Summary
Obstructive sleep apnoea syndrome (OSAS) is a frequent sleep disorder that is known to be an independent risk factor for arterial hypertension (AHT). Potential confounding factors associated with both OSAS and AHT, such as age, diabetes mellitus and obesity, have been explored extensively, and are considered as independent but additive factors. However, these factors are also contributors to left ventricular (LV) hypertrophy (LVH) and LV diastolic dysfunction, both of which are important causes of cardiovascular morbidity, and have been reported to be associated with OSAS for decades. In this review, we present an overview of how OSAS may promote changes in LV geometry and diastolic dysfunction through its best-known cardiovascular complication, arterial hypertension. We also summarize the epidemiological links between OSAS and LVH, outline diastolic dysfunction in OSAS patients, and try to highlight the mechanisms responsible, focusing on the effect of confounding factors.
Résumé
Le syndrome d’apnées du sommeil obstructif (SASO) est un trouble du sommeil fréquent, et un important facteur de risque d’hypertension artérielle (HTA). Les potentiels facteurs confondants associés au SASO ainsi qu’à l’HTA, comme l’âge, le diabète, et l’obésité, ont été largement explorés et sont aujourd’hui considérés comme des facteurs indépendants mais additionnels. De plus, ces facteurs contribuent à la survenue d’hypertrophie ventriculaire gauche et de dysfonction diastolique, deux conditions associées à une forte morbi-mortalité cardiovasculaire et associées au SASO depuis des décennies. Cette revue présente une vue d’ensemble des effets du SASO sur la géométrie ventriculaire gauche et sa fonction diastolique par sa principale complication cardiovasculaire qu’est l’HTA. Nous résumons les liens épidémiologiques entre SASO et HVG, décrivons la fonction diastolique dans le SASO, et présentons les mécanismes physiopathologiques impliqués en tenant compte des facteurs confondants.
Background
Obstructive sleep apnoea syndrome (OSAS) is characterized by repetitive episodes of partial or complete collapse of the upper airway during sleep; this usually terminates in arousal, leading to sleep fragmentation. These periods of obstructed breathing result in intermittent hypoxaemia with underlying sympathetic nerve activity and increases in heart rate and blood pressure (BP). Diagnosis and severity of OSAS are determined by the apnoea hypopnoea index (AHI). The American Academy of Sleep Medicine sets a threshold of five events per hour of sleep, predominantly obstructive, with symptoms such as daytime sleepiness, insomnia or snoring, for the diagnosis of OSAS .
OSAS is a frequent sleep disorder that affects 2% of middle-aged women and 4% of men . Prevalence has reached 20% in some studies , increasing with age and the growing rate of obesity . However, more than 80% of OSAS patients stay undiagnosed, especially women and those with a lower body mass index (BMI) . OSAS is a well-known independent risk factor for arterial hypertension (AHT) , the prevalence of which reaches 50% in this population . Reciprocally, the reported prevalence of OSAS among hypertensive populations ranges from 20% to 40% , and up to 70% . AHT in OSAS patients is more likely to affect diastolic BP in young people than systolic BP in the elderly ; typically, it has a non-dipper or a riser (higher sleep BP than awake BP) pattern that leads to a higher frequency of masked hypertension (around 30% of cases) , and both of these conditions are known to be associated with even worse outcomes . OSAS is also a recognized cause of resistant AHT , where the prevalence of OSAS exceeds 80% . Potential confounding factors associated with both OSAS and AHT, such as age, diabetes mellitus and obesity, have been explored extensively, and are considered as independent but additive factors . However, those factors are also contributors to left ventricular (LV) hypertrophy (LVH) and LV diastolic dysfunction, both of which are important causes of cardiovascular morbidity , and have been reported to be associated with OSAS for decades .
In this review, we present an overview of how OSAS may promote changes in LV geometry and diastolic dysfunction through its best-known cardiovascular complication, arterial hypertension. We also summarize the epidemiological links between OSAS and LVH, outline diastolic dysfunction in OSAS patients and try to highlight the mechanisms responsible, focusing on the effect of confounding factors.
LV geometry and OSAS
Description of LVH and remodelling in OSAS
In 1990, Hedner at al. conducted a case-control study comparing 61 OSAS and 61 control patients. The interventricular septum and LV posterior wall were thicker, and so the LV mass (LVM) and LVM index (LVMi) were significantly higher in OSAS patients. In 1995, Noda et al. provided the first prevalence of LVH in OSAS patients, defined by LV wall thickness ≥ 12 mm. LVH was reported in 42% of the whole cohort ( n = 51), in 31% when the AHI was < 20 and in 50% when the AHI was ≥ 20. Using the same criteria for LVH, Cloward et al. reported a prevalence of LVH of 88% among 25 obese and severe OSAS patients. A dose-response relationship was also observed between the severity of OSAS and the prevalence of LVH, using LVMi (normalized by height) for LVH assessment . The largest cross-sectional study, including more than 2000 subjects (the Sleep Heart Health Study) , confirmed that LVMi (height) was significantly associated with both the AHI and the hypoxaemia index, even after adjustment for age, BMI, systolic BP and diabetes, with an adjusted odds ratio for LVH of 1.78 (95% confidence interval 1.14–2.79) between patients with an AHI < 5 and those with an AHI ≥ 30. In this study, the prevalence of LVH reached 33% in severe OSAS patients.
The reported LVH is often eccentric; for example, Myslinski et al. observed in 2007 that eccentric LVH was the predominant LV geometry in patients with newly diagnosed OSAS (without continuous positive airway pressure [CPAP] treatment). In this study and in the Sleep Heart Health Study , the LV end-diastolic diameter in OSAS patients was significantly higher than in controls (or treated OSAS patients), and correlated positively with the AHI and the desaturation index. Eccentric LVH was also twice as frequent as concentric LVH among treated OSAS patients . Conversely, Cioffi et al. reported that relative wall thickness was positively correlated with the AHI ; in this study, where BP was not different across the OSAS severity groups, LVM did not differ significantly, but LV concentric remodelling was independently associated with moderate-to-severe OSAS (odds ratio 7.6) and BMI (odds ratio 1.09). Likewise, Koga et al. reported that concentric LVH was the most common LV geometry in 37 OSAS patients. Fig. 1 depicts the different types of LVH.
LV remodelling in OSAS beyond AHT
Whether LVH in OSAS is a reality beyond AHT has been explored extensively and remains controversial. In the first case-control study showing LVH in OSAS patients , neither systolic nor diastolic BP correlated with the desaturation index, and LVMi was 15% higher in normotensive OSAS patients compared with normotensive controls, suggesting that OSAS affects LV geometry independent of AHT. Noda et al. were the first to evaluate LV geometry and AHT based on ambulatory blood pressure monitoring. The prevalence of LVH, which was 50% in case of severe OSAS, reached 70% among patients with coexisting severe OSAS and AHT, defined as BP > 160/95 mmHg. All patients with LVH had AHT, whereas no patients without LVH had AHT. Moreover, LVMi (normalized by body surface area [BSA]) correlated significantly and strongly with 24-hour systolic and diastolic BP ( r = 0.60 and r = 0.48, respectively), and markedly weakly with the obesity index ( r = 0.05), suggesting that AHT diagnosed via 24-hour ambulatory blood pressure monitoring is the most important determining factor for LVH in OSAS. These findings were supported by another cross-sectional study of 81 patients referred to a sleep laboratory , which reported that the AHI correlates with 24-hour BP and with LV thicknesses; however, this last correlation did not remain significant after adjustment for 24-hour BP and the use of antihypertensive drugs. In 2001, Niroumand et al. conducted a large observational study among 533 patients, and concluded that LVMi (height) was correlated with age and AHT, but not with OSAS severity variables after multivariable analysis. Likewise, in a study conducted in non-hypertensive subjects, no differences were found in LV thicknesses, LVM or LVMi between controls, mild-to-moderate OSAS patients and severe OSAS patients , whereas a second study reported a higher interventricular septum thickness in patients with an AHI > 15, but these patients were also more obese . However, among 130 newly diagnosed and untreated patients, Baguet et al. found an independent correlation between LVMi (height) and mean nocturnal arterial oxygen saturation ( P = 0.004), in addition to a significant correlation with clinical BP ( P = 0.01). The low prevalence of LVH observed in this study (5–9.5%) may also be explained by less severe AHT (and obesity) compared with the previous study; this is consistent with a recent study reporting an LVH prevalence of 12% among well-controlled hypertensive patients . In a recent study involving 155 resistant hypertensive patients, Dobrowolski et al. also found an independent association between concentric geometry and both systolic BP and OSAS.
LV remodelling in OSAS beyond obesity
As the prevalence of LVH is very high in obese subjects , obesity can be a major confounding factor. In the first study reporting a higher LVMi in OSAS patients, body weight was significantly higher in OSAS patients than in controls . Yet, similar results were reported by Dursunoglu et al. , without differences in BMI across groups. Several studies have reported correlations between the AHI and LV thicknesses that did not remain significant after adjustment for BMI . Besides, the very high prevalence of LVH (88%) among 25 severe OSAS patients found by Cloward et al. did not take into account the fact that all patients were obese (mean BMI 38 ± 11), because LV measurements were not related to obesity. However, when LV geometry was assessed by LVMi, results were not consistent, suggesting the importance of the method used. Indeed, some studies reported a significant correlation between LVMi and OSAS severity variables after multivariable analysis including BMI , whereas others did not . Furthermore, several studies reported that, after multivariable analysis, LVMi was determined by metabolic syndrome, but not by BMI .
Effect of measurement method on LV geometry
Obtaining a reliable measurement of LVM in OSAS patients is difficult because this population is frequently overweight, which may significantly affect results. From one study to another, LVH may be defined as LV thicknesses (interventricular septum and/or posterior wall) ≥ 12 mm or may be based on LVM that can be normalized by BSA or height or not. Studies that assess LVH using LV measurements alone tend to report no association between LVH and OSAS or higher two-dimensional LV dimensions in OSAS patients that do not remain significant after adjustment for obesity . Most recent studies show a persistent independent relationship between OSAS and LVMi (height) after adjustment for confounding factors or multivariable analysis . However, few studies have reported no significant association between LVMi (BSA) or LVMi (height) and OSAS after taking into account confounding factors. Despite these discrepancies, we can reasonably think that OSAS is independently associated with LVMi, and that in addition to this, OSAS, BMI and AHT have significant interacting effects on LVMi .
Today, it is admitted that LVMi normalized by height and based on an obesity-independent measure of body size, is associated with a higher proportion of incident cardiovascular events than LVH detected by normalization for BSA, and is convenient for identification of individuals at high risk in populations with a high prevalence of obesity, such as OSAS populations. This method allows detection of poor prognosis and obesity-related LVH, which is not identified using BSA . In a recent observational study among 188 treated and severe OSAS patients, the prevalence of LVH when assessed by LVMi normalized by height was two times higher than when normalized by BSA (12.4% vs. 6.5%) . We should also mention that a small prospective study of 47 OSAS patients reported uniform results regarding the effects of CPAP on cardiac remodelling, assessed by echocardiography or cardiac magnetic resonance imaging .
Effects of OSAS treatment on LV geometry
Observational studies in obese and severe OSAS patients reported regression of interventricular wall thickness after 6 months of nasal CPAP (nCPAP) (13 mm vs. 12.3 mm, P < 0.02 and 13 mm vs. 10 mm, P < 0.001, respectively), which was not observed for the LV posterior wall . Subgroup analyses in this study showed no effect of nCPAP in non-compliant patients. The findings also suggested a delayed effect of OSAS treatment, as LV dimensions were not decreased after 1 month of treatment, just as in another study after 2–3 months of nCPAP or an oral appliance . However, after 18 months of nCPAP, 41 obese patients did not show LVM improvement, except those with a decreased body weight, suggesting that CPAP treatment does not improve LV remodelling per se .
Physiopathology
Pathophysiological mechanisms involved in OSAS-related LV remodelling are summarized on Fig. 2 . Because of repetitive respiratory disturbances leading to related hypoxia, hypercapnia, negative intrathoracic pressure and microarousals, OSAS patients experience permanent oscillations in their haemodynamic variables and sympathetic activation during sleep. This activation is observed more during sleep, but lasts through the day, leading to peripheral vascular remodelling, increased vascular resistance and then increased afterload. In 2007, Drager et al. compared pulse wave velocity (as a surrogate measure of arterial stiffness) and LV dimensions in non-hypertensive OSAS patients, hypertensive patients without OSAS and controls. OSAS and hypertension were both associated with increased arterial stiffness and LVH; the effects were of similar magnitude, but were also additive, suggesting that increased afterload may contribute to LV remodelling in OSAS patients. These observations were supported by Baguet et al. , who reported a significant correlation between carotid intima-media thickness and pulse wave velocity, and systolic and pulsed BP in OSAS patients. Moreover, repetitive episodes of hypoxaemia and reoxygenation may trigger oxidative stress mechanisms and production of reactive oxygen species, leading to endothelial dysfunction , ischaemia-reperfusion injury and then ventricular remodelling. Besides, intermittent hypoxaemia also triggers activation of the renin-angiotensin-aldosterone system, as studies reported increased angiotensin II and aldosterone concentrations in OSAS patients . This is another pathway that probably contributes to the development of LV remodelling.
