Relationship between the physical activity levels and AUC

In the energy trial, the step counts and MVPA per day during the study period were 9545.6 (4433.5) steps (95% CI 6603.4–1248.7) and 91.5 (39.1) min (95% CI 65.6–117.5). In the balance trial, the step counts and MVPA per day during the study period were 9905.8 (3546.6) steps (95% CI 7552.2–28.0) and 100.6 (28.1) min (95% CI 80.9–26.6–120.3) (Additional file 3: Fig. S3E, F). The correlation between physical activity level and the AUC of postprandial glucose level after dinner was analyzed during the study period. Step counts and MVPA were used as indices of physical activity level. In the energy trial, there was no significant correlation between step counts or MVPA during the study period and the after-dinner AUC in the standard lunch trial (Additional file 1: Fig. S1A, B). In the balance trial, there was no significant correlation between step counts or MVPA during the study period and the after-dinner AUC in the standard trial (Additional file 1: Fig. S1C, D).

Comparison of the glucose levels in the energy trial

The 24-h blood glucose excursion in the energy trial is shown in Additional file 2: Fig. S2A. The monitoring of glucose levels for 4 h after lunch showed an increase in postprandial blood glucose levels in the standard- and high-energy lunch trials (Fig. 2B). Significant differences between energy trials after lunch at each 15-min time interval were examined (Additional file 2: Supplemental Fig. S2C). The AUC for 4 h after lunch was 339.6 (39.4) (95% CI 315.8–363.4) in the no-energy lunch trial, 354.7 (37.5) (95% CI 332.0–377.3) in the low-energy lunch trial, 403.7 (39.7) (95% CI 379.7–427.7) in the standard-energy lunch trial and 418.4 (29.9) (95% CI 400.3–436.5) in high-energy lunch trial. The AUC for 4 h after lunch showed no significant difference between the no and low-energy lunch trials (P = 0.107) and between the standard- and high-energy lunch trials (P = 0.601, 95%CI = 16.2) (Fig. 2C). The peak for 4 h after lunch was 89.9 (12.1) (95% CI 82.6–97.3) in the no-energy lunch trial, 108.5 (11.0) (95% CI 101.9–115.2) in the low-energy lunch trial, 131.0 (18.3) (95% CI 119.9–142.1) in the standard-energy lunch trial and 126.5 (12.6) (95% CI 118.9–134.1) in high-energy lunch trial. The peak for 4 h after lunch showed a significant decrease in no-energy lunch trial compared with the low-, standard- and high-energy lunch trial (P = 0.004, P = 0.001, P = 0.001 vs no-energy lunch trial). The peak for 4 h after lunch showed a significant decrease in the low-energy lunch trial compared with the standard- and high-energy lunch trial (P = 0.001, P = 0.002 vs low-energy lunch trial). (Fig. 2D). Monitoring of glucose levels for 4 h after dinner showed an increase in postprandial blood glucose levels in the no- and low-energy lunch trials (Fig. 2E). Significant differences between energy trials after dinner at each 15-min time interval were examined (Additional file 2: Fig. S2D). The iAUC for 2 h after dinner was 122.6 (59.1) (95% CI 95.5–149.7) in the no-energy lunch trial, 117.0 (55.4) (95% CI 90.8–143.1) in the low-energy lunch trial, 72.7 (44.0) (95% CI 55.4–90.0) in the standard-energy lunch trial and 42.0 (34.9) (95% CI 29.3–54.6) in high-energy lunch trial. On examining the iAUC for 2 h after dinner, the no and low-energy lunch trials showed an increase in iAUC after dinner compared with the standard- and high-energy lunch trials (P = 0.006, P = 0.001 vs none), (P = 0.004, P = 0.001 vs low-energy trial) (Fig. 2F). The AUC for 3 h after dinner was 403.2 (46.5) (95% CI 375.1–431.3) in the no-energy lunch trial, 405.2 (49.2) (95% CI 375.4–434.9) in the low-energy lunch trial, 366.5 (39.0) (95% CI 342.9–390.0) in the standard-energy lunch trial and 332.4 (35.2) (95% CI 311.2–353.7) in high-energy lunch trial. On examining the AUC for 3 h after dinner, the standard-energy lunch trial had a significantly lower value than the no- and low-energy lunch trials (P = 0.046, P = 0.019 vs standard-energy lunch trial), and the high-energy lunch trial had a significantly lower value than the no-, low-, and standard-energy lunch trials (P = 0.002, P = 0.001 vs high-energy lunch trial) (Fig. 2G). The sum of the AUC for 4 h after lunch and dinner was 843.0 (54.8) (95% CI 808.5–877.4) in the no-energy lunch trial, 854.5 (74.7) (95% CI 807.5–901.5) in the low-energy lunch trial, 867.5 (61.1) (95% CI 829.1–906.0) in the standard-energy lunch trial and 847.8 (61.9) (95% CI 808.9–886.8) in high-energy lunch trial. On comparing the sum of the AUC for 4 h after lunch and dinner, the standard-energy lunch trial had a significantly higher value than the no-lunch trial (Fig. 2H). The peak glucose levels after dinner were 183.1 (26.4) (95% CI 167.1–199.0) in the no-energy lunch trial, 181.5 (22.1) (95% CI 168.1–194.8) in the low-energy lunch trial, 162.2 (23.3) (95% CI 148.1–176.2) in the standard-energy lunch trial, and 134.9 (21.9) (95% CI 127.2–148.1) in the high-energy lunch trial. The peak glucose levels after dinner in the no-, low-, and standard-energy lunch trials were significantly higher than that in the high-energy lunch trial (P = 0.000, P = 0.000, P = 0.006, vs high-energy lunch trial) (Fig. 2I).

Fig. 2

Postprandial blood glucose levels for lunch and dinner in the energy trial. All data are presented as the mean (standard deviation). Study protocol for the energy trial (A). Glucose concentration (B), area under the curve (AUC) (C), and peak blood glucose levels (D) for 4-h after lunch. Glucose concentration (E) for 4 h after dinner. Incremental AUC (iAUC) (F) and AUC (G) at 2 and 3 h after dinner. Sum of the AUC for 4 h after lunch and dinner for each trial in the energy trial (H). Peak blood glucose levels at 4 h after dinner (I). Figure 2C and I used One-way ANOVA. D, F, G, and H used the Friedman test.

Comparison of the glucose levels in the balance trial

The 24-h blood glucose excursion in the balance trial is shown in Additional file 2: Fig. S2B. In the balance trial, monitoring glucose levels for 4 h after lunch showed an increase in postprandial blood glucose levels in the carbohydrate-rich trial (Fig. 3B). Significant differences between balance trials after lunch at each 15-min time interval were examined (Additional file 2: Fig. S2E). The AUC for 4 h after lunch was 400.9 (35.6) (95% CI 380.3–421.5) in the standard trial, 367.5 (50.8) (95% CI 338.1–396.8) in the protein-rich trial, 361.2 (33.4) (95% CI 341.9–380.4) in the fat-rich trial and 473.0 (61.9) (95% CI 437.2–508.7) in the carbohydrate-rich trial. The AUC for 4 h after lunch showed a significant difference between the standard and other trials and between the carbohydrate-rich and other trials (P = 0.019 vs protein-rich trial, P = 0.006 vs fat-rich trial) (P = 0.006 vs standard trial, P = 0.004 vs protein-rich trial, P = 0.001 vs fat-rich trial) (Fig. 3C). The peak for 4 h after lunch was 124.9 (14.9) (95% CI 116.3–133.6) in the standard trial, 102.6 (25.1) (95% CI 88.1–117.1) in the protein-rich trial, 104.7 (14.3) (95% CI 96.5–113.0) in the fat-rich trial, and 146.7 (25.9) (95% CI 131.8–161.7) in the carbohydrate-rich trial. The peak for 4 h after lunch showed a significant increase in the carbohydrate-rich trial compared with the standard, protein-rich and fat-rich trial (P = 0.013, P = 0.004, P = 0.001 vs carbohydrate-rich trial). The peak for 4 h after lunch showed a significant increase in the standard trial compared with the protein-rich and fat-rich trial (P = 0.023, P = 0.002 vs standard trial) (Fig. 3D). Monitoring glucose levels for 4 h after dinner showed an increase in postprandial blood glucose levels in the fat-rich trial (Fig. 3E). Significant differences between balance trials after dinner at each 15-min time interval were examined (Additional file 2: Fig. S2F). The iAUC for 2 h after dinner was 57.5 (37.4) (95% CI 25.6–112.4) in the standard trial, 49.4 (28.0) (95% CI 25.6–63.9) in the protein-rich trial, 92.4 (41.4) (95% CI 58.5–114.0) in the fat-rich trial and 60.7 (41.8) (95% CI 29.1–84.9) in the carbohydrate-rich trial. On examining the iAUC for 2 h after dinner, there was a significantly higher rate in the fat-rich trials than those in the standard and protein-rich trials (P = 0.034, P = 0.022 vs fat-rich trials) (Fig. 3F). The AUC for 3 h after dinner was 348.4 (54.2) (95% CI 317.1–379.7) in the standard trial, 333.4 (39.8) (95% CI 310.4–356.4) in the protein-rich trial, 376.6 (62.6) (95% CI 340.4–412.7) in the fat-rich trial and 341.3 (57.3) (95% CI 308.2–374.4) in the carbohydrate-rich trial. A significantly higher AUC for 3 h after dinner was observed in the fat-rich trial than that in the protein-rich trial (P = 0.050 vs protein-rich trial) (Fig. 3G). The sum of the AUC for 4 h after lunch and dinner was 846.5 (80.9) (95% CI 799.8–893.2) in the standard trial, 794.8 (69.4) (95% CI 754.8–834.9) in the protein-rich trial, 832.4 (96.4) (95% CI 776.7–888.1) in the fat-rich trial, and 910.0 (113.2) (95% CI 844.4–975.2) in the carbohydrate-rich trial. On comparing the sum of the AUC for 4 h after lunch and dinner, the carbohydrate-rich trial had significantly higher values than the standard, protein-rich, and fat-rich trials. (P = 0.048, P = 0.002, P = 0.019 vs carbohydrate-rich trial) The number of protein-rich trials was significantly lower than that of the standard and fat-rich trials (P = 0.013, P = 0.041 vs protein-rich trial) (Fig. 3H). The peak blood glucose level after dinner was 141.4 (26.6) (95% CI 126.1–156.8) in the standard trial, 134.6 (16.0) (95% CI 125.3–143.8) in the protein-rich trial, 162.4 (31.8) (95% CI 144.1–180.8) in the fat-rich trial, and 146.9 (31.1) (95% CI 128.9–164.9) in the carbohydrate-rich trial The peak blood glucose level after dinner was significantly higher in the fat-rich trial than that in the protein-rich trial (P = 0.017 vs protein-rich trial) (Fig. 3I).

Fig. 3

Postprandial blood glucose levels for lunch and dinner in the balance trial. All data are presented as the mean (standard deviation). Study protocol for the balance trial (A). Glucose concentration (B), area under the curve (AUC) (C), and peak blood glucose levels (D) for 4 h after lunch. Glucose concentration (E) for 4 h after dinner. Incremental AUC (iAUC) (F) and AUC (G) for 2 and 3 h after dinner. The sum of AUC for 4 h after lunch and dinner for each trial in the balance trial (H). Peak blood glucose levels at 4 h after dinner (I). Figure 3C, D, and H used the Friedman test. F, G, and I used One-way ANOVA.

Relationship between starvation time and blood glucose levelsCorrelation between the difference in starvation time from breakfast to lunch and blood glucose levels

In the energy trial, no correlation was found between the starvation time from breakfast to lunch (95% CI 4.5–5.4 h, 4.1–5.3 h and 4.4–6.2 h in the low-, standard- and high-energy lunch trial) and the AUC for 4 h after lunch (P = 0.949, P = 0.895, P = 0.637) (Fig. 4A, B, and C). No correlation was found between the starvation time from breakfast to lunch (95% CI 4.6–5.5 h, 4.7–5.9 h, 4.6–5.5 h and 4.4–5.4 h in the standard, protein-rich, fat-rich and carbohydrate-rich trial) and the AUC 4 h after lunch in the balance trial (P = 0.770, P = 0.449, P = 0.953, P = 0.191) (Fig. 4D, E, F, and G).

Fig. 4

Correlations between starvation time from breakfast to lunch and postprandial blood glucose levels for lunch. Correlation between starvation time from breakfast to lunch and iAUC for 2 h after lunch in the low- (A), standard- (B), and high- (C) energy trials. Correlations between starvation time from breakfast to lunch and iAUC 2 h after lunch in the standard (D), protein-rich (E), fat-rich (F), and carbohydrate-rich (G) trials.

Correlation between the difference in starvation time from lunch to dinner and blood glucose levels

In the energy trial, on examining the correlation between the starvation time from lunch to dinner (95% CI 10.1–11.5, 5.2–6.2 h, 5.4–6.3 h and 5.1–6.1 h in the no-, low-, standard- and high-energy lunch trial) and the iAUC for 2 h after dinner, a correlation was found only in the high-lunch trial (P < 0.01, r = 0.799) (Fig. 5D). In the balance trial, on examining the correlation between the starvation time from lunch to dinner (95% CI 5.3–6.1 h, 5.3–6.4 h, 5.2–6.1 h and 5.5–6.7 h in the standard, protein-rich, fat-rich and carbohydrate-rich trial) and the iAUC for 2 h after dinner, a correlation was found only in the fat-rich trial (P < 0.01, r = 0.735) (Fig. 5H).

Fig. 5

Correlations between starvation time from lunch to dinner and postprandial blood glucose levels for dinner. Correlation between starvation time from lunch to dinner and iAUC for 2 h after lunch in the no (A), low- (B), standard- (C), and high-energy trials (D). Correlations between starvation time from lunch to dinner and iAUC 2 h after dinner in the standard (E), protein-rich (F), fat-rich (G), and carbohydrate-rich (H) trials.

Comparison of the blood glucose levels after dinner on classifying into two groups by mean starvation time in the energy trial

There was a positive correlation between the length of the starvation time from lunch to dinner and blood glucose levels in the high-energy and high-carbohydrate diets (Fig. 5). Therefore, we divided the subjects into two groups: the shorter-starvation group (n = 6, starvation time range 4.2–5.6 h) and the longer-starvation group (n = 7, starvation time range 5.7–7.1 h) by the median mean of starvation time from lunch to dinner in the energy trial.

In the shorter starvation group, the iAUC for 2 h after dinner was 135.0 (65.2) (95% CI 86.7–183.3) in the no-energy lunch trial, 121.8 (43.4) (95% CI 88.4–155.2) in the low-energy lunch trial, 56.6 (35.0) (95% CI 34.4–78.8) in the standard energy trial and 146.9 (31.1) (95% CI 16.5–39.1) in the high-energy trial. The iAUC for 2 h after dinner was significantly lower in the high-intake trial than that in the no-, low-, and standard-lunch trials (P = 0.028, P = 0.028, P = 0.046 vs high-energy lunch trial). Moreover, it was significantly lower in the standard lunch trial than that in the no and low-energy lunch trials (P = 0.028, P = 0.028, P = 0.046 vs standard-energy trial). In the longer starvation group, the iAUC for 2 h after dinner was 112 (54.8) (95% CI 70.9–153.1) in the no-energy lunch trial, 112.8 (66.1) (95% CI 63.4–162.1) in the low-energy lunch trial, 86.5 (44.2) (95% CI 60.6–112.4) in the standard energy trial, and 54.1 (39.0) (95% CI 35.3–72.9) in the high-energy lunch trial.The iAUC for 2 h after dinner was significantly lower in the high-energy lunch trial than those in the no, low-, or standard-energy lunch trials (P = 0.018, P = 0.018, P = 0.043 vs high-energy lunch trial). Comparing the longer-starvation group with the shorter-starvation group, the standard-energy lunch trial in the shorter starvation group showed lower blood glucose levels after dinner (Fig. 6A, B). We examined the physical characteristics of the shorter- and longer-starvation groups. The starvation times were 5.3 (0.5) and 6.3 (0.1) h in the shorter and longer starvation group. The longer starvation group had significantly longer starvation times (P = 0.000) (Additional file 3: Fig. S3A). In addition, the iAUC for 2 h after dinner in the standard trial was 50.0 (23.8) (95% CI 35.0–65.0) and 78.6 (34.5) (95% CI 57.9–99.3) mg/dl*2 h in the shorter and longer starvation group. Comparing the iAUCs for 2 h blood glucose levels after dinner in the standard-lunch trial, the longer-starvation group showed significantly higher values than did the shorter starvation group (P = 0.024) (Additional file 3: Fig. S3B). There were no significant differences in BMI or daily energy intake between groups (P = 0.587, P = 0.723) (Additional file 3: Fig. S3C, D). The step counts were 9786.6 (3786.3) (95% CI 7405.2–12,168.1) and 9653.6 (1289.2) (95% CI 7064.3–12,242.9) steps/day in the shorter and longer starvation group. The MVPA was 99.2 (31.2) (95% CI 79.5–118.8) and 92.4 (37.4) (95% CI 69.7–115.0) min/day in the shorter and longer starvation group. Physical activity levels, such as step counts and MVPA, tended to be higher in the shorter-starvation group than those in the longer-starvation group (P = 0.645, P = 0.245) (Additional file 3: Fig. S3E, F). The daily intake of dietary fiber was 13.7 (3.0) (95% CI 11.8–15.6) and 13.1 (3.0) (95% CI 11.3–14.9) g in the shorter and longer starvation group. The MEQ score was 64.5 (11.7) (95% CI 57.2–71.9) and 61.6 (7.3) (95% CI 57.2–66.0) in the shorter and longer starvation group. The shorter-starvation group tended to consume more dietary fiber and had higher MEQ scores than the longer-starvation group (P = 0.628, P = 0.461) (Additional file 3: Fig. S3G, H).

Fig. 6

Comparisons of the blood glucose levels after dinner on classifying into two groups by mean starvation time in the energy and balance trials All the data are presented as the mean (standard deviation). iAUC for 2 h after dinner for each trial for the shorter-starvation (A) and longer-starvation (B) groups in the energy trial. iAUC for 2 h after dinner for each trial for the shorter-starvation (C) and longer-starvation (D) groups in the balance trial. Fig A and B used the Friedman test. C used One-way ANOVA.

Comparison of the blood glucose levels after dinner on classifying into two groups by mean starvation time in the balance trial

We divided the patients into two groups: the shorter starvation group (n = 7, starvation time range 4.4–5.9 h) and the longer starvation group (n = 7, starvation time range 5.9–7.3 h) based on the median mean of starvation time from lunch to dinner in the balance trial.

In the shorter starvation group, the iAUC for 2 h after dinner was 44.3 (27.9) (95% CI 18.5–70.1) in the standard trial, 31.3 (24.4) (95% CI 8.8–53.9) in the protein-rich trial, 99.3 (46.5) (95% CI 56.2–142.3) in the fat-rich trial, and 40.3 (34.7) (95% CI 8.2–72.3) in the carbohydrate-rich trial. The iAUC for 2 h after dinner was significantly higher in the fat-rich trial than those in the standard and carbohydrate-rich trials (P = 0.013, P = 0.015 vs fat-rich trial) (Fig. 6C). However, in the longer-starvation group, the iAUC for 2 h after dinner was 70.7 (43.0) (95% CI 5.9–186.3) in the standard trial, 67.5 (18.7) (95% CI50.2–84.7) in the protein-rich trial, 85.6 (37.9) (95% CI50.5–120.7) in fat-rich trial and 81.2 (40.0) (95% CI 44.1–118.2) in the carbohydrate-rich trial. There were no significant differences between the trials (P = 0.224) (Fig. 6D).

Relationship between the percentage of energy intake during the usual lunch and postprandial blood glucose levels after dinner during the trial

The energy intake and percentage of energy intake of participants in the energy trial are shown in Table 3A, B. The relationship between the 2 h after dinner iAUC and peak blood glucose level for each trial in the energy trial and the usual percentage of energy intake was examined and found to have no significant correlation (P = 0.322, P = 0.197 in usual energy intake, P = 0.514, P = 0.398 in percentage of protein intake, P = 0.456, P = 0.394 in percentage of fat intake, P = 0.325, P = 0.321 in percentage of carbohydrate intake) (Additional file 4: Fig. S4).

Table 3 Energy intake and percentage of energy intake during daily life for participants

The energy intake and percentage of energy intake of the participants in the balance trial are shown in Tables 3C and D. The relationship between the 2 h after dinner iAUC and peak blood glucose level for each trial in the balance trial and the percentage of usual lunch intake was examined. The results showed no significant correlation between the percentage of fat and carbohydrate intake during the usual lunch and postprandial blood glucose levels after dinner. However, there was a positive correlation between the usual percentage of protein intake at lunch and the iAUC at 2 h after dinner (P = 0.088, r = 0.473) and a significant positive correlation with the peak blood glucose level at dinner (P = 0.041, r = 0.551) (Additional file 5: Fig. S5A, D).