Detailed descriptions of the study protocol have been published elsewhere [3]. Briefly, our ten male subjects’ average values in age, height, body weight, body mass index and VO_2max were 24.0 ± 3.3 (range: 20–31) years, 170.8 ± 5.0 (162.1 - 175.6) cm, 61.9 ± 5.7 (54.6 - 69.7) kg, 21.2 ± 1.7 (18.4 - 23.3) kg·m^-2, and 3.18 ± 0.45 (2.34 - 3.75) L·min^-1 (or 52.0 ± 9.2 (36.2 - 61.4) ml·kg^-1·min^-1), respectively. The aim and design of this study were explained to every subject before each gave their written, informed consent. This study was conducted in accordance with the guidelines proposed in the Declaration of Helsinki. The Ethical Committee of both JAXA (reference number: 32-2-7) and University of Tsukuba (22–283) reviewed and approved the study protocol.

The subjects completed three experimental sessions separated by approximately 1 week. During each session, the subjects performed one of three exercise trials, i.e., SIT, HIAT, or CAT. The detailed protocols of the three cycling exercises are shown in Additional file 1: Table S1. All aspects of the experimental session remained the same during all three sessions except the actual exercise technique. The three sessions were implemented in random order.

On each session day, the subjects arrived at our laboratory at approximately 7:15 am. They were asked to minimize any physical activity (walking etc.) while en route from their home to the laboratory. Six subjects drove their own cars to the laboratory, and 4 subjects were picked up at their homes and driven to the building by a research assistant because they did not have their own cars. At 7:45 am, a subject would seat himself in a comfortable armchair and remain in a resting (seated) position without movement until 8:20 am. From 8:05 to 8:20 am (15 minutes), the subject was connected to a face mask of the indirect calorimeter for baseline data collection. Average VO₂ data during the final 10 minutes were used as the baseline value. After the baseline measurements, we disconnected the face mask and allowed the subject to take a drink of water before riding on the cycling ergometer (75XL Ш, Konami, Tokyo, Japan) for the exercise phase. After the water, he was reconnected to the indirect calorimeter and started pedaling the ergometer at 8:25 am.

After the exercise phase, the subject returned to the resting position immediately but stayed connected to the indirect calorimeter for 10 minutes to measure immediate post-exercise VO₂. After 10 minutes, we removed the face mask but the subject remained in the resting position for another 170 minutes. During the post-exercise phase, the subject remained in the resting position immobile. Ten minutes of VO₂ measurements were started at 30, 60, 90, 120, 150, and 180 minutes post exercise. For each measurement session, average VO₂ data during the final 5 minutes was used.

All gas exchange measurements were measured using the mixing chamber method with an open-circuit computerized indirect calorimeter (AE-310S, Minato Medical Science, Osaka, Japan). The ventilatory volume, VO₂, and VCO₂ were calculated every 15 seconds. The gas analyzer was calibrated before each trial. Coefficient of variation (CV) from the mean for the three resting VO₂ measurements in the 10 subjects was 4.3%.

To measure each subject’s VO_2max, we used the criteria described by Tanaka et al. [4] Briefly, after a 2-min warm up at 15 Watt (W), the subjects began the actual exercise protocol at a 30 W level. The workload was increased every minute by 15 W until volitional exhaustion. We determined the highest oxygen uptake achieved over 30 seconds as the VO_2max.

To determine each subject’s exercise intensity during the exercise session, we used the VO_2max measurement data (i.e., values per minute for workload and VO₂) and calculated a simple linear regression equation for each subject: Y = β x + c, with Y = workload (W), x = VO₂ (ml), and ß and c as constants. Subsequently, percentage VO_2max data (ml) (e.g., 120% VO_2max for SIT) were applied to the equation, whereupon each subject’s exercise intensity (workload (W)) was determined.

To determine net exercise VO₂ and EPOC, the area under the curve (AUC) for each phase was calculated for each subject using a statistical-software package (SAS version 9.2, SAS Institute Inc, Cary, NC, USA). As for the SIT and HIAT, periods of active rest were included in the calculation of the net exercise VO₂.

To analyze differences among the three sessions, we used one-way analysis of variance (ANOVA), and applied Tukey-Kramer’s post hoc test when the difference was significant (P < 0.05) according to the results of ANOVA. We assessed the relationship between two measurement values with Pearson’s product moment correlation. We used SAS, version 9.2 (SAS Institute Inc, Cary, NC, USA) to analyze the data.

The net VO₂ incurred with the HIAT (31.8 ± 4.1 L) was significantly lower than during the CAT (71.1 ± 10.0 L), but it was significantly greater than the net VO₂ incurred during the SIT (14.7 ± 1.5 L). The EPOC was significantly greater with the SIT (6.8 ± 4.0 L) than with the CAT (2.9 ± 2.8 L), and it also tended to be greater than with the HIAT (4.5 ± 3.3 L) (P = 0.12). We observed no significant differences in the EPOC between HIAT and CAT.

Additional file 2: Figure S1 shows the relationship between subjects’ VO_2max and the ratios of EPOC to the net exercise VO₂ during each exercise session. The ratio of EPOC to net exercise VO₂ during the SIT (r = −0.61, P = 0.06) and the HIAT (r = −0.79, P < 0.01) correlated inversely with VO_2max, whereas, we observed no significant correlation between these two variables for the CAT (r = −0.42, P = 0.23). As for the correlation between absolute EPOC value and VO_2max, we observed a significant correlation for the HIAT (r = −0.67, P = 0.03), but no significant correlation for the SIT (r = −0.47, P = 0.17) and CAT (r = −0.34, P = 0.34).

Ohkawara et al. [5] showed a significant negative correlation between a subject’s fitness level and the EPOC under normal living conditions in a metabolic chamber. The significant correlation was observed not on a day with moderate-intensity physical activity but on a day with vigorous-intensity physical activity. Also, Singh et al. [6] recently showed that VO₂ decline during the first minute of recovery after maximum exercise was faster in children with a higher VO_2max. Our experimental results with a single bout of exercise were consistent with those studies’ results. That is, in the present study, while the ratio of EPOC to net exercise VO₂ correlated inversely with VO_2max, we observed this significant association when subjects performed a high-intensity interval exercise, especially an aerobic-type interval protocol (the HIAT), but not when they performed a moderate-intensity continuous protocol (the CAT). Although several studies [7–9] reported that training level did not significantly impact EPOC, the exercise intensities (30~70% VO_2max) used in those previous studies were relatively low compared to our interval exercise protocols.

While the precise mechanism of EPOC remains unclear, physiological adaptation associated with improved fitness levels has been considered a factor potentially influencing EPOC [1]. For example, an enhanced lactate metabolism with elevated body temperature partly accounts for EPOC [2, 10]. Trained individuals have better thermoregulatory capacities than untrained individuals because physical training enhances the sweating mechanism at a given level of the central sweating drive [11]. Therefore, elevated body temperature in untrained individuals could last longer than in trained individuals [12]. Moreover, subjects with lower VO_2max might produce more lactate than those with higher VO_2max especially during strenuous exercise. An enhanced lactate metabolism requires oxygen consumption for recovery. Thus, fitness level may contribute to the magnitude of EPOC.

In conclusion, we revealed that cardiorespiratory fitness level correlates inversely with the magnitude of EPOC, especially when performing an aerobic-type interval exercise.