Disordered breathing in severe cerebral illness – Towards a conceptual framework

Despite potentially life-threatening symptoms of disordered breathing in severe cerebral illness, there are no clear recommendations on diagnostic and therapeutic strategies for these patients. To identify types of breathing disorders observed in severely neurological comprised patients, to direct further research on classification, pathophysiology, diagnosis and treatment for disordered breathing in cerebral disease. Data including polygraphy, transcutaneous capnometry, blood gas analysis and radiological examinations of patients with severe cerebral illness and disordered breathing admitted to the neurological intensive care were analyzed. Patients (15) presented with acquired central hypoventilation syndrome (ACHS), central bradypnea, central tachypnea, obstructive, mixed and central apneas and hypopneas, Cheyne Stokes respiration, ataxic (Biot's) breathing, cluster breathing and respiration alternans. Severe cerebral illness may result in an ACHS and in a variety of disorders of the respiratory rhythm. Two of these, abrupt switches between breathing patterns and respiration alternans, suggest the existence of a rhythmogenic respiratory network. Polygraphy, transcutaneous capnometry, blood gas analysis and MRI are promising tools for diagnosis and research alike.


Introduction
The neural control of respiration is crucial for survival in humans and animals. Scientific evidence on this topic was collected and established for centuries (Firth, 1897;West, 2014). Most knowledge is derived from animal models, however the last two decades brought progress in clinical, neuroradiological, electrophysiological, anatomic and genetic methods that helped defining a model of respiratory control in humans. In healthy subjects respiratory control maintains a physiological gas exchange in rest and during exercise and coordinates respiration with reflexes like cough, swallowing, choking and sneezing. The respiratory control furthermore integrates voluntary activities as speech and singing into the breathing rhythm and mediates other influences like sleep, stress, infection, emotion and pain. The respiratory rhythm is divided into an inspiratory phase, a post-inspiratory phase, corresponding to passive expiration, and a expiratory phase, corresponding to active expiration, e.g. during exercise or cough (Richter, 1982).
Patients with severe cerebral illnessas defined in the methods section beloware treated on intensive care units, mechanical ventilation weaning and rehabilitation units. Care structures in Germany have been observed and described by Rollnik et al. (2020). However on an international level they are not well studied and known. Breathing irregularities can be observed in acute life-threatening illness as well as during weaning from the ventilator and early rehabilitation. Although brain illness forms a highly important global burden (Roth et al., 2020) only few case reports and small series on disordered breathing in severe cerebral illness have been published (Rollnik et al., 2020;Sivakumar et al., 2018;Windisch et al., 2018). Due to a lack of comprehensive clinical studies in patients with disordered breathing in severe cerebral illness, the consequences of clinically observed breathing irregularities are unclear.
Diagnostic criteria for obstructive and central apneas or hypopneas, cheyne-stokes-respiration, bradypnea and central neurogenic hyperventilation have already been published (Berry et al., 2020;Sweidan et al., 2019). Laboratory criteria of ACHS have yet to be consented, but in reported cases hypercapnia was consistently present (Sivakumar et al., 2018) as it is the case in congenital hypoventilation syndrome (Hou et al., 2018;Maloney et al., 2018;Stankiewicz and Pazevic, 1989). Ataxic breathing, -also called Biot's breathing (Wijdicks, 2007) -, cluster breathing, respiratory alternans and other breathing patterns were also described, but these descriptions still lack quantifiable criteria. Ataxic breathing shows irregular variability of breathing effort and timing, whilst cluster breathing is defined as "regular cycles of deep breaths with variable periodicity". Respiration alternans consists of "small breaths. interposed between full breaths" (Fisher, 1969;Wijdicks, 2007). There is no comprehensive model of disordered breathing in severe cerebral illness to date.

Objective
This study aimed at identifying types of breathing disorders observed in neurological ill intensive care and early rehabilitation patients to direct further research on classification, pathophysiology, diagnosis and treatment for disordered breathing in cerebral disease.

Research questions
Which types of disordered breathing are detected by polygraphy, transcutaneous kapnometry and blood gas analysis in patients with severe cerebral illness?
Does the identification of disordered breathing change the treatment strategy, e.g. result in the indication for initiation of long-term ventilation?

Materials and methods
The study was approved by the Medical Ethics Committee of the University Oldenburg with the trial identifier 2019-050 and registered with the German registry of clinical trials (https://www.drks.de/drks_ web/setLocale_EN.do; clinical trial identifier number DRKS00011391). Written informed consent was obtained from all participants or their guardians, the study conforms with the World Medical Association Declaration of Helsinki (World Medical, 2013). We included patients that were administered to our hospitals Neurological Intensive Care Unit during the years 2015-2021 and had an early rehabilitation Barthel-Index below 35 points or received intensive care treatment due to a severe cerebral disease. Furthermore, patients needed to display at least one of the following symptom-constellations for study inclusion: 1. Existent ACHS, defined as: i. pCO 2 ≥ 45 mmHg at wake or pCO 2 ≥ 50 mmHg in sleep or nighttime, tcCO 2 ≥ 50 mmHg for > 30 min ii. caused by cerebral illness iii. no other illness causing hypercapnia (obesity, COPD and other illnesses of lung and thorax, neuromuscular disease, acute noncerebral illness with indication for intensive-care-unit-treatment) 2. Present central neurogenic hyperventilation, defined as: i. Respiratory alkalosis ii. Hyperventilation or tachypnea persisting in sleep 3. Diagnosed abnormal breathing pattern, defined as: i. apnea-hypopnea-index > 5/h (Berry et al., 2020) Patients being either younger than 18 years or suffering from heart failure, COPD GOLD III-IV or radiological finding of pulmonary emphysema were excluded.
Polygraphy was performed with Somnotouch Resp (SOMNOmedics GmbH, Randersacker, Germany) and validated by a sleep medicine specialist according to the AASM criteria (Berry et al., 2020). For transcutaneous capnometry a Sentec capnometer (SenTec AG, Therwil, Switzerland) was used. Due to the expected size of the cohort a statistical analysis was neither planned nor performed. In case of missing data or fail to collect data at a certain timepoint, they were reported missing. Preparation of the manuscript has been conducted in adherence with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement and guidelines (von Elm et al., 2007). By request from any qualified investigator anonymized patient data not demonstrated in this article will be shared by the corresponding author.

Results
Three Female and twelve male patients, with an age span of 49-78 years, were included. All individuals had a tracheal cannula and were mechanically ventilated on admission ( Table 1). None of these patients was diagnosed with a disorder of respiratory regulation prior to admission.
Six patients were diagnosed with ACHS, one in combination with central tachypnea (pat. 8), one in combination with central bradypnea (pat. 7), one in combination with central apneas and hypopneas (pat. 13) and one in combination with a highly complex disorder of the respiratory rhythm (pat. 2). Confirmed via radiological findings, all patients with ACHS had a lateral medullary stroke (Table 1 and Fig. 1). The most severe case, persistently needing 24 h of mechanical ventilation, had consecutively suffered bilateral medullary strokes (pat. 5). Normocapnia in patients with ACHS was achieved by pressure-controlled ventilation (PCV), whilst pressure support ventilation (PSV) was ineffective.
Three patients with either supramedullary or medullary pathology were diagnosed with cheyne-stokes-respiration. One patient with pontine hemorrhage showed isolated central apneas and hypopneas.
Six patients suffered from highly complex disorders of the respiratory rhythm: Pat. 2 suffered from a ACHS, central apnea and hypopnea, central bradypnea, central tachypnea and respiration alternans. Pat. 3 presented central bradypnea, central tachypnea, central apnea and hypopnea and ataxic breathing in clinically bilateral medullary stroke with only unilateral clearly verifiable radiological presentation. Pat. 10 possessed spasmodic breathing, that interrupted the weaning from the mechanical ventilator at the intensive care unit as it led to violation of the critical values. The spasmodic breathing was clinically attributed to phrenic dystonia that has been successfully treated by oral medication with Gabapentine 3 × 300 mg per day p.o, with an immediate (within 24 h) effect. Under medication polygraphy unveiled a Crescendodecrescendo-breathing not meeting AASM-criteria for cheyne-stokesrespiration (Berry et al., 2020), central bradypnea and respiration alternans. Pat. 11 showed a combination of central apnea and hypopneas, central bradypnea and cluster breathing (Wijdicks, 2007). Pat. 12 presented central apnea and hypopneas, central bradypnea and central tachypnea, and Patient 15 central apnea and hypopneas as well as central bradypnea and ataxic breathing. All six patients with highly complex disorders of the respiratory rhythm featured rapid and fundamental changes of their breathing pattern. Respiration alternans accompanied this switch-over in Pat. 2 and 10 ( Fig. 2).  Four patients were decannulated. In ten patient long-term ventilation was initiated to improve daytime sleepiness and headaches. Longterm mechanical ventilation assured survival in patients 5, 6 and 7.

Discussion
The spectrum of disordered breathing in this study was diverse: Patients showed ACHS, central bradypnea, tachypnea, apnea and hypopnea, cheyne-stokes-respiration, ataxic breathing and respiration alternans, or a combination of the aforementioned (Table 1). Notably, specific breathing patterns may abruptly switch from one to the other, which is in some patients accompanied by respiration alternans. To our knowledge, we present the first polygraphic data in literature of ataxic breathing, cluster breathing, respiratory alternans and of abrupt switches between different breathing patterns (Fig. 2).
The regions of the medulla oblongata generating the respiratory rhythm and controlling CO2-homeostasis are anatomically close to each other, but still distinct (Smith et al., 2013). The here presented data demonstrate that disorders of the respiratory rhythm and of the CO 2 -dependent respiratory drive can exist separately or coexist respectively. The data support a two-axis-model of disordered breathing in severe cerebral disease. Disordered breathing of cerebral origin should be discriminated in respect to those two axes, the respiratory rhythm and the CO 2 -dependent respiratory drive. Central apnea and hypopnea (Berry et al., 2020), tachypnea, bradypnea, ataxic breathing, cluster breathing, respiratory alternans (Fisher, 1969;Wijdicks, 2007Wijdicks, , 2017 and sudden switches between breathing patterns are disorders of the respiratory rhythm, whilst the ACHS is a disorder of the CO 2 -dependent respiratory drive. Disordered respiratory rhythm and CO2-dependent respiratory drive may interact, as it is the case in cheyne-stokes-respiration, presenting as a specific breathing rhythm triggered by an altered CO 2 -sensitive apnea threshold (Cherniack et al., 2005;Cherniack and Longobardo, 1973;Guyenet et al., 2019).
The reduced CO 2 -dependent respiratory drive, reflected by a reduced hypercapnic ventilatory response in lesions of the rostrolateral medulla, as probable cause of ACHS is already well-established(Nattie and Li,  Table 1. Acquisition technique is given as subscript (cranial computer tomography (CCT), Diffusion-weighted magnetic resonance imaging (DWI), fluid-attenuated inversion recovery (FLAIR), magnetization-prepared rapid gradient-echo (mprage), timing of radiofrequency pulse sequences 1 (T1), timing of radiofrequency pulse sequences 2 (T2). Red arrows indicate lesion. Time from according neurological event to radiographic finding: P1 = 14 weeks, P2 = 53 weeks, P3 = 58 weeks, P4 = 4 weeks, P5 = 28 weeks, P6 = 2 weeks, P7 = 9 weeks, P8 = 5 weeks, P9 = 2 weeks, P10 = 1 week, P11 = 2 weeks, P12 = 6 days, P13 = 8 weeks, P14 = 4 weeks and P15 = 13 weeks. All images with subscript allegation of acquisition technique.  Table 1. Vertical lines marking episodes of 10 s. SpO2 is displayed as a percentage, Pulse as numbers of heart rate in beats per minute, the range of the here displayed measurement is given with respect to the upper and lower limit according to the light-grey horizontal lines confining the reading. All other categories as airflow, effort, thorax(-movement), abdomen(-movement) and activity are displayed in arbitrary units. The areas of airflow reading, marked in red represent central apnea. The areas of the airflow reading, marked in green represent hypopnea.The areas of the SpO2 reading, marked in red represent a desaturation of ≥ 3%. Patient 2: respiration alternans with bradypnea (f=6-7/min.) and an underlying tachypnoeic crescendo-decrescendo breathing pattern at the same time, Patient 3: ataxic breathing with highly variable respiratory effort, bradypnea (6/ min.) central apnea and a sudden tachpnea lasting 40 s. Patient 10: tachypnoeic crescendo-decrescendo breathing pattern followed by bradypnea (5-6/min.), both breathing patterns coexist for one minute, Patient 11: cluster breathing with regular cycles of 3-4 breaths separated by central apnea, Patient 12: a rapid and sudden switch from tachypnea to bradypnea (8-9/min.), both separated by apnea, Patient 15: ataxic breathing with irregular variability of breathing effort and timing, less pronounced than in patient 3. 2012). Our data are consistent with the findings of previous case reports and series, demonstrating lateral medullary infarction being among the most frequent causes of ACHS (Hunziker et al., 1964;Kumral et al., 2011), which can be life-threatening in absence of mechanical ventilation. Although it is stated that anatomic correlates of disorders of the respiratory rhythm in severe cerebral illness are more variable and not well understood regions of importance have been identified allowing basic insight to functional connections of anatomical distinct areas (Bjerrum and Rosendal, 2015;Bonnin-Vilaplana et al., 2012;Lee et al., 1976;Rahmanian et al., 2014;Stewart et al., 1996;Wong and Duffy, 1982).
The key regions of respiratory control are the solitary nucleus, located in the dorsal medulla oblongata, the ventral respiratory group (VRG), the retrotrapezoid nucleus and the parafacial respiratory group, located in the ventral medulla, and the Kölliker-Fuse-parabrachialcomplex, located in the pons (Ikeda et al., 2017).
All peripheral afferent projections of mechano-and chemoceptors converge on Neurons of the Nucleus tractus solitarius in the dorsal medulla oblongata, which forwards the processed information to medullary and pontine respiratory neurons (Alheid et al., 2011;Kubin et al., 2006).
The ventral respiratory group consists of a column of neurons along the ventral medulla oblongata, from the most rostrally located Bötzinger-complex, followed in caudal direction subsequently by the pre-Bötzinger-Complex, rostral VRG and caudal VRG (Rybak et al., 2007;Smith et al., 2013). Throughout the VRG there are populations of inspiratory and expiratory neurons. The VRG gets afferent input from the opposite VRG, the Nucleus tractus solitarius and many other brain regions that take part in respiratory regulation (Ikeda et al., 2017;Smith et al., 2013). The pre-Bötzinger-complex is the inspiratory pacemaker (Anderson and Ramirez, 2017;Ikeda et al., 2017). Pre--Bötzinger-complex, parafacial respiratory group and possibly the postinspiratory complex, close to the VRG and caudal to the facial nucleus, are the oscillators generating the respiratory rhythm (Anderson et al., 2016;Anderson and Ramirez, 2017;Ghali et al., 2020;Ikeda et al., 2017;Pisanski and Pagliardini, 2019).
Located in the pons, Kölliker-Fuse-parabrachial-complex regulates the variability of the respiratory rhythm and controls the transition from the inspiratory to the post-inspiratory and probably from the postinspiratory to the late-expiratory phase (Barnett et al., 2018;Dhingra et al., 2017;Dutschmann and Herbert, 2006;Jenkin et al., 2017). Also located in the pons, PB controls the breathing frequency by changing the duration of expiration (Zuperku et al., 2017).
Numerous studies and observations suggested that chemoceptors, baroceptors and the integrative neurons that modulate the sympathic/ parasympathic regulatory areas form a network, so that the generator is of the respiratory rhythm is more likely to be distributed than one distinct anatomical structure (Dhingra et al., 2020(Dhingra et al., , 2019Nattie, 1999). This conclusion of previous work is supported by two observations of our study: Firstly the cerebral imaging data of the here given study, taken together with the polygraphic datademonstrate that different lesion sites may result in similar disorders of the respiratory pattern. Secondly the pattern of respiration alternans documented in our study with two fundamentally different respiratory rhythms executed parallel in one patient can best be explained by the coexistence of at least two different generators of the respiratory rhythm. While the rapid and fundamental changes in respiratory rhythm that we observed in six patients might be explained by a single generator generating different rhythms, this explanation is highly improbable in respiration alternans. One may hypothesize a basal bradypneic generator which can be switched off or overdriven by faster respiratory rhythms. From an evolutionary perspective a redundant organization of the respiratory network protects the individuum from the potentially detrimental consequences of respiratory failure. While none of the patients was pre-diagnosed with a disordered breathing, polygraphy, transcutaneous or end-tidal capnometry, blood gas analysis and MRI have proven to be suitable to diagnose ACHS and disorders of the respiratory rhythm. This approach resulted in the initiation of long-term mechanical ventilation in ten of fourteen surviving patients.
One limitation of our study, that on admission all patients had a tracheostomy cannula, masking possible obstructions of the upper airways. Also, while the definition of ACHS presented in the methods section of our study has the advantage of clear and quantifiable diagnostic criteria, it still needs validation in a higher number of patients. Further on, no information on the incidence of disordered Breathing can be derived from this study. The lack of quantifiable data demonstrating clinical improvement of patients, who were electively put on mechanical ventilation, is a further limitation.

Conclusion
Serial blood gas analyses, transcutaneous or endtidal kapnography, polygraphy or polysomnography and cerebral imaging with focus on pontine and medullary structures potentially reveal a variety of central disorders of breathing. Collection of these parameters will help future studies to quantify criteria for disordered breathing in severe cerebral illness, to correlate disordered breathing to the anatomy of the respiratory network and to evaluate therapeutic options. We identified a broad spectrum of breathing disorders in our patients, suggesting the existence of a neuronal network generating the respiratory rhythm rather than a sole anatomical structure. In order to guarantee an optimum for gas exchange parameters as well as an effective treatment of central apnea, we recommend controlled ventilation in opposition to assisted ventilation in these disorders, due to the inability of the latter to treat hypercapnia and central apnea.

Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.