Sleep disturbance is associated with perturbations in immune-inflammatory pathways in oncology outpatients undergoing chemotherapy

2.1. Patients and settings

This study is part of a larger, longitudinal study of the symptom experience of oncology outpatients receiving chemotherapy [20]. Eligible patients were ≥18 years of age; had a diagnosis of breast, gastrointestinal, gynecological, or lung cancer; had received chemotherapy within the preceding four weeks; were scheduled to receive at least two additional cycles of chemotherapy; were able to read, write, and understand English; and gave written informed consent. Patients were recruited from two Comprehensive Cancer Centers, one Veteran's Affairs hospital, and four community-based oncology programs.

2.2. Study procedures

The study was approved by the Institutional Review Board at each of the study sites. Of the 2234 patients approached, 1343 consented to participate (60.1% response rate). The major reason for refusal was being overwhelmed with their cancer treatments. Eligible patients were approached in the infusion unit during their first or second cycle of chemotherapy by a member of the research team to discuss study participation and obtain written informed consent.

Patients completed the General Sleep Disturbance Scale (GSDS) [21] six times over two cycles of chemotherapy (i.e., prior to chemotherapy administration, approximately 1 week after chemotherapy administration, and approximately 2 weeks after chemotherapy administration). Blood for ribonucleic acid (RNA) isolation was collected at the enrollment assessment. Medical records were reviewed for disease and treatment information. For this study, a total of 717 patients provided a blood sample for the gene expression and pathway perturbation analyses (Supplemental Fig. 1). Of these 717 patients, 198 patients had their samples processed using RNA sequencing (i.e., RNA-seq sample) and 162 patients had their samples processed using microarray (i.e., microarray sample) technologies.

2.3. Instruments

2.4. Data analysis

3.1. RNA-seq performance

Of the 198 patients in the RNA-seq sample, 81 were in the Low and 117 were in the Very High classes (Table 1). Median library threshold size was 6.21. Following the application of quality control filters, 10,825 genes were included in the final analysis. The common dispersion was estimated as 0.21, yielding a biological coefficient of variation of 0.45 well within the expected value for clinical samples [42].

3.2. Microarray performance

Of the 162 patients in the microarray sample, 78 patients were in the Low and 84 were in the Very High sleep disturbance classes (Table 2). All of these samples demonstrated good hybridization performance for biotin, background negative, and positive control assays on the arrays. Limma was used for background correction, quantile normalization, and log2 transformation [43]. Following quality control filters, 42,921 loci were included in the final analysis.

3.3. Differences in demographic and clinical characteristics

Of the 198 patients with phenotypic data in the RNA-seq sample (Table 1), compared to the Low class, the Very High class was significantly younger and more likely to be female, live alone, and had a lower annual income. Patients in the Very High class had lower KPS scores; a higher number of comorbidities; higher SCQ scores; higher MAX2 scores; were more likely to self-report a diagnosis of anemia or blood disease, depression, or back pain; and were less likely to have gastrointestinal cancer and more likely to have gynecological cancer.

Of the 162 patients with phenotypic data in the microarray sample (Table 2), compared to the Low class, the Very High class was significantly younger; more likely to be female and have child care responsibilities; and less likely to be married/partnered, employed, and exercise on a regular basis. Patients in the Very High class had lower KPS scores; a higher body mass index (BMI); a higher number of comorbidities; higher SCQ scores; higher MAX2 scores; were more likely to self-report a diagnosis of anemia or blood disease or depression; and were more likely to receive an antiemetic regimen containing a neurokin-1 receptor antagonist in combination with other two antiemetics.

3.4. Logistic regression analyses

For the RNA-seq sample (Table 3), six variables were retained in the final logistic regression model (i.e., age, gender, lives alone, income, KPS score, SCQ score) and were used as covariates in the gene expression analysis. Patients who were younger, female, lived alone, had a lower KPS score, and a higher SCQ score were more likely to belong to the Very High class. Compared to patients who had an income of $30,000 or less, patients with an annual income between $30,000 and $70,000 and between $70,000 and $100,000 had a 77% and 73% decrease in the odds of belonging in the Very High class respectively.

For the microarray sample (Table 3), eight variables were retained in the final logistic regression model (i.e., age, married or partnered, currently employed, exercise on a regular basis, BMI, MAX2 score, KPS score, self-reported diagnosis of depression) and were used as covariates in the gene expression analysis. Patients who were not married or partnered and had a lower KPS score were more likely to belong to the Very High class. Patients who exercised on a regular basis had a 75% decrease in the odds of belonging to the Very High class. Having a self-reported diagnosis of depression was associated with a 5.19 times increase in the odds of belonging in the Very High class.

3.5. Differentially expressed genes and pathways between the Low and Very High classes

Of the 14 surrogate variables identified for the RNA-seq sample, one was associated with sleep disturbance scores and was excluded from the final model. The final differential expression model for this sample included 13 surrogate variables and six significant demographic and clinical characteristics.

Of the 15 surrogate variables identified for the microarray sample, one was associated with sleep disturbance scores and was excluded from the final model. The final differential expression model for this sample included 14 surrogate variables and eight demographic and clinical characteristics. For both samples, a total of 4090 genes were included in the PIA analyses. As shown in Table 4, using Fisher's combined probability method, across the two samples, thirteen KEGG signaling pathways related to immune-inflammatory mechanisms were significantly perturbed.

4.1. Endocytosis

Endocytosis is a process through which various substances are internalized into a cell. As noted in one review [44], recent evidence suggests that endocytosis occurs across the blood brain barrier (BBB) during sleep. This process results in the clearance of waste products in the perivascular space. Endocytosis may be of clinical importance. For example, in a preclinical study that evaluated if cerebrospinal fluid clearance was increased during sleep [45], findings suggest that the enhanced removal of neurotoxic waste products that accumulate in the awake central nervous system may be a key component of restorative sleep.

Equally important, loss of sleep increases the activity of several pro-inflammatory cytokines (e.g., TNF, IL6) that can lead to increased permeability of the BBB [44,46]. In addition, while the sleep-wake cycle effects endocytotic processes, levels of endocytosis appear to regulate sleep. For example, in one study of Drosophila [46], blocking endocytosis resulted in an increase in total sleep time.

4.2. Phagosome

Phagocytosis is a process through which cells recognize, engulf, and digest a variety of large particles (e.g., apoptotic cells, cellular debris). Within the central nervous system, phagocytosis plays a critical role in the proper development of neural circuits and the maintenance of homeostasis [47]. Specifically, microglial cells are the immune cells that are involved in the phagocytosis of cellular debris and the “orchestration” of neuroinflammation within the brain [48]. Recent hypotheses suggest that inflammation and neurodegenerative processes may occur if phagocytotic processes fail [47].

Albeit limited, a growing body of evidence suggests that microglia play a role in a variety of sleep disorders [48]. For example, during an acute infection with its associated “sickness behavior”, sleep disturbance and fatigue are common symptoms. These symptoms may be associated with an increase in the production of cytokines by microglia [49]. In addition, alterations in the homeostatic functioning of microglia may be one of the underlying mechanisms for narcolepsy [48]. Increases in inflammation and alterations in microglia's phagocytic activity may result in the degeneration of hypocretin neurons that are implicated in the pathophysiology of narcolepsy.

Of note, total or partial sleep disturbance can alter phagocytic activity and increase inflammatory processes [50]. For example, in a preclinical study that investigated whether microglia phagocytose synapses around the time of sleep onset or during the light phase to help facilitate sleep [51], findings suggest that through a process of phagocytosis, weak synapses are eliminated during every phase of sleep.

4.3. Antigen processing and presentation

Antigen processing and presentation is a molecular mechanism by which whole antigens are broken down and loaded onto major histocompatibility complex (MHC) molecules for display on the surface of T and B cells [52–54]. While MHC class I molecules “report” on endogenous events (e.g., intracellular bacteria, cellular transformation) to CD8⁺ T cells, MHC class II molecules present exogenous antigens to CD4⁺ T cells [53]. As noted in a number of reviews [55–57], sleep deprivation is associated with alterations in innate and adaptive immune responses. Specifically, sleep deprivation contributes to altered antigen presentation; lowered Th1 responses; higher Th2 responses; and decreases in antibody production. These changes lead to chronic inflammation and increased risk for and/or worsening of several chronic conditions (e.g., cancer, Alzheimer's disease, cardiovascular disease, diabetes).

4.4. NK cell mediated cytotoxicity

NK cells are a family of cytotoxic lymphocytes that lyse cancer and virally infected cells. Once activated, NK cells can produce biochemical signals and cytokines to stimulate adaptive immune responses [58]. As noted in preclinical [59] and clinical [60,61] studies, sleep disturbance was associated with a significant decrease in NK cell counts. This reduction was associated with increases in levels of TNF-alpha, that suppressed the expression of circadian rhythm genes (i.e., PER1, PER2, PER3) in NK cells.

4.5. Cytokine-cytokine receptor interaction

Cytokines are a diverse group of soluble extracellular proteins that are involved in the regulation of inflammatory processes [62,63]. Cytokines are released in response to a stimulus (e.g., chemotherapy) and use multiple signaling pathways [64]. Upon release, they bind to specific surface receptors; activate signal transduction pathways; and exert a response on target cells [63,65]. Cytokine-cytokine receptor interactions are complex processes that allow cytokines to produce a variety of biological responses with a limited set of surface receptors and effector signaling molecules [65].

Across numerous reviews [56,66], evidence exists to support a role for cytokines in physiologic regulation of sleep, as well as in pathophysiologic processes. In terms of cancer, as noted in one review [67], both patients and survivors experience a variety of sleep problems including insomnia, hypersomnia, somnolence syndrome, and nightmares. While multiple mechanisms are likely to contribute to these heterogenous sleep problems, cytokines released from the tumor and associated treatments can enter the central nervous system, bind to cytokine transporters at the BBB and be transported to the brain. Once in the brain, these cytokines can activate microglia and lead to alterations in sleep and behavior.

4.6. Apoptosis

Apoptosis or programmed cell death, is a crucial homeostatic mechanism that is involved in normal cell turnover, as well as in the death of cells that are damaged by external agents (e.g., chemotherapy) [68]. In terms of sleep disturbance, one area of research is focused on the effects of intermittent hypoxia associated with obstructive sleep apnea. For example, in a murine model of sleep apnea [69], irreversible and functionally significant injury was found in two wake active neural groups (i.e., dopaminergic neurons in the periaqueductal gray; noradrenergic neurons in the locus coeruleus. Equally intriguing is the emerging evidence on the actions of melatonin, a neurohormone that regulates circadian rhythm and sleep-wake cycles. Depending on physiologic conditions, melatonin can promote cell death or survival [70].

4.7. Neutrophil extracellular trap formation

Neutrophil extracellular traps (NETs) are extracellular protein-binding traps composed of DNA, proteins, and enzymes. The formation of NETs, a process called NETosis, is a specific form of cell death performed by neutrophils. While NETs play a crucial role in our innate immune response, recent evidence suggests that they are involved in a variety of inflammatory diseases (e.g., diabetes, rheumatoid arthritis) [71]. While direct evidence to support associations between sleep disturbance and NETs was not found, insomnia disorders and obstructive sleep apnea are associated with increases in inflammatory responses [66]. In addition, increases in inflammatory responses were found in oncology patients with sleep disturbance [72].

4.8. NOD-like receptor signaling pathway

NOD-like receptors are a group of intracellular pattern-recognition proteins that play a critical role in the regulation of immune and inflammatory responses through a variety of mechanisms (e.g., formation of inflammasomes, activation of NF-κ B and mitogen-activated protein kinases (MAPK) pathways, cytokine production, apoptosis) [73]. When NOD-like receptors detect “non-self” molecules produced by pathogens or endogenous damage-associated molecular patterns, they initiate an inflammatory response that facilitates tissue repair/regeneration and homeostasis [74].

Of note, sustained activation of one of the NOD-like receptors, namely, leucine-rich repeat and pyrin domain containing protein 3 (i.e., NLRP3 inflammasome), with exogenous or endogenous triggers, exacerbates a number of chronic inflammatory diseases (e.g., obesity, type 2 diabetes mellitus, atherosclerosis) [74]. In terms of sleep disturbance, NLRP3 regulation within cells exhibits a circadian rhythm. Preclinical studies have described associations between alterations in NLRP3 signaling and sleep regulation [75], chronic obstructive sleep apnea [76], and sleep deprivation [77]. In a study of patients with insomnia with objective short sleep duration [78], increases in plasma levels of NLRP3 were associated with less total sleep time and longer wake after sleep onset times.

4.9. Th17 cell differentiation

Interleukin-17 producing CD4⁺ helper T cells (i.e., Th17 cells) are pro-inflammatory immune cells that protect against bacterial and fungal infections [79]. Alterations in Th17 cells are associated with a number of inflammatory diseases [80–82]. In terms of sleep disturbance, evidence suggests that lineage specification of Th17 cells is under direct circadian control [83]. In addition, in an experimental model of multiple sclerosis [84], melatonin, a hormone that is involved in the regulation of circadian rhythm, interfered with the differentiation of Th17 cells.

4.10. Intestinal immune network for IgA production

The intestine is the largest lymphoid tissue in the body. One striking feature of intestinal immunity is its ability to generate large amounts of noninflammatory IgA antibodies that serve as the first line of defense against bacteria and other pathogens. This immunologic barrier is achieved through IgA antibodies that prevent pathogens from adhering to intestinal epithelial cells; suppress the growth and virulence of pathogens; and neutralize toxins [85]. As part of humoral immunity, a primary function of IgA is to maintain the functioning of a healthy intestinal microbiota.

Recent evidence suggests that a bidirectional relationship exists between the immune system in the gastrointestinal tract and the brain. For example, while gut microbiota may modulate brain function, the brain may alter the gut's microbiota by changing intestinal permeability and gastrointestinal motility [86]. Recent reviews have noted that the gut microbiome is essential for the maintenance of normal sleep. In addition, abnormal sleep patterns and alterations in the duration of sleep can effect the composition, diversity, and function of the gut microbiota [87,88]. Finally, interventions that target the gut microbiome may improve insomnia [89].

4.11. TCR signaling pathway

The TCR is an antigen recognition signaling pathway that is composed of membrane proteins, co-receptors, kinases, and ligands. Following interaction with a foreign antigen that is associated with MHC molecules, a cascade of signaling events occur that results in T cell proliferation, cytokine production, and differentiation of T cells into effector cells [90]. Numerous immunologic functions, including the type and magnitude of various immune responses, are dependent on circadian rhythms and sleep [12]. In a study that investigated how sleep and circadian rhythm influence the number and function of natural regulatory T cells in healthy adults [91], these T cells displayed a circadian rhythm with the highest cell counts during the night and the lowest levels during the day. In addition, sleep deprivation decreased T cell proliferation.

4.12. Complement and coagulation cascades

The complement system is a proteolytic cascade that mediates humoral immunity [92]. The coagulation system is an amplification cascade that maintains hemostasis [93]. Given that these two systems share a common ancestry, evidence suggests that cross talk occurs between these two systems; particularly in the setting of inflammation [93]. In terms of sleep disturbance, as noted in one review [94], the complement system exhibits circadian properties. While not completely understood, disruption in circadian rhythms can induce dysregulation of pro-inflammatory cytokines and complement factors. Of note, in a recent study [95], compared to healthy controls, patients with obstructive sleep apnea had elevated morning and evening levels of plasma C3a. In addition, elevated levels of C3a were correlated with increased severity of obstructive sleep apnea.

4.13. TNF signaling pathway

TNF is a soluble pro-inflammatory cytokine member of the TNF ligand family [96]. TNF can modulate multiple signaling pathways (e.g., apoptosis, inflammation) [97]. As noted in one review [98], in healthy animals and albeit limited studies in healthy humans, acute enhancement or inhibition of TNF in the brain, promotes and inhibits sleep. In addition, most of the clinical studies on the role of TNF and sleep, evaluated patients with chronic elevations in systemic TNF. Across these studies, these patients had disruptions in sleep and normalization of their TNF levels resulted in improvements sleep [98]. Equally important, TNF-alpha plays an important role in the regulation of circadian rhythms by regulating the transcription of clock genes [99].

4.14. Strengths and limitations

This study had several strengths including: a relatively large sample size; rigorous quality controls; and results from independent tests across two samples. However, some limitations must be acknowledged. Because of its cross-sectional design, longitudinal studies are needed that assess for associations between changes in sleep disturbance before, during, and after chemotherapy and pathway perturbations. Second, given that this study is the first to evaluate for associations between sleep disturbance and pathway perturbations in oncology patients undergoing chemotherapy, our findings warrant confirmation in independent samples. Third, this study evaluated for sleep disturbance using a single subjective measure. Future research should explore associations between subjectively and objectively measured sleep disturbance and pathway perturbations. Fourth, we did not evaluate for the use of sleep medications. Lastly, because sleep disturbance was assessed only during chemotherapy, future studies should attempt to replicate these findings in oncology patients undergoing other types of treatment (e.g., radiation therapy, immunotherapy, surgery).

4.15. Conclusions

Consistent with the findings summarized above for preclinical and clinical research in patients without cancer, our findings suggest strong associations between sleep disturbance and perturbations in several immune-inflammatory pathways in oncology patients receiving chemotherapy. However, as illustrated in this analysis, associations between sleep disturbance and immune-inflammatory mechanisms are complex. Therefore, additional research is needed to elucidate the role of these mechanisms in the development and maintenance of sleep disturbance in patients with cancer.