Endurance and Sprint Training Improve Glycemia and V˙O_2peak but only Frequent Endurance Benefits Blood Pressure and Lipidemia

Abstract

Purpose 

Sprint interval training (SIT) has gained popularity as a time-effective alternative to moderate-intensity endurance training (END). However, whether SIT is equally effective for decreasing cardiometabolic risk factors remains debatable, as many beneficial effects of exercise are thought to be transient, and unlike END, SIT is not recommended daily. Therefore, in line with current exercise recommendations, we examined the ability of SIT and END to improve cardiometabolic health in overweight/obese males.

Methods 

Twenty-three participants were randomized to perform 6 wk of constant workload SIT (3 d·wk⁻¹, 4–6 × 30 s ~170% W_peak, 2 min recovery, n = 12) or END (5 d·wk⁻¹, 30–40 min, ~60% W_peak, n = 11) on cycle ergometers. Aerobic capacity (V˙O_2peak), body composition, blood pressure (BP), arterial stiffness, endothelial function, glucose and lipid tolerance, and free-living glycemic regulation were assessed pre- and posttraining.

Results 

Both END and SIT increased V˙O_2peak (END ~15%, SIT ~5%) and glucose tolerance (~20%). However, only END decreased diastolic BP, abdominal fat, and improved postprandial lipid tolerance, representing improvements in cardiovascular risk factors that did not occur after SIT. Although SIT, but not END, increased endothelial function, arterial stiffness was not altered in either group. Indices of free-living glycemic regulation were improved after END and trended toward an improvement after SIT (P = 0.06–0.09). However, glycemic control was better on exercise compared with rest days, highlighting the importance of exercise frequency. Furthermore, in an exploratory nature, favorable individual responses (V˙O_2peak, BP, glucose tolerance, lipidemia, and body fat) were more prevalent after END than low-frequency SIT.

Conclusion 

As only high-frequency END improved BP and lipid tolerance, free-living glycemic regulation was better on days that participants exercised, and favorable individual responses were consistent after END, high-frequency END may favorably improve cardiometabolic health.

Lifestyle-induced chronic diseases represent a substantial cause of mortality in modern society. A growing number of individuals are classified as obese, which is associated with elevated blood pressure (BP), vascular dysfunction, and glucose and lipid intolerance, events directly linked to cardiovascular disease (CVD) and type 2 diabetes (T2D). Low cardiorespiratory fitness associated with these conditions is the primary and most powerful risk factor linked to mortality (¹) and can be prevented by performing regular exercise. Although moderate-intensity continuous endurance training (END) has classically been shown to increase aerobic capacity (V˙O_2peak), literature is accumulating to indicate that high-intensity interval training (HIIT) may be superior to END for improving this parameter (^2,3). However, a large range of HIIT protocols exist that vary in duration, intensity, and volume. Given the commonly cited barrier of “lack of time” to perform exercise (⁴), considerable interest has recently been placed on extremely low-volume, time-effective interval protocols known as sprint interval training (SIT). This approach involves short repetitions (10–30 s) of maximal exercise interspersed with periods of recovery (²). Despite the dramatic reduction in exercise volume, SIT appears comparable with traditional END for inducing skeletal muscle adaptations and increasing V˙O_2peak (^2,5,6). However, the role of SIT for improving cardiovascular and metabolic parameters associated with primary or secondary risk reduction of chronic conditions (CVD, T2D) remains less conclusive. For example, although SIT has been shown to improve glucose tolerance and insulin sensitivity, most consistently in type 2 diabetic or overweight individuals (^5,7,8), the beneficial effects of exercise for decreasing arterial stiffness or improving diastolic BP (DBP) are not always present after higher-intensity exercise (^7,9,10).

Moreover, although exercise intensity and duration are variables commonly manipulated between END and SIT protocols, the frequency of exercise is another important parameter as many cardiovascular and metabolic outcomes associated with exercise are transient in nature. For instance, postexercise improvements in BP (¹¹), insulin sensitivity (⁷), and postprandial lipid metabolism (¹²) appear to return to baseline beyond 24 h after a single exercise bout. Although END is conducive to daily sessions, exercise at both a high frequency and high intensity has been associated with unfavorable outcomes such as systemic inflammation (¹³) and overtraining-induced impairments in cardiac function, arterial stiffness, and performance (¹⁴). As a result, it is only recommended to perform SIT ~3 d·wk⁻¹ to allow adequate recovery between sessions (¹⁵). Consequently, the cumulative benefits of END training when performed with a high frequency, as per general guidelines (¹⁵), may be more effective with respect to disease risk reduction than infrequent SIT. However, many recent studies comparing exercise intensity have frequency-matched END and SIT protocols (⁵) or frequency- and volume-matched END and HIIT protocols (^16–19). Out of necessity, this has restricted END to shorter or less frequent bouts than what is optimal as indicated by general guidelines. In further support, when END is not frequency or volume matched, some reports indicate greater improvements in V˙O_2peak (¹⁰) and visceral fat mass (²⁰) in overweight individuals after END, factors that are linked to CVD risk. Therefore, we examined the ability of 6 wk supervised END (5 d·wk⁻¹) and constant workload SIT (3 d·wk⁻¹), which varied in frequency, intensity, and duration as per general recommendations, to improve clinically relevant cardiovascular and metabolic outcomes in overweight/obese males. Given the transient effects of exercise on these parameters, we hypothesized that END would elicit favorable adaptations as a result of the high-frequency exposure.

METHODS

Participants

Twenty-three overweight/obese males (37.4 ± 15.1 yr; body mass index [BMI], 34 ± 3.4 kg·m⁻²) were recruited. Before enrollment, participants were informed of potential risks and completed a health questionnaire, study screening, and a physical activity readiness questionnaire (PARQ+). Participants were eligible if overweight (BMI > 25 kg·m⁻²), insufficiently active (<100 min physical activity per week), nonsmokers, and between the ages of 18 and 70 yr. Participants were excluded if they smoked, were diagnosed with T2D, or took glucose-lowering medication. All other medications were permitted. Antihypertensive agents (n = 3 END; n = 1 SIT), cholesterol-lowering agents (n = 1 END; n = 2 SIT), antidepressants (n = 2 SIT), thyroid medication (n = 1 SIT), migraine medication (n = 1 END), arthritis medication (n = 1 SIT), and attention disorder medication (n = 1 SIT) were taken daily, and all subjects continued their previous medications throughout the study. All participants gave written informed consent to the study protocol, which was approved by the University of Guelph Human Research Ethics Board (REB no. 17-08-008) in accordance with the Declaration of Helsinki and registered as a human clinical trial (clinicaltrials.gov ID no. NCT03376685).

Experimental protocol

Participants completed two baseline visits before the exercise intervention, separated by at least 48 h. Introductory visits were conducted after a 12-h overnight fast and having refrained from caffeine, alcohol, or exercise for 24 h prior. One visit involved a dual-energy x-ray absorptiometry (DEXA) scan, insertion of a continuous glucose monitor (CGM), cardiovascular assessments (BP, pulse wave velocity [PWV], flow-mediated dilation [FMD]), and an oral glucose tolerance test (OGTT). The second visit involved analysis of fasted lipid profiles and postprandial responses to an oral fat tolerance test (OFTT). After 1 wk of baseline CGM recording, a V˙O_2peak test was conducted, and participants were randomized to either END or SIT in a parallel study design based on age, BMI, and V˙O_2peak. Neither researchers nor participants were blind to intervention group. All exercise training was 6 wk in duration and was completed using supervised in-laboratory cycle ergometers (Racermate Velotron, Seattle, WA). END involved 30–40 min at a power output corresponding to ~60% W_peak performed 5 d·wk⁻¹ (weeks 1–2, 30 min; weeks 3–4, 35 min; weeks 5–6, 40 min). SIT composed of 4–6 × 30 s at a power output corresponding to ~170% W_peak with 2 min recovery between intervals (50 W) and included a 3-min warm-up (50 W) and a 2-min cool-down (50 W). SIT was performed 3 d·wk⁻¹ (weeks 1–2, 4 repetitions; weeks 3–4, 5 repetitions; weeks 5–6, 6 repetitions) (Table 1). All absolute workloads remained constant throughout the training protocol, whereas the duration and/or number of repetitions were increased biweekly to ensure an increasing exercise stimulus. HR was tracked with Polar A300 watches/straps (Polar, Kempele, Finland), and general ratings of perceived exertion (RPE, Borg 6–20 scale) from the entire exercise bout were assessed weekly at the end of exercise. After the training intervention, a V˙O_2peak test was repeated. DEXA scans, cardiovascular assessments, and OGTT were performed ~48 h after the last exercise bout, whereas blood analysis and OFTT were performed ~72 h after the last exercise bout. All posttraining testing was completed with procedures identical to pretraining. Self-reported dietary intake was recorded for 3 d at the beginning of the study (baseline week) and for 3 d during the final week of exercise training (n = 17) using ESHA Food Prodigy (Salem, OR). Nontraining free-living physical activity data were collected on a subset of participants (n = 12) for 2–5 d during the baseline week and during the final week of training (ActivPal Technologies, Glasgow, Scotland, UK).

TABLE 1 - Summary of exercise training protocols.

	Endurance (5 d·wk −1 )			Sprint (3 d·wk −1 )
	Weeks 1–2	Weeks 3–4	Weeks 5–6	Weeks 1–2	Weeks 3–4	Weeks 5–6
Protocol	30 min at 114 ± 25 W (56% ± 3% W peak)	35 min at 118 ± 25 W (58% ± 3% W peak)	40 min at 118 ± 25 W (58% ± 3% W peak)	3 min warm-up (50 W); 2 min cooldown (50 W)
				4 × 30 s at 350 ± 81 W (168% ± 5% W peak)/2 min (50 W)	5 × 30 s at 359 ± 77 W (173% ± 12% W peak)/2 min (50 W)	6 × 30 s at 359 ± 77 W (173% ± 12% W peak)/2 min (50 W)
Daily time	30 min	35 min	40 min	13 min	15.5 min	18 min
Weekly time	150 min	175 min	200 min	39 min	46.5 min	54 min
Daily volume	~190 kJ	~220 kJ	~250 kJ	~70 kJ	~90 kJ	~110 kJ
Weekly volume	~950 kJ	~1100 kJ	~1250 kJ	~210 kJ	~270 kJ	~330 kJ
RPE	12.8 ± 1.1	12.3 ± 1.0	12.6 ± 1.3	16.9 ± 1.1*	16.9 ± 1.6*	16.5 ± 1.8*
Exercise HR (% max)	76.8 ± 4.2	78.3 ± 5.3	77.3 ± 4.8	84.2 ± 5.4*	84.7 ± 4.9*	83.5 ± 6.7*
Peak HR (% max)	84.3 ± 5.2	83.1 ± 6.1	84.2 ± 6.4	94.5 ± 8.1*	96.5 ± 4.4*	96.4 ± 4.8*

Data are analyzed using unpaired two-tailed Student’s t-tests between groups. W, HR, and RPE data are expressed as mean ± SD. n = 11 END; n = 12 SIT.

*P < 0.05 vs END for respective week.

HR, heart rate; RPE, rating of perceived exertion; W, watts; W_peak, power output (watts) at peak aerobic capacity.

DEXA

A whole-body DEXA scan (Christie InnoMed Inc., Mississauga, Canada) was performed for analysis of total lean mass, fat mass, and regional fat mass (Hologic APEX Version 3.2 Software, Bedford, MA). The android region was determined as the top of the iliac crest to 20% of the distance between the pelvic cut and the bottom of the neck line (²⁰). The gynoid upper boundary was below the pelvis by a distance of 1.5 times the android space and was equal to 2 times the height of the android region (²⁰).

Cardiovascular parameters

After 10 min of supine rest and three automated measures of resting BP (BPTru, VSM MedTech, Coquitlam, BC, Canada), arterial stiffness was measured via carotid-femoral PWV (¹⁴). Briefly, PWV was measured using Sphygmocor CPVH (AtCor Medical Ltd., NSW, Australia), calculated as PWV = distance (m)/transit time (s). Consistent wave forms were recorded for 10 s at each site using a high-fidelity Millar tonometer, and transit time was calculated automatically from the ECG-gated pulse waveforms. Two PWV measures were collected, and if the difference exceeded 0.5 m·s⁻¹, a third reading was taken. As per AHA guidelines, the average of two PWV measures, or the median of three PWV measures, was recorded.

After PWV, vascular endothelial function was measured via FMD using a reactive hyperemia model. The participant’s arm was rested in a foam guide with an ultrasound probe placed 2–5 cm above the antecubital fossa and an 11-cm occlusion cuff (D.E. Hokanson, Bellevue, WA) immediately distal. After 1 min of baseline, the cuff was inflated to 200 mm Hg (5 min) and then released for recovery (3 min) while a probe operating at 13 MHz for B-mode and 5 MHz for Doppler collected brachial artery diameter and blood velocity at the lowest possible insonation angle (≤60°). Images were stored on a computer at 60 Hz (DVI2USB 3.0 video grabber; Epiphan Systems Inc., Ottawa, Canada), and semiautomated offline analysis was used (Cardiovascular Suite, Quipu, Pisa, Italy). Mean blood velocity was calculated by adding antegrade and retrograde blood velocities together. Relative FMD was calculated as percent change in artery diameter from baseline (mean over 1 min) to peak 3-s average diameter. Shear rate was calculated as 4 × blood velocity/artery diameter, for each frame. Shear rate AUC (SR_AUC) was calculated as the entire hyperemic stimulus from cuff release to peak diameter.

Oral glucose and fat tolerance tests

A fasted capillary blood sample was taken via finger stick, and blood glucose was analyzed using a handheld glucometer (OneTouch Ultra2; LifeScan, Zug, Switzerland). Participants then consumed a 75-g glucose beverage (TRUTOL; Thermo Scientific, Waltham, MA), and capillary blood glucose measurements were performed at 30, 60, 90, and 120 min.

At a second baseline visit, an intravenous catheter was inserted into a forearm vein, and a baseline blood sample was drawn (5 mL serum separator tube, 3 mL K₂EDTA vacutainer, 4 mL K₂EDTA vacutainer). Participants then consumed a high-fat beverage composed of whipping cream, cocoa powder, and calorie-free sweetener, which provided 845 kcal, 90 g fat, 11 g carbohydrate, and 6 g protein. Blood samples were drawn at 1, 2, 3, 4, 5, and 6 h in 3-mL K₂EDTA vacutainers. Fasting serum tubes were allowed to clot for 30 min, centrifuged at 1734g (10°C) for 10 min (Beckman Coulter Allegra X-22R, Brea, CA), and analyzed for lipid profiles at a certified medical testing facility (LifeLabs, Guelph, Ontario, Canada). Fasting 4-mL K₂EDTA tubes were immediately stored at 4°C and analyzed (LifeLabs) for HbA1C. Fasting and hourly samples (3 mL K₂EDTA) were centrifuged at 1734g (4°C) for 10 min and stored at −80°C for in-house analysis of free fatty acids (FFA) and triglycerides (TAG) using commercially available kits (Wako Diagnostics, Mountain View, CA).

CGM

A CGM (FreeStyle Libre Pro; Abbott Diabetes Care, Wiesbaden, Germany) was inserted onto the back of the upper arm of participants and secured with an adhesive. A CGM was used to record ~7 d of free-living glycemia pretraining, and new sensors were inserted to record ~7 d during the final week of training. Participants were blind to glucose readings.

V˙O_2peak testing

Participants completed an incremental V˙O_2peak test on an electronically braked Velotron cycle ergometer, while HR (Garmin HRM1B, Kansas City, MO) and metabolic parameters were measured via breath-by-breath analysis using a Cosmed metabolic cart (Cosmed Quark CPET, Rome, Italy). V˙O_2peak values were determined as the highest reading over a rolling 30-s average, and power output corresponding to this point was recorded (W_peak).

Statistics

Postprandial OGTT and OFTT responses were calculated as the incremental total area under the concentration versus time curves (AUC), after subtracting baseline values (OGTT) or the lowest value reached (OFTT). Free-living 24-h glucose AUC was calculated as the total area under the concentration versus time curve for each day. Resting MAP was calculated as (2 × DBP + SBP)/3, where SBP is systolic BP and DBP is diastolic BP. All responses to training (pretraining vs posttraining) were analyzed using paired two-tailed Student’s t-tests within each group, with details listed in respective figure legends. We specifically aimed to examine the independent effects of each type of exercise (i.e., using paired Student’s t-tests) to determine the influence of END alone or SIT alone on cardiometabolic health parameters. We did not aim to compare between groups, as a study of this size is underpowered to detect an interactive effect between exercise modalities. Pearson correlations were used to compare the relationship between various cardiovascular parameters. Smallest robust changes for each parameter were determined as the pretraining SD (of both groups combined) multiplied by 0.2 (conservative small effect size) (²¹) to generate the training response overview. Individual change scores in each parameter were then compared with the calculated smallest robust changes in an exploratory nature to determine an increase, decrease, or absence of change in each variable posttraining. To provide 80% power to detect the expected increase (alpha of 0.05) in V˙O_2peak (+~3.5 mL·kg⁻¹·min⁻¹) and glucose tolerance (−~90 mmol·L⁻¹·min⁻¹ OGTT AUC) reported after exercise training protocols (^5,10,22,23), a sample size of ≥11 was required in each group. Data are expressed as mean ± SD, with P < 0.05 considered statistically significant.

RESULTS

Training and participant characteristics

Given the variation in frequency, intensity, and duration between END and SIT protocols as per physical activity guidelines, total weekly volume was approximately fourfold higher with END (Table 1). By contrast, RPE, average HR, and peak HR during exercise were significantly greater during SIT compared with END bouts (Table 1), confirming the high-intensity nature of the SIT protocol.

At baseline, participants randomized to perform END and SIT were matched with respect to age (35.3 ± 15.1 yr END; 39.4 ± 14.9 yr SIT), BMI (Table 2), and aerobic capacity (Table 3). Body weight and BMI did not significantly change after training; however, SIT increased total lean mass, whereas END decreased total, android, and gynoid fat mass (Table 2). Mean blood profiles were not significantly altered in either group (Table 2).

TABLE 2 - Anthropometric and biochemical parameters after 6 wk of END and SIT.

Variable	Endurance		Sprint		Statistics
	Pre	Post	Pre	Post	END	SIT
Weight (kg)	108.4 ± 11.6	107.4 ± 11.6	108.9 ± 13.0	109.9 ± 14.0	0.11	0.09
BMI (kg·m−2)	33.9 ± 2.4	33.6 ± 2.5	34.1 ± 4.3	34.3 ± 4.8	0.19	0.38
Body fat (%)	35.5 ± 4.7	34.7 ± 4.5*	34.7 ± 4.1	34.4 ± 4.5	0.026	0.33
Body fat mass (kg)	38.3 ± 8.5	37.1 ± 8.3*	37.4 ± 7.8	37.5 ± 8.8	0.027	0.84
Lean body mass (kg)	68.8 ± 5.9	68.9 ± 6.0	69.6 ± 7.9	70.5 ± 8.3*	0.86	0.034
Gynoid fat mass (kg)	6.52 ± 1.9	6.24 ± 2.0*	5.78 ± 1.5	5.83 ± 1.7	0.006	0.66
Android fat mass (kg)	4.20 ± 0.92	3.97 ± 0.85*	4.06 ± 0.79	4.07 ± 0.88	0.011	0.84
HbA1C (%)	5.36 ± 0.31	5.31 ± 0.32	5.61 ± 0.36	5.58 ± 0.34	0.11	0.52
TAG (mmol·L−1)	1.59 ± 0.54	1.35 ± 0.41	2.13 ± 1.49	2.63 ± 2.73	0.14	0.34
Cholesterol (mmol·L−1)	4.66 ± 0.76	4.39 ± 0.77	4.83 ± 0.79	4.82 ± 0.92	0.21	0.96
HDL-C (mmol·L−1)	1.09 ± 0.28	1.07 ± 0.26	1.04 ± 0.29	1.06 ± 0.29	0.73	0.71
Cholesterol/HDL ratio	4.49 ± 0.81	4.34 ± 0.88	4.93 ± 1.53	4.89 ± 2.04	0.14	0.89
LDL-C (mmol·L−1)	2.88 ± 0.51	2.73 ± 0.61	2.87 ± 0.54	2.74 ± 0.64	0.27	0.21
Non-HDL-C (mmol·L−1)	3.60 ± 0.58	3.34 ± 0.70	3.78 ± 0.81	3.76 ± 1.02	0.12	0.87
Hs-CRP (mg·L−1)	1.57 ± 1.41	1.32 ± 1.11	1.83 ± 0.98	1.58 ± 0.60	0.23	0.45

Data are analyzed using paired two-tailed Student’s t-tests comparing posttraining vs pretraining in each exercise group. Data are expressed as mean ± SD. For whole-body parameters, n = 11 END and n = 12 SIT; for blood parameters, n = 10–11 END and n = 10–12 SIT. Venous blood samples were unable to be obtained from one SIT participant posttraining, and blood lipid profile results were unable to be reported for one END participant and one SIT participant posttraining.

*P < 0.05 vs pretraining. BMI, body mass index; HbA1C, glycosylated hemoglobin; HDL-C, high density lipoprotein cholesterol; Hs-CRP, high-sensitivity C-reactive protein; LDL-C, low density lipoprotein cholesterol; TAG, triglycerides.

TABLE 3 - Cardiovascular parameters after 6 wk of END and SIT.

Variable	Endurance		Sprint		Statistics
	Pre	Post	Pre	Post	END	SIT
V˙O2peak (L·min−1)	3.05 ± 0.58	3.46 ± 0.49*	2.96 ± 0.66	3.14 ± 0.57*	0.001	0.02
W peak (W)	206 ± 48	246 ± 52*	209 ± 52	237 ± 53*	<0.001	<0.001
V˙O2peak (mL·kg body mass−1·min−1)	28.3 ± 5.8	32.1 ± 4.8*	28.5 ± 5.3	30.1 ± 4.7*	<0.001	0.04
V˙O2peak (mL·kg lean mass−1·min−1)	44.5 ± 8.4	50.3 ± 6.4*	43.2 ± 7.7	45.7 ± 6.4	0.001	0.09
SBP (mm Hg)	128 ± 13	119 ± 17*	123 ± 13	115 ± 7*	0.006	0.02
DBP (mm Hg)	80 ± 9	73 ± 12*	78 ± 6	75 ± 7	0.004	0.20
MAP (mm Hg)	96 ± 10	88 ± 14*	93 ± 8	88 ± 6	0.002	0.07
Resting HR (bpm)	66 ± 10	61 ± 9	69 ± 10	65 ± 9*	0.10	0.002

Data are analyzed using paired two-tailed Student’s t-tests compared posttraining vs pretraining in each exercise group. Data are expressed as mean ± SD. n = 11 END; n = 12 SIT.

*P < 0.05 vs pretraining.

DBP, diastolic blood pressure; HR, heart rate; MAP, mean arterial pressure; SBP, systolic blood pressure; V˙O_2peak, peak aerobic capacity; W_peak, power output (watts) at V˙O_2peak.

Cardiovascular risk factors

END and SIT training both increased W_peak as well as absolute and relative V˙O_2peak (Table 3). However, because SIT increased lean body mass (Table 2), only END increased V˙O_2peak relative to lean body mass (Table 3). END decreased both SBP and DBP and, as a result, elicited an approximately 10% reduction in MAP (Table 3). By contrast, SIT reduced only SBP, but not DBP or MAP (Table 3). Resting HR significantly decreased after SIT, while trended toward a reduction after END (one clear outlier is evident in the END group, (+12 bpm after training), and in the absence of this participant, a decrease in resting HR would be significant (P = 0.02) after END) (Table 3).

On a group level, PWV, an index of arterial stiffness, was not influenced by either training protocol (Fig. 1A); however, dramatic variability in responses existed after SIT. We therefore examined the potential influence of related CVD risk factors; however, PWV responses to training showed no correlation with baseline MAP (Fig. 1B), a common mediating variable. This was likely because participants were primarily normotensive (9/12 SIT, 7/11 END with MAP <97 mm Hg; Fig. 1B) (²⁴), so we next examined non-HDL cholesterol (non-HDL-C) as this parameter is a strong predictor of CVD risk and was elevated in several of our participants (7/12 SIT, 7/11 END with non-HDL-C > 3.4 mmol·L⁻¹) (²⁵). This revealed that changes in PWV after SIT, but not END, were positively correlated with baseline non-HDL cholesterol (Fig. 1C). This may suggest that cardiovascular responses to SIT are influenced by baseline participant characteristics related to CVD risk. By contrast, SIT, but not END, increased normalized FMD (Fig. 1D) and trended toward an increase in relative FMD (P = 0.10; Fig. 1E), suggesting improvements in vascular endothelial function after SIT. Although there were trends for SR_AUC to be lower after training (Fig. 1F), baseline brachial artery diameter was not altered after either END (3.97 ± 0.66 mm pretraining vs 3.99 ± 0.67 mm posttraining, P = 0.92) or SIT (4.22 ± 0.47 mm pretraining vs 4.23 ± 0.51 mm posttraining, P = 0.91).

FIGURE 1: Arterial stiffness and endothelial function after 6 wk of END and SIT. Although neither training protocol influenced PWV (A), dramatic variability in PWV responses to SIT existed. This was not related to baseline MAP (B), but the change in PWV with SIT was positively correlated with baseline non-HDL cholesterol (C). Reference values for MAP (²⁴) and non-HDL cholesterol (²⁵) are based on clinically established ranges. Only SIT increased normalized FMD (D) and trended toward an increase in relative FMD (E). One clear outlier is evident in the SIT group (~6% decrease in relative FMD, E), and in the absence of this participant, the increase in relative FMD would be significant (P = 0.0003) after SIT training. However, we have left this participant in our analysis. Neither training protocol influenced SR_AUC (F). END, moderate-intensity endurance training; FMD, flow-mediated dilation; MAP, mean arterial pressure; non-HDL-C, non-HDL cholestrol; PWV, pulse wave velocity; SIT, sprint interval training; SR_AUC, shear rate area under curve. Data are analyzed using paired two-tailed Student’s t-tests comparing posttraining vs pretraining in each exercise group. *P < 0.05 vs pretraining. Data are expressed as mean ± SD, and gray lines depict individual responses. For PWV in (A–C) n = 11 END, n = 12 SIT. For FMD (D–F), n = 11 END, n = 11 SIT. One SIT participant was excluded from all FMD analysis as a posttraining image was unable to accurately be obtained.

Postprandial lipid and glucose tolerance

Cardiovascular health is closely linked with lipid homeostasis. Indeed, we determined that END, but not SIT, decreased postprandial plasma TAG responses to a lipid tolerance test (OFTT, Fig. 2A,B). However, postprandial free fatty acid responses (Fig. 2C,D) were not altered after either END (P = 0.11) or SIT. Using a similar approach to examine postprandial glucose tolerance (OGTT), we determined that glucose AUC was decreased ~20% after both END and SIT (Fig. 2E,F) without changes in fasting glucose (P = 0.47 END, P = 0.57 SIT; data not shown).

FIGURE 2: Postprandial lipid and glucose tolerance after END and SIT. Postprandial plasma TAG responses during an OFTT were decreased after END (A) but not SIT (B). FFA responses were not altered after either END (C) or SIT (D). Glycemic responses to an OGTT were improved posttraining after both END (E) and SIT (F). AUC, area under curve; END, moderate-intensity endurance training; FFA, free fatty acids; OFTT, oral fat tolerance test; OGTT, oral glucose tolerance test; SIT, sprint interval training; TAG, triglycerides. Data are analyzed using paired two-tailed Student’s t-tests comparing posttraining vs pretraining in each exercise group. *P < 0.05 vs pretraining. Data are expressed as mean ± SD, and gray lines depict individual responses. n = 11 END and n = 10 SIT for OFTT parameters; n = 11 END and n = 12 SIT for OGTT parameters. Venous blood samples (OFTT) were unable to be obtained from one SIT participant posttraining, and one SIT participant was feeling ill on the posttraining testing day therefore was excluded from analysis.

Free-living glycemic regulation

Although an OGTT provides insight into standardized glucose homeostasis under controlled conditions, we next examined free-living glycemic regulation using CGM over several days pre- and posttraining (Fig. 3A,B). We determined that 24-h glucose AUC (Fig. 3C) and average (Fig. 3D) and peak (Fig. 3E) glucose levels were decreased during the final week of training after END (P = 0.03–0.04) and trended toward a decrease after SIT (P = 0.06–0.09). Given the variable frequency of training protocols, and the influence of daily exercise on glycemic regulation, we analyzed glucose parameters on exercise compared with rest days in the final week of training (Fig. 4A,B). This revealed that 24-h glucose AUC (Fig. 4C) and average glucose levels (Fig. 4D) were lower on exercise days in both END and SIT groups, whereas peak glucose levels were not altered (Fig. 4E). Although dietary intake was not standardized at this time, self-reported caloric intake or the percentage of calories from carbohydrates, fat, and protein did not differ after END or SIT (see Supplemental Table 1, Supplemental Digital Content 1, Depicting dietary and physical activity controls, https://links.lww.com/MSS/C228). In addition, free-living nontraining physical activity was similar pre- and posttraining in a subset of END and SIT participants (Supplemental Table 1, Supplemental Digital Content 1, https://links.lww.com/MSS/C228).

FIGURE 3: The influence of 6 wk of END and SIT on free-living glycemic regulation in the final week of exercise training. Representative tracings of daily glycemic regulation over an approximately 7-d period pretraining and during week 6 of training are shown for END (A) and SIT (B). END reduced 24-h glucose AUC (C), average daily glucose levels (D), and peak daily glucose levels (E), while these parameters trended toward a reduction after SIT. AUC, area under curve; END, moderate-intensity endurance training; SIT, sprint interval training. Data are analyzed using paired two-tailed Student’s t-tests comparing week 6 vs pretraining in each exercise group. *P < 0.05 vs pretraining. Data are expressed as mean ± SD, and gray lines depict individual responses. n = 10 END, n = 9 SIT. CGM readings were unable to be obtained in n = 1 END and n = 3 SIT participants due to sensor fault.

FIGURE 4: Free-living glycemic regulation on rest compared with exercise days during the final week of END and SIT. Representative tracings of daily glycemic trends on exercise days vs rest days are shown for END (A) and SIT (B). The 24-h glucose AUC (C) and the average daily glucose (D) were improved on exercise compared with rest days, whereas peak glucose levels (E) were not altered. AUC, area under curve; END, moderate-intensity endurance training; SIT, sprint interval training. Data are analyzed using paired two-tailed Student’s t-tests comparing exercise vs rest days in each exercise group. *P < 0.05 vs rest days. Data are expressed as mean ± SD, and black lines depict individual responses. n = 10 END, n = 9 SIT. CGM readings were unable to be obtained in n = 1 END and n = 3 SIT participants due to sensor fault.

Individual participant responses

Given our abundance of clinically relevant cardiovascular and metabolic parameters, we next examined individual responses in an exploratory nature. When all parameters were combined, positive responses appeared more abundant after END than SIT (Fig. 5A). Furthermore, every individual in the SIT group had at least two risk factors that moved in an unfavorable direction after training. On a group level, although V˙O_2peak was improved after both END and SIT, this response occurred in all END individuals but not all SIT individuals (Fig. 5A,B). Similarly, more consistent beneficial responses with less variability were evident after END when examining DBP (Fig. 5A,C), glucose tolerance (Fig. 5A,D), and non-HDL cholesterol (Fig. 5A,E). Altogether, on an individual level, high-frequency END may elicit more consistent favorable responses in clinically relevant risk factors.

FIGURE 5: Exploratory analysis of individual responses in cardiometabolic parameters after 6 wk of END and SIT. Overall, END appeared to elicit more favorable and consistent improvements in clinically relevant parameters. A, White boxes represent an improvement in respective parameter, gray boxes represent no change, black boxes represent a negative response, and a dash represents data unable to be obtained. Responses were compared with a threshold of the smallest robust change (SRC) calculated as the SD of pretraining values (in both groups combined) multiplied by 0.2 (²¹). B–E, Gray shading depicts the SRC for each variable. AUC, area under curve; CGM, continuous glucose monitor; DBP, diastolic blood pressure; END, moderate-intensity endurance training; FFA, free fatty acids; FMD, flow-mediated dilation; HbA1C, glycosylated hemoglobin; MAP, mean arterial pressure; non-HDL-C, non-HDL cholesterol; OFTT, oral fat tolerance test; OGTT, oral glucose tolerance test; PWV, pulse wave velocity; SIT; sprint interval training; SBP, systolic blood pressure; TAG, triglycerides; V˙O_2peak, peak aerobic capacity. Change scores calculated by subtracting posttraining values from pretraining values for each individual. Group statistics analyzed using paired two-tailed Student’s t-tests comparing posttraining vs pretraining in each exercise group for all parameters.

DISCUSSION

In the current study, we demonstrate that when performed according to physical activity guidelines (¹⁵), both END and SIT improved V˙O_2peak and glucose tolerance. However, glucose homeostasis was better on exercise compared with rest days. In addition, only END improved body fat, BP, and lipid tolerance, and there were fewer negative responses to END. Altogether, these data suggest that high-frequency END may favorably improve whole-body cardiovascular and metabolic outcomes associated with clinically relevant disease risk reduction.

Although both END and SIT increased V˙O_2peak, all individuals responded positively to END but not SIT. This aligns with previous literature determining substantial variability exists in V˙O_2peak responses to SIT performed 3 d·wk⁻¹ (²⁶) and increasing exercise frequency, or volume, attenuates V˙O_2peak nonresponse (^26–28). It is worth noting that our groups did not perform equal amounts of overall work, and thus a differential response is perhaps logical. Similar to our findings, Fisher et al. (¹⁰) determined greater improvements in V˙O_2peak after high-frequency END (5 d·wk⁻¹) compared with low-frequency SIT (3 d·wk⁻¹), suggesting exercise frequency and/or volume influences the magnitude of V˙O_2peak responses. In light of this, the design of studies limiting END to low frequencies of SIT (^5,23) or the volumes of HIIT (^9,19) determining comparable or superior improvements in V˙O_2peak with SIT/HIIT likely underrepresent the beneficial effects of END when performed as per general high-frequency guidelines. Although a recent meta-analysis suggests that low-volume HIIT is superior to END for improving V˙O_2peak (³), this included both HIIT and SIT protocols, and END protocols were generally performed at an equal frequency to SIT/HIIT (i.e., END and SIT/HIIT 3 d·wk⁻¹). Although the present findings that END, but not SIT, consistently improved V˙O_2peak may therefore appear controversial, previous work has also suggested that END is more advantageous for improving V˙O_2peak when END is performed per general high-frequency–based practice (¹⁰). Nevertheless, in the present study, our SIT protocol was clamped at a set intensity (~170% W_peak) throughout the training period. It is possible that the all-out nature of Wingate-based SIT protocols is important for increasing V˙O_2peak, which may have limited the responses observed in our SIT group. Although this remains a possible limitation, several previous reports performing 6 wk of SIT at a similar intensity (7–8 × 20 s at 170% V˙O_2peak with 10 s recovery) (^23,29) have determined substantial increases in V˙O_2peak after SIT despite the non–all-out nature of the SIT protocol. Although SIT-related improvements in V˙O_2peak in our study appear of lower magnitude than those in previous literature (^23,29), this may be due to the influence of individual response/nonresponse in a modest size experimental group, as is typical for intervention-based exercise studies (i.e., n = 6–12). It is known that substantial variability exists in V˙O_2peak responses to exercise training (^26–28,30), and therefore wide-spread studies are needed to determine whether one exercise protocol, performed as per general practice, is “superior” for improving aerobic capacity.

With respect to BP, we determined that END improved all BP parameters, although SIT did not decrease DBP or MAP. Transient postexercise hypotensive responses are accepted to persist ~24 h (¹¹), and thus, we measured BP ~48 h postexercise to capture chronic adaptations. However, other studies have determined an improvement in BP 24 h after 2 wk of SIT, which was normalized to pretraining levels at 72 h (⁷). It is therefore possible that END, but not SIT, could be related to structural changes mediating a persistent decrease in BP. As we determined that resting HR decreased significantly only after SIT, changes in vascular resistance or stiffness may be more likely in response to END. Indeed, END is known to decrease arterial stiffness, which is indicative of positive vascular remodeling (^9,22) and generally represents a prolonged chronic adaptation (³¹). By contrast, the influence of SIT on central arterial stiffness is less well established, as SIT has been shown to both improve (²²) and not alter (⁷) central PWV in young obese males. In addition, HIIT did not improve carotid artery compliance in older individuals whereas END did (⁹). In the current study, although PWV was not altered after either END or SIT, which could be related to the 6-wk duration of training, a dramatic range of individual responses to SIT existed. Physiologically, it is possible that the underlying mechanisms leading to transient increases in catecholamines (³²) and arterial stiffness (³³) after supramaximal Wingate exercise (30 s all-out) could lead to vascular stiffening and counteract the beneficial effects of exercise. The effects of SIT in those at risk for CVD warrants further investigation, particularly as a 1-m·s⁻¹ increase in PWV (in 3/12 of our SIT participants, which was positively correlated with baseline non-HDL cholesterol) is associated with a 14%–15% increase in CVD risk (³⁴).

As blood lipids are associated with CVD risk and lipid-lowering agents can improve BP and arterial stiffness in hypertensive, normocholesterolemic individuals (³⁵), a strong link exists between lipids and cardiovascular function. Exercise frequency may also be important for lipid homeostasis as previous literature has determined that SIT performed 3 d·wk⁻¹, but not 2 d·wk⁻¹, was effective at decreasing blood lipids and body fat (³⁶). With respect to postprandial lipid metabolism, improvements are transient after END or SIT exercise (^12,37) and, therefore, contentious in response to training (³⁸). However, we observed a reduction in TAG AUC with END but not SIT. As a potential mechanism explaining these differences, acute exercise transiently increases gene transcription (³⁹), and therefore repeated stimuli of high-frequency END may be beneficial for upregulating proteins involved in lipid utilization. In support, increases in mitochondrial density appear more robust after high-frequency END (5 d·wk⁻¹) compared with low-frequency SIT (3 d·wk⁻¹) (⁴⁰). Similarly, exercise is capable of transiently increasing lipoprotein lipase mRNA, content, and activity to cleave circulating TAG for uptake into peripheral tissues (^41,42), suggesting that high-frequency END may be superior. Lipid mobilization from adipose tissue is also increased to a greater extent with END compared with SIT given the higher rates of fat oxidation (⁴³), which is linked with improved adipose homeostasis and, as a result, reductions in CVD risk. As we determined that only END decreased total and regional fat mass, high-frequency END appears favorable for improving whole-body lipid homeostasis, similar to our BP findings. By contrast, we observed an improvement in resting endothelial function after SIT but not END. Physiologically, blood flow and shear stress would be increased to a greater extent during a bout of SIT compared with END, which likely is linked to differences in vasodilatory capacity given that FMD testing models this stimulus (¹⁶). This is in line with previous literature suggesting that improvements in FMD are often intensity dependent (¹⁶) and would suggest that, in contrast to our hypothesis, exercise intensity, not frequency or volume, may be more important for endothelial function assessed by FMD. As FMD responses to training did not correlate with changes in PWV (r = 0.15, P = 0.65 SIT; r = −0.22, P = 0.51 END) or MAP (r = 0.34, P = 0.30 SIT; r = 0.38, P = 0.24 END), this suggests that different mechanisms/characteristics of training are influencing vascular structure and function.

Given that exercise is a potent stimulus to improve glucose tolerance, we examined glycemic regulation using two different approaches. To specifically determine the chronic influence of exercise, we performed an OGTT ~48 h after the final training bout and determined that both END and SIT improved OGTT responses. This suggests that all exercise performed in this study was beneficial for certain aspects of glucose tolerance. Mechanistically, the increase in lean mass we observed after SIT could be related to increased glucose disposal (and therefore improved whole-body glucose tolerance). By contrast, in the absence of changes in lean mass with END, improvements in glucose tolerance could be related to decreased adiposity. This notion is further supported by our findings that only high-frequency END improved postprandial lipid homeostasis, which has implications for glycemic regulation as lipids are associated with insulin resistance (⁴⁴).

To examine glycemic regulation in a free-living setting, we used CGM to determine dynamic glycemic trends during the final week of exercise training, which is influenced by both acute (i.e., individual bouts performed within the week) and chronic (i.e., cumulative training over 6 wk) effects of exercise. To date, studies monitoring glycemic regulation using CGM have been acute in nature, being 24 h after a single exercise bout (⁴⁵), or 1–3 d after training with standardized dietary conditions (^46,47), and determining improvements with both END and SIT in T2D individuals. Although the chronic influence of END and SIT on free-living glycemia without strict laboratory standardization has not been examined, we reasoned low-frequency SIT would be less effective for long-term glycemic regulation given the transient effects of exercise on glucose homeostasis. We determined that free-living glycemic parameters were significantly improved after END and only trended toward an improvement after SIT (P = 0.06–0.09). Although the absence of significant changes after SIT is likely related to the low frequency of exposure, we also acknowledge that nutritional factors and differences in energy deficit could dramatically influence glycemic regulation in our nonstandardized approach. Although self-reported dietary intake in our study indicates the absence of overt changes after training, we do not have the resolution to examine how other aspects of dietary consumption (i.e., glycemic index, timing of meals in relation to exercise, and mixed composition of meals) influences glycemic regulation. These complexities of dietary patterns have profound influences on glucose homeostasis. This remains a subject of future research to understand the optimal patterns of exercise type and timing, and dietary intake, for improving glycemic regulation assessed by CGM. In addition, the lower sample size in these measures and the relatively controlled baseline glycemic parameters in both groups may decrease our ability to detect small improvements in glycemic regulation. Regardless, glycemic regulation was better on exercise compared with rest days in both groups, highlighting that daily exercise is important for improving aspects of glycemic regulation. Although the energy deficit induced by an acute bout of END would be greater than SIT, the rapid glycogen depletion during SIT (⁴⁸) could improve glucose regulation on exercise days despite the lower within-bout volume. Although mechanistic insight into the acute influence of SIT on glycemic regulation remains to be directly determined, our data nonetheless suggest that repeated daily exercise is important for improving aspects of glycemic regulation.

Limitations and future research

While designed to examine the efficacy of END and SIT as per general guidelines (¹⁵), our exercise protocols varied in frequency, intensity, duration, and volume. As a result, we are unable to directly differentiate the influence of frequency from other parameters, which remains a subject of future research. Although comparing between low- and high-frequency exercise within a single type of protocol (END or SIT) could provide more direct insight into the role of frequency, we specifically aimed to examine the efficacy of these protocols according to general practice guidelines. Therefore, the ecological validity of this model contributes substantially to the understanding of these variables and their implications for whole-body physiology and exercise prescription. Practically, it remains possible that performing moderate-intensity exercise 5 d·wk⁻¹ is not feasible for some individuals, particularly given the time commitment required. Therefore, as SIT provides a time-efficient benefit, we cannot dismiss the potential importance of SIT in certain populations. It is therefore logical to consider a combined END–SIT regime to maximize exercise frequency and health outcomes.

An obvious limitation when comparing SIT within the literature is the broad range of interval protocols used. Our SIT protocol (4–6 × 30 s at ~170% W_peak with 2-min recovery at 50 W) was performed at a clamped intensity, similar to some previous literature (^23,40) but in contrast to other all-out SIT protocols that naturally increase in intensity over time (⁵). We elected to use a fixed workload SIT protocol to provide a similar, constant, and objective stimulus between intervals/bouts and between participants. In addition to the range of SIT intensity and duration within previous literature (i.e., 170% W_peak—all-out intensity; 20–30 s interval duration), the recovery period between intervals also displays variability (i.e., 10 s to 4 min). The recovery period between SIT intervals is likely important for chronic adaptations (⁴⁹) and metabolic responses to an acute bout (⁵⁰), as oxygen consumption has been shown to peak in the initial 20-s recovery after SIT before rapidly decreasing by 2 min of recovery (⁵⁰). However, it has yet to be determined whether an optimal recovery interval exists when performing SIT. Altogether, substantial variability exists within SIT protocols, and our data, in addition to others examining nonresponse to SIT protocols of varying frequency (²⁶), would suggest that exercise frequency is also an important parameter to consider. In addition to frequency, exercise volume was approximately fourfold higher with END compared with SIT, which could be influencing cardiometabolic outcomes. Although it is not feasible to volume-match between low-volume SIT protocols and END, many previous studies have matched volume between HIIT and END (²). As a result, this likely limits exposure to END in comparison with optimal guidelines of volume and frequency.

Our participants were classified as overweight/obese; however, several other metabolic parameters (glucose tolerance and BP) were relatively controlled at baseline. As a result, our findings cannot define the efficacy of END or SIT as a treatment approach but instead have important implications for understanding exercise as a preventative measure. Our relatively small sample size (n = 11 END, n = 12 SIT) and age range of participants also represents a limitation, particularly given the substantial variability in individual responses that exist in many cardiometabolic parameters after exercise (³⁰). When examining cardiometabolic parameters on an individual level, we did not have the ability to perform repeated testing or compare the variability in training response to that of a control group. Therefore, our exploratory depiction of individual responses does not account for biological and technical error. This warrants future investigation to determine optimal individual exercise recommendations.

CONCLUSION

We determined that low-volume SIT increased V˙O_2peak, glucose tolerance, and endothelial function, findings similar to previous literature (²). However, a major caveat of previous comparisons between END and SIT is the low-frequency of exposure to END compared with what is optimal as indicated by general guidelines (¹⁵). In the current study, when exercise protocols were performed as per general guidelines (high-frequency END, low-frequency SIT), it appears END may be favorable for reducing cardiovascular risk factors, as only high-frequency END improved BP and lipid parameters. Furthermore, the incidence of nonresponse in certain parameters (V˙O_2peak, glycemic regulation, and lipid parameters) was less prevalent after END. Altogether, when all parameters are considered, our data provide evidence that high-frequency END may be more beneficial for improving whole-body metabolism and decreasing cardiometabolic disease risk. This suggests that exercise frequency is a parameter that should be considered between END and SIT protocols.