Holistic burden of illness in patients with endogenous Cushing's syndrome: A systematic literature review

Abstract Objective The objective of this systematic literature review (SLR) was to summarize the latest studies evaluating the burden of illness in endogenous Cushing's syndrome (CS), including the impact of CS on overall and domain‐specific health‐related quality of life (HRQoL) and the economic burden of CS to provide a holistic understanding of disease and treatment burden. Methods An SLR was conducted in PubMed, MEDLINE and Embase using the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) checklist to identify peer‐reviewed manuscripts and conference abstracts published in English from 2015 to December 4, 2020. Results Forty‐five publications were eligible for inclusion; data were extracted from 37 primary studies while 8 SLRs were included for reference only. Thirty‐one studies reported HRQoL using validated patient reported outcome (PRO) measures in pre‐ or post‐surgery, radiotherapy and pharmacotherapy patients. Overall, this SLR found that patients with CS have worse outcomes relative to healthy populations across specific dimensions, such as depression, despite an improvement in HRQoL post‐treatment. These findings reveal that CS symptoms are not fully resolved by the existing care paradigm. Few studies report on the economic burden of CS and currently available data indicate a high direct healthcare system cost burden. Conclusions Patients with CS experience a significant, complex and multifactorial HRQoL burden. Symptom‐specific burden studies are sparse in the literature and the understanding of long‐term CS symptomatic burden and economic burden is limited. This review intends to provide an updated reference for clinicians, payers and other stakeholders on the burden of CS as reported in published literature and to encourage further research in this area.


| INTRODUC TI ON
Endogenous Cushing's syndrome (CS) is a rare condition caused by chronic oversecretion of cortisol.The estimated annual incidence is 1.8-3.2cases per million, 1,2 with approximately 70% of cases occurring in females.Based on aetiology, CS can be classified as adrenocorticotropic hormone (ACTH)-dependent or ACTHindependent.ACTH-independent CS is most commonly caused by a cortisol producing adrenal adenoma, and rarely by adrenal cortical carcinoma, or bilateral disease such as macronodular adrenal hyperplasia or primary pigmented nodular adrenocortical disease. 3TH-dependent CS accounts for 80% of cases and is caused by either pituitary or ectopic ACTH producing tumours.Cushing's disease (CD), the most common form of endogenous CS, is caused by an ACTH-secreting pituitary adenoma that stimulates the adrenal overproduction of cortisol.3 Chronic hypercortisolism can lead to multisystem morbidities due to the ubiquitous downstream effects of cortisol.Patients with CS present with a variety of clinical manifestations of cortisol excess, including mood, reproductive, skin, neurocognitive and metabolic symptoms, many of which are common in the general population.4 Current treatment options for CS include transsphenoidal surgery as first-line treatment, followed by pharmacotherapy, bilateral adrenalectomy, and pituitary radiotherapy, as second-line treatment options.Available medical therapies include steroidogenesis inhibitors (e.g.ketoconazole, levoketoconazole, etomidate, metyrapone and osilodrostat), glucocorticoid receptor antagonists (e.g.mifepristone) and tumour directed therapies (e.g.pasireotide and cabergoline).
These medications are administered either as mono-or combination therapies and may be used to treat CS prior to surgery, when surgery is contraindicated, when remission is not achieved post-operatively, or in the setting of disease recurrence while awaiting the effects of radiation therapy. 5Despite the available treatment options, there remains a clear need to improve the burden of illness in Cushing's syndrome, as it relates to healthrelated quality of life (HRQoL) and economic burden of the disease and its treatment. 6eviously published systematic literature review (SLR) studies have examined the literature on the burden of illness in patients with CS or CD.However, many SLRs were limited in their approach.Several studies describing patients with CD were actually designed to evaluate the burden of illness among patients with all types of pituitary tumour subtypes while other studies focused solely on specific symptoms instead of the holistic burden. 7,8The objective of this SLR is to provide a comprehensive, up-to-date summary of the latest studies evaluating the burden of illness in CS, including the impact of CS on overall HRQoL and on specific domains of health such as anxiety, depression and fatigue.In addition, this study aims to summarize evidence on the economic burden of CS experienced by patients, in terms of the direct (e.g.hospitalization, outpatient visits and pharmacy costs) and indirect costs (e.g.loss of productivity, loss of pay and unpaid work) associated with treatment or management of the condition, with the goal of providing a holistic understanding of both disease and treatment burden and the impact of each on patients.

| MATERIAL S AND ME THODS
This SLR study was conducted between October 2020 and April 2021 according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and conducted by the full team of study authors. 9Select findings from this SLR were presented as posters at the European Neuroendocrine Association (ENEA) Workshop in 2021 and the International Society for Pharmacoeconomic and Outcomes Research (ISPOR) Conference in 2022. 10,11The SLR protocol was not registered as the review aimed to synthesize qualitative data and a follow-up meta-analysis was not planned.

| Data Sources
PubMed (via PubMed.com), MEDLINE (via embase.com) and Embase (via embase.com) were searched systematically using search terminology pertaining to endogenous CS, specifically humanistic (or HRQoL-related) and economic outcomes.Search algorithms for each electronic database were designed using appropriate search strategies including a combination of medical subject headings (MeSH) or Emtree terms and free texts in titles and abstracts of indexed publications (Tables A1 and A2).This includes terms such as 'Cushing syndrome', 'Cushing disease', 'quality of life' and 'resource use' (Tables A1 and A2).A3.

| Study selection
Upon execution of the systematic searches across the three databases, duplicates were removed using Endnote™, version X9, Philadelphia, PA: Clarivate™ (2013).The screening phase was conducted in 2 steps: title and abstract screening followed by full text screening.The screening of titles and abstracts was conducted by a single reviewer (BO), and a random selection of 15% of excluded citations were validated by a second reviewer (AS).Full text screening was conducted by two independent reviewers (AS and BO), thus ensuring cross-validation and consensus.
Both screening steps were conducted on a web-based screening software, Covidence (Veritas Health Innovation; Melbourne, Australia).A complete list of included and excluded citations (including the reasons for exclusion) were generated post the full-text screening phase.

| Data extraction
Data from the included studies were extracted in a Microsoft Excel™ (Microsoft Corporation, United States) based evidence grid by one researcher (BO) and validation of the included data were conducted by a second researcher (AS).Detailed data were extracted for included studies such as publication details, study characteristics, patient characteristics and specific QOL and economic burden of illness outcomes, where available.

| Outcomes measured
Studies measuring outcomes that impact the HRQoL of patients with CS and the economic impact of CS on patients or the healthcare system were included.

| Study selection
The searches identified 301 records across the 2 electronic databases, of which 246 (82%) were eligible for screening after deduplication.Following a comprehensive review of the 246 titles and abstracts, 74 (30%) studies were selected for further full-text review, of which 45 (18% of 246) studies met criteria for qualitative synthesis of the literature (Figure 1). 7,8, Studes were excluded for a variety of reasons throughout the screening process, including but not limited to, the study population, study design or outcomes being out of scope and thus not of interest to the authors (Figure 1).

| Study and patient characteristics
Table 1 summarizes the key study and patient characteristics for all included primary studies.Of the 45 studies that met eligibility criteria, 37 (82%) were primary studies and 8 (18%) were previously published SLRs.0][51][52][53][54] Of the included studies, 21 (57%) were conducted in Europe, 14 (38%) in North America, 1 in South America (3%) and 1 (3%) in Asia.Among the 37 primary research studies, there were no notable qualitative differences between results in studies conducted in different geographies.Sample sizes of patients with CS ranged from 6 to 1852 patients across the 37 studies.Most studies (30; 81%) reported on prior interventions, including transsphenoidal surgery, adrenalectomy, radiotherapy and pharmacotherapy.
In the 29 (78%) studies reporting patients' sex, female patients comprised 18%-100% of the study cohorts.In the 26 (70%) studies reporting patient age, the mean age ranged from 13 to 55 years.

| Cushing's syndrome specific HRQoL
34][35][36][37][38][39][40]42,43,46 CushingQoL is a 12-item, CS-specific HRQoL PRO that examines patients' HRQoL over the past 4 weeks. 8The Tuebingen CD-25 is a multidimensional HRQoL PRO that examines six domains (mood, sexual activity, eating behaviour, social environment, physical and cognitive functions). 30QoL-CD assesses disease progression and evaluates medical treatment alongside other HRQoL attributes such as general, emotional and mental health. 19These PRO measures incorporated disease specific questions focused on psychological, social and physical issues that can impact a patient's day to day life.
CushingQoL was the most commonly utilized PRO measure (Table 2).Using this measure, for which higher scores are associated with improved overall QoL, mean total scores ranged from 27 to 71 (out of 100). 22,23,26,29,30,37,39,42,43CushingQoL scores improved in patients with remission, post-surgery or post-pharmacotherapy compared to those with active disease, pre-surgery and pre-pharmacotherapy, respectively.There was an overall improvement in total CushingQoL scores that cannot be attributed to a singular domain.However, this improvement was not always consistent. 23,37,42ere was an increase in mean CushingQoL scores from baseline to follow-up after treatment with levoketoconazole. 23Of the 60 patients evaluated in this study, 40% met or exceeded the minimally important difference (MID) of 10.1 points for CushingQoL at Month 3, and 47% of the 51 patients met the MID at Month 6. 23 Additionally, one study (3%) reported that mean CushingQoL scores increased from baseline by 67% at Week 12, to 82.3% at Week 24, and 34.4% at Week 48, exceeding the minimal important difference (MID) threshold after treatment with pasireotide. 22

| Depression and anxiety
Results of 8 (26%) studies reporting depression or anxiety using PRO measures are summarized in Table 3. 14,23,33,38,39,41,46,48 One study (9%) reported mean depression scores (± standard deviation (SD)) using the Minnesota Multiphasic Personality Inventory (MMPI-II), a self-reported psychometric questionnaire, for patients with CD and found that both preoperative (59.7 ± 5.0) and 12 months postoperative (63.9 ± 3.6) scores were below the threshold (≥65) indicating a clinical diagnosis of depression. 48ing the Hospital Anxiety and Depression Scale (HAD/HADS-D), a 21-point scale in which scores of 11 or higher indicate depression and anxiety, 1 (9%) study reported mean depression and anxiety scores (5.5 and 7) among patients with CD who underwent pituitary surgery. 46Similarly, using the Freiburg questionnaire on the use of coping strategies (FKV-LIS), a scale with scores ranging from 1 to 5 where higher scores indicate more effective use of coping strategies, another study (9%) reported a mean score of 2.3 among patients with CD post pituitary surgery. 39,58Of the 10 (27%) studies that evaluated depression or anxiety, 5 (50%) studies reported depression using Beck Depression Inventory (BDI I/ II), a depression-specific PRO measure that evaluates the severity of depressive symptoms on a scale ranging from 0 to 63, where higher scores indicate more severe depression and a mean change of ≥3 BDI-II points indicates a clinically meaningful change in depression due to the treatment. 23,33,38,41,48,59,60In a single study (10%) reporting scores for patients with active CS at baseline and F I G U R E 1 PRISMA diagram.PRISMA: Preferred reporting items for systematic reviews and meta-analyses.
was only confirmed in one of the two studies. 38Another study (13%) reported HRQoL outcomes pre-and post-pharmacotherapy (pasireotide) using CushingQoL.Mean HRQoL scores improved from baseline through Week 12 and were sustained until Week 24 but declined at Week 48 compared to baseline.These changes in the CushingQoL scores exceeded the MID threshold. 22

| Other types of HRQoL burden
Fatigue was reported in 1 (3%) study using the Mental Fatigue Scale (MFS-Score) PRO measure and in another study using qualitative reporting by patients with CS. 36 The MFS questionnaire response set consists of 4 alternatives where a rating of 0 reflects normal function, 1 indicates a problem, 2 indicates pronounced symptom and 3 indicates a maximal symptom. 36One study (3%) using MFS-Score reported statistically significant differences between patients with CD in remission and healthy controls for fatigue (1.25 vs. 0.5), mental fatigue (1.25 vs. 0.5), mental recovery (1 vs. 0.5), irritability (0.8 vs. 0.5), light sensitivity (0.5 vs. 0.25) and decreased sleep at night (1 vs. 0.5).A second study reported that patients with diagnosed CD, either active or in remission, experienced both physical and mental fatigue.This corroborates the findings of the first study suggesting patients with CD experience more fatigue than healthy controls and highlights the unmet clinical need in these patients. 12

| Economic burden
Five studies (14%) reported on economic burden for patients with CS, including 3 (60%) studies reporting direct economic burden (including costs directly related to healthcare expenditure, e.g.medical costs, pharmacy costs and inpatient hospital costs) and 2 (40%) studies reporting both direct and indirect burden (including direct costs and costs that are not directly related to healthcare, e.g.,27 One study (20%) reported that the all-cause, total per patient per month (PPPM) cost for a CD patient was $3232 as measured by administrative claims from a large health insurance database.The majority of PPPM costs ($2800) were direct medical costs including inpatient hospital stays, outpatient and emergency room visits, and ambulatory services use. 17Two studies (40%) reported that the mean total annual healthcare system costs associated with CD were $26,269 and $34,992, respectively (SD not reported), with non-pharmacy costs making up most of these costs ($21,704 and $31,395).In line with a previously published SLR, this review found substantial differences between the overall HRQoL of healthy individuals and patients with CS in remission. 51The mean EQ-5D VAS score (range of 0-100, with higher score correlating to improved HRQoL) for patients with CS was 54 at baseline, 66 at first postoperative visit and 69 in longer-term remission. 42The range of mean normative EQ-5D VAS scores (75-84) reported across European countries (including Belgium, Denmark, France, Germany, Greece, Hungary, Italy, Netherlands, Slovenia, Spain and Sweden) indicated that endogenous CS patients continue to have worse HRQoL scores compared to healthy individuals, suggestive of long-term effects of CS. 43,61 Similarly, CS had a substantial adverse impact on the physical function dimension in patients, relative to UK population norms. 62is study reported key findings on HRQoL and mood preand post-treatment.One study, focused on CD patients who did not achieve remission, reported a worsened depression score 1-year post-transsphenoidal surgery. 46A second study examined CD patients in remission, and reported worse depression scores among CD patients an average of 8.4 years following bilateral adrenalectomy, compared to CD patients in remission for an average of 2.7 years who had not undergone bilateral adrenalectomy. 36wever, of the eight studies that evaluated pre-and post-surgery outcomes, six studies demonstrated an overall HRQoL improvement post-surgery. 18,19,37,38,43,484][65] Additionally, comorbidities impacted overall QoL for CS patients even post-treatment. 66A clinically relevant aspect of the patient's treatment experience is the onset of cortisol withdrawal symptoms following rapid decreases in circulating cortisol levels after prolonged exposure.[65] While several studies in the past 5 years reported HRQoL preand post-surgery, no study has reported HRQoL outcomes pre-and post-radiotherapy and few studies report on the progression of HRQoL pre-and post-pharmacological interventions.CushingQoL scores were found to exceed the MID threshold of ≥10.1 from baseline to follow-up after pharmacotherapy with levoketoconazole and pasireotide, suggesting improvements in QoL with treatment. 22,23tably, compared to patients with active disease, QoL was found to be better in patients in remission across all interventions. 37However, both groups of patients with CS still had worse HRQoL than their counterparts without CS. 37This confirms previous key findings that patients with active CS have substantially lower QoL scores than those in remission, regardless of sex, age or time in remission, and emphasizes the importance of attaining biochemical control of cortisol production.Further research is needed to understand the benefits and limitations of using pharmacotherapy for cortisol lowering, and to identify those who are most likely to benefit from it.
This study also aimed to describe sex differences in the presentation of disease burden.The prevalence of CS is higher in female compared to male patients.The studies reporting HRQoL and sex association data demonstrate worse HRQoL in females compared to males within subdomains of SF-36 scores, specifically physical role functioning, emotional role functioning, vitality and mental health. 35,39However, conclusions about sex differences in HRQoL are limited due to small study sample sizes and thus additional research is needed to improve our understanding of sex specific differences in HRQoL in CS, and to better describe the burden in male patients. 35,39tients with CS not only experience a significant HRQoL burden throughout their patient journey, but they also experience economic burden at individual and systemic levels.Although previous SLRs have examined various aspects of the burden of CS, few have described the economic burden of the disease and none have highlighted societal costs related to lost work or decreased productivity due to impairment.ing the high cost burden of disease in patients with CS. 67 The limited availability of economic data coupled with limited information on how the existing economic evidence was compiled, highlights the lack of a holistic economic burden analysis.Future economic burden studies should not only evaluate the annual long-term economic burden incurred by patients, but also how that economic burden is impacted by delays in diagnosis and impairments in long-term HRQoL, despite successful treatment.

| Limitations of this SLR
This SLR has several limitations to note.In most of the included studies, the analysis was conducted only in patients with CS and the studies did not include a normative population comparator.Published manuscripts and conference abstracts included in this review were given equal weighting in the qualitative synthesis.
Furthermore, studies published on the drug osilodrostat are not captured in this study.Though the drug was approved by the FDA in March 2020, studies reporting on the impact of osilodrostat on patients' disease burden were not published by the search date.It will be important for future SLRs capturing the disease burden of CS, to include studies that document the impact of osilodrostat on the CS treatment paradigm as indicated.Not all included studies reported total scores for HRQoL-related metrics; some only reported sub-domain scores.Therefore, our estimated averages of total scores are from a subset of studies and may not represent the overall CS patient population.Additionally, studies that did not report quantitative results for either HRQoL or economic burden were not included.Due to the limited number of studies reporting pre-and Peer-reviewed manuscripts published from 2015 to the search date (December 4, 2020) and conference abstracts and posters (subject to data availability and indexed via embase.com) published in the 2 years prior to search date were eligible for inclusion.Only studies published in English were included.Studies focusing on endogenous CS and CD were included and studies focusing on exogenous CS, or any other conditions were excluded from the SLR.Studies eligible for inclusion were primary research studies, economic analyses and models.Previously published SLRs meeting the eligibility criteria were included for reference purposes only.There were no geographical limits on the search.Details of eligibility criteria based on population, intervention, comparator, outcomes and study design criteria (PICOS) are described in Table
23e mean scores were 17.1 and 12.5, respectively, indicating the presence of minimal depression but a meaningful improvement in depression scores post-treatment given that they exceeded the MID threshold.23Inanother study (10%), the mean total BDI-II score reported in patients with CS in remission (11.3 ± 10.2) was General HRQoL assessed the patients' perceived physical and mental health over time using validated generic (not specific to endogenous CS) PRO measures.Of the 31 (84%) studies that reported HRQoL, 21 (68%) reported general HRQoL using SF questionnaires (SF-36, SF-HLQ, SF12v2Score) (n = 15; 71%), EuroQoL version 5D (EQ-5D) (n = 4;19%), Work Role Functioning Studies reporting CS specific HRQoL on CushingQoL.Table includes data for only selected studies that used CushingQoL.
TA B L E 2Abbreviations: CD, Cushing's disease, CS, Cushing's syndrome, NR, not reported; SD, standard deviation.a After diagnosis of CS. b Mean % change from baseline.c Mean change from baseline.Note: Studies measuring depression or anxiety using PROs.
It is possible that other studies of interest have been published as the search date for this SLR.In accordance with the published literature, our findings demonstrate that patients with endogenous CS experience a substantial, complex and multisymptomatic HRQoL burden.Even though the HRQoL of patients with CS improves post-treatment, it does not normalize and may 15,164 | DISCUSS IONTo the authors' knowledge, this systematic review is the most up-to-date and comprehensive qualitative synthesis of scientific evidence published on both the HRQoL and economic burden of patients with CS.