Cheyne-Stokes respiration in ischemic stroke

Cheyne-Stokes respiration (CSR) is thought of as a neurogenic alteration of the respiratory pattern typically accompanying bilateral or deeply seated brain damage ^[1-4]. However, a previous study ^[5] and our personal experience suggested that CSR was frequent in stroke in general. We investigated the respiratory patterns and arterial oxygenation changes in patients with brain infarcts of varying location to assess possible prognostic and therapeutic implications of pathologic breathing.

Methods. We examined 32 consecutive conscious inpatients admitted within three days after the onset of ischemic stroke (10 women and 22 men; age range, 19 to 89 years; median age, 64 years). None showed evidence of acute cardiopulmonary dysfunction or had received sedative drugs. Eight patients were hemiplegic, 11 were hemiparetic, three were monoparetic, 10 were dysarthric and ataxic, and four of the latter showed additional hemihypesthesia. Cranial CTs were available from all patients and did not reveal evidence of hemorrhage or transtentorial herniation. Infarct location was left supratentorial in 10 patients, right supratentorial in 12, and infratentorial in another 10 patients. We studied as a control group 20 age-matched subjects without a history of cerebrovascular disease. In all patients, chest and abdominal wall motion and arterial oxygen saturation were monitored over at least 1 hour using strain gauges and finger pulse oximetry. End-tidal CO₂ concentration of the expired air was monitored in four patients, using infrared photometry. The breathing pattern (respiratory motion) was evaluated employing fast Fourier transformed respiratory frequency analysis. CSR was defined as periodic modulation of the respiratory motion amplitude of greater than 50% amplitude depth occurring more than 10 times per hour.

Results. Compared with controls, stroke patients were slightly tachypneic (mean respiratory rate, 18.6/min; table 1). Respiratory monitoring revealed CSR in 13 of 22 patients (59%) with supratentorial stroke and in four of 10 patients (40%) with infratentorial stroke. The mean frequency of the respiratory amplitude modulation in patients with CSR was 0.77/min (SD 0.23/min). CSR was associated with concomitant periodic drops in arterial oxygen saturation below normal values (table, figure 1). Intravenous theophylline ethylenediamine (250 mg) was administered over 1 hour during respiratory monitoring in seven patients with CSR, and oxygen inhalation by face mask (2 l/min) was administered over 1 hour during respiratory monitoring in five patients with CSR. In all patients, breathing pattern and oxygen saturation normalized during treatment (figure 2). No adverse effects were noted.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Periodicity of arterial oxygen saturation (SaO₂; upper trace), chest wall motion (middle trace), and CO₂ concentration in the expired air (lower trace) in a stroke patient with Cheyne-Stokes respiration. The phase shift between upper and middle trace is due to the sampling time of the pulse oximeter of approximately 40 seconds. The drops in CO₂ concentration during hypopnea are due to dead space ventilation

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Normalization of SaO₂ (upper trace) and breathing pattern (lower trace) in a patient with Cheyne-Stokes respiration after start of oxygen inhalation (2 l/min by face mask)

Discussion. Although Cheyne based his classic description on observations of a hemiplegic stroke patient, ^[6] only one systematic study ^[5] reported the frequency of CSR in acute supratentorial brain infarction. We confirmed this study's finding of an approximately 50% frequency in our patient group. In addition, we found CSR-related blood oxygenation changes (table, figure 1). The periodic drops in oxygenation may constitute an additional hazard to the hypoperfused peri-infarct tissue of the ischemic penumbra ^[7] that may be particularly vulnerable to repeated episodes of relative hypoxia ^[8]. Theophylline ethylenediamine has been long known to abolish CSR ^[9]. That this agent as well as simple oxygen application normalized respiratory pattern and arterial oxygen saturation in all patients treated should encourage further studies of the therapeutic efficacy of these measures in stroke patients with CSR.

Experimentally, large parts of the suprapontine nervous system subserve respiratory control. The involved structures include mesencephalic and diencephalic nuclei and fronto-orbital, cingulate, insular, anterior temporal, and sensorimotor cortices ^[10]. In the present study, the occurrence of CSR was unrelated to infarct location. Among the patients with CSR, frequency variation of their respiratory amplitude modulation was low (mean, 0.77/min; SD 0.24/min), suggesting that CSR represents a relatively uniform response to injury of diverse parts of the CNS. Thus, our findings accord with the experimental concepts and provide clinical evidence that a large-scale distributed network underlies respiratory control in the human brain.