Does Effective Management of Sleep Disorders Reduce Cancer-Related Fatigue?

Abstract

Cancer and cancer therapy are often associated with symptoms such as fatigue and sleep disturbances, before, during and after therapy. These symptoms of fatigue and poor sleep often occur in parallel having a significant impact on the physical functioning of patients with cancer. A strong correlation between cancer-related fatigue (CRF) and sleep has been observed in several studies, suggesting that they may be reciprocally related. The co-clustering of these symptoms suggests that they may have similar underlying aetiology and that treatments targeting either symptom may positively affect the other. Studies examining these clusters have shown that these symptoms often co-vary together. The potential mechanisms that link the relationship between insomnia and CRF are intriguing but require further investigation. Despite the high prevalence of insomnia and the often bidirectional relationship between poor sleep and fatigue, there are limited data to support the use of sleep management interventions as a means to reduce fatigue in patients with cancer. Assessment of the available evidence across trials is complicated by different study designs, patient selection criteria, stage of cancer treatment and by the nature of the interventions studied. Improvements from baseline in both sleep parameters and CRF have been documented in a limited number of studies, including two randomized-controlled trials using cognitive behavioural therapy for insomnia (CBT-I). In contrast, the efficacy of pharmacological therapies in reducing both insomnia and CRF is largely lacking. Clearly, treating clinically significant insomnia is likely to have benefits for the patient with cancer and for those who are recovering from cancer. In particular, pharmacotherapies for insomnia, singly or in combination with CBT-I, should be evaluated in multicentre randomized clinical trials to examine their efficacy in improving sleep quality and reducing associated CRF.

1. Introduction

Cancer and cancer therapy are often associated with fatigue and sleep disturbances, before, during and after treatment that significantly impact upon quality of life.^[1–5] These symptoms often occur together,^[6–10] having significant impact on the mental and physical functioning of patients with cancer.^[11] For example, a retrospective analysis of data from 263 patients with cancer undergoing chemotherapy showed that insomnia, fatigue, depression and anxiety were positively correlated with each other, and negatively correlated with health-related quality of life.^[12] Furthermore, compared with patients with cancer having no co-morbid conditions, those with two or three co-morbid conditions had a greater risk of lower physical functioning.^[11]

Cancer-related fatigue (CRF) has been reported in 33–66% of patients with cancer.^[8,13,14] According to the National Comprehensive Cancer Network (NCCN), CRF is defined as ‘a distressing persistent subjective sense of physical, emotional and/or cognitive tiredness or exhaustion related to cancer or cancer treatment that is not proportional to recent activity and interferes with usual functioning.^[6] Although the aetiology of CRF is unknown, a variety of biological mechanisms have been proposed, including serotonin (5-HT) neurotransmitter dysregulation, vagal afferent activation, alterations in muscle and adenosine triphosphate metabolism, hypothalamic–pituitary–adrenal (HPA) axis dysfunction, circadian rhythm disruption, anaemia and inflammation.^[15–22]

Sleep disturbances, such as difficulty falling asleep, problems maintaining sleep and poor sleep quality, are also frequently reported in patients with cancer.^[14,23,24] For example, of 100 oncology outpatients undergoing palliative care, 72% experienced sleep disturbances, the most frequent being not feeling rested (72%), difficulty staying asleep (63%) and difficulty falling asleep (40%).^[23] Factors that may potentially affect sleep in patients receiving cancer therapy include side effects of chemotherapy and radiation, immunological response to neoplastic growth, and other cancer-related physical symptoms such as decreased activity levels due to CRF and pain, in addition to the prevalent psychological responses of anxiety and depression in these patients.^[25]

A strong correlation between CRF and sleep has been observed in many studies, suggesting that they may be reciprocally related. In a literature review by Roscoe and colleagues,^[25] 23 of 24 studies supported the correlation between the two symptoms and suggested that sleep disorders may be both a cause of and caused by fatigue. Cause and effect is difficult to determine, but these two conditions are often co-morbid. The co-clustering of these two symptoms suggests a common underlying aetiology and that treatments targeting either condition could positively affect the other.^[25] Given the strong relationship between fatigue and insomnia with depression,^{[8,23,25–32]} however, this association could be explained solely by the presence of co-morbid depression. In support of this interpretation, a study in patients receiving chemotherapy for cancer showed that although insomnia, fatigue, depression and anxiety were positively correlated with one another, multiple regression analyses showed that most of the variance was explained by depression, with only 4% explained by the insomnia and CRF.^[12] In addition, studies using antidepressants for hot flushes and associated symptoms in patients with cancer also show benefits in CRF and insomnia symptoms.^[31,33,34] On the other hand, in patients who are not depressed, sertraline was ineffective at relieving fatigue in patients with cancer,^[35] suggesting that depression and CRF are indeed separable entities in this population.

Despite the uncertainty surrounding the precise relationships between sleep and CRF, the NCCN has identified sleep disorders as one of the seven factors that often influence CRF, and recommended treatment of sleep disturbances in patients undergoing active cancer treatment, patients on long-term follow-up and patients undergoing palliative and hospice care.^[6]

The 6th annual meeting of The International Sleep Disorders Forum: The Art of Good Sleep, held in 2008, evaluated the level of evidence demonstrating that effective management of sleep disorders reduces fatigue symptoms in patients with cancer. Based on the available published evidence, discussion focused primarily on the effects of cognitive-behavioural therapy for insomnia (CBT-I, a non-pharmacological intervention) and modafinil (a pharmacological intervention indicated in excessive daytime sleepiness in narcolepsy, sleep apnoea and shift work sleep disorders) for reducing symptoms of fatigue in patients with cancer.

Given already existing reviews and literature on meta-analyses of the non-pharmacological^[36–38] and pharmacological treatment of fatigue in patients with cancer,^[39,40] this article will focus primarily on studies that have examined the effect of non-pharmacological and pharmacological sleep interventions on associated fatigue symptoms and insomnia. In addition, pharmacological interventions for insomnia as a whole in patients with cancer will be reviewed as, despite the widespread use of hypnotic agents (such as benzodiazepines, non-benzodiazepine hypnotic agents, sedating anti-depressants, anti-histamines, neuroleptics and chloral hydrate) in palliative care practice,^[41] it is surprising that there have been no large scale randomized controlled trials of hypnotic medications in patients with cancer and insomnia.^[42] There is also some evidence to suggest that such medications may be associated with adverse effects. For example, in a study of 909 outpatients and inpatients with cancer, the use of sleeping pills and/or tranquillizers in the preceding week was associated with substantially lower quality of life indices and increased severity of symptoms (fatigue, pain, dyspnoea, constipation and paradoxically insomnia itself), even after confounding factors (female gender, analgesic usage, cardiovascular disease and malignant disease severity) were taken into account.^[43] Adverse effects of hypnotic agents for the treatment of insomnia in patients with cancer may result from a number of factors, including, impaired hepatic metabolism, resulting exacerbation of adverse effects such as daytime sedation, interaction of sedation with CRF amplifying the impact of this symptom, and impaired cognitive functioning reducing the capacity to manage day-to-day life with cancer.

2. Methods

During The Art of Good Sleep 2008 meeting, a workshop entitled ‘What is the level of evidence to demonstrate that effective management of sleep disorders may reduce fatigue symptoms in patients with cancer?’ was held in order to discuss the topic and to identify areas of focus. The workshop was moderated by the authors of this article and was attended by an international panel of sleep specialists (see Acknowledgements). Some of the most important studies on this topic were reviewed and areas for further research were defined. During the workshop a number of articles that were relevant to the topic were discussed. Subsequent to the workshop, a structured literature search was conducted to identify other relevant articles to complement those discussed in the workshop.

2.1 Search Strategy

English language articles limited to adults were searched in March 2009 using PubMed (no date restriction) with the following terms: ‘behavior therapy’ OR ‘cognitive therapy’ OR ‘hypnotics and sedatives’ OR ‘benzodiazepines’ OR ‘narcotics’ OR ‘antidepressive agents’ OR ‘neurotransmitter uptake inhibitors’ OR ‘trazodone’ OR ‘mirtazapine’ OR ‘antipsychotic agents’ OR ‘histamine antagonists’ OR ‘chloral hydrate’ OR ‘melatonin’ OR ‘ramelteon’ OR ‘serotonin antagonists’ OR ‘receptors, GABA-A’ OR ‘GABA agonists’ AND ‘sleep disorders’ OR ‘sleep’ OR ‘fatigue’ OR ‘fatigue syndrome, chronic’ OR ‘fatigue’ AND ‘neoplasms’ OR ‘neoplasms’ OR ‘cancer’ OR ‘malignant disease’ OR ‘terminal care’ AND ‘clinical trial’ OR ‘randomized controlled trial’ OR ‘meta-analysis’ AND ‘adult’.

This strategy identified 79 articles for further study. Additional publications and abstracts were identified from the workshop held at The Art of Good Sleep 2008 meeting and the reference lists of the articles identified.

3. Results

Fully published articles and abstracts of studies were selected from the literature search assessing the effect of non-pharmacological and pharmacological interventions on the interaction between sleep disorders and fatigue in patients with cancer, together with articles and abstracts assessing pharmacological interventions in insomnia as a whole. There is an extensive literature on the effect of non-pharmacological interventions such as exercise programmes and other complementary therapies (e.g. massage therapy, yoga and muscle relaxation) on CRF,^[6,36] but discussion of these studies is beyond the scope of this review unless associated sleep outcomes were also reported. Similarly, a number of studies have examined the effect of pharmacological interventions (e.g. methylphenidate) in CRF without insomnia,^[39,40] so again these studies fall outside the scope of this review.

Table I summarizes the results of clinical trials investigating non-pharmacological therapy for insomnia and its impact on CRF: six using CBT-I,^[44–49] one using behaviour therapy,^[50] one using mindfulness-based stress reduction (MBSR),^[51] one using psycho-education^[52] and one using an exercise intervention.^[53] An additional study of a CBT-I trial in patients with breast cancer was identified, but although fatigue was mentioned in the text, outcomes for this domain were not reported, so the results are not presented here.^[54]

Table I.

Efficacy of non-pharmacological treatments for insomnia and associated fatigue symptoms in patients with cancer

Study [ref] (design)	Cancer type (treatment)	Treatment group [intervention] (n)	Treatment dose	Outcomes
Berger et al.[50] (30-day, r, c, mc, 8-week f/up)	Breast (chemotherapy)	BT [SC/SH/SR/RT](113) Control [HE] (106)	Week 1: 30 minutes Week 2: 15 minutes Week 3: 15 minutes Week 4: 30 minutes Week 5–8: none	BT significantly improved sleep quality compared with HE but had no impact on CRF symptoms at endpoint The BT group showed improvements in sleep quality on the PSQI (group × time interaction, p = 0.049) and reductions in night awakenings on actigraphy (group × time interaction, p = 0.03). Fatigue symptoms measured using the PFS actually increased in both groups equally during the intervention but returned to baseline at endpoint
Carlson and Garland[51] (8-week, u)	Mixed (not specified)	MBSR [MBSR] (63) No control group	Week 1–8: 90 minutes	MBSR was associated with significant changes between pre and post-intervention for both sleep and CRF symptoms Significant changes post-intervention were found for PSQI scores (p < 0.001), POMS fatigue (p < 0.001), POMS depression (p = 0.001) and stress (SOSI, p < 0.001). There was no significant relationship between improvements in fatigue and sleep, but there was for stress and depression
Davidson et al.[49] (8-week, u)	Mixed (completed)	CBT-I [SC/SH/RT/CT] No control group	Week 1: 60 minutes Week 2: 90 minutes Week 3: 90 minutes Week 4: 90 minutes Week 5: 60 minutes Week 8: 60 minutes	CBT-I significantly improved insomnia and CRF symptoms versus baseline Sleep diary measures of number of awakenings, wake time after sleep onset were reduced and total sleep time and quality increased significantly at both 4 and 8 weeks. Using the CQoLQ-30, fatigue was also significantly improved at 8 weeks (p = 0.022)
Dirksen and Epstein[44] (6-week, r, c)	Breast (completed)	CBT-I [SC/SR/SH/CT] (40) Control [SH only] (41)	Week 1: 180 minutes Week 2: 60 minutes Week 3: 60 minutes Week 4: 60 minutes Week 5: 15 minutes Week 6: 15 minutes	CBT-I significantly improved insomnia severity and CRF versus baseline The ISI was improved from baseline by both CBT-I (14.38 versus 23.91) and control treatment (16.31 versus 22.71). Fatigue (POMS), anxiety (STAI) and depression (CES-D) measures improved significantly (p ≤ 0.05) with CBT-I relative to baseline 2 weeks after therapy, unlike the control group. HR-QoL measures in both groups improved significantly (p ≤ 0.05) from baseline. Statistically significant (p < 0.05) post-treatment correlations of insomnia severity with fatigue were noted for both CBT-I (0.66) and control treatment (0.40), but not with depression
Espie et al.[45] (5-week, r, c, mc, 6 month f/up)	Mixed (completed)	CBT-I [SC/SR/CT](100) Control [TAU] (50)	Week 1–5: 50 minutes	CBT-I was significantly better than normal clinical practice (control) for sleep and CRFa Sleep diary measures of sleep onset latency and wake time after sleep onset were significantly reduced (p < 0.001), and sleep efficiency significantly increased (p < 0.001), with CBT-I relative to control after 5 weeks of treatment, with the effects sustained at 6-month f/up (p ≤ 0.001). Objective actigraphic assessments confirmed reductions in sleep onset latency but total sleep time was significantly decreased after CBT-I therapy reflecting the SR component of treatment. Fatigue (FSI), anxiety and depression (HADS) and HR-QoL significantly improved after treatment with CBT-I relative to control, with the effects sustained at the 6-month f/up
Payne et al.[53] (14-week, r, c)	Breast (hormone)	Exercise [EX] (10) Control [NT] (10)	Week 1–14: 20 minutes 4 times a week	Exercise significantly improved sleep but had no effect on CRF The exercise intervention resulted in significant improvements in PSQI scores (p = 0.007) with reduced actigraphic wake time (p= 0.02) and movement during sleep (p = 0.002) compared with control. Fatigue symptoms were not improved. No change in depressive symptoms (CES-D) was noted but baseline levels were in the normal range
Quesnel et al.[47] (8-week, u, 6 month f/up)	Breast (completed)	CBT-I [SC/SR/SH/CT/FM] (10) No control group	Week 1–8: 90 minutes	CBT-I significantly improved sleep and CRF symptoms Sleep efficiency and total wake time, assessed by both diary measures and polysomnography, were significantly improved post-intervention and at 6-month f/up. General (p = 0.016) and physical (p = 0.008) fatigue (MFI), depression (BDI, p = 0.004) and quality of life (CQoLQ-30, p = 0.016) all significantly improved post-intervention and at 6-month f/up
Savard et al.[46] (8-week, r, c)	Breast (completed)	CBT-I [SC/SR/SH/CT/FM] (27) Control [WLC] (30)	Week 1–8: 90 minutes	CBT-I significantly better than wait list control for sleep; no significant difference for CRFa Sleep diary measures such as sleep efficiency (p < 0.0001), total wake time (p < 0.001), sleep onset latency (p < 0.05), wake after sleep onset (p < 0.0001) and ISI (p < 0.05) improved significantly with CBT-I relative to control; total sleep time did not increase significantly after CBT-I therapy. These improvements were maintained at 12-month f/up. No significant differences were found on polysomnographic assessment. Anxiety and depression (HADS) and global HR-QoL improved significantly after treatment with CBT-I relative to control. Although fatigue measures were reduced from baseline, there was no significant difference between CBT-I and control groups
Simeit et al.[48] (4-week, c, 6 month f/up)	Mixed (rehabilitation immediately post-treatment)	CBT-I [SC/SH/CT/PMR] (80) CBT-I [SC/SH/CT/AT] (71) Control [TAU] (78)	Day 1–3: 60 minutes of SC/SH/CT Week 1–4: daily PMR or AT	CBT-I including relaxation training significantly improves sleep, CRF and quality of life. This study compared CBT-I using two forms of relaxation training, PMR and AT with controls. PSQI scores showed reduced sleep latency, increased sleep duration, sleep efficiency, sleep quality (p < 0.001), reduced use of sleep medication and increased daytime functioning (p < 0.05) over the 6 month period in the intervention group. CQoLQ-30 scores showed improvements in global quality of life and fatigue post-intervention which were maintained at 6-month f/up
Williams and Schreier[52] (r, c, 3 month f/up)	Breast (chemotherapy)	Education [HE/EX/RT] (38) Control (33)	40-Minute audiotape	Self-education comprising healthy eating, exercise and relaxation training halved sleeping difficulties but had no effect on CRF. Using a self-care diary measure, the intervention group showed »50% reduction in sleeping difficulties. Fatigue increased in both groups

^aBased on intention to treat population.

AT= autogenic training; BDI = Beck Depression Inventory; BT = behaviour therapy; c = controlled study; CBT-I = cognitive behaviour therapy for insomnia; CES-D = Center for Epidemiologic Studies – Depression Scale; CQoLQ-30 = Cancer Quality of Life Questionnaire-30; CRF= cancer-related fatigue; EX = exercise; FM= fatigue management; FSI = Fatigue Symptom Inventory; f/up = follow-up; HE= healthy eating; HADS= Hospital Anxiety and Depression Scale; HR-QoL = health-related quality of life; ISI = Insomnia Severity Index; MBSR= mindfulness-based stress reduction; mc= multicentre; MFI = Multidimensional Fatigue Inventory; NT = no treatment; PFS Piper Fatigue Scale; PMR= progressive muscular relaxation; POMS= Profile of Mood States; PSQI = Pittsburgh Sleep Quality Index; r = randomized; RT = relaxation therapy; SC = stimulus control; SH= sleep hygiene; SOSI = Symptoms of Stress Inventory; SR = sleep restriction; STAI = State-Trait Anxiety Inventory; TAU = treatment as usual (no help with insomnia); u = uncontrolled study; WLC= Wait List Control.

CBT-I, a well-established treatment that has shown comparable efficacy to pharmacotherapy in the treatment of primary insomnia,^[55] was the approach used in the majority of the studies identified. CBT-I consists of a behavioural and cognitive component. The behavioural components include stimulus control, sleep restriction, sleep hygiene and in some cases relaxation training. The cognitive component of CBT-I addresses dysfunctional beliefs and excessive rumination about sleep difficulties that unless challenged can interfere with the capacity to implement behaviour change.

Other non-pharmacological interventions used in these studies included behaviour therapy (as for CBT-I, but without the cognitive component), MBSR, exercise and education. MSBR is based on the construct of mindfulness, defined as both the capacity to attend fully to the present moment and the ability to maintain an attitude of non-judgemental acceptance over experience. MBSR aims to develop these cognitive skills through a combination of educational approaches, the teaching of meditation and yoga, and the use of group support and problem solving. In a meta-analysis, MSBR has been shown to be effective for stress reduction in a wide range of clinical disorders.^[56]

Table II summarizes the characteristics of clinical trials investigating both the pharmacological treatment of insomnia and trials investigating insomnia and associated fatigue in patients with cancer. Only two trials investigating the use of benzodiazepine hypnotic agents for insomnia in patients with cancer were identified and both of these assessed insomnia exclusively in the terminal care setting.^[57,58]

Table II.

Efficacy of pharmacological treatments for insomnia and/or associated fatigue symptoms in patients with cancer

Study [ref] (design)	Cancer type (treatment)	Co-morbid condition	Treatment group [route] (n)	Treatment dose	Outcomes
Insomnia outcomes as prime focus
Ehsanullah et al.[57] (5/5-day, db, r, c, co, sc)	Mixed (terminal)		Diazepam [rectal] (24) Diazepam [rectal] (24)	5 mg 10 mg	Patients with terminal cancer received 5mg diazepam suppositories for 5 days, followed by a 2-day wash out period, then 10mg for a further 5 days. Seven patients died during the 12-day study period. 15/19 in the 5mg group and 17/21 in the 10mg group were rated by nurses to have slept well. There was a very high incidence of side effects; 58% experienced daytime drowsiness, 29% dry mouth and 21% amnesia
Matsuo and Morita[58] (retrospective audit, mc)	Mixed (terminal)		Midazolam [intravenous] (104) Flunitrazepam [intravenous] (59)	≤18 mg ≤2 mg	Efficacy measurement by physician-agreed 3-point rating scale (good, fair or poor) with established interrater reliability (κ = 0.68). Efficacy, defined as a good or fair rating, was 91% for midazolam and 81% for flunitrazepam. Adverse events were prominent; hangover effects (34% versus 19%), delirium (~12% in both groups), respiratory depression (3.8% versus 17%, p = 0.007), and dose escalation (11% versus 2.6%, p = 0.015)
Insomnia outcomes reported where another co-morbid condition was the prime focus
Kim et al.[59] (4-week, ol, u, sc)	Mixed (not reported)	Depression	Mirtazapine [oral] (42)	15–45 mg	MADRS depression ratings fell (33.2 ± 10.7 versus 22.5 ± 12.5, p < 0.001) reciprocally with improvements in quality of life EQ-5D VAS (41.9 ± 20.9 versus 54.1 ± 23.5, p < 0.01). Using the C-LSEQ, total sleep time increased from 3.6 ± 1.9 to 6.8 ± 2.5 hours (p < 0.001). Initial exacerbation of daytime sleepiness was seen in 36% of patients that resolved by the end of the study period. In addition, there were substantial drop-outs (59%) from the study limiting its findings
Stearns et al.[60] (6-week, ol, u, sc)	Breast (completed)	Hot flushes	Paroxetine [oral] (27)	10–20 mg	Over the 6-week study period, mean reduction in the frequency of hot flushes was 67%(95%CI 56–79%) and in hot flush severity was 75%(95% CI 66–85%). In addition, depression (CES-D mean change −4.2, 95% CI −0.7 to −7.6, p = 0.02), sleeping difficulties (MOS mean change −14.6, 95%CI −6.4 to −22.8, p = 0.0002), anxiety (HADS mean change −1.8, 95% CI −0.9 to −2.7, p = 0.0005) and quality of life (EuroQoL mean change 5.9, 95% CI 2.1–9.7, p = 0.004) all improved. There were substantial side effects with approximately double the numbers of women reporting somnolence (60%), dizziness (43%) and nausea (30%) at study end compared with baseline
Insomnia and CRF outcomes reported
Grassi et al.[61] (8-week, ol, u, sc)	Breast (not reported)	Depression	Reboxetine [oral] (20)	2–10 mg	Over the 8-week study period reboxetine showed significant falls in HAM-D ratings (p < 0.001), hopelessness (p < 0.001) and anxious preoccupation (p < 0.01); together with global improvements in quality of life (p < 0.001) and sleep (p < 0.01) but not fatigue as measured by the CQoLQ-30. Despite sleep improving overall, transient initial insomnia was reported in one third of patients
Carpenter et al.[33] (6/6-week, db, pc, r, co, mc, 12-month f/up)	Breast (completed)	Hot flushes	Venlafaxine/placebo [oral] (26) Placebo/venlafaxine [oral] (26) Venlafaxine/placebo [oral] (9) Placebo/benlafaxine [oral] (9)	37.5 mg 37.5 mg 75 mg 75 mg	In this study, patients with depression were specifically excluded. Hot flushes were measured objectively by 24-hour ambulatory sternal skin conductance (SSC) monitoring and subjectively by event monitoring and diary measures. Low-dose venlafaxine was associated with a 22% reduction in hot flush frequency using SSC compared with placebo (effect size, 0.16; p ≤ 0.001). Equivalent figures for the high-dose regime were 14% (effect size 0.22, p = 0.013). No significant changes in fatigue (POMS fatigue subscale) or sleep (PSQI) were demonstrated. Although side effect profiles were not significantly different between active treatment and control groups, most patients discontinued use by 12-month f/up
Loprinzi et al.[62] (5-week, ol, u, sc)	Breast and prostate (maintenance tamoxifen or antiandrogens)	Hot flushes	Venlafaxine [oral] (28)	12.5mg twice daily	Average daily frequency of hot flushes fell from 6.6 at baseline to 4.3 at 5 weeks, with severe flushes falling from 1.4 to 0.1 per day (p < 0.0002). Over the study period, trouble sleeping fell (52% versus 22%) but not significantly (p = 0.07) and fatigue symptoms were reduced (48% versus 8%, p= 0.01)
Morrow et al.[63] (1-month, ol, u, sc)	Breast (completed)	Not Reported	Modafinil [oral] (51)	200 mg	At 1 month, the mean fatigue severity index (0–10 scale) was significantly reduced from baseline (3.7 versus 6.9; p < 0.01) with 86% of patients having a ‡1-point improvement. The mean global effectiveness rating was 5.0 (7-point scale; 1 = no benefit, 7 = great improvement). Approximately one-half of patients reported improvements in sleep and less daytime drowsiness, whereas two-thirds reported improvements in general activity, mood, walking ability, normal work ability, social relations and enjoyment of life
Stearns et al.[34] (4/4-week, db, pc, r, co, mc)	Breast (completed)	Hot flushes	Paroxetine/placebo [oral] (37) Placebo/paroxetine [oral] (39) Paroxetine/placebo [oral] (38) Placebo/paroxetine [oral] (37)	10 mg 10 mg 20 mg 20 mg	Paroxetine 10 mg reduced hot flush frequency by 40.6% compared to 13.7% for placebo (p = 0.0006). Paroxetine 20 mg reduced hot flush frequency 51.7% compared with 26.6% for placebo (p = 0.002). Although both doses appeared equally efficacious, patients were more likely to discontinue 20 mg paroxetine, possibly because the 20 mg dose was associated with increased nausea above baseline (41% versus 10%, p < 0.001). Paroxetine 10 mg was associated with a significant improvement in sleep compared with placebo (p = 0.01), but no change in fatigue symptoms could be demonstrated
Weitzner et al.[31] (4-week, ol, u, sc)	Breast (chemotherapy ~ 60%; completed ~ 40%)	Hot flushes	Paroxetine [oral] (13)	10–20 mg	Mean ratings of hot flush severity decreased significantly between pre and post-intervention assessments (3.62 versus 2.08, p = 0.002). Improvements in fatigue using the MFSI (35.7 versus 20.2, p = 0.01), sleep using the PSQI (1.85 versus 0.77, p = 0.0003) and depression using the CES-D (25.67 versus 10.83, p = 0.0009) were also found to be significant. Side effects were not commented on

c = controlled study; CES-D = Center for Epidemiologic Studies Depression Scale; C-LSEQ = Chonnam National University Hospital–Leeds Sleep Evaluation Questionnaire; co = crossover; CQoLQ-30 = Cancer Quality of Life Questionnaire 30; CRF= cancer-related fatigue; db = double-blind; EQ-5D VAS= EuroQol-5D visual analogue scale; f/up = follow-up; HADS= 7-item anxiety sub-scale from the Hospital Anxiety and Depression Scale; HAM-D = 17-item Hamilton Rating Scale for Depression; MADRS= Montgomery Å sberg Depression Rating Scale; mc= multicentre; MFSI = Multidimensional Fatigue Symptom Inventory; MOS= Medical Outcomes Study 9-Item Sleep Scale; ol = open-label; pc = placebo-controlled; r = randomized; sc = single centre; u = uncontrolled study.

Two other trials reported insomnia outcomes in patients with cancer in the context of treatment for other co-morbid conditions, notably the use of mirtazapine for the treatment of nausea in depressed patients^[59] and paroxetine for the treatment of hot flushes.^[60] Hot flushes are distressing symptoms that often disrupt sleep in cancer survivors, occurring primarily in women whose treatment has resulted in ovarian failure with consequent premature menopause. These symptoms can also occur in men, for example in prostate cancer following orchidectomy or the chemical suppression of androgen secretion. Six further trials were identified that reported both insomnia and fatigue outcomes in patients with cancer; one using reboxetine in the treatment of co-morbid depression,^[61] two using venlafaxine for the treatment of hot flushes,^[33,62] two using paroxetine for the treatment of hot flushes^[31,34] and one using modafinil for the specific treatment of CRF.^[63] A final trial of mirtazapine reporting insomnia outcomes in the treatment of hot flushes was identified but, as approximately 40% of the study participants were post-menopausal women with no history of breast cancer, the trial was excluded.^[64]

4. Discussion

Many of the studies identified in this review involved patients with endocrine cancers, in particular breast cancer. It is possible that the effects of non-pharmacological and pharmacological interventions on associated fatigue symptoms and insomnia may differ depending of the anatomical location of the cancer and specific treatment demands.

4.1 Non-Pharmacological Treatments

Two trials examined the effect of interventions for insomnia and associated fatigue symptoms in breast patients with cancer undergoing active chemotherapy. Whereas both of these trials, one using an educational^[52] and one a behaviour therapy approach,^[50] showed efficacy for insomnia symptoms, CRF levels increased. This may be due to the adverse effects of concurrent medications as fatigue levels returned to baseline upon cessation of chemotherapy. Future trials also need to focus on assessing the efficacy of insomnia interventions in populations of patients with cancer who are undergoing cancer therapy or who have completed their treatment.

The remaining trials, largely but not exclusively on populations of breast cancer survivors who had completed treatment, varied substantially in methodological quality. Three uncontrolled trials, one using MBSR,^[51] two using CBT-I^[47,49] and one controlled but non-randomized trial using CBT-I,^[48] all showed improvements in both insomnia and CRF symptoms. The results of three randomized controlled trials of CBT-I, however, were mixed; two showing positive effects for both insomnia and CRF^[44,45] and one solely for insomnia symptoms.^[46] A final trial in this group using an exercise intervention showed benefit for insomnia but not CRF symptoms.^[53] Of the trials employing CBT-I as the prime intervention, five out of six showed positive effects on both insomnia and CRF symptoms, providing substantial evidence in support of the hypothesis that an intervention aimed primarily at insomnia can have secondary effects on CRF. In addition, two studies employed a specific fatigue management component to the CBT approach, with one showing benefit^[47] and the other not.^[46] These mixed results highlight the complex relationship between sleep disturbances and CRF, and indicate that, although sleep is likely to play an important role in CRF, and the two conditions are concurrently co-morbid, CRF is likely to result from multiple aetiologies.

Whether there is a direct, reciprocal relationship between the relief of insomnia and improvement in CRF cannot as yet be determined, as this association may be better explained by a third variable correlated with both, notably depressed mood. Of the above studies in which elevated levels of depressive symptoms were present, all showed beneficial effects of insomnia intervention on these symptoms, which could have provided an alternative explanation for the mechanism of CRF reduction. Interestingly, one study showed that improvements in CRF symptoms were related to improvements in stress and depression scores but not to subjective measures of sleep quality.^[51] The same authors provide evidence that the intervention used in their study also resulted in reduced salivary cortisol secretion.^[65] This raises the possibility that depression and associated excessive HPA axis activation may at least partly underpin the association between insomnia and fatigue in patients with cancer. Further longitudinal and intervention studies of insomnia treatment in patients with cancer, with and without associated depression and measuring appropriate biomarkers, will be needed to improve our understanding of the direction of these effects.^[66,67]

4.2 Pharmacological Interventions

As outlined in the introduction, there is a paucity of trials examining the efficacy of pharmacotherapies for the treatment of insomnia in patients with cancer. Only two studies were identified, one using rectal diazepam^[57] and the other intravenous midazolam/flunitrazepam^[58] both in terminal care settings. Although these treatment approaches were deemed efficacious in terms of night sedation, they carried with them a high incidence of adverse effects, particularly in terms of daytime cognitive impairment. This evidence, together with the finding that patients with terminal cancer show improvements in cognitive function with no rebound insomnia upon withdrawal of hypnotic medication,^[68] raises concerns about sedative hypnotic use in this population.

Studies in non-terminal patients with cancer also varied substantially in methodological quality. Two open-label, uncontrolled trials reported improvements in insomnia symptoms in the co-morbid treatment of depression with mirtazapine^[59] and in the co-morbid treatment of hot flushes using paroxetine.^[60] In the former study there were substantial dropouts limiting the interpretability of the findings, together with a large number of patients experiencing initial daytime sleepiness, and in the latter, side effects of daytime somnolence, dizziness and nausea were doubled compared with baseline.

No study directly addressed the question of whether pharmacological treatment of insomnia specifically reduces CRF. Only associated improvements in both symptom domains during the course of treatment for co-morbid depression and hot flushes, or CRF itself have been reported. Four uncontrolled open-label studies were identified in this group, three of which reported improvements in both insomnia and CRF^[31,62,63] and one reported only improvements in insomnia symptoms.^[61] It is notable, however, that the two well-designed randomized controlled trials, showed improvements in insomnia, but failed to demonstrate an improvement in CRF symptoms.^[33,34] This is in contrast to the positive effect on both insomnia and CRF symptoms in the majority of the CBT-I trials, a finding that may be explained by the daytime somnolence associated with many of the hypnotic agents, further exacerbating CRF. Interestingly, a trial using the wake-promoting agent modafinil for CRF showed reduced insomnia symptoms.^[63]

In addition, there is a need to evaluate the efficacy of non-sedating pharmacological treatments for insomnia, such as selective 5-HT_2A and orexin antagonists and melatonin receptor agonists,^[69] singly or in combination with CBT-I, for the relief of insomnia in patients with cancer with associated fatigue. Interestingly, mirtazapine, which has potent 5-HT_2A antagonist activity,^[70] is effective in relieving insomnia and nausea in patients with cancer with depression but, as a result of its non-selective activity on other receptor classes, is associated with a high incidence of sedative effects.^[59,71] Clearly, definitive clinical trials in patients with cancer with co-morbid insomnia are needed to identify alternative non-sedating pharmacotherapies for the treatment of insomnia in this population.^[43]

Finally, co-morbid depression may influence the associations between treatment effects and improvements in insomnia and CRF symptoms in both behavioural and pharmacological treatment studies. It is noteworthy that no changes in insomnia or CRF symptoms could be demonstrated in the randomized controlled trial examining the effect of venlafaxine for the treatment of hot flushes in patients with cancer conducted by Carpenter et al.,^[33] in which patients with depression were specifically excluded. This suggests that the improvement in depressive symptoms might be the key issue leading to an improvement in insomnia and CRF. Whereas one study reported improvements in insomnia and CRF in association with improvements in depressive symptoms,^[31] another study showed improvements in insomnia and CRF without associated improvements in depressive symptoms.^[61] Therefore, whether or not associated depression is indeed a putative explanatory variable remains an unresolved issue.

5. Conclusion

The assessment of the available evidence across trials for a direct cause–effect relationship between sleep disorders and fatigue in patients with cancer, and/or that the effective management of sleep disorders can reduce CRF, is confounded by different study designs, patient selection criteria, stage of cancer treatment and by the nature of the interventions studied. The evidence that CBT-I improves both insomnia and associated fatigue in patients with cancer is well substantiated, with five out of six trials showing a positive effect, whereas the evidence for pharmacotherapy is very limited. The question remains as to whether the improvement in CRF is caused by an improvement in sleep or whether other factors such as depression and physical activity changes may better explain this association. Despite their widespread use in clinical practice, there is limited evidence for the efficacy of sedative hypnotic agents in this population. Moreover, adverse effects such as daytime sedation and impaired cognitive functioning raise potential concerns. Therefore, randomized controlled studies on the use of non-sedating pharmacological interventions, singly or in combination with CBT-I, are needed to determine whether improving sleep will also improve fatigue and quality of life in patients with cancer.

Areas of particular interest for further study and discussion are:

• What exactly do we mean by fatigue in the context of patients with cancer (classification, terminology, assessment methods, applicability or otherwise of standard sleep medicine concepts)?

• How is any apparent relationship between sleep and CRF driven by the various mechanisms that affect both (e.g. inflammatory markers, circadian rhythm disturbances, depressed mood, HPA axis dysregulation)?

• Should studies on sleep and CRF focus on symptom clusters or single variables?

• What form of pharmacotherapy for insomnia in patients with cancer (for example sedating versus non-sedating medications) is effective without unacceptable side effects?

• How can CBT-I and pharmacotherapy be optimally combined?

• What mechanisms underlie the effects of various cytokines in sleep-related pathologies and how are these influenced by pharmacological agents?

• What are the health economic impacts of fatigue in patients with cancer and how can these be assessed better in order to aid decision making?

Regardless of the true nature of the relationship between sleep and CRF, treating significant sleep disorders may nonetheless have benefits beyond just improving sleep.