Summary of Evidence

The literature provides a robust, multi-layered case linking obstructive sleep apnoea (OSA) to left ventricular (LV) diastolic dysfunction. Evidence spans mechanistic animal and human studies, cross-sectional and cohort observational data, meta-analyses, a Mendelian randomisation study, and randomised controlled trials (RCTs) of CPAP. The association is dose-dependent, extends to preserved-EF heart failure (HFpEF), and is at least partially reversible with treatment.

1. Epidemiological & Observational Evidence

Prevalence Data

Multiple cross-sectional studies consistently demonstrate that LV diastolic dysfunction (LVDD) is significantly more prevalent in OSA patients than matched controls, even before systolic dysfunction appears:

Fung et al. (2002, Chest) — one of the earliest landmark studies — found severe OSA is independently associated with LVDD, establishing the association before obesity and hypertension were fully explored as confounders. DOI: 10.1378/CHEST.121.2.422
Makris et al. (2019) studied 42 patients with first-diagnosed severe OSA (AHI >30) and no cardiovascular comorbidities. Diastolic dysfunction was significantly more prevalent in OSA patients, with OSA severity independently predicting LVDD — suggesting OSA acts as a primary risk factor. DOI: 10.1177/1753466619880076
D'Andrea et al. (2020, BMC Pulmonary Medicine) examined 55 OSA patients with preserved LVEF using exercise stress echocardiography. LV global longitudinal strain was significantly reduced (−13.4 ± 3.8 vs −18.4 ± 3.3 in controls, p < 0.001) and E/E' ratios rose sharply on exertion (ΔE/E' +87.5% vs +25.4% in controls, p < 0.0001), demonstrating subclinical diastolic reserve impairment not visible at rest. DOI: 10.1186/s12890-020-1099-9
Lovic et al. (2014) prospectively enrolled 125 patients with newly diagnosed OSA and normal LVEF, demonstrating early systolic and diastolic dysfunction — suggesting these changes precede clinically apparent cardiac disease. DOI: 10.1155/2014/898746
Raut et al. (2020, Sleep Medicine) found that OSA severity correlates with higher grades of LVDD and rising high-sensitivity troponin I, consistent with a "vicious cycle" of recurrent myocardial injury driving progressive diastolic impairment. DOI: 10.1016/j.sleep.2020.10.024

Dose-Dependence and Hypoxia Indices

Maiolino et al. (2023, Journal of Hypertension) — in 162 patients from two Italian cohorts — showed that oxygen desaturation index (ODI) and AHI independently predicted E/A ratio at multivariate regression. The percentage of time with SaO₂ < 90% predicted LV end-diastolic volume. Nocturnal hypoxic burden is the most relevant metric, not just apnoea frequency. DOI: 10.1097/01.hjh.0000941928.48403.24
Zhang & Zhu (2025, Medicine) used AutoStrain LV speckle-tracking in 126 patients and found E/A and E/E' differed significantly between OSAS subgroups and controls; LVGLS correlated most strongly with AHI (r = −0.732, p < 0.001). DOI: 10.1097/MD.0000000000042309
Wang et al. (2014) found AHI during REM sleep ≥32.3/h was the only independent predictor of LVDD (AUC 0.647) after adjusting for age, sex, hypertension, and BMI. DOI: 10.5664/jcsm.3360

Paediatric Data

Ahmadu et al. (2024) found LVDD present in 6.7% of children with OSA vs 0% of controls (p = 0.002), significantly more common in those with abnormal LV geometry. DOI: 10.4103/heartviews.heartviews_58_23

2. Pathophysiological Mechanisms

Four major pathways converge on diastolic impairment:

A. Intrathoracic Pressure Swings

During obstructive apnoeas, efforts against a closed airway generate large negative intrathoracic pressures (−30 to −80 cmH₂O):

Increase LV transmural pressure and afterload
Cause paradoxical septal wall motion
Reduce LV stroke volume per beat
Promote LV wall stress → hypertrophy → impaired relaxation

Source: Javaheri et al. (2017, JACC) DOI: 10.1016/j.jacc.2016.11.069; Javaheri et al. (2024, JACC) DOI: 10.1016/j.jacc.2024.02.059

B. Sympathetic Nervous System Activation

Recurrent hypoxaemia and arousal trigger reflex sympathetic discharge, sustained even during waking hours:

Elevated systemic vascular resistance and afterload
Increased resting heart rate → reduced diastolic filling time
LV hypertrophy via chronic pressure overload
Direct catecholamine-mediated cardiomyocyte dysfunction and apoptosis

Source: Liu et al. (2023) DOI: 10.31083/j.rcm2412342

C. Intermittent Hypoxia / Reoxygenation

Cycles of intermittent hypoxia (IH) activate HIF-1α, NF-κB, and reactive oxygen species (ROS):

Upregulates pro-inflammatory cytokines: TNF-α, IL-6, IL-8
Promotes myocardial fibrosis via TGF-β and NHE-1 activation
Impairs cardiomyocyte ATP production (mitochondrial dysfunction)
Activates thrombospondin-1 and extracellular matrix remodelling

Wei et al. (2016) demonstrated in an OSA rat model that chronic IH causes cardiac inflammation (↑HIF-1α, NF-κB, IL-6, MMP-2), cardiomyocyte apoptosis, and fibrosis. DOI: 10.7555/JBR.30.20160110

Li et al. (2018) found serum Pentraxin-3 (PTX-3) was independently correlated with LVDD (E/Em ratio and LAVI) in OSA patients. DOI: 10.23937/2378-3516/1410097

D. RAAS Activation and Myocardial Fibrosis

OSA activates the renin-angiotensin-aldosterone system, elevating aldosterone levels:

Myocardial collagen deposition and stiffening
Endothelial dysfunction
Sodium retention → increased preload and pulmonary congestion

Mineralocorticoid receptor antagonists (spironolactone, finerenone) normalise BP and improve cardiac function in animal IH models. Badran et al. (2025) DOI: 10.1093/sleep/zsaf090.0405

3. Meta-Analyses and Systematic Reviews

LV Structure and Function Meta-Analysis

Yu et al. (2019, Herz) — 17 studies, 747 OSA patients, 426 controls. DOI: 10.1007/s00059-019-04850-w

Parameter	WMD (95% CI)	p-value
LV end-diastolic diameter	+1.24 mm (0.68–1.80)	<0.001
LV end-systolic diameter	+1.14 mm (0.47–1.81)	0.001
LV mass	+35.34 g (20.67–50.00)	<0.001
LVEF	−3.01% (−1.90 to −0.79)	0.001
LA diameter	+2.13 mm (1.48–2.77)	<0.001
LA volume index	+3.96 ml/m² (3.32–4.61)	<0.001

CPAP and Diastolic Function Meta-Analysis

Al-Sadawi et al. (2022, Respiration) — 9 RCTs, 833 participants. DOI: 10.1159/000519406

Parameter	Pooled Effect	p-value
Deceleration time	−39.49 ms (−57.24 to −21.74)	<0.001
Isovolumic relaxation time	−9.32 ms (−17.08 to −1.57)	0.02
E/e' ratio	−1.38 (−2.60 to −0.16)	0.03

CPAP and Myocardial Strain Meta-Analysis

Castro et al. (2025, Journal of Hypertension) — 10 studies, 385 patients. DOI: 10.1097/HJH.0000000000004171

Parameter	Effect	p-value
LV-GLS	−1.92% (−2.63 to −1.21)	<0.01
RV-GLS	−1.88% (−2.77 to −0.99)	<0.01
PASP	−5.23 mmHg (−8.54 to −1.92)	0.002
E/e' ratio	−0.95 (−2.42 to 0.53)	0.21 (NS)

4. OSA as a Gateway to HFpEF

Abdullah et al. (2018, Am J Cardiol): OSA present in 16.8% of HFpEF patients. After propensity-score matching, OSA was associated with a 2.2-fold increased risk of HFpEF hospitalisation (RR 2.2, 95% CI 2.12–2.21). DOI: 10.1016/j.amjcard.2018.04.052
Stewart et al. (2023, Sleep Advances): In a specialist HFpEF clinic, 100% of 28 consecutive patients had OSA on polysomnography, 57% with moderate-severe disease. Only 21% were symptomatic by Epworth Sleepiness Scale. DOI: 10.1093/sleepadvances/zpad035.038
Giannoni et al. (2019, Front Cardiovasc Med): Obstructive apnoeas predominate in HFpEF (53% at night). HFpEF patients with OSA had higher LA volumes and more severe diastolic dysfunction. DOI: 10.3389/fcvm.2019.00125
Connolly et al. (2023, AHA Circulation): OSA severity correlated with LV mass index (p=0.04), LA volume index (p=0.03), and LV relative wall thickness (p=0.008). High H₂FPEF risk scores had >17-fold greater odds of prevalent OSA. DOI: 10.1161/circ.148.suppl_1.14281

5. Causal Evidence: Mendelian Randomisation

Yang et al. (2023, Frontiers in Genetics) — Bidirectional two-sample MR using GWAS data (16,761 OSA cases). DOI: 10.3389/fgene.2023.1266869

Genetically predicted OSA → LV end-diastolic volume: β = 0.114 (95% CI 0.034–0.194), p = 0.006
Genetically predicted OSA → LV stroke volume: β = 0.111 (95% CI 0.031–0.191), p = 0.007
Reverse direction: genetically predicted ↓LVEF → ↑OSA risk (OR 1.161, p = 0.015)

This bidirectional relationship supports OSA as both a cause and consequence of LV dysfunction, with a potential vicious cycle.

6. Treatment Response and Reversibility

Tamulėnaitė et al. (2023): 3 months CPAP significantly reduced prevalence of LVDD (p = 0.023) and improved LA reservoir strain. DOI: 10.1093/ehjci/jead119.285
Januskevicius et al. (2022, Medicina): 3 months CPAP improved LV-GLS and LA reservoir function; significant reductions in galectin-3 and sST2. DOI: 10.3390/medicina58111511
Kuchmin et al. (2023): In severe OSA, 3 months CPAP improved E/A ratio across both ventricles and reduced PASP. DOI: 10.26442/20751753.2023.1.202163
Al-Sadawi et al. (2022) meta-analysis: IVRT (−9.32 ms), deceleration time (−39.49 ms), and E/e' (−1.38) all improved with PAP treatment across 9 RCTs. DOI: 10.1159/000519406
Yakovlev et al. (2024): In 207 men with HFpEF + OSAS, CPAP reduced hospitalisations by 16% at 12 months. DOI: 10.26442/00403660.2024.01.202563

7. Sex Differences

Maier et al. (2021, Front Med): HFpEF and diastolic dysfunction (E/e') were significantly more frequent in women with SDB. SDB severity correlated with diastolic dysfunction in women but not men — suggesting women with OSA may be underdiagnosed while bearing a disproportionate diastolic burden. DOI: 10.3389/fmed.2021.675987

8. Key Echocardiographic Markers Across Studies

Parameter	Direction in OSA	Notes
E/A ratio	↓	Most consistent finding; correlates with AHI and ODI
E/e' ratio	↑	Reflects elevated filling pressures; improves with CPAP
IVRT	↑	Impaired LV relaxation; reverses with CPAP
Deceleration time	↑	Reverses with CPAP
LA volume index	↑	Structural marker; slower to reverse
LV mass index	↑	Predicts diastolic impairment risk
LV-GLS	↓	Sensitive subclinical marker; improves with CPAP
NT-proBNP	↑	Elevated in OSA with LVDD (PPV 70%, NPV 97.3%)

9. Critical Appraisal and Limitations

Confounding: Obesity, hypertension, and T2DM are shared risk factors for both OSA and LV dysfunction. Residual confounding remains a concern in observational studies despite multivariate adjustment and MR design.
Heterogeneity in OSA diagnosis: Studies use varying AHI thresholds (3% vs 4% desaturation), home testing vs. polysomnography, and different severity stratifications.
Echocardiographic standardisation: Diastolic grading has evolved (ASE/EACVI 2016 guidelines); older studies used E/A and IVRT without E/e' or strain analysis, limiting comparability.
CPAP RCT limitations: Many trials suffer from poor adherence (<4 hrs/night), small samples, and short follow-up, likely underestimating treatment benefit.
Directionality gap: The Mendelian randomisation study confirms OSA causes LV structural changes, but specific diastolic filling parameters (E/A, E/e') were not phenotyped in the GWAS used.

Conclusions

The evidence robustly supports a causal, dose-dependent, and partially reversible relationship between OSA and LV diastolic dysfunction. The link is mechanistically coherent (intrathoracic pressure, sympathetic activation, intermittent hypoxia, RAAS/fibrosis), epidemiologically consistent across age groups and both sexes, and supported at the causal level by Mendelian randomisation.

Diastolic dysfunction appears to be an intermediate phenotype on the pathway from OSA to HFpEF, with nocturnal hypoxic burden (ODI, SpO₂ <90% time) being more predictive than AHI alone. CPAP treatment produces measurable improvements in functional diastolic parameters (IVRT, DT, E/e') within 3–6 months, with structural markers (LA volume, LV mass) lagging behind.

Clinical implications: There is a strong case for routine echocardiographic assessment of diastolic function in moderate-severe OSA patients, and for OSA screening in all patients presenting with HFpEF or unexplained diastolic dysfunction.

OSA and Diastolic Dysfunction: A Deep Literature Review