- © 2004 by American Society of Clinical Oncology
Appetite and Cancer-Associated Anorexia: A Review
- From the Harry R. Horvitz Center for Palliative Medicine, Department of Hematology/Medical Oncology; the Cleveland Clinic Foundation, and the Cleveland Clinic Taussig Cancer Center Cleveland, OH
- Address reprint requests to Mellar P. Davis, MD, FCCP, Cleveland Clinic Foundation, 9500 Euclid Ave, R35, Cleveland, OH 44195; e-mail: davism6{at}ccf.org
Abstract
Appetite is governed by peripheral hormones and central neurotransmitters that act on the arcuate nucleus of the hypothalamus and nucleus tactus solitarius of the brainstem. Cancer anorexia appears to be the result of an imbalance between neuropeptide-Y and pro-opiomelanocortin signals favoring pro-opiomelanocortin. Many of the appetite stimulants redress this imbalance. Most of our understanding of appetite neurophysiology and tumor-associated anorexia is derived from animals and has not been verified in humans. There have been few clinical trials and very little translational research on anorexia despite its prevalence in cancer.
INTRODUCTION
Anorexia is one of the most common symptoms in advanced cancer. Anorexia is the result of a failure of usual appetite signals. Cachexia, the debilitating state of involuntary weight loss, is frequently described with anorexia as the anorexia-cachexia syndrome. Cancer causes disengagement of the normal balance of energy intake to caloric expenditure in the face of increasing catabolism. Yet weight loss does not always correlate with anorexia, and the pathophysiology may be different.1 The spectrum of clinical presentations ranges from predominant anorexia to predominant sarcopenia (muscle loss). Weight loss is not accounted for by diminished nutritional intake alone in that both symptoms occur in most patients.2 Clinically, they may be inseparable and their effects difficult to measure. Anorexia may consist of appetite loss, satiety, combined satiety and appetite loss, or altered food preferences. Satiety may occur during meals, indicating a lack of gastric accommodation, or afterward, as a result of gastroparesis and delayed antropyloric transit. The difference may be seen clinically as premature discontinuation of feeding, due to satiety from poor gastric accommodation or postprandial bloating or nausea from antropyloric motility delay.
Multiple reviews on anorexia have been written in the last 10 years.3–23 A search of OvidWeb Medline Database (Ovid Technologies, NY) using “Anorexia and Cachexia” (1992 to 2002) revealed 21 general reviews published from 1992 to 2002. Most intermingle anorexia with cachexia. The proportion of cancer patients treated for anorexia is unknown. The most commonly reported appetite stimulants were megestrol acetate and corticosteroids. Listed less frequently were dronabinol and melatonin. Metoclopramide is recommended for early satiety.24 The only medication exposed to dose-response trials has been megestrol acetate. Prospective comparative trials failed to demonstrate an advantage of megestrol acetate plus dronabinol over megestrol acetate alone.23 Fluoxymesterone seems less effective than corticosteroids and megestrol acetate; the latter two agents are equally effective, though with long-term use, corticosteroids have more side effects.25
A French working group developed the following guidelines, which are a reasonable synopsis of clinical experience: (1) corticosteroids and synthetic progesterones are appetite stimulants; (2) both drug groups are useful for cancer-associated anorexia, particularly in the palliative setting, despite potential side effects; (3) the most effective way of using them is unknown; clinical trials are therefore important; (4) cyproheptadine, metoclopramide, nandrolone, and pentoxifylline should not be used outside clinical trials; and (5) hydrazine sulfate is not indicated.26,27
Despite the paucity of evidence-based clinical data for human anorexia, since the discovery of leptin, a wealth of experimental observations have been made on the basic animal neurophysiology and neurochemistry that governs appetite. Hypotheses about the mechanisms of cancer-associated anorexia have evolved from animal tumor models. Unfortunately, there is little corroboration of the proposed pathophysiology in cancer patients. This is a state-of-the-art review of our current knowledge about normal appetite signals and tumor-associated anorexia pathophysiology, derived primarily from animal studies and some human studies. Possible mechanisms by which appetite stimulants work can be proposed from animal models, which may provide an avenue for future clinical research.
Normal Appetite Signals
Peripheral appetite signals.
Normal hunger-satiety cycles consist of four phases (Tables 1 and 2). A gastric motility phase, mediated by vagus afferents through the nucleus tractus solitarius to the subadjacent dorsal motor nucleus, governs gastric motility. A postabsorptive phase is mediated by duodenal release of cholecystokinin, which binds to vagal cholecystokinin (CCK)-A receptors, reducing motility via the nucleus tractus solitarius. The metabolic phase combines the hepatic release of glucose and insulin and adipocyte release of leptin, which downmodulates neuropeptide-Y (NPY) neurons in the arcuate nucleus of the hypothalamus. Finally, an ileal phase involves glucagon-like peptide-1, which inhibits gastric motility by inhibiting hypothalamic NPY release.28–35
Peripheral Appetite Signals
Neurohumoral Effects on Appetite
Central appetite control.
There is a balance between energy input and resting energy expenditure governed by a parallel system of NPY and pro-opiomelanocortin (POMC) neurons arising from the arcuate nucleus of the hypothalamus (Tables 1, 2, and 3). The main normal regulators of NPY and POMC are leptin and serotonin. Downstream secondary effectors include thyroid releasing hormone, corticotropin releasing hormone, oxytocin, orexin, and melanocortin-concentrating hormone.36
Appetite Receptors
Neuropeptide–and agouti-gene–related transcript (AGRP).
The main appetite stimulating central neurotransmitters are NPY and AGRP (Tables 1, 2, and 3). Both originate from the same arcuate nucleus neurons near the hypothalamic median eminence.36–39 NPY binds to postsynaptic Y1 and Y5 receptors, and AGRP is released from NPY neurons.38–40 A separate but synergistic appetite response occurs through NPY and AGRP release.
Pro-opiomelanocortin and cocaine and amphetamine transcript peptide.
A parallel system arcuate of POMC neurons from the arcuate nucleus innervates both paraventricular and lateral hypothalamus (Tables 1, 2, and 3). Melanocortins oppose NPY. POMC reduces appetite by binding to the paraventricular nucleus and lateral hypothalamus and postsynaptic melanocortin receptor 4 receptors.35–37 Leptin, as the major appetite suppressant hormone, sustains POMC gene expression and suppresses NPY. Within the paraventricular nucleus, secondary negative mediators of appetite are released: thyroid-releasing hormone, corticotropin-releasing hormone, and oxytocin.41 POMC neurons coexpress cocaine and amphetamine–related transcript peptide (CART), which is a potent appetite suppressant.
Leptin.
The dominant negative peripheral NPY regulator is leptin derived from white adipose tissue.42,43 Leptin-induced anorexia is also mediated by corticotrophin-releasing hormone (CRH).44,45
Serotonin and appetite secondary signals.
A separate serotonin system inhibits NPY neuron activity through postsynaptic 5HT1b and 5HT2c receptors, resulting in satiety. However, 5HT1a receptors on NPY neurons facilitate NPY release.46
Energy expenditures, NPY, and POMC signals.
NPY neurons increase parasympathetic output and decrease resting energy expenditure, whereas POMC stimulates sympathetic activity and increases resting energy expenditure.32,47 This normally results in a coordination of appetite signals with basal metabolic rate.
In summary, normal appetite involves a balance between peripheral hormones and neuropeptides and central appetite signals that influence hypothalamic NPY and POMC neurons. Such a balance coordinates appetite signals with gastrointestinal motility by way of the nucleus tractus solitarius.
Pathophysiology of Cancer-Associated Anorexia
Based on animal data, it has been hypothesized that cancer anorexia is the end result of altered central and peripheral neurohormonal signals that govern appetite (Table 4). This proposed hypothesis involves qualitative and quantitative changes in the hypothalamus and gastric signals as a result of systemic and regional proinflammatory interleukin-1 (IL-1), tumor necrosis factor alpha (TNF-α), ciliary neurotrophic factor (CNF), and IL-6.47–53 However, there are few data available in advanced cancer patients to validate this hypothesis.
Proposed Pathophysiology of Cancer-Induced Anorexia Based on Animal Research
IL-1.
IL-1 induces satiety and influences meal size, meal duration, and meal frequency in rats.54,55Cytokines will influence food intake more than water ingestion, and meal size is reduced more than meal duration or frequency.56 There is an imbalance of pro- and antikinetic hormones favoring gastroparesis. Ghrelin, a motilin-like hormone secreted by the stomach, is downregulated by interleukin-1.34 Hypothalamic IL-1 is increased either through access from the median eminence (a circumventricular nucleus without a blood-brain barrier proximal to the arcuate nucleus) or is generated within the hypothalamus.56–58 IL-1 stimulates serotonin release within the serotonin or is generated from tryptophan derived from proteolysis. IL-1 decreases NPY neurotransmission directly and secondarily increases corticotropin releasing hormone in tumor bearing rats.49–52 NPY mRNA expression is reduced by IL-1-β and is site specific, occurring only within the arcuate nucleus.53 Interleukin-1-β, IL-6, and TNF-α interfere with NPY release and neurotransport to a greater degree than gene expression.59,60
Qualitative and quantitative changes in anorexia.
When NPY is infused into the cerebral ventricles of tumor-bearing rats, a blunted appetite response was observed compared with normal rats. This indicates there is either reduced binding of NPY to the neuropeptide-Y1 and Y5 receptors or subcellular alterations in secondary signal transduction that cause persistent adenylyl cyclase activity.61,62 Secondary signals that involve G-proteins and adenylyl cyclase appear less responsive to NPY.63,64 The normal inhibitory effect of NPY on adenylyl cyclase is lost perhaps through uncoupling of receptor associated G-proteins.64 There is a demonstrated decrease in hypothalamic NPY immunostaining in tumor-bearing rats compared with normal controls.65
TNF-α.
TNF-α induces anorexia without influencing NPY neurotransmission. However, in experimental animals, anorexia occurs before hypothalamic TNF-α levels rise.63 Anorexia is accentuated through combinations of cytokines (IL-1, IL-8, and TNF-α).56 Leptin levels in tumor bearing animals are usually low because of chronic TNF-α–induced suppression of leptin release. Leptin does not appear to play a role in cancer associated anorexia.58
CRH.
CRH, which is normally reduced in the paraventricular nucleus and arcuate nucleus during fasting, fails to be downregulated in tumor-bearing rats, despite decreased caloric intake.66
Melanocortin and CNF.
Alpha melanocortin signaling via melanocortin receptor 4 receptors is increased in tumor-bearing rats and is not decreased by intracerebroventricular ghrelin and NPY infusions.67 CNF, a cytokine produced by neurons and Schwann cells, belongs to the IL-6 cytokine family and shares the same gp130 receptor as leptin.68 Ciliary neurotropic factor is increased in the hypothalamus of tumor-bearing animals and causes anorexia through downregulation NPY gene expression, inhibition of NPY release, and downregulation of NPY-Y1 receptor gene.69,70
In summary, the proposed peripheral mechanisms that generate cancer anorexia are based on experimental animal data. Peripheral signals become unbalanced because of reduced ghrelin and facilitated CCK neurotransmission. Leptin appears to be uninvolved. Central events are qualitative and quantitative reductions in NPY and Y1 receptors, facilitated melanocortin neurotransmission upregulation of POMC, and disinhibition of corticotropin releasing hormone in the face of caloric deprivation. The proposed primary mediators of anorexia are IL-1, serotonin, and ciliary neurotropic factor; secondary mediators are TNF-α, IL-6, and CRH.
These findings suggest that there are multiple mediators responsible for tumor associated anorexia (if animal models reflect human experience). How much each plays a role in anorexia is unknown. A major factor limiting our understanding of anorexia is the lack of knowledge about altered central appetite signals in the cancer patient.
Cancer Anorexia in Humans
Ghrelin.
Ghrelin plasma levels are increased in lung cancer patients with anorexia. Ghrelin levels increase with anorexia after chemotherapy, which is at odds with the findings in animal studies.71
Leptin.
Serum leptin increase with acute inflammation and in response to interleukin-1 and TNF.72–74 Tachyphylaxis develops quickly such that, in advanced cancer, serum leptin levels are found to be lower than normal.58 There is no significant difference in leptin circadian rhythms between cancer patients and normal individuals.58 An inverse relationship occurs between serum leptin and interleukin-6 levels in cancer patients.58 Therefore, it appears that leptin does not play a role in cancer associated anorexia.
IL-1, IL-6, and TNF-α.
Serum IL-1 alpha levels, C-reactive protein IL-6, and TNF-α are increased in cancer patients.58,75–80 The association of serum cytokines with anorexia is controversial. In one study, megestrol acetate reduced anorexia and improved weight and the benefits to appetite were inversely correlated with serum IL-6 levels. In another study, megestrol acetate reduced anorexia independent of serum IL-6 levels.80,81 In another study, anorexia involving untreated cancer patients anorexia did not correlate with circulating cytokines.82 The Prognostic Inflammatory and Nutritional Index, which incorporates C-reactive protein levels, albumin, prealbumin, and alpha-1 acid glycoprotein, is high in advanced cancer patients compared with the normal population but does not correlate with weight loss over time.83 There is no reported association between this index and anorexia.
Serotonin.
Anorectic cancer patients have high fasting serum and cerebrospinal fluid tryptophan levels.84,85 However, no correlation exists between cytokine production (by stimulated peripheral mononuclear cells) and serum tryptophan levels or anorexia.86 Contrary to the reported increased serum tryptophan in cancer anorexia, patients treated with IL-2 develop anorexia with reduced baseline serum tryptophan.87
In summary, conflicting data exist regarding the relationship between serum cytokines, serotonin, and anorexia in cancer. Cytokine levels are increased in cancer but are unassociated with anorexia or weight loss. Serum tryptophan, a precursor to serotonin, is increased in the serum of cancer patients but is unrelated to anorexia. There is no evidence to confirm the central neurohumoral changes associated with anorexia which are based on a hypothesis generated through studies in tumor bearing animals.
Orexigenic Medications and Proposed Sites of Action
Medications to treat cancer associated anorexia include corticosteroids, dronabinol, erythropoietin, melatonin, mirtazapine and progesterone.25,88–93 There are no human data regarding their neurohumoral mechanisms of action in cancer. Speculation can only be based on non–tumor bearing animal models as a theoretical rationale.
Animal Models: Mechanisms of Action
Progesterones.
Progesterones reduce tyrosine hydroxylase activity and dopamine, which are negative modulators of NPY neurotransmission through D2 receptors.94 Progesterone also downregulates neuropeptide Y2 receptors and thus reduces negative NPY feedback inhibition.95
Corticosteroids.
Corticosteroids reduce proinflammatory cytokines and increase mRNA-neuropeptide-Y, preproneuropeptide-Y mRNA and mRNA to neuropeptide Y1 receptor expression. Secondarily, corticosteroids reduce rat corticotropin releasing hormone levels.53,96–99
Cannabinoids.
Cannabinoids theoretically stimulate appetite through hypothalamic cannabinoid-1 receptors.99–101 Gamma aminobutyric acid and dynorphin are increased by cannabinoids, which are secondary appetite stimulants.99–102 Peripherally, CB1 receptors are coexpressed on CCK receptor containing interneurons and oppose CCKA receptor actively in experimental animals.103
Melatonin.
Melatonin activates 5HT1A receptors in the rat hypothalamus.103 Pharmacologic doses reduce hypothalamic serotonin release, uptake and serotonin receptor binding in experimental animals.104–106 There is mutual opposing feedback between the serotonin-melatonin system.107–109 Melatonin also reduces serotonin stimulated release of CRH.107 Exogenous melatonin also increases hypothalamic gamma aminobutyric acid (GABA).106
Appetite Stimulants in Humans
Mirtazapine.
Mirtazapine blocks 5HT2 post synaptic receptors and facilitates transmission through 5HT1a receptors. Increased appetite and weight gain are common clinical side effects with mirtazapine.110 There is little experience with mirtazapine in cancer anorexia.92
Erythropoietin.
Erythropoietin combined with nonsteroidal anti-inflammatory drugs causes weight gain and increased appetite in advanced cancer.93 Erythropoietin increases serum NPY concentrations.93,111,112 Only anecdotal evidence exists for its benefit in cancer anorexia.
In summary, several appetite stimulants may theoretically favor various neuropeptides, neurotransmitters and receptors in the hypothalamus and reduce anorexia in advanced cancer.
Drawbacks to the proposed mechanisms are the lack of supporting clinical evidence and theories, derived mostly from non–tumor-bearing animal studies, that may be substantially different in both tumor bearing animals and humans. Such proposed drug actions may in fact be irrelevant.113 There are few studies of appetite stimulants in animal tumor models. The mechanism by which corticosteroids, progesterones, melatonin, and cannabinoids actually stimulate appetite in cancer patients or tumor-bearing animals is unknown.
DISCUSSION
Clinical use of appetite stimulants is empiric and not derived from translational research. There are major drawbacks to our understanding of appetite and anorexia: (1) Appetite neurophysiology is based on animal data, given that there are no data to corroborate such findings in humans. Many clinicians assume that animal models reflect anorexia in cancer patients. (2) Appetite is governed predominately by CNS signals, which restricts research methodology in humans. (3) There are few appetite receptor–containing eukaryotic cell lines that substitute as research models. (4) There is no transitional connection between orexigenic drug studies and the neurophysiology of appetite and tumor associated anorexia derived from animal models. (5) There is only speculation on how orexigenic drugs actually work. (6) There are inconsistencies between proposed modulators of appetite and clinical experience. For example, dopamine through D2 receptors suppresses NPY neurotransmission in rats; however, L-dopa improves appetite in advanced cancer.48,113 Hypothalamic serotonin is proposed to be derived from proteolysis and subsequent neural transportation of circulating tryptophan, yet in cytokine-treated cancer patients, anorexia is associated with reduced serum tryptophan levels.86,87,114 Ghrelin is proposed to be reduced with anorexia, yet plasma levels are increased in lung cancer patients who suffer from anorexia.114 (7) Anorexia in advanced cancer is governed by multiple secondary causes and responses to treatment can take days to weeks to be fully appreciated. (8) There is a complete absence of understanding regarding the central neurohumoral changes that cause human anorexia. (9) Cachexia and anorexia are frequently confused with one other, as are their outcome measures. Yet each may have a distinctly different pathophysiology. (10) There is no consensus about outcome nor the significance of anorexia and few validated tools to measure response.115–117 (11) The place of appetite stimulants in anorexia has not been established by prospective studies. (12) It is unlikely that treating a single symptom like anorexia will influence global quality of life since anorexia is a symptom that is relatively less distressing to patients than many other cancer symptoms.118 (13) Universal evidence-based guidelines to guide treatment of anorexia in advanced cancer are lacking.
Appetite is subjectively compared with “normal” for a particular individual, yet there are no thresholds to determine at what point anorexia is present nor a critical duration of time to consider it abnormal.119 The symptom distress of appetite loss may in fact be greater with early satiety than the absence of hunger.119 In addition, anorexia frequently causes greater anxiety in family members than in patients.119 As a result, requests to intervene may come more often from family concerns than from patients.
The French National Federation of Cancer Centres working group has reviewed standards, options, and recommendations for appetite stimulants, based on the literature.120 Levels of evidence were graded A, B, C, or D, and “expert agreement” based on the best available studies (Table 5). 120 Recommendations based on the various options were weighted based on clinical evidence.120 The strength of recommendations was directly proportional to the evidence. Based on the working group's literature review, corticosteroids, megestrol acetate, and medroxyprogesterone are recommended with level B evidence.120 Cyproheptadine, dronabinol, metoclopramide, nandrolone, and pentoxyfylline had level C evidence for benefit.120 By expert agreement it was recommended that level C drugs were not to be used outside clinical trials.120 Based on prospective trials and at least one small randomized study, melatonin has level B2 evidence.91,121 Mirazapine, reported in only a small case series, has level D evidence.92 Hydralazine sulfate should not be used (level A evidence).120
Definitions of Levels of Evidence
Appetite stimulants were recommended for incurable disease (level C evidence) and for any tumor type (expert agreement). Appetite stimulants may be given with nutritional supplements (expert agreement). The optimal mode of administration and appropriate time to start appetite stimulants is unknown.120
In conclusion, appetite signals have been well studied in rodent models, but little is known about human appetite neurophysiology. Proposed mechanisms for tumor-induced anorexia include quality and quantitative alterations in NPY and POMC signals. These mechanisms, based on animal models, have yet to be validated in humans. Conflicting evidence exists in proposed pathophysiology of anorexia and clinical findings. Although several appetite stimulants are available, most have not been studied in formal phase I, II, or III trials. Only a few appetite stimulants have been prospectively studied in clinical trials, and thus evidence-based guidelines cannot be formulated. As a result, there is much that needs to be systematically researched for this common symptom in advanced cancer.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgments
The authors acknowledge the help of Michele Wells and Deborah Davis in preparing the manuscript.
Footnotes
-
A WHO project in palliative medicine.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
- Received March 17, 2003.
- Accepted January 21, 2004.
