Adaptive Servo-Ventilation: A Comprehensive Descriptive Study in the Geneva Lake Area

Background: Use of adaptive servo-ventilation (ASV) has been questioned in patients with central sleep apnea (CSA) and chronic heart failure (CHF). This study aims to detail the present use of ASV in clinical practice. Methods: Descriptive, cross-sectional, multicentric study of patients undergoing long term (≥3 months) ASV in the Cantons of Geneva or Vaud (1,288,378 inhabitants) followed by public or private hospitals, private practitioners and/or home care providers. Results: Patients included (458) were mostly male (392; 85.6%), overweight [BMI (median, IQR): 29 kg/m2 (26; 33)], comorbid, with a median age of 71 years (59–77); 84% had been treated by CPAP before starting ASV. Indications for ASV were: emergent sleep apnea (ESA; 337; 73.6%), central sleep apnea (CSA; 108; 23.6%), obstructive sleep apnea (7; 1.5%), and overlap syndrome (6; 1.3%). Origin of CSA was cardiac (n = 30), neurological (n = 26), idiopathic (n = 28), or drug-related (n = 22). Among CSA cases, 60 (56%) patients had an echocardiography within the preceding 12 months; median left ventricular ejection fraction (LVEF) was 62.5% (54–65); 11 (18%) had a LVEF ≤45%. Average daily use of ASV was [mean (SD)] 368 (140) min; 13% used their device <3:30 h. Based on ventilator software, apnea-hypopnea index was normalized in 94% of subjects with data available (94% of 428). Conclusions: Use of ASV has evolved from its original indication (CSA in CHF) to a heterogeneous predominantly male, aged, comorbid, and overweight population with mainly ESA or CSA. CSA in CHF represented only 6.5% of this population. Compliance and correction of respiratory events were satisfactory. Clinical Trial Registration: www.ClinicalTrials.gov, identifier: NCT04054570.


INTRODUCTION
Adaptive servo-ventilation (ASV) was proposed for managing Cheyne-Stokes breathing (CSB) in the late 90's, 20 years after the first report of treatment of obstructive sleep apnea syndrome (OSAS) by continuous positive airway pressure (CPAP) (1,2). The initial algorithm was based on a fixed expiratory pressure (EPAP) and a variable pressure support, with the aim of controlling the "crescendo-decrescendo" of tidal volume and normalizing nocturnal breathing. Some devices targeted a minute ventilation set at 90% of spontaneous ventilation to allow nocturnal PaCO 2 to increase slightly above the apnea threshold (3). Since their original design, options such as auto-titration of EPAP, of pressure support, and of back-up respiratory rate (BURR) have been added to the initial algorithms (4)(5)(6)(7)(8).
Very little information exists as to the present use of ASV in clinical practice in an unselected population (5,(25)(26)(27). The following study aims to describe the use of ASV in clinical practice in all patients within an area with a longstanding experience in home non-invasive ventilation (NIV) and treatment of sleep-disordered breathing (SDB) (28). The aims of this study are: 1/to detail "real life" present indications for ASV, and the prevalence of its use; 2/to describe the population under ASV, and its major comorbidities; 3/to provide detailed data on settings, compliance, correction of respiratory events and other items reported by ventilator software, as well as modalities of medical follow-up.

Study Design
A multicenter cross-sectional observational study performed in the Cantons of Geneva and Vaud (1,288,378 inhabitants in 2017) was designed to include all subjects under NIV followed by every possible structure involved: university hospitals, regional general hospitals, pulmonary rehabilitation centers, and pulmonologists in private practice.
Our definition of home NIV included all patients using bi-level positive pressure devices, multi-mode devices, ASV, volumetric ventilators, who were treated at home or in a long-term care facility (not a hospital) for ≥3 months. A cross-checking of patients treated in our area through healthcare providers, and ventilator manufacturers guaranteed a comprehensive assessment of the targeted population.
Ethical approval was granted by the Cantonal Commission for Research Ethics (CCER) in Geneva, Switzerland (no. PB_2016-00925/15-275) in agreement with the amended Declaration of Helsinki. Trial was registered at clinicaltrials.gov (No: NCT04054570). Identification, screening, and data collection were performed by two investigators between June 1, 2016 and July 10, 2018.

Inclusion/Exclusion Criteria
The present study focuses exclusively on patients treated by ASV. Patients were excluded if they used any other device, and if they (or their pulmonologist) refused to participate in this study.

Data Collected
Anthropometric data, indication for ASV, major co-morbidities, pulmonary function tests, pulse-oximetry, arterial blood gases, ASV devices used, and information downloaded from devices were collected from medical records. No additional investigation was performed by the investigators. Availability of recent pulmonary function tests, pulse-oximetry, and arterial blood gases depended on "real-life" follow-up procedures and medical records. Results of echocardiographies were collected for patients with central sleep apnea (CSA) only. Data recorded were the most recent measurements performed within the 12 months prior to data collection.

Statistical Analysis
Analyses included mainly descriptive statistics. Qualitative parameters were described as frequencies and percentages and quantitative parameters were described as mean and standard deviation (SD) or as median and quartiles (IQR).
Univariate and multivariate analyses regarding modality of follow up (hospital-based vs. private practitioners), and average daily use are described in on-line supplement ( Table 1S).
Missing data were not replaced but simply reported, thus reflecting "real life" practices.
Statistical significance was assessed at the two-sided 0.05 level for all analyses. All analyses were performed using SPSS or R softwares (R Foundation for Statistical Computing, Vienna, Austria).

Prevalence and Clinical Characteristics
Two university hospitals, one regional general hospital, one pulmonary rehabilitation center, and 38 of the 43 pulmonologists in private practice in the Cantons of Geneva and Vaud participated in data collection, i.e., all possible structures and specialists involved (albeit for 5 pulmonologists).
Of a total of 1,014 patients treated by home NIV, 500 (49%) had ASV devices (prevalence: 39/10 5 inhabitants). By comparison, at the end of 2018, 19,350 patients were treated by CPAP in the same area (prevalence: 1,502/10 5 inhabitants; ratio CPAP/ASV: 39:1) (personal communication, provided by health care providers listed in acknowledgments). Patients under ASV thus represent 2.5% of the population treated by CPAP for OSAS in the same area.
Clinical characteristics of the 458 patients under ASV and their co-morbidities are detailed in Table 1 and
Patients reported as "CSA related to medication" were under opiates (methadone, morphine sulfate), administered either for chronic pain (n = 2), or for substitutive therapy of prior

AHI and Other Data Provided by Ventilator Software
Relevant data retrieved from ventilators are described by device in Table 4. Apnea-hypopnea index (AHI) under ASV (available in 428 patients) was normalized (i.e., < 10/h) in 94% (n = 401).
Frontiers in Medicine | www.frontiersin.org  AHI was slightly lower with nasal masks, than with facial or nasal pillows: this difference had no clinical relevance (Figure 3).

Follow-Up (Hospital-Based vs. Private Practitioners)
Data are provided in on-line supplement.

DISCUSSION
The present report is to our knowledge the largest report of unselected patients treated by long term ASV. Prevalence of ASV is presently very close to that of NIV for chronic hypercapnic respiratory failure in our area. The ASV population represents 2.5% of the population treated by CPAP for sleep-disordered breathing (SDB). The typical profile of the "ASV patient" in this study is an overweight or obese male subject in his early 70's, with emerging sleep apnea uncontrolled by CPAP, and several cardiac, respiratory and/or neurological comorbidities. Compliance is higher than previously reported, with a very good control of SDB, based on data downloaded from the ASV devices. Our data show that ASV devices are prescribed in clinical situations which extend far beyond the initial target of ASV, i.e., Cheyne-Stokes Breathing (CSB): indeed, 4 years after the publication of the SERVE-HF study (9), CSB in CHF represents only 1.5% of the population treated by ASV, and only 6.6% of all subjects on ASV (30/458) have CSA related to CHF. In spite of SERVE-HF (9), a third (36%) of cardiac patients under ASV still had a LVEF ≤45%. While all but 4 patients with CHF (26/30; 87%) had undergone an echocardiography within the prior 12 months, this was the case for only 56% (n = 60/108) of the whole ASV population treated for CSA. ASV patients are mostly followed by pulmonologists in private practice (67%). Outcomes (AHI, compliance) are similar irrespective of follow-up by a pulmonologist in private practice or hospital-based (On-Line Supplement).
The majority of patients were prescribed ASV for emergent or persistent CSA (74%) under CPAP. Using a large populationbased sample of CPAP device data, the dynamic nature of CSA occurring during CPAP therapy has been recently highlighted. A big data analysis (n = 133,006) using telemonitoring data of US CPAP devices identified a variety of CSA trajectories occurring during CPAP therapy (29). Previous studies in the field with relatively small sample sizes and heterogeneous populations in terms of titration procedures provided inconsistent results (30). The Liu et al. "real life" analysis allowed to delineate the true prevalence of CSA at 3.5% of CPAP treated OSA. CSA was transient, persistent, or emergent in 55.1, 25.2, and 19.7%, respectively (29). CSA under CPAP, whatever the subtype, was also clearly associated with a higher risk of therapy discontinuation (29) suggesting that these patients should be identified and specific phenotypes associated with emergent CSA more clearly described. These figures are in accordance with our results: close to 20,000 patients are currently treated by CPAP in our area: subjects under ASV represent 2.5% of those under CPAP, and most of them were previously under CPAP. In another big data analysis, patients demonstrated a better control of residual events and an improvement in adherence early after switching from CPAP to ASV (14). Compliance and AHI levels in this study are in agreement with these observations. The ASV population described differs from previous reports, which either focus on CHF (9, 31), or on specific neurological disorders (Chiari malformation (32), Multiple System Atrophy (33), opiate or other drug toxicities (16,21,22,34,35). Two retrospective studies (25,26) describe relatively large populations under ASV. Carnevale et al. (25) included 74 patients: 33 with CHF and 41 with various neurological disorders or idiopathic CSA. ASV was effective in reducing AHI, sleepiness, and improving SpO 2 in both groups. A German study analyzed 285 patients receiving ASV and undergoing diagnostic polysomnography (PSG). The most common indications were CHF and emergent CSA (67%), CSA in CHF (22%), idiopathic CSA (10%), and drug-induced CSA (0.4%) (26). The patients described in our study are long term users of ASV [median 40 months, IQR (22; 62)] which suggests at least a subjective benefit (although no formal assessment of health-related quality of life or symptom score were performed during this study). Interestingly, compliance rates are high (Figures 2, 3, Table 4,  Table 2S), and treatment of underlying SDB is effective based on data downloaded by the ventilators (IAH) and nocturnal pulse oximetry. Correlation between AHI from ventilator software and PSG, in the absence of major leaks, has been reported as appropriate for clinical use (36)(37)(38).
An interesting group of patients is that of drug-induced SDB. These patients are under opioids, either for treatment of chronic pain or for substitutive treatment of prior drug abuse, often in combination with benzodiazepines. This population is rapidly increasing, reaching epidemic proportions in the USA (39). OSA, CSA, and nocturnal hypoventilation have all been reported under long term opiate usage (21,22,35). ASV controls opiate-induced CSA in 58% of patients: it is more effective than CPAP, and equivalent to bi-level positive pressure ventilation (22). Patients under ASV for opiate-induced SDB in the present study were treated for central SDB. Prevalence of SDB is known to be high in these patients and is often symptomatic (40,41). Although a few clinical studies have shown that ASV could control SDB (CSA) in patients under chronic high dose opioid treatment (18,19,42), there are no randomized trials showing a long term improvement in either symptom scales, HRQL or survival. Many questions therefore remain, such as: should we screen patients under opioids for SDB? Should treatment be proposed systematically? What threshold levels of AHI are relevant? Before the use of ASV explodes in this population, with the associated costs, these issues warrant prospective trials.
For patients with CHF, SERVE-HF has radically changed the clinical approach to CSB in CHF. In this study, a significant percentage of patients with CHF, LVEF ≤45% and SDB remain under ASV (11/30: 37% of all cardiac cases). Almost all of these patients had an echocardiography within the preceding 12 months. This shows that despite SERVE-HF, a certain number of unwarranted prescriptions of ASV persist. The data collected do not allow us to determine to what extent ASV treatments were interrupted after SERVE-HF, or how many patients who are within the high-risk population identified by SERVE-HF, requested to remain under ASV because of improvement of their symptoms (43).
This study has several limitations: first of all, it is a descriptive study and must not be considered as a validation of ASV in the indications reported. Our aim was to better define practices regarding ASV and to identify subgroups of interest for whom prospective randomized studies should be conducted (i.e., CSA related to medication, opiates, CSA associated to neurological disorders, or "idiopathic" CSA). Secondly, there are missing values in our data collection: all result of the "real life" design of this study and reflect the variability of follow-up of ASV patients. Thirdly, other than compliance and AHI, there are no outcome measures and no subjective scores. It was the aim of this study to describe a new population in the field of home ventilatory support, and indeed, the results provide a new insight to the use of ASV in chronic home care and future areas of research. Finally, we recorded echocardiographic data only in patients corresponding to the SERVE-HF trial inclusion criteria (i.e., subjects with "de novo" CSA excluding patients under prior CPAP and thus ESA) (9) to determine to what extent patients with a LVEF <45% in this specific group of ASV users were still being treated. Because left ventricular dysfunction and chronic heart failure have also been documented in patients with emergent sleep apnea (ESA), the prevalence of cardiac dysfunction in the ASV population studied may be underestimated (26,31,44). Indeed 13% of ESA patients had a history of chronic heart failure and cardiovascular risk factors were highly prevalent. In a recent French multicentric descriptive study of 177 patients under ASV, 20.3% had emergent CSA: 75% of these patients had a cardiac disease and 14% had a LVEF<45% (45). Future studies should focus on ESA, clinical and echocardiographic findings in this group and prognosis under ASV.
In summary, use of ASV targets a much wider population than that of CSB in CHF. Although there are rationales for proposing ASV in the different populations described, and data presented suggest that ASV is effective in these groups (excellent compliance, correction of SDB according to AHI), SERVE-HF has shown to what extent evidence can be counter-intuitive. Because of the numeric importance of some of the patient groups (i.e., chronic opioid treatments or addiction in the US), future prospective randomized controlled trials must explore benefits of ASV in terms of symptom control, HRQL, and survival. The description of the population treated by ASV is a first step to initiate prospective randomized studies and validate-or notthe use of ASV in specific indications.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethical approval was granted by the Cantonal Commission for Research Ethics (CCER) in Geneva, Switzerland (no. PB_2016-00925/15-275) in agreement with the amended Declaration of Helsinki. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
CC, PP, and J-PJ contributed conception and design of the study. CC and PP organized the database. AP performed the statistical analysis. CC and J-PJ wrote the first draft of the manuscript. CC, PP, DA, and J-PJ wrote sections of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.

ACKNOWLEDGMENTS
We wish to thank the Ligue Pulmonaire Genevoise (www. lpge.ch) for its financial support. J-LP was supported by the French National Research Agency in the framework of the Investissements d'avenir program (ANR-15-IDEX-02). We are grateful to all the pulmonologists in private practice who accepted to participate in this study and made available their medical files. We also wish to express our gratitude to all the health care providers and device manufacturers in our area for their help in obtaining a complete database of patients under home mechanical ventilation in the Cantons of Geneva and Vaud as well as detailed information of the population under CPAP: The Pulmonary leagues (Ligue Pulmonaire Genevoise, Ligue Pulmonaire Vaudoise) ResMed Schweiz, Philips Respironics, the CENAS sleep center, SOS Oxygène Suisse SA, and Sleep & Health.