You are on page 1of 16

GASTROINTESTINAL CANCER AND OBESITY

REVIEwS
Obesity and gastrointestinal cancer:
the interrelationship of adipose and
tumour microenvironments
Jacintha O’Sullivan1,2, Joanne Lysaght1,2, Claire L. Donohoe1 and John V. Reynolds1*
Abstract | Increasing recognition of an association between obesity and many cancer types exists,
but how the myriad of local and systemic effects of obesity affect key cellular and non-​cellular
processes within the tumour microenvironment (TME) relevant to carcinogenesis, tumour
progression and response to therapies remains poorly understood. The TME is a complex cellular
environment in which the tumour exists along with blood vessels, immune cells, fibroblasts, bone
marrow-​derived inflammatory cells, signalling molecules and the extracellular matrix. Obesity ,
in particular visceral obesity , might fuel the dysregulation of key pathways relevant to both the
adipose microenvironment and the TME, which interact to promote carcinogenesis in at-​risk
epithelium. The tumour-​promoting effects of obesity can occur at the local level as well as
systemically via circulating inflammatory , growth factor and metabolic mediators associated with
adipose tissue inflammation, as well as paracrine and autocrine effects. This Review explores key
pathways linking visceral obesity and gastrointestinal cancer, including inflammation, hypoxia,
altered stromal and immune cell function, energy metabolism and angiogenesis.

Large-​scale epidemiological studies have demonstrated and 110 million children worldwide are obese3. In the
a consistent and compelling association between the risk United States, the obesity prevalence in men and women
of cancer development and obesity (defined by BMI) for is 35% and 40%, respectively, and approximately 5% of
several cancers1 (summarized in Table 1). An umbrella men and 10% of women have a BMI >40 kg/m2 (ref.4).
review of the literature comprising 204 meta-​analyses Obesity is estimated to contribute to approximately 9%
of large studies with limited heterogeneity or evidence of of all cancers5 and between 5% and 20% of all cancer-​
bias was published in 2017 (ref.2). For gastrointestinal related deaths6. Furthermore, the prevalence of obesity
cancer, the analysis revealed a strong and potentially in the United States in the 2–19-year age group was
causal association between obesity and oesophageal estimated to be 17% in 2014, with a BMI ≥40 kg/m2
adenocarcinoma, colon cancer in men, biliary tract can- in 5.8%; additionally, the prevalence of overweight in
cer and pancreatic cancer. The evidence is also highly those <10 years of age is up to 40% in some European
suggestive of an association between liver cancer and countries, presenting a concerning vista for the future7–9.
adenocarcinoma of the gastric cardia (the oesophago- Globally, excess body weight is third behind smoking
gastric junction)2. In 2016, the International Agency and infection as an attributable risk factor for cancer, and
for Research on Cancer (IARC) reported relative risks second to smoking in Western populations10.
(RRs) of 1.5 to 1.8 for the highest BMI (up to ≥40 kg/m2) In addition to promoting carcinogenesis, obesity
compared with a normal BMI (18.0–24.9 kg/m2) for might also affect the tumour biology of established can-
these tumour sites1. Oesophageal adenocarcinoma was cers. For instance, poorer survival is reported in patients
1
Trinity Translational Medicine notable for a progressive increase in RR for each 5 kg/m2 with colon cancer and pancreatic cancer who are over-
Institute, Trinity College increase in BMI (RR 4.8 for a BMI ≥40 kg/m2), suggest- weight or obese at the time of diagnosis11,12. The risk
Dublin and St. James’s
ing a dose–response effect for this tumour type. The of dying is increased in both patients with overweight
Hospital, Dublin, Ireland.
IARC report concluded that the absence of excess body (HR 1.26, 95% CI 0.94–1.69, P = 0.04) and with obe-
2
These authors contributed
equally: Jacintha O’Sullivan,
fat reduces the risk of most cancers1. sity (HR 1.86, 95% CI 1.35–2.56, P < 0.001) with pan-
Joanne Lysaght. These epidemiological reports are in the context of creatic cancer12 and in men with obesity treated with
*e-​mail: reynoljv@tcd.ie an increase in the global prevalence of obesity between adjuvant chemotherapy for colon cancer (HR 1.16, 95%
https://doi.org/10.1038/ 1975 and 2014 from 3.2% to 10.8% in men and from CI 1.01–1.33, P = 0.0297)11. Although this finding might
s41575-018-0069-7 6.4% to 14.9% in women; currently, 640 million adults be partially due to later diagnosis and reduced uptake of

Nature Reviews | Gastroenterology & Hepatology


REvIEwS

most reviews to date have largely focused on hormone-​


Key points
sensitive cancers, in particular breast and prostate
• Epidemiological evidence implicates obesity as a risk factor for the development of cancer24, or on specific adipokines and growth factors,
cancers at multiple sites in the gastrointestinal tract, including the oesophagus, liver, including leptin, adiponectin, insulin and insulin-​like
colon, gastric cardia and pancreas. growth factor 1 (IGF1) (refs25,26). In this Review, we
• Immune, metabolic and inflammation-​associated properties of excess adiposity are discuss the relationships between obesity and gastro­
increasingly understood, but how they influence the tumour microenvironment intestinal cancers (oesophagus, stomach, pancreas, liver
(TME), which comprises tumour cells, blood vessels, immune cells, fibroblasts and the
and colorectal), with a particular focus on key cellular
extracellular matrix, remains unclear.
components, such as immune cells and adipose-​derived
• Adipocyte stem cells are increased in adipose tissue of individuals with obesity;
stem cells (ASCs), and common themes in both the TME
in mouse models, these cells migrate to tumour sites and differentiate into cell types
and the obese adipose microenvironment, including
that might affect the TME.
inflammation, hypoxia, angiogenesis and altered energy
• Interstitial fibrosis within the TME can influence cytokine signalling, epithelial cell
metabolism. In addition, the prospect of therapeutically
morphology and stem cell differentiation. Adipocyte stem cells might be a source of
cancer-​associated fibroblasts, and obesity might play a part in fibrosis-​associated targeting obesity to reduce cancer risk or to improve
mechanosignalling within tumours. patient outcomes is considered.
• Metabolic reprogramming in the TME is associated with obesity, hypoxia and
angiogenesis. These processes are tightly interconnected and represent a potential Obesity-​associated inflammation
target affecting not only tumours but also immune or inflammatory cells within Adipose tissue is largely located in two main depots:
the TME. subcutaneous and visceral. The main function of adi-
• Bariatric or metabolic surgery is associated with an improvement in the metabolic pose tissue is in energy homeostasis, but it also acts as
profile of patients with obesity and a marked decrease in the incidence of cancer an endocrine organ, releasing hormones, growth factors
development. and adipokines, the cell signalling proteins produced
by adipose tissue27. Excess caloric intake leads to alter-
ations in insulin sensitivity and cellular stress within
screening or diagnostic tests in populations with obe- adipocytes owing to surplus lipid accumulation, with
sity13,14, it is also linked to a reduction in the efficacy of consequent release of inflammatory cytokines and
conventional therapies, including chemotherapy, radio- adipokines from adipocytes and infiltrating immune
therapy and agents that target angiogenesis6,7,12,15–17. This cells28. Visceral adipose tissue (VAT), which includes
negative association with outcomes, treatment effects central depots including the omentum, mesenteric,
and prognosis is acknowledged by a position statement epiploic, gonadal, epicardial and retroperitoneal fat
in 2014 from the American Society of Clinical Oncology pads, is thought to be a key source of obesity-​derived
(ASCO), which committed to tackling the problem of inflammation and altered metabolism in comparison
obesity and cancer through increasing education and with subcutaneous adipose tissue29. The largest and
awareness, the provision of tools and resources for best studied VAT depot is the omentum, the so-​called
providers to address obesity with their patients, and fatty apron that connects the stomach and colon and
the development of robust research models, as well as lies anterior to the intestines29. Increased waist circum-
increasing support for public health policies to prevent ference is a proxy measure of VAT and is a predictor of
and treat obesity18. elevated health risk in individuals who are overweight
A clear need therefore exists for scientific knowl- (BMI 25.0–29.9 kg/m2) or obese (BMI ≥30 kg/m2)30.
edge that aligns compelling epidemiological associ- Individuals with increased VAT deposits are more
ations between obesity and gastrointestinal cancer likely to have hypertension, type 2 diabetes mellitus
with a better understanding of the altered biological (T2DM) and dyslipidaemia, features that characterize
processes that occur in the context of obesity within the metabolic syndrome and are also associated with
adipose tissue, the microenvironment of preneoplastic a risk of cardiovascular disease30,31. Although the epi­
and tumour tissues, and immunoregulatory organs, in demiological evidence for the association between waist
particular the liver. However, research in this area is circumference and cancer incidence is limited by the
in its infancy. One focal point for research should be the absence of prospective cohort studies, there is an asso-
tumour microenvironment (TME), the cellular environ- ciation between increased abdominal obesity, measured
ment in which a tumour exists. The TME includes blood by waist circumference, and the incidence of colorectal
vessels, immune cells, fibroblasts, bone-​marrow-derived cancer32, as well as obesity-​related cancers combined33.
inflammatory cells, signalling molecules and the extra- An important aspect unique to VAT is its anatomical
cellular matrix (ECM) (Fig. 1). The key processes linking link to the liver via portal vein drainage, and the VAT–
obesity and cancer that are commonly dysregulated in liver axis is thought to account for the greater influence
both adiposity and in the TME include inflammation, that VAT has on insulin resistance and inflammation
hypoxia, energy metabolism, angiogenesis and epithe- compared with subcutaneous adipose tissue34. The
lial to mesenchymal transition (EMT)19–23. The effect of so-​called portal hypothesis describes how the liver is
obesity on the TME can occur both at the local level directly exposed to increasing amounts of free fatty
and systemically via circulating inflammatory, growth acids and pro-​inflammatory factors released from VAT
factor and metabolic mediators associated with adipose directly into the portal vein of individuals with obesity,
tissue inflammation. ultimately increasing the risk of liver inflammation and
Although the broad scientific principles that link nonalcoholic fatty liver disease (NAFLD), as well as
obesity and cancer apply across many tumour sites, promoting insulin resistance35.

www.nature.com/nrgastro
G A S T R O I N T E S T I N A L C A N C E R A N D OR BE v
E ISEIw
TYS

Table 1 | relative risk of gastrointestinal cancer at high BMI low-​grade chronic inflammation in neighbouring and
distal organs (Fig. 2).
site relative risk of 95% CI strength of
the highest BMI evidencea
category versus Obesity and the hallmarks of cancer
normal BMI The modern molecular understanding of cancer was
proposed in 2000 by Douglas Hanahan and Robert
Oesophageal 4.8 3.0–7.7 Strong
adenocarcinoma Weinberg46, with six original hallmarks that included
sustained angiogenesis, invasion and metastasis, growth
Colon 1.3 1.3–1.4 Strong in men,
suggestive in women and replication, self-​sufficiency, resistance to anti-​
growth signals and the evasion of apoptosis46. An update
Gallbladder 1.3 1.2–1.4 Strong
published in 2011 added the emerging hallmarks and
Pancreas 1.5 1.2–1.8 Strong enabling characteristics of genomic instability, tumour-​
Liver 1.8 1.6–2.1 Not reported promoting inflammation, reprogramming of energy
Gastric cardia 1.8 1.3–2.5 Suggestive
metabolism and the evasion of immune destruction47.
It is now accepted that it is not only the functional
Rectum 1.3 1.3–1.4 Suggestive capabili­ties of cancer cells that define a cancer but also the
Breast: postmenopausal 1.1 1.1–1.2 Strong influence of its microenvironment, and in this context,
Corpus uteri 7.1 6.3–8.1 Strong inflammation provides a key added dimension; a so-​
called seventh hallmark48. Inflammation, first mooted
Ovary 1.1 1.1–1.2 Suggestive
as a factor relevant to carcinogenesis in 1863 by
Renal cell carcinoma 1.8 1.7–1.9 Strong Rudolf Virchow, is now widely accepted as predispos-
Meningioma 1.5 1.3–1.8 Not reported ing to cancer, whether mediated by extrinsic or intrinsic
Multiple myeloma 1.5 1.2–2.0 Strong influences48. It can lead to tissue damage and genetic
instability and can disrupt homeostasis within the
Thyroid 1.1 1.0–1.1 Suggestive
immune system and tissue microenvironments.
Data from a meta-​analysis reported by the International Agency for Research on Cancer
(IARC) working group1. aThe strength of evidence is reported from an umbrella meta-​analysis2
The simplest scientific explanation for the asso-
with quality determined as strong when P < 10−6 from a meta-​analysis random effects model ciation of obesity with an increased risk of cancer of
(>1,000 cancer cases) and suggestive if P < 10−3 in a random effects model (>1,000 cancer any type is that the microenvironment of precancer
cases). Cancer sites such as the oesophagus, colon, gallbladder, pancreas, breast, corpus uteri
and renal cell carcinoma are often referred to as obesity-​related cancers, distinguishing them or cancer is altered and enabled by obesity-​associated
from tumours at other sites that have not been found to be associated with obesity. inflammation, particularly when obesity is associated
with dysmetabolism and characterized by the metabolic
Within VAT, adipocytes comprise ~90% of the adi- syndrome, insulin resistance or T2DM49. Inflammation
pose tissue volume, but adipose tissue is also rich in linked to obesity might be fuelled by systemic adi-
other cells from the stromal–vascular fraction, includ- pokines, pro-​inflammatory cytokines or endocrine and
ing innate and adaptive immune cells, endothelial cells, metabolic influences, as well as pleiotropic influences
peri­cytes, fibroblasts and stem and progenitor cells36–39. within the TME, such as the modulation of stemness
VAT has more pleiotropic properties influencing car- and the promotion of EMT19. Although 10–50% of
cinogenesis than subcutaneous fat, including differ- individuals with obesity might remain free of metabolic
ent gene expression profiles of adipocyte stem cells40. abnormalities and be ‘metabolically healthy’ (refs50–52),
Within adipose tissue, the increased number and altered these individuals often have chronic systemic inflam-
activation phenotype of immune cells, including mac- mation with elevated circulating levels of inflammatory
rophages41, are an important source of activated inflam- cytokines, such as TNF, IFNγ and C-​reactive protein53,54,
matory pathways in individuals with obesity29, with clear which might have key enabling roles in driving carcino-
links between inflammation and dysregulated metabolic genesis from initiation to progression and invasion
pathways, such as TNF production correlating with or metastasis19,55.
insulin resistance42.
In addition, the serum levels of the adipokine leptin, Antitumour immunity in obesity
which is derived from adipocytes, increase with obe- The link between inflammation and obesity within the
sity and play a central part in controlling appetite and TME should be viewed in the context of disruption of
energy expenditure43,44. Leptin has extensive immuno- the cancer-​immunity cycle, as proposed by Chen and
modulatory roles through the stimulation of inflamma- Mellman56,57, which describes the basic antitumour
tory cytokines including IL-6, TNF, IL-8 and IL-12 from immune response. This cycle begins with maturing
macrophages, dendritic cells, and natural killer (NK) dendritic cells presenting processed tumour antigens on
cells, as well as through fuelling T helper 1 (TH1)-type major histocompatibility complex I (MHC I) or MHC II
T cell responses from the adaptive immune system43. molecules to naive T cells in sentinel lymph nodes, lead-
Leptin is also involved in many pro-​cancer processes, ing to the activation and clonal expansion of effector
including the promotion of cell proliferation, metastasis, T cells56. The activated effector T cells then migrate
anti-​apoptosis and angiogenesis, via pathways synergis- to the tumour in response to chemokine signals and
tic with vascular endothelial growth factor (VEGF)45. infiltrate the tumour. Upon encountering their cognate
Thus, elevated levels of inflammatory cytokines and antigen, cytotoxic CD8+ T cells will kill their targets,
adipokines enter the circulation from VAT in obe- resulting in the release of additional tumour antigens
sity, leading to both local inflammation and systemic for dendritic cells to process, and the cycle begins

Nature Reviews | Gastroenterology & Hepatology


REvIEwS

Tumour microenvironment Obesity-​induced alterations in visceral adipose tis-


Neutrophil or sue immune cell profiles. Although local interactions
granulocyte
between immune cells, adipocytes and tumour cells
Carcinoma cell
occur within the TME, systemic alterations first occur
Adipocyte at an organ level in adipose depots, in particular VAT,
Blood vessel as obesity develops. The normal homeostatic profiles
Macrophage of innate and adaptive immune cell populations within
Fibroblast, VAT and the liver are substantially altered with increas-
myofibroblast ing levels of obesity, either through induced expansion
or MSC in situ or active recruitment from blood, bone marrow
ECM and neighbouring tissues63. In VAT, factors including
T cell lipogenesis, lipotoxicity, polyunsaturated fatty acid
ASC
composition and adipose tissue fibrosis are implicated
in obesity-​associated inflammation through alterations in
Tumour microenvironment in obese state
immune cell subset composition and the production
of cytokines and growth factors, including IL-6, leptin,
VEGF and IGF1 (refs64–67). In lean individuals, VAT
is characterized by low numbers of activated neutro-
phils, mast cells, γδ T cells and a prevalence of M2-type
macrophages, NK cells, invariant NK T (iNKT) cells,
regulatory T cells (Treg cells) and CD4+ TH2 cells, with
IL-4, IL-10 and adiponectin exerting important homeo­
static anti-​i nflammatory functions12,68–72 (Fig.  3). In
humans, excessive caloric intake associated with obe-
sity can result in a several-​fold increase in adipose tis-
sue volume, and as a result, the immune cell profile is
substantially altered with a decrease in CD4+ TH2 cells,
M2-type macrophages and FOXP3+ Treg cells and a con-
comitant increase in M1 macrophages, CD8+ T cells,
NK cells, B cells, mast cells and OX40+ Treg cells68,70,73,74
(Fig. 3). This profile is largely pro-​inflammatory, with a
dominance of TH1 cells, with consequent IFNγ prod­
Fig. 1 | the tumour microenvironment in obesity. The tumour microenvironment within
uction being of considerable importance as a potential
an obese state is characterized by increased numbers of adipocyte-​derived stem cells
driver of macrophage polarization to the classically acti-
(ASCs). There are also increased numbers of fibroblasts and increased extracellular matrix
(ECM) deposition along with an increase in numbers and activation of different immune vated inflammatory M1 phenotype. Other character-
cell populations. ASCs can differentiate into fibroblasts, and this process might be istics of the obesity-​associated profile include elevated
promoted by factors secreted by tumour cells, such as transforming growth factor-​β or IL-6, TNF, IL-1β and leptin and reduced production
platelet-​derived growth factor. Obesity stimulates interstitial fibrosis with alterations in of adiponectin. Overall, the predominant TH2-like
ECM mechanics noted in the tumours of obese mice. ASC-​derived pro-​angiogenic factors anti-​inflammatory immune cell phenotype in healthy,
might also affect tumour vascularity. Tumour vascularity is increased in obese mice versus lean individuals is lost with obesity, and the ensuing
lean controls, particularly in niches rich in ASCs, in which the proliferative index of cells pro-​inflammatory state leads to a dysregulation of
adjacent to vessels and adipocytes is increased. MSC, mesenchymal stem cell. immune homeostasis.

again56. However, in reality, this cycle might be circum- Obesity-​induced alterations in liver immune cell
vented as tumours develop multiple evasion strategies profiles. In obesity, the liver receives increased lev-
and can avoid immunosurveillance owing to physio- els of secreted inflammatory adipokines, particularly
logical characteristics of the TME, including hypoxia, from VAT via the portal vein, as well as associated
acidosis and nutrient deprivation58,59, which all counter growth factors, hormones, cytokines, chemokines and
effective antitumour immunity60. The key to unlock- free fatty acids35. In healthy, lean individuals, the liver
ing the link between obesity and cancer might be an maintains immunological tolerance by producing anti-​
improved understanding of how obesity affects immune inflammatory factors, predominantly IL-10, with selec-
cell function within the TME and the pre-​malignant tive expansion of hepatic Treg cells75 (Fig. 3). Resident
epithelium to both impair antitumour responses and hepatic lymphocytes are diverse and, similar to their
promote inflammatory processes that can affect all the innate counterparts, reflect their unique functional
hallmarks of cancer. In this context, studies in colorec- phenotype, including NK cells, NKT cells, mucosal-​
tal cancer and prostate cancer have demonstrated that associated invariant T (MAIT) cells, γδ T cells, CD4+ and
at the local TME level, tumour cell–adipocyte inter­ CD8+ T cells and B cell populations76–79. Excess visceral
actions within an adipocyte-​rich microenvironment obesity is associated with NAFLD, which can progress
can lead to tumours becoming dependent on adipocyte-​ with hepatic inflammation and immune cell activation
derived excess nutrients, growth factors and inflamma- to nonalcoholic steatohepatitis (NASH), which is a major
tory cytokines, potentially offering novel avenues for risk factor for liver fibrosis and hepatocellular carci-
therapeutic targeting61,62. noma (HCC)80. Increased triglyceride content within

www.nature.com/nrgastro
G A S T R O I N T E S T I N A L C A N C E R A N D OR BE v
E ISEIw
TYS

hepatocytes, fibrosis, increased levels of free fatty acids studies suggest that VAT and the liver are closely linked
and hepatocyte cell death are all thought to result in in a local and systemic milieu of altered immune, inflam-
enhanced local inflammation and immune cell activation matory and metabolic processes that might synergize and,
in the liver of individuals with obesity81,82. As a result, this when combined, provide a compelling mechanistic link to
normally tolerogenic organ contains increased numbers altered metabolism and the immune and inflammatory
of Kupffer cells83, neutrophils84, dendritic cells85, TH17 changes associated with the evident risk of carcinogenesis.
CD4+ T cells86, CD8+ T cells87 and NKT cells88.
In studies of the VAT and liver from patients with Obesity-​induced alterations in circulating immune
oesophageal adenocarcinoma undergoing surgery, parallel cell profiles. In the circulation, the factors associated
alterations in immune and inflammatory pathways were with a systemic inflammatory state in obesity include
identified, including activation of T cells expressing IFNγ leptin, IL-6, IFNγ, TNF, growth factors, visfatin, resistin,
and TNF, as well as activation of TH17 cells70,82,89,90. Such ghrelin, C-​reactive protein and a range of chemokines

Oesophageal adenocarcinoma Gastric cancer


• Increased IL-4, IL-10, IL-1β • Increased leptin, ghrelin,
• NF-κB activation Liver CD4+ TH17 cells, MDSCs
• STAT3 activation
Stomach
Hepatocellular carcinoma
• Increased IGF1, IL-8, IL-6 Colorectal cancer
• Activated RAS–MAPK • Increased IL-1, IL-6,
signalling TNF, IL-8, IL-17A, IL-22
• Dysregulated PI3K and/or • Increased COX2 and
AKT–mTOR signalling VEGF expression
Colon • Increased MAIT cells
and neutrophils

Visceral adipose tissue secretes bioactive Tumour-adjacent adipose tissue functions


compounds into systemic circulation in a paracrine manner
Macrophage recruitment
T cell
Adipocyte infiltration

Tumour
microenvironment

Systemic Tumour
circulation vascular supply
Increased visceral adiposity Pro-tumorigenic environment
• ↓ Adiponectin • ↑ IL-10 • ↑ CCL2 • ↑ Inflammation
• ↑ Leptin • ↑ IFNγ • ↑ Free • ↑ Insulin resistance
• ↓ IL-4 • ↑ TNF IGF1 • ↑ Angiogenesis
• ↑ IL-6 • ↑ Insulin

Fig. 2 | obesity induces a local inflammatory state within the omentum that can drive carcinogenesis at multiple
sites. A local inflammatory state is induced by obesity in the omentum, leading to the secretion of inflammatory cytokines,
chemokines and growth factors into the circulation that then lead to systemic inflammation. This systemic inflammation
affects neighbouring organs, including the liver, through the accumulation of fat and the passive drainage of inflammatory
mediators and potential active migration of immune cells via the portal system, further exacerbating systemic
inflammation through the release of acute phase proteins and inflammatory mediators. Chronic systemic inflammation
also affects organs of the gastrointestinal tract at distal sites including the stomach, colon and oesophagus, through the
activation of resident immune cells and triggering of inflammatory signalling pathways that drive local carcinogenesis.
AKT, protein kinase B; CCL2, CC-​chemokine ligand 2; COX2, cyclooxygenase 2; IGF1, insulin-​like growth factor 1; MAIT,
mucosal-​associated invariant T; MAPK , mitogen-​activated protein kinase; MDSC, myeloid-​derived suppressor cell; mTOR ,
mechanistic target of rapamycin; NF-​κB, nuclear factor-​κB; PI3K , phosphoinositide 3-kinase; STAT3, signal transducer and
activator of transcription 3; TH17 , T helper 17; VEGF, vascular endothelial growth factor.

Nature Reviews | Gastroenterology & Hepatology


REvIEwS

including CC-​chemokine ligand 2 (CCL2) and CCL5 which promotes carcinogenesis through increased
(refs91,92) (Fig. 3). Obesity itself, in the absence of can- mutational rates, angiogenesis and dysregulation in the
cer, induces changes in the frequencies of circulating normal immune cell infiltrate within tissues. Obesity
innate and adaptive immune cells, including a decrease directly leads to alterations in the immune cell pheno-
in activated CD8+ cytotoxic T lymphocytes, Treg cells and type and inflammatory profile within various organs,
Vδ2 T cells68 and an increase in M1-type macrophages, which can affect the ability of the immune system to
CD4+ T cells, B cells and NK cells93. However, changes in control cancer growth. Although substantial progress
circulating immune cells appear dependent on the extent has been made in identifying specific cellular changes
of obesity and the cellular subset examined. induced as a result of obesity within adipose tissue, much
In the cancer context, a better understanding of the research is needed to identify the changes in immune
effect of obesity on systemic immunity is important, par- cells and their functions within the TME as a direct result
ticularly owing to an increasing interest in the prognostic of visceral obesity or tumour-​adjacent adipose depos-
and therapeutic implications of the immune cell infil- its. Developing immune-​based therapies for patients
trate in solid tumours. For example, the Immunoscore, with obesity-​associated cancer presents a unique chal-
which measures the densities of CD3+ and cytotoxic lenge for cancer immunologists: to successfully dampen
CD8+ T cells in the tumour core and in the invasive mar- pathological adipose tissue-​derived inflammation while
gin94,95, has clearly demonstrated the applied potential of simultaneously not interfering with ongoing antitumour
introducing immune cell parameters into cancer clas- immune responses at distal sites.
sification and as a prognostic tool in colorectal cancer.
Whether obesity, with its potential effect on the TME, Adipose-​derived stem cells in the TME
influences the T cell phenotype within tumours and Although the majority of research to date has focused on
response to therapies remains unclear60,65,96,97. the systemic effects of excess adiposity, adipocytes per se
In summary, the appropriate activation of immune are active players in the TME. Adipocytes in tumours are
cells is required to prevent, control and remove malig- in close contact with cancer cells, with a bidirectional
nant cells. Conversely, inappropriate activation of the crosstalk being a putative contributor to carcinogen-
immune system can lead to sustained inflammation, esis98,99 (Fig. 4). In the appropriate context, adipocytes
within the TME, as in VAT and other depots, are a
potential source of both growth factors and signalling
Circulation in obesity molecules, including pro-​inflammatory adipokines and
• Increase in IL-6, IFNγ, TNF, CRP, cytokines, pro-​angiogenic factors and components of the
visfatin, grehlin, resistin,
ECM100–102. Within both fat depots and the TME stroma,
CCL2 and CCL5
• Decrease in activated CD8+ T cells adipocyte progenitors known as ASCs represent approx-
• Increase in M1 macrophages, imately 15–30% of cells in the stromal–vascular fraction
CD4+ T cells, B cells and NK cells of adipose tissue and can be mobilized in response to
chemokines such as CC-​chemokine receptor 2 (CCR2)
Lean liver
• Normal resident Treg cells,
and CXC-​chemokine receptor 4 (CXCR4) ligands from
CD56hi NK cells, the TME103–106. The TME within an obese state is charac­
iNKT cells, CD161+Vα7.2+ Lean omentum terized by increased numbers of ASCs, which might fuel
MAIT cells, Vδ1 and • Enrichment of CD4+ carcinogenesis and tumour progression107. ASCs are
Vδ3 T cells, CD4+ T cells, TH2 cells, iNKT cells, phenotypically similar to bone-​marrow-derived mesen-
CD8+ T cells and B cells regulatory Treg cells and
M2 macrophages chymal stem cells (MSCs), although their proteasome
Obese liver
• Increase in CD14+ Obese omentum
and immunomodulatory ability might differ108. ASCs,
Kupffer cells, • Hypoxia, cellular stress immunologically privileged by lacking MHC II expres-
MOP+ neutrophils, and rupturing hypertrophic sion and lacking a response to co-​stimulatory molecules,
TNF+ dendritic cells, adipocytes might also exert an immunomodulatory role at the inter-
TH17 CD4+ T cells, • Enrichment of CD4+ TH1 face between tumours and immune cells in the TME, in
CD8+ T cells and and CD8+ T cells, NK cells,
CD3+CD56+Vα24+ B cells, M1 macrophages particular through decreased NK cell, cytokine and B cell
iNKT cells and neutrophils responses109. ASCs from patients with obesity compared
with normal weight individuals differ functionally; for
Fig. 3 | Immune cell composition in lean versus obese omentum, liver and blood. example, obesity enhances the ability of ASCs to differen-
Lean adipose tissue of the omentum is characterized by an enrichment of anti-​ tiate into adipocytes with consequent reduction in tissue
inflammatory innate and adaptive immune cells, including CD4+ T helper 2 (TH2) cells, repair capabilities and an upregulation of inflammatory
invariant natural killer T (iNKT) cells, regulatory T cells (Treg cells) and M2 macrophages, genes110,111. In mouse models of diet-​induced obesity,
which maintain local homeostasis.  In the obese omentum, this homeostasis is disrupted ASCs derived from the visceral tissue of humans resulted
owing to hypoxia, cellular stress and rupturing hypertrophic adipocytes that release free in increased growth of tumour cells implanted in the peri­
lipids. As a result, there is an accumulation of pro-​inflammatory immune cells, including toneum112. In other mouse models, ASCs migrate via the
CD4+ TH1 and CD8+ T cells, natural killer (NK) cells, B cells, M1 macrophages and
systemic circulation into tumour sites and can undergo
neutrophils. Accompanying this infiltration of inflammatory cells is the increased release
of pro-​inflammatory cytokines, adipocytokines and chemokines into the blood and differentiation into endothelial cells, pericytes and peri­
directly into the liver via the portal system, which adds to local and systemic tumoural adipocytes, putatively important for neovas-
inflammation and changes in the immune cell populations normally resident in healthy cularization and maintenance of the tumour niche113.
lean liver and blood. CCL , CC-​chemokine ligand; CRP, C-​reactive protein; MAIT, ASC-​derived pro-​angiogenic factors, including hepato-
mucosal-associated invariant T. cyte growth factor and VEGF114,115, might also influence

www.nature.com/nrgastro
G A S T R O I N T E S T I N A L C A N C E R A N D OR BE v
E ISEIw
TYS

TME the ECM, as in cancer, might be permissive of uncon-


trolled tumour cell proliferation or evasion from apopto-
sis. In tumours associated with genetically induced and
Adipokines
Cytokines
diet-​induced obesity, ASCs differentiate into myofibro-
Adipocyte Angiogenic factors blasts, which leads to a stiffer ECM with increased depo-
sition of fibronectin and type I collagen20. Consequently,
increased rigidity results in a reciprocal signalling loop
between the ECM and cancer cells, enhancing invasive-
Differentiation CXCR4 ness126, and this process might occur via mechanotrans-
Chemokines Adipose tissue duction, whereby mechanical stimuli are converted
into chemical signals, promoting cancer progression
CCR2
ASC and invasion20.
A dense and stiff ECM results in changes in cytokine
signalling, epithelial cell morphology with a loss of polar-
Bone marrow ity and stem cell differentiation125,127,128. Receptors such as
periostin seem to be required for tissue to be enriched
Bone with cancer stem cells129, creating a tumorigenic niche
that is thought to be important for the development
Fig. 4 | Interplay between adipocytes, stem cells and the tumour microenvironment. of metastatic disease. CAFs are pro-​inflammatory; for
Chemokines produced in the tumour microenvironment (TME), such as CXC-​chemokine example, CAFs from a mouse model of squamous skin
ligand 1 (CXCL1), CXCL8, CXC-​chemokine receptor 4 (CXCR4) and CC-​chemokine carcinoma have a pro-​inflammatory genetic signature,
receptor 2 (CCR2), result in the recruitment of adipose-​derived stem cells (ASCs) from mediated via nuclear factor-​κB (NF-​κB)130. In gastric
adipose tissue, as well as potentially from the bone marrow107. ASCs act both as a source cancer, an inflammatory IL-6–STAT3 (signal transducer
of adipocytes (upon differentiation) and as a source of adipokines, cytokines and and activator of transcription 3) signalling cascade and
angiogenic factors, which in turn might influence the TME. Adipocytes also act in a associated MSCs result in the activation of neutro-
paracrine manner to secrete factors that favour tumour progression by promoting phils, and crosstalk between MSCs and neutrophils
mechanisms such as angiogenesis, tumour fibrosis, altered extracellular matrix
favours transformation of MSCs into CAFs131. Thus,
mechanics and mechanosignalling103,246,247.
the stroma is a key player in creating the environmen-
tal conditions that support the survival of tumour cells,
tumour vascularity, which is increased in obese mice with obesity promoting an environment conducive to
compared with lean controls, particularly in niches rich tumour development.
in ASCs, where the proliferative index of cells adjacent
to vessels and adipocytes is augmented116. Epithelial to mesenchymal transition. Another impor-
tant link between ASCs, the adipose microenvironment
Fibroblasts and the extracellular matrix in the TME. and the TME is the promotion of EMT in the TME,
Within the TME, ASCs promote fibrosis and provide which results in an invasive tumour phenotype that
a source of cancer-​associated fibroblasts (CAFs)117,118 might also be relatively chemoresistant132–134. EMT is the
(Fig. 5). CAFs are the dominant cell type in the stroma process by which epithelial cells acquire mesenchymal-​
surrounding tumour cells119. CAFs are characterized like properties — an increased ability to migrate, evade
by the expression of α-​smooth muscle actin (α-​SMA), apoptosis and produce ECM components — and is
neural/glial antigen 2 (NG2), tenascin C and platelet-​ hypothesized to be a key method by which tumour cells
derived growth factor receptor α (PDGFRα), and they acquire the ability to metastasize135. At the time of EMT,
might be associated with a more aggressive tumour bio­ there is a loss of apical–basal polarity and cell–cell junc-
logy and poorer survival119–121. For example, in a study tions in epithelial cells, with cytoskeletal reorganization
of 183 patients with oesophageal adenocarcinoma who and associated changes in signalling and gene expres-
underwent surgery, 93% of tumours contained CAFs, sion132. The process of EMT generally progresses through
and their density correlated with poor survival (HR 7.1, conserved pathways with minor variations according to
95% CI 1.7–29.4, P = 0.0016)122. The ECM or intra-​ tissue type136. Triggering factors for EMT are complex,
tumour stroma content is also relevant. For instance, in and although transcription factors drive EMT, several
stage II and III colon cancer, it is an independent predic- master regulators are involved in the process, includ-
tor of overall survival, with overall and disease-​free sur- ing epigenetic changes and microRNAs137. Other fac-
vival times being substantially lower in patients whose tors within the TME promoting EMT include hypoxia,
tumours have a high stromal content, defined as >50%123. acidosis, leptin and low glucose 138. Inflammatory
In mice, obesity stimulates interstitial fibrosis medi- cytokines fuel EMT in several different cancer cell lines,
ated by fibroblasts, with alterations in the ECM mechan- including HCC139. For example, in a mouse model of
ics. For example, a fibrotic phenotype has been reported HCV-​associated HCC, diet-​induced obesity increased
in pancreatic neoplasia in a mouse model124 as well as in inflammatory signalling via STAT3, and this finding was
a study of patients with oesophageal cancer122. The ECM associated with larger tumours with a cancer-​stem-cell-​
undergoes continuous remodelling, and the dynamics of like phenotype140. Oesophageal adenocarcinoma cell
the process are thought to be crucial to the normal func- lines co-​cultured with fat taken from patients with obe-
tion of tissue by providing an organizational platform for sity resulted in the upregulation of EMT programmes141
cellular migration and cell growth125. Disorganization of and expression of matrix metalloproteinases (MMPs)142.

Nature Reviews | Gastroenterology & Hepatology


REvIEwS

Differentiation From a therapeutic viewpoint, treatment of obesity-​


ASC MSC induced pancreatic tumours in mice using an angioten-
sin 1 inhibitor, losartan, reduced mechanical stress on
the tumour cells and decreased tumour growth124. In a
study comparing ASCs from lean adults and those with
CAF
obesity, ASCs from the cohort with obesity expressed
higher levels of CAF markers, suggesting that obesity
• Growth factors resulted in more rapid conversion of ASCs to CAFs
(TGFβ, HGF, IGF) within the TME99. Although conceptually appealing, the
• Angiogenesis factors role of ASCs and CAFs is dynamic and remains incom-
(VEGF, PDGF, HGF)
• Inflammatory factors pletely understood, and it might be context-​dependent,
(IL-6, IL-1, ATP) as shown by studies with HCC153,154, colon cancer155
ECM deposition • Hypoxia and pancreatic cancer156 cell lines, in which MSCs were
and remodelling (HIF1α, LDH) found to both promote and inhibit cancer growth157.
Moreover, the specific link between VAT, the TME and
Altered immune Tumour cell altered tumour biology is unclear. In addition, it should
environment
be acknowledged that the majority of studies to date
on ASCs and cancer have been in non-​gastrointestinal
↑ Proliferation ↑ Motility Angiogenesis tumour sites103.

Processes linking obesity and the TME


Fig. 5 | the role of cancer-​associated fibroblasts in the tumour microenvironment. Inflammation. Although inflammation in the TME can
Cancer-​associated fibroblasts (CAFs) are stromal cells of the tumour microenvironment be influenced by systemic factors derived from VAT and
(TME) that are derived from mesenchymal stem cells (MSCs) and also potentially from fatty liver, as already discussed, tumour-​adjacent adi-
adipose-​derived stem cells (ASCs)106,108. CAFs might be influenced by ASCs via
pose tissue might also act as an energy source, supply­
bidirectional crosstalk , whereby ASCs migrate to the TME under the influence of factors
produced by CAFs, and ASCs cause the activation of resident quiescent fibroblasts into ing non-​esterified fatty acids and growth factors not
CAFs. The pro-​tumorigenic effects of ASCs (secretion of growth factors, angiogenic only for tumour cells but also for immune and stromal
factors, inflammatory factors and hypoxia factors) might be mediated via CAFs rather cells within the TME158,159. The majority of studies on
than directly by ASCs. CAFs secrete and remodel extracellular matrix (ECM), but they peritumoural fat have been in breast cancer, but many
also secrete numerous factors, such as growth factors and pro-​angiogenic factors, tumour types, including gastric cancer and colon cancer,
which facilitate increased motility and proliferation of tumour cells as well as increased develop in close contact with adipose tissue, suggesting
tumour angiogenesis109,248,249. CAFs contribute to an inflammatory and metabolically that dysregulated peritumoural adipocytes affect the
dysregulated TME through the production of hypoxia-​inducible factor 1α (HIF1α) and TME through the release of local inflammatory medi­
lactate dehydrogenase (LDH), as well as inflammatory cytokines that lead to effects on ators158. For example, Trevellin et al. highlighted a poten-
the local immune cell infiltrate and on tumour cells. CAFs, perhaps with infiltrating
tial role for peritumoural adipose tissue in promoting
immune cells, might also play a role in the development of tumour fibrosis, a factor
thought to favour the persistence of cancerous epithelial cells through the production of lymph node metastasis in oesophageal adenocarcinoma,
growth factors or ECM components. Mouse models of obesity demonstrate increased with human adipocyte-​derived leptin influencing the
fibrosis within the TME and altered ECM mechanics that, in turn, influence local tumour expression of the EMT regulatory genes α-​SMA and
cells20,112. HGF, hepatocyte growth factor ; IGF, insulin-​like growth factor ; PDGF, platelet-​ E-cadherin and promoting metastasis in the oesophageal
derived growth factor; TGFβ, transforming growth factor-​β; VEGF, vascular endothelial adenocarcinoma cell line OE33 (ref.160).
growth factor. The robust epidemiological data linking obesity and
carcinogenesis suggest that obesity is permissive to can-
ASCs (or MSCs) from the stroma of adipose tissue cer development in pre-​malignant disease, which has
can induce EMT-​like changes in cancer cell lines143,144 led to studies in experimental models and in patients
including colon cancer145 and pancreatic cancer146. The with gastrointestinal pre-​malignant conditions, such
adipokine leptin has also been found to mediate as Barrett oesophagus, Helicobacter pylori-​associated
the in vitro induction of EMT markers by ASCs from gastritis, NAFLD or NASH, and pancreatic intraepi-
donors with obesity in breast cancer models 147,148. thelial neoplasia. Rebours et al., for instance, evaluated
Nonetheless, the link between an EMT–inflammation obesity and pancreatic intraepithelial neoplasia and
axis and cancer progression in the context of obesity reported significant correlations of the lesions with
remains unclear and inconsistent149. Furthermore, some VAT (P = 0.02), pancreatic fatty infiltration (extralobular
studies demonstrate that the majority of cells within (P = 0.01) and intralobular (P < 0.0001)) and intralobular
metastases are not derived from those that undergo fibrosis (P = 0.003)161.
EMT but rather are derived from circulating epithelial Additionally, hepatic steatosis is thought to sensi-
cells133,134, which might enter the circulation through tize the liver parenchyma in humans for progression to
collective epithelial cell migration150,151 or tumour frag- NASH and HCC as a result of multiple subsequent hits,
mentation152 and survive even without association with including dysregulated adipokine production, lipotoxic-
other cells of the TME and in the absence of mesenchy- ity, insulin resistance, oxidative stress, generation of reac-
mal markers. In this context, whether EMT is a common tive oxygen species (ROS), defective mitochondrial ATP
source of metastasis or merely a marker of metastatic activity and hypoxia162,163. Oxidative stress in the liver
potential requires future study, as does the influence of results in increased IL-8 expression and the chemoat-
obesity on the EMT–inflammation phenotype. traction of neutrophils, which can further exacerbate

www.nature.com/nrgastro
G A S T R O I N T E S T I N A L C A N C E R A N D OR BE v
E ISEIw
TYS

inflammation164,165. Hyperinsulinaemia and enhanced in the serum of patients than in healthy individuals
bioavailability of IGF1 associated with obesity lead to (9.3 ± 2.1 pg/ml versus 4.4 ± 0.8 pg/ml). Moreover, IL-6
the downstream activation of insulin receptor substrate 1 levels correlate with tumour grade and are associated
(IRS1), which activates a range of signalling path- with a poorer overall survival184,185. IL-1β can induce the
ways including p53, mitogen-​activated protein kinase production of other inflammatory, chemoattractant and
(MAPK) and phosphoinositide 3-kinase–protein pro-​angiogenic factors such as IL-6, TNF, IL-8, cyclooxy­
kinase B (PI3K–AKT) that might induce proliferation genase 2 (COX2) and VEGF, which are all reported to
and anti-​apoptotic pathways relevant to both liver can- have a role in colorectal cancer pathogenesis186. These
cer and carcinogenesis at other sites166,167. Leptin is both factors are abundant within human obese VAT and are
pro-​inflammatory and a mitogen, with anti-​apoptotic released into the circulation where they can potentially
and pro-​angiogenic properties within the TME through influence tissue homeostasis within the lower gastrointes-
Janus kinase (JAK)/STAT-​dependent mechanisms168. In tinal tract. In addition, conditioned media from the obese
the liver, leptin is also reported to have a role in driving adipose tissue of patients with oesophageal, gastric and
key features of NASH, including fibrosis, local inflam- colorectal cancer significantly (P < 0.05) enhance the pro-
mation and angiogenesis169,170, and it also results in HCC liferation of colorectal and oesophageal adenocarcinoma
cell cycle progression, inhibition of endoplasmic reticu- cells in vitro compared with conditioned media from
lum stress-​associated apoptotic pathways and the induc- lean adipose tissue of patients with or without cancer,
tion of autophagy, resulting in decreased apoptosis171. with VEGF and IL-6 among many factors implicated187.
STAT3 activation driven by IL-6 and leptin enhances Figure 2 summarizes the connection between obesity and
HCC development through the promotion of tumour the induction of the inflammatory state within visceral fat
cell proliferation and inhibition of apoptosis172. In mouse and its influence on different gastrointestinal cancer sites.
models of HCC, including Mdr2 or Tsc1 knockouts, key
inflammatory mechanisms including NF-​κB activation, Hypoxia. Hypoxia in VAT and in the TME is a key con-
endoplasmic reticulum stress and ROS accumulation tributor to the local and systemic inflammation charac-
have been reported as drivers of carcinogenesis55,173,174. teristic of obesity and constitutes a potential link between
Barrett oesophagus and oesophageal adenocarci- the adipose microenvironment and the TME179,188,189.
noma are strongly associated with acid, bile reflux and Adipose tissue depots become hypoxic in obesity as adi-
obesity, and pro-​inflammatory changes in immune cell pocytes grow and exceed the diffusion capacity of oxygen,
phenotypes are increased in VAT, liver and oesophageal and this effect can initiate and promote an inflammatory
tissue from patients with these diseases70,82,89,90. In Barrett process involving macrophage recruitment, enhanced
oesophagus, the pathological precursor to oesophageal adipokine secretion, increased glucose utilization, insulin
adenocarcinoma, visceral obesity is an independent pre- resistance and lactate production190–193.
dictor of progression from the adaptive epithelium of In the TME, hypoxia induces a metabolic switch
intestinal metaplasia to high-​grade dysplasia or adeno­ from oxidative metabolism to anaerobic glycolysis (dis-
carcinoma (P < 0.0001)175. A murine model using a cussed further in the next section)194. A key process is
human IL-1β transgene that mirrors human oesophagitis, the activation of hypoxia-​inducible factor 1α (HIF1α)
metaplasia, dysplasia and adenocarcinoma progression in response to hypoxia. HIF1α promotes angiogenesis,
was developed by Quante and colleagues176. This model neovascular­ization and sustained inflammation193,195–197.
is dependent on IL-1β expression in the oesophagus and HIF1α expression in tumours is also associated with a
on systemic IL-6. In this context, both IL-1β and IL-6 are poor prognosis and a reduced response to standard ther-
elevated in the adipose tissue and blood of individuals apies. For instance, in human HCC, co-​overexpression of
with obesity, suggesting a possible link in the progression programmed cell death 1 ligand 1 (PD-​L1) and HIF1α
of Barrett oesophagus to cancer177–179. In addition, IL-6 in tumour tissue was associated with an increased risk
and leptin both drive STAT3 activation, the constitutive of recurrence or metastasis and death, suggesting a
activation of which plays a key role in carcinogenesis of benefit from HIF1α inhibitors in conjunction with
numerous cancers, including oesophageal adenocarci- PD-​L1 blockade198–200.
noma, by inducing tumour cell proliferation, survival and Hypoxia also increases expression of the obesity-​
invasion, promoting inflammation via NF-​κB activation related pro-​inflammatory cytokine IL-32 in VAT in
and suppressing antitumour immunity180,181. patients with colon cancer and obesity201. IL-32 expres-
The inflammation-​to-cancer model for gastric cancer sion has been correlated with metastatic potential and
occurs via H. pylori infection and chronic gastritis182. In the regulation of cancer-​stem-cell-​like properties, and it
a murine model of H. pylori-​driven gastric adenocarci- is a potent inducer of prostaglandin E2 in colorectal can-
noma using Helicobacter felis, obesity induced by a high-​ cer201,202. In the TME of hypoxic colorectal cancer tumours,
fat diet resulted in significantly increased inflammation, HIF1α increases the transcription of the cell-​surface glu-
metaplasia and dysplasia in infected mice through cose transporter GLUT1, inducing key glycolytic enzymes
increased recruitment of myeloid-​derived suppressor such as lactate dehydrogenase, and inhibits acetyl-​CoA
cells (MDSCs) and TH17 cells in the stomach183. In addi- formation and oxidative phosphorylation197,203.
tion, increased macrophages were observed in adipose Hypoxia and angiogenesis are closely linked, and
tissue of obese mice with Helicobacter infection but not HIF1α induces angiogenic growth factors including
in Helicobacter-​infected lean mice183. leptin, VEGF, MMP2 and MMP9204. Specifically, in
In colorectal cancer, a prominent role for IL-6 is gastric carcinomas, MMP2, MMP9 and VEGF largely
well documented, and higher levels of IL-6 are detected contribute to the angiogenic profile of the tumour and

Nature Reviews | Gastroenterology & Hepatology


REvIEwS

disease progression205. The process of angiogenesis plays can promote oxidative phosphorylation and dampen
an instrumental role in adipose tissue homeostasis; for glycolysis in cells; disruption of this balance is asso-
instance, adipose tissue produces angiogenic factors ciated with mutations in p53 and oncogenic transfor-
including VEGF206,207, and inhibition of adipocyte differ- mation224. In human Barrett oesophagus, for instance,
entiation reduces angiogenesis208–210. Studies performed in indirect measurements of hypoxia by HIF1α in the
diet-​induced obese mice or in ob/ob mice that cannot pro- tissue microenvironment correlate with p53 status and
duce leptin showed that inhibition of VEGF substantially energy metabolism, specifically oxidative phosphoryl-
reduces weight gain211,212, which highlights the impor- ation, and are associated with visceral obesity225. These
tance of angiogenesis regulation in the adipose micro­ studies suggest that HIF-1α is a useful biomarker to help
environment. In this regard, VEGFR1–PGF (placental profile metabolic activity in tumours and has potentially
growth factor) signalling appears to be involved in the important therapeutic implications.
link between obesity and angiogenesis, both in the TME A further intriguing dimension in the TME is the
and adipose tissue microenvironment, suggesting cognate parallel metabolic changes, particularly aerobic glycoly-
influences213. Leptin, similarly to VEGF, is also induced by sis, in both tumour cells and activated immune cells such
hypoxia or HIF1α and promotes angiogenesis214,215. Leptin as M1 macrophages and pro-​inflammatory T cells226,227.
activity is mediated through the leptin receptor (ObR), The metabolic switch in tumour and immune cells
and expression of both can be stimulated by hypoxia, with in the TME rapidly provides ATP and intermediates
increased leptin levels and ObR and HIF1α expression for the biosynthesis of nucleotides, lipids and proteins
observed in human colorectal cancer, highlighting that as well as certain tricarboxylic acid intermediates that
overexpression of the leptin pathway is tightly connected can activate HIF1α228. Consequently, there is a parallel
with the abundance of HIF1 (refs216,217). In oesophageal metabolic shift not only in tumour cells but also in the
adenocarcinoma, obesity has also been associated with surrounding stromal compartment of the TME, which
an upregulation of the ObR receptor within the tumour includes immune cells, T cells and macrophage infiltrate.
tissue, as well as with advanced tumour stage218. In human Barrett oesophagus, increasing adiposity was
The development of hypoxia within both expanded positively associated with glycolysis and negatively
obese adipose tissue and the TME results in similar pro-​ associated with oxidative phosphorylation within the
inflammatory responses, which can fuel and sustain the metaplastic oesophageal tissue225. In addition, condi-
carcinogenic process. A lack of oxygen forces all cells, tioned media from VAT from patients with obesity and
including adipocytes, tumour cells and immune cells, to oesophageal adenocarcinoma significantly (P < 0.05)
adapt their metabolic process to survive. Consequently, altered mitochondrial biology and energy metabolism
this interrelationship should be a key focus of research profiles in oesophageal cancer cell lines, and metabo-
on obesity-​related carcinogenesis. lomic profiling revealed elevated lactate production as
a consequence of increased glyco­lysis229. Expression
Metabolism. Cellular metabolism, in particular the of a glycolytic marker, pyruvate kinase M2 (PKM2),
balance between oxidative phosphorylation and glyco­ which catalyses the final rate-​limiting step of glycoly-
lysis, is also a common theme that links the adipose sis and stimulates HIF1α, was also positively associated
microenvironment and the TME and is relevant to not with obesity229. This finding might be of therapeutic
only tumour and pre-​malignant cells but also immune relevance, as PKM2 also plays a role in regulating mac-
cells. Glycolysis is stimulated by hypoxic conditions, rophages, T cells and dendritic cells in the TME and is
and common mutations in many cancers, such as in the required for the expression of the immune checkpoint
p53-encoding gene TP53, support glycolytic pathways PD-​L1 in immune cells and in CT26 murine colon
by promoting GLUT1 translocation to the plasma mem- tumours, highlighting its potential value as a marker
brane219. The Warburg effect is a metabolic phenotype in associated with tumour metabolism that is relevant to
cancer cells referring to a metabolic shift in ATP gener­ obesity-​associated cancer230.
ation from oxidative phosphorylation to glycolysis, even With the enormous interest in understanding the
under aerobic conditions220. This effect is observed in immune microenvironment of tumours, both for prog-
routine oncological clinical practice with 18fluorodeoxy­ nostication and targeting with immunotherapeutic
glucose (FDG) PET in combination with CT, which is agents, the potential to target metabolism within the TME
utilized to optimally stage many cancer types221. Within is an attractive therapeutic concept for all cancers, with
the TME, energy production from stromal cells can perhaps greater potential gain in the context of obesity, in
reverse the Warburg effect, a process known as metabolic which altered metabolic processes relevant to T2DM and
symbiosis, in which lactate from a hypoxic glycolytic cardiovascular disease might already exist. For example,
tumour can be a metabolite for oxidative phosphoryl- the insulin-​sensitizing drug metformin, which targets
ation and ATP generation222. The survival-​promoting mechanistic target of rapamycin complex 1 (mTORC1)
effects of fat cells, through pathways that are currently signalling via AMP-​activated protein kinase (AMPK)-
unknown but possibly via metabolic symbiosis, might dependent and AMPK-​independent pathways, might
play a role in cancer progression223. reduce tumour cell growth and impair mitochondrial
The gene TP53, which is commonly mutated in chro- function and energy metabolism, with the greatest effects
mosomally unstable tumours including gastrointestinal observed in the setting of obesity231. Further understand-
cancers, represents an interesting link with metabolism ing of the interactions between metabolites in the obese
and hypoxia, responding to challenging conditions in TME and their oncogenic nature will provide insight
the tissue microenvironment224. It is proposed that p53 into the development of novel antimetabolic therapies

www.nature.com/nrgastro
G A S T R O I N T E S T I N A L C A N C E R A N D OR BE v
E ISEIw
TYS

Adipose tissue
Obese state T cell infiltration
Inflamed
adipose tissue M1 macrophage

Lean state

M2 macrophage
Blood vessel Peritumoral adipose tissue
• ↑ Leptin • ↑ Angiogenic factors
• ↓ Adiponectin • ↑ Metabolic mediators
• ↑ Inflammatory mediators • ↑ Resistin

Mitochondria

Altered tumour metabolism Tumour hypoxia Active tumour angiogenesis


• Shift to glycolytic metabolism: Warburg • ↑ HIF1α • ↑ VEGF • VEGF • ANG1
• ↑ PKM2 • ↑ Metabolic genes • ↑ Leptin • VEGFR1 • ANG2
• ↑ Mitochondrial mass • ↑ IL-32 • VEGFR2 • ↑ CD31: blood vessel numbers

Altered tumour metabolism, hypoxia and dysregulated tumour angiogenesis

Altered tumour microenvironment drives disease progression and alters treatment responses in tumours

Fig. 6 | links between obese adipose tissue and cellular process in the tumour microenvironment. The obese state
leads to inflamed adipose tissue that stimulates the secretion of many inflammatory , metabolic and angiogenic mediators
from adipocytes and immune cells, in addition to modulating leptin and adiponectin levels. These mediators can, in turn,
stimulate the upregulation and dysregulation of a number of key cellular processes in the tumour micro­environment,
including altered energy metabolism, hypoxia and dysregulated angiogenesis, which all promote a tumour microenvironment
favourable for driving disease progression and influencing treatment response. ANG, angiopoietin; HIF1α, hypoxia-​
inducible factor 1α; PKM2, pyruvate kinase M2; VEGF, vascular endothelial growth factor ; VEGFR , vascular endothelial
growth factor receptor.

that target the obese TME. Figure 6 illustrates how the in the loss of 4% of body weight; however, bariatric sur-
obesity state can influence the different cellular processes gery generally leads to between 15% and 30% weight loss,
in the TME, specifically hypoxia, energy metabolism and maintained usually for at least 3 years and associated with
angiogenesis, three processes that are interlinked. a high rate of resolution of dysmetabolism, particularly
T2DM232–237. Intriguingly, for cancer, although no rand-
Targeting obesity to prevent cancer omized data exist, there is considerable indirect evidence
If the obesity and cancer paradigm is founded on con- from observational studies that weight loss in the popu-
vincing epidemiological data, and the science discussed lation with obesity reduces cancer risk. In a prospective
herein is increasingly consistent in linking the adipose observational study from Sweden, the Swedish Obese
and tumour microenvironments, an obvious question is Subjects (SOS) study, which included 2,010 patients with
whether a reverse paradigm exists: does treatment of obe- obesity who underwent bariatric surgery, and a well-​
sity or key associated metabolic and endocrine pathways defined matched comparison group of 2,037 individuals
reduce cancer risk? Clinical prospective studies or rand- with obesity who received standard care with no specific
omized clinical trials to address this question, requiring intervention, there were fewer overall cancer diagnoses
many thousands of patients and years of follow-​up, might in the group who underwent surgery (HR 0.67, 95%
not be feasible, and most indirect evidence is provided CI 0.53–0.85, P = 0.0009). The effect was primarily
from long-​term follow-​up of patients who have had sur- observed in women, but the study was too small to inves-
gery for morbid obesity, so-​called bariatric surgery, or tigate site-​specific changes or to assess the effects on men,
metabolic surgery when performed for obesity compli- who comprised 29% of the cohort238. In another study
cated by T2DM. Diet and lifestyle changes usually result from Utah, in the United States, wherein men represented

Nature Reviews | Gastroenterology & Hepatology


REvIEwS

14% of the study population, individuals with obesity Most scientific research exploring the association
who received bariatric surgery were compared with con- of cancer and obesity originates from mouse models
trols with obesity who were recruited from the state driver or indirectly in humans where inflammatory mark-
licence database239. A risk reduction was observed for all ers are studied. One systematic review in 2017 of the
cancers (HR 0.76, 95% CI 0.65–0.89, P = 0.0006), as well effect of lifestyle interventions on adipose tissue gene
as obesity-​related cancers (HR 0.62, 95% CI 0.49–0.78, expression profiles demonstrated that interventions,
P <0.0001) and cancers in women only (HR 0.73, 95% such as reduced dietary intake and/or increased phys-
CI 0.62–0.87, P = 0.0004). In another large retro­spective ical activity that lead to weight loss, resulted in altered
case–control study using an integrated health insurance gene expression of adipokines and inflammatory
database, matched patients undergoing bariatric sur- markers241, suggesting that weight loss reduces the
gery (n = 22,198) and matched controls not undergo- activity of key inflammation and tumour-​associated
ing surgery (n = 66,427) were followed for a median of transcription factors. In humans, weight loss is asso-
3.5 years240. The incidence of obesity-​related cancers1 ciated with decreased levels of C-​reactive protein, IL-6
(Table 1) was reduced in the bariatric surgery cohort (HR and TNF and reduced activity of tumour-​promoting
0.59, 95% CI, 0.51–0.69, P < 0.001), and the reduction in transcription factors, including NF-​κB, STAT3 and AP1
incidence was greater than for cancers of other sites not (refs55,242–244). Dietary modification might also have an
associated with obesity. Similar to other studies, differ- effect; for example, in obese diabetic mice, adipose
ences were noted only in female patients, who repre- tissue inflammation induced by a high-​fat diet can be
sented 80% of the overall cohort, reflecting the increased prevented by a diet containing n-3 polyunsaturated
prevalence of bariatric surgery in females. For gastro- fatty acids, possibly via reducing the production of
intestinal cancer, surgery was associated with reduced pro-​inflammatory arachidonic acid metabolites, with
colon cancer (HR 0.59, 95% CI 0.36–0.97, P = 0.04) and downstream effects on NF-​κB activation and related
pancreatic cancer (HR 0.46, 95% CI 0.22–0.97, P = 0.04), pathways245. These and other studies suggest that weight
whereas there were no cases of oesophageal adenocarci- loss is an effective strategy to reverse energy imbalance
noma in the surgical group compared with 16 cases in the and white adipose tissue inflammation, with second-
non-​operative control group (P = 0.02)240. ary consequences in the precancer microenvironment
Although some methodological limitations might or the TME that might reduce cancer risk or improve
apply, including unmeasured differences with control cancer outcomes.
groups and perhaps a greater health focus on cancer
diagnosis and prevention, including diet, exercise and Conclusions
screening, the consistency across studies, with the haz- The epidemiological association between obesity and
ard risks remaining proportional over time, suggests many cancers is now firmly established, and there is
causality and supports the hypothesis of a reduction in increasing acceptance that bariatric or metabolic sur-
cancer incidence with a reduction in body mass. It is gery in individuals with morbid obesity might affect
unclear how long the lead time for a potential decrease cancer risk. Notwithstanding this clear and compelling
in obesity-​related cancers would be, and this effect might link, as well as an understanding of how obesity, par-
also reflect a bias in these retrospective studies. It is also ticularly visceral obesity, might affect all the hallmarks
unknown whether the reduction of smaller amounts of of cancer, research is still in its infancy, and to date, no
weight than is usual for bariatric surgery affects can- obesity-​specific marker of risk or specific target has been
cer incidence or whether the duration of excess weight identified. Major unanswered questions are summarized
before weight loss affects cancer risk. Large prospec- in Box 1. It is clear, however, that the factors distinguish-
tive studies with high-​quality longitudinal data will be ing adipocytes and the adipose microenvironment in
required to answer these questions. individuals with obesity, particularly patients who are
metabolically unhealthy, compared with non-​obese
individuals are linked with pro-​inflammatory and pro-​
Box 1 | Unanswered questions in obesity and cancer research
tumorigenic events that have a potential effect in the
• What is the explanation for the profound gender differences in some obesity-​ microenvironment of preneoplastic and neoplastic dis-
associated cancers, in particular oesophageal adenocarcinoma, which predominates ease, affecting all established and emerging hallmarks
in males? of cancer. Parallel processes in obesity and the TME,
• What is the effect of obesity and its aligned tumour microenvironment in response to including hypoxia, altered immune function and adi-
cancer therapies, including chemotherapy, radiation therapy, targeted therapy and pokine and cytokine production, ASCs, CAFs, hypoxia,
immune-​based therapies? metabolism and angiogenesis, represent key areas for
• Does immunotherapy in the context of obesity have a differential response compared future research to unravel the biological mechanisms
with in a non-​obese context? specific to obesity-​associated pathways in carcinogenesis
• Does immunotherapy in the context of obesity exacerbate pro-​inflammatory and to identify potential targets for prevention and ther-
processes at key sites such as the visceral adipose tissue and liver? apy. At this time, however, primary cancer prevention
• Do approaches affecting metabolic pathways in cancer and immune cells, through public health policies to reduce the prevalence
such as metformin, have differential effects in patients with obesity compared with of obesity and secondary prevention through appropri-
non-obese patients? ate management of severe obesity, particularly if associ-
• Does bariatric or metabolic surgery decrease the risk of cancer in patients with ated with systemic inflammation and dysmetabolism,
obesity and high-​risk pre-​malignant conditions, such as fatty liver disease or liver are compelling approaches towards a reduction in the
fibrosis and dysplastic Barrett oesophagus?
incidence and mortality from many cancers that have

www.nature.com/nrgastro
G A S T R O I N T E S T I N A L C A N C E R A N D OR BE v
E ISEIw
TYS

increased commensurate with the obesity epidemic. development and progression. Moreover, in studies of
For the research community, accurately characterizing targeted and immune therapies for obesity-​associated
the obese state based on the site of adipocyte depots cancers, subgroup analysis of the TME in patients with
(such as visceral compared with subcutaneous) as well obesity compared with non-​obese individuals should be
as the associated metabolic health and biomarkers an important component of translational research and
of an adipose tissue inflammatory profile, including, clinical trials.
for instance, CRP, IL-6 and insulin resistance, might
help to identify patients at increased risk of cancer Published online xx xx xxxx

1. Lauby-​Secretan, B. et al. Body fatness and cancer — mammary tumor models. Cancer Prev. Res. 5, the subcutaneous and visceral adipose tissue in
viewpoint of the IARC Working Group. N. Engl. J. Med. 930–942 (2012). humans. PLOS ONE 8, e57892 (2013).
375, 794–798 (2016). 22. Santander, A. M. et al. Paracrine interactions between 41. Lumeng, C. N., Bodzin, J. L. & Saltiel, A. R. Obesity
2. Kyrgiou, M. et al. Adiposity and cancer at major adipocytes and tumor cells recruit and modify induces a phenotypic switch in adipose tissue
anatomical sites: umbrella review of the literature. macrophages to the mammary tumor macrophage polarization. J. Clin. Invest. 117,
BMJ 356, j477 (2017). microenvironment: the role of obesity and 175–184 (2007).
3. NCD Risk Factor Collaboration. Trends in adult body-​ inflammation in breast adipose tissue. Cancers 7, 42. Xu, H. et al. Chronic inflammation in fat plays a crucial
mass index in 200 countries from 1975 to 2014: 143–178 (2015). role in the development of obesity-​related insulin
a pooled analysis of 1698 population-​based 23. Ackerman, D. & Simon, M. C. Hypoxia, lipids, and resistance. J. Clin. Invest. 112, 1821–1830 (2003).
measurement studies with 19.2 million participants. cancer: surviving the harsh tumor microenvironment. 43. Pérez-​Pérez, A. et al. Role of leptin as a link between
Lancet 387, 1377–1396 (2016). Trends Cell Biol. 24, 472–478 (2014). metabolism and the immune system. Cytokine Growth
4. Flegal, K. M., Kruszon-​Moran, D., Carroll, M. D., 24. Cozzo, A. J., Fuller, A. M. & Makowski, L. Factor Rev. 35, 71–84 (2017).
Fryar, C. D. & Ogden, C. L. Trends in obesity among Contribution of adipose tissue to development of 44. Friedman, J. M. & Halaas, J. L. Leptin and the
adults in the United States, 2005 to 2014. JAMA cancer. Compr. Physiol. 8, 237–282 (2017). regulation of body weight in mammals. Nature 395,
315, 2284–2291 (2016). 25. Brahmkhatri, V. P., Prasanna, C. & Atreya, H. S. 763–770 (1998).
5. Arnold, M. et al. Obesity and cancer: an update of the Insulin-​like growth factor system in cancer: novel 45. Lin, T. C. et al. Leptin signaling axis specifically
global impact. Cancer Epidemiol. 41, 8–15 (2016). targeted therapies. Biomed. Res. Int. 2015, 538019 associates with clinical prognosis and is
6. Calle, E. E., Rodriguez, C., Walker-​Thurmond, K. & (2015). multifunctional in regulating cancer progression.
Thun, M. J. Overweight, obesity, and mortality from 26. Pollak, M. Insulin and insulin-​like growth factor Oncotarget 9, 17210–17219 (2018).
cancer in a prospectively studied cohort of U.S. adults. signalling in neoplasia. Nat. Rev. Cancer 8, 915–928 46. Hanahan, D. & Weinberg, R. A. The hallmarks of
N. Engl. J. Med. 348, 1625–1638 (2003). (2008). cancer. Cell 100, 57–70 (2000).
7. Ogden, C. L. et al. Trends in obesity prevalence 27. Crewe, C., An, Y. A. & Scherer, P. E. The ominous triad 47. Hanahan, D. & Weinberg, R. A. Hallmarks of cancer:
among children and adolescents in the United States, of adipose tissue dysfunction: inflammation, fibrosis, the next generation. Cell 144, 646–674 (2011).
1988–1994 through 2013–2014. JAMA 315, and impaired angiogenesis. J. Clin. Invest. 127, 48. Colotta, F., Allavena, P., Sica, A., Garlanda, C. &
2292–2299 (2016). 74–82 (2017). Mantovani, A. Cancer-​related inflammation, the
8. Abarca-​Gómez, L. et al. Worldwide trends in body-​ 28. Vegiopoulos, A., Rohm, M. & Herzig, S. Adipose seventh hallmark of cancer: links to genetic instability.
mass index, underweight, overweight, and obesity tissue: between the extremes. EMBO J. 36, Carcinogenesis 30, 1073–1081 (2009).
from 1975 to 2016: a pooled analysis of 2416 1999–2017 (2017). 49. Iyengar, N. M., Gucalp, A., Dannenberg, A. J. &
population-​based measurement studies in 128.9 29. Meza-​Perez, S. & Randall, T. D. Immunological Hudis, C. A. Obesity and cancer mechanisms:
million children, adolescents, and adults. Lancet 390, functions of the omentum. Trends Immunol. 38, tumor microenvironment and inflammation.
2627–2642 (2017). 526–536 (2017). J. Clin. Oncol. 34, 4270–4276 (2016).
9. Ahrens, W. et al. Prevalence of overweight and obesity 30. Janssen, I., Katzmarzyk, P. T. & Ross, R. Body mass 50. Stefan, N., Haring, H. U. & Schulze, M. B.
in European children below the age of 10. Int. J. Obes. index, waist circumference, and health risk: evidence Metabolically healthy obesity: the low-​hanging fruit in
38, S99 (2014). in support of current National Institutes of Health obesity treatment? Lancet Diabetes Endocrinol. 6,
10. Renehan, A. G. & Soerjomataram, I. Obesity as an guidelines. Arch. Intern. Med. 162, 2074–2079 249–258 (2018).
avoidable cause of cancer (attributable risks). (2002). 51. Hinnouho, G. M. et al. Metabolically healthy obesity
Recent Results Cancer Res. 208, 243–256 (2016). 31. Klein, S. et al. Waist circumference and and risk of mortality: does the definition of metabolic
11. Dignam, J. J. et al. Body mass index and outcomes in cardiometabolic risk: a consensus statement from health matter? Diabetes Care 36, 2294–2300
patients who receive adjuvant chemotherapy for colon shaping America’s health: Association for Weight (2013).
cancer. J. Natl Cancer Inst. 98, 1647–1654 (2006). Management and Obesity Prevention; NAASO, the 52. Phillips, C. M. Metabolically healthy obesity across
12. Liu, J. et al. Genetic deficiency and pharmacological Obesity Society; the American Society for Nutrition; the life course: epidemiology, determinants, and
stabilization of mast cells reduce diet-​induced and the American Diabetes Association. Obesity 15, implications. Ann. NY Acad. Sci. 1391, 85–100
obesity and diabetes in mice. Nat. Med. 15, 1061–1067 (2007). (2017).
940–945 (2009). 32. Dong, Y. et al. Abdominal obesity and colorectal 53. Schmidt, F. M. et al. Inflammatory cytokines in general
13. Maruthur, N. M., Bolen, S., Brancati, F. L. & Clark, J. M. cancer risk: systematic review and meta-​analysis of and central obesity and modulating effects of physical
Obesity and mammography: a systematic review and prospective studies. Biosci. Rep. 37, BSR20170945 activity. PLOS ONE 10, e0121971 (2015).
meta-​analysis. J. Gen. Intern. Med. 24, 665–677 (2017). 54. Iglesias Molli, A. E. et al. Metabolically healthy obese
(2009). 33. Freisling, H. et al. Comparison of general obesity and individuals present similar chronic inflammation level
14. Maruthur, N. M., Bolen, S., Gudzune, K., Brancati, F. L. measures of body fat distribution in older adults in but less insulin-​resistance than obese individuals with
& Clark, J. M. Body mass index and colon cancer relation to cancer risk: meta-​analysis of individual metabolic syndrome. PLOS ONE 12, e0190528
screening: a systematic review and meta-​analysis. participant data of seven prospective cohorts in (2017).
Cancer Epidemiol. Biomarkers Prev. 21, 737–746 Europe. Br. J. Cancer 116, 1486 (2017). 55. Font-​Burgada, J., Sun, B. & Karin, M. Obesity and
(2012). 34. Kabir, M. et al. Molecular evidence supporting the cancer: the oil that feeds the flame. Cell Metab. 23,
15. Sinicrope, F. A., Foster, N. R., Sargent, D. J., portal theory: a causative link between visceral 48–62 (2016).
O’Connell, M. J. & Rankin, C. Obesity is an adiposity and hepatic insulin resistance. Am. J. 56. Chen, D. S. & Mellman, I. Oncology meets
independent prognostic variable in colon cancer Physiol. Endocrinol. Metab. 288, E454–E461 (2005). immunology: the cancer-​immunity cycle. Immunity 39,
survivors. Clin. Cancer Res. 16, 1884–1893 (2010). 35. Item, F. & Konrad, D. Visceral fat and metabolic 1–10 (2013).
16. Reeves, G. K. et al. Cancer incidence and mortality in inflammation: the portal theory revisited. Obes. Rev. 57. Chen, D. S. & Mellman, I. Elements of cancer
relation to body mass index in the Million Women 13 (Suppl. 2), 30–39 (2012). immunity and the cancer-​immune set point. Nature
Study: cohort study. BMJ 335, 1134 (2007). 36. Nishimura, S. et al. CD8+ effector T cells contribute 541, 321–330 (2017).
17. Prizment, A. E., Flood, A., Anderson, K. E. & to macrophage recruitment and adipose tissue 58. Deng, B. et al. Intratumor hypoxia promotes immune
Folsom, A. R. Survival of women with colon cancer in inflammation in obesity. Nat. Med. 15, 914–920 tolerance by inducing regulatory T cells via TGF-​beta1
relation to precancer anthropometric characteristics: (2009). in gastric cancer. PLOS ONE 8, e63777 (2013).
the Iowa Women’s Health Study. Cancer Epidemiol. 37. Yang, H. et al. Obesity increases the production of 59. Huber, V. et al. Cancer acidity: an ultimate frontier
Biomarkers Prev. 19, 2229–2237 (2010). proinflammatory mediators from adipose tissue T cells of tumor immune escape and a novel target of
18. Ligibel, J. A. et al. American Society of Clinical and compromises TCR repertoire diversity: immunomodulation. Semin. Cancer Biol. 43, 74–89
Oncology position statement on obesity and cancer. implications for systemic inflammation and insulin (2017).
J. Clin. Oncol. 32, 3568–3574 (2014). resistance. J. Immunol. 185, 1836–1845 (2010). 60. Anderson, K. G., Stromnes, I. M. & Greenberg, P. D.
19. Olson, O. C., Quail, D. F. & Joyce, J. A. Obesity 38. Weisberg, S. P. et al. Obesity is associated with Obstacles posed by the tumor microenvironment
and the tumor microenvironment. Science 358, macrophage accumulation in adipose tissue. J. Clin. to T cell activity: a case for synergistic therapies.
1130–1131 (2017). Invest. 112, 1796–1808 (2003). Cancer Cell 31, 311–325 (2017).
20. Seo, B. R. et al. Obesity-​dependent changes in 39. Caspar-​Bauguil, S. et al. Adipose tissues as an 61. Amor, S. et al. Peritumoral adipose tissue as a
interstitial ECM mechanics promote breast ancestral immune organ: site-​specific change in source of inflammatory and angiogenic factors in
tumorigenesis. Sci. Transl Med. 7, 301ra130 (2015). obesity. FEBS Lett. 579, 3487–3492 (2005). colorectal cancer. Int. J. Colorectal Dis. 31, 365–375
21. Dunlap, S. M. et al. Dietary energy balance modulates 40. Perrini, S. et al. Differences in gene expression and (2016).
epithelial-​to-mesenchymal transition and tumor cytokine release profiles highlight the heterogeneity of 62. Huang, J. et al. Adipocyte p62/SQSTM1 suppresses
progression in murine claudin-​low and basal-​like distinct subsets of adipose tissue-​derived stem cells in tumorigenesis through opposite regulations of

Nature Reviews | Gastroenterology & Hepatology


REvIEwS

metabolism in adipose tissue and tumor. Cancer Cell and function. J. Leukoc. Biol. 100, 1435–1442 113. Zhang, Y. et al. White adipose tissue cells are recruited
33, 770–784 (2018). (2016). by experimental tumors and promote cancer
63. Saltiel, A. R. & Olefsky, J. M. Inflammatory 90. Kavanagh, M. E. et al. The esophagitis to progression in mouse models. Cancer Res. 69,
mechanisms linking obesity and metabolic disease. adenocarcinoma sequence; the role of inflammation. 5259–5266 (2009).
J. Clin. Invest. 127, 1–4 (2017). Cancer Lett. 345, 182–189 (2014). 114. Rehman, J. et al. Secretion of angiogenic and
64. Lafontan, M. Adipose tissue and adipocyte 91. Cabia, B., Andrade, S., Carreira, M. C., Casanueva, F. F. antiapoptotic factors by human adipose stromal cells.
dysregulation. Diabetes Metab. 40, 16–28 (2014). & Crujeiras, A. B. A role for novel adipose tissue-​ Circulation 109, 1292–1298 (2004).
65. Khan, T. et al. Metabolic dysregulation and adipose secreted factors in obesity-​related carcinogenesis. 115. Zhang, Y., Bellows, C. F. & Kolonin, M. G. Adipose
tissue fibrosis: role of collagen VI. Mol. Cell. Biol. 29, Obes. Rev. 17, 361–376 (2016). tissue-​derived progenitor cells and cancer. World J.
1575–1591 (2009). 92. Maurizi, G., Della Guardia, L., Maurizi, A. & Poloni, A. Stem Cells 2, 103–113 (2010).
66. D’Archivio, M. et al. ω3-PUFAs exert anti-​inflammatory Adipocytes properties and crosstalk with immune 116. Zhang, Y. et al. Stromal progenitor cells from
activity in visceral adipocytes from colorectal cancer system in obesity-​related inflammation. J. Cell. endogenous adipose tissue contribute to pericytes
patients. PLOS ONE 8, e77432 (2013). Physiol. 233, 88–97 (2017). and adipocytes that populate the tumor
67. Caer, C. et al. Immune cell-​derived cytokines 93. Pecht, T., Gutman-​Tirosh, A., Bashan, N. & Rudich, A. microenvironment. Cancer Res. 72, 5198–5208
contribute to obesity-​related inflammation, Peripheral blood leucocyte subclasses as potential (2012).
fibrogenesis and metabolic deregulation in human biomarkers of adipose tissue inflammation and 117. Jotzu, C. et al. Adipose tissue derived stem cells
adipose tissue. Sci. Rep. 7, 3000 (2017). obesity subphenotypes in humans. Obes. Rev. 15, differentiate into carcinoma-​associated fibroblast-​like
68. Donninelli, G. et al. Distinct blood and visceral 322–337 (2014). cells under the influence of tumor derived factors.
adipose tissue regulatory T cell and innate lymphocyte 94. Galon, J. et al. Towards the introduction of the Cell. Oncol. 34, 55–67 (2011).
profiles characterize obesity and colorectal cancer. ‘Immunoscore’ in the classification of malignant 118. Bochet, L. et al. Adipocyte-​derived fibroblasts
Front. Immunol. 8, 643 (2017). tumours. J. Pathol. 232, 199–209 (2014). promote tumor progression and contribute to the
69. Elgazar-​Carmon, V., Rudich, A., Hadad, N. & Levy, R. 95. Galon, J. et al. Cancer classification using the desmoplastic reaction in breast cancer. Cancer Res.
Neutrophils transiently infiltrate intra-​abdominal fat Immunoscore: a worldwide task force. J. Transl Med. 73, 5657–5668 (2013).
early in the course of high-​fat feeding. J. Lipid Res. 49, 10, 205 (2012). 119. Kalluri, R. The biology and function of fibroblasts in
1894–1903 (2008). 96. Pagès, F. et al. International validation of the cancer. Nat. Rev. Cancer 16, 582 (2016).
70. Lysaght, J. et al. T lymphocyte activation in visceral consensus Immunoscore for the classification of colon 120. Nishida, T. et al. Low stromal area and high stromal
adipose tissue of patients with oesophageal cancer: a prognostic and accuracy study. Lancet 391, microvessel density predict poor prognosis in
adenocarcinoma. Br. J. Surg. 98, 964–974 (2011). 2128–2139 (2018). pancreatic cancer. Pancreas 45, 593–600 (2016).
71. Wouters, K. et al. Circulating classical monocytes are 97. Taieb, J. et al. Evolution of checkpoint inhibitors for 121. Spaeth, E. L. et al. Mesenchymal stem cell transition
associated with CD11c+ macrophages in human the treatment of metastatic gastric cancers: current to tumor-​associated fibroblasts contributes to
visceral adipose tissue. Sci. Rep. 7, 42665 (2017). status and future perspectives. Cancer Treat. Rev. 66, fibrovascular network expansion and tumor
72. Wu, D. et al. Eosinophils sustain adipose alternatively 104–113 (2018). progression. PLOS ONE 4, e4992 (2009).
activated macrophages associated with glucose 98. Strong, A. L., Burow, M. E., Gimble, J. M. & Bunnell, B. A. 122. Underwood, T. J. et al. Cancer-​associated fibroblasts
homeostasis. Science 332, 243–247 (2011). Concise review: the obesity cancer paradigm: predict poor outcome and promote periostin-​
73. Mraz, M. & Haluzik, M. The role of adipose tissue exploration of the interactions and crosstalk with dependent invasion in oesophageal adenocarcinoma.
immune cells in obesity and low-​grade inflammation. adipose stem cells. Stem Cells 33, 318–326 (2015). J. Pathol. 235, 466–477 (2015).
J. Endocrinol. 222, R113–R127 (2014). 99. Strong, A. L. et al. Obesity enhances the conversion of 123. Huijbers, A. et al. The proportion of tumor-​stroma
74. Kratz, M. et al. Metabolic dysfunction drives a adipose-​derived stromal/stem cells into carcinoma-​ as a strong prognosticator for stage II and III colon
mechanistically distinct proinflammatory phenotype associated fibroblast leading to cancer cell cancer patients: validation in the VICTOR trial.
in adipose tissue macrophages. Cell Metab. 20, proliferation and progression to an invasive Ann. Oncol. 24, 179–185 (2013).
614–625 (2014). phenotype. Stem Cells Int. 2017, 9216502 (2017). 124. Incio, J. et al. Obesity-​induced inflammation and
75. Ju, C. & Tacke, F. Hepatic macrophages in homeostasis 100. Park, J., Morley, T. S., Kim, M., Clegg, D. J. & desmoplasia promote pancreatic cancer progression
and liver diseases: from pathogenesis to novel Scherer, P. E. Obesity and cancer—mechanisms and resistance to chemotherapy. Cancer Discov. 6,
therapeutic strategies. Cell. Mol. Immunol. 13, underlying tumour progression and recurrence. 852–869 (2016).
316–327 (2016). Nat. Rev. Endocrinol. 10, 455–465 (2014). 125. Lu, P., Weaver, V. M. & Werb, Z. The extracellular
76. Kenna, T. et al. Distinct subpopulations of γδ T cells 101. Hefetz-​Sela, S. & Scherer, P. E. Adipocytes: impact on matrix: a dynamic niche in cancer progression.
are present in normal and tumor-​bearing human liver. tumor growth and potential sites for therapeutic J. Cell Biol. 196, 395–406 (2012).
Clin. Immunol. 113, 56–63 (2004). intervention. Pharmacol. Ther. 138, 197–210 (2013). 126. Levental, K. R. et al. Matrix crosslinking forces tumor
77. Toubal, A. & Lehuen, A. Lights on MAIT cells, a new 102. Toren, P., Mora, B. C. & Venkateswaran, V. Diet, progression by enhancing integrin signaling. Cell 139,
immune player in liver diseases. J. Hepatol. 64, obesity, and cancer progression: are adipocytes the 891–906 (2009).
1008–1010 (2016). link? Lipid Insights 6, 37–45 (2013). 127. Frantz, C., Stewart, K. M. & Weaver, V. M. The
78. Kurioka, A., Walker, L. J., Klenerman, P. & Willberg, C. B. 103. Freese, K. E. et al. Adipose-​derived stems cells and extracellular matrix at a glance. J. Cell Sci. 123,
MAIT cells: new guardians of the liver. Clin. Transl their role in human cancer development, growth, 4195–4200 (2010).
Immunol. 5, e98 (2016). progression, and metastasis: a systematic review. 128. Hynes, R. O. The extracellular matrix: not just pretty
79. Rohr-​Udilova, N. et al. Deviations of the immune cell Cancer Res. 75, 1161–1168 (2015). fibrils. Science 326, 1216–1219 (2009).
landscape between healthy liver and hepatocellular 104. Arendt, L. M. et al. Obesity promotes breast cancer 129. Malanchi, I. et al. Interactions between cancer stem
carcinoma. Sci. Rep. 8, 6220 (2018). by CCL2-mediated macrophage recruitment and cells and their niche govern metastatic colonization.
80. Reccia, I. et al. Non-​alcoholic fatty liver disease: a sign angiogenesis. Cancer Res. 73, 6080–6093 (2013). Nature 481, 85 (2012).
of systemic disease. Metabolism 72, 94–108 (2017). 105. Eterno, V. et al. Adipose-​derived mesenchymal stem 130. Erez, N., Truitt, M., Olson, P. & Hanahan, D. Cancer-​
81. Narayanan, S., Surette, F. A. & Hahn, Y. S. The cells (ASCs) may favour breast cancer recurrence via associated fibroblasts are activated in incipient
immune landscape in nonalcoholic steatohepatitis. HGF/c-​Met signaling. Oncotarget 5, 613–633 (2014). neoplasia to orchestrate tumor-​promoting
Immune Netw. 16, 147–158 (2016). 106. Zhao, B. C., Zhao, B., Han, J. G., Ma, H. C. & inflammation in an NF-​κB-dependent manner.
82. Conroy, M. J. et al. Parallel profiles of inflammatory Wang, Z. J. Adipose-​derived stem cells promote Cancer Cell 17, 135–147 (2010).
and effector memory T cells in visceral fat and liver of gastric cancer cell growth, migration and invasion 131. Zhu, Q. et al. The IL-6–STAT3 axis mediates a
obesity-​associated cancer patients. Inflammation 39, through SDF-1/CXCR4 axis. Hepatogastroenterology reciprocal crosstalk between cancer-​derived
1729–1736 (2016). 57, 1382–1389 (2010). mesenchymal stem cells and neutrophils to
83. Ogawa, Y. et al. Soluble CD14 levels reflect liver 107. Zhang, T. et al. CXCL1 mediates obesity-​associated synergistically prompt gastric cancer progression.
inflammation in patients with nonalcoholic adipose stromal cell trafficking and function in the Cell Death Dis. 5, e1295 (2014).
steatohepatitis. PLOS ONE 8, e65211 (2013). tumour microenvironment. Nat. Commun. 7, 11674 132. Kalluri, R. & Weinberg, R. A. The basics of epithelial-​
84. Rensen, S. S. et al. Increased hepatic myeloperoxidase (2016). mesenchymal transition. J. Clin. Invest. 119, 1420
activity in obese subjects with nonalcoholic 108. Strioga, M., Viswanathan, S., Darinskas, A., Slaby, O. (2009).
steatohepatitis. Am. J. Pathol. 175, 1473–1482 & Michalek, J. Same or not the same? Comparison of 133. Zheng, X. et al. EMT program is dispensable for
(2009). adipose tissue-​derived versus bone marrow-​derived metastasis but induces chemoresistance in pancreatic
85. Ibrahim, J. et al. Dendritic cell populations with mesenchymal stem and stromal cells. Stem Cells Dev. cancer. Nature 527, 525 (2015).
different concentrations of lipid regulate tolerance and 21, 2724–2752 (2012). 134. Fischer, K. R. et al. Epithelial-​to-mesenchymal
immunity in mouse and human liver. Gastroenterology 109. Razmkhah, M., Mansourabadi, Z., Mohtasebi, M. A., transition is not required for lung metastasis but
143, 1061–1072 (2012). Talei, A. R. & Ghaderi, A. Cancer and normal adipose-​ contributes to chemoresistance. Nature 527, 472
86. Rau, M. et al. Progression from nonalcoholic fatty liver derived mesenchymal stem cells (ASCs): do they have (2015).
to nonalcoholic steatohepatitis is marked by a higher differential effects on tumor and immune cells? Cell 135. Brabletz, T., Kalluri, R., Nieto, M. A. & Weinberg, R. A.
frequency of Th17 cells in the liver and an increased Biol. Int. 42, 334–343 (2017). EMT in cancer. Nat. Rev. Cancer 18, 128 (2018).
Th17/resting regulatory T cell ratio in peripheral blood 110. Pérez, L. M., Bernal, A., San Martín, N. & Gálvez, B. G. 136. Lamouille, S., Xu, J. & Derynck, R. Molecular
and in the liver. J. Immunol. 196, 97–105 (2016). Obese-​derived ASCs show impaired migration and mechanisms of epithelial–mesenchymal transition.
87. Gadd, V. L. et al. The portal inflammatory infiltrate angiogenesis properties. Arch. Physiol. Biochem. 119, Nat. Rev. Mol. Cell Biol. 15, 178 (2014).
and ductular reaction in human nonalcoholic fatty liver 195–201 (2013). 137. De Craene, B. & Berx, G. Regulatory networks defining
disease. Hepatology 59, 1393–1405 (2014). 111. Oñate, B. et al. Stem cells isolated from adipose EMT during cancer initiation and progression.
88. Tajiri, K., Shimizu, Y., Tsuneyama, K. & Sugiyama, T. tissue of obese patients show changes in their Nat. Rev. Cancer 13, 97–110 (2013).
Role of liver-​infiltrating CD3+CD56+ natural killer transcriptomic profile that indicate loss in stemcellness 138. Li, L. & Li, W. Epithelial–mesenchymal transition in
T cells in the pathogenesis of nonalcoholic fatty liver and increased commitment to an adipocyte-​like human cancer: comprehensive reprogramming of
disease. Eur. J. Gastroenterol. Hepatol. 21, 673–680 phenotype. BMC Genomics 14, 625 (2013). metabolism, epigenetics, and differentiation.
(2009). 112. Zhang, Y. et al. Stromal cells derived from visceral and Pharmacol. Ther. 150, 33–46 (2015).
89. Conroy, M. J. et al. The microenvironment of visceral obese adipose tissue promote growth of ovarian 139. Ricciardi, M. et al. Epithelial-​to-mesenchymal
adipose tissue and liver alter natural killer cell viability cancers. PLOS ONE 10, e0136361 (2015). transition (EMT) induced by inflammatory priming

www.nature.com/nrgastro
G A S T R O I N T E S T I N A L C A N C E R A N D OR BE v
E ISEIw
TYS

elicits mesenchymal stromal cell-​like immune-​ 163. Yu, J., Shen, J., Sun, T. T., Zhang, X. & Wong, N. 189. Trayhurn, P., Wang, B. & Wood, I. S. Hypoxia in
modulatory properties in cancer cells. Br. J. Cancer Obesity, insulin resistance, NASH and hepatocellular adipose tissue: a basis for the dysregulation of tissue
112, 1067 (2015). carcinoma. Semin. Cancer Biol. 23, 483–491 (2013). function in obesity? Br. J. Nutr. 100, 227–235
140. Uthaya Kumar, D. B. et al. TLR4 signaling via NANOG 164. Poli, G. Pathogenesis of liver fibrosis: role of oxidative (2008).
cooperates with STAT3 to activate Twist1 and promote stress. Mol. Aspects Med. 21, 49–98 (2000). 190. Ratushnyy, A., Lobanova, M. & Buravkova, L. B.
formation of tumor-​initiating stem-​like cells in livers of 165. Wang, Z., Li, Z., Ye, Y., Xie, L. & Li, W. Oxidative stress Expansion of adipose tissue-​derived stromal cells
mice. Gastroenterology 150, 707–719 (2015). and liver cancer: etiology and therapeutic targets. at “physiologic” hypoxia attenuates replicative
141. Allott, E. H. et al. Elevated tumor expression of PAI-1 Oxid. Med. Cell. Longev. 2016, 7891574 (2016). senescence. Cell Biochem. Funct. 35, 232–243
and SNAI2 in obese esophageal adenocarcinoma 166. Malaguarnera, M., Di Rosa, M., Nicoletti, F. (2017).
patients and impact on prognosis. Clin. Transl & Malaguarnera, L. Molecular mechanisms 191. Trayhurn, P. Hypoxia and adipocyte physiology:
Gastroenterol. 3, e12 (2012). involved in NAFLD progression. J. Mol. Med. 87, implications for adipose tissue dysfunction in obesity.
142. Allott, E. H. et al. MMP9 expression in oesophageal 679–695 (2009). Annu. Rev. Nutr. 34, 207–236 (2014).
adenocarcinoma is upregulated with visceral obesity 167. Chen, J. S. et al. Involvement of PI3K/PTEN/AKT/ 192. Kang, Y. E. et al. The roles of adipokines,
and is associated with poor tumour differentiation. mTOR pathway in invasion and metastasis in proinflammatory cytokines, and adipose tissue
Mol. Carcinog. 52, 144–154 (2013). hepatocellular carcinoma: association with MMP-9. macrophages in obesity-​associated insulin resistance
143. Giannoni, E. et al. Reciprocal activation of prostate Hepatol. Res. 39, 177–186 (2009). in modest obesity and early metabolic dysfunction.
cancer cells and cancer-​associated fibroblasts 168. Mullen, M. & Gonzalez-​Perez, R. R. Leptin-​induced PLOS ONE 11, e0154003 (2016).
stimulates epithelial-​mesenchymal transition and JAK/STAT signaling and cancer growth. Vaccines 4, 26 193. Rasouli, N. Adipose tissue hypoxia and insulin
cancer stemness. Cancer Res. 70, 6945–6956 (2016). resistance. J. Investig. Med. 64, 830–832 (2016).
(2010). 169. Tsochatzis, E., Papatheodoridis, G. V. & 194. Weljie, A. M. & Jirik, F. R. Hypoxia-​induced metabolic
144. Martin, F. et al. Potential role of mesenchymal stem Archimandritis, A. J. The evolving role of leptin and shifts in cancer cells: moving beyond the Warburg
cells (MSCs) in the breast tumour microenvironment: adiponectin in chronic liver diseases. Am. J. effect. Int. J. Biochem. Cell Biol. 43, 981–989
stimulation of epithelial to mesenchymal transition Gastroenterol. 101, 2629–2640 (2006). (2011).
(EMT). Breast Cancer Res. Treat. 124, 317–326 170. Polyzos, S. A., Kountouras, J. & Mantzoros, C. S. 195. van Uden, P., Kenneth, N. S. & Rocha, S. Regulation of
(2010). Leptin in nonalcoholic fatty liver disease: a narrative hypoxia-​inducible factor-1α by NF-​κB. Biochem. J.
145. Shinagawa, K. et al. Mesenchymal stem cells enhance review. Metabolism 64, 60–78 (2015). 412, 477–484 (2008).
growth and metastasis of colon cancer. Int. J. Cancer 171. Nepal, S. & Park, P. H. Modulation of cell death and 196. Bakirtzi, K. et al. The neurotensin-​HIF-1α-​VEGFα
127, 2323–2333 (2010). survival by adipokines in the liver. Biol. Pharm. Bull. axis orchestrates hypoxia, colonic inflammation,
146. Kabashima-​Niibe, A. et al. Mesenchymal stem cells 38, 961–965 (2015). and intestinal angiogenesis. Am. J. Pathol. 184,
regulate epithelial–mesenchymal transition and tumor 172. He, G. et al. Hepatocyte IKKbeta/NF-​kappaB inhibits 3405–3414 (2014).
progression of pancreatic cancer cells. Cancer Sci. tumor promotion and progression by preventing 197. Koppenol, W. H., Bounds, P. L. & Dang, C. V. Otto
104, 157–164 (2013). oxidative stress-​driven STAT3 activation. Cancer Cell Warburg’s contributions to current concepts of cancer
147. Strong, A. L. et al. Leptin produced by obese adipose 17, 286–297 (2010). metabolism. Nat. Rev. Cancer 11, 325–337 (2011).
stromal/stem cells enhances proliferation and 173. Menon, S. et al. Chronic activation of mTOR complex 198. Baba, Y. et al. HIF1A overexpression is associated with
metastasis of estrogen receptor positive breast 1 is sufficient to cause hepatocellular carcinoma in poor prognosis in a cohort of 731 colorectal cancers.
cancers. Breast Cancer Res. 17, 1–16 (2015). mice. Sci. Signal. 5, ra24 (2012). Am. J. Pathol. 176, 2292–2301 (2010).
148. Yao-​Borengasser, A. et al. Adipocyte hypoxia promotes 174. Pikarsky, E. et al. NF-​κB functions as a tumour 199. Lin, D. & Wu, J. Hypoxia inducible factor in
epithelial-​mesenchymal transition-​related gene promoter in inflammation-​associated cancer. Nature hepatocellular carcinoma: atherapeutic target.
expression and estrogen receptor-​negative phenotype 431, 461–466 (2004). World J. Gastroenterol. 21, 12171–12178 (2015).
in breast cancer cells. Oncol. Rep. 33, 2689–2694 175. Di Caro, S. et al. Role of body composition and 200. Dai, X. et al. Association of PD-​L1 and HIF-1α
(2015). metabolic profile in Barrett’s oesophagus and coexpression with poor prognosis in hepatocellular
149. Suarez-​Carmona, M., Lesage, J., Cataldo, D. & Gilles, C. progression to cancer. Eur. J. Gastroenterol. Hepatol. carcinoma. Transl Oncol. 11, 559–566 (2018).
EMT and inflammation: inseparable actors of cancer 28, 251–260 (2016). 201. Catalan, V. et al. IL-32α-​induced inflammation
progression. Mol. Oncol. 11, 805–823 (2017). 176. Quante, M. et al. Bile acid and inflammation activate constitutes a link between obesity and colon cancer.
150. Clark, A. G. & Vignjevic, D. M. Modes of cancer cell gastric cardia stem cells in a mouse model of Barrett-​ Oncoimmunology 6, e1328338 (2017).
invasion and the role of the microenvironment. like metaplasia. Cancer Cell 21, 36–51 (2012). 202. Yang, Y. et al. Dysregulation of over-​expressed IL-32 in
Curr. Opin. Cell Biol. 36, 13–22 (2015). 177. Moschen, A. R. et al. Adipose and liver expression of colorectal cancer induces metastasis. World J. Surg.
151. Friedl, P. & Wolf, K. Tumour-​cell invasion and interleukin (IL)-1 family members in morbid obesity Oncol. 13, 146 (2015).
migration: diversity and escape mechanisms. and effects of weight loss. Mol. Med. 17, 840–845 203. Wincewicz, A., Sulkowska, M., Koda, M. & Sulkowski, S.
Nat. Rev. Cancer 3, 362 (2003). (2011). Clinicopathological significance and linkage of the
152. Aceto, N. et al. Circulating tumor cell clusters are 178. Kern, P. A., Ranganathan, S., Li, C., Wood, L. & distribution of HIF-1α and GLUT-1 in human primary
oligoclonal precursors of breast cancer metastasis. Ranganathan, G. Adipose tissue tumor necrosis factor colorectal cancer. Pathol. Oncol. Res. 13, 15–20
Cell 158, 1110–1122 (2014). and interleukin-6 expression in human obesity and (2007).
153. Qiao, L. et al. Suppression of tumorigenesis by human insulin resistance. Am. J. Physiol. Endocrinol. Metab. 204. Lolmede, K., Durand de Saint Front, V., Galitzky, J.,
mesenchymal stem cells in a hepatoma model. 280, E745–E751 (2001). Lafontan, M. & Bouloumie, A. Effects of hypoxia on
Cell Res. 18, 500 (2008). 179. Divella, R., De Luca, R., Abbate, I., Naglieri, E. & the expression of proangiogenic factors in
154. Lu, Y. R. et al. The growth inhibitory effect of Daniele, A. Obesity and cancer: the role of adipose differentiated 3T3-F442A adipocytes. Int. J. Obes.
mesenchymal stem cells on tumor cells in vitro and tissue and adipo-​cytokines-induced chronic Relat. Metab. Disord. 27, 1187–1195 (2003).
in vivo. Cancer Biol. Ther. 7, 245–251 (2008). inflammation. J. Cancer 7, 2346–2359 (2016). 205. Zheng, H. et al. Expressions of MMP-2, MMP-9 and
155. Ohlsson, L. B., Varas, L., Kjellman, C., Edvardsen, K. & 180. Yu, H., Pardoll, D. & Jove, R. STATs in cancer VEGF are closely linked to growth, invasion, metastasis
Lindvall, M. Mesenchymal progenitor cell-​mediated inflammation and immunity: a leading role for STAT3. and angiogenesis of gastric carcinoma. Anticancer Res.
inhibition of tumor growth in vivo and in vitro in Nat. Rev. Cancer 9, 798–809 (2009). 26, 3579–3583 (2006).
gelatin matrix. Exp. Mol. Pathol. 75, 248–255 181. O’Sullivan, K. E., Reynolds, J. V., O’Hanlon, C., 206. Lemoine, A. Y., Ledoux, S. & Larger, E. Adipose tissue
(2003). O’Sullivan, J. N. & Lysaght, J. Could signal transducer angiogenesis in obesity. Thromb. Haemost. 110,
156. Cousin, B. et al. Adult stromal cells derived from and activator of transcription 3 be a therapeutic target 661–668 (2013).
human adipose tissue provoke pancreatic cancer cell in obesity-​related gastrointestinal malignancy? 207. Neels, J. G., Thinnes, T. & Loskutoff, D. J. Angiogenesis
death both in vitro and in vivo. PLOS ONE 4, e6278 J. Gastrointest. Cancer 45, 1–11 (2014). in an in vivo model of adipose tissue development.
(2009). 182. Chang, W. J., Du, Y., Zhao, X., Ma, L. Y. & Cao, G. W. FASEB J. 18, 983–985 (2004).
157. Wu, X.-B. et al. Mesenchymal stem cells promote Inflammation-​related factors predicting prognosis 208. Fukumura, D. et al. Paracrine regulation of
colorectal cancer progression through AMPK/mTOR-​ of gastric cancer. World J. Gastroenterol. 20, angiogenesis and adipocyte differentiation during
mediated NF-​κB activation. Sci. Rep. 6, 21420 4586–4596 (2014). in vivo adipogenesis. Circ. Res. 93, e88–e97 (2003).
(2016). 183. Ericksen, R. E. et al. Obesity accelerates Helicobacter 209. Miyazawa-​Hoshimoto, S. et al. Roles of degree of fat
158. Guaita-​Esteruelas, S., Guma, J., Masana, L. & Borras, J. felis-​induced gastric carcinogenesis by enhancing deposition and its localization on VEGF expression in
The peritumoural adipose tissue microenvironment immature myeloid cell trafficking and TH17 response. adipocytes. Am. J. Physiol. Endocrinol. Metab. 288,
and cancer. The roles of fatty acid binding protein Gut 63, 385–394 (2014). E1128–E1136 (2005).
4 and fatty acid binding protein 5. Mol. Cell 184. Galizia, G. et al. Prognostic significance of circulating 210. Engin, A. Adipose tissue hypoxia in obesity and its
Endocrinol. 462, 107–118 (2018). IL-10 and IL-6 serum levels in colon cancer patients impact on preadipocytes and macrophages: hypoxia
159. Dirat, B. et al. Cancer-​associated adipocytes exhibit an undergoing surgery. Clin. Immunol. 102, 169–178 hypothesis. Adv. Exp. Med. Biol. 960, 305–326
activated phenotype and contribute to breast cancer (2002). (2017).
invasion. Cancer Res. 71, 2455–2465 (2011). 185. Knupfer, H. & Preiss, R. Serum interleukin-6 levels 211. Tam, J. et al. Blockade of VEGFR2 and not VEGFR1
160. Trevellin, E. et al. Esophageal adenocarcinoma and in colorectal cancer patients—a summary of can limit diet-​induced fat tissue expansion: role of
obesity: peritumoral adipose tissue plays a role in published results. Int. J. Colorectal Dis. 25, local versus bone marrow-​derived endothelial cells.
lymph node invasion. Oncotarget 6, 11203–11215 135–140 (2010). PLOS ONE 4, e4974 (2009).
(2015). 186. Voronov, E. & Apte, R. N. IL-1 in colon inflammation, 212. Sun, K. et al. Dichotomous effects of VEGF-​A on
161. Rebours, V. et al. Obesity and fatty pancreatic colon carcinogenesis and invasiveness of colon cancer. adipose tissue dysfunction. Proc. Natl Acad. Sci. USA
infiltration are risk factors for pancreatic precancerous Cancer Microenviron. 8, 187–200 (2015). 109, 5874–5879 (2012).
lesions (PanIN). Clin. Cancer Res. 21, 3522–3528 187. Lysaght, J. et al. Pro-​inflammatory and tumour 213. Incio, J. et al. PlGF/VEGFR-1 signaling promotes
(2015). proliferative properties of excess visceral adipose macrophage polarization and accelerated tumor
162. Peverill, W., Powell, L. W. & Skoien, R. Evolving tissue. Cancer Lett. 312, 62–72 (2011). progression in obesity. Clin. Cancer Res. 22,
concepts in the pathogenesis of NASH: beyond 188. Rausch, L. K., Netzer, N. C., Hoegel, J. & Pramsohler, S. 2993–3004 (2016).
steatosis and inflammation. Int. J. Mol. Sci. 15, The linkage between breast cancer, hypoxia, and 214. Anagnostoulis, S. et al. Human leptin induces
8591–8638 (2014). adipose tissue. Front. Oncol. 7, 211 (2017). angiogenesis in vivo. Cytokine 42, 353–357 (2008).

Nature Reviews | Gastroenterology & Hepatology


REvIEwS

215. Bouloumie, A., Drexler, H. C., Lafontan, M. 229. Lynam-​Lennon, N. et al. Excess visceral adiposity 241. Campbell, K. L., Landells, C. E., Fan, J. & Brenner, D. R.
& Busse, R. Leptin, the product of Ob gene, promotes induces alterations in mitochondrial function and A systematic review of the effect of lifestyle
angiogenesis. Circ. Res. 83, 1059–1066 (1998). energy metabolism in esophageal adenocarcinoma. interventions on adipose tissue gene expression:
216. Koda, M. et al. Expression of the obesity hormone BMC Cancer 14, 907 (2014). implications for carcinogenesis. Obesity 25 (Suppl. 2),
leptin and its receptor correlates with hypoxia-​ 230. Palsson-​McDermott, E. M. et al. Pyruvate kinase M2 40–51 (2017).
inducible factor-1α in human colorectal cancer. is required for the expression of the immune 242. Bai, Y. & Sun, Q. Macrophage recruitment in obese
Ann. Oncol. 18 (Suppl. 6), 116–119 (2007). checkpoint PD-​L1 in immune cells and tumors. adipose tissue. Obes. Rev. 16, 127–136 (2015).
217. Koda, M., Sulkowska, M., Kanczuga-​Koda, L., Front. Immunol. 8, 1300 (2017). 243. Park, E. J. et al. Dietary and genetic obesity promote
Surmacz, E. & Sulkowski, S. Overexpression of the 231. Ikhlas, S. & Ahmad, M. Metformin: insights into its liver inflammation and tumorigenesis by enhancing
obesity hormone leptin in human colorectal cancer. anticancer potential with special reference to AMPK IL-6 and TNF expression. Cell 140, 197–208 (2010).
J. Clin. Pathol. 60, 902–906 (2007). dependent and independent pathways. Life Sci. 185, 244. Pendyala, S., Neff, L. M., Suarez-​Farinas, M.
218. Howard, J. M. et al. Associations between leptin and 53–62 (2017). & Holt, P. R. Diet-​induced weight loss reduces
adiponectin receptor upregulation, visceral obesity 232. Jensen, M. D. et al. 2013 AHA/ACC/TOS guideline for colorectal inflammation: implications for colorectal
and tumour stage in oesophageal and junctional the management of overweight and obesity in adults: carcinogenesis. Am. J. Clin. Nutr. 93, 234–242
adenocarcinoma. Br. J. Surg. 97, 1020–1027 a report of the American College of Cardiology/ (2011).
(2010). American Heart Association Task Force on Practice 245. Todoric, J. et al. Adipose tissue inflammation induced
219. Zhang, C. et al. Tumour-​associated mutant p53 Guidelines and The Obesity Society. J. Am. Coll. by high-​fat diet in obese diabetic mice is prevented by
drives the Warburg effect. Nat. Commun. 4, 2935 Cardiol. 63, 2985–3023 (2014). n-3 polyunsaturated fatty acids. Diabetologia 49,
(2013). 233. LeBlanc, E. S., O’Connor, E., Whitlock, E. P., 2109–2119 (2006).
220. Epstein, T., Gatenby, R. A. & Brown, J. S. The Warburg Patnode, C. D. & Kapka, T. Effectiveness of primary 246. Rowan, B. G. et al. Human adipose tissue-​derived
effect as an adaptation of cancer cells to rapid care-​relevant treatments for obesity in adults: stromal/stem cells promote migration and early
fluctuations in energy demand. PLOS ONE 12, a systematic evidence review for the US Preventive metastasis of triple negative breast cancer xenografts.
e0185085 (2017). Services Task Force. Ann. Intern. Med. 155, 434–447 PLOS ONE 9, e89595 (2014).
221. Paspulati, R. M. & Gupta, A. PET/MR imaging in (2011). 247. Orecchioni, S. et al. Complementary populations of
cancers of the gastrointestinal tract. PET Clin. 11, 234. Colquitt, J. L., Pickett, K., Loveman, E. & human adipose CD34+ progenitor cells promote
403–423 (2016). Frampton, G. K. Surgery for weight loss in adults. growth, angiogenesis, and metastasis of breast cancer.
222. Pisarsky, L. et al. Targeting metabolic symbiosis to Cochrane Database Syst. Rev. 8, CD003641 (2014). Cancer Res. 73, 5880–5891 (2013).
overcome resistance to anti-​angiogenic therapy. 235. Gloy, V. L. et al. Bariatric surgery versus non-​surgical 248. Bellows, C. F., Zhang, Y., Chen, J., Frazier, M. L. &
Cell Rep. 15, 1161–1174 (2016). treatment for obesity: a systematic review and meta-​ Kolonin, M. G. Circulation of progenitor cells in obese
223. Nakajima, E. C. & Van Houten, B. Metabolic symbiosis analysis of randomised controlled trials. BMJ 347, and lean colorectal cancer patients. Cancer Epidemiol.
in cancer: refocusing the Warburg lens. Mol. Carcinog. f5934 (2013). Biomarkers Prev. 20, 2461–2468 (2011).
52, 329–337 (2013). 236. Reges, O. et al. Association of bariatric surgery using 249. Baglioni, S. et al. Functional differences in visceral and
224. Vousden, K. H. & Lane, D. P. p53 in health and laparoscopic banding, roux-​en-Y gastric bypass, subcutaneous fat pads originate from differences in
disease. Nat. Rev. Mol. Cell Biol. 8, 275–283 (2007). or laparoscopic sleeve gastrectomy versus usual care the adipose stem cell. PLOS ONE 7, e36569 (2012).
225. Phelan, J. J. et al. Examining the connectivity between obesity management with all-​cause mortality. JAMA
different cellular processes in the Barrett tissue 319, 279–290 (2018). Author contributions
microenvironment. Cancer Lett. 371, 334–346 237. Sjöström, L. et al. Bariatric surgery and long-​term J.O., J.L., C.L.D. and J.V.R. researched data for the article. All
(2016). cardiovascular events. JAMA 307, 56–65 (2012). authors contributed equally to the discussion of content, writ-
226. Huang, D., Li, C. & Zhang, H. Hypoxia and cancer 238. Sjöström, L. et al. Effects of bariatric surgery on ing of the article and the reviewing and/or editing of the
cell metabolism. Acta Biochim. Biophys. Sin. 46, cancer incidence in obese patients in Sweden (Swedish manuscript before submission.
214–219 (2014). Obese Subjects Study): a prospective, controlled
227. Johnson, A. R., Milner, J. J. & Makowski, L. The intervention trial. Lancet Oncol. 10, 653–662 (2009). Competing interests
inflammation highway: metabolism accelerates 239. Adams, T. D. et al. Cancer incidence and mortality after The authors declare no competing interests.
inflammatory traffic in obesity. Immunol. Rev. 249, gastric bypass surgery. Obesity 17, 796–802 (2009).
218–238 (2012). 240. Schauer, D. P. et al. Bariatric surgery and the risk of Publisher’s note
228. Krzywinska, E. & Stockmann, C. Hypoxia, metabolism cancer in a large multisite cohort. Ann. Surg. https:// Springer Nature remains neutral with regard to jurisdictional
and immune cell function. Biomedicines 6, 56 (2018). doi.org/10.1097/SLA.0000000000002525 (2017). claims in published maps and institutional affiliations.

www.nature.com/nrgastro

You might also like