Obesity

Despite many decades of research and close to 100 000 journal papers with obesity in the title identified on PubMed (in January 2023), the precise causes of obesity and the best management/treatment strategies remain an area of controversy and vigorous debate. So much so that the Obesity Medicine Association recently published a Clinical Practice Statement 2022 addressing thirty obesity myths, misunderstandings and/or oversimplifications to assist clinicians in the care of people with obesity^{(Reference Bays, Golden and Tondt8)}. Gone are the days when obesity is simply attributed to an imbalance between energy intake and energy expenditure, notwithstanding views to the contrary^{(Reference Doroghazi9)}. Although, at the most basic level, the energy balance equation is still accepted by many (though not all) scientists working in obesity research, comprehensive reports⁽¹⁰⁾ and excellent commentaries^{(Reference Dhurandhar, Petersen and Webster11)} detail how a wide variety of diverse but interrelated factors influence energy intake, energy expenditure and body composition. The Obesity Society⁽¹²⁾ has identified sixty-six potential contributors to obesity which operate ‘inside the person’ or ‘outside the person’ to either increase energy intake, decrease energy expenditure or both. These contributors are broadly classified into seven categories: (1) environmental pressures on physical activity, (2) biological/medical, (3) maternal/developmental, (4) economic, (5) food and beverage behaviour/environment, (6) psychological and (7) social (Table 1).

Table 1. Potential* contributors to obesity

EI, energy intake; EE, energy expenditure.

* Potential contributors indicate anything that has been put forth in the research literature as a question of investigation and is not intended to be a verification of whether; or the extent to which, each may or may not contribute.

The column headed ‘Both?’ indicates factors that may increase energy intake and decrease energy expenditure, or the effect is unknown.

(I) Indicates that the factor operates inside (within) the person.

(O) Indicates that the factor operates outside of the person.

Adapted from The Obesity Society (2015).

Although some argue that we still do not have a clear explanation for the obesity epidemic^{(Reference Mozaffarian13)} the prevailing view in the literature is that the primary driver of obesity is excess energy availability^{(Reference Barrera and Shively14)}. Some individuals are strongly genetically predisposed to obesity^{(Reference Bouchard15)} while a small percentage are genetically resistant to it^{(Reference Riveros-McKay, Mistry and Bounds16)}. The genetic risk for obesity may be broadly classified into two categories: (1) monogenic obesity which is typically rare, early onset and involves single-gene defects, and (2) polygenic (common) obesity which involves hundreds of polymorphisms that each have a small effect. The central nervous system and neuronal pathways that control hedonic aspects of food intake (and hence body weight) are the main mechanisms underlying both monogenic and polygenic obesity^{(Reference Loos and Yeo17)}.

The most effective treatment for obesity/severe obesity is bariatric surgery which can elicit 50–60 % excess weight loss for 10 years or more^{(Reference Eisenberg, Shikora and Aarts18,Reference Kheniser, Saxon and Kashyap19)} . Pharmacotherapy for obesity is rapidly catching up due to the development of glucagon-like peptide 1 (GLP-1) receptor agonists^{(Reference Brown, Heerspink and Cuthbertson20–Reference Kumar and D'Alessio22)} which effectively control food intake as demonstrated by recent trials with tirzepatide^{(Reference Jastreboff, Aronne and Ahmad23)} and semaglutide^{(Reference Wadden, Bailey and Billings24–Reference Weghuber, Barrett and Marrientos-Perez26)}. The amylin analogue, cagrilintide, is also effective for weight loss^{(Reference Lau, Erichsen and Francisco27)} while antibody blockade of activin type II receptor signalling via bimagrumab promotes fat mass loss and simultaneously increases lean mass^{(Reference Heymsfield, Coleman and Miller28)}. These developments herald a new era in pharmacotherapy for people with obesity^{(Reference Firman and Batterham29)}.

After bariatric surgery and pharmacotherapy, diet has been shown to be effective for the prevention and management of obesity although long-term maintenance of weight loss with diet alone is a major issue. There is still disagreement on the most effective dietary strategy for weight control and recent debates about the carbohydrate insulin model^{(Reference Ludwig, Aronne and Astrup30–Reference Speakman and Hall32)} have the potential to confuse the public not to mention academics working in the field. Evidence to support the merits of any one dietary macronutrient for weight control is limited^{(Reference Churuangsuk, Hall and Reynolds33–Reference Schutz, Montani and Dulloo35)} while evidence to support another recent dietary trend, time-restricted eating, is mixed and inconclusive^{(Reference Jamshed, Steger and Bryan36–Reference Paar, Devlin and Hawley38)}. Physical activity/exercise has largely been relegated to the fourth line of ‘defence’ against overweight and obesity although some suggest that it has little value in this respect^{(Reference Luke and Cooper39,Reference Malhotra, Noakes and Phinney40)} . This will be examined in the next three sections the first of which addresses prevention.

Physical activity and prevention of obesity

Physical inactivity may be both a cause and a consequence of obesity. A large volume of observational evidence demonstrates that obesity prevalence/likelihood is associated with low levels of physical activity and/or high amounts of sedentary behaviour^{(Reference King, Stensel, Hardman and Gill41)}. The cause-and-effect nature of this association is difficult to establish. Examples of such literature include studies examining energy expenditure in domestic tasks^{(Reference Archer, Shook and Thomas42)} and occupational activity^{(Reference Church, Thomas and Tudor-Locke43)} as well as studies of time spent in sedentary pursuits^{(Reference Healy, Matthews and Dunstan44)}. Archer and colleagues^{(Reference Archer, Shook and Thomas42)} examined time-use diary data to quantify household management energy expenditure from 1965 to 2010 in adult women (aged 19–64 years) living in the United States. They estimated that household management energy expenditure decreased by 42 % in non-employed women and 30 % in employed women across this time – equivalent to a decrement of 10·5 MJ/week (2518 kcal/week) or 1·5 MJ/d (360 kcal/d) in non-employed women and 3·9 MJ/week (923 kcal/week) or 0·55 MJ/d (132 kcal/d) in employed women. They observed that much of this reduction in ‘active pursuits’ (i.e. housework) was reallocated to sedentary pastimes (e.g. watching TV) and concluded that this could have contributed to the increase in obesity prevalence in women in the United States between 1965 and 2010. Similarly, Church and colleagues^{(Reference Church, Thomas and Tudor-Locke43)} examined occupation-related physical activity over nearly five decades (1960–2006) in United States adults (aged 40–50 years) using data from the National Health and Nutrition Examination Surveys. They estimated that occupation-related energy expenditure decreased by 594 kJ/d in men during this time and this was sufficient to explain an increase in body mass from 77 kg at baseline to 90 kg at follow-up which closely matched the actual increase in body mass (77–92 kg) during this period. Similar trends were observed in women in this study. Another study employing National Health and Nutrition Examination Survey data^{(Reference Healy, Matthews and Dunstan44)} has demonstrated a positive association between prolonged sedentary time and waist circumference (a marker for central adiposity) in adults, while a recent review of twenty-one observational studies examining physical activity levels among pregnant women concluded that active pregnant women gain less weight during pregnancy than inactive pregnant women^{(Reference Hamann, Deruelle and Enaux45)}.

A rather novel study design employed to examine the influence physical activity on body composition is the study of monozygotic (identical) twins discordant for physical activity^{(Reference Kujala, Leskinen and Rottensteiner46)}. This ‘natural experiment’ involved comparisons of seventeen Finnish twin pairs found to be discordant for physical activity over a prolonged period. The twins with higher physical activity were shown to have lower body fat, visceral fat and liver fat than their less active counterparts. Another interesting and thought-provoking study conducted by Levine and colleagues^{(Reference Levine, Lanningham-Foster and McCrady47)} examined non-exercise activity thermogenesis, i.e. the energy expended in posture and movement associated with the routines of daily life (excluding energy expended through purposeful exercise). Inclinometers and triaxial accelerometers were used to assess time spent lying, siting, standing and ambulating in ten lean individuals and ten individuals with obesity over a 10-d period resulting in about 25 million data points on posture and movement for each person. Doubly labelled water was used to quantify energy expenditure. The group classified with obesity was found to sit for 2 h longer each day than the group classified as lean. It was estimated that the difference in non-exercise activity thermogenesis between groups equated to 1464 kJ/d. Interestingly all individuals participating in this study (from both groups) were self-proclaimed ‘couch potatoes’. Posture allocation did not change after an 8-week intervention during which the lean group gained weight by overfeeding and the group with obesity lost weight by energy restriction. This raises the intriguing possibility that non-exercise activity thermogenesis is ‘biologically determined’^{(Reference Levine, Lanningham-Foster and McCrady47)} and resistant to change although this research is limited by a small sample size and a relatively short follow-up period.

The finding that people with obesity are generally less active than people who are lean leads to the assertion that increases in physical activity can be helpful for preventing and managing obesity, but Pontzer and colleagues have conducted a series of studies over the past decade countering this proposition^{(Reference Pontzer, Raichlen and Wood48–Reference Rimbach, Yamada and Sayagama54)}. These studies have employed the doubly labelled water technique to assess energy expenditure in various human and primate groups. One such study^{(Reference Pontzer, Raichlen and Wood48)} demonstrated that although the physical activity level of Hadza hunter–gatherers was higher than that of Westerners, total daily energy expenditure did not differ between groups after accounting for differences in body size. Pontzer and colleagues concluded that ‘the similarity in metabolic rates across a broad range of cultures challenges current models of obesity suggesting that Western lifestyles lead to decreased energy expenditure’^{(Reference Pontzer, Raichlen and Wood48)}. Another study by Pontzer's group observed similar levels of energy expenditure in wild and captive primate populations in contrast to the expectation that wild primates would be more active^{(Reference Pontzer, Raichlen and Gordon49)}. In view of such findings Pontzer proposed a Constrained Total Energy Expenditure hypothesis stating ‘rather than increasing with physical activity in a dose-dependent manner, experimental and ecological evidence suggests … total energy expenditure is a relatively constrained product of our evolved physiology’^{(Reference Pontzer50)}. This hypothesis was supported by a study of 322 adult human subjects living in five populations which found that although physical activity was positively correlated with total energy expenditure this relationship was much stronger at low levels of energy expenditure and plateaued in the upper range of physical activity^{(Reference Pontzer, Brown and Raichlen51)}. Pontzer hypothesised that in contrast to primate groups human subjects utilise a greater portion of their energy expenditure to fuel their larger brains and that human subjects have a greater predisposition to accumulate adipose tissue which is an advantage for their evolution^{(Reference Pontzer, Durazo-Arvizu and Dugas52)}.

More recently, Pontzer's group has used longitudinal data from The International Atomic Energy Agency doubly labelled water database to demonstrate that ‘energy compensation’ occurs when physical activity levels are elevated, i.e. basal energy expenditure is reciprocally reduced^{(Reference Careau, Halsey and Pontzer53)}; and that low adjusted total energy expenditure does not predict gains in body fat over time, and higher adjusted total energy expenditure does not protect against weight gain^{(Reference Rimbach, Yamada and Sayagama54)}. Collectively, the work of Pontzer and colleagues is provocative and may suggest that physical activity/exercise will be ineffective for the prevention/management of obesity. In contrast, exercise has been shown to be effective for weight loss in laboratory-based studies of people with obesity, and physical activity may have a role to play in weight loss maintenance as will be explored in the next section.

Physical activity and management of obesity

There is good evidence from randomised controlled trials that exercise is effective for reducing body mass and body fat in people with overweight/obesity. A notable example is a 12-week randomised controlled trial study conducted by Ross and colleagues^{(Reference Ross, Dagnone and Jones55)} comprising four study groups: (1) control, (2) diet-induced weight loss, (3) exercise-induced weight loss and (4) exercise without weight loss. Participants in the diet-induced weight loss group reduced their energy intake by 2929 kJ/d while those in the exercise-induced weight loss group maintained their usual energy intake and expended 2929 kJ/d in exercise (brisk walking or light jogging). Body mass decreased by 7·5 kg (8 %) in both weight loss groups with a greater reduction in body fat in the exercise-induced weight loss group. There were similar reductions in visceral adipose tissue in both weight loss groups and some reduction in visceral adipose tissue also in the exercise without weight loss group. The authors concluded that physical activity without energy restriction substantially reduces obesity – particularly abdominal obesity^{(Reference Ross, Dagnone and Jones55)}. Ross and colleagues subsequently published a review of eight studies which directly compared diet and exercise interventions for weight loss concluding that ‘exercise without diet restriction is an effective strategy for reducing obesity and related co-morbidities’ and recommending 45–60 min of moderate intensity exercise each day^{(Reference Ross, Freeman and Janssen56)}. This is broadly consistent with the recommendations of the American College of Sports Medicine Position Stand on Appropriate Physical Activity Intervention Strategies for Weight Loss and Prevention of Weight Regain in Adults, i.e. 150–250 min/week of moderate intensity exercise to prevent weight gain/provide modest weight loss and >250 min/week of moderate intensity exercise to elicit clinically significant weight loss^{(Reference Donnelly, Blair and Jakicic57)}.

Although physical activity alone can produce substantial and clinically significant weight loss it is important to note that this is not the norm. Indeed, a recent review of twelve systematic reviews involving 149 studies concluded that exercise training interventions typically induce weight loss of 1·5–3·5 kg^{(Reference Bellicha, van Baak and Battista58)}. Various factors may explain this modest weight loss including metabolic and behavioural compensation. Metabolic compensation occurs via physiological changes in metabolism while behavioural compensation relates to changes in dietary and exercise behaviours to ‘compensate’ for the perturbation in energy balance caused by the intervention. (The term ‘behavioural compensation’ is controversial because changes in ‘behaviour’ may be determined by physiological/metabolic factors.) Dhurandhar and colleagues estimate that exercise studies typically result in 55–64 % less weight loss than expected due to compensation^{(Reference Dhurandhar, Kaiser and Dawson59)}. Consistent with this prediction are the findings of a recent study showing that treadmill/cycling exercise over a 24-week period induced only half the expected weight loss in adult men and women with obesity due to reductions in 24-h energy expenditure, i.e. RMR and/or non-exercise activity thermogenesis^{(Reference Broskey, Martin and Burton60)}.

Aside from metabolic compensation another, under-researched, factor which is rarely assessed in energy balance studies is energy excretion, i.e. consumed food which is excreted in faeces or filtered through the kidneys. The routinely used kilocalorie equivalents per g of carbohydrate, fat and protein are 4, 9 and 4, respectively. These represent the ‘metabolisable energy’ per unit of mass for a given macronutrient but there is variability between individuals in the extent to which full oxidation of metabolisable energy occurs and this variation is due (at least in part) to differences in fat faecal excretion. This can average between 0·5 and 5 % of energy intake with an interindividual variation of 2–9 %^{(Reference Lund, Gerhart-Hines and Clemmensen61)}. Whether energy excretion changes during exercise interventions is unknown but it is feasible that it could, and this would be another source of compensation reducing the effectiveness of the intervention. Nearly 30 years ago, Bouchard and colleagues^{(Reference Bouchard, Tremblay and Despres62)} demonstrated that some people lose more weight, body fat and visceral fat than others in response to exercise. In their study average exercise-induced weight loss among seven pairs of identical twins after a 93-d exercise intervention (2 h each day, 9 out of every 10 d) was 5 kg, but weight loss ranged from <2 kg to about 8 kg. There was less variation in response within twin pairs than there was between twin pairs suggesting that responses to exercise are determined, at least in part, by genetic factors. More recent studies have confirmed that there is interindividual variability in body composition changes in response to exercise^{(Reference Thomas, Marsh and Maslen63,Reference Barber, Ruiz-Ramie and Robbins64)} although the precise contribution of genetic and environmental factors to this variability remains to be identified.

Although exercise alone can induce weight loss in people with obesity it should ideally be used in combination with other therapies. It has been shown that the combination of high-intensity interval training and time-restricted eating induce superior reductions in total and visceral fat than either intervention alone^{(Reference Haganes, Silva and Eyjólfsdóttir65)}. Similarly, the combination of exercise and liraglutide therapy improved healthy weight/fat loss (achieved via energy restriction) more than either therapy alone^{(Reference Lundgren, Janus and Jensen66)}. This latter study highlights that exercise has an important role to play in the maintenance of long-term weight loss as an adjunct to other therapies including bariatric surgery and pharmacotherapy^{(Reference Kheniser, Saxon and Kashyap19)}. The challenge of maintaining long-term weight loss is made particularly difficult due to metabolic adaptation induced by changes in hormones that influence appetite^{(Reference Busetto, Bettini and Makaronidis67)} and there is evidence that exercise may assist with long-term weight loss by improving appetite regulation^{(Reference Butryn, Crane and Lufburrow68)}, an issue examined in the next section.

Exercise and appetite

There has been a proliferation of interest in the relationship between physical activity/exercise and appetite in recent decades. Research examining this relationship has been facilitated by an expansion of research and understanding of the factors influencing food preferences^{(Reference Berthold, Morrison and Ackroff69)}, the physiological mechanisms controlling eating and different types of eating, i.e. deficit-induced eating, habitual eating and opportunistic eating^{(Reference Watts, Kanoski and Sanchez-Watts70)}, and the factors contributing to satiation and meal termination^{(Reference Cunningham and Rolls71)}. For centuries the hypothalamus has been the centre of attention as a key regulator of appetite^{(Reference Azevedo, Ivan and Friedman72)} and within the arcuate nucleus of the hypothalamus, agouti-related peptide neurones are thought to govern the drive to eat^{(Reference Deem, Faber and Morton73)}. A key hormone influencing agouti-related peptide neuronal activity is ghrelin which has attracted widespread attention due to its unique role in stimulating hunger and food intake^{(Reference Theilade, Christensen and Vilsbøll74)}. Aside from ghrelin many other hormones influence appetite mostly with appetite suppressing effects; these include leptin, peptide YY (PYY), insulin, pancreatic polypeptide, amylin, oxyntomodulin, cholecystokinin, GLP-1 and adiponectin^{(Reference Theilade, Christensen and Vilsbøll74)}. The most widely studied of these in relation to exercise are ghrelin, PYY and GLP-1 due to the potency with which these hormones act on appetite and their relative ease of measurement. Due to the crucial role appetite plays in the development of overweight and obesity it is essential to understand how exercise affects appetite and hence the most effective way to implement exercise as a therapy for obesity.

The effects of physical activity on the regulation of appetite and food intake have been extensively reviewed^{(Reference Dorling, Broom and Burns75–Reference Ropelle, da Silva and Cintra77)} and may be broadly categorised as acute (responses to a single bout of exercise) and chronic (responses to exercise training) effects. Most research in human subjects has focused on the peripheral (hormonal) influences of exercise rather than central (brain) responses although some evidence is available in relation to the latter^{(Reference Ropelle, da Silva and Cintra77)}. Single bouts of exercise induce a short-term energy deficit without stimulating compensatory increases in appetite or food intake. Many studies have shown that vigorous-intensity acute exercise suppresses ghrelin, notably the active form of the hormone – acylated ghrelin – while simultaneously increasing concentrations of GLP-1 and PYY^{(Reference Dorling, Broom and Burns75)}. Responses are generally consistent in groups who are lean and those with obesity. For example, Douglas and colleagues^{(Reference Douglas, King and Clayton78)} demonstrated that total GLP-1 and total PYY were increased in both groups during the hours after treadmill exercise and these elevations coincided with suppressed appetite perceptions. Simply interrupting sitting with short breaks of brisk walking elevates total GLP-1 and total PYY as demonstrated by a recent acute study involving men and women with central overweight/obesity^{(Reference Chen, Walhin and Hengist79)}. Exercising in a fasted, rather than fed, state is more effective for eliciting a short-term decrease in energy intake but may lead to increased hunger subsequently^{(Reference Frampton, Edinburgh and Ogden80)}. Individual variability in appetite perceptions and appetite-related hormone (acylated ghrelin, total PYY) responses to exercise has been demonstrated using a robust replicated crossover design suggesting some individuals will experience greater appetite suppressing effects from acute exercise than others^{(Reference Goltz, Thackray and King81)}.

The significance of acute changes in appetite and appetite-related hormones for long-term weight management is uncertain. In theory, if the acute appetite suppressing responses to exercise were repeated regularly this would lead to long-term reductions in appetite and food intake but evidence to support this hypothesis is limited due to the difficulty of performing well-controlled intervention studies with accurate measures of free-living food intake. Indeed, with repeated energy deficits created through exercise some form of compensation is likely to occur and there is evidence that attentional bias towards food cues is increased in response to acute exercise^{(Reference Flack, Anderson and McFee82)}. At least one chronic exercise study conducted by a prolific group of appetite researchers at the University of Leeds has suggested a ‘dual-process action’ of exercise on appetite control^{(Reference King, Caudwell and Hopkins83)}. This study involved a 12-week, laboratory based, supervised exercise intervention (energy expenditure = 10 460 kJ/week) in fifty-eight men and women with overweight/obesity. The intervention resulted in an average weight loss, fat mass loss and waist circumference reduction of 3·2 kg, 3·2 kg and 5 cm, respectively^{(Reference King, Caudwell and Hopkins83)}. There was an increase in fasting hunger but an increase also in the satiety quotient of a breakfast meal (calculated by the change in appetite scores relative to meal content). The authors concluded that while exercise-induced weight loss increased the drive to eat it also enhanced the satiating effect of a fixed meal^{(Reference King, Caudwell and Hopkins83)}.

The Leeds group has also conducted two innovative studies examining energy balance with differing levels of energy expenditure in men and women^{(Reference Stubbs, Sepp and Hughes84,Reference Stubbs, Hughes and Johnstone85)} . The first of these used a within-subject, repeated measures design to compare energy intake and balance over 7 d in six lean women across three interventions: control (0 MJ/d), medium exercise level (1·9 MJ/d) and high exercise level (3·4 MJ/d). Participants self-weighed their ad libitum food intake while energy expenditure was estimated using continuous heart rate monitoring. Daily energy expenditure increased across conditions: 9·2, 11·0 and 12·1 MJ/d for control, medium exercise level and high exercise level, respectively. The corresponding values for energy intake were 8·9, 9·2 and 10·0 MJ/d. It was concluded that increasing energy expenditure through exercise led to significant but partial compensations in energy intake (about 33 % of the energy expenditure due to exercise) and that accurate adjustments in energy intake to acute increases in energy expenditure would take weeks rather than days^{(Reference Stubbs, Sepp and Hughes84)}. A second 7-d study by the Leeds group employed a whole-body indirect calorimeter to examine energy expenditure responses to sedentary (1·4 × RMR) and moderately active (1·8 × RMR) conditions in men while energy intake was continuously monitored. Energy expenditure was 9·7 MJ/d in the sedentary trial and 12·8 MJ/d in the moderately active trial. Corresponding values for energy intake were 13·5 and 14·4 MJ/d, respectively. Thus, energy intake was higher than energy expenditure in both trials, but cumulative energy balance was 2·4 times higher in the sedentary trial (26·3 MJ) than the moderately active trial (11·1 MJ) causing the authors to conclude that a sedentary routine does not induce compensatory reductions in energy intake, leading to a positive energy balance, most of which is stored as fat^{(Reference Stubbs, Hughes and Johnstone85)}.

The studies outlined earlier conducted by the Leeds group provide valuable information on the effects of exercise on energy balance but their application to the field of obesity is limited because participants in these studies were lean. More recently, members of the Leeds research group in collaboration with other researchers have conducted a systematic review and meta-analysis examining the effects of exercise training interventions on energy intake and appetite control in adults with overweight and obesity^{(Reference Beaulieu, Blundell and van Baak86)}. Forty-eight relevant articles were identified, and it was concluded that exercise training in people with overweight or obesity does not induce a substantial change in food intake or appetite during the period of training although it was acknowledged that the self-report nature of the food intake data and the small number of studies rated as fair or good quality were limitations.

Recent decades have seen a proliferation of research and understanding about the effects of exercise on appetite. Much of this has focused on hormonal control of appetite which clearly provides only partial understanding. Methods to quantify energy intake under free-living conditions remain woefully inadequate and technological developments for non-intrusive assessment of energy intake are essential to address this. It is also clear that a variety of endocrinology factors beyond the most frequently studied appetite-related gut hormones (i.e. ghrelin, GLP-1, PYY) play a role in the interaction between exercise and appetite. These include the recently discovered blood-borne signalling metabolite N-lactoyl-phenylalanine which is associated with reduced food intake^{(Reference Li, He and Contrepois87)}, various myokines^{(Reference Grannell, Kokkinos and le Roux88)}, growth differentiation factor 15^{(Reference Klein, Kleinert and Richter89)}, lactate^{(Reference McCarthy, Islam and Hazell90)}, orexin and serotonin^{(Reference Mavanji, Pomonis and Kotz91)} and a newly proposed ‘gravitostat’ operating via a sensor released from osteocytes^{(Reference Jansson, Palsdottir and Hägg92)}. We are also learning more about how genetic factors, exercise and appetite interact. For example, there is a potentially greater benefit of exercise in those with obesity risk variants of the fat mass and obesity associated (FTO) gene^{(Reference Karra, O'Daly and Choudhury93–Reference Hiraike, Yang and Liu95)}. Further study of these issues is crucial to provide a more complete understanding of how exercise and appetite interact. At the beginning of this section, it was noted that the hypothalamus is central to appetite control and more research in this area is vital. Recent evidence from animal studies suggests that exercise may beneficially modify anorexigenic signals and food intake via mechanisms involving myokines released during muscle contraction and a reduced central anti-inflammatory response^{(Reference Ropelle, da Silva and Cintra77)} but there is a need for longitudinal studies in human subjects to better understand how exercise affects hypothalamic regulation of appetite.

Physical activity, ageing and chronic disease

So far, this review has focused on the therapeutic potential of physical activity in relation to obesity due to the prolonged upward trend in the prevalence of this condition globally and the ensuing public health burden. Aside from obesity there is a wealth of evidence to support the benefits of physical activity for the prevention and management of chronic diseases associated with ageing, most notably CVD, cancer, and type 2 diabetes among others^{(Reference Stensel, Hardman and Gill96,Reference Hargreaves97)} . Knowledge and understanding about the mechanisms by which physical activity interacts with chronic disease risk is accumulating at pace. Many ‘exerkines’ (signalling moieties released in response to acute/chronic exercise) have been identified. These exert their effects through endocrine, paracrine and autocrine pathways involving multiple organs, cells and tissues including skeletal muscle (myokines), the heart (cardiokines), liver (hepatokines), white adipose tissue (adipokines), brown adipose tissue (baptokines) and neurones (neurokines) and have the potential to assist in the prevention and management of chronic disease^{(Reference Chow, Gerszten and Taylor98)}. Future decades are likely to see profound increases in our mechanistic understanding in relation to physical activity and chronic disease.

The field of accelerometer-based wearable devices has facilitated important advances in the quality of research linking physical activity/inactivity with all-cause mortality and chronic disease risk. Recent examples include: (1) a report from the UK Biobank (78 500 participants followed for a median of 7 years) demonstrating an association between higher daily step count and a lower risk of all-cause mortality, CVD mortality and cancer mortality as well as a lower incidence of CVD and cancer up to a threshold of approximately 10 000 steps daily^{(Reference Del Pozo Cruz, Ahmadi and Lee99)}, (2) a report from the coronary artery risk development in young adults study showing that those who took 7000 or more steps/d experienced a lower mortality rate than those taking <7000 steps/d – an association which held in Black and White men and women^{(Reference Paluch, Gabriel and Fulton100)}, (3) an estimation using accelerometer data from the National Health and Nutrition Examination Survey that 110 000 deaths per year could be avoided if United States adults aged 40–85 years or older increased their moderate-to-vigorous intensity physical activity by 10 min/d, a finding which applied in all groups studied, i.e. Mexican American, non-Hispanic Black and non-Hispanic White adults^{(Reference Saint-Maurice, Graubard and Troiano101)} and (4) a report from the All of Us Research Program in the United States that higher daily step counts are associated with a lower risk of various conditions including obesity, sleep apnoea, gastroesophageal reflux disease, diabetes and hypertension^{(Reference Master, Annis and Huang102)}.

The past decade has seen increased focus on the adverse health effects of sedentary behaviour specifically and independently of overall physical activity levels. A recent publication, for example, involving over 105 000 men and women followed for a median of 11·1 years, observed that sitting for ≥8 h/d compared with sitting for <4 h/d was associated with an increased risk of all-cause mortality and CVD events and this association was stronger in low-income and lower-middle-income countries than in high-income and upper-middle income countries^{(Reference Li, Lear and Rangarajan103)}. Such findings confirm earlier reports from the community based Kungsholmen Project in Stockholm, Sweden, that physical activity is an essential healthy lifestyle component associated with extended mortality where benefits are maintained even in those in their 70s and 80s^{(Reference Rizzuto, Orsini and Qiu104)}. These and other studies demonstrate the potential for lowering chronic disease burden through increased physical activity and reduced sedentary behaviour. An analysis by Lee and colleagues^{(Reference Lee, Shiroma and Lobelo105)} incorporating population attributable fractions associated with physical inactivity estimates that worldwide physical inactivity causes 6 % of heart disease, 7 % of diabetes, 10 % of breast cancer, 10 % of colon cancer and 9 % of premature mortality. According to the calculations of Lee and colleagues, decreasing global physical inactivity by 25 % could avert 103 million deaths/year worldwide^{(Reference Lee, Shiroma and Lobelo105)}.

Perhaps the disease most closely linked to ageing is dementia. Here again, evidence is accumulating to support the therapeutic potential of physical activity. A notable study is the 1-year randomised controlled trial conducted by Erickson and colleagues^{(Reference Erickson, Voss and Prakash106)} which investigated hippocampus shrinkage in late adulthood due to its association with impaired memory and dementia risk. In this trial 120 women and men (mean age 67 years) were randomly allocated to either a walking intervention group or a stretching control group. Walking increased left and right hippocampus volume (assessed by MRI) by 2 % while volumes declined in the control group. It was estimated that walking reversed age-related hippocampus volume loss by 1–2 years. Increases in hippocampus volume were correlated with increases in fitness (maximum oxygen uptake), increases in serum concentrations of brain-derived neurotrophic factor, a mediator of neurogenesis, and increases in memory performance. It was concluded that aerobic exercise training is effective for reducing hippocampal volume loss and improving memory function in late adulthood.

The implication of the study by Erickson and colleagues^{(Reference Erickson, Voss and Prakash106)} is that exercise may reduce the risk of developing dementia, or at least delaying the onset of dementia. Many studies support this contention. Recent examples include a report from the Chicago Health and Ageing Project^{(Reference Dhana, Franco and Ritz107)} which developed a healthy lifestyle score based on five modifiable lifestyle factors: healthy diet, regular physical activity, participation in late life cognitive activities, non-smoking and light-to-moderate alcohol consumption. Participants with four or five healthy factors lived longer and lived a larger proportion of their remaining years without Alzheimer's dementia than those with zero or one healthy lifestyle factor. This is one of many studies suggesting that physical activity, as part of a ‘package’ of healthy lifestyle factors, has potential for reducing dementia risk. The extent to which physical activity alone (independent of other healthy lifestyle factors) can reduce the risk of dementia is debatable but data from the UK Biobank^{(Reference Del Pozo Cruz, Ahmadi and Naismith108)} (78 430 adults aged 40–79 years followed for 6·9 years) reveal that higher accelerometer-derived step count is associated with a lower risk of all-cause dementia. The optimal dose in this report is just under 10 000 steps/d and stronger associations were noted when steps were performed at a higher intensity.

The associations between physical activity, sedentary behaviour and chronic disease risk described earlier suggest there may be great benefit in encouraging middle-aged and older adults to maintain/increase their physical activity levels and reduce the amount of time spent sedentary. This may be particularly important in those aged 60 years and above. A recent analysis of daily energy expenditure through the life course employing doubly labelled water assessments in over 6400 males and females (aged between 8 d and 95 years) from twenty-nine countries found energy expenditure was stable between 20 and 60 years of age but a gradual decline occurs after age 60^{(Reference Pontzer, Yamada and Sagayama109)}. As has been noted ‘it cannot be coincidence that the increase in incidence of noncommunicable diseases and disorders begins in this same time frame’^{(Reference Rhoads and Anderson110)}. More research is required to understand the extent to which the decline in physical activity/energy expenditure beyond age 60 is biologically determined and immutable and the extent to which it is amenable to change by conscious effort.