Epidemiology

The epidemiological evidence underpinning the association between obesity and cancer incidence is now compelling. Large population-based prospective studies have demonstrated convincingly consistent increased cancer incidences per 5 kg/m² increase in BMI⁽ Reference Calle, Rodriguez, Walker-Thurmond and Thun ^{4 ,} Reference Renehan, Tyson, Egger, Heller and Zwahlen ^{8 , 11 –} Reference Reeves, Pirie, Beral, Green, Spencer and Bull ^{13 )}. For example, a recent UK population based prospective cohort of 5·2 million patients with a median follow-up of 6 years recorded 166 966 incident cancers across twenty-two of the most common cancers⁽ Reference Bhaskaran, Douglas, Forbes, Dos-Santos-Silva, Leon and Smeeth ^{12 )}. BMI was positively associated with cancers at the following sites: colon, rectum, liver, gallbladder, pancreas, breast (post-menopausal only), cervix, ovary, uterus, kidney, thyroid and leukaemia⁽ Reference Bhaskaran, Douglas, Forbes, Dos-Santos-Silva, Leon and Smeeth ^{12 )}. The derived hazard ratios from this study are displayed in Table 1. The administrative dataset used in this study was unable to subtype cancers, for example both squamous and adenocarcinomas of the oesophagus were pooled as oesophageal cancer. This probably accounts for the lack of association between oesophageal adenocarcinoma and obesity, which has been consistent and marked in other studies (approximate hazard ratios 4·76, 95 % CI 2·96, 7·66; for patients with BMI ≥ 40 compared with normal BMI)⁽ Reference Renehan, Tyson, Egger, Heller and Zwahlen ^{8 ,} Reference Calle and Kaaks ^{14 )}. A meta-analysis prospective study of 282 000 cancer cases from population of 4·8 million patients demonstrated similarly consistent findings across disease sites and sexes⁽ Reference Renehan, Tyson, Egger, Heller and Zwahlen ^{8 )}. Applying the Bradford-Hill hypotheses of strength, consistency, specificity, temporality, biological gradient, plausibility, coherence, experimental evidence and analogy, obesity is a strong candidate to be a causal factor in the development of cancer as detailed in a review by Renehan et al. ⁽ Reference Hill ^{15 ,} Reference Renehan, Egger and Zwahlen ^{16 )}.

Table 1.

Hazard ratio (HR) for cancers per 5 kg/m² increase in BMI⁽ Reference Bhaskaran, Douglas, Forbes, Dos-Santos-Silva, Leon and Smeeth ^{12 )}

Further epidemiological studies may lead to refinement of the associated increases in relative risk of incident cancers, especially cancers at sites with a less frequent incidence. Current estimates are subject to wide CI at disease sites where few incident cancers are reported. Heterogeneity in already reported relative risks at different cancer sites indicates that there is probably not a universal mechanism by which obesity drives cancer development. For example, the attributable fraction of endometrial cancers is approximately 41 % v. 10 % for cancers of the colon⁽ Reference Bhaskaran, Douglas, Forbes, Dos-Santos-Silva, Leon and Smeeth ^{12 )}. The main caveat concerning the association between obesity and cancer is the fact that a number of other environmental factors including dietary composition, energy intake and physical activity represent important confounding factors in obese patients, and no specifically designed prospective studies have yet to demarcate the relative contribution of each.

Pathological inflammation in obesity is associated with insulin resistance and altered cellular energetics

Obesity-associated inflammation and cancer are inextricably linked and despite robust epidemiological evidence connecting obesity and cancer, the pathophysiological mechanisms underpinning the association remain poorly characterised⁽ Reference Donohoe, Doyle and Reynolds ^{17 ,} Reference Donohoe, O'Farrell, Doyle and Reynolds ^{18 )}. The expanded adipose tissue mass associated with an obese phenotype is deposited in both the intra-abdominal compartment, where it is known as central or visceral adiposity, and beneath the skin, where it is called subcutaneous adipose tissue. Visceral adipose tissue (VAT) is more highly correlated with an altered glucose and lipid metabolic profile as well as increased risk of CVD than subcutaneous adipose tissue⁽ Reference Fujioka, Matsuzawa, Tokunaga and Tarui ^{19 –} Reference Nakamura, Tokunaga, Shimomura, Nishida, Yoshida and Kotani ^{21 )}. Obese patients often have a relative abundance of fat in one or other compartment⁽ Reference Vague ^{22 )}. Those with excess visceral fat have an increased risk (independent of BMI) of breast cancer, oesophageal adenocarcinoma, colorectal adenocarcinoma and colorectal adenoma⁽ Reference Beddy, Howard, McMahon, Knox, De Blacam and Ravi ^{9 ,} Reference Schapira, Clark, Wolff, Jarrett, Kumar and Aziz ^{23 –} Reference Nam, Kim, Han, Ryu, Park and Kim ^{26 )}.

Adipose tissue in obese subjects has altered endocrine function and a secretory profile with predominantly pro-inflammatory cytokines, which are secreted by both adipocytes (and called adipokines) and immune cells. This results in a state of chronic low-grade inflammation which is thought to be pro-tumourigenic⁽ Reference Esposito and Giugliano ^{27 –} Reference Das ^{29 )}. Inflammation provides the selective pressure that may fuel the accumulation of mutations driving tumour growth and survival and could result in poor immune surveillance of tumours once they develop⁽ Reference Coussens and Werb ^{30 )}. The relative contribution of each individual adipokine to this systemic inflammation is not understood and it is likely that a complex interplay between the various pro- and anti-inflammatory immune cells and their secreted cytokines contributes to tumourigenesis.

Another consequence of systemic inflammation in the obese state is insulin resistance and T2DM⁽ Reference Xu, Barnes, Yang, Tan, Yang and Chou ^{31 )}. These conditions are associated with increased levels of inflammatory mediators including acute phase reactants such as fibrinogen, C-reactive protein, IL-6, plasminogen activator inhibitor-1⁽ Reference Hotamisligil, Arner, Caro, Atkinson and Spiegelman ^{32 –} Reference Antuna-Puente, Feve, Fellahi and Bastard ^{34 )}. The prevalence of insulin resistance increases as body weight increases and can be reversed with weight loss⁽ Reference DeFronzo ^{35 ,} Reference Buchwald, Avidor, Braunwald, Jensen, Pories and Fahrbach ^{36 )}. Epidemiological studies on the role of T2DM in carcinogenesis report consistent increases in cancer incidence at certain sites, which display considerable overlap with those sites associated with obesity-associated cancers as previously described, including liver in particular, as well as pancreas, endometrium, breast, colorectal, bladder, non-Hodgkin's lymphoma and kidney cancers⁽ Reference Johnson, Carstensen, Witte, Bowker, Lipscombe and Renehan ^{37 )}.

Diet-induced insulin resistance develops as a metabolic adaptation to increased circulating levels of non-esterified fatty acids (NEFA), which are constantly released from adipose tissue, especially from visceral fat stores⁽ Reference Calle and Kaaks ^{14 )}. Increased NEFA levels force liver, muscle and other tissues to shift towards increased storage and oxidation of fats for their energy production⁽ Reference Bergman and Ader ^{38 )}. The compensatory effect is a reduced capacity of these tissues to absorb, store and metabolise glucose. Patients with insulin resistance also exhibit a reduction in cellular insulin-receptor levels and reduced responsiveness of some intracellular transduction pathways mediating the effects of insulin binding to its receptor⁽ Reference Moller and Flier ^{39 )}. Interestingly, insulin resistance leads to increased insulin production and the mitogenic effects of insulin have been associated with several cancers⁽ Reference Schoen, Tangen, Kuller, Burke, Cushman and Tracy ^{24 ,} Reference Trevisan, Liu, Muti, Misciagna, Menotti and Fucci ^{40 ,} Reference Colangelo, Gapstur, Gann, Dyer and Liu ^{41 )}. The insulin-like growth factor (IGF) axis comprising two growth factors (IGF-1 and IGF-2) as well as a number of binding proteins and two receptors (IGF-1R and IGF-2R), sharing homology with the insulin receptor, has been cited as a mediator of hyperinsulinaemia on cell growth and energy metabolism⁽ Reference Pollak ^{42 )}. Increased expression of IGF-1R is reported in a number of cancer subtypes but a large systematic review and meta-analysis of studies of serum concentrations of IGF-1 and insulin as biomarkers reveals there to be significant bias found in the studies which have been published in the literature⁽ Reference Hellawell, Turner, Davies, Poulsom, Brewster and Macaulay ^{43 –} Reference Tsilidis, Travis, Appleby, Allen, Lindstrom and Schumacher ^{46 )}. New prospective specifically-designed epidemiological studies are warranted to help clarify the relative contribution of insulin resistance, inflammation and obesity to cancer development.

Changes in cellular energetics and dysregulated metabolism are hallmarks of cancer and it is not surprising that associations between metabolic changes and obesity are being uncovered⁽ Reference Hanahan and Weinberg ^{47 )}. Recent findings of our group show that secreted factors from VAT of obesity-associated cancer patients induce metabolic changes in oesophageal adenocarcinoma cell lines; increasing mitochondrial mass, mitochondrial membrane potential and inducing a shift to glycolysis⁽ Reference Lynam-Lennon, Maher, Maguire, Phelan, Muldoon and Reynolds ^{48 )}. Moreover, these studies identified a positive correlation between obesity and the expression of the glycolytic marker pyruvate kinase M2 in the tumours of oesophageal adenocarcinoma patients supporting a shift to glycolysis in obesity. Further analysis revealed that the VAT of obese oesophageal adenocarcinoma patients exhibited a glycolytic profile with increased expression of lactate, an established carcinogen⁽ Reference Lynam-Lennon, Maher, Maguire, Phelan, Muldoon and Reynolds ^{48 ,} Reference Groff, Stugard, Mays, Koopmans and DiGirolamo ^{49 )}. Therefore, these data present a possible link between obesity and the dysregulated bioenergetics that are characteristic of cancer, thus providing a mechanism through which increased VAT mass might promote tumourigenesis.

The sex differential in cancer incidence in obese patients has also been repeatedly reproduced in epidemiological studies⁽ Reference Renehan, Tyson, Egger, Heller and Zwahlen ^{8 ,} Reference Bhaskaran, Douglas, Forbes, Dos-Santos-Silva, Leon and Smeeth ^{12 )}. This, as well as the marked increased incidence in endometrial cancers in obese females in particular, has led to the hypothesis that sex hormones play a role in the carcinogenic process in obesity, at least for some sites. There is a well-recognised association between the development of endometrial cancer and oestrogen excess⁽ Reference Lukanova, Lundin, Zeleniuch-Jacquotte, Muti, Mure and Rinaldi ^{50 ,} Reference Zeleniuch-Jacquotte, Akhmedkhanov, Kato, Koenig, Shore and Kim ^{51 )}. A factor analysis of inflammatory, insulin-regulated physiological axes and sex steroids measured from the pre-diagnostic serum levels of patients enrolled in the European Prospective Investigation into Cancer and Nutrition cohort compared patients who developed endometrial cancer v. matched controls⁽ Reference Dossus, Lukanova, Rinaldi, Allen, Cust and Becker ^{52 )}. This allowed an analysis of these highly correlated biomarkers for their relative contribution to endometrial cancer development and demonstrated a contribution to cancer risk in all three of these axes.

In light of the metabolic changes that occur with increasing visceral adipose tissue mass, it stands to reason that obesity may be a confounding factor when developing immune-based anti-cancer therapies for obesity-associated malignancies. However, an improved understanding of the interplay between the immunity–obesity–cancer axis may allow novel insights into cancer development, and can allow improvement of current strategies for personalised therapy.

Adipokines and immune function

In addition to their direct roles in cellular energetics and metabolism⁽ Reference Cao ^{53 )}, emerging evidence demonstrates roles for adipokines in immune system regulation. For example, leptin, produced by adipose tissue, is a long-term regulator of weight and acts via its receptor to decrease appetite and food intake and increase energy consumption⁽ Reference Maffei, Halaas, Ravussin, Pratley, Lee and Zhang ^{54 )}. Leptin is also a pro-inflammatory cytokine inducing activation of T cells and Th1 cell differentiation⁽ Reference Lord, Matarese, Howard, Baker, Bloom and Lechler ^{55 )}. It also modulates the function of all other immune cell lineages (but only in the presence of other non-specific immunostimulants)⁽ Reference Matarese, Moschos and Mantzoros ^{56 )}. Leptin is induced by inflammatory cytokines such as IL-1, IL-6 and lipopolysaccharide⁽ Reference Faggioni, Feingold and Grunfeld ^{57 )}. It could be hypothesised that the relative lack of some subpopulations of T cells (CD3⁺, CD4⁺CD45RO⁺, CD8⁺)⁽ Reference Tanaka, Isoda, Ishihara, Kimura and Yamakawa ^{58 )} could be related to the leptin resistant state prevalent in obese subjects⁽ Reference Considine, Sinha, Heiman, Kriauciunas, Stephens and Nyce ^{59 )}. Adiponectin is an abundant adipokine secreted by adipocytes in VAT. Circulating levels are inversely associated with obesity⁽ Reference Kadowaki and Yamauchi ^{60 )} and decreased expression of adiponectin is closely correlated with insulin resistance⁽ Reference Weyer, Funahashi, Tanaka, Hotta, Matsuzawa and Pratley ^{61 ,} Reference Yamauchi, Kamon, Waki, Terauchi, Kubota and Hara ^{62 )}. Adiponectin inhibits macrophage phagocytosis and reduces macrophage production of pro-inflammatory IL-6 and TNF-α and increases production of anti-inflammatory IL-10 and IL-1R⁽ Reference Lago, Dieguez, Gómez-Reino and Gualillo ^{63 )}.

Innate and adaptive immunity in obesity and cancer

Obesity induces a state of pathological chronic inflammation, marked by elevated levels of NEFA, abnormal cytokine production and activation of inflammatory signalling pathways⁽ Reference Fain ^{64 –} Reference Wellen and Hotamisligil ^{68 )}. The expanded vasculature and increased blood supply that accompanies the expansion of VAT in obesity, results in increased inflammatory immune cell infiltration, enhanced pro-inflammatory cytokine release and ultimately local and systemic tumourigenic inflammation⁽ Reference Karin ^{69 ,} Reference Ramos-Nino ^{70 )}. The alterations in immune cell repertoires in the peripheral blood, VAT and proximal tissues, in particular liver, continue to be investigated in an effort to elucidate the extent of immune dysregulation in obesity. Immunotherapeutic strategies to reset the balance between pro- and anti-inflammatory cells could also be crucial to restoring immune control, attenuating obesity-associated inflammation and treating/preventing resultant disease. However the immune dysregulation in obesity with underlying malignancy could significantly negatively impact on anti-tumour immunity. In fact, ideal immunotherapeutic approaches in obesity-associated cancer should attenuate chronic systemic inflammation while simultaneously augmenting specific immunity at the tumour site. Research is required to fully characterise the immune cell repertoires in the periphery, VAT and tumour of obesity-associated cancer patients to distinguish between the subsets that drive tumourigenic inflammation, those that control it and those that are crucial for tumour eradication. For the purpose of this review, we will focus on the present knowledge of conventional T lymphocytes, innate lymphocytes and macrophages in obesity and cancer (Table 2).

Table 2.

Summary of immunological impact of obesity in human and potential therapeutic targets

iNKT, invariant natural killer T cells; MAIT, mucosal associated invariant T cells; NK, natural killer; VAT, visceral adipose tissue.

Conventional T cells

CD4⁺ and CD8⁺ T cells comprise the antigen-specific effector arm of the adaptive immune response and elicit their effector functions through cytokine production and cytotoxic activity. In the CD4⁺ T cell compartment, Th1 cells produce interferon (IFN)-γ, TNF-α and IL-2, while T helper type 2 (Th2) cells express IL-4, IL-5 and IL-13⁽ Reference Ikeda, Chamoto, Tsuji, Suzuki, Wakita and Takeshima ^{91 ,} Reference Foster, Martinez-Moczygemba, Huston and Corry ^{92 )}. Inflammatory T helper type 17 (Th17) cells express IL-17 and IL-22 and regulatory T (T_reg) cells express IL-10 and transforming growth factor β ⁽ Reference Ciric, El-behi, Cabrera, Zhang and Rostami ^{93 ,} Reference Cottrez and Groux ^{94 )}. CD8⁺ T cells are the key cytotoxic T lymphocytes but are also potent cytokine producers. Therefore in the context of malignancy, tumour infiltration of the appropriate CD4⁺ and CD8⁺ T cell milieu is crucial for tumour eradication⁽ Reference Kalams and Walker ^{95 )}. In fact, the ratio of cytotoxic T lymphocytes to T_reg cells can be predictive of outcome⁽ Reference Kono, Kawaida, Takahashi, Sugai, Mimura and Miyagawa ^{96 ,} Reference Shen, Zhou, Wang, Li, Zhong and Liang ^{97 )}. In obesity, our group and others have reported enrichments of activated and effector pro-inflammatory CD4⁺ and CD8⁺ T cells in VAT and have identified these cells as key players in inflammatory macrophage recruitment and inflammatory cytokine production, thus contributing to the initiation and maintenance of obesity-associated inflammation⁽ Reference Lysaght, Allott, Donohoe, Howard, Pidgeon and Reynolds ^{65 ,} Reference Winer, Chan, Paltser, Truong, Tsui and Bahrami ^{73 ,} Reference Nishimura, Manabe, Nagasaki, Eto, Yamashita and Ohsugi ^{80 )}. Furthermore, we have previously reported significantly higher frequencies of such IFN-γ-producing T cells in the VAT of obesity-associated cancer patients, compared with non-cancer control subjects⁽ Reference Lysaght, Allott, Donohoe, Howard, Pidgeon and Reynolds ^{65 )}. In recent unpublished studies, we have also observed enrichments of TNF-α and IL-17-producing T cells in the VAT, demonstrating that more than one inflammatory T cell subset is contributing to adipose tissue inflammation in obesity and obesity-associated cancer. In contrast, while T_reg cells are enriched in lean VAT, their frequency is significantly diminished in obese VAT⁽ Reference Feuerer, Herrero, Cipolletta, Naaz, Wong and Nayer ^{98 )}. Therefore, VAT in obesity is rich in pro-inflammatory cytokine-producing effector T cells and a relative deficiency of T_reg cells⁽ Reference Lysaght, van der Stok, Allott, Casey, Donohoe and Howard ^{66 ,} Reference Nishimura, Manabe, Nagasaki, Eto, Yamashita and Ohsugi ^{80 ,} Reference Feuerer, Herrero, Cipolletta, Naaz, Wong and Nayer ^{98 )}. Such approaches to reset the balance between regulatory and inflammatory T cell subsets may potentially attenuate pathological inflammation in obesity.

Natural killer cells

Natural killer (NK) cells are MHC-unrestricted lymphocytes and comprise a major component of innate lymphocytes in human blood and liver, comprising 10–15 and 30 % of total lymphocytes, respectively⁽ Reference Robertson and Ritz ^{99 )}. NK cells display powerful cytotoxic activities once activated, and are potent producers of pro-inflammatory cytokines including IFN-γ and TNF-α which can contribute to the modulation of adaptive immune responses and direct anti-tumour effects⁽ Reference Robertson and Ritz ^{99 –} Reference Ing and Stevenson ^{101 )}. In the context of inflammatory disease, NK cells have been implicated in the pathogenesis of inflammatory bowel disease and psoriasis⁽ Reference Yadav, Chen and Liu ^{102 ,} Reference Gardiner and Dunphy ^{103 )}. However as of yet, their role has not been fully elucidated in the chronic inflammation that is characteristic of obesity. Lynch et al. have reported significant reductions of circulating NK cells in metabolically unhealthy obese subjects, suggesting that a lack of NK cells in the periphery is detrimental for metabolic health⁽ Reference Lynch, O'Connell, Kwasnik, Cawood, O'Farrelly and O'Shea ^{83 )}. Lynch et al. also reported significant expansions of circulating inhibitory-receptor bearing and CD69⁺ NK cells in metabolically unhealthy obese individuals, compared with healthy controls, suggesting that peripheral NK cells in metabolically unhealthy obese patients were activated but unable to elicit effector functions, making such individuals more susceptible to disease⁽ Reference Lynch, O'Connell, Kwasnik, Cawood, O'Farrelly and O'Shea ^{83 )}. Furthermore, circulating NK cells from obese subjects were found to be less effective at killing tumour cells than their counterparts in lean subjects, suggesting that obese individuals have impaired defences against malignancy⁽ Reference O'Shea, Cawood, O'Farrelly and Lynch ^{104 )}. More recently, a study by O’ Rourke et al. revealed expansions of the CD56^BRIGHT (main cytokine producing subset) of NK cells in obese VAT, indicating that cytokine-producing fractions of NK cells are more prevalent in obesity⁽ Reference O'Rourke, Gaston, Meyer, White and Marks ^{105 )}. Moreover, NK cells were identified as key players in macrophage recruitment to adipose tissue but did not affect overall secreted inflammatory cytokine levels⁽ Reference O'Rourke ^{106 )}. In fact, the macrophage population recruited to the VAT by NK cells in this study was not the macrophage subset implicated in VAT inflammation and therefore, the role of NK cells in adipose tissue inflammation was unclear. However, new data gathered by Wensveen et al. demonstrates that IFN-γ-producing NK cells are pivotal in M1 macrophage polarisation in the adipose tissue of obese mice and subsequent insulin resistance, but similar results have yet to be reported in human subjects⁽ Reference Wensveen, Jelenčić, Valentić, Šestan, Wensveen and Theurich ^{107 )}. Interestingly, this study was the first to report a direct interaction between NK cells and adipocytes through the mouse equivalent of NK receptor P46 (NKP46) natural cytotoxicity receptor 1 (NCR1), and shows that NK cells and adipocytes together drive inflammatory macrophage recruitment and contributes to adipose tissue inflammation in mice. With respect to therapeutic avenues, NKP46 might be a potential target to attenuate NK cell and adipocyte-mediated inflammation. Since mouse and human NK cells are characterised differently, it is first important to fully investigate the NK repertoires in human blood and VAT, elucidate their interactions with adipocytes, their cytokine profiles and subsequent contribution to inflammatory cell recruitment and adipose tissue inflammation. Furthermore, the investigation of the NK cell milieu in blood, VAT and tumour of patients with obesity-associated malignancy is warranted to completely understand the changes that NK cells undergo in human obesity and how this impacts anti-tumour immunity and patient outcomes.

Invariant natural killer T cells

Invariant NKT (iNKT) cells are a subset of MHC-unrestricted innate lymphocytes that respond rapidly, elicit potent IFN-γ production and cytotoxic activity and modulate adaptive immune responses, thus making them key players in anti-tumour immunity. They express a restricted T cell receptor repertoire consisting of a Vα14Jα18 α-chain in mice and a Vα24Jα18 α-chain in human subjects, paired with a limited number of β-chains⁽ Reference Bendelac, Rivera, Park and Roark ^{108 )}. Such cells also express a number of cell-surface markers typically expressed on NK cells, hence the name NKT cell, and recognise lipid antigen presented by the MHC-like glycoprotein, CD1d⁽ Reference Bendelac, Rivera, Park and Roark ^{108 –} Reference Gumperz ^{111 )}. The most potent activator of iNKT cells isolated to date is the marine sponge-derived glycolipid α-galactosylceramide (α-GalCer), which has been shown to induce anti-tumour cytotoxic activity and cytokine production in mice leading to the inhibition of tumour growth⁽ Reference Kawano, Cui, Koezuka, Toura, Kaneko and Motoki ^{112 )}. However, α-GalCer has proven much less effective in human subjects due to the lower abundance of iNKT cells in human subjects compared with mice⁽ Reference Kenna, Golden-Mason, Norris, Hegarty, O'Farrelly and Doherty ^{113 )}. VAT is the only human tissue reported to have a high proportion of iNKT cells, averaging 10 % in healthy subjects⁽ Reference Lynch, O'Connell, Kwasnik, Cawood, O'Farrelly and O'Shea ^{83 )}. However, in the setting of obesity the frequencies of iNKT cells are diminished and this depletion is reversed following weight loss⁽ Reference Lynch, Nowak, Varghese, Clark, Hogan and Toxavidis ^{114 )}. VAT iNKT cells preferentially secrete higher levels of IL-10, compared with the Th1 cytokine profile they exhibit in other tissues, suggesting that iNKT cells in adipose tissue predominantly serve an immune regulatory role⁽ Reference Lynch, Nowak, Varghese, Clark, Hogan and Toxavidis ^{114 )}. Furthermore, the adoptive transfer of iNKT cells into obese mice can induce weight loss and improve insulin sensitivity, implying that iNKT cells are pivotal for metabolic health and protective against insulin resistance. This may suggest that metabolically unhealthy individuals might benefit from iNKT cell-based therapies and that IL-10-producing VAT iNKT cells may prove useful in the attenuation of obesity-associated inflammation. However, iNKT cell therapeutic potential in obesity-associated cancer has yet to be uncovered.

Gamma/delta T cells

Human γδ T cells are another type of innate or unconventional T cell subset, typically comprising 1–5 % of circulating adult T cells, and are also found within the gut⁽ Reference Dunne, Elliott, Hussey, Mahmud, Kelly and Doherty ^{115 )}, skin⁽ Reference Laggner, Di Meglio, Perera, Hundhausen, Lacy and Ali ^{116 )}, lung⁽ Reference Lahn ^{117 )} and uterus ⁽ Reference Exley and Boyson ^{118 )}. In human subjects there are three main γδ T cell types, defined by their T cell receptor delta chain (called Vδ1, Vδ2 or Vδ3, respectively), which can be paired with various gamma chains⁽ Reference Hayday ^{119 )}. γδ T cells respond rapidly to infection or stress signals and display anti-viral, anti-bacterial, anti-parasitic and anti-tumour functions, and a strong Th1 predominance⁽ Reference Dunne, Mangan, Madrigal-Estebas and Doherty ^{120 )}. The main subset of circulating human γδ T cells (Vγ9Vδ2 T cells) recognises pyrophosphate antigens derived from bacterial or endogenous sources. The precise antigens recognised and modes of antigen recognition for other γδ T cell subsets are presently unclear but it is known that MHC-mediated antigen presentation is not required⁽ Reference Crowley, Reich, Mavaddat, Altman and Chien ^{121 ,} Reference Singhal, Mori and De Libero ^{122 )}. In mouse models, γδ T cells produce copious amounts of IL-17⁽ Reference Lockhart, Green and Flynn ^{123 )}, and comprise the predominant T cell subset found in tissues⁽ Reference Bergstresser, Sullivan, Streilein and Tigelaar ^{124 )}, whereas this is not the case in human subjects. Furthermore, the predominant γδ T cell subset in human blood, the Vγ9Vδ2 T cell subset, is only found in human subjects and other higher primates. All three γδ T cell subsets in the human demonstrate potent and specific anti-tumour cytotoxicity against a range of different tumour types⁽ Reference Hayday ^{119 ,} Reference Siegers, Dhamko, Wang, Mathieson, Kosaka and Felizardo ^{125 ,} Reference Mangan, Dunne, O'Reilly, Dunne, Exley and O'Shea ^{126 )} and as a result, their use as immunotherapeutic agents is presently under investigation⁽ Reference Gomes, Martins and Silva-Santos ^{88 )}. Meta-analysis of clinical trial data shows limited success with present immunotherapeutic approaches used to date however, most of which involve adoptive transfer of Vγ9Vδ2 T cells. It is hoped that an improved understanding of γδ T cell biology and subset differences will improve future immunotherapeutic strategies.

In obese individuals, there is an inverse correlation between the frequency of circulating γδ T cells and increasing BMI⁽ Reference Costanzo, Taylor, Dutt, Han, Fujioka and Jameson ^{87 )}, with obese individuals having, on average, more than four times fewer circulating Vγ9Vδ2 T cells than non-obese donors. Remaining γδ T cells have a skewed maturation phenotype similar to that observed in aged populations, a blunted IFN-γ response and reduced IL-2 receptor α chain expression. Constanzo and colleagues reported that this dysregulation could be overcome by stimulation of cells with pyrophosphate antigens and addition of IL-2⁽ Reference Costanzo, Taylor, Dutt, Han, Fujioka and Jameson ^{87 )}. IL-2 acts as a growth factor for γδ T cells, suggesting that cellular dysfunction in obesity may be via growth factor deprivation rather than cell incapacitation. In mouse models of obesity, γδ T cells are shown to promote insulin resistance and inflammation via recruitment of macrophages to adipose tissue⁽ Reference Mehta, Nuotio-Antar and Smith ^{127 )}, though whether this also applies in human subjects remains to be investigated. The role of γδ T cells in the underlying inflammation and anti-tumour responses in the setting of obesity-associated cancer has yet to be elucidated. Our group are currently investigating the distribution and functional phenotype of γδ T cell subsets in the periphery, VAT and tumour of obesity-associated cancer patients in order to identify their usefulness in the prophylaxis and treatment of such malignancies.

Mucosal associated invariant T cells

Mucosal associated invariant T (MAIT) cells are a population of invariant T cells comprising 1–5 % of T cells in human blood, and are defined by expression of an invariant Vα7·2-Jα33 chain and high levels of the NK cell marker CD161⁽ Reference Treiner, Duban, Bahram, Radosavljevic, Wanner and Tilloy ^{128 )}. MAIT cells are enriched in human adipose tissue and in mucosal and inflamed tissues, where they recognise vitamin B₂ metabolites produced by bacteria and yeasts⁽ Reference Dusseaux, Martin, Serriari, Péguillet, Premel and Louis ^{129 )}. MAIT cells thus recognise and kill bacteria-infected cells⁽ Reference Le Bourhis, Dusseaux, Bohineust, Bessoles, Martin and Premel ^{130 )}. While MAIT cells have been described in breast cancer and are reportedly resistant to chemotherapy treatment⁽ Reference Dusseaux, Martin, Serriari, Péguillet, Premel and Louis ^{129 )}, their role in cancer is as yet unknown. MAIT cell frequency is dramatically decreased in the blood and adipose tissues of patients with T2DM and/or severe obesity compared with control donors, and this depletion is associated with diabetic status⁽ Reference Carolan, Tobin, Mangan, Corrigan, Gaoatswe and Byrne ^{89 ,} Reference Magalhaes, Pingris, Poitou, Bessoles, Venteclef and Kiaf ^{90 )}. Moreover, in both patient groups, circulating MAIT cells display an activated phenotype that is associated with elevated Th1 and Th17 cytokine production. In obese patients, MAIT cells are more abundant in adipose tissue than in blood and exhibit a striking IL-17 profile, a cytokine implicated in insulin resistance. Similar results were also observed in an obese paediatric cohort⁽ Reference Carolan, Tobin, Mangan, Corrigan, Gaoatswe and Byrne ^{89 )}. MAIT cell frequencies increase while their cytokine production decreases in obese patients after bariatric surgery, in line with improvements in metabolic parameters. These studies reveal profound abnormalities in MAIT cells in patients harbouring metabolic disorders, suggesting their potential role in these pathologies, yet future work remains to determine the role for MAIT cells in obesity-associated cancers.

Macrophages

While macrophages play a crucial role in innate immunity, their polarisation to an M1 phenotype positions them as key drivers of pathological inflammation, while a dominant M2 phenotype can facilitate their immune suppressive role in the tumour microenvironment⁽ Reference Mantovani, Sica, Sozzani, Allavena, Vecchi and Locati ^{131 ,} Reference Verreck, de Boer, Langenberg, Hoeve, Kramer and Vaisberg ^{132 )}. M1, or classically-activated macrophages, are regarded as the more inflammatory subset, while M2 macrophages are considered anti-inflammatory and immunoregulatory. Macrophages recruited to the tumour microenvironment can maintain an M1 phenotype and elicit potent IFN-γ, TNF-α, IL-12 and chemokine production, promote T cell activation and mediate anti-tumour activity⁽ Reference Lamagna, Aurrand-Lions and Imhof ^{133 )}. However, upon recruitment to the tumour microenvironment, macrophages are often polarised toward an M2 phenotype and become tumour-associated macrophages which produce immunomodulatory and pro-angiogenic factors including IL-10, TGF-β, vascular endothelial growth factor and matrix metallopeptidase 9⁽ Reference Lamagna, Aurrand-Lions and Imhof ^{133 )}. In the obese setting, several studies have implicated M1 macrophages as drivers of adipose tissue inflammation, with T cells and NK cells contributing to their recruitment and polarisation⁽ Reference Xu, Barnes, Yang, Tan, Yang and Chou ^{31 ,} Reference Nishimura, Manabe, Nagasaki, Eto, Yamashita and Ohsugi ^{80 ,} Reference Wensveen, Jelenčić, Valentić, Šestan, Wensveen and Theurich ^{107 ,} Reference Weisberg, McCann, Desai, Rosenbaum, Leibel and Ferrante ^{134 )}. In fact, M1 macrophages are believed to be the main producers of IL-1β and TNF in obesity and key players in insulin resistance⁽ Reference Patsouris, Li, Thapar, Chapman, Olefsky and Neels ^{135 )}. Furthermore, while the chemokine system plays a central role in macrophage recruitment to adipose tissue and the development of insulin resistance, the inhibition of one chemokine alone is not enough to block macrophage accumulation in murine models of obesity⁽ Reference Surmi, Webb, Ristau and Hasty ^{136 ,} Reference Ota ^{137 )}. Therefore, future studies might investigate the inhibition of more than one chemokine pathway to block M1 macrophage accumulation in obese VAT and alleviate chronic inflammation. Blocking chemokine-guided recruitment of macrophages to tumour might attenuate tumour-associated macrophage-mediated immune suppression and angiogenesis in cancer and prove useful in cancer immunotherapy. However, it is now understood that tumour-associated macrophage phenotype and function is critically influenced by the tumour microenvironment and therefore exploiting such environmental cues might be a preferable therapeutic approach⁽ Reference Ostuni, Kratochvill, Murray and Natoli ^{138 )}.

Immunometabolism in obesity and cancer

In recent years, research has revealed how immune cell metabolism supports and controls immune function, and has led to a focus on exploiting immunometabolism to treat inflammation and cancer. In brief, more activated and rapidly dividing immune cells rely on glycolysis to maintain their biomass and energy levels, while more quiescent and regulatory immune cells mainly utilise oxidative phosphorylation (OxPhos)⁽ Reference Mockler, Conroy and Lysaght ^{139 )}. Depending on the stage of their development, T cell subsets possess strikingly different metabolic profiles which reflect their energetic and biosynthetic requirements, which in turn support their function⁽ Reference Fox, Hammerman and Thompson ^{140 ,} Reference Pearce and Pearce ^{141 )}. Naive T cells are usually metabolically quiescent and use OxPhos as the main means of generating ATP, while rapidly proliferating effector CD4⁺ Th1, Th2 and Th17 cells and CD8⁺ cytotoxic T cells utilise significantly elevated levels of glycolysis to meet their energy requirements⁽ Reference Fox, Hammerman and Thompson ^{140 –} Reference Bental and Deutsch ^{146 )}. Following resolution of the immune response, the remaining pool of memory T cells mainly utilise OxPhos and lipid oxidation to meet their lower turnover rates and to support their longevity, as do T_reg cells⁽ Reference Pearce, Walsh, Cejas, Harms, Shen and Wang ^{147 ,} Reference Michalek, Gerriets, Jacobs, Macintyre, MacIver and Mason ^{148 )}. Manipulation of such cellular metabolic pathways might represent a novel means of therapeutically skewing immune responses. Since our group and others have shown that pro-inflammatory Th1 cells are key players in obesity-associated inflammation while T_reg cells are depleted, a targeted therapy to alter T cell metabolic profile and attenuate T cell-mediated inflammation in the VAT may present favourable outcomes in obesity and obesity-associated cancer⁽ Reference Lysaght, van der Stok, Allott, Casey, Donohoe and Howard ^{66 ,} Reference Nishimura, Manabe, Nagasaki, Eto, Yamashita and Ohsugi ^{80 ,} Reference Gregor and Hotamisligil ^{149 ,} Reference Gilbert and Slingerland ^{150 )}. Similarly, inflammatory M1 macrophages, other key players in adipose tissue inflammation, and antibody-secreting B cells also rely on glycolytic processes to fuel their inflammatory functions while M2 macrophages use fatty acids to support OxPhos⁽ Reference Weisberg, McCann, Desai, Rosenbaum, Leibel and Ferrante ^{134 ,} Reference O'Neill ^{151 ,} Reference Murray, Allen, Biswas, Fisher, Gilroy and Goerdt ^{152 )}.

Exploitation of immune cell metabolism might be utilised as a prophylactic to prevent obesity-associated disease or to treat obesity-associated cancer. For instance, inhibition of both the mammalian target of rapamycin and mitogen-activated protein kinase (extracellular signal-regulated kinase) pathways using rapamycin and the mitogen-activated protein kinase inhibitor PD325901 robustly blocked effector CD4⁺ T cell proliferation and decreased disease severity in a mouse model of arthritis⁽ Reference Lin, Stein, Wong, Kalpathy, Su and Utz ^{153 )}. Since our group have found pro-inflammatory effector T cells to be enriched in the inflamed omentum in obesity, extracellular signal-regulated kinase and mammalian target of rapamycin inhibitors might elicit a similar anti-inflammatory effect in obesity⁽ Reference Lysaght, van der Stok, Allott, Casey, Donohoe and Howard ^{66 )}. Others have found that beauvericin-mediated inhibition of T cell activation and pro-inflammatory cytokine production in an animal model of colitis was achieved by targeting P13 K and AKT and resulted in reduced weight loss, diarrhoea and mortality⁽ Reference Wu, Wu, Huang, Liu, Ye and Huang ^{154 )}. More recently, inhibition of AKT in expanded tumour-infiltrating lymphocytes resulted in increased rates of OxPhos and fatty acid oxidation⁽ Reference Crompton, Sukumar, Roychoudhuri, Clever, Gros and Eil ^{155 )}. This led to enhanced in vivo persistence of such memory T cells and augmented anti-tumour immunity following adoptive transfer⁽ Reference Crompton, Sukumar and Restifo ^{156 )}. Therefore, AKT inhibition might serve to dampen pro-inflammatory effector T cells in the VAT and simultaneously promote tumour-specific memory T cell responses at the tumour site, thus possibly offering a novel immuno-metabolomic therapeutic approach for chronically inflamed obesity-associated cancer patients. However, there is first an urgent need to perform focussed studies to elucidate the metabolic profiles of pro- and anti-inflammatory immune cells in the VAT and tumour of such patients.

Future prospects for obesity-associated cancer: prevention v. treatment

It is estimated that one in every four cancer cases is preventable by implementing lifestyle changes. Furthermore, studies such as the Women's Intervention Nutrition Study have shown that reduction of fat in the diet of women with breast cancer and associated weight loss reduced the relative risk of cancer recurrence by 24 %⁽ Reference Chlebowski, Blackburn, Thomson, Nixon, Shapiro and Hoy ^{157 )}. Prevention and management of obesity is therefore a major therapeutic goal. Intervention before the pathological inflammatory cascade ensues and immune system becomes excessively dysregulated is the ideal prophylactic approach. However, there have already been multiple government campaigns, legislation amendments and educational reforms to promote healthier lifestyle but with escalating obesity rates, a more aggressive approach is certainly needed. Griggs & Sabel argue for the benefits of aggressive promotion of lifestyle changes by physicians as part of the cancer patient treatment plan⁽ Reference Griggs and Sabel ^{158 )}. However, obesity is a complex variable, and BMI is not always a reliable measure of metabolic health. Schmitz et al. suggest that further clarification of the biological-social-environmental feedback loop is required in order to elucidate the combined and independent contributions of race, ethnicity, co-morbidities and obesity on cancer survival and adverse treatment effects⁽ Reference Schmitz, Neuhouser, Agurs-Collins, Zanetti, Cadmus-Bertram and Dean ^{159 )}.

This review highlights the critical importance of understanding the obesity–immunity–cancer axis and the fact that obesity driven immunological dysregulation should no longer be ignored when developing immunotherapeutic strategies. Immunotherapies such as pembrolizumab (Merck), nivolumab and ipilimumab (Bristol Meyer Squibb) have recently gained Food and Drug Administration approval and are so far, proving effective for a number of different cancers (www.clinicaltrials.gov). However, their suitability has not been thoroughly scrutinised under the pre-existing inflammatory condition of obesity. The challenge of using checkpoint inhibitors in obesity-associated cancer is to alleviate immune suppression and augment anti-tumour responses without inducing excessive pro-tumourigenic inflammatory cytokine production by T cells, which have already been shown to drive adipose tissue inflammation in obesity⁽ Reference Lysaght, Allott, Donohoe, Howard, Pidgeon and Reynolds ^{65 ,} Reference Nishimura, Manabe, Nagasaki, Eto, Yamashita and Ohsugi ^{80 )}. In the context of chronic obesity-associated inflammation, there is cause for concern with such immunotherapies, since blocking programmed death-ligand 1 has previously been shown to enhance monocyte function by decreasing IL-10 and enhancing inflammatory cytokine production⁽ Reference Curiel, Wei, Dong, Alvarez, Cheng and Mottram ^{160 )}. Furthermore, aged obese mice were observed to have the most extreme pathological responses to anti-CD40 and IL-2 therapy, causing cytokine storms, organ pathology and eventual death⁽ Reference Mirsoian, Bouchlaka, Sckisel, Chen, Pai and Maverakis ^{161 )}. Therefore, the use of such elegant immunotherapeutic approaches might be theoretically desirable for obesity-associated malignancy but first, their potential to exacerbate inflammation in obesity must be fully investigated. In the context of obesity-associated cancer, targeting the correct tissue rather than a systemic immunotherapeutic approach is also of paramount importance, when aiming to enhance anti-tumour immunity without contributing to adipose tissue and systemic inflammation. For instance, augmenting anti-tumour immunity without exacerbating inflammation might be achieved through the adoptive transfer of T or NK cells, already primed with a chemotactic signal toward the tumour site and not the adipose tissue, as demonstrated by Wennerberg et al. ⁽ Reference Wennerberg, Kremer, Childs and Lundqvist ^{162 )}. Similarly, approaches might combine adoptive transfer with chemokine treatments to enhance anti-inflammatory cell migration to the VAT and attenuate pathological inflammation. For instance, M2 macrophages, T_reg cell and iNKT cell numbers might be replenished in the obese VAT to restore immune balance and prevent progression to inflammation-driven disease. Already, adoptive transfer of iNKT cells has been shown to reduce weight loss and improve insulin sensitivity⁽ Reference Lynch, Nowak, Varghese, Clark, Hogan and Toxavidis ^{114 )}. Furthermore, the Th2 proportion of CD4⁺ T cells adoptively transferred into mouse models of obesity has also been shown to reverse weight gain and insulin resistance⁽ Reference Winer and Winer ^{163 )}. However the specific chemokine pathways guiding the key T, NK and iNKT cells to the tumour and to inflamed tissue sites must first be elucidated before such combination immunotherapies can be considered.

It is now evident that several critical factors must be taken into consideration when developing successful immunotherapeutic approaches. Firstly, evaluation of the immunological environment must be considered. Overwhelming data shows a dearth of innate lymphocytes or their inactivation and general dysregulation in obesity (e.g. γδ T cells and MAIT cells are reduced with increasing BMI)⁽ Reference Costanzo, Taylor, Dutt, Han, Fujioka and Jameson ^{87 ,} Reference Carolan, Tobin, Mangan, Corrigan, Gaoatswe and Byrne ^{89 ,} Reference Magalhaes, Pingris, Poitou, Bessoles, Venteclef and Kiaf ^{90 )}. It may be useful to evaluate the level of immune involvement in the tumour microenvironment, and evaluate the degree of immune cell dysregulation prior to devising a patient treatment plan. Indeed, evaluation of immunological parameters is presently in development as a prognostic aid, and shows superior predictive ability compared with traditional histopathological staging methods in colorectal cancer⁽ Reference Pagès, Berger, Camus, Sanchez-Cabo, Costes and Molidor ^{164 –} Reference Park, McMillan, Powell, Richards, Horgan and Edwards ^{166 )}. Galon et al. show that evaluation of the type, density and location of immune cell infiltration into tumours can allow prediction of patient survival⁽ Reference Galon, Costes, Sanchez-Cabo, Kirilovsky, Mlecnik and Lagorce-Pagès ^{165 )}. A greater understanding of the underlying biology behind immunotherapy is also urgently required. Checkpoint inhibitors such as ipilimumab and tremelimumab have been hailed as breakthrough drugs, but show no benefit in the majority of patients and there is a paucity of published data in the obesity-associated cancer space⁽ Reference Snyder, Makarov, Merghoub, Yuan, Zaretsky and Desrichard ^{167 )}. Recent work shows there may be a genetic basis for clinical responses in a cohort of melanoma patients⁽ Reference Snyder, Makarov, Merghoub, Yuan, Zaretsky and Desrichard ^{167 )}. Improved understanding of the immunological landscape in obesity and cancer will undoubtedly allow development of further breakthroughs, for example combination immunotherapy approaches may be advantageous, combining selective cytokine or chemokine replenishment or depletion strategies with checkpoint inhibitors, adoptive transfer and traditional chemoradiotherapy strategies or hormone therapies for maximum effect. Some success has already been reported when anti-PD-1 (programmed cell death protein 1) and anti-CTLA4 (cytotoxic T lymphocytes antigen-4) therapies are administered together⁽ Reference Curran, Montalvo, Yagita and Allison ^{168 )}. Replenishment of growth factors that are lacking in obesity may be a useful addition to existing immunotherapeutic strategies, such as reconstituting IL-2 levels in order to boost γδ T cell responses. Simultaneous neutralisation of certain pro-inflammatory cytokines (e.g. IL-17, IL-1, IL-6) may be also be warranted in obese patients to fully overcome immunological dysregulation. Optimal dosing and timing strategies will also require further consideration, particularly in regard to treating patients with elevated BMI⁽ Reference Drake ^{169 )}.

In conclusion, global health is now plagued by pathologies that have arisen from an obesity epidemic that shows little sign of abating. This review has focused on the potential role of a dysregulated immune system in a growing number of obesity-associated cancers. With a new generation of immunotherapies which includes the checkpoint inhibitors nivolumab and ipilimumab anticipated to change the face of cancer treatment and improve survival for a number of cancer types, it is important that the chronic inflammation underpinning obesity-associated cancer is better understood. Over the past 10 years, there has been an increase in studies in human subjects investigating the phenotypes and functions of immune cells in adipose tissue, together with their role in obesity-associated disease, including cancer, however at this time there is scant understanding of how this knowledge can be used to therapeutic advantage with targeted approaches. Accordingly, further research investigating the immune compartments and chemokine and cytokine networks underpinning inflammation-driven obesity-related malignancy in human subjects is of immediate relevance and importance. Also, in the wake of multiple campaigns which have failed to halt growing obesity rates, more hard-line approaches are urgently required to promote lifestyle change and nutritional education. Childhood obesity must be tackled to prevent a further upsurge in obesity-associated cancer incidence among future generations. In addition to laboratory-based research, a joint effort between clinicians, research scientists, dietitians and governing bodies is needed to focus on the preventability of obesity through diet and lifestyle. Exploring these avenues will allow a multi-pronged approach to combating this burgeoning health crisis.