Cancer as a Homeostatic Challenge: The Role of the Hypothalamus

Abstract

The initiation, progression, and metastatic spread of cancer elicits diverse changes in systemic physiology. In this way, cancer represents a novel homeostatic challenge to the host system. Here, we discuss how the hypothalamus, a critical brain region involved in homeostasis senses, integrates, and responds to cancer-induced changes in physiology. Through this lens, cancer-associated changes in behavior (e.g., sleep disruption) and physiology (e.g., glucocorticoid dysregulation) can be viewed as the result of an inability to re-establish homeostasis. We provide examples at each level (receptor sensing, integration of systemic signals, and efferent regulatory pathways) of how homeostatic organization becomes disrupted across different cancers. Finally, we lay out predictions of this hypothesis, and highlight outstanding questions that aim to guide further work in this area.

Cancer, the Brain, and Systemic Homeostasis

Tumors disrupt the activity of cells within their local microenvironment in order to evade immune destruction and meet metabolic demands for growth, proliferation, and metastasis. Less well described, however, is the ability of cancers to alter distal organ function (e.g., the brain) to complement local paracrine/autocrine signaling. Evidence has accumulated demonstrating that diverse cancers may elicit neurological symptoms that can be devastating to a patient’s quality of life, and in some cases, these symptoms are associated with cancer mortality. For example, patients with breast cancer frequently experience disruptions in sleep and wakefulness (e.g., insomnia, low sleep efficiency, and daytime fatigue). These problems independently predict cancer mortality even when controlling for covariates like age, hormone receptor expression, cortisol concentrations, depression, and metastatic spread [1]. Critically, these cancer-associated behavioral comorbidities are often present prior to diagnosis and become exacerbated throughout the course of surgery, treatment, and remission/recurrence [2][3] (see Box 1). Despite the large amount of clinical evidence supporting the existence of neurological sequelae secondary to cancer, only a few mechanistic studies have tackled this multi-system crosstalk (that is, independent of cancer treatments) [4]. Presumably, this gap in knowledge is partially due to the way scientific disciplines are divided. Neuroscientists and cancer biologists are frequently siloed into separate departments with limited interaction, and they use increasingly specialized language in their respective fields, likely hampering collaborative efforts. Additionally, the tools required to tease apart complex multi-tissue interactions in health and disease have only recently been developed, and in many cases still require substantial improvement. However, a growing number of papers in recent years has galvanized interest at the intersection of these two fields, which has yielded the nascent research area of ‘cancer neuroscience’ [5][6].

Box 1: The role of psychological stress.

In this article, we focus primarily on the effects of the physiological “internal” stress that cancer represents. Another important element to consider is that of external stress (e.g., emotional or psychosocial), which can contribute both to the patient’s subjective experience and potentially to the progression of cancer. While a detailed discussion of external stress in cancer is beyond the scope of the current article, we provide a brief overview of its role here. The types of external stress associated with cancer are varied. For example, the diagnosis of cancer itself can cause an enormous stress response in some individuals, and can result in some of the physiological changes discussed above (e.g., elevated glucocorticoid and glucose concentrations) [127]. Consequently, cancer patients who undergo psychological therapy (e.g., supportive group therapy, mindfulness meditation) not only show improved psychological outcomes, but also reduced cancer progression, rates of relapse, and increased survival [130] [131]. However, it is important to note that cancer-induced changes in physiology and behavior could start years before diagnosis [3][128], with the additional psychological stress of having cancer further exacerbating these problems. The role of psychological state on cancer progression can also be observed in rodent models. For example, investigators demonstrated that mice with ‘high anxiety’ have decreased immunity and increased tumor progression and metastasis [129] independent of external stressors. To what extent the stress of a cancer diagnosis compared to cancer-induced changes in the physiology that began before diagnosis (e.g., alterations in the HPA axis) contribute to systemic problems faced by cancer patients needs to be further evaluated.

The hypothalamus is a critical brain structure responsible for establishing and maintaining homeostasis in systemic physiology. It is essential for the regulation of sleep/wake behavior, body temperature, sexual behavior, appetitive and consummatory phases of feeding, systemic energy balance, and circadian rhythms, among other functions [7]. Most, if not all, genetically-identified cell types within the hypothalamus can serve an integrative function, receiving inputs from a wide variety of internal and external stimuli to tune homeostatic responses [8][9]. Many physiological cues that hypothalamic neurons sense are frequently altered in the context of cancer, including inflammatory cytokines/chemokines (e.g., IL-6) [10] [11], metabolic fuel sources and intermediates (e.g., glucose) [12], and endocrine hormones (e.g., leptin, ghrelin) [13] [14].

Here, we provide an overview of evidence supporting the hypothesis that some cancer-related physiological/behavioral co-morbidities may arise due to aberrant homeostatic responses to the cancer itself, where the hypothalamus plays a key role. We discuss examples from pre-clinical studies on how cancer can disrupt systemic signals, how the central nervous system detects these changes, integrates them, and responds in order to influence downstream physiology and behavior (see Fig. 1). Finally, we lay out predictions of this hypothesis and discuss outstanding questions that remain.

Figure 1: Disrupted hypothalamic feedback loops in cancer.

(a) Tumor-associated immune signaling or metabolic stress activates paraventricular hypothalamic CRF-expressing neurons, which act as the first node in the HPA axis. Downstream signaling and release of glucocorticoids (cortisol, corticosterone) subsequently impairs anti-tumor immunity and alters cancer progression. The normal negative feedback of glucocorticoids on the PVN is attenuated because of continued stimulus from tumor-derived factors or those of the host response. (b) Neurons in the arcuate nucleus (POMC, NPY, AgRP) and in the lateral hypothalamus (HO) sense changes in systemic energy balance via peripheral and central chemoreceptors and humoral receptors (ligands include glucose, ghrelin, leptin, insulin). These become altered due to metabolic changes elicited by the tumor, likely through an inflammatory mechanism (e.g., IL-6). The hypothalamus responds via the autonomic nervous system (ANS) to normalize peripheral physiology (via NE or ACh signaling), but is unable to do so effectively due to continued stimulus from the tumor. (c) Schematic of the proposed homeostatic framework at the level of an individual hypothalamic neuron. Diverse inputs (e.g., metabolic, immune) with different strengths (weights, ω) converge onto genetically-defined neurons which then integrate/summate these signals. If the net input strength surpasses threshold, the neuron fires and elicits downstream effects via the HPA axis or autonomic nervous system. Cancer disrupts the inputs (e.g., ghrelin, leptin, insulin, cytokines) to produce aberrant responses. Figure created with bioRender.com.

Sensing Cancer-Related Physiological Disruption

The brain is exquisitely sensitive to the physiological state of the body, putatively providing a real-time map of the internal environment across sub-conscious and conscious levels of organization (processes often referred to as interoception [15]). Conscious awareness of interoceptive processing typically occurs only when homeostasis is sufficiently perturbed (e.g., when feeling hunger, thirst, or a full bladder), or when a change in physiology poses a significant threat to the body (e.g., in the cases of nausea or angina). To monitor the body’s internal environment, the CNS uses multiple receptor modalities selective for discrete signals. Below, we outline several of these pathways, and discuss evidence for how the signals that they detect are often disrupted in cancer.

Chemoreceptors

The body employs an array of receptors sensitive to a variety of chemicals and substances (e.g., chemoreceptors). These are complemented by receptors that sense changes in extracellular glucose (glucoreceptors), tissue deformation (mechanoreceptors), changes in pressure (baroreceptors), temperature (thermoreceptors), ion concentrations (osmoreceptors), and pH (acid sensing channels) that act to relay information about the body to the brain [16][17][18]. Complementary (central) chemoreceptors are also present in the brain itself, with substantial enrichment in the hypothalamus and brainstem [19][20]. Therefore, sensitivity to the metabolic and chemical conditions of the body can be jointly conveyed by both peripheral and central chemoreceptors. Many of the ligands for these receptors become increasingly disrupted throughout the course of cancer progression. For example, it has long been known that patients with cancer often display impaired glucose tolerance. Freund made the original observation (in 1885) that ~89% of cancer patients were spontaneously hyperglycemic [21], and more recently a study of 850 cases revealed that hyperglycemia (blood glucose > 200 mg/dL) was observed 3x more frequently in patients with cancer than in age-matched controls [12]. These findings have been replicated in several preclinical models, providing support for the notion that this hyperglycemic phenotype may be independent of other extraneous factors like stress, age, diet, or treatment modality [22][23][25]. This increase in systemic glucose likely facilitates tumor progression, as cancer cells preferentially use glucose to generate ATP via glycolysis, rather than via mitochondrial oxidative phosphorylation, even in properly oxygenated tissues (i.e., the Warburg effect; [26]). As glucose is the brain’s primary fuel source [27], it keeps a careful watch on local and systemic glucose concentrations in order to maintain metabolic homeostasis. Peripheral and central glucoreceptors are distributed throughout the body and brain, which can collectively act to adaptively regulate glucose concentrations across space and time. In the periphery, glucoreceptors are present on vagal afferents (which terminate in the nucleus of the solitary tract; NTS), and putatively use a strategy similar to pancreatic ß-cells to sense glucose [28]. This involves relying on glucokinase and/or ATP-sensitive potassium channels containing the Kir6.2 subunit (K_ATP), closure of which causes neuronal depolarization in response to changes in extracellular glucose [29].

Glucose-sensitive neurons in the hypothalamus were first described in the 1960s, where they were found to reside in the lateral, arcuate, and ventromedial hypothalamic nuclei [30][31]. Direct sensing of glucose occurs in at least a dozen subsets of neurons within the hypothalamus, including those expressing pro-opiomelanocortin (POMC), agouti-related peptide (AgRP), and neuropeptide Y (NPY) in the arcuate nucleus, along with hypocretin/orexin (HO) and melanin-concentrating hormone (MCH) neurons in the lateral hypothalamus [32] [33] [34]. HO-expressing neurons, which are directly inhibited by elevations in extracellular glucose [34], use a unique strategy to sense glucose and other chemical signals such as pH and oxygen. A specialized tandem-pore potassium channel (K_2P), which carries a leak potassium current, promotes membrane hyperpolarization and neuronal silencing in response to physiological changes in extracellular glucose [35]. In a reciprocal fashion, these neurons depolarize and fire in response to insulin-induced hypoglycemia [36]. Critically, changes in the activity of these neurons links alterations in systemic metabolic state with behavioral arousal, as HO neurons regulate the activity of diverse neuronal systems largely involved in both sleep/wake states (e.g., locus coeruleus noradrenergic neurons) [37] and energy balance (e.g., arcuate POMC neurons) [38]. A detailed description of other chemoreceptors is beyond the scope of this review, but we refer the reader to [15] [39] for further discussion.

Humoral Receptors

Specialized receptors for endocrine and immune signals complement the actions of chemoreceptors in the regulation of systemic homeostasis. Consistent with the hyperglycemic phenotype observed in many patients with cancer (discussed above), studies have demonstrated an association between different humoral components of systemic energy balance and cancer [40], specifically leptin [41], ghrelin [42], and insulin [43], among others. Leptin, an adipokine hormone primarily produced by white adipose tissue (WAT), acts centrally within the hypothalamus to inhibit food intake and regulate energy balance around its homeostatic range [44]. Leptin signals at the plasma membrane through its gp130-coupled cognate receptor (LEP-R/OB-R), which engages several secondary messenger cascades, including the phosphatidylinositol 3-kinase (PI3K), extracellular-signal regulated kinase (ERK1/2), and signal transducer and activator of transcription-3 (JAK/STAT3) pathways, which are critical in promoting changes in the expression of hundreds to thousands of genes [45]. Within the hypothalamus, arcuate POMC neurons are major (but not the only) targets for leptin’s action [46]. These neurons project to the dorsomedial nucleus (DMH), paraventricular nucleus (PVN), and the lateral hypothalamus and presumably act to transduce the adipocyte-derived signal into a neural code [47]. In relation to leptin, apart from its canonical role in energy balance, numerous studies have described altered leptin signaling directly in tumors and (more broadly) in the systemic circulation of cancer patients. Overexpression of both leptin and its receptor (LepRb isoform) have been noted in multiple cancers, including breast [48], colorectal [49], liver [50], and thyroid cancer [51]. Overexpression of leptin is consistent with anorexia observed in some cancer patients, although it remains to be empirically determined whether leptin-responsive cells within the hypothalamus are essential for this phenomenon. Leptin overexpression is generally pro-tumorigenic, and is associated with increased expression of anti-apoptotic proteins (e.g., BCL-xL), thus aiding in tumor cell survival and proliferation [52][53]. Interestingly, increasing hypothalamic brain-derived neurotrophic factor (BDNF) signaling (artificially or through environmental enrichment) can promote downregulation of adipocyte leptin production and inhibition of cancer cell proliferation [54]. Studies like these highlight the potential breadth of leptin’s involvement in neuronal regulation of cancer in the body.

Ghrelin, a peptide hormone typically produced and secreted in the stomach and brain, operates in a reciprocal fashion to that of leptin. Indeed, it promotes food intake and stimulates growth hormone secretion [55]. Ghrelin was identified as the endogenous ligand for the growth hormone secretagogue receptor (GHSR), and acts primarily through stimulating NPY neurons in the arcuate nuclei to regulate feeding behavior [55] through protein kinase A (PKA) and N-type calcium channel-dependent mechanisms [56]. In a manner similar to leptin, ghrelin, its receptor (isoforms GHSR1a and 1b), and the activity of its catalytic enzyme ghrelin-O-acyl transferase (GOAT) are often deregulated in patients with diverse malignancies [57][58]. Indeed, tumors or cells within the tumor microenvironment can directly secrete ghrelin to aid in metastatic spread and cellular proliferation [59][58]. GHSRs are expressed on vagal afferent fibers and neurons within the nodose ganglion (where vagal afferent cell bodies reside [60]). Consequently, ghrelin-induced feeding and additional physiological changes can be largely attenuated or fully abolished by vagotomy [61]. How ghrelin acts on peripheral receptors and central targets in cancer is an area that requires additional research, although emerging evidence suggests divergent roles for the acylated and unacylated form of ghrelin in cancer [62].

Insulin, a peptide hormone produced in large quantities by pancreatic ß-cells, functions to coordinate glucose uptake into cells or promote its oxidation. Accumulating evidence suggests that overactivation of the insulin receptor (IR) is a common trait in cancer cells and that it can potentially aid in the progression of tumor growth [63] [64]. Sustained systemic elevations in insulin positively correlate with cancer mortality [65]. Within the hypothalamus, insulin receptors are expressed in both arcuate POMC and NPY neurons [66], where ligand binding activates the insulin receptor substrate-2 (IRS2)–PI3K signaling pathway [67]. This action is critical for negative feedback of hepatic glucose production, ensuring relative stability in peripheral energy balance [68]. In the periphery, changes in insulin concentrations are conveyed to the brain via insulin receptors present on vagal afferent fibers, including those that directly innervate the pancreas [69], as well as on endothelial cells lining brain vasculature [70]. How insulin signaling within the tumor microenvironment distally regulates these insulin sensing components remains undefined.

Another major class of humoral signals generally disrupted in cancer are messaging molecules of the immune system (i.e., chemokines and cytokines). A common finding across inflammatory conditions (including cancer) is an increase in the concentrations of circulating interleukin-6 (IL-6), a pleiotropic cytokine that acts on many target tissues throughout the body and brain [71][72]. In ‘triple-negative’ (basal) breast cancer, for instance, stromal and tumor cells can directly secrete IL-6, which acts on lymphatic endothelial cells (LECs) within the primary tumor and distant sites to promote metastasis [73]. IL-6-induced CCL5 expression recruits CCR5+ cancer cells into lymphatic circulation and allows for tumor cell dissemination. Classical IL-6 signaling acts through its cognate receptor (IL-6Rα) which consists of an 80-kDa IL-6 binding protein and glycoprotein 130 (gp130) primarily in the liver [74]. Similar to leptin, IL-6 engages the gp130/JAK/STAT3 signaling pathway which primarily induces tumor growth, hinders anti-tumor immunity, and in patients, promotes poor clinical outcomes [75] [76]. IL-6 signaling in the liver further disrupts glucose uptake, as an IL-6/STAT3 target gene product, SOCS3, can directly interact with insulin receptor substrates (IRS1/2) for ubiquitin-mediated degradation [77]. Additionally, a large number of IL-6-, IL-6R-, JAK-, and STAT3-inhibitors are currently undergoing clinical and preclinical investigations as anti-cancer therapies [78]. IL-6 acts as a general inflammatory alarm, highlighted by its primary role in orchestrating the acute phase response [79]. With sufficient stimulation, this signal is relayed to brain endothelial cells which can propagate the inflammatory signal via prostaglandin release to elicit fever responses [72]. Within the hypothalamus, IL-6 trans signaling (via the soluble IL-6R) engages gp130 expressed on CRF neurons in the paraventricular nuclei to activate the HPA axis [80]. These direct and indirect sensing mechanisms allow the brain to respond to peripheral inflammatory cues in real-time [81][82][83]. We further discuss the HPA-axis below.

The Hypothalamus as a Systemic Integrator

Which neurons sense and integrate physiological cues disrupted by cancer in the periphery, and how are these signals interpreted? The hypothalamus is among the most phylogenetically conserved regions in the vertebrate brain, reflecting its essential role in maintaining physiological and behavioral homeostasis. Its functions rely on a diverse and complex network of nuclei comprised of specialized circuits and neurotransmitter systems [84]. Critically, the hypothalamus regulates a multitude of functions frequently disrupted in patients with cancer, including arousal, feeding, energy balance, stress responses, circadian rhythms, and motivated behavior [7]. During the course of cancer progression, changes in systemic physiology are sensed via various routes (discussed above), which the hypothalamus must integrate and respond to in an attempt to restore homeostasis [85] [86]. How this occurs on a molecular level is largely unknown, as this integration occurs across multiple temporal scales (seconds, hours, days) and spatial domains (various hypothalamic sub-nuclei and higher order structures).

A good example of a (conscious) consequence of hypothalamic integration may be the subjective feeling of hunger, which often doesn’t come about immediately, but arises slowly over time. There is evidence that discrete neuron types within the hypothalamic paraventricular nucleus (PVN) employ a ‘slow’ integrative process that allows them to operate on physiological timescales (minutes to hours). This is achieved by their slow ‘off’ kinetics driven by a specialized voltage gated sodium channel (Na_V1.7), conferring ‘near perfect’ integration properties to these neurons [87]. This allows these cells to summate inputs over longer timescales than those that use more conventional (‘leaky’) integration strategies. As the conscious awareness of a developing problem in the body typically comes about rather slowly (i.e., hours, days, or even weeks), it follows that hypothalamic integration of immune and metabolic factors disrupted by cancer may operate on similar timescales. Additional work is required to systematically test this empirically.

Hypocretin/orexin (HO)-expressing neurons within the lateral hypothalamic area serve a non-redundant role in ensuring the stability of wakefulness, highlighted by the fact that their ablation or loss of their neurotransmitter content results in the debilitating sleep disorder narcolepsy [88]. As we discussed above, sleep disruption is a common phenomenon among patients with cancer. In mouse models of breast cancer, HO neurons were found to be hyperactive (increased cFos immunoreactivity) during the development of sleep fragmentation during the later stages of tumor growth [22]. This was associated with reductions in systemic leptin concentrations, as well as disruptions in ghrelin sensitivity and elevations in systemic IL-6 and hyperglycemia. As HO neurons seem to integrate these and other physiological signals altered in cancer, it follows that fragmented sleep was a result of altered input strengths to HO neurons in tumor-bearing mice [89]. Consistent with this idea, attenuation of HO signaling via dual receptor antagonism promoted deep, restorative sleep in tumor bearing mice, suggesting that dysfunctional output of these neurons (as a response to cancer in the body) is capable of driving sleep disruption in breast cancer. Additionally, this intervention attenuated the hyperglycemic phenotype observed in these mice [22]. This could be recapitulated by sympathetic denervation (using the noradrenergic neurotoxin 6-OHDA), implicating altered HO neuronal activity in sleep and metabolic disruption (via the SNS) in non-metastatic breast cancer. A clear understanding of how these processes works at the population, cellular, and subcellular level is an area that requires additional work [90].

Efferent Pathways Regulating Peripheral Physiology in Cancer

Autonomic Nervous System

The brain elicits control over the body via neural and humoral pathways. The autonomic nervous system (ANS), usually partitioned into sympathetic (SNS) and parasympathetic (PNS) branches, acts as a major neural output of the brain regulating systemic physiology. In recent years, the role the ANS plays in cancer processes has slowly come into focus [91][6]. It has long been known that the hypothalamus regulates autonomic output, as Shimazu and colleagues demonstrated its role in autonomic regulation of liver glycogen metabolism in the 1960s [92]. The hypothalamus contains pre-autonomic neurons that are intricately linked to downstream output nuclei in the brainstem [93][94]. For example, HO neurons in the lateral hypothalamus directly innervate the dorsal vagal complex (DVC) and raphe pallidus (RPa) to regulate respiration, body temperature, and peripheral organ function [95][96][97]. A detailed discussion of hypothalamic-autonomic anatomy is beyond the scope of this review, but we refer the reader to the following resources for more information [98][99].

Both sympathetic and parasympathetic nerves infiltrate prostate tumors, where they can act synergistically to facilitate early cancer development and subsequent metastatic spread [100]. Adrenergic signaling from sympathetic nerves can act directly on cancer cells or stromal cells in the tumor microenvironment [101][102]. In prostate cancer, SNS-derived norepinephrine (NE) promotes changes in endothelial cell metabolism which results in the enhancement of angiogenesis and energy supply to the tumor parenchyma [101]. In the gastrointestinal tract (which receives major PNS input), cholinergic signaling from nerves and local tuft cells drives neoplasia via the induction of nerve growth factor (NGF), which engages a feedforward loop promoting further neural innervation and carcinogenesis. This effect is dependent on muscarinic (M3) acetylcholine (ACh) receptor signaling, suggesting antagonism of muscarinic receptors to inhibit the ACh-NGF axis as a possible strategy in the treatment of gastric cancer [103].

In breast cancer, sympathetic nerves can directly promote cancer metastasis via beta-adrenergic signaling [91], while parasympathetic nerves can elicit tumor inhibition [104]. In primary patient samples, the presence of sympathetic nerves, or absence of parasympathetic nerves, is associated with poor clinical outcomes in breast cancer [104]. The mechanisms by which different arms of the autonomic nervous system elicit these opposing effects is an area of active research. It is possible that the effects depend on how adrenergic and cholinergic signals regulate the activity or mobility [105] of local immune cells (e.g., CD8+ T cells) [106][107], rendering tumors more or less responsive to checkpoint blockade (e.g., anti-PD-1/PD-L1) or other immunotherapies. Consequently, inhibition of beta-adrenergic signaling is now being tested as an anticancer strategy in a variety of malignancies, including breast cancer [108].

Hypothalamus-Pituitary-Adrenal (HPA) Axis

A major humoral output of the brain regulating systemic physiology relevant for cancer is the HPA axis. Corticotropin-releasing factor (CRF)-expressing neurons in the PVN act as the first node in this pathway, and these neurons are activated in response to diverse stressful stimuli, including psychological, metabolic, and inflammatory ones [109]. CRF neurons release both CRF and arginine vasopressin (AVP) into the anterior pituitary via the hypophysial portal veins to elicit adrenocorticotrophic hormone (ACTH) release into circulation. Downstream binding of ACTH to melanocortin type 2 receptors (MC2-R) in the adrenal cortex promotes glucocorticoid (e.g., corticosterone/cortisol) release [110]. Negative feedback in the hypothalamus resolves this ‘stress response’ in order to prevent runaway amplification of glucocorticoid production [111][112]. Glucocorticoids act on virtually all cells throughout the body primarily via the phylogenetically conserved nuclear glucocorticoid receptor (GR) to facilitate glucose mobilization [113], immune suppression [114], and setting circadian oscillators [115], among other functions. Indeed, optogenetic stimulation of CRF neurons in the PVN is able to rapidly induce systemic signatures of immunosuppression [116] and alter B-cell antibody production [117] via their connections to the splenic nerve.

Altered HPA axis function is a common phenomenon in patients with cancer [118][119]. Blunted or disrupted rhythms in glucocorticoid secretion can predict mortality in breast [120], lung [121], ovarian [122], and colorectal cancer patients [123]. This may be mediated by the largely immunosuppressive actions of glucocorticoids. In preclinical models of cancer cachexia, inflammation-induced changes in hepatic ketogenesis promotes metabolic stress and elevated corticosterone concentrations in circulation [124]. This directly impairs anti-tumor immunity and increases the failure rate of checkpoint inhibitor immunotherapy [124]. Findings from immunodeficient mice with breast cancer indicate that increased systemic glucocorticoid concentrations during cancer progression promotes the activation of the GR at distal metastatic sites, which is associated with increased tumor cell colonization and decreased survival [125]. Given the widespread use of glucocorticoids to alleviate diverse symptoms in patients with cancer, studies like these emphasize the need for extreme caution when using glucocorticoids (e.g., dexamethasone), as this may result in paradoxically worse outcomes [126]. It should be emphasized, as mentioned earlier, that the potential connections between the HPA axis and various cancers seems to be independent of the stress of a cancer diagnosis or other relevant covariates. Nonetheless, in the broader picture, these factors can be important contributors to patient health and cancer outcomes. Another important factor to consider is the psychological burden associated with cancer, discussed briefly in Box 1.

Concluding Remarks

Patients with cancer frequently experience debilitating symptoms ultimately controlled by the central nervous system, including sleep/wake disruption, changes in appetite and feeding behavior, reduced motivation, and cognitive impairment, among others. We presented a framework through which these symptoms can be viewed as an aberrant homeostatic response to cancer-induced changes in systemic physiology. These changes are likely detected by central and peripheral chemo- and humoral receptors which relay information to the brain and more specifically, the hypothalamus. These stimuli can be integrated within this brain structure or passed onto additional neuronal circuits in higher-order regions (e.g., insular cortex). Finally, the brain produces an output (e.g., via the HPA axis or the autonomic nervous system) in an attempt to restore systemic homeostasis. We suggest that failure to restore balance can result in the development of cancer-associated behavioral and physiological abnormalities (see Figs 1, 2).

Figure 2: Conceptual framework of altered hypothalamic function driving physiological changes in cancer.

(a) Tumor progression successively disrupts host physiology (cytokines, metabolites…) which alter the function of local cells and then gets propagated to the brain via neural and humoral routes. T hypothalamus is a major integrator of these incoming signals either directly or via its action as a relay to higher order brain structures (e.g., insula). Reciprocal outputs (HPA axis and ANS) are used in an attempt to restore systemic homeostasis. (b) This aberrant hypothalamic activity results in diverse physiological consequences secondary to the primary tumor growth. We hypothesize that attenuation of these aberrant signaling pathways could resolve some of these problems, which are frequently reported by patients across a variety of cancers. Figure created with bioRender.com.

This hypothesis lays out several testable predictions that we anticipate will galvanize further research in this area (and see Outstanding Questions): (1) Cancer-induced changes in systemic physiology and behavior should be detectable at each level of homeostatic regulation (receptor, central integrator, and effector system); (2) normalization of homeostatic feedback loops (or removal of the malignancy) should attenuate physiological and behavioral co-morbidities; (3) conscious awareness of something wrong (e.g., daytime fatigue) should be secondary to hypothalamic dysfunction in cancer; and (4) transduction of peripheral physiology into a hypothalamic neural code may be detectable using brain imaging techniques. Additionally, we hope to garner interest from computational neuroscientists to develop models of cancer-induced physiological disruption, which can be used to develop testable hypotheses and interpret high-dimensional physiological datasets. Using this integrative strategy, treatments and surgical techniques directly targeting the cancer may be complemented by those that ensure the optimal health of the patient and limit the side-effects of therapy. Together, we believe that this holistic approach will provide the greatest benefit to those suffering from this devastating disease.

Outstanding Questions.

• What receptors are important for relaying cancer-related changes in physiology to the brain?

• Are these pathways unique to different types of cancer, or common across different malignancies?

• What hypothalamic nuclei and other brain areas receive and integrate these signals?

• How do these pathways evolve throughout the course of cancer initiation, growth, and metastasis?

• Can normalizing these dysregulated feedback loops improve outcomes in cancer?

• When and how do cancer-induced changes in physiology reach conscious awareness (e.g., nausea, pain, fatigue)?

• How do cancer treatments alter these pathways, and can treatment modalities be developed that aim to restore normal homeostatic function without compromising anticancer efficacy?

• Is it possible to identify a central ‘neural code’ for cancer in the body that is readable with current technologies?

Highlights.

• Cancer represents a homeostatic challenge to the host system.

• The brain detects cancer-induced changes in physiology via multiple receptor modalities present on afferent nerves, as well as centrally within the hypothalamus and other brain structures.

• Integration of peripheral signals elicits downstream changes in neuronal output.

• The hypothalamus attempts to restore homeostasis via the HPA axis and autonomic nervous system.

• Failure to restore homeostasis can result in deleterious phenotypes associated with cancer, such as blunted glucocorticoid responses and sleep/wake disruption.