Timing matters: early time-restricted feeding (6h) improves insulin sensitivity in the absence of weight loss

I have previously written about the apparent effect of intermittent fasting (IF, or time-restricted feeding, TRF) on IGF-1 and inflammation. Specifically, eating all food during an 8 hour window appeared to lower IGF-1 levels in the absence of significant weight loss. And despite some critics arguing that subjects lost a not-significant amount of weight during the trial, I have also shown that weight loss by itself does not reduce IGF-1 levels, and thus, the reduction in IGF-1 appears to be due to TRF.

A common criticism (which in my opinion, is valid) of TRF studies is that usually, TRF leads to spontaneous calorie restriction (CR), making the interpretation of the findings difficult (are the results due to CR or TRF?). This holds true for markers that improve with weight loss (and thus, CR) like insulin sensitivity.

A new study just published is the most rigorously calorie-matched study to date comparing the effects of TRF on metabolic markers.

TRF comes in different flavors, depending on the time of the day in which food is eaten. Early TRF (eTRF) involves eating most of the food before sunset, ideally before 3–4 pm, and its based on the hypothesis that this type of feeding is aligned to our circadian rhythm. While there are several lines of evidence that suggest this to be beneficial, there has not been a direct testing of the effects of eTRF versus a “normal” eating pattern while rigorously controlling for calories.

The study

The authors studied the effect of eTRF (defined by them as “(…) a subtype of TRF in which dinner is eaten in the mid-afternoon”), in which all calories were consumed in a 6 hour window (18h of fasting/6h of eating) versus eating the same meals in a 12 hour window (12h of fasting/12h of eating). Importantly, this was a crossover study, which means that the same subjects (n = 8, overweight with prediabetes) underwent the two different feeding regimens for 5 weeks, after a 7 week wash-out period (time in between interventions):

All meals were provided by the authors so the two different feeding trials were meal-by-meal matched; that is, they ate exactly the same foods and the only thing that changed was the timing of feeding. Neither treatment produced significant weight loss (-1.4 ± 1.3 kg in eTRF vs. -1.0 ± 1.1 kg in control), which also suggests that eTRF does not affect energy expenditure significantly.

Interestingly, relative to eating the 3 meals in 12 hours, the eTRF intervention:

  • Reduced mean (-26 ± 9 mU/L) and peak (-35 ± 13 mU/L) levels of insulin during an oral glucose tolerance trial. Fasting insulin levels were reduced by 3.4 ± 1.6 mU/L.
  • Increased the insulinogenic index (a marker of beta-cell responsiveness) by 14 ± 7 U/mg and decreased insulin resistance by 36 ± 10 U/mg (3h incremental AUC ratio).
  • Reduced dramatically systolic (11 ± 4 mm Hg) and diastolic (10 ± 4 mm Hg) blood pressure, to a level comparable to some anti-hypertensive medications even in the absence of weight loss and only in 5 weeks.
  • Increased fasting triglycerides by 57 ± 13 mg/dL, which could be attributed to the larger fasting period before the test in the eTRF group.
  • Decreased levels of 8-isoprostane, a marker of oxidative stress, by 11 ± 5 pg/mL. Other inflammatory markers were not affected.
  • Decreased appetite and increased fullness in the evening.


  • eTRF appeared to have lasting effects on insulin sensitivity, as most of the participants who first completed the eTRF arm entered the control arm with lower (> 25%) postprandial insulin levels, despite 7 weeks of washout period between interventions.
  • One participant in which insulin levels worsened after eTRF had a long history of overnight shift work, which suggests that people who work during night shifts might benefit from alternative eating schedules.
  • The beneficial effects on insulin sensitivity and blood pressure where observed despite the diet being less than optimal. On the flip side, the diet that was used is similar to what a typical person would eat daily, which makes the intervention more feasible in the real world; just eating all your food before 3 pm without any further dietary change might be beneficial (although I would be very cautious with this interpretation).


I have been very skeptical of the idea that eating mostly during morning hours is necessarily better than eating in the afternoon when feeding windows are matched (so fasting periods are the same and therefore, feeding-fasting periods are time-equated). While this study does not answer that question (I have come across only two studies in rodents which control for feeding-fasting periods that show a very slight, inconclusive evidence in favor of eating mostly in the beginning of active hours), it certainly shows that just by limiting food intake to a short period during the morning/early afternoon (without any other dietary change), insulin sensitivity, blood pressure and oxidative stress improve in obese, prediabetic subjects.

It makes biological sense to limit food intake during the resting period (which circadian-wise, starts after the sunset in the absence of artificial/modern zeitgebers). It is now a very well established fact that insulin sensitivity/secretion follows a circadian pattern, being higher in the morning and decreasing towards the evening/night (see also this very nice, recent study on the topic). Thus, carbohydrate metabolism appears to be optimized for eating during the morning/early afternoon.

The results of this study reinforce what I have recommended based on current evidence: in sedentary subjects, it might be better to restrict the eating period to a short window of time (< 12h) during the day, ideally finishing eating before sunset.

However, it remains to be seen if the same happens for subjects undergoing any training program, specially resistance exercise during the evening. For instance, in the Moro et al. study, fasting insulin and glucose decreased significantly (suggesting improved insulin sensitivity) despite the fact that the feeding window was in the afternoon (from 1 to 8 pm). Importantly, training was performed between 4 and 6 pm, which strongly suggests that resistance exercise modifies the insulin sensitivity profile during the evening, as a single bout of resistance exercise improves insulin sensitivity for at least 24 hours. Thus, the importance of early feeding windows might be primarily (even maybe restricted) to sedentary people who do not engage in resistance exercise training (or any other kind of high intensity training). In accordance with this idea, eating 1 meal per day without CR between 4 and 8 pm (without exercise) worsens glycemic control and first-phase beta cell function compared to 3 meals per day.

Finally, the composition of the diet (in terms of macro and micronutrients, and food sources) might also impact in how much of an improvement is seen with eTRF. For instance, increasing protein and reducing carbohydrates in the diet might additionally improve glucose control in type 2 diabetic subjects. Probably, a combination with eTRF might be synergistic.


In obese, prediabetic subjects, eTRF appears to improve insulin sensitivity, blood pressure and oxidative stress in the absence of significant weight loss. Eating during a window earlier in the day might benefit sedentary subjects looking to improve their metabolic health, but this benefit might be greatly diminished (or even abolished) in subjects engaged in resistance exercise, specially during the evening. Additional dietary practices might also change the metabolic response to TRF.