Sub‐maximal aerobic exercise training reduces haematocrit and ameliorates symptoms in Andean highlanders with chronic mountain sickness

New Findings What is the central question of this study? What is the effect of sub‐maximal aerobic exercise training on signs and symptoms of chronic mountain sickness (CMS) in Andean highlanders? What is the main finding and its importance? Aerobic exercise training (ET) effectively reduces haematocrit, ameliorates symptoms and improves aerobic capacity in CMS patients, suggesting that a regular aerobic ET programme might be used as a low‐cost non‐invasive/non‐pharmacological management strategy of this syndrome. Abstract Excessive erythrocytosis is the hallmark sign of chronic mountain sickness (CMS), a debilitating syndrome associated with neurological symptoms and increased cardiovascular risk. We have shown that unlike sedentary residents at the same altitude, trained individuals maintain haematocrit within sea‐level range, and thus we hypothesise that aerobic exercise training (ET) might reduce excessive haematocrit and ameliorate CMS signs and symptoms. Eight highlander men (38 ± 12 years) with CMS (haematocrit: 70.6 ± 1.9%, CMS score: 8.8 ± 1.4) from Cerro de Pasco, Peru (4340 m) participated in the study. Baseline assessment included haematocrit, CMS score, pulse oximetry, maximal cardiopulmonary exercise testing and in‐office plus 24 h ambulatory blood pressure (BP) monitoring. Blood samples were collected to assess cardiometabolic, erythropoietic, and haemolysis markers. ET consisted of pedalling exercise in a cycloergometer at 60% of V˙O2peak for 1 h/day, 4 days/week for 8 weeks, and participants were assessed at weeks 4 and 8. Haematocrit and CMS score decreased significantly by week 8 (to 65.6 ± 6.6%, and 3.5 ± 0.8, respectively, P < 0.05), while V˙O2peak and maximum workload increased with ET (33.8 ± 2.4 vs. 37.2 ± 2.0 ml/min/kg, P < 0.05; and 172.5 ± 9.4 vs. 210.0 ± 27.8 W, P < 0.01; respectively). Except for an increase in high‐density lipoprotein cholesterol, other blood markers and BP showed no differences. Our results suggest that reduction of haematocrit and CMS symptoms results mainly from haemodilution due to plasma volume expansion rather than to haemolysis. In conclusion, we show that ET can effectively reduce haematocrit, ameliorate symptoms and improve aerobic capacity in CMS patients, suggesting that regular aerobic exercise might be used as a low‐cost non‐invasive and non‐pharmacological management strategy.


INTRODUCTION
The excessive production of red blood cells (excessive erythrocytosis; EE) is the hallmark feature of chronic mountain sickness (CMS) or Monge's disease, a highly prevalent and incapacitating syndrome in Andean and other high-altitude populations around the world . CMS is defined by the presence of EE associated with severe hypoxaemia, neurological sequelae and sleep disorders, also often accompanied by other complications such as pulmonary hypertension and cardio-cerebrovascular accidents due to adverse changes in blood rheology (Penaloza & Arias-Stella, 2007;Villafuerte & Corante, 2016). In addition, we and others have shown that EE associates with increased cardiovascular disease risk factors and cardiometabolic disorders such as in-office and ambulatory hypertension, insulin resistance, dyslipidaemia and metabolic syndrome (Bilo et al., 2020;Corante et al., 2018;De Ferrari et al., 2014;Gonzales & Tapia, 2013;Miele et al., 2016) in various high-altitude regions across the world (Gonzales & Tapia, 2013;Okumiya et al., 2010Okumiya et al., , 2011Sherpa et al., 2011). The long-term burden of CMS equates to a loss of 3 months of healthy life per year at high altitude (4000 m) (Pei et al., 2012). It is estimated that 5-10% of the world's population living at high-altitude may develop this condition, and its prevalence increases with altitude and age (Leon-Velarde et al., 2014; Monge-C et al., 1989). Above 4300 m in the central Andes of Peru, more than 30% of highlanders by their mid-50s suffer from EE (Leon-Velarde et al., 1997; Monge-C et al., 1989Monge-C et al., , 1992. CMS signs and symptoms disappear completely when patients descend to sea-level conditions and following bloodletting or haemodilution at their native altitude of residence, suggesting that the underlying symptoms are secondary to EE. Therefore, treatment strategies for CMS aim to reduce haematocrit and blood viscosity, either by reducing the number of red blood cells or reducing the hypoxic stimulus. However, there is little evidence supporting the safety and efficacy of any long-term pharmacological  Reis et al., 1997). Moreover, people that practice exercise regularly have lower Hb concentrations than sedentary people (Mairbaurl, 2013;Telford et al., 2003). This difference is even more pronounced in athletes (Bonilla et al., 2005;Lippi & Sanchis-Gomar, 2019;Schobersberger et al., 1990;Telford et al., 2003).
Studies in Andean dwellers suggest that physical exercise may reduce haematocrit at high altitude. Schmidt et al. (1990) and Cornolo et al. (2005) have shown that native high-altitude athletes living and training above 4000 m have sea-level Hb concentrations in contrast to sedentary individuals living at the same altitude. Moreover, a recent

New Findings
• What is the central question of this study?
What is the effect of sub-maximal aerobic exercise training on signs and symptoms of chronic mountain sickness (CMS) in Andean highlanders?
• What is the main finding and its importance?
Aerobic exercise training (ET) effectively reduces haematocrit, ameliorates symptoms and improves aerobic capacity in CMS patients, suggesting that a regular aerobic ET programme might be used as a low-cost non-invasive/non-pharmacological management strategy of this syndrome. study by our group has shown that exercise training (ET) reduces haematocrit in a rat model of chronic hypoxia-induced erythrocytosis mainly due to exercise-induced haemolysis without changes in plasma volume (Macarlupu et al., 2021).
Although exercise fatigue and reduced exercise capacity have been commonly reported in CMS patients, allegedly due to systemic or pulmonary haemodynamic burden and O 2 -diffusion impairment as a consequence of excessive haematocrit and blood viscosity (Letcher et al., 1981;Monge, 1943;Ostergaard, 2020;Pratali et al., 2012;Soria et al., 2019;Stuber et al., 2010;Winslow & Monge-C, 1987), several studies indirectly suggest that adaptive mechanisms exist to maintain O 2 transport in the face of a high haematocrit (Juvonen et al., 1991;Lindenfeld et al., 1985). We have previously shown that

Ethical approval
The study was approved by the Institutional Ethics Committee of Universidad Peruana Cayetano Heredia (CIEH-UPCH approval no. 606-25-18, SIDISI no. 100520), and was conducted in accordance with the principles of the Declaration of Helsinki (except for registration in a database). All participants received a detailed explanation of the experimental protocol before consent, and were provided with a consent form in Spanish to be signed before participation in the study.

Study participants
Ten untrained men with CMS were recruited for the study, and two of them withdrew before the first session of the ET protocol. Thus, we were able to follow eight CMS patients throughout the study (Table 1).

Preliminary screening, haematocrit and Qinghai CMS score
Clinical examination was performed during a preliminary screening session and general health information was collected. During this session, an ECG (Quark C12x, Cosmed, Albano Laziale, Italy) and spirometry (Pony FX, Cosmed, Albano Laziale, Italy) were performed, and pulse O 2 saturation (S pO 2 ) and heart rate (HR) were measured using a Nellcor N-560 oximeter (Nellcor Puritan Bennet Inc., Pleasanton, CA, USA), and systolic and diastolic blood pressure (SBP and DBP, respectively) using a validated oscilometric device (UA-767Plus, A&D, Tokyo, Japan; Verdecchia et al., 2004

2.4
Blood samples

Ambulatory blood pressure monitoring
Ambulatory blood pressure monitoring (ABPM) is currently recognised by international guidelines as a key instrument for out-of-office blood pressure (BP) measurement and in the diagnosis and management of hypertension (Whelton et al., 2018;Williams et al., 2018 Participants were asked to stay still during the recordings and keep a standardised activity journal. Valid ABPM recordings were those with at least 70% of expected readings available and which did not contain two or more consecutive hours without valid readings. Variables obtained from the recordings were systolic, diastolic and mean daytime (awake), night-time (sleep) and 24-h blood pressure. production (V CO 2 ), minute ventilation (V E ), and end-tidal P O 2 and P CO 2 (P ETO 2 and P ETCO 2 , respectively). Twelve-lead ECG and HR were recorded continuously (Quark C12x, Cosmed, Albano Laziale, Italy).
SBP and DBP during exercise was measured non-invasively using sphyngomanometry at rest and then during the final minute of each workload increment. Participants performed a maximal preliminary CPET with a step-incremental protocol at the beginning of the study to determine their aerobic capacity or peakV O 2 (V O 2 peak ). In a later session, a baseline five-step CPET at 25%, 50%, 75% and 100% of each participant's pre-determinedV O 2 peak was performed. All measurements were repeated after 4 and 8 weeks of ET.

Sub-maximal aerobic exercise training
The training scheme consisted of sessions every other day with 2 days in a row weekly (4 days/week) to complete 32 sessions in 8 weeks. The protocol allowed one session/week as the maximum number of absences. Sub-maximal aerobic ET consisted of pedalling exercise in a cycle-ergometer at 60% ofV O 2 peak for 1 h/day. After 4 weeks, a maximal CPET was performed to readjust 60% ofV O 2 peak for the following 4 weeks. Training sessions were supervised by research personnel to ensure a proper workload (load and cadence) and duration. Each participant had a resting period of ∼48 h before performing CPET, and the day of the test counted as a training session.
S pO 2 , HR and in-office SBP and DBP were measured in duplicate after 5 min of rest once the participant was seated on the cycle ergometer before each session and immediately after. No modification of diet or additional exercise activity was advised.

Sample size and statistical analysis
Assuming comparable differences and corresponding large effect sizes previously observed in haematocrit (η 2 = 0.72) and CMS score (η 2 > 1) in studies of CMS management with acetazolamide (Richalet et al., 2008), our primary end-outcome variables after 8 weeks of ET in the present study required a sample size of eight participants in order to achieve a power of 0.80 at P < 0.05.
After testing for normality of data, repeated-measures one-way

Cardiopulmonary exercise test
Ambient conditions during CPET sessions were stable with barometric pressure of 456 ± 0.3 mmHg, mean room temperature of 20 ± 0.3 • C and relative humidity of 61 ± 0.7%. Table 2 shows CPET measurements together with S pO 2 and BP values at rest before the incremental exercise protocol and during peak exercise at baseline and after 4 and 8 weeks.
After 4 weeks, a maximal exercise test was repeated to readjust 60% ofV O 2 peak for the following 4 weeks.V O 2 peak increased after 8 weeks  All values are presented as means ± SD. Comparisons were made using paired t-test or Wilcoxon matched-pairs signed rank test. Abbreviations: DBP, diastolic blood pressure; HR, heart rate; MAP, mean arterial blood pressure; SBP, systolic blood pressure.
at peak exercise at week 4 (P < 0.05), but no statistically significant reduction after 8 weeks.

Ambulatory blood pressure monitoring
ABPM results are shown in Table 3. ABPM parameters showed no differences after 8 weeks of aerobic ET compared to baseline measurements.

Cardiometabolic risk, erythropoietic and haemolysis blood markers
Average glycaemia, insulinaemia or HOMA-IR index values, as well as EPO and iron profile markers, showed no differences after 8 weeks of ET. Lipid profile analyses showed no differences except for an increase in high-density lipoprotein cholesterol (HDL-C) at the end of the study (Table 4).
At baseline, free haptoglobin showed a negative correlation with haematocrit (r = 0.72, P < 0.05, Figure 3a) which lost statistical

DISCUSSION
We showed that sub-maximal aerobic ET for 8 weeks reduces haematocrit and ameliorates CMS signs and symptoms. Despite some evidence regarding diminished exercise capacity (Winslow & Monge-C, 1987;Winslow et al., 1985), or exercise being counterproductive for highlanders with CMS (Pratali et al., 2012;Soria et al., 2019;Stuber et al., 2010), we show that an exercise programme at 60% of maximal effort for 4 days a week during 8 weeks is a potential management approach for CMS.
Bloodletting and haemodilution have been the traditional management strategies for CMS at high altitude (Klein, 1983;Sedano et al., 1988;Winslow & Monge-C, 1987;Winslow et al., 1985). Interestingly, signs and symptoms resume within hours when haematocrit is reduced by these methods despite environmental hypoxia, showing that CMS symptomatology is secondary to EE. Pharmacological treatment strategies have been used to reduce haematocrit either by reducing the erythropoietic stimulus or erythropoiesis itself. These interventions included angiotensin converting enzyme inhibitors (Plata et al., 2002;Vargas et al., 1996), dopaminergic antagonists (Leon-Velarde et al., 2003) and ventilatory stimulants such as medroxyprogesterone (Kryger et al., 1978) and almitrine (Villena et al., 1985). However, only a few studies have shown evidence for safety and efficacy in the treatment of CMS. The most recent and longer-term randomised controlled trials with clinical significance used acetazolamide, a systemic carbonic anhydrase inhibitor, as a potential treatment of the syndrome. Two randomised, doubleblind, placebo-controlled studies assessed the safety and efficacy of acetazolamide treatment for up to 6 months in CMS patients in Cerro de Pasco, Peru (Richalet et al., 2008;Richalet et al., 2005).
Results showed that acetazolamide increased P aO 2 , decreased serum EPO and decreased haematocrit by 5%. Treatment also decreased pulmonary vascular resistance, increased nocturnal S pO 2 and reduced sleep-disordered breathing episodes. Overall, acetazolamide reduced hypoventilation, blunted erythropoiesis and improved pulmonary circulation without adverse effects throughout the duration of the trials. Although its implementation as a treatment appears efficient and safe, longer trials would be required to assess any development of tolerance or potential long-term consequences of the chronic inhibition of carbonic anhydrase in the different organs and systems.
For this reason, non-pharmacological approaches are an interesting avenue to explore, especially in resource-constrained populations.
Native highlanders who exercise regularly, or highly trained highlander athletes, have significantly lower haematocrit values compared to the sedentary population at the same altitude. We showed that 8-week submaximal aerobic ET reduced haematocrit by 7% and ameliorated CMS signs and symptoms significantly, reducing CMS score down to values usually observed in the healthy highlander population. Our results paralleled those obtained with acetazolamide treatment, but without the improvement on S pO 2 . In addition, we did not detect any significant change in P ETO 2 or P ETCO 2 . The reduction of haematocrit might be explained by the expansion of plasma volume (PV) and the consequent haemodilution, as this has been well documented in both cross-sectional and longitudinal endurance ET studies (Convertino, 2007;Schmidt & Prommer, 2008). PV expansion can account for nearly all of the ET-induced hypervolaemia up to 2-4 weeks; after this time expansion may be distributed equally between plasma and red cell volumes (Convertino, 2007). Hypervolaemia may provide larger vascular volume and filling pressure for greater cardiac stroke volume, and thus cardiac output (Q) during exercise.
This, together with greater ET-induced O 2 extraction and increased muscle mitochondrial oxidative capacity (Skattebo et al., 2020) might also contribute to the increased aerobic capacity after 8 weeks of ET despite decreased haematocrit and unchanged S pO 2 . In addition, we have recently shown that aerobic capacity is supported by adrenergic and non-adrenergic vasoconstriction of non-active skeletal muscle in individuals with CMS, which likely aids in central redistribution of blood volume during exercise (Hansen et al., 2021). Also, heightened α-adrenergic signalling restrains vasodilatation within active skeletal muscle to better match O 2 delivery (Hansen et al., 2021).
Interestingly, lower haematocrit (lower Hb concentration) might also contribute to increased O 2 diffusion and O 2 extraction between muscle microcirculatory vessels and mitochondria (Wagner, 1996).
When Hb concentration is reduced, time to diffusive equilibration in these vascular beds is shortened. Piiper & Scheid (1981)  be magnified by ET-induced mechanical haemolysis due to excessive haematocrit levels (Macarlupu et al., 2021;Telford et al., 2003). In fact, we have recently shown that haemolysis might be an additional mechanism for ET-induced haematocrit reduction in a rat model of high-altitude erythrocytosis (Macarlupu et al., 2021). Haemolysis due to mechanical stress might lead to a reduction in haematocrit without significant modification of PV or total blood volume (Schobersberger et al., 1990;Selby & Eichner, 1986;Telford et al., 2003). This mechanism is particularly expected to be relevant in impact sports such as running.
However, there is also evidence of haemolysis induced by exercise in disciplines where impact is reduced or absent (Lippi & Sanchis-Gomar, 2019;Schobersberger et al., 1990;Selby & Eichner, 1986). Haemolysis would occur to a greater degree when blood shows a higher viscosity, and therefore it would be favoured when haematocrit is excessive. As haptoglobin binds and combines with free plasma Hb, our finding of an inverse relationship of free haptoglobin and haematocrit at baseline suggests a steady-state background haemolysis due to EE. However, this relationship is lost after 4 and 8 weeks of ET once haematocrit decreased, and total plasma free haptoglobin showed a trend to rise suggesting a marginal increased haemolysis consequent to ET.
One other important effect of ET is the reduction of BP. Studies at sea-level have shown that aerobic ET for 8 weeks (30-min session; 3 times/week at 80-90% of ventilatory threshold) decreases 24-h SBP and DBP, with the largest reduction mainly at night-time (Carvalho et al., 2009). However, our results show no reduction in office-based BP or ABPM at rest after 8 weeks of aerobic ET. This lack of change might be explained by a possible concurrent effect of the autonomic characteristics of CMS highlanders, the ET-induced expansion of PV and the ET-induced drop in peripheral vascular resistance (Fagard, 2006 (Hansen et al., 2021). This effect is probably augmented by ET as we observed a significant drop in SBP and DBP at peak exercise after 4 weeks. Nevertheless, the expression of this phenotype is possibly modest and might be lost over time (Bilo et al., 2020;Corante et al., 2018).
A compensatory or offsetting effect on HR might be also taking place during ET. Our results show that although not statistically significant, HR at rest shows a trend of continuous reduction after 8 weeks of ET compared to baseline values. Normally, ET induces a reduction in HR and the effect occurs after only a few months, with about three training sessions per week (Genovesi et al., 2007;Whyte et al., 2008;Zavorsky, 2000). Meta-analyses on the effect of ET over HR indicate that endurance training decreases the resting HR between 4 and 6 bmp (Huang et al., 2005;Reimers et al., 2018). Similarly, HR peak is reduced following regular aerobic exercise in sedentary adults and endurance athletes. The overall effect of aerobic training on HR peak is moderate as it can be altered by 3-7% (Zavorsky, 2000). It is possible that severe hypoxaemia and basal enhanced sympathetic activity in CMS highlanders prevent the full reduction of resting HR in CMS and may lead to augment HR at peak exercise.
Lastly, aerobic ET is also associated to improved cardiometabolic risk profile (Lin et al., 2015). In previous studies we have identified independent associations between EE and 24 h ambulatory hypertension including systolic-diastolic and isolated diastolic hypertension (Bilo et al., 2020;Corante et al., 2018), and a significant proportion of masked hypertension, the latter linked to increased cardiovascular morbidity and mortality in lowlanders (Pierdomenico & Cuccurullo, 2011). We have also identified independent associations between EE, insulin resistance, hyperglycaemia, and dyslipidaemia. These findings agree with other studies at high altitude in the Peruvian Andes that have identified independent relationships between EE and hypertension, hypertriglyceridaemia (Gonzales & Tapia, 2013;Jefferson et al., 2002;Leon-Velarde & Arregui, 1994) and metabolic syndrome (De Ferrari et al., 2014). Several RCTs have shown reductions in blood triglycerides, glucose and insulin, and increases in HDL-C as the most common biomarker changes following ET programmes (Lin et al., 2015).
Circulating HDL-C levels are usually significantly increased with ET in a manner that is dependent on the duration and intensity of the training plan (Durstine et al., 2001;Kodama et al., 2007). Our results showed a significant increase in HDL-C, a trend to triglyceride reduction and no changes in glycaemic control markers. Studies with training plans from 4 to 104 weeks with mild or moderate intensity reported changes in some cardiometabolic marker levels (Lin et al., 2015), such as an increase in HDL-C even as early as 4 weeks of ET (Banz et al., 2003;Cox et al., 1993;Haskell, 1984;Kraus et al., 2002;LeMura et al., 2000). On the other hand, other blood lipid, as well as glycaemic control markers, frequently tend to improve over a period ≥12 weeks of ET (Cho et al., 2011;Libardi et al., 2012;Watkins et al., 2003) suggesting that possibly the length of our study was not enough to observe such changes.

Limitations
One main limitation of our study is that although it is difficult to maintain a group of participants training for 8 weeks, the protocol might have been controlled with the inclusion of CMS highlanders without ET and/or a group of non-CMS highlanders subjected to a similar ET programme. Either of the two comparisons would have been interesting and informative. However, we believe that this does not preclude the conclusions of our study as each participant was their own baseline control and the main aim of the study was to investigate whether ET can reduce haematocrit and alleviate symptoms in CMS highlanders.
The inclusion of male highlanders only is an additional limitation of our study. CMS has a very low prevalence in pre-menopausal women (Azad et al., 2021;Leon-Velarde et al., 1997;Leon-Velarde et al., 2001), and therefore forming comparable groups covering the same age range to avoid any confounding effects of age, and of menopause itself, would have been difficult. For these and cultural reasons, women were not included in the study and therefore our findings cannot be extrapolated to female CMS highlanders.

Conclusions
In conclusion, we show that sub-maximal aerobic ET can effectively reduce haematocrit and ameliorate symptoms in CMS patients. In addition, despite the common belief of reduced tolerance to exercise in CMS highlanders, we show that ET can successfully improve aerobic capacity and exercise workload in these patients maintaining BP homeostasis and improving cardiovascular disease risk markers such as HDL-C. Overall, our results suggest that a regular aerobic ET might be used as a low-cost non-invasive and non-pharmacological practical management strategy for CMS.