Effects of nasal high flow on sympathovagal balance, sleep, and sleep-related breathing in patients with precapillary pulmonary hypertension

Patients with Nice class I pulmonary arterial hypertension or class IV chronic thromboembolic PH (collectively referred to as precapillary PH) were consecutively recruited from May 2018 to April 2019. The study protocol conforms to the 1975 Declaration of Helsinki and was approved by the institution’s human research committee (Ethikkommission der Ärztekammer Westfalen Lippe und der Medizinischen Fakultät der Westfälischen Wilhelms-Universität Münster). All participants gave written informed consent to participate in the study. The project has been registered and updated prospectively under the German Clinical Trials Registry (drks.de Identifier: DRKS00013907 and DRKS00013907).

Both males and females participated in the study. Participants had to be at least 18 years of age and able to consent. Diagnosis of precapillary PH was established according to the most recent Guidelines of the European Society of Cardiology (ESC) [24]. Inclusion criteria comprised diagnosis of precapillary PH at least 12 weeks before recruitment and no hospitalization for heart failure (HF) within 4 weeks prior to enrolment. Patients had to have had optimal medical therapy in accordance with the most recent ESC guidelines [24] with combination therapy and no change in medication in the last 4 weeks. If patients were not treated with combination therapy, the reason was documented. PH patients had to be diagnosed with sleep-disordered breathing (SDB) with predominant OSA (apnea-hypopnea index ≥ 10/h; ≥ 50% of apneic events being obstructive).

Exclusion criteria were as follows: secondary (postcapillary) pulmonary hypertension (mean pulmonary arterial hypertension > 25 mm Hg; PCWP > 15 mm Hg); chronic obstructive pulmonary disease; insulin-dependent diabetes; severe renal impairment; severe neurological preexisting conditions including prior stroke; intake of opioids; severe mental disease. Detailed inclusion and exclusion patient criteria are available in the German Clinical Trials Registry ID DRKS00013906 and DRKS00013907.

All study participants underwent standard 2-dimensional Doppler echocardiography (LOGIQ S8-XD clear^TM, GE Healthcare, London, United Kingdom) [25, 26]. Patients with PH also underwent a 1.5 Tesla MRI scan (Ingenia^TM, Philips Healthcare, Best, The Netherlands) to quantify right ventricular systolic function [27], which was classified as normal or impaired based on age- and gender-specific reference values [27]. Patients also underwent noninvasive hemodynamic and respiratory monitoring in the afternoon in the sleep laboratory, and then spent a diagnostic night (night 1) and a study night (night 2) during attended polysomnography (PSG), as explained below (Fig. 1).

To ascertain SDB with predominant OSA, patients underwent full diagnostic PSG (Somno HD^TM, Somnomedics, Randersacker, Germany) in the academic sleep laboratory at Münster University Hospital. Respiratory recordings included nasal airflow (Ternimed^TM, Bielefeld, Germany), thoracic and abdominal effort, and peripheral oxygen saturation (SpO₂). Respiratory events were scored according to the guidelines of the American Academy of Sleep Medicine (AASM) [28]. Apnea was defined as a reduction in nasal airflow by > 90% of baseline for > 90% of the event’s duration and > 10 s. Apneas were classified as obstructive, if there was a continued or increased inspiratory effort throughout the entire period of absent airflow. If this was not the case, apneas were classified as central. Hypopnea was defined as a > 30% fall from baseline in airflow signal for 90% of the event’s duration of at least 10 s and 3% desaturation from the preevent baseline. The oxygen desaturation index (ODI) was based on decreases in peripheral oxygen saturation of 3% at least.

Sleep stages were recorded and analyzed using a twelve-lead electroencephalography (EEG) in accordance with the most recent AASM guidelines [28‐35].

During the next night, full PSG was combined with transcutaneous carbon dioxide (CO₂) monitoring (SenTec, Therwil, Switzerland). The night was split into four parts of 2-h duration each. The first part was recorded without treatment of SDB (with some patients requiring oxygen), then PAP, NHF at 20 L/min (NHF20), and NHF 50 L/min (NHF50) were given for 2 h each in a randomized order. NHF (PrismaVent 50^TM, Löwenstein Medical, Bad Ems, Germany) was administered using a nasal cannula (Optiflow^TM, Fisher & Paykel, Schorndorf, Germany), and flow rates were chosen based on previous studies showing differential effects on ventilation and oxygenation [17, 20, 21]. Automatic PAP (5–12 cm H₂O) was delivered through a nasal mask. Noninvasive hemodynamic monitoring (detailed below) was performed throughout the second night. All measurements were fully attended (B. B. and J. S.), and sleep stages, respiration, and SVB were evaluated online and on a “beat-to-beat” basis. Four 10-min sleep segments for each patient (one each from the NT, PAP, NHF20, and NHF50 treatment periods) were chosen by visual identification. Each segment had to be characterized by stable N2 (nonrapid eye movement stage 2) sleep and sinus rhythm with fewer than 5% ectopic beats.

The acute daytime effects of NHF therapy on SVB and hemodynamics were assessed in the late afternoon with the patient awake. Initially, subjects were asked to breathe spontaneously for 10 min to obtain baseline parameters. Thereafter, patients were exposed to 10 min each of NHF20 and NHF50 in a random order, separated by periods of spontaneous breathing until heart rate, blood pressure, transcutaneous carbon dioxide, and oxygen saturation had returned to baseline. Patients were asked to rate their subjective well-being at baseline and after each NHF intervention using a visual analogue scale ranging from 0 to 10 (0 = worst overall wellbeing imaginable; 10 = best wellbeing imaginable).

SVB and hemodynamics were measured using a noninvasive monitor device (Task Force Monitor^TM, CNSystems, Graz, Austria), as previously validated [29‐31]. For assessment of SVB, diastolic blood pressure variability (BPV) and heart rate variability (HRV) were analyzed using the continuous noninvasive arterial blood pressure signal (CNAP^TM technology, sampling rate 100 Hz) and a 3-lead electrocardiogram (sampling rate 1000 Hz), both implemented in the Task Force Monitor^TM. Diastolic BPV (measured in mm Hg²) and HRV (i.e., variability in RR intervals measured in ms²) were both continuously recorded and then normalized for total power spectra with the resulting unit being %. Data were computed by frequency domain analysis (adaptive autoregressive parameter model) and presented as the high frequency component (HF: 0.15–0.40 Hz), low frequency component (LF: 0.04–0.15 Hz), and relative ratio (LF/HF) for both HRV and diastolic BPV [29‐32]. For both measurements, a higher ratio indicates increased sympathetic drive because the LF component is believed to mainly reflect sympathetic drive, and the HF component is believed to exclusively reflect parasympathetic drive [29‐31, 33].

Baroreceptor reflex sensitivity (BRS) was measured using the sequence method [29‐32, 34]. The time-constant of this stimulus-response relationship primarily reflects, particularly with respect to the upsequences, vagal but not sympathetic responsiveness [29‐32, 34].

Hemodynamic measurements included beat-to-beat systolic (sBP) and diastolic (dBP) blood pressure; these were recorded and validated against periodic measurements obtained every 15 min by oscillometric recording from the upper contralateral arm. Transthoracic impedance measurements were used to estimate cardiac stroke volume index (SVI), cardiac index (CI), and systemic vascular resistance (SVR), from which the total peripheral resistance index (TPRI) was calculated (mean blood pressure divided by CI). Bioimpedance-based measurements have been previously validated against invasive hemodynamic monitoring in the catheter laboratory [29‐31].

All signals were simultaneously acquired and displayed in real time using personal computer running DOMINO 2.9.0 software.

All analyses were performed using Sigma Plot^TM software (Version 13.0, Systat Software GmbH, Erkrath, Germany). Assuming a two-sided significance level of 0.05 (alpha) and 80% power (beta), a sample size of twelve patients per group was calculated to allow detection of a 20% change in the LF/HF ratio of HRV [6, 7]. Values for the mean and standard deviation of the HRV LF/HF ratio during 10 min of N2 sleep were obtained from previous preliminary data. Furthermore, it was also known from our previous measurements that intraindividual variation in the LF/HF ratio component of HRV during N2 sleep is ~ 5% at most.

Results were expressed as mean and standard deviation for continuous variables with a normal distribution, and median and interquartile range for continuous variables with a skewed distribution. Categorical variables were expressed as percentages unless otherwise specified. Respiratory parameters, SVB, and hemodynamic measures recorded during the different interventions were compared with baseline values using a paired t test or Wilcoxon rank sum test, as appropriate. For all tests a p value ≤ 0.05 was considered statistically significant.

Compared with no treatment, hemodynamic parameters remained unchanged during NHF20. There was a shift towards predominance of parasympathetic drive over sympathetic drive during N2 sleep, shown by an increase (~ 15%) in the HRV HF component and a decrease (by ~ 25%) in the HRV LF component, and a decrease (by ~ 40%) in the HRV LF/HF ratio with delivery of NHF20 during N2 sleep (Table 2 and Fig. 2). There was also a trend towards better sleep efficiency during use of NHF20 (from ~ 58 to ~ 65% (p = 0.076) (Table 3 and Fig. 3). NHF20 had no significant effects on the severity of SDB (Table 3 and Fig. 4).

NHF50 was poorly tolerated at night, with 6/12 patients refusing therapy and the other six having lower sleep efficiency (Table 3 and Figs. 3 and 4). Compared with no treatment, NHF50 had no significant effects on hemodynamics, SVB (Table 2 and Fig. 2) or OSA severity (Table 3 and Fig. 4). Similar to NHF20, NHF50 had no significant effects on markers of nocturnal SDB severity (Table 3 and Fig. 4).

APAP had no effect on hemodynamic parameters compared with no treatment. The effects of APAP on SVB were heterogenous, with a decrease in the LF/HF ratio of HRV and a trend towards an increase in the LF/HF ratio of dBPV (Table 2 and Fig. 2). APAP had neutral effects on sleep parameters, but significantly improved markers of SDB severity (Table 3 and Figs. 3 and 4).

Although NHF20 and NHF50 led to a significant and flow-dependent decrease in subjective well-being (Supplemental Table 1), application of NHF20 and NHF50 during daytime was associated with significant flow-dependent increases in SpO₂ and to a nonsignificant decrease in tcCO₂ when compared with no treatment (Supplemental Table 1, Supplemental Figure 1). Furthermore, NHF20 and NHF50 were associated with flow-dependent shift of SVB towards increased sympathetic drive (change in median LF/HF ratio from 0.9 to 1.5 with NHF20 and 1.6 NHF50, respectively; all p < 0.05) (Supplemental Table 1, Supplemental Figure 2). NHF20 and NHF50 were also associated with flow-dependent increased measures of cardiac afterload, as documented by the observed increase in diastolic blood pressure and by the slight increase of the total peripheral resistance index (Supplemental Table 1, Supplemental Figure 2).

Overall, NHF20 decreased indices of sympathetic drive and increased indices of parasympathetic drive, with a trend towards increased sleep efficiency. However, the level of PAP generated during NHF20 was not sufficient to reduce obstructive apneic or hypopneic events in adults with PH. This is consistent with previous studies showing that every 10 L increase in NHF flow rate generates about 1 cm H₂O of positive airway pressure [36]. Current guidelines recommend treating adults with OSA with 5–12 cm H₂O of PAP, and NHF20 would likely only produce about 2 cm H₂O of PAP [11]. However, NHF can have significant effects on obstructive events on children, who require lower PAP levels [17].

Based on the low levels of PAP delivered, the reduced sympathetic drive observed in sleeping PH patients treated with NHF20 can only be explained by the effects on dead space ventilation [20]. It has previously been shown that wasted ventilation (i.e., dead space ventilation) contributes to chronic hyperventilation, dyspnea, and air hunger in patients with PH, causing physical exhaustion and sympathetic overactivity [21, 37]. NHF has been shown to reduce dead space ventilation and make breathing more efficient, as previously demonstrated in ten healthy volunteers using scintigraphy and directly measuring oxygen (O₂) and CO₂ in the upper airways [22].

We found that NHF50 was poorly tolerated by ambulatory patients with PH. This appears to contradict previous studies showing marked improvement of respiratory parameters and subjective well-being during treatment with NHF50 in patients with chronic obstructive pulmonary disease (COPD), heart failure, or acute hypercapnic respiratory failure [38‐42]. In contrast, the patients with stable PH in our study complained about the noise and uncomfortable nasal sensations during use of NHF50. This discomfort was so severe that half of all patients requested to discontinue treatment after several minutes. The fact that sleep efficiency was reduced in the six patients who tolerated use of NHF50 for 2 h reinforces the observation that this flow rate is poorly tolerated at night. However, at a flow rate of 50 L/min, NHF has the potential to generate up to 5 cm H₂O of PAP, which is at the lower end of the range of pressured recommended for PAP therapy of OSA [11]. Unfortunately, the decreased power resulting from the withdrawal of half the patients during use of NHF50 means that definitive conclusions about PAP delivery and treatment of OSA cannot be made.

As expected, use of PAP improved SDB in patients with PH but did not lead to improved sleep in our study. The effects of PAP therapy on SVB are heterogenous. A previous study in patients with left-sided heart failure showed that PAP may have hypotensive effects [15], leading to unwanted increases in sympathetic drive that may counteract the positive effects associated with reversal of upper airway obstructions [15]. This complex SVB scenario is covered by our data. On the one hand, there is a decrease in the LF/HF ratio based upon HRV analysis (which may reflect the overall picture of SVB best, showing an overall desired decrease in SVB derived from upper airway stability through PAP). On the other hand, there is a trend towards an increase in the LF/HF ratio of dBPV analysis (which may reflect the negative effect of PAP on the right heart due to decreased venous return) [29‐31, 33]. It has been speculated that PAP therapy has hypotensive effects in patients with right ventricular dysfunction (with increased sympathetic drive as a response) [15]. Olsson and colleagues showed this in a study on patients with PH [43]. Although the present study did not reveal adverse effects of PAP on blood pressure, the effects of PAP therapy on blood pressure may have been there, which was slightly higher during PAP compared with no treatment. This may apply to automatically titrating PAP therapy (as used in the present study) in particular (compared with continuous PAP therapy, which delivers a more constant and, on average, probably lower level of pressure) [44]. As a result, the present study adds to the growing body of evidence that initiation of PAP therapy in patients with PH and obstructive sleep apnea should be a highly individualized decision.

Flow-dependent favorable effects of NHF on gas exchange have been previously described [22]. Flow-dependent improvements in oxygenation, reduced anatomical dead space, improved breathing efficiency and generation of PAP have been discussed as potential mechanisms [22]. We hypothesized that clearance of anatomical dead space by NHF would translate into improved wellbeing and a decrease in sympathetic drive in the awake state. In fact, we found the exact opposite, with SVB parameters showing a relative, flow-dependent increase in sympathetic activity. Two main mechanisms may help explain these results. Firstly, the effects of NHF on oxygen saturation may differ between daytime and nighttime. During the day, NHF20 increased mean SpO₂ from approximately 91 to 93%, while at nighttime, where hypoxia has previously been described to be more severe [8, 10], mean oxygen saturation was approximately 88% in the same patients. Faster and greater increases in oxygen saturation have been shown to be associated with increased levels of oxygen radicals, which in turn are associated with poor functional status in patients with heart failure [43, 45‐47]. It appears that differences in the effects of NHF20 on oxygen saturation also account for the unwanted increase in sympathetic drive during daytime and a beneficial reduction in sympathetic drive at night in the present study. Additionally, patient frustrations and lack of tolerance with NHF20 and NHF50 usage during the day may trigger cortical mechanisms of sympathetic activation, overcoming the beneficial effects on dead space. Conversely, this was only true for NHF50 usage at night, whereas NHF20 was better tolerated and beneficial effects on dead space did occur.

Despite the comprehensive approach taken, our study has limitations that should be considered. Firstly, the sample size was small, and although the study was adequately powered to investigate the effects of NHF and PAP on SVB, it was probably underpowered for assessment of effects on sleep-related breathing. This is particularly true for the effects of NHF50 at night because half of all patients declined treatment. Also in that sense and taking into account the statistical assumption behind the calculation of the Bonferroni post hoc corrections for multiple t tests actual p values reported in Tables 2 and 3 and Supplemental Table 1 should be interpreted with caution, but statistically significant p values on HRV obtained and reported for HRV were still significant even taking into account the Bonferroni post hoc correction

Secondly, the duration of interventions was limited to 2 h, and it is possible that longer periods of NHF may have had different effects on study outcomes. Future studies with longer use of NHF are needed to test the hypothesis that NHF20 favorably alters sympathetic drive and improves the functional status of patients with precapillary PH. Likewise, we do not know whether longer periods of PAP treatment of OSA in PH might decrease sympathetic drive.

Thirdly, sympathetic drive was not measured invasively (e.g., by direct recording of muscle sympathetic nerve activity [MSNA]). Noninvasive recording of HRV and BPV can only provide a general estimation of SVB, which includes the vagal component. However, close correlation between MSNA and HRV/BPV measurements (LF/HF ratios in particular) have previously been shown in healthy volunteers and patients with heart failure [33, 48, 49], and the use of this methodology under different breathing and ventilation-related experimental conditions has previously been established by our group [50].

Fourthly, PAP therapy was not manually titrated to optimal pressures, which may explain why the pressure range used (5–12 cm H₂O) did not completely normalize the AHI. Ultimately, the results obtained in the present study in patients with PH cannot be applied to different patient populations such as patients with COPD wherein NHF may correct hypoxia and dead space ventilation; the result of which may result in decreased sympathetic drive [51, 52].

Fifthly, another limitation of the study was that the 10 min N2 intervals chosen were not adjusted for respiratory rate. However, as the segments were only 2 h long, it was extremely difficult to find distinct 10 min recordings of constant stable N2 with no artifacts and comparable respiratory rates. Furthermore, in a nighttime setting, it would not have been possible to additionally adjust for respiratory rate as previously established by our group [53, 54].

Finally, scoring of PSG raw data was performed using an unblinded approach regarding the intervention being used (no treatment, NHF20, NHF50, APAP). Ideally, the scorers would have been blinded to this information.