Food Timing, Circadian Rhythm and Chrononutrition: A Systematic Review of Time-Restricted Eating’s Effects on Human Health

The collection and selection of articles was carried out by three independent researchers from September 2019 to September 2020 using the keys words time-restricted feeding, time-restricted eating and time-restricted nutrition.

The screened databases were MEDLINE, Web of Science, Scopus, Science Direct and the Cochrane Library. In addition, the following specialized journals were analyzed: Nutrition Reviews from the Oxford Academy, Nutrition Reviews from the Wiley Online Library, Obesity from the Wiley Online Library, the American Journal of Clinical Nutrition, Nutrition—Annual Review of Nutrition and Clinical Nutrition. The research equations used for each database are listed in Supplemental Method S1.

The included articles were all clinical trials, published between January 2014 and September 2020, in English, evaluating TRE on adult human populations.

Literature reviews, meta-analyses and any other type of publication were excluded, as were studies on Muslim fasting.

Once the articles were collected, they were selected through a process including three steps. The first was the selection of articles based on a reading of their titles. The remaining articles were then selected based on their abstract. Finally, the full article was read completely before being included or excluded according to the abovementioned criteria.

All the selection steps were carried out blindly by three operators. In the case of disagreement, a discussion between the three operators and the supervisor was carried out until consensus was reached.

For the constitution of the database, the authors used the Rayyan QCRI web application, specialized in systematic literature reviews [24].

The data were extracted from the original articles and classified in a table () according to the following themes: reference of the article, type of study, characteristics of the participants, main outcomes, main results and level of evidence. These data were then synthesized in a table for analysis. Data extraction and analysis was carried out by a single operator, the author of the review. Supplemental Table S1

All the included articles will be referred to in this paper as RXX, with XX being a number.

Levels of evidence were classified according to the Downs and Black checklist [25], which is a 27-item validity score assessing all the methodological aspects of clinical trials. Two questions about the double-blind method (14 and 15) were uncounted, which brought the score to 25 (Supplemental Table S2).

The second indicator was the classification provided by the High French Health Authority (HAS) [26] (Supplemental Table S3—HAS gradation).

The combination of these two indicators allowed for the determination of three levels of evidence—high, medium and low.

Preliminary research identified four hundred and ninety-four articles, including clinical trials, literature reviews and other relevant work. After the exclusion of one hundred and six duplicates, three hundred and eighty-eight articles remained, from which 22 have been finally selected [27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49]. An article was added during the full article reading stage (R15—[41]), making a total of 23 articles included for analysis (see flow chart in Figure 1).

The literature on TRE is limited, with only 23 articles having been published, reporting the results of 22 trials. The studies took place over periods ranging from four days to four months and involved groups of 8 participants for the smallest sample and 105 participants for the largest one. The overall level of evidence was low to medium, with 10 controlled randomized trials with short intervention periods (average of 33.5 days) and small samples (average n = 21) for the medium-quality evidence studies. The other 11 studies were of low quality, with three non-randomized controlled studies and eight single-arm non-controlled trials. One study had a high level of evidence, with a controlled and randomized protocol and the largest sample of the review trials (105 participants) (R23).

However, it is important to note that 28 experiments evaluating TRE were still in progress at the closure of the collection period (March 2020). Among these 28 trials, 26 are randomized and involve larger samples (mean n = 81), which will provide better quality evidence on the subject. These observations indicate that TRE is a research subject of growing interest and knowledge.

Time-restricted feeding is an intermittent fast which consists in limiting the daily feeding period [17]. However, the definition of the restriction is still imprecise in studies on humans and varies from 8 to 12 h, depending on the sources.

First, the nomenclature used to describe the type of time-restricted feeding practiced is standardized: TRE followed by the number of hours of feeding and fasting in a 24-h period, separated by a slash. For example, TRE 8/16 indicates a restriction of the daily feeding period to 8 h for a 16 h fast.

Among the 22 studies in the review, only one practiced a restriction equal to 12 h or less (R13), all the others having limited daily nutrition to a window of 10 h or less. These data clarify the definition of TRE in humans and indicate that it starts from a period of 10 h, a daily fast of 14 h and more.

The 8/16 variant is the most common form of TRE, with 50% of the trials in the review (n = 12). One TRE protocol (R1) did not impose any hours of fasting but required participants to shift their first meal and advance the time of their last daily meal by an hour and a half.

The food window can also vary during the day. For example, Sutton et al. (R4) introduced the concept of early TRE (eTRE), in which the feeding period started earlier in the day (R2, R4 and R5). By extension, delayed or late TRE (dTRE) delays the intake of the first meal (R3, R7, R8, R9, R12, R14 and R23). Hutchinson et al. (R11) have evaluated both early and late TRE. Three studies (R16, R17 and R18) allowed the self-selection of the food intake period by the participants.

TRE is generally an every-day (except for R12 and R13), water (except for R13) and ad libitum fast, which means that participants can eat until satiety without any restriction on the quality of quantity of food intake, except for four studies which used iso-caloric protocols (R2, R4, R19, R21).

These characteristics show that TRE could be adaptable to the participants’ eating habits, which may positively influence dietary adherence.

TRE was overall well-tolerated, with an adherence rate superior to 80% in eight studies (R6, R10, R12, R16, R17, R18, R20, R21) on the 10 that have evaluated adherence (Table 1). However, Gabel et al. (R7) have reported a dropout rate of 24% and in another feasibility study (R22), the compliance to the 9/15 TRE was 72 ± 24% (i.e., ~5 days/week).

Otherwise, adherence to meal timing was self-declared by the participants in diet diaries or interviews for most of the studies (R1, R3, R6, R7, R10, R11, R12, R13, R14, R16, R17, R19, R22, R23). Food intake measurement was also based on participants’ self-declarations (R1, R7, R8, R9, R12, R13, R14, R19, R22, R23). The food intake was not even tracked in seven of the studies (R3, R6, R10, R11, R16, R17, R20). These limitations could lead to declaration and confusion bias.

This limitation is an inherent difficulty in measuring health behaviors like food intake in free-living conditions. Gill and Panda (R15) developed a smartphone application which combines a photograph and an optional notification added by the participant (myCircadianClock, mCC—www.mycircadianclock.org↗). This method, also used in Wilkinson et al. (R18) and Chow et al. (R20), can represent an interesting way to reduce the methodological bias involved in food tracking. Studies have thereby shown that using a mobile application is more accurate for monitoring compliance to a dietary protocol than a traditional diet diary, and that it can also improve adherence to this protocol [50].

A recent literature review indicated that the main determinants of adherence to a diet are the ability to reduce the desire to eat and to conform to the patient’s eating habits [51]. In the present review, hunger remained stable in four studies (R4, R5, R11, R21), and in Ravussin et al. (R5) participants reported an improvement in several components of appetite by reducing hunger discomfort and the desire to eat. In addition, time-restricted eating appears less restrictive than classic diets, with no limitation in the quality or quantity of the food, and thus more adaptability to the participants’ eating behaviors.

Not only the adherence, but the long-term adherence is the most important factor of the success of a diet on overall health benefits [52]. The studies in this review were four days to 4 months long, which is a relatively short time to conclude. In the trial conducted by Wilkinson et al. (R18), 63% of the participants were somehow engaged in TRE 16 months after the end of the 12-week intervention. This result suggests that TRE could be a sustainable dietary intervention and it would be interesting to evaluate adherence over a longer period (12 to 24 months).

One may have expected that the restriction of feeding time would increase the calorie intake. Inversely, TRE reduced caloric consumption by 20% on average, without an alteration of macronutrient distribution (Table 1). Interestingly, this calorie restriction has been described as not intentional by the participants, which is likely to protect them against cognitive restriction. A study revealed that cognitively-restrained eaters increase their energy intake compared to baseline and lose less body weight than unrestrained eaters after a 10-month dietary intervention [53].

Moreover, TRE provides a similar level of calorie restriction to other voluntary restrictive dietary measures, such as continuous caloric restriction. Indeed, the CALERIE is the largest ever study to evaluate continuous caloric restriction in a non-obese population [54] and its results indicated that even the most motivated subjects were unable to maintain a 25% calorie restriction over two years. Put together, these findings suggest that (1) TRE could represent a more sustainable strategy for patients who want to reduce their caloric intake and (2) TRE could produce a more sustainable body weight loss than voluntarily restrictive diets.

It has been established in the literature that intermittent fasting produces beneficial metabolic effects via caloric restriction [55,56]. This principle is confirmed by the results of the review, with an average weight loss of 3% and fat mass perdition (Table 1), body fat loss and a decrease in the level of fasting glucose concentration (R1), weight loss and BMI reduction in (R7, R9, R15, R18), a drop in systolic blood pressure (R7, R9, R18), a reduction in waist circumference and a decrease in cholesterol concentration (R18).

Interestingly, six studies on TRE have shown beneficial metabolic effects, regardless of the calorie restriction (Table 1). For example, a randomized iso-caloric study evaluating eTRE showed a decrease in the average blood sugar level and reduced insulin resistance (R2). Likewise, a crossover randomized trial (R21) demonstrated that short-term TRE improved nocturnal glycemic control. An iso-caloric trial on eTRE 8/16 (R4) showed an improvement in glucose tolerance and a major decrease in systolic blood pressure, comparable to the results obtained with pharmacological treatment with an ACE inhibitor [57]. Moro et al. reported that a combination of TRE 8/16 with regular physical activity reduced fat mass, decreased blood glucose levels, improved insulin resistance and lowered triglyceride levels (R8). Grant et al. (R14) showed a reduction of 4%–7% of fat mass, despite an increase in caloric intake between groups, but only on a per protocol analysis. McAllister et al. showed body weight and fat mass loss, a reduction in systolic blood pressure and an increase in HDL cholesterol in two groups applying 8/16 TRE in ad libitum and isocaloric conditions, with no difference between groups (R19). The authors suggested that this effect was due to the increase in adiponectin levels. Adiponectin is an adipokine of which the secretion increases during fasting [58] and which could produce a loss of fat mass by increasing energy expenditure [59,60] and by activating AMPK, a kinase which stimulates lipolysis and insulin-sensitivity [61,62,63]. An adiponectin level increase was also observed in the randomized study by Moro et al. (R8). These findings are interesting and suggest that the TRE could produce positive metabolic effects independently of energy balance.

Although metabolic changes are the result of an imbalance between the energy content of food eaten and energy expended partly via physical activity [64,65], only three of the six mentioned studies measured physical activity (R8, R14, R21). In the Moro et al. trial, for example (R8), 8-week TRE associated with resistance training (RT) produce the abovementioned metabolic effects in comparison with a group of adults who performed only strength training, with no difference in food intake and physical activity. In this study, the physical activity was only measured during the training session, thus the stated effects could be due to an unmeasured energy expenditure increase outside of training sessions. Tinsley et al. also evaluated the effects of TRE associated with resistance training (R4). They measured physical activity during and outside training sessions, which limits the potential bias. It is necessary for further studies on TRE to systematically control the impact of physical activity to confirm these interesting effects.

It has also been established that the beneficial metabolic effects of a diet are induced by the weight and/or fat loss it produces [56]. This assertion has been confirmed in seven studies of this review (R1, R3, R7–9, R8, R17, R18, R19—Table 1). Wilkinson et al., for example, performed a 12-week 10/14 TRE on 19 adults with metabolic syndrome, and recorded LDL and non-HDL cholesterol reduction, a decrease in systolic blood pressure associated with a body weight loss of 3%, body fat perdition and the reduction in BMI and waist circumference. In contrast to other diets, TRE generates metabolic benefits regardless of any loss of weight or fat. For example, Hutchinson et al. showed an improvement in the glucose tolerance and a decrease in triglycerides in participants practicing eTRE for two weeks (R11). Furthermore, Sutton et al. recorded an improvement in glucose-tolerance and a decrease in insulin levels and blood pressure (R4). These promising results suggest that TRE generates an intrinsic metabolic effect, independently of weight loss or fat mass.

The main mechanism hypothesized to explain this intrinsic effect is its action on the body’s circadian rhythm [66]. The circadian system represents all the physiological processes involved in a 24-h cycle, such as the sleep/wake cycle, blood pressure, heart rate, hormone secretion, cognitive performance and mood regulation [66]. The system is regulated by a number of environmental stimuli such as food intake, light exposure and physical activity [67]. Growing evidence shows that the disruption of the circadian system is the cause of many metabolic pathologies like obesity [68,69]. Thus, limiting the time of food consumption seems to readjust the food intake with the circadian clock [70,71,72,73]; this is a fundamental principle of chrononutrition, which is the study of relationships between circadian clocks and food intake [74] and which suggests that meal timing affects metabolism [75].

Several results support this hypothesis. First, adiponectin has been established to play a key role in all the metabolic effects described above, such as the loss of fat mass [62], promotion of glycemic and lipid metabolism [60,63] and reduction in blood pressure [76]. This molecule is known to actively regulate the action of the circadian system [77,78]. In addition, a randomized controlled iso-caloric study involving large-scale gene expression analysis showed that eTRE affected the expression of six genes involved in circadian rhythm (R2). Finally, a study indicated that TRE produced a significant and durable improvement of sleep quality (R15) and a 12-week single-arm trial evaluating 10/14 TRE on 19 adults recorded an increase in sleep duration and efficiency in 84% of the participants. Moreover, the improvement of sleep quality is known to be directly linked to the enhancement of the circadian system [79,80]. However, the exact action of TRE on circadian systems remains unclear in humans, and studies are needed to better understand the mechanisms involved.

A growing number of observations on animal models hypothesize that the positive effects of physical activity on cardiovascular markers are also due to the ability to restore the circadian rhythm [81,82]. Moro et al. (R8) compared a group combining a TRE diet and an RT program (three sessions/week) with a group practicing RT alone for eight weeks. Only the TRE + RT group recorded a decrease in total body mass and fat mass as well as an improvement in metabolic markers (glycemic tolerance, decrease in triglycerides). However, as mentioned above, energy expenditure through physical activity was only recorded during training sessions, which could lead to a classification bias. Future studies should systematically control this factor and could test TRE alone vs. TRE + RT. Again, smartphone-based methods represent an interesting strategy to measure physical activity and energy expenditure in normal living conditions [83].

Other beneficial effects were explored in our review. Intermittent fasting has been shown to prevent cancer by lowering the IGF-1 level in animals [84] and humans [84]. This effect has been stated in two review studies (R2 and R8) and suggests a protective effect of TRE in carcinogenesis. In addition, researchers have shown that an intermittent one-month fast leads to a decrease in pro-inflammatory cytokines [85]. This result was also confirmed by a review study which reported a drop in TNF-α and IL-1-β levels after 8 weeks of TRE associated with physical training (R8). The decrease in these mediators of inflammation is otherwise hypothesized to play a role in circadian rhythm improvement through the action of adiponectin [86]. Gene expression analysis has shown that TRE increases the expression of four genes involved in autophagy and longevity (R2). This study was also the first to show that TRE increases the secretion of BDNF, a protective factor against the development of neurodegenerative diseases [87]. In addition, two studies exploring the effects of TRE in the elderly showed an improvement in walking speed, a predictor of geriatric robustness [88], as well as an attenuation of immuno-senescence process (R10 and R13). Furthermore, Sutton et al. (R4) showed that TRE causes a significant decrease in 8-isoprostane, a biomarker of oxidative stress [89]. Although they are promising, these observations are based on biological criteria whose clinical implications are limited. Longitudinal studies are needed to assess the impact of TRE on the morbidity and mortality linked to these.