Introduction

Given the high prevalence of impaired glucose tolerance (IGT) and type 2 diabetes mellitus (T2DM), coupled with their significant role as risk factors for cardiovascular disease (CVD) and all-cause mortality, these conditions have emerged as a major public health challenge.1,2 Research indicates that postprandial glycemic response (PPGR) is not only a critical determinant of overall glycemic control in diabetic patients,3 but also serves as an important predictor of both microvascular and macrovascular complications.4,5 Therefore, effective management of postprandial blood glucose (PBG) holds substantial practical significance for improving public health.

Continuous glucose monitoring (CGM) shows comprehensive glucose profiles by measuring interstitial glucose levels 24 hours a day. This continuous data stream captures the direction, periodicity, and magnitude of glucose fluctuations related to meals, exercise, and medication, providing a crucial overview of glycemic patterns. Currently, CGM is increasingly used to evaluate PPGRs to food in both individuals with and without diabetes mellitus (DM).6–13

Carbohydrate has a more profound impact on PBG than any other dietary component.14 In Asia, particularly in China, refined carbohydrate-based foods such as white rice and wheat products (eg, steamed bread, noodles) serve as major staple foods,15 which constitute the main dietary source of carbohydrates. Furthermore, oats have been widely recognized for their potential health benefits, which include reducing glycemic response, regulating blood cholesterol levels, improving gut microbiota balance, and aiding in blood pressure regulation.16 As a result, they have become an increasingly common staple in daily diets. Although the glycemic index (GI) of major staple foods such as rice, noodles, and oats has been extensively documented in the literature,17,18 it is important to note that GI values are predominantly established under standardized laboratory conditions in healthy individuals. Emerging research consistently show that identical meals trigger markedly heterogeneous PPGRs across different individuals.6–8,11–13 In real-life dietary contexts, there remains a notable research gap regarding the relative effects of these staples on PPGR.6–8,11–13 This gap is particularly pronounced in populations that require precise glycemic control, mostly those with IGT and DM. Although chrononutrition studies have demonstrated that consuming the same meals in the late evening leads to higher PPGR compared to earlier in the daytime,9,19–22 few studies have simultaneously compared PPGR differences across breakfast, lunch, and dinner (with the same food consumed at each meal) within the same cohort of patients with DM.

Individuals with dysglycemia ultimately return to everyday life. We therefore developed a streamlined, real-world framework using CGM to map their PPGRs to common foods under free-living conditions, as a first step toward precision nutrition. To address the gap in simultaneously evaluating meal timing and staple food type within the same dysglycemic population, we developed a study focusing on these two most readily modified and variable components of daily eating. We characterized PPGRs at both group and individual levels in out-patients with abnormal glucose metabolism and preliminarily examined how demographic, biochemical, pharmacological and lifestyle factors contribute to inter-individual differences. Although dietary fat and protein also modulate PPGRs, the present work focused specifically on carbohydrate-driven effects.

MethodsStudy Design

This one-stop framework captures real-world PPGRs in type 2 diabetes outpatients: a single clinic visit delivers dietary guidance and CGM sensor insertion; participants then consume the test meals at home and self-remove the device at completion—no return trips, minimal burden, maximum real-life data.

Within this framework, we conducted standardized glucose and staple-food challenges: this was a prospective observational study. A total of 33 participants with IGT and T2DM were enrolled. Over a one-week free-living period, participants completed 7 standardized meal tests. Concurrently, CGM was utilized to precisely assess: 1) changes in PPGRs when the same glucose was ingested at different meal times and 2) differences in PPGRs when various staple food types were ingested at the same lunch time (to mirror customary Chinese eating patterns - where lunch is typically the heaviest carbohydrate meal - we scheduled the staple-food challenge at lunch mealtime).

The study protocol was approved by the Ethics Committee of Peking Union Medical College Hospital (No.: JS 2450) and registered at clinicaltrials.gov (NCT04562454). All participants provided written informed consent prior to enrollment. The study was conducted in accordance with the latest version of the Declaration of Helsinki.23

Participant Recruitment

Participants were recruited from patients diagnosed with IGT or T2DM at the Department of Endocrinology outpatient clinic from March 2024 to September 2025. Inclusion criteria were as follows: age between 18 and 75 years, body mass index (BMI) ranging from 18 to 35 kg/m2, glycated hemoglobin (HbA1c) <11%, either untreated or on a stable medication regimen for more than 3 months. Permitted medications mainly included metformin, dipeptidyl peptidase-4 inhibitors (DPP-4i; eg, sitagliptin, linagliptin), sodium-glucose cotransporter 2 inhibitors (SGLT2i; eg, dapagliflozin, empagliflozin), and glucagon-like peptide-1 receptor agonists (GLP-1RA; eg, semaglutide, liraglutide). Additionally, participants were required to have no history of using α-glucosidase inhibitors or exogenous insulin within 3 months prior to enrollment.

Exclusion criteria included: 1) pregnancy or lactation; type 1 or specific types of DM; 2) uncontrolled hyperthyroidism or hypothyroidism; history of malignancy within the past 3 years; chronic gastrointestinal disorders; eating disorders; any form of gluten intolerance or wheat allergy; hypersensitivity to medical adhesives; chronic anemia; acute cardiovascular events or acute glycemic events (such as diabetic ketoacidosis or hyperosmolar hyperglycemic state) within 6 months before enrollment; 3) unwillingness to comply with the study diet, inability to fully understand the study procedures or unwillingness to participate.

Standardized Meals

Participants were instructed to complete a meal test once per day over one week, totaling 7 tests. For each test, they were asked to consume a standard meal containing 50g carbohydrates. These tests included: glucose solutions at breakfast, lunch, and dinner (anhydrous glucose powder, Fuzhou Haiwang Fuyao Pharmaceutical Co., China); steamed wheat bread (steamed until fully cooked and fluffy), Japonica rice (cooked with 250 mL of water in a rice cooker until done), wheat noodles (boiled in 500 mL of water for 5–8 minutes until al dente), and oats (pure oat flakes, Seamild, China) at lunch. The meals were consumed at each participant’s habitual mealtimes. No washout interval of ≥2 days was implemented between the tests. To ensure freshness and food safety, as well as to better emulate real-world conditions, participants were encouraged to prepare steamed bread, rice and noodles by themselves according to standardized guidelines (see Appendix Method 1). Participants were required to remain at rest for 4 hours before and after each test meal, refraining from strenuous physical activity and avoiding additional food intake. Throughout the study period, aside from the standardized test meals, participants were asked to maintain their usual daily routines, dietary habits, and glucose-lowering medications.

Characteristics of Participants

During the recruitment phase, participants received comprehensive training in the outpatient department regarding the study’s objectives, detailed procedures, and specific requirements for the standardized diets. Prior to the application of CGM, researchers collected extensive data from each participant, including anthropometric measurements (eg, height, weight), biochemical analyses, as well as responses from questionnaires covering lifestyle and demographic characteristics (eg, gender, age, education, birthplace, residence, smoking status, alcohol consumption, sleep duration, and dietary habits). Height and weight were measured under fasting conditions, with participants in lightweight clothing, without shoes, and after emptying their bladder. BMI was calculated using the formula: BMI (kg/m2) = body weight (kg)/height2 (m2). Obesity was classified as BMI levels of 28 kg/m2 or higher.24

Biochemical analyses were performed in accordance with standard protocols using blood samples obtained after an 8-hour fasting. Analytes included glucose, insulin, HbA1c, triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C), creatinine (Cr), uric acid (UA), and albumin-to-creatinine ratio (ACR) of urine. The homeostasis model assessment (HOMA) was applied to evaluate insulin resistance and beta-cell function using the following formulas:

HOMA of insulin resistance (HOMA-IR) = [fasting blood glucose (FBG, mmol/L) × fasting insulin (FINS, mIU/L)]/22.5

HOMA of beta-cell function (HOMA-β) = [20 × FINS (mIU/L)]/[FBG (mmol/L) - 3.5] (%)

CGM

The CGM (Guardian™ Sensor 3, Medtronic) was attached to the posterior aspect of the non-dominant upper arm of each participant. The device automatically recorded interstitial glucose concentrations every 5 minutes and was removed on the day of the final test meal.

Data Processing and Cleaning

To minimize self-reporting bias, the following measures were implemented: 1) participants were encouraged to take photographs of each meal and record the exact time of intake; 2) kitchen scales were provided for precise measurement; 3) a standardized protocol was established to regulate food preparation and consumption procedures (see Appendix Method 1).

Due to the initialization delay of CGM, participants who attached the CGM in the morning started the standardized meal test in the evening of the same day, while those attached in the afternoon began the test in the morning of the next day. Test meal data were excluded during cleaning phase based on the following criteria: 1) deviation from the protocol in self-reported mealtime or food weight; 2) PBG failed to rise effectively, remaining below baseline values; 3) despite a carbohydrate content exceeding 40g, iAUC120 was below 18 mmol/L*min, suggesting a discordantly low glycemic response.6 Therefore, in alignment with our aim to characterize PPGRs to meal timing and staple foods, these data were excluded to ensure the analysis focused on participants with a measurable postprandial glucose signal. 4) Missing sequences of fewer than five consecutive points were imputed using the average of adjacent time points, while longer sequences (exceeding five points) were excluded due to their potential to substantially distort the data.6

Statistical Analysis

Continuous data are presented as mean ± SD for normally distributed variables and as median (interquartile range) for non-normally distributed variables, and differences between the two groups were assessed using Student’s t-test or Mann–Whitney U-test. Qualitative data were presented as frequency (percentage), and the differences between groups were compared by Chi-square test.

In this study, PPGR was characterized using the following metrics: iAUC120, relative peak glucose120 (peak glucose-baseline glucose), glucose fluctuation amplitude120 (peak glucose-minimum glucose), time to peak120, coefficient of variation120 (CV), iAUC180, relative peak glucose180, glucose fluctuation amplitude180, time to peak180, and CV180. The iAUC120 was selected as the primary outcome, as it is widely regarded as the optimal indicator for assessing PPGR.8 The baseline glucose was defined as the median glucose during the 30-minute period preceding each meal.6,7 This value served as the reference for calculating the iAUC.

In the analysis of meal timing and staple food type effects, linear mixed-effects models with subject-specific random intercepts were employed to assess the associations between exposures and PPGR metrics. The exposures included mealtime (breakfast, lunch, and dinner) and different carbohydrate-based staple food types (glucose, steamed bread, rice, noodles, and oats) consumed during lunch. Covariates adjusted for in the models consisted of age, sex, BMI, HbA1c, and baseline blood glucose. Post-hoc pairwise comparisons were conducted using the Tukey method to account for multiple testing.

Additionally, to preliminarily explore the factors influencing inter-individual heterogeneity in PPGR, we calculated the iAUC120 difference (∆) between each of the four daily staple foods (steamed bread, rice, noodles, and oats) and glucose for each participant. For instance, ∆iAUC120 (steamed bread) = iAUC120 (steamed bread) − iAUC120 (glucose).

Based on these differences, four indices were derived: 1) meal timing variability index: the standard deviation (SD) of the three iAUC120 values of glucose at breakfast, lunch, and dinner for each participant; 2) refined staple food sensitivity index: the mean of ∆iAUC120 values for the three refined staple foods (steamed bread, rice, and noodles) for each participant; 3) oat sensitivity index: defined directly as ∆iAUC120 (oat); 4) staple food type variability index: the SD of the four ∆iAUC120 values for each participant. To examine the associations between these indices and target variables, we handled missing data using multiple imputation (MI). Linear regression models were fitted on each imputed dataset, and results were pooled using Rubin’s rules to obtain final effect estimates, 95% confidence intervals (CIs), and p-values.

Statistical significance was defined as a p-value <0.05, and all analyses were carried out using R Studio 2025.05.1+513 (R version 4.5.1). In the R-based analysis, the “pracma” (version 2.4.4) package was used to calculate iAUC; the “lme4” (version 1.1.37) and “lmerTest” (version 3.1.3) packages were applied to fit linear mixed-effects models. The “mice” (version 3.18.0) package was employed to generate 10 imputed datasets with 200 iterations per imputation. To account for the clustering effect inherent in the repeated measures design (eg, multiple observations per subject), the cluster argument was specified with the subject identifier (“id”) to preserve the hierarchical structure of the data during imputation. The convergence of the imputation process was visually assessed using trace plots, which indicated basically acceptable We conducted sensitivity analyses using an alternative imputation method-random forest imputation-to evaluate the robustness of our study findings. The detailed R package information can be found in Appendix Method 2.

Results

The participant enrollment and exclusion process were summarized in Appendix Figure 1. Initially, 35 participants were enrolled; 2 subsequently withdrew due to schedule conflicts, resulting in 33 participants included in the final analysis.

Overall, participants deemed the framework simple, well tolerated, and unanimously expressed willingness to return for further outpatient PPGR tests with different meals. For data cleaning, a total of 231 meal records (7 standard meals × 33 participants) were collected. Of these, 200 records (86.58%) were retained and 31 (13.42%) were excluded for the following reasons: 1) non-adherence to the study protocol (n = 12); 2) baseline glucose higher than PBG (n = 4); 3) missing data (n = 9); 4) iAUC120 below 18 mmol/L*min (n = 6).

Baseline Characteristics

Baseline characteristics of the participants were presented in Table 1. The mean age of the participants was 43.21 ± 10.47 years, and 16 (48.48%) were female. There were 28 (84.85%) participants with T2DM and 5(15.15%) with IGT. The median disease duration was 3.00 (1.00, 5.00) years, and the median HbA1c was 6.20 (5.57, 6.78) %. The mean BMI was 27.21 ± 4.42 kg/m2, with 14 participants (42.42%) classified as obesity. A total of 26 participants (78.79%) received glucose-lowering medications, while the remaining 7 were managed with lifestyle intervention alone.

Table 1 The Characteristics of Participants at Baseline

Compared with female participants, males had a significantly higher proportion of smokers (47.06% vs 6.25%, p = 0.017), a higher prevalence of habitual refined grain consumption (94.12% vs 62.50%, p = 0.039), a greater rate of dyslipidemia (88.24% vs 50.00%, p = 0.026), and higher mean serum Cr levels (79.80 ± 16.56 vs 56.77 ± 9.23 μmol/L, p < 0.001). No significant differences were observed between genders in age, disease duration, HbA1c, or use of glucose-lowering medications (all p > 0.05).

Effects of Meal Timing on Glycemic Responses

Patients with missing glycemic data from two or more time points (breakfast, lunch, or dinner) were excluded from this analysis. Consequently, a total of 33 participants were included in the time-effect analysis.

As shown in Figure 1, we described the glucose response curves over the 120-minute and 180-minute periods, as well as the distribution of iAUC120 and iAUC180 following glucose consumption at breakfast, lunch, and dinner.

Figure 1 The glucose response curves during 120min (a) and 180min (b) and distribution of iAUC120(c) and iAUC180(d) following glucose consumption at breakfast, lunch, and dinner.

Based on a linear mixed-effects model with breakfast as the reference group, we assessed the effects of lunch and dinner timing on various PPGR indices over 120 and 180 minutes. The results are summarized in Table 2. For the 120-minute period, the overall effect of meal timing on iAUC120 was not statistically significant (P = 0.110). Although dinner time was associated with a significant reduction in iAUC120 compared to breakfast (β = −59.49, 95% CI: −114.68 to −4.30; unadjusted p = 0.039), this difference was not significant after adjustment for multiple comparisons (adjusted p = 0.096). No significant difference in iAUC120 was observed between lunch and breakfast (β = −40.20, 95% CI: −96.76 to 16.36; adjusted p = 0.352). Similarly, meal timing showed no significant overall effects on relative peak glucose, glucose fluctuation amplitude, time to peak, or CV (all P for time effect >0.05).

Table 2 Effects of Meal Timing on Postprandial Glycemic Response to Glucose at Breakfast, Lunch, and Dinner

For the 180-minute period, the overall effect of meal timing on iAUC180 was not statistically significant (P for time effect = 0.097). Dinner was associated with a significantly lower iAUC180 compared to breakfast (β = −76.15, 95% CI: −144.04 to −8.25; unadjusted p = 0.032), and this difference was not significant after adjustment (adjusted p = 0.080). No significant difference was found between lunch and breakfast (β = −46.05, 95% CI: –115.73 to 23.62; adjusted p = 0.405). Meal timing did not significantly affect other PPGR metrics, including relative peak glucose, glucose fluctuation amplitude, time to peak, or CV (all P for time effect >0.05).

Effects of Staple Food Types on Glycemic Responses

Participants with missing data for three or more of the five staple food types (glucose, steamed bread, rice, noodles, and oats) were excluded from the analysis. Consequently, a total of 30 participants were included.

As illustrated in Figure 2, we present the glucose response curves during 120-minute and 180-minute period and the distribution characteristics of iAUC120 and iAUC180 after consumption of glucose, steamed bread, rice, noodles, and oats at lunchtime.

Figure 2 The glucose response curves during 120min (a) and 180min (b) and basic distribution of iAUC120(c) and iAUC180 (d) following glucose, steamed bread, rice, noodles, and oats at lunch.

As summarized in Table 3, using a linear mixed-effects model with glucose as the reference group, we analyzed the effects of four staple food types on various PPGR metrics over 120 and 180 minutes. For the 120-minute period, the overall effect of staple food types on iAUC120 was highly significant (p < 0.001). Specifically, iAUC120 values following consumption of steamed bread, rice, noodles, and oats were all significantly lower than glucose. Noodles demonstrated the strongest reducing effect (β = −179.53, 95% CI: −228.84 to −130.22, adjusted p < 0.001), followed by steamed bread (β = −142.99, 95% CI: −195.33 to −90.65, adjusted p < 0.001), oats (β = −129.88, 95% CI: −181.07 to −78.70, adjusted p < 0.001), and rice (β = −100.58, 95% CI: −152.04 to −49.13, adjusted p = 0.002). Among the staple foods, noodles resulted in a significantly lower iAUC120 than rice (adjusted p = 0.025), while no other significant differences were observed between the remaining staples. Regarding relative peak glucose and glucose fluctuation amplitude, all staple foods led to significantly lower values compared to glucose (all P for staple food type effect p < 0.001), with no significant differences among the four staple types. The overall effect of staple food type was not significant for time to peak or CV (all P for staple food type effect > 0.05).

Table 3 Effects of Staple Food Types on Postprandial Glycemic Response to Glucose, Steamed Bread, Rice, Noodles, and Oats at Lunch

For the 180-minute period, the overall effect of staple food types on iAUC180 remained highly significant (p < 0.001). Specifically, iAUC180 values following steamed bread, noodles, and oats were significantly lower than glucose. Noodles again showed the strongest reducing effect (β = −181.98, 95% CI: −250.74 to −113.21, adjusted p < 0.001). Although the iAUC180 for rice was lower than glucose before adjustment (unadjusted p = 0.022), the difference was not significant after multiple comparison correction (adjusted p = 0.143). No other significant differences in iAUC180 were observed among the staple foods. For relative peak glucose and glucose fluctuation amplitude, all staples resulted in significantly lower values than glucose (all P for staple food type effect p < 0.001), with no significant differences among themselves. The overall effect of staple food type was not significant for time to peak or CV (all P for staple food type effect > 0.05).

Associations of Candidate Variables with Meal Timing Variability Index, Refined Staple Food Sensitivity Index, Oat Sensitivity Index, and Staple Food Type Variability Index

In both unadjusted and adjusted models, HOMA-IR was positively associated with meal timing variability index (unadjusted: β = 6.92, 95% CI: 2.21 to 11.63, P = 0.006; adjusted: β = 9.10, 95% CI: 3.01 to 15.20, P = 0.005) in Appendix Table 1. Additionally, compared with participants born in southern China, those born in northern China had a significantly higher refined staple food sensitivity index (unadjusted: β = 268.18, 95% CI: 78.81 to 457.54, P = 0.007; adjusted: β = 267.14, 95% CI: 59.18 to 475.09, P = 0.014) in Appendix Table 2. In the sensitivity analysis, the aforementioned findings remained robust in Appendix Table 5 and Appendix Table 6; however, the association between sleep, dietary regularity, and meal timing variability index was only observed in the sensitivity analysis in Appendix Table 5. The remaining candidate variables-including age, sex, BMI, HbA1c, medication use, smoking-exhibited no statistically significant associations with the meal timing variability index, refined staple food sensitivity index, oat sensitivity index, or staple food type variability index in either model (all P-values > 0.05) (Appendix Tables 3–8).

Discussion

This prospective observational study aimed to achieve two primary objectives. Firstly, to validate the practical utility of the framework for conveniently assessing PPGRs by leveraging CGM under real-life conditions, with a focus on patient acceptability and feasibility. Secondly, to utilize this framework to investigate the impact of meal timing and staple food types on PPGR in individuals with IGT and T2DM, thereby addressing a notable gap in real-world evidence. The main findings are as follows: 1) Regarding time effects: at the group level, consuming glucose at different meal times did not significantly affect PPGR indicators at 120 or 180 minutes; at the individual level, HOMA-IR, sleep duration, and dietary regularity appeared to be positively associated with iAUC120 variability across different meal timings. 2) Regarding staple food type effects: at the group level, consuming different types of staple foods (glucose, steamed bread, rice, noodles, and oats) standardized to 50g carbohydrates at lunch primarily influenced the 120 and 180-minute iAUC, relative peak glucose, and glucose fluctuation amplitude; specific differences will be elaborated on in detail in subsequent sections; no significant effects were observed on time to peak or CV; at the individual level, being born in northern China, compared to southern China, was associated with higher PPGR to refined staple food.

Chrononutrition has emerged as a significant field within metabolic research.22 A growing body of evidence indicates that PPGR induced by nighttime eating is generally higher than that following morning intake.9,19–22 For example, Leung et al demonstrated that healthy populations who consumed 75g glucose at 8:00 PM exhibited a significantly higher iAUC120 compared to the same dose ingested at 8:00 AM.20 This conclusion was further supported by a meta-analysis incorporating 8 randomized clinical trials involving 116 healthy non-shift workers, which found that the iAUC after consuming isocaloric, standardized carbohydrate meals in the evening was significantly greater than in the morning (standardized mean difference = 1.30, 95% CI: 1.01–1.59, I2 = 0%, p < 0.00001).21 These diurnal variations in glycemic response align with inherent rhythmic fluctuations in glucose metabolism observed in healthy populations: glucose tolerance typically peaks during the daytime light phase—usual eating periods—and declines during the nighttime fasting phase. This rhythm is likely influenced by circadian-regulated changes in insulin secretion and sensitivity, as well as periodic variations in hormones such as cortisol.22

However, current research on the impact of meal timing (not just nighttime) on PPGR, particularly in populations with IGT or T2DM, remains limited. In one study involving individuals with T2DM not using insulin, participants consumed a standardized mixed meal at 8:00AM, 12:00, and 4:00PM. The results showed that iAUC following the morning meal was significantly higher than that after the noon and afternoon meals, and the iAUC at noon was significantly lower than in the afternoon. The investigators hypothesized that this pattern may be attributed to disruptions in the circadian rhythm of insulin sensitivity in T2DM patients.25 Our study accounted for individual variations in chronotype and underlying metabolic characteristics,9,22 with participants encouraged to adhere to their habitual meal timings. Following adjustment for subject-specific random intercepts via a linear mixed-effects model, the results revealed that meal timing did not exert a significant overall effect on PPGR. This finding diverged from observations under strictly controlled laboratory conditions.25 We attribute this discrepancy primarily to fundamental contextual and methodological distinctions: unlike controlled protocols that standardize meal schedules, our real-world design allowed natural behavioral variability (eg sleep-wake cycles, stress levels), which likely introduced compensatory mechanisms and masked isolated circadian effects. Moreover, our cohort of younger outpatients following personalized meal timings across multiple days differed demographically and procedurally from previous studies using older participants under fixed, single-day conditions.25 Rather than a mere limitation, this finding underscores the critical role of context in translational glucose metabolism research and highlights the need for future studies that systematically compare tightly controlled experimental paradigms with free-living settings to clarify whether—and through what mechanisms—real-world factors modulate circadian meal-timing effects across diverse populations.

Furthermore, at the individual level, we investigated factors influencing the variability of iAUC120 in response to meal timing. Results showed that participants with higher HOMA-IR exhibited less stable iAUC120 to meal timing. Previous studies have reported that in free-living conditions, HOMA-IR is negatively correlated with parameters of circadian rhythm stability in individuals with obesity or prediabetes.26 This suggests that individuals with insulin resistance may exhibit greater daily fluctuations in glucose metabolism, while the insufficient compensation of pancreatic β-cell function in patients with DM may further amplify the impact of mealtime on glycemic responses. Additionally, only sensitivity analysis revealed that both longer sleep duration and greater dietary regularity were positively associated with mealtime response variability. It remains unclear whether these associations represent chance findings or reflect a true biological relationship whereby adequate sleep and regular meal patterns enhance endogenous circadian rhythms, thereby exaggerating metabolic differences across meal times. This question merits further investigation in future studies.

As a monosaccharide, glucose is directly absorbable by small intestinal epithelial cells without prior digestion, resulting in a particularly prominent PPGR. A study involving healthy females demonstrated that when subjects consumed either glucose or rice-both containing 50g carbohydrates-the glucose group exhibited significantly higher iAUC, relative peak glucose, and SD compared to the rice group. This trend was consistent across both 120-minute and 180-minute periods.27 In our findings of individuals with IGT and T2DM, glucose induced significantly higher iAUC120, relative peak glucose (at both 120 and 180 minutes), and glucose fluctuation amplitude (at both 120 and 180 minutes) compared to rice. The only exception was iAUC180, where no significant difference was observed. When further compared to other staple foods-including steamed bread, noodles, and oats-glucose consistently showed significantly higher values in iAUC, relative peak glucose, and glucose fluctuation amplitude at both 120 and 180 minutes. Notably, no statistically significant differences were detected between glucose and any of these staples in terms of time to peak or CV. These observations suggest that regardless of an individual’s glycemic metabolic status-whether normal glucose tolerance, IGT, or T2DM-glucose may trigger the strongest postprandial glycemic impact among common carbohydrate sources. Over the 180-minute observation window, the differences in iAUC between various carbohydrates appeared to be partially attenuated in diabetic patients. This phenomenon may be attributed to impaired insulin secretion and insulin resistance, which are characteristics of T2DM. Additionally, in populations with IGT and T2DM, the type of staple food seemed to exert a more substantial influence on glycemic load-related metrics (such as iAUC, relative peak glucose, and glucose fluctuation amplitude) than on time to peak or CV.

A clinical study simulating real-world dietary conditions in individuals with type 1 DM compared PPGRs to three meals of 42g carbohydrate: high-protein pasta, regular pasta, and rice. Both high-protein and regular pasta resulted in significantly lower peak blood glucose and 5-hour total glucose area under the curve (AUC) compared to rice.10 The authors proposed that the mechanisms may include pasta’s compact texture and larger particle size after mastication, which delays gastric emptying, as well as its protein matrix encapsulating starch granules, thereby slowing carbohydrate digestion.28,29 In contrast, traditional Chinese noodles, due to their higher water content and looser structure during processing, allow starch to be more readily digested, potentially leading to a higher PPGR. In our study, although the iAUC for noodles was significantly lower than that for rice at 120 minutes, this difference was no longer statistically significant at 180 minutes. Furthermore, within wheat-based staples, the impact of fermented (eg, steamed bread) versus non-fermented (eg, noodles) products on PPGR is of interest. One study of overweight adults showed no significant difference in the 180-minute total glucose AUC or gut peptide responses (including C-peptide, glucagon, PYY, GIP, and GLP-1) after consuming 75g of either noodles or steamed bread.30 Extending these findings to populations with IGT and T2DM, our study demonstrated no statistically significant differences in any PPGR metrics between noodles and steamed bread, regardless of whether the observation period was 120 or 180 minutes. Oat products are stratified by composition and morphology into categories including whole oat kernels, thick rolled oats, oat flour, and other derivatives like oat bran. Such diversity in product forms is associated with varying processing techniques, and the intensity of processing can modify nutritional profiles and physical properties, ultimately exerting a significant influence on PPGRs.16 A meta-analysis concluded that, in contrast to refined grains, intake of whole oat kernels and thick flakes is associated with a significant reduction in PPGR, whereas thin or instant flakes demonstrate no significant beneficial effects on glycemic or insulinemic responses.31 In this study, we used instant oats for practicality. This choice likely explains why no significant differences were observed in PPGRs between oats and the other three staples (steamed bread, rice, noodles) at 120 or 180 minutes.

At the individual level, our analysis revealed that, although all participants currently resided in Northern China, those born in the North exhibited a higher PPGRs to refined staple food compared to Southern-born counterparts. This observation was supported by a nationally representative cross-sectional study (2015–2017) across 31 provinces in mainland China, which highlighted significant geographical variation in diabetes mellitus prevalence, with the highest rates observed in the North.32 The observed link between birthplace (North/South China) and PPGRs requires further mechanistic exploration. We propose two complementary hypotheses: first, that early‑life exposure to region‑specific staple diets (eg, wheat in the North, rice in the South) may program lasting metabolic adaptations; second, that regionally shaped gut microbiota could mediate this effect via metabolic interactions. Future studies should adopt prospective, trans‑regional cohort designs. Tracking migrants (South‑to‑North or vice‑versa) alongside non‑migrants, with serial PPGR, gut microbiome, and metabolome, would help distinguish the effects of genetic, early‑life environmental, and current lifestyle factors. In the present study, no factors associated with either oat sensitivity index or staple food type variability index were identified. We hypothesize that these two indices may be more closely linked to intrinsic gastrointestinal function and gut microbiota composition than to conventional biochemical markers or overt lifestyle factors.12,13 However, this hypothesis awaits validation through subsequent research.

The primary aim of our study was to develop and implement a practical outpatient framework for PPGR assessment by leveraging CGM in real-world settings. The novelty of our work lies in its pragmatic, integrated methodological approach within a real-world clinical context, rather than in elucidating new biological mechanisms. We have framed it as an important step towards translatable precision nutrition. Applying this framework, we found that despite modest differences at the group level, individual responses nevertheless varied considerably, indicating that it is necessary to focus not only on the general characteristics of PPGRs observed at the group level but also to analyze the heterogeneity of PPGRs exhibited at the individual level. This discovery underscores the critical importance of adopting personalized strategies when translating generalized nutritional principles into clinical practice. The research perspective should, therefore, shift from “when is the best time to eat” to “when is the best time to eat for a specific individual,” and from “which type of staple food is better” to “which type of staple food is better for a specific individual.” Future research should focus on identifying the characteristics of individuals who are more responsive to variations in meal timing and staple food type, thereby facilitating tailored dietary guidance that aligns with the patient’s own circadian biology for diabetes management.

This study has several strengths: 1) we developed a streamlined and real-world framework and examined the effects of meal timing and staple food type on PPGRs at both group and individual levels, extended the observation period (120 and 180 minutes), and incorporated multiple PPGR metrics for a comprehensive assessment. 2) the study was conducted under real-life conditions where participants consumed standardized meals in their free-living environment, enhancing the external validity of the findings. 3) rigorous quality control was implemented, including a 30-minute pre-meal stabilization period to confirm baseline glucose stability, thereby strengthening data reliability.

Several limitations should also be noted: 1) given the critical reliance of study outcomes on patient adherence, to enhance the clinical applicability of the framework and ensure reliable data, we propose the following strategies: firstly, we recommend the combined use of validated standardized questionnaires and low-burden digital tools (eg, daily short message service or in-app reminders) to systematically document meal timing and type. Secondly, a clear adherence threshold (eg, ≥80% of prescribed meals) should be established. For patients whose adherence falls below this threshold, a structured clinical review should be initiated to identify and address specific barriers (eg, meal preparation difficulties or misunderstanding of instructions), rather than directly utilizing their data for decision-making. 2) we acknowledge that applying a pre‑specified iAUC120 threshold to exclude low responders may introduce selection bias. While this criterion, drawn from prior literature,6 aimed to ensure a measurable postprandial signal, it could systematically omit individuals with low glycemic responses. This limitation affects result interpretation by narrowing the characterized response spectrum-especially for hypo‑responders-and potentially overestimating mean group responses, thus warranting caution when generalizing findings to broader dysglycemic populations. 3) the generalizability of our findings is primarily limited to our cohort of relatively young, motivated outpatients with dysglycemia, and should be extrapolated with caution to older, more comorbid, or inpatient populations. Nevertheless, the simplified model we developed provides a pragmatic tool for capturing real-world glycemic response patterns. The critical next step is therefore to validate this model in larger, more diverse cohorts through prospective, multicenter, long-term studies, which would directly assess the robustness and broader applicability of both the findings and the framework. 4) mechanistic insights were constrained by the lack of insulin and incretin response data, which were crucial for understanding underlying physiological pathways. 5) potential confounding factors include carryover effects due to short inter-meal intervals (≤2 days), possible breakfast interference on lunch tests, and the influence of minimal oil-free, sugar-free vegetables accompanying meals—though their protein/fiber content was low and likely had negligible impact on the core findings. In addition, although we standardized the preparation process, inherent inconsistencies in ingredient preparation or timing variations among participants during home cooking inevitably introduced residual variability that could not be fully eliminated—reflecting, nonetheless, the pragmatic nature of our real-world study design. 6) although the premise that individual PPGR to the same meal was reproducible was supported by existing literature,6,8,11,13 the single measurement of each meal/staple type necessitates further replication over longer periods to minimize the impact of intra-individual variability.

Conclusion

In conclusion, our study provides a novel methodological framework and contextual evidence for personalizing dietary advice in dysglycemic outpatients, which can serve as a foundation for future mechanistic investigations. Future research should expand sample size and diversity, while incorporating biomarkers such as gut microbiota and gut hormones, to validate the current conclusions and further elucidate the underlying mechanisms.

Abbreviations

IGT, impaired glucose tolerance; T2DM, type 2 diabetes mellitus; CVD, cardiovascular disease; PPGR, postprandial glycemic response; PBG, postprandial blood glucose; CGM, Continuous glucose monitoring; DM, diabetes mellitus; GI, glycemic index; BMI, body mass index; HbA1c, glycated hemoglobin; DPP-4i, Dipeptidyl peptidase-4 inhibitors; SGLT2i, sodium-glucose cotransporter 2 inhibitors; GLP-1RA, glucagon-like peptide-1 receptor agonists; TG, triglycerides; TC, total cholesterol; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; Cr, creatinine; UA, uric acid; ACR, albumin-to-creatinine ratio; HOMA, homeostasis model assessment; HOMA-IR, HOMA of insulin resistance; HOMA-β, HOMA of beta-cell function; FBG, fasting blood glucose; FINS, fasting insulin; iAUC, incremental area under the curve; CV, coefficient of variation; MI, multiple imputation; CIs, confidence intervals; SD, standard deviation.

Data Sharing Statement

The datasets used and analyzed during the current study are available from the corresponding author Tao Yuan on reasonable request.

Acknowledgments

The authors were deeply grateful to the study participants for their valuable contribution.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

This study was supported by “13th Five-Year” National Science and Technology Major Project for New Drugs under Grant No. 2019ZX09734001 (to Weigang Zhao).

Disclosure

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References

1. Sun H, Saeedi P, Karuranga S, et al. IDF diabetes atlas: global, regional and country-level diabetes prevalence estimates for 2021 and projections for 2045. Diabet Res Clin Pract. 2022;183:109119. doi:10.1016/j.diabres.2021.109119

2. Tabák AG, Herder C, Rathmann W, Brunner EJ, Kivimäki M. Prediabetes: a high-risk state for diabetes development. Lancet. 2012;379(9833):2279–17. doi:10.1016/S0140-6736(12)60283-9

3. Ketema EB, Kibret KT. Correlation of fasting and postprandial plasma glucose with HbA1c in assessing glycemic control; systematic review and meta-analysis. Arch Public Health. 2015;73(1):43. doi:10.1186/s13690-015-0088-6

4. Cavalot F, Pagliarino A, Valle M, et al. Postprandial blood glucose predicts cardiovascular events and all-cause mortality in type 2 diabetes in a 14-year follow-up: lessons from the San Luigi Gonzaga diabetes study. Diabetes Care. 2011;34(10):2237–2243. doi:10.2337/dc10-2414

5. Shiraiwa T, Kaneto H, Miyatsuka T, et al. Postprandial hyperglycemia is a better predictor of the progression of diabetic retinopathy than HbA1c in Japanese type 2 diabetic patients. Diabetes Care. 2005;28(11):2806–2807. doi:10.2337/diacare.28.11.2806

6. Mendes-Soares H, Raveh-Sadka T, Azulay S, et al. Model of personalized postprandial glycemic response to food developed for an Israeli cohort predicts responses in Midwestern American individuals. Am J Clin Nutr. 2019;110(1):63–75. doi:10.1093/ajcn/nqz028

7. Xiong X, Xue Y, Cai Y, He J, Su H. Prediction of personalised postprandial glycaemic response in type 1 diabetes mellitus. Front Endocrinol. 2024;15:1423303. doi:10.3389/fendo.2024.1423303

8. Zeevi D, Korem T, Zmora N, et al. Personalized nutrition by prediction of glycemic responses. Cell. 2015;163(5):1079–1094. doi:10.1016/j.cell.2015.11.001

9. Stutz B, Krueger B, Goletzke J, et al. Glycemic response to meals with a high glycemic index differs between morning and evening: a randomized cross-over controlled trial among students with early or late chronotype. Eur J Nutr. 2024;63(5):1593–1604. doi:10.1007/s00394-024-03372-4

10. Zavitsanou S, Massa J, Deshpande S, et al. The effect of two types of pasta versus white rice on postprandial blood glucose levels in adults with type 1 diabetes: a randomized crossover trial. Diabetes Technol Ther. 2019;21(9):485–492. doi:10.1089/dia.2019.0109

11. Berry SE, Valdes AM, Drew DA, et al. Human postprandial responses to food and potential for precision nutrition. Nat Med. 2020;26(6):964–973. doi:10.1038/s41591-020-0934-0

12. Tily H, Patridge E, Cai Y, et al. Gut microbiome activity contributes to prediction of individual variation in glycemic response in adults. Diabetes Ther. 2022;13(1):89–111. doi:10.1007/s13300-021-01174-z

13. Shilo S, Godneva A, Rachmiel M, et al. Prediction of personal glycemic responses to food for individuals with type 1 diabetes through integration of clinical and microbial data. Diabetes Care. 2022;45(3):502–511. doi:10.2337/dc21-1048

14. Bell KJ, King BR, Shafat A, Smart CE. The relationship between carbohydrate and the mealtime insulin dose in type 1 diabetes. J Diabetes Compl. 2015;29(8):1323–1329. doi:10.1016/j.jdiacomp.2015.08.014

15. Hu EA, Pan A, Malik V, Sun Q. White rice consumption and risk of type 2 diabetes: meta-analysis and systematic review. BMJ. 2012;344(mar15 3):e1454. doi:10.1136/bmj.e1454

16. Zhang K, Dong R, Hu X, Ren C, Li Y. Oat-based foods: chemical constituents, glycemic index, and the effect of processing. Foods. 2021;10(6):1304. doi:10.3390/foods10061304

17. Atkinson FS, Foster-Powell K, Brand-Miller JC. International tables of glycemic index and glycemic load values: 2008. Diabetes Care. 2008;31(12):2281–2283. doi:10.2337/dc08-1239

18. Atkinson FS, Brand-Miller JC, Foster-Powell K, Buyken AE, Goletzke J. International tables of glycemic index and glycemic load values 2021: a systematic review. Am J Clin Nutr. 2021;114(5):1625–1632. doi:10.1093/ajcn/nqab233

19. Leung GKW, Huggins CE, Ware RS, Bonham MP. Time of day difference in postprandial glucose and insulin responses: systematic review and meta-analysis of acute postprandial studies. Chronobiol Int. 2020;37(3):311–326. doi:10.1080/07420528.2019.1683856

20. Leung GKW, Huggins CE, Bonham MP. Effect of meal timing on postprandial glucose responses to a low glycemic index meal: a crossover trial in healthy volunteers. Clin Nutr. 2019;38(1):465–471. doi:10.1016/j.clnu.2017.11.010

21. de Almeida RS, Marot LP, Latorraca C de CDOC, Oliveira R de Á, Crispim CA. Is evening carbohydrate intake in healthy individuals associated with higher postprandial glycemia and insulinemia when compared to morning intake? A systematic review and meta-analysis of randomized crossover studies. J Am Nutr Assoc. 2023;42(4):349–360. doi:10.1080/07315724.2022.2043199

22. Henry CJ, Kaur B, Quek RYC. Chrononutrition in the management of diabetes. Nutr Diabetes. 2020;10(1):6. doi:10.1038/s41387-020-0109-6

23. World Medical Association. World medical association declaration of Helsinki: ethical principles for medical research involving human participants. JAMA. 2025;333(1):71–74. doi:10.1001/jama.2024.21972.

24. Chen C, Lu FC. The guidelines for prevention and control of overweight and obesity in Chinese adults. Biomed Environ Sci. 2004;17 Suppl:1–36.

25. Peter R, Dunseath G, Luzio SD, Chudleigh R, Roy Choudhury S, Owens DR. Daytime variability of postprandial glucose tolerance and pancreatic B-cell function using 12-h profiles in persons with Type 2 diabetes. Diabet Med. 2010;27(3):266–273. doi:10.1111/j.1464-5491.2010.02949.x

26. Drummen M, Tischmann L, Gatta-Cherifi B, Raben A, Adam T, Westerterp-Plantenga MS. Reproducibility and associations with obesity and insulin resistance of circadian-rhythm parameters in free-living vs. controlled conditions during the PREVIEW lifestyle study. Int J Obes Lond. 2021;45(9):2038–2047. doi:10.1038/s41366-021-00873-8

27. Liu A-S, Fan Z-H, Lu X-J, et al. The characteristics of postprandial glycemic response patterns to white rice and glucose in healthy adults: identifying subgroups by clustering analysis. Front Nutr. 2022;9:977278. doi:10.3389/fnut.2022.977278

28. Granfeldt Y, Björck I. Glycemic response to starch in pasta: a study of mechanisms of limited enzyme availability. J Cereal Sci. 1991;14(1):47–61. doi:10.1016/S0733-5210(09)80017-9

29. Kim EH-J, Petrie JR, Motoi L, et al. Effect of structural and physicochemical characteristics of the protein matrix in pasta on in vitro starch digestibility. Food Biophys. 2008;3(2):229–234. doi:10.1007/s11483-008-9066-7

30. Shah S, Brown PDS, Mayengbam S, et al. Metabolic and gut microbiota responses to sourdough pasta consumption in overweight and obese adults. Front Nutr. 2020;7:615003. doi:10.3389/fnut.2020.615003

31. Musa-Veloso K, Noori D, Venditti C, et al. A systematic review and meta-analysis of randomized controlled trials on the effects of oats and oat processing on postprandial blood glucose and insulin responses. J Nutr. 2021;151(2):341–351. doi:10.1080/07420528.2019.1683856

32. Li Y, Teng D, Shi X, et al. Prevalence of diabetes recorded in mainland China using 2018 diagnostic criteria from the American diabetes association: national cross sectional study. BMJ. 2020;369:m997. doi:10.1136/bmj.m997