Sleep disturbance of adults with a brain tumor and their family caregivers: a systematic review

Abstract

The high incidence and psychophysiological morbidities of sleep disturbance in cancer have been increasingly recognized. Yet, more detailed understanding of sleep disturbance and options for management have been neglected areas in both clinical care and research. Brain tumor patients have been particularly overlooked. A systematic search of the literature from 1990 to 2015 was performed to review sleep disturbance in adults with primary or secondary brain tumor and their family caregivers. Fifty eligible studies were identified, of which 12 focused on sleep, 37 reported sleep items within a health-related quality of life measure and 1 reported caregivers’ sleep. No sleep intervention has been developed or tested for brain tumor patients. Sleep disturbance and somnolence were frequently reported as the most severely rated symptoms within health-related quality of life across the disease course or treatments, along with fatigue. However, sleep-focused studies yielded inconsistent results in small samples of mostly benign brain tumors in long-term remission from total tumor resection. The research using standardized, multifaceted sleep assessments, particularly in patients with malignant brain tumor and caregivers who are undergoing treatment, is seriously lacking. A more systematic examination of sleep disturbance is warranted to inform the development of better symptom management programs in this population.

Sleep is a vital circadian process closely linked to immune system function, homeostatic mechanisms, psychological well-being, cognitive functioning, as well as quality of life in people.^1–4 These functions of sleep are critical in the oncology population throughout the disease trajectory, to maximize health and well-being.^5,6 The high incidence and psychophysiological morbidities of sleep disturbance in general cancer patients have been discussed in the literature,^5,7–11 highlighting its impact on quality of life^5,6 and the clinical consequences for patients and their caregivers, such as poorer treatment adherence and disease outcomes.^8,12

Unfortunately, brain tumor (BT) populations have largely been overlooked in this research. Although there are several reviews of sleep disturbance and its management in the oncology population in general,^8,13,14 studies that assessed sleep disturbance in BT populations were rarely included in data synthesis, nor have they been considered previously within a specific systematic review. A high prevalence of sleep disturbance has been increasingly recognized in clinical practice. Evidence also suggests a unique pathophysiology of disrupted sleep-wake rhythms related to BT itself or BT-related treatments.^15,16 Yet, understanding sleep disturbance and developing options for management have been neglected areas in both neuro-oncology care and research.

This systematic review aims to assimilate current evidence on sleep disturbance in adult BT patients and their caregivers, to allow further progress in the understanding of epidemiology and possible targets of intervention, and to underpin future research. It focuses on identifying the: (i) prevalence and clinical associates of sleep disturbance; (ii) relationship between objective and patient-reported outcomes; (iii) current practice of sleep interventions; and (iv) association between patient and caregiver in the dyad’s sleep.

Methods

This systematic review followed the established guidelines for systematic reviews and meta-analyses (the Preferred Reporting Items for Systematic Reviews and Meta-Analyses [PRISMA] statements).¹⁷

Search Strategy

The complete search method is outlined in Supplementary Table S1. A search of the scientific literature was conducted on 7 electronic databases (eg, PubMed, Embase) for published articles. In the gray literature, dissertations, theses, and conference proceedings and abstracts were searched to circumvent the possible influence of publication bias on review results. Search terms used were keywords or Medical Subject Headings terms related to sleep, cranial cancer, and sleep-targeted intervention, spreading the search to their related or stemmed words. Initial searches using a combination of all 3 search terms yielded a smaller number of reports, compared with the search results using sleep and cancer terms only. Therefore, a combination of sleep and cancer terms was used in searches. Searches were limited to human studies published in English between January 1990 and September 2015. Additional manual searches identified potentially eligible studies from reviews or meta-analyses and from the reference lists of relevant studies. Paper copies of the journal for which full text was not available electronically were searched.

Study Selection

The authors independently reviewed the potential relevance of all titles and abstracts identified from the searches and then full-text copies of all articles judged potentially relevant from the initial screening based on the eligibility criteria for inclusion. Inclusion criteria were: (i) adults (≥18 y) with a primary (malignant, nonmalignant) or secondary (metastatic) BT and their adult family caregivers; (ii) empirical study report or case report; and (iii) reporting sleep outcomes using one of the following measures:

• validated self-report sleep assessments

• validated health-related quality of life (HRQoL) scales with ≥1 sleep item

• validated psychological (eg, mood) assessments with ≥1 sleep item

• validated symptom assessments with ≥1 sleep item

• sleep log or diary kept for ≥3 consecutive sleep-wake cycles

• actigraphy (ACT), polysomnography (PSG), or sleep electroencephalogram (EEG)

Exclusion criteria were: (i) reviews, book chapters, or opinions; (ii) no sleep outcome reported; (iii) use of nonvalidated sleep assessment; (iv) adult survivors of pediatric brain cancer or caregivers of a child with BT; (v) unable to distinguish BT sleep data from mixed cancer samples; and (vi) duplicate reports of identical results.

Data Extraction and Quality Assessment

Data were extracted for the study objective, design, sample size, demographic and clinical characteristics of the study population, intervention (or exposure), sleep assessment tool, and outcomes related to sleep. Data extraction was aided by using a data extraction template and NVivo qualitative data analysis software (QSR International v11 Pro, 2015). All included studies were assessed against each of the questions in a quality assessment checklist (see Supplementary Table S2) and subsequently given an overall quality rating of poor, fair, good, or cannot determine. The overall quality rating of 10 of 50 studies was conducted independently by 2 authors (M.J. and M.A.) and the results compared. As the ratings agreed for all 10 studies, quality assessments for the rest of the included studies were conducted by one author (M.J.). All 3 authors confirmed the quality rating results and discussed and came to consensus on the final quality rating results. Only the studies of fair or good quality were included in the data synthesis to answer the review questions.

Data Synthesis

Studies were categorized as (i) sleep intervention study, (ii) descriptive sleep study, or (iii) descriptive HRQoL study, according to the main focus or objective of each study for this review. A study which aimed to assess an intervention intended to bring about a change in the sleep-related symptoms or interference was referred to as a sleep intervention study. A study, without a sleep-specific intervention, which described the participants in terms of sleep or sleep within HRQoL was referred to as a descriptive sleep study or a descriptive HRQoL study, respectively. Any systematic difference between study groups, or pre/posttreatment, within a specific descriptive study was noted as an exposure variable and its effect on sleep outcomes was reviewed if applicable.

In order to review the prevalence and severity of sleep disturbance, the cutoff score of a number of standardized, multi-item sleep assessments was used, namely the Pittsburgh Insomnia Quality Index (PSQI),¹⁸ the Insomnia Severity Index (ISI),¹⁹ and the Epworth Sleepiness Scale (ESS).²⁰ These self-reported outcome measures assess different facets of sleep disturbance, are not single-item or subscales of multisymptom HRQoL assessments, and have demonstrated psychometric properties and clinical utility for implying the clinically significant impairment. The physiological measures of sleep, such as ACT and PSG, were reviewed to determine the presence of impairments in sleep quantity, circadian rhythms, or sleep-related breathing.

Results

Electronic database searches identified a total of 1002 records for published articles and 37 records for the gray literature. After the initial screening, 89 articles remained for full-text review, of which 48 articles were excluded as they did not meet eligibility criteria (43 articles), were duplicate reports of the same study results (3 articles), or were unobtainable as full texts (2 dissertations). An additional 32 articles were identified by the manual search, of which 9 were included after full-text assessment. Therefore, a total of 50 articles were included in this review (Figure 1).

Fig. 1.

Study selection process and outcomes.

Description of Studies

All 50 included studies were published in the form of journal articles. Publication years ranged between 1997 and 2015, with a noticeable increase in the number of studies reporting sleep and HRQoL outcomes of BT populations in the past 5 years (29 studies). Studies were spread globally, with most conducted in Europe and North America (43 studies) and some in Asia (7 studies).

All studies included were categorized into either descriptive sleep studies (n = 13) or descriptive HRQoL studies (n = 37). None of the studies could be categorized as a sleep intervention. Of 13 descriptive sleep studies, 4 examined sleep disturbance in mixed glioma patients, and 8 in benign BT such as pituitary adenoma or craniopharyngioma (Table 1 summarizes findings). One study examined sleep characteristics of family caregivers of patients with primary malignant BT.²¹

Table 1.

Findings of descriptive sleep studies examining sleep disturbance of BT patients

Study	Study Type	Sample	Sleep Assessment	Sleep Outcome
Biermasz et al, 2011 25	Cross-sectional	N = 17 NFMA. Location: pituitary. All treated with transsphenoidal surgery, median survival: 8 y, Netherlands.	PSG (1 night) ACT (7 days) PRO: Sleep diary, Berlin Questionnaire, PSQI, ESS, and Clinical Symptom score for sleep disorders.	PSG: Patients had poorer sleep efficiency, more time awake in bed, a disruption in sleep stage distributions than controls. No sign of sleep apnea or abnormal periodic limb movements. ACT: Patients had more time in bed, more actual sleep time and more naps than controls, but similar sleep efficiency. PRO: Patients reported more sleep disturbance and awakenings during the night. Patients showed an increased disruption of a clear separation of high activity during the day and low activity during the night (a disruption in the circadian clock). Patients showed higher risk of developing sleep apnea.
Borodkin et al, 2005 62	Case report	N = 1 Pituitary microadenoma, male, age 21 y, obese (BMI 30), survival: 2 y, Israel.	PSG (1 night) ACT (9 days). Clinical assessment by a sleep specialist	PSG: Patient had no disruptions in sleep architecture. ACT: Patient showed irregularities in rest-activity pattern. Advanced sleep onset time by 1 h. Shorter than 24 h periodicity. Irregular patterns in occurrence of sleep episodes. Diagnosis of circadian rhythm sleep disorders with irregular sleep-wake patterns was subsequently given.
Crowley et al, 2011 31	Cross-sectional	N = 28 Craniopharyngioma, Ireland. (N = 23 obese or overweight controls)	PSG (1 night) PRO: ESS	Significantly more number of patients reported having somnolence or fatigue than controls. ESS: Significant daytime somnolence (ESS > 10) was more prevalent in patients than controls. ESS scores were higher in patients than controls. PSG: No difference in rate of diagnosis of sleep apnea in patients and controls. Patients had longer sleep time and better sleep efficiency than obese controls.
Dempsey et al, 2003 61	Case report	N = 1 Acromegalic pituitary adenoma, male, age 60 y, Caucasian, survival: 40 y, UK.	PSG (1 night)	PSG: Patient had sleep onset latency of 5.5 minutes and began REM sleep. Total sleep time of 6 h 58 min. Normal number of micro arousals. No evidence of periodic limb movements. Patient had a diagnosis of narcolepsy with cataplexy and sleepiness as most prominent symptoms.
Faithfull & Brada, 1998 16	Cohort	N = 19 Primary brain tumor (79% HGG, 16% pituitary adenoma, 5% medulloblastoma), UK	PRO: Daily diary with 11 VAS including drowsiness and sleep items and clinical interview	16/19 developed somnolence syndrome of mild to severe levels. Somnolence had a cyclical pattern consisting of 2 phases. Onset of symptoms at week 2 post RT. Worst on days 11 to 21, then resolved. Reappearance at week 5, lasting for 2 to 5 days. Consistently across all reported symptoms. Most prevalent symptoms included drowsiness. Drowsiness and fatigue correlated.
Krieg et al, 2012 26	Cross-sectional	N = 9 Pineocytoma grade 1. All received pinealectomy, mean survival 28 months, Germany. (N = 9 each control group of craniotomy or lumbar discectomy, matched for age, gender, and date of surgery)	PRO: PSQI, ISI, ESS, and the European Diagnosis Questionnaire of Sleep Impairment	Overall, sleep quality reduced after craniotomy. Insomnia severity was scored higher in craniotomy group than other surgery groups. Somnolence was rated similarly in all groups. No groups showed onset of sleep apnea, narcolepsy, or restless leg syndrome.
Leistner et al, 2015 27	Cross-sectional	N = 247 Pituitary adenoma (21% nonfunctioning). Mean survival: 12.27 y, Germany. (N = 757 healthy, matched controls)	PRO: PSQI	There was no difference between patients and controls in sleep duration, sleep onset latency, or subjective sleep efficiency. Patients had slightly poorer sleep than controls.
Mainio et al, 2013 45	Cohort	N = 70 Primary brain tumor (24% LGG, 30% HGG, 31% meningioma), Finland.	PRO: Sleep subscale (5 yes/no items) of Nottingham Health Profile	Effects of operation and tumor laterality on sleep were considered. A significant improvement in sleep was found at 3 months after surgery and persisted up to one year after surgery. Among tumor laterality groups, patients with bilateral tumor had poorer sleep than left or right laterality group at all time points. However, when controlling for depression, the laterality effect was not so evident.
Pickering et al, 2014 28	Cross-sectional	N = 15 Craniopharyngioma. Mean survival: 9.7 y, Denmark. (N = 14 controls)	ACT (14 days) PRO: Sleep diary, PSQI, and ESS	ACT: Patients woke up an hour earlier in the morning. No other differences were found. PRO: More patients reported longer sleep latency and daytime dysfunction than controls. The general level of daytime sleepiness was similar in patients and controls.
Pollak et al, 2004 46	Cohort	N = 11 Primary brain tumor (45% glioma, 55% meningioma), Israel.	PSG within 1 week before and 1 month after operation	PSG: No significant changes after operation were found in total sleep time, sleep efficiency, sleep latency, number of arousals, or periodic leg movements. The duration of REM sleep stage increased significantly after operation while duration of other sleep stages was unchanged. Sleep-related breathing improved significantly after tumor resection. A clinically significant apnea was found in 7. Resection improved this in all patients. No relationship found between the tumor size, location, mass effect, or clinical outcome and sleep apnea.
Powell et al, 2011 49	Cohort	N = 68 Primary brain tumor (18% AA, 59% GBM, 9% LGG, 14% other), UK.	PRO: Littman somnolence syndrome scale and Somnolence VAS	Patients reported worse somnolence as the RT initiated. There was a peak in the score beginning at week 6 and continuing until week 12, and then return to the initial baseline. A significant increase in somnolence from 3 weeks through to 12 weeks was demonstrated; 34/68 patients had mild to severe somnolence. The cyclic pattern of somnolence after RT was similarly shown in 2 outcome measures. No effect of demographic and tumor variables was shown.
van der Klaauw et al, 2007 33	Cross-sectional	N = 76 NFMA. All received total tumor resection, mean survival 10 y, Netherlands.	PRO: ESS and Münchener Chronotype Questionnaire	Patients showed increased daytime sleepiness compared with controls. There were 19 patients who had excessive daytime somnolence (ESS > 10) compared with 7 controls (26% vs. 10%, respectively). Somnolence correlated with majority of quality of life subscales of psychophysiological domains. No alteration in sleep patterns found in the NFMA patients, suggesting that the increased daytime sleepiness and fatigue are not due to change in sleep duration or timing of sleep.

Abbreviations: NFMA, nonfunctioning macroadenoma; VAS, visual analogue scale; REM, rapid eye movement; RT, radiotherapy; PRO, patient reported outcome; GBM, glioblastoma; LGG, low grade glioma; AA, anaplastic astrocytoma; BMI, body mass index.

Several standardized HRQoL assessments developed for cancer populations include a sleep subscale. Of the 37 HRQoL studies, 17 used the European Organisation for Research and Treatment of Cancer (EORTC) Core Quality of Life Questionnaire (QLQ-C30), Brain Neoplasm module (QLQ-BN20), and a shorter version of the core questionnaire for palliative care (C15-PAL). Five studies used the MD Anderson Symptom Inventory–Brain Tumor module (MDASI-BT). In both HRQoL tools, sleep disturbance (also called insomnia) and drowsiness were important facets of the BT patients’ HRQoL,^22–24 although only assessed by a single question.

The quality assessment rated 10 studies as good and 36 as fair. Four studies were rated as poor in quality and were excluded from the synthesis of findings (3 HRQoL and 1 sleep study; see Supplementary Table S3).

Type, Prevalence, Severity, and Potential Contributing Factors of Sleep Disturbance in Patients with Brain Tumor

Types of sleep disturbance

A wide variety of sleep-related terms were used across the studies, often interchangeably, to describe sleep disturbance. Sleep terms were not defined but included in the HRQoL assessments. The terms were largely integrated into 5 types of sleep disturbance (see Supplementary Table S4 for sleep-related terms identified under each type): (i) dissatisfaction with sleep quality; (ii) nonspecific disturbed sleep; (iii) insomnia symptoms (disruption in sleep quantity, pattern, or architecture); (iv) sleep apnea; and (v) somnolence. Nonspecific disturbed sleep and somnolence were the most frequently reported sleep complaints, largely due to the large number of studies that assessed these types of sleep complaints. Of note, the sleep item, conveniently labeled insomnia, of the EORTC QLQ assessments was regarded as nonspecific disturbed sleep, as this single item asks whether patients had trouble sleeping.

Prevalence and severity of sleep disturbance using the PSQI, ISI, and ESS

The prevalence and severity of sleep disturbance were reviewed from studies that used standardized, multiconstruct sleep assessments: PSQI, ISI, and ESS. Mean scores of these measures and percentages of the sample whose scores were above the clinical thresholds are shown in Table 2. PSQI global score greater than 5 is indicative of significant sleep disturbance.¹⁸ Mean PSQI global scores in brain tumor samples were to some extent above the cutoff in all 4 studies^25–28 and were significantly higher than the controls in 2 studies of benign BT patients.^26,27 None of the studies computed the percentage of the sample above the cutoff. Given seemingly large variations in the scores, it was difficult to determine the prevalence and severity based on the PSQI. The ESS assesses daytime somnolence, and scores above 10 indicate severe somnolence that would require clinical interventions.^20,29 Baseline mean scores varied between the studies ranging from 5.9 to 10.3, also with large standard deviations. The prevalence rates of clinically impaired somnolence ranged between 21% and 29% in 5 studies.^25,30–33 One study assessed insomnia in pineocytoma and low-grade BTs using the ISI. Low-grade tumor patients who underwent craniotomy reported a greater level of insomnia than pineocytoma and non-BT control groups.²⁶

Table 2.

Baseline PSQI, ESS, and ISI scores and percentage of sample above clinical thresholds (PSQI > 5, ESS > 10, or ISI ≥ 8 in cancer⁷⁰)

Study	Tumor	n	ISI		PSQI Global		ESS
			Mean ± SD	%	Mean ± SD	%	Mean ± SD	%
Leistner et al, 2015 27	Pituitary adenoma	247	—		6.75 ± 4.17*	—	—
Biermasz et al, 2011 25	NFMA	17	—		5.82 ± 3.0	—	6.53 ± 5.8	29
Krieg et al, 2012 26	Pineocytoma	9	5.9 ± 1.9	—	6.6 ± 1.3*	—	6.6 ± 1.5	—
	LG tumors	9	10.4 ± 3.1*	—	7.2 ± 2.0*	—	9.9 ± 1.9	—
Pickering et al, 2014 28	Craniopharyngioma	15	—		6.5 ± 3.3	—	10.3 ± 5.0	—
Brown et al, 2006 30	HGG	185	—		—		Median 67^	21^
Crowley et al, 2011 31	Craniopharyngioma	28	—		—		Median 7.5*	25*
Roemmler-Zehrer et al, 2015 68	Craniopharyngioma	26	—		—		9.0 ± 3.9	—
	NFPA	31					7.8 ± 4.9	—
Valko et al, 2015 32	GBM	65	—		—		5.9 ± 4.3	22
van der Klaauw et al, 2007 33	NFMA	76	—		—		7.6 ± 4.6*	26*

Abbreviations: NFMA, nonfunctioning macroadenoma; LG, low grade; NFPA, nonfunctioning pituitary adenoma; GBM, glioblastoma.

* Significant difference (P < .05) from the control group within the study.

^ ESS scores were transformed to scores 0–100, 0 being the worst somnolence. The cutoff of ≤50 was used for computing % of clinically impaired somnolence.

Prevalence and severity of sleep disturbance reported in the HRQoL studies

Sleep disturbance and/or drowsiness were repeatedly reported as one of the most prevalent and severe symptoms, along with fatigue and psychological distress in many of the HRQoL studies involving malignant or metastatic BT samples.^34–41 Mean scores of insomnia and drowsiness scales of the EORTC QLQs of the included studies are shown in Table 3. Mean scores of sleep symptom items of the MDASI-BT are similarly shown in Table 4. The severity of sleep disturbance items were similarly rated among the studies, despite wide variation in BT patients. Moderate to severe symptom distress was seen in 27%–34% for sleep disturbance and in 22%–32% for drowsiness.^34–36 These sleep symptoms were often ranked the first or second most severe symptom reported with fatigue.^34,37–39 Fatigue appeared to be the most frequent complaint among brain tumor patients, and there was a significant association between fatigue and sleep disturbance.^32,39,41,42 No further comparison or statistical analysis was possible given the heterogeneity in the samples.

Table 3.

Insomnia and drowsiness subscale scores of the EORTC QLQ assessments

Study	Tumor	n		Insomnia	Drowsiness
				Mean ± SD	Mean ± SD
Aaronson et al, 2011 22	LGG	195		—	—
Bunevicius et al, 2014 48	Mixed PBT	100		44.0 ± 40.4	23.3 ± 31.2
Caissie et al, 2012 50	MBT	108	Pre-WBRT	34	33
		70	Post (1 mo )	19	39 ± 30
Cheng et al, 2010 43	Glioma	92		21.7 ± 32.6	—
			Age >50 y	30.5	—
			<50 y	16.4
			KPS <80	38.3	—
			≥80	15.6
Giesinger et al, 2009 37	Mixed PBT	42		27.8 ± 32.0	38.9 ± 32.9
Gustafsson et al, 2006 38	LGG	39		26.3 ± 29.2	—
Kim et al, 2012 39	Mixed PBT	25	Pre-rehab	50.0 ± 27.2	—
			Post (1 mo )	47.9 ± 27.1	—
Koo et al, 2013 51	MBT	15	Pre-upfront SRS	31.1	31.1
			Post (1 mo )	20.0	44.4
		15	Pre-salvage SRS	24.4	21.4
			Post (1 mo )	22.2	28.9
Minniti et al, 2013 53	GBM	65	Pre RT + TMZ	21.6 ± 25.3	39.8 ± 22.7
			Post (6 mo )	—	48.9
Nguyen et al, 2013 60	MBT	68		—	—
Osoba et al, 1997 24	Glioma	105	Baseline	20.2 ± 25.6	—
		92	Follow-up	22.8 ± 27.5	—
Osoba, Brada, Prados et al, 2000 69	AA	145		21 ± 26	34 ± 30
Osoba, Brada, Yung et al, 2000 56	Recurrent GBM	109	Phase II TMZ	21.5 ± 31.1	32.7 ± 28.4
		89	Phase III TMZ	22.6 ± 29.4	41.0 ± 27.7
		90	Phase III PCB	23.4 ± 26.5	39.6 ± 25.9
Reddy et al, 2013 54	GBM	24	Pre RT	30.6 ± 35.3	33.3 ± 19.7
		16	Post (6 mo )	10.5	18.7
Taphoorn et al, 2005 55	GBM	248	RT	26.4 ± 1.9	—
		242	RT + TMZ	27.4 ± 2.1	—
Yavas et al, 2012 59	HGG	118		27.9 ± 28.2	28.3 ± 23.3

Abbreviations: LGG, low grade glioma; AA, anaplastic astrocytoma; GBM, glioblastoma; RT, radiotherapy; TMZ, temozolomide; PCB, procarbazine.

Table 4.

Disturbed sleep and drowsiness item scores of the MDASI-BT

Study	Tumor	n	Disturbed Sleep	Drowsiness
			Mean ± SD	Mean ± SD
Armstrong et al, 2006 35	Mixed PBT	201	2.82 ± 3.23	3.24 ± 3.05
Armstrong et al, 2009 34	MBT	124	2.65 ± 2.92	2.40 ± 2.83
Armstrong et al, 2011 36	Ependymoma	118	—	—
Tankumpuan et al, 2015 40	Mixed PBT	88	3.1 ± 3.0	—
Taychakhoonavudh et al, 2014 52	MBT	46	Baseline 2.30 ± 2.93	—
		17	Follow-up (6 mo) 2.00 ± 2.69

Clinical associations with sleep disturbance

Tumor Characteristics.

This review included studies of heterogeneous groups of the brain tumor population ranging from benign or malignant primary brain tumors (PBTs) to metastatic brain tumors (MBTs). They also varied in the point in the disease trajectory, as some studies examined newly diagnosed patients, while others examined long-term survivors or those in remission. No systematic patterns or differences in the nature of sleep disturbance between BT subgroups were noticeable. Indeed, studies of mixed BT samples included the tumor characteristic variables, such as tumor type or diagnosis, in their covariate or multivariate regression analyses but found no significant association of tumor-related variables with HRQoL outcomes, including sleep.^30,43 Furthermore, sleep-focused studies were limited to small samples of benign BT. There was little attempt to examine sleep symptoms using multiconstruct, standardized sleep assessments in the samples of malignant PBT and MBT patients.

Some studies explored whether BT location and laterality had a particular association with sleep regulation or related HRQoL. Patients with PBT have diverse clinical manifestations and impairments due to variability in underlying neurobiology, locations, and treatment regimens.^32,44 For instance, tumors located near the neural substrates that regulate sleep-wake rhythms and hormonal secretions may disrupt the patient’s sleep patterns or quality. However, such an effect of tumor laterality or location on sleep or the HRQoL outcomes was not found in PBT patients.^43,45–47 Two studies conducted in patients with PBTs in Finland reported that poorer sleep was described by patients with anterior than posterior⁴⁷ and bilateral than unilateral tumors.⁴⁵ However, such location or laterality effects on sleep were mediated by restlessness or depression.^45,47 There was no difference in sleep between tumor histology groups.⁴⁷

Surgery.

Cranial surgery was the initial treatment for brain tumors in almost all participants. Three studies specifically examined sleep and HRQoL before and after surgery^45,46 or different surgery types,²⁶ and 2 studies examined pre-surgery states of sleep and HRQoL.^43,48 Patients who underwent craniotomy showed poorer sleep quality and more sleep disturbance than the matched control patients who received lumbar discectomy, showing a negative impact of surgery in the brain.²⁶ In contrast, sleep disturbance improved significantly at 3 months and 1 year following the tumor resection surgery in within-subject comparisons.⁴⁵ However, the PSG conducted in a sample of 11 mixed glioma patients showed no changes in sleep quantity or patterns before and after the surgery.⁴⁶ Nevertheless, sleep disturbance and drowsiness were highly prevalent in PBT patients prior to surgery, as assessed by the QLQ-C30 and QLQ-BN20.^43,48 Other studies had patients with different surgical treatment profiles, and it was unclear whether previous surgery had any relationship with reported sleep disturbance.

Radiotherapy.

An interesting pattern of somnolence was seen in PBT patients following cranial irradiation treatment. In a cohort of 19 PBT patients (79% high-grade gliomas [HGGs]), 16 developed mild to severe somnolence.¹⁶ Somnolence had a distinct trajectory consisting of 2 phases. Its onset was seen at 2 weeks post-irradiation, reaching a peak over the next week, then it resolved. The symptom reappeared at 5 weeks post-irradiation lasting 2–5 days. This cyclic pattern of somnolence was similarly shown in a larger cohort of 68 PBT patients, with greater somnolence from 3 weeks and peaking at 6–12 weeks post-irradiation before resolving.⁴⁹ Both studies were conducted by the same research group, and it is unknown whether this pattern can be replicated in other centers.

The effects of whole brain radiotherapy (WBRT) or stereotactic radiosurgery (SRS) on sleep disturbance and drowsiness were studied in cohorts of MBT patients. Drowsiness was prevalent in MBT patients before and after WBRT.^42,50 In a slightly older cohort, significantly more drowsiness was reported at 12 weeks posttreatment than baseline.⁴² Studies assessing 2 different types of SRS in MBT patients had either a small sample size⁵¹ or short survival times.⁵² Consequently, groups in 2 SRS arms were combined for analysis, and it was not possible to draw a conclusion about their effects on sleep.

Radiotherapy with Concurrent Chemotherapy.

Radioth erapy with concurrent chemotherapy was a widely used treatment option for HGG patients. In a cohort of elderly glioblastoma patients (n = 65; age range, 70–81 y), changes in HRQoL, including sleep disturbance and drowsiness, were assessed pre- and postradiotherapy with concurrent chemotherapy.⁵³ While mean scores were unchanged for sleep disturbance, worse drowsiness was reported at a 6-month follow-up. Sleep disturbance improved in 29% for 6 months after treatment. Similar improvements in sleep disturbance and drowsiness were seen in a different cohort of GBM patients at different time points following treatment. However, improvements in sleep symptoms did not persist.⁵⁴

An international cohort of 544 glioblastoma patients treated with either radiotherapy alone (control) or concurrent radiotherapy and chemotherapy had their HRQoL assessed using the EORTC QLQs.⁵⁵ While the 2 groups did not differ on any of the HRQoL scales, including sleep and drowsiness at baseline, overall HRQoL was impaired in both groups. The most substantial impairments were seen in fatigue and sleep disturbance in glioblastoma patients. During treatment and follow-up assessments, there was no group difference in HRQoL. Patients in both treatment arms showed a trend to improved HRQoL that was not statistically significant.⁵⁵

Chemotherapy.

One study⁵⁶ compared the effects of 2 chemotherapy agents, temozolomide and procarbazine, on HRQoL outcomes of recurrent glioblastoma patients recruited for 3 different trials. Sleep disturbance scores (insomnia from QLQ-C30) did not change from baseline in patients receiving temozolomide regardless of tumor progression. However, drowsiness became worse in those patients who had tumor progression but improved in those without progression. Patients receiving procarbazine reported greater decline in HRQoL, including drowsiness, compared with temozolomide. Regardless of tumor progression, most HRQoL scores became worse in patients in the procarbazine trial. Although temozolomide had more favorable results than procarbazine in this study, more evidence is needed to determine the effects of temozolomide on sleep symptoms in BT patients.

Corticosteroids.

The use of corticosteroids in BT patients may disrupt patients’ sleep.⁵⁷ Dexamethasone use during radiotherapy was reported in a number of studies, but none found or reported significant effects of dexamethasone on sleep.^{32,42,55,58,59} One study examined the associations between dexamethasone use and HRQoL in a cohort of MBT patients (n = 68) undergoing WBRT.⁶⁰ Patients received a mean dose of 13.8 mg/day (range = 0.5–32) dexamethasone and concurrent palliative WBRT. Sleep was assessed by using 2 single-item sleep scales of HRQoL. They found that patients on a higher dose than 16 mg/day had a more disturbed sleep at week 2 during the treatment. But no other significant changes in sleep related to dexamethasone dose or duration were observed over time. Only a small number of patients remained at high doses of dexamethasone after week 2, rendering further interpretation difficult.

Physiological Assessments and Patient-Reported Outcomes of Sleep Disturbance

This review also aimed to address whether, and to what extent, objective and subjective assessments of sleep in BT patients agree with each other. A total of 6 studies used an objective assessment of sleep, 5 had patients with a benign BT involving the pituitary region. Among these 5 studies, 2 were case reports of 2 male patients with different subtypes of pituitary adenoma. They reported rare clinical presentations of narcolepsy developed post-irradiation⁶¹ or circadian rhythm sleep disorders which led to diagnosis of the pituitary tumor.⁶² The cases differed greatly in both demographic and clinical characteristics. In both cases, the PSG or ACT outcomes were used diagnostically and not compared with patient-reported outcomes.

The other 3 studies reported objective sleep outcomes obtained by PSG and/or ACT, as well as patient-reported outcomes using the PSQI, ESS, or sleep diary. These studies drew comparisons with matched control groups. There was no clear pattern of sleep disturbance among the objective data or between objective and subjective outcomes, even within the same study. For example, sleep of 17 nonfunctioning macroadenoma patients in long-term remission was assessed using the PSG, ACT, and sleep questionnaires.²⁵ They found that while PSG data showed normal duration of sleep and latency but poorer sleep efficiency of patients compared with controls, ACT data showed increased time spent in bed and sleeping, and a normal level of sleep efficiency in patients. However, patients self-reported having to wake more often during the night and suffering from more sleep disturbance than controls.

One study using PSG in a small cohort of glioma and meningioma patients (n = 11) focused on sleep-related breathing patterns and changes before and after tumor resection.⁴⁶ Clinically significant sleep apnea was detected in 7 patients. Sleep parameters were mostly unchanged post-operation, while sleep-related breathing improved in all patients. Although reduced mean sleep time (<5 h) overall and an increased rapid eye movement sleep percentage after resection was shown, no discussion of other sleep parameters was attempted. There was a lack of evidence of objectively assessed sleep outcomes, particularly in patients with malignant PBT or MBT, to answer the review question.

Sleep Intervention Targeted at Brain Tumor Patients and Caregivers

We failed to identify evidence of the development or examination of any interventions for managing sleep symptoms of BT patients and caregivers. However, a couple of studies briefly reported the use of sleep interventions. A case report revealed that oral melatonin was administered for 3 months (5–10 mg/day at 21:30 or 23:00 h) to improve irregular sleep-wake patterns of a pituitary microadenoma patient in Israel; however, no detailed outcomes were reported.⁶² Oral melatonin was stopped after 3 months due to ineffectiveness. Another study described a trial of treatment with continuous positive air pressure therapy or modafinil in craniopharyngioma patients with diagnoses of sleep apnea or somnolence, with effectiveness reported on an individual basis without formal sleep assessment (n = 13).³¹

Sleep Disturbance of Patients and Caregivers

The final review question asked how, and to what extent, sleep disturbance of patients and caregivers would be comparable and affect each other. Only one study²¹ examined sleep characteristics of 133 adult family caregivers within 1 or 2 months following the patients’ diagnosis of malignant PBT. Caregivers’ sleep was assessed using a wrist accelerometer (ACT) and the sleep quality subscale of the PSQI. In comparison to the parameters of healthy sleep for adults (ie, sleep duration of 7 hours for restoration, with a sleep onset <15 minutes after lying down, and <10% wake time during the nocturnal sleep period [the American Academy of Sleep Medicine]), the caregivers took significantly longer time (mean ± SD = 35 ± 34.5 min, P < .01) to fall asleep, and slept for less time (mean ± SD = 5 h 57 min ± 84.6 min, P < .01) than recommended. Caregivers had frequent awakenings during sleep (wake after sleep onset = 15.1%, P value not reported), and 48% of caregivers required a nap during the daytime. Poor sleep quality was reported by 32% (n = 42) of the caregivers. Significant associations were found between caregivers’ sleep disturbance and their level of anxiety and the functional status (KPS) of the patient. The patient’s tumor type or patient–caregiver relationship type was not associated with the caregiver’s sleep. Unfortunately, it was not possible to compare the patient and caregiver sleep disturbances and their influence on each other’s sleep, as no sleep outcomes were reported for the patients.

Discussion

The primary objective of this review was to gather evidence to identify types of sleep disturbance that are prevalent in adult brain tumor patients and report the level of severity. To our knowledge, this is the first systematic review to synthesize all evidence available from a broad range of nonpediatric brain tumor groups, from benign, to malignant, to metastatic BTs, captured by all measures of sleep disturbance. The majority of studies reviewed (41 of 50) were published in the last decade, suggesting recent growing awareness of sleep-related complications in BT patients and the clinical importance of their impact on quality of life.^43,63

Certain brain tumors are incurable and require ongoing treatments during the course of the illness. Toxicity or side effects of cancer treatment can cause a considerable amount of symptom distress in BT patients.⁴⁰ The use of patient-centered HRQoL assessments as an endpoint in clinical trials or palliative care settings is essential, as the potential impact on quality of life becomes increasingly important when choosing the optimal treatment.^55,63

This review demonstrated that sleep is an important element of HRQoL in patients with BT, and it is commonly disturbed. A high prevalence and severity of sleep disturbance and drowsiness in BT patients, along with fatigue, were indicated by the results of many HRQoL studies included in this review. This finding is consistent with the findings of a review of HRQoL in patients with PBTs and MBTs.⁶⁴ The authors performed a data synthesis of 14 studies to compute average scores of the EORTC QLQ-C30 and QLQ-BN20 scales using the total number of participants in each study as the weighted variable in an analysis of variance.⁶⁴ The mean (SD) insomnia (ie, disturbed sleep) scores of the QLQ-C30 were 22.16 (65.07) and 37.49 (28.40), which were the third and second highest symptom scores after fatigue, in PBT (n = 602) and MBT (n = 154), respectively.⁶⁴ Similarly, the mean (SD) drowsiness scores of the QLQ-BN20 were 24.83 (86.98) and 34.21 (34.08), which were the third and second highest symptom scores after future uncertainty, in PBT (n = 1579) and MBT (n = 154), respectively.⁶⁴

There are several limitations in the studies included in this review. Studies used a variety of measures in heterogeneous samples with many different clinical and treatment features. It was difficult to identify prevalent types, or clear clinical associations, of sleep disturbance because the results largely depended on an assessment of specific type of sleep disturbance in each study. Sleep outcomes were frequently reported as part of HRQoL outcomes of clinical trials, but as secondary and optional endpoints, and in an observational manner. The use of a single-item sleep scale within an HRQoL measure was common. Group differences in treatment arms on the sleep outcome were not usually reported. In addition, many studies suffered from the lack of statistical power due to small sample sizes or high attrition rates in longitudinal designs. This rendered study groups with a different exposure or clinical variable of interest to be later combined, and often only descriptive data were reported. This issue is particularly critical for research involving HGGs or brain metastases, which will have poor prognosis and rapid decline in physical and cognitive functioning. Caution is required when interpreting the results, as patients with higher functional status (KPS) and better prognosis tended to remain in the study longer for follow-up assessments. Patients with glioblastoma in one trial tended to drop out at the time of disease progression.⁵⁴ Compliant patients with HGG were more likely to have longer survival than noncompliant patients.⁵⁵ As a result, poorer performing patients may have been underrepresented, and overall performance of the population on HRQoL domains may be overestimated.

Evidence from comprehensive sleep-focused studies is scarce, particularly in patients with a malignant PBT or MBT. Formal sleep assessments were mostly conducted in small samples of benign tumors around the pituitary gland region. This part of the brain may be of interest given its neurophysiological involvement in the regulation of sleep. However, it will be difficult to exploit such data in drawing clinical implications for sleep management of patients with malignant brain tumors. In these studies, patients were in a long-term remission with nonfunctioning tumors or had their tumors surgically removed. These tumors lack clinical features of malignant tumors that may affect sleep. In fact, gliomas are the most common primary CNS cancer in adults,⁶⁵ 75% of which are HGG.⁶⁶ Glioblastoma itself comprises 46% of all malignant PBTs in the US.⁶⁶ Also, brain metastases are common from lung or breast cancers.⁴² The fact that HGG and MBT generally carry a poor prognosis and low functional status will hinder the use of overnight PSG assessment in a sleep laboratory. The use of standardized, multifaceted patient-reported sleep outcome assessments in these patients will be of value for research, as well as for potential use in routine cancer care.

There was no study which examined an intervention specifically targeted at improving the sleep of BT patients and/or their caregivers. It is not surprising considering that little research has been done to determine the nature, prevalence, and etiology of sleep disturbance in neuro-oncology populations. We cannot rule out the possibility of effectiveness and successful implementation of sleep interventions in BT populations that have been used in the general or other cancer populations (see a review⁶⁷ of treatment options of insomnia in cancer). However, it is important to build a strong evidence base for the development of a suitable and effective intervention or management program tailored to patients with a brain tumor and their caregivers.

In conclusion, this systematic review highlights an emerging picture from studies reporting HRQoL outcomes that sleep disturbance is a highly prevalent and severe symptom causing distress in patients with brain tumor. Much of the available evidence is limited to describing the presence of sleep disturbance as part of overall performance status of patients across different treatment trajectories. There is a lack of research establishing a comprehensive understanding of the patterns, prevalence, severity, and risk factors of sleep disturbance in patients with malignant brain tumor and their caregivers and little assessing associations in the patient–caregiver sleep experience. Future research is warranted to perform a more systematic examination of sleep disturbance using standardized, multifaceted sleep assessments to inform the development of better symptom management programs in this population.

Supplementary Material

Supplementary material is available at Neuro-Oncology online.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sector.

Conflict of interest statement. The authors have no conflicts of interest to declare.