|Home | About | Journals | Submit | Contact Us | Français|
Patients with chronic thromboembolic pulmonary hypertension (CTEPH) often have substantial right ventricular dysfunction. The resulting low cardiac index might predispose to sleep disordered breathing (SDB) by increasing ventilatory instability. The prevalence of SDB and potential association with impaired cardiac index was examined in patients with CTEPH. Patients referred for evaluation for pulmonary thromboendarterectomy surgery were recruited. Subjects underwent a sleep study, unless already using positive airway pressure (PAP) therapy. Hemodynamic data were obtained from contemporaneous right heart catheterization. A total of 49 subjects were included. SDB – defined as ongoing PAP use or apnea-hypopnea index (AHI) ≥ 5/h – was found in 57% of subjects. SDB was generally mild in severity, with respiratory events mainly consisting of hypopneas. Cardiac index was found to be significantly lower in subjects with SDB than those without (2.19 vs. 2.55 L/min/m2; P=0.024), while no differences were observed in other characteristics. Additionally, cardiac index was independently predictive of AHI. In a subgroup of subjects with an elevated percentage of central events, both cardiac index and lung to finger circulation time correlated with AHI. In conclusion, SDB is prevalent in patients with CTEPH, and might decrease with treatments that improve cardiac index.
Patients with chronic thromboembolic pulmonary hypertension (CTEPH) have variable degrees of right ventricular dysfunction due to chronic blood clots and remodeling of the pulmonary vasculature leading to pulmonary hypertension. CTEPH patients may therefore have impaired cardiac index, but typically lack elevated left heart pressures. Sleep disordered breathing (SDB) is prevalent in patients with left heart failure, with prior research suggesting this is due to elevated left heart filling pressures leading to increased chemosensitivity.1–3 However, the independent effect of low cardiac output might additionally predispose to SDB, and therefore CTEPH patients may also be at risk.4–7 To our knowledge, there is no dedicated study to examining SDB in CTEPH.8–13 The aims of this study are:
Patients referred for evaluation for pulmonary thromboendarterectomy surgery were sequentially recruited for participation. Exclusion criterion was an expert clinician assessment that the etiology of pulmonary hypertension was not CTEPH. The institutional review board at the University of California San Diego approved the protocol. Written, informed consent was obtained from all subjects.
Enrolled participants underwent sleep testing with an ApneaLink Plus (ResMed, San Diego, CA), consisting of nasal pressure sensor, respiratory effort band, and pulse oximeter worn on the finger. For safety of participants and consistency in baseline oxygenation, subjects on nocturnal supplemental oxygen were maintained on their prescribed setting, with the sensor placed above the oxygen cannula. Similarly, for safety of participants, subjects with a prior diagnosis of SDB actively using positive airway pressure (PAP) therapy were regarded as having SDB; no sleep study was performed.
Studies were scored by a blinded, registered polysomnographic technologist using modified American Academy of Sleep Medicine Chicago Criteria.14 Specifically, apnea was defined as a >90% reduction in airflow, and hypopnea defined as a >30% reduction in airflow, lasting greater than 10 seconds, and associated with a ≥3% oxyhemoglobin desaturation. Obstructive apneas were associated with thoracic effort and/or associated flow limitation. Hypopneas were scored as undifferentiated events.
Subjects underwent right heart catheterization as part of a standard clinical evaluation. Briefly, patients were brought into the catheterization laboratory and were measured while resting supine. A Swan-Ganz catheter was advanced to the pulmonary artery under fluoroscopic guidance. With appropriate respiratory timing, right sided pressures were transduced hydrostatically. Cardiac output was measured by thermodilution of room temperature saline.
Lung to finger circulation time (LFCT) has been previously described as a measure of circulatory delay in patients with SDB.15, 16 Flow and saturation waveforms from sleep study data were analyzed in Spike2 software (CED, Cambridge, UK). In a blinded fashion, all recordings were screened for scorable events, defined as having a distinct end of hypopnea/apnea with an associated desaturation of ≥3%. LFCT was measured from the end of a respiratory event to the onset of resaturation. 10 events from the start and 10 events from the end of the recording were selected. If fewer than 20 events were available, all events were used; however, subjects with less than four events were excluded from analysis. The mean LFCT for each subject was reported.
Analysis was performed using SigmaPlot (Systat Software Inc, San Jose, CA) and SAS Studio (SAS Institute, Cary, NC). Means were compared using a t-test or rank sum test, and proportions compared using a chi-squared test. Pearsons product-moment or Spearman rank-order correlation was used as appropriate. Linear regression and logistic regression were performed with variables of interest. A p-value of <0.05 was considered statistically significant.
Of 75 patients screened for inclusion, 19 declined, three did not have CTEPH, two had missing sleep study data, and two did not undergo cardiac catheterization. Thus, complete data were obtained in 49 subjects, all of whom had a diagnosis of CTEPH following clinical and radiographic evaluation by expert clinician consensus. 45 subjects subsequently elected to undergo pulmonary thromboendarterectomy surgery, allowing for pathological confirmation of CTEPH.
The prevalence of SDB amongst 49 total subjects is reported in Table 1. Using a definition of SDB as either PAP use at enrollment or an apnea-hypopnea index (AHI) ≥ 5/h, the prevalence of SDB was 57%. Characterization of SDB found on sleep studies is reported in Table 2, revealing primarily mild SDB with events consisting primarily of hypopneas. There were no differences with respect to age, gender, BMI, or sedative use between those with and without SDB (Table 1).
Of the 39 subjects not using PAP, 44% were using supplemental oxygen upon enrollment. The percentage of subjects using oxygen did not differ between those with and without SDB on sleep studies (Table 2). There was no significant difference in the AHI between those using oxygen and those who were not (6 versus 13 events/h; P=0.229). Similarly, no significant difference was found between the two groups with respect to baseline saturation (94% versus 92%; P=0.141), nadir saturation (83% versus 79%; P=0.402), or percentage time spent with saturation <90% (38% versus 52%; P=0.308), although numeric differences are noted.
Compared to 18 subjects with an AHI ≥ 5/h on sleep testing, the 10 subjects using PAP at enrollment were found to be older (50 versus 63 years; P = 0.016); otherwise there were no significant differences with respect to gender, body mass index (BMI), Epworth sleepiness score, or hemodynamic data (data not shown).
All 49 subjects underwent right heart catheterization. Cardiac index (i.e. cardiac output divided by estimated body surface area) was significantly lower in those with SDB compared to those without (Table 1). No differences were observed in cardiac output, right atrial (RA) pressure, pulmonary vascular resistance, or pulmonary artery wedge pressure.
Thirty-nine subjects underwent sleep testing, which was performed within 2 days of catheterization in 85% of subjects. Linear regression was performed to examine the effect of cardiac index on AHI, with natural log transformation applied to AHI in order to satisfy the normality assumption, which revealed a significant association (Figure 1). BMI and age were not found to correlate with log AHI (P=0.133 and P=0.648 respectively), nor was baseline saturation (P=0.610). In 24 subjects in whom LFCT was scorable, there was a significant association between cardiac index and LFCT (Figure 2A), using an inverse first order model (i.e. y=y0+a/x), determined a priori based on an expected hyperbolic relationship. Despite association between LFCT and cardiac index, and between cardiac index and log AHI, no significant correlation was found between LFCT and log AHI (P=0.129). In addition, there was no significant difference in LFCT between those with AHI < 5/h and those with AHI ≥ 5/h (Figure 2B).
In subjects with a highly collapsible upper airway, SDB might be expected regardless of respiratory control stability; therefore, an a priori subgroup analysis was performed on subjects with a percentage of events that were central apneas ≥1% (i.e. ≥ the cohort median), which we believed were likely to have only a mildly collapsible airway. A linear relationship between cardiac index and log AHI in this subgroup [N=20] was found (Figure 3A), and the mean cardiac index was lower in those with AHI ≥ 5/h compared to those with AHI < 5/h (2.18 vs 2.55 L/min/m2; P=0.036). Additionally, there was a significant association between LFCT [N=14] and log AHI (β=0.117, R2=0.310; P=0.039). Within the subgroup, LFCT was significantly shorter in subjects with AHI < 5/h compared to those with AHI ≥ 5/h (Figure 3B).
Multiple linear regression was performed with variables known or suspected to account for severity of SDB (age, BMI, baseline saturation, RA pressure, pulmonary artery wedge pressure, and cardiac index); only BMI (P=0.050) and cardiac index (P=0.028) were significantly associated with log AHI. In all 49 subjects, the probability of SDB was analyzed using multiple logistic regression with covariates of age, BMI, RA pressure, pulmonary artery wedge pressure, and cardiac index. The only significant predictor of SDB in this model was cardiac index (P=0.046).
The key findings of this study are: 1) SDB is highly prevalent among patients with CTEPH, and 2) Cardiac index strongly predicted the presence and severity of SDB.
These results suggest that the prevalence of SDB is higher in CTEPH patients than in the general population. Most apneas were obstructive, consistent with obstructive sleep apnea, although the majority of events were hypopneas, which cannot reliably be distinguished on portable testing, and therefore estimates of the prevalence of obstructive versus central sleep apnea cannot be made. Nonetheless, the major focus of this study is the potential contribution of low cardiac index to unstable breathing, which may be relevant in SDB of both obstructive and central nature.
Subjects with SDB had a significantly lower cardiac index compared to those without, and an inverse correlation between cardiac index and AHI was observed. This relationship does not appear to be due to confounding by body mass index, which was shown to have no correlation with cardiac index, as is consistent with prior studies.17 As the circulatory delay is the postulated mechanism by which right ventricular impairment might lead to SDB, cardiac index was utilized, rather than cardiac output. Cardiac index would be expected to reflect more accurately the time it takes blood to travel to from the lungs to carotid and/or central chemoreceptors, as both cardiac index and time are a function of velocity (i.e. output) and distance (i.e. body size). This assertion is further supported by the strong correlation observed between cardiac index and LFCT. The effect of cardiac index on SDB does not appear to be the result of confounding by fluid overload and rostral fluid shifts, as no difference was observed in supine RA pressure, although measures of fluid redistribution were not available.18 In addition, the association between cardiac index and SDB remained significant in multiple regression models that accounted for other known causes of SDB. Therefore, these results suggest that low cardiac index due to right ventricular impairment contributes to SDB.
While LFCT was well correlated with cardiac index, it did not appear to predict SDB when examined in all subjects. The most probable explanation is the imprecision in LFCT measurements and the potential influence of factors other than cardiac index, such as pulmonary function and signal noise or averaging. A trend towards correlation between LFCT and AHI was observed, so a larger sample size might have overcome the large variance in LFCT to reach significance. Nonetheless, further refining LFCT or an alternative non-invasive technique to identify those with prolonged circulatory delay as a contributing factor to SDB would likely have clinical utility.
CTEPH patients were observed to have a low baseline oxygen saturation, which likely contributed to the high prevalence of SDB. Patients with CTEPH have considerable ventilation/perfusion mismatching.19 As a result, the oxyhemoglobin saturation is on the steeper part of the dissociation curve, whereby relatively small perturbations in ventilation can result in desaturation. Therefore, SDB may be ‘overdiagnosed’ in these patients relative to those with normal oxygenation, given that desaturation is part of the criteria for respiratory events. On the other hand, hypoxemia itself has been postulated to contribute to chemosensitivity and ventilatory instability.20 There was no correlation between baseline saturation and AHI, which suggests that differences in ventilatory instability, not simply propensity to desaturate, accounts for the observed variability of SDB.
Importantly, the correlation between circulatory delay (both by cardiac index and LFCT) and SDB was more pronounced when examining the subgroup of patients with a higher proportion of central apneas. This finding supports the concept of ‘effect modification’, in which the importance of ventilatory instability in causing SDB depends on the upper airway anatomy and function.21 Thus, right ventricular dysfunction and resultant circulatory delay is likely most relevant to the pathogenesis of SDB in certain patients. Mechanisms by which increased circulatory delay might contribute to SDB in these patients include larger negative feedback oscillations, predisposition to upper airway collapse, and ventilatory overshoot leading to arousals.22–24 Full polysomnography and physiological characterization were not possible in this referral patient group, but these pilot findings appear to justify further investigation.
The consequences of SDB in CTEPH are unknown. Evidence is beginning to clarify the importance of specific manifestations of SDB, such as hypoxemia, arousals, or negative intrathoracic pressure, on outcomes such as diabetes and hypertension, which might impact this population. In addition, the presence of SDB and overall low nocturnal saturation might impact right ventricular function, although the magnitude of this effect remains controversial.25–27
SDB has been associated with a small but increased risk of thromboembolism and therefore could be a risk factor for CTEPH.28 This study is not specifically designed to address this possibility. However, one would expect that more severe sleep apnea would have a higher likelihood of thromboembolism, whereas in this study most subjects were found to have mild SDB. In addition, while SDB might contribute to right ventricular dysfunction, other known SDB risk factors (eg, BMI, male gender, age) were equivalent between those with and without SDB in this study, arguing against this direction of causality.
As a pilot study, there are methodological limitations. Portable sleep testing is potentially less accurate than polysomnography, although generally biases towards underestimation of AHI. Referral bias in this cohort could reduce generalizability of the findings, but since surgical evaluation is broadly indicated, the effect is likely small. Intensive physiological study of SDB was not possible here, but would be needed to more definitively control for all confounding factors. Follow up data showing a reduction in SDB after pulmonary thromboendarterectomy surgery with associated right ventricular function improvement would further validate the findings from this study. Furthermore, understanding whether improving cardiac index and decreasing circulatory delay results in improvement in SDB would have broad clinical relevance.
The authors thank Bradley A. Edwards, PhD and Scott A. Sands, PhD for discussions regarding ventilatory control theory.
Funding Sources: NIH/NHLBI T32 HL098062-05
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.