Sleep disruption in patients with active and treated endogenous Cushing’s syndrome

Abstract

Context

The hypothalamic-pituitary-adrenal axis is a critical regulator of circadian rhythm in humans. Impaired sleep adversely affects metabolic, emotional, and cognitive health.

Objective

To characterize sleep disturbances in patients with active and treated Cushing’s syndrome (CS), and identify factors associated with impaired sleep in treated patients.

Design

Single-center cross-sectional study.

Methods

Patients with pituitary or adrenal CS enrolled in an observational study completed Nottingham Health Profile (NHP), CushingQoL, and Hospital Anxiety and Depression assessments. Cross-sectional analysis was conducted including patients with active and treated disease.

Results

113 (94 female) patients with CS were included, 104 pituitary and 9 adrenal, with mean age at diagnosis of 43.9 ± 13.4 years. Mean and maximum duration of follow up was 5.1 and 23 years. Mean NHP sleep score was lower (i.e., improved) in patients with treated vs. active disease (29.6 ± 30.2 vs. 51.9 ± 30.9, p = 0.0005), as was CushingQoL sleep score (p = 0.015), but 41.5% of patients with treated disease stated they often or always had trouble sleeping. The proportion of treated vs. active patients taking medication for sleep, mood, or pain was not different. Neither NHP nor CushingQoL pain scores were lower in treated vs. active patients (p = 0.39 and 0.53). In patients with treated CS, anxiety and depression correlated with worse sleep scores.

Conclusions

Patients with treated CS report improved sleep quality compared to those with active disease, but almost half of treated patients still report sleep challenges. The need for sleep medications, reported by one third of patients, was not different after CS treatment. Ongoing mood disturbances may play a role in persistent sleep disruption. Further work should focus on determinants of sleep impairments in treated CS patients.

Introduction

The hypothalamic pituitary adrenal (HPA) axis is a critical regulator of circadian rhythm in humans. Insomnia and disrupted sleep are reported by over 50% of patients with hypercortisolemia due to endogenous Cushing’s syndrome (CS), a rare disorder characterized by excess adrenal glucocorticoid production caused by a pituitary, ectopic, or adrenal tumor [1, 2]. Sleep disturbances can range from insomnia disorder, characterized by chronic difficulty initiating and maintaining sleep despite adequate opportunity, to more general sleep disturbances, poor sleep quality, and/or unrefreshing sleep. Patients with CS also experience other significant morbidities including anxiety, diabetes, obesity, and impaired quality of life [3–7]. Normalization of cortisol levels does not always result in complete remission of clinical, metabolic, and neuropsychiatric abnormalities, and long-term quality of life remains impaired [3, 8–13].

A defining feature of CS is absence of the HPA axis circadian rhythm, which is normally characterized by high circulating cortisol levels in the early morning, with decreases in the afternoon and evening, reaching a nadir around midnight [14]. Cortisol is a key circadian clock zeitgeber (“time-giver”) for the body [15] and the HPA axis has a bidirectional relationship with sleep, with the cortisol nadir occurring during the first half of the night when deep sleep predominates. As CS is associated not only with hypercortisolemia, but also with a flattened circadian rhythm, circulating cortisol levels are abnormally elevated at night [14]. Available data suggest that people with CS experience reduced slow wave sleep, increased sleep latency, greater wake time, and sleep fragmentation [16–18].

Research characterizing the extent and quality of sleep disturbances in CS has been limited, and it is not clear if successful treatment can reverse or even improve impaired sleep in CS patients [3, 7, 10–12]. Given the known associations between insomnia, neurocognitive impairment, and metabolic dysregulation, sleep disturbance over time may worsen the neuropsychological and body composition abnormalities that are well documented in treated CS patients [12, 19–23]. Sleep disturbance may also contribute to elevated cardiovascular and mortality risk [6, 24–26].

The aims of the current study were to characterize sleep disturbances in patients with chronic hypercortisolemia due to CS, quantify differences in patient reported sleep outcomes in active versus treated patients, and identify factors associated with impaired sleep in treated CS patients. We examined patient-reported sleep outcomes in a cross-sectional cohort of patients with endogenous pituitary and adrenal CS at different stages in their treatment journey. Based on cortisol’s critical role in maintaining circadian rhythm, we hypothesized that sleep quality and quantity would improve after successful treatment of CS.

Methods

Study design

The current analysis includes cross-sectional data from patients who are enrolled in an ongoing prospective cohort study including patients with endogenous CS at Memorial Sloan Kettering Cancer Center (MSKCC) starting in 2017. In this protocol, CS patients being treated at the MSK Pituitary and Skull Base Tumor Center are enrolled at any time during their treatment journey, and assessed, when available, at treatment stage 1 (active disease, including initial presentation or recurrent disease), treatment stage 2 (6 months after surgical remission or after achieving medical control), treatment stage 3 (6 months after discontinuing post-operative glucocorticoid replacement or approximately 1–1.5 years after achieving endocrine control on medical therapy), and every 5 years thereafter. The above time points for treatment stage 2 and 3 were based on an effort to reassess patients according to the recovery of their HPA axis rather than at a specific chronological time, given that HPA axis recovery is variable in CS patients after surgical remission, and time to achieve endocrine control on medical therapy is also variable. Chronological time since surgery or initiation of medical therapy at each treatment stage therefore varied by patient.

ACTH-dependent pituitary and ACTH-independent adrenal Cushing’s was confirmed according to Endocrine Society guidelines [14]. Criteria for remission were based on Endocrine Society consensus statement and recent guidelines [27, 28]. Stage 3 study visits included patients who had achieved surgical remission, or had achieved endocrine control, as defined by a normal 24 h UFC on medical therapy, and based on clinical review and assessment by EBG. For the purposes of this study, all visits treatment stage 3 and beyond were included in the treatment stage 3 category.

At each study visit, patients’ clinical and biochemical data were collected according to standard of care. Patients completed the Nottingham Health Profile (NHP), CushingQoL, and Hospital Anxiety and Depression (HADS) assessments.

In the current study, cross-sectional analysis of 113 patients (104 pituitary and 9 adrenal) was conducted, including patients at any treatment stage.

The study was approved by the Institutional Review Board at Memorial Sloan Kettering Cancer Center. All subjects gave written informed consent before participation.

Outcome measurements

Nottingham health profile (NHP)

The NHP evaluates general health and QoL in patients. This assessment consists of 38 yes/no questions that are subdivided into six subscales, which assess impairments: pain (eight items), energy level (three items), sleep (five items), emotional reaction (nine items), social isolation (five items) and physical mobility (eight items). Each question is assigned a weighted value; the sum of all weighted values in a given subarea adds up to 100. A higher score indicates greater impairment [29]. The five sleep questions are: ‘I take pills to help me sleep,’ ‘I’m waking up in the early hours of the morning,’ ‘I lie awake for most of the night,’ ‘It takes me a long time to get to sleep,’ and ‘I sleep badly at night.’

CushingQol

The CushingQol is a validated disease-specific questionnaire consisting of 12 questions in a five-point Likert scale ranging from always to never (for 10 questions) or very much to not at all (for two questions). The total score ranges from 12 to 60, with a lower score indicating greater negative impact on health-related QoL. This is converted to a 0-100 scale, 0 indicating worst and 100 indicating the best QoL [30]. The CushingQoL has a single question assessing the frequency of trouble sleeping with response items of Always, Often, Sometimes, Rarely, and Never.

Hospital anxiety and depression score (HADS)

The HADS consists of 14 items pertaining to anxiety and depression (seven items each), with each item measured on a four-point scale, scored from 0 to 3. Total scores for the anxiety and depression subscales range from 0 to 21. A higher score indicates greater symptomatology [31]. The reliability, validity, and factor structure of the HADS has been established in a variety of clinical populations, including pituitary Cushing’s patients [32]. Scores of 11 or more on either subscale are considered abnormal (i.e. significant morbidity), scores of 8–10 represent “borderline,” and 0–7, “normal.” Prior work has grouped borderline and abnormal together [33].

Clinical variables

Detailed information was abstracted from patients’ medical records, including demographics (age, race, ethnicity), clinical and treatment variables including type of Cushing’s (ACTH-dependent or independent); comorbidities, BMI, diagnostic test results, relevant medical therapy (including use of medications for anxiety/depression, sleep, pain, and Cushing’s) time since diagnosis, previous and current treatments (e.g., surgery, radiation therapy, medical therapy), confirmation of remission status, and Cushing’s testing at the time of each study visit (including plasma ACTH, serum cortisol, late night salivary cortisol, and 24 h urine free cortisol values, and dose of hydrocortisone replacement, if applicable). Hormone testing was assessed per clinical laboratory used, either MSKCC or local laboratory. MNSC values were reported as a mean of two values and recorded as abnormal or normal per assay used.

Statistical analysis

Descriptive statistics was used to assess demographic and clinical characteristics at baseline. For categorical baseline clinical characteristics, we used Pearson’s chi-squared test or Fisher’s exact test to compare among different treatment stages. For continuous clinical characteristics, we used Kruskal–Wallis test to compare among different treatment stages due to non-parametric distribution of the variables.

To examine symptom differences cross-sectionally by treatment stage, we used One-Way ANOVA for normally distributed symptom scores to determine if symptom scores vary by treatment stage. Kruskal–Wallis test was used for non-parametric symptom scores.

To explore the relationship between sleep and each individual comorbid symptom such as depression, anxiety and pain in treated (stage 3) patients, we used a linear regression model with sleep as the dependent variable and depression/anxiety/pain as independent variables separately, followed by multivariate analysis. All statistical tests were two-sided. Statistical significance was set at P < 0.05. All statistical analyses were conducted using STATA (version 15.0; STATA Corporation, College Station, TX).

Results

Study participants

Clinical characteristics

From a cohort of 169 endogenous Cushing’s and silent ACTH tumor patients enrolled in an ongoing MSKCC prospective cohort study between the dates of 2017–2022, 56 patients were excluded due to missing data, treatment stage 3 data not meeting criteria for remission due to elevated 24-hour urine free cortisol (UFC), or diagnosis of silent ACTH-secreting tumor or ectopic ACTH-dependent CS. The current study included 113 patients (104 pituitary and 9 adrenal) who were analyzed cross-sectionally. Patients at any treatment stage were included, with 67 treatment stage 1 (active disease) patients, 50 treatment stage 2 patients, and 65 treatment stage 3 and beyond patients included. Patient demographic information is shown in Table 1. Of the sample, 94 (83.2%) were female and 19 (16.8%) were male. Mean age at diagnosis was 43.9 ± 13.4 years.

Table 1.

Sociodemographic characteristics of participants (N = 113 unique patients)

	Total
	N	%
Age (Mean ± SD)
Treatment Stage 1: 43.9 ± 13.4	67
Treatment Stage 2: 42.4 ± 12.5	50
Treatment Stage 3: 45.9 ± 14.1	65
Sex
Male	19	16.8
Female	94	83.2
Race
White	86	77.5
Black	15	13.5
Asian	3	2.7
Other	7	6.3
Ethnicity
Hispanic	9	8.3
Non-Hispanic	100	91.7
Type of Cushing’s
Pituitary	104	92.0
Adrenal	9	8.0

Baseline and follow up clinical characteristics and comorbidities are shown in Table 2. Mean duration of time since diagnosis at follow up was 5.1 ± 5.2 years, with a maximum follow up of 23 years. At treatment stage 3, 63.5% of patients had achieved surgical remission due to transsphenoidal surgery, 19% had not achieved surgical remission, and 17.5% had not undergone transsphenoidal surgery (e.g., nonsurgical treatment or adrenal surgery if indicated). 85.5% of patients in treatment stage 3 were not being treated with a Cushing’s medical therapy (i.e., had achieved surgical remission) and 14.5% were taking medical therapy. Of note, eight patients in treatment stage 3 were taking hydrocortisone replacement due to hypopituitarism or primary adrenal insufficiency from prior bilateral adrenalectomy.

Table 2.

Clinical characteristics of participants (n = 113 unique patients)

	Treatment stage 1		Treatment stage 2		Treatment stage 3		P-value
	N	Mean±SD or %	N	Mean±SD or %	N	Mean ± SD or %
Diabetes							0.17
No	35	53.0	29	60.4	45	71.4
Yes	21	31.8	12	25.0	15	23.8
At risk	10	15.2	7	14.6	3	4.8
Hypertension							0.71
No	27	40.9	22	45.8	24	38.1
Yes	39	59.1	26	54.2	39	61.9
Heart disease							0.87
No	62	93.9	46	95.8	59	93.6
Yes	4	6.1	2	4.2	4	6.4
Prior adrenalectomy
No	66	100.0	45	93.8	55	87.3
Yes	0	0.0	3	6.2	8	12.7
Prior radiation
No	64	97	46	95.8	57	90.5
Yes	2	3	2	4.2	6	9.5
Surgical remissiona
No	10	14.9	9	18.4	12	19
Yes	2b	3.0	32	65.3	40	63.5
N/A	55	82.1	8	16.3	11	17.5
BMI	59	32.9 ± 6.2	46	30.3 ± 7.1	59	30.7 ± 8.4	0.026
Cushing untreated years (delay in diagnosis)	61	3.2 ± 4.2	46	4.3 ± 7.9	57	3.6 ± 5.1
Years since cushing diagnosis	62	1.0 ± 2.5	46	2.1 ± 3.0	59	5.1 ± 5.2
24 Hr free cortisol (mcg)	55	391.4 ± 1647.3 (4.9−12346)	–	–	34	19.7 ± 13.5 (2.5-58)	0.0001
Plasma ACTH (pg/mL)
Pituitary	49	68.7 ± 54.6 (5−332)	–	–	–	–
Adrenal	4	6.2 ± 2.5 (5−10)	–	–	–	–
Hydrocortisone replacement (mg)	0		30	18.0 ± 7.3 (5.0-40.0)	8	14.4 ± 6.8 (5-25)	0.21
Cushing's medication							0.008
No	65	98.5%	40	81.6%	53	85.5%
Yes	1	1.5%	9	18.4%	9	14.5%
Midnight salivary cortisol							< 0.001
Normal	3	7.3%		–	20	74.1%
Abnormal	38	92.7%	–	–	7	25.9%
Sleep medication							0.67
No	41	62.1%	33	67.4%	43	69.3%
Yes	25	37.9%	16	32.6%	19	30.6%
Anxiety/depression medication							0.48
No	46	69.7%	39	79.6%	45	72.6%
Yes	20	30.3%	10	20.4%	17	27.4%
Pain medication							0.41
No	48	72.7%	30	61.2%	43	69.4%
Yes	18	27.3%	19	38.8%	19	30.6%

^aIncludes data from all patients but refers to remission from pituitary surgery specifically, i.e. adrenal patients are N/A

^bPrior surgical remission with subsequent recurrence

Twenty seven patients at treatment stage 3 had available midnight salivary cortisol (MNSC) values, and of these, 7 patients (25.9%) had abnormal MNSC values, despite achieving a normal 24-hour UFC. In addition to achieving normal 24-hour UFC, these 7 patients were determined to be in remission/endocrine control based on review of biochemical testing and clinical presentation (including comorbidities and symptoms) by EBG.

Comorbid conditions

Mean BMI at treatment stage 1 (active disease) was 32.9 ± 6.2, compared to 30.7 ± 8.4 at treatment stage 3 (remission/medical control, p = 0.026). The proportion of patients with diabetes at treatment stage 1 was 31.8%, compared to 23.8% of patients at treatment stage 3 (p = 0.17). 59.1% of patients had hypertension at treatment stage 1 compared to 61.9% of patients at treatment stage 3 (p = 0.71).

Patient reported sleep quality and quantity

Table 3 shows sleep scores and comorbid conditions by treatment stage. Out of a range from 0 to 100, mean NHP sleep score in patients with active disease (stage 1) was 51.9 ± 30.9 compared to 29.6 ± 30.2 with treated disease (stage 3), with a higher score indicating worse symptoms. NHP sleep score was significantly lower in patients after treatment (p = 0.0005), indicating less severe symptoms (Fig. 1).

Table 3.

Sleep scores at each treatment stage

	Treatment stage 1		Treatment stage 2		Treatment stage 3		P-value
	N	Mean±SD or %	N	Mean±SD or %	N	Mean±SD or %
Nottingham Health Profile
Sleep	59	51.9 ± 30.9	44	35.8 ± 31.6	58	29.6 ± 30.2	0.0005
CushingQoL questionnaire
Have trouble sleeping							0.015*
Always	24	38.1%	11	23.4%	9	14.5%
Often	17	27.0%	9	19.2%	17	27.4%
Sometimes	15	23.8%	15	31.9%	19	30.6%
Rarely	4	6.4%	9	19.2%	11	17.7%
Never	3	4.8%	3	6.4%	6	9.7%
'never' = no

*p-value for stage 3 vs. stage 1, using dichotomized values, with 'always' and 'often' = yes, and 'sometimes,' 'rarely,' 

Fig. 1.

NHP sleep and pain scores at each treatment stage: Stage 1 (active disease); Stage 2 (~ 6 months after treatment); Stage 3 (at least 6 months after discontinuation of post-operative glucocorticoid replacement). Mean NHP sleep but not pain scores were significantly lower (i.e., improved) after treatment, p = 0.0005 and 0.39, respectively

Per CushingQoL, 65.1% of patients with active disease stated they often or always had trouble sleeping, compared to 41.9% of treated patients (p = 0.015, Fig. 2).

Fig. 2.

Sleep and pain assessments per CushingQoL at each treatment stage. 65.1% of patients with active disease stated they often or always had trouble sleeping, compared to 41.9% of treated patients (p = 0.015). 34.9% of patients with active disease reported that they were often or always in pain, compared to 25.8% of treated patients (p = 0.53)

As shown in Fig. 3, the proportion of patients taking a medication for sleep did not decrease with treatment: 37.9% of patients in with active disease took a medication for sleep, compared to 30.6% after treatment (stage 3, p = 0.67).

Fig. 3.

Percent of patients requiring medication for sleep, anxiety/depression or pain at each treatment stage.37.9% of patients in with active disease (stage 1) took a medication for sleep, compared to 30.6% after treatment (stage 3, p = 0.67); 30.3% of patients with active disease took a medication for anxiety or depression, compared to 27.4% after treatment (p = 0.48); and 27.3% of patients with active disease took a medication for pain compared to 30.6% after treatment (p = 0.41). Use of all three medication categories did not change with treatment

Comorbid conditions: anxiety, depression, pain

As shown in Table 4, mean HADS anxiety and depression scores in treatment stage 1 (active disease) were 9.6 ± 4.5 and 7.6 ± 4.9, respectively, both considered borderline abnormal (scores > 7). The mean of both scores were significantly lower at treatment state 3 (remission/biochemical control), 6.5 ± 4.4 (anxiety) and 4.8 ± 4.3 (depression), p = 0.0006 and p = 0.0029, respectively, indicating improvements in anxiety in depression with treatment (Fig. 4).

Table 4.

Comorbid symptom scores at each treatment stage

	Treatment stage 1		Treatment stage 2		Treatment stage 3		P-value
	N	Mean±SD or %	N	Mean±SD or %	N	Mean±SD or %
Hospital anxiety and depression scale
Anxiety	60	9.6 ± 4.5	44	7.1 ± 4.4	58	6.5 ± 4.4	0.0006
Depression	60	7.6 ± 4.9	44	5.4 ± 4.7	58	4.8 ± 4.3	0.0029
CushingQoL questionnaire irritable							<0.001*
Always	9	14.5%	2	4.3%	1	1.6%
Often	18	29.0%	5	10.9%	9	14.5%
Sometimes	23	37.1%	16	34.8%	18	29.0%
Rarely	5	8.1%	19	41.3%	21	33.9%
Never	7	11.3%	4	8.7%	13	21.0%
Pain							0.53*
Always	5	7.9%	5	10.6%	7	11.3%
Often	17	27.0%	10	21.3%	9	14.5%
Sometimes	18	28.6%	12	25.5%	10	16.1%
Rarely	10	15.9%	12	25.5%	15	24.2%
Never	13	20.6%	8	17.0%	21	33.9%
Nottingham health profile
Emotional reaction	59	40.9 ± 28.9	43	20.2 ± 24.5	57	15.9 ± 22.9	< 0.001
Social isolation	59	28.3 ± 29.3	44	15.0 ± 20.6	58	14.7 ± 24.9	0.0068
Pain	59	27.7 ± 30.2	44	32.6 ± 34.6	58	24.1 ± 28.4	0.39
Physical abilities	59	28.1 ± 24.4	44	20.0 ± 18.8	58	16.8 ± 19.8	0.015
Energy level	59	62.8 ± 38.0	44	46.3 ± 39.5	58	28.6 ± 36.1	< 0.001
'never' = no

*p-value for stage 3 vs. stage 1, using dichotomized values, with 'always' and 'often' = yes, and 'sometimes,' 'rarely,' 

Fig. 4.

HADS depression and anxiety scores at each treatment stage: Stage 1 (active disease); Stage 2 (~ 6 months after treatment); Stage 3 (at least 6 months after discontinuation of post-operative glucocorticoid replacement). Mean anxiety and depression scores were significantly lower (i.e., improved) after treatment, p = 0.0006 and p = 0.0029, respectively

Reported irritability per CushingQoL improved with treatment. 43.5% of patients with active disease reported often or always experiencing irritability, compared to 16.1% of treated (stage 3) patients, p < 0.001 (Table 4).

Mean pain scores did not improve with treatment, as measured by NHP and CushingQoL. Mean NHP pain score in active disease was 27.7 ± 30.2, compared to 24.1 ± 28.4 after treatment (stage 3), p = 0.39 (Fig. 1). Per CushingQoL, in active disease, 34.9% of patients reported that they were often or always in pain, compared to 25.8% of patients after treatment (stage 3), p = 0.53 (Fig. 2).

As shown in Fig. 3, the proportion of patients taking a medication for anxiety/depression or pain did not decrease with treatment: 30.3% of patients with active disease took a medication for anxiety or depression, compared to 27.4% after treatment (stage 3, p = 0.48); and 27.3% of patients with active disease took a medication for pain compared to 30.6% after treatment (stage 3, p = 0.41).

Factors associated with impaired sleep in treated (stage 3) patients

Comorbidities

In bivariate linear regression analyses, anxiety, depression, and pain were individually found to correlate with worse sleep scores in patients with treated CS (N = 58). After multivariate analysis, HADS anxiety and depression, but not pain, remained significantly correlated with NHP sleep scores in treated patients. With one point increase in HADS depression score, NHP sleep score increased 2.15 points (coefficient = 2.15, p = 0.037, 95% CI 0.14–4.16) indicating that worse sleep correlated with higher depression scores. With one point increase in HADS anxiety score, NHP sleep score increased 1.95 points (coefficient = 1.95, p = 0.035, 95% CI 0.14–3.75), indicating worse sleep correlated with higher anxiety scores.

No significant correlations were identified between sleep and age, BMI, presence of diabetes or hypertension.

Midnight salivary cortisol values

Treated patients who had abnormal MNSC values tended to have higher NHP sleep scores compared to those with normal MNSC values (62.6 ± 43.4 vs. 30.0 ± 28.15) but the difference did not reach statistical significance (p = 0.07).

Use of medications

Treated patients who used sleeping medications had higher NHP sleep scores compared to treated patients not on sleeping medications (52.7 vs. 20.2, p = 0.0001), with higher scores indicating greater sleep impairment. Similarly, treated patients taking pain medications had higher NHP sleep scores compared to those not taking pain medications (43.4 vs. 24.2, p = 0.029). The use of medications for anxiety or depression was not correlated with NHP sleep scores in treated patients (mean NHP sleep score in patients taking anxiety/depression medications was 38.3, compared to 27.0, p = 0.23).

Discussion

The current study aimed to characterize sleep disturbances in patients with active and treated CS, and identify factors associated with impaired sleep in treated CS patients. Analysis of patient reported outcomes from a cohort of 113 patients with endogenous CS indicates that, while treatment improves self-reported sleep quality, nearly half (41.9%) of treated patients continue to report often or always having trouble sleeping. We also found that the proportion of patients taking a medication for sleep, mood, or pain did not change after treatment, and in treated CS patients (i.e., those who had achieved surgical remission or biochemical control with medical therapy), anxiety and depression were associated with worse sleep scores.

Insomnia and comorbid conditions including anxiety and fatigue are among the most common and distressing symptoms for patients living with CS [21]. These co-occurring symptoms also negatively impact patients’ quality of life and functional performance, resulting in ongoing economic impact and utilization of healthcare resources [34, 35]. Whether sleep deficits, reflected by assessment of sleep architecture or by patient report, improve with treatment of CS has not been previously established. While the proportion of patients reporting that they often or always had trouble sleeping decreased with treatment (65.1% versus 41.9%), the ongoing sleep disturbance in nearly half of patients despite treatment may contribute to poor overall long-term health-related quality of life (HRQOL) [36].

Our data support recent findings from a nationwide Swedish longitudinal registry study showing that use of sleeping pills among treated CS patients remained the same as at diagnosis (22%) even after long-term sustained remission [37]. Sleeping pill use was higher in our cohort, and usage did not change even after long term remission (37.9% with active disease and 30.6% after treatment reported taking sleeping pills, P = 0.67). While there is evidence that some sleeping aids have long-term efficacy [38], continued use of sedative-hypnotics has been associated with an increased risk of overall mortality [38]. Our findings indicate ongoing sleep impairments in patients with treated CS, and serve as a reminder for providers to regularly ask patients about their use of sleep aids, including over the counter medications and supplements, to allow for a comprehensive assessment of ongoing symptoms.

A striking finding in the current data was the lack of improvement in pain scores, as assessed by NHP and CushingQol, in treated compared to active disease. This contrasts with significant improvements in depression and anxiety scores. These findings are consistent with our recent cross-sectional study of 55 patients with treated CS who completed 5 validated PROMS, which found that anxiety and pain scores did not decrease with treatment [34]. We found that almost 70% of patients reported moderate or severe fatigue, and almost 90% reported taking over the counter analgesics for pain. This study was limited by potential recall bias, since patients were asked to report on their symptoms, as they recollect, at diagnosis compared to the present moment, and lack of biochemical confirmation of current remission status [34]. Both of these limitations have been addressed in the current study, as patients at each time point in their treatment journey were asked to report on their current symptoms, and remission/endocrine control was established with biochemical testing. Our data indicate that pain is a clinically relevant symptom that patients with CS experience over time despite treatment. Prevalence, severity, and predictors of pain need more investigation in active and treated CS.

Using multivariate analysis, we found correlations between impaired sleep with anxiety and depression scores in patients with treated disease. Similarly, treated patients who required sleep and pain medications had worse sleep scores compared to those who did not require these medications, indicating a need for multidisciplinary assessment of these ongoing comorbidities in patients even after endocrine remission. It is currently not clear if ongoing HPA axis dysregulation contributes to impaired sleep and mood in treated patients, or if ongoing pain, anxiety, and depression impact sleep quantity and quality. Further work should clarify these relationships, as well as possible differences between surgical remission and endocrine control from medical treatment, as well as the potential role of glucocorticoid withdrawal syndrome in the symptom burden among treated patients [39].

Additional predictors of impaired sleep in patients with treated CS remain unknown. Our data failed to show correlations with demographics, treatment type, metabolic/CVD comorbidities (e.g., BMI, hypertension, diabetes), or age. We also did not find an association between abnormal MNSC values in treated patients and impaired sleep, although the relationship did approach statistical significance (p = 0.07). This negative finding may be due to insufficient patient sample size. Notably, 25% of our treated patients had abnormal MNSC values, despite achieving normal 24 h UFC. As restoration of normal circadian rhythm in treated CS patients is of critical clinical importance, further work should investigate MNSC values in a larger cohort and with objective sleep assessments.

There are several limitations to the current study, including cross-sectional analysis of patients at different stages in their treatment journey. Longitudinal analysis is needed to quantify the proportion of patients who experience a clinically relevant change in sleep and comorbid conditions, by quantifying the minimal clinically important difference (MID) for each PROM. In addition, while the PROMs used were generic and disease specific, they were not specific for sleep, limiting our ability to identify clinically relevant insomnia scores. The inclusion of patients who have achieved surgical remission and those with endocrine control from medical therapy may be considered a limitation, as this could introduce confounders such as the potential (but not yet clarified) effect of CS medications on sleep [40], but analyzing a ‘real world’ cohort also strengthens the generalizability of the findings. Finally, the clinical data included, while extensive, were extracted from patient charts and therefore some data points are missing and assay methodologies may vary.

In summary, our study provides patient reported data on sleep impairments that active and treated patients with CS experience, and the relationships between sleep, mood, and pain. We found that some measures of sleep improve with treatment, but nearly half of treated patients report ongoing sleep challenges, and one third report taking sleeping pills. Disrupted sleep may be related to ongoing pain, depression, and anxiety. Further studies should prospectively quantify sleep quality using objective as well as subjective assessments, use validated PROMs specific for sleep, and identify clinically relevant predictors of sleep disruption among patients with CS. These data are needed to develop effective management strategies for insomnia and comorbid conditions, with an overall goal to lessen the long-term burden of CS that exists even after successful treatment.

Author contributions

EBG designed the study and wrote the main manuscript text. IG, HR, VC, HB, CG and MS contributed to conducting study visits and entering data. QL prepared the figures. All authors reviewed the manuscript.

Funding

This research was funded by NIH/NCI Cancer Center Support Grant P30 CA008748.

Data availability

Data is provided within the manuscript, tables and figures.

Declarations

Conflict of interest

EBG reports serving as an editor for Pituitary and is an investigator for research grants to MSKCC from Recordati, Sparrow, and Xeris, and serving as an occasional consultant to HRA Pharma, Sparrow, Xeris, and Lundbeck, none of which conflict with the current work. No other authors report a conflict.