Sleep Disturbances based on Patient Reported Outcomes in Patients with Breast Cancer

Abstract

Purpose.

Sleep disturbances are common in patients with breast cancer, but comprehensive evaluations with patient-reported outcomes (PRO) and sleep evaluation with polysomnography (PSG) are lacking. This study describes sleep disruption using PROs and PSG to identify underlying sleep disorders.

Methods.

A retrospective review of patients with breast cancer undergoing formal sleep evaluation from 4/1/2009 to 7/31/2014 was performed. Clinical characteristics, PROs using Pittsburgh Sleep Quality Index (PSQI) and Epworth Sleepiness Scale (ESS), and PSG data were reviewed.

Results.

404 patients were identified with 43% early, 30% locally advanced and 17% metastatic disease. PSQI revealed poor sleep in 75%, and ESS demonstrated daytime sleepiness in 55%. Sleep aid use was reported by 39%, and pain medication use in 22%. Most patients (50.2%) had multiple sleep disorders. Insomnia (54.5%) was the most frequent, followed closely by obstructive sleep apnea (OSA) (53.7%). PSG was performed in 74%. Multivariate analysis linked poor sleep to use of sleep aids [OR 7.7, 95% CI 3.9 to 15.2], anxiety disorder [OR 4.8, 95% CI 1.7 to 14.0], and metastatic disease [OR 2.8, 95% CI 1.1 to 6.6]. Daytime sleepiness correlated with known diagnosis of OSA [OR 1.9, 95% CI 1.0 to 3.3] and sleep aid use [OR 0.6, 95% CI 0.4 to 0.9].

Conclusions.

Poor sleep was associated with sleep aid use, anxiety disorder and metastatic disease. Insomnia was the most common, followed by OSA mostly mild. Education about sleep health and proactive screening for sleep symptoms would be beneficial in patients with breast cancer.

Introduction

Sleep disruption poses a pervasive challenge for patients with breast cancer, and estimates of poor sleep ranges from 20% to 70%¹. Breast cancer accounts for 30% of new cancer diagnoses in women in the United States, with a 90% 5-year survival rate owing to advancements in detection and treatments². Sleep symptoms in women may vary with hormonal changes, and both perimenopausal and early menopause can be induced with cancer therapy. The risk of sleep-disordered breathing increases in post-menopausal women. Unfortunately, the deleterious impacts on sleep may linger throughout cancer treatment and even persist years beyond treatment completion^3–5. Although there has been significant data evaluating risk of breast cancer and sleep duration and sleep-related breathing disorder, there remains a paucity of data evaluating both screening of sleep disruption and sleep disorders in patients with breast cancer^6–9. The purpose of our study was to characterize sleep disturbances in patients with breast cancer using Patient Reported Outcomes (PROs). Our secondary objective was to identify risk factors in those with underlying sleep disorders including insomnia and obstructive sleep apnea (OSA).

Materials and Methods

We retrospectively reviewed all patients with breast cancer that underwent sleep consultation at our institution from April 1, 2009 to June 30, 2014. The study was approved by the Institutional Review Board (DR09–414). Study data were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture hosted at The University of Texas at M.D. Anderson Cancer Center¹⁰.

Clinical characteristics were collected and included: basic demographics, comorbidities, breast cancer data (date of diagnosis, menopausal status, tumor receptor status, clinical stratification, recent therapy), sleep symptoms, physical examination, surveys (Pittsburgh Sleep Quality Index, PSQI; Epworth Sleepiness Scale, ESS); polysomnography (PSG) parameters (total sleep time, sleep position, sleep efficiency, sleep stage, apneas, hypopneas, oxygen saturation nadir, positive pressure therapy), underlying sleep disorders and treatment of sleep disorder. Classification of early, locally advanced and metastatic disease status was based on clinical stage at the time of the sleep consult and reviewed by oncology collaborators. In those with multiple breast cancers, the active diagnosis at the time of sleep evaluation was selected. Menopausal status was determined at the time of cancer diagnosis as pre-menopausal or post-menopausal and followed clinical history and National Comprehensive Cancer Network (NCCN) guidelines¹¹.

PROs included sleep surveys which were obtained at the time of initial consultation. PSQI estimates sleep quality and disruption over the past month, and ESS assesses daytime sleepiness.^12,13 Sleep surveys were dichotomized as follows: PSQI, good sleep < 8, poor sleep ≥ 8; ESS, no daytime sleepiness < 10, daytime sleepiness ≥ 10. The higher cut-off of 8 is based on Carpenter and colleagues data that this may be more appropriate to determine poor sleep in clinical populations^14,15.

Sleep disorders were determined based on American Academy of Sleep Medicine (AASM) guidelines¹⁶. Insomnia disorder was confirmed when all four of the following criteria are met: symptoms including difficulty initiating sleep, difficulty maintaining sleep, or waking up too early; sleep difficulties occur despite adequate opportunity for sleep; daytime impairment attributable for sleep difficulties including fatigue, irritability, daytime sleepiness; sleep-wake difficulty not better explained by another sleep disorder. Sleep-related breathing disorders encompassed those characterized by abnormal respiratory patterns or abnormal hypoventilation during sleep. The presence or absence of OSA, central sleep apnea (CSA) and sleep-related hypoventilation was determined via PSG based on the AASM guidelines.¹⁷ The severity of OSA or CSA was rated as mild (5 to <15), moderate (15 to < 30) or severe (≥30). Sleep-related breathing disorder also includes those that did not meet criteria for OSA or CSA and included: 1) sleep-related hypoventilation with desaturations related to underlying metastatic disease or pulmonary disease; 2) patients with Respiratory-Event Related Arousals (RERAs) associated with daytime sleepiness, and an elevated Respiratory Disturbance Index (RDI, total number of apneas, hypopneas and RERA per hours of sleep) classified as upper airway resistance syndrome. Patients with significant sleep related breathing disorder underwent PSG with positive airway pressure (PAP) titration utilizing continuous positive airway pressure (CPAP), bi-level positive airway pressure (BPAP) and auto-servoventilation (ASV). Positive pressure was titrated to eliminate snoring and sleep disordered breathing. A split study was conducted when AHI ≥ 30 in the first two hours of sleep, so the second portion of the study was used to titrate positive pressure therapy. Adherence to PAP therapy included use of at least 4 hours per night 70% of the time¹⁸.

Polysomnography (PSG).

PSG was performed in those with concern for sleep-disordered breathing, excessive daytime sleepiness and/or periodic limb movement disorder. An all-night-attended, comprehensive PSG was performed using a computerized polygraph to monitor electroencephalogram (O2-A1, O2-A2, C3-A2, C4-A1), left and right electro-oculogram, electrocardiogram, chin and anterior tibialis electromyogram, abdominal and chest wall excursion using impedance plethysmography, airflow by nasal pressure and nasal and oral thermistors, and oxygen saturation (SaO₂) by pulse oximetry. Apnea was defined as cessation of breathing for at least 10 seconds, and hypopnea was defined as decreased effort to breathe at least 30% less than baseline and with at least a 4% decrease in SaO₂. A respiratory effort related arousals (RERAs) was characterized by increasing respiratory effort followed by an arousal from sleep, which does not meet the criteria for apnea or hypopnea.

Statistical Analysis.

Summary statistics were used to describe the general population and by hormonal therapy use in the last three months. T-test, Wilcoxon-rank sum test, chi-squared test, and Fisher’s exact test were used to compare characteristics by hormone therapy use. Logistic regression models were conducted to assess associations between characteristics and sleep disruption (PSQI ≥ 8), sleepiness (ESS ≥ 10), insomnia and OSA. Multivariable models were conducted including those characteristics with p values < 0.15 in the univariate setting. This study is considered exploratory, so statistical significance was determined with p < 0.05 and no adjustments for multiple testing were conducted. Stata v16 (College Station, TX) was used for all analysis.

Results

During the study period at our institution, 12,343 patients received active treatment for breast cancer, of which 3.27% were referred to the sleep clinic. Patient characteristics are listed in Table 1. Body mass index (BMI) classification included: 0.2% underweight (BMI < 18.5 kg/m²), 13.9% healthy weight (BMI 18.5 to 24.9 kg/m²), 23.3% overweight (BMI 25.0 to 29.9 kg/m²) and 62.6% obese (BMI ≥ 30.0 kg/m²). The majority had locally advanced or metastatic disease. Most patients had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 (60.6%), and the remainder 1 to 2 (39.4%). Twenty-two (5.4%) died less than one year after sleep consultation, and all but one were undergoing active therapy.

Table 1.

Patient Characteristics

Characteristic	Patients (n = 404)	Percentage (%)
Age (years)
Median (range)	56.75 (30.0 – 84.0)
Mean ± standard deviation	56.0
Gender
Female	400	99.0
Male	4	1.0
Body mass index (kg/m2)
Median (range)	32.25 (18.3 – 74.2)
Mean ± standard deviation	33.1
Race
Caucasian	239	59.1
Black	96	23.8
Hispanic	38	9.4
Asian	25	6.2
Other	6	1.5
Cancer stratification
Early	213	52.7
Locally advanced	121	30.0
Metastatic	70	17.3
Menopausal status
Premenopausal	133	32.9
Postmenopausal	267	66.1
Not applicable	4	1.0
Hormonal receptor status ^
Estrogen positive	263	65.1
Progesterone positive	191	47.3
HER2-neu positive	280	69.3
Triple negative	71	17.6
Active therapy in past 3 months ^	282	69.8
Multi-modality*	58	14.4
Chemotherapy	104	25.7
Hormonal	163	40.3
Surgery	41	10.1
Radiation	47	11.6

^{^}Totals will vary since patients may be in multiple categories

^*Includes surgery, radiation, chemotherapy or hormonal therapy

Data from sleep consult is detailed in Table 2. Most were referred for OSA symptoms (246, 60.9%), sleep disturbance (98, 24.3%), insomnia (89, 22.0%) and others for fatigue (37, 9.2%) and movement disorder (9, 2.2%). Sleep symptoms were present before cancer diagnosis in 29.2%, whereas 24.8% reported development after cancer diagnosis and unclear in the remainder. In those with a prior diagnosis of OSA, less than 1/3 were compliant with PAP therapy. Of those undergoing hormonal therapy (163), many (101) reported hot flashes around the time of sleep evaluation. Approximately half of the patients had multiple sleep disorders (Table 3). The most common sleep disorder was insomnia followed by OSA. Interventions for those with insomnia included counseling on behavioral therapies, and 132 patients were formally referred for cognitive behavioral therapy for insomnia (CBT-I) while 42 were prescribed sleep aids.

Table 2.

Sleep Consult Information

Parameter	All (404)
	N	%
Sleep symptoms
Snoring	335	82.9
Fatigue	293	72.5
Daytime sleepiness	270	66.8
Nocturnal awakenings	241	59.7
Sleep onset insomnia	194	48.0
Naps	179	44.3
Gasping and/or choking arousals	146	36.1
Witnessed apneas	130	32.2
Sleep maintenance insomnia	130	32.2
Neuropathy	61	15.1
Pain disrupting sleep	60	14.9
Restless legs	46	11.4
Timing of sleep symptoms to cancer
Sleep symptoms existed before cancer diagnosis	118	29.2
Sleep symptoms developed after cancer diagnosis	100	24.8
Unknown if related to cancer diagnosis	186	46.0
Medications
Sleep aid	158	39.1
Depression/anxiolytic	118	29.2
Pain	91	22.5
Wake-stimulating	14	3.5
Co-morbid conditions
Hypertension	195	48.3
Hypothyroidism	101	25.0
Depression	99	24.5
Diabetes	80	19.8
Anxiety	78	19.3
Prior diagnosis of OSA	75	18.6
Neuropathy	58	14.4
Sleep surveys
PSQI
Available	330	81.7
Poor sleep (PSQI ≥ 8)	248	75.4
ESS
Available	361	89.4
Daytime sleepiness (ESS ≥ 10)	199	55.3

OSA, obstructive sleep apnea; PSQI, Pittsburgh Sleep Quality Index; ESS, Epworth Sleepiness Scale

Table 3.

Sleep Disorders

Parameter	All (n=404)
	N	%
Diagnostic polysomnography performed	299	74.0
Sleep Disorders
Multiple	203	50.2
Single	185	45.8
None	4	1.0
Lost to follow up	12	3.0
Insomnia	220	54.5
Obstructive sleep apnea	217	53.7
Mild	111
Moderate	50
Severe	53
Central sleep apnea	5	1.2
Sleep-related hypoventilation	10	2.5
Other sleep-related breathing disorder	41	10.1
Restless leg syndrome	39	9.7
Periodic Limb Movement Disorder	32	7.9
Disorder of hypersomnolence	2	0.5

Diagnostic PSG was performed in 299 (74.0%) of patients with 280 baseline and 19 split-night studies (Table 3). In addition, 243 positive airway pressure titrations were performed. Sleep disorders could not be identified in 3.0% that were lost to follow up. OSA was present in 217 (53%), and half were (111) were mild and the other half either moderate (50) or severe (54). Sleep-related hypoventilation was found in 10 patients, and these were related to lung disease: chronic obstructive pulmonary disease in 3, obesity hypoventilation syndrome in 1, interstitial lung disease in 1, metastatic disease to the lung in 5. There were 5 patients with central sleep apnea (CSA), all related to pain medication use. Sleep-related breathing disorder was noted in 41 patients with Respiratory Disturbance Index ranging from 9 to 60 episodes/hour, with REM-related OSA in 6 patients and mild supine-related OSA in 3 patients. PAP therapy was prescribed in 224 patients (195 CPAP, 28 BIPAP, 1 ASV). In others, 37 were clinically observed, and a few were prescribed an oral appliance (3), desensitization to PAP (1) or avoidance of supine sleep (1). Seven patients were prescribed oxygen therapy. At the time of follow up, 63.4% (142) patients were adherent with PAP therapy, 20.1% (45) were not and 16.5% (37) were lost to follow up.

Movement disorders were noted in 39 patients with restless legs syndrome (RLS) and 32 patients with periodic limb movement disorder (PLMD). Lab work up was normal in 43 with thyroid dysfunction detected in 3 and iron deficiency in 1, and prescription for dopamine agonist was given to 12. Disorder of hypersomnolence was noted in 2 patients, of whom 1 had a narcolepsy previously diagnosed. The other patient had persistent hypersomnia despite adherence with PAP therapy for mild OSA and cancer-related fatigue, and she was prescribed wake-promoting therapy with symptomatic improvement.¹⁹

Table 4 details associations with poor sleep based on PSQI. Those with anxiety disorder (OR = 4.84; p = 0.004), on depression/anxiety medications (OR = 7.66; p < 0.001) and those with metastatic disease (OR = 2.75; p = 0.023) were more likely to have poor sleep. A PSQI ≥ 5 was also applied, but there were no significant associations. Table 5 includes the results assessing associations with hypersomnia. Those with a prior diagnosis of OSA were more likely to have hypersomnia (OR = 1.86; p = 0.035). Table 6 includes results assessing associations with insomnia. Those with anxiety disorder or on sleep aids were more likely to have insomnia. Table 7 includes the results of associations with OSA, and those with higher BMI and hypertension were more likely to have OSA.

Table 4.

Associations with poor sleep based on Pittsburgh Sleep Quality Index (PSQI)

	Univariate			Multivariable
Characteristic	OR	95% CI	p-value	OR	95% CI	p-value
BMI	1.01	0.98 – 1.04	0.705
Age (y)	0.96	0.94 – 0.99	0.004	1.00	0.97 – 1.04	0.841
Time from Diagnosis to Sleep Consult (y)	0.96	0.91 – 1.01	0.133	0.97	0.91 – 1.04	0.374
Post-menopausal	0.58	0.34 – 1.00	0.050	0.63	0.28 – 1.42	0.263
Hypertension	1.03	0.63 – 1.67	0.911
Hypothyroidism	0.55	0.32 – 0.94	0.028	0.62	0.33 – 1.19	0.150
Atrial fibrillation	0.14	0.03 – 0.69	0.016	0.16	0.02 – 1.07	0.058
Diabetes	1.10	0.58 – 2.07	0.779
Prior diagnosis of OSA	1.13	0.61 – 2.11	0.697
Depression	1.75	0.96 – 3.19	0.068	1.11	0.35 – 3.51	0.860
Anxiety disorder	4.89	2.23 – 10.74	0.000	4.84	1.67 – 13.98	0.004
Neuropathy	1.88	0.88 – 3.99	0.101	1.88	0.80 – 4.45	0.149
Use of sleep aid	6.83	3.73 – 12.53	<0.001	7.66	3.87 – 15.17	<0.001
Use of anti-depressant or anti-anxiety medication	1.81	1.04 – 3.14	0.036	0.63	0.20 – 2.01	0.434
Pain medication	1.09	0.61 – 1.95	0.764
Locally advanced cancer	0.88	0.51 – 1.52	0.657	0.56	0.29 – 1.09	0.088
Metastatic cancer	1.86	0.90 – 3.83	0.095	2.75	1.15 – 6.56	0.023
Active therapy	1.06	0.62 – 1.79	0.838
Chemotherapy	1.45	0.78 – 2.69	0.240
Radiation treatment	0.88	0.41 – 1.92	0.753
Hormonal therapy	0.82	0.45 – 1.49	0.507

BMI, body mass index; OSA, obstructive sleep apnea

Table 5.

Associations with daytime sleepiness based on Epworth Sleepiness Scale (ESS)

	Univariate			Multivariable
Characteristic	OR	95% CI	p-value	OR	95% CI	p-value
BMI	1.04	1.01 – 1.06	0.010	1.02	0.99 – 1.05	0.175
Age (y)	0.99	0.97 – 1.01	0.372
Time from Diagnosis to Sleep Consult (y)	0.99	0.94 – 1.03	0.541
Post-menopausal	0.83	0.54 – 1.30	0.425
Hypertension	1.67	1.10 – 2.54	0.016	1.40	0.89 – 2.21	0.146
Hypothyroidism	1.11	0.68 – 1.79	0.681
Atrial fibrillation	1.71	0.42 – 6.93	0.456
Diabetes	1.64	0.96 – 2.79	0.069	1.18	0.66 – 2.10	0.576
Prior diagnosis of OSA	2.29	1.30 – 4.00	0.004	1.86	1.04 – 3.33	0.035
Depression	0.84	0.52 – 1.36	0.485
Anxiety disorder	0.78	0.46 – 1.31	0.346
Neuropathy	1.10	0.62 – 1.97	0.738
Use of sleep aid	0.50	0.33 – 0.77	0.002	0.56	0.36 – 0.87	0.011
Use of anti-depressant or anti-anxiety medication	0.80	0.51 – 1.26	0.336
Pain medication	1.20	0.73 – 1.98	0.461
Locally advanced	1.28	0.80 – 2.06	0.306
Metastatic	1.18	0.66 – 2.12	0.577
Active treatment	1.28	0.81 – 2.01	0.285
Chemotherapy	1.43	0.85 – 2.42	0.181
Radiation treatment	0.62	0.32 – 1.22	0.170
Hormonal therapy	1.09	0.66 – 1.80	0.747

BMI, body mass index; OSA, obstructive sleep apnea

Table 6.

Associations with insomnia

	Univariate			Multivariable
Characteristic	OR	95% CI	p-value	OR	95% CI	p-value
BMI	0.96	0.93 – 0.98	0.002	0.97	0.93 – 1.00	0.076
Age (y)	0.97	0.95 – 0.99	0.001	0.97	0.93 – 1.01	0.106
Time from Diagnosis to Sleep Consult (y)	0.96	0.91 – 1.010	0.095	0.94	0.84 – 1.05	0.273
Post-menopausal	0.71	0.46 – 1.08	0.107	1.25	0.55 – 2.85	0.590
Hypertension	0.46	0.30 – 0.68	0.000	0.61	0.32 – 1.17	0.137
Hypothyroidism	0.83	0.52 – 1.33	0.446
Atrial fibrillation	0.13	0.02 – 1.03	0.054
Diabetes	0.54	0.32 – 0.93	0.025	1.16	0.52 – 2.60	0.710
Prior diagnosis of OSA	0.38	0.21 – 0.67	0.001	0.41	0.17 – 1.00	0.050
Depression	1.60	1.01 – 2.53	0.044	1.55	0.62 – 3.89	0.353
Anxiety disorder	3.71	2.21 – 6.24	0.000	3.74	1.61 – 8.70	0.002
Neuropathy	1.17	0.67 – 2.05	0.585
Use of sleep aid	7.87	5.00 – 12.37	0.000	5.52	3.04 – 10.03	0.000
Use of anti-depressant or anti-anxiety medication	1.89	1.23 – 2.93	0.004	0.87	0.33 – 2.30	0.784
Use of pain medication	1.04	0.65 – 1.68	0.858
Locally advanced	1.29	0.82 – 2.02	0.274
Metastatic	0.89	0.51 – 1.56	0.677
Active treatment	1.03	0.67 – 1.59	0.891
Chemotherapy	1.16	0.71 – 1.90	0.558
Radiation treatment	0.91	0.48 – 1.74	0.786
Hormonal therapy	0.92	0.57 – 1.50	0.746

Table 7.

Associations with Obstructive Sleep Apnea (OSA)

	Univariate			Multivariable
Characteristic	OR	95% CI	p-value	OR	95% CI	p-value
BMI	1.04	1.01 – 1.06	0.010	1.02	0.99 – 1.05	0.175
Age (y)	0.99	0.97 – 1.01	0.372
Time from Diagnosis to Sleep Consult (y)	0.99	0.94 – 1.03	0.541
Post-menopausal	0.83	0.54 – 1.30	0.425
Hypertension	1.67	1.10 – 2.54	0.016	1.40	0.89 – 2.21	0.146
Hypothyroidism	1.11	0.68 – 1.79	0.681
Atrial fibrillation	1.71	0.42 – 6.93	0.456
Diabetes	1.64	0.96 – 2.79	0.069	1.18	0.66 – 2.10	0.576
Prior diagnosis of OSA	2.29	1.30 – 4.00	0.004	1.86	1.04 – 3.33	0.035
Depression	0.84	0.52 – 1.36	0.485
Anxiety disorder	0.78	0.46 – 1.31	0.346
Neuropathy	1.10	0.62 – 1.97	0.738
Use of sleep aid	0.50	0.33 – 0.77	0.002	0.56	0.36 – 0.87	0.011
Use of anti-depressant or anti-anxiety medication	0.80	0.51 – 1.26	0.336
Use of pain medication	1.20	0.73 – 1.98	0.461
Locally advanced	1.28	0.80 – 2.06	0.306
Metastatic	1.18	0.66 – 2.12	0.577
Active treatment	1.28	0.81 – 2.01	0.285
Chemotherapy	1.43	0.85 – 2.42	0.181
Radiation treatmemt	0.62	0.32 – 1.22	0.170
Hormonal therapy	1.09	0.66 – 1.80	0.747

Discussion

In our study, patients with breast cancer of all stages had significant sleep disturbances and sleep disorders. The majority of our patient cohort comprised those with early stage malignant disease (52.7%), and most were undergoing active treatment (69.3%). Interestingly almost 1/3 of the patients had sleep symptoms prior to cancer diagnosis whereas a quarter of the cohort identified sleep disruption starting after cancer diagnosis. Snoring, fatigue, daytime sleepiness and nocturnal awakenings were the most commonly reported symptoms. PROs with PSQI and ESS were available in over 80% of our cohort, and associations with use of sleep aids were prominent in both. Multiple sleep disorders were found in many, and insomnia emerged as the most prevalent in 54.5% followed very closely by OSA in 53.7% (which was mostly mild). Our experience with sleep disturbance and sleep disorders in patients with breast cancer using PROs and subsequent workup is the largest reported to date.

Sleep disruption contributed to significant symptom burden in our cohort. Poor sleep based on PSQI was significantly increased in patients using sleep aids [OR 7.66, 95% CI 3.87 to 15.17], with an anxiety disorder [OR 4.84, 95% CI 1.67 to 13.98], or with metastatic disease [OR 2.75, 95% CI 1.15–6.56]. Symptom clusters include two or more concurrent symptoms which may have shared underlying mechanisms and outcomes²⁰. More than half of studies on symptom clusters in patients with cancer describe one that is psychoneurological with fatigue, anxiety/depression, sleep disruption and pain. Previous studies examining cancer-related symptom clusters have found the sleep disruption is coincident with fatigue, mood disturbance, and pain.²¹ In patients with breast cancer, the composition of symptom clusters may change across the various stages of cancer treatment, but fatigue-sleep disturbance and the psychological cluster (such as anxiety, depression, nervousness, irritability) remain the most commonly reported symptom cluster in this cohort²¹. Aside from snoring (82.9%), the most prominent symptom in our cohort included fatigue (72.5%), thus, our data corroborates the association between poor sleep, mood disorder and fatigue²⁰. Further, more than one-third (39.1%) of our patients were taking sleep aids. In a longitudinal, descriptive randomized control trial of women receiving outpatient chemotherapy, Moore and colleagues report 20% of participants took at least one sleep aid before the first cycle of chemotherapy, and usage decreased over time²². In a larger study evaluating the risk of fractures with sleep aid use, Haque and associates reported sleep medication use in 40% of women breast cancer survivors²³. Thus, use of sleep aids may be more pervasive in breast cancer, underscoring the lack of widely available non-pharmacologic sleep management approaches such as CBT-I²⁴. The association between poor sleep with anxiety and depression is well-established, with potentially a common psychoneuroimmunological mechanism, often, often reciprocal or cyclic, and challenging, as mood disorders may also presents as insomnia²⁵. However there may be more significant sequelae, for in a large collection of patients with breast cancer using PROs Measurement Information System (PROMIS), anxiety was seen as a potential driver of impaired quality of life ²⁶. Moreover, 29.2% of our cohort identified sleep issues before cancer diagnosis, but 24.8% noted these developed after cancer diagnosis. Thus, addressing those with previous sleep issues is paramount, for the sleep disruption engendered with cancer therapies is well established^27,28. Since a quarter of our cohort developed sleep issues after diagnosis, screening using PRO or at the least initiating the conversation may potentially identify and ameliorate sleep disruption with interventions. Finally, metastatic cancer comprised only 17.3% of our cohort, but symptom burden in advanced cancer can be immense, often related to widespread inflammation, pain or side effects of pain medications, symptoms related to site of metastases, and aberrant sleep-wake cycle exacerbating circadian disruption^27,29. Increasing evidence from clinical and basic research suggests that chronic circadian disruption contributes to both tumor initiation and progression³⁰. Thus, poor sleep may adversely affect not only quality of life, but tumor biology and clinical outcomes³¹. With the advent of immunotherapy and targeted therapies, survival will hopefully continue to improve, but sleep symptoms may persist into survivorship.

Daytime hypersomnolence in cancer patients can be sequelae from anti-neoplastic therapy, medication side effects, sedative and opioid medications, disruption of normal sleep-wake cycles, advanced cancer, or undiagnosed underlying sleep disorders^32,33. In our cohort, daytime sleepiness based on ESS was associated with a prior diagnosis of OSA [OR 1.86, 95% CI 1.04 to 3.33] and use of a sleep aid [OR 0.56, 95% CI 0.36 to 0.87]. Interestingly, most of our patients were referred for sleep-disordered breathing, and in 18.6% with a prior diagnosis of OSA, less than one-third were adherent with PAP therapy. Use of sleep aid correlated with significant daytime hypersomnolence, but there was no association of daytime sleepiness in the 22.5% using pain medication Two patients did have disorders of hypersomnolence, and one was related to cancer-related fatigue, and the other had a narcolepsy¹⁹. PSG was performed in 74% of patients due to concern for underlying sleep disorder. OSA was diagnosed by PSG in 53.7% (217 patients) of our cohort, and most had mild OSA. Current data estimates of OSA in 10% to 15% of the population, with 3% among women ages 30 to 49 years and 9% among women 50 to 70 years of age, so risk of OSA increases in with age and menopause in women^34–36. It is postulated that diminishing levels of progesterone, known to be a respiratory stimulant and upper airway dilator, as well as increased body weight may both contribute.^37,38 Weight gain is frequently seen as a result of adjuvant therapy and although Sella and colleagues report weight gain modestly increases at 1 and 3 years postdiagnosis, it may further dispose breast cancer survivors to OSA^39–41. Further, presentation of OSA in women is atypical with more fatigue and insomnia, rather than snoring or sleepiness^36,38,42. Women may have milder OSA and/or REM predominant OSA including events with less severe desaturations or duration³⁶. In our cohort, 10% (41 patients) had mild sleep-related breathing disorder with significant RDIs. Sleep-related hypoventilation was noted in 2.5%, but this should also be considered in those with metastatic disease to the lung. More importantly, OSA may have other cardiovascular and metabolic sequelae, and even REM-related OSA is independently associated with hypertension, metabolic syndrome and diabetes, and it has been linked to both increased cancer incidence and cancer-related mortality^43–45. Thus, screening for daytime hypersomnolence and subsequent evaluation for OSA or other sleep-related breathing disorders is important and a valuable intervention in this population.

Sleep disorders represent potentially modifiable conditions that can alleviate symptom burden in cancer patients. In our cohort, many had either one (46%) or multiple (50%) sleep disorders. Insomnia was associated with use of sleep aids [OR 5.52, 95% CI 0.17 to 1.00], anxiety disorder [OR 3.74, 95% CI 1.61–8.70], and prior diagnosis of OSA [OR 0.41, 95% CI 0.17 to 1.00], and OSA was associated with prior diagnosis of OSA [OR 1.86, 95% 1.04 to 3.33] and use of sleep aids [OR 0.56, 95% CI 0.36 to 0.87]. Insomnia was the most common underlying sleep disorder in our series. Susceptibility to insomnia in patients with breast cancer has been well described, and it may be attributed to a variety of reasons that disrupt sleep including: increased frequency and severity of hot flashes associated with treatment, increased depression, anxiety or fatigue related to diagnosis or treatment, pre-existing sleep issues, or mood disorders¹. Palesh and colleagues demonstrated that insomnia was three times more common in patients undergoing chemotherapy than the general population²⁸. Fleming and associates also demonstrated an increased risk for persistent insomnia in those undergoing chemotherapy [OR 0.08, 95% CI 0.02 to 0.29] and increased pre-diagnosis Insomnia Severity Index scores [OR 1.13/unit increase, 95% CI 1.05 to 1.21]²⁴. Cancer and associated treatments may also contribute to dysregulated circadian rhythms resulting in abnormal sleep-wake patterns^27,46. Emerging data indicates sleep regularity is a stronger predictor of mortality risk than sleep duration, and high sleep regularity is associated with lower risk of all-cause cancer and cardiometabolic mortality⁴⁷. In our cohort, many were counseled on sleep health, and some were referred for formal cognitive and behavioral therapy. In those with OSA, PAP therapy was prescribed, and interestingly, CPAP adherence in our cohort was 67%. Since CPAP adherence is postulated to be a complex interplay among self-efficacy, treatment expectancy and risk perception, we hypothesize that these higher rates of adherence may be related to a motivated oncologic population^48,49. Our findings further highlight the importance for health care providers to engage in discussions regarding sleep health, sleep-related symptoms and potentially impactful interventions. Prospectively screening this population for sleep disruption, counseling for sleep hygiene and use of CBT-I could be helpful⁵⁰. These may be good adjuncts to treatment for patients with breast cancer in all stages to improve quality of life⁵¹.

Our study has limitations inherent to most retrospective studies. Only patients referred to our sleep center (from another service or self-referred) were included, so a referral bias exists. Although most patients had surveys at the time of evaluation, some surveys were missing, and post- intervention assessments were not available. However, our data was obtained through the electronic medical record in addition to the sleep laboratory database, so loss of patient data was minimized. Insomnia Severity Index scores were not available. Finally, in patients with breast cancer, hormonal changes can exacerbate nocturnal awakening due to vasomotor changes or hot flashes, but interestingly post-menopausal status did not significantly influence poor sleep on PSQI. In addition, some anti-depressants are used to treat hot flashes, so their impact on sleep may be underestimated. Further prospective studies systematically evaluating sleep symptoms in all patients with breast cancer with more comprehensive PRO including sleep surveys would be helpful. Our study represents the largest cohort evaluating sleep disturbances using both PRO with sleep surveys and PSG when indicated in patients with breast cancer to date. The significance of our findings in this large cohort sets a foundation for future research to both study and improve sleep health among this population.

Sleep disturbance can contribute significantly to symptom burden in patients with breast cancer. A quarter of our cohort developed sleep disruption after their cancer diagnosis, and a little less than 1/3 of our cohort identified sleep issues prior to cancer diagnosis. Poor sleep was associated with use of sleep aids, anxiety disorder and metastatic disease. Insomnia was the most common sleep disorder, followed closely by OSA (mostly mild). The impact of sleep disruption, suboptimal sleep health and untreated sleep disorders may have a detrimental effect on quality of life. Proactive screening using PRO and counseling to optimize sleep health would be beneficial in patients with breast cancer.

Abbreviations

AASM

American Academy of Sleep Medicine (AASM)

ASV

auto-servoventilation

BMI

body mass index

BPAP

bi-level positive airway pressure

CBT-I

cognitive behavioral therapy for insomnia

CSA

central sleep apnea

ECOG

Eastern Cooperative Oncology Group

ESS

Epworth Sleepiness Scale

NCCN

National Comprehensive Cancer Network

OR

odds ratio

OSA

obstructive sleep apnea

PAP

positive airway pressure

PLMD

periodic limb movement disorder

PSG

polysomnography

PSQI

Pittsburgh Sleep Quality Index

PRO

patient-reported outcomes

REDCap

research electronic data capture

RLS

restless leg syndrome

RERA

respiratory effort related arousal

RDI

Respiratory Disturbance Index

SaO₂

oxygen saturation

Data availability.

All the data generated and/or analyzed during the current study are included in this article and are available from the corresponding author on reasonable request.