Excess Morbidity Persists in Patients With Cushing’s Disease During Long-term Remission: A Swedish Nationwide Study Free

Abstract

Cushing’s disease (CD) is associated with excess mortality (1-6). Despite remission, excess mortality in CD is doubled, mainly due to cardiovascular diseases (6). Furthermore, active CD is associated with thromboembolism, fractures, and severe infections (7,8). Whether long-term remission reduces the risk of cardiovascular disease, thromboembolism, fractures, and sepsis has not yet been fully elucidated.

Only 2 epidemiologic studies have addressed multisystem morbidity related to Cushing’s syndrome (CS). The first study, a nationwide study from New Zealand that included 253 patients with CS, showed a high prevalence of cardiovascular disease, including ischemic heart disease, even though 80% to 90% of the patients were in biochemical remission (2). The second study, a nationwide population-based study from Denmark that included 343 patients with CS, showed an increased risk for cardiovascular events, infections, and fractures compared to the background population, with the risk of myocardial infarction remaining elevated during long-term follow-up (9). Nevertheless, large cohort studies evaluating the incidence of comorbidities associated with CD, as well as the influence of remission state, are still lacking.

Herein, we present a large, nationwide, population-based study with the aim of investigating the incidence of cardiovascular diseases (myocardial infarction, ischemic heart disease, heart failure, atrial fibrillation, thromboembolism, and stroke), sepsis, fractures, and cancer in patients with CD with special emphasis on long-term morbidity.

Materials and Methods

Study design

This was a nationwide, retrospective, observational study of patients with CD. Data were obtained from the Swedish National Patient Register and the Swedish Cancer Register. The Swedish National Patient Register contains information on every hospital visit or admission since 1987. The Swedish Cancer Register contains information on all malignant tumors diagnosed in Sweden since 1958. Both registries, held by the National Board of Health and Welfare, have national coverage and contain high-quality data (10,11).

Study population

Patients who were diagnosed with CD between 1987 and 2013 were identified in the Swedish National Patient Register. Eligible for evaluation were patients who had received the diagnostic code for CS (255A) between 1987 and 1996 according to the ninth edition of the International Classification of Diseases (ICD-9) and those who had received the diagnostic code for CD (E24.0) or CS (E24.9) together with the code for benign pituitary adenoma (D35.2), between 1997 and 2013 according to the tenth edition of the International Classification of Diseases (ICD-10).

Experienced endocrinologists systematically reviewed medical records for all patients who were identified to validate the diagnosis of CD. A standardized protocol was used to retrieve information about clinical features, biochemical analyses, diagnostic imaging, results from inferior petrosal sinus sampling, and histopathological diagnosis (12). Evaluation of remission status was also based on review of medical records. Remission was defined as resolution of clinical features of CS together with one or more of the following: normalization of urinary free cortisol, normalization of serum or salivary cortisol at midnight, adequate cortisol suppression after dexamethasone suppression test, adrenal insufficiency, and/or bilateral adrenalectomy (12). After evaluation, the patients were divided into 3 groups: (i) in remission, (ii) not in remission, and (iii) unknown remission status.

Collection of morbidity data

A list of patients with confirmed CD was sent to the National Board of Health and Welfare and information on specific comorbidities was retrieved from the Swedish National Patient Register and from the Swedish Cancer Register. Information on the following predefined comorbidities, diagnosed between 1987 and 2014, were obtained for all patients: stroke; including hemorrhagic and ischemic (ICD-10 I61-I64; ICD-9 431-434, 436), myocardial infarction (ICD-10 I21.; ICD-9 410.), ischemic heart disease (ICD-10 I20-I25; ICD-9 410-414), heart failure (ICD-10 I50.; ICD-9 428.), atrial fibrillation (ICD-10 I48.9; ICD-9 427.3), pulmonary embolism (ICD-10 I26; ICD-9 415.), deep vein thrombosis (ICD-10 I80.-I82.; ICD-9 451.-453.), sepsis (ICD-10 A40.; A41, ICD-9 038), fractures (ICD-10 S02., S12., S22., S32., S42., S52., S62., S72., S82., S92.; ICD-9 800-829), and cancer (ICD-7 140-205). All patients were followed until the end of the study (December 31, 2014) or death.

To study morbidity in relation to CD diagnosis and remission, the follow-up period of the patients who were in remission at last follow-up was divided into 3 periods. The first period was defined as 3 years before CD diagnosis. Patients who received their diagnosis from January 1, 1990 until December 31, 2013 were eligible. The second period was from diagnosis until 1 year after remission (peri-treatment period) and the third period started at 1 year after remission until the end of the study (long-term remission). For patients who developed recurrence of CD, only the time after the last successful treatment was included in the long-term remission period.

Ethical considerations

The Regional Ethical Review Board in Gothenburg approved the study (reference number 145/11; approved April 25, 2013).

Statistics

Descriptive data are presented as mean ± standard deviation or median (interquartile range [IQR]).

Person-years at risk were calculated from study inclusion until death or end of the study, and were stratified according to gender, 5-year age groups, and 1-year calendar periods. The expected number of cases for each stratum was calculated by using the Swedish general population for every calendar year, 5-year age group, and gender as reference. The observed number of each comorbidity among patients with CD was compared to those expected in the general population and presented as standardized incidence ratios (SIRs) together with 95% confidence intervals (CIs). The 95% CIs were calculated assuming a Poisson distribution of the observed numbers. SIRs for each diagnosis were calculated on a yearly basis with the exception of ischemic heart disease, atrial fibrillation, and heart failure, which were analyzed based on the first event only. SIR was not calculated if fewer than 3 events were observed.

Fisher’s exact test was used to study the influence of glucocorticoid replacement therapy on thromboembolism and sepsis and the influence of radiotherapy on stroke. The level of significance was set to P < 0.05.

The analyses were performed using SPSS (version 22) and STATA (version 15).

Results

Of 1317 patients identified in the Patient Registry, 534 (41%) had the diagnosis of CD confirmed (12). At the time of the morbidity analysis, the CD diagnosis had not been validated in 32 patients. Thus, we included 502 patients with available morbidity data in this analysis.

Of the 502 patients with confirmed CD, 419 (83%) were in remission, 40 (8%) were not in remission, and 43 (9%) had unknown remission status at the last follow-up visit (Table 1). The median follow-up duration was 13 (IQR 6 to 23) years, resulting in 7160 person-years. Median time in remission was 10 (IQR 4 to 21) years. Patient characteristics are detailed in Table 1. Time at risk was 830 person-years in the first period (3 years prior to diagnosis), 678 person-years in the second period (from diagnosis until 1 year after remission), and 673 person-years in the third period (>1 year in remission).

Morbidity in the whole cohort during the whole study period

In the whole cohort (N = 502), a 3-fold increased SIR was observed for stroke, 2-fold for myocardial infarction, and 7-fold for sepsis (Table 2). The incidence for thromboembolism was also increased: 5-fold for pulmonary embolism and 3-fold for deep vein thrombosis. SIRs were increased for ischemic heart disease, heart failure, and fractures, but not for cancer.

Morbidity trends from 3 years before diagnosis and during the peri-treatment period among patients in remission at last follow-up

SIR for thromboembolism was increased in both the period before diagnosis (11.5; 95% CI 4.2 to 25.0) and the peri-treatment period (18.3; 95% CI 7.9 to 36.0). More specifically, for deep vein thrombosis, SIR was increased during both periods, while SIR for pulmonary embolism was elevated during the peri-treatment period (Table 3).

SIRs for myocardial infarction, ischemic heart disease, and fractures were increased only during the 3 years prior to diagnosis.

SIR for stroke was increased 5-fold during the peri-treatment period (4.9; 95% CI, 1.3 to 12.5), but not before the diagnosis (Table 3). SIRs for sepsis and atrial fibrillation were also increased during the peri-treatment period (Table 3).

Morbidity in patients during long-term remission

Two-hundred ninety-six patients (232 women and 64 men) had been in remission for at least 1 year and were followed until the end of the study. Median time in remission was 7 (IQR 3 to 14) years. The incidence of stroke (3.1; 95% CI 1.8 to 4.9), thromboembolism (4.9; 95% CI 2.6 to 8.4), and sepsis (6.0; 95% CI 3.1 to 10.6) remained elevated during long-term remission (Table 3; Fig. 1).

Figure 1.

Standardized incidence ratios for stroke, thromboembolism, and sepsis from 3 years before CD diagnosis until long-term remission stratified by follow-up duration.

Open in new tabDownload slide

Four out of 62 (6%) patients who had been treated with pituitary radiotherapy were diagnosed with stroke during long-term remission compared to 10 of 233 (4%) not treated with radiotherapy (P = 0.5).

One hundred and thirty-four patients in long-term remission were receiving glucocorticoid replacement. Eleven patients were diagnosed with sepsis: 8 of 134 (6%) on glucocorticoid replacement therapy and 2 of 140 (1%) not on replacement (P = 0.06). Three of 134 (2%) patients receiving glucocorticoid replacement therapy were diagnosed with thromboembolism during long-term remission compared to 5 of 140 (4%) not receiving glucocorticoids (P = 0.7).

Discussion

In this large, nationwide, population-based study, we found an increased overall incidence of stroke, myocardial infarction, ischemic heart disease, heart failure, pulmonary embolism, deep vein thrombosis, sepsis, and fractures in a well-validated cohort of 502 patients with CD. Importantly, the incidences of thromboembolism, sepsis, and stroke remained increased during long-term follow-up of patients in remission.

Hypercortisolism leads to hypertension, insulin resistance, hypercoagulability, visceral adiposity, osteoporosis, and impairment of the immune system (7,8,13-15). All these conditions may contribute to excess morbidity in patients with active CD. Since CD is a rare disease (1,2,12,16) limited data are available on long-term morbidity. A nationwide, population-based study including 343 patients with CS from Denmark with a mean follow-up duration of 12.1 years showed increased multisystem morbidity (9). In particular, patients with CS had increased risk for thromboembolism, myocardial infarction, stroke, and infections. Risks of myocardial infarction and infections were still elevated during long-term follow-up. Although the cohort was relatively large, there are some differences compared to the current study. The Danish cohort included both patients with CD (N = 211) and adrenal CS (N = 132). Also, a limitation was that the diagnosis and estimation of remission were based on registration according to ICD codes and not on review of medical records as in the current study.

Cardiovascular diseases are the main cause of death in patients with CD (6). In our study, the incidences of myocardial infarction and ischemic heart disease were elevated only before the patients were diagnosed with CD. In contrast, the incidence of stroke was elevated both in the peri-treatment period as well as during long-term remission. The reason for the discrepancy between the incidence of stroke and myocardial infarction remains unknown. In an earlier cross-sectional study of patients with CD, who were followed 5 years after achieving remission, the prevalence of atherosclerosis was increased compared to controls (17). In addition, cardiovascular risk factors such as obesity, hypertension, impaired glucose tolerance, and hyperlipidemia were still present, suggesting an increased cardiovascular risk (17). However, another study showed that the presence of atherosclerosis in patients with CD in remission did not differ from that of healthy controls if comorbidities related to atherosclerosis were absent or adequately treated (18). Therefore, the differences in incidences of stroke and myocardial infarction might be related to inadequately treated risk factors such as hypertension, the strongest risk factor for stroke (19).

In the present study, we found an increased overall incidence of heart failure in patients with CD. Changes in cardiac structure and function in patients with CD are reported to be reversible after remission has been achieved (20). Thus, this possibly explains why the incidence of heart failure was not elevated following long-term remission. The incidence of atrial fibrillation was only increased during the peri-treatment period, suggesting that surgery and not hypercortisolism per se is the main risk factor.

An important finding in our study was the 5-fold increased incidence of thromboembolism during long-term remission. Active CD is associated with high levels of procoagulant factors such as factor VIII and IX and von Willebrand factor as well as impaired fibrinolysis (21,22). In a multicenter, cohort study, the risk for thromboembolism was elevated during active disease as well as after pituitary surgery (23). A recent large, single-center, retrospective study showed that patients with CS have a higher risk for thromboembolism for approximately 30 to 60 days in the postoperative period (24). Another study showed that these abnormalities did not normalize 1 year following successful surgery (25). Indeed, this is supported by our findings, where the incidence of thromboembolism was increased around the time of treatment. Thus, ours and previous studies support the use of perioperative thromboprophylaxis, although the dosage and duration of treatment remains to be determined (26). Moreover, the increased incidence of thromboembolism during long-term remission warrants further studies on benefit-risk assessment of a prolonged thromboprophylaxis in patients with CD.

The incidence of sepsis was elevated during the peri-treatment period, probably due to an increased risk related to the surgery itself in combination with increased risk due to hypercortisolism (9). Hypercortisolism impairs the immune system, which, together with hyperglycemia and vascular damage, makes the patient with CD prone to severe infections (8,27). Indeed, severe infections are a significant clinical problem in patients with active CS (9,28-30) and one of the most common perioperative causes of death (31). In children and adolescents with CS, the risk for infection is especially increased among patients receiving glucocorticoid replacement (32). In the present study, we also observed an increased incidence of sepsis in patients with CD in long-term remission. In our cohort, the association between glucocorticoid replacement and sepsis during long-term remission could not be confirmed. However, our observation is based on relatively small number of events (8 vs 2 patients receiving glucocorticoids or not, respectively) with a trend for statistical significance (P = 0.06). Hence, the role of glucocorticoid replacement as a risk factor of sepsis, as well as of other comorbidities such as thromboembolism and fractures during long-term remission needs further investigations, both in terms of glucocorticoid type and dosage.

At the time of diagnosis, 38% to 50% of patients with CS have osteoporosis and 15% to 50% have had fractures (33,34). In a clinical study investigating vertebral fractures with spinal radiographs, asymptomatic fractures were identified in 48% of patients with CS (35). Bone mass and turnover were partially reversed 2 years after remission in patients with CD (36) while restoration of coupled bone remodeling and normalization of bone mineral density were observed at long-term follow-up (37,38). Only 2 previous population-based studies have assessed fracture risk in patients with CS, showing an increased risk 3 years before diagnosis, but not 3 years after treatment, which is in accordance with our findings (9,39).

The main strength of our study is the large number of patients with verified CD diagnosis and known remission status based on a systematic review of all medical records as well as the long duration of follow-up. In addition, the study included all patients in Sweden diagnosed with CD, which reduces the risk for selection bias. Moreover, the reliability of data from the Swedish National Patient Register for numerous diagnoses is high (11) with a positive predictive value of 98% for myocardial infarction (40), 97% for atrial fibrillation, 99% for stroke, 82% for heart failure (11), and 81% for pulmonary embolism (41).

A limitation of the study is the retrospective design, which is almost inevitable when studying long-term comorbidities in CD. Also, detailed information about peri-treatment glucocorticoid replacement and other cardiovascular risk factors, such as hyperlipidemia and visceral obesity, as well as the prevalence of diabetes mellitus and hypertension in the reference population was not available. The reason for the latter is that the Swedish National Patient Register does not contain robust information about diseases such as diabetes mellitus and hypertension that are mainly diagnosed and managed within the primary health care system. The same accounts for most psychiatric disorders. Another potential limitation is our definition of remission. Biochemical tests used for follow-up of patients with CD have changed considerably through the years, with urinary free cortisol most commonly used earlier and late-night salivary cortisol more recently. We are however confident that patients in the remission group were indeed in remission. If there was a doubt about remission status, it was categorized as “unknown.” The definition of the peri-treatment period can also be questioned, especially in patients treated with radiotherapy where remission is delayed. Therefore, long-term remission was defined, arbitrarily, to be from 1 year after curative treatment to avoid including complications related to surgery as well as patients with active disease following radiation.

In conclusion, this nationwide study on patients with CD shows a high incidence of multiple serious comorbidities before, during, and after treatment. Notably, we found an increased risk for thromboembolism, sepsis, and stroke during long-term follow-up. Thus, early identification and management of risk factors for these comorbidities are essential during long-term follow-up.

Abbreviations

Abbreviations
 
• CD

  Cushing’s disease

 
• CI

  confidence interval

 
• CS

  Cushing’s syndrome

 
• BMI

  body mass index

 
• ICD

  International Classification of Diseases

 
• IQR

  interquartile range

 
• SIR

  standardized incidence ratio

Acknowledgments

Editorial support was provided by Peter Todd of Tajut Ltd. (Kaiapoi, New Zealand).

Financial Support: The study was financed by grants from the Swedish state under the agreement between the Swedish government and the county councils, the ALF-agreement (ALFGBG-593301) and a grant from the Gothenburg Society of Medicine.

Additional Information

Disclosure Summary: DSO has received consultancy fees from Pfizer, Ipsen, and Sandoz and unrestricted research grants from Sandoz. DC has received lecture honoraria from Otsuka Pharma Scandinavia AB, Sanofi Genzyme, and Shire. PB has received consultancy fees from Pfizer; independent research grants from Novartis; and lecture fees from Ipsen, Pfizer, and Novartis. DP has received lecture fees from NovoNordisk, Boehringer-Ingelheim, and Sanofi. GJ has received lecture fees from Novartis, Novo Nordisk, Pfizer, Sandoz, Merck Serono, and Otsuka as wells as consultancy fees from Astra Zeneca and Shire. OR has received lecture fees from Novo Nordisk, Ipsen, Sandoz, and Pfizer; an unrestricted research grant from HRA-pharma; and consultancy fees from Novartis, Alnylam, and HRA-pharma. EP, PD, ES, TO, MP, KB, SB, CH, BEE, LB, CF, EME, JW, BE, A-KÅ, ES, and I-LB have nothing to declare.

Data Availability: Access to raw data is available upon request addressed to the corresponding author.

References