Five weeks of heat training increases haemoglobin mass in elite cyclists

What is the central question of this study? Do haemoglobin mass and red blood cell volume increase in elite cyclists training in a hot environment compared to a control group training at normal temperature? What is the main finding and its importance? Five weeks of heat training increases haemoglobin mass in elite cyclists. There are small to intermediate effect sizes for exercise parameters favouring heat training.

Hb mass may be increased following exercise training in previously untrained individuals, with improvements averaging 10% (Bonne et al., 2014;Montero et al., 2015Montero et al., , 2017. A detailed review of exercise training-induced erythropoiesis has been published recently (Montero & Lundby, 2018). Hb mass may, however, be augmented by means other than exercise training, and environmental influence seems the most well established. Since chronic exposure to high altitude may augment Hb mass (Siebenmann et al., 2015;Siebenmann, Robach, & Lundby, 2017b), elite athletes who already possess very high Hb mass (Lundby & Robach, 2015) may aim at obtaining even higher values by means of altitude training, although this practice has been questioned in more recent times (Lundby & Robach, 2016). In contrast to chronic altitude exposure, in which the expansion of Hb mass is preceded by a rapid reduction in plasma volume (PV), exposure to a hot environment leads to an expansion of PV of up to 20% within days, which thereafter stabilizes at around +10% (Sawka, Convertino, Eichner, Schneider, & Young, 2000). While such an expansion may not influence exercise capacity despite potentially facilitating a higher maximal cardiac output, due to a concomitant reduction in blood O 2 content, we speculate that if the PV expansion can be sustained for several weeks, this may eventually also lead to an increase in Hb mass , which in turn could facilitate exercise performance.
The underlying rationale for our hypothesis is that the kidney may be seen as a 'critmeter' , where both PV and RBCV are proposed to be counter-regulated in response to changes in either, with the aim to keep the haematocrit (Htc) within a normal range (Donnelly, 2001;Montero & Lundby, 2018). In essences, RBCV is regulated through erythropoietin (EPO) synthesized in response to a drop in tissue O 2 pressure by peritubular fibroblast-like cells upon stabilization of hypoxia-inducible factor-2α within the juxtamedullary region.
Uniquely for the kidney, tissue O 2 pressure can be modulated by (1) changes in renal O 2 consumption, fundamentally dependent on sodium reabsorption, which in turn is proportional to glomerular filtration rate and renal blood flow, and (2) O 2 delivery to the proximal tubule, which is determined by arterial O 2 content as well as blood flow.
Since flow contributes to both sides of the balance governing tissue O 2 pressure, EPO production must be determined by the difference between O 2 consumption and arterial O 2 content (Montero & Lundby, 2018). Thus, O 2 sensors located within the juxtamedullary apparatus are proposed to regulate Htc via modulation of renal EPO production according to arterial O 2 content-dependent changes in tissue O 2 pressure (Donnelly, 2001;Montero & Lundby, 2018). Despite our critmeter focus, other factors also could facilitate, or at least contribute to, a potential increase in RBCV. While the master regulator of the hypoxia inducible factor (HIF) system clearly is oxygen, also heat shock protein (HSP) expression may facility HIF stabilization and hence EPO synthesis. For an extensive account of this topic, the reader is referred to recent reviews (Ely, Lovering, Horowitz, & Minson, 2014;Hawley, Lundby, Cotter, & Burke, 2018).
Based on the above we therefor speculate that a heat traininginduced increase in PV, and accordingly a reduction in Htc, may stimulate EPO synthesis and thereby expand Hb mass . Since the increase in Hb mass may take several weeks with exercise training (Montero

New Findings
• What is the central question of this study?
Do haemoglobin mass and red blood cell volume increase in elite cyclists training in a hot environment compared to a control group training at normal temperature?
• What is the main finding and its importance?
Five weeks of heat training increases haemoglobin mass in elite cyclists. There are small to intermediate effect sizes for exercise parameters favouring heat training. et al., 2017), 3-4 weeks with altitude exposure (Siebenmann et al., 2015), and 2-3 weeks following injections of supra-physiological levels of recombinant human EPO (Lundby, Achman-Andersen, Thomsen, Norgaard, & Robach, 2008), we have opted for a heat training period lasting 5 weeks. In the present study we applied a heat training strategy which is similar, albeit longer in total length, to a protocol with which we previously have observed an expansion in PV of approximately 200 ml (or 6%) following 10 days' intervention . We have furthermore tested the approach within a limited number of active cyclists, where Hb mass became significantly increased by +34 g vs. +2 g in control group, although no statistical group differences could be established . The lack of statistical difference between groups was likely the result of a rather large individual response. Nonetheless, considering the initial promising results, but also realizing that this type of training may have little relevance in amateur sport, and that physiological adaptations may be even more difficult to induce in elite athletes, it is necessary to also test a given training regime in elite athletes before such a training modality is implemented in this population.
With the aim to test whether heat training can increase Hb mass in elite endurance athletes, we enrolled a group of elite cyclists to a heat training group and a control group. Both groups underwent an extensive test battery before and after the 5-week intervention period including blood volume and exercise performance assessments.

Ethical approval
The study was performed according to the ethical standards established by the Declaration of Helsinki 2013 and was approved by the local ethical committee at Lillehammer University College (MR241018). The study was not registered in a database. All cyclists signed an informed consent form prior to participation.

Participant characteristics
The number of subjects to be included was based on experience from a previous study with a similar heat exposure profile . Twenty-three male cyclists (V O 2 max = 76.2 ± 7.6 ml min −1 kg −1 ) expressed an interest in volunteering for the study. They were matched in pairs based onV O 2 max and randomly assigned to either control or intervention group, and none of them dropped out. Based on the peak aerobic power output (W max ),V O 2 max and training characteristics, the cyclists were regarded elite (Jeukendrup, Craig, & Hawley, 2000). The cyclists were matched to create two homogeneous groups based oṅ V O 2 max : a heat training group (HEAT; n = 11, age = 19 ± 2 years, body height = 178 ± 8 cm, body mass = 68.6 ± 6.9 kg) and a control group (CON; n = 12, age = 19 ± 3 years, body height = 179 ± 5 cm, body mass = 70.8 ± 5.6 kg). Two of the cyclists in CON did not perform the exercise testing due to logistic problems with weekend scheduling and hence for these measurements n = 10 in CON while for the remaining measurements n = 12.

Experimental design
Participants were tested (test procedures described later in this section) before and after a 5-week intervention period.

Training intervention
The HEAT group performed the heat session as an afternoon session 5 times a week for the first 4 weeks and 4 times during the fifth week using their own bicycle connected to electromagnetically braked rollers (Computrainer Lab, Racer Mate Inc., Seattle, WA, USA). The session started with 5 min easy spinning before the rollers were calibrated to quantify and adjust wheel-ergometer rolling resistance to 1.0-1.4 kg, as prescribed by the manufacturer. Thereafter the participants aimed at performing 50 min cycling at 45% of power output at 4 mmol l −1 [La − ]. However, due to individual differences, the power output was upregulated with 25 W for the next session if rating of perceived exertion (RPE) on the 6-20 scale (Borg, 1982) was ≤11 and downregulated with 20 W for the next session if RPE was ≥15.
During all heat sessions, participants consumed 0.5 litre water (at no specific time point or rate). The mean temperature and humidity during the first 2.5 weeks was 37.5 ± 0.3 • C and 66 ± 2%, respectively, while it was 38.5 ± 0.2 • C and 64 ± 1%, respectively, for the last 2.5 weeks.
The participants were encouraged to rehydrate after the heat session and to consume extra fluid until their urine had a normal pale yellow or straw colour (Armstrong et al., 1994). The CON group also performed a supervised afternoon session 5 times per week for the first 4 weeks and 4 times during the fifth week with the same duration and RPE target and measurements as HEAT using their own bicycle connected to electromagnetically braked rollers (Computrainer Lab). In addition to the afternoon sessions, both HEAT and CON perform normal training during the morning session.
During the training intervention there were no differences between HEAT and CON in mean weekly duration of the endurance training and the distribution of this training into low intensity training (zone 1, <55% of functional threshold power (FTP); 6.52 ± 1.75 vs.

Exercise tests
An overview of the exercise test protocol is given in Figure 1. (1) 5 min warm-up, (2) 1RM leg press, (3) 5 min easy cycling, (4) lactate profile test, (5) 5 min rest/easy cycling, (6)V O 2 max test, (7) 5 min rest/easy cycling, (8) 30 min at power output equal to 2.0 mmol l −1 [La − ], (9) 5 min bouts at power output equal to 3rd and 2nd last workload from lactate profile test, and (10)  respectively) with 1 min rest between sets were performed before the 1RM assessment. The smallest increase in 1RM load allowed was 5 kg and the test was terminated after failure of two lifts in a row. After a 5 min recovery the blood lactate profile test was started.
The cycling testing started with a blood lactate profile initiated with 5 min cycling at 125 W followed by 50 W increases every 5 min.
Blood was sampled from a fingertip at the end of each 5-min bout and  (Noordhof, Skiba, & de Koning, 2013), and this was divided by the power output and multiplied by 100.

Haematology
Hb mass and intravascular volumes were assessed on four occasions: two pre-intervention (Pre-1 and Pre-2) and again upon termination of the intervention period (Post-1, obtained 1-2 days after intervention; and Post-2, obtained 3-4 days after last intervention). Pre-1 and Pre-2 were averaged into one single point, Pre, and also Post-1 and Post-2 were averaged into a single point, Post. All measurements were completed with an automated blood volume analyser based on CO rebreathing (Detalo Performance, Detalo Health, Birkerod, Denmark).
After arrival at the laboratory, the participant was placed in supine position for 20 min and capillary blood was sampled from a fingertip and venous blood from a forearm vein. Blood was analysed immediately in triplicates for (1)  Linear regression analyses were performed to assess correlates of changes in Hb mass . Furthermore, the effect size (ES) was calculated as Cohen's d by using the mean pre-post change in HEAT minus the mean pre-post change in CON, divided by the pooled pre-test standard deviation (Morris, 2008). A two-tailed P-value less than 0.05 was considered significant.
F I G U R E 2 Hb mass and blood volumes. (a) Hb mass (g), (b) plasma volume (PV, mL), (c) red blood cell volume (RBCV, mL), and (d) total blood volume (BV, mL) in Heat and Control participants before (pre) and after (post) the 5-week intervention period. Each dashed line represents one study participant and the thick line represents the group average. *P < 0.05 for time and group interaction

Haematological adaptations
The main haematological parameters determined in this study are presented in Table 1 and Figure 2. There were no differences between groups in any of these variables prior to the start of the intervention.
While the determinations of [Hb], Htc, total protein and albumin revealed no changes pre-vs. post-intervention, blood osmolality analysis revealed a time-group interaction (P = 0.005). Following the 5- week intervention period, Hb mass was +4.6% higher in the HEAT group but remained unchanged (+0.5%) in the CON group, and statistical analysis revealed a time-group interaction (Figure 2, P < 0.05). A time effect was observed for plasma volume, which was altered by 4.8% (188 ml ± 246) and 2.1% (84 ml ± 2.1) in HEAT and CON, respectively (P = 0.034). A group effect could, however, not be found. There was a positive correlation between changes in plasma volume and changes in Hb mass across the two groups (r = 0.54, P = 0.01). Red blood cell volume and blood volume remained unaffected by the intervention. Similar results were found using ANCOVA with baseline values as covariates (see Table 3).
There were no changes in W max in any of the groups. HEAT increased power output at 4 mmol l −1 [La − ] (P = 0.035), while no changes occurred in CON (9.1 ± 12.4 vs. −0.4 ± 5.1%, respectively), but there were no group differences. However, the effect size showed a small benefit of HEAT vs. CON (ES = 0.34; Table 2). There were no changes or group differences in GE fresh and [La − ] fresh , but HEAT achieved a larger reduction in [La − ] fatigue than CON (−15.7 ± 24.9 vs. 15.9 ± 26.7%, respectively, P = 0.038), which was accompanied by a large effect size favouring HEAT (ES = 1.04; Table 2). There were no changes or group differences in GE fatigue from pre to post, but there was an intermediate effect size in favour of HEAT (ES = 0.52; Table 2). Both HEAT and CON increased mean power output during the 15 min cycling trial (6.9 ± 8.4 vs. 3.4 ± 5.1%, respectively), with no differences between groups ( Table 2) (Table 2).

TA B L E 2
Data from the submaximal-, maximal incremental exercise and 15 min cycling tests before (Pre) and after the intervention period (Post) in the heat training group (HEAT) and the control group (CON); the magnitude of improvements of HEAT vs. CON is also shown HEAT (n = 11) CON (n = 10) ANOVA

Pre Post Pre Post
HEAT vs.

CON (ES) Time Group Interaction
Body mass (kg) 67.6 ± 6.9 67.7 ± 6.8 70. Values are means ± SD. * Different from Pre (P < 0.05). The relative change from Pre is larger than in CON (P < 0.05). Power at 4 mmol l −1 : power output at a blood lactate concentration of 4 mmol l −1 ; %V O 2 max at 4 mmol l −1 : fractional utilization ofV O 2 max at the power output at 4 mmol l −1 ; GE fresh and [La − ] fresh : gross economy and blood lactate concentration at the second last bout of the blood lactate profile; GE fatigue and [La − ] fatigue : gross economy and blood lactate concentration at the power output from second last bout of the blood lactate profile after the prolonged cycling.V O 2 max : maximal oxygen consumption; RER: respiratory exchange ratio; HR peak : peak heart rate; [La − ] end : blood lactate concentration 1 min after exercise; RPE: rate of perceived exertion; W max : peak aerobic power output; Power 15min : mean power output during 15 min cycling test;V O 2 mean : mean oxygen consumption during the 15 min cycling test; %V O 2 max at 15 min test: fractional utilization ofV O 2 max during the 15 min cycling test.
Both HEAT and CON increased 1RM leg press (from 315 ± 39 to 327 ± 39 kg and from 296 ± 46 to 309 ± 42 kg, all P > 0.05), with no group differences. Similar results were found using ANCOVA with baseline values as covariates (Table 3).

DISCUSSION
In the present study elite cyclists were exposed to 5 weeks of heat training (

Heat training and blood volume expansion
Acclimatization to a hot environment leads to hypervolaemia, which for a long time was thought exclusively to be the response of an increase in PV (Sawka et al., 2000). Recent studies, however, suggest that also RBCV becomes elevated (Karlsen et al., 2015a;Oberholzer et al., 2019).
With chronic heat exposure albumin moves from the intervascular to the extravascular space and as a consequence water follows and PV is increased (Sawka et al., 2000). However, other mechanisms may also be involved and chronic heat exposure is not necessarily needed to expand PV as PV also becomes increased when exercise training is conducted in a hot environment while at the same time residing in a cool environment -i.e. repeated, intermittent short duration heat exposure combined with exercise training. Accordingly, Nielsen et al. (1993) found PV to increase by 13% following 9-12 days with exhausting exercise training conducted at 40 • C. Similarly, Lorenzo and co-workers (2010) found PV to increase by 6.5% following 10 days with light exercise training conducted at 40 • C. In a similar manner, we have previously found PV to be elevated by 6% following 10 days of similar heat training as in the present study . The magnitude of acute increase in PV has been suggested to be related to the magnitude of fluid loss during heat exercise (Akerman, Lucas, Katare, & Cotter, 2017). Despite the increase in PV being a robust observation following heat training, we have recently conducted a 5- week heat training study (i.e. the longest heat training intervention in duration to date) in which both control and heat group experienced a PV increase (188 and 304 ml, respectively) with no statistical differences between groups . In line with the latter, but at odds with most previous acclimatization studies, we observed a main effect of time for PV increases, but no differences between groups. It should also be noted that we previously have validated the CO rebreathing approach for the determination of PV (Keiser et al., 2017) and that we have used this approach in both studies Oberholzer et al., 2019). The lack of measurable effects on PV with these long duration studies could be related to the fact that with chronic heat exposure a peak in PV is typically reached after about 5-10 days (20% increase) whereafter PV declines again (∼10% increase) (Sawka et al., 2000). If this were the case in the present study and if at the same time accepting the critmeter theory, it must also be assumed that the erythropoietic stimulus is greatest in the initial phase of heat acclimatization. One other obvious difference between previous studies (Karlsen et al., 2015b;Oberholzer et al., 2019) and the current study is that the participants in the present study were much fitter, with an averageV O 2 max of 76.2 ± 7.6 ml min −1 kg −1 .
In this regard, we have previously reported that the expansion of red blood cell volume at altitude is less in already fit individuals with high blood volumes , but whether this observation can be transferred to PV and heat exposure remains speculative.
Instead of exposing study volunteers to heat in a climatic chamber as done in the current and above-mentioned studies, Karlsen and coworkers (Karlsen et al., 2015b) exposed non-heat acclimated Danish cyclists to outdoor exercise training in Qatar (36 • C) while for the reminder of the day the volunteers where kept indoors and under airconditioned conditions. Interestingly, after the 2-week intervention period they reported an increase in PV of 13.5% and a concomitant increase in Hb mass of nearly 60 g. The latter is to the best of our knowledge the first study demonstrating an increase in Hb mass (or red blood cell volume) with heat acclimatization. In a similar manner, we have recently observed a tendency (time × group interaction: P = 0.061) for a larger increase in Hb mass following 5 weeks of heat training (5 sessions per week for 1 h at 40 • C) compared to a control group . While the average response tended to be higher in the intervention group, it should be noted that individual responses varied markedly (+34 ± 36 g in HEAT and +2 ± 33 g in CON).
In the current study, we found Hb mass to increase by 42 g (4.6%) after 5 weeks of heat acclimatization (RBCV P = 0.081 for it to increase in HEAT). Intriguingly, the observed increase in Hb mass in the present study is similar to the 5% improvement we have observed after 4 weeks' acclimatization to 3450 m altitude in healthy but untrained study volunteers (Siebenmann et al., 2015), but far greater than what we have observed in endurance athletes with normobaric hypoxic live high-train low (LHTL) (Siebenmann et al., 2012) or LHTL at natural altitude where we have observed no changes (Robach et al., 2018).

What could be the mechanisms behind the increase in Hb mass ?
Our hypothesis was that the increase in Hb mass would be preceded by an increase in PV and hence reduction in Htc and that this would facilitate the synthesis of EPO according to the kidneys 'critmeter' function (Montero & Lundby, 2018). In line with our hypothesis, there was a positive correlation between changes in plasma volume and changes in Hb mass across the two groups (r = 0.54, P = 0.01). In a previous heat study, we have also observed a positive correlation (r = 0.493, P = 0.023) between changes in PV and Hb mass when pooling the heat and the control group . In line with the latter, we have also in a previous exercise training study in untrained individuals observed that the increase in Hb mass was preceded by decreases in Htc . While our hypothesis was based on the critmeter function of the kidney, other mechanisms may be involved of course. The rate of EPO synthesis is mainly regulated by the HIF system that is globally known to be ubiquitinated in normoxic conditions but stabilized in hypoxic conditions. However, the HIF system has also been shown to stabilize with increased HSP expression (Maloyan et al., 2005) and it may hence be speculated that the augmented Hb mass occurred secondary to a heat exposure-induced increase in HSP. For a detailed discussion on HSP expression with heat exposure and its relation to exercise performance, the reader is referred to two recent reviews on that matter (Ely et al., 2014;Hawley et al., 2018).
Since the current study was conducted in elite athletes, blood sampling was kept at a strict minimum which circumscribed the determination of EPO or other candidate erythropoietic hormones such as angiotensin-2, which could have shed light on the underlying mechanisms for the expansion in Hb mass . It has also been suggested that noradrenaline has erythropoietic capabilities (Schrier, 1974), and although we did not determine noradrenaline levels, it is well established that hyperthermia increases sympathetic nervous activity (Nielsen et al., 1993).

Heat acclimatization and exercise performance in normal temperature conditions
The larger increase in Hb mass in HEAT in the present study did not lead to a greater increase inV O 2 max or W max . Considering that for each gram of haemoglobinV O 2 max may increase about 4 ml, it would be expected that exercise performance would become increased in those studies in which an increase in Hb mass is reported (Prommer & Schmidt, 2010). Thus the 60 g increase observed by Karlsen et al. (2015b) could theoretically increaseV O 2 max by a substantial 240 ml min −1 , but did not. Similarly, the 34 g increase observed by Oberholzer et al. (2019) did not result in a concomitant increase inV O 2 max . However, in the present study the change inV O 2 max was in the same direction as the development of Hb mass , i.e. the 42 g increase in HEAT was accompanied by a 225 ml min −1 increase inV O 2 max , while the 6 g difference in CON was accompanied by 161 ml min −1 increase inV O 2 max (non-statistical differences between groups). The lack of difference between groups inV O 2 max improvements despite differences in augmented Hb mass is at odds with exercise training studies in previously untrained individuals conducted in a thermoneutral environment in which most of the improvement inV O 2 max can be ascribed to concomitant improvements in blood volume/RBCV (Bonne et al., 2014;Montero et al., 2015).
While there were no differences between groups in [La − ] fresh , HEAT reduced [La − ] fatigue to a larger extent than CON. The latter observations are in line with previous findings in well-trained endurance athletes, indicating that favourable adaptations at submaximal exercise intensities are more challenging to induce in the fresh state than after prolonged exercise (Øfsteng et al., 2018;Rønnestad et al., 2011;Vikmoen, Rønnestad, Ellefsen, & Raastad, 2017). The lack of an effect of HEAT on GE in the fresh state is in accordance with heat training studies lasting 2-5 weeks performed on trained cyclists (Karlsen et al., 2015b;Mikkelsen et al., 2019). Despite minor statistical differences between groups, the numerical differences observed at submaximal exercise favours HEAT, with ∼3% increase in power output Taking the high training status of the cyclists and the relatively short intervention period into account, these numerical differences can be viewed as relevant for elite exercise performance and this is supported by the effect size which elucidates a small to intermediate effect of HEAT vs. CON. In accordance with the current study, previous studies on endurance-trained participants have observed improved power output at lactate threshold in normal temperature after heat training (Lorenzo, Halliwill, Sawka, & Minson, 2010), although two of these did not include a control group (Neal, Corbett, Massey, & Tipton, 2016;Rendell et al., 2017). Similarly, a reduction in aerobic metabolism, glycogen utilization and [La − ] in normal temperature has been observed after a period of heat training (Young, Sawka, Levine, Cadarette, & Pandolf, 1985). However, due to lack of control group and the rather poor training status of the included participants, care should be taken in the interpretation of these results and the relevance for elite exercise performance may be questioned.
The present observation of a small effect size of HEAT on power output at 4 mmol l −1 [La − ] can be speculated to be related to the increase in Hb mass , which increases the oxygen availability to the exercising skeletal muscles. Furthermore, it has been observed that repeated exposure to passive heat stress increases skeletal muscle mitochondrial function (Hafen, Preece, Sorensen, Hancock, & Hyldahl, 2018) and capillarization (Hesketh et al., 2019). Even though the latter studies were performed on sedentary participants, they nonetheless indicate that HEAT may have induced peripheral adaptations that may have contributed to the small beneficial effect of HEAT. Lastly, it has been suggested that increased blood volume, as observed in HEAT (with a small ES), may lead to increased blood flow through the splanchnic circulation and thereby increasing lactate removal (Lorenzo et al., 2010) or may also simply be related to a dilution of lactate from blood volume expansion, both with a consequence of delayed blood lactate accumulation.
Power output at lactate threshold has been shown to be a good predictor of performance in endurance events (Coyle et al., 1991;Jacobs et al., 2011) and is sensitive to changes in performance (Jones, 1998).
The observed small effect size of HEAT on mean power output during the final 15 min all-out test was therefore expected. The larger oxygen delivery capacity due to the ∼5% increase in Hb mass in HEAT is an obvious candidate to partly explain these improvements. In support thereof, we observed a positive correlation between increases in mean power output during the 15 min cycling test and increases in Hb mass (r = 0.47, P = 0.03). Although differences in 15 min test between groups did not reach statistical significance, it can be speculated that the numerical increase observed in HEAT of ∼7% has performance relevance compared to the ∼3% increase in CON in this group of highly trained cyclists. In line with the latter, it has previously been documented that heat training induces larger improvements in endurance performance than in a control setting (Lorenzo et al., 2010). In support, studies have also observed ergogenic effects of heat training on endurance performance in trained cyclists (Neal et al., 2016;Rendell et al., 2017), although the already mentioned lack of control groups requires careful interpretations. To the contrary, other studies on trained cyclists have not found any difference in endurance exercise capacity between a heat-trained group and a control group (Karlsen et al., 2015b;Keiser et al., 2015;Mikkelsen et al., 2019). The study by Mikkelsen et al. (2019) applied a similar intervention duration as the present study, but did not find any advantage of heat training compared to their control group. However, in the present study we observe a small effect size of HEAT on certain physiological determinants of endurance performance (mean power during 15 min trial) which were not tested in the previous study. The reason(s) for this discrepancy remains unknown, but we can point at some small differences that might contribute to explaining it. The present participants had a higher fitness level (V O 2 max : ∼76 vs. ∼60 ml min −1 kg −1 ; W max : ∼6.3 vs. ∼5.1 W kg −1 ), and were younger (∼19 vs. ∼39 years). During the heat sessions, the participants in the present study were limited to drinking 0.5 litres of water (to ensure dehydration), while they were allowed to consume water ad libitum in the Mikkelsen et al. (2019) study. The latter could have induced a smaller degree of dehydration with a potentially smaller stimulus to increase Hb mass and potentially contribute to explaining the slightly larger Hb mass increase in the present study (∼5 vs. ∼3%). Lastly, endurance performance in the present study was measured during slightly warmer conditions than in the Mikkelsen et al. (2019) study (∼18 vs. ∼14 • C) and was performed at the end of a prolonged exercise protocol, while it was performed on a separate day in Mikkelsen et al. (2019).
Finally, since it is established that both normal endurance training, especially high intensity aerobic training (Laursen, 2010), and heavy strength training (Rønnestad & Mujika, 2014) can improve endurance performance, it is important to stress that there were no difference between HEAT and CON in conducted training volume, intensity distribution or amount of heavy strength training. Accordingly, there were no group differences in 1RM leg press. Therefore, is it likely that the numerical advantage and small to intermediate effect size of HEAT is due the heat training.

Perspective
The usual pitfall with studies on interventions that may or may not favour elite exercise performance is that these are not always conducted in elite individuals. In the current study we argue that this limitation was overcome by selecting participants with an already highV O 2 max of 76.2 ml min −1 kg −1 . Albeit this is the first study to demonstrate increases in Hb mass following laboratory-based heat training in such athletes, it is tempting to advocate this strategy to Olympic calibre/champion athletes also. However, as with other new/unfamiliar training strategies for elite performance, this should be adequately tested by the given athletes before it is implemented before important sporting events.
be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.