Article Text

Download PDFPDF

Original article
Obstructive sleep apnoea is associated with myocardial injury in patients with refractory angina
  1. Glaucylara R Geovanini1,
  2. Alexandre C Pereira2,
  3. Luis H W Gowdak3,
  4. Luciana Oliveira Cascaes Dourado3,
  5. Nilson T Poppi3,
  6. Gabriela Venturini2,
  7. Luciano F Drager1,4,
  8. Geraldo Lorenzi-Filho1
  1. 1Sleep Laboratory, Pulmonary Division, Heart Institute (InCor), University of São Paulo (USP), Sao Paulo, Brazil
  2. 2Laboratory of Genetics and Molecular Cardiology, InCor, USP, Sao Paulo, Brazil
  3. 3Refractory Angina Research Group, InCor, USP, Sao Paulo, Brazil
  4. 4Hypertension Unit, InCor, USP, Sao Paulo, Brazil
  1. Correspondence to Dr Geraldo Lorenzi-Filho, Director of Sleep Laboratory, Pulmonary Division, Heart Institute (InCor), University of São Paulo, Medical School, Av Dr Enéas Carvalho de Aguiar, 44—Laboratório do Sono, São Paulo 05403-000, Brazil; geraldo.lorenzi{at}


Objective To investigate the association between obstructive sleep apnoea (OSA) severity with markers of overnight myocardial injury in patients with refractory angina.

Methods Patients with refractory angina were characterised clinically and they underwent ischaemia imaging stress tests by single-photon emission computed tomography (SPECT) and/or cardiac MRI. The patients were admitted to the hospital, remained under resting conditions for blood determination of high-sensitivity cardiac troponin T (hs-cTnT) at 14:00, 22:00 and after overnight polysomnography at 7:00.

Results We studied 80 consecutive patients (age: 62±10 years; male: 66%; body mass index (BMI): 29.5±4 kg/m2) with well-established diagnosis of refractory angina. The mean apnoea–hypopnoea index (AHI) was 37±29 events/h and OSA (AHI >15 events/h) was present in 75% of the population. Morning detectable hs-cTnT and above 99th percentile was present in 88% and 36% of the population, respectively. Patients in the first to third quartiles of OSA severity did not have circadian variation of hs-cTnT. In contrast, patients in the fourth quartile (AHI ≥51 events/h) had a circadian variation of hs-cTnT with a morning peak of hs-cTnT that was two times higher than that in the remaining population (p=0.02). The highest quartile of OSA severity remained associated with the highest quartile of hs-cTnT (p=0.028) in multivariate analysis.

Conclusion Very severe OSA is common and independently associated with overnight myocardial injury in patients with refractory angina.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Video abstract


Refractory angina (RA) is characterised by recurrent symptoms of angina that do not respond to conventional treatment in patients with advanced chronic coronary artery disease (CAD) who are unsuitable for further interventions.1 These patients are not candidates for further revascularisation, because of inadequate coronary anatomy, such as a lack of graft conduits, multiple coronary restenosis, poor distal targets and severe diffuse atherosclerosis.1 ,2 This population is considered a ‘no option’ group, because they complain of angina triggered by routine daily activities despite optimal treatment. Therefore, the search for unrecognised comorbid that may contribute to ischaemia is extremely important among patients with RA.

Obstructive sleep apnoea (OSA) is characterised by repetitive episodes of apnoeas or hypopneas during sleep, promoting exaggerated negative intrathoracic pressure, arousals from sleep and intermittent hypoxia.3 OSA triggers a cascade of deleterious effects to the cardiovascular system, including sympathetic activation, oxidative stress, systemic inflammation and endothelial dysfunction.4 OSA has been independently associated with atherosclerosis5 and worse cardiovascular outcomes.6 OSA is common in the general population and is extremely frequent among patients with cardiovascular disease.7–11 For instance, the prevalence of OSA among patients with CAD ranges from 30% to 54%.12 ,13 We have recently shown that among consecutive patients with an established diagnosis of RA, OSA was present in 73% of the population.14 Because patients with RA have an imbalance between supply and consumption of oxygen, we hypothesised that severe OSA triggers overnight myocardial injury. To this end, the circadian variation of high-sensitivity cardiac troponin T (hs-cTnT), a well-established biomarker of myocardial injury, was measured in a cohort of patients with RA.


Study population

From August 2011 to February 2014, we evaluated consecutive patients with a well-stablished diagnose of RA referred to a specialised outpatient clinic from the Heart Institute (InCor), University of Sao Paulo Medical School. RA was defined according to the current guidelines as a chronic condition (>3 months) characterised by angina due to coronary insufficiency in the setting of CAD, which cannot be controlled by a combination of medical therapy, percutaneous coronary intervention (PCI) and coronary artery bypass grafting (CABG) surgery.15 ,16 We excluded patients with Canadian Cardiology Society (CCS) angina symptom class <grade 2, previous stroke with disability or an unstable clinical condition. On average, patients had suffered myocardial infarction with an interval varying from 5 months up to 33 years (mean±SD 121±95 months). All patients read and signed an informed consent. The study was approved by the local Institutional Review Board (protocol number: 0137/11).

Sleep study

All subjects underwent full polysomnography (EMBLA; Flaga hf. Medical Devices; Reykjavik, Iceland) performed at the Sleep Laboratory. Sleep staging and OSA diagnosis were scored using standard Task Force American Academy.17 Hypopnoea was defined as a 50% decrement in airflow lasting ≥10 s associated with oxygen desaturation of 3% or with arousal. Apnoea was defined as cessation of airflow for ≥10 s (whether central, obstructive or mixed). Due to an expected high pretest probability, OSA was defined according to an apnoea–hypopnoea index (AHI) of >15 events/h. Because of an anticipated high prevalence of OSA in patients with RA14 we divided this population according to AHI quartiles. No patient was treated before or after the sleep study. This population is linked to our public health system and unfortunately has no access to OSA standard treatment. We are currently applying for a grant in order to determine the impact of OSA treatment in this population.

Ischaemia imaging tests

Single-photon emission computed tomography (SPECT) and/or cardiac MRI, accepted methods for ischaemia evaluation according to current guidelines,18 ,19 were used. Briefly, SPECT imaging was determined using an intravenous radiotracer (Technetium-Sestamibi (99Tc-Setamibi)) before (rest) and after pharmacological stress (dipyridamole or adenosine). Cardiac MRI was performed by intravenous injection of contrast (gadolinium), and images at rest and stress induced by vasodilator (dipyridamole) were compared. The number of ischaemic segments was determined by evaluation of the quality of opacification in each of the walls of the heart, the magnetic resonance score (MR score) that represents the ischaemic burden in the MRI.

hs-cTnT blood measurement

EDTA-treated plasma samples were obtained and stored at −80°C. The hs-cTnT was determined by means of an auto-analyser (Roche Diagnostics). This is the fifth generation Elecsys troponin T assay that enables the measurement of concentrations well below those detectable by conventional assays. The detection limit of the assay is ≥5 ng/L and the value at the 99th percentile is ≥14 ng/L.20 ,21

Study design

To ensure a controlled environment and resting conditions, the patients were admitted to the hospital on the day before the overnight polysomnography. All patients underwent a clinical history, physical examination and answered the Epworth Sleepiness Scale questionnaire. A score >10 points was used to define excessive daytime sleepiness (EDS).22 Nocturnal angina was defined by the presence of at least one self-reported episode of angina over the past month that awoke the patient during sleep. Demographics, blood pressure (BP), heart rate (HR), waist and neck circumferences and body mass index (BMI) were evaluated. The average of the last two arterial BP measures was used. Blood sample for analysis of lipid profile, renal function and glucose levels were also determined and glomerular filtration rate (creatinine clearance) was estimated by the Cockcroft–Gault formula. The blood samples for determination of hs-cTnT were drawn at 14:00, 22:00 and on the following morning at 7:00 after polysomnography.

Statistical analysis

Continuous comparisons were analysed using the Student's t test or the Mann–Whitney U test, when appropriate. Categorical comparisons were analysed by using the χ2 or Fisher's exact test. Due to the expected high frequency of OSA as well as of detectable hs-cTnT, the population was divided according to AHI quartiles and hs-cTnT quartiles and analysis of variance and or Kruskal–Wallis were used for further analysis. The circadian variation of troponin among patients in the fourth quartile was compared with the remaining population in the lower AHI quartiles (first to third quartiles). The statistical analysis was performed based on distribution by the logistic regression model and Wald's test. The general linear model was used to determine whether the means (hs-cTnT levels at 14:00, 22:00 and 7:00) of two groups (OSA fourth quartile × remaining OSA quartiles) differ. Variables with p<0.1 on univariate analysis were entered into a multivariate analysis. Multiple logistic regression analysis was used to determine the characteristics independently associated with OSA severity and with hs-cTnT (quartile distribution). The tests were performed with a significance level of 5%. Data were analysed with SPSS (V.20.0; SPSS, Chicago, Illinois, USA) statistical software.


Patient characteristics

We initially evaluated 89 patients with RA, but 9 were excluded for the following reasons: refused to participate (n=3), had angina CCS <grade 2 (n=3), had stroke disability (n=1), had unstable angina (n=1) or was engaged in another study (n=1). Therefore, the final sample consisted of 80 patients. A previous history of at least one intervention (CABG surgery or PCI) was present in 94% of the population. A total of 60 patients (75%) had OSA (AHI>15 events/h). The demographic and clinical characteristics of the entire population divided according to AHI quartiles are presented in table 1. The patients were heavily medicated and the majority was not currently smoking, had normal ejection fraction and had resting HR and BP well controlled. As compared with patients with no OSA (≤15 events/h), patients with OSA were more obese (BMI: 30.5±4 vs 27.5±4 kg/m2, p=0.01), had a higher proportion of hypertension (98 vs 80%, p=0.01) and diabetes mellitus (68 vs 40%, p=0.02), respectively. The average levels of lipid profile, fasting glucose and other blood analyses were similar in patients according to OSA quartiles. Urea and creatinine was higher in patients in the fourth OSA quartile (see online supplement 1). Nocturnal angina was common in the study population (58%) and similar among patients with and without OSA (55 vs 65%, p=0.43) as well as according to OSA severity quartiles (table 1). The entire population had low sleep efficiency, poor quality of sleep shown by interrupted sleep (arousals and awakenings) and superficial sleep (low percentage of slow-wave sleep stage). On average, the population presented with EDS, but no difference in patients with and without OSA was observed (see online supplement 2).

Table 1

Clinical and demographic characteristics of the entire population with refractory angina as well as divided according to AHI quartile

Supplemental material

Supplemental material

Ischaemia imaging tests

All patients but one (who had claustrophobia) (n=79) underwent ischaemia evaluation by at least one imaging method: MRI or SPECT. If patients were positive in at least one of these tests, we classified them as ischaemia-positive. Only two patients had different results (positive in SPECT and negative in MRI) nevertheless, because SPECT test were more recent than MRI, so we considered them as ischaemia-positive. The number of patients underwent either only SPECT (n=21), only MRI (n=13) or both imaging tests (n=45). Of the 79 patients evaluated for ischaemia, only 6 patients (7.6%) were negative for ischaemia (two by MRI, one by SPECT and three by both methods). The frequency of positive test evaluated by SPECT or MRI was greater in patients with OSA than in those without OSA (98% vs 73%, respectively, p=0.003). The presence of ischaemia remained associated with OSA (AHI >15 events/h), in a logistic regression analysis, after adjustment for gender, age and BMI (OR: 35.15, CI 2.86 to 432.44, p=0.005).This association between ischaemia by imaging tests and OSA was also demonstrated when the patients were divided by AHI quartile (table 1). The number of ischaemic segments (MR score) was not statistically significantly different, according AHI quartile (table 1). To illustrate, according to ischaemia territory distribution: 51% (front wall), 48% (lower wall) and 1% (lateral wall). The majority of ischaemia was distributed around infarct territory. There was no difference among AHI quartiles according to ischaemia distribution (p=0.179 to SPECT and p=0.952 to MRI results).

hs-cTnT measurement

All patients included in the study had blood samples collected for the circadian variation of hs-cTnT. Patients in the highest AHI quartile (≥51 events/h) had significantly higher morning levels of hs-cTnT (figure 1). Morning detectable limit of hs-cTnT (≥5 ng/L) was present in the vast majority of the entire population (88%) (table 1). The proportion of patients with morning detectable limit was higher in the fourth AHI quartile comparing with the remaining population but without statistical significance (95% vs 85%, p=0.437) and also the fourth AHI quartile had higher levels of morning 99th percentile (hs-cTnT ≥14 ng/L), significantly higher in the fourth AHI quartile than in the remaining population (figure 2). Conversely, hs-cTnT was relatively stable during the 24 h period in patients in the lowest OSA quartiles severity. In contrast, there was a circadian variation of hs-cTnT among patients in the fourth AHI quartile. Patients in the fourth AHI quartile had a morning peak of hs-cTnT (figure 3). Because the levels of morning hs-cTnT were on average very high in this population, we also divided the population according to hs-cTnT quartile distribution in order to cross-check the variables associated with morning peak of hs-cTnT. Patients on the fourth AHI quartile (≥51 events/h) were the only variable associated with the highest hs-cTnT quartile (≥18 ng/L) in univariate (table 2) and multivariate analyses (table 3), after adjusted for OSA traditional confounders (age, male gender and obesity).

Table 2

Variables associated with quartile distribution of morning hs-cTnT (7:00)

Table 3

Variables independently associated with the highest quartile of morning (at 7:00) hs-cTnT levels in a logistic regression analysis

Figure 1

Mean level of high-sensitivity cardiac troponin T (hs-cTnT) at 7:00 is shown in each quartile distribution of obstructive sleep apnoea by apnoea–hypopnoea index (AHI).

Figure 2

Comparison of percentage of patients with positive morning high-sensitivity cardiac troponin T (hs-cTnT) levels in patients in the fourth quartile of obstructive sleep apnoea versus the remaining population (≥5 and ≥14 ng/L, respectively). AHI, apnoea–hypopnoea index; p99, above percentile 99.

Figure 3

Circadian variation of high-sensitivity cardiac troponin T (hs-cTnT) levels is shown among patients in the fourth apnoea–hypopnoea index (AHI) quartile (≥51 events/h) compared with patients in the first to third AHI quartiles (<51 events/h).


Our study confirms previous findings showing that OSA is extremely common and severe in patients with RA.14 We extend this finding by showing evidence that 25% of patients in the most spectrum of OSA severity (AHI ≥51 events/h) are exposed to overnight subclinical myocardial damage as evidenced by the following observations: (1) patients in the fourth OSA quartile had a mean morning hs-cTnT that was twofold greater than the mean levels of the population in lower OSA severity (figure 1); (2) patients in the fourth OSA quartile had a frequency of morning hs-cTnT ≥14 ng/L that was two times higher than patients in the remaining OSA quartiles (figure 2); (3) hs-cTnT was stable during the 24 h period in patients in the lower quartiles of OSA severity; in contrast, there was a circadian variation among patients in the fourth OSA quartile with a morning hs-cTnT peak (figure 3) and (4) the highest quartile of hs-cTnT was independently associated with the highest quartile of OSA after adjustment for age, gender and BMI (table 3). Finally, a high proportion of the population with RA presented with nocturnal angina symptom (58%), which could be a plausible explanation of the reason of OSA not being associated with nocturnal angina symptom (p=0.563). Taken together, the present results suggest that a significant proportion of patients with RA have very severe OSA and experience subclinical overnight myocardial injury.

The potential role of OSA in promoting myocardial injury has been previously investigated and has shown conflicting results. One study that did not use hs-cTnT assays found no correlation between OSA and overnight myocardial injury in men with CAD.23 In contrast, recent studies that used a high-sensitivity assay of a newer generation of troponin that brings higher sensitivity for diagnosing myocardial damage and detects even microscopic zones of myocyte cells found that OSA is associated with elevated morning hs-cTnT24 ,25 as well as with troponin I.26 Our study is therefore in line with recent evidence but extends these previous findings in several ways. The previous studies24–26 evaluated the general population with no overt CAD and correlated OSA with low limit detection levels of troponin. The clinical significance of these previous findings is controversial. In contrast, we evaluated a high-risk population for CAD, homogeneous population well-established diagnosis of RA and we adopted the hs-cTnT cut-off of ≥14 ng/L that represents the value above at 99th percentile and may represent a better precision cut-off to detect myocardial injury.20 ,21 The proportion of patients with morning hs-cTnT (≥14 ng/L) was extremely high in the entire population (36%). Values of cardiac troponin exceeding the 99th percentile of a normal reference population are recommended for the diagnosis of myocardial infarction.27 Our findings are compatible with the fact that a CAD high-risk population was studied. More importantly, the proportion of elevated troponin levels above 99th percentile was twofold greater among patients in the highest quartile of OSA severity than that in the remaining population (figure 2). In contrast to previous studies24–26 that evaluated troponin in one time point, we investigated the circadian variation of hs-cTnT under well-controlled conditions. One novel finding was that the circadian variation of hs-cTnT was unique in patients in the highest quartile of OSA severity (AHI ≥51 events/h). In contrast to the patients in the lower OSA quartiles, who had relatively stable hs-cTnT during the 24 h cycle, those in the highest quartile of OSA severity had a clear circadian variation, with a morning peak of hs-cTnT (figure 3). This observation is, therefore, consistent with the hypothesis that OSA severity may contribute to an imbalance between oxygen supply and demand during the night that will ultimately be translated into overnight myocardial damage in these patients with advanced CAD.

The clustering between the most severe forms of underling cardiovascular disease and OSA has been found in patients with hypertension,7 ,8 ,10 metabolic syndrome4 ,9 and CAD.12–14 In line with these previous observations, we found a higher proportion of patients with myocardial ischaemia by imaging tests among patients with OSA than in patients without OSA (98% vs 73%, p=0.003, respectively) as well as among those in the highest OSA quartile severity as compared with patients in the first OSA severity quartile (table 1). Moreover, the presence of ischaemia by imaging test remained associated with the highest OSA quartile in a logistic regression analysis. This observation is in line with the hypothesis that OSA may contribute to the progression of coronary atherosclerotic burden.5 ,28 These findings should, however, be interpreted with caution. The absence of positive ischaemia on imaging tests in a small proportion of the studied patients with well-characterised RA may be due to imaging test limitations.

The absence of typical traits may help in explaining why the association between OSA and RA has so far been overlooked in the literature. Our study has shown no clear clinical correlates between OSA, poor objective sleep quality and symptoms of EDS (see online supplement 2). This observation is in line with previous studies that found little or no association between EDS and OSA in consecutive patients with established cardiovascular disease, such as hypertension7 and metabolic syndrome,9 for instance. In contrast to the typical patients with OSA and other cardiovascular diseases, our patients with RA frequently had complaints of EDS. However, the Epworth Sleepiness Scale of patients with OSA and patients without OSA was similar (see online supplement 2). Therefore, the symptoms of EDS are not associated with OSA. The high frequency of EDS may be due to a high prevalence of depression in this population, probably due to a low quality of life caused by incapacitating angina.14 Another important finding is that the symptoms of nocturnal angina were similar in patients with OSA and patients without OSA (table 1). We speculate that the association is blurred, because all patients had severe CAD, and more than half of the population (58%) presented with nocturnal angina. Taken together, our study suggests that patients with RA should be systematically investigated for OSA due to the high pretest probability, the low correlation with clinical symptoms and the potential implications of diagnosing occult OSA in RA population. Potential underdiagnoses of OSA have also been described in the cardiology outpatient setting.29

Study limitations

This study has some limitations. On the one hand, this is a cross-sectional study and the associations must be interpreted with caution. On the other hand, we used a well-established marker of myocardial injury and the morning levels of hs-cTnT found in the patients in the fourth OSA quartile are frequently used for the diagnosis of myocardial infarction. Moreover, the independent association between the fourth quartile of hs-cTnT and the fourth OSA quartile is also reassuring. Another caveat of our study is that the morning peak hs-cTnT was found among patients in the fourth OSA quartile. Therefore, our data allow the conclusion that only patients with extremely severe OSA (AHI≥51 events/h) experience overnight myocardial injury (figure 3). Nonetheless, the morning hs-cTnT levels of patients in the third AHI quartile that corresponds to the level used to define severe OSA (AHI>30 events/h) was slightly higher than in the lower AHI quartiles but did not reach statistical difference. Another limitation was the time interval between imaging tests and sleep study. The interval was variable (23±22 months for SPECT and 18±17 months for MRI). However, since RA is characterised by recurrent symptoms of angina that do not respond to conventional treatment, imaging tests for ischaemia are not required for diagnosis. We showed these results to better characterise our study population. The main objective of the present study was to show the association between OSA and overnight myocardial injury.


Therefore, our study indicates that one-fourth of the RA population have OSA severe enough to damage their heart during the night. OSA triggers several pathways that are deleterious to the cardiovascular system and intermittent hypoxia is certainly the main plausible mechanism. However, our study does not allow any further insight about the exact mechanisms linking severe OSA with overnight myocardial injury. The treatment of OSA with continuous positive airway pressure has been shown to be beneficial in other settings, such as hypertension, early markers of atherosclerosis and future cardiovascular events.6 ,30 Therefore, future studies are necessary to investigate the impact of the treatment of OSA with continuous positive airway pressure in the population of patients with RA.

Key messages

What is already known on this subject?

  • Consistence evidence suggests that obstructive sleep apnoea (OSA) is prevalent in refractory angina population.

What might this study add?

  • Very severe OSA is common and independently associated with overnight myocardial injury in patients with refractory angina.

How might this impact on clinical practice?

  • Exploring potential pathways and improving recognition of OSA in coronary heart disease population might have benefits in the reduction of cardiovascular morbidity and mortality.



  • Twitter Follow Glaucylara Geovanini at @gal.geovanini

  • Contributors Conception and design: GRG, ACP, LHWG, LFD and GL-F. Analysis and Interpretation: GRG, ACP, LHWG, LOCD, NTP, GV, LFD and GL-F. Drafting the manuscript for important intellectual content: GRG, ACP, LFD and GL-F.

  • Funding This manuscript is supported by Fundação de Amparo à Pesquisa do Estado de São Paulo and Fundação Zerbini, Brazil.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Heart Institute Ethics Committee (protocol number: 0137/11). GRG, GL-F and co-authors.

  • Provenance and peer review Not commissioned; externally peer reviewed.