Cardiovascular and cardiorespiratory effects of high-intensity interval training in body fat responders and non-responders

This study aimed to investigate cardiovascular and cardiorespiratory adaptations to exercise intervention among participants who showed higher (responders–RsBFP) and lower (non-responders–NRsBFP) levels of body fat percentage (BFP) responsiveness. Adolescents (42.5% males) participated in a ten-week school-based high-intensity interval training (HIIT), followed by a comparison of BFP, blood pressure (BP), and cardiorespiratory fitness (CRF). RsBFP age of 16.15 ± 0.36 years, body height 170.82 ± 8.16 cm, weight 61.23 ± 12.80 kg, and BMI 20.86 ± 3.29 kg/m2. Meanwhile, NRsBFP age of 16.04 ± 0.36 years, body height 168.17 ± 8.64 cm, weight 57.94 ± 8.62 kg, and BMI 20.47 ± 2.24 kg/m2. HIIT intervention impacted BFP, with a higher decrease in the RsBFP than the NRsBFP (ΔBFPRs = − 2.30 ± 3.51(10.34%) vs. ΔBFPNRs = 1.51 ± 1.54(6.96%) p < 0.001). The primary comparison showed a statistically significant interaction effect in relation to CRF (F(1,71) = 14.12; p < 0.001). Detailed comparisons showed large and significant CRF changes in RsBFP (7.52%; d = 0.86; p < 0.001) but not in NRsBFP (2.01%; d = 0.11; p = 0.576). In addition, RsBFP and NRsBFP benefited equally in SBP (5.49%, d = 0.75; p < 0.001; 4.95%, d = 0.74; p < 0.001, respectively). These findings highlight that exercise benefits on body fat may be mainly related to gains in CRF. Due to substantial intra-individual variability in adaptation, there is a need for personalized intervention tailored for those with different reaction thresholds in body mass components.


Material and methods
The current work is part of the project "Physical activity and nutritional education in preventing civilization diseases-theoretical aspects and practical implications for the secondary school physical education program, " which was carried out in a secondary school in Wroclaw (a city in the Lower Silesia region of Poland) over ten weeks.Before running the project, G*Power software (version 3.1.)was used to calculate the a priori sample size.Considering a mixed effect analysis of variance (ANOVA) as the main base analysis, an effect size (ES) of 0.25 (medium ES), a p-value of 0.05, a power of 0.80, four groups, and two measurements, the suggested sample size was 179 participants.

Participants
The current article is the second study measuring specific anthropometric and physiological features in Rs and NRs to HIIT.Participants comprised 73 EG adolescents, including 31 males and 42 females.

Procedures
The measurements were recorded before and after the ten-week intervention on the same day between 8:00 a.m. and 1:00 p.m. Participants were asked to excrete, avoid PA and excessive drinking of liquids, and keep their typical morning patterns directly before measurement.

Intervention
A PE lesson (total duration of 45 min) started with a standardized ten-minute warm-up of a five-minute slow jog and five minutes of stretching (dynamic and static).The lesson's main activity was a 14-min Tabata Training Protocol (TAP) comprising three four-minute sessions.Each session's Tabata protocol consisted of eight cycles of two exercises, including push-ups and high knees.The second session included dynamic lunges and a spider crawl, while the third involved a plank to push-up and side squeeze 23 .Each cycle started with a maximumintensity exercise lasting 20 s, with participants completing as many repetitions as possible, followed by a tensecond active rest.To verify exercise intensity during the TAP, maximum heart rate (HRmax) was determined with the formula HRmax = 208 − 0.7 × age (16 years).The calculated HRmax (197 bpm) was used to compute the high-intensity exercise ranging from 75 to 80% of HRmax (145-157 bpm).Students' HRs were monitored during the first TAP PE lesson using a Polar H1 HR monitor (Polar Electro, Kempele, Finland).The monitors were fitted to the student's chest, level with the xiphoid process, and underneath clothing.HR was displayed on the Polar H1 watch screens during TAP exercises to encourage users to maintain an adequate intensity level.The EG achieved an average HR of 155.8 beats per minute (bpm) (± 18.2; 95% confidence intervals [CIs] 121-184).In subsequent lessons of the Tabata protocol, the exercise intensity of HR measurement was similar to that recorded during the first PE lesson.

Classification of responders and non-responders
Rs and NRs were defined as individuals who did or did not experience benefits following the completion of the exercise training intervention 43 .Classifying the participants as Rs and NRs based on BFP used the TE, similar to recent studies 16,18 .The following equation was used: where TE is the typical error, and SD diff is the standard deviation of the difference (change) between the postintervention and preintervention values.
Rs were classified as participants who demonstrated a greater than two-fold TE from zero decrease in BFP 18 .Thus, the BFP cutoff values were 6.298% for males and 3.245% for females.

Statistics
The Shapiro-Wilk assessed the normality of data distribution.All quantitative variables were presented as mean, standard deviation, and 95% CIs, while frequencies were presented as numbers and percentages.
Comparisons between baseline variables for sample characteristics of Rs BFP and NRs BFP utilized Student's unpaired t-test, while analysis of covariance assessed differences in postintervention baseline BFP between Rs BFP and NRs BFP , with BFP as a covariate.The flow of BFP in the Rs BFP and NRs BFP between response categories of physiological outcomes was presented with a Sankey diagram.To test differences in frequencies between Rs BFP and NRs BFP , a chi-squared test of independence was conducted.The odds ratio (OR) and Cramer's V for ES were then calculated.
To test changes and disparities in preintervention and postintervention values between groups, a mixed effect ANOVA was conducted, with homoscedasticity and sphericity tested using Levene's and Mauchly's tests, respectively (with Greenhouse-Geisser correction when the sphericity assumption was violated).The main effects were response classification and intervention.Analysis of the Rs BFP and NRs BFP preintervention and postintervention measurements was initially performed by within-subject comparisons and then by between-subject comparisons.Detailed post hoc comparisons used Bonferroni's correction.A Student's paired t-test was performed to assess Δ between both response categories.Cohen's d ES was also calculated.
Linear regression assessed the relationship between preintervention BFP and physiological outcomes when significant changes between respondence categories were observed.Regression equations were calculated using the formula y = b 0 + b 1 × BFP (where b 0 is the constant and b 1 is the slope).Statistically significant models are presented in figures with regression lines drawn.Then, multiple regression analysis tested the regression constants and slopes.The procedure determines if two regression lines are significantly shifted up or down (test for intercepts) and whether there is an interaction (test for slopes).
To determine if Rs BFP and NRs BFP reacted differently to the HIIT stimulus in physiological outcomes while considering preintervention BFP, the present study assessed intra-individual variability by examining correlation coefficients with repeated measures.To avoid violating the assumptions for linear relationship analysis conducted for repeated measures 44 , the procedure (repeated measure correlation) described by Bakdash and Marusich 45,46 was used.
The alpha level was fixed at p < 0.05 for all tests of statistical significance.Calculations employed Statistica 13.0 (StatSoft Poland 2018, Cracow, Poland) and R software with RStudio (PBC, Boston, MA, USA URL http:// www.rstud io.com/ (accessed on 15 May 2023)).Repeated measures analysis was performed with the rmcorr package 45,46 .

Ethical approval
The Senate Research Ethics Committee at the Wroclaw University of Health and Sport Sciences Poland approved this study (consent No. 33/2018 on the 31st October 2018), which followed human experiments' institutional ethical requirements and the Declaration of Helsinki.The school principal, parents, and study participants gave informed consent before participating.The participants were informed about the study purpose and type, the methods used, and the conditions of their participation.The surveys were conducted by Wroclaw University of Health and Sport Sciences researchers.

Frequencies of responders and non-responders in systolic blood pressure, diastolic blood pressure, and fitness index response categories based on body fat percentage
The potential flow of the Rs BFP in the response categories for all outcomes (systolic BP [SBP], diastolic BP [DBP], and fitness index [FI]) was studied based on BFP.As shown in Fig. 1, most of the Rs BFPs and NRs BFPs (males and females) are located in corresponding SBP, DBP, and FI categories.However, a trend of increasing participation of the NRs BFP in the Rs BFP categories was also observed across subsequent variables (SBP, DBP, and FI).Therefore, any differences in proportions were not statistically significant (p > 0.05) (Fig. 1).

High-intensity interval training intervention effects on cardiovascular and cardiorespiratory characteristics
Table 3 presents characteristics of the preintervention, postintervention, and Δ in measured parameters.Figure 2 provides an overview of the within and between-responder classification adaptations in the vascular and CRF parameters.

Inter-individual differences in the relationship between preintervention body fat percentage and changes in cardiorespiratory fitness of the responders and non-responders
Inter-individual variability in the relationship between changes in physiological outcomes and preintervention BFP was studied using linear regression.Linear modeling indicated the different potential of BFP at baseline in influencing the cardiovascular parameters and CRF.Baseline BFP did not significantly influence SBP or DBP, as there was no marked shift between regression lines (SBP: b = − 0.04, p = 0.961; DBP: b = − 1.67, p = 0.069).These findings suggest equal changes in SBP and DBP in relation to preintervention BFP in both response categories.Similar slopes between groups confirmed a non-significant interaction term (SBP: b = − 0.17, p = 0.507; DBP: b = − 0.02, p = 0.093).
There was a significant difference in intercepts of the FI relationship to BFP in both response categories (b = − 1.62; p < 0.001) (Fig. 3), though a significant vertical shift showed that the change was higher in Rs BFP compared to NRs BFP based on preintervention BFP.However, non-significant interactions (b = − 0.02; p = 0.827) suggested that one-unit changes in the predictor were associated with similar mean response changes in Rs BFP compared to NRs BFP .

Intra-individual variability in fitness index in response to high-intensity interval training based on body fat percentage
Repeated measures correlation analysis revealed different associations between the investigated variables in both response categories.The strongest (significant and negative) relationship was found for BFP and FI (r rm = − 0.44; p = 0.002) in Rs BFP , whereas the same relationship was lowest and non-significant in NRs BFP (r rm = − 0.08; p = 0.567) (Fig. 4).Also, decreasing BF was moderately and positively related to decreasing SBP in Rs BFP (r rm = 0.39; p = 0.007) and poorly aligned with decreased DBP (r rm = 0.22; p = 0.007).In addition, there were moderate negative correlations between BFP and SBP (r rm = − 0.43; p = 0.018) and DBP (r rm = − 0.48; p = 0.008) in NRs BFP .

Discussion
To the best of the authors' knowledge, this is the first study to investigate differences in cardiovascular parameters and CRF in Rs BFP and NRs BFP in a school-based HIIT intervention, although the strategy was similar to the approach of Maturana et al. 19,39 .Our results showed that HIIT during one PE lesson per week for ten weeks impacted participants' BF in various ways, leading to substantial individual variability in Rs BFP and NRs BFP .Rs had a greater increase in CRF (large effect), evidenced by higher final postintervention levels (moderate effect), compared to NRs, though there were no differences in the number of changes between both response categories, including for BP parameters.Both categories decreased in SBP after HIIT, while only NRs did for DBP.Interindividual analysis of the relationship between preintervention BF and changes in physiological outcomes showed similar individual variability in both response categories, resulting in a lack of interaction between categories.In addition, the variability in any outcome was not associated with initial BF mass in either category, indicated by the constant, horizontal regression line, slopes, and coefficients.The only significant effect was an association between BF at baseline and FI responsiveness in Rs BFP .Additionally, the BF response increase in pre and postdifferences was accompanied by positive changes (increasing differences) in BP parameters and FI.Meanwhile, the opposite relationship was observed for NRs BFP for BP and FI.
Findings support the latest exercise guidelines for individuals with overweight/obesity, emphasizing the benefits of a multicomponent exercise approach for cardiometabolic health.Multicomponent exercise, including bodyweight drills and resistance-based activities, may enhance musculoskeletal fitness in inactive, middle-aged adults with overweight/obesity 47,48 .For children, a multidimensional physical activity intervention could also be effective in improving musculoskeletal and cardiovascular fitness, and the term HIIT seems to be effective 49 .
We observed the impact of HIIT on fat tissue reduction.There was such substantial individual variability that it was possible to distinguish Rs and NRs based on BFP.Study conducted by Dominic and Kishore 50 showed the effectiveness of HIIT on BF reduction, which was confirmed in a meta-analysis by Khodadadi et al. 51 .With joint improvement in body composition, increased CRF 52 and motor fitness 53 are observed.However, multidirectional analyses are less common 54 .The benefits of HIIT are often achieved in a few areas, though there is a lack of deeper insight into individual responses, which may differ due to physiological processes 55 .
Despite the mean results confirming improvement in some parameters, some subjects did not respond or had a varied effect in their response magnitude 56 .Therefore, the question arises of how to produce beneficial gains in NRs, particularly since the observations showed inconsistent results.Juric et al. 57 found high variability in body tissue despite a significant impact on CRF after HIIT intervention.Nonetheless, the lack of consistency in observations should be explained by inter-individual analyses.A study by Montero and Lundy 43 suggested that non-reactive subjects should receive a higher intervention load.On the other hand, a high dose of any PA may lead to overfatigue or overtraining 58 .Moreover, adolescents may reject PA if the intensity and/or volume is too high 59 .As such, it is essential to start from an appropriate level and then increase the load over time 60 .Our study showed that improvement in BFP was associated with better CRF, which was also observed in studies by Lan et al. 61 and Guo et al. 62 .Generally, aerobic metabolic processes use fatty acids, which may explain the links between losing BF and joint improvement and increased CRF 63 .
The methodological approach used to study HIIT effects on BP and CRF in Rs BFP and NRs BFP was unusual, meaning that comparing the results to other works is challenging.The same approach was presented by Maturano 19,39 , though they examined variability in H-RF parameters by measuring VO 2 max.Their findings demonstrated the influence of exercise intensity and biological variability on an individual's VȮ 2 max response after MIT and HIIT, highlighting the importance of personalizing interventions 64 .
Individualization of training intervention by load optimization, despite the target (sports results or health), is an effective way to achieve goals 19,39,64 .Training modifications are needed when progression is not visible, though appropriate increases must be introduced 65 .From a practical point of view, addressing the training intervention on an individual level is challenging when targeted at a population.However, their effectiveness could grow when the factors distinguishing individual responses are identified 66 .Baker et al. 67 showed that body composition influences individual intervention responses.Also, a study by Andrade-Mayorga et al. 68 showed that physiological characteristics impacted changes.These observations demonstrate that factors modify individual responses based on physical variables.As such, some of them should be the basis for PA program development.
Another aspect was the analysis of inter-individual variation in the relationship between changes in physiological outcomes and preintervention BF.Intra-individual variability was similar in Rs BFP and NRs BFP , with no dependence of baseline BFP on changes in SBP, DBP, and FI.Similar studies conducted by Maturano 39 (for other variables) found relationships between both response categories, with more substantial increases in peak power output, lactate threshold, and microvascular responsiveness in Rs, whereas a more significant increase in cycling efficiency was observed in NRs.In addition to differences in VȮ 2 max, a greater increase in microvascular responsiveness was observed in Rs compared to NRs.Furthermore, Rs and NRs did not exhibit differences in metabolic adaptations.As such, there is a growing need for personalized training plans based on the desired clinical outcome.
A study on intra-individual variation in the relationship between differences in BF and physiological outcomes showed varied association patterns in Rs BFP and NRs BFP .Rs BFP had positive gains in BF and improvement in all physiological outcomes, while the NRs BFP did not display such an association.There are few methods for studying such a relationship, as the individual response may be due to physiology, hormone levels, and muscle tissue type.As such, creating a multivariable profile of individuals to find the best-fitting training program for achieving optimal physical outcome improvements is required.
The current study was not cross-over in design, which would have provided more robust data.Another limitation was the small number of participants separated into Rs and NRs, which forced sex agglomeration.Indeed, analyzing sexes separately would be more advantageous when inferring the findings.A further limitation is the criteria for identifying Rs and NRs, which were defined by a change in the typical error (TE) of the main outcomes and co-variables from baseline to follow-up (null hypothesis testing).Future analysis should include prior knowledge (distribution) on the intervention's effects and could apply (combine) the Bayesian method (ROPE + HDI decision-making in posterior distribution) to identify the response categories, an approach that was verified as considerably more advantageous 19,69 .Furthermore, using VO 2 max to assess CRF would have provided more precise information on adaptation than the Harvard Step Test (HST) method used.Also, using a DEXA device for body morphology measurements would provide more accurate data due to its reference method in body morphology measurements.There is also a need to verify the effectiveness of interventions tailored to non-responders groups.Also interesting would be the identification of factors that affect the response after HIIT.

Conclusions
Our study showed the effectiveness of school-based HIIT in decreasing BF mass, though there was substantial individual variation.Also, BFP response categories differed from those for other physiological outcomes.Therefore, adaptation in body mass composition does not guarantee a positive response in physiological parameters.
Our findings highlight BF adaptation and benefits from intensive exercise intervention during PE lessons, which may be related to gains in CRF, although BP did not have the same response.However, there were visible positive changes in BP parameters independent of the BF response, though the BF changes may appear before other health-related outcomes are modified.There was substantial intra-individual variability in adaptation.Therefore, there is a need for personalized interventions (structure and load) tailored for persons with different reaction thresholds in body mass components.

Figure 1 .
Figure 1.A Sankey diagram showing the switch of body fat percentage responders (Rs) and non-responders (NRs) (males and females) across the systolic blood pressure (SBP), diastolic blood pressure (DBP), and fitness index (FI) response subgroups.Block sizes represent the proportion of sample allocation between subgroups of Rs and NRs.Trajectories highlighted in dark grey belong to the responders to cardiovascular and cardiorespiratory features.Chi-squared independence test statistics are presented on the right.

Figure 2 .
Figure 2. Within and between groups differences between body fat percentage responders and non-responders to the high-intensity interval training intervention in vascular (systolic and diastolic blood pressure) and cardiorespiratory (fitness index) parameters.Mixed effect analysis of variance results in the title of each diagram shows an interaction term (F-value) together with corresponding p-values and generalized eta square (ɳ 2 GES ), which is particularly suitable in a repeated measures design.Besides the effect size derived from the whole model, the within and between-subject comparisons are displayed as Cohen's d-effect size (d-values with 95% confidence intervals) and p-values derived from pairwise Bonferroni post hoc comparisons.The black dots represent the individual values connected to PRE and POST through continuous grey lines.Horizontal solid lines present mean values and boxes are 95% confidence intervals for mean values.ANOVA analysis of variance; CI confidence interval; DBP diastolic blood pressure; FI fitness index (cardiorespiratory fitness parameter); Rs BFP responders, NRs BFP non-responders; PRE preintervention; POST postintervention; SBP systolic blood pressure.

Figure 4 .
Figure 4.A scatterplot with multiple groups (responders and non-responders at two time points) displaying the relationship between increased body fat percentage (preintervention and postintervention measurements) and changes (decrease or increase) in physiological outcomes.Individual variations are represented by solid regression lines (rmcorr fit for each participant), while dashed grey lines represent common general trends in intra-individual variability, which were calculated from a multilevel mixed model with random intercepts and fixed slopes.Graphs are supplemented with r rm -values and their corresponding p-values.

Table 1 .
An overview of participants' baseline anthropometric and body fat percentage measurements.
Unpaired Student's t-test results are presented as t-values and p-values.BH body height; BMI body mass index; BW body weight; CI confidence interval; Rs BFP responders; NRs BFP non-responders; p p-value; sd standard deviation; t t-value.

Table 2 .
Baseline, postintervention, and change (Δ) values in body fat percentage responders and nonresponders.The p-values are derived from Bonferroni's post hoc tests (preintervention and postintervention) and Student's paired t-test (Δs).CI confidence interval; Δ change; Δ% change in %; Rs BFP responders, NRs BFP non-responders; p p-value; PRE preintervention (baseline) measurements; POST postintervention measurements; ESb effect size between groups; ESw effect size within the group.

Table 3 .
Characteristics of the preintervention, postintervention, and change (Δ) values in systolic blood pressure, diastolic blood pressure, and fitness index between body fat percentage responders and non- responders.Δ change; Δ% change in %; CI confidence interval; DBP diastolic blood pressure; FI fitness index; Rs BFP responders, NRs BFP non-responders; PRE preintervention (baseline) measurements; POST postintervention measurements; SBP systolic blood pressure; sd standard deviation; ESb effect size between groups; ESw effect size within the group.