Published ahead of print on March 2, 2006, doi:10.1164/rccm.200503-320OC
© 2006 American Thoracic Society doi: 10.1164/rccm.200503-320OC
Left Ventricular Structural Adaptations to Obstructive Sleep Apnea in Dilated CardiomyopathySleep Research Laboratories of the Toronto Rehabilitation Institute; Toronto General Hospital of the University Health Network; Harrowston Heart Failure Clinic, Mount Sinai Hospital; and Centre for Sleep Medicine and Circadian Biology of the University of Toronto, Toronto, Ontario, Canada Correspondence and requests for reprints should be addressed to T. Douglas Bradley, M.D., Toronto General Hospital/UHN, 9N-943, 200 Elizabeth Street, Toronto, ON, M5G 2C4 Canada. E-mail: douglas.bradley{at}utoronto.ca
Rationale and Objectives: Obstructive sleep apnea is common among patients with heart failure and exposes the left ventricle to trophic mechanical and adrenergic stimuli. We hypothesized that in heart failure patients with nonischemic dilated cardiomyopathy (a condition characterized by eccentric hypertrophy), those with obstructive sleep apnea would have a higher prevalence of left ventricular hypertrophy by wall thickness criteria (  12 mm), and greater septal thickness than those without obstructive sleep apnea. Methods and Results: We performed echocardiography and polysomnography in 47 patients with nonischemic dilated cardiomyopathy. Obstructive sleep apnea was present in 45% of these patients. The prevalence of left ventricular hypertrophy was greater in those with than in those without obstructive sleep apnea (47.6 vs. 15.4%, p = 0.016). Interventricular septal thickness (p < 0.001) and relative wall thickness (p = 0.011) were significantly greater in those with than in those without obstructive sleep apnea. However, there was no significant difference in posterior wall thickness between the groups. The frequency of obstructive apneas and hypopneas during sleep was the only significant independent correlate of septal thickness (p = 0.001). Conclusions: In patients with nonischemic dilated cardiomyopathy, the presence of obstructive sleep apnea is associated with an increased prevalence of left ventricular hypertrophy. The higher relative wall thickness and interventricular septal thickness in patients with obstructive sleep apnea indicate that the left ventricle is relatively less eccentric than in patients without obstructive sleep apnea, and that such remodeling affects mainly the septum. These structural adaptations may reflect unique nocturnal mechanical and adrenergic stimuli associated with obstructive sleep apnea.
Key Words: heart failure • left ventricular hypertrophy • obstructive sleep apnea Heart failure (HF) is a debilitating condition affecting approximately 4,700,000 Americans (1). Intrinsic to the progression of HF is adverse cardiac remodeling, which involves left ventricular hypertrophy (LVH) and/or dilatation. A number of factors can contribute to the remodeling process, including hypertension, elevated sympathetic nervous system activity, and myocardial infarction (2–5). One potential stimulus to left ventricular (LV) remodeling that has received little attention is obstructive sleep apnea (OSA). A large population-based study identified OSA as an independent risk factor for HF (6). In addition, OSA often coexists with HF (7, 8), where its treatment by continuous positive airway pressure reduces blood pressure and improves LV systolic function (9, 10). These observations suggest that OSA contributes to the pathogenesis of HF in such subjects. Obstructive apneas are accompanied by recurrent hypoxia and arousals from sleep, which activate the sympathetic nervous system, and cause repetitive surges in heart rate and blood pressure (11–13). In combination with the generation of exaggerated negative intrathoracic pressure against the occluded upper airway, this markedly increases LV afterload during sleep (14, 15). In addition, the consequent increase in myocardial oxygen demand is accompanied by a reduction in oxygen supply because of apnea-related hypoxia (12). Thus, the coexistence of OSA in patients with HF exerts unique nocturnal mechanical, adrenergic, and metabolic stresses that could result, cumulatively, in greater adaptive LV remodeling than would occur in patients with HF, but without OSA. Furthermore, OSA places unique stresses on the interventricular septum (16, 17). These include increased pulmonary artery pressure during sleep due to hypoxic pulmonary vasoconstriction (12, 16), which increases right ventricular afterload and systolic septal wall tension without affecting posterior wall tension. Generation of exaggerated negative intrathoracic pressure during obstructive apneas increases venous return, causing distension of the right ventricle and leftward shift of the septum during diastole (11, 12, 16, 18). We therefore hypothesized, that compared with HF patients without OSA, those with OSA would have a higher prevalence of LVH, by wall thickness criteria, and more pronounced septal thickening. To explore this hypothesis, in patients with HF, we compared LV structure between those with and those without OSA. Because myocardial infarction and ischemia can affect myocardial remodeling independently, we confined our study to subjects with nonischemic dilated cardiomyopathy, in which LV remodeling is usually eccentric (i.e., dilated and thin walled) (19). Some of the work described herein has been published in abstract form (20).
Subjects As part of a prospective epidemiologic study, we performed polysomnography on all newly referred patients to the Mount Sinai Hospital heart failure clinic during the 5-yr period between 1997 and 2002. For this study, we included only subjects with nonischemic dilated cardiomyopathy to avoid regional variations in LV wall thickness (LVWT) that could arise because of scarring in subjects with ischemic cardiomyopathy. Entry criteria included the following: (1) HF of at least 6 mo duration based on a history of exertional dyspnea and peripheral edema; (2) LV systolic dysfunction (LV ejection fraction  45% by two-dimensional echocardiography); (3) nonischemic dilated cardiomyopathy based on the absence of significant coronary artery lesions on coronary angiography or absence of ischemic changes on exercise thallium scintigraphy, and an LV end-diastolic dimension by M-mode echocardiography of 27 mm/m2 body surface area (19); and (4) appropriate medical therapy for HF with stable medications for at least 1 mo before participation. Exclusion criteria were as follows: (1) a prior history of sleep apnea, (2) predominantly central sleep apnea (CSA), (3) ischemic cardiomyopathy, (4) primary valvular heart disease, and (5) obstructive lung disease. The protocol was approved by the Research Ethics Board of the University of Toronto and all subjects provided written, informed consent before their participation.
Echocardiography
Polysomnography
Statistical Analyses
Characteristics of the Subjects Fifty-nine consecutive subjects with nonischemic dilated cardiomyopathy underwent polysomnnography during the 5-yr period between 1997 and 2002 (Figure 1). During this time, several patients who were screened for the study had an improvement in their LV ejection fraction to more than 45% after optimization of medical therapy. These patients did not undergo polysomnography. Of those who did, 26 (42%) did not have OSA (non-OSA group), 21 (44%) had OSA (OSA group), and eight (14%) had CSA. The OSA group had a moderate degree of OSA and a significantly higher AHI, by design, than the non-OSA group (Table 1). Minimum O2 saturation was significantly lower in the OSA than in the non-OSA group (p = 0.002), but there were no significant differences in age, sex, body mass index, New York Heart Association class, systolic or diastolic blood pressure, mean O2 saturation, or LV ejection fraction. Only a minority of subjects had a past history of hypertension, and the prevalences were similar in the non-OSA and OSA groups (23 vs. 33%, respectively; p = 0.65). The type of medications used did not differ significantly between the two groups, although there was a tendency for greater diuretic use in the OSA group (p = 0.059; Table 2).
Echocardiographic Findings The OSA group had a 3.1-fold higher prevalence of LVH by wall thickness criteria than the non-OSA group (p = 0.016; Figure 2). LVWT (p = 0.005), IVST (p < 0.001), and RWT (p = 0.011) were all significantly greater in the OSA than the non-OSA group (Table 3), indicating that they had a greater summed IVST and PWT for a given end-diastolic dimension than the non-OSA group. However, LVEDD, LV end-systolic internal dimension, or PWT did not differ significantly between the two groups (Table 3). Because LVEDD was slightly, but not significantly, greater in the non-OSA patients, mean LV mass index was similar in the two groups (177.2 ± 53.0 vs. 182.5 ± 38.1 g/m2, in the non-OSA and OSA groups, respectively; p = 0.712). Because the mean LV mass index was greater than 150 g/m2, both groups generally met LV mass index criteria for LVH (24, 31).
Single regression analyses were performed on indices of LVWT. LVWT was related significantly to obstructive AHI (p = 0.001), body mass index (p = 0.001), and male sex (p = 0.004; Table 4). IVST was related significantly to obstructive AHI (p < 0.001) and body mass index (p = 0.009), and RWT was related significantly to obstructive AHI (p = 0.017) and to body mass index (p = 0.028). The relationship between IVST and obstructive AHI is plotted in Figure 3. In contrast, there was no significant relationship between PWT and obstructive AHI (R2 = 0.025, p = 0.492). Multiple regression analyses revealed that the only significant independent correlate of LVWT was male sex (p = 0.023), whereas for IVST it was the obstructive AHI (p = 0.002; Table 5). There were no statistically significant independent correlates of RWT. Right ventricular systolic pressures were similar in non-OSA and OSA patients (39 ± 9 vs. 41 ± 10 mm Hg, respectively; p = 0.50).
Several of the present findings provide evidence that OSA may influence the process of LV remodeling in patients with HF. First, although nonischemic dilated cardiomyopathy is usually associated with a dilated, thin-walled LV (i.e., eccentric hypertrophy), the prevalence of LVH, defined on the basis of LV wall thickening, was much greater in those with OSA than in those without. Second, we found that LVWT, IVST, and RWT were significantly greater in those with OSA than in those without. These findings indicated a greater summed IVST and PWT for a given end-diastolic dimension (i.e., a less eccentric pattern of hypertrophy) in subjects with OSA. Third, there was a significant direct relationship between IVST and the severity of OSA, as quantified by the AHI. This relationship was independent of potential confounding factors including age, sex, body mass index, O2 desaturation, and blood pressure. These observations indicate a significant relationship between OSA and the presence of both LVH and the degree of LV thickening. Prospective data from the Framingham Heart Study demonstrate incremental risk of cardiovascular events with increasing LV thickness (32). Among studies in which potential relationships between OSA and LV structure in adults have been examined (23, 33–35), none included patients with HF. In general, those studies found increased LV thickness or mass in association with OSA. Nevertheless, in most of those studies, once differences in body weight and blood pressure were taken into account, relationships between OSA and increased LV thickness or mass were no longer significant. In one of these studies, Arias and colleagues (35) found that, although both IVST and PWT were greater in subjects with, than in those without OSA, these differences were not independent of confounding factors, and treatment of OSA by continuous positive airway pressure for 3 mo had no effect on PWT or IVST. Thus, a clear-cut association between OSA and LV enlargement, independent of other confounding factors, has not been established in adults without HF. Our data also suggest a relationship between body mass index and indices of global LV enlargement. However, multiple regression analyses (Table 5) indicate that OSA has an additional independent effect on LV thickening that preferentially affects the septum. Thus, in our patients, part of obesity's effect on the septum is through its association with OSA as a comorbidity. The only study that has demonstrated an association between OSA and LV thickening or hypertrophy, independently of confounding factors, was performed in children with normal LV systolic function (36). In that study, Amin and colleagues reported that children with OSA had significantly greater LV mass and RWT than those without OSA. Our study differed from that of Amin and colleagues in that we examined the relationship between OSA and LV structure in adults with HF and nonischemic dilated cardiomyopathy. Furthermore, our detailed assessment of LV structure through separate examinations of the septum and the posterior wall allowed us to determine that OSA is specifically associated with a greater summed IVST and PWT for a given end-diastolic dimension, indicating relatively less eccentric hypertrophy than in patients without OSA. This thickening affects mainly the septum. The higher prevalence of LVH in our patients with OSA was present even after controlling for age, sex, body mass index, and history of hypertension. Accordingly, the main novel finding of our study was that the pattern of LV thickening in patients with OSA differed substantially from that in the non-OSA group. LV mass index did not differ significantly between the patients with and without OSA, and the overall pattern of LVH was eccentric (i.e., RWT < 0.41), which is characteristic of nonischemic dilated cardiomyopathy (19). However, the greater summed IVST and PWT for a given end-diastolic dimension in the OSA group are consistent with the observation that a greater degree of LV thickening is associated with higher nocturnal blood pressure, as occurs in patients with OSA (14, 37). Although we did not measure nocturnal blood pressure, we have previously demonstrated that it rises from wakefulness to sleep in HF patients with OSA (14). Also, IVST was greater in the OSA than in the non-OSA group, whereas PWT was not. IVST correlated significantly with the obstructive AHI (Figure 2 and Table 5), while PWT did not. The finding of more pronounced septal thickening in subjects with OSA may be explained by nocturnal cardiovascular stresses unique to OSA. In normal non-REM sleep, metabolic rate and central sympathetic outflow decline in comparison to wakefulness (15, 38). However, OSA counteracts this quiescent state. Intermittent apnea-related hypoxia can contribute to myocardial injury (39, 40), and also stimulates repetitive surges in sympathetic nervous system activity that are potentiated by apnea, CO2 retention, and arousals from sleep (41). Excessive sympathetic activity and circulating catecholamines promote cardiac hypertrophy by stimulating myocyte growth, and promoting myocyte injury and fibrosis (32, 39–42). In addition, these surges in sympathetic activity cause cyclic increases in blood pressure that further increase LV afterload, a potent stimulus to myocardial metabolic demands and hypertrophy (39, 40, 42). Finally, patients with OSA generate exaggerated negative intrathoracic pressure against an occluded upper airway. Negative intrathoracic pressure increases LV transmural pressure, a major determinant of LV afterload (11, 15, 16, 42). Exaggerated negative intrathoracic pressure also increases venous return, causing the right ventricle to distend and the septum to shift paradoxically leftward during diastole (15–17). As a consequence, septal wall tension increases, preventing normal diastolic relaxation of the left ventricle. In addition, although right ventricular systolic pressures did not differ between the OSA and non-OSA groups, it is important to emphasize that Doppler studies were obtained with subjects awake. However, it is during sleep that obstructive apneas cause intermittent hypoxia and repetitive increases in pulmonary artery pressure, which increase right ventricular afterload, and therefore septal tension (12, 16). One would not expect right ventricular systolic pressures to be greater during wakefulness in the patients with OSA than in the patients without OSA because OSA, per se, seldom causes daytime pulmonary hypertension or right heart failure in the absence of coexisting chronic obstructive lung disease accompanied by daytime hypoxia (43–45). However, patients with obstructive lung diseases were excluded from our study. Hence, during sleep, the septum is exposed uniquely to the combined effects of increased right ventricular preload and afterload, and increased LV afterload, whereas the posterior wall is only exposed to increased LV afterload. These factors likely combine to make the septum particularly susceptible to any remodeling forces imposed by OSA. Although it is possible that right-sided forces could contribute to septal thickening as part of a generalized thickening of the right ventricle, it was not possible to obtain reliable transthoracic echocardiographic measurements of right ventricular free-wall thickness in our subjects because of their obesity, thick chest walls, and markedly dilated left ventricles. We did not find any significant relationship between the degree of nocturnal O2 desaturation and indices of LVH. Indeed, it has been demonstrated in patients with OSA that administration of supplemental O2 to prevent hypoxia has no influence on nocturnal apnea-induced elevations in systemic blood pressure (46). It therefore appears that other factors, such as generation of negative intrathoracic pressure or elevations in system blood pressure during sleep, as discussed above, are more important stimuli to LVH than O2 desaturation per se. It is possible that the greater degree of LV thickening in patients with OSA could be related to a longer duration of HF, or to a greater undetected prevalence of hypertension. However, we were unable to determine the duration of HF in most patients. Nevertheless, a longer duration of HF should also have been associated with a greater degree of LV dilation, but there was no difference in LVEDD between the OSA and non-OSA groups. There is also no reason to believe that a history of hypertension would be missed more frequently in patients with than in those without OSA. The lack of a relationship between hypertension and indices of LV thickening in the OSA group is probably due to a similar prevalence of a history of hypertension in both groups, and that only a minority in subjects had a history of hypertension. The 33% prevalence of known hypertension in the patients with OSA is consistent with a previous large epidemiologic study of the general population (47). In conclusion, we have demonstrated that in patients with HF and nonischemic dilated cardiomyopathy, those with OSA have a significantly higher prevalence of LVH by wall thickness criteria, and a different pattern of LV remodeling than patients without OSA, despite similar LV mass index. This pattern is characterized by greater RWT, indicating a relatively less eccentric pattern of LVH than in the non-OSA group, and by greater thickening of the septum, which was proportional to the frequency of obstructive apneas and hypopneas. Our data suggest that LV remodeling associated with OSA in patients with nonischemic dilated cardiomyopathy is characterized by superimposition of LV wall thickening, particularly affecting the septum, on a dilated LV chamber. Because our data were cross-sectional in nature, they do not establish whether the relationship between OSA and LV thickening was causal. Nevertheless, this greater degree of LV thickening may put HF patients with OSA at greater risk for adverse cardiovascular events than those without OSA (2, 31, 36, 48). These observations add to the growing evidence that OSA can have detrimental effects on LV structure and function (9, 11, 32). Because treatment of OSA by continuous positive airway pressure in patients with HF causes marked improvements in LV systolic function (9, 10), our findings also suggest the potential for therapy for OSA to contribute to reverse remodeling of the left ventricle in patients with nonischemic dilated cardiomyopathy. Randomized trials will be required to test this hypothesis.
The authors thank George Tomlinson for his assistance in the statistical analysis of the data.
Supported by grants from the Ontario Thoracic Society and the Canadian Institutes of Health Research (MOP 11607). K.U. and C.M.R. were supported by unrestricted research fellowships from Respironics, Inc.; J.D.P. and J.S.F. by Career Investigator Awards from the Heart and Stroke Foundation of Ontario; and T.D.B. was supported by a Senior Scientist Award from the Canadian Institutes of Health Research. Originally Published in Press as DOI: 10.1164/rccm.200503-320OC on March 2, 2006 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form March 1, 2005; accepted in final form February 28, 2006
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
2 tests or the Fisher exact test, as appropriate, for categoric data, and by unpaired two-tailed t tests for continuous data. To determine whether there were any significant relationships between the indices of LV thickness and the independent variables (clinical characteristics and polysomnographic factors), single regression analyses were performed. Independent variables were then entered into a multiple linear regression model with indices of LV thickness as the dependent variables only if they changed the slope of the relationship between obstructive AHI and the dependent variable by at least 10% with use of a statistical program (SPSS version 13; SPSS, Inc., Chicago, IL). A p value less than 0.05 was considered statistically significant. 
