OUP user menu

Obstructive sleep apnoea inhibits the recovery of left ventricular function in patients with acute myocardial infarction

Hiroshi Nakashima, Toshiro Katayama, Chisa Takagi, Kentarou Amenomori, Masahiko Ishizaki, Yukiharu Honda, Shin Suzuki
DOI: http://dx.doi.org/10.1093/eurheartj/ehl219 2317-2322 First published online: 6 September 2006


Aims It has been suggested that obstructive sleep apnoea syndrome (OSA) may be a direct cause of left ventricular (LV) systolic dysfunction. This study was designed to examine our hypothesis that OSA inhibits the recovery of LV function in patients with acute myocardial infarction (AMI).

Methods and results Our 86 consecutive first-AMI patients underwent primary percutaneous coronary intervention (PCI). All patients underwent polysomnography and OSA was defined as an apnoea–hypoapnoea index (AHI) ≥15 events/h, of which more than 50% were obstructive. Left ventriculograms immediately after PCI and at 21 days were used to evaluate LV ejection fraction (LVEF), LV end-diastolic volume index, and regional wall motion (RWM) within the infarct area. OSA was observed in 37 patients (43%). All three indices of LV function after primary PCI were comparable between the two groups. Increases in LVEF and RWM during admission were significantly lower in OSA patients than those without OSA (delta LVEF: −0.3±9.6 vs. 7.4±7.2%, P<0.001; delta RWM: 0.26±1.04 SD/chord vs. 1.16±1.20 SD/chord, P=0.002). Multiple regression analysis showed that AHI correlated negatively with delta LVEF and delta RWM.

Conclusion The novel finding is that OSA may inhibit the recovery of LV function in patients with AMI.

  • Obstructive sleep apnea
  • Acute myocardial infarction
  • Primary percutaneus coronary intervention


Obstructive sleep apnoea (OSA) is characterized by repetitive air-flow reduction or cessation caused by upper airway collapse in spite of respiratory effort. There is increasing evidence that OSA is a risk factor for the development of cardiovascular disease.14 The deteriorative effects of OSA on cardiac function have received attention recently because OSA has been found to increase cardiac after-load,5,6 stimulate sympathetic nerve activity,7 and induce myocardial ischaemia.8,9 Two major epidemiological studies have also found that the incidence of OSA among patients with chronic heart failure ranged from 11–37%.10,11 Left ventricular (LV) systolic dysfunction defined as LV ejection fraction (LVEF) <50% was noted in 7.7% of OSA patients.12 Continuous positive airway pressure (CPAP) treatment, which dramatically decreases apnoea or hypoapnoea events, improved LVEF in patients with dilated cardiomyopathy and OSA.13 These findings suggest that OSA may be a direct cause of LV systolic dysfunction.

Successful reperfusion is the principal treatment for acute myocardial infarction (AMI).14,15 Reperfused myocardium is sensitive to tissue perfusion grade, haemodynamic changes, and neurohormonal activation. OSA has frequently been noted in patients with AMI, but the effects of OSA on the time course of LV function after reperfusion have not been understood. In this study, we have hypothesized that OSA may inhibit the recovery of LV function in AMI patients who underwent primary percutaneous coronary intervention (PCI).


Study population

Between July 2003 and December 2004, 120 patients admitted to Nagasaki Citizens Hospital with a first ST-segment elevation myocardial infarction were initially considered for inclusion in the study. AMI was defined as ischaemic symptoms lasting >30 min with ST-segment elevation ≥1 mm in at least two contiguous limb leads or ≥2 mm in precordial leads. The inclusion criteria for the study were as follows: (i) infarct-related artery (IRA) location in either the left anterior descending artery (LAD) or right coronary artery (RCA) at initial coronary arteriography; (ii) primary PCI achieved <12 h from the onset of AMI; (iii) right anterior oblique view of left ventriculogram achieved after primary PCI and at 21 days; (iv) polysomnography was performed and the findings showed normal sleep studies or OSA.

A total of 12 patients who could not undergo primary PCI within 12 h from the onset because of delayed admission were excluded from this study. We excluded 19 patients with IRA location in the left circumflex artery because right anterior oblique view of left ventriculogram cannot precisely estimate the LV wall motion surrounded by left circumflex artery. Two patients expired before the sleep study and three patients showed a central sleep apnoea. Therefore, a total of 36 patients were excluded from this study and our study population consisted of the remaining 86 patients. All patients gave written informed consent, and the study was approved by all relevant committees at our institution.

PCI procedure

All patients received oral administration of aspirin (200 mg) and a bolus injection of heparin (10 000 U) as a matter of routine prior to PCI. Primary PCI was performed in the IRA according to conventional methods within 60 min of admission. Spontaneous recanalization of the IRA was defined as initial TIMI grade 2 or 3 flow and good collateral as ≥2 using the Rentrop classification.16 In the event of inadequate lesion dilatation and flow-limiting dissection, the patient was stented. Angiographic no-reflow after PCI was taken to suggest flow <TIMI 3 in the absence of residual stenosis >30%, a dissection, or distal thrombo-embolization.17 Procedural success for PCI was defined as diameter stenosis <30% with a TIMI 3 flow.

Immediately after PCI, all patients were started on a 4 mg/h dose of intravenous nicorandil, which was continued for 2 days. Oral ticlopidine (200 mg/day) was given to all patients who underwent stent implantation. Angiotensin-converting enzyme (ACE) inhibitors or angiotensin II receptor antagonists were given to all patients who did not show hypotension, defined as <90 mmHg.

Analysis of LV function

Right anterior oblique views of left ventriculograms were taken immediately after PCI and at 21 days and analysed by a single physician (T.K.) unaware of the clinical specifics of each patient. The area–length method was used to calculate the LVEF and LV end-diastolic volume index (LVEDVI).18 Regional wall motion (RWM) within the infarct area was analysed using the centerline method. The extent of severely depressed wall motion was expressed by the number of consecutive chords >2 SD below norm. According to a previous study, the territories of the LAD and the RCA extend from chords 10 to 66 and 51 to 80, respectively.19 Serial changes in LV function were calculated by subtracting LVEF, LVEDVI, and RWM after PCI from these values at 21 days, and expressed as delta LVEF, delta LVEDVI, and delta RWM.


Between 14 and 21 days, all patients underwent overnight polysomnography using the standard technique. Electroencephalograms, electroocculograms, and chin electromyograms were recorded to determine the stages of sleep. States of sleep and arousal were scored according to standard criteria. Naso-oral air-flow and thoraco-abdominal excursions were analysed to define the type of apnoea. Arterial oxyhaemoglobin saturation (SaO2) was recorded using a pulse oximeter and electrocardiogram recordings in a single precordial lead were taken.

Apnea was defined as complete cessation of air-flow for ≥10 s. Hypoapnoea was defined as a ≥50% reduction in air-flow lasting ≥10 s associated with a 4% decrease in oxygen saturation and a state of arousal. Patients' apnoea–hypoapnoea index (AHI) was defined as the number of apnoeas and hypoapnoeas per hour. An obstructive apnoea was defined as the absence of air-flow in spite of respiratory movement or exertion. A central sleep apnoea was defined as the absence of both air-flow and respiratory movement. OSA was defined as ≥15 AHI events per hour, of which more than 50% were obstructive. Mean SaO2, minimum SaO2, and cumulative time with SaO2<90% during the study were also evaluated and recorded.


Continuous variables are expressed as mean ± SD and were compared using unpaired Student's t-test between control patients and patients with OSA. Dichotomous variables are expressed as counts with percentages and were compared using χ2 test. To investigate the effects of stratified AHI values on delta LVEF and delta RWM, patients were subdivided on the basis of AHI values (AHI<5 events/h, 5≤AHI<15 events/h, 15≤AHI<30 events/h, AHI≥30 events/h) and one-way analysis of variance followed by multiple comparisons using the Williams' test was used. Multiple linear regression analysis was used to assess the independent impact of AHI on the recovery of delta LVEF and delta RWM, respectively, while adjusting for other variables. The following parameters that can clinically affect LV function2022 were entered into the model: body mass index (BMI), time to admission, anterior infarct, peak creatine kinase (CK) value, and no-reflow. To decrease the inflation of the Type 1 error rate due to multiple testing, statistical significance was defined as two-sided P<0.01.


Thirty-seven of 86 patients (43%) fulfilled the criteria for OSA and 49 (control) patients presented with an AHI <15 events/h. The patient characteristics of the two groups are presented in Table 1. There were no significant differences between the two groups in terms of age, gender, coronary risk factors, time to admission, infarct location, and haemodynamic findings on admission. Coronary angiographic findings including IRA distribution, spontaneous recanalization, multivessel disease, and the availability of good collateral flow were similar between the two groups. Incidence of angiographic no-reflow was higher in patients with OSA than in control patients (21.6 vs. 6.1%, P=0.033). All eight OSA patients and two of the three control patients who showed no-reflow improved to TIMI grade 3 flow after receiving intracoronary papaverine or verapamil injection.

View this table:
Table 1

Patient characteristics

Control (n=49)OSA(+) (n=37)P-value
Male gender33 (67)30 (81)0.154
BMI (kg/m2)23.1±2.823.1±2.50.976
Hypertension36 (73)26(70)0.743
Hypercholesterolemia21 (43)15 (41)0.829
Diabetes15 (31)12 (32)0.857
Smoking24 (49)16 (43)0.598
Time to admission (h)4.1±3.64.4±4.60.752
LAD/RCA25 (51)/24 (49)20 (54)/17 (46)0.780
Peak CK (IU/L)2543±19562743±25140.679
Spontaneous recanalization19 (39)15 (41)0.765
Multivessel disease26 (53)14 (38)0.161
Good collaterals3 (6)6 (16)0.130
No-reflow3 (6)8 (22)0.033
Final TIMI 3 flow48 (98)36 (97)0.383
  • Figures in brackets represent percentage.

Sleep studies

Sleep disorders were frequently observed in control patients too because the symptoms which relate to sleep disorder do not differ between the two sets of patients (Table 2). The polysomnographic findings are presented in Table 3. Patients with OSA showed significantly higher AHI than control patients. There were no significant differences in mean SaO2 and minimum SaO2 during the sleep in the two groups. The total duration of arterial desaturation defined as SaO2<90% was significantly longer in patients with OSA than in controls. After predischarge left ventriculogram was performed, CPAP treatment was started in 17 of 37 OSA patients who were tolerable with and agreed with this treatment.

View this table:
Table 2

Sleep-related symptoms

Control (n=49)OSA(+) (n=37)P-value
Snoring20 (41)20 (54)0.223
Insomnia14 (29)11 (30)0.587
Day-time sleepiness7 (14)8 (22)0.223
Witnessed apnea14 (29)11 (30)0.586
  • Figures in brackets represent percentage.

View this table:
Table 3

Polysomnographic data

Control (n=49)OSA(+) (n=37)P-value
Total sleep time (min)352.1±108.7341.2±72.80.608
AHI (events/h)5.8±4.231.7±13.6<0.001
Mean SaO2 (%)95.0±2.494.5±2.60.356
Minimum SaO2 (%)88.0±9.085.8±4.00.176
Cumulative time of SaO2<90% (min)2.6±6.910.9±14.70.001
Arousal index (events/h)7.3±6.523.9±14.7<0.001

LV function

A comparison of LV function in the two groups is shown in Table 4. There were no significant differences in terms of LVEF, RWM, number of chord showing <−2 SD/chord or LVEDVI after PCI in the two groups, but at 21 days, patients with OSA showed a significantly decreased LV systolic function compared with control patients. The changes in LV function presented in Table 5 indicate that both delta LVEF and delta RWM were significantly lower in patients with OSA. Figures 1 and 2 showed the changes of LVEF and RWM based on the AHI level. Patients with AHI≥30 events/h and those with 15≤AHI<30 events/h showed a tendency for a lower delta LVEF and delta RWM compared with patients with AHI<5 events/h. These findings suggest that OSA defined as AHI ≥15 events/h may inhibit the recovery of LV systolic function. In fact, after adjustment in the multivariable model, AHI (P=0.008) was identified as an independent contributor to lower delta LVEF (Table 6). In addition, AHI (P=0.003) and peak CK value (P<0.001) were identified as independent factors associated with lower delta RWM (Table 6).

Figure 1

Difference of delta LVEF by the four stratified AHI levels.

Figure 2

Difference of delta RWM by the four stratified AHI levels.

View this table:
Table 4

LV function

Control (n=49)OSA(+) (n=37)P-value
After PCI
 LVSBP (mmHg)121±26122±190.749
 Heart rate (b.p.m.)76±1380±170.283
 LVEDP (mmHg)13±712±60.683
 LVEF (%)53±1254±120.795
 RWM (SD/chord)2.29±1.022.20±0.870.699
 Number of chords <−2 SD/chord29±1627±170.728
 LVEDVI (mL/m2)68±1862±130.130
After 21 days
 LVSBP (mmHg)125±23125±250.936
 Heart rate (b.p.m.)71±1569±120.677
 LVEDP (mmHg)13±612±60.677
 LVEF (%)59±1352±120.022
 RWM (SD/chord)−1.27±1.38−1.92±1.340.051
 Number of chords <−2 SD/chord17±1826±200.071
 LVEDVI (mL/m2)75±2067±170.091
  • LVSBP, LV systolic blood pressure.

View this table:
Table 5

Changes in LV function

Control (n=49)OSA(+) (n=37)P-value
Delta LVEF (%)7.4±7.2−0.3±9.6<0.001
Delta RWM (SD/chord)1.16±1.200.26±1.040.002
Delta LVEDVI (mL/m2)5.2±14.64.8±15.40.913
View this table:
Table 6

Multiple regression analysis

Dependent variableExplanatory variableRegression coefficient (95% confidence interval)P-value
Delta LVEFAHI−0.170 (−0.301 to −0.039)0.008
Peak CK−0.001 (−0.002 to 0.00007)0.070
Anterior MI5.62 (1.61 to 9.63)0.007
Time to admission−0.279 (−0.782 to 0.224)0.271
BMI−0.389 (−1.17 to 0.39)0.323
No-reflow−5.23 (−11.78 to 1.33)0.116
Delta RWMAHI−0.023 (−0.039 to −0.008)0.003
Peak CK−0.0002 (−0.0003 to −0.00008)<0.001
Anterior MI0.730 (0.260 to 1.20)0.003
Time to admission−0.053 (−0.117 to 0.010)0.096
BMI−0.063 (−0.155 to 0.029)0.178
No-reflow−0.308 (−1.07 to 0.454)0.421


We used the AHI ≥15 events/h to diagnose OSA. In fact, our control patients include mild OSA patients with 5≤AHI<15 events/h. Although the generally accepted AHI cut-off value that can show worse clinical and functional outcomes is not yet determined, published studies that investigated the effects of OSA and/or CPAP on LV function and haemodynamics employed AHI cut-off values of 10, 15, and 20 events/h.6,12,13 As illustrated in Figures 1 and 2, patients with AHI<5 events/h and 5≤AHI<15 events/h showed similar recovery of LV function. While, patients with 15≤AHI<30 events/h and AHI≥30 events/h showed a poor improvement of LV function compared with patients with AHI<5 events/h.

The most important finding in this study is that OSA may inhibit the recovery of LV systolic function in patients with AMI who have undergone primary PCI. We observed no significant differences between our two groups in terms of LVEF and RWM within the infarct area after primary PCI. However, the improvement in LVEF and RWM were less marked in patients with OSA than in our control patients. Multiple regression analysis showed that AHI was an independent predictor of delta LVEF and delta RWM. We discuss here the several possible mechanisms for poorer functional outcomes in patients with OSA.

Ischaemia-reperfusion directly injures the cardiac muscle because of exposure to myocardial necrosis, apotosis, and stunning.2325 Repetitive oxygen desaturation/reoxygenation stimulated the production of blood reactive oxygen metabolite in patients with severe OSA (AHI>20 events/h) more than in controls.26 Patients with OSA had increased intracellular reactive oxygen species production in some monocyte and granulocyte sub-populations.27 In addition, antioxidant capacity was reduced in patients with severe OSA defined as AHI>20 events/h.28 In the clinical setting of AMI, the greater free radical production and their decreased elimination may, in patients with OSA, increase the size of the infarct zone and/or result in stunned myocardium. In our study subjects, we know that infarct sizes were not significantly different because peak CK values were comparable between the two groups. Therefore, the inferior recovery of LV function in our OSA patients may be attributable to higher degrees of stunned myocardium. The higher incidence of angiographic no-reflow immediately after PCI in our OSA patients, a possible clinical sign of reperfusion injury, may support this speculation.

It is well known that LV systolic function is highly dependent on cardiac preload and afterload. In our study subjects, LV end-diastolic pressure (LVEDP) and LVEDVI at two measurements were comparable between the two groups and the grade of LV remodelling (delta LVEDVI) was not different. LV systolic pressure and heart rate were also similar between the two groups. Therefore, loading condition at measurements did not seem to affect the results of LV function.

However, profound changes in loading condition may occur during apnoea in patients with OSA. Both the heart rate and blood pressure rise during apnoea, and rise further at the termination of apnoea.5 Strongly negative intrathoracic pressure swings due to upper airway obstruction results in an increase in systolic LV transmural pressure as an index of LV afterload.6 Acute increase in afterload can reduce LV systolic function.29,30 CPAP treatment which significantly reduces the LV afterload is effective in the improvement of LVEF in OSA patients with dilated cardiomyopathy.13 In addition, an abrupt fall in negative intrathoracic pressure may increase venous return and dilate right ventricular cavity and, consequently, disturb LV filling. In fact, it was recently reported that OSA patients without an overt cardiac disease showed right ventricular enlargement and elevated pulmonary artery pressure compared with control obese patients.31 This finding seems to indicate that preload reserve is decreased in patients with OSA. Acute increase of afterload during apnoea under the limited preload reserve may be a key mechanism for the decreased recovery of LV systolic function in our OSA patients.

As the apnoea episode progresses, sympathetic nerve activity increases and reaches its peak level at the end of the apnoea.7 Sympathetic nerve activity is also very high during daytime in sleep apneic patients. On the basis of the proven clinical benefits of β-blockers and ACE-inhibitors for treating patients with AMI,32,33 it is highly probable that the increases in LV afterload and high sympathetic nerve activity have an adverse effect on the functional improvement of the reperfused myocardium. Recurrent increases in LV afterload during sleep and systemic and coronary vasoconstriction due to increased sympathetic nerve activity may induce an imbalance between myocardial oxygen demand and the supply of oxygen and consequently inhibit the recovery of LV systolic function in patients with OSA. In addition, repetitive myocardial ischaemia caused by the fall in SaO2 associated with apnoea may further deteriorate LV function.


OSA showed deteriorative effects on the recovery of LV function in patients with AMI. The largest improvements in LV function occurred in the 14 days after admission.34 If patients with OSA do not receive treatment such as CPAP, it is unlikely that LV function in patients with OSA can improve to the levels in control patients. Early diagnosis of OSA seems to be essential because CPAP may attenuate the adverse effects of OSA. However, patients with AMI in the presence or absence of OSA frequently showed symptoms relating to sleep disorder. Patients with OSA did not present with increased rates of obesity because the BMI was not significantly different between the two groups. It is known that the majority of Asian patients with OSA are non-obese.35 We would like to recommend that patients with AMI undergo overnight polysomnography because the incidence of OSA is high among patients with AMI and it is unlikely that physical examination and patient-history taking alone can identify high-risk patients for OSA in a Japanese population.


It is probable that the area at risk was comparable between the two groups because the distribution of IRA and the initial extent of severely depressed wall motion within infarct area were comparable. We speculate that poorer recovery of LV function in patients with OSA may be owing to the greater extent of stunned myocardium because the peak CK value was comparable between the two groups, and this value, at least in part, was related to the infarct size. However, we did not assess the myocardial viability study using modalities such as single-photon emission computed tomography and could not determine the precise infarct size. There is a possibility that more myocardial necrosis may occur in patients with OSA.

This study was not designed to evaluate the effects of CPAP. Our nine of 17 patients who used CPAP and 13 of 20 patients who did not receive any specific treatment for OSA underwent follow-up left ventriculogram at 9 months after discharge. The better improvement of LVEF during follow-up period was noted in patients treated by CPAP compared with non-treatment patients with OSA. But this difference did not reach a statistical significance (8.7±11.6 vs. 0.7±11.6%, P=0.225). Another study may be needed to evaluate whether CPAP provides beneficial effects on LV function in OSA patients with AMI treated by primary PCI.

Conflict of interest: none declared.


View Abstract