Mortality in Cushing's syndrome: declining over 2 decades but remaining higher than the general population

Abstract

Objective

Patients with endogenous Cushing's syndrome (CS) have elevated mortality, particularly during active disease. A recent meta-analysis reported reduced mortality rates after 2000 in adrenal CS and Cushing disease (CD), though many studies lacked population-matched controls.

Methods

Nationwide retrospective study (2000-2023) in Israel using the Clalit Health Services database to assess all-cause mortality in patients with endogenous CS matched 1:5 with controls by age, sex, socioeconomic-status, and body mass index (BMI). Primary outcome was all-cause mortality. Secondary outcomes included cause-specific mortality, impact of hypercortisolism remission, disease source, and mortality risk factors.

Results

The cohort included 609 cases with CS (mean age 48.1 ± 17.2 years; 65.0% women) and 3018 matched controls (47.9 ± 17.2 years; 65.4% women). Over a median follow-up of 16 years, 133 cases (21.8%) and 472 controls (15.6%) died (HR = 1.44, 95% CI, 1.19-1.75). Both patients with CD (HR = 1.73, 95% CI, 1.27-2.36) and adrenal CS (HR = 1.31, 95% CI, 1.00-1.81) had increased mortality risk. Patients without remission within 2 years had a higher mortality risk than those achieving remission (HR = 1.44, 95% CI, 1.00-2.17). Mortality was similar for CD and adrenal CS (HR = .83, 95% CI, .56-1.24). Older age, male gender, and prior malignancy were independent risk factors for mortality.

Conclusion

This is the largest national cohort study on mortality risk in CS over the past 2 decades, showing a significantly higher risk compared to matched controls in a homogeneous database. While etiology had no impact, remission significantly affected mortality, highlighting the importance of disease control for long-term survival.

Introduction

Endogenous Cushing's syndrome (CS) results from increased glucocorticoid production and extended exposure to cortisol and is classified into 2 main groups: Adrenocorticotropic hormone (ACTH)-dependent CS accounts for 70%-80% of cases (60%-70% Cushing disease [CD] and 5%-10% ectopic CS) and ACTH-independent CS accounts for 20%-30% of cases (20%-25% cortisol-producing adrenal adenoma, 5%-7% bilateral adrenal hyperplasia, 1% adrenocortical carcinoma).^1-5

While numerous studies have explored all-cause and cause-specific mortality in patients with endogenous CS, the rarity of the condition poses significant challenges.^6-20 Many of these studies are limited by small sample sizes and/or short follow-up periods, which undermine their ability to reliably estimate mortality risks. Furthermore, a substantial portion of the literature was published prior to 2000, before advancements in modern management.

The largest analysis to date on mortality risk in patients with CS included 20 study cohorts with a total of 3691 patients, spanning the years 1958-2018.²¹ Though the mean follow-up time for the studies included in the meta-analysis was not explicitly stated, it ranged between 3.1 and 16.8 years. The death rate was 10% in studies published before 2000 and 3% in those from 2000 onward and the leading causes of death included atherosclerotic diseases and thromboembolism (43.4%), infection (12.7%), and malignancy (10.6%). Despite improved outcomes in the post-2000 era, the study confirmed that patients with active CD still face significantly higher mortality risks compared to those in remission. The findings also highlighted that earlier cohorts included patients treated with now-outdated approaches, such as bilateral adrenalectomy, which may have influenced mortality outcomes.²¹ The significant reduction in the proportion of deaths noted post-2000 compared to pre-2000, in both the adrenal and CS groups, may suggest better outcomes from modern management.

As studies included in the meta-analysis had significant heterogeneity, our long-term study assessing mortality in patients with CS diagnosed in recent decades using a nationwide retrospective cohort study with individually matched controls offers a unique perspective on impact of recent treatment landscape on mortality. We have evaluated all-cause and cause-specific mortality risks in patients diagnosed with CS between 2000 and 2023, adding another 5 years of modern management versus prior data. The impact of disease remission, disease source, and baseline characteristics associated with increased mortality risk has also been analyzed.

Methods

Study design

This retrospective matched-cohort study utilized the electronic health record database of Clalit Health Services (CHS), the largest health maintenance organization in Israel, serving over 4.8 million members. Clalit Health Services insures a geographically and socioeconomically diverse population across Israel. Its geographic diversity is evident in the provision of healthcare services across major metropolitan areas, small towns, villages, and rural regions, serving Jewish, Arab, Druze, and Bedouin communities—each with unique healthcare needs and lifestyle factors. Clalit Health Services covers individuals across all life stages, from newborns and children to working-age adults and the elderly, as well as people from diverse socioeconomic backgrounds, including low-income populations, middle-class families, and high-income individuals. This diversity enhances the representativeness of its nationwide database for public health research and healthcare policy decisions. The database contains comprehensive demographic and clinical information, including hospital and outpatient diagnoses, laboratory test results, and medical treatments. Diagnoses and dates^22,23 were identified using ICD-10 codes. Data extraction was performed via the CHS research platform, powered by MDClone. The study protocol, including data collection methods, has been previously published.^22,23

The study complies with the Declaration of Helsinki and was approved by the Ethics Review Boards of Rabin Medical Center and granted a waiver of written informed consent, as data were anonymized and no active patient involvement was required.

Participants

Our cohort focuses on CD and adrenal CS only, with the exclusion of CS secondary to adrenal or any other ACTH secreting malignancy. The inclusion criteria for the cohort and methods for identifying patients with pituitary or adrenal causes of CS have been detailed in previous publications.²² Following initial screening, patients with an ICD-10 diagnosis of CS were required to meet at least one of the following criteria: (1) 24-h urinary free cortisol (UFC) ≥ 4 times the upper limit of normal (×ULN), (2) 24-h UFC ≥3 ×ULN with surgical intervention to remove a pituitary or adrenal adenoma, or (3) 24-h UFC ≥2 ×ULN with treatment using metyrapone, ketoconazole, osilodrostat, cabergoline, or pasireotide. All patients with CS and non-suppressed ACTH levels who did not receive pituitary-directed therapy and were diagnosed with a malignancy possibly causative of ectopic CS, including small cell lung carcinoma, bronchial and thymic carcinoids, medullary thyroid carcinoma, neuroendocrine tumors, or pheochromocytoma, were suspected of ectopic CS and were excluded from this study. Patients diagnosed with adrenocortical carcinoma before or within 5 years of CS diagnosis were excluded. Patients treated with supraphysiological doses of corticosteroids at the time of diagnosis were also not included in the study. Diagnosis was retrospectively ascertained by 2 study authors (AA, YR), at the time of data collection, based on the documented biochemical and imaging tests, as well as management and follow-up details. Each patient with CS was matched in a 1:5 ratio with age, sex, socioeconomic status, and body mass index (BMI)-matched controls from the general population who had never been tested for hypercortisolism. The matching was performed based on the date of Cushing syndrome diagnosis. Participants were followed from the time of CS diagnosis until death or censored at the data collection date, on June 30, 2023. All baseline data were recorded at the time of CS diagnosis.

Definitions and study outcomes

The primary outcome was time to all-cause mortality following a CS diagnosis during the follow-up period. Mortality data were obtained from the hospital's mortality database, systematically updated from the Ministry of the Interior Population Registry. Mortality data was collected until November 30, 2024.

Socioeconomic status was determined based on the classification of the Central Bureau of Statistics according to the city and street of the clinic to which the patient was assigned. The Central Bureau of Statistics provides a socioeconomic index of the population at the levels of local authority, settlement, and statistical area, based on demographic composition, education and schooling, standard of living, employment, and social benefits. These data are derived from various administrative sources, including the Central Bureau of Statistics, the National Insurance Institute, the Ministry of Finance, the Ministry of Education, the Ministry of Transport and Road Safety, and the Population and Immigration Authority.

The date of CS diagnosis was based on the first recorded instance of CS in the medical record, elevated UFC levels, or adrenal/pituitary surgery. Remission was defined as a normal 24-h UFC without medical therapy for hypercortisolism or presence of hypocortisolism requiring glucocorticoid replacement after surgery, within 2 years from CS diagnosis. In this study, a 2-year cutoff was established following a CS diagnosis to define the timeline for surgical remission. This time period provided sufficient time to complete a full diagnostic workup, implement treatment plans, conduct surgeries, and assess treatment responses. Due to the retrospective design of this multicenter study with multiple physicians involved, there was no standardized method for assessing remission and, in addition, some patients had their first UFC measurement several months after surgery, so the 2-year cutoff was applied to ensure consistency. In analyses that took into account the remission status, follow-up began 2 years after the initial diagnosis. This approach ensured that we only included patients who were alive at the 2-year mark, thus minimizing the risk of bias. For patients who achieved remission within the first 2 years, recurrence was defined as any evidence of hypercortisolism (ie, 24-h UFC ≥2 ×ULN or medication to lower cortisol levels) after the 2-year mark.

We compared the risk for all-cause mortality between patients who achieved biochemical remission and those who did not. Additionally, we compared mortality risk between patients with and without remission against individually matched controls. The analysis also included cause-specific mortality, differences in mortality risk based on disease source, and baseline characteristics that could be linked to an increased mortality risk. Data on cause-specific mortality was derived from hospital discharge notes at the time of admission that ultimately resulted in the patient's death.

Statistical analysis

Statistical analysis was generated using SAS Software, Version 9.4, SAS Institute Inc., Cary, NC, USA. Continuous variables were presented by mean ± standard deviation or median (interquartile range [IQR]). Categorical variables were presented by (N, %). The t-test, the Mann–Whitney test, and the χ² test were used for the comparison of normally distributed, non-normal, and categorical variables, respectively. The Cox proportional hazard model was used to calculate hazard ratios (HR) adjusted for age, gender, diabetes mellitus, hypertension, ischemic heart disease, cerebrovascular disease, and prior malignancy. Median and mean survival times were assessed with Kaplan-Meier survival analysis. The appropriateness of the proportional hazard assumption was assessed visually. Two-sided P-values less than .05 were considered statistically significant.

Results

Study cohort

Between January 1, 2000, and June 30, 2023, 609 patients with CS [396 (65.0%) women, mean age 48.1 ± 17.2 years) were included in the study. Each patient was matched with up to 5 controls based on age, sex, socioeconomic status, and BMI, resulting in a total of 3018 controls. The baseline characteristics of both the 609 patients and 3018 controls have been reported previously.^22,23

The etiology of hypercortisolism was CD in 251 (41.2%) patients, adrenal CS in 200 (32.8%), and indeterminate in 158 (25.9%) cases (Table 1).

At baseline, patients with CS had higher rates of diabetes mellitus, hypertension, dyslipidemia, ischemic heart disease, chronic kidney disease, and prior malignancy compared to controls (P < .05). Smoking rates were similar between the 2 groups (Table 1).

Mortality risk: Cushing syndrome vs matched controls

At the end of follow-up, with a median duration of 16.0 years (IQR, 11.3-21.6 years) for patients with CS and 16.1 years (IQR, 11.3-21.6 years) for controls, 133 of 609 patients with CS had died, compared to 472 of 3018 controls [21.8% vs 15.6%, respectively; hazard ratio (HR), 1.44, 95% CI, 1.19-1.75) (Table 2). Kaplan–Meier survival curves are presented in Figure 1. The mean age at the time of death was significantly younger in patients with CS compared to controls (72.6 ± 13.0 years vs 76.3 ± 11.0 years, respectively; P = .004) (Table 3).

The assessment of 5-year and 10-year mortality rates was restricted to individuals with a follow-up period of at least 5 and 10 years, respectively. Accordingly, the 5-year mortality analysis included 583 patients with CS and 2888 matched controls with available follow-up data, while the 10-year mortality analysis included 487 patients with CS and 2411 controls who had follow-up data for at least 10 years postdiagnosis. At 5 years following the CS diagnosis, the mortality rate was 3.6% (21 deaths) in patients with CS and 2.9% (83 deaths) in controls (HR 1.27, 95% CI, .79-2.05). At 10 years, the mortality rate increased to 10.7% (52 deaths) in patients with CS, compared to 7.6% (184 deaths) in controls (HR 1.41, 95% CI, 1.04-1.92) (Table 2).

Cause-specific mortality

Cause of death data were available for 90 patients with CS. The most common causes of mortality were infectious diseases (38 cases, 28.6%), followed by malignancies (17 cases, 12.8%), cardiovascular disease (13 cases, 9.8%), and cerebrovascular diseases (7 cases, 5.3%). In the control group, cause-specific mortality data was available for 285 patients. The leading causes of death were infectious diseases (101 cases, 21.4%), cardiovascular disease (47 cases, 10.0%), malignancies (43 cases, 9.1%), and cerebrovascular disorders (25 cases, 5.3%) (Table 3). Of note, 8 cases of non-fatal suicide attempts were recorded in patients with CS.

All-cause mortality by disease etiology

Of 251 patients with CD, 53 (21.1%) died during follow-up vs 159 (12.8%) of 1246 matched controls (HR 1.73, 95% CI, 1.27-2.36). At 5 years following CS diagnosis, mortality rates were 4.2% vs 3.1%, respectively (HR 1.40, 95% CI, .69-2.82), and at 10 years following diagnosis, respective mortality rates were 10.6% vs 6.2% (HR 1.73, 95% CI, 1.04-2.88).

Of 200 patients with adrenal CS, 47 (23.5%) died during follow-up vs 181 (18.3%) of 991 matched controls (HR 1.31, 95% CI, 1.00-1.81). At 5 years following CS diagnosis, mortality rates were 4.2% vs 2.8%, respectively (HR 1.49, 95% CI, .68-3.27), and at 10 years following disease diagnosis, respective mortality rates were 10.7% vs 9.8% (HR 1.09, 95% CI, .65-1.85).

Mortality risk: disease remission vs no remission

All remission analyses were updated to include only subjects with at least 2 years of follow-up. Data on remission status at 2 years was available for 426 patients with CS, with 291 (68.3%) achieving remission and 135 (31.7%) not achieving surgical remission.

Among the 291 patients who achieved remission within 2 years of diagnosis, 47 (16.1%) died during follow-up, compared with 181 (13.0%) of 1387 matched controls (HR 1.27, 95% CI, 1.00-1.75). In contrast, among the 135 patients who did not achieve remission by 2 years, 36 (26.7%) died during follow-up, compared with 84 (12.6%) of 664 matched controls (HR 2.29, 95% CI, 1.55-3.38) (Table 4). Notably, 61 patients (20.2%) who were in remission at 2 years later experienced disease recurrence during follow-up. Among them, 11 patients (18.0%) had died by the end of follow-up, compared with 36 of 241 patients (14.9%) who remained in continuous remission throughout the median follow-up period of 13.7 years (IQR, 5.5-20.9 years) (P = .55).

The mean age at death was significantly younger for patients with CS in remission compared to controls (71.6 ± 13.5 years vs 76.6 ± 11.6 years, P < .001), and for those not in remission compared to their matched controls (70.6 ± 14.2 years vs 75.6 ± 10.5 years, P < .001) (Table 3).

Compared with patients in remission, patients who did not attain remission within 2 years following CS diagnosis were at 1.44-fold increased risk for mortality at the end of follow-up (HR 1.44, 95% CI, 1.00-2.17).

At 5 years, the mortality rate was 3.7% among matched controls, 3.8% among patients in remission at 2 years, and 5.4% among patients not in remission. At 10 years, mortality was 8.6% among controls, 11.5% among patients in remission, and 13.7% among those not in remission (Table 4).

Data on glucocorticoid treatment following surgery for CS was available for 105 patients in remission. Among these, 81 were on replacement (physiologic) doses, while 24 were on supraphysiologic doses. The median treatment duration for those on physiologic doses of glucocorticoids was 12.0 (IQR, 3.6-26.7) months, with 4 patients on permanent long-term replacement treatment and 13 patients on replacement at the end of the follow-up. Mortality was significantly higher in the supraphysiologic dose group, with 9 deaths (37.5%) compared to 12 deaths (14.8%) in the replacement dose group (P = .015).

All-cause mortality by disease etiology and remission status

Analysis by disease source and remission status showed that among patients with CD, 5 of 102 (4.9%) in remission died within 5 years of diagnosis, compared to 4 of 77 (5.2%) who were not in remission. At 10 years, 8 of 78 (10.3%) in remission had died, compared to 8 of 58 (13.8%) not in remission. For patients with adrenal CS, 3 of 123 (2.4%) in remission died within 5 years, compared to 3 of 41 (7.3%) who did not achieve remission. At 10 years, 8 of 101 (7.9%) in remission had died, compared to 6 of 35 (17.1%) not in remission. Due to the small number of fatalities, statistical analysis was not performed, and the data is presented descriptively.

Mortality rates for patients with CD in remission were numerically higher than matched controls at 5 years (4.9% vs 4.4%, respectively), 10 years (10.3% vs 7.8%), and at the end of follow-up (16.2% vs 11.1%, respectively), though the difference was not statistically significant (HR = 1.46, 95% CI, .87-2.48) (Table 4). For patients with CD who did not attain remission, mortality risk was almost 3-fold higher than controls (HR = 2.91, 95% CI, 1.75-4.82).

Predictors of mortality

Univariate analysis identified several factors associated with an increased risk of all-cause mortality among patients with CS (Table 5). Male gender was a significant risk factor, with 29.1% of male patients succumbing during the follow-up period compared to 17.9% of female patients (HR = 1.68, 95% CI, 1.20-2.37). Advanced age at diagnosis also emerged as a strong predictor, with each additional year of age increasing the risk of mortality by 8% (HR = 1.08, 95% CI, 1.06-1.09) (Table 5).

While body mass index did not significantly predict mortality risk, several comorbid conditions markedly increased mortality risk. A history of malignancy before CS diagnosis was associated with an 8-fold increase in all-cause mortality (Table 5). Hypertension and ischemic heart disease were each associated with a fourfold higher risk, diabetes mellitus was linked to more than a twofold increased risk of death (Table 5, Supplementary Figures).

The Kaplan-Meier survival curves of cases and controls with hypertension, diabetes, and both comorbidities combined are shown in Figure S1, Figure S2, and Figure S3, respectively.

Regarding disease etiology, no significant difference in mortality risk was observed between patients with CD and those with adrenal CS (HR = .83, 95% CI, .56-1.24) (Table 5).

In the multivariate analysis that included age, gender, hypertension, diabetes mellitus, ischemic heart disease, dyslipidemia, chronic kidney disease, and prior malignancy, each additional decade of age at the time of CS diagnosis was linked to a 60% higher risk of all-cause mortality by the end of follow-up. Male gender was associated with a 1.5-fold increase in mortality risk, chronic kidney disease was associated with a 3-fold increased risk while a prior history of malignancy before a CS diagnosis was associated with more than 4-fold increase in all-cause mortality risk (Table 5).

Discussion

This study represents the largest to date examining all-cause mortality risk in patients with CS over the last 2 decades. Our findings highlight a significantly elevated mortality risk in patients with CS compared to controls matched for age, gender, BMI, and socioeconomic status, even though we have excluded patients with ectopic CS and adrenal carcinoma. Interestingly, similar to Dekkers et al.,⁸ but in contrast to the most recent meta-analysis,²¹ all-cause mortality rates were comparable between patients with CD and those with adrenal CS. While disease etiology had no significant impact on mortality risk, success in achieving remission was associated with a reduced risk of mortality during follow-up. This underscores the critical role of disease control in improving long-term outcomes for patients with CS, despite therapeutic innovations and quality of care improvements in recent years.^24,25

Although multiple studies have evaluated all-cause mortality risk in patients with endogenous CS, the rarity of the condition often limits these investigations to either small cohorts or short follow-up periods.^6-20 To address these limitations, Graversen et al. conducted a meta-analysis over a decade ago that included 797 patients from 7 studies, comprising patients with CD (n = 688) and adrenal CS (109 patients). They reported a standardized mortality ratio (SMR) of 1.8 (95% CI, 1.3-2.7) for CD and 1.9 (95% CI, .9-3.9) for CS due to adrenal adenomas. Interestingly, no significant difference in mortality was observed between surgically cured CD patients and the general population.²⁶ More recently, a meta-analysis by Limumpornpetch et al.²¹ evaluated mortality risk across 20 study cohorts, including 3691 patients with CS over a time span from 1958 to 2018, and reported a pooled SMR of 3.00 (95% CI, 2.33-3.85) for all CS cases. The SMR for CD was 2.8 (95% CI, 2.1-3.7), significantly lower than that for adrenal CS (3.34; 95% CI, 1.68-6.63), indicating a higher mortality burden for adrenal CS. Limumpornpetch et al. emphasized that studies with shorter follow-up durations of 1-10 years did not reveal a significant association between follow-up length and mortality rates. However, when the median follow-up period extended beyond 10 years, mortality rates were significantly higher compared to cohorts with less than 5 years of follow-up.

Notably, the mortality rate observed in our cohort (21.8%), was higher than that reported in the meta-analysis by Limumpornpetch et al., which is likely attributable to the significantly longer follow-up period in our study (16 years vs 6 years in the meta-analysis). However, when we analyzed mortality rates specifically at 5 years following CS diagnosis, the rates closely mirrored those reported in the meta-analysis, with a mortality rate of 3.6%. Over time, this rate rose to 10.7% at 10 years, underscoring the importance of extended follow-up to fully capture the long-term mortality burden associated with CS. The mortality rates in our cohort align closely with those reported by Ragnarsson et al.²⁷, who observed 133 deaths among 502 patients with CD over a mean follow-up period of 13 years in a similarly sized cohort. Furthermore, our findings align with the trends noted in the meta-analysis regarding reduced mortality risk in recent decades; rates decreased significantly over time, with 10% of deaths reported in studies published before 2000 compared to 3% in studies published from 2000 to 2021, reflecting an overall 71% reduction in mortality risk. Consistent with this trend, our study was limited to patients diagnosed with CS after 2000, supporting the notion of improved outcomes in more recent years, possibly due to advancements in modern diagnostic testing that have enabled earlier and more accurate diagnoses, while surgical techniques and medical treatments have also progressed. These improvements are further supported by the growing volume and depth of publications, particularly since 2000.^21,28,29

Our data show that mortality rates at the end of follow-up were higher for patients with CS in remission compared with matched controls, similar to prior reports indicating mortality risk remains increased despite biochemical remission,^30,31 and the meta-analysis reporting that mortality was still 2.3 times higher in patients with CD in remission than in an age- and sex-matched general population.²¹ The ongoing elevated mortality risk, despite biochemical remission, may result from irreversible effects of hypercortisolemia,³² disease recurrence, long-term development of malignancy²² or venous thromboembolism,^33-37 or prolonged use of supraphysiologic glucocorticoid doses.³⁸ While most patients in remission in our cohort received temporary glucocorticoid replacement dose and treatment was stopped over the course of the follow-up, a minority had adrenal insufficiency at the end of follow-up, which may contribute to the mortality risk in those in CS remission.³⁸ Unsurprisingly, mortality rates were almost 3-fold higher for patients with CD not attaining remission within 2 years, compared with those in remission.

While Limumpornpetch et al.²¹ reported that mortality estimates for men and women were similar, results from the European Registry on Cushing's syndrome (ERCUSYN) registry indicated patients who died were more often males.¹¹ In our study men had indeed an increased mortality risk, and additional studies are required to assess potential gender differences in long-term survival of patients with CS. Additional factors associated with increased mortality risk included advanced age, hypertension, diabetes mellitus and ischemic heart disease. Interestingly, similar to Graversen et al.²⁶ we did not find any difference in mortality risk between patients with CD or those with adrenal CS, but this is in contrast with Limumpornpetch et al, who reported a higher mortality risk with adrenal CS vs CD. Yet it should be noted that the period of patient enrollment in many studies included in the meta-analysis occurred before 1990 for adrenal CS and involved diagnostic methods such as pre-CT, pre-MRI, and pretranssphenoidal surgery, potentially contributing to higher mortality. Additionally, the 95% confidence intervals for all subgroups of adrenal CS were wider compared to CD, likely due to smaller sample sizes.²¹

Older age and male gender were independent risk factors for mortality in patients with CS in our study, consistent with findings from a previous study.³⁹ Older age is obviously associated with mortality in any adult patient population. However, a study from the ERCUSYN registry found that older patients with CS often lack several typical symptoms of hypercortisolism, which may lead to delayed diagnosis and prolonged exposure to excessive amounts of cortisol and a heightened mortality risk.^40,41 A recent study of 130 patients with ACTH-dependent CS from the Netherlands found a nonsignificant trend of higher mortality rates among male patients with CS (HR = 2.35, 95% CI, .73-7.51),⁴² while results from the ERCUSYN registry revealed that patients who died were more frequently male.¹¹ This finding was explained by older age at diagnosis and a slightly higher prevalence of CS-related comorbidities in males. Lastly, we found here that prior malignancy is an independent risk factors for mortality in patients with CS. As there are growing evidence that CS is associated with higher rates of malignancy, and that high cortisol levels may promote cancer development,^22,43 further large-scale studies are needed to confirm this finding, which suggests excess mortality in patients with CS who develop malignancy after the diagnosis of CS, and raises questions about a more aggressive course of malignancies in CS. Chronic hypercortisolism, a hallmark of CS, can lead to immunosuppression, which may reduce the body's ability to surveil and eliminate malignant cells.⁴⁴ Additionally, hypercortisolism can induce metabolic disturbances such as obesity and diabetes, which are known risk factors for various cancers.^45,46 Furthermore, glucocorticoids can modulate gene expression through epigenetic mechanisms, potentially leading to long-lasting changes that predispose to malignancy in patients with CS.⁴⁷ Cause-specific mortality for malignancy was higher in patients with CS than in controls, supporting the potential role of malignancies as a mediator of increased mortality in this group.

We found that the most common causes of death were infectious diseases (28.6%), malignancies (12.8%), and cardiovascular diseases (9.8%), similar to results from the ERCUSYN registry¹¹ but different from Limumpornpetch et al.²¹ meta-analysis, where cardiovascular diseases were found to be the most common cause of death (27.4%) and the second most common cause of death was infectious diseases (12.7%). The specific causes of death were examined in this study according to the reason for the last hospitalization that preceded the patients' death. This method may have biased our results and led to an increase in the number of infections (a common cause of hospitalization) compared with patients who died of ischemic heart disease, in which death sometimes occurs before the patient has had enough time to reach the hospital. Furthermore, the definition for cardiovascular mortality may differ between different studies as some included cardiac causes, cerebrovascular disease, and thromboembolism as cardiovascular mortality, while in our study cardiovascular disease implied myocardial infarction, ischemic heart disease, and congestive heart failure.⁴⁸

This study has several inherent strengths, including large size, long-term follow-up for a rare disease, strict criteria for CS diagnosis, disease etiology, remission status, and a well-matched control group design, used to minimize the effect of age, sex, socioeconomic status, and BMI on cancer risk. Moreover, our database represents the complete population and we assume that screening for cardiovascular disorders as well as treatment for diabetes mellitus, hyperlipidemia, and hypertension is done in all patients with CS and matched control per national guidelines. However, retrospective registry-based research has several limitations. Missing data could have prevented us from determining disease etiology or remission status in some cases, as well as glucocorticoids dose and/or duration throughout the course of the follow-up. Lack of data on disease duration prior to diagnosis is another significant limitation, as has previously been identified as a potential risk factor for mortality.³⁹ The causes of death in this study should be interpreted with caution, as 32.3% of the deaths in CS cases were categorized as unknown, and even when causes were reported, the true etiology could not always be confirmed. Another limitation is the lack of consideration for the severity of comorbidities, their control status, and whether they improved following remission of CS.

Finally, the sample size of the study and the small number of outcomes may have influenced the results, so that differences that were non-significant would have been found to be significant if the sample size had been larger.

In conclusion, presently, despite significant advancement of diagnostic and therapeutic options, we have shown in this large nationwide retrospective matched-cohort study that patients with CS are at an increased mortality risk compared with matched controls. Surgical remission within 2 years may attenuate the risk of all-cause mortality in patients with CS, who continue to have a higher mortality risk even following disease remission. We have also identified several potential risk factors for mortality, though disease etiology was not found to be an important predictor of mortality risk. Further research in larger multinational databases is needed to confirm our findings, and to elucidate possible mechanisms for this increased mortality risk.

Supplementary material

Supplementary material is available at European Journal of Endocrinology online.

Funding

None.

Authors’ contributions

Amit Akirov (Conceptualization [lead], Data curation [lead], Formal analysis [lead], Investigation [lead], Methodology [lead], Writing—original draft [lead], Writing—review & editing [lead]), Maria Fleseriu (Validation [supporting], Writing—review & editing [supporting]), Hiba Masri Iraqi (Methodology [supporting], Writing—review & editing [supporting]), Tzipora Shochat (Formal analysis [supporting], Methodology [supporting]), Shiri Kushnir (Data curation [equal], Methodology [supporting]), Ilan Shimon (Methodology [supporting], Writing—review & editing [supporting]), and Yaron Rudman (Formal analysis [supporting], Methodology [supporting], Writing—original draft [supporting], Writing—review & editing [supporting]).

Data availability

The data that support the findings of this study are available from CHS. Restrictions apply to the availability of these data, which were used under license for this study. Deidentified individual participant-level data sharing will be considered by the corresponding author of this study, with the permission of CHS. All applicants will be asked to sign a data access agreement. All requests will be discussed, whether data sharing is appropriate, based on the scientific rigor of the proposal.

References

Author notes