Stress and cancer: mechanisms, significance and future directions

Abstract

The notion that stress and cancer are interlinked has dominated lay discourse for decades. More recent animal studies indicate that stress can substantially facilitate cancer progression through modulating most hallmarks of cancer, and molecular and systemic mechanisms mediating these effects have been elucidated. However, available clinical evidence for such deleterious effects is inconsistent, as epidemiological and stress-reducing clinical interventions have yielded mixed effects on cancer mortality. In this Review, we describe and discuss specific mediating mechanisms identified by preclinical research, and parallel clinical findings. We explain the discrepancy between preclinical and clinical outcomes, through pointing to experimental strengths leveraged by animal studies and through discussing methodological and conceptual obstacles that prevent clinical studies from reflecting the impacts of stress. We suggest approaches to circumvent such obstacles, based on targeting critical phases of cancer progression that are more likely to be stress-sensitive; pharmacologically limiting adrenergic–inflammatory responses triggered by medical procedures; and focusing on more vulnerable populations, employing personalized pharmacological and psychosocial approaches. Recent clinical trials support our hypothesis that psychological and/or pharmacological inhibition of excess adrenergic and/or inflammatory stress signalling, especially alongside cancer treatments, could save lives.

Full text

0123456789();:
For decades, stress has been suggested to affect can-
cer incidence and cancer progression1,2. However,
both epidemiological studies and clinical trials have
yielded mixed results, or indicated small or clinically
insignificant effects of stress on cancer progression.
Consequently, current medical routines do not include
measures to prevent stress responses as a means to
improve cancer survival. Within the medical commu-
nity, this may reflect a disbelief that stress is a signifi-
cant biological factor underlying cancer aetiology and
progression.
By contrast, in recent years, animal studies have pro-
vided solid evidence that stress can facilitate growth and
metastasis of many types of cancer. Most importantly,
numerous endocrine, cellular and molecular mecha-
nisms underlying these effects have been identified. For
example, animal models have shown that stress factors
can promote most established hallmarks of cancer2,
and that stress responses can facilitate cancer growth and
metastasis via directly affecting molecular characteris-
tics of the malignant tissue3–5, its microenvironment6,
antitumour immune activity4,7–9 and other indirect
modulators of cancer progression10,11. In patients with
cancer, stress has been shown to activate many of these
processes8,10–13, supporting the clinical significance of
these findings.
We suggest that the discrepancy between preclinical
studies and clinical or epidemiological studies stems
from two sources. First, preclinical studies can syn-
chronize stress or stress- reducing interventions with
critical periods along cancer progression that are highly
susceptible to the impacts of stress. Second, conceptual
and methodological difficulties in conducting clini-
cal studies may obscure the impact of stress on cancer
progression.
In this Review, we describe and discuss stress and
stress responses at the organism level and in the context
of cancer. We further explain mechanisms via which
stress can facilitate cancer initiation, impair cancer treat-
ments and promote cancer growth and metastasis, based
on animal studies and on parallel human correlative or
causative studies. We also review epidemiological studies
and clinical trials in patients with cancer, and discuss
why we believe many of these studies are predisposed to
show minor or no effects, and then suggest approaches
that we hypothesize will provide more conclusive evi-
dence on whether stress significantly affects long- term
cancer outcomes in humans.
Stress and cancer: mechanisms,
significance and future directions
AnabelEckerling , ItayRicon- Becker , LiatSorski , EladSandbank
and ShamgarBen- Eliyahu
✉
Abstract | The notion that stress and cancer are interlinked has dominated lay discourse for
decades. More recent animal studies indicate that stress can substantially facilitate cancer
progression through modulating most hallmarks of cancer, and molecular and systemic
mechanisms mediating these effects have been elucidated. However, available clinical evidence
for such deleterious effects is inconsistent, as epidemiological and stress- reducing clinical
interventions have yielded mixed effects on cancer mortality. In this Review, we describe and
discuss specific mediating mechanisms identified by preclinical research, and parallel clinical
findings. We explain the discrepancy between preclinical and clinical outcomes, through pointing
to experimental strengths leveraged by animal studies and through discussing methodological
and conceptual obstacles that prevent clinical studies from reflecting the impacts of stress. We
suggest approaches to circumvent such obstacles, based on targeting critical phases of cancer
progression that are more likely to be stress- sensitive; pharmacologically limiting adrenergic–
inflammatory responses triggered by medical procedures; and focusing on more vulnerable
populations, employing personalized pharmacological and psychosocial approaches. Recent
clinical trials support our hypothesis that psychological and/or pharmacological inhibition of
excess adrenergic and/or inflammatory stress signalling, especially alongside cancer treatments,
could save lives.
Sagol School of Neuroscience
and School of Psychological
Sciences, Tel Aviv University,
Tel Aviv, Israel.
✉e- mail:
shamgar@tauex.tau.ac.il
https://doi.org/10.1038/
s41568-021-00395-5
NATure revIewS
|
CAnCER
Reviews
VOLUME 21
|
DECEMBER 2021
|
767

0123456789();:
Stress and stress responses
Hans Selye in 1956 (REF.14) described stress as a response
of the body to the demands made upon it in an attempt
to return to homeostasis. Meeting the demands of life,
spanning from day to day tasks to major threats such
as the diagnosis and treatment of cancer, requires
mobilization of metabolic energy to sustain necessary
physiological adaptive responses. This is achieved by
activation of the sympathetic nervous system (SNS) and
the hypothalamic–pituitary–adrenal (HPA) axis, leading to
local and systemic secretion of adrenergic factors from
sympathetic nerve endings and the adrenal medulla
(mostly noradrenaline (also known as norepinephrine)
and adrenaline (also known as epinephrine), respec-
tively), the release of glucocorticoids (such as cortisol)
from the adrenal cortex, and the secretion of opioids,
oxytocin and other stress mediators (FIG.1).
The stress responses described above are initiated by
the central nervous system (CNS) following processing
of various stimuli, including physiological inner- body
responses to various conditions, such as tissue damage
(including during surgery and under anaesthesia), or
being subjected to low temperature; external stimuli,
such as being attacked by an assailant with a weapon
or being informed of having cancer; or ongoing CNS
activities, resulting from being anxious or ruminating
about financial insecurity, social isolation, interpersonal
relationships or having cancer (FIG.1). Notably, both
depression and social isolation involve activation and/or
dysregulation of the HPA axis, and are characterized by a
pro- inflammatory state15,16, which triggers similar path-
ways to stress responses (discussed below). Last, stress
and depression promote each other17, and most animal
models of depression are based on stress exposure18.
Stress can be both beneficial and deleterious (BOX1).
The effects of stress on the capacity of an organism to
cope with challenges typically follow an inverted U
shape19 — when the intensity, duration or nature of the
stressor is moderate, stress facilitates adaptive natural
changes, but when stress exceeds the resources of the
individual to cope, and becomes ‘toxic stress’, the risk for
disease increases20. McEwen and Stellar defined allosta-
sis as the naturally occurring continuous adaptations
towards different homeostatic states21. When allostasis
becomes strenuous, and the allostatic load increases to
the point of overload, patients are at greater risk21.
Notably, the intensity and duration of stress responses
to internal or external stimuli markedly differ between
individuals, and depend on physiological factors, includ-
ing genetic and developmental variations19, and physical
fitness (BOX2); individual psychosocial characteristics,
including perceived social support22, perceived ability to
cope23 and other personal traits; and the characteristics
of the stressful life events previously experienced24–26,
including childhood adversity27. It follows that stressors
such as cancer diagnosis, treatment and survivorship are
likely to be differentially experienced by patients, provok-
ing different stress responses. Thus, stress- management
therapies, behavioural or pharmacological, should
be individually tailored. Additionally, understanding
specific physiological mechanisms mediating dele-
terious (or beneficial) effects of stress responses may
point to effective downstream pharmacological thera-
peutic approaches, which may also surpass individual
differences at higher psychological/cognitive levels.
Critical periods in cancer progression
Normal cells transform into malignant cells through
acquisition of unique characteristics with evolution-
ary advantages, known as the ‘hallmarks’ of cancer28,29.
These characteristics include resistance to apoptotic
signals, independence from external growth signals, the
capacity to attract vascularization, evasion of immune
destruction and the acquisition of invasive properties
into distant organs with a permissive microenvironment
to form metastases. Importantly, along this transforma-
tion, pre- malignant or malignant foci may be eliminated,
may become dormant or slowly progressing30, or may
advance to a clinical manifestation.
Theoretically, some phases may be more critical
along this multimodal non- linear process. Examples
include activation of the ‘angiogenic switch’ that ena-
bles increased growth or escape from dormancy31;
initial interactions with immune cells following neo-
vascularization and/or release of damage- associated
molecular patterns32; the passage of circulating tumour
cells through pulmonary or hepatic capillaries, where
highly active marginating natural killer cells recognize
and eliminate such aberrant cells33–35; survival of circu-
lating tumour cells in the circulation and extravasation
into new organs36; and the capacity of a micrometastasis
to grow independently of the primary tumour37.
Stress may have greater impact during such potential
critical phases. Moreover, whether stress will exacerbate
or mitigate malignant processes may depend on the
phase of malignant progression, specific tumour char-
acteristics and the spectrum of stress responses. Also,
immune system–tumour interactions may either impair
or promote tumour growth38, and stress hormones can
regulate both processes7,9. Thus, interactions between
stress and cancer are expected to be non- linear, and the
impact of stress could depend on the phase of cancer
progression.
Hypothetically, an acute or chronic stress episode that
is synchronized with a critical phase may bear a greater
impact on cancer progression than non- synchronized
episodes. Studies in animal models, more than clinical
or epidemiological studies, can focus on a critical phase,
employing specific tumour types, and/or stress para-
digms, and thus maximize our ability to observe the
potential impact of stress. For example, stressing animals
shortly before and after intravenous tumour cell inocu-
lation maximizes the deleterious impact of stress on the
capacity of marginating pulmonary natural killer cells to
prevent experimental lung metastasis33,39,40. In breast can-
cer mouse models, chronic stressors did not affect growth
of primary tumours but did promote their dissemina-
tion and metastatic growth41,42. Last, subjecting mice to
chronic social isolation before mammary tumour inoc-
ulation exerted no effects on primary tumour growth,
whereas if initiated when tumours were palpable, primary
tumour growth was increased43.
In the clinical setting, some critical phases cannot be
recognized but others, especially those related to cancer
Sympathetic nervous
system
(SNS). Part of the autonomic
nervous system that is
involuntarily activated by
stressors (for example,
a dangerous or stressful
situation) and orchestrates the
‘fight or flight’ response through
adrenergic innervation of the
adrenal medulla and of various
organs (for example, the heart)
through systemic and local
release of adrenaline and
noradrenaline, respectively.
Hypothalamic–pituitary–
adrenal (HPA) axis
A neuroendocrine system
with negative feedback
that increases systemic
glucocorticoid (for example,
cortisol) levels in various
circumstances, including
stressful conditions.
Hypothalamic corticotropin-
releasing hormone (CRH)
elevates systemic release of
adrenocorticotropic hormone
(ACTH) from the anterior
pituitary, which triggers the
release of glucocorticoids from
the adrenal cortex, which also
trigger negative feedback
through the pituitary and
hypothalamic levels.
Damage- associated
molecular patterns
Endogenous host- derived
molecules that are released
by damaged and dying cells.
They are recognized by pattern
recognition receptors on
numerous cells, which lead to
migration and activation of
various immune cells and
consequent innate and
adaptive immune responses.
www.nature.com/nrc
Reviews
768
|
DECEMBER 2021
|
VOLUME 21

0123456789();:
treatment, are known to impact cancer progression
(BOX3), and can perhaps be exploited to mitigate the
effects of stress on cancer progression.
Mechanisms of stress impacts on cancer
As briefly reviewed below, a vast body of literature indi-
cates that stress can promote most hallmarks of cancer2,
and mechanisms mediating these effects by specific
stress hormones, their receptor systems and intracellular
molecular mechanisms have been identified (reviewed
in REFS3,4,6,44). We herein discuss and refer to tumour
initiation as transformation from non- malignant to
malignant tissue, in contrast to tumour progression
that follows this transformation, although most hall-
marks of cancer can affect both initiation and progres-
sion of the disease. We present causative findings from
Ongoing CNS-derived processes
Anxiety and rumination regarding:
• interpersonal relationships
• financial hardships
• having cancer
Stimuli external to the CNS
• Work stress
• Attack by an assailant
• Cancer diagnosis
• Malignant tissue
Physiological and psychological stressors
IL-6 and
IL-1β
Tumour
Adrenal gland
Systemic level
GlucocorticoidsAdrenaline Noradrenaline
• Endorphins
• Oxytocin
• Prolactin
• Vasopressin
Systemic
ACTH
CRH
• Surgery
• Tissue damage and inflammation
• Low body temperature
• Social isolation
CNS
Hypothalamus
Pituitary gland
Sympathetic nerve endings
Innervated lymphoid organs
Spleen Bone
marrow
Lymph
node
Stress
responses
Fig. 1 | Stress responses and reciprocal stress–cancer interactions. Physiological and psychological stressors including
stimuli external to the central nervous system (CNS), such as being informed of having cancer, undergoing surgery or the
presence of malignant tissue and its related inflammation, and ongoing CNS- derived processes (for example, anxiety
and rumination about cancer) are perceived and processed by the CNS and trigger stress responses. Consequently, the
pituitary gland releases endorphins, oxytocin, prolactin, vasopressin, adrenocorticotropic hormone (ACTH) and other stress
mediators, and activation of the hypothalamic–pituitary–adrenal (HPA) axis through hypothalamic corticotropin- releasing
hormone (CRH) and systemic ACTH release leads to secretion of glucocorticoids (for example, cortisol) from the adrenal
cortex. Simultaneously, the CNS activates the sympathetic nervous system (SNS), leading to secretion of adrenergic factors
from the adrenal medulla (mostly adrenaline) and sympathetic nerve endings (mainly noradrenaline). The latter also
innervate lymphoid organs (for example, spleen and lymph nodes), bone marrow and various organs. These stress factors
promote most hallmarks of cancer through impacting the malignant tissue, its microenvironment, immunity, lymphatic
flow and distant potential pre- metastatic niches (FIG.2). Malignant tissue can facilitate stress responses through local
and systemic inflammation (for example, through interleukin-6 (IL-6) and IL-1β) that affects the CNS, dysregulates HPA
axis activity220,221 and promotes depression, sleep disturbances and cancer- related fatigue. Overall, CNS- initiated stress
responses may lead to exacerbated tumour growth and spread, and to peripheral stress–inflammatory–cytokine responses,
which feed back to the CNS, altering cognition and mood, and facilitating stress responses, creating a vicious cycle.
NATure revIewS
|
CAnCER
Reviews
VOLUME 21
|
DECEMBER 2021
|
769

0123456789();:
animal studies, which often are followed by parallel
clinical findings, complementing each other in terms of
methodological robustness and clinical relevance.
Cancer initiation
DNA damage. Specific stress factors have been shown to
cause DNA damage and jeopardize DNA repair, poten-
tially facilitating malignant transformation. Specifically,
in a mouse fibroblast cell line, serum derived from
stressed mice, or adrenaline, noradrenaline and cor-
tisol (each factor alone as well as synergistically when
combined), increased DNA damage and/or reduced
DNA repair following UV irradiation45. In murine and
human non- cancer cell lines, β- adrenergic receptor
(β- AR)- mediated generation of reactive oxygen species
and β- arrestin–MDM2- dependent p53 degradation
increased DNA damage and inhibited DNA repair46.
Corresponding invivo studies confirmed that chronic
stress induces these two β- AR- mediated processes47,
and that glucocorticoid- mediated response can also
cause MDM2- dependent p53 downregulation and
increase resistance to apoptosis following ionization
irradiation48. In humans, several studies indicated that
psychological stress is associated or causatively linked
to increased DNA damage49, and several human cancer
cell lines exhibited accelerated DNA damage invitro fol-
lowing β- adrenergic and glucocorticoid signalling50–52,
in part through activation of the ATR–p21 pathway52.
Nevertheless, it should be noted that DNA damage alone
is insufficient to cause tumour initiation, as mutations
need to be maintained and accumulated across repeated
cell divisions, and should lead to acquisition of resistance
to apoptosis and to increased proliferation, among other
characteristics.
Oncogenic viruses. Thirteen to 15% of human cancer
incidence is attributed to carcinogenic infections53,54,
and stress can also increase the risk for cancer initia-
tion by promoting the prevalence and outbreak of
oncogenic viruses. Following invitro infection of var-
ious human cell lines, major oncogenic human viruses
were shown to be reactivated by either glucocorticoids
or catecholamines, including human papillomaviruses
(HPVs), Epstein–Barr virus, Kaposi sarcoma- associated
herpesvirus and hepatitis B and C viruses55. Additionally,
stress hormones were shown to stimulate oncogene
expression in human cells infected with oncogenic
viruses, as well as to suppress expression of type I inter-
ferons (IFNα and IFNβ) in leukocytes, impairing anti-
viral immunity55–57. In humans, academic examination
stress in cadets, and/or activation of the SNS or HPA
axis, was associated with reactivation of latent onco-
genic viruses58,59; higher levels of perceived stress were
associated with impaired HPV- specific Tcell responses
in women with cervical dysplasia60; and loss of a child
predicted increased risk for HPV- associated cancers in
a cohort of more than four million parents in Sweden61.
Tumorigenesis. Several invivo animal studies assessed
the effects of stress on actual tumorigenesis, rather than
on interim indices, such as DNA damage or reactiva-
tion of oncogenic viruses. Repeated restraint stress48,62,
social isolation63 and cold ambient temperature64 pro-
moted carcinogen- induced tumorigenesis. In transgenic
models of spontaneous cancer, repeated restraint stress
increased pancreatic tumorigenesis through β2- AR
signalling65, whereas sympathetic denervation decreased
tumorigenesis in a prostate cancer model66. However, in
such models that are based on accelerated induction of
cancer, it is hard to distinguish between effects of stress
on tumour initiation and its effect on tumour progres-
sion, as the time course of stress largely overlaps with
both initiation and progression periods48,65,67,68. Thus,
stress can potentially exacerbate the effects of carcino-
genic exposure, yet it is unclear whether stress is a signif-
icant factor in tumour initiation in the absence of known
exposure to carcinogens.
Cancer progression
Direct effects on tumour cells. Stress hormones,secreted
systemically or released locally in the tumour micro-
environment from sympathetic nerve endings, immune
cells69,70 or tumour cells71–73, can directly affect tumour
cells, promoting their malignant characteristics.
Specifically, noradrenaline and adrenaline were shown
invitro to promote tumour cell proliferation74–76, survival
(anti- apoptosis)74,75,77, migration74,78,79, invasion74,78–81,
epithelial–mesenchymal transition (EMT)42,78,82,83 and
production of prostaglandins76,79 and matrix metallo-
proteinases (MMPs)76,80,81 (FIG.2). Accordingly, behav-
ioural or physiological stressors (for example, social
Catecholamines
A family of molecules that are
characterized by a catechol
and an amine group in their
chemical structure, and
function as neurotransmitters
and hormones within the body.
These include dopamine,
noradrenaline and adrenaline,
all of which are synthesized
from the amino acid tyrosine.
Restraint stress
An experimental stress
paradigm, where the animal
is placed in a confined space
(a tube- shaped apparatus
perforated for air exchange)
that prevents free movement
but does not press or induce
pain to the animal. Such
restraint can last minutes to
hours and can be repeated
daily for several weeks as a
chronic stress paradigm.
Sympathetic denervation
Refers to experimental
methods for ablation of
sympathetic nerves (also called
sympathectomy), by either
surgical cut of sympathetic
nerve fibres or chemical
ablation (for example, using
6- hydroxydopamine).
Box 1 | Acute and chronic stress
Acute stress is defined as lasting minutes to hours, whereas chronic stress can last days,
weeks, months or longer271. A short- term transient stress response can be adaptive,
as part of the ‘fight or flight’ response, where sympathetic nervous system (SNS) and
hypothalamic–pituitary–adrenal (HPA) axis activation increases the heart rate, blood
pressure and glucose availability. Such stress responses can also promote the release of
pro- inflammatory cytokines (for example, interleukin-6 (IL-6) and IL-1β) and trafficking
of leukocytes to the skin following stress cessation272,273, potentially to allow skin pathogen
resistance in the case of injury271. By contrast, long- lasting or repeated stress exposures
can lead to HPA axis dysregulation, glucocorticoid resistance and/or insensitivity to HPA
axis negative feedback274. These may lead to chronic inflammation secondary to disrupted
HPA axis- induced inhibition of pro- inflammatory responses274. Nevertheless, chronic
elevated levels of glucocorticoids contribute to immunosuppression275. Moreover, animal
studies have demonstrated that both acute and chronic stress paradigms can suppress
immunity40,273 and promote certain anti- inflammatory responses, such as decreased
plasma IL-12 levels276.
Notably, the distinction between acute and chronic stress is often ambiguous. Chronic
stress paradigms in animals are often based on repeated41 or alternating85 acute stressors,
rather than continuous stressors. Furthermore, there is no unified definition of acute or
chronic stress62,85,92,135, with 3–5 consecutive days of repeated acute stressors defined
both as acute92 and as chronic85 stressors. Also, continuous chronic social isolation
was found to increase reactivity to acute restraint stress67,137, demonstrating mutual
interdependence between acute and chronic stress. In humans, acute events can
generate a chronic threat perception and/or chronic stress responses277, especially given
pre- event anticipation and post- event ruminations27. In the context of cancer treatment,
the overlapping nature of acute and chronic medical and psychological stressors,
and the psychological consequences of these events, may mask the distinction between
acute and chronic stress and their impact on cancer progression. Moreover, some
naturally adaptive responses to acute stress, such as redistribution of leukocytes to the
skin at the expense of internal organs, may increase the risk for internal organ metastasis,
as indicated by animal studies employing acute stressors40,133. Thus, the intricacies of
acute and chronic stress responses in the context of cancer progression and treatment
suggest caution in making any generalizations.
www.nature.com/nrc
Reviews
770
|
DECEMBER 2021
|
VOLUME 21

0123456789();:
confrontation, restraint and surgery) in animal models
were shown to increase tumour growth and metastasis
through activation of tumour β- AR, as indicated by
their specific pharmacological41,74,80,84–86 or molecular87,88
blockade, or by genetic knockout84.
Recent studies have indicated the contribution of
tumour innervation to tumour progression89. Tumours
can secrete neuronal growth factors, increasing sym-
pathetic tumour innervation. This creates a feedfor-
ward loop that promotes cancer progression under
stress- induced sympathetic activation, as a result of
higher tumoural noradrenaline levels65. Correspondingly,
numerous human cancers were found to express
β- AR65,74,75,78,79,81,83,90, and their higher expression74,75,78,79,83
or higher tumour noradrenaline91 and/or plasma
adrenaline82 levels were correlated with larger tumour
size, advanced stage, lymph node metastasis and/or
reduced survival in several cancer types. Interestingly,
low social support in patients with ovarian cancer
predicted higher tumour levels of noradrenaline91.
Behavioural stress can also promote tumour growth
through glucocorticoid secretion48,92, and synthetic glu-
cocorticoid receptor (GR) agonists (for example, dexa-
methasone) promoted metastasis and reduced survival
in xenograft and syngeneic breast cancer models93. In
patients with breast cancer, higher tumour expression
levels of GR and GR- regulated kinases predicted poorer
survival93,94.
Angiogenesis and lymphangiogenesis. Invitro findings
indicated that noradrenaline and adrenaline increase
tumour cells’ expression and secretion of several angi-
ogenic factors, including vascular endothelial growth
factor (VEGF), interleukin-6 (IL-6) and IL-8 (REFS81,95–97),
and that noradrenaline- mediated angiogenesis is rein-
forced following direct contact between tumour cells
and endothelial cells98. In stressed nude mice orthotop-
ically implanted with human ovarian carcinoma cells,
β2- AR–cyclic AMP (cAMP)–protein kinase A (PKA)
signalling increased tumour expression of VEGF, and
tumour vascularization and growth87. Similar findings
were confirmed in pancreatic cancer99, colorectal cancer
(CRC)100 and breast cancer41,101 models. Stress- induced
β- AR signalling also inhibited the anti- angiogenic fac-
tor thrombospondin 1 (TSP1) in prostate cancer xeno-
grafts through epigenetic modulation102. In patients with
ovarian carcinoma, lower social well- being and elevated
distress or depressive symptoms correlated with higher
plasma and tumour VEGF levels103,104, and higher ascites
and plasma IL-6 levels105,106.
Tumour lymphatic vessel density and lymphangi-
ogenic growth factors are associated with metastases
and with reduced survival in patients with cancer107.
Chronic restraint stress in mice, through β- AR signal-
ling, increased expression of the lymphangiogenic fac-
tor VEGFC in tumour and stromal cells, and increased
expression of cyclooxygenase 2 (COX2; also known as
PTGS2) in tumour- associated macrophages (TAMs).
These changes led to elevated lymphatic vessel density
and increased metastasis108. In patients with cancer,
acute blockade of SNS activity reduced lymph flow in
patients with cervical carcinoma108, and breast tumours
in socially isolated women exhibited increased density
of lymphatic vessels109.
Immunomodulation and inflammation. Stress has
been shown to promote both inflammation and
immune evasion8. Most immune cells express β- ARs110,
prostaglandin receptors111 and GRs44, and the effects of
stress on their activity and distribution have been exten-
sively studied in animal models and in patients with
cancer7–9,110.
In murine models, natural killer cell activity
against tumour cells was suppressed by stress- induced
β- adrenergic signalling or β- adrenergic agonists33,40,112,113,
and a stress- induced increase in lung metastases was
shown to be mediated by suppression of natural killer
cells114. In patients with ovarian cancer, lower social sup-
port and higher distress correlated with lower natural
killer cytotoxicity115. Stress was also shown to induce a
shift from T helper 1 cell (TH1 cell)- type to T helper 2 cell
(TH2 cell)- type cytokine production, to increase
tumour growth in mouse models of CRC116 and squa-
mous cell carcinoma62, as well as to increase tumour
growth through β- AR- mediated suppression of CD8+
Tcells in mammary and melanoma mouse models84.
Correspondingly, in patients with ovarian carcinoma,
depressed and anxious mood correlated with a reduced
TH1 cell/TH2 cell- type cytokine ratio117. Additionally, in
mouse models, a stress- induced β- adrenergic response
promoted tumour growth by upregulation of suppressive
Prostaglandin receptors
A class of cell surface
G- protein- coupled
receptors that bind different
prostaglandins and are
expressed on various cell
types, including immune cells;
for example, prostaglandin E2
binds to the prostaglandin E2
receptor 1–4 subtypes.
T helper 1 cell
(TH1 cell). A CD4+ Tcell
that participates in the
pro- inflammatory type 1 or
cellular immune response
against intracellular pathogens
and malignant cells. Naive
Tcells are differentiated into
the type 1 phenotype following
exposure to interleukin-12
(IL-12), and are known for the
secretion of interferon- γ (IFNγ),
which is also involved in the
effector functions of cytotoxic
Tcells.
Box 2 | Physical exercise, stress and cancer
Physical exercise exerts a challenge to whole- body homeostasis, promoting extensive
adaptations of cells, tissues and organs278. Moderate physical exercise is known to
improve cardiometabolic indices, to increase cognitive performance and to improve
numerous health conditions and support their treatment, including cancer279. Physical
exercise increases the levels of stress hormones (for example, adrenaline, endorphins and
cortisol) for the duration of the exercise, blunts hormone responses to stress280,281
and modulates inflammatory status and cytokine levels during exercise (for example,
increased interleukin-6 (IL-6), IL-10 and IL1Rα, but not TNF and IL-1β)282. In the context of
cancer, physical exercise was shown to have beneficial impact on quality of life, fatigue,
anxiety, depressive symptomatology and psychological distress283–287. The effect of
exercise on inflammation is complex. In the general population, physical exercise is
generally associated with reduced inflammation282, whereas in patients with cancer this
association is more limited288. Importantly, prospective correlational studies indicated
that physically active patients have significantly lower mortality rates than non- active
patients289,290. Interestingly, whereas stress responses exert numerous pro- tumorigenic
effects (as reviewed herein), physical exercise- induced stress factors exhibit
antitumorigenic properties291. For example, in preclinical studies, exercise- conditioned
serum, derived from healthy humans and patients with cancer, had growth- inhibitory
effects on breast cancer cell lines invitro and invivo292. Moreover, mice s ubjected to
voluntary physical exercise had attenuated tumour growth and enhanced antitumour
activity via β- adrenergic signalling292–294. Hypothesized explanations for the apparent
contradictory beneficial and deleterious effects of β- adrenergic signalling include the
rapid and transient increase and decrease of adrenergic responses to exercise; inh ibited
stress responses following physical exercise; and the rapid exercise- related mobilization
of cytotoxic immunocytes (for example, CD8+ Tcells, natural killer cells)295 to the
circulation, as opposed to stressors and their aftermath that induce immunosuppression.
Additionally, physical exercise was shown to exert the production of di hydroxyphenylala-
nine (DOPA) and dopamine (as part of the catecholamine response)296 that were reported
to antagonize tumour progression10, whereas these responses are generally not induced
by stressors. Overall, these results warrant further studying of the mechanisms by which
physical exercise improves psychological indices, physical adaptation to stress and
malignant conditions, and devising suitable exercise regimens for patients with cancer to
potentially improve short- term and long- term outcomes.
NATure revIewS
|
CAnCER
Reviews
VOLUME 21
|
DECEMBER 2021
|
771

0123456789();:
immune cells such as myeloid- derived suppressor cells
(MDSCs) and regulatory Tcells62,84,100,118, whereas in
pa tients with breast cancer, higher levels of stress corr elated
with reduced numbers of circulating MDSCs119.
With respect to inflammation, stress- induced
β- adrenergic signalling in preclinical studies was shown to
promote COX2 expression and prostaglandin secretion in
both tumour cells and TAMs41,79,108, to stimulate secretion
of pro- inflammatory cytokines (for example, IL-6)95,97 and
to increase tumour recruitment of macrophages and their
M2 polarization41,90,120,121. Correspondingly, in patients
with cancer, social isolation correlated with upregulation
of M2 polarization in breast tumours109, higher levels of
depression were associated with higher levels of prosta-
glandins in ovarian tumours79, and tumour expression
levels of genes encoding β2- AR and prostaglandins
predicted reduced survival79.
Metastasis. Metastases are promoted by many of the
aforementioned mechanisms, as well as by additional
stress- induced processes. For example, in mice, stress-
induced β- AR activation promoted migration of cir-
culating tumour cells to the bones, through increased
expression of receptor activator of nuclear factor- κB
ligand (RANKL) by bone marrow stem cells (BMSCs)122,
or to the lungs by CC- chemokine ligand 2 (CCL2)–CC-
chemokine receptor 2 (CCR2)- mediated attraction of
macrophages85, consequently forming pre- metastatic
niches and increasing organ- specific metastasis. Add-
itionally, stress increased tumour cell EMT42,82,83, tum our
and stromal cell secretion of MMPs41,74,80,99 and tum-
our cell resistance to anoikis77, promoting malignant
cell detachment, invasiveness and survival in the
circulation123. In patients with breast and ovarian cancer,
perceived stress, depressive symptoms or social isolation
predicted higher tumour expression of EMT- related
genes109,124, and higher MMP9 levels in tumour cells and/or
TAMs104. Importantly, β- AR blockade reduced stress-
induced metastasis in many murine models, of both
experimental and spontaneous metastases33,41,74,85,108,122,125.
Correspondingly, in patients with gastric and lung can-
cer, tumour β- AR expression levels correlated with lymph
node metastasis74,126, and incidental use of β- blockers was
associated with decreased metastasis or recurrence in
patients with breast and ovarian cancer108,127,128 and with
improved survival in melanoma and breast cancer129,130,
but not in lung and ovarian cancer131,132. These diverse
outcomes are expected given differences between the
indices studied (for example, metastasis versus survival),
diverse cancer types and the uncontrolled settings of cor-
relational studies, and call for randomized controlled tri-
als (RCTs) to test the effects of β- blockers on long- term
cancer outcomes.
Acute and chronic stressors. Although most animal
studies report that stress, whether acute or chronic,
promotes primary tumour growth and metastasis, a
few studies report that stress can decrease primary
tumour growth. For example, several paradigms of
acute stress were reported to increase primary tumour
growth and metastasis in rodents, including restraint
stress92, 16- h tilt–light stress133, 30–60 min of inter-
mittent swim stress39,40,113, laparotomy100,114,133 or 7 h of
social confrontation stress134, whereas other studies showed
that acute restraint stress135 and foot shock stress136 can
inhibit primary tumour growth. Heterogeneity of the
acute stressors, tumour models, animal species and
phase of tumour progression during stress exposure
may underlie this apparent inconsistency (as discussed
above). With respect to chronic stress, and examining
a more standardized setting of chronic social isolation
in breast cancer models, stress exposure increased67,137,
decreased68,138 or had transient43 effects on primary
tumour growth. Although there are physiological dif-
ferences between acute and chronic stress (BOX 1),
comparison between acute and chronic restraint stress
showed that whereas the stress duration had differen-
tial effects on spleen T lymphocytes, neither acute nor
chronic stress affected the growth of primary mammary
tumours but both increased blood vessel density in
meta static foci101. Additionally, chronic social isolation,
but not chronic restraint, reduced survival of mammary
tumour- bearing mice101. As elaborated in BOX1, there
is ambiguity regarding the definitions of acute and
chronic stressors, and some adaptive characteristics of
Box 3 | Critical time frames during cancer treatment
Along the course of cancer treatment, there are recognized critical phases where
susceptibility to the impacts of stress may be heightened. These include the surgical
removal of the primary tumour, and neoadjuvant and adjuvant therapies. Specifically,
during the short perioperative period (days before and after surgery), surgical excision
of the malignant mass may increase shedding of tumour cells to the circulation297,298,
terminate primary tumour- related secretion of anti- angiogenic factors299,300 and induce
the release of growth factors301,302. These processes cumulatively or synergistically
increase the risk of metastatic disease198,199. Moreover, stress and inflammatory responses
are elevated as a result of psychological distress, tissue damage, hypothermia, blood
transfusions, pain and specific analgesic/anaesthetic approaches198,199. These neuro-
endocrine responses, especially catecholamine and prostaglandin signalling, suppress
antitumour immunity9,303, and directly facilitate progression of residual disease, as
elaborated in the main text. Most importantly, as the short perioperative period holds a
delicate balance between pro- metastatic and anti- metastatic processes, stress responses
during this time can tilt the balance towards the pro- metastatic direction, creating a
‘snowball effect’ that impacts long- term cancer outcomes186. Indeed, several clinical
perioperative events (for example, anastomosis leak and/or secondary surgery) or
specific medical routines (for example, use of the sedative dexmedetomidine) were
associated with worse long- term cancer outcomes304, and animal studies provided
causative evidence that such events can increase the deleterious impacts of stress on
cancer metastasis305. Additionally, a recent study in rodents reported that the effects
of pre- surgical behavioural stress exacerbate the deleterious effects of surgery on
lung metastasis133.
The peri- adjuvant time frame also constitutes a critical period of cancer progression.
Adjuvant therapies and their side effects are accompanied by psychological distress306,
induce inflammatory responses307 and can promote tumorigenic and metastatic
processes308. For example, the chemotherapies cisplatin and paclitaxel activate the
pro- inflammatory nuclear factor- κB (NF- κB) pathway, inducing the expression of
various pro- tumorigenic and pro- metastatic factors such as interleukin-6 (IL-6) and
IL-8, and promoting angiogenesis and tumour cell proliferation, survival and epithelial–
mesenchymal transition (EMT)307. Adjuvant therapies can lead to selection of drug-
resistant tumour clones, and to host- derived responses that promote cancer
recurrence308. Thus, as adjuvant therapies have both pro- tumour and antitumour
effects, and as stress during cancer therapy can impair its efficacy (as discussed in
the main text), stress may have greater impact during the peri- adjuvant time frame.
As the short perioperative and the peri- adjuvant time frames are characterized by
excessive stress and inflammatory responses and by accelerated tumour progression,
they could be exploited therapeutically for anti- metastatic approaches, and specifically
interventions that reduce stress and inflammation.
T helper 2 cell
(TH2 cell). A CD4+ Tcell that
participates in type 2 or
humoral immune response
against extracellular pathogens
(for example, helminths) and
allergens. Naive Tcells are
differentiated into a type 2
phenotype following exposure
to interleukin-4 (IL-4), and are
known for the secretion of IL-4,
IL-13 and IL-5, and promotion
of the production of antibodies.
β- Blockers
A class of drugs with
antagonistic activity towards
β- adrenergic receptors (β- ARs).
The drugs vary in specificity
to the different β- ARs (β1- AR,
β2- AR and β3- AR) and are
classified as selective or
non- selective to a certain
receptor subtype.
www.nature.com/nrc
Reviews
772
|
DECEMBER 2021
|
VOLUME 21

0123456789();:
acute stress responses in the natural setting may promote
cancer progression. Importantly, no generalization can
be drawn regarding stress chronicity and cancer progres-
sion, and other aspects of stress–cancer interactions may
be more critical. Overall, the majority of animal studies
report that stress promotes primary tumour progression,
rather than inhibits it. The impact of stress on metasta-
sis seems even more consistent, with the great majority
of studies reporting increased pro- metastatic processes,
and few reporting no impact.
In summary, the effects of stress, acute or chronic,
on tumour progression and metastasis are robust; are
mediated by β- adrenergic signalling; are mediated to
a lesser degree by HPA axis signalling114; and occur
through affecting tumour cells, and their microenvi-
ronment, including immune and stromal cells (FIG.2;
TABLE1). Notably, β- AR signalling that promotes
tumour progression corresponds with natural adrener-
gic effects on healthy/non- malignant tissue, including
adrenergic effects on EMT139,140, inflammation141–143 and
angiogenesis144,145 (not in the context of cancer). Last,
whereas rodent models and invitro cell culture pro-
vide causal evidence for specific links between stress
responses and tumour progression, findings from stud-
ies in patients are mostly correlative, with the exception
of a few intervention studies reviewed below.
Epidemiological studies
Stress and cancer incidence
A comprehensive meta- analysis of 142 prospective
studies published in 2008 (REF.146) (average sample size
of 87,000 people per study) indicated that psychoso-
cial stress predicts a 6% increase in cancer incidence
(hazard ratio = 1.06; 95% CI 1.02–1.11, P = 0.005). Of
note, depression was a major factor in this effect, rather
than stressful life events. However, the meta- analysis
identified a significant publication bias, suffered from
marked heterogeneity in the outcomes of the included
studies and was criticized for meta- analytic methodo-
logical flaws147. Moreover, 76% of the studies reported a
Tilt–light stress
An experimental stress
paradigm in which the home
cage of rodents is placed in a lit
room in a 45° tilted position,
starting before the onset of the
animals’ dark period, resulting
in reduced available floor
space and disruption of the
dark–light cycle.
Swim stress
An experimental stress
paradigm where a weight is
attached to the tail of rodents
(usually rats, up to 2.5% of
total body weight), which are
then placed in a room
temperature water tank for few
minutes, followed by a rest
period. This swim–rest cycle is
usually repeated several times.
b Proliferation and survival
c DNA damage
d Pro-metastasis effects:
increased EMT, MMPs
and invasion
e Pre-metastatic niche
f
NoradrenalineGlucocorticoids
Nerve growth
factors
MMPs
CCL2
RANKL
Adrenaline
β-AR
GR
Blood vessel Sympathetic
nerves
Macrophage BMSC
↓ Natural killer
cells
↑ MDSCs
↓ Effector
T cells
↑ M2 TAMs
MDM2
p53
h Immune evasion
CCR2
Angiogenesis and
lymphangiogenesis
↑ VEGF, IL-8 and IL-6
g Inflammation
↑ Prostaglandins and IL-6
Prostaglandin
receptor
a Stress factor exposure
Tumour cell
↑ Regulatory
T cells
Fig. 2 | Effects of stress on the tumour and its microenvironment.
Malignant tissue is exposed to systemic stress factors, including adrenaline,
noradrenaline and glucocorticoids (for example, cortisol in humans), and
to locally released noradrenaline through sympathetic tumour innervation
(part a). Tumours can also release nerve growth factors that increase their
sympathetic innervation and noradrenaline exposure, creating a feedforward
loop. Through membrane- bound β- adrenergic receptors (β- ARs), which bind
adrenaline and noradrenaline, and intracellular glucocorticoid receptors
(GRs), all of which are expressed by tumour, immune and stromal cells, stress
factors promote most hallmarks of cancer. Tumour cell proliferation and
resistance to cell death are increased (part b). In addition, activation of β- ARs
and GRs also induces activation of the E3- ubiquitin ligase MDM2 and
consequent degradation of p53, which leads to impaired genome
maintenance and accumulation of DNA damage (part c). Stress factors
promote invasion and metastasis by inducing tumour epithelial–
mesenchymal transition (EMT) and the release of matrix metalloproteinases
(MMPs) (part d). Furthermore, activation of β- ARs promotes the formation of
organ- specific pre- metastatic niches through processes such as
CC- chemokine ligand 2 (CCL2)–CC- chemokine receptor 2 (CCR2)- mediated
attraction of macrophages to the lung and receptor activator of nuclear
factor- κB ligand (RANKL) secretion by bone marrow stem cells (BMSCs),
which attract circulating tumour cells (part e). Stress factors promote the
release of various pro- angiogenic (part f) and inflammatory (part g) factors,
such as vascular endothelial growth factor (VEGF), interleukin-8 (IL-8), IL-6
and prostaglandins, from tumour and stromal cells, all of which promote
tumour progression. Stress- induced immune suppression facilitates tumour
immune evasion by upregulation of myeloid- derived suppressor cells
(MDSCs), regulatory Tcells and M2 tumour- associated macrophage (TAM)
polarization, and through downregulation of effector Tcell and natural killer
cell activity (part h). Activation of prostaglandin receptors and activation of
β- ARs each induces the same intracellular downstream processes (not
shown), including the cyclic AMP (cAMP)–protein kinase A (PKA) pathway,
suggesting that simultaneous blockade of β- adrenergic and prostanoid
signalling might be important to improve cancer treatment.
NATure revIewS
|
CAnCER
Reviews
VOLUME 21
|
DECEMBER 2021
|
773

0123456789();:
null effect, whereas 18% indicated harmful effects and
6% indicated protective effects.
More recent studies linked various specific stressors,
including a cold climate148, bereavement61, war149 and
depression150, to higher incidence of various cancer types,
yet other studies reported null effects151–153. Focusing on
work stress as a risk factor, two meta- analyses yielded
inconsistent conclusions: the first154 reported null effects
of prospective studies, whereas the second155 reported
elevated relative risk (of 1.24 and 1.36 in lung cancer
Table 1 | Biological effects of stress on cancer progression: preclinical studies and related observations in patients with cancer
Cancer Model Stressor Effect (location) Mediator Refs
Angiogenesis
Melanoma,
breast,
ovarian
Human cells invitro Adrenaline or
noradrenaline
↑ Angiogenesis; ↑ VEGF;↑ IL-6;
↑ IL-8
Tumour–endothelial cell contact
(β2- AR–Jagged 1–Notch); tumour
cell β1- AR and/or β2- AR–cAMP–
PKA signalling
81,95–98
Ovariana,
pancreatica,
colorectalb,
mammaryb,
prostatea
Mice; human or
mouse cells invitro
Social isolation,
chronic restraint,
audio of screaming
rats, laparotomy or
orisoproterenol
↑ Tumour vascularization;
↑ tumour VEGF; ↑ tumour growth;
↓ TSP1
β2- AR–cAMP–PKA signalling;
↑HIF1α; ↓CXCL4; macrophage
recruitment; β- AR–CREB–HDAC2
pathway
41,87,
99–102
Ovarian Patients with
cancerc
Low social supportd
or helplessness
↑ Plasma VEGF; ↑ tumour VEGF NA 103,104
Ovarian Patients with
cancerc
Low social attachmentd
or vegetative depressiond
↑ IL-6 (plasma, ascites);
↑ nocturnal cortisol (saliva)
NA 105,106,
220
Lymphatic modulation
Breasta,b Mice Chronic restraint ↑ Tumour VEGFC; ↑ tumour LVD;
↑ lymphatic dilation, flow;
↑ lymph node metastasis
β- AR; ↑ COX2; macrophage
recruitment
108
Breastc,
cervical
Patients with
cancer
Social isolationd
or SNS activity
↑ Tumour LVD; ↑ lymphatic flow NA 108,109
Inflammation and immunity
Mammaryb,
leukaemiab
Rats; blood samples
from stressed rats
studied ex vivo
Laparotomy, swim stress,
wet cage, metaproterenol
or adrenaline
↓ NKCC β1- AR and/or β2- AR 33,40,
112–114
Colorectalb,
squamous cell
carcinoma b,
mammaryb or
melanomab
Mice; mouse cells
invitro
Chronic restraint, 22 °C
housing temperature,
audio of screaming mice
or laparotomy
↓ TH1 cell/TH2 cell- type cytokine
ratio (serum); ↓ effector CD8+
and CD4+ TILs; ↑ tumour growth;
↑ MDSCs (tumour, spleen);
↑ regulatory Tcells (tumour, blood)
↓CXCL4; β- AR; β2- AR–STAT3
signalling
62,84,100,
116,118
Ovarian or
breast
Patients with
cancerc
Low social supportd, high
distressd, depressed/
anxious moodd
or psychological stress
↓ NKCC (tumour, blood);
↓ TH1 cell/TH2 cell- type cytokine
ratio (blood, ascites, tumour);
↓ MDSCs (blood)
NA 115,117,119
Breasta,b
or ovarian a
Mice; human or
mouse cells invitro
Chronic restraint or social
isolation
↑ Macrophage recruitment;
↑ prostaglandin (tumour cells,
TAMs); ↑ TAM M2 polarization
β- AR; β2- AR/NF- κB–prostaglandin
E2 axis; β- AR–cAMP–PKA– MCP1
production
41,79,90,
108,120
Ovarian or
breast
Patients with
cancerc
Psychological stress,
depression or social
isolationd
↑ Plasma IL-1Rα; ↑ tumour
prostaglandin; ↑ M2 polarization
of TAMs
NA 79,109,119
Metastasis
Breasta,b,
gastrica or
pancreatica
Mice; human or
mouse cells invitro
Chronic restraint,
alternating stressors, or
audio of screaming rats
↑ Pre- metastatic niche; ↑ EMT;
↑ MMPs (tumour, stroma) β- AR–RANKL; β- AR–CCL2/CCR2
axis; miR-337-3p–STAT3
41,42,74,
80,82,85,
99,122
Breasta,b or
gastrica
Mice or rats Chronic restraint,
laparotomy, alternating
stressors, wet cage or
swim stress
↑ Spontaneous and experimental
metastasis β1- AR and/or β2- AR 33,40,41,
74,85,108,
114,122,125
Ovarian or
breast
Patients with
cancerc
Perceived stressd, social
isolationd or depressiond
↑ Tumour EMT genes;
↑ TAM MMP9
NA 104,109,124
All findings are causal, except those indicated as correlational findings in the ‘Model’ or ‘Cancer’ column. ↑, increase; ↓, decrease; β- AR, β- adrenergic receptor;
cAMP, cyclic AMP; CCL2, CC- chemokine ligand 2; CCR2, CC- chemokine receptor 2; COX2, cyclooxygenase 2; EMT, epithelial–mesenchymal transition; HDAC2,
histone deacetylase 2; IL-6, interleukin-6; LVD, lymphatic vessel density; MDSC, myeloid- derived suppressor cell; MMP, matrix metalloproteinase; NA, not
applicable; NKCC, natural killer cell cytotoxicity; NF- κB, nuclear factor- κB; PKA, protein kinase A; RANKL, receptor activator of nuclear factor- κB ligand; SNS,
sympathetic nervous system; TAM, tumour- associated macrophage; TH1 cell, T helper 1 cell; TH2 cell, T helper 2 cell; TIL, tumour- infiltrating lymphocyte;
TSP1, thrombospondin 1; VEGF, vascular endothelial growth factor. aXenograft. bSyngeneic. cCorrelational findings. dAdjusted for disease stage.
www.nature.com/nrc
Reviews
774
|
DECEMBER 2021
|
VOLUME 21

0123456789();:
and CRC, respectively), but the latter also included
case–control studies that are susceptible to retrospec-
tive recall and interpretation bias. Last, it is important
to note that in humans, malignant transformation is a
prolonged process and subclinical cancer dormancy is
highly prevalent30. Thus, cancer incidence may be ele-
vated not only by initiation of the disease but also by
escape from dormancy or faster progression of cancer
to clinical manifestation. Indeed, animal studies report
that stress and stress factors can induce escape from
dormancy in tumour cells156–158.
Stress and cancer progression
Effects of stress on cancer progression are commonly
studied by assessing survival rates in patients diagnosed
with cancer. The overall hazard ratio indicated by 157
prospective studies included in the 2008 meta- analysis
discussed above146 was 1.03 (95% CI 1.02–1.04, P < 0.001),
with more than 73% of studies reporting null findings.
This small effect should be interpreted with caution.
First, stress (for example, life events) was commonly
assessed irrespective of its timing relative to cancer
detection, and the specific impact of stress while having
cancer, including the critical perioperative period, was
not assessed. Second, most patients with cancer experi-
ence some levels of cancer- related distress159,160, which
may suffice to generate a similar effect on cancer pro-
gression, irrespective of whether patients were catego-
rized with low versus high stress levels. This could mask
relations between stress levels and cancer progression
in such circumstances, but nevertheless could enable
marked beneficial effects of stress- reducing interven-
tions. Third, although comprehensive, this meta- analysis
is 13 years old, and has narrowed down analyses to either
distinct cancer types or defined stressors. More recent
meta- analyses have focused on more specific condi-
tions, and have reported larger effect sizes. Specifically,
depression in patients with breast cancer predicted
29% elevated risk for cancer- specific mortality161, and
low levels of perceived social support, a smaller social
network, being unmarried or being depressed predicted
a 12–25% elevated relative risk for cancer mortality in
various cancer types162,163.
Indeed, recent studies confirmed that the effects
of stress on survival are stressor- specific and cancer-
specific. For example, depression that followed cancer
diagnosis predicted decreased survival in breast161 and
renal164 cancers, but not in ovarian cancer165. Low social
support and low social attachment predicted decreased
survival in patients with ovarian cancer165, breast
cancer166 or CRC167, whereas work stress had no effect152.
Importantly, previous life history of stress and adversi-
ties may interact with post- diagnosis stress168,169, as early
adverse experiences can shape maladaptive responses to
stressors27.
Overall, given the small and inconsistent effects
reported by epidemiological studies, and the heteroge-
neous methodological approaches, populations studied
and type of stressors, it remains uncertain whether stress
can increase cancer incidence, and to what extent it facil-
itates cancer progression. Potentially, stress has a larger
impact in certain conditions or populations. Clearly,
epidemiological studies face significant obstacles. The
subjective perception of stress in patients with cancer
is influenced by the physical and mental burden of the
disease, and therefore studies that retrospectively assess
pre- diagnostic or post- diagnostic stress by subjective
reports are biased147. On the other hand, objective expo-
sure to adverse life events (for example, based on national
registries of divorces or deaths) does not include the
individual subjective experience. As described below,
the use of stress- reducing interventions in RCTs can
circumvent many of these obstacles.
Stress management in patients with cancer
The most methodologically sound approach to test
in humans whether stress affects cancer progression
would be RCTs, where the intervention is a verified
stress- management approach and the outcomes include
psychological indices, interim biomarkers and, most
importantly, long- term cancer outcomes. Such RCTs are
not practical for studying cancer incidence but are feasi-
ble for studying cancer progression and mortality. Such
psychological and pharmacological RCTs have been
conducted during the last four decades, as discussed
below.
Psychological RCTs: long- term outcomes
Recent meta- analyses170–173 have cumulatively identi-
fied 22 RCTs that employed psychosocial interventions
as being methodologically stringent, using Cochrane or
other criteria. Most interventions were initiated at least
a month postoperatively (16/22 RCTs) and/or were
conducted in patients with metastatic disease (12/22
RCTs); and most studies employed group interven-
tions (14/22 RCTs), rather than individual (7/22 RCTs)
approaches. Importantly, most interventions did yield
improvement in psychological indices (TABLE2), and a
few improved physiological biomarkers (for example,
natural killer cell activity)174–176 (BOX4). Based on these
meta- analyses (each considering 11–15 trials)170–173 and
our own assessment of all 22 studies (TABLE2), there is
no clear evidence for improvements in long- term cancer
outcomes171,172, but the results are nevertheless inform-
ative. Specifically, there seems to be an agreement that
some interventions can delay disease progression during
the first post- intervention years, but less so or not at all
beyond this initial period171,172. Psychosocial interven-
tions may have temporary effects either because their
impact on tumour biology is short- lasting or because
patients’ adherence to the psychological intervention
reduces along the follow- up period. It is suggested that
some patients, more than others, may benefit from
psychological interventions, specifically patients who
are older, unmarried and psychologically vulnerable or
stressed8,170, as well as patients in earlier disease stages
(for example, early- stage melanoma)177. It should be
noted that some of these studies have been criticized
for having methodological flaws178–180 (but also see
the response to criticism)181, including not having the
statistical power to study cancer mortality, employing
only 30–150 patients per group, which may lead to
exaggerated effect sizes. Some interventions have been
suggested to act through improving patients’ treatment
Laparotomy
An experimental stress
paradigm in which a
midline abdominal incision is
performed under anaesthesia,
and often the small intestine is
externalized and left hydrated
in a soaked gauze pad for
30 min. The intestine is then
internalized and the abdomen
is sutured.
Social confrontation stress
An experimental stress
paradigm where an intruder
rodent (a non- cage- mate
animal) is introduced into a
home cage populated with
several stable cage- mates.
The intruder is usually attacked
by the residents cage- mates
and/or displays submissive
behaviour.
Foot shock stress
An experimental stress
paradigm that is executed
in an apparatus containing an
electrified grid floor, in which
the animal is exposed to
electric shocks of varying
intensity and duration.
The paradigm can be acute
or chronic, and is also used
for fear- conditioning.
Hazard ratio
The ratio of the probability
of events in a treatment group
to the probability of events
in a control group.
Publication bias
The tendency to publish
a study based on its results
(positive rather than negative
findings or significant rather
than non- significant findings).
Existence of this bias can be
statistically assessed in
meta- analyses by Egger’s
linear regression test.
Cochrane
A non- profit organization
(maintaining no conflict of
interests), which, among
other activities, publishes
methodologies and guidelines
to produce high- quality
systematic reviews and
meta- analyses.
NATure revIewS
|
CAnCER
Reviews
VOLUME 21
|
DECEMBER 2021
|
775

0123456789();:
Table 2 | Psychosocial stress- reducing interventions in RCTs and long- term cancer outcomes
Study Patient
numbers
Intervention (setting; timing;
duration (weeks)a; treatment type)
Psychological
benefit
Survival
effect
Survival effect sizeb
Early- stage breast cancer
Burton etal.
(1995)187
n = 200,
4 groups of
50 eachc
Individual; preoperative; 1;
one interview + 30- min
psychotherapeutic intervention
Yes No First- year recurrence rates: T = 7–10%;
C = 14%; simple contrast between
control and intervention groups; NS
Kissane etal.
(2004)257
n = 303, T = 154 Group; post surgery; 20;
CBT- supportive therapy sessions + 3
relaxation sessions
Yes No Median survival time (months): T = 81.9,
C = 85.5; multivariate Cox analysis,
HR = 1.35, NS
Andersen etal.
(2008)258
n = 227 , T = 114 Group; post surgery; 16 weekly
sessions + 8 monthly sessions;
stress management
Yes Yes Mortality, 11- year follow- up: T = 24/114,
C = 30/113; multivariate Cox analysis,
HR = 0.44; P = 0.016
Median time to recurrence (months):
T = 33.6, C = 26.4; multivariate Cox
analysis, HR = 0.55, P = 0.034
Boesen etal.
(2011)259
n = 210, T = 105 Group; post surgery; 8; comprehensive
psychoeducation + supportive therapy
No No Mortality, 4- year follow- up: T = 6/105,
C = 3/105; statistical analysis not
preformed due to low event number
Stagl etal.
(2015)260
n = 240, T = 120 Group; post surgery; 10;
cognitive- based stress management
Yes Yes dMortality, 8–15- year follow- up:
T = 15/120, C = 15/120; multivariate
Cox analysis using four covariatesd,
HR = 0.21, P = 0.04
Metastatic breast cancer
Spiegel etal.
(1989)182
n = 86, T = 50 Group; post surgery; 52;
supportive- expressive
therapy + self- hypnosis
Yes Yes Mean survival time (months): T = 36.6,
C = 18.9; log- rank test, P < 0.0001
Cunningham
etal. (1998)261
n = 66, T = 30 Group; post surgery; 35;
supportive + CBT
No No Median survival time (months): T = 28.8,
C = 23.6; log- rank test, P = 0.35
Edelman etal.
(1999)262
n = 124, T = 62 Group; post surgery; 8 weekly
sessions + 3 sessions once a month; CBT
YeseNo Median survival time (months): T = 11.64,
C = 12.84; log- rank test, NS
Goodwin etal.
(2001)184
n = 225, T = 158 Group; replication study, similarf to
Spiegel etal. (1989)182
YesgNo Median survival time (months): T = 17.9,
C = 17.6; Cox univariate analysis,
HR = 1.06, NS
Kissane etal.
(2007)263
n = 227 , T = 147 Group; similar to Spiegel etal.
(1989)182 + 3 relaxation classes
Yes No Median Survival time (months): T = 24,
C = 18.3; univariate Cox analysis,
HR = 0.92, NS
Spiegel etal.
(2007)183
n = 125, T = 64 Group; replication study, same as
Spiegel etal. (1989)182
Yes No/yeshMedian survival time (months):
exploratory subgroup findings
(n = 25 ER- negativeh); T = 29.8, C = 9.3;
Multivariate Cox analysis, P = 0.002
Andersen etal.
(2010)264
n = 62, T = 29
(a subgroup of
patients from
Andersen etal.
(2008)258)i
Group; same as Andersen etal. (2008)258 Yes Ye s Mortality after recurrence: T = 19/29,
C = 25/33; median survival after
recurrence (months): T = 38.4, C = 20.4;
multivariate Cox analysis, HR = 0.41,
P = 0.014
Melanoma
Fawzy and
Fawzy (2003)177
n = 68, T = 34 Group; post surgery; 6;
health education + stress
management + coping
skills + psychological support
Yes No Mortality, 5–6- year follow- up: T = 3/34,
C = 10/34; log- rank test, P = 0.03
Mortality, 10-year follow- up: T = 9/34,
C = 11/34; log- rank test, NS
Boesen etal.
(2007)185
n = 262, T = 131 Group; replication study, similar to
Fawzy and Fawzy (2003)177
YesjNo Mortality, 4–6- year follow- up: T = 8/128,
C = 8/130; univariate Cox analysis,
HR = 0.99, NS
Other cancer types
Linn etal.
(1982)265 (several
cancer types)
n = 120, T = 62 Individual; NR; NR; supportive therapy Yes No Mean survival time (months), 1- year
follow- up: T = 3.7 , C = 4.37; life table
method, χ2 test, NS
Ilnyckyj etal.
(1994)266 (several
cancer types)
n = 127 , four
groups:k T = 31,
30, 35, C = 31
Group; NR; 24; supportive discussion
group sessions
No No Mean survival time (months), 10- year
follow- up: T = 70.7 , C = 82.4; log- rank
test, NS
www.nature.com/nrc
Reviews
776
|
DECEMBER 2021
|
VOLUME 21

0123456789();:
adherence, patients’ health behaviour and quality of
the medical treatment (for example, additional surveil-
lance and care) following improved communication
with medical personnel180. Thus far, no research group
has replicated previously reported positive outcomes,
although given the objective difficulties of intervention
trials and lack of funding, only a few replications have
been attempted177,182–185. Notably, each of the 22 studies
used a different treatment protocol, initiated treatment
at different times during cancer progression, provided
treatment for a different duration and/or studied a
different patient population and cancer type (TABLE2).
These heterogeneities may be the source of inconsistent
outcomes, and different results of meta- analyses. At the
single study level, 8 of the 22 interventions reported a
significant survival advantage of a psychosocial inter-
vention (TABLE2). Beyond the legitimate debate of the
validity of specific studies, eight successful demonstra-
tions could indicate promising outcomes. However, as
only eight such demonstrations have been reported,
and the results of these eight interventions have not
been replicated in published studies, combined with
the likelihood of unpublished studies with null effects,
this raises questions regarding the effectiveness of these
psychosocial interventions in improving cancer survival.
However, we believe that these inconsistent out-
comes are expected apriori, given the following consid-
erations. First, as discussed above and in BOX3, critical
time periods, such as the immediate perioperative time
frame in patients undergoing surgery, may bear a
non- proportional high impact on the fate of metastatic
disease, especially in patients harbouring only scattered
tumour cells and micrometastases186. Psychological
interventions have been commonly initiated weeks fol-
lowing surgery, which would miss this critical period
(only 3/22 studies in TABLE2 are perioperative)187–189.
Such delayed interventions may impact metastases at
a more advanced and therapeutically resistant stage,
thus confronting a greater challenge in preventing
metastatic disease, but still having the ability to delay
a metastatic outbreak171,172,174,177. Second, many medical
procedures, including surger y and chemotherapy, induce
stress- related inflammatory responses of local physio-
logical origin, including cellular responses of injured tis-
sue (for example, increased levels of damage- associated
molecular patterns and prostaglandins) (BOX 3).
Psychosocial interventions alone are unlikely to signif-
icantly reduce such local responses, which may mask
the potential beneficial effects of psychosocial interven-
tions during medical procedures (BOX4). Third, in many
Study Patient
numbers
Intervention (setting; timing;
duration (weeks)a; treatment type)
Psychological
benefit
Survival
effect
Survival effect sizeb
Other cancer types (cont.)
Ratcliffe
etal. (1995)267
(lymphoma)
n = 63, T = 36 Individual; post third cycle of
chemotherapy; NR; relaxation training
with or without hypnosis
Yes Yes Mortality, 5- year follow- up: T = 14/36,
C = 13/27; multivariate Cox analysis,
HR = 0.66, P = 0.06
Kuchler etal.
(2007)188
(gastrointestinal
cancers)
n = 271, T = 136 Individual; pre surgery to discharge
from hospital; 2–25 sessions;
individually tailored psychological
support
Yes Yes Survival, 2- year follow- up: T = 69/136,
C = 45/135; log- rank test, P = 0.002;
survival, 10- year follow- up: T = 29/136,
C = 13/135; log- rank test, P = 0.006
Ross etal.
(2009)268
(colorectal
cancer)
n = 249, T = 125 Individual; post surgery; 10 meetings
over 24 months; home visits by a
medical doctor or nurse providing
emotional support or information
No No Mortality, 6.5–9.5- year follow- up:
T = 75/125, C = 73/124; log- rank test, NS
Temel etal.
(2010)269
(metastatic
non- small cell
lung cancer)
n = 151, T = 77 NR; intervention group patients were
assigned to early palliative carel
Yes Yes lMedian survival time (months): T = 11.6,
C = 8.9; log- rank test, P = 0.02
Guo etal.
(2013)270 (several
cancer types)
n = 178, T = 89 Individual; during radiotherapy;
4–6; psychoeducation + CBT
+ supportive- expressive therapy
Yes No % survival, 2- year follow- up: T = 83.1%,
C: = 84.3%; log- rank test, NS
Zhang etal.
(2013)189
(oesophageal
cancer)
n = 60, T = 31 Individual; pre surgery; 3 weeks,
sessions every other day; health
education, psychological support,
stress management, coping strategies
and behaviour training
Yes No Survival, 4- year follow- up: T = 15/27 ,
C = 18/28; log- rank test, NS
C, control group; CBT, cognitive behavioural therapy; ER, oestrogen receptor; NR, not reported; NS, not significant; RCT, randomized controlled trial; T, treatment
group. aOne weekly session, unless otherwise specified. bLog- rank test and univariate Cox analyses address differences between groups that are driven only
by group assignment, whereas multivariate Cox analyses incorporate additional factors into the statistical model beyond group assignment. cThe different groups
were: preoperative interview; preoperative interview + 30- min preoperative psychotherapeutic intervention; preoperative interview + chat (attention); and routine
hospital care control. dEquivalent number of deaths between groups; difference was statistically significant in a Cox multivariate analysis addressing age at
diagnosis, disease stage, tumour size, HER2 status and hormonal treatment. eImproved psychological measures at the end of the intervention were not sustained at
3 and 6- month follow- up. fRelaxation techniques were taught instead of self- hypnosis. gIn patients with high baseline of distress. hCox proportional hazard analysis
showed a significant interaction between ER status and treatment, indicating that ER- negative patients allocated to the intervention survived longer than control
patients. iParticipants in this study were patients who previously participated in Andersen etal. study258. jPsychological benefits were only evident shortly after the
intervention, and enrolled patients exhibited low baseline levels of psychological distress. kThe different groups were: group meetings professionally guided by a
social worker for 6 months; group meetings professionally guided for 3 months + 3 months of unguided meetings; unguided group meetings; and control (no group
meetings). lEarlier initiation of palliative care, also addressing individual psychosocial needs of the patients.
Table 2 (cont.) | Psychosocial stress- reducing interventions in RCTs and long- term cancer outcomes
NATure revIewS
|
CAnCER
Reviews
VOLUME 21
|
DECEMBER 2021
|
777

0123456789();:
patients, psychosocial interventions cannot be expected
to be effective, either given low stress levels at study
entry or given individual characteristics of psycholog-
ical needs or coping style, not addressed by prevalent
standardized group therapies. Last, if one expects the
effect size of psychological intervention to be similar to
those of chemotherapy or hormonal therapy, hundreds
of patients of the same cancer type would need to be
included. We assert that, given appropriate funding, all
of these obstacles can be overcome, as detailed below,
enabling better assessment of the efficacy of stress
management for improving cancer survival.
Pharmacological RCTs: cancer biomarkers
Recently, several biomarker RCTs have employed phar-
macological interventions to antagonize stress responses
in patients with cancer, all employing the non- selective
β- blocker propranolol. Among other reasons, this drug
was chosen based on its early promising outcomes in
animal models of stress or surgery- induced cancer
progression41,74,85,108,125, t he involvement of both β1- AR and
β2- AR in various pro- malignant mechanisms6,40,65,87
and its high safety profile relative to other adrenergic anta-
gonists, especially regarding potential cardiovascular and
tissue healing- related complications190–192. Among other
positive prognostic outcomes in treated patients, pro-
pranolol downregulated the expression of mesenchymal
genes, the EMT transcription factors Snail and Slug, and
activity levels of the inflammatory transcription factors
nuclear factor- κB (NF- κB) and AP-1 in primary breast
tumours193, facilitated a decrease in CA-125 serum levels
in ovarian cancer194 and decreased classical monocyte
activation in haematopoietic cell transplant recipients195.
Propranolol is also currently being tested in combination
with immunotherapy in patients with melanoma196.
It is important to note that adrenergic stress res-
ponses and inflammatory responses often intertwine,
especially during cancer treatments, as perioperative
stress, tissue damage and other medical procedures
simultaneously induce both adrenergic and prostanoid
responses197–199 (BOX3), because each response facilitates
the other197 and because β- adrenergic and prostaglan-
din receptors activate the same intracellular immuno-
suppressive and tumour- promoting mechanisms (for
example, cAMP–PKA signalling)197. Therefore, it may
be necessary to simultaneously block β- adrenergic
and inflammatory responses to overcome the met-
astatic promoting effects of stress and/or medical
procedures. Indeed, several preclinical studies indi-
cated that simultaneous blockade of β- AR and COX2
activity (using propranolol and etodolac, respectively)
was synergistically more effective than each approach
alone in preventing immunosuppression and cancer
metastasis33,125,200,201.
These insights have been recently implemented clin-
ically in the context of curative oncological surgeries,
in two RCTs that have initiated combined proprano-
lol and etodolac treatment 5 days before surgery, for a
total of 11–20 days, in patients with breast cancer190,202
or CRC203. In resected tumours from both RCTs, the
treatment decreased EMT and the activity of several
pro- metastatic and pro- inflammatory transcription
factors (for example, those of the GATA, STAT, EGR
and CREB families), and improved the profile of infil-
trating leukocytes and tumour proliferation markers (for
example, Ki-67)190,202,203. In patients with breast cancer,
where repeated perioperative blood samples were also
analysed, treatment improved systemic inflammatory
and immunological markers, including IL-6, C- reactive
protein (CRP) and natural killer cell CD11a expression,
before and/or after surgery190,202. Although not pow-
ered to assess survival, the treatment improved 3- year
disease- free survival (DFS) in patients with CRC who
were protocol compliant203, and our as yet unpublished
data also show improved 5- year DFS.
Overall, these clinical findings indicate that
β- adrenergic blockade, with or without COX2 inhi-
bition, can significantly improve numerous biomark-
ers of cancer progression, and justify larger RCTs to
test long- term cancer outcomes of pharmacological
stress management, as currently being conducted
(NCT03838029 (REF.204), NCT03919461 (REF.205)).
Additional pharmacological approaches were also
studied. Specifically, the use of anxiolytic and anti-
depressant drugs (for example, selective serotonin reup-
take inhibitors and serotonin and noradrenaline reuptake
inhibitors) in patients with cancer is prevalent and effec-
tive in reducing anxiety and depression10,16. Nevertheless,
epidemiological studies assessing their impact on cancer
Box 4 | Behavioural stress management and its impact on short- term cancer-
related indices
Multiple psychological, behavioural and physiological interventions have been used to
target different aspects of stress in patients with cancer, such as massage, acupuncture,
yoga, tai chi, mindfulness and cognitive behavioural stress- reduction interventions
(reviewed in REFS8,10,12). Such interventions were shown to reduce stress, anxiety and
depression, and to improve quality of life309,310 in patients with cancer (for example,
in breast cancer174 and melanoma175). Accordingly, current guidelines for optimal
oncological care include screening and addressing psychosocial concerns311.
Importantly, Antoni and Dhabhar8 suggested that stress- management interventions
can have physiological protective effects against tumour progression through
improving protective immunity (for example, immunosurveillance), reducing chronic
inflammatory processes and inhibiting immunosuppressive mechanisms (for example,
regulatory Tcell activity). Indeed, in breast cancer survivors, yoga and tai chi reduced
pro- inflammatory processes312,313, and mindfulness- based stress reduction increased
the T helper 1 cell (TH1)/T helper 2 cell (TH2) ratio314, decreased nuclear factor- κB (NF- κB)
activity and increased anti- inflammatory signalling and gene expression of type 1
interferon315. Similar effects were noted by Antoni, studying the effects of a cognitive
behavioural therapy (CBT)- based stress- management intervention in patients with
breast cancer following surgery316. In addition to significant psychological benefits,
the intervention enhanced protective immunity (that is, increased gene expression
of type 1 interferon, and serum levels of interferon- γ (IFNγ) and interleukin-2 (IL-2)),
and reduced inflammatory processes (for example, reduced expression of the genes
encoding IL-1β, IL-6 and TNF, and increased prevalence of glucocorticoid receptor
(GR) response elements)176,317.
Missing from these studies are specific assessments of sympathetic activity and
potential reduction of tumour- associated noradrenaline and/or systemic adrenaline
levels in treated patients with cancer. Correlative studies in patients with cancer do
suggest association of these indices with stressors such as social isolation91.
Taken together, these changes may predict favourable prognosis for a broad range
of patients with cancer, and were suggested by Antoni and Dhabhar8 to explain
the beneficial effects of stress- management interventions on long- term cancer
survival258,260,264. Such interventions should be initiated as early as possible after cancer
diagnosis, and potentially before cancer surgery197, to improve their impact on both
mental health and long- term cancer outcomes.
www.nature.com/nrc
Reviews
778
|
DECEMBER 2021
|
VOLUME 21

0123456789();:
survival yielded inconsistent results206–208, and no effects
on cancer survivorship were noted when causally assessed
in an RCT that enrolled patients with advanced cancer
of various types209. Additionally, the effects of anxiolytic
and antidepressant drugs on cancer- related biomarkers
is largely unknown, and their impact on such indices
in controlled preclinical studies is contradictory210–212.
Thus, more preclinical and clinical research is needed to
assess the impact of such pharmacological approaches
on cancer- related biomarkers and long- term outcomes.
Stress and cancer reciprocal relations
In the clinical setting, stress and cancer can promote
each other. Patients with cancer often experience peaks
of stress on initial diagnosis, on cancer treatment and on
cancer recurrence159,160,198,213,214. Throughout cancer sur-
vivorship, anxiety decreases in some patients but persists
in others169, and patients with cancer show increased risk
for anxiety and depressive disorders214–216. Consequently,
stress responses and affective disorders may acceler-
ate cancer progression through various mechanisms
detailed above. Indeed, among patients with breast
cancer, higher anxiety, stress, depressive symptoms or
elevated diurnal cortisol levels were found to predict
suppressed antitumour cell- mediated immunity217–219;
and perceived stress, social isolation and depression pre-
dicted increased tumour cell EMT and levels of MMPs
in patients with ovarian and breast cancer (controlling
for disease parameters) (TABLE1).
Simultaneously, the malignant tissue itself may
heighten local and systemic stress responses, through
tumour- induced increases in sympathetic tumour
innervation and noradrenaline release65, and through
local and systemic inflammation that affects the CNS,
dysregulates HPA axis activity220,221 and facilitates
depression, sleep disturbances and cancer- related
fatigue222–224. Together with cancer- related cognitive
impairments225,226, these symptoms may induce or exac-
erbate stress responses227, perpetuating a vicious cycle of
stress and cancer (FIG.1).
Importantly, the brain, tumours and the immune
system all affect each other bidirectionally, either pro-
moting or hindering tumour progression. For example,
artificial activation of the brain reward system in mice
was found to decrease a suppressive MDSC phenotype
through reduced SNS signalling, resulting in attenuated
tumour growth228. Crosstalk between stress and cancer
is prominent within the perioperative period. In patients
with breast, colorectal or ovarian cancer, plasma cortisol
levels and/or stress inflammatory indices were elevated
even before surgery, presumably due to psychological
distress or tumour- derived inflammation190,203,220, which
may sensitize pain responses and worsen psychologi-
cal stress197. Pharmacological blockade of stress and/or
inflammatory responses before surgery reduces these
indices, as well as tumour EMT and other pro- metastatic
molecular indices in the malignant tissue190,202,203.
Stress impairs cancer treatments
Stress was reported in both preclinical and clinical
studies to impair adjuvant and neoadjuvant cancer
treatments, including chemotherapy, radiotherapy and
immunotherapy, through mediation of glucocorticoids
and/or catecholamines. Specifically, in murine models,
behavioural and/or surgical stress impaired the capac-
ity of the (clinically studied)229 immunostimulating
agents, CpG class C (CpG- C) and glucopyranosyl lipid-
A stable emulsion (GLA- SE), to reduce experimental
metastases in mammary cancer and CRC models133,230,231;
and invitro, corticosterone suppressed IL-12 secre-
tion from leukocytes following CpG- C or GLA- SE
stimulation133,232. Social disruption stress or β- AR acti-
vation in melanoma and lymphoma mouse models com-
promised several immunotherapies through impairing
CD8+ Tcell responses233,234; and restraint stress, cat-
echolamines or glucocorticoids impaired the efficacy of
chemotherapy in human breast and ovarian cancer cell
lines, both invitro and in xenograft models52,235.
Additionally, treatment with cytotoxic therapy or
sunitinib (an inhibitor of several tyrosine- kinase recep-
tors exerting both anti- angiogenic and direct anti-
tumoureffects) was impaired by chronic restraint stress
or administration of noradrenaline or adrenaline in
CRC, prostate cancer and melanoma mouse models236–238.
In mammary, pancreatic, melanoma, colon and lung
cancer models, β- AR signalling, induced by ambi-
ent temperature stress, jeopardized cytotoxic thera-
pies (cisplatin and nab- paclitaxel chemotherapies and
TRAIL (TNF- related cytokine which induces apop-
tosis by binding to cell surface death receptors))239,
radiotherapy240 and PD1- targeted immunotherapy84.
Activation of β- AR also induced resistance to the HER2
targeted therapy trastuzumab in gastric and breast
cancer mouse models241,242. Social disruption and acute
restraint stress impaired chemotherapy and immuno-
therapy in lung cancer, CRC and fibrosarcoma mouse
models, through glucocorticoid- induced expression of
the immunosuppressive transcription factor TSC22D3
in dendritic cells, and consequent impairment of anti-
tumour immunity92. Administration of the synthetic
glucocorticoid dexamethasone induced chemotherapy
and hormone- therapy resistance in prostate and breast
cancer mouse models93,243–245, as well as invitro in
breast cancer tumour samples and numerous human
carcinoma cell lines243,246. Last, in mice, blockade of GR
in combination with chemotherapy or hormone therapy
potentiated invivo therapeutic responses244,245.
Corresponding clinical observations have been
reported in patients. In breast cancer, tumour expres-
sion of β- AR negatively correlated with response to
trastuzumab242, and in patients with prostate cancer,
increased GR expression in bone metastases following
treatment with enzalutamide (anti- androgen recep-
tor therapy) predicted poorer therapeutic response244.
Retrospective observations indicated that incidental
β- blocker usage with anti- angiogenic agents, immuno-
therapy, radiation and/or chemotherapy extended patient
DFS and overall survival247–250.
In sum, ample preclinical studies indicate that stress,
noradrenaline, adrenaline and glucocorticoids can jeop-
ardize adjuvant and neoadjuvant therapies, although
clinical studies have not sufficiently addressed this
important issue. Also concerning is the prevalent use of
synthetic glucocorticoids (including dexamethasone) in
CpG class C
(CpG- C). A synthetic oligodeox-
ynucleotide (ODN) that
functions as a Toll- like
receptor 9 (TLR9) agonist
and induces a physiological
host- dependent activation of
the immune system.
Glucopyranosyl lipid- A
stable emulsion
(GLA- SE). A synthetic agonist
of Toll- like receptor 4 (TLR4).
For administration, GLA is
dissolved in an oil–water stable
emulsion that serves as an
adjuvant delivery system.
NATure revIewS
|
CAnCER
Reviews
VOLUME 21
|
DECEMBER 2021
|
779

0123456789();:
patients with cancer. These agents are routinely employed
to reduce chemotherapy- induced emesis (nausea and
vomiting)251, to potentiate chemotherapy in lymphoid
cancers246,252 and to counteract inflammatory or autoim-
mune responses to immunotherapy253. Use of synthetic
glucocorticoids in solid malignancies, which may or may
not express GRs, could jeopardize adjuvant treatments
and promote cancer progression. Indeed, in patients
with non- small cell lung cancer, synthetic glucocorti-
coid use predicted decreased response to immune check-
point inhibitors (including anti- PDL1 immunotherapy),
and decreased DFS and overall survival254–256.
Conclusions and perspectives
Although the evidence that stress promotes cancer
initiation is inconsistent, there is robust evidence that
stress can facilitate cancer progression through modu-
lating most hallmarks of cancer. Molecular and systemic
mechanisms mediating these effects have been identi-
fied in animal studies, and most have been recognized
in patients with cancer. SNS- derived adrenergic stress
responses, and adrenergic–inflammatory responses in
the context of medical procedures, are key mediators of
these deleterious effects of stress. The use of synthetic
steroids, and stress- induced glucocorticoid release in
some models, were also shown to promote cancer pro-
gression, and to reduce efficacy of adjuvant therapies.
However, it should be noted that animal studies leverage
their ability to synchronize stress exposure with specific
phases of cancer growth and metastasis that are critically
prone to stress. By contrast, epidemiological studies and
most clinical trials assessing stress- reducing psychoso-
cial interventions did not focus on stress- prone phases,
some of which cannot be identified and addressed clin-
ically. Thus, it is not a surprise that epidemiological and
clinical intervention studies have shown small effect
size or mixed outcomes. Importantly, psychological
interventions have the potential to individually address
patients’ unique sources of stress responses, may exert
enduring post- treatment effects without drug adverse
effects and are feasible in patients with contraindications
to drug therapy. Based on our current understanding
of cancer biology, stress and the complex interactions
between them along critical time frames in the contin-
uum of cancer, we hypothesize that stress- management
interventions can reduce cancer recurrence and mor-
tality, especially in patients undergoing curative onco-
logical surgery. To facilitate such beneficial effects, we
suggest that stress- management interventions should
be tested during critical periods affecting cancer pro-
gression, especially the short perioperative period and
adjuvant treatments, and compared with other time
periods; should be accompanied by pharmacologi-
cal approaches to overcome stress and inflammatory
responses that are unavoidably triggered by medical
procedures; and should include individualized mod-
ules to accommodate patient- unique characteristics and
needs, and focus on patients with higher manifestation
of stress symptomology. Such studies should be powered
similarly to testing a new drug therapy, and will likely
require prioritization by non- profit funding organiza-
tions. Recent biomarker clinical trials, including phar-
macological stress- reducing interventions, indicate the
potential capacity of such approaches to reduce cancer
mortality. Based on the current data, we believe that
such approaches should be tested through large collab-
orative multicentre RCTs, assessing the impact of uni-
fied interventions on long- term cancer outcomes, with
similar rigour to that employed when studying a new
agent for cancer therapy.
Published online 10 September 2021
1. LeShan, L. Psychological states as factors in the
development of malignant disease: a critical review.
J. Natl Cancer Inst. 22, 1–18 (1959).
2. Mravec, B., Tibensky, M. & Horvathova, L. Stress and
cancer. Part I: mechanisms mediating the effect of
stressors on cancer. J. Neuroimmunol. 346, 577311
(2020).
This review describes mechanisms by which stress
affects specific hallmarks of cancer, emphasizing
how stress is an integral part of cancer biology.
3. Cole, S. W. & Sood, A. K. Molecular pathways:
β- adrenergic signaling in cancer. Clin. Cancer Res. 18,
1201–1206 (2012).
4. Eng, J. W.-L. etal. A nervous tumor microenvironment:
the impact of adrenergic stress on cancer cells,
immunosuppression, and immunotherapeutic
response. Cancer Immunol. Immunother. 63,
1115–1128 (2014).
5. Armaiz- Pena, G. N., Cole, S. W., Lutgendorf, S. K.
& Sood, A. K. Neuroendocrine influences on cancer
progression. Brain Behav. Immun. 30, S19–S25
(2013).
6. Cole, S. W., Nagaraja, A. S., Lutgendorf, S. K.,
Green, P. A. & Sood, A. K. Sympathetic nervous
system regulation of the tumour microenvironment.
Nat. Rev. Cancer 15, 563 (2015).
This review describes the contribution of
adrenergic signalling to cancer progression,
focusing on the tumour microenvironment.
7. Armaiz- Pena, G. N., Colon- Echevarria, C. B. &
Lamboy- Caraballo, R. Neuroendocrine regulation of
tumor- associated immune cells. Front. Oncol. 9, 1077
(2019).
This review examines the effects of sympathetic
and/or glucocorticoid signalling on various
tumour- associated immune cells.
8. Antoni, M. H. & Dhabhar, F. S. The impact of
psychosocial stress and stress management on
immune responses in patients with cancer. Cancer
125, 1417–1431 (2019).
This review discusses both preclinical and clinical
studies and summarizes the effects of stress and
stress management on immune indices in cancer,
suggesting potential optimal strategies for stress
management in patients with cancer.
9. Neeman, E. & Ben- Eliyahu, S. Surgery and stress
promote cancer metastasis: new outlooks on
perioperative mediating mechanisms and immune
involvement. Brain Behav. Immun. 30, S32–S40
(2013).
10. Cui, B. etal. Cancer and stress: NextGen strategies.
Brain Behav. Immun. 93, 368–383 (2020).
11. Lutgendorf, S. K. & Andersen, B. L. Biobehavioral
approaches to cancer progression and survival:
mechanisms and interventions. Am. Psychol. 70,
186–197 (2015).
12. Mravec, B., Tibensky, M. & Horvathova, L.
Stress and cancer. Part II: therapeutic implications
for oncology. J. Neuroimmunol. 346, 577312
(2020).
13. Moreno- Smith, M., Lutgendorf, S. K. & Sood, A. K.
Impact of stress on cancer metastasis. Future Oncol.
6, 1863–1881 (2010).
14. Selye, H. The Stress of Life (McGraw- Hill, 1956).
15. Cacioppo, J. T., Cacioppo, S., Capitanio, J. P. &
Cole, S. W. The neuroendocrinology of social isolation.
Annu. Rev. Psychol. 66, 733–767 (2015).
16. Bortolato, B. etal. Depression in cancer: the many
biobehavioral pathways driving tumor progression.
Cancer Treat. Rev. 52, 58–70 (2017).
17. Liu, R. T. & Alloy, L. B. Stress generation in depression:
a systematic review of the empirical literature and
recommendations for future study. Clin. Psychol. Rev.
30, 582–593 (2010).
18. Wang, Q., Timberlake, M. A. II, Prall, K. & Dwivedi, Y.
The recent progress in animal models of depression.
Prog. Neuropsychopharmacol. Biol. Psychiatry 77,
99–109 (2017).
19. Sapolsky, R. M. Stress and the brain: individual
variability and the inverted- U. Nat. Neurosci. 18,
1344 (2015).
20. McEwen, B. S. Neurobiological and systemic
effects of chronic stress. Chronic Stress 1,
2470547017692328 (2017).
21. McEwen, B. S. & Stellar, E. Stress and the individual:
mechanisms leading to disease. Arch. Intern. Med.
153, 2093–2101 (1993).
22. McEwen, B. S. & Gianaros, P. J. Central role of the
brain in stress and adaptation: links to socioeconomic
status, health, and disease. Ann. NY Acad. Sci. 1186,
190 (2010).
23. Lazarus, R. S. & Folkman, S. Stress, Appraisal,
and Coping (Springer, 1984).
24. Holmes, T. H. & Rahe, R. H. The social readjustment
rating scale. J. Psychosom. Res. 11 , 213–218
(1967).
25. McEwen, B. S., Gray, J. D. & Nasca, C. Recognizing
resilience: learning from the effects of stress on the
brain. Neurobiol. Stress. 1, 1–11 (2015).
26. Fava, G. A. etal. Clinical characterization of allostatic
overload. Psychoneuroendocrinology 108, 94–101
(2019).
27. Kiecolt- Glaser, J. K., Renna, M. E., Shrout, M. R.
& Madison, A. A. Stress reactivity: what pushes
us higher, faster, and longer — and why it matters.
Curr. Dir. Psychol.Sci. 29, 492–498 (2020).
28. Hanahan, D. & Weinberg, R. A. Hallmarks of cancer:
the next generation. Cell 144, 646–674 (2011).
www.nature.com/nrc
Reviews
780
|
DECEMBER 2021
|
VOLUME 21

0123456789();:
29. Fouad, Y. A. & Aanei, C. Revisiting the hallmarks
of cancer. Am. J. Cancer Res. 7, 1016 (2017).
30. Manjili, M. H. Tumor dormancy and relapse: from
a natural byproduct of evolution to a disease state.
Cancer Res. 77, 2564–2569 (2017).
31. Bergers, G. & Benjamin, L. E. Tumorigenesis and
the angiogenic switch. Nat. Rev. Cancer 3, 401–410
(2003).
32. Patidar, A. etal. DAMP–TLR–cytokine axis dictates
the fate of tumor. Cytokine 104, 114–123 (2018).
33. Melamed, R. etal. Marginating pulmonary- NK activity
and resistance to experimental tumor metastasis:
suppression by surgery and the prophylactic use of a
β- adrenergic antagonist and a prostaglandin synthesis
inhibitor. Brain Behav. Immun. 19, 114–126 (2005).
34. Melamed, R. etal. The marginating- pulmonary
immune compartment in rats: characteristics of
continuous inflammation and activated NK cells.
J. Immunother. 33, 16–29 (2010).
35. Sorski, L. etal. Prevention of liver metastases through
perioperative acute CpG- C immune stimulation.
Cancer Immunol. Immunother. 69, 2021–2031
(2020).
36. Strilic, B. & Offermanns, S. Intravascular survival
and extravasation of tumor cells. Cancer Cell 32,
282–293 (2017).
37. Shaashua, L. etal. Spontaneous regression of micro-
metastases following primary tumor excision: a critical
role for primary tumor secretome. BMC Biol. 18,
1–13 (2020).
38. Gonzalez, H., Hagerling, C. & Werb, Z. Roles of the
immune system in cancer: from tumor initiation to
metastatic progression. Genes Dev. 32, 1267–1284
(2018).
39. Rosenne, E. etal. Inducing a mode of NK- resistance
to suppression by stress and surgery: a potential
approach based on low dose of poly I- C to reduce
postoperative cancer metastasis. Brain Behav. Immun.
21, 395–408 (2007).
40. Ben- Eliyahu, S., Shakhar, G., Page, G. G., Stefanski, V.
& Shakhar, K. Suppression of NK cell activity and
of resistance to metastasis by stress: a role
for adrenal catecholamines and β- adrenoceptors.
Neuroimmunomodulation 8, 154–164 (2000).
41. Sloan, E. K. etal. The sympathetic nervous system
induces a metastatic switch in primary breast cancer.
Cancer Res. 70, 7042–7052 (2010).
This preclinical study in a breast cancer model
reports that whereas chronic stress does not
promote primary tumour growth, it promotes its
metastatic dissemination, demonstrating specific
interactions between stress and unique stages
in cancer progression.
42. Du, P. etal. Chronic stress promotes EMT- mediated
metastasis through activation of STAT3 signaling
pathway by miR-337-3p in breast cancer. Cell Death
Dis. 11, 1–13 (2020).
43. Madden, K. S., Szpunar, M. J. & Brown, E. B.
Early impact of social isolation and breast tumor
progression in mice. Brain Behav. Immun. 30,
S135–S141 (2013).
44. Volden, P. A. & Conzen, S. D. The influence of
glucocorticoid signaling on tumor progression.
Brain Behav. Immun. 30, S26–S31 (2013).
45. Flint, M. S., Baum, A., Chambers, W. H. & Jenkins, F. J.
Induction of DNA damage, alteration of DNA repair
and transcriptional activation by stress hormones.
Psychoneuroendocrinology 32, 470–479 (2007).
46. Hara, M. R. etal. A stress response pathway regulates
DNA damage through β2-adrenoreceptors and
β- arrestin-1. Nature 477, 349–353 (2011).
47. Hara, M. R., Sachs, B. D., Caron, M. G. &
Lefkowitz, R. J. Pharmacological blockade of a
β2AR–β- arrestin-1 signaling cascade prevents the
accumulation of DNA damage in a behavioral stress
model. Cell Cycle 12, 219–224 (2013).
48. Feng, Z. etal. Chronic restraint stress attenuates p53
function and promotes tumorigenesis. Proc. Natl
Acad. Sci. USA 109, 7013–7018 (2012).
49. Gidron, Y., Russ, K., Tissarchondou, H. & Warner, J.
The relation between psychological factors and
DNA- damage: a critical review. Biol. Psychol. 72,
291–304 (2006).
50. Lamboy- Caraballo, R. etal. Norepinephrine- induced
DNA damage in ovarian cancer cells. Int. J. Mol. Sci.
21, 2250 (2020).
51. Flaherty, R. L. etal. Glucocorticoids induce production
of reactive oxygen species/reactive nitrogen species
and DNA damage through an iNOS mediated pathway
in breast cancer. Breast Cancer Res. 19, 1–13 (2017).
52. Reeder, A. etal. Stress hormones reduce the efficacy
of paclitaxel in triple negative breast cancer through
induction of DNA damage. Br. J. Cancer 11 2 ,
1461–1470 (2015).
53. Plummer, M. etal. Global burden of cancers
attributable to infections in 2012: a synthetic analysis.
Lancet Glob. Health 4, e609–e616 (2016).
54. de Martel, C., Georges, D., Bray, F., Ferlay, J. &
Clifford, G. M. Global burden of cancer attributable
to infections in 2018: a worldwide incidence analysis.
Lancet Glob. Health 8, e180–e190 (2020).
55. Antoni, M. H. etal. The influence of bio- behavioural
factors on tumour biology: pathways and mechanisms.
Nat. Rev. Cancer 6, 240–248 (2006).
56. Irwin, M. R. & Cole, S. W. Reciprocal regulation
of the neural and innate immune systems. Nat. Rev.
Immunol. 11, 625–632 (2011).
57. Collado- Hidalgo, A., Sung, C. & Cole, S. Adrenergic
inhibition of innate anti- viral response: PKA blockade
of type I interferon gene transcription mediates
catecholamine support for HIV-1 replication.
Brain Behav. Immun. 20, 552–563 (2006).
58. Cacioppo, J. T. etal. Autonomic and glucocorticoid
associations with the steady- state expression of latent
Epstein–Barr virus. Hormones Behav. 42, 32–41
(2002).
59. Glaser, R. etal. The differential impact of training
stress and final examination stress on herpesvirus
latency at the United States Military Academy at West
Point. Brain Behav. Immun. 13, 240–251 (1999).
60. Fang, C. Y. etal. Perceived stress is associated with
impaired T- cell response to HPV16 in women
with cervical dysplasia. Ann. Behav. Med. 35, 87–96
(2008).
61. Fang, F. etal. Risk of infection- related cancers
after the loss of a child: a follow- up study in Sweden.
Cancer Res. 71, 116–122 (2011).
62. Saul, A. N. etal. Chronic stress and susceptibility
to skin cancer. J. Natl Cancer Inst. 97, 1760–1767
(2005).
63. Sumis, A. etal. Social isolation induces autophagy
in the mouse mammary gland: link to increased
mammary cancer risk. Endocr. Relat. Cancer 23,
839–856 (2016).
64. Kokolus, K. M. etal. Baseline tumor growth and
immune control in laboratory mice are significantly
influenced by subthermoneutral housing temperature.
Proc. Natl Acad. Sci. USA 11 0 , 20176–20181 (2013).
65. Renz, B. W. etal. β2 adrenergic- neurotrophin
feedforward loop promotes pancreatic cancer.
Cancer Cell 33, 75–90.e7 (2018).
This research demonstrates the reciprocal relations
between the malignant tissue and its direct
sympathetic innervation in preclinical pancreatic
cancer models.
66. Magnon, C. etal. Autonomic nerve development
contributes to prostate cancer progression. Science
341, 1236361 (2013).
67. Hermes, G. L. etal. Social isolation dysregulates
endocrine and behavioral stress while increasing
malignant burden of spontaneous mammary tumors.
Proc. Natl Acad. Sci. USA 10 6, 22393–22398
(2009).
68. Hasen, N. S., O’Leary, K. A., Auger, A. P. &
Schuler, L. A. Social isolation reduces mammary
development, tumor incidence, and expression
of epigenetic regulators in wild- type and p53-
heterozygotic mice. Cancer Prev. Res. 3, 620–629
(2010).
69. Nguyen, K. D. etal. Alternatively activated
macrophages produce catecholamines to sustain
adaptive thermogenesis. Nature 480, 104–108
(2011).
70. Flierl, M. A. etal. Phagocyte- derived catecholamines
enhance acute inflammatory injury. Nature 449,
721–725 (2007).
71. Wong, H. P. S. etal. Nicotine promotes cell
proliferation via α7-nicotinic acetylcholine receptor
and catecholamine- synthesizing enzymes- mediated
pathway in human colon adenocarcinoma HT-29 cells.
Toxicol. Appl. Pharmacol. 221 , 261–267 (2007).
72. Shi, M. etal. The β2-adrenergic receptor and Her2
comprise a positive feedback loop in human breast
cancer cells. Breast Cancer Res. Treat. 125, 351–362
(2011).
73. Amaro, F. etal. β- Adrenoceptor activation in breast
MCF-10A cells induces a pattern of catecholamine
production similar to that of tumorigenic MCF-7 cells.
Int. J. Mol. Sci. 21, 7968 (2020).
74. Zhang, X. etal. Chronic stress promotes gastric cancer
progression and metastasis: an essential role for
ADRB2. Cell Death Dis. 10, 1–15 (2019).
75. Zhi, X. etal. Adrenergic modulation of AMPK-
dependent autophagy by chronic stress enhances cell
proliferation and survival in gastric cancer.
Int. J. Oncol. 54, 1625–1638 (2019).
76. Wong, H. P. etal. Effects of adrenaline in human colon
adenocarcinoma HT-29 cells. Life Sci. 88, 1108–1112
(2011).
77. Sood, A. K. etal. Adrenergic modulation of focal
adhesion kinase protects human ovarian cancer
cells from anoikis. J. Clin. Invest. 120, 1515–1523
(2010).
78. Liu, H. etal. Activation of adrenergic receptor
β2 promotes tumor progression and epithelial
mesenchymal transition in tongue squamous cell
carcinoma. Int. J. Mol. Med. 41, 147–154 (2018).
79. Nagaraja, A. S. etal. Sustained adrenergic signaling
leads to increased metastasis in ovarian cancer via
increased PGE2 synthesis. Oncogene 35, 2390–2397
(2016).
80. Kim- Fuchs, C. etal. Chronic stress accelerates
pancreatic cancer growth and invasion: a critical
role for β- adrenergic signaling in the pancreatic
microenvironment. Brain Behav. Immun. 40, 40–47
(2014).
81. Moretti, S. etal. β- Adrenoceptors are upregulated
in human melanoma and their activation releases
pro- tumorigenic cytokines and metalloproteases
in melanoma cell lines. Lab. Invest. 93, 279–290
(2013).
82. Pu, J. etal. Adrenaline promotes epithelial- to-
mesenchymal transition via HuR–TGFβ regulatory axis
in pancreatic cancer cells and the implication in cancer
prognosis. Biochem. Biophys. Res. Commun. 493,
1273–1279 (2017).
83. Liu, J. etal. A novel β2-AR/YB-1/β- catenin axis
mediates chronic stress- associated metastasis in
hepatocellular carcinoma. Oncogenesis 9, 1–14
(2020).
84. Bucsek, M. J. etal. β- Adrenergic signaling in mice
housed at standard temperatures suppresses an
effector phenotype in CD8+ Tcells and undermines
checkpoint inhibitor therapy. Cancer Res. 77,
5639–5651 (2017).
85. Chen, H. etal. Chronic psychological stress promotes
lung metastatic colonization of circulating breast
cancer cells by decorating a pre- metastatic niche
through activating β- adrenergic signaling. J. Pathol.
244, 49–60 (2018).
86. Lamkin, D. M. etal. Chronic stress enhances
progression of acute lymphoblastic leukemia via
β- adrenergic signaling. Brain Behav. Immun. 26,
635–641 (2012).
87. Thaker, P. H. etal. Chronic stress promotes tumor
growth and angiogenesis in a mouse model of ovarian
carcinoma. Nat. Med. 12, 939–944 (2006).
This is the first preclinical study to demonstrate the
effects of chronic stress on tumour angiogenesis,