Reversal of Obesity: The Quest for the Optimum Dietary Regimen

Adult bodies have a weight ‘set point’, a level about which weight may vary under normal conditions but which represents a stable range, despite variation in energy intake and expenditure. With any deviation from the weight set point, metabolic homeostatic mechanisms such as hunger or satiety ensure the return of the body to its set point range in a negative feedback mechanism. The arcuate nucleus of the hypothalamus integrates various signals from around the body to regulate energy balance through the secretion of gut and adipose tissue hormones, including ghrelin and leptin, and hypothalamic neurotransmitters such as neuropeptide Y, which induce or inhibit feeding. Although partially determined by genetics, nevertheless constant overeating or other factors may force the weight set point upwards, from where it is difficult to lower it because the homeostatic mechanisms will try to maintain the body at its new set point. It is thought that this explains why less weight is lost than expected during caloric deprivation. Continual secretion of the hormones results in insulin, leptin and/or ghrelin resistance, which leads to increased caloric consumption, further raising the set point. Much remains to be learned about this mechanism and it is hypothesised that all diets ultimately fail because no non-surgical mechanism is currently known to lower the set point to its original level. ²³, ²⁴, ²⁵, ²⁶

A few studies have investigated why weight loss programmes succeed or fail and determined that older age (particularly when first overweight), being female, fewer self-implemented weight loss attempts, greater initial weight loss, being unmarried, keeping a diary, positive mental health, frequency of GP visits, more exercise and less sedentary behaviour as well as greater dietary restraint were independent predictors of successful weight loss; studies were divided over the effect, if any, of education and ethnicity ²⁷, ²⁸, ²⁹, ³⁰. LaRose et al found a difference in motivation according to age, with successful younger adults being motivated by appearance and social acceptance, whereas older adults were more motivated by health aspects ³¹.

Furthermore, those weight loss approaches that work in clinical trials often involve supplied food portions and regular meetings with physicians, dieticians and therapists to keep subjects on track; where there appears to be some recidivism there may be ‘motivational phone calls’. This is very labour- and therefore cost-intensive for the healthcare profession and is not realistic for the large numbers of patients that need to lose weight and maintain the weight loss. In trials where food is not supplied, or during the weight loss maintenance phase, these weight loss regimens can be complicated and time consuming in terms of calorie counting. In addition, participants in weight loss studies are known for compensatory increases in hunger and may also experience cold, irritability and low energy.

Caloric restriction (CR) is a reduction of caloric intake without deprivation of essential nutrients. CR was first noted as a weight loss strategy in the 1980s with the advent of very low calorie diets (VLCDs) comprising fewer than 800 calories per day ³². These continuous caloric restriction (CCR) diets proved very effective at attaining modest weight loss in the short term by achieving a negative energy balance but they rarely achieved the desired or expected weight loss and were difficult to sustain long term ³². Furthermore, there was concern that the weight loss might lower resting metabolic rate and cause the depletion of lean mass, possibly affecting bone density or muscle mass; several studies have shown that lower lean mass is associated with reduced resting metabolic rate, which is disadvantageous for weight loss and its maintenance ³³. There is also the risk of nutrient deficiencies, electrolyte abnormalities and binge eating once normal eating is resumed ³⁴; the phenomenon of dieting resulting in increased weight from baseline is well known. These CCR diets are particularly problematic for patients because firstly they must count calories rigorously, they are usually permanently hungry and the food, being generally depleted of energy, is not satisfying ^{35, 36}; fatigue from lack of energy intake could also be a problem ³⁷. Furthermore, it was found that metabolic adaptation occurs when caloric intake is greatly reduced, which results in reduced thyroid hormone, catecholamine and leptin concentrations and a slower resting metabolic rate (RMR), causing subjects to be less physically active, which appears to make the body resistant to further weight loss and may cause weight regain ³⁸,³⁹.

A strategy which may address some of these issues is intermittent caloric restriction (ICR), where the limited intake generally lasts for any period ≥20 hours but is time limited, following which the individual may eat ad libitum for a day or two, after which ICR resumes. Most of the studies comparing CCR with ICR attempt to determine whether it is safe and equally effective. For some, this type of regimen may be easier to follow than CCR and its duration may be short enough to make compliance more successful and to overcome the problems of metabolic adaption ⁴⁰. Furthermore, ICR may carry additional benefits over and above those found in CCR, such as improved retention of lean body mass and less risk of malnutrition. As with CCR, there have been concerns that ICR would result in disordered eating patterns and over-consumption on non-fasting days, although short term studies have found no evidence of this ⁴¹,⁴².

A popular approach in recent trials is alternate day fasting (ADF), a variant of ICR, where calories are severely restricted (ranging from zero calories per day to <800 calories/day or 25% of energy requirements) for 24 hours, but the following day the individual may eat ad libitum. Although many rodent studies consistently show that it has beneficial effects on weight, glycaemic control and metabolism, there are, as yet, few RCTs that investigate the efficacy of ADF. Yet many researchers consider that it may hold the key to patient compliance, since one is only fasting for today and tomorrow one can eat as much as is wanted, returning a greater element of control to the individual ^{43, 44}.

Fasting and caloric restriction promote longevity and delay or slow progression of numerous age-related diseases in most laboratory animal studies ^{45, 46, 47, 48, 49}. The same result has been seen in human observational studies of long-lived populations, such as those in Okinawa, Japan ⁵⁰ and in studies of individuals who have been practising caloric restriction for some years ^{51, 52}. Even a reduction of 5% of body weight is associated with a marked decrease in incidence and progression of age-related conditions such as hypertension, cardiovascular disease, dementia and cancer ^{53, 54, 55}. In animal studies CCR and ICR appear equally effective in preventing mammary tumours ^{56, 57}, delaying a prostate cancer diagnosis ⁵⁸, protecting against cardiovascular disease (CVD) ⁵⁹, increasing resistance to neuronal damage ⁶⁰ and consequent cognitive impairment ⁶¹ and increasing lifespan ^{62, 63}. In humans, there are few randomised controlled trials (RCTs) of ICR but early results suggest it is as effective as CCR for overall health benefits ⁶⁴. Some have hypothesised that ICR may be even more effective at disease prevention than CCR due to increased cellular stress resistance and through brief energy restriction upregulating beneficial antioxidant enzymes, proteins and other protective molecules ⁶⁵.

Obesity is a manifestation of the increased storage of fatty acids as triglycerides in adipose tissue, particularly within the abdominal cavity. Obesity may be assessed by measuring weight but also by considering adiposity (fat mass), whether through BMI, waist circumference or some other measure. It has been shown that development of adiposity and the distribution of body fat may be a more accurate predictor of obesity-related pathology than weight gain. There are two types of adipose tissue: white adipose tissue (WAT), which stores energy, and brown adipose tissue (BAT), which generates body heat. WAT develops in size through increased storage of the fatty acids in triglycerides. This may occur either as an increase in the number of adipocytes or an increase in the size of adipocytes, although weight loss is believed to reduce the size of adipocytes, rather than their number ⁶⁶. The extent of visceral adipose tissue may be estimated from the extent of abdominal adipose tissue, with which it is strongly correlated. Both may be assessed by waist circumference measurement; a high waist circumference (the ‘apple’ shape) is a known risk factor for T2D and CVD, particularly in females, whereas a higher hip and thigh circumference (the ‘pear’ shape) is associated with a lower risk. A high waist circumference is also associated with elevated plasma triglycerides, as higher visceral lipid metabolism delivers high concentrations of non-esterified free fatty acids into the portal circulation. For this reason, many consider that the goal of treatment should be to reduce waist circumference rather than weight alone, especially in patients with T2D. ⁶⁷, ⁶⁸, ⁶⁹

But adipose tissue is not just a passive reservoir for storage of energy. All WAT acts as an endocrine organ, secreting hormones, pro-inflammatory adipokines and cytokines, such as oestrogen, leptin, adiponectin, retinol-binding protein-4 (RBP4) and tumour necrosis factor-α (TNF- α), which promote development of insulin resistance by interrupting insulin signalling and metabolism and can lead to cardiovascular disease (CVD). Visceral adipose tissue, the adipose tissue surrounding the organs, has higher endocrine and metabolic activity relative to subcutaneous adipose tissue, and is associated with higher cardiometabolic risk and incidence of insulin resistance. Higher subcutaneous relative to visceral adipose tissue is associated with a lower risk of obesity-related conditions. ⁶⁷, ⁶⁸, ⁶⁹

The relationship between obesity and insulin resistance has long been recognised, with studies showing a significant and broadly linear relationship between degree of insulin resistance and BMI in all ethnicities, including among the elderly. Insulin resistance correlated better with BMI relative to weight or other measures of adiposity. The only possible exception is among patients who are morbidly obese (BMI >40 kg/m²), suggesting that the relationship holds for all those with a BMI of 20-40 kg/m^2.. There is also a strong association between insulin resistance and visceral (abdominal) adiposity, which is stronger than with total adiposity. ³⁴,⁷⁰,⁷¹ Insulin resistance is found in virtually all T2D patients and in many of those who are obese but not diagnosed diabetic, since it occurs early in the condition ⁷². Nevertheless, there are ethnic differences in the incidence of insulin resistance, which occurs less frequently in African Americans compared to Caucasians or Hispanics ⁷³. Weight loss invariably leads to reduced insulin resistance, no matter how it occurs ⁷¹.

The human genotype has evolved over hundreds of thousands of years, when food sources were scarce and there were inevitably long time gaps between meals. Furthermore, intermittent fasting and increased interval between meals have been used in various religions and for health benefit for millennia ⁷⁴. Consequently we have developed genes adapted to this way of life, with increased storage of fat following a meal, which is slowly released to provide energy during periods with no food; there is no suggestion that our prehistoric ancestors were obese. Yet it is the general belief that to maintain a healthy weight, food should be taken in three meals per day, with snacks in between if hungry; in fact many doctors and dieticians recommend small, frequent meals (grazing) as the answer to weight loss on the basis that it increases satiety and reduces hunger. Yet this is counter-intuitive and there is concern that the practice may lead to over-consumption; furthermore, the continual stimulation of insulin secretion by constant eating, which generates a sustained level of blood glucose, could lead to or perpetuate insulin resistance.

There are a number of studies investigating the efficacy of small, frequent meals for weight loss but these are mostly observational and cross-sectional and the results are mixed. Furthermore, prospective studies which have compared the effects of 1-2 meals per day versus 3-5 meals per day have all been short term so the long term impact is unknown. Despite this, the American Dietetic Association recommends that ‘total caloric intake should be spread throughout the day, with the consumption of 4-5 meals’. ¹²,⁷⁵ However, since increased meal frequency appears in general to lead to a positive energy balance, this seems unlikely to benefit weight or fat mass, although usually there is no reduction in lean mass. The obverse of grazing is time restricted feeding (TRF), where food intake is ad libitum but is restricted to narrow windows of time (normally 4-13 hours in the day). This has also been explored in rodents and proved to reduce body weight and counteract a high fat diet ⁷⁶.

Bariatric surgery, normally Roux-en-Y gastric bypass (RYGB), has had an astonishing success in reducing weight and reversing T2D almost immediately following surgery but the mechanisms are uncertain. There has been controversy over whether the success is due to the surgery per se or to the calorically restricted post-surgery diet, or some combination of the two; furthermore, the reversal of T2D and normalisation of glucose metabolism and insulin sensitivity may itself be due to the weight and/or fat loss, with lowered release of adipokines.

A new approach to weight reduction is urgently needed, one in which weight loss is relatively rapid, cheap, simple to carry out, provides the essential nutritional requirements, has minimal adverse effects, can be carried out safely at home without any support or intervention from the health care profession and can be continued indefinitely without risk. Some of the recent clinical trials looking at intermittent caloric restriction, and particularly alternate day fasting, may point the way towards just such as approach. This article investigates RCTs comparing CR to other weight loss programmes and ICR versus CCR, the types of CCR or ICR diet, the addition of oral supplements, optimal meal timing, CR versus bariatric surgery, CCR versus ICR in unsupervised weight loss maintenance and the special case of ADF in order to determine what time of regimen is optimal for weight loss and weight loss maintenance. Recommendations will then be made for future studies.

Pubmed was searched for ‘fasting’ (ignoring studies relating to fasting blood measurements), ‘caloric restriction’ and ‘calorie restriction’. Studies were not included in this literature review unless there was an element of food abstention or caloric/energy restriction and the study lasted for longer than 1 week; studies of religious fasting were excluded. The studies were then categorised into those investigating some form of CR vs no CR, CCR vs ICR, types of diet, optimal meal timing, the efficacy of food supplements, CR vs bariatric surgery and weight loss maintenance. Only RCTs are included in the tables and detailed analysis but other studies are discussed if they provide additional clarification. Results are not mentioned unless there is a statistically significant difference between groups.

Ten RCTs (Table 1) looked at the efficacy of caloric restriction versus no caloric restriction or other forms of therapy for reduction in weight, BMI and fat mass over time periods ranging from 12 weeks up to 2 years. Shai et al⁷⁷, one of the 2 year studies, investigated overweight or obese middle aged adults with T2D or coronary heart disease, who were randomised to a CR low fat diet, a CR Mediterranean diet or a low carbohydrate diet; those randomised to the CR Mediterranean diet or the low carbohydrate diet lost significantly more weight than the CR low fat diet. Another, Brehm et al⁷⁸, a 6 month trial which randomised obese females to an ad libitum very low carbohydrate diet or a CR low fat diet, found that those on the low carbohydrate diet lost significantly more weight than those calorically restricted.

The remaining eight studies all found that CR was significantly more effective than the alternatives. These included studies of overweight Asian females with T2D, calorically restricted by 30% for 12 weeks ⁷⁹, normal or overweight Caucasians, calorically restricted by 25% for 2 years ⁸⁰,⁸¹, overweight or obese male Caucasians with T2D, calorically restricted by 25% for 6 months ⁸², normal or overweight subjects given 400-600 kcals/day ADF for 12 weeks ⁸³. The two exceptions ⁷⁷,⁷⁸ merely point up the fact that low carbohydrate diets have consistently proved to be more successful than low fat diets for weight loss, while adding CR does not improve the ability of a low fat diet to achieve weight loss. These two studies also suggest that it is not caloric restriction, per se, that causes the weight loss but that CR and low carbohydrate diets both help to trigger the mechanism of weight loss. Choi et al⁷⁹ also investigated the effect of CR on types of fat in Asian females and found significant reductions in total fat mass, body fat percentage and abdominal, visceral and subcutaneous fat. Although several of these studies showed lean body mass decreasing with body weight ⁷⁹, ⁸⁰, ⁸¹, Brehm et al found that this did not affect bone mineral content ⁷⁸.

Other benefits were seen with several, but not all, of the successful CR diets: significant reduction in blood pressure, heart rate, fasting glucose, HbA1c, insulin resistance, total and LDL cholesterol, leptin, liver enzymes and inflammatory markers, with significant increase in HDL cholesterol and LDL particle size (indicating a less harmful form of LDL) ⁷⁹, ⁸⁰, ⁸¹, ⁸², ⁸³. Triglyceride levels tended only to fall with low carbohydrate diets or ADF ⁷⁷, ⁷⁸, ⁷⁹, ⁸⁰, ⁸¹, ⁸², ⁸³, rather than CCR. The effect of CR on resting metabolic rate (RMR) and total daily energy expenditure was inconclusive ⁸⁰, ⁸¹, ⁸².

Caloric restriction is more effective for weight and fat loss than other weight loss regimen. One drawback to CR is that lean mass may also reduce.

Of the seven studies shown in Table 2, most show no difference between CCR and ICR, indicating that both would be of equal efficacy in promoting weight and fat loss. Two studies showed a greater reduction of BMI or body fat with ICR ^{84, 85}, while Arguin et al found increased loss of lean body mass with ICR compared to CCR ⁸⁶. The small study by Catenacci et al⁸⁷ employed a zero calorie ADF regimen versus CCR over 8 weeks and found that resting metabolic rate, absolute weight loss and change in fat and lean mass did not differ between groups, but there was a significantly greater decrease in fasting glucose with zero calorie ADF compared to CCR. Although there was no difference in degree of weight loss between groups, nevertheless Harvie et al⁸⁵ analysed those achieving ≥5% weight loss over the 3 months of the study and found that this was 65% in the 25% ICR group, 58% in the 25% CCR group and 40% in the 15% ICR group, suggesting that the greater the caloric restriction the better, regardless of whether it is CCR or ICR. Fasting insulin and insulin resistance declined to a significantly greater extent in the 25% ICR group compared to the CCR group, with no difference relative to the 15% ICR group. Compliance was significantly better in the ICR group vs CCR and intermittent restriction did not lead to disordered eating and overconsumption on non-restricted days. The earlier study by the same team also found that compliance among the ICR group vs the CCR group was increased ⁶⁵, with a similarly improved result for insulin and insulin resistance but the ICR group had a higher frequency of headaches, lack of energy and problems fitting the diet into their daily routine. Two of the studies showed a greater reduction in fasting glucose, total cholesterol and triglycerides with ICR vs CCR ^{84, 86}. Ash et al⁴² found that subjects who attained normal glycaemia had significantly greater weight loss than subjects whose HbA1c level remained >6%; they calculated that a 1% reduction in HbA1c was associated with a 6.5% weight reduction.

Both CCR and ICR are generally similarly effective for weight and fat loss although a few studies showed greater efficacy with ICR and greater improvement in blood glucose, insulin, insulin resistance and lipids. Compliance was improved with ICR vs CCR.

Table 3 shows the 12 studies looking at various dietary permutations of a caloric restriction regimen: 8 investigating CCR and 4 studying ICR. The 2 studies of ICR comparing low and high fat showed no difference between the groups ^{88, 89} and similarly with CCR, Brinkworth et al⁹⁰ and Kirk et al⁹¹ found no difference between a high fat, low carbohydrate diet and a low fat, high carbohydrate diet for weight, fat mass, fat-free mass or intra-abdominal fat and time to achieve 7% weight loss was the same. Nevertheless, Pascale et al⁹² found that the addition of low fat to CR enhanced weight loss overall but when analysed into those with T2D and those without but with a family history of the condition, it appeared that the fat restriction only assisted weight loss in those with T2D and made no difference in those without. To support this conclusion, the study by Brinkworth et al⁹⁰ and the 2 studies of ICR ^{88, 89} which showed not difference, all compared high and low fat diets in subjects without T2D. In addition, Das et al³⁵ found that in young, overweight adults there was no difference in weight, percentage fat loss, fat-free mass or resting metabolic rate between 30% CR with a low glycaemic or a high glycaemic load.

Three studies investigated whether a semi-liquid diet was more effective than solid food. Metzner et al⁶⁷ investigated overweight or obese females randomised to 1200 kcal/day solid food or isocaloric meal replacements for two meals, and found similar reduction in weight, BMI, waist circumference, fat mass and lean mass in both groups but there were a significantly greater number of subjects in the meal replacement group who lost >5% of baseline weight. Curiously, females with waist circumference ≤88 cm lost significantly more weight in the meal replacement group but there was no significant difference in weight change in females with waist circumference >88 cm. Wadden et al⁹³ found that a continuous liquid 420 kcal/day programme was significantly more helpful for weight loss than continuous 1,200 kcal/day of solid food after 1 year, while Kroeger et al⁹⁴ and Klempel et al⁹⁵, using the same subject group, gave an intermittent partially liquid diet (total calories 880-1080 per day) versus an isocaloric solid food diet for 8 weeks and showed that the liquid diet group had a greater decrease in weight and waist circumference compared to the solid food group but a similar reduction in BMI, fat mass and visceral fat; lean mass and abdominal subcutaneous fat were unchanged in both ICR groups. The remaining studies all investigated whether lower caloric intake was more effective than higher. Wing et al⁹⁶ studied subjects with T2D and found that those on a continuous diet of 400 calories/day lost significantly more weight than those on 1000 calories per day, although Foster et al⁹⁷ studied three different levels of continuous caloric restriction and Harvie et al⁸⁵ investigated 2 caloric levels plus carbohydrate restriction in non-diabetics and neither study found any difference between the groups.

Those studies that investigated cardiovascular or diabetes markers mostly found little difference between groups, although Brinkworth et al⁹⁰ found that total, LDL and HDL cholesterol increased and triglycerides decreased to a greater extent with the low carbohydrate, high fat diet compared to the high carbohydrate, low fat diet. Similarly, Harvie et al⁸⁵ showed that greater caloric plus carbohydrate restriction resulted in greater reductions in serum insulin and HOMA-IR, while Kirk et al⁹¹ found no difference in the change in fasting glucose, c-peptide, leptin, adiponectin between a low carbohydrate, high fat diet compared to a high carbohydrate, low fat diet, although fasting insulin and HOMA-IR were significantly decreased with the low carbohydrate, high fat diet and plasma 3-hydroxybutyrate was increased 10-fold. Pascale et al⁹² found no difference among diabetics if a fat restriction was added to the caloric restriction but among non-diabetics there was a greater decrease in total cholesterol in the low fat group.

Studies of ICR semi-liquid diet versus isocaloric ICR solid food generally found that fasting glucose, insulin, homocysteine, total and LDL cholesterol, triglycerides, inflammatory markers and liver enzymes, blood pressure and heart rate decreased and LDL particle size increased to a greater extent with the liquid diet; LDL particle size increased in the liquid food group, whereas there was no change in the solid food group ^{67, 94, 95}, although Metzner et al⁶⁷ found that while serum homocysteine increased in the CCR group, folate increased in the meal replacement group and a similar study by Noakes et al showed that both folate and β-carotene were higher in the meal replacement group ⁹⁸, possibly indicating the benefit of added micronutrients in the meal replacements, not present in self-selected food. Wing et al found that a lower calorie group had improved glycaemic control and insulin sensitivity after loss of 11% body weight compared to a higher calorie group ⁹⁶. There was generally little difference in compliance between groups, which was good or reasonable, and incidence of binge eating declined with time ⁹³. Nevertheless, Das et al³⁵, who compared CR with high or low glycaemic load, found that in the high glycaemic load group only there was a decrease in satisfaction with the provided foods and a significant increase in the desire to eat non-study foods during the first three months of the study.

Diet composition has little effect on weight and fat loss in caloric restriction, although a low carbohydrate diet improves the blood lipids, insulin and insulin resistance. Taking calories in liquid form can be helpful for weight and fat loss and improves biochemistry but greater caloric restriction was not necessarily more effective than less caloric restriction, except possibly in T2D.

Table 4 shows the six RCTs investigating the effect of CR with or without additional oral supplementation, three of which investigate ω3 fish oils, known to enhance lipid oxidation in healthy humans. The study by Su et al⁹⁹ shows that with the addition of protein and/or ω3 fish oils to CR, most groups had significant reductions in BMI and waist circumference but the reductions in weight and total body mass were not significant and the addition of protein and/or fish oils did not cause significantly greater reduction in BMI and waist circumference relative to CR alone, even after 12 weeks. However, the study by Munro et al¹⁰⁰ showed that ω3 fish oils can significantly enhance reduction in weight and BMI in calorically restricted obese adults, possibly through reducing insulin resistance, although there was no difference between groups in fat mass or hip measurement. Kunesova et al¹⁰¹ also found that the addition of ω3 fish oils resulted in significantly increased loss of weight, BMI and hip circumference. The reason for the beneficial effects of ω3 fish oils in the studies by Munro et al and Kunesova et al but none in the study by Su et al may be due to the much higher dosage of docosahexaenoic acid (DHA) in the latter two studies, since others have shown benefit for weight loss when DHA intake is considerably greater than that of eicosapentaenoic acid (EPA) ¹⁰², possibly due to its anti-inflammatory or lipogenesis suppression effects ¹⁰⁰.

Coker et al showed that the addition of whey protein and essential amino acids to CCR had no effect on weight loss but increased fat reduction by 30% relative to CCR alone in an 8 week study of the obese elderly ¹⁰³. Furthermore, in two studies by Georg-Jensen et al investigating the effect of CR with or without low viscous alginate fibre, the earlier ¹⁰⁴, which comprised 24 subjects and lasted two weeks, failed to find any benefit for reduction of weight and waist circumference with the additional fibre but the later study ¹⁰⁵, which comprised 80 subjects and lasted 12 weeks, showed a significant benefit of the alginate fibre on weight and percentage of body fat.

Su et al⁹⁹ showed that, despite the lack of effect on weight and BMI, CR supplemented with protein and ω3 fish oils resulted in a 1.5-fold greater recovery from metabolic syndrome (as shown by the Z-score) compared to CR alone. On the other hand, Munro et al¹⁰⁰ found no significant difference between CR with and without fish oil in the reduction in fasting glucose, total, LDL and HDL cholesterol, triglycerides, leptin and inflammatory markers. Kunesova et al¹⁰¹ found that the addition of ω3 fish oils to CR generated a higher increase in β-hydroxybutyrate, indicating higher ketogenesis and possibly higher fatty acid oxidation or decreased lipogenesis. In the study of the elderly by Coker et al¹⁰³, the addition of whey protein and essential amino acids did not significantly alter the change in fasting glucose or lipids relative to CR alone. The larger study of CR with or without alginate fibre found no difference in fasting glucose, insulin, HOMA-IR, total, LDL or HDL cholesterol, triglycerides, ghrelin, inflammatory markers or heart rate but there was a greater reduction in the alginate fibre group in HbA1c and systolic and diastolic blood pressure ¹⁰⁵.

The addition of ω3 fish oils to CR aided weight and fat loss, with greater ketogenesis, provided the dosage of DHA was sufficiently high. Whey protein and fibre may also be of benefit.

Table 5 shows the three RCTs that have investigated optimal meal timing with caloric restriction. Hoddy et al¹⁰⁶, investigating ADF with lunch only, dinner only or three small isocaloric meals, found no difference in weight between meal timing groups but systolic blood pressure was significantly lower in the small meals group only, while heart rate decreased in the lunch group only and resting metabolic rate reduced in the dinner group only, suggesting that dinner should not be the one meal of the day. A study by Williams et al¹⁰⁷ compared obese diabetics randomised to CR either one day per week per five weeks or to five consecutive days every five weeks, with behavioural therapy provided to achieve the goal of 1,500-1,800 kcal/day in order to determine the optimum interval for ICR but there was no difference between the two groups with respect to weight or fasting plasma glucose.

Stote et al investigated three meals per day compared with an isocaloric one meal per day and found that the one meal per day was significantly more beneficial for weight and fat mass reduction, although blood pressure increased, as well as total, LDL and HDL cholesterol and liver enzymes, which remained in normal range, but cortisol and blood urea nitrogen were decreased ¹⁰⁸. This was a short study, carried out in healthy normal weight subjects and it remains to be seen what the longer term outcome would have been and whether the same intervention would have raised glucose, lipid and enzymes levels in obese or diabetic subjects.

The studies are unclear whether one meal per day is more effective for weight and fat loss than three meals of the same total calories.

Although not all of the seven studies shown in Table 6 are RCTs and the study period is usually short, they do provide an interesting comparison of bariatric surgery versus caloric restriction. Four studies ¹⁰⁹,¹¹² found no difference between Roux-en-Y gastric bypass (RYGB) and caloric restriction of 700 or 1000 kcals/day or the standard post-surgery protocol (around 500 kcals/day), but little information is provided about the content of the diet. One study ¹¹³ found that a higher level of daily caloric allowance provided greater reduction in weight and BMI than RYGB, with a steady daily weight loss during caloric restriction but rapid loss following RYGB; this was a study of predominantly African American females, whereas the majority of the other study participants were Caucasians and it is not known if the ethnicity would have made a difference. Two studies ⁶⁸,¹¹⁴found that RYGB was more effective at weight and BMI reduction compared to CR but one of these ⁶⁸ was a low fat, high carbohydrate diet, which has already been shown to have little effect on weight; little information was provided on the diet content in the other study.

All studies of bariatric surgery focus mainly on incretins but those investigating other parameters generally found no difference in changes in fasting plasma glucose, insulin, HOMA-IR, C-peptide, thyroid hormones, leptin and other adipokines, inflammatory markers and liver enzymes ⁶⁸,¹⁰⁹,¹¹²,¹¹⁴. The fact that glycaemic control tends to be similar in the CR and RYGB groups suggests that it is the weight loss following caloric restriction after RYGB that brings about the improvements, rather than an incretin-mediated mechanism due to bypassing the duodenum ¹¹¹, ¹¹³. Nevertheless, Lingvay et al¹¹³, who found CR more effective than RYGB in African American females, also showed that fasting glucose and HbA1c were lower in the CR group vs RYGB, despite lower caloric intake in the RYGB group; insulin requirements were also significantly lower. Although Steven et al found greater weight loss with RYGB than CR, triglycerides rose after RYGB but fell in the CR group ¹¹⁴, while in the only study to test fasting glucagon, it was found to decrease to a greater extent with diet than RYGB ¹¹⁰.

Since these studies show that in most cases the results from CR on weight, fasting glucose, insulin and other biomarkers of T2D, are indistinguishable from those produced by RYGB plus CR, this suggests that rather than undergo expensive and potentially dangerous surgery with its known long term adverse effects, it may be preferable to attempt CR in the first instance. Where RYGB shows greater success than CR is in maintenance of weight loss, possibly since the consequences of overeating for RYGB patients are so unpleasant.

There is little difference in weight and fat loss, glycaemic control or other biochemistry between RYGB and CR, suggesting that the health benefits following RYGB are due to the caloric restriction following the surgery rather than the surgery itself.

The benefits of CR for weight loss maintenance was seen in a 2001 meta-analysis ¹¹⁵ of 29 US studies of subjects who had carried out a structured weight-loss programme. It found that continuous very low energy diets (VLEDs) of <800 kcal/day were significantly more successful at weight loss maintenance than continuous hypoenergetic balanced diets (HBDs) (not precisely defined) after 5 years. Table 7 shows 11 RCTs, the majority of which show no difference between CCR and ICR in respect of weight or fat loss maintenance, regardless of a low carbohydrate, low fat or low glycaemic load element or whether the calories were provided as liquid or solid food. The only factors that appear to make a difference are the addition of protein to a very low energy diet, which reduced weight regain ¹¹⁶,¹¹⁷ and zero calorie ADF, which slowed fat and lean mass regain ⁸⁷.

Most studies did not measure other parameters but three found no difference in resting metabolic rate, fasting insulin, glucose, insulin sensitivity, total, LDL and HDL cholesterol, triglycerides, leptin or ghrelin between the two study groups ³⁵,⁸⁶,⁸⁷, although after follow-up brain-derived neurotrophic factor (BDNF), important for cognition and weight loss, had increased in ADF but decreased in CCR ⁸⁷. Two studies of very low energy intake, with or without added protein, found that leptin and triglycerides were significantly lower in the protein group ¹¹⁶,¹¹⁷, while others found that all parameters had returned to baseline in both study groups ⁸⁷,⁹²

In an interesting study by Westerterp-Plantenga et al¹¹⁸, higher caffeine intake was associated with increased reduction in weight, BMI, waist circumference, fat mass and percentage body fat during the weight loss period, with lower resting energy expenditure and increased satiety and fat oxidation. Nevertheless, during the weight maintenance phase, a caffeine and epigallocatechin gallate (green tea) supplement was associated with significantly lower weight, BMI, waist circumference, fat mass, percentage body fat, fasting insulin and triglycerides and significantly increased fat oxidation and β-hydroxybutyrate among those with a normally low, but not high, caffeine intake and in fact this group regained no weight at all, while the other three groups regained varying amounts. This effect in low caffeine consumers only was hypothesised to be because the effect of caffeine appears to depend upon habitual intake, possibly due to lack of sensitivity with a habitually high intake.

Among the non-RCT studies, the beneficial effect of protein is echoed in a study by Arciero et al¹¹⁹, who investigated 24 overweight or obese adults who had been on a high protein CR diet for 10 weeks, resulting in significant decreases in body weight and fat mass. The subjects then chose whether to continue with this diet or switch to a traditional ‘healthy heart’ diet of low fat, low protein and high carbohydrate for a further 52 weeks. Those in the high protein CR group regained significantly less weight (<1% increase), total body fat and abdominal fat compared to the ‘healthy heart’ diet group (who gained 6.1% of their weight). Fasting glucose and insulin were unchanged from the beginning of the 52 weeks, indicating that the subjects retained their enhanced insulin sensitivity from the weight loss phase, despite the ‘healthy heart’ group regaining the weight and total body fat, suggesting that factors other than body weight and fat mass may mediate insulin resistance. An interesting study by Mutch et al¹²⁰ investigated 40 female Caucasians who followed an 8 week low calorie diet after which they all lost weight. After a six month weight maintenance phase, they were then stratified according to whether they were weight maintainers (0-10% of weight regain) or weight regainers (50-100% weight regain). Part of the difference between the two groups was genetic but the remainder was because weight maintainers experienced a significant reduction in insulin secretion in response to an oral glucose tolerance test, whereas no changes in insulin secretion were observed in the weight regainers.

There is little difference between CCR and ICR or diet composition in terms of weight loss maintenance, although adequate protein and zero calorie ADF proved to be most beneficial. Increased caffeine and green tea among those with normally low caffeine intake may also help. Whether or not an individual experiences a reduction in insulin secretion may help determine whether or not they will maintain the weight loss.

The only RCT comparing ADF with CCR was by Catenacci et al using zero calorie ADF; the results proved to be just as successful as CCR for weight loss over eight weeks and in preventing weight regain over 26 weeks in obese subjects ⁸⁷. Another zero calorie fasting study investigated severely obese patients prescribed either VLCD or zero calorie ADF but there was again no significant difference in the reduction in weight experienced by both groups ¹²¹. In a study to investigate the feasibility of zero calorie ADF in non-obese subjects for 22 days, participants were told that on the non-fasting day they should double their usual intake to make up for the calories lost the previous day. Despite these instructions, participants lost 2.5% of baseline weight and 4% of baseline fat mass and fat-free mass; insulin was also significantly reduced but resting metabolic rate did not vary, indicating that separate 24 hour fasts did not cause adaptation by the body. Hunger increased on day 1 but did not increase further, whereas feelings of fullness increased over the course of the study; no subjects left the study and compliance appeared to be good. ¹²² Similarly, a small study of obese vs non-obese young female patients in a Korean medical centre found that after nine days of zero calorie fasting there was a significant decrease in weight and BMI in both groups; although both experienced hunger, this decreased only in the non-obese subjects ¹²³.

ADF studies employing 25% of energy needs on the fasting day include one by Hoddy et al showing that in obese subjects ADF generated significant reductions in body weight, fat mass, fat-free mass and visceral fat mass; hunger had not increased from Day 1 and feelings of fullness on the fasting day increased as the study progressed ¹²⁴. An Iranian study also investigated ADF with 20-30% of normal energy intake for six weeks in obese females and found significant decreases in body weight, BMI, fat mass, waist circumference, systolic blood pressure and diastolic blood pressure ¹²⁵.

Unsupervised ADF for weight loss maintenance was also been investigated in the RCT by Catenacci et al⁸⁷. Using zero calorie ADF versus CCR, they found that after 24 weeks there was no difference in weight regain between the two groups but the ADF group showed a lower amount of fat mass regain. Other ADF studies include Klempel et al⁴⁴, who studied obese subjects undergoing controlled 500 kcal/day ADF for four weeks, followed by a further four weeks of unsupervised feeding and found that after the unsupervised feeding, body weight reduced further than was found after the controlled feeding; hunger decreased after approximately two weeks and there was no hyperphagic response on the non-fasting day. Similar studies were carried out by Varady et al⁴³ and Bhutani et al⁶⁴, both using a 25% energy requirement ADF, which significantly reduced body weight, waist circumference, percentage of body fat and fat mass, with a similar amount of weight lost in the supervised and unsupervised periods, indicating that this form of ADF can be maintained away from a clinically controlled environment, although there were weekly meetings with a dietician. There was no change in fat-free mass, suggesting that ADF preserved lean mass. Compliance was good at 86% during the controlled feeding and 89% during the unsupervised feeding; there were no complaints of fatigue ⁴³,⁶⁴

Alhamdan et al¹²⁶ carried out a meta-analysis in 2016 comprising 10 quality studies of overweight or obese healthy adults aged 18-70, comparing the efficacy of interventions lasting 3-12 weeks using ADF (4 studies) or CCR using very low calorie diets (VLCDs) of <800 calories per day (6 studies). Because of the few published ADF studies to date, the authors included non-randomised clinical trials in this category but all studies are recent (published after 1999). Interestingly, the authors noted that the ADF studies were all grant funded, while the VLCD studies were all industry funded. Some of these studies focused on weight maintenance after weight loss, rather than weight loss per se, but overall there was no significant difference in mean body weight loss or lean mass between ADF and CCR with VLCDs but ADF subjects lost significantly more fat mass than VLCD subjects. The authors also noted that in some VLCD studies, subjects had suffered headaches, fatigue, dizziness and hair loss, which had not been experienced in the ADF studies. Furthermore, they observed that ADF had been found to decrease hunger and increase satiety and dieting satisfaction, which may increase compliance, and there was no hyperphagic response on the non-fasting day. The authors commented that another reason for the increased compliance with ADF may be that no change in food type is required but merely a change in meal timing. Heilbronn et al¹²² had previously observed that unsupervised individuals who do not have their food portion provided may have great difficulty estimating energy intake in CCR, whereas with zero calorie ADF, there is no estimation required.

The principal concern about CR is the reduction in lean mass along with fat mass, with loss of bone and muscle during proteolysis, as reflected in increased blood urea nitrogen, which is often seen during fasting but usually only in fasts lasting >60 hours. In a 2015 review of ICR studies, Tinsley et al³⁴ showed that while ADF studies consistently showed reduction in weight and fat mass, among those that measured lean mass, 50% showed no change while 50% showed a reduction, although two of these three studies required zero calorie ADF, suggesting that a small protein-rich meal should be added to the regimen if ADF is continued for any length of time.

Although there are few studies, zero calorie ADF appears to be at least as beneficial as other forms of CR for weight and fat loss but may not cause a reduction in metabolic rate or lean mass. ADF may be more helpful for weight/fat loss maintenance, with good compliance and fewer adverse effects.

The RCTs show that a significantly greater amount of reduction in BMI and fat mass, as well as abdominal, visceral and subcutaneous fat, is experienced with CR compared to other weight loss regimens. There is generally little difference in reduction in adiposity between CCR and ICR, either in the initial weight loss period or during an unsupervised period, although in a few studies more fat loss may be found with ICR; a zero calorie ADF regimen slowed fat mass regain in the unsupervised period. More fat mass was lost with one meal per day rather than an isocaloric three meals per day.

Other studies show that CR of mean 1437 kcal/day for 3 weeks results in a significant decrease in waist circumference and intra-abdominal fat area ¹²⁷. Similarly, a VLCD administered for two weeks gave a significant reduction in BMI, abdominal adipose tissue and visceral adipose tissue but there was no difference in subcutaneous tissue ¹²⁸, while obese females given a VLCD for four weeks had a two-fold increase in lipolysis of abdominal adipose tissue, potentially contributing to a reduction in adipocyte size ¹²⁹. A study of 15% or 25% CR for 24 weeks could also reduce visceral fat cell size in proportion to weight loss ¹³⁰. A 2011 review by Varady ⁶⁹ noted that weight loss through CR was roughly proportional to loss of visceral fat mass, with CCR and ICR being similarly effective at reducing visceral fat mass. Animal studies reflect the effect of CR on adipocyte size, reducing white adipose tissue but increasing ‘beige’ fat deposits ¹³¹ and preventing visceral adipose tissue accumulation ¹³²; greater reductions in body weight were associated with more fat loss but not necessarily reduction in fat cell number ¹³³.

The initial effect of CR is water loss as a result of depleted glycogen stores ¹⁴. Following this, CR appears to function through increased lipolysis resulting in net fat loss from cells, leading to decreased adipocyte size and therefore reduced secretion of leptin and other pro-inflammatory adipokines ¹⁴,¹³³. Interestingly, Johnstone notes that some studies have shown that the slowest rate of weight loss (with an LCD, rather than a VLCD) is associated with the greatest loss of fat mass and the smallest loss of lean mass ¹⁴.

There is little difference in the effectiveness of ICR and CCR for adipose tissue reduction, although zero calorie ADF may be of advantage in slowing fat mass regain. CR appears to reduce the more dangerous visceral fat, decreasing adipocyte size, while having less effect on subcutaneous fat. The slowest rate of weight loss may have the greatest impact on fat mass.

Several of the RCTs of CR versus no CR show lean body mass decreasing with body weight and fat mass ^{79, 80, 81}, although one showed that this did not affect bone mineral content. Some of the RCTs also showed a reduction in resting metabolic rate and total daily energy expenditure. In comparisons of CCR with ICR, generally a similar amount of lean mass is lost, although one found increased loss of lean mass with ICR compared to CCR ⁸⁶. This occurred particularly among postmenopausal females, where the reduction was twice as great in ICR as in CCR, although in this study there was no association between lean mass reduction and change in resting metabolic rate. Among studies of weight regain during an unsupervised period, zero calorie ADF slowed both fat and lean mass regain ⁸⁷. Despite this possibly adverse effect of ICR vs CCR, a 2011 review by Varady ⁶⁹ noted that a higher proportion of lean mass was lost with CCR relative to ICR, although this could in part be due to different fat measurement techniques.

A non-randomised study by Arciero et al¹¹⁹ put 40 overweight or obese adults on a high protein CR diet for 10 weeks, following which there were significant decreases in weight and fat and lean mass. Curiously, resting metabolic rate increased, whereas in most other CR studies it decreased, causing the authors to hypothesise that the addition of increased protein prevented RMR from falling. Lean body mass is the single greatest predictor of resting metabolic rate and these two moved in tandem during the initial 10 weeks.

Both CCR and ICR can generate loss of lean as well as fat mass, although studies differ on which type of CR is more detrimental to lean mass. The addition of protein to CR was able to increase resting metabolic rate and lean body mass.

Although almost all the RCTs analysed above investigated subjects who are aged <65, CR appears to be effective in all age groups. Yet in general, studies of the elderly show that they may be more successful with dietary restriction than younger individuals, possibly due to altered biochemistry but maybe also because they are more prepared to follow the instructions of health professionals ¹³⁴. It is also thought that the usual definitions of obesity and overweight based on BMI may not be applicable in the elderly due to the change in stature and body composition that accompanies ageing and that waist circumference may be a better measure of adiposity in this age group ⁷⁰. Since in the elderly, overweight or mild obesity can protect against osteoporosis, fractures and mortality, there is particular concern that CR may impact bone density and muscle mass, triggering or worsening the known ageing condition of sarcopenic obesity, where depleted muscle mass is combined with a redistribution of body fat to the abdomen ⁷⁰. One study in this review focusing specifically on the obese elderly showed that the addition of protein may aid fat reduction and should also promote lean mass retention, and the addition of resistance exercise may also be of benefit. There are as yet no long term RCTs of the effects of CR in the elderly.

Virtually all studies employ female, or predominantly female, subjects, yet there is no suggestion in any study that gender impacts the effectiveness of CR or any form of CR. The three exclusively male RCTs showed that CR was effective for weight and BMI reduction in those who were obese or had T2D and that the findings were in line with those for females in studies lasting 12 weeks to two years. Few studies analyse ethnicity but where details were given, it appears that the results are similar regardless of whether the subjects are predominantly Caucasian or Asian (Japanese, Taiwanese, Korean), although a study which compared ethnicities found that Caucasians aged 50-59 achieved significantly greater weight loss than other age and ethnic groups ¹³⁴. One of the RCTs analysed above ¹¹³ was a comparison of RYGB or CR in African American females and was the only one of the bariatric surgery versus CR studies to demonstrate that CR alone was significantly more effective for weight reduction than the CR following RYGB. Another study provided a comparison of Caucasian and African American premenopausal females who were calorically restricted for 12 weeks and found an ethnic difference only among those with greater baseline weight, where Caucasians lost significantly more weight than African Americans but African Americans had a greater decrease in waist:hip ratio ¹³⁵. However, a study of Caucasian males and matched South Asians showed that the South Asians were more insulin resistant at baseline, although the results of an eight day VLCD were the same for weight loss ¹³⁶.

CR is beneficial for weight and fat loss in all age groups but particularly the elderly, with added protein to aid lean mass retention. It appears helpful for both males and females and in all ethnicities, but possibly particularly for Caucasians.

There are many RCTs investigating caloric restriction and exercise but since this paper is concerned with the optimum dietary regimen, only an overview of the exercise studies will be provided. Because of the multitude of these studies, only those published after 2011 (10 RCTs in total) have been considered. Seven of these show that, although significant weight loss occurred in all groups, there was no difference in the weight loss between diet, exercise or diet + exercise ^{137, 138, 139, 140, 141, 142, 143}. Two studies investigating body composition found that body mass did not decrease in the resistance training group but in the resistance training + CR group there was a significant reduction, particularly in abdominal obesity ⁷⁰,¹⁴⁴; similarly, the resistance training + CR group showed a significantly greater reduction in metabolic syndrome prevalence, VLDL, triglycerides and systolic and diastolic blood pressure relative to resistance training alone ¹⁴⁵. Only one study found that the combination of exercise + ADF was more effective than either alone, although exercise on its own was significantly less effective than ADF ¹⁴⁶.

In general, exercise has little or no impact on weight loss when compared with or combined with CR.

Among the RCTs, those comparing CR with no CR found no serious decline in mood or cognition among the CR subjects and no increase in serious adverse clinical events. Similarly, where there was a difference between CCR and ICR, compliance was slightly improved in the ICR group and there was no disordered eating or overconsumption on non-restricted days ^{65, 85}. Incidence of binge eating in ICR, where present, declined with time ⁹³. A notable feature of some of these RCTs is that subjects consistently fail to achieve the target calorie or energy restriction; in Ruggenenti et al⁸² the goal was 25% caloric restriction but only 15% was achieved, while in Ravussin et al⁸⁰ the goal was again 25% caloric restriction but only 12% was achieved. Although significant weight loss nevertheless occurred, this provides an indication of how difficult a goal of 25% caloric restriction can be unless all meals are provided. Hunger was in general higher at the start of the study and declined as the body adapted to lower intake, however, a small study of young female Koreans found that after nine days of zero calorie fasting, all had experienced hunger but this decreased only in the non-obese but not in obese subjects ¹²³. Increased protein intake is normally found to aid satiety.

In studies of ADF, compliance was generally good, both during controlled and unsupervised feeding ⁴³,⁶⁴,⁸³, although hunger was reported on fasting days early in the study but lessened with time ¹²²; other studies reported increased feelings of fullness ¹⁴⁶. There were no complaints of fatigue. The meta-analysis by Alhamdan et al¹²⁶ noted that with VLCDs some subjects suffered headaches, fatigue, dizziness and hair loss, which were not experienced in the ADF studies. Furthermore, they observed that ADF had been found to decrease hunger and increase satiety and dieting satisfaction, which may increase compliance, and there was again no hyperphagic response on the non-fasting days. The authors commented that a reason for the increased compliance in ADF may be that no change in food type was required but merely a change in meal timing. Nevertheless, Hoddy et al¹²⁴ noted that these were subjective ratings of hunger and fullness, taken in the evening of a fast day, so that a post-prandial assessment before and after ADF were not given. Objective measures of hunger and satiety, such as ghrelin or PYY were not assessed. In their study, Hoddy et al¹²⁴ showed that despite considerable weight loss after eight weeks of ADF, ghrelin, an indicator of hunger, was unchanged from baseline and PYY, an indicator of fullness, increased during the study, corresponding with the subjective assessments. Nevertheless, Hoddy et al recognise that these findings are not in accordance with the usual findings in CR studies. Johnstone ¹⁴ notes that some ADF studies appear to suggest that there may be a difference in perception of hunger between those of normal weight, who continue to feel hunger throughout the study, and obese subjects, whose hunger declines as the body adapts, although a small study by Halberg et al showed that non-obese males habituated to ADF after approximately two weeks and felt more satisfied with the diet after about four weeks ¹⁴⁷.

Finally, Johnstone noted that in an unsupervised study comparing fasting, LCD and VLCD, that after 12 months it was the fasting group that best maintained the initial weight loss. This could be explained by the subjects’ reports that they had confidence that they could restrict food intake for a period of time with no ill effects and they now employed the technique to maintain their weight, although Johnstone pointed out that this was not an RCT and the subjects had volunteered for the fasting group, suggesting that this more radical approach might suit them better.¹⁴

CR appears to carry no significant risk of serious clinical events or decline in mood or cognition. ICR generated slightly improved compliance with no disordered eating or overconsumption on non-restricted days. Headaches, fatigue and dizziness were avoided with ADF. Hunger generally declined with time, especially with higher protein intake.

The RCTs analysed above comprise a mixture of subjects who are of normal weight, overweight, obese and/or with T2D. Although no study has set out to compare the impact of CR between diabetics and non-diabetics or groups with different BMI, CR appears to benefit all. Yet there are a few differences, which it may be worth highlighting. Firstly, Pascale et al⁹² found that CR with a low fat diet enhanced weight loss in those with T2D but made no difference in those without, whereas three other RCTs comparing CR with high and low fat diets in non-diabetics all showed no difference. Pascale et al further found that total cholesterol was reduced to a greater extent in non-diabetics on CR with low fat, compared to diabetics. The fact that a low fat diet can aid weight loss in diabetics is a somewhat surprising finding since in general a low fat diet has not been found to support weight reduction. Furthermore, in non-diabetics the degree of CR appears to make little difference to the extent of weight, BMI and fat loss but in diabetics those on 400 kcals/day lost significantly more weight than those on 1000 kcal/day. A difference in effect between diabetics and non-diabetics is not a phenomenon that has previously been highlighted but the number of studies in which it has been found are small and this should be tested in larger studies. With respect to baseline weight, Varady ⁶⁹ noted that whether a subject was overweight or obese did not affect the relative amount of fat to lean mass lost during weight loss using CCR, although this meant that obese subjects lost more weight in absolute terms than overweight subjects.

Although none of the RCTs investigated type 1 diabetes, in which the incidence of obesity also appears to be growing, nevertheless a study of obese type 1 diabetics showed that caloric restriction for 21 days in a controlled environment could safely result in reduction in weight, BMI, fat mass and waist circumference and allowed a significant reduction in insulin dose ¹⁴⁸. There is, however, a recognised risk of hypoglycaemia with caloric restriction in type 1 diabetes, so regular monitoring would be required.

The RCTs analysed above are not homogeneous, since some consider energy deficit while other prescribe a set number of calories, but it appears that a 25% calorie deficit has a better weight loss response relative to a 15% deficit ⁸⁵, while zero calorie ADF resulted in less fat regain during unsupervised weight maintenance than a 400 kcal/day deficit ⁸⁷. Yet other than these, and the study showing that among diabetics only a lower calorie intake led to greater weight loss, the majority of studies show that there is little difference between greater or lesser caloric restriction for reduction of weight and fat mass.

However, the Dietary Guidelines for Americans recommend that an energy deficit of at least 500 kcal/day is necessary for weight loss. This was tested in 54 overweight or obese adults, who were given the goal to lose 5% of their total body weight in 14 weeks through creating an energy deficit of ≥500 kcal/day. It was found that those who averaged an energy deficit of >500 kcal/day lost nearly 4 times the weight of individuals whose energy deficit was smaller, while those who achieved the goal of 5% weight loss carried out self-monitoring nearly twice as often as those who failed. ¹⁴⁹ In terms of absolute amounts, a retrospective study of 1887 outpatients found that 520 kcal/day versus 850 kcal/day for 12 weeks produced significantly greater weight loss and improved weight maintenance, although after controlling for baseline body weight there was no significant difference between diets for initial weight loss or weight regain. In fact the 850 kcal/day diet had a lower incidence of adverse events and less need for medical monitoring ¹⁵⁰. Several other studies have also shown that the degree of energy restriction is not always reflected in the weight loss results, which may in part be due to techniques to measure food intake ⁶⁹.

A meta-analysis of six RCTs by Tsai et al³² compared VLCDs to LCDs, with a follow-up of at least 1 year weight maintenance period. VLCDs usually comprise a liquid or partially liquid meal replacement with added protein to preserve lean mass, normally providing <800 kcal/day; they are normally administered only under medical supervision in the US, although this is not a requirement in Europe. The meta-analysis results showed that VLCDs produced significantly greater short term weight loss but similar results at follow-up after the weight maintenance phase due to greater weight regain among the VLCD group; similar numbers dropped out of each group and there were no adverse events in any study. These results show that there is no long term advantage to VLCDs over LCDs, whereas on a VLCD, subjects were more likely to experience increased risk of gallstones, cold intolerance, hair loss, headache, fatigue, dizziness, volume depletion (with electrolyte abnormalities), muscle cramps and constipation, although these symptoms are usually mild and easily managed.

So while it is unclear whether or not an energy deficit of at least 500 kcal/day is necessary for weight loss and maintenance, the reduced need for medical monitoring and lower reports of adverse events may make a higher caloric allowance for CCR more practical, although use of ICR may avoid some of the medical and supervision problems of CCR. Nevertheless, Tsai et al make the point that if an easy and reliable method of unsupervised weight loss maintenance could be found, the ideal combination might be a supervised short term VLCD programme followed by the LCD unsupervised weight maintenance programme ³². This is echoed by Klempel et al⁹⁵, who suggest combining one day per week of zero calorie fasting with six days per week of 20% energy restriction, possibly through liquid meal replacement for one meal per day, a regimen shown to be highly successful for weight loss in animal models. Furthermore, use of meal replacements largely overcomes the recognised problem of portion size estimation

The RCTs analysed above failed to find a consistent answer to the optimum meal timing. Nevertheless, in perhaps the simplest study, in which three meals per day were compared with an isocaloric one meal per day i.e. time-restricted feeding (TRF), Stote et al showed that the one meal per day was significantly more beneficial for weight and fat mass reduction. The authors note that this may be because the participants failed to eat the same number of calories as they would have in three meals per day due to extreme fullness, although the drop-out rate was relatively high in this group. ¹⁰⁸ The few other human studies investigating this issue tend to confirm the lack of consistency. Garrow et al showed that consuming one meal per day for a week resulted in greater weight loss than an isocaloric diet consumed in five meals per day ¹⁵¹ and Belinova et al found that in type 2 diabetics a hypocaloric diet consumed at breakfast and lunch compared to the same diet divided into six small meals per day had a significantly greater effect on weight, plasma ghrelin, hepatic fat content, insulin sensitivity and feelings of hunger ¹⁵², although there was no difference in weight reduction in obese males consuming an isocaloric diet in one, three or six meals per day for five weeks ¹⁵³ or in obese females consuming breakfast and dinner versus 3-5 isocaloric meals per day for four weeks ¹⁵⁴. A study which combined TRF and ADF found no difference in weight after 15 days ¹⁴⁷, although when TRF was investigated in resistance-trained male athletes, who were required to consume 100% of energy needs in three meals during an eight hour period for eight weeks, compared to an isocaloric control group, the TRF group had lost a greater amount of fat mass, with no difference in lean mass or blood lipids ¹⁵⁵. A similar trial by the same group, in which the athletes were required to consume all energy needs in a four hour window for four days per week found no difference in body fat composition compared to the control group, despite a reduction in total energy intake ¹⁵⁶.

Animal studies reveal a circadian rhythm in the development of obesity, as the time of feeding and fasting affects several genes which are key regulators of glucose and lipid metabolism. This has become known as ‘chrononutrition’. Most studies have been carried out on mice, which are mainly nocturnal feeders, making any translation of results to humans problematic. In general, 24 hour fasting mice showed a large reduction in hepatic gene transcripts compared to normal feeding, while day-fed mice demonstrated an altered oscillation of rhythmic transcripts to normal night-fed mice and developed obesity, perturbation in glucose and lipid homeostasis and insulin resistance. Animal models of obesity regularly display disruption in the feeding/fasting rhythms and circadian gene expression, with obese mice choosing to feed during the day as well as at night. Increased propensity to obesity and metabolic syndrome has also been observed among human night shift workers; in a laboratory setting, where subjects had induced dyssynchrony between feeding and fasting times and a disrupted circadian clock, weight gain and metabolic disturbance ensued, with decreased levels of satiety hormones, elevated post-prandial glucose, insulin resistance and mean arterial pressure. ¹⁵⁷ Furthermore, there appear to be daily rhythms in glucose homeostasis and insulin sensitivity, which naturally decline over the course of the day. These rhythms are driven by molecular ‘clocks’ in the hypothalamus, which are affected by light, but peripheral clocks are located in other tissues involved in lipid and glucose metabolism, such as adipose tissue and the liver, which are controlled mainly by feeding times. ¹⁵⁸ Another aspect of this is ‘food entrainment’, the internal mechanism through which the circadian clock genes are controlled by daily scheduled food availability. This allows the body to realign the timing of behavioural and physiological functions focused on the anticipation of food (known as food anticipatory activity). Caloric restriction, and also presumably TRF, have a strong impact on food entrainment, which appears to influence dopaminergic pathways which enhance locomotor sensitisation. ¹⁵⁹

Studies show that in animals with diet induced obesity, ad libitum feeding extends for the full 24 hours, with development of metabolic disorders and down-regulation of genes associated with metabolism. Nevertheless, isocaloric TRF can restore the normal feeding pattern and hepatic gene transcripts and protect against or reverse obesity, hyperinsulinaemia, leptin resistance, hepatic steatosis and inflammation, reducing white adipose tissue and protecting brown adipose tissue from ‘whitening’; the benefits of TRF were directly proportional to the duration of food restriction. In addition, TRF restores circadian expression of the down-regulated genes. ^{157, 158} In addition, more fat was lost with all calories eaten in one meal a day than ad libitum, while brown adipose tissue content increased, although insulin resistance was also increased. ¹⁶⁰

A review by Rothschild et al found that the effect of TRF on weight differs between animals and humans, although the decrease in glucose, insulin, total cholesterol and triglyceride concentrations is comparable. Paradoxically, while a shorter eating window induces considerable weight loss in animals it has no impact on humans, while a 10-12 hour window, which produces inconsistent results in animals produces consistent weight loss and reduction in LDL cholesterol in humans. The reason for this is unknown but the authors note that the animal studies were all RCTs, whereas the human studies had no control group. Interestingly, fewer participants withdrew from TRF studies relative to intermittent fasting studies. ¹⁶¹ Other studies have shown that TRF in humans can increase brown adipose tissue at the expense of white, as well as synchronising circadian rhythms and metabolism, although the mechanism for this is not yet understood. While the optimum time window for TRF is still unclear, studies have shown that morning feeding is more beneficial to weight loss, fasting glucose, insulin sensitivity and lipid profile than evening feeding and that any meal consumed after 1500 hours is more likely to increase weight than a meal consumed before that time; late night eating is also associated with development of coronary heart disease. This may be because gastric emptying is more robust during the daytime and metabolism of glucose is slower in the evening than morning. In addition it is thought that some fasting programmes impose a diurnal rhythm in food intake, leading to improved oscillation in circadian gene expression which reprogramme energy metabolism and body weight regulation. ^{12, 76, 162, 163}

The RCTs show that most of the successful CR regimens found a decrease in fasting plasma insulin and insulin resistance, where measured, which in some cases was greater with ICR compared to CCR ⁸⁵ and was aided by a low carbohydrate diet, a semi-liquid diet and lower versus higher caloric intake ^{91, 96}. The RCT by Wing et al⁹⁶ suggested that decreased insulin sensitivity is dependent on degree of caloric restriction and the magnitude of weight loss, which are independent of each other.

Two further studies suggest that the role of insulin is closely linked to successful weight loss. Mutch et al¹²⁰ found that during an unsupervised six month weight loss maintenance phase, those who regained only 0-10% of lost weight experienced a significant reduction in insulin secretion in response to an oral glucose tolerance test, whereas no changes in insulin secretion were observed in those who regained 50-100% of lost weight. Furthermore, Hoddy et al analysed their obese non-diabetic subjects into degrees of insulin resistance and found that ADF for eight weeks generated a significantly greater reduction in fasting insulin and HOMA-IR in subjects in the highest insulin resistance tertile, compared to subjects in the lowest tertile, despite equal reduction in weight and fat mass in each tertile ¹⁶⁴. This would indicate that CR provides the greatest improvement in insulin sensitivity to those with the greatest degree of insulin resistance. This is in contrast to another study, which found that those obese non-diabetics with the highest baseline plasma insulin concentrations and the greatest insulin resistance lost the least weight with CCR and regained it more readily ¹⁶⁵; nevertheless, this could perhaps indicate the greater efficacy of ADF vs CCR.

Since any food raises blood glucose and triggers secretion of insulin, which inhibits lipolysis, it is thought that the reduction in food intake and periods of abstinence from food force a switch of fuel substrate from oxidation of glucose to oxidation of fatty acids once glycogen stores are extinguished. This is seen in the elevated levels of plasma β-hydroxybutyrate and fatty acid concentrations, as stored triglycerides are broken down and adipose tissue lipolysis releases fatty acids into the circulation; a decrease in fasting triglycerides is usually also seen. Some studies suggest that the fast must last for >18 and preferably 24 hours to obtain the full benefit, including increased resting energy expenditure. ³⁴

Very few studies have been carried out on the weight set point; it is almost impossible to assess in humans and in animals the set point must be calculated from hoarding behaviour. Nevertheless in the few studies which have calculated it, obese rodents were found to have a higher weight set point but caloric restriction was found to reverse ghrelin resistance and alter neuropeptide secretion on a temporary basis but the lost weight was eventually regained, particularly with a high fat diet, although a low fat diet may have more success. ^{25, 166} Other factors which appear to lower the set point, at least temporarily, in animals include nicotine ¹⁶⁷ and acute, but not chronic, exercise ¹⁶⁸. Consequently, although temporary weight loss can be achieved in rodents by caloric restriction, this may not be sufficient to alter the weight set point permanently, leading to rebound weight gain. This suggests that some form of caloric restriction must be maintained for life.