Cardiovascular effects of obstructive sleep apnoea and effects of continuous positive airway pressure therapy: evidence from different study models

Abstract

Background

Cardiovascular consequences of obstructive sleep apnoea (OSA) and the effects of continuous positive airway pressure (CPAP) therapy on blood pressure, endothelial dysfunction and major adverse cardiovascular events (MACE) have been studied over decades using different study designs. However, clinical findings from different study models on cardiovascular outcomes are sometimes contradictory.

Methods

A literature search was conducted in PubMed for randomised controlled trials, meta-analyses, population-based epidemiological studies and OSA cohort studies up to September 2023 investigating the cardiovascular effects of OSA and CPAP in adults with OSA on the following cardiovascular end-points: blood pressure, arterial hypertension, endothelial function and MACE (myocardial infarction, stroke, transient ischaemic attack or cardiovascular death). The level of evidence for these outcomes was discussed on the basis of different study models.

Results and conclusions

There is high-level evidence of a causal relationship between OSA and arterial hypertension and endothelial dysfunction, as well as on higher MACE incidence among subgroups of patients with untreated OSA. The cardiovascular effects of OSA depend on the severity of OSA, symptoms, phenotype and comorbidities. The blood pressure-lowering effect of CPAP is mainly observed in uncontrolled and treatment-resistant hypertension. The MACE risk reduction in OSA depends on good long-term CPAP adherence. Younger, sleepy patients with more severe OSA, higher hypoxaemic burden and without overt cardiovascular end-organ disease may particularly benefit from CPAP treatment in terms of cardiovascular risk reduction. Randomised controlled trials of CPAP or other effective OSA treatments in primary cardiovascular prevention and in patients at highest risk are lacking.

Shareable abstract

There is solid evidence on the causal relationship between OSA and hypertension, endothelial dysfunction and MACE. Although influenced by OSA phenotypes, CPAP improves blood pressure control and good long-term CPAP adherence reduces the incidence of MACE. https://bit.ly/48lbea8

Lessons for clinicians

• In order to advise on the effect of continuous positive airway pressure (CPAP) on cardiovascular risk, a comprehensive assessment encompassing disease severity, symptom burden, patient characteristics, comorbidities, blood pressure control (ideally using 24-h ambulatory blood pressure monitoring and considering nocturnal blood pressure) is required.

• Strategies to improve CPAP adherence or to provide effective and well-tolerated non-CPAP treatments for obstructive sleep apnoea are important to improve cardiovascular outcomes, particularly in specific subgroups at risk such as sleepy, non-elderly patients with more hypoxaemic burden and poor blood pressure control.

Introduction

Obstructive sleep apnoea (OSA) is highly prevalent in the general population [1–3]. Its hallmark is a repetitive upper airway collapse during sleep despite continued inspiratory effort, resulting in complete airflow cessation (apnoea) or airflow reduction (hypopnoea) accompanied by intrathoracic pressure oscillations and arousals from sleep as well as intermittent oxygen desaturations and thus hypoxaemia and eventually hypercapnia [3, 4]. These pathophysiological consequences can lead to autonomic imbalance and overactivity of the sympathetic nervous system, an increase in blood pressure, and ultimately to endothelial dysfunction [4].

The most effective treatment for OSA is continuous positive airway pressure (CPAP), which prevents pharyngeal collapse. Counteracting the primary pathophysiological source of the cardiovascular consequences of OSA, the role of CPAP in cardiovascular risk reduction has been investigated in different patient cohorts using different study designs (e.g. randomised controlled trials (RCTs), meta-analyses, population-based observational studies and OSA cohort studies) [5–19].

Since the first description of OSA in the 1960s [20] and further insights into the pathophysiology of upper airway collapse and the haemodynamic effects of OSA in the 1980s [21], as well as the introduction of CPAP as an effective treatment [22], the cardiovascular consequences of OSA and the effects of treatment on these have been the focus of research. The first epidemiological studies in the late 1980s and 1990s on the cardiovascular effects of OSA, and in particular its association with arterial hypertension, were followed in the late 1990s by RCTs on CPAP effects, particularly on systemic blood pressure [23, 24]. We now also have RCTs on the CPAP effect on hard cardiovascular end-points, so-called major adverse cardiovascular events (MACE) [7, 25–28], although the results cannot be generalised due to the population included [29]. The results of epidemiological observational studies and OSA cohort studies are also taken into account in order to avoid the natural limitations of such RCTs with a long follow-up period [23, 30–34], e.g. the inclusion of mostly less symptomatic patients who then also make poor use of CPAP therapy. In addition, study-level [17, 35] and individual patient data [36–38] meta-analyses are being used to identify predictors of a benefit of OSA treatment on blood pressure control or MACE, as OSA is a heterogeneous disease in terms of pathophysiology, symptoms, comorbidities and long-term outcomes. Nowadays, the focus is on phenotyping, as it has also been recognised that long-term outcome differs between different OSA phenotypes.

Although there is an extensive literature on this topic spanning over three decades, there are sometimes conflicting results arising from different study models, particularly on the effects of CPAP on cardiovascular outcomes, which cannot be explained solely by the heterogeneity of the population studied. Clinical research questions may be addressed by different study designs and each design shows specific advantages and limitations. The level of evidence for the role of OSA and its treatment on different cardiovascular outcomes and patient populations varies, and some results are difficult to extrapolate to a general OSA population [29, 39, 40].

In this narrative review of human studies, we therefore address the state of knowledge on the cardiovascular consequences of OSA, focusing on the main cardiovascular outcomes (blood pressure control/arterial hypertension, endothelial dysfunction and MACE (defined as myocardial infarction, stroke, transient ischaemic attack or cardiovascular death)), as well as the effects of CPAP therapy on these and the impact of confounders.

The aim is to provide a structured overview of the knowledge and evidence on OSA-related cardiovascular outcomes and the effects of CPAP therapy from different study models (prospective population-based, prospective OSA cohort-based, case–control, uncontrolled interventional trial, RCT, and study-level and individual patient-level data meta-analyses of RCTs) and to identify the knowledge gaps on which further research should focus.

A structured review of this kind, which looks at the evidence in a differentiated and systematic way for 1) the most important cardiovascular risk factor hypertension, 2) for the intermediate cardiovascular outcome endothelial dysfunction as a precursor to arteriosclerosis and 3) for hard cardiovascular end-points, and on the basis of the various study designs together with their limitations, does not yet exist and should contribute to understanding the gaps in this field of research.

Methods

A literature research was conducted on PubMed from 1956 to September 2023 using the search terms ((obstructive sleep[Title/Abstract] OR sleep-disord*[Title/Abstract])) AND (blood press*[Title/Abstract] OR hyperten*[Title/Abstract] OR cardiovasc*[Title/Abstract] OR endoth*[Title/Abstract])) AND (CPAP[Title/Abstract] OR continuous positive[Title/Abstract])). We considered all RCTs and meta-analyses (including RCTs only) in adults written in English and reporting data on the effect of CPAP on at least one of the following cardiovascular outcomes: blood pressure/arterial hypertension, endothelial dysfunction and MACE, defined as myocardial infarction, stroke, transient ischaemic attack, cardiovascular death, as well as mortality. For study designs other than RCTs and meta-analyses, inclusion was based on methodological solidity, follow-up and sample size in relation to the incidence of outcomes. For each study design (prospective population-based epidemiological study, prospective OSA cohort study, case–control study, uncontrolled interventional trial, RCT, and study-level and individual patient-level data meta-analyses of RCTs), we discuss the evidence for the aforementioned cardiovascular outcomes (blood pressure/arterial hypertension, endothelial function and MACE), the generalisability of the results, and the strengths and limitations of the study design. Given the narrative nature of this study, the results and discussion sections were combined to allow for a better understanding, followed by a summary and conclusions section. We followed the Scale for the Assessment of Narrative Review Articles (SANRA) for structuring this narrative review [41].

Results and discussion

Prospective population-based observational studies

Cross-sectional and longitudinal analyses from prospective population-based epidemiological studies are a valid approach to assess associations of OSA and its severity grades with cardiovascular outcomes in a real-life population. Cross-sectional studies are mainly used to study prevalence but do not allow to establish an association between cause and effect. The most relevant strengths of this approach are the possibility to include large samples to be followed up for a long time compared with RCTs, for example, and to account for many potential confounders. However, they do not allow a direct causal relationship to be established between the disease and the outcome, but rather the generation of hypotheses. Such epidemiological studies often form the datasets for later analyses that were not planned at the time the study was designed [42, 43]. The main findings of the most relevant population-based observational studies in the field of OSA and cardiovascular effects are discussed [1, 31, 44, 45].

A dose–response relationship between the incidence of hypertension and OSA severity according to the apnoea–hypopnoea index (AHI) has been found in the Wisconsin Sleep Cohort Study, a prospective study including 709 participants. The adjusted odds ratio for the presence of incident hypertension at the 4-year follow-up was 2.9 when patients had moderate-to-severe OSA (AHI >15 events·h⁻¹) [46]. A consistent finding on the dose–response relationship was found in the Sleep Heart Health Study, a prospective population-based study including 6132 participants aged >40 years. The relative risk for hypertension in severe OSA (AHI >30 events·h⁻¹) compared with absent OSA was 1.4 (95% CI 1.0–1.8) in a cross-sectional analysis and the odds ratio of hypertension increased with the increase in AHI [23]. However, a dose–response relationship between OSA severity by AHI and between prevalence or incidence of hypertension has not been consistently shown and is mostly absent for other cardiovascular outcomes. However, in severe OSA defined by AHI ≥30 events·h⁻¹, there is a stronger association with many vascular end-points than with AHI <30 events·h⁻¹, e.g. for stroke risk [47].

In the Wisconsin Sleep Cohort Study, patients with moderate-to-severe OSA, when compared with patients without OSA, showed a higher risk of developing a non-dipping nocturnal blood pressure profile [48, 49], which has been associated with adverse cardiovascular outcomes [50]. Moreover, patients with severe OSA showed a significantly higher cardiovascular mortality controlling for age, sex and body mass index [45]. In the Sleep Heart Health Study, men in the highest AHI quartile (AHI ≥20 events·h⁻¹) had an adjusted hazard ratio of 2.9 (95% CI 1.1–7.4) [51]. In a prospective observational in women free of cardiovascular disease (AHI <10 events·h⁻¹: control group (n=258); AHI >10 events·h⁻¹: untreated OSA (n=268); AHI >10 events·h⁻¹ and CPAP: CPAP-treated OSA (n=441)), untreated OSA was associated with an increased incidence of MACE, particularly stroke, and showed a stronger association with incident stroke than with coronary heart disease [52].

In a 20-year follow-up analysis of the Penn State Adult Cohort, the association of mild-to-moderate OSA with all-cause mortality and the modifying effect of age and presence of hypertension or cardiovascular disease was studied [33]. All-cause mortality was increased in patients aged <60 years with mild-to-moderate OSA but not in older patients. The presence of hypertension or cardiovascular disease was mainly relevant for an increased risk of all-cause mortality in younger patients [33].

A prospective observational study by Marin et al. [31] including 1651 subjects (healthy controls, treated and untreated OSA) showed that men with untreated severe OSA had a significantly increased risk of fatal and non-fatal cardiovascular events. The association of OSA symptom subtypes (disturbed sleep, minimally symptomatic, excessively sleepy and moderately sleepy) and incident and prevalent cardiovascular disease was studied in more than 1200 patients with moderate-to-severe OSA in the Sleep Heart Health Study [53]. The excessively sleepy had the highest risk for incident cardiovascular disease [53].

The findings from the prospective observational study by Marin et al. [31] suggest that CPAP treatment lowers this risk [31]. These findings were corroborated by having matched patients with OSA with healthy controls for their age and body mass index. In the prospective observational study in women, CPAP reduced the risk of vascular events [52].

In summary, population-based epidemiological observational studies have shown a dose–response relationship between OSA severity and incident hypertension (but not with other vascular outcomes), an association between severe untreated OSA and cardiovascular events, particularly in younger and in sleepy patients, and suggested a risk reduction for cardiovascular events by treating severe OSA with CPAP. No data on endothelial dysfunction are available.

This type of study design has some inherent limitations. Although large sample sizes allow for consideration of many potential confounders and ultimately a high incidence of the outcome studied during a long follow-up period, prespecified control of biases that might be observed during data collection is not practically feasible. Similarly, an a priori power calculation or sample size estimation is difficult. In some studies, the analysis encompasses sleep disordered breathing in general and not specifically OSA only. Another relevant limitation is the limited data on female patient cohorts, which currently represents an important knowledge gap in terms of gender medicine.

Prospective OSA cohort studies

Prospective OSA cohorts also allow the investigation of incidence and prognosis and the relationship between OSA and cardiovascular outcomes, and in particular the impact of treatment on these outcomes, in a representative OSA population within a clinical setting, although less selected than in RCTs. These studies have provided valuable information about potential predictors of increased cardiovascular risk, as well as the benefits of CPAP therapy in terms of cardiovascular risk reduction.

Data from the Pays de la Loire Sleep Cohort, a large prospective OSA cohort study, were analysed in combination with health administrative data on MACE (11% in 5358 patients during a median follow-up of 78 months) in patients without known cardiovascular disease at the time of diagnosis of OSA, to evaluate the association of symptom subtypes and hypoxic burden with MACE [32]. Hypoxic burden was defined as the total area under the oxygen desaturation curve associated with the respiratory event in units of %·min⁻¹ [54]. While there was no association between the symptom subtypes and MACE, hypoxic burden and sleep time with oxygen saturation below 90% (t<90) were independently associated with MACE, and this association was stronger in younger and in female patients [32]. In a cohort study investigating the prognostic significance of various polysomnography indices and CPAP therapy in different OSA groups for cardiovascular events (MACE), 278 MACE were documented in more than 1500 OSA patients during a median follow-up period of 8.3 years [34]. AHI was not associated with MACE, but t<90 and nocturnal heart rate were.

Data on CPAP effects are also available from OSA cohort studies. In the Pays de la Loire Sleep Cohort, the relationship between hours of CPAP use and MACE was also investigated. A 25% reduction in MACE was observed with CPAP use >6 h per night compared with the reference group with CPAP use <4 h per night. This association was stronger in excessively sleepy patients, in males and in patients without overt cardiovascular disease at diagnosis of OSA [55]. Similarly, a cohort study of women with OSA showed a risk reduction of MACE under CPAP therapy with an adherence >4 h per night, although not specifying the influence of menopause on the results [56]. In moderate and severe OSA patients who had an indication for CPAP (n=1108), CPAP prescription was not associated with a reduction in MACE [34]. However, a cluster analysis identified a subgroup (n=333) who were younger, more obese and had more severe OSA (higher AHI and t<90) in whom CPAP therapy reduced the risk of MACE (hazard ratio 0.49, 95% CI 0.25–0.95) [34].

In summary, OSA cohort studies have shown an association between higher levels of nocturnal hypoxaemia (OSA-specific nocturnal hypoxaemic burden and t<90) and adverse cardiovascular outcome, and highlighted the importance of good CPAP adherence in reducing MACE. They also indicated that the benefit of CPAP in reducing MACE is higher in primary cardiovascular prevention, younger patients and more severe OSA. There are no conclusive data on arterial hypertension and endothelial dysfunction from OSA cohort studies.

Cohort studies also have several potential sources of bias. A link between the allocation of the intervention and the studied outcome introduces a major bias in an observational study. An important bias in both population-based epidemiological and OSA cohort studies is the “healthy adherer” effect that is difficult to measure [57]. Favourable outcomes due to patient behaviour in “healthy adherers” may be falsely attributed to CPAP therapy. However, cohort studies help to identify vulnerable patient groups and predictors of favourable treatment outcomes.

Case–control studies

Case–control studies compare groups retrospectively and allow identification of outcome predictors. They compare two matched groups, in our case patients with and without OSA, and thus partially control for some conditions prior to conducting the analysis. Two case–control studies that included patients with OSA of different severity showed an independent association between OSA and arterial hypertension. This association was further strengthened when analysing only male patients, with an increasing odds ratio across AHI tertiles. However, the majority of women included were post-menopausal, so that no reliable gender comparison was possible [58, 59].

In addition, a nested case–control analysis showed an independent association between arterial stiffness and OSA in a dose–response relationship with the AHI, regardless of whether or not concomitant diabetes or cardiovascular comorbidities were present [60]. No conclusive data on MACE and CPAP effects on the assessed cardiovascular outcomes are available.

The main limitation of case–control studies is selection bias, which can be reduced with a nested approach. One of the advantages of case–control studies, as of other observational studies, compared with RCTs is the possibility to study a large number of patients with limited costs, thus enabling a broader extension of the findings to the general OSA population [29].

Uncontrolled interventional trials

Uncontrolled interventional trials are single-arm studies in which the effect of the intervention (e.g. CPAP) is quantified before and after its administration in a group of OSA patients. Due to the lack of a control group, the level of evidence of such studies is low, as they cannot prove the causal nature of the intervention and the observed change. Nonetheless, this study design can be used to identify a trend of intervention effect in a specific population, which then needs to be supported by more robust evidence, e.g. an RCT. Due to the large number of available uncontrolled CPAP intervention studies and the low level of evidence for this approach, we do not discuss the results of such individual studies.

Randomised controlled interventional trials

RCTs are prospective controlled interventional studies in which the effect of an intervention is measured and quantified by comparison with a control group, whereby the influence of bias is minimised by randomisation and blinding. These studies enable the evaluation of cause–effect relationships [61].

There are more than 80 RCTs reporting the effect of CPAP on different measures of blood pressure in heterogeneous populations with OSA. These RCTs sometimes describe different effects of CPAP on blood pressure, which may depend primarily on the initial blood pressure control and the different OSA populations studied [17]. In many RCTs, CPAP use was insufficient, which leads to an underestimation of the treatment effect. For example, a CPAP withdrawal study with patients who had excellent adherence to therapy until the CPAP break showed a significantly greater increase in blood pressure on re-activation of OSA than the blood pressure-lowering effect of CPAP observed in many conventional CPAP RCTs [10, 17]. One of the earlier important RCTs found a mean blood pressure-lowering effect of CPAP of 2.5 mmHg in 24-h mean ambulatory blood pressure measurement (ABPM) in middle-aged obese patients with moderate-to-severe OSA [15]. Arterial hypertension was not an eligibility criteria. However, the treatment effect was larger in patients on antihypertensive drugs and with severe OSA [15]. A multicentre RCT in patients with moderate-to-severe OSA with arterial hypertension, but controlled blood pressure at baseline, found a similar treatment effect of CPAP on blood pressure [62]. A multicentre RCT by the Spanish Sleep Network in patients with OSA and resistant hypertension found a slightly larger blood pressure-lowering effect of CPAP on 24-h-mean and diastolic blood pressure and an increase in the proportion of patients with nocturnal dipping pattern [63]. Other RCTs in patients with OSA and treatment-resistant hypertension have shown larger blood pressure-lowering effects of CPAP on 24-h blood pressure [18]. In the RCTs investigating the effect of CPAP therapy on blood pressure, it is important to note the characteristics of the patient populations that were included (e.g. sleepy OSA versus minimally symptomatic OSA; quality of initial blood pressure control; comorbidities; age) and which blood pressure parameters were used as outcomes (office blood pressure versus 24-h ABPM; 24-h versus daytime blood pressure versus nocturnal blood pressure), as well as to specify CPAP adherence, to allow for a weighted data interpretation. Office blood pressure measurements are less reliable as an outcome parameter than 24-h blood pressure, and the blood pressure-lowering effect of CPAP seems to be more pronounced on nocturnal blood pressure [17]. Limitations of many of these RCTs are the inclusion of patients with and without arterial hypertension and the lack of detailed information on antihypertensive treatment during the study (e.g. antihypertensive medication dosage, compliance and change in antihypertensive therapy during the study).

Several RCTs have shown a significant effect of CPAP on endothelial function improvement compared with the control group [64–66], even in minimally symptomatic patients with OSA [67], or a worsening of endothelial function in response to CPAP withdrawal and, thus, recurrence of OSA [68].

RCTs on the effect of CPAP on MACE are difficult to conduct because a large patient population and a long follow-up period are required to achieve a sufficiently high incidence of MACE. This leads to a selection of less symptomatic patients with OSA, as it would be unethical to assign symptomatic patients to a control group for years. Sleepiness in patients with OSA seems to be associated with worse cardiovascular outcome [53] and the patients most likely to benefit from treatment were frequently excluded from these RCTs. With the exception of one RCT on the incidence of hypertension and MACE (but not powered for MACE) in non-sleepy OSA [25], the available RCTs investigate secondary prevention effects of CPAP on the recurrence of cardiovascular or cerebrovascular events or mortality in mostly non-sleepy OSA [7, 26–28]. Intention-to-treat analyses have shown no statistically significant effect of CPAP on the combined end-point of MACE, but treatment adherence to CPAP was modest or low [40]. This observation may be explained by selection bias, i.e. the inclusion of patients who had not presented with OSA symptoms. Subanalyses of an RCT with only CPAP-adherent patients were underpowered for the assessment of cardiovascular events [40]. In the SAVE trial, CPAP use decreased over time and was limited with 3.3±2.3 h during follow-up. Blood pressure was already well controlled under antihypertensive medication at the beginning and remained so during the follow-up, even if it decreased slightly on CPAP during follow-up (∼2 mmHg) [69].

RCTs are the best approach to study the intended effect of CPAP on blood pressure, endothelial function and, in theory, MACE. Several sources of bias and confounding can best be avoided by conducting an RCT. In evidence-based medicine, RCTs are considered the highest level of evidence after meta-analyses of RCTs, followed by cohort studies and case–control studies. An important limitation of most RCTs on MACE is statistical power, which has been deemed as insufficient to properly assess most of the individual cardiovascular outcomes (most had a combined end-point, and cerebrovascular and cardiovascular outcomes may differ). A propensity score-matched analysis of the SAVE trial indicated beneficial effects of CPAP on stroke but not cardiac events, and it was hypothesised that combined end-points might dilute the treatment effect on cerebrovascular outcomes [19, 26]. In addition, the findings cannot be extended to the general OSA population due to too selective inclusion and exclusion criteria [29]. RCTs on CPAP effects on MACE are intrinsically difficult to carry out in view of different factors: ethical issues in withholding an effective treatment to symptomatic patients, selection bias due to including a highly selected patient population and problems achieving good CPAP adherence in the intervention arm. Thus, the results of observational studies should also be considered.

To summarise, the RCTs investigating the effect of CPAP therapy on MACE in patients with OSA examined cardiovascular and cerebrovascular secondary prevention in non-sleepy patients with very low CPAP adherence. In the intention-to-treat analysis, no benefit of CPAP for the secondary prevention of MACE was found due to low adherence to CPAP in this selected, less symptomatic OSA population. However, this result cannot be generalised. RCTs on MACE in patients expected to benefit more from CPAP (e.g. younger, sleepy patients with uncontrolled hypertension and lack of vascular end-organ damage) are lacking.

Study-level meta-analyses of randomised controlled interventional trials

The literature search identified more than 20 study-level meta-analyses of RCTs looking at the effect of CPAP on cardiovascular end-points in OSA [13, 17, 35, 70–87]. The number of included RCTs ranged between four and 68. The mean study population size was 1937 patients. Meta-analyses have shown that CPAP therapy has a mean blood pressure-lowering effect of ∼2 mmHg in a heterogeneous OSA population, with and without hypertension [17, 70, 71, 77, 79, 81–87]. The pooled treatment effect was larger in subgroups such as severe OSA or difficult-to-control or resistant hypertension [17, 82, 83]. In addition to providing a more precise estimate of the treatment effect, meta-analyses pool studies with different population characteristics and are therefore also suitable for identifying predictors of the intervention effect by applying meta-regressions. For example, the largest meta-analysis to date on the effect of CPAP therapy on blood pressure, including over 8000 patients with OSA, showed that CPAP therapy has a greater blood pressure-lowering effect in those with uncontrolled baseline blood pressure, age <60 years and more severe nocturnal oxygen desaturations [17]. Baseline blood pressure appeared to be the most important predictor of blood pressure response to CPAP.

A small meta-analysis of RCTs found a significant, clinically relevant improvement in endothelial function as measured by flow-mediated dilatation [13].

There were no conclusive data on the preventive effect of CPAP on MACE in patients with OSA from study-level meta-analyses [35, 75, 84, 86]. Because study-level meta-analyses have the same limitations as the included RCTs, it was concluded that the highly selected population of the included RCTs and low CPAP adherence hindered the generalisability of the results [35]. Sensitivity analyses indicated that increased CPAP usage might lead to a risk reduction for MACE [84].

Even if the meta-analyses from RCTs offer the highest level of evidence, knowledge gaps in this area remain. Certain patient groups such as patients with severe OSA or sleepy patients with OSA could not be specifically studied. Furthermore, there is little data on sex differences, as no meta-analyses on female patients with OSA have been published so far.

Individual patient-level data meta-analyses of randomised controlled interventional trials

Individual patient data meta-analyses, which pool individual patient data rather than summary statistics on a study level, are the gold standard for identifying predictors of treatment response and are therefore helpful in defining study groups of interest for future RCTs. The literature search identified three individual patient data meta-analyses of RCTs on blood pressure or MACE [36–38]. An individual patient data meta-analysis including almost 1000 patients from CPAP RCTs in OSA indicated that the presence of uncontrolled hypertension was associated with a greater blood pressure-lowering effect of CPAP, independently of OSA severity [37]. Another meta-analysis based on individual patient data of four RCTs with minimally symptomatic OSA patients indicated a trend towards an increase in systolic blood pressure in those patients using CPAP <4 h per night [36].

An individual patient data meta-analysis based on three RCTs on the effect of CPAP in non-sleepy patients with OSA on cardiovascular events and mortality in secondary prevention examined the risk of recurrent major adverse cardiovascular and cerebrovascular events (MACCE) [38]. A total of 691 (16.5%) MACCE were observed in 4186 patients during a median follow-up of 3.3 years. The incidence of MACCE was comparable in the CPAP and control groups, but the on-treatment analysis revealed a lower risk of MACCE in CPAP-adherent patients (38.5% in the CPAP group), defined as ≥4 h per day, with a hazard ratio of 0.66 (95% CI 0.52–0.85) [38]. Patients with pre-existing cardiovascular disease and OSA and good CPAP adherence had a significantly lower risk of MACCE than patients who used CPAP <4 h per day. This led to the conclusion that treatment adherence is a key factor for cardiovascular secondary prevention in patients with OSA.

Summary and conclusions

The assessment of the relationship between OSA and major cardiovascular events, as well as the impact of CPAP on them, is a multifaceted picture with various interrelated factors. This intrinsic complexity requires analysing the same question using different study models. Different study designs offer advantages that need to be weighed against bias and specific pitfalls. An overview of the strengths and limitations of each study design is provided in table 1. There is convincing evidence confirming a causal relationship between OSA and arterial hypertension. CPAP treatment has been showed to improve blood pressure control, especially in patients with uncontrolled blood pressure at baseline, difficult-to-treat hypertension or resistant hypertension. However, OSA is a heterogeneous disease in terms of pathophysiological features, symptoms and cardiovascular effects, and all these aspects affect, with a different weight in the specific patient, the therapeutic benefit that CPAP treatment can offer. Therefore, a comprehensive assessment of OSA phenotypes, beyond the sleep study alone, is required when assessing therapeutic options for patients with OSA. Comorbidities and blood pressure control must be taken into account in order to advise patients on the potential benefits of OSA therapy in terms of cardiovascular risk. Similarly, a CPAP-mediated improvement of endothelial dysfunction, another relevant cardiovascular risk determinant, is supported by good evidence.

TABLE 1.

Overview on the strengths and limitations of different study designs

Study design	Strengths	Limitations
Population-based observational study	Large samples; accounts for many confounders	Retrospective; no a priori power calculation
OSA cohort study	Large, selected samples; risk predictors analysis; identification of vulnerable patient groups	Retrospective; less selected samples than RCT; “healthy adherer” effect
Case–control study	Large samples; partially controlled setting	Retrospective; not totally controlled environment
Uncontrolled interventional trial	Prospective; smaller logistic and financial burden	Control group missing; low level of evidence
RCT	Prospective; controlled setting; causal assessment; high-quality evidence	Smaller samples; sometimes reduced generalisability; ethics; higher logistic and financial burden
Meta-analysis (study level)	High-quality evidence	Secondary analysis; depending on the characteristics of included studies
Meta-analysis (individual patient data)	High-quality evidence; better interpretation of meta-analysed data; identification of predictors of treatment response	Secondary analysis; depending on the characteristics of included studies; greater analysis burden (time and resources)

OSA: obstructive sleep apnoea; RCT: randomised controlled trial.

Current evidence shows a higher incidence of MACE in some subgroups of patients with untreated OSA and highlights the importance of good long-term CPAP adherence in reducing cardiovascular risk. Subgroup analyses have identified a potential OSA phenotype who may benefit most from CPAP: hypertensive, non-elderly, male, sleepy patients with more severe OSA and without vascular end-organ damage. In addition to the direct effects of OSA (e.g. heart rate variability and cyclic intermittent oxygen desaturations in association with obstructive respiratory events) and disease-specific symptoms and comorbidities, studies on MACE should consider the role of blood pressure, endothelial dysfunction and the effect of CPAP therapy on these as intermediate outcomes and modifiable cardiovascular risk factors. When designing future studies, care should be taken to ensure that the exact dosage of antihypertensive medications, as well as drugs that counteract other cardiovascular risk factors, is documented and reported at the beginning and during the course of the study. A summary of these findings is provided in table 2.

TABLE 2.

Summary of findings (arterial hypertension, endothelial dysfunction, major adverse cardiovascular events (MACE) and related continuous positive airway pressure (CPAP) treatment effect) of the analysed study designs

Study design	Findings
Population-based observational study
Hypertension	Dose–response relationship between OSA severity and hypertension incidence
Endothelial dysfunction	No data
MACE	Association between untreated OSA and cardiovascular events, especially in younger, male sleepy patients; CPAP-mediated risk reduction of cardiovascular events in severe OSA
OSA cohort study
Hypertension	No conclusive evidence
Endothelial dysfunction	No data
MACE	Association between nocturnal hypoxaemia and adverse cardiovascular outcomes; good CPAP adherence reduces MACE incidence, especially in primary prevention setting, younger patients and severe OSA
Case–control study
Hypertension	Association between hypertension and OSA severity
Endothelial dysfunction	Association between arterial stiffness and OSA severity
MACE	No conclusive evidence
Uncontrolled interventional trial	Not analysed
Randomised controlled trial
Hypertension	CPAP-mediated BP improvement or no difference; effect size depends on cohort characteristics
Endothelial dysfunction	CPAP-mediated improvement
MACE	No effect of CPAP on MACE
Meta-analysis (study level)
Hypertension	CPAP-mediated BP improvement, especially in difficult-to-control and resistant hypertension
Endothelial dysfunction	CPAP-mediated improvement
MACE	CPAP-mediated MACE incidence reduction possible (sensitivity analysis)
Meta-analysis (individual patient data)
Hypertension	Baseline BP as response predictor of CPAP-mediated BP reduction
Endothelial dysfunction	No data
MACE	Good CPAP adherence reduces MACCE incidence in patients with moderate-to-severe OSA and cardiovascular disease

OSA: obstructive sleep apnoea; BP: blood pressure; MACCE: major adverse cardiovascular and cerebrovascular events.

A good balance between the level of evidence and potential sources of bias, the primary outcome, the appropriate sample size for sufficient statistical power, the patient population of interest, and the feasibility of studies in terms of patient tolerability, acceptability and ethical aspects should be considered when assessing the evidence. Despite a substantial number of well-designed studies, some research questions in the field of cardiovascular effects of OSA and its treatment remain unanswered. Further well-designed, large studies with broad inclusion criteria for generalisability and sufficient power to detect an effect on individual cardiovascular outcomes are needed, as well as RCTs in specific risk populations for cardiovascular consequences of OSA, and thus relevant OSA phenotypes. Moreover, patient stratification within several arm RCTs should be considered. Furthermore, studies investigating the role of OSA and the effects of its treatment on cardiovascular outcomes in pre-menopausal and post-menopausal women are needed. In addition, different OSA therapies could also be used and compared in RCTs, e.g. CPAP, mandibular advancement device, maxillo-mandibular repositioning osteotomy and electrical stimulation, provided they are effective in controlling OSA, safe and well tolerated.

In conclusion, there is high-level evidence on the causal relationship between OSA and hypertension and OSA and endothelial dysfunction, as well as convincing evidence on higher MACE incidence in some subgroups of patients with untreated OSA (figure 1). The cardiovascular effects of OSA differ among the heterogeneous group of patients with OSA, depending on disease severity, symptoms, comorbidities and phenotype characteristics. The blood pressure-lowering effect of CPAP is mainly observed in uncontrolled and treatment-resistant hypertension. Patients with OSA that benefit most from CPAP treatment in terms of MACE risk reduction have been identified as younger, sleepy, with more severe OSA, higher nocturnal hypoxaemic burden and without overt cardiovascular end-organ disease (figure 2). Good long-term CPAP adherence is essential for reducing the risk of MACE in patients with OSA. Providing effective and well-tolerated treatment options for OSA remains an important goal. RCTs investigating specific treatments for OSA in primary cardiovascular prevention and in patients at highest risk are currently lacking.

FIGURE 1.

Overview on the association of obstructive sleep apnoea (OSA) on the incidence of hypertension, endothelial dysfunction and major adverse cardiovascular events (MACE) in different OSA types.

FIGURE 2.

Overview on the effect of continuous positive airway pressure (CPAP) on blood pressure (BP), endothelial dysfunction and major adverse cardiovascular events (MACE) in different obstructive sleep apnoea (OSA) types. RCT: randomised controlled trial.

Questions for future research

• Need for a gender medicine approach and well-designed studies on cardiovascular effects of OSA and its treatment in female OSA patients at pre- and post-menopausal age.

• Identification of further OSA phenotypes that might particularly benefit from OSA treatment in terms of cardiovascular risk reduction.

• Studies with high-level evidence in at-risk OSA phenotypes for adverse cardiovascular outcomes.