Overweight, obesity and risk of hormone-related cancers

Being overweight or obese has been associated with a reduced risk of breast cancer in the premenopausal years (McTiernan, Reference McTiernan2003), although a loss of at least 4·5 kg of body weight between the ages of 18 and 30 years was recently reported to halve the risk of developing breast cancer between the age of 30 and 49 years (Kotsopoulos et al. Reference Kotsopoulos, Olopado and Ghadirian2005). In postmenopausal women, being overweight or obese is associated with a 30–50 % increased risk of developing breast cancer (for a review, see McTiernan, Reference McTiernan2003), though the role of central adiposity in breast cancer risk is more equivocal (Harvie et al. Reference Harvie, Hooper and Howell2003). In addition, a positive association between the risk of oestrogen receptor positive/progesterone receptor positive (ER+/PR+) but not ER − / PR −  breast cancer and BMI was reported for postmenopausal women (Enger et al. Reference Enger, Ross, Paganini-Hill, Carpenter and Bernstein2000), suggesting a hormone-mediated mechanism. Up to 60 % of women diagnosed with breast cancer experience an increase in body weight associated with chemotherapy and treatment-related menopause (Holmes & Kroenke, Reference Holmes and Kroenke2004) and there is evidence that heavier women and women who gain weight after diagnosis have a poorer prognosis and an increased risk of disease recurrence and death compared with normal-weight women, irrespective of menopausal status (Kroenke et al. Reference Kroenke, Chen, Rosner and Holmes2005).

The association between obesity and prostate cancer risk seems less clear, with studies reporting a positive association between BMI and prostate cancer risk, no association and even decreased risk in men under 60 years of age, or those with a family history of prostate cancer (for a review, see Freedland & Aronson, Reference Freedland and Aronson2005). The recent inverse association that has been reported between BMI and prostate cancer risk (Giovannucci et al. Reference Giovannucci, Rimm, Liu, Leitzmann, Wu, Stampfer and Willett2003) is particularly intriguing, as obesity is characterised by lower circulating testosterone levels in men (Field et al. Reference Field, Colditz, Willett, Longcope and McKinlay1994) which could provide a mechanistic link. However, a recent case–control study showed that high visceral fat area and high visceral fat:subcutaneous fat ratio (as quantified by computer tomography) is associated with a more than four-fold increased risk of prostate cancer (von Hafe et al. Reference von Hafe, Pina, Perez, Tavares and Barros2004). There is also a growing body of evidence which shows that obesity is an independent predictor of prostate cancer recurrence, adverse pathological features and biochemical progression in prostate cancer patients following treatment with radical prostatectomy (Bassett et al. Reference Bassett, Cooperberg, Sadetsky, Silva, DuChane, Pasta, Chan, Anast, Carroll and Kane2005; Freedland et al. Reference Freedland, Grubb, Yiu, Humphreys, Nielsen, Mangold, Isaacs and Partin2005).

Physical activity and risk of common hormone-related cancers

Prospective epidemiological studies have shown that physical fitness is a strong predictor of cancer mortality in men but not women, with men in the highest physical fitness category having significantly lower risk of cancer mortality in comparison with their sedentary counterparts (Kampert et al. Reference Kampert, Blair, Barlow and Kohl1996; Evenson et al. Reference Evenson, Stevens, Cai, Thomas and Thomas2003). However, epidemiological evidence of an inverse relationship between physical activity level and risk of hormone-related cancers has been more consistent for breast cancer than for prostate cancer. A comprehensive review of the literature by Gammon et al. (Reference Gammon, John and Britton1998) showed that eighteen out of twenty-five studies reported a decreased risk of breast cancer among women who were most active at work (risk reduction of 18–52 %) or during their leisure time (risk reduction of 12–60 %) in comparison with sedentary women, although a dose–response relationship was not evident in most of the studies (Gammon et al. Reference Gammon, John and Britton1998). It was unclear whether all physically active women are at decreased risk, or whether the risk reduction is restricted to premenopausal or postmenopausal women only. Furthermore, the wide variety of study designs (including prospective and case–control studies) and problems associated with accurately assessing physical activity status contributed to the conflicting findings on the optimal time period, duration, frequency, or intensity of physical activity required to minimise breast cancer risk. For prostate cancer, the evidence is more inconsistent, with about half of the prospective cohort and case–control studies so far completed reporting a reduction in risk in the range of 10–30 % for the most physically active men, but with an increased risk of prostate cancer being observed among the most physically active men in some studies (Thune & Furberg, Reference Thune and Furberg2001; Friedenreich & Orenstein, Reference Friedenreich and Orenstein2002).

Recent evidence has shown that women who have been treated for breast cancer can improve their chances of survival by engaging in a physically active lifestyle (Holmes et al. Reference Holmes, Chen, Feskanich, Kroenke and Colditz2005). Compared with women who were least physically active, the risk of death over an average of 8 years of follow-up was halved in breast cancer patients who reported participating in the equivalent of 3–5 h of walking exercise per week at average pace. The benefit of physical activity was particularly apparent in women whose tumours were over-expressing oestrogen and progesterone receptors, consistent with a hormone-mediated mechanism. No detrimental effects of vigorous physical activity participation were found, but neither was there any evidence that higher weekly energy expenditures were associated with increased benefit (Holmes et al. Reference Holmes, Chen, Feskanich, Kroenke and Colditz2005). No epidemiological studies have investigated the effects of physical activity on disease-free survival in patients recovering from prostate cancer treatment.

Lifestyle change and cancer risk: the role of sex steroid hormones and insulin-like growth factor axis proteins

Since circulating sex steroid hormones and growth factors can be influenced by lifestyle factors such as diet, physical activity level and BMI (Friedenreich & Orenstein, Reference Friedenreich and Orenstein2002; Barnard & Aronson, Reference Barnard and Aronson2005), an increasing number of intervention studies are investigating how lifestyle changes that promote negative energy balance could affect the risk of hormone-related cancers by modulating circulating levels of sex steroid hormones and insulin-like growth factor (IGF) axis proteins. Intervention studies have focused on the effects of increased physical activity levels, dietary fat restriction and combined exercise and dietary interventions. Consumption of low-fat diets is associated with reduced energy intake in comparison with diets with higher fat content (Lissner et al. Reference Lissner, Levitsky, Strupp, Kalkwarf and Roe1987). However, the independent effects of dietary fat restriction and other changes to the quality of the diet, including fibre intake, cannot be overlooked even though the link between dietary fat consumption and risk of hormone-related cancers is unproven (Schuurman et al. Reference Schuurman, van den Brandt, Dorant and Goldbohm1999; Chan et al. Reference Chan, Stampfer, Ma, Gann, Gaziano and Giovannucci2001; Michaud et al. Reference Michaud, Augustsson, Rimm, Stampfer, Willet and Giovannucci2001; Smith-Warner et al. Reference Smith-Warner, Spiegelman and Adami2001).

Sex steroid hormones and risk of hormone-related cancers

Reduced dietary fat, physical activity and circulating sex steroid hormones

A meta-analysis of dietary fat intervention studies, in which percentage energy from fat intake was changed to 18–25 % in eleven studies and 10–12 % in two studies, showed that reduced dietary fat intake was associated with decreases in serum oestradiol of 7·4 % in premenopausal women (measured at various points of the menstrual cycle) and 23 % in postmenopausal women (Wu et al. Reference Wu, Pike and Stram1999). In premenopausal women, postmenopausal women and postmenopausal breast cancer survivors, decreases in circulating oestrone and oestrone sulfate and oestradiol have all been observed following short-term low-fat dietary interventions in which the total energy intake from fat was restricted to 10–25 % from 3 weeks to 6 months duration (Rose et al. Reference Rose, Boyar, Cohen and Strong1987, Reference Rose, Chlebowski, Connolly, Jones and Wynder1992, Reference Rose, Connolly, Chlebowski, Buzzard and Wynder1993; Boyar et al. Reference Boyar, Rose, Loughridge, Engle, Palgi, Laakso, Kinne and Wynder1988; Woods et al. Reference Woods, Gorbach, Longcope, Goldin, Dwyer and Morrill-LaBrode1989, Reference Woods, Barnett, Spiegelman, Trail, Hertzmark, Longcope and Gorbach1996; Heber et al. Reference Heber, Ashley, Leaf and Barnard1991; Goldin et al. Reference Goldin, Woods, Spiegelman, Longcope, Morrill-LaBrode, Dwyer, Gualtieri, Hertzmark and Gorbach1994). However, in these populations, there is also evidence that SHBG decreases in response to low-fat diets or moderate restriction of energy intake (Rose et al. Reference Rose, Connolly, Chlebowski, Buzzard and Wynder1993; Goldin et al. Reference Goldin, Woods, Spiegelman, Longcope, Morrill-LaBrode, Dwyer, Gualtieri, Hertzmark and Gorbach1994; Crave et al. Reference Crave, Fimbel, Lejeune, Cugnardey, Dechaud and Pugeat1995; Woods et al. Reference Woods, Barnett, Spiegelman, Trail, Hertzmark, Longcope and Gorbach1996), potentially increasing sex steroid hormone bioavailability, although more severe short-term restriction of energy intake (1380 kJ (330 kcal)/d) increases circulating SHBG levels (Franks et al. Reference Franks, Kiddy, Hamilton-Fairley, Bush, Sharp and Reed1991).

In postmenopausal breast cancer survivors, some recent preliminary evidence shows that eating a low-fat diet could significantly improve relapse-free survival. Women who reduced their dietary fat intake to 20 % of the total energy were compared with a control group eating a standard diet containing a higher proportion of fat. After 5 years, breast cancer had recurred in 9·8 % of the women in the intervention group in comparison with 12·4 % in the controls but the dietary effect appeared to be greater in women with hormone receptor-negative disease (Chlebowski et al. Reference Chlebowski, Blackburn, Elashoff, Thomson, Goodman, Shapiro, Giuliano, Karanja, Hoy and Nixon2005). This suggests that the apparent protective effect of the lower-fat diet was not associated with a hormone-mediated mechanism. These preliminary data did not control for fruit and vegetable consumption, chemotherapy treatment or the effects of weight loss, which averaged nearly 2 kg in the intervention group.

In middle-aged men, change from a high-fat, low-fibre diet to a low-fat, high-fibre diet over a period of 8 weeks resulted in a decrease in circulating testosterone and adrenal androgens (androstenedione and DHEA), and smaller decreases in circulating concentrations of oestradiol and SHBG without altering urinary clearance (Wang et al. Reference Wang, Catlin, Starcevic, Heber, Ambler, Berman, Lucas, Leung, Schramm, Lee, Hull and Swerdloff2005). A decrease in circulating total testosterone and SHBG-bound testosterone has also been reported in men aged 19–56 years after changing from a high-fat, low-fibre diet to a low-fat, high-fibre diet (Dorgan et al. Reference Dorgan, Judd and Longcope1996a).

Participation in physical activity is reported to be associated with changes in menstrual cycle duration and circulating oestrogens that could potentially influence breast cancer risk. In premenopausal women, regular exercise is associated with longer menstrual cycles which might be due to suppressed release of gonadotrophin-releasing hormone (Walberg-Rankin et al. Reference Walberg-Rankin, Franke and Gwazdauskas1992; Cooper et al. Reference Cooper, Sandler, Whelan and Smith1996) and can delay the onset of the menstrual cycle in adolescent females or cause anovulation so that exposure to oestrogen and progesterone is reduced or delayed (Bernstein et al. Reference Bernstein, Ross, Lobo, Hanisch, Krailo and Henderson1987; Merzenich et al. Reference Merzenich, Boeing and Wahrendorf1993). In postmenopausal women, increased levels of physical activity were reported to be associated with decreased serum concentrations of oestrone (but not oestradiol or testosterone) (Nelson et al. Reference Nelson, Meredith, Dawson-Hughes and Evans1988), whereas 4–5 months of moderate-intensity physical exercise resulted in decreased circulating levels of SHBG (Caballero & Maynar, Reference Caballero and Maynar1992; Caballero et al. Reference Caballero, Mahedero, Hernandez, Alvarez, Rodriguez, Rodriguez and Maynar1996), which could act to increase sex steroid hormone bioavailability.

There is limited evidence from short-term combined programmes of exercise and low-fat diets in premenopausal or postmenopausal women or in middle-aged men. However, in a study of postmenopausal women following a low-fat (10 % of total energy) high-fibre (8·4–9·6 g dietary fibre/100 kJ (35–40 g dietary fibre/1000 kcal) per d) complex-carbohydrate diet combined with exercise (walking exercise 6–7 d/week), serum insulin was decreased by 39 and 19 % in those taking and not taking hormone replacement therapy, respectively, and was associated with corresponding increases in SHBG of 39 and 42 % in the two groups, which could decrease sex steroid hormone bioavailability (Tymchuk et al. Reference Tymchuk, Tessler and Barnard2000). However, a small feasibility study, involving ten breast cancer survivors who were mainly postmenopausal, only reported slight non-significant changes in free oestradiol, oestrone sulfate, total testosterone, androstenedione and DHEA after a combined 8-week low-fat (20 % of total energy) and aerobic exercise intervention (McTiernan et al. Reference McTiernan, Ulrich, Kumai, Bean, Schwartz, Mahloch, Hastings, Gralow and Potter1998a) in which a moderate 1·2 kg loss of body weight was also observed (Table 1). In middle-aged men following a combined 3-week intensive low-fat-diet (10 % fat; 10–15 % protein; 75–80 % high-complex carbohydrate) and exercise intervention (daily walking and supervised exercise), serum oestradiol was halved but without change in serum testosterone levels (Rosenthal et al. Reference Rosenthal, Barnard, Rose, Inkeles, Hall and Pritikin1985). In another study by the same group (Tymchuk et al. Reference Tymchuk, Tessler, Aronson and Barnard1998), an increase in serum SHBG was accompanied by a decrease in fasting insulin in middle-aged men following the same intervention (Table 1).

Table 1 The effects of combined low-fat and aerobic exercise interventions on circulating levels of insulin, sex steroid hormones and/or insulin-like growth factor (IGF) axis proteins

DE, combined low-fat diet and aerobic exercise intervention; C, control group; ↑ , increase; HRT, hormone replacement therapy; ST, short-term; ↓ , decrease; SHBG, sex hormone-binding globulin; PSA, prostate-specific antigen; LT, long-term; LNCaP, androgen-dependent prostate cancer cell line; FBS, fetal bovine serum; IGFBP, insulin-like growth factor binding protein; E, aerobic exercise-only intervention; DHEA, dehydroepiandrosterone; CRP, C-reactive protein.

The urinary products of oestrogen metabolism, 16α-hydroxyoestrone and 2-hydroxyoestrone, have also been studied in relation to breast cancer risk and lifestyle interventions associated with negative energy balance in premenopausal and postmenopausal women. Two published prospective studies reported a trend for a higher urinary 2-hydroxyoestrone:16α-hydroxyoestrone ratio to be associated with a reduced risk of breast cancer in premenopausal and postmenopausal women (Meilahn et al. Reference Meilahn, De, Allen, Fentiman, Bradlow, Sepkovic and Kuller1998; Muti et al. Reference Muti, Bradlow and Micheli2000). However, studies that have investigated the effect of exercise and dietary interventions on this oestrogen metabolite ratio (to establish whether the pathway of oestrogen metabolism can be manipulated by lifestyle changes) have yielded inconsistent results (Longcope et al. Reference Longcope, Gorbach, Goldin, Woods, Dwyer, Morrill and Warram1987; Pasagian-Macaulay et al. Reference Pasagian-Macaulay, Meilahn, Bradlow, Sepkovic, Buhari, Simkin-Silverman, Wing and Kuller1996; Atkinson et al. Reference Atkinson, Lampe and Tworoger2004; Matthews et al. Reference Matthews, Fowke, Dai, Leon, Jin, Shu, Gao, Longcope, Hebert and Zheng2004; Campbell et al. Reference Campbell, Westerlind, Harber, Friedenreich and Courneya2005).

Of all the possible combinations of negative lifestyle interventions (low-fat diet with or without exercise, exercise alone), the most consistent evidence for an oestrogen-lowering effect exists for restricted dietary fat consumption. A number of mechanisms could explain the decrease in circulating oestrogen levels following reduced dietary fat consumption. Decreased rates of aromatisation in adipose tissue, leading to decreased oestradiol production, could result from a decrease in fat mass or through the conditions of negative energy balance modulating cellular enzyme activity (Heber et al. Reference Heber, Ashley, Leaf and Barnard1991). In addition, a reduction in dietary fat intake as part of an energy-restriction diet can affect β-glucoronidase activity in the intestinal flora (Goldin & Gorbach, Reference Goldin and Gorbach1976), which deconjugates biliary oestrogens so they can be reabsorbed from the intestinal tract, with the remaining conjugated oestrogens being excreted (Goldin et al. Reference Goldin, Adlercreutz, Gorbach, Warram, Dwyer, Swenson and Woods1982). β-Glucoronidase activity increases with high-fat diets, especially high saturated fat intake (Goldin & Gorbach, Reference Goldin and Gorbach1976) which would decrease excretion. Increased faecal oestrogen excretion in vegetarian compared with omnivorous women may also support and emphasise an important role for dietary fibre in the control of oestrogen levels (Goldin et al. Reference Goldin, Adlercreutz, Gorbach, Warram, Dwyer, Swenson and Woods1982). Added benefits of increasing fibre intake during the consumption of low-fat diets might accrue from the presence in high-fibre diets of naturally occurring substances with weak oestrogenic and anti-oestrogenic activity, which may induce production of SHBG in the liver and positively influence sex hormone metabolism and the resulting biological effects (Adlercreutz et al. Reference Adlercreutz, Fotsis, Bannwart, Wahala, Makela, Brunow and Hase1986).

Insulin-like growth factor axis proteins and risk of hormone-related cancers

The IGF system consists of the IGF ligands (IGF-1 and IGF-2), cell surface receptors that mediate the biological effects of the IGF, including the IGF-1 and IGF-2 receptors, the insulin receptor and a family of IGF binding proteins (IGFBP-1–6) (LeRoith & Roberts, Reference LeRoith and Roberts2003). The IGF have an important role in normal growth and development, as well as in a variety of pathological situations, including tumorigenesis (Khandwala et al. Reference Khandwala, McCutcheon, Flyvbjerg and Friend2000). IGF are potent mitogens for diverse cancer cell lines in vitro (Maloney et al. Reference Maloney, McLaughlin, Dagdigian, Garrett, Connors, Zhou, Blattler, Chittenden and Singh2003) and also suppress apoptosis (Parrizas & LeRoith, Reference Parrizas and LeRoith1997; Heron-Milhavet & LeRoith, Reference Heron-Milhavet and LeRoith2002). Extensive evidence shows that breast cancer cells are responsive to exogenous IGF and tamoxifen reduces circulating IGF-1 levels (Pollak, Reference Pollak1998). IGF also exert strong mitogenic and anti-apoptotic effects on normal and transformed prostatic epithelial cells both in vitro and in vivo (Cohen et al. Reference Cohen, Peehl, Lamson and Rosenfeld1991; Angelloz-Nicoud & Binoux, Reference Angelloz-Nicoud and Binoux1995; Torring et al. Reference Torring, Vinter-Jensen, Pedersen, Sorensen, Flyvbjerg and Nexo1997) and modulate growth of the prostate carcinoma cell lines LNCaP, PC-3 and DU-145 in vitro (Pietrzkowski et al. Reference Pietrzkowski, Mulholland, Gomella, Jameson, Wernicke and Baserga1993; Ngo et al. Reference Ngo, Barnard, Leung, Cohen and Aronson2003). Up regulation of IGF-1 receptor expression could augment the response to IGF-1. Compared with equivalent healthy tissues, the IGF-1 receptor is over-expressed in human malignancies, including cancers of the breast and prostate (Hellawell et al. Reference Hellawell, Turner, Davies, Poulsom, Brewster and Macaulay2002; Perks & Holly, Reference Perks and Holly2003).

Hepatic production of IGF axis proteins is mainly dependent on growth hormone secretion (Janssen et al. Reference Janssen, Wildhagen, Ito, Blijenberg, Van Schaik, Roobol, Pols, Lamberts and Schroder2004), but circulating levels are also affected by age and sex, insulin, nutritional status, the host response to systemic diseases such as cancer and endocrine therapy, whereas the locally expressed components are controlled by factors specific to each individual tissue (Rajaram et al. Reference Rajaram, Baylink and Mohan1997; Perks & Holly, Reference Perks and Holly2003). IGF-1 is the major bioactive component among the IGF and shares significant structural homology with insulin (Yu & Rohan, Reference Yu and Rohan2000). IGF-1 bioavailability is influenced by circulating free levels and production in the tissues, as well as tissue expression of the IGF receptors. However, IGF bioavailability in the circulation and extracellular fluids is also modulated by the IGFBP, which bind with varying relative affinity for IGF-1 and IGF-2 (Foulstone et al. Reference Foulstone, Prince, Zaccheo, Burns, Harper, Jacobs, Church and Hassan2005). The most frequently studied binding proteins in relation to cancer risk are IGFBP-1 and IGFBP-3, which are secreted primarily by the liver (Goodwin et al. Reference Goodwin, Ennis, Pritchard, Trudeau, Koo, Hartwick, Hoffma and Hood2002a). Over 90 % of the circulating IGF-1 is bound in a ternary complex with IGFBP-3 (the most abundant circulating binding protein) and a glycoprotein, the acid-labile subunit, and this complex, which constitutes a storage pool of IGF-1 in the blood (Allen et al. Reference Allen, Appleby, Kaaks, Rinaldi, Davey and Key2003), is too large to cross the endothelial membrane. IGFBP-1 is a smaller protein structure that does not form a ternary complex and can pass to the extracellular tissue. Thus, IGFBP-1 is found predominantly in the tissues (although detectable levels circulate in the blood) and acts primarily to inhibit the physiological effects of IGF-1 (Rajaram et al. Reference Rajaram, Baylink and Mohan1997). Changes in IGFBP production resulting from interventions or lifestyle changes affect IGF bioavailability. IGFBP are also subject to potentially regulating proteolysis by various proteases secreted by prostate and breast cancer cells that may act as growth stimulators by increasing local IGF bioavailability (Baxter, Reference Baxter2000).

A recent systematic review and meta-regression analysis of twenty-six datasets from twenty-one studies conducted by Renehan et al. (Reference Renehan, Zwahlen, Minder, O'Dwyer, Shalet and Egger2004) showed that high circulating levels of IGF-1 (highest quartile v. lowest quartile) was associated with a 49 % increased risk of prostate cancer and a 65 % increased risk of premenopausal breast cancer, whereas no association was found for postmenopausal breast cancer. This comprehensive review considered evidence from fifteen prospective studies (usually nested within larger cohorts) and six case–control studies. Prospective studies are preferred because circulating levels of factors that could be implicated in cancer risk can be increased by the disease process itself, and this can undermine the predictive validity of case–control studies. However, re-analysis of the systematic review and meta-analysis presented by Renehan et al. (Reference Renehan, Zwahlen, Minder, O'Dwyer, Shalet and Egger2004) showed that the strengths of association for prostate cancer and premenopausal breast cancer were increased when the case–control studies were omitted from the analysis (Holly, Reference Holly2004).

No consistent association between higher circulating levels of IGFBP-3 and reduced risk of hormone-related cancers was found in a recent systematic review and meta-analysis (Renehan et al. Reference Renehan, Zwahlen, Minder, O'Dwyer, Shalet and Egger2004) and the association between circulating IGFBP-1 and cancer risk has been less well studied. In postmenopausal breast cancer patients, circulating levels of IGFBP-1 significantly predicted distant disease recurrence and death (but the effects were not independent of fasting insulin levels), whereas circulating IGFBP-3 concentrations only predicted distant disease recurrence (Goodwin et al. Reference Goodwin, Ennis, Pritchard, Trudeau, Koo, Hartwick, Hoffma and Hood2002a). However, no association between circulating IGFBP-3 and disease recurrence or disease-free survival was reported in another study (Rocha et al. Reference Rocha, Hilsenbeck, Jackson, VanDenBerg, Weng, Lee and Yee1997). In prostate cancer patients, serum IGFBP-3 levels were low in comparison with normal controls (Kanety et al. Reference Kanety, Madjar, Dagan, Levi, Papa, Pariente, Goldwasser and Karasik1993) and lower in patients with advanced disease v. those with localised tumours and benign prostate hyperplasia (Miyata et al. Reference Miyata, Sakai, Hayashi and Kanetake2003). There is evidence that IGFBP-1 and IGFBP-3 can inhibit the growth of breast and prostate cancer cell lines in vitro (Figueroa et al. Reference Figueroa, Lee, Jackson and Yee1995; Butt et al. Reference Butt, Fraley, Firth and Baxter2002; Wilson et al. Reference Wilson, Birnbaum, Poot, Quinn and Swisshelm2002). As IGFBP limit IGF bioavailability, increased circulating and/or tissue levels could blunt the proliferative effect of IGF on normal and transformed cancer cells. In addition, binding protein proteases such as PSA can modulate IGF bioavailability in the cellular milieu (Pollak, Reference Pollak1998). In this respect, the IGFBP-3:PSA ratio was identified as an independent predictor of relapse-free and cause-specific survival in advanced prostate cancer patients who were treated with hormonal therapy (Miyata et al. Reference Miyata, Sakai, Hayashi and Kanetake2003). However, in some cellular environments IGFBP-3 is anti-apoptotic, as shown with the Hs578T breast cancer cell line (McCaig et al. Reference McCaig, Perks and Holly2002) and this might explain the recently reported positive association between higher IGFBP-3 and premenopausal breast cancer risk in a recent systematic review and meta-analysis (Renehan et al. Reference Renehan, Zwahlen, Minder, O'Dwyer, Shalet and Egger2004).

Dietary energy restriction, physical activity and circulating insulin-like growth factor axis proteins

IGF-1 levels decrease in response to fasting and protein–energy malnutrition, but levels are restored in response to improvements in energy intake and are increased above normal levels and overfeeding (Yu & Rohan, Reference Yu and Rohan2000). IGFBP-1 levels are also influenced by nutritional state, with serum levels being highest during fasting and lowest in the fed state, and an inverse association between circulating IGFBP-1 concentration and insulin secretion (Musey et al. Reference Musey, Goldstein, Farmer, Moore and Phillips1993; Ferry et al. Reference Ferry, Katz, Grimberg, Cohen and Weinzimer1999). A 50 % reduction in energy intake over 6 d caused a significant increase in IGFBP-1 in normal adults (Smith et al. Reference Smith, Underwood and Clemmons1995). Conversely, circulating IGFBP-1 levels are chronically reduced in conditions of elevated fasting insulin levels, such as obesity and insulin resistance (Kaaks & Lukanova, Reference Kaaks and Lukanova2001). Although circulating IGFBP-3 may be less acutely affected by the metabolic state (Underwood et al. Reference Underwood, Thissen, Lemozy, Ketelslegers and Clemmons1994; Smith et al. Reference Smith, Underwood and Clemmons1995), circulating IGFBP-3 decreases with prolonged energy and/or protein restriction (Smith et al. Reference Smith, Underwood and Clemmons1995; Underwood, Reference Underwood1996).

Short-term vigorous aerobic exercise training (in the absence of frank over-feeding) appears to decrease the circulating IGF-1, even in weight-stable individuals (Nemet et al. Reference Nemet, Connolly, Pontello-Pescatello, Rose-Gottron, Larson, Galassetti and Cooper2004). However, evidence for an attenuating effect of regular moderate intensity physical activity on circulating IGF-1 levels is equivocal, as the association between a physically active lifestyle and circulating IGF-1 axis proteins is complex (for a review, see Yu & Rohan, Reference Yu and Rohan2000), being confounded by dietary patterns, age and type of physical activity performed. In premenopausal women, a non-significant trend for increased levels of physical activity to be associated with lower IGF-1 and IGFBP-3 levels and increased levels of IGFBP-1 was reported by Voskuil et al. (Reference Voskuil, Bueno de Mesquita, Kaaks, van Noord, Rinaldi, Riboli, Grobbee and Peeters2001), whereas a 12-month moderate-intensity aerobic exercise programme (5 d/week) had no effect on serum concentrations of IGF-1 or IGFBP-3 (or the ratio of the two) in overweight postmenopausal women (McTiernan et al. Reference McTiernan, Sorensen, Yasui, Tworoger, Ulrich, Irwin, Rudolph, Stanczyk, Schwartz and Potter2005). Furthermore, there were no changes in resting circulating levels of IGF-1, IGFBP or testosterone following a 6-month programme of resistance training in women aged 60–85 years (Borst et al. Reference Borst, Vincent, Lowenthal and Braith2002) or following a 12-week programme of endurance exercise in elderly women (Lange et al. Reference Lange, Lorentsen, Isaksson, Juul, Rasmussen, Christensen, Bulow and Kjaer2001). Similar findings have been reported following a strength training programme in postmenopausal breast cancer survivors (Schmitz et al. Reference Schmitz, Ahmed, Hannan and Yee2005a). However, significantly greater changes in circulating levels of IGF-1 (decrease), IGFBP-3 (increase) and the IGF-1:IGFBP-3 ratio (decrease) were observed following a 15-week aerobic exercise intervention in women recovering from breast cancer treatment v. a patient control group (Fairey et al. Reference Fairey, Courneya, Field, Bell, Jones and Mackey2003). In this study, there was no difference between the increases in circulating IGFBP-1 recorded for the exercise (+5·6 %) or control (+4·2 %) groups and weight loss was minimal, showing that potentially beneficial changes in IGF axis proteins can be induced by exercise without loss of body weight (Fairey et al. Reference Fairey, Courneya, Field, Bell, Jones and Mackey2003). The authors also highlighted the similarity of changes in IGF axis proteins evoked by oestrogen receptor modulation therapies (raloxifene and tamoxifen) in postmenopausal breast cancer survivors. The strengths of the latter two studies include the randomised controlled designs, high exercise adherence rate, and with minimal loss to follow-up, although subject numbers were limited to between twenty-five and thirty-six in the exercise and control groups.

In middle-aged men, a series of insightful studies have investigated the effects of exercise, low-fat diets and combined low-fat diet and exercise programmes on IGF axis markers and the mechanisms by which these interventions might help to reduce the risk of prostate cancer. The effects of short- and long-term combined programmes of low-fat diet and aerobic exercise on circulating IGF axis proteins and insulin were compared in overweight middle-aged men by Ngo et al. (Reference Ngo, Barnard, Tymchuk, Cohen and Aronson2002). Men recruited for the short-term programme consumed a low-fat diet ( < 10 % fat, 10–15 % protein and 75–80 % carbohydrate) and participated in daily aerobic exercise (60 min/d) for 11 d. These were compared with men who had been following a similar regimen for an average of 14·2 years. The researchers also examined the effects of changes in circulating IGF axis proteins and sex steroid hormones on prostate cancer cell growth in vitro by incubating baseline and post-intervention serum with the androgen-dependent prostate cancer cell line LNCaP in cell culture. Fasting insulin levels were 25 and 68 % lower than baseline levels following the short- and long-term combined programmes, respectively, corresponding to decreases in serum IGF-1 of 20 and 55 %, increases in IGFBP-1 of 53 and 150 % and decreases in body weight of 4 and 38 %, respectively. There was no change in circulating IGFBP-3 concentration in either group. LNCaP apoptosis was increased and cell growth inhibited by 30 % (short-term intervention) and 44 % (long-term intervention) when cells were incubated with post-intervention serum v. baseline serum in the cell-culture experiments. Serum levels of IGF-1 and IGFBP-1 were positively and negatively correlated with LNCaP growth, respectively (Table 1).

In another study by the same group, similar positive changes in fasting insulin and IGF axis proteins to those observed following the long-term combined diet and exercise programme were apparent in men who had been engaged in a long-term exercise intervention for an average time of 10 years (Barnard et al. Reference Barnard, Ngo, Leung, Aronson and Golding2003). However, the increase in circulating IGFBP-1 was reported to be greater in men who had been following the long-term combined programme than in men engaged in long-term exercise only. LNCaP cell growth was inhibited to a similar degree when incubated with serum from men on the two interventions in relation to control serum in cell culture, but significantly greater apoptosis was observed when LNCaP cells were incubated in serum from men on the combined diet and exercise intervention. This later finding could have been influenced by the greater change in IGFBP-1 observed in the group following the combined programme as all other changes in serum factors were similar (Table 1).

Other studies by the same group showed that even though the long-term combination of low-fat diet and regular exercise reduced circulating levels of sex hormones in the middle-aged men (Rosenthal et al. Reference Rosenthal, Barnard, Rose, Inkeles, Hall and Pritikin1985), changes in IGF axis proteins were shown to be mainly responsible for the ability of post-intervention serum to inhibit prostate cancer cell growth in the cell-culture experiments. Adding sex steroid hormones and insulin back to the diet and exercise serum, so that levels were comparable with baseline measures before incubation with the LNCaP cells, showed that reductions in serum oestradiol, free testosterone and insulin induced by the diet and exercise intervention accounted for only half of the inhibitory effects on cell growth observed (Tymchuk et al. Reference Tymchuk, Barnard, Ngo and Aronson2002). However, the inhibition of LNCaP cell growth was completely eliminated by adding back IGF-1 (Barnard et al. Reference Barnard, Ngo, Leung, Aronson and Golding2003). Furthermore, when IGFBP-1 was added to baseline serum, LNCaP growth was reduced and apoptosis induced (Ngo et al. Reference Ngo, Barnard, Leung, Cohen and Aronson2003). This series of studies has yielded some very interesting data on the effects of low-fat dietary and moderate-intensity exercise interventions on IGF axis proteins and sex steroid hormones, and the mechanisms by which such changes could influence prostate cancer development. However, small subject numbers, cross-sectional designs or the lack of a control group and the problem of extrapolating in vitro cell-culture data to the in vivo situation are significant limitations.

In a more recent randomised controlled trial in men with early biopsy-proven prostate cancer who had chosen not to undergo any conventional treatment (thus controlling for the confounding effects of treatment interventions), the effects of a 1-year very-low-fat vegan diet and moderate intensity exercise intervention on PSA, treatment trends and serum-stimulated LNCaP growth were investigated (Ornish et al. Reference Ornish, Weidner and Fair2005). After 1 year, PSA had increased by 6 % in the controls and decreased by 4 % in the intervention group. Body weight decreased by 4·5 kg in the intervention group and was unchanged in the controls. Six of the forty-nine control patients went on to conventional treatment (due to rising PSA) compared with none of the forty-four patients in the intervention group. In the serum-stimulated LNCaP assay, cell growth was reduced by 9 % in the controls and by 70 % in the intervention group, but there was no difference between the groups in apoptosis or serum testosterone concentration. The intention is to follow the patients for a longer time period to establish the effect of the intervention on disease recurrence rates and mortality.

Negative energy balance interventions and insulin resistance

Exercise with or without dietary interventions that promote negative energy balance has the potential to improve insulin resistance, which could profoundly influence any beneficial changes in circulating sex steroid hormones and/or IGF axis proteins with respect to cancer risk. Insulin stimulates the hepatic production of IGF and their growth-promoting effects and inhibits IGFBP-1 production (Thissen et al. Reference Thissen, Ketelslegers and Underwood1994; Yu & Rohan, Reference Yu and Rohan2000). By inhibiting IGFBP production, insulin may further increase IGF-1 bioavailability. The regulatory effect of insulin and IGF-1 on SHBG production and sex hormone bioavailability could also have important implications for the development of hormone-related cancers. Insulin and IGF-1 suppress hepatic production of SHBG, resulting in more free testosterone and oestradiol to stimulate cell growth (Plymate et al. Reference Plymate, Matej, Jones and Friedl1988; Singh et al. Reference Singh, Hamilton-Fairley, Koistinen, Seppala, James, Franks and Reed1990; Pasquali et al. Reference Pasquali, Casimirri, de Iasio, Mesini, Boschi, Chierici, Flamia, Biscotti and Vicennati1995). Insulin is also a potent mitogen for various cell lines, mediating glucose transport across the cell membrane to provide an energy source and directly activating cellular RNA and protein production via the mitogen-activating protein kinase pathway (King & Kahn, Reference King and Kahn1981). Insulin has been shown to induce mitogenic effects on normal and malignant breast epithelial cells (Belfiore et al. Reference Belfiore, Frittitta, Costantino, Frasca, Pandini, Sciacca, Goldfine and Vigneri1996; Papa & Belfiore, Reference Papa and Belfiore1996) and prostate cancer cells (Polychronakos et al. Reference Polychronakos, Janthly, Lehoux and Koutsilieris1991) in vitro and elevated circulating insulin is reported to be a risk factor for breast cancer (Bruning et al. Reference Bruning, Bonfrer, van Noord, Hart, de Jong-Bakker and Nooijen1992; Hirose et al. Reference Hirose, Toyama, Iwata, Takezaki, Hamajima and Tajima2003) and prostate cancer (Hsing et al. Reference Hsing, Chua, Gao, Gentzschein, Chang, Deng and Stanczyk2001, Reference Hsing, Gao, Chua, Deng and Stanczyk2003). In addition, high fasting insulin levels have been associated with distant recurrence and death in breast cancer survivors (Goodwin et al. Reference Goodwin, Ennis, Pritchard, Trudeau, Koo, Madarnas, Hartwick, Hoffman and Hood2002b) and visceral fat area and high visceral fat:subcutaneous fat ratio (as quantified by computer tomography) was shown to be associated with a greater than four-fold increased risk of prostate cancer (von Hafe et al. Reference von Hafe, Pina, Perez, Tavares and Barros2004).

Lower BMI, higher levels of physical activity and lower energy intake were all independently associated with lower fasting insulin levels in a recent large-scale study of mostly healthy postmenopausal women (Chlebowski et al. Reference Chlebowski, Pettinger, Stefanick, Howard, Mossavar-Rahmani and McTiernan2004). Dietary-induced weight loss improves insulin sensitivity in obese men (Ross et al. Reference Ross, Dagnone, Jones, Smith, Paddags, Hudson and Janssen2000). However, regular exercise without dietary energy restriction has also been shown to improve insulin sensitivity (Ross et al. Reference Ross, Dagnone, Jones, Smith, Paddags, Hudson and Janssen2000, Reference Ross, Janssen, Dawson, Kungl, Kuk, Wong, Nguyen-Duy, Lee, Kilpatrick and Hudson2004). Improvements in insulin sensitivity following negative energy balance interventions may be mediated by weight loss (Niskanen et al. Reference Niskanen, Uusitupa, Sarlund, Siitonen, Paljarvi and Laakso1996; Goodpaster et al. Reference Goodpaster, Kelley, Wing, Meier and Thaete1999), although improved insulin sensitivity has not been accompanied by weight loss in all exercise intervention studies (Duncan et al. Reference Duncan, Perri, Theriaque, Hutson, Eckel and Stacpoole2003). Furthermore, changes in fasting serum insulin levels have been observed after only 2–3 weeks of diet and exercise interventions, even though the individuals remained obese (BMI>30 kg/m²) (Barnard et al. Reference Barnard, Ugianskis, Martin and Inkeles1992; Tymchuk et al. Reference Tymchuk, Tessler, Aronson and Barnard1998; Ngo et al. Reference Ngo, Barnard, Tymchuk, Cohen and Aronson2002).

Exercise-induced effects on insulin sensitivity that are attributable to weight loss might be associated with the effects of negative energy balance on visceral fat stores. Evidence suggests that weight loss from energy restriction alone causes diffuse fat and lean tissue mass reductions, sometimes with minimal effects on visceral fat (Wood, Reference Wood1993) which is the most metabolically active fat depot (Krotkiewski et al. Reference Krotkiewski, Bjorntorp, Sjostrom and Smith1983). However, aerobic exercise preferentially causes reduction in visceral fat in men (Schwartz et al. Reference Schwartz, Shuman, Larson, Cain, Fellingham, Beard, Kahn, Stratton, Cerqueira and Abrass1991) and women (McTiernan et al. Reference McTiernan, Ulrich, Slate and Potter1998b) and a 1–2 kg reduction in visceral fat can have profound effects on glucose tolerance, fasting insulin and blood lipid levels (Krotkiewski et al. Reference Krotkiewski, Bjorntorp, Sjostrom and Smith1983). In postmenopausal women and middle-aged men, interventions comprising low-fat, high-carbohydrate diets in conjunction with moderate-intensity exercise training that resulted in a decrease in fasting insulin concentration, were accompanied by an increase in circulating SHBG (Tymchuk et al. Reference Tymchuk, Tessler, Aronson and Barnard1998, Reference Tymchuk, Tessler and Barnard2000). However, in postmenopausal breast cancer survivors, positive changes in circulating IGF axis proteins (including IGF-1 and IGFBP-3) after 15 weeks of moderate-intensity aerobic exercise training were independent of changes in fasting insulin or insulin resistance (Fairey et al. Reference Fairey, Courneya, Field, Bell, Jones and Mackey2003).

Other potential benefits of negative energy balance interventions in cancer survivors

Continued advances in early-detection and effective treatments have resulted in the hope of longer survival and even cure for many cancer patients. Despite an increase in UK cancer incidence in the 10-year period between 1994 and 2003, overall mortality from cancer decreased (Cancer Research UK, 2005b). Furthermore, survival data for cancer diagnosis during the period 1996–9 in the UK (Cancer Research UK, 2005c) showed that both breast and prostate cancer fell into the highest 5-year survival category (77 % for breast cancer; 65 % for prostate cancer). With the increasing ageing population, the number of elderly cancer survivors is expected to double over the next 50 years (Demark-Wahnefried et al. Reference Demark-Wahnefried, Clipp, Morey, Pieper, Sloane, Snyder and Cohen2004), which means that disease recurrence and the risk of secondary primary cancers is becoming an important issue in the management of cancer patients.

Given the increase in life expectancy following a cancer diagnosis, promotion of healthy lifestyle behaviours among survivors might help them to enjoy a higher level of physical functioning, improved cardiovascular health and better health-related quality of life. Older individuals diagnosed with cancer are at increased risk of other cancers and chronic age-related conditions and are susceptible to functional losses that can threaten independent living (Aziz, Reference Aziz2002; Hewitt et al. Reference Hewitt, Rowland and Yancik2003). It has been argued that physical activity, diet and/or weight-control interventions hold considerable promise for ameliorating the adverse sequelae of cancer (Aziz, Reference Aziz2002). Studies have shown that participation in regular physical activity and the consumption of healthy low-fat diets during and/or after treatment are associated with higher levels of physical functioning, reduced feelings of fatigue and improved health-related quality of life among breast and prostate cancer survivors (Demark-Wahnefried et al. Reference Demark-Wahnefried, Clipp, Morey, Pieper, Sloane, Snyder and Cohen2004; Knols et al. Reference Knols, Aaronson, Uebelhart, Fransen and Aufdemkampe2005; Schmitz et al. Reference Schmitz, Holtzman, Courneya, Masse, Duval and Kane2005b). Nevertheless, the specific beneficial effects of exercise may vary as a function of disease stage, treatment approach and current lifestyle of the patient (Knols et al. Reference Knols, Aaronson, Uebelhart, Fransen and Aufdemkampe2005) and the full range of positive effects resulting from physical activity is as yet unknown (Schmitz et al. Reference Schmitz, Holtzman, Courneya, Masse, Duval and Kane2005b). In addition, the methodological quality of most studies has been moderate, highlighting the need for more robustly designed randomised controlled trials with larger populations of cancer survivors (Aziz, Reference Aziz2002; Knols et al. Reference Knols, Aaronson, Uebelhart, Fransen and Aufdemkampe2005).

The beneficial effects of physical activity might also extend to anti-tumour defences in cancer patients, which could reduce the risk of disease recurrence and second cancers (Aziz, Reference Aziz2002). Chemotherapy following a cancer diagnosis suppresses immune function, including a reduction in circulating T-helper (CD4⁺) cells and an impairment of natural killer (NK) cell function (Head et al. Reference Head, Elliott and McCoy1993; Sewell et al. Reference Sewell, Halbert, Robins, Galvin, Chan and Blamey1993; Hakim et al. Reference Hakim, Cepeda, Kaimei, Mackall, McAtee, Zujewski, Cowan and Gress1997). In addition to the effects of treatment per se, the physiological effects of stress associated with a cancer diagnosis might further inhibit cellular immune responses that are relevant to cancer prognosis, including NK cell toxicity and T-cell responses (Andersen et al. Reference Andersen, Farrar, Golden-Kreutz, Kutz, MacCallum, Courtney and Glaser1998). The adverse effect of psychosocial stressors on immune function is considered to be mediated by excess secretion of the stress hormone cortisol and the catecholamines (McEwen & Stellar, Reference McEwen and Stellar1993; Madden & Felten, Reference Madden and Felten1995). Elevated cortisol and catecholamine levels evoked by psychological stressors can significantly influence immune function, including lymphocyte proliferation and NK cell activity (Banu et al. Reference Banu, Vaidya and Udupa1988; Benschop et al. Reference Benschop, Nieuwenhuis, Tromp, Godaert, Ballieux and van Doornen1994; Bryla, Reference Bryla1996; Kronfol et al. Reference Kronfol, Nair, Zhang, Hill and Brown1997). Self-reported negative mood states have been associated with lower NK cell activity (Levy & Herberman, Reference Levy and Herberman1985; Levy et al. Reference Levy, Herberman, Lippman, D'Angelo and Lee1991) and symptoms of depression have been linked with increased salivary cortisol (Ehlert et al. Reference Ehlert, Patalla, Kirschbaum, Piedmont and Hellhammer1990), impaired lymphocyte proliferation (Schleifer et al. Reference Schleifer, Keller, Meyerson, Raskin, Davis and Stein1984; Maes, Reference Maes1995) and reduced NK cell cytotoxicity against tumour cells (Nerozzi et al. Reference Nerozzi, Santoni, Bersani, Magnani, Bressan, Pasini, Antonozzi and Frajese1989; Caldwell et al. Reference Caldwell, Irwin and Lohr1991; Maes et al. Reference Maes, Stevens, Peeters, DeClerck, Scharpe, Bridts, Schotte and Cosyns1992). Studies have reported abnormal circadian rhythmicity of cortisol in breast cancer patients (Touitou et al. Reference Touitou, Bogdan, Levi, Benavides and Auzeby1996; van der Pompe et al. Reference van der Pompe, Antoni and Heijnen1996; Sephton et al. Reference Sephton, Sapolsky, Kraemer and Spiegel2000; Abercrombie et al. Reference Abercrombie, Giese-Davis, Sephton, Epel, Turner-Cobb and Spiegel2004) and flattened or abnormal diurnal salivary cortisol rhythms have recently been associated with earlier mortality in breast cancer patients (Sephton et al. Reference Sephton, Sapolsky, Kraemer and Spiegel2000) and persistent fatigue in survivors 1–5 years after initial diagnosis (Bower et al. Reference Bower, Ganz, Dickerson, Petersen, Aziz and Fahey2005).

Regular physical activity can have a positive effect on psychological health status and quality of life in cancer survivors (Baldwin & Courneya, Reference Baldwin and Courneya1997; Courneya & Friedenreich, Reference Courneya and Friedenreich1997) that could enhance immune function through normalisation of stress hormone levels. Alternatively, changes in circulating sex hormone levels might influence the functioning of immune cells in peripheral blood. Sex hormones have been reported to exert immunoregulatory effects both in vivo and in vitro (Paavonen, Reference Paavonen1994), with evidence of suppressive effects for progesterone and the androgens, whereas oestrogens can be suppressive or stimulatory depending on the situation (van Vollenhoven & McGuire, Reference van Vollenhoven and McGuire1994). Peters et al. (Reference Peters, Lotzerich, Niemeier, Schule and Uhlenbruck1994; Reference Peters, Lotzerich, Niemeir, Schule and Uhlenbruck1995) reported an increase in the percentage of NK cells and an improvement in NK cell function and monocyte phagocytic capacity in breast cancer survivors that was accompanied by an increase in ‘satisfaction of life’ score after 7 months of moderate exercise training. Two recent studies have consolidated this finding with reports of improved immune function in postmenopausal women treated for breast cancer following a period of aerobic exercise or combined aerobic exercise and resistance training lasting for 4–6 months (Fairey et al. Reference Fairey, Courneya, Field, Bell, Jones and Mackey2005; Hutnick et al. Reference Hutnick, Williams, Kraemer, Orsega-Smith, Dixon, Bleznak and Mastro2005). An increase in T-helper cell activation was observed in patients randomised to an exercise group after receiving chemotherapy in comparison with control non-exercising patients who had received chemotherapy (Hutnick et al. Reference Hutnick, Williams, Kraemer, Orsega-Smith, Dixon, Bleznak and Mastro2005). In the other study, an improvement in NK cell cytotoxic activity and in unstimulated lymphocyte proliferation was observed in breast cancer patients randomised to aerobic exercise training following surgery, radiotherapy and/or chemotherapy with or without hormone therapy, in comparison with non-exercising control patients (Fairey et al. Reference Fairey, Courneya, Field, Bell, Jones and Mackey2005). Although the underlying mechanisms for exercise-induced improvements in immune function are poorly understood, further studies are warranted, as lifestyle interventions which could have an impact on anti-tumour defences could have a significant influence on disease-free survival in cancer survivors.

Summary and conclusions

Obesity, physical activity status and circulating levels of sex steroid hormones and IGF axis proteins are intrinsically linked to energy balance. Epidemiological studies have reported a positive association between obesity and breast cancer risk for postmenopausal women, but with the role of central adiposity being more equivocal. However, recent evidence suggests that increased central adiposity is strongly associated with the risk of prostate cancer. In addition, obesity is associated with poorer prognosis following a breast or prostate cancer diagnosis. Evidence for a protective effect of increased physical activity levels on hormone-related cancer risk appears to be stronger for breast cancer than prostate cancer at the present time, with recent data also showing that physically active breast cancer survivors have an improved chance of longer-term survival. The link between physical activity status and disease-free survival has not yet been studied in prostate cancer patients, although one study (Ornish et al. Reference Ornish, Weidner and Fair2005) reported positive results in prostate cancer patients undergoing a combined exercise and low-fat diet intervention. A stronger association between circulating oestrogen levels and breast cancer risk has been reported for postmenopausal than premenopausal women and is linked to an increased aromatase activity in heavier postmenopausal women. Higher circulating testosterone is associated with an increased risk of prostate cancer and the role of oestrogen metabolism in the development of this disease warrants further study. Higher circulating IGF-1 is also associated with an increased risk of developing prostate cancer and breast cancer in premenopausal but not postmenopausal women, whereas a high circulating level of IGFBP-3 is associated with an elevated risk of premenopausal breast cancer.

An increasing number of intervention studies in ‘at-risk’ individuals and in patients recovering from cancer treatment are now investigating the effects of lifestyle interventions that promote negative energy balance on circulating levels of sex hormones and IGF axis proteins as surrogate markers of cancer risk. In relation to such lifestyle interventions, the most consistent evidence for a sex steroid hormone-lowering effect exists for oestrogen following restricted dietary fat consumption with or without regular exercise. For circulating proteins of the IGF-axis, evidence derived predominantly from middle-aged ‘at-risk’ men suggests that combining very-low-fat diets ( ≤ 10 % of total energy) with regular physical activity can induce significant changes in IGF-1 and IGFBP-1 that could protect against the development or progression of prostate cancer. One recent randomised controlled trial also reported positive results in breast cancer survivors (Fairey et al. Reference Fairey, Courneya, Field, Bell, Jones and Mackey2003). Recent evidence also suggests that lifestyle interventions associated with negative energy balance can lead to improvements in physical function and reduced feelings of fatigue, improved anti-tumour defences and an enhancement of health-related quality of life in cancer survivors. Nevertheless, the methodological quality of most intervention studies has been limited due to small subject numbers, lack of adequate control groups or non-randomised designs and a number of significant questions remain. For example, it is not known which combination of specific dietary and physical activity interventions work best for reducing the risk of hormone-related cancers in younger and older obese and non-obese populations, nor whether it is possible to get similar protective effects from dietary energy restriction and physical activity alone in these populations. Second, the optimal dose–response relationship for reducing cancer risk by dietary energy restriction and/or physical activity interventions in these different groups is unknown. Most intervention studies investigating the combined effect of diet and exercise have involved the consumption of very-low-fat-diets ( ≤ 10 % of total energy) and the effect of less intensive reductions of dietary fat and energy intake, with and without simultaneous exercise participation is currently unknown. Third, more studies investigating the association between lifestyle interventions that promote negative energy balance and disease recurrence and/or secondary primary tumours in cancer survivors are needed following promising results from very recent trials. Fourth, the predictive validity of some of the circulating biomarkers being used to assess the risk of hormone-related cancers requires further investigation. Improved knowledge about the degree to which changes in a given surrogate end-point reflects changes in clinically meaningful end-points, such as cancer diagnosis or disease-free survival, is urgently needed. More intervention studies with randomised controlled designs, higher subject numbers and longer-term follow-up measures in ‘at-risk’ populations and survivors are necessary to answer these questions and to establish the magnitude of change required in surrogate markers to induce a protective effect.