Twenty-five young healthy sedentary or recreationally active men were recruited to participate in this study, with none engaged in a structured endurance training program. Subjects were randomly allocated to one of two parts of the study. Sixteen subjects (mean ± SD: 21 ± 2 y; 82 ± 17 kg; 1.83 ± 0.08 m; BMI: 23.7 ± 3.1 kg·m^-2; VO₂max: 48 ± 9 ml·kg^-1·min^-1) were allocated to the main part of the study, and completed the full experimental procedures. The remaining nine subjects (mean ± SD: 23 ± 8 y; 73 ± 9 kg; 1.78 ± 0.09 m; 23.0 ± 1.4 kg·m^-2; VO₂max: 47 ± 11 ml·kg^-1·min^-1) took part in a separate experiment to determine intra-individual variation in response to an oral glucose tolerance test, and did not perform HIT. There were no significant differences in the age, height, weight, BMI or VO₂max between the two groups of subjects. Subjects were informed of the experimental protocol both verbally and in writing before giving informed consent. Furthermore, all subjects were informed about how potential life-style changes could affect the results of the study, and were requested to maintain their normal diet and levels of physical activity (apart from the training program) throughout the duration of the study. The study protocol was approved by the institutional Ethics Committee and conducted in accordance with the Helsinki Declaration.

Baseline aerobic performance and health parameters were determined over a 2-week period prior to commencement of the training program.

Subjects refrained from performing any strenuous physical activity for 2 days prior to the OGTT, and attended the laboratory having fasted overnight. Venous blood samples were collected by venepuncture before, and 60, 90 and 120 min after ingestion of 75 g glucose (Fisher Scientific, Loughborough, UK) dissolved in 100 ml of water. Plasma was separated by centrifugation (10 min at 1600 g) and stored at -20°C until analysis of glucose, insulin and NEFA concentrations. Plasma glucose concentrations were measured using an automatic analyzer (YSI Stat2300, Yellow Spring Instruments, Yellow Spring, OH) and plasma insulin concentrations were determined by ELISA (Invitrogen, UK). Plasma insulin was measured only for samples taken at 0, 60, and 90 min. Plasma NEFA concentrations were determined by a colorimetric assay (Wako Chemicals, Germany) using a modified protocol. Briefly, 3.75 μl of plasma samples and standards of known concentration were pipetted into a 96-well plate. 75 μl of colour reagent A was added to each well and incubated at 37°C for 10 min. 150 μl of colour reagent B was added and incubated for a further 10 min at 37°C. The plate was then removed from the incubator and allowed to cool to room temperature prior to the absorbance being read at 550 nm. Coefficients of variation (CV) for duplicate samples were 3% for glucose, 5% for insulin, and 8% for NEFAs.

On a separate occasion, subjects performed an exhaustive incremental cycling test (Lode Excalibur Sport, Groningen, the Netherlands) to determine maximal power output (Wmax) and maximal oxygen uptake capacity (VO₂peak) using an online gas analysis system (SensorMedics, Bilthoven, the Netherlands). After cycling at 30 W for 1 min, power output was increased by 30 W·min^-1 until volitional exhaustion. VO₂peak was determined as the highest value achieved over a 20-s period.

Endurance performance was determined to provide an integrated physiological readout to facilitate comparison of the present study with previous studies which provided data from muscle biopsy samples 15. Subjects performed two self-paced cycling time trials in which they were instructed to complete 250 kJ of work as fast as possible. The linear factor was chosen to produce a power output corresponding to 75% of Wmax at a pedal rate of 90 rpm. No encouragement was given, and subjects were blinded from information on time, power output and pedal frequency. The amount of work (kJ) completed was called out every 25 kJ. Time trials were spaced at least two days apart. Although the first of the two pre-training trials was mainly used as a familiarisation trial, the fastest time achieved in the two trials was considered to best represent the pre-training performance level and used in the analysis (19 out of 25 subjects performed better in the second trial than in the familiarisation trial).

The sprint training protocol was similar to that used previously by Burgomaster et al. 414. Six sessions of sprint interval exercise were spread over 14 days, with 1 or 2 days of rest between each session. Each training session consisted of 4–6 repeated 30-s all-out cycling efforts against a resistance equivalent to 7.5% of body weight (Wingate tests), with 4 min of recovery between sprints. During recovery, subjects remained on the bike and either rested or cycled at a low cadence without resistance. The number of sprints increased from 4 during the first two sessions, to 5 in the third and fourth sessions, and 6 in the last two sessions. Total time commitment was 17–26 min per session, involving only 2–3 minutes sprint exercise.

A second OGTT was performed after completion of the training program. In order to determine whether potential changes were due to acute effects attributable to the last training session, subjects were tested either two (n = 10), or three (n = 6) days after the last bout of exercise. One day after the second OGTT subjects performed a third cycling time trial in order to determine changes in aerobic performance.

In order to assess normal intra-individual variation in response to an OGTT over a period of several weeks as used in the present study, nine subjects performed an identical protocol to the training group but without performing the six training sessions. Coefficients of variation (CV) for repeated measurements within individual subjects were determined for baseline glucose and NEFA levels, for glucose and NEFA area under the plasma curve, and for time trial performance.

Area under the plasma curve (AUC) was calculated using the conventional trapezoid rule. The Cederholm index, which represents peripheral insulin sensitivity 16, was calculated using the formula:

ISI_Cederholm = 75000 + (G₀-G₁₂₀) × 1.15 × 180 × 0.19 × BW/120 × G_mean × log (I_mean)

Where BW is body weight, G₀ and G₁₂₀ are plasma glucose concentration at 0 and 120 min (mmol·l^-1), and I_mean and G_mean are the mean insulin (mU·l^-1) and glucose (mmol·l^-1) concentrations during the OGTT.

All data are presented as means ± SEM. Plasma glucose, insulin, and NEFA responses to the pre-training and post-training OGTTs were analyzed using two-way repeated measures ANOVA with post hoc Student Newman-Keuls tests. Differences between pre-training and post-training data for time trial performance, AUCs for plasma glucose, insulin, and NEFA levels, and insulin sensitivity as measured by the Cederholm index were analyzed using paired sample t-tests. Pearson's correlation coefficient was used to assess bivariate correlations between baseline values of, and changes in the variables performance, AUCs of glucose, insulin, and NEFAs, and the Cederholm Index. Significance was accepted at P < 0.05.

Fasting plasma glucose concentrations were unaltered after 2 weeks of HIT. In the pre-training OGTT, plasma glucose concentration was elevated 60 min after the 75 g glucose bolus (Figure 1A; 0 min: 5.0 ± 0.1 v 60 min: 6.5 ± 0.4 mmol·l^-1; P < 0.0001), but not post-HIT (Figure 1A; 0 min: 5.0 ± 0.1 v 60 min: 5.0 ± 0.2 mmol·l^-1). The plasma glucose area under the curve (AUC) was significantly reduced post-training (Figure 1A: AUC pre 664 ± 103 v AUC post 585 ± 65 mmol·min·l^-1; P < 0.001).

Fasting plasma insulin concentrations were unaltered after 2 weeks of HIT. In the pre-training OGTT, plasma insulin concentration was elevated 60 min after the 75-g glucose bolus (Figure 1B; 0 min: 10.5 ± 1.6 v 60 min: 74.0 ± 8.9 mU·l^-1; P < 0.05). Post-training this increase was significantly attenuated (Figure 1B; 0 min: 10.6 ± 1.6 v 60 min: 42.6 ± 6.0 mU·l^-1; P < 0.01). Although plasma insulin levels were still elevated 90 min after the 75-g glucose bolus both pre- and post-training, levels were no longer significantly higher than at 0 min (pre: 38.8 ± 6.3 mU·l^-1; post: 27.7 ± 2.7 mU·l^-1). The plasma insulin AUC was significantly reduced post-training (Figure 1B: AUC pre 4226 ± 1912 v AUC post 2654 ± 1252 mU·min·l^-1; P < 0.001). Insulin sensitivity significantly improved following 2 weeks of HIT (Cederholm index: pre 80 ± 6 v post 98 ± 5 mg·l²·mmol^-1·mU^-1·min^-1, P < 0.01).

There was a trend towards a decrease in baseline plasma NEFA concentrations following HIT (Figure 1C; pre: 350 ± 36 μmol·l^-1 v post: 290 ± 39 μmol·l^-1, P = 0.058). In the pre-training OGTT, plasma NEFA concentration was decreased 60 min after the 75 g glucose bolus (Figure 1C; 0 min: 350 ± 36 μmol·l^-1 v 60 min: 255 ± 48 μmol·l^-1; P < 0.01), and to a greater extent post-HIT (Figure 1C; 0 min: 290 ± 39 μmol·l^-1 v 60 min: 153 ± 17 μmol·l^-1, P < 0.001; pre 60 min: 255 ± 48 μmol·l^-1 v post 60 min: 153 ± 17 μmol·l^-1 P < 0.05). The plasma NEFA AUC was significantly reduced post-training (Figure 1C: AUC pre 31584 ± 13205 v AUC post 23370 ± 8630 μmol·min·l^-1; P < 0.001), whereas despite the decreased insulin AUC post-training the incremental NEFA AUC was similar pre- and post-training (pre: -12748 ± 2752 μmol·min·l^-1; post: -13513 ± 2541 μmol·min·l^-1).

To bench-mark our results with recently published studies, we determined the impact of HIT on performance. Work done in the 250-kJ time trial was significantly increased by an average of 6% following 2 weeks of HIT (mean difference: 75 s, 95% CI: 21–126 s; P < 0.01). At baseline, the only physiological parameter which related to a metabolic parameter, was aerobic capacity vs. NEFA response to the OGTT (R² = 0.43, P < 0.001). There was a modest correlation between changes in glucose and insulin AUC (R² = 0.25, P < 0.05). Changes in performance did not correlate with baseline values for, or changes in the AUCs of glucose, insulin, and NEFAs, and insulin sensitivity as measured by the Cederholm index. We also considered whether the timing of the post-training assessment impacted on the observed metabolic changes. Improvements in glucose, insulin, and NEFA responses were similar whether assessed two or three days following the final training session (mean reduction in glucose AUC: 2 d post 12 ± 10%, 3d post 19 ± 4%; insulin AUC: 2 d post 34 ± 9%, 3d post 38 ± 9%; NEFA AUC, 2d post 23 ± 21%, 3d post 27 ± 9%).

In subjects not performing the HIT program, no significant differences between the first and second OGTTs were observed for plasma glucose (AUC: 671 ± 47 v 659 ± 40 mmol·min·l^-1) and NEFAs (AUC: 24035 ± 1611 v 22599 ± 2544 μmol·min·l^-1), nor for time trial performance (1477 ± 214 v 1491 ± 245 s). Coefficients of variation for the repeated measurements were 2.1% for the time trial, 4.9% and 7.0% for fasting glucose and NEFA concentrations, and 8.1% and 8.2% for glucose and NEFA AUCs respectively.