Increased maximal oxygen uptake after sprint-interval training is mediated by central haemodynamic factors as determined by right heart catheterization

There is a lack of knowledge regarding the contribution of central and peripheral factors to the increases in VO 2max following sprint-interval training (SIT). This study investigated the importance of maximal cardiac output (Q max ) in relation to VO 2max improvements following SIT and the relative importance of the hypervolemic response on Q max and VO 2max . We also investigated whether systemic O 2 extraction increased with SIT as has been previously suggested. Healthy men and women ( n = 9) performed 6 weeks of SIT. State-of-the-art measurements: right heart catheterization, carbon monoxide rebreathing and respiratory gas exchange analysis were used


Introduction
The effectiveness of sprint-interval training (SIT) in improving maximal oxygen uptake (VO 2max ) is by now undisputed (Gist et al., 2014;Milanović et al., 2015). Numerous studies have shown that the increases in VO 2max induced by SIT are comparable to those after traditional endurance training (TET), despite the lower time commitment (Gist et al., 2014;Milanović et al., 2015). However, uncertainties remain regarding the key adaptive features that contribute to the increase in VO 2max induced by SIT.
The Fick equation states that any change in VO 2max can be explained by a change in maximal cardiac output (Q max ) or/and arteriovenous oxygen difference (a-vO 2 diff). This implies that the training effects observed with SIT should be manifested in central and/or peripheral improvements. Peripheral adaptations involved in oxygen utilization have been extensively studied in the context of SIT. Capillary density (Bonafiglia et al., 2017;Raleigh et al., 2018;Scribbans et al., 2014), maximal citrate synthase activity (Burgomaster et al., 2005(Burgomaster et al., , 2006(Burgomaster et al., , 2008Gillen et al., 2016;Granata et al., 2016;Larsen et al., 2016;Parra et al., 2000;Raleigh et al., 2018), mitochondrial respiration (Saltin et al., 1976;Granata et al., 2016) and mitochondrial density (Larsen et al., 2016) are some of the factors that have been shown to improve following SIT interventions. Some investigators have even gone so far as to claim that the VO 2max improvements induced by SIT can be explained solely by peripheral adaptations, with increased a-vO 2 diff proposed as the main mediator of the improvements (Macpherson et al., 2011;Raleigh et al., 2018). The importance of central adaptations for VO 2max after SIT has been studied to a lesser extent and with conflicting results. There are studies that found no changes in Q max (Macpherson et al., 2011;Raleigh et al., 2018), BV (Jacobs et al., 2013;Laursen et al., 2005) or cardiac remodelling (Jalaludeen et al., 2020) after SIT interventions. Recently, however, we found that a 6-week SIT intervention resulted in improvements in Q max , total haemoglobin mass (tHb), blood volume (BV) and plasma volume (PV) (Mandić et al., 2022), in parallel to the expected increase in VO 2max . Similarly, two other recent studies have demonstrated central adaptations following SIT (Bostad et al., 2021;Satiroglu et al., 2021).
Elucidating the contributions of central and peripheral adaptations to the increase in VO 2max after exercise interventions presents several difficulties, the most obvious being the different indirect methods used to estimate Q max and a-vO 2 diff. Reliance on indirect methods inevitably leads to large variability in the collected data, which, together with the generally small sample sizes, makes experiments underpowered in detecting changes after exercise training interventions. In addition, some of the used methods underestimate the absolute values for Q max , making it challenging to evaluate the findings in a broader physiological context, exemplified by our recent publication (Mandić et al., 2022).
Given the effects on central adaptations found in our previous study (Mandić et al., 2022) and the peripheral adaptations previously reported with SIT (Bonafiglia et al., 2017;Burgomaster et al., 2005Burgomaster et al., , 2006Burgomaster et al., , 2008Gillen et al., 2016;Granata et al., 2016;Jacobs et al., 2013;Larsen et al., 2016;Parra et al., 2000;Raleigh et al., 2018;Scribbans et al., 2014), further investigation of the contribution of these factors is warranted. Accordingly, in this study we aimed to examine the importance of Q max relative to VO 2max improvements after SIT and the relative importance of the hypervolemic response after SIT on Q max and VO 2max . We also examined whether any of the parameters constituting a-vO 2 diff changed with SIT, as has been previously hypothesized (Jacobs et al., 2013;Macpherson et al., 2011;Raleigh et al., 2018). The gold standard method with blood samples from the pulmonary and radial artery during right heart catheterization and the direct Fick method was used to determine a-vO 2 diff and Q max . Phlebotomy was performed to evaluate the significance of exercise-induced hypervolemia. Based on our recent results, we hypothesized that the improvements in stroke volume (SV) and Q max would revert to baseline levels when BV was restored, resulting in a significant decrease in VO 2max . In addition, we hypothesized that a-vO 2 diff would increase following the training intervention and stay unaffected by phlebotomy, mitigating some of the decrease in VO 2max .

Ethical approval
All subjects were informed verbally and in writing about the study before providing written informed consent to participate. The study was approved by the Swedish Ethical Review Authority (ref 2019−0 4492). The study conformed to the standards set by the Declaration of Helsinki, except for registration in a database.

Subjects
In total 12 subjects were recruited for participation in the study. From the 12 subjects, nine (5 female; VO 2max , 30.8 ± 3.2 ml kg min −1 and 4 male; VO 2max , 39.8 ± 5.1 ml kg min −1 ) completed the whole study. The loss of three participants was due to the mitigation strategies employed during the SARS-CoV-2 pandemic which forced us to terminate their participation. Healthy adults with no regular interval training or other structured exercise programme were included in the study (VO 2max , 34.8 ± 6.1 ml kg min −1 ; age, 27 ± 5 years; height, 175 ± 14 cm; weight, 72 ± 16 kg). With the exception of the exercise intervention, subjects were instructed to maintain their usual lifestyle unchanged throughout the study period. They were also asked to refrain from any physical activity or training that could affect the outcome of the study.

Training intervention
The training intervention consisted of 6 weeks of supervised training sessions as previously described (Mandić et al., 2022). The training sessions were performed three times per week and consisted of 10 min of unloaded cycling interspersed with three 30 s sprint-intervals against a braking force equivalent to 7.5% of body weight J Physiol 601.12 on a mechanically braked cycle ergometer (Monark 894E, Varberg, Sweden). The subjects were instructed to pedal as fast as possible against the inherent resistance of the cycle ergometer. Braking force was applied manually when the maximum cadence was reached. Before the first sprint-interval, the subjects completed a short, unloaded warm-up (2.5 min). The subjects were strongly encouraged to exert maximum effort during the intervals. The sprints were separated by a 2 min rest period with unloaded cycling. After the last interval, the subjects cycled for 2 min unloaded, intended as a cool-down.

Pre-and post-intervention measurements
Pre-intervention and post-intervention tests were performed in the week before and the week after the intervention. All procedures were identical in terms of time of day and practice. An overview of the study design can be found in Fig. 1.
The optimized carbon monoxide rebreathing method was used to determine tHb, from which BV and PV later were calculated using haemoglobin concentration [Hb] and haematocrit (Hct) (Keiser et al., 2017). Subjects rested for 15 min before a venous blood sample was taken from the median cubital vein. The sample was immediately analysed for baseline carboxyhaemoglobin (%HbCO), [Hb] and Hct (microcentrifugation, 4 min at 11903 g). End-tidal carbon monoxide (CO) was measured at baseline and after rebreathing using a CO gas analyser (Dräger, PAC 700, Lübeck, Germany). Throughout rebreathing, the same gas analyser was used to check for CO leaks. Subjects breathed a gas mixture of chemically pure (99.97%) CO (0.8 ml kg −1 ) and medical oxygen (AGA, Stockholm, Sweden) for 2 min before disconnecting from the spirometer (Blood tec GmbH, Bayreuth, Germany). Two venous blood samples (1 ml) were taken, one before rebreathing and one 7 min after administration of CO. The samples were then analysed in duplicate for %HbCO (ABL 800, Radiometer A/S, Copenhagen, Denmark),. tHb was calculated as previously described (Prommer & Schmidt, 2007). The coefficient of variation for this method in our laboratory is 1.0-1.6%, which is in agreement with previous publications (Gore et al., 2005;Steiner & Wehrlin, 2011).
To determine VO 2max and maximal workload (W max ), subjects performed an incremental cycling test to volitional fatigue on an electronically braked ergometer (Monark LC6, Vansbro, Sweden) with identical preand post-intervention protocols. Using an online gas collection system (COSMED Quark CPET, Rome, Italy), fractions of inspired and expired O 2 and CO 2 were measured continuously and recorded as breath-by-breath values. The subjects started the test with 50 or 100 W for 5 min as a warm-up. Subsequently, the resistance was increased by 1 W every 3 s, corresponding to an increase of 20 W per minute, until the subjects reached volitional fatigue. VO 2max was considered to be the highest 20 s average reached during the test. Criteria for the test were either a plateau in VO 2 or respiratory exchange ratio (RER) > 1.15, a heart rate within 10 beats of the age-related maximum and a rating of perceived exertion ≥18.
For the pulmonary artery catheterization and radial artery cannulation, subjects were placed in a supine position on the examination table, and the two access sites were sterilely cleaned and covered. Under continuous electrocardiogram and blood pressure monitoring, a flow-guided pulmonary artery catheter was placed percutaneously guided by ultrasound (Vivid E9, GE Healthcare Co., Milwaukee, WI, USA) under local anaesthesia (Carbocain 20 mg ml −1 , Aspen Nordic, Denmark). Using a double Seldinger technique, access to the internal jugular vein was created via an 18-gauge needle (Cordis, Miami Lakes, FL, USA), followed by insertion of a 0.038 inch guidewire. After vein placement was confirmed, an introducer sheath (7F, St Jude/Abbott Medical, Plymouth, MN, USA) was slid over the wire, which was then removed. Finally, a Swan-Ganz balloon-tipped catheter (7F, Edwards Lifescience, Irvine, CA, USA) was inserted and advanced through the right atrium, right ventricle and pulmonary artery in a wedge position before being returned to the pulmonary artery. Combined fluoroscopy (OEC Fluorostar 7900, GE Healthcare Co., Milwaukee, WI, USA) and pressure wave morphology was used to confirm correct catheter placement.
After subcutaneous local anaesthesia at the right wrist (Carbocain 20 mg ml −1 , Aspen Nordic, Denmark), the radial pulse was palpated and the catheter was placed using a guide wire after intra-arterial cannulation. The needle was removed, and the arterial catheter (6F, St Jude/Abbott Medical, Plymouth, MN, USA or 20G, Med FloSwitch, Greiner Bio-One, Kremsmunster, Austria) was inserted into the radial artery via the wire. The guidewire was then removed, the arterial catheter was secured, and the placement of the catheter was checked by flushing with saline. Insertion of the arterial catheter was performed under ultrasound guidance (Vivid E9, GE Healthcare Co., Milwaukee, WI, USA) when palpation was considered difficult. At baseline, a blood sample was obtained from the pulmonary and radial arteries to determine the subject's oxygen saturation and [Hb].
VO 2max and Q max were measured simultaneously at each test occasion, for a total of three times: before the training intervention, after the intervention, and within 10 min following phlebotomy. This means that two VO 2max and Q max measurements were performed on the same day after the intervention. The two measurements were separated by a rest period of 2.5 h. During the rest period, all subjects received a standardized snack containing 1 g of carbohydrate per kilogramme of body weight. Q max was determined according to Fick's principle based on the measured VO 2max and a-vO2diff derived from pulmonary and radial artery blood samples (Table 2). Q max was calculated as VO 2 × (c a O 2 -c v O 2 ) −1 and cO 2 = (1.36 × [Hb] × SO 2 ) + (0.003 × PO 2 ) where [Hb] is the haemoglobin concentration, SO 2 haemoglobin O 2 saturation and PO 2 partial pressure of O 2 , with the samples nearest coinciding with VO 2max used for the calculations. Arterial and venous blood samples were collected simultaneously in 2 ml heparinized syringes (SET Bloodgas, SET Medikal, Istanbul, Turkiye) at near maximal and continuously up until maximal effort during the VO 2max test and were immediately analysed in duplicates (ABL 800, Radiometer A/S, Copenhagen, Denmark). Sampling rate and timepoint were coordinated in order to ensure that the samples taken from the arterial and pulmonary catheter were synchronized.
Phlebotomy was performed through the central venous introducer catheter placed during right heart catheterization. The total amount of blood drawn during phlebotomy was based on each subject's individual increase in tHb with the goal of returning tHb to pretraining levels. On average, 276 ± 206 ml of blood was removed during phlebotomy, representing 5.2% ± 4.1% (range 1.0-13.0) of the subject's total BV. Blood sample volumes, including potential blood loss during the insertion of jugular and radial catheters, prior to phlebotomy, were documented and then taken into account when returning BV to baseline (range 19.5-72 ml, average 37.6 ml). All blood samples were followed by re-infusion of the blood remaining in the catheters' dead space, with saline.

Statistics
All data are presented as means ± SD unless otherwise stated. The effects of training and phlebotomy on VO 2max , Q max , SV, c a O 2 , c v O 2 , W max, and blood gas parameters were derived by mixed linear models with time as a fixed effect and subject/time as a random effect to account for the repeated measures design. Alpha level was set to 5% (P < 0.05) and Tukey's post hoc test was used to assess the differences between the different timepoints. Analysis was performed using the lme4 (Bates et al., 2015) and lmerTest (Kuznetsova et al., 2017) packages in R version 4.1.1. The difference over time in power output during each SIT session was also analysed by linear mixed model. Session × bout was set as the fixed effect and session/subject as the random effect. Here, 'bout' denotes the separate intervals 1, 2 and 3 in each SIT session. Each dependent variable was treated as an independent hypothesis test. To correct for multiple hypothesis tests the Benjamini-Hochberg false discovery rate was calculated, where a false discovery rate of 1% was considered significant. Effects of training on tHb, BV and PV were analysed with paired t tests. Pearson's correlation was used for all correlations presented. Due to a problem with the pulmonary artery sampling during right heart catheterization in one subject, data on Q max , SV, c a O 2, and c v O 2 contain eight observations.

Discussion
Six weeks of SIT increased VO 2max with concomitant increases in BV and Q max . The increases in VO 2max and Q max were causally linked to the hypervolemic response as evident from the normalization of VO 2max and Q max once the SIT-induced increase in BV was restored to pre-intervention levels by phlebotomy. In addition to the intervention-induced effect on central adaptations, c v O 2 decreased following the training and systemic O 2 extraction increased, indicating simultaneous peripheral adaptations.
Based on previous studies where TET has been used as training stimuli, there is compelling evidence that BV and A B C Q max are important determinants of VO 2max (Montero, Cathomen et al., 2015;Montero, Diaz-Cañestro et al., 2015;Sawka et al., 2000). Numerous experiments in which BV has been manipulated either by phlebotomy or artificial expansion, acutely or following TET interventions, have demonstrated strong causal relationships between the manipulated BV and VO 2max (Balke et al., 1954;Bonne et al., 2014;Celsing et al., 1987;Ekblom et al., 1972;Krip et al., 1997;Montero, Cathomen et al., 2015;Skattebo et al., 2021). In the present study, using a similar experimental set-up involving phlebotomy previously used to evaluate the effects of exercise-induced hypervolemia on VO 2max and Q max changes following TET (Bonne et al., 2014;Montero, Cathomen et al., 2015), we observed expected increases in BV, Q max and VO 2max ( Fig. 2 and Fig. 3). Furthermore, the observed increases in VO 2max and Q max were eliminated when BV was normalized by phlebotomy, whereas the training-induced increase in systemic O 2 extraction and decrease in c v O 2 were unaffected by phlebotomy (Fig. 3C, G). Thus, despite the difference in training regimen, our results are similar to earlier reports using TET as training stimuli (Bonne et al., 2014;Montero, Cathomen et al., 2015). Our findings suggest that the underlying adaptations leading  to an increase in VO 2max following SIT and TET are more similar than previously reported and show that central adaptations play an important role in SIT-induced increases in VO 2max . However, there are some interesting discrepancies in the literature. Recently, in a 10-week TET intervention, phlebotomy did not have any effect on VO 2max or Q max (Skattebo, Bjerring et al., 2020). The discrepancy between these results and others (Bonne et al., 2014;Montero, Cathomen et al., 2015), including the present study, may be explained by the differences in BV expansion and thus by the smaller reduction required to return BV to baseline values. In the mentioned study, 181 ml (3.7%) blood had to be drawn to normalize BV (Skattebo, Bjerring et al., 2020), compared with 382 ml (7%) (Bonne et al., 2014) and 310 ml (5.9%) (Montero, Cathomen et al., 2015) after 6 weeks of TET or, in our case 276 ml after 6 weeks of SIT. Elegantly, the same group of investigators later demonstrated that phlebotomy of 450 ml but not 150 ml impaired VO 2max and exercise capacity, suggesting the existence of compensatory mechanisms able to  counteract the effects of small BV losses (Skattebo et al., 2021). This is supported by previous work showing that small PV losses due to sweating do not affect VO 2max or Q max (Saltin, 1964;Saltin & Stenberg, 1964). In addition, it should be noted that there is a possibility that our subjects were slightly hypovolemic during phlebotomy because of an initial calculation error that resulted in an inflated phlebotomy corresponding to 9 g of haemoglobin. The effect of this error was most likely partially mitigated since all blood loss during the post-intervention assessment, preceding the phlebotomy trial by 2.5 h, was documented and then accounted for when returning BV to baseline. With this in mind, it is important to consider that the presented non-significant difference in VO 2max between pre-intervention and after phlebotomy might be a product of the mild hypovolemia that the subjects experienced in combination with the sample size. However, we do not believe that this changes the main conclusion that key adaptations such as BV must be included in an explanatory model aiming to explain the improvements in VO 2max after SIT.
Highlighting the similarity between SIT-and TET-induced adaptations and their importance for VO 2max , the present data agree well with the linear regression and the curvilinear fit proposed by Skattebo and colleagues to describe the relationship between BV reductions and VO 2max (Skattebo et al., 2021) (Fig. 6). Although peripheral adaptations are undoubtedly important for long-term adaptations after any exercise intervention that is expected to increase VO 2max and aerobic performance, as evident from the data on systemic O 2 extraction and c v O 2 in the present study, our results refute previous hypotheses of a solely peripherally driven increase in VO 2max after SIT (Macpherson et al., 2011;Raleigh et al., 2018). It is surprising how Figure 6. The relationship between amount of blood phlebotomized and its effect on VO 2max Figure adapted from Skattebo et al. (2021). Studies (Balke et al., 1954;Bonne et al., 2014;Krip et al., 1997;Montero, Cathomen et al., 2015;Skattebo, Bjerring et al., 2020 included in the analysis, in addition to the data presented in the current study. much attention has been paid to the potential effects of peripheral adaptation on VO 2max after interval training. The distinct differences between SIT and TET when it comes to training variables such as duration and intensity may be the reason for the great emphasis that has been put on peripheral adaptations in the context of SIT (Jacobs et al., 2013;Macpherson et al., 2011;Raleigh et al., 2018). Numerous studies have demonstrated that peripheral adaptations occur concomitantly with increases in VO 2max , but no study has shown that the increases in, for example, capillary density (Bonafiglia et al., 2017;Raleigh et al., 2018;Scribbans et al., 2014), maximal citrate synthase activity (Burgomaster et al., 2005(Burgomaster et al., , 2006(Burgomaster et al., , 2008Gillen et al., 2016;Granata et al., 2016;Parra et al., 2000;Raleigh et al., 2018) or mitochondrial respiration (Granata et al., 2016) translate to an increase in systemic a-vO 2 diff that could explain improvements in VO 2max of ≈ 10%, which are frequently observed after SIT interventions in untrained or moderately trained subjects (Rosenblat et al., 2022). In the present study, mixed venous oxygen saturation (S v O 2 ) at VO 2max prior to the exercise intervention was on average 23.9 ± 6.2% and decreased to 21.0 ± 6.1% following the training intervention. Simultaneously, we observed an increase in systemic O 2 extraction (Fig. 3G), driven by the decreased c v O 2 (Fig. 3C) and in contrast to SV and Q max , it was unaffected by phlebotomy. While studies in athletes, sampling blood from the right atrium, have shown that S v O 2 and c v O 2 can reach values as low as 10% and 2.01 ml dl −1 respectively, it seems unlikely that SIT interventions lasting 2−12 weeks would elicit peripheral improvements matching the data from elite cross-country skiers (Calbet et al., 2005). In line with our findings, increases of similar magnitude in O 2 extraction have also been observed following other types of exercise interventions (Beere et al., 1999;Boushel et al., 2014;Klausen et al., 1982;Roca et al., 1992;Rud et al., 2012). Our data also fit fairly well with earlier proposed relationships between S v O 2 , c v O 2 and VO 2max (Skattebo, Calbet et al., 2020). Taken together this would imply that the peripheral adaptations commonly reported after SIT result in improvements reflected by increased O 2 extraction. However, without a substantial increase in O 2 delivery in the form of increased Q max , the magnitude of these adaptations cannot explain all of the increases in VO 2max observed with SIT (Rosenblat et al., 2022). In the current study, the BV expansion accounted for a substantial part (≈ 55%) of the increase in VO 2max . Factoring in the potential additive role of cardiac remodelling and/or changes in afterload to SIT-induced improvements in VO 2max (Matsuo et al., 2014) underscores the importance of central adaptations to the improvements in VO 2max following SIT.
One explanation for the conflicting results in cardiovascular changes after different exercise interventions in general and for SIT research in particular is the difficulty J Physiol 601.12 in measuring cardiovascular adaptations such as Q max , SV and a-vO 2 diff. One of the main strengths of the present study is that factors constituting a-vO 2 diff were assessed directly by performing right heart catheterization with blood sampling from the pulmonary artery. Most studies use the Fick equation with measurements of VO 2max and indirect estimates of Q max . The calculated and indirect methods are subject to a number of limitations, the most problematic being the measurement of Q max . There are various methods for indirect estimation of Q max , but common to all are methodological problems that result in either consistent over-or underestimation of Q max (Siebenmann et al., 2015) which in turn has a significant impact on calculated a-vO 2 diff. Although the direct Fick method, through right heart catheterization, is the gold standard method for assessment of Q max and a-vO 2 diff, the method comes with a few limitations. Firstly, the invasive nature of the measurement makes it hard to obtain large sample sizes and the catheter can on rare occasions slip away from the pulmonary artery during cycling, resulting in inadequate blood sampling, as evident from the missing data in one of our subjects. To make the results as externally valid as possible with regards to applicability to the general population, subjects of both sexes with varying baseline fitness were included, but the subjects investigated are both younger and healthier than the general population which also warrants consideration when generalizing the findings of this study. From a statistical standpoint, n = 9 is adequate from a power perspective due to the relatively large effect sizes of the main outcome variables (BV, Q max and VO 2max) but quantitatively smaller and less consistent changes in secondary variables such as arterial O 2 content or heart rate will go undetected. In addition, we recognize that our measurements have not assessed peripheral a-vO 2 diff or changes in local blood flow regulation. Future studies examining these factors using femoral catheters in the context of SIT are awaited.
In conclusion, the current study demonstrates a causal relationship between the training-induced changes in BV, Q max, and VO 2max . This confirms and underscores the importance of central haemodynamic factors such as BV and Q max in SIT-induced increases of VO 2max . In addition, the decrease in c v O 2 caused by an increase in systemic O 2 extraction highlights that peripheral adaptations occur in parallel, which also contribute to the improvement in VO 2max following SIT. More importantly, our results suggest that the contribution of central and peripheral factors to the improvement in VO 2max is similar between SIT and what has been reported for other training modalities. This encourages future research to investigate regulatory mechanisms that could explain how such different training modalities can lead to similar improvements in VO 2max and Q max despite the large differences in training stimuli.