Secondary Logo

Journal Logo

ORIGINAL ARTICLES

Longitudinal Analysis of Sleep Disturbance in Breast Cancer Survivors

Yang, Gee Su; Starkweather, Angela R.; Lynch Kelly, Debra; Meegan, Taylor; Byon, Ha Do; Lyon, Debra E.

Author Information
doi: 10.1097/NNR.0000000000000578
  • Free

Abstract

Breast cancer is the foremost cause of cancer death (42,170 estimated deaths in 2020) after lung cancer in women in the United States (Siegel et al., 2020). With the advances in cancer treatment and early detection, the 5-year relative survival rate for breast cancer is 90%, which is much higher than esophagus, lung, liver, and pancreas cancers (9%–20%; Siegel et al., 2020). Extended survivorship has brought challenges to breast cancer survivors (BCS); they tend to live with unremitting symptoms, such as sleep disturbances, fatigue, pain, depression, anxiety, and/or cognitive impairment during and after cancer treatments (Bodai & Tuso, 2015).

BCS often report that poor quality of sleep and staying awake throughout the night are among the greatest challenges experienced (Lowery-Allison et al., 2018), with more than 50% of BCS experiencing sleep disturbances in the first 3 months after chemotherapy and radiation treatment (Lester et al., 2015; Schreier et al., 2019). Sleep disturbances have been found to begin or worsen throughout cancer treatment and continue for as long as 10 years after the cessation of cancer treatment (Lowery-Allison et al., 2018). Sleep disturbances among BCS have been described as multifactorial, with numerous risk factors and environmental conditions affecting their prevalence.

The physical and emotional effect of cancer diagnosis and treatment often stems from heightened anxiety, depression, fear of reoccurrence, and fatigue, which directly affect sleep quality (Strollo et al., 2020). These symptoms may share common underlying mechanisms, such as inflammatory signaling processes or posttreatment circadian misalignment (Ancoli-Israel et al., 2014; Liu et al., 2012; Whisenant et al., 2017). Notably, cancer treatments, such as chemotherapy and steroids, are associated with impaired sleep, and the increased prevalence of vasomotor symptoms can have a profound effect on sleep quality (Carreira et al., 2018). Abrupt onset of vasomotor symptoms, such as hot flashes and night sweats, correlates with increased incidence of depression and sleep disturbances in BCS (Accortt et al., 2015). Furthermore, prior studies found that Black women and women of minority status are at increased risk for experiencing sleep disturbances, which may signal the role of race plays in our understanding of sleep disturbances among BCS (Budhrani et al., 2015; Schreier et al., 2019).

BCS may be at additional risk for sleep disorders because of epigenetic factors, including telomere shortening. Telomere length shortening is a normal occurrence associated with aging; it is accelerated with inflammation and oxidative stress (Alhareeri et al., 2020; Garland et al., 2014). Telomere shortening is associated with breast cancer-specific mortality (Duggan et al., 2014). There is increased DNA damage and a reduction in telomerase activity in BCS after treatment (Scuric et al., 2017). Garland et al. (2014) reported that telomere lengths of BCS with insomnia symptoms were shorter, but not significantly, compared to those without insomnia symptoms. Evidence supports that poor sleep quality is associated with telomere shortening in the wider population (Cribbet et al., 2014; Tempaku et al., 2015).

Long-term sleep disturbances affect the overall health of BCS in a multitude of ways. Poor sleep has been shown to have a long-lasting negative effect on quality of life and is predictive of adverse health outcomes, such as poor healing, cognitive impairment, and increased economic distress (Henneghan et al., 2018; Strollo et al., 2020). Although the significance of long-term sleep disturbances on overall health of BCS warrants increased clinical awareness and further investigation, only a few studies have examined long-term changes of sleep disturbances and their relationships with multiple potential risk factors, such as demographics, cancer- and treatment-related factors, lifestyle, symptom characteristics, and biological factors in a well-defined cohort of BCS. Despite other research with similar aims to this study, a long-term examination of salient factors that may affect sleep quality in BCS is greatly warranted to optimize symptom management strategies over the active treatment and early survivorship. Thus, this study aimed to elucidate characteristics of sleep disturbances and determine possible predictors that may contribute to sleep disturbances in BCS through 2 years postchemotherapy.

METHODS

Study Design and Participants

The parent study (EPIGEN) of this secondary analysis was a longitudinal, prospective study to examine interplays among sociodemographic variables, psychoneurological symptoms (i.e., anxiety, depression, fatigue, cognitive dysfunction, pain, and sleep disturbances), and epigenetic alterations in adult women diagnosed with early-stage breast cancer over 2 years. The data collection time points were (a) baseline before starting the chemotherapy (T1), (b) the midpoint of chemotherapy (T2), (c) 6 months postchemotherapy (T3), (d) 1 year postchemotherapy (T4), and (e) 2 years postchemotherapy (T5). Recruitment occurred at a university-affiliated national cancer center and four regional collaborative sites located in central Virginia. Participants were eligible if they were diagnosed with Stage I–III breast cancer, scheduled to initiate chemotherapy treatment, and aged 21 years or older. Participants were ineligible if they had a history of chemotherapy for cancer, severe mental health problems including dementia or active psychosis, or immune diseases. An institutional review board approved the study (Virginia Commonwealth University IRB HM 13194), and written informed consent was provided before the initiation of the study. The ethical standards for human subjects were met according to the 1964 Helsinki Declaration through all procedures in this study.

Measures

Sleep disturbance conditions (a total score and seven subscale scores) were considered dependent variables, and sociodemographic, cancer-related factors, health behaviors, symptom characteristics, and biological factors were included as independent variables.

General Sleep Disturbance

Sleep quality was evaluated using the General Sleep Disturbance Scale (GSDS), which consists of 21 items on an 8-point rating scale and includes seven subscales, including quality of sleep, quantity of sleep, sleep onset latency, midsleep awakenings, early awakenings, use of medications for sleep, and fatigue and alertness at work (excessive daytime sleepiness). Negatively worded items were reverse-scored to make higher scores correspond to more sleep disturbance symptoms. A total score of 43 or greater and subscale scores of 3 or greater indicate a significant presence of sleep disturbance symptoms (Lee, 1992).

Sociodemographic and Cancer-Related Factors

Age, race, educational level, income, employment, marital status, human epidermal growth factor receptor 2 positive (HER2+), estrogen receptor, reproductive stage, cancer stage (I–III), surgery (biopsy/simple vs. lumpectomy/segmental), and treatment regimen (e.g., adjuvant endocrine therapy) were collected from interviews/surveys with study participants and their medical records.

Lifestyle

Information on alcohol drinking, smoking status, body mass index, antihypertension medication, antipsychotropic medication (e.g., depression and anxiety), and leisure activity was collected. The Godin Leisure–Exercise Questionnaire measured leisure activity to evaluate the frequency and intensity of weekly leisure activity score during free time, in which the Godin scale score is divided into three levels: active (24 units or more), moderately active (14–23 units), and insufficiently active/sedentary (less than 14 units) out of 99 units (Godin, 2011).

Symptom Characteristics

Symptoms were measured with reliable and valid tools; the tools used have been frequently used in studies with BCS.

Depressive Symptoms

Anxiety and depression were assessed with the Hospital and Anxiety and Depression Scale, which includes 14 items on a scale of 0–3 and comprises two subscales for respective depression and anxiety symptom severity (Snaith, 2003). A participant was considered normal, borderline, or clinical case for an obtained score of 0–7, 8–10, or 11 or greater, respectively (Snaith, 2003). Scores of negatively worded items were reversed to indicate that higher scores mean higher levels of depression and anxiety.

Fatigue

Fatigue severity and interference were measured with the Brief Fatigue Inventory, which includes nine items on a scale of 11 points, with 0 indicating no fatigue and 10 indicating as bad as you can imagine (Mendoza et al., 1999). Scores of 1–3, 4–7, and 8–10 correspond to mild, moderate, and severe fatigue, respectively.

Pain

Pain was assessed with the nine-item Brief Pain Inventory short form on a scale of 0–10 (Cleeland, 1989). This measure consists of the severity of pain and its interference with daily function. A higher average score for each category indicates greater pain severity or effects on interference on daily function (Cleeland, 1989).

Cognitive Function

Neurocognitive function was measured by the noninvasive performance-based computerized test Central Nervous System Vital Signs (https://www.cnsvs.com; Gualtieri & Johnson, 2006). A neurocognitive index score was obtained by calculating an average of five representative domains, including composite memory, psychomotor speed, reaction time, complex attention, and cognitive flexibility, as a proxy of global neurocognitive function, with higher scores indicating better neurocognitive function.

Perceived Stress

The 10-item Perceived Stress Scale was employed to evaluate the stress level appraised by various situations in daily life during the last month (Cohen et al., 1983). Scores of 0–13, 14–26, and 27–40 were considered low, moderate, and high perceived stress, respectively (Cohen et al., 1983).

Epigenetic Factors

Telomere Length

Genomic DNA was extracted from whole blood using a Puregene DNA Isolation Kit (Qiagen, Valencia, CA). The telomere (T) and single copy housekeeping gene (S) values were obtained by performing monochrome multiplex quantitative polymerase chain reaction assay, which used a set of target and reference primers in a single reaction and was run on a BioRad CFX96 (Dahlgren et al., 2018). If more than 10% variance was observed in the T/S ratio values between all three measures, the assay was repeated to meet quality control standards (Alhareeri et al., 2020).

DNA Methylation Age Analysis

Genomic DNA was extracted from whole blood using the Puregene DNA isolation kit (Qiagen, Valencia, CA); an aliquot of 500 μg of DNA was bisulfite converted using an EZ DNA Methylation Kit (Zymo Research, Irvine, CA) and then hybridized into a genome-wide Infinium HumanMethylation450K BeadChip (HM450K; Illumina, San Diego, CA) at HudsonAlpha Institute for Biotechnology (Yang et al., 2020). DNA methylation age was estimated using R Version 3.6.1. To eliminate systematic bias and reduce technical variability, the quantile normalization plus beta-mixture quantile normalization method was used to preprocess the raw data (Wang et al., 2015). Afterward, the methylation age was calculated from the preprocessed data by the Horvath method using the watermelon package (Pidsley et al., 2013).

Statistical Analysis

We computed descriptive statistics to report means, standard deviation, frequencies, and percentages for all applicable variables. The temporal changes across five time points in telomere length (ratio of relative telomere to single copy gene [T/S ratio]), DNA methylation age, leisure activity, psychoneurological symptoms (anxiety, depression, fatigue, pain, and neurocognitive function), perceived stress, and sleep disturbances (total and subscale scores) were obtained by the F tests included in linear mixed-effects models with a random intercept to account for the dependency in repeated measures over multiple time points. To identify significant predictors of sleep disturbances at each time point, bivariate and linear regression models were fitted with the GSDS total score as the outcome and other variables as the predictors. We tested multicollinearity of variables included in the current analysis and excluded variables showing high variance inflation factor (VIF) scores (VIF > 10). Potential predictors were chosen from the bivariate analyses if the p value was <.25 and then moved to the multivariable linear regression model (Grant et al., 2019). Statistical significance was considered at a p value of ≤.05. SPSS Version 25 was used to conduct the analysis.

RESULTS

Participant Characteristics

Baseline characteristics of women with early-stage breast cancer in the EPIGEN study are shown in Table 1. Among 77 participants enrolled, 74 women were included for the analysis (three were dropped because of missing data). The average age was 51.3 (SD = 10.3) years old, and 70% (n = 52) were non-Hispanic White. Most participants attained high educational level (any education beyond high school; 78.4%), were employed (62.2%), were nonsmokers (79.7%), and received radiation therapy (78.4%). Nearly half reported drinking alcohol (55.4%), earned an income of less than $60,000 (45.9%), received biopsy/simple surgery (50.7%), and received adjuvant endocrine therapy (52.9%). The average body mass index was 29.9 kg/m2 (SD = 7.5), indicating overweight. Nineteen percent of participants were HER2+, and 28% were triple negative (estrogen receptor, progesterone receptor, and HER2). Nearly one third (35.1%) took antihypertensive medication, and 37.0% used antipsychotropic medication for depression and anxiety.

TABLE 1 - Baseline Demographic and Cancer-Related Characteristics of Women With Breast Cancer (N = 74)
Characteristic n (%)
Age in year (minimum–maximum) M = 51.3, SD = 10.3 (23–71)
Race
 Non-Hispanic White 52 (70.3)
 African American 22 (29.7)
Educational level
 Did not finish high school 16 (21.6)
 Any education beyond high school 58 (78.4)
Employment
 Employed (part time/full time) 46 (62.2)
 Unemployed/disabled/retired/student 28 (37.8)
Marital status
 Married/partner 46 (62.2)
 Single/divorced/separated 28 (37.8)
Income
 <$60,000 34 (45.9)
 ≥$60,000 40 (54.1)
Alcohol drinking
 Yes 41 (55.4)
 No 33 (44.6)
Smoking status
 Yes 15 (20.3)
 No 59 (79.7)
BMI in kg/m2 (minimum–maximum) M = 29.9, SD = 7.5 (19.1–54.3)
Reproductive stage
 Pre- and perimenopause 32 (43.2)
 Postmenopause 42 (56.8)
HER2+
 Yes 14 (18.9)
 No 60 (81.1)
Triple negative
 Yes 21 (28.4)
 No 53 (71.6)
Antihypertensive medication
 Yes 26 (35.1)
 No 48 (64.9)
Psychotropic medication (e.g., depression, anxiety)
 Yes 27 (37.0)
 No 46 (63.0)
Cancer stage
 I 20 (27.0)
 IIA 31 (41.9)
 IIB 15 (20.3)
 IIIA 8 (10.8)
Surgery
 Biopsy/simple 37 (50.7)
 Lumpectomy/segmental 36 (49.3)
Radiation therapy
 Yes 58 (78.4)
 No 16 (21.6)
Adjuvant endocrine therapy
 Yes 36 (52.9)
 No 32 (47.1)
Note. M = mean; SD = standard deviation; BMI = body mass index; HER2+ = human epidermal growth factor receptor 2 positive.

Temporal Changes in Potential Predictor Variables

As shown in Table 2, participants were assessed on epigenetic factors, lifestyle, and symptom characteristics over 2 years. The analysis included telomere length (T/S ratio) and DNA methylation age for epigenetic factors as biological predictors. DNA methylation age was decreased during chemotherapy treatment (T2–T1), persisting until 6-month follow-up time point, and gradually improved over time. DNA methylation age was significantly changed over 2 years (F = 6.2, p < .001), whereas no significant change in telomere length was observed (F = 0.6, p = .683). For lifestyle factors, it appeared that leisure activities decreased during chemotherapy and at 6-month follow-up time point compared to baseline; however, they increased over time at 1- and 2-year follow-up time points after the initiation of chemotherapy. There was a difference in the leisure activity score across time points (F = 5.1, p < .001). Participants experienced temporal changes of symptom characteristics, including anxiety (F = 9.0, p = .007), depression (F = 4.9, p < .001), fatigue (F = 6.6, p < .001), pain (F = 2.7, p = .032), neurocognitive function (F = 2.8, p = .024), and perceived stress (F = 2.9, p = .022). In general, these symptoms worsened at T2 and T3; however, the neurocognitive index gradually increased over time regardless of chemotherapy treatment.

TABLE 2 - Temporal Changes of Potential Predictors: Leisure Activity, Symptom Characteristics, and Biological (Epigenetic) Factors Over 2 Years
Variables Baseline (T1)
M (SD)
4th chemotherapy (T2)
M (SD)
6-month follow-up (T3)
M (SD)
1-year follow-up (T4)
M (SD)
2-year follow-up (T5)
M (SD)
F (p)
Leisure activity (Godin) 29.2 (32.4) 20.6 (20.3) 25.4 (21.1) 33.3 (26.1) 33.3 (36.5) 5.1 (.001)
Anxiety (HADS-A) 8.5 (3.5) 6.8 (3.3) 7.0 (3.9) 6.9 (3.5) 6.5 (3.5) 9.0 (<.001)
Depression (HADS-D) 3.7 (3.3) 4.7 (3.1) 4.1 (3.6) 3.5 (3.1) 3.2 (3.0) 4.9 (.001)
Fatigue (BFI) 2.0 (2.6) 3.3 (2.7) 2.9 (2.8) 2.3 (2.4) 2.3 (2.5) 6.6 (<.001)
Pain (BPI) 1.7 (2.4) 1.9 (2.4) 2.3 (2.6) 1.6 (2.4) 1.9 (2.4) 2.7 (.032)
Neurocognitive function (CNSVS neurocognitive index) 99.8 (14.4) 102.9 (15.1) 105.3 (14.8) 105.9 (15.7) 107.4 (14.6) 2.8 (.024)
Perceived stress (PSS) 16.8 (8.0) 14.7 (7.6) 14.2 (7.9) 13.5 (7.3) 12.8 (7.6) 2.9 (.022)
Telomere length (T/S ratio) 1.5 (0.5) 1.5 (0.6) 1.5 (0.6) 1.6 (0.6) 1.6 (0.6) 0.6 (.683)
DNA methylation age 52.3 (12.5) 50.3 (11.7) 50.2 (12.0) 52.2 (11.5) 54.4 (10.2) 6.2 (<.001)
Note. SD = standard deviation; Godin = Godin Leisure–Exercise Questionnaire; HADS-A = Hospital Anxiety and Depression Scale–Anxiety; HADS-D = Hospital Anxiety and Depression Scale–Depression; BFI = Brief Fatigue Inventory; BPI = Brief Pain Inventory; CNSVS = Central Nervous System Vital Signs (computerized neurocognitive assessment software program); PSS = Perceived Stress Scale; T/S ratio = ratio of relative telomere to single copy gene.

General Sleep Disturbances

The summary of total and subscale scores of GSDS is shown in Table 3. Participants showed significant temporal changes of sleep disturbances over 2 years (F = 7.4, p < .001). The total score of GSDS was highest at T2 (mean = 48.9, SD = 22.9) and decreased over time (T3: mean = 44.6, SD = 21.6; T4: mean = 42.9, SD = 21.7; T5: mean = 37.4, SD = 20.4), indicating that participants experienced a significant level of sleep disturbances (a total score ≥43) during chemotherapy treatment. They reported clinically meaningful sleep disturbances in sleep onset latency, midsleep awakenings, early awakenings, and quality of sleep at T2. Midsleep awakenings (F = 3.4, p = .010), early awakenings (F = 2.9, p = .024), and fatigue at work (excessive daytime sleepiness; F = 9.3, p < .001) showed significant temporal changes.

TABLE 3 - Temporal Changes of General Sleep Disturbances Over Time
Visit Baseline (T1)
M (SD)
4th chemotherapy (T2)
M (SD)
6-month follow-up (T3)
M (SD)
1-year follow-up (T4)
M (SD)
2-year follow-up (T5)
M (SD)
F (p)
N 74 73 74 71 68
GSDS total 39.1 (22.9) 48.9 (22.9) 44.6 (21.6) 42.9 (21.7) 37.4 (20.4) 7.4 (<.001)
1. Sleep onset latency 2.8 (2.4) 3.1 (2.7) 3.0 (2.5) 2.8 (2.4) 2.4 (2.5) 1.1 (.357)
2. Midsleep awakenings 4.6 (2.7) 5.2 (2.4) 4.5 (2.4) 4.5 (2.4) 4.1 (2.3) 3.4 (.010)
3. Early awakenings 3.3 (2.5) 3.9 (2.6) 3.4 (2.6) 3.2 (2.3) 2.9 (2.3) 2.9 (.024)
4. Quality of sleep 3.1 (2.1) 3.5 (2.2) 3.4 (2.1) 3.5 (2.0) 2.9 (1.9) 2.2 (.070)
5. Quantity of sleep 1.8 (1.2) 2.0 (1.3) 1.8 (1.2) 1.8 (1.1) 1.8 (1.1) 0.5 (.704)
6. Fatigue and alertness at work 1.9 (1.4) 2.8 (1.6) 2.4 (1.5) 2.2 (1.4) 1.9 (1.3) 9.3 (<.001)
7. Use of sleep medication 0.4 (0.6) 0.5 (0.7) 0.5 (0.8) 0.5 (0.8) 0.4 (0.6) 0.8 (.558)
Note. M = mean; SD = standard deviation; GSDS = General Sleep Disturbance Scale.

Bivariate and Multivariable Regression Analyses at Each Time Point

Individual demographic and cancer-related factors, lifestyle, symptom characteristics, and epigenetic factors were added as potential predictors into the bivariate and multivariable regression analyses to determine their associations with sleep disturbances at each time point. Results from bivariate analyses and multivariable analyses are shown in Table 4. All bivariate and multivariable analyses were adjusted for radiation therapy, surgery, and adjuvant endocrine therapy. At T1, the bivariate association of perceived stress, anxiety, depression, pain, and fatigue with sleep disturbances was supported. In the multivariable analysis, anxiety (β = .36, p = .005) was identified as a significant predictor for sleep disturbances, whereas smoking status (β = −.10, p = .325), perceived stress (β = .01, p = .973), depression (β = −.02, p = .870), pain (β = .27, p = .096), and fatigue (β = .32, p = .080) were not associated with sleep disturbances (model summary: adjusted R2 = .513, F = 8.7, p < .001). At T2, HER2+ status, taking antipsychotropic medication, perceived stress, anxiety, depression, pain, and fatigue showed bivariate associations with sleep disturbances. In the multivariable analysis, anxiety was a significant predictor (β = .28, p = .013); though, other variables were not associated with sleep disturbances (model summary: adjusted R2 = .562, F = 9.1, p < .001). At T3, the bivariate associations between all symptom characteristics, income, employment, taking antipsychotropic medication, and sleep disturbances were supported (p < .25). In the multivariable analysis, perceived stress (β = .33, p = .028) and fatigue (β = .61, p < .001) were significant predictors for developing sleep disturbances (model summary: adjusted R2 = .723, F = 16.0, p < .001). At T4, taking antipsychotropic medication and all symptom characteristics showed bivariate associations with sleep disturbances. In the multivariable analysis, anxiety (β = .41, p = .005) and fatigue (β = .72, p < .001) were significant predictors for sleep disturbances (model summary: adjusted R2 = .605, F = 10.4, p < .001). At T5, according to bivariate analyses, education level, income, cancer stage, leisure activity, and all symptom characteristics were associated with sleep disturbances; still, no significant association was found in the multivariable analysis.

TABLE 4 - Bivariate and Multivariable Analyses at Baseline Prior to Chemotherapy, the Midpoint of Chemotherapy, 6-Month, 1-Year, and 2-Year Follow-Up Time Points Postchemotherapy
Time point Variables Bivariate analysis (controlling for radiation therapy, surgery, and endocrine therapy) Multivariable analysis (controlling for radiation therapy, surgery, and endocrine therapy; variables of p < .25 from the bivariate analysis)
β (p) β (p)
T1 Demographic characteristics
 Age −.03 (.816)
 Race .04 (.748)
 Age × Race .05 (.692)
 Educational level −.14 (.288)
 Income −.07 (.570)
 Employment −.03 (.804)
 Marital status .05 (.675)
Disease characteristics
 HER2+ .12 (.341)
 Triple negative .05 (.771)
 Reproductive stage .10 (.446)
 Stage Stage III: −.06 (.630)
Stage II: .06 (.670)
Stage I: —
Lifestyle
 Alcohol drinking −.06 (.635)
 Smoking status .19 (.137) −.10 (.325)
 Leisure activity −.12 (.329)
 BMI .04 (.746)
 Antihypertension medication −.02 (.847)
 Antipsychotropic medication .14 (.268)
Biological factors
 Telomere length .03 (.803)
 DNA methylation age .02 (.905)
Symptom characteristics
 Perceived stress .52 (< .001)*** .01 (.973)
 Anxiety .53 (< .001)*** .36 (.005)**
 Depression .52 (< .001)*** −.02 (.870)
 Pain −.58 (< .001)*** .27 (.096)
 Fatigue .66 (< .001)*** .32 (.080)
 Neurocognition index −.15 (.255)
Model summary Adjusted R 2 = .513,
F(9, 57) =8.7,
p < .001
T2 Demographic characteristics
 Age −.04 (.713)
 Race −.05 (.678)
 Age × Race −.06 (.662)
 Educational level .13 (.280)
 Income −.04 (.753)
 Employment −.06 (.612)
 Marital status .07 (.579)
Disease characteristics
 HER2+ .15 (.235) .16 (.104)
 Triple negative −.03 (.845)
 Stage Stage III: .004 (.972)
Stage II: −.05 (.717)
Stage I: —
Lifestyle
 Leisure activity −.06 (.615)
 BMI .05 (.897)
 Antihypertension medication −.09 (.482)
 Antipsychotropic medication .25 (.041)* .08 (.387)
Biological factors
 Telomere length −.19 (.331)
 DNA methylation age −.05 (.697)
Symptom characteristics
 Perceived stress .55 (<.001)*** .21 (.090)
 Anxiety .53 (<.001)*** .28 (.013)*
 Depression .64 (<.001)*** .20 (.136)
 Pain .31 (.012)* −.08 (.455)
 Fatigue .54 (<.001)*** .20 (.120)
 Neurocognition index −.13 (.340)
Model summary Adjusted R 2 = .562,
F(10, 53) = 9.1,
p < .001
T3 Demographic characteristics
 Age −.06 (.621)
 Race .08 (.534)
 Age × Race .07 (.617)
 Educational level −.05 (.719)
 Income −.22 (.078) .10 (.236)
 Employment −.16 (.210) −.08 (.325)
 Marital status .12 (.369)
Disease characteristics
 HER2+ .03 (.793)
 Triple negative .09 (.608)
 Stage Stage III: −.15 (.258)
Stage II: .11 (.391)
Stage I: —
Lifestyle
 Leisure activity −.004 (.973)
 BMI .09 (.497)
 Antihypertension medication −.10 (.416)
 Antipsychotropic medication .18 (.150) −.04 (.581)
Biological factors
 Telomere length −.10 (.411)
 DNA methylation age .04 (.759)
Symptom characteristics
 Perceived stress .67 (<.001)*** .33 (.028)*
 Anxiety .51 (<.001)*** .17 (.186)
 Depression .51 (<.001)*** −.17 (.136)
 Pain .61 (<.001)*** .08 (.485)
 Fatigue .77 (<.001)*** .61 (<.001)***
 Neurocognition index −.10 (.453)
Model summary Adjusted R 2 = .723,
F(11, 52) = 16.0,
p < .001
T4 Demographic characteristics
 Age −.09 (.482)
 Race .07 (.557)
 Age × Race .07 (.605)
 Educational level −.14 (.266)
 Income −.12 (.342)
 Employment −.06 (.617)
 Marital status .05 (.675)
Disease characteristics
 HER2+ .03 (.837)
 Triple negative .02 (.917)
 Stage Stage III: .09 (.466)
Stage II: −.02 (.894)
Stage I: —
Lifestyle
 Leisure activity −.10 (.451)
 BMI .01 (.935)
 Antihypertension medication −.06 (.627)
 Antipsychotropic medication .25 (.044)* −.01 (.884)
Biological factors
 Telomere length .10 (.426)
 DNA methylation age −.05 (.702)
Symptom characteristics
 Perceived stress .65 (<.001)*** .12 (.453)
 Anxiety .59 (<.001)*** .41 (.005)**
 Depression .52 (<.001)*** −.22 (.131)
 Pain .45 (<.001)*** −.23 (.127)
 Fatigue .63 (<.001)*** .72 (<.001)***
 Neurocognition index −.17 (.192) −.002 (.981)
Model summary Adjusted R 2 = .605,
F (10, 51) = 10.4,
p < .001
T5 Demographic characteristics
 Age −.08 (.499)
 Race .13 (.291)
 Age × Race .12 (.351)
 Educational level −.21 (.095) −.08 (.549)
 Income −.23 (.065) .08 (.497)
 Employment −.06 (.642)
 Marital status .26 (.045)* .14 (.229)
Disease characteristics
 HER2+ .03 (.801)
 Triple negative .05 (.794)
 Stage Stage III: −.06 (.669)
Stage II: .24 (.057)
Stage I: —
Stage III: −.05 (.655)
Stage II: .08 (.501)
Stage I: —
Lifestyle
 Leisure activity −.33 (.006)** −.13 (.268)
 BMI −.01 (.954)
 Antihypertension medication .04 (.748)
 Antipsychotropic medication .03 (.822)
Biological factors
 Telomere length .13 (.299)
 DNA methylation age −.07 (.628)
Symptom characteristics
 Perceived stress .66 (<.001)*** .21 (.221)
 Anxiety .51 (<.001)*** .24 (.109)
 Depression .54 (<.001)*** −.03 (.877)
 Pain .56 (<.001)*** .16 (.363)
 Fatigue .64 (<.001)*** .29 (.140)
 Neurocognition index −.16 (.205) .15 (.214)
Model summary Adjusted R 2 = .570,
F(15, 42) = 6.1,
p < .001
Note. Covariate references include African American (race), “did not finish high school” (education), <$60,000 (income), unemployed/disabled/retired/student (employment), single/divorced/separated (marriage), no HER2+, no triple negative, no reproductive stage, cancer Stage I, no alcohol drinking, no smoking, no antihypertension medication, and no antipsychotropic medication. HER2+ = human epidermal growth factor receptor 2 positive; BMI = body mass index; T1 = baseline prior to chemotherapy; T2 = fourth chemotherapy; T3 = 6-month follow-up postchemotherapy; T4 = 1-year follow-up postchemotherapy; T5 = 2-year follow-up postchemotherapy. Reproductive stage, alcohol drinking, and smoking status were included in only T1.
*p < .05.
**p < .01.
***p < .001.

DISCUSSION

This study addressed temporal changes of sleep disturbances and potential predictors that may be associated with sleep disturbances in BCS from baseline through 2-year survivorship after initiating chemotherapy. Our results demonstrated that BCS experienced poor sleep quality during the period of chemotherapy treatment and gradually improved over time. Specifically, significant temporal changes were observed in midsleep awakenings, early awakenings, and fatigue at work (excessive daytime sleepiness), with disturbances being elevated at T2. Symptom characteristics were associated with higher levels of sleep disturbances across all time points in the bivariate analyses. Anxiety (T1, T2, and T4), fatigue (T3 and T4), and perceived stress (T3) were significant predictors of sleep disturbances across treatment and survivorship trajectories in the multivariable analyses.

Sleep disturbances in BCS were found to be most prevalent during active treatment (T2) and steadily decreased following treatment through the 2-year follow-up time point. Ancoli-Israel et al. (2014) reported similar sleep symptom trajectories; peak sleep disturbance and fatigue levels were observed following four cycles of chemotherapy and then improved by 1-year postchemotherapy (Ancoli-Israel et al., 2014). The most common sleep disturbance characteristics reported in this study consisted of sleep onset latency, midsleep awakenings, early awakenings, and overall sleep quality (Ancoli-Israel et al., 2014). In a sample of BCS up to 10 years posttreatment, Lowery-Allison et al. (2018) found that greater than one third of the population took twice as long to fall asleep, had twice as many awakenings during the night, and slept for 2 hours less on average per night (Lowery-Allison et al., 2018). In an integrative review including 12 studies on sleep disturbances, findings detailed that at 1-year posttreatment, sleep quality remained poor in BCS, indicating a continued negative effect of chemotherapy on sleep, and the most common symptoms experienced consisted of increased nocturnal awakening and waking after sleep onset (Budhrani et al., 2015). Liu et al. (2012) also found that there were overall positive relationships between changes in interleukin-6 and interleukin-1 receptor antagonist and changes in sleep disturbances scores as well as between C-reactive protein and total wake time during the night in BCS who received at least four cycles of chemotherapy. This may implicate that inflammation may mediate chemotherapy-related sleep disturbances (Liu et al., 2012). Similarly, the meta-analysis on sleep deprivation that includes 72 studies showed the positive association of shorter sleep duration and extremely long sleep duration with higher levels of C-reactive protein (Irwin et al., 2016). Our finding on sleep disturbance patterns may be informative for an educational intervention for this population.

As previously discussed, epigenetics factors are thought to play a role in sleep disturbances in BCS. Current literature supports that telomere length may be associated with poor sleep quality in BCS (Cribbet et al., 2014; Tempaku et al., 2015); however, we showed nonsignificant results in telomere length and DNA methylation age in association with sleep disturbances partially because of a small sample size. Given that telomere length and DNA methylation age are age-associated biological measures, this nonsignificant result could be expected because age was not a significant factor for developing sleep disturbances across all time points in this study. Furthermore, the deviation of age in this cohort is slight, which would make it difficult to determine the effect of age or age-related biological factors on sleep disturbances. Further research is needed to investigate the association between these variables and sleep disturbances in a large sample size.

In this study, sleep disturbances in BCS were closely associated with anxiety, fatigue, and perceived stress, although the influence of these factors was shown to fluctuate at different time points. Specifically, both anxiety and fatigue frequently appeared to be associated with sleep disturbances. Multiple studies suggested that anxiety of cancer reoccurrence is associated with sleep disturbances in cancer patients (Hall et al., 2014; Lowery-Allison et al., 2018; Strollo et al., 2020). BCS reported considerable anxiety when diagnosed with cancer (Jones et al., 2014). We identified that fatigue was significantly associated with sleep disturbances at 6-month and 1-year follow-up time points. In contrast, a recent study showed that fatigue co-occurs with severe sleep disturbances before and during chemotherapy (Fox et al., 2020). Furthermore, increased levels of pain were associated with sleep disturbances prior to chemotherapy. This correlation is consistent with results from several other studies investigating sleep disturbances in BCS (Lowery-Allison et al., 2018; Syrjala et al., 2014).

Clusters of both physical and emotional symptoms can contribute to perceived distress in BCS. Physical and emotional symptoms causing distress in this study included fatigue, anxiety, and pain, all of which affected sleep disturbances. These findings are supported by current literature, which has reported that perceived distress extends beyond treatment into early survivorship and is related to clusters of distressing symptoms (Lester et al., 2015; Schreier et al., 2019; Strollo et al., 2020). In addition, Strollo et al. (2020) reported that, at 9 years posttreatment, sleep disturbances of cancer survivors were significantly affected by physical, emotional, and economic distress (Strollo et al., 2020).

Our results demonstrated that the association between pain and sleep disturbances was supported in the bivariate analyses at all time points. Current literature suggests that circadian rhythms are frequently disrupted in cancer patients, worsening symptoms experienced in this population (Ancoli-Israel et al., 2014; Roveda et al., 2019). It is known that sleep and pain interact bidirectionally and that circadian rhythms play a role in this relationship (Palada et al., 2020). For example, circadian genes, such as CLOCK and BMAL1, regulate the expression of pain-related genes, such as TAC1, which encodes for substance P (Zhang et al., 2012). In addition, circadian clock rhythmicity may regulate the expression of nociceptive neurons (e.g., voltage-gated calcium channel and NMDA glutamate receptor) in the dorsal root ganglions as well as the expression of μ-opioid receptors in part of the brainstem that is involved in descending pain inhibition (Takada et al., 2013). Through these mechanisms, disruption of the circadian rhythm in BCS negatively affects sleep quality and can also affect their experience of pain.

Potential limitations should be noted. This study had a relatively small sample size with no control group, affecting the generalizability of findings. However, because of the nature of longitudinal study design, intrinsic and extrinsic factors could be controlled when comparing variables between the baseline and other time points. In addition, we included only composite or total scores of questionnaires as predictor variables in the regression models; although, more detailed information can be obtained if subscale scores are used. Lastly, sleep disturbances were self-reported; future studies could consider a combination of subjective and objective measurement tools, such as actigraphy, for more robust results.

Conclusion

This study indicates that the severity of sleep disturbances changes over time, with the most severe sleep disturbances during the chemotherapy phase of treatment. In addition, the predictors of sleep disturbances changed over time, with anxiety being a factor earlier in the treatment trajectory (prechemotherapy) and continuing over time with fatigue and perceived stress being involved later in the treatment trajectory. These findings show that symptom management strategies to address sleep disturbances should be tailored to address the temporal factors that may change in severity over the active treatment and early survivorship period. Notably, sleep improved after the cessation of active treatment for most participants. Findings gained from this study on sleep disturbance patterns and the potential risk factors can be incorporated into clinical practice in planning education and developing interventions.

FU1

ORCID iDs

Gee Su Yang https://orcid.org/0000-0002-4602-1152

Angela R. Starkweather https://orcid.org/0000-0001-7168-0144

Debra Lynch Kelly https://orcid.org/0000-0002-0017-7821

Taylor Meegan https://orcid.org/0000-0002-7905-062X

Ha Do Byon https://orcid.org/0000-0003-4095-3108

Debra E. Lyon https://orcid.org/0000-0003-3067-3962

REFERENCES

Accortt E. E., Bower J. E., Stanton A. L., Ganz P. A. (2015). Depression and vasomotor symptoms in young breast cancer survivors: The mediating role of sleep disturbance. Archives Women’s Mental Health, 18, 565–568. 10.1007/s00737-015-0504-5
Alhareeri A. A., Archer K. J., Fu H., Lyon D. E., Elswick R. K. Jr., Kelly D. L., Starkweather A. R., Elmore L. W., Bokhari Y. A., Jackson-Cook C. K. (2020). Telomere lengths in women treated for breast cancer show associations with chemotherapy, pain symptoms, and cognitive domain measures: A longitudinal study. Breast Cancer Research, 22, 137. 10.1186/s13058-020-01368-6
Ancoli-Israel S., Liu L., Rissling M., Natarajan L., Neikrug A. B., Palmer B. W., Mills P. J., Parker B. A., Sadler G. R., Maglione J. (2014). Sleep, fatigue, depression, and circadian activity rhythms in women with breast cancer before and after treatment: A 1-year longitudinal study. Supportive Care in Cancer, 22, 2535–2545. 10.1007/s00520-014-2204-5
Bodai B. I., Tuso P. (2015). Breast cancer survivorship: A comprehensive review of long-term medical issues and lifestyle recommendations. The Permanente Journal, 19, 48–79. 10.7812/TPP/14-241
Budhrani P. H., Lengacher C. A., Kip K., Tofthagen C., Jim H. (2015). An integrative review of subjective and objective measures of sleep disturbances in breast cancer survivors. Clinical Journal of Oncology Nursing, 19, 185–191. 10.1188/15.CJON.185-191
Carreira H., Williams R., Müller M., Harewood R., Stanway S., Bhaskaran K. (2018). Associations between breast cancer survivorship and adverse mental health outcomes: A systematic review. Journal of National Cancer Institute, 110, 1311–1327. 10.1093/jnci/djy177
Cleeland C. S. (1989). Measurement of pain by subjective report. In Chapman C. R., Loeser J. D. (Eds.), Advances in pain research and therapy (Vol. 12, pp. 391–403). Raven Press.
Cohen S., Karmarck T., Mermelstein R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385–396.
Cribbet M. R., Carlisle M., Cawthon R. M., Uchino B. N., Williams P. G., Smith T. W., Gunn H. E., Light K. C. (2014). Cellular aging and restorative processes: Subjective sleep quality and duration moderate the association between age and telomere length in a sample of middle-aged and older adults. Sleep, 37, 65–70. 10.5665/sleep.3308
Dahlgren P. N., Bishop K., Dey S., Herbert B.-S., Tanaka H. (2018). Development of a new monochrome multiplex qPCR method for relative telomere length measurement in cancer. Neoplasia, 20, 425–431. 10.1016/j.neo.2018.02.007
Duggan C., Risques R., Alfano C., Prunkard D., Imayama I., Holte S., Baumgartner K., Baumgartner R., Bernstein L., Ballard-Barbash R., Rabinovitch P., McTiernan A. (2014). Change in peripheral blood leukocyte telomere length and mortality in breast cancer survivors. Journal of the National Cancer Institute, 106, dju035. 10.1093/jnci/dju035
Fox R. S., Ancoli-Israel S., Roesch S. C., Merz E. L., Mills S. D., Wells K. J., Sadler G. R., Malcarne V. L. (2020). Sleep disturbance and cancer-related fatigue symptom cluster in breast cancer patients undergoing chemotherapy. Supportive Care in Cancer, 28, 845–855. 10.1007/s00520-019-04834-w
Garland S. N., Palmer C., Donelson M., Gehrman P., Johnson F. B., Mao J. J. (2014). A nested case-controlled comparison of telomere length and psychological functioning in breast cancer survivors with and without insomnia symptoms. Rejuvenation Research, 17, 453–457. 10.1089/rej.2014.1586
Godin G. (2011). The Godin–Shephard Leisure-Time Physical Activity Questionnaire. Health & Fitness Journal of Canada, 4, 18–22. 10.14288/hfjc.v4i1.82
Grant S. W., Hickey G. L., Head S. J. (2019). Statistical primer: Multivariable regression considerations and pitfalls. European Journal of Cardio-Thoracic Surgery, 55, 179–185. 10.1093/ejcts/ezy403
Gualtieri C. T., Johnson L. G. (2006). Reliability and validity of a computerized neurocognitive test battery, CNS vital signs. Archives of Clinical Neuropsychology, 21, 623–643. 10.1016/j.acn.2006.05.007
Hall D. L., Mishel M. H., Germino B. B. (2014). Living with cancer-related uncertainty: Associations with fatigue, insomnia, and affect in younger breast cancer survivors. Supportive Care in Cancer, 22, 2489–2495. 10.1007/s00520-014-2243-y
Henneghan A. M., Carter P., Stuifbergan A., Parmelee B., Kesler S. (2018). Relationships between self-reported sleep quality components and cognitive functioning in breast cancer survivors up to 10 years following chemotherapy. Psycho-Oncology, 27, 1937–1943. 10.1002/pon.4745
Irwin M. R., Olmstead R., Carroll J. E. (2016). Sleep disturbance, sleep duration, and inflammation: A systematic review and meta-analysis of cohort studies and experimental sleep deprivation. Biological Psychiatry, 80, 40–52. 10.1016/j.biopsych.2015.05.014
Jones S. L., Hadjistavropoulos H. D., Gullickson K. (2014). Understanding health anxiety following breast cancer diagnosis. Psychology, Health & Medicine, 19, 525–535. 10.1080/13548506.2013.845300
Lee K. A. (1992). Self-reported sleep disturbances in employed women. Sleep, 15, 493–498. 10.1093/sleep/15.6.493
Lester J., Crosthwaite K., Stout R., Jones R. N., Holloman C., Shapiro C., Andersen B. L. (2015). Women with breast cancer: Self-reported distress in early survivorship. Oncology Nursing Forum, 42, E17–E23. 10.1188/15.ONF.E17-E23
Liu L., Mills P. J., Rissling M., Fiorentino L., Natarajan L., Dimsdale J. E., Sadler G. R., Parker B. A., Ancoli-Israel S. (2012). Fatigue and sleep quality are associated with changes in inflammatory markers in breast cancer patients undergoing chemotherapy. Brain, Behavior, and Immunity, 26, 706–713. 10.1016/j.bbi.2012.02.001
Lowery-Allison A. E., Passik S. D., Cribbet M. R., Reinsel R. A., O’Sullivan B., Norton L., Kirsh K. L., Kavey N. B. (2018). Sleep problems in breast cancer survivors 1–10 years posttreatment. Palliative and Supportive Care, 16, 325–334. 10.1017/S1478951517000311
Mendoza T. R., Wang X. S., Cleeland C. S., Morrissey M., Johnson B. A., Wendt J. K., Huber S. L. (1999). The rapid assessment of fatigue severity in cancer patients: Use of the Brief Fatigue Inventory. Cancer, 85, 1186–1196. 10.1002/(sici)1097-0142(19990301)85:5<1186::aid-cncr24>3.0.co;2-n
Palada V., Gilron I., Canlon B., Svensson C. I., Kalso E. (2020). The circadian clock at the intercept of sleep and pain. Pain, 161, 894–900. 10.1097/j.pain.0000000000001786
Pidsley R., Wong C. C., Volta M., Lunnon K., Mill J., Schalkwyk L. C. (2013). A data-driven approach to preprocessing Illumina 450K methylation array data. BMC Genomics, 14, 293. 10.1186/1471-2164-14-293
Roveda E., Bruno E., Galasso L., Mulè A., Castelli L., Villarini A., Caumo A., Esposito F., Montaruli A., Pasanisi P. (2019). Rest-activity circadian rhythm in breast cancer survivors at 5 years after the primary diagnosis. Chronobiology International, 36, 1156–1165. 10.1080/07420528.2019.1621330
Schreier A. M., Johnson L. A., Vohra N. A., Muzaffar M., Kyle B. (2019). Post-treatment symptoms of pain, anxiety, sleep disturbance, and fatigue in breast cancer survivors. Pain Management Nursing, 20, 146–151. 10.1016/j.pmn.2018.09.005
Scuric Z., Carroll J. E., Bower J. E., Ramos-Perlberg S., Petersen L., Esquivel S., Hogan M., Chapman A. M., Irwin M. R., Breen E. C., Ganz P. A., Schiestl R. (2017). Biomarkers of aging associated with past treatments in breast cancer survivors. NPJ Breast Cancer, 3, 50. 10.1038/s41523-017-0050-6
Siegel R. L., Miller K. D., Jemal A. (2020). Cancer statistics, 2020. CA: A Cancer Journal for Clinicians, 70, 7–30. 10.3322/caac.21590
Snaith R. P. (2003). The Hospital Anxiety and Depression Scale. Health and Quality of Life Outcomes, 1, 29. 10.1186/1477-7525-1-29
Strollo S. E., Fallon E. A., Gapstur S. M., Smith T. G. (2020). Cancer-related problems, sleep quality, and sleep disturbance among long-term cancer survivors at 9-years post diagnosis. Sleep Medicine, 65, 177–185. 10.1016/j.sleep.2019.10.008
Syrjala K. L., Jensen M. P., Mendoza M. E., Yi J. C., Fisher H. M., Keefe F. J. (2014). Psychological and behavioral approaches to cancer pain management. Journal of Clinical Oncology, 32, 1703–1711. 10.1200/JCO.2013.54.4825
Takada T., Yamashita A., Date A., Yanase M., Suhara Y., Hamada A., Sakai H., Ikegami D., Iseki M., Inada E., Narita M. (2013). Changes in the circadian rhythm of mRNA expression for μ-opioid receptors in the periaqueductal gray under a neuropathic pain-like state. Synapse, 67, 216–223. 10.1002/syn.21633
Tempaku P. F., Mazzotti D. R., Tufik S. (2015). Telomere length as a marker of sleep loss and sleep disturbances: A potential link between sleep and cellular senescence. Sleep Medicine, 16, 559–563. 10.1016/j.sleep.2015.02.519
Wang T., Guan W., Lin J., Boutaoui N., Canino G., Luo J., Celedón J. C., Chen W. (2015). A systematic study of normalization methods for Infinium 450K methylation data using whole-genome bisulfite sequencing data. Epigenetics, 10, 662–669. 10.1080/15592294.2015.1057384
Whisenant M., Wong B., Mitchell S. A., Beck S. L., Mooney K. (2017). Distinct trajectories of fatigue and sleep disturbance in women receiving chemotherapy for breast cancer. Oncology Nursing Forum, 44, 739–750. 10.1188/17.ONF.739-750
Yang G. S., Mi X., Jackson-Cook C. K., Starkweather A. R., Lynch Kelly D., Archer K. J., Zou F., Lyon D. E. (2020). Differential DNA methylation following chemotherapy for breast cancer is associated with lack of memory improvement at one year. Epigenetics, 15, 499–510. 10.1080/15592294.2019.1699695
Zhang J., Li H., Teng H., Zhang T., Luo Y., Zhao M., Li Y. Q., Sun Z. S. (2012). Regulation of peripheral clock to oscillation of substance P contributes to circadian inflammatory pain. Anesthesiology, 117, 149–160. 10.1097/ALN.0b013e31825b4fc1
Keywords:

breast cancer; chemotherapy; longitudinal study; sleep disturbance

Copyright © 2022 Wolters Kluwer Health, Inc. All rights reserved.