The objective of this scoping review was to identify and synthesize the available evidence on CSA syndromes in adults with ischemic stroke. The primary research question was: What is the extent of evidence on the prevalence, clinical manifestations, diagnostic methods, management strategies, and outcomes of CSA syndromes in adults with ischemic stroke across different care settings and disease phases?

Population: Studies were eligible if they included adults ($\ge$18 years) diagnosed with ischemic stroke confirmed by clinical and neuroimaging criteria, and if CSA syndromes were reported, identified through polysomnography or equivalent diagnostic testing. Although the prespecified eligibility criterion for CSA relied on a central apnea index threshold, we also retained studies that used pattern-based descriptions of central breathing instability, such as central periodic breathing or Cheyne–Stokes respiration, when they provided relevant CSA-related data. This approach was necessary because the literature uses heterogeneous definitions and reporting standards, and excluding these studies would have omitted clinically informative evidence. Accordingly, the review captures the full spectrum of CSA-related evidence reported after ischemic stroke, while explicitly acknowledging definitional heterogeneity.

Concept: Eligible studies explored the impact of CSA syndromes on stroke patients, including aspects of pathophysiology, diagnosis, treatment, clinical progression, or prognostic significance.

Context: Studies conducted in any healthcare setting (acute hospital units, rehabilitation centers, or community environments) and at any stage of illness (acute 7 days, subacute 1–3 months, or chronic $>$3 months) were accepted.

All types of quantitative, qualitative, and mixed‑methods research were eligible. These included randomized controlled trials, non‑randomized controlled studies, before‑and‑after or interrupted time‑series designs, prospective and retrospective cohorts, case‑control and cross‑sectional studies, as well as descriptive studies such as case series. Qualitative research using phenomenology, grounded theory, or ethnographic approaches was also accepted. In addition, systematic or scoping reviews meeting the inclusion criteria were considered for inclusion.

Exclusion criteria included:

Studies involving pediatric or non‑human populations.

Editorials, commentaries, and letters to the editor.

Conference abstracts, protocols, or documents without full text.

Papers without specific data on CSA syndromes or without confirmed ischemic stroke populations.

No restriction regarding geographic region was applied.

The literature search covered PubMed (https://pubmed.ncbi.nlm.nih.gov), Scopus (https://www.scopus.com/), Web of Science (Core Collection) (https://www.webofscience.com/wos/woscc/basic-search), and the Cochrane Library (https://www.cochranelibrary.com) databases from inception to 31 August 2025. Searches were restricted to articles published in English or Spanish. An initial pilot search was performed in PubMed to identify key Medical Subject Headings (MeSH) terms and free‑text related to two broad blocks of search terms: “sleep apnea, central” (“periodic breathing”, or “Cheyne-Stokes respiration” or “central breathing disorders”) and ischemic stroke (“brain stem infarctions” or “cerebral infarction” or “brain infarction” or “brain ischemia” or “cerebrovascular disorders”). The final PubMed strategy was adapted for the other databases (Supplementary Material I). To ensure completeness, the reference lists of all included papers and relevant systematic reviews were manually examined for additional studies. Grey literature and non‑peer‑reviewed publications were excluded.

All references retrieved from electronic databases were imported into Zotero 7 for Windows (v. 7.0.22) (Corporation for Digital Scholarship, https://www.zotero.org/) for reference management and duplicate removal. The selection process was conducted in two stages.

First, two independent reviewers screened article titles and abstracts for relevance against the predetermined inclusion criteria. Second, full‑text versions of all potentially eligible studies were obtained and evaluated independently by the same reviewers. Any disagreements were resolved by discussion or consultation with a third reviewer. Data extraction was conducted independently by two reviewers using a customized Microsoft Excel‑based extraction form designed specifically for this review (Supplementary Material II). Extracted data included the authorship, year of publication, country, study design, sample size, stroke type, lesion location, phase of disease, diagnostic methods, quantitative respiratory indices (CAI, Cheyne–Stokes respiration percentage, total AHI), comorbidities (atrial fibrillation, congestive heart failure, pulmonary disease), and reported outcomes including neurological severity (NIHSS), functional recovery (mRS, Barthel Index), mortality, and therapeutic interventions (e.g., CPAP or adaptive servo‑ventilation). Any inconsistencies identified during data extraction were discussed and resolved by consensus. Revisions to the extraction form made during the process were documented to ensure transparency.

Extracted data were organized according to the following variables:

Study characteristics (design, country, publication year).

Population data and stroke characteristics (phase, lesion location).

Diagnostic procedures applied for identifying CSA syndromes.

Main clinical and neurological outcomes (NIHSS, mRS, mortality).

Reported management and therapeutic strategies for CSA syndromes.

Each article’s main findings were summarized to describe the scope and nature of evidence addressing the predefined research question.

Because of the methodological heterogeneity of included studies, a quantitative meta‑analysis was not feasible. Instead, findings were synthesized in a narrative framework and systematically analyzed according to level of evidence, following the hierarchical standards of evidence‑based medicine.

Across the 55 included studies Table 1 (Ref. [3, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57]), 20 provided original CSA or central periodic breathing (CPB)/Cheyne–Stokes respiration (CSR) data in adult patients with ischemic stroke or mixed stroke cohorts including ischemic stroke [3, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17] to larger observational cohorts (n = 139 [6]; n = 156 [7], 2006; n = 182 [11], 2018; n = 166 [12], 2020) and a population‑based home sleep apnea test (HSAT) cohort of 1346 ischemic stroke patients [17].

Most CSA data came from the acute or early subacute phase. Recordings within 24 hours to 10 days after stroke onset were reported in several cohorts using polysomnography (PSG) or cardiorespiratory monitoring [5, 6, 7, 8, 9, 11, 13, 14, 15]. Subacute or early rehabilitation phase CSA was described around 40–44 days after stroke in a rehabilitation unit cohort [10], and within approximately one week in another full‑night PSG study [16]. Chronic post‑stroke CSA or coexisting central sleep apnea was assessed $\ge$3 months after first ischemic stroke in a monocentric cohort of 185 patients [3], in patients treated with adaptive servo‑ventilation (ASV) a median of 11 months after stroke [18], and at 3–6 months in a longitudinal unilateral lateral medullary infarction series [13]. Study designs included prospective observational cohorts, case–control studies, single‑patient case reports and small case series, as well as narrative and systematic reviews that pooled central versus obstructive patterns but did not always provide separate CSA indices [19, 20]. Several studies enrolled mixed ischemic and hemorrhagic stroke populations, but where CSA indices were reported separately or ischemic stroke predominated, only the CSA‑related ischemic stroke data were considered [7, 12].

Definitions and diagnostic modalities for CSA varied substantially between studies. Many cohorts used full‑night attended PSG, including EEG and standard respiratory channels, to identify central versus obstructive events by the absence of thoraco‑abdominal effort [10, 12, 13, 16, 23, 44]. Central sleep apnea was commonly defined as CSA when $\ge$50% of respiratory events were central and the apnea–hypopnea index (AHI) exceeded a threshold (e.g., AHI $\ge$5 or $\ge$10 events per hour), with a central AHI (cAHI) or central apnea index (CAI) used to quantify central burden [10, 17, 18, 44].

CSR and central periodic breathing during sleep (CPBS) were usually defined by a crescendo–decrescendo pattern of tidal volume and central apneas/hypopneas. Kim et al. [11] defined CSR as $\ge$3 consecutive central apneas/hypopneas with a crescendo–decrescendo breathing pattern, cycle length $\ge$40 seconds, and $\ge$5 central events per hour recorded over at least 2 hours. Hermann et al. [8] and Siccoli et al. [9] defined CPBS/CPB as $\ge$3 cycles of crescendo–decrescendo breathing with $\ge$50% reduction in nasal airflow lasting $\ge$10 seconds, with central patterns confirmed by reduced respiratory effort. Some acute studies relied on limited‑channel cardiorespiratory polygraphy without EEG [7, 9, 14, 15], using nasal pressure, thoracic and abdominal bands, and pulse oximetry to classify central events and CPB/CSR, but did not always provide explicit CAI thresholds for CSA as a syndrome. The large ischemic stroke HSAT cohort used a home sleep apnea test and defined CSA as CAI $\ge$5 events per hour with $\ge$50% of respiratory events being central [17].

Several case reports and small series documented CSA conversion or REM‑predominant CSA after stroke using PSG or portable monitoring. De Paolis et al. [26] reported an increase in central apnea index from 3 to 56 events per hour three days after a cardioembolic ischemic stroke, with total AHI 70 events per hour. Qu et al. [22] reported REM‑predominant CSA after medullary infarction with a baseline central AHI of 70.1 events per hour. Older acute ischemic stroke series assessed CSR using bedside respiratory motion and pulse oximetry without EEG or detailed CAI; Nachtmann et al. [5] defined CSR as periodic modulation of respiratory amplitude $>$50% of depth occurring more than 10 times per hour, without a numeric central index.

In several reviews [19, 20, 30, 31], CSA was conceptually defined as central apneas with absent respiratory effort and often as $\ge$50% central events within total AHI, but explicit numeric CAI cut‑offs specific to stroke cohorts were not consistently provided. A number of included stroke sleep‑disordered breathing studies focused on obstructive sleep apnea and total AHI without a separate central definition, and therefore did not contribute detailed CSA diagnostic criteria.

In acute ischemic stroke cohorts, the proportion of patients with CSR ranged from approximately 19% to more than 50% when defined by pattern rather than CAI alone. In an acute ischemic stroke cohort of 182 patients assessed with overnight sleep apnea testing, CSR was present in 35 patients (19.1%) [11]. In a 156‑patient acute stroke cohort monitored within 24 hours of onset, CPB was present in 33 of 138 patients with usable breathing recordings (24%) when defined as central crescendo–decrescendo periodic breathing occupying a quantifiable fraction of recording time [7]. In a 32‑patient acute ischemic stroke series with bedside respiratory monitoring, CSR was detected in 17 of 32 patients (53%) [5].

When CSA was defined by cAHI or CAI thresholds, prevalence estimates tended to be lower. In a rehabilitation cohort of 93 stroke patients, CSA (central AHI $\ge$10 events per hour) was present in 19% [10]. In the ischemic stroke HSAT cohort (1346 patients), CSA (CAI $\ge$5 events per hour and $\ge$50% of events central) was found in 1.4% (19/1346) [17]. In the multinational SAS‑CARE 1 PSG cohort including mainly ischemic strokes and transient ischemic attacks, CSA prevalence was reported at 13–17%, with CSR present in 21–25% at baseline, decreasing at 3‑month follow‑up [12].

Central indices also showed broad ranges. In Siccoli et al. [9], the central apnea index in 74 acute first‑ever ischemic stroke patients averaged 7 $\pm$ 12 events per hour (range 0–61), and in those with sleep apnea (AHI $>$10 events per hour) the CAI averaged 12 $\pm$ 14 events per hour (range 0–61). In Martínez García et al. [6], among 139 acute ischemic stroke patients studied within 72 hours, the central apnea index was 2.4 $\pm$ 2.3 events per hour overall, with higher values in documented cerebrovascular accidents versus transient ischemic attacks (2.9 $\pm$ 2.2 vs 1.9 $\pm$ 2.2 events per hour) and in nocturnal strokes versus daytime strokes (3.2 $\pm$ 1.8 vs 1.7 $\pm$ 1.5 events per hour).

In lesion‑specific cohorts, central indices were often high. In 28 patients with unilateral lateral medullary infarction, median acute AHI was 16.5 (interquartile range 5.5–43.3), cAHI 11.0 (3.3–29.1), and CAI 8.3 (2.2–23.8) events per hour; CSA alone was present in 43% and mixed patterns in additional patients [13]. In Riglietti et al. [15], dominant CSA ($\ge$70% of events central and central AHI $\ge$5 events per hour) in an acute cardiovascular–autonomic stroke cohort had mean central AHI 27.5 $\pm$ 15.1 events per hour, with central AHI across the cohort decreasing from 9.4 $\pm$ 12 to 4.0 $\pm$ 5.3 events per hour over 3 months.

Case‑level CSA severity was very high in some reports. In two fatal CSA cases with medullary lesions, 38 and 35 central apneas per hour were documented [21], and REM‑predominant CSA after bulbar infarction showed central AHI 70.1 events per hour [22]. In a case of CSA conversion after stroke, central apnea index increased from 3 to 56 events per hour three days after a cardioembolic ischemic stroke, with total AHI 70 events per hour [26].

One meta‑analysis that included mixed stroke types reported CSA prevalence 10.0% (95% confidence interval 6.5–14.9%) among stroke patients when CSA was defined as $\ge$50% central events within AHI, while obstructive events accounted for a higher proportion [20]. That pooled estimate did not provide ischemic stroke‑only prevalence and combined different phases and etiologies.

Cardiac comorbidities were frequently evaluated. In Nopmaneejumruslers et al. [10], CSA (central AHI $\ge$10 events per hour) was associated with lower mean nocturnal transcutaneous carbon dioxide tension (39.5 vs 43.0 mmHg) and a higher prevalence of left ventricular ejection fraction (LVEF) $\le$40% (22% vs 5%) compared with non‑CSA stroke patients. In the 182‑patient acute ischemic stroke cohort [11], CSR patients were about nine years older on average and had higher frequencies of atrial fibrillation (31.4% vs 9.5%) and heart failure (8.6% vs 0.0%) than non‑CSR patients, as well as larger left atrial size and left‑atrial volume index and a higher proportion with LVEF 50%.

In multivariable logistic regression, Kim et al. [11] found that previous modified Rankin Scale (mRS) score, bilateral hemispheric lesions, left‑atrial volume index (per 10 mL/m²) and LVEF 50% remained independently associated with CSR, whereas age, hypertension, and smoking were not significant after adjustment. CSR was uncommon in small‑vessel occlusion strokes (2.86%) compared with other Trial of Org 10172 in Acute Stroke Treatment (TOAST) subtypes in that study.

Rowat et al. [7] reported that CPB presence was associated with more severe strokes (higher National Institutes of Health Stroke Scale, NIHSS), congestive heart failure, dysphagia, and reduced consciousness, while severity of prior cerebrovascular disease was not associated with CPB. Siccoli et al. [9] found that more severe CPBS was associated with older age, higher stroke severity, electrocardiographic abnormalities, and lower LVEF.

Age, sex and body mass index (BMI) were recurrent correlates of sleep‑disordered breathing overall, although not always specifically of CSA. In the chronic Baillieul cohort [44], moderate–severe sleep‑disordered breathing (SDB), including coexisting or central sleep apnea defined by cAHI/AHI ratios, was associated with male sex, obesity, age $\ge$65 years, and infratentorial lesions; coexisting/central sleep apnea had a higher frequency of cerebellar lesions and increased hypercapnic ventilatory response compared with no or mild SDB. In the ischemic stroke HSAT cohort, CSA cases (1.4% overall) were more often male and non‑obese, and CSA was not significantly associated with NIHSS, heart failure or most other comorbidities in that dataset [17].

Other vascular risk factors showed inconsistent relationships with CSA across studies. In Kim et al. [11], hypertension was more frequent in the CSA group in univariate analysis (71.4% vs 55.8%) but was not independently associated with CSA, and smoking was less frequent in the CSA group but did not remain significant after adjustment. In Nopmaneejumruslers et al. [10] and some other cohorts, stroke type and location were not significantly related to CSA, whereas cardiac function parameters were. Some physiological cohorts described CSA in relation to autonomic measures. Riglietti et al. [15] reported that CSA and mixed CSA/OSA patterns in acute stroke were associated with specific autonomic profiles (e.g., higher baroreflex gain and autonomic nervous system indexes) compared with obstructive patterns, but numerical autonomic values were study‑specific and not directly comparable with other cohorts. In an asymptomatic carotid stenosis cohort (not post‑stroke), CSA prevalence of 39% among patients with extracranial internal carotid artery stenosis was linked to severe stenosis and autonomic imbalance [27], but this population was not included in the ischemic stroke clinical outcome analyses.

In large acute hemispheric and mixed stroke cohorts, some studies did not find strong associations between CSA and specific vascular territories, while others highlighted bilateral or infratentorial involvement. In the study of Kim et al. [11], CSA presence was not significantly associated with vascular territory, supra‑ versus infratentorial location, or laterality when considered separately; however, bilateral hemispheric lesions were more frequent in the CSA group and remained independently associated with CSA in multivariable analysis. In Nachtmann et al. [5], CSA was present in 59% of supratentorial and 40% of infratentorial ischemic strokes, and the authors reported no clear dependence of CSA occurrence on infarct location when categorized simply as supratentorial versus infratentorial.

Martínez García et al. [6] reported that central apnea index and CSA were relatively uncommon (five patients with Cheyne–Stokes breathing) and that polysomnographic variables, including the central apnea index, did not vary significantly according to stroke site (left hemisphere, right hemisphere, vertebrobasilar territory, or undetermined) or lacunar versus non‑lacunar extent. Nocturnal‑onset strokes in that cohort, however, had higher total AHI and higher central apnea index than daytime strokes.

Some studies focused on specific lesion locations. Hermann et al. [8] described three patients with CPBS among 31 acute first‑ever stroke patients; lesions involved the left cingulate cortex, left insula, and right paramedian thalamus on diffusion‑weighted MRI, and CPBS occupied 18–24% of total sleep time, improving on follow‑up recording. Pavšič et al. [13] investigated 28 patients with unilateral lateral medullary infarction and found that CSA, defined by central AHI $\ge$5 events per hour with obstructive AHI 5 events per hour, and mixed CSA/OSA were very frequent in this brainstem infarction cohort; central indices decreased over 3–6 months.

Rowat et al. [7] reported that CPB was associated with large acute cerebral hemispheric lesions and severe mass effect on neuroimaging, though detailed volumetric data were not provided; CPB was not associated with the severity of prior cerebrovascular disease. Siccoli et al. [9] found that CPBS was more severe in extensive hemispheric strokes and less frequent in strokes involving the left insula and mesencephalon, based on imaging classifications. In the cohort of Baillieul et al. [44], coexisting or central sleep apnea among patients with moderate–severe SDB was associated with a higher frequency of cerebellar lesions, and moderate–severe SDB in general was associated with infratentorial lesions.

In the observational cohort of Kim et al. [11], a subgroup analysis limited to large‑artery atherosclerosis without cardiac disease showed that patients with CSA were older and more likely to have both supra‑ and infratentorial lesions, but in multivariable analysis only the previous mRS score remained significant in that subgroup. Several narrative reviews [30, 31, 40] discussed medullary, pontine, insular and thalamic involvement in relation to CSA, but did not contribute new, independent lesion–CSA datasets beyond the primary studies already described.

A substantial proportion of the 55 studies did not systematically relate CSA to detailed neuroimaging findings; in many of those studies, lesion location was either not reported in a form usable for CSA‑specific correlation or was described without stratifying respiratory patterns by location.

In Rowat et al. [7], CPB presence was independently associated with death or dependency (modified Rankin Scale $\ge$3) at 3 months; the adjusted odds ratio for this combined endpoint was approximately 5.9, and patients with CPB had higher mortality rates and lower Barthel Index scores than those without CPB. CPB was also associated with acute complications such as dysphagia and reduced consciousness.

In the SAS‑CARE 1 cohort [12], higher baseline AHI (including central and obstructive events) was associated with poor functional outcome (mRS $\ge$3) at 3 months, and the mean AHI decreased modestly between baseline and follow‑up; CSA proportions declined over this interval but CSA‑specific outcome analyses were not separated from overall AHI. Thus, a direct estimate of functional or mortality risk attributable exclusively to CSA was not reported.

In Mekky et al. [16], stroke patients had higher AHI, CSA index and desaturation index and lower minimal oxygen saturation compared with age‑matched controls; within the stroke group, lower minimal SpO₂ at one week correlated with worse NIHSS at one month. The study did not report functional outcomes stratified solely by CSA status, and no mortality analysis was presented specific to CSA.

In Nopmaneejumruslers et al. [10], clinical outcomes such as mortality or long‑term disability by CSA status were not the primary focus; the study concentrated on CSA prevalence and its relationship with cardiac function and transcutaneous carbon dioxide tension (CSA‑specific outcomes not reported). In Pavšič et al. [13], outcomes of interest were changes in AHI and cAHI over time in unilateral lateral medullary infarction; functional or mortality endpoints stratified by CSA were not detailed (not reported).

Kim et al. [11] reported baseline NIHSS and premorbid mRS differences between CSR and non‑CSR groups but did not present separate long‑term functional outcome or mortality rates according to CSR status beyond the logistic regression models for CSR determinants (CSR‑specific prognostic data not reported). In Martínez García et al. [6] and Nachtmann et al. [5], respiratory patterns were described and comparisons with general population or control data were made, but CSA‑specific mortality or disability statistics were not reported.

Narrative reviews and meta‑analyses [19, 20, 31, 39, 40] commonly stated that CSA after stroke are associated with more severe neurological impairment or higher mortality, but did not provide new numerical outcome data for CSA‑only groups beyond the primary studies already summarized.

Therapeutic data were limited to small observational reports and case series; one retrospective series found that adaptive servoventilation reduced apnea–hypopnea index and improved symptoms in post-acute stroke patients with persistent CSA, while other reports described partial spontaneous resolution or treatment-related changes in central indices [18, 22, 26].

Overall, among the 55 studies included in this scoping review, a limited subset provided quantitative outcome measures that were explicitly stratified by CSA, CPB or CSR status in ischemic stroke; in many cohorts, outcomes were reported for sleep‑disordered breathing as a whole or were not separated by respiratory pattern, resulting in “not reported” for CSA‑specific functional status, complication rates or mortality in a considerable proportion of the evidence base.

The variability in CSA prevalence and phenotype can be explained by the different ways stroke and comorbid disease alter ventilatory control. Unilateral lesions in insula, cingulate cortex, thalamus and medulla were sufficient to generate CPB or CSA in patients without severe heart disease, with partial resolution over weeks or months, supporting a direct neurogenic contribution from damage to central respiratory networks [8, 13, 21]. At the same time, several cohorts showed that central patterns concentrate in patients with reduced left ventricular ejection fraction, enlarged left atrium, lower nocturnal carbon dioxide and atrial fibrillation, indicating that circulation delay and high loop gain frequently interact with stroke‑related injury to produce CSA [10, 11]. Longitudinal data—central indices falling in some lesion‑focused series but rising in broader stroke cohorts—suggest that as the brain recovers and cardiac status evolves, the balance between obstructive and central events shifts, which accounts for apparently conflicting time‑course findings [8, 12, 13, 15, 57].

The pathophysiology of CSA in ischemic stroke appears multifactorial, involving disruption of central respiratory networks and interaction with cardiac dysfunction and ventilatory instability. From a clinical standpoint, identifying CSA may help risk-stratify stroke patients, but the evidence for specific management strategies remains limited and largely non-randomized.

These mechanisms help interpret why central patterns carry prognostic information in several studies. Because CSA tends to occur in patients with more severe neurological deficits and worse cardiac status, their presence identifies individuals who already have a high risk profile, which explains the observed associations with death or dependency rather than implying direct causality [7, 12, 23]. Early sleep testing captures transient but prognostically relevant central instabilities, yet feasibility constraints mean that the sickest patients are often not studied, so current data likely underestimate the burden of CSA in real‑world stroke [11, 58]. Case reports and small series showing conversion from obstructive to central patterns and response to bilevel or adaptive servo‑ventilation illustrate that management may need to adapt dynamically as central instability emerges, but also highlight that some central patterns resolve spontaneously, supporting careful, individualized decision‑making [18, 22, 26] (Fig. 2).

Fig. 2.

Conceptual framework of central sleep apnea after ischemic stroke. The figure summarizes the main pathways described in the literature, including disruption of central respiratory networks, altered chemoreceptive and autonomic control, ventilatory instability, and the interaction with cardiac dysfunction, which may lead to central apneas, periodic breathing, and adverse clinical outcomes.

The way the existing evidence is structured points directly to what is needed next. Harmonized, stroke‑specific definitions that integrate central indices and pattern descriptors would reduce artificial heterogeneity and make future studies comparable [9, 10, 11, 44, 55]. Prospective multicenter cohorts that include severe strokes and repeat standardized sleep studies from the acute phase into the chronic stage are required to disentangle transient neurogenic CSA from more persistent, heart‑failure–related phenotypes and to clarify how lesion topography, cardiac function and autonomic markers jointly shape central instability [8, 12, 13, 15, 57]. Finally, pragmatic interventional trials embedded in stroke services—comparing diagnostic strategies and treatment pathways for patients with prominent central patterns—are needed to test whether identifying and treating CSA after ischemic stroke can meaningfully change functional outcomes or vascular risk, rather than simply describe another marker of severity [7, 18, 22, 26].

The current evidence supports a prioritized research agenda. Future studies should first adopt standardized, stroke-specific definitions of CSA and related central breathing phenotypes. They should then incorporate longitudinal phenotyping across the acute-to-chronic stroke continuum, stratify results by stroke subtype and lesion location, and evaluate potential modifiers such as sex, age, and cardiac function. Finally, pragmatic interventional studies are needed to determine whether identification and treatment of CSA after ischemic stroke improve functional recovery and other clinical outcomes.