The main finding of this study is that greater energy intake after 5 pm is associated with poorer glucose tolerance in adults with obesity and diet or metformin-controlled prediabetes or T2D, independently of higher body weight or fat mass, diet composition or greater energy intake.

Our data confirm the association of late eating with worse glucose tolerance shown in previous studies in individuals without obesity [1]. Adding to previous findings on the detrimental effect of late eating on BMI and metabolism [2] and its association with poorer diet [2, 3], we now observed that the association of LE with poorer glucose tolerance is independent of greater body weight, fat mass, calorie amount, or poorer diet composition.

In previous studies associating eating late with poorer glucose metabolism, later eaters had higher BMI, higher body fat [5, 6], as well as lower satiety and greater hunger [7, 8] which may have explained their greater daily calorie intake. Food consumed in the evening, compared to morning, is typically higher in energy density resulting in an overall higher total energy intake [2], which may explain why late eating is associated with greater body weight and fat mass. Therefore, the glucose benefits observed when energy intake is distributed earlier in the day may be explained by a lower body weight. However, even in individuals reporting to consume the same total daily calorie amount, late eaters can present higher BMI/fat mass and poorer glucose metabolism, highlighting the potential role of meal timing per se, independently of calorie amount, on poorer metabolism [9,10,11,12].

Our study shows that older individuals with prediabetes or early T2D who are habitual later eaters have poorer glucose tolerance, independently of body weight or fat mass and energy intake. This is in agreement with short-term intervention trials (1–14 days) in healthy volunteer. Participants consuming an isocaloric diet aiming at weight stability showed worse glucose tolerance and lower resting-energy expenditure when calories were consumed later in the day [13, 14]. This may be related to previously reported higher postprandial glucose response after dinner compared to breakfast [14,15,16]. The importance of late eating on glucose was also shown in a prospective observational epidemiological study of 2642 women at risk of T2D: eating after 9 pm was associated with 1.5 times higher 5-year risk of developing T2D [17].

A combined intervention of caloric restriction with AM versus PM distribution of daily calories glucose and HbA1C decreased more, and insulin response was higher when calories were consumed in the morning compared to the evening [18, 19], highlighting the importance of meal timing on glucose metabolism in individuals with T2D. However, another weight loss study in 23 individuals with obesity and prediabetes or T2D, showed no differences in weight or metabolism when 50% of the total daily calories were consumed in the morning versus the evening [20].

Diet composition is also a well-established determinant of T2D risk. Observational studies have shown that late eaters tend to select highly processed high-carbohydrates and/or fats meals in the evening [2, 3]. Our study supports those findings. LE consumed more carbohydrates and fats after 5 pm compared to EE. This behavior has previously been associated with worse overnight glucose metabolism and may result in desynchronization of the peripheral circadian system [2] that can lead to even worse glucose tolerance.

Bias and limitations include the inclusion criteria of the NY-TREAT study [4], focusing on individuals with a prolonged ≥14-h eating window, introduces a potential bias. However, given the prevalence of such eating patterns in the general population, the data may still be representative. Despite real-time data collection via a smartphone app, there remains an element of self-reporting as participants need to remember to photograph their meals, though validations suggest a minimal 10% error rate [3]. The study’s small sample size is a limitation but, for pilot studies such as this one, former power calculation is not always possible. However, caution is advised in generalizing findings, as the cohort specifically targets individuals with prediabetes or T2D and obesity. Replicating the study in more diverse populations and age groups would enhance external validity, contributing to a broader understanding of the results’ applicability beyond the studied demographic.