosis and management of Cushing’s disease difficult. The aim of the present study was to investigate the prevalence of cyclicity and variability in patients with Cushing’s disease, and to identify putative distinctive features. Design: Retrospective case-note study. Methods: We analysed the case-records of 201 patients with Cushing’s disease in a retrospective case-note study. Cyclicity was considered as the presence of at least one cycle, defined as a clinical and/or biochemical hypercortisolaemic peak followed by clinical and biochemical remission, and by a new clinical and/or biochemical hypercortisolaemic peak. The fluctuations of mean serum cortisol levels, as assessed by a 5-point cortisol day curve, defined the variability. Results: Thirty (14.9%; 26 females) patients had evidence of cyclicity/variability. ‘Cycling’ patients were older but no difference in sex or paediatric distribution was revealed between ‘cycling’ and ‘non-cycling’ patients. The median number of cycles was 2 for each patient, and 4 years was the median intercyclic period. A trend to lower cure rate post-neurosurgery and lower adenoma identification was observed in ‘cycling’ compared to ‘non-cycling’ patients. In multivariate analysis, older patients, longer follow-up, female sex, and no histological identification of the adenoma were associated with an increased risk of cyclic disease. Conclusions: This large population study reveals that cyclicity/variability is not an infrequent phenomenon in patients with Cushing’s disease, with a minimum prevalence of 15%. Physicians should be alert since it can lead to frequent problems in diagnosis and management, and no specific features can be used as markers. Key words: Cushing’s disease - cyclicity - variability - prevalence 2 Page 3 of 33 INTRODUCTION Cushings syndrome (CS) due to excess endogenous cortisol production has been occasionally described as cyclical (1) but with variable other descriptions such as periodic hormonogenesis (2), unpredictable hypersecretion of cortisol (3), fluctuating or fluctuability steroid excretion (4,5) or intermittent Cushing’s syndrome (6). It has been more frequently described in pituitary-dependent Cushing’s syndrome, Cushing’s disease, having being characterised by periodic increases and decreases in cortisol levels, clinically and/or biochemically documented (2,7, 8). However, it has been considered to be an uncommon phenomenon, since only a few small series and several case reports confirm its presence (9,10). It is important to consider that acute investigation may be associated with episodes of lowered activity and delays in definitive diagnosis. Similarly, apparent ‘cure’ after surgical treatment may simply reflect a nadir in activity and be falsely reassuring. Furthermore, fluctuating disease activity may render medical control extremely difficult. Intermittent as well as sustained cortisol hypersecretion might be associated with considerable morbidity and mortality, necessitating effective definitive treatment (11,12). Although most experienced endocrine physicians recognise this phenomenon, it has proven difficult to define robustly. Cyclicity has been commonly defined as the occurrence of three peaks and two troughs of hypercortisolaemia (9), but this cannot be readily applied to all patients since the intercyclic period may vary and may be so long as to defy a clear assessment. Cyclicity has also been reported in patients with adrenal tumours (13) and ectopic ACTH syndrome (14), but it is most described in patients with Cushing’s disease. Due to the lack of any comprehensive data on the prevalence of the phenomenon, the aim of the present study was to investigate the presence of 3 Page 4 of 33 ‘cyclicity’ (defined as the presence of clinical and biochemical peaks and troughs) and of ‘variability’ (biochemical fluctuations of cortisol secretion) in a large cohort of patients with Cushing’s disease. In addition, some characteristics of this discrete patient population have been investigated. 4 Page 5 of 33 MATERIALS AND METHODS We analysed retrospectively the data in 201 patients with Cushing’s disease, consecutively admitted from 1946 through to 2007 to the Department of Endocrinology, St Bartholomew's Hospital London, with the authorisation of our institutional case-note review committee (registration number 08/76). The diagnosis of Cushing’s disease was based on the presence of typical symptoms and signs based on a fixed departmental protocol, along with detailed biochemical evaluation, including circadian rhythm studies, low- and high-dose dexamethasone suppression tests (LDDST, HDDST), the corticotrophin releasing hormone (CRH) stimulation test, and bilateral inferior petrosal sinus sampling (BIPSS) with CRH (a routine procedure in our Department from 1985). Confirmation of Cushing’s disease was based on the histopathological diagnosis of a corticotroph tumour and/or clinical and biochemical remission following trans-sphenoidal surgery (TTS), and/or biochemical confirmation of ACTH-dependent Cushing’s syndrome with a central gradient on BIPSS (15). For the present study the following demographic data were extracted from the case records of each patient: gender, age at presentation, symptoms and signs of hypercortisolism, years of follow-up, diagnostic work-up, therapeutic modalities, pituitary imaging studies and pathology results. A cycle was defined a clear-cut clinical and/or biochemical hypercortisolaemic peak followed by a clinical and biochemical remission and at least by a new clear-cut clinical and/or biochemical hypercortisolaemic state. Cyclicity was considered to be the presence of at least one cycle. The clinical peak was documented by the appearance and disappearance of specific Cushingoid features, according to the detailed clinical records. This assessment was based on the registration of changes 5 Page 6 of 33 (improvement or exacerbation) observed using a clinical assessment system with the advantage of collecting information methodically from a single institution on the basis of a consistent protocol in which specific symptoms/ signs were assigned as either present or absent. This to some extent eliminates individual variations of the interviewing physicians in data accrual (16,17), but is similar to many lists extant in the literature (15); this list of features is included in a table as part of the results section. The biochemical peak, characterised by the biochemical confirmation of hypercortisolaemia as previously described (15), was assessed using the cortisol daycurve (CDC, involving measuring serum cortisol levels at 5 fixed times through a day) (18). This assessment was consistently performed in our department, and is based on a normal range established in a normal control group and supported by an isotopic dilution technique (18). Each cycle was then further characterised as clinical, biochemical, or both. When the hypercortisolaemic state was not present as a cycle but as fluctuations of mean serum cortisol levels, as assessed by a 5-point CDC, it was defined as variability. Besides the presence of cortisol fluctuations, it was arbitrarily decided to confirm variability by the presence of doubling (mean level increase >200% over basal values) or halving (mean level decrease <50% from basal) of the serum cortisol during CDC performance, in the absence of therapy or without a change in therapy in order to minimise non-physiological changes due to possible assay variability error (9,19-21). As an example of this calculation, we found one patient who first presented to her general practitioner with clinical Cushing’s syndrome which then remitted; 3.5 years later her clinical features recurred and she had increased urine free cortisol (UFC); two months later her UFC was normal and she suppressed on a LDDST). A further two months later her UFC was again increased and she was admitted for diagnostic work-up. She was administered mitotane and the dose remained stable with a 6 Page 7 of 33 CDC between 363 and 423; however, some 6 years from her previous hypercortisolemic peak on a stable dose of mitotane for 4 years she presented with a CDC 700 nmol/L (Figure 1). A second patient was not cured by surgery and was treated medically while awaiting the delayed effects of radiotherapy; however, during this period she was difficult to control medically because of variation in her CDC, and after 6 years she finally had bilateral adrenalectomy. As shown in Figure 2, a hypercortisolaemic peak was present in November 1996 and another in August 1997 on unchanged therapy. Two subsequent cycles on unchanged medication revealed peaks in September 1997, September 1998 and December 1999. Finally, as an example of cyclicity before the admission for the confirmation of Cushing’s syndrome, a patient presented with clinically florid Cushing’s syndrome to her general practitioner which remitted in two weeks; two years later another episode of clinically florid Cushing’s syndrome lasted two months. The third peak presented 27 months later when the patient was fully investigated and cured by surgery and radiotherapy. Cyclicity/ variability was recorded if present: 1) prior to the first presentation of the investigation of Cushing’s syndrome and consequently before any therapeutic intervention, based on clinical ground or biochemical data provided by the previous medical history, or 2) after the confirmation of the diagnosis and following any form of acute or chronic therapeutic modality. The length of cycle was determined as the duration from one apparent peak of clinical or biochemical Cushing’s syndrome activity until the following peak. The case records of all patients were carefully evaluated according departmental protocols. Cases with simple progression from a mild form to the overt disease, the on-going effects of chronic treatment such as radiotherapy or mitotane withdrawal, the self-changing treatment doses or non-compliance with medication, significant alcohol consumption, or significant diet-induced weight or incomplete data recording were excluding criteria from the definition of cyclicity or variability. 7 Page 8 of 33 During the follow-up period clinical parameters and cortisol secretion were assessed at least at 3-monthly intervals for clinical purposes, or more frequently, usually monthly, when required, using the mean cortisol levels through the day (CDC). Follow-up assessment was performed yearly after 3 years of confirmed remission. In order to perform a uniform and consistent data analysis, other hormonal data were not analysed since only the CDC was formally assessed in the whole population studied. While the possibility of ectopic Cushing’s syndrome was considered during follow-up in all the cases of Cushing’s disease with negative histology, no case of ectopic disease was identified (15). No case of adrenal insufficiency was recorded during the periods of remission in any of the patients studied. Post-operative cure was defined by biochemical and clinical remission of Cushing’s syndrome. The paediatric population was considered to include any patient with Cushing’s disease less than 18 years old. Statistical Analysis Values are presented as mean value ± standard error (±S.E.). Statistical significance in the results was accepted at a P-value <0.05. The normal distribution of continuous variables was assessed by applying the non-parametric KolmogorovSmirnov test. Comparisons between groups were made by an independent-sample ttest. The Mann–Whitney U test was performed for variables which were not normally distributed. Correlations between categorical variables were estimated by the chisquare test or with Fisher’s Exact Test when appropriate. Odds Ratio (OR) and 95% Confidence Interval (CI) for binary outcomes in univariate and multiple logistic 8 Page 9 of 33 models were calculated and reported. Cyclic disease presence was the dependent variable. Analysis was performed using SPSS (version 11.01; SPSS, Inc., Chicago, IL, USA) for Windows XP (Microsoft Corp.). 9 Page 10 of 33 RESULTS The whole group of the 201 patients (154 women, 76.6%) initially presented at a mean age of 37.6±1.15 (range: 7-95) years and were followed-up for a mean period of 14.1±0.7 (range: from presentation-52) years; 8.5% (17/201) of the total group were less than 18 years old at presentation. Nine patients of the total cohort were studied only at presentation. Evidence of cyclicity or variability was observed in 30 patients, of which 26 were female. These patients had a mean age of 42.3±2.0 years at presentation and were followed-up for a mean period 14.8±1.6 (range: from presentation-30) years. The ‘cycling’ did not differ from the ‘non-cycling’ group in sex ratio or years of follow-up, but did differ in age at presentation. Cyclicity/variability was present in 16.88% (26/154) of females and in 8.52% (4/47) of males; 1/17 (6.25%) of the pediatric patients showed cyclicity (Tables 1 and 2). In 27 (90%) of these patients there was recorded at least one cycle (median: 2; range: 1-4) and variability in 12 (40%). In a total of 54 recorded cycles, 8 were only clinical, 39 clinical and biochemical, 7 biochemical alone. The mean cycle length was 3.8±0.6 years. Only in one patient was a stable intra-cyclic period every 1.5 years in two cycles observed. Evidence of cyclic disease before diagnosis was present in 12 patients, before diagnosis and during follow-up in 9 patients, and during follow-up in 9 patients. Consequently, in the whole ‘cycling’ subgroup, evidence of cyclic disease before diagnosis was present in 21 (70%) patients, and in the follow-up in 18 (60%) (Table 1). In the 19 patients with cyclicity/ variability submitted to trans-sphenoidal surgery, 6 were cured (post-operative serum cortisol at 09.00h <50nmol/L), while 4 had remission with recurrence. Of the cured patients only two had post-operative 10 Page 11 of 33 serum cortisol levels <50nmol/L in the immediate post-operative period, while 4 patients had higher levels which fell gradually over 3-7days. In one ‘not-cured’ patient, radiotherapy was performed immediately after the documentation of the detectable post-operative serum cortisol levels, confounding further interpretation. A trend for lower cure rates and a lower rate of pituitary adenoma identification in histology was found in the patients showing cyclicity. The presence of a macroadenoma (2 in ‘cycling’ and 28 in the ‘non-cycling’ patients) did not differ between the groups. Eight of the ‘cyclic’ patients required bilateral adrenalectomy compared to 47 ‘non-cycling’ patients; the median follow-up period in those subgroups was 23 years (range: 9-30 years) and 20.5 years (range:0-52 years) respectively; Nelson’s syndrome was diagnosed in two (of 8) ‘cyclic’ patients and one (of 47) ‘non-cyclic’. Two ‘cyclic’ patients were controlled with ‘block-andreplacement’ modalities but no patients in the ‘non-cycling’ group required this treatment. Interestingly, imaging studies suggested or confirmed the presence of an adenoma (including macroadenomas) at the first presentation in 19 ‘cycling’ patients and 156 ‘non-cycling’ (Table 2). In 4 (13.3%) patients we noted a clear paradoxical rise in serum cortisol level following either the low- or high-dose dexamethasone suppression tests. Variability in the signs and symptoms of hypercortisolism are reported in Table 3. Only gastrointestinal symptoms and galactorrhoea in the females appeared to differ between ‘cyclic’ and ‘non-cyclic’ patients, but after allowance for multiple comparisons this was no longer statistically significant. Older patients were found more likely to have a cyclic disease in univariate analysis. In multivariate analysis, however, after adjusting for other covariates, older patients, longer follow-up and female sex were associated with an increased risk of 11 Page 12 of 33 cyclic disease, whereas ‘adenoma identification’ was associated with a significantly lower risk for cyclic disease (Table 4). No difference was found between the patients with only cyclicity, with cyclicity and variability, or only variability, in terms of age, years of follow-up, cure rates after neurosurgery, imaging or pathologic findings. However, the presence of cyclicity was more prevalent before diagnosis as opposed to being seen in the followup period (P=0.02); variability was equally present before as well as after treatment. 12 Page 13 of 33 DISCUSSION In this study, it has been shown that cyclical or variable Cushing’s syndrome as defined by specific diagnostic criteria had a prevalence of approximately 15% in a large series of patients with Cushing’s disease, of whom 70% showed cyclicity or variability before the diagnosis was made and any therapy initiated. It is interesting that very similar percentages of 17% (42% with the first evidence post-operatively) (22) and 18% (with the first evidence 50% post-operatively) (23) were reported in small series of patients submitted to neurosurgical treatment, although a somewhat higher figure was suggested in an early series from Northern Ireland (24). If cyclicity is a feature of the patient’s presentation, confirmation of the diagnosis can be difficult, particularly if the patient is evaluated during the period of ‘cycling out’ (24). The diagnostic work-up may be particularly difficult in this ‘trough’ period, as at that time the routine diagnostic tests may all be apparently normal. Such variability may also interfere with effective treatment before definitive cure; in the first place, post-operative ‘cure’ may be mistakenly ascribed to effective surgery when the disease is simply in an inactive phase. Furthermore, the use of medical therapy may be complicated by variability in control which may lead to alternating periods of over- and under-treatment, increasing the necessity for more definitive therapy such as bilateral adrenalectomy. However, the finding of an approximately similar frequency of cyclicity being present before and after treatment does not indicate any particular period in the natural history to be more important, even if cycles were more commonly identified before treatment. We noted that most often two cycles, three peaks, were seen in each patient with an intercyclic period of approximately 4 years. Interestingly, a stable intra-subject periodicity of cycles was not observed, other than in a single patient. This would imply that regular cyclicity 13 Page 14 of 33 with ’12 hours to 85 days’ periodicity, as has been previously suggested, can clearly occur, but seems much less frequent than more irregular and unpredictable fluctuations. (2,10,25). Isolated clinical and biochemical cycles showed similar frequencies: clinical along with biochemical cycles were the most commonly identified, implying no additive information from laboratory data for the investigation of cyclicity. This finding suggests the importance of the signs and symptoms as being a harbinger of a new hypercortisolaemic peak. On the other hand, laboratory data are clearly important in confirming such variability, and this might be aided by the use of home-testing with salivary cortisol (26); this has been previously been shown to be particularly useful in establishing the definitive diagnosis and follow-up of ‘cyclic’ patients with rapid fluctuations in hormonal secretion (3). The population demonstrating cyclicity and variability do not seem to show any special defining characteristics, although the group as a whole had a slight but significantly higher mean age; this may reflect the impact of such variability in delaying diagnosis. There was no obvious difference between the adult and paediatric groups, but the number of patients in the latter was small. The cure rates in patients with cyclicity/variability showed a trend to be lower compared to the ‘non-cycling’ population with Cushing’s disease, but the numbers are again small. Low surgical cure rates have been previously reported in cyclic Cushing’s disease, and are in contrast with the commonly referred 80%-90% cure rate of Cushing’s disease (27,28). However, the apparent low cure rates in the total population may simply reflect the fact that patients were recruited over a prolonged period of time, with low cure rates being seen in the early years that this form of surgery was used. The same trend towards a difference between cyclic and non-cyclic disease was observed in terms of the presence of histological confirmation of an 14 Page 15 of 33 adenoma, which was less commonly seen in patients with cyclical disease. These findings may be prognostically useful in advising such patients, and may even be a determinant of the type of pituitary operation attempted, i.e. total hypophysectomy as opposed to microadenomectomy. Furthermore, the percentage of patients requiring bilateral adrenalectomy did not differ from the general population, although ‘blockand-replacement’ to control adrenal function appeared to be used in patients with cyclic disease. It has been previously suggested that pituitary-directed medical treatment, such as bromocriptine (29) or sodium valproate (30), might be particularly valuable in these patients, but the evidence for this is slim and in general these agents are not notably effective (31). Imaging studies do not seem to be useful in identifying cyclic disease since the proportion of patients with abnormal imaging was also broadly concordant between the groups. While in general the signs and symptoms also did not seem to differ between the two populations, there appeared to be a difference in gastrointestinal symptoms and in galactorrhea in females, but it is unclear if those are truly of clinical significance. We did note the previously-described (25) paradoxical responses to dynamic tests observed in a small percentage of this population, further emphasising the limitations of dynamic testing in these patients (6,7,9,32,33): this response has mainly been reported in patients with ACTH-independent primary pigmented nodular adrenal dysplasia (34). However, while clearly clinicians need to be aware of this pitfall in diagnosis, most centres emphasise the utility of the BIPSS in patients with ACTH-dependent Cushing’s syndrome, as well as the vital necessity of assessing the serum cortisol on the day of testing to confirm that the disease is active at the time of the procedure (18). As noted above, salivary cortisol assessment may prove a useful 15 Page 16 of 33 option, offering a more rapid home-assessment of fluctuations in cortisol secretion (3,35). Notably, female sex, longer follow-up period and the older age of patients at presentation seem to increase the risk of identifying ‘cyclic’ disease. However, it is difficult to conclude that these features can be considered specific, although the need for life-long follow-up is clear. In addition, histological confirmation of an adenoma was found to be less associated with cyclicity. Interestingly, no difference was observed between cyclicity and variability in any the parameters studied, other than the time of diagnosis, implying that there is no reason to analyse these parameters of cortisol fluctuations differently. Such ‘unpredictable hypersecretion’, which may or not be periodic, may represent a common patho-physiologic process. The molecular basis for this significant variability in hormonal release has been little studied in molecular terms (36). Several hypotheses have been suggested to explain the phenomenon, including episodic haemorrhage, the synchronous growth and death of tumour cells (9,10), or fluctuations in adrenal-pituitary axis feedback (14). It seems unlikely that there is abnormal hypothalamic control as such regulation should be absent in the presence of an autonomous tumour (9,28,37-40). One might speculate that the innate circadian rhythmicity of the tumorous corticotrophs is disturbed, which may account for the changes in ACTH output, although the long time course is difficult to reconcile with this theory. There are clearly a number of limitations to this study. Firstly, it is retrospective and relied totally on patient recall of the clinical situation before diagnosis, and the reports by the physicians in the medical case records. Hence, the data presented are descriptive. However, we believe that they are clinically useful as 16 Page 17 of 33 they seek to quantify, as far as is possible in a retrospective survey (since no prospective studies have been published), such oscillations in clinical and biochemical activity. In addition, the cyclicity or variability of the patients might have been underdiagnosed since we adopted very strict specific criteria. Nevertheless, all patients were followed-up in a single unit with unified protocols, such that there was a relatively uniform assessment. Consequently, it can be emphasised that the calculated prevalence reflects the minimum prevalence in this idiosyncratic group of patients. In addition, most patients with a diagnosis of Cushing’s disease are immediately treated and surgically cured, such that residual cyclicity and variability will no longer be seen. However, in spite of these limitations this is the first large-scale survey of this phenomenon, and its relatively high frequency suggests that there is a necessity for large prospective studies where more biochemical and clinical data can be properly evaluated. While the CDC is not in common use, we have found that it provides a good approximation to biochemical activity (18). The alternative would have been to have used 24h urinary cortisol excretion, but these assays are less reliably established and subject to errors in collection (19). For a large retrospective survey such day curves might currently represent the best estimate of biochemical activity we have available considering the unknown validity of the other diagnostic tests in ‘cyclic’ disease (32,33). In summary, we report a prevalence of approximately 15% of cyclicity and variability in patients with Cushing’s disease on clinical and/or biochemical criteria, with no particularly specific characteristic defining this population which could help differentiate them from patients lacking cyclicity. Such phenomena may lead to problems in confirmation of the diagnosis and/or result in dilemmas in deciding on initial treatment, particularly when surgery fails requiring bilateral adrenalectomy or 17 Page 18 of 33 medical treatment. We suggest that this is a minimum prevalence, and marked fluctuations in cortisol secretion might be a more common characteristic of Cushing’s disease than previously considered. The absence of commonly accepted criteria for the definition of cyclicity implies the necessity for prospective studies and for the setting of common guidelines. Physicians should be alert to the presence of cyclical Cushing’s syndrome since its presence influences the confirmation of diagnosis as well as therapeutic decisions in to order to optimise the effective management of the disease, and to effectively control the potential long-term sequelae. Disclosure There is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported. This research did not receive any specific grant from any funding agency in the public, commercial or not-for-profit sector. 18 Page 19 of 33 REFERENCES 1. Oates TW, McCourt JP, Friedman WA, Agee OF, Rhoton AL, Thomas WC Jr. Cushing’s disease with cyclic hormonogenesis and diabetes insipidus. Neurosurgery 1979 5 598–603 2. Bailey RE. Periodic hormonogenesis–a new phenomenon. Periodicity in function of a hormone-producing tumor in man. Journal of Clinical Endocrinology and Metabolism 1971 32 317–327 3. Hermus AR, Pieters GF, Borm GF, Verhofstad AA, Smals AG, Benraad TJ, Kloppenborg PW. Unpredictable hypersecretion of cortisol in Cushing's disease: detection by daily salivary cortisol measurements. Acta Endocrinologica (Copenh) 1993 128 428-432. 4. Zondek H, Zondek GW, Leszynsky HE. Fluctuability of steroid excretion. Acta Endocrinologica 1957 26 91–95. 5. Bassoe HH, Emberland R, Stoa KF. Fluctuating steroid excretion in Cushing’s syndrome. Acta Endocrinologica 1958 28 163–168 6. Brooks RV, Jeffcoate SL, London DR, Prunty FT, Smith PM. Intermittent Cushing’s syndrome with anomalous response to dexamethasone. Journal of Endocrinology 1966 36 53–61 7. Atkinson AB, McCance DR, Kennedy L, Sheridan B. Cyclical Cushing's syndrome first diagnosed after pituitary surgery: a trap for the unwary. Clinical Endocrinology (Oxford) 1992 36 297-299 8. Vagnucci AH, Evans E. Cushing’s disease with intermittent hypercortisolism. American Journal of Medicine 1986 80 83–88 9. Meinardi JR, Wolffenbuttel BH, Dullaart RP Cyclic Cushing's syndrome: a clinical challenge. European Journal Endocrinology 2007 157 245-254 19 Page 20 of 33 10. Mantero F, Scaroni CM, Albiger NM. Cyclic Cushing's syndrome: an overview. Pituitary 2004 7 203-207 11. Etxabe J, Vazquez JA. Morbidity and mortality in Cushing’s disease: An epidemiological approach. Clinical Endocrinology (Oxford) 1994 23 479–484 12. Mancini T, Kola B, Mantero F, Boscaro M, Arnaldi G High cardiovascular risk in patients with Cushing's syndrome according to 1999 WHO/ISH guidelines. Clinical Endocrinology (Oxford) 2004 61 768-777 13. Blau N, Miller WE, Miller ER Jr, Cervi-Skinner SJ Spontaneous remission of Cushing’s syndrome in a patient with an adrenal adenoma. Journal of Clinical Endocrinology and Metabolism 1975 40 659-663 14. Estopiñán V, Varela C, Riobo P, Dominguez JR, Sancho J. Ectopic Cushing's syndrome with periodic hormonogenesis--a case suggesting a pathogenetic mechanism. Postgraduate Medical Journal 1987 63 887-889 15. Newell-Price J, Bertagna X, Grossman AB, Nieman LK. Cushing's syndrome. Lancet. 2006 367 1605-1617 16. Trainer PJ, Besser GM: St. Bartholomew’s Hospital: The Bart’s Endocrine Protocols. Eds Churchill-Livingstone. Edinburgh, 1995 17. Isidori AM, Kaltsas GA, Pozza C, Frajese V, Newell-Price J, Reznek RH, Jenkins PJ, Monson JP, Grossman AB, Besser GM. The ectopic adrenocorticotropin syndrome: clinical features, diagnosis, management, and long-term follow-up. Journal of Clinical Endocrinology and Metabolism 2006 91 371-377 18. Trainer PJ, Eastment C, Grossman AB, Wheeler MJ, Perry L, Besser GM. The relationship between cortisol production rate and serial serum cortisol 20 Page 21 of 33 estimation in patients on medical therapy for Cushing's syndrome. Clinical Endocrinology (Oxford) 1993 39 441-443 19. Hellman L, Weitzman ED, Roffwarg H, Fukushima DK, Yoshida K. Cortisol is secreted episodically in Cushing’s syndrome. Journal of Clinical Endocrinology and Metabolism 1970 30 686–689 20. Krieger DT. Circadian pituitary adrenal rhythms. Advances in Experimental Medicine and Biology 1975 54 169–189. 21. Curtis GC. Long-term changes in corticosteroid excretion. In Biorhythms and Human Reproduction, pp 417–423. Eds M Ferin, F Halberg, RM Richart, RL Vande Wiele, New York: John Wiley, 1974. 22. McCance DR, Gordon DS, Fannin TF, Hadden DR, Kennedy L, Sheridan B, Atkinson AB. Assessment of endocrine function after transsphenoidal surgery for Cushing's disease. Clinical Endocrinology (Oxford) 1993 38 79-86 23. Streeten DH, Anderson GH Jr, Dalakos T, Joachimpillai AD. Intermittent hypercortisolism: a disorder strikingly prevalent after hypophysial surgical procedures. Endocrine Practice 1997 3 123-129 24. Atkinson AB, Kennedy AL, Carson DJ, Hadden DR, Weaver JA, Sheridan B. Five cases of cyclical Cushing's syndrome. British Medical Journal (Clinical Research Ed.) 1985 291 1453-1457 25. Shapiro MS, Shenkman L. Variable hormonogenesis in Cushing’s syndrome. The Quarterly Journal of Medicine 1991 79 351–363 26. Arnaldi G, Angeli A, Atkinson AB, Bertagna X, Cavagnini F, Chrousos GP, Fava GA, Findling JW, Gaillard RC, Grossman AB, Kola B, Lacroix A, Mancini T, Mantero F, Newell-Price J, Nieman LK, Sonino N, Vance ML, Giustina A, Boscaro M Diagnosis and complications of Cushing's syndrome: 21 Page 22 of 33 a consensus statement. Journal of Clinical Endocrinology and Metabolism 2003 88 5593-5602 27. Hofmann BM, Hlavac M, Martinez R, Buchfelder M, Müller OA, Fahlbusch R. Long-term results after microsurgery for Cushing disease: experience with 426 primary operations over 35 years. Journal of Neurosurgery 2008 108 9-18 28. Yasuda K, Miura K. Unique type of Cushing's disease in clinical profile: cyclic Cushing's syndrome and Cushing's disease with favorable outcome to a high daily dose of bromocriptine. Nippon Naibunpi Gakkai Zasshi 1994 70 1116 29. Adachi M, Takayanagi R, Yanase T, Sakai Y, Ikuyama S, Nakagaki H, Osamura Y, Sanno N, Nawata H 1996 Cyclic Cushing’s disease in long-term remission with a daily low dose of bromocriptine. Internal Medicine (Tokyo, Japan) 35 207-211 30. Beckers A, Stevenaert A, Pirens G, Flandroy P, Sulon J, Hennen G 1990 Cyclical Cushing's disease and its successful control under sodium valproate. Journal of Endocrinological Investigation 13 923-929 31. Alexandraki KI, Grossman AB. Pituitary-targeted medical therapy of Cushing's disease. Expert Opinion on Investigational Drugs 2008 17 669-677 32. James VH, Landon J, Wynn V. Oral and intravenous suppression tests in the diagnosis of Cushing's syndrome. The Journal of Endocrinology 1965 33 515524. 33. French FS, Macfie JA, Baggett B, Williams TF, Van Wyk JJ. Cushing's syndrome with a paradoxical response to dexamethasone. The American Journal of Medicine 1969 47 619-624. 22 Page 23 of 33 34. Stratakis CA, Sarlis N, Kirschner LS, Carney JA, Doppman JL, Nieman LK, Chrousos GP, Papanicolaou DA. Paradoxical response to dexamethasone in the diagnosis of primary pigmented nodular adrenocortical disease. Annals of Internal Medicine 1999 131 585-591 35. Yaneva M, Mosnier-Pudar H, Dugu´e MA, Grabar S, Fulla Y, Bertagna X. Midnight salivary cortisol for the initial diagnosis of Cushing’s syndrome of various causes. Journal of Clinical Endocrinology and Metabolism 2004 89 3345-3351 36. Dahia PL, Grossman AB. The molecular pathogenesis of corticotroph tumors. Endocrine Reviews 1999 20 136-155 37. Jordan RM, Ramos-Gabatin A, Kendall JW, Gaudette D, Walls RC. Dynamics of adrenocorticotropin (ACTH) secretion in cyclic Cushing’s syndrome: evidence for more than one abnormal ACTH biorhythm. Journal of Clinical Endocrinology and Metabolism 1982 55 531–537 38. Watanobe H, Aoki R, Takebe K, Nakazono M, Kudo M. In vivo and in vitro studies in a patient with cyclical Cushing’s disease showing some responsiveness to bromocriptine. Hormone Research 1991 36 227–234 39. Scapagnini U, Van Loon GR, Moberg GP, Preziosi P, Ganong WF. Evidence for central norepinephrine-mediated inhibition of ACTH secretion in the rat. Neuroendocrinology 1972 10 155–160, 40. Hsu TH, Gann DS, Tsan KW, Russell RP. Cyproheptadine in the control of Cushing’s disease. Johns Hopkins Medical Journal 1981 149 77–83 23 Page 24 of 33 TABLES Table 1. Summary of the characteristic features of cyclicity and variability in the population studied. Parameters studied Cyclic population Overall prevalence of cyclicity or variability 30/201 (14.9%) - prevalence of cyclicity 27/201 (13.43%) - prevalence of variability 12/201 (5.97%) Female population 26/30 (86.67%) Male population 4/30 (13.33%) Paediatric population (17 pts, <18years), % 1/17 (6.25%) Median number of cycles (range) 2 (1-4) Length of cycles (mean±SE, years) 3.8±0.6 (0.2-26) Type of 54 cycles Clinical cycles 8 (14.8%) Clinico-biochemical cycles 39 (72.20%) Biochemical cycles 7 (13.0%) Time of diagnosis Before diagnosis 12 (40%) After treatment 9 (30%) Before diagnosis + after treatment 9 (30%) - prevalence of cyclicity + variability 9 (30%) - prevalence of cyclicity only 18 (60%) - prevalence of variability only 3 (10%) 1 Page 25 of 33 Table 2. Summary of the distinctive and common characteristics found in patients with cyclic and non-cyclic disease. P<0.05 is taken as statistically significant. Parameters studied Cyclic (N=30) Non-cyclic (N=171) P Mean age (range), years 42.3±2.0 (17-72) 36.8±1.1 (7-95) 0.02 Mean follow up (range), years 14.8±1.6 (0-30) 14.0±0.8 (0-52) 0.68 Female, % 26 (86.67%) 154 (75.7%) 0.24 Pediatric population, % 1/30 (3.33%) 16/171 (9.36%) 0.48 Macroadenoma presence 2 (6.67%) 67 (16.37%) 0.53 Bilateral adrenalectomy 8 (26.67%) 47 (27.49%) 0.90 2 (25%) 19 (40.4%) 0.70 2 (6.67%) (-) N=19 N=127 Cured 6 (31.58%) 67 (52.76%) Recurrence 4 (21.05%) 12 (9.45%) Not cured 9 (49.37%) 48 (37.80%) Pathology N=19 N=127 P 9 (47.37%) 83 (65.35%) 0.09 N=30 N=156 P 19 (63.3%) 67 (69.87%) 0.52 Nelson’s syndrome Block and replacement treatment Neurosurgery Adenoma identification Imaging Suggestive/diagnostic findings P 0.09 2 Page 26 of 33 Table 3. Symptoms and signs of hypercortisolism in the two populations (P<0.05 is taken as statistically significant before adjustment for multiple comparisons) Parameters studied Cyclic (N=30) Non-cyclic (N=171) P 22 (73.3%) 139 (82.7%) 0.22 21 (70%) 86 (51.2%) 0.07 Weight gain 19 (63.3%) 111 (66.1%) 0.84 Hypertension 18 (60%) 118 (70.2%) 0.29 17 (56.7%) 83 (49.4%) 0.55 Depression 14 (46.7%) 50 (29.8%) 0.09 Obesity/ central adiposity 13 (43.3%) 101 (60.1%) 0.11 Thin skin 12 (40%) 56 (33.3%) 0.54 Change in exercise ability 12 (40%) 64 (38.6%) 1.00 Headaches/ Migraines 11 (36.7%) 55 (32.7%) 0.68 Oedema 11 (36.7%) 51 (30.4%) 0.53 Gastrointestinal symptoms 11 (36.7%) 30 (17.9%) 0.03 10 (33.3%) 62 (36.9%) 0.84 Striae 9 (30%) 71 (42.3%) 0.23 Skeletomuscular/ joints aches 9 (30%) 52 (31%) 1.00 Disuric symptoms 9 (30%) 36 (21.7%) 0.35 Infection 9 (30%) 41 (24.4%) 0.50 Skin pigmentation 9 (30%) 31 (18.5%) 0.15 Cardiovascular disease/symptoms 9 (30%) 37 (22%) 0.35 Cushingoid appearance Easy bruising/ Ecchimosis Proximal myopathy/ proximal muscles wasting Buffalo hump/ cervival or intrascapular fat pad 3 Page 27 of 33 Fatigue 8 (26.7%) 32 (19%) 0.33 Acne 7 (23.3%) 49 (29.2%) 0.66 Other psychiatric disorders 7 (23.3%) 35 (20.8%) 0.81 Dizziness/ Drops attacks/ Lethargy 7 (23.3%) 29 (17.3%) 0.44 Visual disturbances 6 (20%) 38 (22.6%) 1.00 Abnormal carbohydrate metabolism 6 (20%) 29 (17.3%) 0.80 Emotional lability 5 (16.7%) 21 (12.5%) 0.56 Skin disorders 5 (16.7%) 34 (20.2%) 0.81 Sleep disorders 4 (13.3%) 17 (10.1%) 0.53 Flushing/ Red face 4 (13.3%) 27 (16.1%) 1.00 Sexual dysfunction 4 (13.3%) 35 (20.8%) 0.46 Osteoporosis/ fracture 4 (13.3%) 20 (11.9%) 0.77 Neurological symptoms 4 (13.3%) 23 (13.7%) 1.00 Anxiety/ Panic 3 (10%) 18 (10.7%) 1.00 Increased sweating 3 (10%) 14 (8.3%) 0.73 Hair thin/ dry 3 (10%) 12 (7.1%) 0.71 Poor healing 2 (6.7%) 4 (2.4%) 0.23 Fundus abnormalities 2 (6%) 10 (6.7%) 1.00 Loss of weight 1 (3%) 5 (3%) 1.00 Renal calculi 1 (3.3%) 7 (4.2%) 1.00 Thirst/ Polydipsia 1 (3.3%) 10 (6%) 1.00 Thyroid disorders 1 (3.3%) 3 (1.8%) 0.49 0 2 (6.7%) 0.22 Mental alterations concentration (memory/ disturbances, 4 Page 28 of 33 slowness) Poor growth/short stature 0 9 (56.3%) N=26 N=128 Hirsutism 23 (81.1%) 103 (88.5%) 0.57 Virilisation 6 (23.1%) 35 (27.6%) 1.00 Menstrual irregularity 12 (46.2%) 63 (49.6%) 0.83 Galactorrhoea 5 (19.2%) 6 (4.7%) 0.02 Female population 0.47 5 Page 29 of 33 Table 4. The association of the parameters studied in terms of the presence of cyclic disease. P<0.05 is atken as statistically significant. Factors studied Results from Univariate Analysis Results from Multiple Analysis OR 95%CI P-value OR 95%CI P-value Age at presentation 1.03 1.00-1.05 0.05 1.04 1.00-1.09 0.03 Years of follow up 1.01 0.97-1.05 0.72 1.08 1.00-1.16 0.05 Sex 2.10 0.69-6.37 0.19 6.18 1.12-34.27 0.04 Adenoma identification 0.43 0.17-1.17 0.09 0.30 1.00-0.96 0.04 2.06 0.65-6.48 0.216 2.12 0.70-6.41 0.18 0.75 0.33-1.69 0.48 0.70 0.18-2.79 0.61 0.94 0.39-2.25 0.88 0.52 1.00-2.75 0.44 0.53 0.11-2.43 0.41 0.62 0.10-3.72 0.60 on pathologic specimen Cure (versus recurrence/ persistent disease) Radiologic evidence of pituitary abnormality Bilateral adrenalectomy performed Macroadenoma present 6 Page 30 of 33 Page 31 of 33 Page 32 of 33 Figure 1: Cortisol ‘Day Curve’ (nmol/L) assessment of a patient under long-term stable medical treatment with varying levels of cortisol. CDC: Blood samples were taken at 09.00h, 12.00h, 15.00h, 18.00h and 21.00h from the patient, the serum cortisol measured, and the mean value calculated. Individual values and the mean level are shown in the graph. Previous data based on normal volunteers and an isotopic dilution technique demonstrated that in normal subjects the mean of these 5 values should lie between 150 and 300 nmol/L. See text for references. Page 33 of 33 Figure 2: Cortisol Day Curve (nmol/L) levels in a patient after the confirmation of the diagnosis and following medical treatment after surgical failure. The arrows indicate the time of surgery and following radiotherapy before medical treatment, and the hypercortisolaemic peaks that can be compared as the patient was on the same treatment at these 4 time-points.