Exercise-induced anorexia

Strenuous exercise (≥60 % of maximum oxygen uptake (VO₂ max)) has consistently been found to acutely supress appetite during and shortly after the exercise bout. This is known as exercise-induced anorexia and has been demonstrated during a variety of exercise modes including running⁽ Reference Broom, Stensel and Bishop ^{12 –} Reference King, Miyashita and Wasse ^{14 )}, cycling⁽ Reference Becker, Macedo and Cunha ^{15 –} Reference Ueda, Yoshikawa and Katsura ^{21 )}, swimming⁽ Reference King, Wasse and Stensel ^{22 )} and resistance exercise⁽ Reference Broom, Batterham and King ^{23 )}. Although significant changes in appetite have been reported in these studies, values tend to return to control values within 30 min of the cessation of exercise. Such a transient effect is unlikely to influence energy intake in the hours after exercise but may delay the initiation of feeding when food is provided immediately after exercise⁽ Reference King, Wasse and Stensel ^{24 )}. It therefore remains most important to understand the influence of exercise on resting appetite and energy intake responses in the hours after exercise.

Appetite responses in the immediate post-exercise period (0–120 min)

The majority of studies that have investigated the appetite response to exercise have employed an observation period lasting up to 2 h after the exercise bout. The consensus among these studies is that, after the recovery from exercise-induced anorexia, appetite during the post-exercise period does not differ from a resting control trial⁽ Reference Burns, Broom and Miyashita ^{13 ,} Reference Becker, Macedo and Cunha ^{15 ,} Reference Evero, Hackett and Clark ^{17 –} Reference Martins, Morgan and Bloom ^{19 ,} Reference Erdmann, Tahbaz and Lippl ^{25 –} Reference Ueda, Yoshikawa and Katsura ^{34 )}. However, although appetite perceptions do not appear to decrease during the post-exercise period, it must be acknowledged that some investigations have reported elevated appetite perceptions after exercise compared with a control trial⁽ Reference Jokisch, Coletta and Raynor ^{35 –} Reference Verger, Lanteaume and Louis-Sylvestre ^{37 )}.

The majority of the studies listed earlier have investigated the appetite response to exercise in healthy non-obese men and demonstrated no differences in appetite between exercise and control trials during the 2 h after a range of exercise modes including walking⁽ Reference Farah, Brunstrom and Gill ^{26 )}, cycling⁽ Reference Becker, Macedo and Cunha ^{15 ,} Reference Evero, Hackett and Clark ^{17 –} Reference Martins, Morgan and Bloom ^{19 ,} Reference Jokisch, Coletta and Raynor ^{35 )} and running⁽ Reference Burns, Broom and Miyashita ^{13 ,} Reference Gonzalez, Veasey and Rumbold ^{28 ,} Reference Kelly, Guelfi and Wallman ^{30 )}. The use of male participants in the majority of investigations has led some authors to postulate that sex-based differences may occur in the appetite response to exercise due to the critical relationship between energy balance and reproductive function in females⁽ Reference Wade and Jones ^{38 )}. However, the present research suggests that females do not exhibit increased appetite perceptions during the 2 h period after exercise⁽ Reference Laan, Leidy and Lim ^{18 ,} Reference Martins, Morgan and Bloom ^{19 ,} Reference Hagobian, Yamashiro and Hinkel-Lipsker ^{29 ,} Reference Larson-Meyer, Palm and Bansal ^{31 –} Reference Unick, Otto and Goodpaster ^{33 )}. Furthermore, a recent study by Hagobian et al.⁽ Reference Hagobian, Yamashiro and Hinkel-Lipsker ^{29 )} directly compared the appetite response to exercise in male and female participants and concluded that 80 min cycling at 70 % VO₂ max did not stimulate increases in appetite in either sex during the 40 min after exercise.

As obesity is the result of a chronic excess of energy intake over energy expenditure, it is also logical to consider that appetite regulation in response to exercise may differ between lean and obese individuals. However, the present research suggests that exercise also fails to stimulate immediate compensatory increases in appetite in overweight and obese populations. In this regard, Unick et al.⁽ Reference Unick, Otto and Goodpaster ^{33 )} reported that walking at 70–75 % of the maximum heart rate to expend 13 kJ/kg body mass did not stimulate any compensatory increases in appetite during the 60 min after exercise in overweight and obese women. Similarly, Tsofliou et al.⁽ Reference Tsofliou, Pitsiladis and Malkova ^{32 )} did not observe any increases in appetite during the 60 min after 20 min brisk walking in ten obese healthy women. Furthermore, in a direct comparison between obese and normal weight young men, Ueda et al.⁽ Reference Ueda, Yoshikawa and Katsura ^{34 )} demonstrated that 60 min cycling at 50 % VO₂ max did not stimulate any increases in appetite during the 60 min after exercise in either group.

Energy intake responses in the immediate post-exercise period (0–120 min)

The majority of studies detailed in the previous section also assessed the food intake response to exercise by providing participants with an ad libitum meal ≤2 h after the exercise bout. In accordance with the appetite responses detailed in the previous section, the majority of these studies demonstrated that energy intake was unaffected by exercise⁽ Reference George and Morganstein ^{27 –} Reference Kelly, Guelfi and Wallman ^{30 ,} Reference Tsofliou, Pitsiladis and Malkova ^{32 ,} Reference Unick, Otto and Goodpaster ^{33 ,} Reference Balaguera-Cortes, Wallman and Fairchild ^{39 –} Reference Maraki, Tsofliou and Pitsiladis ^{41 )}. However, some studies have demonstrated increases⁽ Reference Laan, Leidy and Lim ^{18 ,} Reference Martins, Morgan and Bloom ^{19 ,} Reference Verger, Lanteaume and Louis-Sylvestre ^{37 ,} Reference Shorten, Wallman and Guelfi ^{42 )} or decreases⁽ Reference Ueda, Yoshikawa and Katsura ^{34 ,} Reference Almada, Cataldo and Smalley ^{43 ,} Reference Westerterp-Plantenga, Verwegen and Ijedema ^{44 )} in energy intake after exercise.

An early study to investigate the effect of exercise on subsequent food intake responses was performed by King et al. in 1994⁽ Reference King, Burley and Blundell ^{40 )}. In this study, cycling at 30 or 70 % VO₂ max to expend ∼1460 kJ did not affect energy intake at an ad libitum buffet meal 15 min after exercise. Furthermore, in a second group of healthy young men, energy intake was unaffected by either 26 or 52 min cycling at 75 % VO₂ max. Many of these findings were replicated by Erdmann et al.⁽ Reference Erdmann, Tahbaz and Lippl ^{25 )} in a combined sample of normal-weight men and women, as energy intake was unaffected by 30 min cycling at either 50 or 100 W. However, although energy intake was unchanged in response to 60 min cycling at a fixed work rate of 50 W, increasing the duration to 120 min stimulated an increase in energy intake. Although these findings suggest that the energy intake response to exercise is dependent upon exercise duration, all trials commenced after a 12 h overnight fast with the ad libitum meal provided 15 min after exercise. Therefore the observed increases in energy intake may have been a result of the extended overnight fast rather than exercise duration per se.

Although these studies may be criticised for the provision of a buffet meal so close to the cessation of exercise, other authors have also demonstrated no change in energy intake after exercise when the ad libitum meal was provided up to 1 h after exercise. This relationship has been demonstrated in young men and women⁽ Reference Gonzalez, Veasey and Rumbold ^{28 ,} Reference Kelly, Guelfi and Wallman ^{30 ,} Reference Balaguera-Cortes, Wallman and Fairchild ^{39 ,} Reference Maraki, Tsofliou and Pitsiladis ^{41 )}, overweight and obese women⁽ Reference Tsofliou, Pitsiladis and Malkova ^{32 ,} Reference Unick, Otto and Goodpaster ^{33 )} and in response to a variety of exercise modes. Furthermore, Hagobian et al.⁽ Reference Hagobian, Yamashiro and Hinkel-Lipsker ^{29 )} reported no change in energy intake in response to exercise in men and women matched for age and VO₂ max.

The potential influence of exercise mode on the energy intake response to exercise has been demonstrated by Larson-Meyer et al.⁽ Reference Larson-Meyer, Palm and Bansal ^{31 )}. In this study, energy intake from an ad libitum meal that was provided 2 h after exercise was not affected by 60 min running at 70 % VO₂ max but increased in response to 60 min walking exercise at the same relative intensity in a separate group of participants. Although this study suggests that the exercise mode may influence energy intake responses, it is plausible that the higher percentage body fat and lower VO₂ max in the walking group may have confounded the results.

In this regard, Finlayson et al.⁽ Reference Finlayson, Bryant and Blundell ^{45 )} demonstrated that body fat and physical activity levels may influence the energy intake response to exercise in females. This novel study analysed the energy intake response to 50 min cycling exercise and separated the participants into two groups: compensators and non-compensators. Compensators were defined as the participants that increased energy intake beyond the energy cost of exercise, whereas non-compensators consumed less energy than that expended during exercise. Analysis of between-group differences revealed a significantly higher BMI and percentage body fat and a lower habitual exercise frequency in the compensators. Alternatively, recent evidence suggests that this relationship may not occur in men as Jokisch et al.⁽ Reference Jokisch, Coletta and Raynor ^{35 )} demonstrated that 45 min cycling at 65–75 % of the maximum heart rate decreased energy intake in a sample of inactive normal weight young men but not active normal weight young men.

Other studies have also demonstrated a decrease in energy intake but are confounded by the provision of the ad libitum meal within 10 min of exercise completion⁽ Reference Almada, Cataldo and Smalley ^{43 ,} Reference Westerterp-Plantenga, Verwegen and Ijedema ^{44 )}, which may have prevented the recovery from exercise-induced anorexia prior to feeding. However, in support of these findings, Ueda et al.⁽ Reference Ueda, Yoshikawa and Katsura ^{34 )} observed a decrease in energy intake at an ad libitum meal that was provided 60 min after a 60 min bout of cycling at 50 % VO₂ max. Furthermore, this reduction in energy intake occurred in both lean and obese male participants. Other authors have also reported similar energy intake responses in normal weight and overweight participants⁽ Reference George and Morganstein ^{27 )} but this is not a universal finding⁽ Reference Kissileff, Pi-Sunyer and Segal ^{46 )}.

Some authors have demonstrated an increase in energy intake during the immediate post-exercise period⁽ Reference Laan, Leidy and Lim ^{18 ,} Reference Martins, Morgan and Bloom ^{19 ,} Reference Verger, Lanteaume and Louis-Sylvestre ^{37 ,} Reference Shorten, Wallman and Guelfi ^{42 )}. The reasons for such variations in the energy intake response to exercise are unclear but may be influenced by a variety of factors including participant differences, the composition and timing of the ad libitum meal, variations in exercise mode and time-of-day effects. Although measures of absolute energy intake provide important information regarding feeding behaviour, King et al.⁽ Reference King, Burley and Blundell ^{40 )} suggested that it may be more relevant to express the energy intake response to exercise as ‘relative energy intake’ after deducting the net energy cost of exercise. Subsequently, all increases in absolute energy intake reported thus far in this review are negated after accounting for the energy expenditure of exercise. Although this approach provides an overview of energy balance, the short monitoring period of the studies described thus far bias the results towards a lower relative energy intake during exercise trials as food intake is unlikely to be up-regulated sufficiently at a single feeding episode to overturn substantial energy deficits. Therefore investigations into the energy intake response to multiple ad libitum meals are important and are discussed later.

Appetite responses beyond a single test meal (2–9 h)

Although some authors have suggested that exercise may stimulate appetite in response to a standardised meal in the post-exercise period⁽ Reference Malkova, McLaughlin and Manthou ^{47 )}, the majority of studies demonstrate that appetite does not increase above control values during the 2–9 h after exercise. This includes a variety of exercise modes including running⁽ Reference King, Miyashita and Wasse ^{14 ,} Reference Broom, Batterham and King ^{23 ,} Reference King, Lluch and Stubbs ^{48 –} Reference Wasse, Sunderland and King ^{50 )}, cycling⁽ Reference Wasse, Sunderland and King ^{50 )}, walking⁽ Reference Borer, Wuorinen and Chao ^{51 –} Reference Pomerleau, Imbeault and Parker ^{54 )} and resistance exercise⁽ Reference Broom, Batterham and King ^{23 )} in both male and female populations.

However, exceptions have been observed within the literature as King et al.⁽ Reference King, Wasse and Stensel ^{22 )} demonstrated that 60 min intermittent swimming exercise stimulated an increase in appetite compared with a resting control trial from 1·5 to 6 h after exercise. This contrasts with previous findings from the same author as 60 min brisk walking exercise⁽ Reference King, Wasse and Broom ^{53 )} and 90 min running exercise⁽ Reference King, Miyashita and Wasse ^{14 )} did not influence resting appetite perceptions during the 7 and 8·5 h post-exercise period, respectively. These differences occurred despite utilising a similar experimental protocol and participant population of physically active young men, which suggests a potential influence of exercise mode on subsequent appetite responses. However, subtle changes in meal timing and the recruitment of different participants confound any inferences regarding exercise mode.

The potential confounding influence of these factors is demonstrated by the findings of two studies from Broom et al.⁽ Reference Broom, Stensel and Bishop ^{12 ,} Reference Broom, Batterham and King ^{23 )}, as 60 min running at 70 % VO₂ max in young physically active males stimulated an increase in hunger from 2 to 8 h after exercise in the former but not the latter of the two studies. Such discrepancies highlight the need for within-measures study designs when comparing the appetite response to different exercise protocols.

Energy intake responses beyond a single test meal (2–9 h)

Few studies have monitored food intake in a laboratory setting for more than 2 h after an acute exercise bout. However, the available research suggests that exercise does not stimulate any changes in energy intake during the subsequent 22·5 h⁽ Reference King, Miyashita and Wasse ^{14 )}. This finding is also supported by studies that have used self-report measures of food intake, which have failed to discover any changes in energy intake during the 24 h⁽ Reference Hanlon, Larson and Bailey ^{55 )}, 48 h⁽ Reference King, Lluch and Stubbs ^{48 )} and 72 h⁽ Reference Pomerleau, Imbeault and Parker ^{54 )} after exercise.

Studies that have successfully performed prolonged monitoring of food intake in a laboratory setting have typically provided participants with one or two buffet meals during the 3–7·5 h after exercise. Studies from our laboratory have reported that energy intake is unchanged by exercise under these laboratory conditions at any of the provided feeding opportunities⁽ Reference King, Miyashita and Wasse ^{14 ,} Reference King, Wasse and Stensel ^{22 ,} Reference Wasse, Sunderland and King ^{49 ,} Reference King, Wasse and Broom ^{53 )}. Furthermore, the provision of an overnight food bag upon leaving the laboratory demonstrated that energy intake continued to remain unchanged for 22·5 h after exercise⁽ Reference King, Miyashita and Wasse ^{14 )}.

The importance of such prolonged monitoring of energy intake is demonstrated by Pomerleau et al.⁽ Reference Pomerleau, Imbeault and Parker ^{54 )}. In this study, energy intake at an ad libitum buffet meal was significantly higher 1 h after walking at 70 % VO₂ max compared with a resting control. However, after the provision of an additional ad libitum meal 6·5 h after exercise and an overnight snack bag, energy intake did not differ significantly between the trials.

One weakness of the earlier studies is that meals were provided to participants at pre-defined time points during the trials, which constrains the opportunities for food intake and may hinder the detection of differences in energy intake. This issue was recently addressed by King et al.⁽ Reference King, Wasse and Stensel ^{24 )} who allowed participants unlimited access to a buffet meal during the 6 h after 60 min of running at 70 % VO₂ max. In support of previous findings, energy intake remained unchanged after exercise compared with a resting control trial and resulted in a substantially lower relative energy intake after exercise.

Manipulating exercise protocols to produce the most beneficial responses

This review has demonstrated a consensus among the majority of the scientific literature that an acute bout of land-based exercise does not stimulate compensatory increases in appetite and energy intake during the following hours. Subsequently, recent efforts have focussed on manipulating exercise protocols to most effectively reduce appetite and energy balance in the hours after exercise.

In this regard, Cheng et al.⁽ Reference Cheng, Bushnell and Cannon ^{16 )} recently reported that moderate intensity cycling exercise induced a more prolonged suppression of hunger when performed 2 h after a high-fat breakfast compared with after a 12 h overnight fast. However, a subsequent investigation from our research group at Loughborough University suggested that this effect is limited to the period of exercise-induced anorexia, as appetite and energy intake did not differ from a resting control trial during the 5–9 h period after fasted or postprandial exercise despite greater appetite suppression during the postprandial exercise bout⁽ Reference Deighton, Zahra and Stensel ^{56 )}.

More commonly, investigators have manipulated the intensity of exercise when comparing exercise protocols but initial studies yielded equivocal findings. In this regard, Thompson et al.⁽ Reference Thompson, Wolfe and Eikelboom ^{20 )} demonstrated greater appetite suppression during cycling exercise at 68 % VO₂ max compared with 35 % VO₂ max but found no differences in energy intake after exercise. Conversely, Imbeault et al.⁽ Reference Imbeault, Saint-Pierre and Alméras ^{57 )} found no differences in appetite but reported a decrease in energy intake after running at 75 % VO₂ max compared with energy-matched walking at 35 % VO₂ max. Alternatively, Ueda et al.⁽ Reference Ueda, Yoshikawa and Katsura ^{21 )} and King et al.⁽ Reference King, Burley and Blundell ^{40 )} found no differences in the appetite or energy intake responses to cycling at 50 v. 75 % VO₂ max and 30 v. 70 % VO₂ max, respectively.

Interest in the influence of exercise intensity on appetite and energy intake responses has been renewed in recent years due to the popularity of high-intensity intermittent exercise (HIIE). In this regard, several authors within both the scientific literature and the media have suggested that HIIE may elicit greater weight loss than traditional endurance exercise due to greater reductions in appetite during the post-exercise period⁽ Reference Boutcher ^{58 –} Reference Bendoris ^{63 )}. However, despite these postulations, research from our laboratory has demonstrated that a recently popularised sprint interval exercise protocol, consisting of six 30 s maximal sprints on a cycle ergometer, stimulated compensatory increases in appetite during the 5 h monitoring period after exercise that did not occur in response to 60 min continuous cycling at 60 % VO₂ max. Although these differences in appetite did not influence ad libitum food intake on the day of exercise, similar absolute food intake between trials resulted in a substantial negative energy balance after endurance exercise compared with sprint exercise and a resting control trial due to a significantly higher energy expenditure during the endurance exercise than sprint exercise (mean 2640 (sd 418) v. 594 (sd 50) kJ; P<0·0005)⁽ Reference Deighton, Barry and Connon ^{64 )}.

In order to further investigate this issue, a high-volume, submaximal HIIE protocol was investigated in a subsequent study and it was determined that compensatory increases in appetite and energy intake did not occur during the 5 h monitoring period after ten 4 min cycling intervals at 85–90 % VO₂ max or 60 min continuous cycling at 60 % VO₂ max⁽ Reference Deighton, Karra and Batterham ^{65 )}. The absence of any compensatory increases in appetite in this study suggests that previously observed increases after supramaximal interval exercise are likely to be a result of the extreme intensity rather than the intermittent nature of exercise. It remains plausible that a threshold exercise intensity may exist for the stimulation of appetite during the post-exercise period but this requires further investigation. It is also important to note that these studies employed exercise protocols that represented both extremes of the HIIE spectrum (i.e. very low-volume supramaximal interval exercise and high-volume, submaximal interval exercise) and therefore provides convincing evidence that HIIE does not elicit lower appetite perceptions during the hours after exercise than traditional endurance exercise.