Time in range during caloric restriction in type 2 diabetes with obesity

ABSTRACT

Introduction

To evaluate glycemic changes during caloric restriction with continuous glucose monitoring (CGM)‐derived time in range (TIR) in individuals with type 2 diabetes and obesity.

Materials and Methods

This 12‐week single‐arm intervention consisted of 6 weeks of home‐delivered meals (800–1,200 kcal/day), followed by 6 weeks of self‐managed diet (1,500–1,800 kcal/day for men; 1,200–1,500 kcal/day for women). CGM (14‐day sensor) was performed at baseline, weeks 5–6, and weeks 11–12. A 75‐g oral glucose tolerance test was conducted at baseline, week 6, and week 12 to calculate the C‐peptide index (CPI) and Matsuda index.

Results

Participants had a median age of 46.0 years [38.0, 53.5], body mass index (BMI) of 29.2 kg/m² [26.8, 31.2], HbA1c 6.6% [6.0, 7.13], and diabetes duration 2.01 years [0.91, 3.65]. Over 12 weeks, TIR improved from 84.3% to 90.3% (P = 0.041), and BMI decreased from 29.3 to 26.7 kg/m² (P < 0.001). Weight reduction was associated with improved insulin sensitivity, whereas changes in CPI were not significant. CPI showed a stronger association with TIR than the Matsuda index, underscoring the importance of insulin secretion capacity in glycemic control. The association between CPI and TIR was more pronounced participants with higher insulin sensitivity (P = 0.011), suggesting that adequate peripheral sensitivity is required to influence glycemic outcomes.

Conclusions

In individuals with type 2 diabetes and obesity, caloric restriction was associated with improved glycemic profiles and reduced body weight. Enhanced insulin sensitivity appears to be the predominant contributor to improved TIR, while preserved β‐cell function remains essential for achieving optimal glycemic outcomes.

INTRODUCTION

Obesity has surged in recent years, largely driven by the increased consumption of ultra‐processed and high‐calorie foods, and has emerged as a major public health concern ¹ . In parallel, the prevalence of type 2 diabetes continues to rise steadily ² . A fundamental strategy for improving glycemic control and alleviating stress on pancreatic β‐cells is dietary calorie restriction. Accordingly, lifestyle modification—particularly dietary intervention—is the cornerstone of type 2 diabetes management ³ . Medical nutrition therapy has been shown to be as effective as pharmacologic treatment in improving glycemic outcomes, and it further enhances quality of life by promoting weight loss in individuals with diabetes ⁴ .

Previous studies have demonstrated that substantial weight loss induced by a very low‐calorie diet can lead to remission of type 2 diabetes, defined as HbA1c <6.5% for at least 3 months after withdrawal of therapy ⁵ . In an early study, patients on a 600 kcal/day diet showed rapid improvements in fasting glucose and insulin response ⁶ . In a subsequent 8‐week intervention of 30 patients, about 40% achieved remission, and this trial established the role of diabetes duration in reversibility, informing the 6‐year cut‐off used later ⁷ . More recently, a large UK primary care trial in individuals with obesity and newly diagnosed type 2 diabetes (<6 years) reported remission rates up to 86% among those losing ≥15 kg ⁸ . Collectively, these findings highlight the potential of intensive dietary interventions to induce diabetes remission through substantial weight loss.

Glycated hemoglobin (HbA1c), the most commonly used glycemic index, reflects average plasma glucose levels over the previous 2–3 months. However, it has limitations in capturing glycemic responses to short‐term dietary changes. Therefore, we incorporated continuous glucose monitoring (CGM) during the study period involving a low‐calorie diet intervention aimed at inducing diabetes remission ⁹ . Among CGM‐derived metrics, time in range (TIR)—the percentage of time that blood glucose levels remain within the target range of 70–180 mg/dL—has emerged as a clinically relevant indicator of glycemic control. TIR offers a more precise evaluation of glycemic control by reflecting day‐to‐day glycemic fluctuations and hypoglycemia risk ¹⁰ .

This study sought to characterize the trajectories of TIR and corresponding changes in insulin secretion and sensitivity during a calorie‐restricted dietary intervention in patients with newly diagnosed type 2 diabetes and obesity.

MATERIALS AND METHODS

Study participants

Adults aged 18–60 years with a diagnosis of type 2 diabetes within the past 6 years, a body mass index (BMI) ≥25 kg/m², and no current insulin use were eligible for participation. Exclusion criteria included a history of cardiovascular disease or cancer, and hospitalization for acute illness within the past 3 months. All participants had normal renal, hepatic, thyroid, and adrenal function confirmed during the screening period. The primary outcome was CGM‐derived TIR (glucose range 70–180 mg/dL) measured at the end of the 12‐week intervention. Secondary outcomes included other CGM‐derived metrics, such as mean glucose and glycemic variability, and their associations with oral glucose tolerance test (OGTT)‐derived indices of insulin secretion and insulin sensitivity. Twenty‐seven participants were enrolled; three discontinued the study before week 6, and the remaining 24 completed the intervention and were included in the final analysis. Baseline characteristics are available in Table S1. Outcomes related to body weight and remission—defined as HbA1c <6.5%—have been reported previously ¹¹ .

Intervention

This study investigated glycemic improvement following a 12‐week nutritional intervention alone, implemented after the discontinuation of all antidiabetic medications. The intervention consisted of two sequential phases. During the first 6 weeks, participants received home‐delivered, calorie‐restricted, nutritionally balanced meals, starting at 800 kcal/day and gradually increasing to 1,200 kcal/day. The meals, developed in collaboration with registered dietitians to align with dietary guidelines for individuals with type 2 diabetes, consisted of protein shakes, fresh salads, and steamed rice‐based bowls (Table S2). Participants were instructed to consume only the provided meals, and adherence was monitored through daily supervision by the study team. In the following 6‐week phase, participants transitioned to a self‐managed diet guided by nutritional education, with daily calorie goals of 1,500–1800 kcal for men and 1,200–1,500 kcal for women. This phase included a one‐time in‐person nutrition counseling session, supplemented by weekly remote support to encourage adherence and maintain glycemic control ¹² . Dietary intake was monitored by a study dietitian (S.‐J.P.) throughout the study period. Adherence to the prescribed energy target, defined as energy intake within 10% of the target, was observed in 83% of participants during the total food replacement period and in 85.5% during the self‐managed diet period. Detailed information on patient‐reported energy intake is provided in Table S3. To reduce the risk of hypoglycemia during caloric restriction, all glucose‐lowering agents were withdrawn prior to the intervention and withheld throughout the study period. Antidiabetic medications were discontinued at the time of CGM initiation following screening, and the dietary intervention was initiated 1 week later.

Ethical approval

The Institutional Review Board of Yeouido St. Mary's Hospital approved the study protocol (IRB No. SC23ENSE0036), and written informed consent was obtained from all participants. The study was conducted in accordance with the principles of the Declaration of Helsinki. This protocol was registered at ClinicalTrials.gov (NCT05754775).

Measurements and definitions

CGM was performed during weeks 0–1, 5–6, and 11–12 (Figure S1). CGM was conducted using the FreeStyle Libre Flash Glucose Monitoring System (Abbott Diabetes Care, Alameda, CA, USA), an intermittently scanned CGM device that provides up to 14 days of interstitial glucose data without the need for fingerstick calibration. CGM data were obtained over 14 consecutive days at each study visit, and all available data across the entire monitoring period were included in the analyses. Although no predefined exclusion threshold based on CGM data capture was applied, data capture rates (%) were calculated for each study period to allow assessment of data completeness, particularly given the intermittently scanned nature of the CGM system. TIR, time above range (TAR), time below range (TBR), and coefficient of variation (CV) were calculated from CGM data based on international consensus criteria ¹³ . TIR was defined as the percentage of time with glucose levels between 70 and 180 mg/dL, TAR as the percentage of time with glucose levels >180 mg/dL, and TBR as the percentage of time with glucose levels <70 mg/dL. In addition, time in tight range (TITR) was defined as the percentage of time with glucose levels between 70 and 140 mg/dL, reflecting tighter glycemic control.

Routine blood chemistry tests and a 75‐g OGTT were conducted at baseline, week 6, and week 12. The OGTT involved venous blood sampling at 0, 30, 60, 90, and 120 min over a 2‐h period. To ensure consistency and minimize inter‐assay variability, all blood samples were analyzed at an independent central laboratory (SML Medical Foundation, Seoul, Korea). The homeostasis model assessment indices were used to estimate β‐cell function (HOMA‐β) and insulin resistance (HOMA‐IR) based on fasting glucose and insulin concentrations. HOMA‐β was calculated as 360 × fasting insulin (μU/mL) divided by fasting glucose (mg/dL) minus 63. HOMA‐IR was calculated by multiplying fasting glucose (in mg/dL) by fasting insulin (in μU/mL), and then dividing the product by 405.

We evaluated insulin secretory function using the early‐phase insulin response during the 75‐g OGTT, calculated as the C‐peptide index (CPI), defined as the increment in C‐peptide from 0 to 30 min, consistent with previous definitions ¹⁴ :

$$\mathrm{CPI}=\left(\mathrm{C}-{\text{peptide}}_{30}-\mathrm{C}-{\text{peptide}}_0\right)/\left({\text{Glucose}}_{30}–{\text{Glucose}}_0\right)$$

(C‐peptide in ng/mL, glucose in mg/dL).

Insulin sensitivity was assessed using the Matsuda index, a validated OGTT‐derived measure, calculated as ¹⁵ :

$$\text{Matsuda index}=\begin{array}{l}\mathrm{10,000}/\surd [\left(\mathrm{FPG}\times \mathrm{FPI}\right)\\ \times (\text{mean OGTT glucose}\\ \times \text{mean OGTT insulin})]\end{array}$$

(FPG: fasting plasma glucose in mg/dL; FPI: fasting plasma insulin in μU/mL; mean OGTT values based on all available time points).

A modified disposition index was defined as the product of CPI and the Matsuda index (CPI × Matsuda index) to reflect β‐cell function adjusted for insulin sensitivity. This metric captures the compensatory capacity of β‐cells in response to insulin resistance and is consistent with the established hyperbolic relationship between insulin secretion and insulin sensitivity ¹⁶ .

Statistical analysis

For descriptive statistics, continuous variables are presented as median (interquartile range), and categorical variables as frequency (percentage). Longitudinal changes in CGM‐derived outcomes were assessed using linear mixed‐effects models with visit (Baseline, Week 6, and Week 12) included as a fixed effect and individual participants modeled as random intercepts to account for within‐participant correlation arising from repeated measurements. Model‐based estimated means and corresponding 95% confidence intervals were reported for each time point. Associations between clinical variables and CGM‐derived outcomes were evaluated using linear mixed‐effects models to appropriately account for repeated measurements across study visits. Effect modification by insulin sensitivity was assessed using a linear mixed‐effects model including CPI, Matsuda group (median split), their interaction term (CPI × Matsuda group), and visit as fixed effects with a random intercept for participant. All statistical tests were two‐tailed, and a P value <0.05 was considered statistically significant. Statistical analyses were conducted using Python (version 3.11) and R (version 4.4.1).

RESULTS

Changes in CGM‐derived metrics

Table 1 presents changes in glycemic indices over time, analyzed using linear mixed‐effects models to account for within‐subject repeated measures. Glycemic control significantly improved over the 12‐week intervention. Model‐estimated TIR (95% confidence interval) increased from 84.3% (78.3–90.3) at baseline to 87.4% (81.4–93.5) at week 6 and 90.3% (84.2–96.3) at week 12 (P = 0.041). Among other CGM‐derived metrics, TAR declined significantly (P = 0.007), and glycemic variability, measured by CV, was also reduced (P < 0.001). Although TITR showed an increasing trend over time, the change did not reach statistical significance (P = 0.233).

Table 1.

Indices of glycemic control across time points during the 12‐week dietary intervention (n = 24)

Variables	Baseline	Week 6	Week 12	P
Blood chemistry
FPG (mg/dL)	127.8 (117.5–138.1)	108.2 (88.6–127.8)	110.2 (90.6–129.9)	<0.001
2 h‐glucose (mg/dL)	246.3 (217.9–274.7)	230.9 (179.0–282.8)	209.0 (157.1–260.9)	0.002
HbA1c (%)	7.0 (6.6–7.4)	6.5 (5.7–7.2)	6.1 (5.4–6.8)	<0.001
Metrics derived from continuous glucose monitoring
TIR (%)	84.3 (78.3–90.3)	87.4 (81.4–93.5)	90.3 (84.2–96.3)	0.041
TITR (%)	63.9 (54.0–73.9)	69.6 (59.7–79.6)	69.5 (59.6–79.5)	0.233
MG (mg/dL)	136.9 (125.8–147.9)	127.3 (116.3–138.4)	129.7 (118.6–140.7)	0.069
GMI (%)	6.6 (6.3–6.8)	6.4 (6.0–6.7)	6.4 (6.1–6.7)	0.096
TAR (%)	15.5 (9.7–21.4)	10.9 (5.1–16.8)	9.5 (3.7–15.4)	0.007
TBR (%)	0.5 (−0.7–1.7)	1.7 (0.5–2.9)	0.2 (−1.0–1.4)	0.171
CV (%)	26.7 (24.7–28.8)	24.9 (22.8–26.9)	21.5 (19.4–23.6)	<0.001

Estimated mean values and 95% confidence intervals at each visit were derived using linear mixed‐effects models. P‐values indicate the statistical significance of visit‐wise changes in each variable. 2 h‐glucose, 2‐h plasma glucose level during the 75‐g OGTT; CV, coefficient of variation; FPG, fasting plasma glucose; GMI, glucose management indicator; HbA1c, glycated hemoglobin; MG, mean glucose; TAR, time above range; TBR, time below range; TIR, time in range.

Weight loss and improvement in β‐cell function

BMI significantly decreased over the 12‐week intervention period. The model‐estimated BMI values (95% confidence interval) at each time point were 29.64 (28.15–31.12) at baseline, 27.37 (25.92–28.81) at week 6, and 27.04 (25.60–28.49) at week 12 (P < 0.001 for overall trend). The modified disposition index, calculated as the product of the CPI and Matsuda index, significantly increased over time (P < 0.001). Pairwise comparisons confirmed significant improvements from baseline at both week 6 (P = 0.014) and week 12 (P = 0.002). Model‐estimated means (95% confidence intervals) were 0.060 (0.029–0.091) at baseline, 0.121 (0.061–0.181) at week 6, and 0.125 (0.065–0.185) at week 12 (Figure 1).

Figure 1.

Changes in body mass index (BMI) and β‐cell function across Baseline, Week 6, and Week 12 visits. Blue line represents predicted BMI and red line represents predicted modified disposition index, based on linear mixed‐effects models. Error bars indicate 95% confidence intervals. Asterisks mark time points with statistically significant differences from baseline. P‐values reflect the overall significance of time effects for each variable (BMI: P < 0.001; disposition index: P < 0.001).

CPI, a marker of early‐phase insulin secretion calculated as the ratio of the incremental C‐peptide to glucose responses at 30 min during the OGTT, showed a significant time‐dependent change (P = 0.153 for the overall time effect), with a significant increase observed at week 6 compared with baseline (P = 0.041) that was not sustained at week 12. Model‐estimated means (95% confidence intervals) were 0.016 (0.011–0.022) at baseline, 0.022 (0.013–0.031) at week 6, and 0.019 (0.010–0.027) at week 12. Insulin sensitivity, assessed by the Matsuda index, showed a significant time‐dependent change (P = 0.004), with a marked improvement observed at week 12 compared with baseline (P = 0.002). The model‐estimated mean (95% confidence interval) Matsuda index was 4.98 (2.98–6.97) at baseline, 6.34 (4.26–8.42) at week 6, and 8.18 (6.01–10.36) at week 12 (Figure 2).

Figure 2.

Changes in C ‐ peptide index (CPI, insulin secretion) and Matsuda Index (insulin sensitivity) across Baseline, Week 6, and Week 12 visits. The yellow and green lines represent the predicted CPI and Matsuda Index, respectively, based on linear mixed‐effects models. Error bars indicate 95% confidence intervals. Significant changes from baseline were observed for CPI at Week 6 and for the Matsuda Index at Week 12 (asterisks). However, only the Matsuda Index showed a significant overall time effect (CPI: P = 0.153; Matsuda Index: P = 0.004).

In addition, changes in the insulinogenic index and the corresponding conventional disposition index were analyzed. The insulinogenic index showed only a modest, non‐significant trend over time, without a clear increasing pattern. Model‐estimated means (95% confidence intervals) were 0.188 (0.093–0.282) at baseline, 0.236 (0.141–0.330) at week 6, and 0.205 (0.110–0.299) at week 12. The overall effect of time was not significant (P = 0.472), and pairwise comparisons showed no significant differences between baseline and week 6 (P = 0.227) or baseline and week 12 (P = 0.670).

Consistent with this pattern, the conventional disposition index derived from the insulinogenic index and the Matsuda index demonstrated a more modest response, with a significant improvement observed only at week 6 and no significant overall time effect. Model‐estimated means (95% confidence intervals) were 0.639 (0.206–1.072) at baseline, 1.239 (0.806–1.672) at week 6, and 1.007 (0.574–1.440) at week 12. The overall effect of time showed a borderline association (P = 0.078). Pairwise comparisons demonstrated a significant increase at week 6 compared with baseline (P = 0.025), whereas the difference at week 12 was not statistically significant (P = 0.169).

Clinical variables associated with TIR

Associations between clinical factors and CGM‐derived TIR were evaluated using linear mixed‐effects models with participant‐specific random intercepts to account for repeated measurements across study visits, and the results are presented in Table 2. TIR was significantly higher among participants with a shorter duration of diabetes. In univariable linear mixed‐effects models, TIR was significantly associated with fasting glucose, C‐peptide, insulin, and HOMA‐β levels. Indices reflecting early‐phase insulin secretory function, including the insulinogenic index and the CPI, were both significantly associated with TIR. Notably, CPI demonstrated greater explanatory power for TIR than the insulinogenic index, as reflected by a higher marginal R² (0.433 vs. 0.379). In contrast, markers of insulin sensitivity—such as BMI, HOMA‐IR, and the Matsuda index—were not significantly associated with TIR, and these findings were consistent in multivariable analyses.

Table 2.

Clinical and metabolic correlates of time in range

Variables	β	P	R2
Univariable analysis
Patient characteristics
Age	−0.120	0.618	0.040
Body mass index	0.959	0.170	0.081
Duration of diabetes	−3.418	0.003	0.246
CGM‐derived glycemic metrics
Mean glucose	−0.468	<0.001	0.813
Coefficient of variation	−0.733	0.015	0.420
OGTT‐based β‐cell function parameters
Fasting glucose	−0.283	<0.001	0.541
Fasting c‐peptide	2.876	0.179	0.370
Fasting insulin	0.350	0.223	0.368
HOMA‐β	0.102	0.002	0.415
HOMA‐IR	0.300	0.735	0.357
Insulinogenic index*	12.367	0.096	0.379
C‐peptide index*	451.037	0.001	0.433
Matsuda index	−0.182	0.520	0.358
Multivariable analysis †
C‐peptide index	446.652	0.001	0.430
Matsuda index	−0.060	0.828

Associations between clinical variables and CGM‐derived time in range (TIR) were evaluated using linear mixed‐effects models with participant‐specific random intercepts, accounting for repeated measurements across study visits.

^*The C‐peptide index and insulinogenic index were calculated to assess the first‐phase insulin secretory response during the OGTT, defined as the ratio of the incremental change in C‐peptide or insulin to the corresponding change in glucose between 0 and 30 min.

^† The multivariable model included the C‐peptide index and the Matsuda index.

Stratified association between insulin secretion and TIR by insulin sensitivity

To explore whether the association between CPI and TIR varied by insulin sensitivity, participants were stratified according to the baseline Matsuda index. The association between CPI and TIR was evaluated using linear regression separately for each group, as visualized in Figure 3. A significant CPI × Matsuda group interaction was identified using a linear mixed‐effects model with visit as a fixed effect and participant as a random intercept (P for interaction = 0.011).

Figure 3.

Interaction between insulin secretion and insulin sensitivity in relation to TIR. Scatter plots represent individual observations. Matsuda groups were defined using a median split of the baseline Matsuda index. Lines indicate fitted linear regression lines within each stratum. The interaction between CPI and Matsuda group was tested using a linear mixed‐effects model including CPI, Matsuda group, their interaction term, and visit as fixed effects with a random intercept for participant (P for interaction = 0.011).

DISCUSSION

This was a prospective, single‐arm exploratory study utilizing CGM to evaluate the glycemic effects of a calorie‐restricted meal plan in patients with early‐stage type 2 diabetes and obesity. Despite discontinuation of prior antidiabetic therapy, TIR significantly improved from 84.1% to 90.4% over the 12‐week study period. Analysis of factors associated with TIR during the intervention period revealed that the early‐phase insulin response, as measured by the CPI, was the strongest predictor of TIR in patients with type 2 diabetes. During the initial 6‐week period of total food replacement, a transient increase in the CPI was observed, suggesting a potential role of caloric restriction in preserving β‐cell function. Moreover, improvements in insulin sensitivity associated with weight loss during the 12‐week nutritional intervention appeared to enhance the interaction between CPI and TIR, thereby contributing to further increases in TIR.

Individuals with similar HbA1c values may have significantly different TIRs, suggesting that TIR may provide more individualized and actionable insights into glycemic control ¹⁷ . Therefore, previous studies analyzing CGM patterns in individuals with type 2 diabetes have suggested that distinct CGM‐derived features may offer valuable insights into the heterogeneous pathophysiology of type 2 diabetes and contribute to the advancement of personalized medicine ¹⁸ . Although there is a recommended target of TIR of approximately 70% corresponding to an HbA1c of 7% ¹⁹ , no clear consensus has been established regarding the optimal TIR cut‐off to define remission in type 2 diabetes ⁵ . Despite being diagnosed with type 2 diabetes using conventional criteria, participants in this study exhibited a relatively high baseline TIR of 89% (interquartile range, 84–92%). By the end of the 12‐week intervention, TIR had further improved to 97% (interquartile range, 92–99%), with most individuals maintaining TIR levels above 70% throughout the study period. Therefore, it may be necessary to propose new glycemic goals, either to aid in the diagnosis of type 2 diabetes or to guide early‐stage management. In addition, while TITR may theoretically provide greater sensitivity when TIR approaches a ceiling, our findings did not demonstrate such incremental sensitivity. In early‐stage type 2 diabetes with obesity, caloric restriction improves insulin sensitivity, thereby attenuating postprandial glucose excursions. Accordingly, the observed improvement in this study was largely attributable to reductions in time spent in hyperglycemic ranges (>180 mg/dL) rather than to a substantial redistribution into the normal range (70–140 mg/dL).

When evaluating clinical factors associated with TIR using CGM and OGTT data collected throughout the study period, CPI—an indicator of early‐phase insulin secretory function—emerged as the strongest predictor of TIR (Table 2). Other markers of insulin secretion, including HOMA‐β and the insulinogenic index, were also significantly associated with TIR. In contrast, indices reflecting insulin sensitivity—such as BMI, HOMA‐IR, and the Matsuda index—showed no significant association, consistent with previous findings ²⁰ , ²¹ . In type 1 diabetes, it is well established that preservation of insulin secretory function is strongly associated with higher TIR ²² . However, the pathogenesis of type 2 diabetes is multifactorial and more heterogeneous, making such associations less certain. The findings of this study reinforce the notion that, even in type 2 diabetes, insulin secretory capacity remains a principal determinant of glycemic control. However, given the central pathophysiological role of insulin resistance in type 2 diabetes, these findings should be interpreted with caution ²³ . Subgroup analysis revealed a clear difference in the association between CPI and TIR according to insulin sensitivity (Figure 3). In the group with a higher Matsuda index, increases in CPI were more strongly associated with improvements in TIR. Although this association remained statistically significant in the low Matsuda index group, its magnitude was relatively modest. These results suggest that preserved peripheral insulin sensitivity may be a critical prerequisite for insulin secretory function to exert its full effect on glycemic control.

To account for the interplay between insulin secretion and sensitivity in the pathophysiology of type 2 diabetes, the disposition index is commonly used as a quantitative measure of β‐cell function ²⁴ . The disposition index, calculated as the product of insulin secretion and sensitivity indices, reflects the compensatory relationship whereby reduced insulin sensitivity requires increased insulin secretion. In this study, a modified disposition index—calculated as the product of CPI and the Matsuda index—improved approximately twofold over the 12‐week dietary intervention. This increase appears to be primarily driven by improvements in insulin sensitivity, with relatively modest changes in insulin secretion indices. Nevertheless, the resulting increase in the disposition index suggests an overall improvement in β‐cell function relative to insulin sensitivity ²⁵ . The magnitude of this improvement is within the range reported following short‐term pharmacologic interventions such as SGLT2 inhibitor therapy ²⁶ .

When examining changes in the components of the disposition index, both CPI and the Matsuda index increased by approximately 30% after 6 weeks of a calorie‐restricted meal plan incorporating total food replacement. However, while the Matsuda index continued to rise throughout the 12‐week intervention—reaching an overall increase of about 70% in parallel with weight loss—the improvement in CPI was transient and returned to baseline levels after regular meals were reintroduced (Figure 2). In individuals with obesity, caloric restriction appears to improve glycemic control primarily through enhanced insulin sensitivity. Nonetheless, from the perspective of diabetes remission, there may be a need to develop nutritional intervention strategies specifically targeting improvements in insulin secretory function.

In this study, CPI was used as a representative marker of insulin secretion. The early‐phase insulin response—reflected by insulin secretion within the first 30 min of glycemic stimulation—plays a crucial role in suppressing postprandial glucose excursions and inhibiting hepatic glucose production. Impairment of this early insulin response is a well‐recognized feature of the early pathophysiology of type 2 diabetes ²⁷ , ²⁸ . CPI was significantly associated with TIR in linear mixed‐effects models (marginal R² = 0.473, P = 0.001), showing a higher explanatory power than HOMA‐β (marginal R² = 0.456, P = 0.001), a conventional marker of insulin secretory function. Indeed, other metabolic indices reflecting insulin secretory function, such as HOMA‐β or the insulinogenic index, are based on plasma insulin levels. In contrast, C‐peptide–based calculations may more accurately reflect insulin secretion, as C‐peptide has a longer half‐life and is not metabolized by the liver, resulting in more stable peripheral concentrations ²⁹ . Notably, the post‐load C‐peptide to glucose ratio demonstrated a stronger association with TIR compared to the fasting status (Table S4). This suggests that it may serve as a practical alternative in situations where the calculation of the C‐peptide index is not feasible due to its complexity ³⁰ , ³¹ .

Additional analysis of clinical factors associated with CPI revealed a negative correlation between diabetes duration and CPI, independent of age and BMI. Notably, a decline in early‐phase insulin secretion with increasing disease duration was evident even among participants with newly diagnosed type 2 diabetes in this study (Table S5). Consistent with this observation, previous studies have shown that sustained remission achieved through calorie restriction depends on the recovery of β‐cell function, which is more likely to occur in individuals with a shorter duration of diabetes ³² . Together, these findings highlight the progressive nature of β‐cell dysfunction early in the course of type 2 diabetes and underscore the importance of initiating dietary interventions promptly after diagnosis in order to modify the natural course of the disease.

This study has several limitations. First, it was designed as a single‐arm, non‐randomized intervention without a parallel control group, which limits the ability to draw causal inferences regarding the observed improvements in glycemic outcomes. Second, the relatively small sample size and single‐center recruitment are inherent to the exploratory design of this study and may limit statistical precision. Third, although the use of CGM provided detailed and dynamic glycemic data, it was applied intermittently over three 14‐day periods. This approach, while informative, may not fully reflect long‐term glycemic variability or capture fluctuations beyond the monitoring windows. Lastly, given that pharmacologic effects should be considered when interpreting glycemic control and β‐cell function, the 1‐week washout period may have been relatively short. Most glucose‐lowering agents have relatively short half‐lives, and their pharmacologic effects generally diminish within several days after discontinuation. In contrast, thiazolidinediones exert durable metabolic effects on insulin sensitivity that may persist for several weeks after cessation ³³ . However, no participants in our study were treated with these agents. In addition, as many glucose‐lowering agents are primarily eliminated via the kidneys, impaired renal function may prolong drug exposure ³⁴ . Notably, no participants in our study had chronic kidney disease, thereby minimizing the potential impact of residual drug effects on our findings. Despite these limitations, this study is among the first to evaluate glycemic responses to caloric restriction using CGM‐derived metrics in individuals with newly diagnosed type 2 diabetes. Our findings offer novel insights into the pathophysiological mechanisms of glycemic regulation in the context of lifestyle intervention and support the broader application of CGM‐derived metrics in both clinical practice and research.

In conclusion, caloric restriction appears to be an effective strategy for improving glycemic control in individuals with type 2 diabetes and obesity. Weight loss achieved through dietary intervention improved TIR, primarily by enhancing insulin sensitivity. While changes in β‐cell secretory efficacy were transient, the overall increase in the disposition index suggests a coordinated adaptation of insulin secretion in response to improved sensitivity. This study delineates the range and temporal dynamics of TIR in newly diagnosed type 2 diabetes, underscoring its clinical value as a complementary metric and highlighting the potential role of integrating CGM with dietary interventions to optimize metabolic outcomes ³⁵ .

FUNDING

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (2021R1A2C2013890); the funders had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.

DISCLOSURE

The authors declare no conflict of interest.

Approval of the research protocol: The study protocol was approved by the Institutional Review Board of Yeouido St. Mary's Hospital (IRB No. SC23ENSE0036).

Informed consent: Written informed consent was obtained from all participants.

Approval date of Registry and the Registration No. of the study/trial: ClinicalTrials.gov registration number: NCT05754775; First posted on March 6, 2023.

Animal Studies: N/A.

Supporting information

Table S1 Baseline characteristics of the patients.

Table S2 A representative daily meal plan.

Table S3 Participant‐reported dietary intake during the intervention.

Table S4 Comparative analysis of insulin secretion markers in relation to time in range.

Table S5 Univariable and multivariable regression models for C‐peptide index.

Figure S1 Study design.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.