Hypoxia matters: comparison of external and internal training load markers during an 8-week resistance training program in normoxia, normobaric hypoxia and hypobaric hypoxia

Purpose To compare external and internal training load markers during resistance training (RT) in normoxia (N), intermittent hypobaric hypoxia (HH), and intermittent normobaric hypoxia (NH). Methods Thirty-three volunteers were assigned an 8-week RT program in either N (690 m, n = 10), HH (2320 m, n = 10), or NH (inspired fraction of oxygen = 15.9%; ~ 2320 m, n = 13). The RT program (3x/week) consisted of six exercises, with three sets of six to 12 repetitions at ~ 70% of one repetition maximum (1RM) with the first session of each week used for analysis. 1RM in back squat and bench press was used to evaluate muscle strength before and after the program. External load was assessed by the volume load relative to body mass (RVL, kg·kg−1). Internal load was assessed by the ratings of perceived exertion (RPE) and heart rate (HR). Results Smaller relative improvements were found for the back squat in the N group (11.5 ± 8.8%) when compared to the NH group (22.2 ± 8.2%, P = 0.01) and the HH group (22 ± 8.1%, P = 0.02). All groups showed similar RVL, HR responses and RPE across the program (P˃0.05). However, reduced HR recovery values, calculated as the difference between the highest HR value (HRpeak) and the resting heart rate after a two min rest, were seen in the N and NH groups across the program (P \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document}< 0.05). Conclusion It seems that 8 weeks of intermittent RT in hypoxic environments could maximize time-efficiency when aiming to improve strength levels in back squat without evoking higher levels of physiological stress. Performing RT at hypobaric hypoxia may improve the cardiorespiratory response, which in turn could speed recovery.


Introduction
Altitude training is widely used by coaches and athletes to complement normal training at sea level.Applied to strength, studies have shown how hypertrophy resistance training (R T ) under moderate hypoxic conditions promote the mechanisms related to maximal strength and muscle mass development (Kurobe et al. 2015;Kon et al. 2010).The alternatives to performing R T under moderate hypoxic conditions are numerous and will depend on the combination of the dosage of hypoxia (chronic vs. intermittent) and the type of hypoxia (hypobaric hypoxia [HH] or real altitude vs. normobaric hypoxia [NH]).Regarding dosage, intermittent exposure, such as "living low-training high", seems to be more advantageous than permanent exposure.It has been observed that permanent exposure can lead to a reduction in muscle cross-sectional area and a decrease in the size of muscle fibers in humans (Narici and Kayser 1995).It can also result in increased neuromuscular fatigue during exercise by impairing neuromuscular transmission during contractions (Amann et al. 2006), and by decreasing the muscle fibers' capacity for relaxation (Allen et al. 2008).Regarding the types of hypoxia: HH can be achieved by ascending to a real moderate altitude, while NH can be achieved by reducing the oxygen pressure in the inspired air (nitrogen dilution or oxygen filtration).For economical and logistical reasons, NH is often used as a laboratory alternative to HH, because athletes do not have to ascend to a higher altitude nor spend time acclimating.Alternative techniques can also be employed to trigger a hypoxic response in muscles.Applying pressure to the blood vessels near the skeletal muscles leads to insufficient oxygen delivery (hypoxia) to the muscle tissue (Patterson et al. 2019).Instead of decreasing atmospheric oxygen globally, hypoxia can be localized by exerting external pressure on the limbs, causing partial restriction of blood flow, a method known as blood flow restriction.
The efficiency, safety, and effectiveness of R T training programs are paramount for sport conditioning.Coaches try to identify the optimal doses of the training load (product of exercise intensity and volume) since, if it is insufficient, adaptation might not occur, while excessive stress might impair performance and potentially increase the risk of injury (Halson 2014).For this reason, it becomes really important to monitor parameters of both internal (i.e., the athlete's individual responses, such as heart rate [HR] and ratings of perceived exertion [RPE]) and external (i.e., the total volume of load lifted: sets × reps × kg) loads to quantify the training stress and hence estimate the training efficacy (McLaren et al. 2017).When evaluating the response to training stress, internal load provides a more precise and individualized measurement compared to external load.However, factors such as training status, nutrition, health, psychological status, environment (e.g., altitude, temperature) and genetics may lead to variation in their internal load when providing the same external load (Impellizzeri et al. 2018).The optimal dose of training intensity for R T under hypoxic conditions may differ, as the environmental condition per se seems to be physiologically more demanding than at sea level (Rodríguez-Zamora et al. 2019).In addition, when comparing the hypoxic conditions (HH vs. NH), it has been reported that HH represents a more severe physiological stimulus than NH at the same theoretical altitude (Millet and Debevec 2020;Timon et al. 2022).Thus, the quantification of the training load could be distinct depending on the R T environment.
Ratings of perceived exertion and HR monitoring have been widely used to verify the internal load in athletes.While RPE seems to be a valid surrogate measure of the internal load in R T (McGuigan et al. 2004;Sweet et al. 2004), the use of HR monitoring as a direct indicator is scarce (Apkarian 2019;Moreira et al. 2017).Heart rate monitoring during R T serves as a practical and costeffective method to gauge the physiological response to exercise.It provides real-time insights into the cardiovascular demands of R T , aiding in the assessment of internal training load (de Beukelaar and Mantini 2023).Research suggests that HR can serve as a reliable indicator of exercise intensity and metabolic stress during R T sessions (de Beukelaar and Mantini 2023).This approach offers a noninvasive means to monitor, and tailor individualized training programs, enhancing both safety and effectiveness.However, it is essential to consider individual variations and factors influencing HR response to ensure accurate interpretation in the context of R T (Scott et al. 2017).
R T involves a blend of static and dynamic contractions that elicits hemodynamic responses (Apkarian 2019;Hill and Butler 1991).When parasympathetic activity is decreased and sympathetic activity increased there is tachycardia and increased cardiac output, increased total peripheric resistance, as well as an increase in both systolic and diastolic blood pressures (Hill and Butler 1991).However, the magnitude of this response may be affected by several factors such as the exercise load, movement velocity, rest interval duration, muscle mass involved, exercise duration, age, training status and altitude exposure (Apkarian 2019;Mourot 2018).
Unfortunately, to date, there are no longitudinal investigations comparing training load markers during intermittent hypoxic R T in normobaric vs. hypobaric hypoxia with the equivalent training in normoxia.Therefore, we aimed to compare external and internal training load markers during an 8-week R T program in normoxia, intermittent normobaric hypoxia, and intermittent hypobaric hypoxia.We hypothesized that: 1) the two hypoxic groups (HH and NH) will lift higher loads across the R T program compared to normoxia; and 2) real moderate altitude would lead to higher psychophysiological stress (HR and RPE) values when compared to the other two conditions.

Participants
Thirty-three physically trained Sport Science students familiarized with the R T exercises employed volunteered to participate in the study.The mean ± SD of age, training experience, height, body mass, BMI, fat mass, and fat free mass were: 22.4 ± 3.2 years, 2.9 ± 1.8 years, 176.5 ± 6.5 cm, 73.8 ± 10.1 kg, 23.7 ± 2.8 kg•m −2 , 8.1 ± 4.2 kg, and 65.7 ± 6.8 kg, respectively.To be included, volunteers could have no self-reported health issues or muscle injuries and not have been exposed to more than three to four consecutive days of altitudes higher than 1500 m above sea level (MASL) for at least 2 months before the study.Participants were provided with information on the research protocol, and they provided signed informed consent.This study was approved by the local Ethics Committee (PEIBA/2018) and was conducted in accordance with the Declaration of Helsinki.

Design
A longitudinal quasi-experimental design was employed to compare changes in the relative volume load (RVL), ratings of perceived exertion (RPE, a.u), and the heart rate (HR, bpm) response during an 8-week R T program.Participants were grouped according to the following conditions: living and training in normoxia (N; 690 m; n = 10); living in normoxia and training at moderate altitude (HH; 2.320 m; n = 10); and living in normoxia and training at normobaric hypoxia (NH; inspired fraction of oxygen [F i O 2 ] = 15.9%;~ 2320 m; n = 13).The characteristics of each group are shown in Table 1.To see the effect of the R T program on one repetition maximum (RM) in the three conditions, we compared the relative one RM values when executing back squat and bench press exercises before and after the program in normoxic conditions (PiO 2 = 149 mmHg).The training program comprised three R T sessions•week −1 (24 in total) at the same time of day with two additional weeks allocated for pre-and post-muscular strength and anthropometric assessments.The first session of each week (Mondays = control session) was used for the analysis.Both the N and NH groups trained at the faculty laboratory (below 700 MASL), while the HH group conducted the training sessions at the High-Performance Center in Sierra Nevada, Granada, Spain (2320 MASL).All volunteers took part in a pre-intervention 2-week conditioning-training program.To ensure the essential amino acids intake for performance during the training periods subjects were provided with a standard protein shake after each training session.To minimize the potential for instruction bias, the researchers remained the same for all the groups.Participants were instructed to abstain from intense exercise and alcohol intake, and to

Methodology
The normobaric hypoxic condition The NH condition was created in a normobaric tent (CAT 430,Colorado Altitude Training,USA), where oxygendepleted air was pumped from a hypoxic generator with a semi-permeable filtration membrane (nitrogen filter technique; CAT 310, Louisville, Colorado, USA) to simulate hypoxic conditions (FiO 2 = 15.9%,equivalent to 2320 MASL).Barometric pressure in the tent was equivalent to that of University of Granada campus (704 mmHg).Ambient O 2 was continuously monitored by a digital controller (Handi + , Maxtec, USA) to maintain the hypoxic conditions in the tent.Immediately before and after each first R T session of the week, arterial oxygen saturation (SpO 2 ) was measured per duplicate using a pulse oximeter (Wristox 3100; Nonin, Plymouth, MN, USA).

The training session
Each R T session encompassed ~ 15 min of warm-up and the execution of six exercises, with three sets of six to 12 repetitions at 65-80% of 1RM with 90 s of rest between sets and exercises (Fig. 1) to promote muscle growth (Kraemer et al. 1991).During the recovery, participants were asked to sit for 2 min.The lifted load was individually adjusted in every session so the participants could perform ± 2 rep.If a participant could perform ± 3 reps, the training load was adjusted by ± 5% accordingly (Kraemer et al. 2009).All the training sessions were supervised by experts to guarantee proper technique and safe execution.Because volume load is largely related to measures of internal load and physiological stress during resistance exercise at various intensities (Genner and Weston 2014), the relative volume load (RVL, kg•kg −1 ) as well as its percentage of change (ΔRVL) were calculated for every R T session and were considered for further analysis.(Relative volume load was calculated as a product of the total number of repetitions performed and the amount of mass lifted divided per body weight; ΔRVL was defined as the difference in RVL between session one and sessions two to eight, expressed in %).

Strength performance
One RM was estimated for each of the six main exercises of the training program (back squat, bench press, deadlift, pulldown prone grip, barbell row, and shoulder press) using a progressive trial and error procedure (Brown and Weir 2001).Three sets of three to six repetitions at increasing loads were completed before performing one set of two to three repetitions to failure.Rest periods between sets were kept to 5 min.Using the two-three RM load, one RM values were predicted using Brzycki's equation (Brzycki 1993).
After the R T program, subjects repeated the same procedure for the back squat and the bench press for upper-and lowerbody comparison.These exercises, back squat and bench press, were selected due to differences in training machine models between locations for the deadlift, pulldown prone grip, barbell row, and shoulder press.Relative one RM (kg•kg −1 of body weight) as its percentage of change (Δ%) in back squat and bench press were considered for further analysis.

Body composition
A segmental multi-frequency bioimpedance analyzer (Tanita BC 418 segmental, Tokyo, Japan) was used to assess weight

Oxygen saturation
Immediately before and after each control session, peripheral arterial oxygen saturation (SpO 2 ) was measured per duplicate using a pulse oximeter fingertip sensor (Wristox 3100; Nonin, Plymouth, MN, USA).

Heart rate
Heart rate (HR) was continuously recorded beat-by-beat by a Polar Memory Belt HR monitor (Polar, Vantaa, Finland) and a Polar M430 watch (Polar Electro Oy, Kempele, Finland).
The belt was placed on the participant's chest ten min before the H T session and removed after the recovery assessment.
To minimize potential instrumentation bias, participants wore the HR monitor during the pre-intervention period.
Then, data were transferred to the software Polar Flow Sync 3 (version for Windows) and exported to an Excel file.Preexercise HR (HR pre ) was the average HR for the minute immediately before the start of the R T session; HR peak and HR min were the highest and lowest HR values during the R T session, while HR mean was the arithmetic mean for R T session.The value for heart rate recovery (HRR) was calculated as the difference between HR peak and the resting heart rate after a 1 min rest (HRR 1 ) and two min rest (HRR 2 ) (Cole et al. 2000(Cole et al. , 1999)).

Ratings of perceived exertion
Rating of perceived exertion (RPE) was assessed after the R T session using the Borg CR-10 scale (Noble et al. 1983).
To ensure the quality of the data collected, all subjects used the CR-10 scale on three different occasions 1 week prior to starting the training program (Psycharakis 2011).

Statistical analysis
Data were presented as means and standard deviations (± SD) unless otherwise indicated.Differences in relative and absolute RM for each exercise (back squat and bench press) were assessed using two-way mixed ANOVA, with one between-subject factor (condition: N vs. NH vs. HH) and one within-subject factor (time: pre-post-R T program) and interaction terms (condition x time).Differences in Δ% were assessed using a one-way ANOVA, with one betweensubject factor (condition: N vs. NH vs. HH).Differences in HR parameters, RVL, and ΔRVL were assessed using a two-way mixed ANOVA, with one between-subject factor (condition: N vs. NH vs. HH) and one within-subject factor (session: from one to eight) and interaction terms (condition x session).A three-factor mixed model ANOVA with a between-subject factor (N vs. NH vs. HH) and two within-subject factors ([session: from 1 to 8] and time [pre-post-training session]) were applied on SpO 2 .Post hoc comparisons, when justified, were performed using Bonferroni correction for multiple pairwise comparisons.Partial eta squared from the ANOVA (η 2 p ) was interpreted using the following classification: 0.02 (small), 0.13 (medium), and 0.26 (large) (Bakeman 2005).A two-tailed Pearson product moment correlation coefficient (r) was used to look for relationships among RVL and HR parameters during the session.The following scales were used to interpret the magnitude of the correlations: < 0.1 trivial, 0.1-0.3small, 0.31-0.5 moderate, 0.51-0.7 large, 0.71-0.9very large, > 0.9 nearly perfect (Cohen 1988).Differences in RPE were determined by an ART two-way mixed ANOVA with one between-subject factor (condition: N vs. NH vs. HH) and one within-subject factor (session: from one to eight) and interaction terms (condition x session).The original RPE data first underwent ART (ARTool for Windows, version 2.1.2),and then, the ranked RPE data were analyzed with the usual ANOVA procedure (Feys 2016).The level of significance was set a priori at P < 0.05.Statistical analyses were conducted using SPSS Statistics for Windows (v.22; IBM Corp., Armonk, NY).

Discussion
This is the first study to monitor and compare both external and internal load markers during an 8-week R T program in three different environments: normoxia, intermittent  1.2 ± 0.2 * 8.9 ± 6.1 0.9 ± 0.2 1 ± 0.2 * 11.9 ± 5.2 0.9 ± 0.2 1.1 ± 0.2 * 14.5 ± 5.8 Table 3 Relative volume of load lifted (RVL) and mean values of heart rate (HR) per session Data are mean ± SD (n = 33) for the 8-week R T program.RVL relative volume of load lifted calculated as a product of the total number of repetitions performed and the amount of mass lifted divided per body weight, HR pre average HR for the minute immediately before the session, HR mean the arithmetic mean for the session HR min and HR peak the lowest and highest HR value during the session, HRR 1 the difference between the HR peak and the resting heart rate after a 1 min rest, HRR 2 the difference between the HR peak and the resting heart rate after a 2 min rest.Significant differences (P < 0.05) among sessions were noted: ¥ vs. 1st; δ vs. 1st, 4th and 5th; ** vs. 1st, 2nd, 3rd and 4th; § vs. 1st, 2nd, 3rd, 4th, 5th and 6th; * vs. 2nd, 3rd, 5th, 6th, 7th, and 8th;  † vs. 1st, 2nd, 3rd,  4th, 6th, and 7th; α vs. 5th, 6th, 7th, and 8th; β vs. 4th, 5th, 6th, 7th, and 8th; and Ω vs. 3rd, 4th   Data are mean ± SD (n = 33) for the 8-week R T program.HRR 2 : the difference between the HR peak and the resting heart rate after a 2 min rest.Significant differences (P < 0.05) among sessions were noted: * vs. 3rd, 5th and 6th; † vs. 3rd and 6th; ‡ vs. 4th, 6th and 7th; § vs. 4th and 6th  hypobaric hypoxia, and intermittent normobaric hypoxia.Our research led to three principal findings: (1) the RVL dynamics were similar in all three groups across the 8-week program, with the hypoxic groups achieving larger improvements in relative 1RM back squat; (2) the indices of physiological stress (HR and RPE) during the R T program seemed to be similar in all three groups; and (3) the exposure to hypobaric hypoxia seems to speed recovery during R T .
In our study, all three groups showed similar RVL dynamics during the R T program (Table 3), with an increased RVL at the end (Fig. 2).In agreement with recent studies (Inness et al. 2016;Guardado et al. 2020), the HH and NH groups showed higher values of relative 1RM in back squat compared to the N group after the program.It is well known that R T in intermittent hypoxic conditions produces some physiological adaptations resulting in higher levels of both muscle hypertrophy and maximal strength compared to the same R T performed in normoxia (Kurobe et al. 2015;Kon et al. 2010).So, why did the relative one RM for bench press not improve?The benefit of R T under hypoxic conditions is based on the large accumulation of metabolic byproducts due to hypoxia, such as blood lactate, protons (H + ), calcium, and inorganic phosphorus, among others, derived from the increase in anaerobic metabolism to compensate the loss of oxygen availability (Kon et al. 2012;Kurobe et al. 2015;Schoenfeld 2013).In the back squat, where more muscle mass is at work, a larger accumulation of metabolic waste byproducts occurs with the concomitant larger improvement in the strength levels.In addition, the observed distinctions in back squat performance could also be attributed to the program's design, particularly the inclusion of two lowerbody push exercises vs. one upper-body push exercise as illustrated in Fig. 1.
Concerning the HR, all three groups showed similar HR responses across the R T program.This is in line with Scott et al. (2017), who reported similar HR responses between a normoxic group (FiO 2 = 21%) and a hypoxic group (FiO 2 = 16%) during a R T session (Scott et al. 2017).It is well known that there is a linear relationship between SpO 2 and HR.While at sea level, SpO 2 hovers around 98-100%; above 3100 MASL (FiO 2 = 14%), SpO 2 drops to 80% due to the lower PaO 2 , requiring the heart to increase the HR to maintain appropriate oxygen delivery to tissues (Cardiology 2012).In our study, on average, the difference in SpO 2 between the hypoxic groups and the N group both before (-3.8%) and after the session (-3.3%) would not have been enough to significantly modify the HR pattern.Another possible explanation could rely on the intrasubject variation.Expressed in bpm, the day-to-day variations in submaximal HR could be estimated at around five-eight bpm (Lamberts et al. 2004).This means that the expected difference between normoxia and hypoxia in terms of the HR pattern at moderate altitude (for instance, our 2320 MASL) could be hidden by the spontaneous fluctuations of HR for a given training load.Furthermore, it is well known that other factors such as endurance training experience and genetics could have also affected the HR response of our participants (Mollard et al. 2007;Masschelein et al. 2015).
In the present study, the HH group showed larger HRR 2 values during a few sessions of the R T program (Table 4) showing faster recovery.According to the literature, higher levels of HRR 2 are indicative of a more efficient cardiovascular system, suggesting that the heart can rapidly return to baseline levels after exertion, which is generally associated with better cardiovascular fitness and health (Cole et al. 2000).HRR 2 is influenced by the gradual decrease in sympathetic activity and the removal of metabolites (such as epinephrine, lactate, and H + ) from the bloodstream, both of which are consequences of engaging in intense exercise (Imai et al. 1994).Terrestrial altitude in comparison to simulated leads to grater hypocapnia and blood alkalosis for the equivalent "altitude" due to the barometric pressure reduction (Savourey et al. 2003).In this sense, performing 2 months of R T at moderate hypobaric hypoxia may produce cardiorespiratory adaptations which in turn would speed recovery (Álvarez-Herms et al. 2012;Savourey et al. 2003).Our findings are consistent with those of Bhattarai et al. (2018), who observed comparable basal HR but significantly faster recovery of HR after step test in highlanders (~ 3000 MASL) vs. lowlanders (Bhattarai et al. 2018) and with those of Álvarez-Herms et al. (2012), who found better HRR 2 values after 4 weeks R T in a hypobaric chamber compared to normoxia (Álvarez-Herms et al. 2012).
Regarding RPE, the Borg scale CR-10 has been proposed as a surrogate measure for the internal load during R T (Sweet et al. 2004;McGuigan et al. 2004).The lack of differences in RPE among groups could be explained by the same intensity applied (~ 70% 1RM) in all three conditions (Gearhart et al. 2002) and the dosage of hypoxia.Progressive arterial hypoxemia and increases in ventilation have been identified as the primary cues for determining RPE in moderate (FiO 2 = 15.2%) and severe hypoxic (FiO 2 = 11.4%)environments (Jeffries et al. 2019).It seems that in our study, the averaged SpO 2 achieved in the hypoxic groups (HH: 93.5 ± 1.3 and NH: 92.5 ± 2.7) was not enough to trigger a decline in the perceptual responses.Attenuated RPE after hypoxic training has been described (Brocherie et al. 2017), suggesting an improved tolerance or acclimatization to hypoxia after only one session.In contrast, R T in hypoxia has been perceived as tougher (Inness et al. 2016;Rodríguez-Zamora et al. 2019) or without effect (Scott et al. 2015).These discrepancies may be attributed to the Borg scale used (6-20 vs. CR-10) and the athletic experience (Noble and Robertson 1996;Barroso et al. 2014).Future work should explore which parameters can explain the perceived exertion in R T under hypoxic conditions and to determine which internal load markers could potentially help the most to provide information about the athlete's physiological stress in hypoxic environments.
Certain issues and limitations regarding the design, methodology, and overall validity of this study need to be considered.To include as many participants as possible in each group, it was decided to make the groups according to the subjects' availability.Hence, there could have been some intrinsic characteristics of the groups (differences in motivation, personality, stress, etc.) that could have influenced the results of the study.In addition, the exact timing for parasympathetic reactivation post-exercise is likely to be highly impacted by an individual's health and training status, and thus it is likely to vary from person to person.However, the within-subject design would have helped to minimize any potential variations attributed to that fact.In terms of methodology, the inclusion of markers for muscle activation and blood/urine samples could have offered valuable insights into the physiological responses to our intervention.However, a conscious decision was made to streamline our study design, prioritizing feasibility and practicality to address our primary research question.Studies with a larger sample size are recommended.Finally, due to technical problems with the HR devices, HRR 2 for session two was not calculated.

Conclusions
The results of this study suggest that 8 weeks of intermittent R T in a hypoxic environment could help to maximize time-efficiency when aiming to improve strength levels in back squat.However, further research is needed.It seems that athletes may lift similar loads (%RM) compared to what can be lifted in normoxia without evoking higher levels of physiological stress.In addition, performing intermittent R T at hypobaric hypoxia may improve the cardiorespiratory response, which in turn could speed recovery.This is of considerable relevance for coaches because it highlights the importance of not only applying the appropriate training loads but also selecting the proper muscular environment during R T .

Fig. 1
Fig. 1 Training regime for the 8 weeks of the resistance training program ) in back squat (BS) and bench press (BP) before (pre) and after (post) the 8-week R T program per group Data are mean ± SD (n = 33) for the 8-week H T program.N = normoxia group, HH = intermittent hypobaric hypoxia group, NH = intermittent normobaric hypoxia group, BS = back squat, BP = bench press, Δ% = the percentage of change (pre-post).* Significant differences between pre and post (P ≤ 0.005).Significant differences (P < 0.05) among groups were noted: α N vs.

Fig. 2
Fig. 2 Change in the training load for the 8 weeks of the resistance training program

Table 4
Heart rate recovery at 2 min (HRR 2 ) per group and per session