When you eat, move, and sleep could matter as much as what you do, this study uncovers how the timing of daily habits influences your risk for type 2 diabetes, opening doors to truly personalized prevention.
Study: High-resolution lifestyle profiling and metabolic subphenotypes of type 2 diabetes. Image Credit: Nattapat.J / Shutterstock
In a recent study published in the journal npj Digital Medicine, researchers investigated the association between habitual lifestyle behaviors and metabolic physiology in individuals at risk of type 2 diabetes (T2D).
T2D incidence continues to rise worldwide, affecting 589 million adults globally and 38 million individuals in the United States (US). Further, 88 million adults in the US have prediabetes, with 70% projected to develop T2D within four years. Therefore, preventing this transition remains a major public health priority. Studies suggest that lifestyle modification is a robust means to manage and prevent T2D.
Diet, physical activity, and sleep are core modifiable lifestyle behaviors that are essential to metabolic health. Further, growing evidence suggests close interactions between the circadian clock system and lifestyle behaviors. Sleep deprivation adversely impacts glucose levels, and circadian desynchronization due to mistimed lifestyle behaviors could impair physiological responses and increase T2D risks.
The study and findings
The present study explored the relationship between habitual lifestyle behaviors and metabolic physiology in people at risk for T2D. Two cohorts were included; 36 healthy adults were included in the primary cohort, and 10 individuals were included in the independent validation cohort. In the primary cohort, 16 and 20 individuals were classified into normoglycemia and prediabetes/T2D groups, respectively, based on glycated hemoglobin (HbA1c) levels.
Habitual lifestyle data were collected using real-time digital health technologies. Dietary intake was logged using a real-time food tracking app. Data on physical activity and sleep were collected using a Fitbit Ionic band, though this data was only available for 24 of the 36 participants due to a product recall during the study period. Continuous glucose monitoring (CGM) was performed using a Dexcom G4 CGM device. An oral glucose tolerance test (OGTT), an isoglycemic intravenous glucose infusion test, and an insulin suppression test were performed.
These tests determined participants’ metabolic sub-phenotypes, such as incretin function, insulin resistance, and beta-cell dysfunction. The prediabetes/T2D group had significantly higher sensor-glucose (from CGM), sensor-glucose variation, and spent more time in the hyperglycemic range than the normoglycemia group.
Meal timing profiles were determined by stratifying food and beverage intake into six time frames, reflecting major food intake periods. Participants exhibited high interindividual variability in meal timing patterns. A principal component analysis based on the meal timing features delineated the cohort by their HbA1c levels into two clusters.
Individuals with elevated HbA1c had lower energy intake from meals consumed between 14:00 and 17:00 hours and higher energy intake from meals consumed between 17:00 and 21:00 hours than those with lower HbA1c. Additionally, the cohort was clustered by incretin function, and individuals with decreased incretin function exhibited higher energy intake during the 11:00–14:00 and 17:00–21:00 hours periods, and lower energy intake during the 14:00–17:00 and 21:00–5:00 hours periods.
Associations between sleep, physical activity, dietary features, and CGM and metabolic outcomes were assessed using the least absolute shrinkage and selection operator (LASSO) combined with regression models. Energy intake from meals between 14:00 and 17:00 hours was inversely associated with fasting plasma glucose (FPG).
Higher energy intake from meals during 17:00–21:00 hours was associated with more time spent in hyperglycemia, less time in the target glucose range at nighttime, and higher next-day mean glucose levels. Notably, these associations were not due to differences in total daily caloric intake, which was similar between groups, suggesting that the timing of meals itself was a key factor. Higher intake of carbohydrates from non-starchy vegetables was associated with reduced next-day mean glucose, whereas that from starchy vegetables was related to higher FPG and HbA1c.
Furthermore, greater variability in sleep efficiency was associated with higher nighttime glucose levels, a higher mean glucose level the next day, and a longer duration spent in the nighttime hyperglycemic range. In addition, higher variability in wake-up duration after sleep onset was associated with higher two-hour OGTT glucose. An earlier wake-up time was related to lower incretin effects. A longer sedentary duration during the day was associated with more time spent in hyperglycemia.
A higher step density after the last meal was associated with less time in nighttime hyperglycemia. Steps taken between 8:00 and 11:00 hours were associated with lower next-day glucose levels in the insulin-resistant (IR) group. Steps between 00:00 and 5:00 hours were positively correlated with higher glucose for the next 48 hours in the IR and insulin-sensitive (IS) groups. Steps taken between 14:00 and 17:00 hours showed a negative correlation with CGM values over the next 48 hours in the IS group.
Next, the team performed a permuted correlation network analysis between sleep, physical activity, and diet features, wherein all lifestyle factors were time-matched. This analysis showed significant correlations among lifestyle factors. Higher rice intake was associated with longer sleep latency and decreased sleep efficiency, whereas higher legume intake was associated with longer total sleep duration and shorter latency.
Additionally, higher intakes of fruits, potassium, and fiber were correlated with longer sleep durations. Longer fasting windows and higher energy intake from meals between 8:00 and 11:00 hours were correlated with longer sleep times. Further, the team built integrated lifestyle machine learning models to predict metabolic sub-phenotypes based on demographic and lifestyle data.
Higher carbohydrate intake from sweets and starchy vegetables, as well as increased energy intake during 17:00–21:00 hours, was associated with prediabetes and higher HbA1c levels. In contrast, higher carbohydrate intake from fruits was associated with normoglycemia. Older age, higher carbohydrate intake from noodles and pasta, increased protein intake, and higher energy intake between 17:00 and 21:00 hours were predictive of incretin dysfunction. Longer exercise duration predicted normal beta-cell function.
Finally, the team evaluated the reproducibility of prediction models using the independent validation cohort, focusing on incretin function, as other metabolic sub-phenotypes were highly skewed. This cohort also underwent continuous lifestyle monitoring and metabolic tests. Application of the prediction model to this cohort yielded 80% accuracy, with a misclassification error of 0.2, indicating robust and consistent predictive performance across cohorts.
It is important to note that the study's authors acknowledge some limitations. These include the modest sample size and the observational nature of the data, which means the findings show strong associations rather than direct causation. The research was also conducted in a single geographic area, indicating that more diverse populations should be studied in the future.
Conclusions
In summary, the findings provided a unique characterization of how habitual lifestyle patterns are related to metabolic susceptibility to type 2 diabetes (T2D). Habitual meal timing was linked to insulin resistance, lower incretin function, and hyperglycemia. Irregular sleep timing and efficiency were associated with higher glucose levels and IR. Crucially, the study revealed that the optimal timing for physical activity may depend on an individual's metabolic profile, with morning activity being more beneficial for individuals who are insulin-resistant and afternoon activity more beneficial for those who are insulin-sensitive. Overall, the findings highlight novel physiological connections between lifestyle behaviors and metabolic risk, informing the development of personalized lifestyle modifications and precision prevention strategies for the prevention of type 2 diabetes.