Highlanders with HAPH (HAPH+) and HH, 21–75 y of age, both genders were recruited and studied at the Aksay health post (3250m, barometric pressure 515±1mmHg, room temperature 24.2±3.2°C), Kyrgyzstan. Participants had to be of Kyrgyz ethnicity, born, raised and currently living at >2500m. The diagnosis of HAPH was established by typical symptoms and a mPAP >30mmHg in the absence of excessive erythrocytosis (hemoglobin concentration in females <19g/dl, in males <21g/dl), and other present or diagnosed diseases that lead to hypoxemia (i.e., such as cardio-pulmonary diseases).12 Healthy lowlanders (LL), 21–75 y of age, both genders, of Kyrgyz ethnicity, born, raised and currently living in Bishkek (760m barometric pressure 692±2mmHg, room temperature 25.8±1.2°C), formed the lowlander control group. To eliminate the influence of smoking on the cardiovascular system, current heavy smokers (>10cigarettes per day), and subjects with a history of smoking >20pack-years were excluded from the study. Other exclusion criteria were diagnosed comorbidities or drug therapies that may have interfered with control of breathing, or pulmonary hypertension other than HAPH, in particular from left ventricular failure, lung disease or other conditions listed in the guidelines on pulmonary hypertension of the European Society of Cardiology.2 Data for the current study have been collected during the conduct of a previous study on HAPH, sleep apnea14 and cerebral oxygenation15 but the analysis of the ECG and pulse wave, the focus of the current study, have not been published. This study was conducted in accordance with the Declaration of Helsinki. Informed written consent was obtained, and the ethics committee of the National Center of Cardiology and Internal Medicine, Bishkek, Kyrgyzstan, approved the study (No. 01-7/219).

Highlanders traveled from their homes in the remote mountain areas by horse or car within one day to the study site at the Aksay health post. LL were studied in the National Center of Cardiology and Internal Medicine of Bishkek. The study design was according to a randomized single-blinded crossover protocol as detailed below. Participants rested in supine position breathing through a face-mask equipped with a reservoir bag and a one-way exhalation valve (non-rebreather oxygen mask, P.J. Dahlhausen & Co. GmbH, Cologne, Germany). Either ambient air (FiO2 0.21) delivered by a continuous positive airway pressure generator (REMstar, Philips Respironics, Zofingen, Switzerland), or oxygen (FiO2 1.0) at a flow rate of 10l/min from a pressurized bottle were breathed through the mask. After a baseline period of at least 10min, a 10min recording period took place with ambient air (or oxygen, according to randomization). After a 10min wash-out period with ambient air breathing (if FiO2 1.0 was used first), the procedure was repeated with the alternate gas. Patients were not allowed to sleep during the experiment. Blinding of participants was achieved by placing the breathing gas source out of their view.

A 4-lead electrocardiogram and the finger photoplethysmogram from pulse oximetry were continuously monitored with a sampling rate of 200Hz. Arterial oxygen saturation (SpO2) was assessed by finger pulse oximetry (Alice 5, Philips Respironics, Zofingen, Switzerland). QTc was calculated by the Bazett's formula.16 Stiffness index (SI) and wave D to wave A ratio (DA) of the second derivative of the finger plethysmogram, both predictors for CVD were assessed.17,18 SI was calculated as height of the participant divided by the time from the systolic to diastolic inflection point of the pulse waveform19 whereas DA was calculated as described by Takada et al.20 and Imanaga et al.21 A clinical examination, measurement of systemic blood pressure, echocardiography and arterial blood gas analyses were obtained as reported previously.14

Data are presented in median (quartiles). A per-protocol analysis was undertaken for the present study. Comparisons between groups were made with one-way ANOVA followed by post-hoc Scheffé multiple-comparisons or Kruskal–Wallis one-way analysis of variance followed by Mann–Whitney-U test as appropriate. Within group comparisons were made with dependent t-test or Wilcoxon signed ranks test as appropriate. ECG and pulse wave contour analysis was performed beat – by – beat over the last 2min of breathing FiO2 0.21 or 1.0, respectively. ECG/pulse wave analysis was performed with customized interactive software programmed in MatLab on mean waveforms ensemble-averaged by ECG triggering over 2min. The number of participants exceeding the critical threshold of QTc prolongation of >440ms was evaluated by the Pearson's Chi-squared test.5,7

In order to evaluate the effect of HAPH and residence at high altitude on QTc and SI independent of known confounders, such as age, gender, body mass index (BMI), mean arterial pressure and SpO2 under ambient air, univariable and multivariable regression analyses were employed. A significance level of p<0.05 was considered statistically significant.

Rimoldi et al. in 2012 and Bailey et al. in 2013, showed that highlanders with CMS had systemic vascular dysfunction, evidenced by impaired flow-mediated dilatation, increased vascular stiffness and increased carotid intima-media thickness, predisposing these patients to higher cardiovascular morbidity and mortality.10,11 A correlation between SpO2 and flow-mediated dilatation (r=0.62) and its improvement 1h after oxygen breathing in highlanders with CMS and in HH with moderate hypoxemia (SpO2<90%) led the authors to conclude that severity of hypoxemia was one of the underlying – and reversible – mechanisms responsible for the vascular dysfunction. However, they presented no results about mPAP and a possible correlation between mPAP and vascular dysfunction in CMS patients. In the current study we found high values of SI in HAPH+ comparable to those reported in patients with two or more cardiovascular risk factors,19 but SI were only weakly correlated with SpO2 (R2=0.045, p=0.001). The multivariable regression analysis revealed HAPH+, age and BMI as independent predictors of SI (Table 2). We found no impairments in the HH control group, which was in accordance with the cited study.11 It has been shown that chronic hypoxemia induces vasodilatation in the peripheral blood vessels,27 therefore, we assumed that baseline vessel diameter in the HAPH+ and HH would be larger compared to LL and would respond to breathing FiO2 1.0 by hyperoxia-induced vasoconstriction and elevation of SI in HAPH+ and HH compared to LL. Interestingly, breathing FiO2 1.0 increased SI in HH but not in HAPH+. This finding indicates that mildly hypoxemic HH have preserved vasomotor reactivity to breathing FiO2 1.0. Indeed, acute supplemental oxygen has shown to increase vascular stiffness and reactivity in healthy subjects and in diseases associated with mild hypoxemia, such as in COPD,28 indicating that breathing FiO2 1.0 modulates the vasomotor activity by enhancing parasympathetic activity and reducing sympathetic activity (indicated in our study by a hyperoxia-induced median reduction in heart rate of −6bpm (IQR, −8 to −5) and −7bpm (IQR, −9 to −5) in HH and HAPH+, respectively), or by reducing the bioavailability of nitric oxide due to increased oxidative stress.29 These changes towards higher vascular stiffness have been associated with enhanced spontaneous baroreflex sensitivity and restoration of vascular tone towards normal physiology in otherwise hypoxemic individuals.28 Although, heart rate decreased similarly under breathing FiO2 1.0 compared to room air in HH and HAPH+, the absence of hyperoxia-induced vascular stiffness as seen in HAPH+ might be due to only 20min hyperoxic gas breathing compared to 30min or 1h in the cited studies,11,29 or due to blunted vasoconstrictive mechanisms or even structural remodeling caused by more severe hypoxemia or pulmonary hypertension compared to HH. In LL, SI may possibly have remained unchanged due to the already well oxygenated blood under ambient air.

SI and wave D to wave A ratio of the second derivative of the finger plethysmogram, both predictors for CVD, correlated good to each other (R2=0.48, p<0.001).

In addition to the elevated mPAP, systemic blood pressure and heart rate were also elevated in HAPH+ compared to HH and LL. These findings are consistent with an elevated sympathetic tone.

QTc was prolonged in HAPH+ compared to HH and LL. Furthermore, QTc prolongation could be reduced with oxygen administration, but remained prolonged compared to the two control groups. Multivariable regression analysis revealed HAPH+, gender and age as independent factors influencing QTc. Our results are in accordance with past studies, showing an association between QTc and elevated pulmonary arterial pressure,30 gender31 and age.32 Right ventricular strain and the association with sleep apnea with cyclic nocturnal deoxygenation contribute to impaired repolarization of cardiac muscles and may therefore reflect an increased risk of ventricular tachycardia and fibrillation.

For logistical reasons, the current study in the remote mountain region could only include a relatively small sample of highlanders compared to previous low altitude studies.33 Instead of a 12-lead ECG, we used a 4-lead ECG and were only able to analyze lead II ECG wave contour. SI assessed by digital volume pulse as a predictor for arterial stiffness is a common noninvasive technique to assess arterial stiffness and has been successfully validated against flow mediated dilatation34 or tonometry-derived augmentation index.19,34 HAPH+ were significantly older compared to the two control groups, however, in multivariable regression models controlling for age, gender and other confounders, HAPH+ remained an independent predictor for SCD and CVD.