In sleep-related breathing disorders (SBD), Cheyne-Stokes respiration (CSR) represents a distinct form of central sleep apnea event. Current research has demonstrated its association with heart failure (HF), atrial fibrillation, renal diseases, and neurological disorders, with particularly extensive studies focusing on its relationship with congestive heart failure (1). However, current understanding of central sleep apnea (CSA) and CSR remains limited, particularly regarding their underlying mechanisms and clinical implications.

Most studies suggest that the combination of CSA with CSR plays an important role in the progression of heart function deterioration and is a strong independent indicator of the mortality rate of patients with atrial fibrillation (AF) and heart failure (2–8). May (6) demonstrated that CSR serves as an independent predictor of AF onset, particularly among older participants. Calvin (3) found that Heart failure patients with CSA and CSR exhibited left atrial enlargement and were older compared to those without CSA. Lo (4) found that CSR were associated with lower ejection fraction and larger left atrial size. Bitter (5) found that in patients with heart failure and preserved ejection fraction (HFpEF), cases of CSA, as a proportion of SBD, increased alongside increases in the degree of impairment of left ventricular diastolic function. Saito (7, 8) further validated the correlation between CSR and both AF and HF during long-term continuous positive airway pressure (CPAP) monitoring, and proposed that tracking CSR changes may facilitate early-stage prediction of HF onset and deterioration. Alternative perspectives propose that CSA with CSR may be a compensatory response to impending heart failure, and CSR could be a protection mechanism in HF (9, 10).

Besides, Most of the aforementioned studies were conducted using an unattended ambulatory system rather than standard polysomnography (PSG) to screen for Obstructive Sleep Apnea (OSA) or CSA and did not further identify or discuss the impact of respiratory events. For example, although mixed apnea events are classified under the obstructive sleep apnea index (OAHI), their potential inclusion of CSR (11, 12) has rarely been discussed in clinical contexts.

While existing studies on SBD with CSR have predominantly targeted advanced heart failure populations, research in patients without a confirmed diagnosis of heart failure remains limited. Notably, CSR events have been clinically documented in patients with neither a confirmed diagnosis of heart failure nor related symptoms.

Therefore, we propose to analyze echocardiography, BNP/NT-proBNP level, PSG and medical history in patients without a confirmed diagnosis of heart failure and presenting no related symptoms, with the aim of investigating whether CSR events correlate with specific cardiac structural/functional alterations and potentially indicate occult heart failure states.

This was a cross-sectional study that collected data including medical history, PSG and echocardiography from hospitalized patients from Jan. 2018 to Jan. 2025 at Guangdong Provincial Hospital of Traditional Chinese Medicine. This retrospective analysis was granted exemption by the Guangdong Provincial Hospital of Chinese Medicine Ethics Committee, as it utilized de-identified data from a parent study with existing informed consent.

Inclusion criteria: 1.

Patients aged ≥45 years.

This study enrolled a total of 171 patients. Group SBD + CSR demonstrated significantly higher mean age (χ² = 24.687, p < 0.001) and prevalence of Af (χ² = 11.078, p = 0.004) compared to the other groups. In contrast, Group nonSBD exhibited a markedly lower mean BMI (χ² = 12.839, P = 0.002) and prevalence of HBP (χ² = 20.182, p < 0.001) relative to the remaining cohorts. Of the 78 patients who completed the 2-year follow-up, 20% (4/20) in Group SBD + CSR were clinically diagnosed with heart failure during the follow-up period (Table 1).

Significant differences were observed across groups in both BNP/Nt-proBNP levels (Figure 1 and Supplementary Table 1) and structural cardiac parameters (Table 2). Group SBD + CSR exhibited markedly higher BNP/Nt-proBNP levels compared to Groups SBD-CSR and nonSBD (χ² = 22.869, p < 0.001, Figure 1). Echocardiographic analysis further revealed that Group SBD + CSR had larger left atrial dimensions (LAd: 3.5[3.1, 3.9] cm vs. 3.3[2.9, 3.5]/3.0[2.9, 3.2] cm in Group SBD-CSR/nonSBD, p = 0.004/p < 0.001; LADi: 2.0[1.8, 2.3] cm/m² vs. 1.8[1.6, 1.9]/1.8[1.6, 2.0] cm/m² in Group SBD-CSR/nonSBD, p < 0.001/p < 0.001) and higher average E/e' ratios [12.9[10.7, 14.5] vs. 8.7[7.4, 9.8]/7.6[6.7, 8.8] in Group SBD-CSR/nonSBD, p < 0.001/p < 0.001]. Conversely, Group nonSBD demonstrated reduced IVSd [1.0[0.9, 1.1] vs. 1.1[1.0, 1.2]/1.1[1.0, 1.2] in Group SBD + CSR/SBD-CSR, p < 0.001/p < 0.001] and LVPW [1.0[1.0, 1.1] vs. 1.1[1.1, 1.2]/1.1[1.0, 1.1] in Group SBD + CSR/SBD-CSR, p < 0.001/p = 0.008] relative to other groups (Figure 2).

Ordinal logistic regression analysis of variables associated with increased BNP/Nt-proBNP levels revealed that age [OR = 1.09 per year, 95%CI (1.05–1.13), p < 0.001], CSR Events [OR = 2.02, 95%CI (1.02–3.98), p = 0.044], lowest SpO2 [OR = 0.93, 95%CI (0.86–1.00), p = 0.044] and atrial fibrillation [OR = 5.11, 95%CI (1.96–13.32), p = 0.001] were independent predictors (Table 3). Analysis of echocardiography parameters, LAd revealed consistent associations for age [OR = 1.07 per year, 95%CI (1.04–1.10), p < 0.001], BMI [OR = 1.15, 95%CI (1.05–1.26), p = 0.003] and atrial fibrillation [OR = 6.18, 95%CI (1.11–34.44), p = 0.038] (Table 4). Extending this framework to LADi categories demonstrated overlapping risk profiles: age [OR = 1.08 per year, 95%CI (1.05–1.12)] and BMI [OR = 0.90(0.83–0.99)] maintained significance, with left atrial diameter index uniquely linked to CSR Events [OR = 3.15, 95%CI (1.50–6.64), p = 0.002] (Table 5). Similar patterns emerged in average E/e' analysis for age [OR = 1.05 per year, 95%CI (1.01–1.08), p = 0.006] and CSR Events [OR = 15.32, 95%CI (6.48–36.22), p < 0.001] (Table 6). Separate parallel lines tests for all model confirmed the proportional odds assumption, ensuring consistent coefficient interpretation across outcome categories.

Further linear regression analyses explored associations between CSR/TST% and cardiac markers previously identified as significant in ordinal logistic regression in Group SBD + CSR. CSR/TST% showed positive linear relationships with NT-proBNP (β = 0.40, R² = 0.16, p = 0.006, Supplementary Figure 1), LADi (β = 0.30, R² = 0.09, p = 0.047, Supplementary Figure 2), and average E/e' ratio (β = 0.41, R² = 0.17, p = 0.005, Supplementary Figure 3).

Binary logistic regression for IVSd abnormality (Female >1.0 cm/Male >1.1 cm) identified BMI [OR = 1.18, 95%CI (1.04–1.35), p = 0.013], hypertension [OR = 3.93, 95%CI (1.71–9.00), p = 0.001] and atrial fibrillation [OR = 0.19, 95%CI (0.04–0.98), p = 0.047] as significant predictors (Supplementary Table 2). For LVPW abnormality (same thresholds), age [O = 1.10 per year, 95%CI (1.01–1.10), p = 0.002] and BMI [OR = 1.25, 95%CI (1.09–1.44), p = 0.017] showed significant associations (Supplementary Table 3). Both models passed the Hosmer-Lemeshow test (p > 0.05) with Omnibus Tests confirming superiority over null models (p < 0.001). In contrast, the PA abnormality (>2.6 cm) analysis showed non-significant Omnibus Tests (p = 0.204) despite acceptable goodness-of-fit (Hosmer-Lemeshow p > 0.05), indicating insufficient evidence for a valid predictive model (Supplementary Table 4).

This study is the first to specifically investigate patients exhibiting CSR events without confirmed heart failure diagnosis or heart failure symptoms. We aim to explore potential associations between CSR events and cardiac structural/functional parameters in this population, with the objective of determining whether CSR may indicate specific cardiac remodeling patterns, or serve as a preclinical marker for impending heart failure. The results demonstrated that those patients exhibited older age, elevated BNP levels, higher AF prevalence, and distinct cardiac remodeling patterns featuring left atrial enlargement and increased E/e’ ratio.

Prior investigations (3, 5, 17) consistently report advanced age and predisposition to left atrial enlargement in HF patients exhibiting CSR. Our study extends these observations to non-HF individuals. Although the more sensitive parameter of left atrial volume index (LAVi) was not utilized, ordinal logistic regression models adjusted for established LADi determinants demonstrated that CSR remains independently associated with LA enlargement [including age and BMI (18, 19)].

The pivotal finding of our study demonstrates significant positive correlations between CSR events and both NT-proBNP levels and E/e', with the latter showing stronger clinical relevance. As a validated hemodynamic marker of diastolic dysfunction, E/e' synergizes with elevated BNP/NT-proBNP and LA enlargement to form a diagnostic triad for early cardiac decompensation (15, 20). Notably, the co-occurrence of these parameters—even in patients with preserved left ventricular ejection fraction (LVEF ≥50%) and normal ventricular dimensions—may signal preclinical heart failure with preserved ejection fraction (HFpEF) (15, 21). While prior research linked SBD and CSR to diastolic impairment in confirmed heart failure populations (17), our novel contribution reveals that: (1) These cardiac anomalies exhibit robust associations with CSR in non-HF individuals and (2) patients with CSR during 2-year follow-up appear to be more likely to develop confirmed heart failure clinically. This work pioneers the hypothesis that CSR may serve as a dynamic biomarker of progressive diastolic stress, marking the transition from pre-HF to overt HF or HFpEF. Given the diagnostic challenges posed by the clinically silent nature of early HFpEF, CSR detection could provide critical phenotyping clues. The temporal sequence from CSR onset to HFpEF diagnosis requires validation through multicenter prospective studies incorporating continuous cardiopulmonary monitoring.

From a pathophysiological perspective, existing studies primarily attribute the pathogenesis of CSR to abnormalities in pulmonary circulation among cardiac patients (22, 23): insufficient pulmonary blood perfusion leads to ventilation/perfusion mismatch, which subsequently establishes a periodic breathing pattern through the chemoreceptor-respiratory center feedback loop. Additionally, atrial fibrillation and progressive left atrial structural remodeling (dilation and fibrosis) reduce cardiac output, inducing circulatory delay and enhanced chemosensitivity, ultimately triggering CSR (24). This study proposes a novel mechanism whereby diastolic dysfunction triggers CSR events: in individuals with essentially normal cardiac structure, postural changes during sleep may exacerbate the adverse effects of diastolic dysfunction on pulmonary circulation hemodynamics, thereby inducing CSR episodes. This discovery opens new avenues for early cardiac function assessment research.

Furthermore, we observed that patients with concomitant sleep apnea demonstrated elevated levels of BMI, hypertension prevalence, interventricular septum thickness, and left ventricular wall thickness compared to those without sleep apnea. Subsequent statistical analysis of our longitudinal data revealed that interventricular septum thickness and left ventricular wall thickness showed stronger correlations with BMI, hypertension prevalence, and age—a pattern aligning with previous findings regarding the association between hypertension and SBD (25), as well as its impact on cardiac structural remodeling (26). These findings substantiate the association between sleep apnea, hypertension, cardiac remodeling. Although our study detected only hypertension-associated cardiac structural alterations in SBD patients without CSR, the future progression warrants attention. Untreated OSA may perpetuate cardiac injury, while respiratory control instability—manifested as elevated loop gain following CPAP therapy—could precipitate new-onset CSA or even CSR (27). These potential developments demand clinical vigilance.

This study pioneers the investigation of patients exhibiting CSR events without confirmed HF diagnosis. Key findings demonstrated that individuals with CSR events presented with advanced age, elevated BNP/Nt-proBNP levels, and echocardiographic characterized by left atrial dimension enlargement and increased E/e' ratio. Multivariable regression models adjusting for potential confounders revealed independent associations between CSR occurrence and BNP/NT-proBNP levels, LADi, and E/e' ratio. Further analysis established significant positive linear correlations between CSR/TST% and NT-proBNP levels, LADi, and E/e'. These findings collectively suggest that CSR events may serve as dynamic biomarkers of progressive left ventricular diastolic dysfunction, potentially functioning as an early warning signal for the pathological transition from subclinical diastolic impairment to overt HF or HFpEF.

There were some limitations that should be considered when interpreting the results. First, although studies have demonstrated good inter-observer agreement for LAD, LADi, and LAVi measurements, LAVi has been shown to be more accurate than LADi in evaluating left atrial dimensions. To better monitor disease progression, LAVi assessment remains essential in subsequent studies. Besides, the retrospective and cross-sectional design of this study constrains our understanding of the association between SBD, CSR, age, AF, HF, and echocardiographic changes. A longitudinal study is warranted to follow up with enrolled patients for elucidating the specific mechanisms and temporal processes underlying the relationship between CSR and cardiac structural remodeling.