High intensity interval training (HIIT) has been widely studied in athletic and non-athletic populations, demonstrating considerable positive benefits, such as, improved training time efficiency relative to increases in aerobic and anaerobic capacity or maintenance of endurance performance during low-volume training periods (Billat et al., 2001; García-Pinillos et al., 2017; Gunnarsson and Bangsbo, 2012; Lindsay et al., 1996). HIIT comes in many different forms and aims to improve distinct sport performances (Laursen and Jenkins, 2002). A number of authors have investigated the optimal HIIT stimulus for improvements in endurance performance such as the intensity and duration of sessions (Etxebarria et al., 2014; Gunnarsson and Bangsbo, 2012; Laursen and Jenkins, 2002; Stepto et al., 1999). Despite the fact that HIIT appears to enhance a number of physiological variables, such as, maximal oxygen consumption (VO₂max), peak power output (PPO), time to exhaustion at maximal velocity, first and second ventilatory threshold and vertical jumping performance (García-Pinillos et al., 2017; Helgerud et al., 2007; Laursen et al., 2002; Laursen and Jenkins, 2002; Mallol et al., 2018) the majority of research examining the effects of HIIT has been conducted with male athletes (Billat et al., 1999; Laursen et al., 2002; Lindsay et al., 1996) or with mixed male and female athletes sample (Farley et al., 2016; Koral et al., 2018; Mallol et al., 2016; Menz et al., 2015) while a group of exclusively female participants is less frequently studied. Some studies have reported less beneficial outcomes to HIIT training in females compared to males and hypothesized this may be due to a greater disposition to aerobic metabolism in females compared to males (Gibala et al., 2014; Gratas-Delamarche et al., 1994). Previous authors supported that gender might be a differentiating element in endurance performance, such as cross-country skiing, cycling and marathon running (Gibala et al., 2014; Gratas-Delamarche et al., 1994). Other studies have concluded that VO₂max, maximal aerobic power, power at lactate threshold (LT), power at onset blood lactate accumulation (OBLA) and peak of speed levels showed significant differences between genders. Such that the male subjects obtained greater values (Hopker et al., 2010; Reaburn et al., 2011; Sandbakk et al., 2012). Additionally, anthropometric and morphological variables such as lipid accumulation may also be an influential factor (Reaburn et al., 2011).

The limited studies on female athletes have shown greater improvements in aerobic and anaerobic capacity after a HIIT program compared with continuous training in female soccer players (Rowan et al., 2012). The authors emphasised the time saving benefits obtained from HIIT sessions, helping to focus on teamwork and sport specific skills (Rowan et al., 2012). Kinnunen et al. (2017) found that a HIIT program applied in pre-season helped to enhance the maximal and explosive strength capacity, improving neuromuscular performance in female ice-hockey players. Recently, Mallol et al. (2018) observed short supramaximal sets of HIIT enhanced maximal VO_2max values and submaximal power [W] at the first and second ventilatory threshold capacity. However, no performance improvements were observed during a triathlon race simulation (Mallol et al., 2018). The context of the study was in relation to maintenance of fitness during periods of reduced training. There is minimal research examining the effects of HIIT in females. Therefore, further research is needed to achieve a greater understanding of this training method in females.

Previous researchers concluded that an intensified period of training in running results in a higher level of cumulative fatigue, greater muscle damage and potential injury in runners (Burt et al., 2012; Del Coso et al., 2013). An intensified training program, high volume and/or intensity, presents a demanding stimulus requiring careful planning and monitoring. Running can lead to higher levels of muscle damage and cumulative fatigue owing to eccentric muscle contractions which is evidenced by higher biochemical and perceptual markers of muscle damage and soreness such as creatine kinase (CK) (Burt et al., 2012; Cipryan, 2017; Keane et al., 2015; Quinn and Manley, 2012). In particular, muscle damage is typically observed in the periods after HIIT sessions. Elevated levels of CK, myoglobin and lactate dehydrogenase were induced by 15 s, 30 s and 60 s HIIT running protocols (Cipryan, 2017). Moreover, the muscular eccentric component of down-hill running training involved higher CK values and greater exercise induced inflammatory responses in female runners (Köhne et al., 2016). In female sports teams such as soccer, rugby and netball, female athletes presented an increased level of muscle damage after high intensity sprint training (Le Meur et al., 2011). Intensified HIIT in cycling, which is a more concentric based activity than running, may not result in the same cumulative level of soreness. Therefore, the current study focused on the comparing the residual effects of HIIT in cycling and running modalities. Additionally, Burt et al. (Burt et al., 2012) have presented data showing that different levels of exercise-induce muscle damage were evident following running and cycling exercises. However, as a general form of training HIIT in cycling might nonetheless induce performance changes similar to those shown following running HIIT programs (Burt et al., 2012; Millet et al., 2002) but this has not been extensively investigated. Several studies in well trained runners have observed positive effects on running performance when a part of their run training volume was replaced by cycle training sessions (Etxebarria et al., 2014; Tanaka, 1994; White et al., 2003). Overall, relatively few investigations have been conducted with female athletes in order to examine how an intensified running training program affects CK, as a biomarker of muscle damage. There is also limited research showing how exercise induced CK varies between running and cycling HIIT modes after a period of HIIT in running or cycling.

Therefore, the current investigation focused on examining the difference in physiological responses, performance outcomes and muscle damage, as acute effects, occurring between run and cycle HIIT modes in female athletes.

A group of fourteen recreational middle-aged female athletes (age = 42 a 10 yr, height = 1.67 0.06 m, body mass = 61.6 10.4 kg, body mass index (BMI) = 22.2 3.4 kg.m^-2) were recruited from a number of community clubs and institutions. Participants were randomly distributed in two groups: a running HIIT group (HIIT_run, n = 7, age = 41, 7 yr, height = 1.64 0.07 m, body mass = 60.7 9.3 kg, BMI = 22.6 2.3 kg.m^-2) who completed two run HIIT sessions per week, and a cycling HIIT group (HIIT_bike, n = 7, age = 43, 13 yr; height = 1.70 0.03 m; body mass = 62.5 12.1 kg; BMI = 21.8 4.5 kg.m^-2) who performed an identical HIIT session protocol on the cycle ergometer. The inclusion criteria were: participants were habitual and active runners, ≥ two running sessions per week, and were able to run 10 km in < 70 min. Participants were excluded if they had no running training in the previous one month, or had an injury that prevented them from participating in training or testing.

Participants were informed of the protocols and experimental procedures and signed a formal written consent. The study followed the guidelines established by the Declaration of Helsinki (2013) and was approved by the Human Research Ethics Committee (HREC code 334.16).

The study examined the physiological and performance benefits, as well as, the muscle damage generated by four weeks of HIIT using running or cycling in female runners. Participants attended Flinders University Exercise Physiology Laboratory twice weekly at the same time of day, to perform the HIIT sessions. To determine differences between the HIIT programs, participants were divided into two groups: HIIT_run who performed supervised running HIIT sessions on an outdoor grass running track and HIIT_bike who performed supervised cycling ergometer HIIT sessions in the laboratory. All participants performed identical testing procedures before and after the four-week intervention period. Each subject undertook the initial and final tests (laboratory testing and 10 km running time trial) at approximately the same time of day and were asked to follow a similar protocol for test preparation. The environmental conditions in the laboratory were maintained between 20-22ºC and 55-65% humidity.

Each participant completed a ‘fast’ incremental exercise test to exhaustion using treadmill running in the week prior to and a week after the training intervention. The test procedure included (a) 10 minutes of their usual warm-up intensity on the treadmill (b) an incremental running test where the initial stage was set at 8 km·h^-1. Each stage lasted one min and the speed was increased by 1 km·h^-1 until exhaustion (Noakes et al., 1990). Expired gases were analysed by TrueOne2400 (ParvoMedics, Utah, USA), to determine maximal oxygen consumption (VO_2max) on a treadmill, the gas was analysed every 5 seconds but the 2 highest consecutive values over 30s was used. Before the warm up, blood lactate concentration was obtained from a fingertip sample and analysed using a portable lactate analyser (Lactate Pro, Arkray, KDK Corporation, Kyoto, Japan). Heart rate (HR), maximal speed achieved during the last stage completed (speed_max) and rating of perceived exertion (RPE) were recorded in the final 15 s of every stage. HR was recorded using Polar RS400 series (Kempele, Finland) HR monitor until test completion. The Borg Scale (from 0 to 10) (Borg, 1982) was employed to monitor the RPE at the end of each stage. Participants were previously familiarised to the perceived exertion method and the 10- point Borg Scale.

After a 60 min break from the incremental running test the participants completed a 10 km TT individually on a 400 m grass running track. Distance was previously measured and marked on the track by the researchers, participants were asked to complete 25 laps. Average and maximum speed (speed_av and speed_max), and HR (HR_av and HR_max) were recorded (Garmin Forerunner 910XT Olathe, Kansas, USA). RPE average (RPE_av) was calculated from values obtained every 2 km throughout the trial. Instantaneous pace and HR were recorded every 2 km. Immediately after the test, lactate concentration was measured from a fingertip sample (Lactate Pro, Arkray, KDK Corporation, Kyoto, Japan).

Participants randomly allocated to the HIIT groups (HIIT_run and HIIT_bike) attended supervised training sessions twice a week. The session structure was similar for all HIIT training: a 10 min warm-up where participants determined their warm-up intensity from the maximal test, followed by 6 x 2 min at 95% of maximal heart rate (HR_max) and 4 x 1 min all out (Bogdanis et al., 1996) followed by a 5 min cool-down. The recovery periods for the 1 and 2 min interval sets were 1 min 30 s and 2 min, respectively, with athletes continuing active recovery at a low intensity. HIIT_run performed the training outdoors around the same track as used for the 10 km TT. Individual training intensity at 95% of HR_max and maximal intensity for the HIIT_run group were determined for each participant based on their pre-intervention testing. The HITT_bike sessions were performed using a cycle ergometer (Wattbike, Nottingham, UK). The HIIT_bike intensity were extrapolated from the treadmill test based on Millet et al. (Millet et al., 2009) review were researchers concluded that HR_max obtained from a maximal cycle ergometer test is about 5% (between 6-10 bpm) lower than HR_max recorded in a maximal treadmill test. The intensity was also corroborated with RPE values (Basset and Boulay, 2000; Norton et al., 2010) which showed a similar pattern for HR during cycling interval training (Green et al., 2006). Participants could change the cycle resistance as they needed as long as they achieved the stipulated HR. In order to recognise their usual training load, they were asked to complete a 1-week training diary before the intervention to ensure that no significant differences existed between participants. During intervention weeks participants recorded their individual training sessions outside of the HIIT program. Every session was documented in a personal training diary which included the running training duration, distance and RPE_av allowing for the session RPE-min to be calculated (Table 1).

The physiological responses to the first (1^st) and last HIIT (8^th) session in each mode (HIIT_run and HIIT_bike) were measured. Blood lactate (Lactate Pro Analyser, Arkray, Japan) and CK concentrations (Reflotron Plus system, Rotkreuz, Switzerland) were measured before and immediately after the training session from fingertip blood samples. Participants came back to the laboratory 24 h after the HIIT sessions in order to measure CK_24h (Quinn and Manley, 2012). Delayed onset muscle soreness (DOMS) was recorded using a CR-10 scale (Lau et al., 2015) before, after and 24 h after session. During each HIIT_run session the average and peak HR, HR during recovery intervals, pace, distance covered and average/maximal RPE values were recorded (Borg, 1982). During the HIIT_bike average and maximal HR, recovery HR intervals and average and maximal RPE values (Borg, 1982) were recorded. Speed, power and cadence average, maximal power and cadence were also recorded for the HIIT_bike group.

Results are presented as mean ± SD. A t-test for independent samples was used to analyze the differences between HIIT_run and HIIT_bike at baseline (pre-test). The between-group (HIIT_run vs HIIT_bike) comparison from pre-test to post-test or 1^st and 8^th HIIT sessions for data obtained in the laboratory tests and 10 km TT was calculated using a 2-way mixed ANOVA (group x time). In addition, a t-test for paired samples was used to analyze the differences between the pre-test and post-test independently for each group (HIIT_run or HIIT_bike). Cohen’s effect size (Cohen, 1988) was calculated to assess a practical significance between the pre-test and post-test in each group. Effect sizes (ES) of above 0.8, between 0.8 and 0.5, between 0.5 and 0.2, and lower than 0.2 were considered as large, moderate, small, and trivial, respectively (Cohen, 1988). Data analysis was performed using the Statistical Package for Social Sciences (SPSS Inc, version 24.0 for Windows, Chicago, IL, USA). The statistical significance was set at p < 0.05. Despite the fact that in some cases, a variable showed a p value > 0.05, whereas, ES was greater than 0.5 (moderate), was considered practical difference.

No significant differences were found pre-intervention between HIIT_run and HIIT_bike for VO_2max, HR_max and Speed_max in the incremental treadmill test or any variable obtained during the 10 km TT. After 4-weeks of the HIIT intervention, HIIT_run improved VO_2max significantly (p < 0.01, ES = 0.6, moderate), and decreased HR_max (p = 0.09, ES = - 0.5, moderate) whereas the Speed_max values were maintained (p = 0.11, ES = 0.3, small). In the HIIT_bike, no changes were observed after 4-weeks for VO_2max, HR_max and Speed_max (p = 0.16 to 0.26, ES = -0.2 to 0.3, trivial to small) (Table 2). According to the two-way mixed ANOVA analysis (group x time), only VO_2max showed a statistically significant difference. The HIIT_run group enhanced the VO_2max result in the post-test (p = 0.01) while the HIIT_bike group showed no change.

After 4-weeks of the intervention, neither HIIT_run nor HIIT_bike changed either the time to complete the 10 km TT (p = 0.06 to 0.84), ES = 0.1 to -0.2, trivial to small), or average heart rate (HR_av) (p = 0.14 to 0.58, ES = -0.2 to -0.3, small) (Table 3). HIIT_run resulted in a highe lactate concentration at the end of the test (Lactate_post) (p = 0.18 to 0.05, ES = 3.6, very large). There was a significant increase in the average and maximum rating of perceived exertion (RPE_av) (p = 0.04, ES = 1.7, large) and RPE_max (p = 0.04, ES = 1.6, large), whereas, HR_av did not change significantly (p > 0.05, ES = -0.2, trivial) (Table 3). In the HIIT_bike group the average pace decreased significantly (p = 0.02, ES = -0.3 small) and the maximal rate of perceived exertion (RPE_max) increased during the 10 km TT post-test (p = 0.04, ES = 0.9, large) (Table 3). There were no group x time differences.

Table 3 and Figure 1 show the results for the 1^st and 8^th HIIT sessions. HIIT_run demonstrated a significant increase for RPE_av (p = 0.03, ES = 0.8, moderate) and an increase in lactate concentration values immediately after the session (Lactate_post) (p = 0.14, ES = 0.6, moderate) (Table 4) and creatine-kinase (CK) concentration before session (CK_pre) (p = 0.15, ES = 1.0, large) (Figure 1). No significant differences were observed for the remaining variables. The HIIT_bike group showed a significant improvement in Speed_av, P_av, and P_max (p = 0.01 to 0.03, ES = 0.6 to 0.7, moderate), while increases in RPE_av and delayed onset muscle soreness immediately after training (DOMS_post) were also observed (p = 0.02, ES = 1.4 to 2.2, large) (Table 4). The maximal heart rate obtained during HIIT sets and recovering sets (HR_max-work and HR_maxrecovery) and Lactate_post demonstrated practical increases although these were not significant (p = 0.16 to 0.34, ES = 0.6 to 2.0, moderate to large). The average heart rate recorded during recovery intervals (HR_{av recovery}) and the concentration of CK 24-hour after the HIIT session (CK_24after) decreased practically (p = 0.46 to 0.49, ES = -0.6 to -1.0, moderate to large) (Figure 1). According to the two-way mixed ANOVA analysis (group x time), only DOMS_post showed statistical significance (p = 0.017).

Previous studies have analysed the effects of HIIT in individual exercise modes on a single mode of activity (Cook et al., 2010; Laursen et al., 2002; Sijie et al., 2012). However, the effects of this training method in running or cycling on performance in a single mode such as running in female athletes has not been studied. In the current investigation, only the HIIT_run group evoked significant improvement in VO_2max during a maximal treadmill test. Neither HIIT_run nor HIIT_bike significantly enhanced the time to complete 10 km TT, despite a significant decrease in average pace for the HIIT_bike participants. Running HIIT generated a significantly greater level of muscular fatigue in non-competitive female runners compared to cycling HIIT.

As noted above, VO_2max improved significantly in the HIIT_run group while the remaining maximal test variables in each training group were not significantly modified. This situation may be due in part to an excessive training amount accumulated by runners who may not have been accustomed to structured HIIT. Previous investigations observed improvements in VO_2max and maximal test variables such as HR and speed or power, however the performance level changed from the current investigation. Rowan et al. (2012) observed an enhancement of 4.73% in VO_2max after 5-week of 5 x 30 s with 4.5 min recovery in female soccer players, however, similar results were obtained from the control group who performed 40 min continuous running at 80% of VO_2max. In another study, Gunnarsson and Bangsbo (2012) showed that 7-weeks of short interval HIIT improved VO_2max in moderately trained male and female runners. Mallol et al. (2018) concluded that after 4-weeks of a cycling HIIT program in moderately trained triathletes, participants improved 6.7% in VO_2max and 15% in peak power. Finally, Lesmes et al. (1978) after 8-weeks of two types of supra maximal interval training (short duration at 170% of velocity at VO_2max (vVO_2max) and long duration at 130% at vVO_2max) concluded that the frequency of training, interval distances and intensities were independent of changes in aerobic power and submaximal HR in females, whereas, interval training intensity was essential to improve these variables in males rather than frequency and interval distance. In our study, HIIT_run achieved greater changes than HIIT_bike and this may be due to the specificity of the activity. For future investigations, it would be relevant to consider the effects of cycling HIIT during a maximal test using a cycle ergometer. Additionally, comparing different HIIT interventions of different intensity and work to rest characteristics.

During the 10 km TT the HIIT_run performance remained unchanged but RPE_av increased significantly indicating a greater perception of effort. At the same time, HR_max decreased after 4-week of HIIT_run program. Perceived exertion may have increased during post -intervention running 10 km TT because of individual fatigue accumulated after the HIIT program possibility due to insufficient recovery after the period of intensified training. Controversially, other researchers have observed improvements in running performances after a HIIT program. Gunnarsson and Bangsbo (2012) obtained an improvement of 6% and 4% in 1500 m and 5 km running tests, respectively. The protocol employed included 7-weeks of interval running working at intensities of 90% of HR_max with a 54% training volume reduction in moderately trained females. In addition Paavolainen et al. (1999) showed an improvement in 5 km run time in well-conditioned athletes with no changes in VO_2max, similar to that observed in the current study results for HIIT_bike participants. Furthermore, Bangsbo et al. (2009) observed a decrease of 1 min in 10 km performance (from 37 min to 36 min) after a 6-9-week training period with intervals near maximal speed with a 30% reduced total training volume. However, Iaia et al. (2009) determined that after 4-week of 8-12 x 30 s maximal speed intervals and a 64% reduction in total training volume, but there were no improvements in 10 km TT run in these endurance trained participants. The studies mentioned above enhanced running performance after HIIT interventions, however, a number of these studies replaced a part of the participants’ usual training volume with HIIT sessions. In the current research, the intensities employed during HIIT intervals were maximal or close to maximal which may have generated an excessive training stress, and therefore, excess residual fatigue to the amateur female runners.

By contrast, the HIIT_bike participants significantly increased performance (Pace_av, min/km) during the 10 km TT run with an increased RPE_max. Similarly, Mikesell and Dudley (1984) noted a decrease of 81 s on 10 km distance in well-trained runners after an intensive aerobic program, combining 40 min run “all out” 3 days a week with 5 x 5 min at VO_2max intervals with 5 min jogging on the treadmill as recovery. A number of investigations conducted with cyclists concluded that cycling HIIT significantly enhances cycling performance in a range of testing protocols and competitive simulations (Laursen et al., 2002; Lindsay et al., 1996; Stepto et al., 1999). The use of HIIT using identical activity modes to that of competition helped to improve the sports performance. Despite the fact that the HIIT_bike group did not show significant differences for time to complete 10 km TT, the Pace_av manifested a significant decrease, hence, performance during the 10 km TT post-test was enhanced. Whilst the mechanisms accounting for this performance improvement are unknown, it is possible that a gain in lower limb muscular power because of the cycle HIIT sessions led to this improvement. It is also possible that the exercise mode of running presented a higher physiological demand due to a greater muscle recruitment, contraction type and accumulative fatigue which was associated with the different performance adaptations between modes (Le Meur et al., 2011). Nevertheless, the present study results suggest that the HIIT_run program did not result in an improvement in 10 km performance considering that a greater lactate_post concentration and RPE_max were observed post-test despite a similar HR_av and similar time to complete the time trial.

This study is novel in that biochemical and perceptive markers associated with muscle soreness were measured in the acute stages following the first (1^st) and last (8^th) HIIT sessions. Following the HIIT_run intervention RPE_av, lactacte_post, and specifically, CK_pre were elevated, whereas, distance completed and Pace_av remained the same from the 1^st HIIT compared with the 8^th HIIT session. This potentially indicates a greater level of muscle damage and lack of assimilation to the HIIT running program.By contrast, HIIT_bike participants improved Speed_av and Power_av from the 1^st to 8^th HIIT session. This enhancement was associated with more positive responses for RPE_av, RPE_max, DOMS_post, lactate_post and HR_max, although HR_av decreased during recovery, suggesting that HIIT_bike athletes were able to perform more effectively in the session. At the same time both groups showed greater CK_pre during the 8^th HIIT. However, the difference between sessions were smaller, once again this may have occurred due to fatigue accumulation and lack of assimilation of the high intensity training. Furthermore, CK_24h values after the 8^th HIIT session decreased compared with the 1^st HIIT in the HIIT_bike group, which may show a physiological adaptation such that HIIT_bike runners were able to achieve a superior level of recovery from muscular fatigue.

Previous studies have focused on markers of muscle damage such as, CK concentration and DOMS after a high intensity training program in runners and cyclist while performing running and cycling testing in isolation (Nieman et al., 2014). In these studies, it was shown that muscle damage was related to activity mode. Nieman et al. (2014) analysed differences between running and cycling performances on runners and cyclists after 3 consecutive days of an ‘overreaching’ training program, concluding that the eccentric contractions intrinsic to running, resulted in 133% greater CK concentration and 87% greater DOMS in runners immediately after 3 days of running training compared with cyclists performing similar training on a cycle ergometer. Additionally, CK concentrations remained more elevated in runners than in cyclists 1, 14 and 38 h after the training program. Similarly, DOMS presented a comparable pattern to CK at 1 and 14 h and by 38 h post intervention; the values were similar in both groups. Likewise, Bruunsgaard et al. (1997) and Proske and Allen (2005) concluded that eccentric activities generated higher levels of muscle damage (CK, DOMS and myoglobin) compared with isometric and concentric contractions. In our study, CK_pre before the first and last HIIT were greater in the cycling group than the running group. However, the difference between sessions for CK_post were similar in both groups. For its part, CK24h results showed that HIIT_bike might be able to recover faster than HIIT_run after a HIIT program.

One of the current research limitations relates to sample size. Longitudinal studies, which require the completion of exhaustive and multiple assessments during an intervention, are onerous on participants, particularly taking into account work and family commitments. This can make it difficult to recruit large numbers of participants and contributed to some of the drop out in the current study. Another limitation relates to the timing of the 10 km TT on the same day of, and following, the maximal incremental test, the completion of multiple, exhaustive tests may have detrimentally affected the performance outcomes for subsequent tests on the same day. This is somewhat ameliorated in that all participants undertook the same testing process both before and after the intervention.

After 4-weeks of a HIIT program, only HIIT_run participants improved VO_2max whereas no improvements were observed for 10 km TT run performance potentially due to fatigue accumulation generated by the HIIT training itself. Although no significant group by time differences were observed, the HIIT_bike participants demonstrated improved 10 km TT run performance (Pace_av) indicative of positive cross training transfer. This occurred without changes in VO_2max during maximal incremental running tests. Both HIIT modes evoked some muscle damage although HIIT_bike seems to have achieved faster muscular recovery 24 h after HIIT session completion. Therefore, it appears that a HIIT_run program in recreational female athletes produces excessive stress, fatigue and muscle damage which may have resulted in inadequate stimulus for enhancement of running performance, whereas, HIIT_bike may be a beneficial training modality that can be used to improve running performance.