INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Several groups have reported elevations in daytime pulmonary artery pressure (Ppa) among patients with obstructive sleep apnea (OSA) (1). Most studies have included significant numbers of patients with coexisting lung disease (1, 10) and, not surprisingly, strong associations have been reported between pulmonary hypertension (PH) and lung function abnormalities and/or awake hypoxemia. Little or no association has been found between the severity of sleep-disordered breathing and Ppa. This has led some investigators to conclude that PH does not occur in OSA unless there is coexisting, clinically recognizable, hypoxemic lung disease (i.e., the so-called "overlap" syndrome) (10, 11). However, in some earlier reports patients with OSA were found who had PH but not awake hypoxemia or significant lung function impairment (1, 3, 4). We recently confirmed this finding in a group of patients with OSA who had no history or lung function evidence of respiratory disease (9). Subsequently, Laks and colleagues (8) also identified several patients with OSA and daytime PH who were normoxemic awake.

In our previous study (9), mild PH was found during wakefulness in approximately 40% of patients with OSA selected because they had neither lung nor cardiac disease. The severity of sleep-disordered breathing, lung function, and body mass was the same in patients with and those without PH. Pulmonary hypertensive patients did, however, have more ventilation perfusion inequality and were slightly hypoxemic while at rest, awake. We argued that the level of hypoxemia was unlikely to fully explain the elevation in Ppa. These data, along with the reports of normoxemic, pulmonary hypertensive patients with OSA in other studies (1, 3, 4, 8), suggested that repetitive hypoxemia during sleep may be sufficient in some patients with OSA to lead to structural changes in the pulmonary vascular bed (so-called pulmonary vascular remodeling) sufficient to cause sustained PH. Intermittent hypoxia has been shown to produce pulmonary vascular remodeling in experimental animals (12), but, to our knowledge, it has not been demonstrated in humans. Also, since daytime PH did not seem to relate to the severity of sleep apnea among patients with OSA we speculated (9) that individual differences in pulmonary arterial reactivity to alveolar hypoxia might be important in the development of PH in OSA.

The aims of the present study were: to confirm, using a new sample population, that daytime PH can occur in OSA in the absence of lung disease; to explore the evidence for vascular remodeling in pulmonary hypertensive patients with OSA by comparing the Ppa responses to increased pulmonary blood flow in pulmonary hypertensive and nonhypertensive patients; to determine whether PH in OSA is associated with a heightened pulmonary pressor response to alveolar hypoxia; and to confirm that ventilation-perfusion inequality is associated with PH in patients with OSA.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection

Thirty-two patients (30 male) with OSA were studied at diagnosis and before treatment. All patients satisfied the following selection criteria: OSA defined as the presence on polysomnography of > 10 obstructive or mixed apneas/hypopneas per hour of sleep associated with a history of loud snoring and excessive daytime sleepiness; no clinical or lung function evidence of chronic lung disease (FEV₁ and FVC > 80% predicted, FEV₁/FVC > 75%), or any evidence of cardiovascular disease known to affect pulmonary hemodynamics; and clear and reproducible Doppler and echocardiographic images (see "pulsed Doppler echocardiography" below). Patients with angina or known coronary artery disease were excluded from the study. None of the patients were receiving antihypertensive medication or drugs affecting cardiac or pulmonary function.

Lung Function

The following measurements were made with patients in the sitting position: forced expiratory lung volumes, static lung volumes and single-breath carbon monoxide diffusing capacity (Jaeger Compact Lab system with body plethysmograph; Erich Jaeger BmbH & Co. KG, Würzburg, Germany); closing capacity (CC) by single-breath nitrogen testing using the method of Buist and Ross (13) (dry rolling-seal spirometer, Warren E. Collins, Braintree, MA; X-Y recorder, RW Series Model 83, Rikadenki Kogyo Co. Ltd., Tokyo, Japan; N₂ analyzer Model No. 47302A Vertek series, Hewlett-Packard, Waltham, MA); and resting arterial blood gases (ABL 3 Blood Gas Analyzer; Radiometer, Copenhagen, Denmark) during room-air breathing. The difference between FRC and CC in the sitting position was computed as an indication of the likelihood of small airways closure during resting tidal breathing.

Polysomnography

Sleep recordings were made using a computerized data acquisition system (Sleepwatch; Compumedics, Melbourne, Australia) and displayed on a high resolution 20-inch monitor (NEC multisync 6FG; NEC, Tokyo, Japan). Monitored variables included: EEG, EOGs, and submental EMG (digital sampling speeds: 125 Hz for EEG, EOG, and EMG). A two-lead ECG, pulse oximetry (Criticare 504; Criticare Systems, Inc., Waukesha, WI), leg movements (movement sensors; Compumedics), body position (mercury switch position sensor; Compumedics), and thoracic and abdominal movements (inductance plethysmography) were also recorded (digital sampling speed, 50 Hz). Sleep was scored manually in 30-s epochs using the system of Rechtschaffen and Kales (14). Hypopnea was defined as a 50% or greater decrease from baseline in abdominal and thoracic excursions and/or airflow lasting 10 s or more. Apnea was defined as cessation of airflow for 10 s or more characterized as obstructive, mixed, or central. Sleep-disordered breathing was quantitated by calculating the apnea-hypopnea index (AHI; apneas plus hypopneas per hour of sleep) and various indices of arterial O₂ saturation (Sa_O2) during sleep (lowest Sa_O2 and percent time Sa_O2 < 90%).

Pulmonary Hemodynamics and Cardiac Output

Pulsed Doppler echocardiography (Acuson Computed Sonography system; Acuson, Mountain View, CA) was performed using 2.5 and 3.5 mHz transducers and with the patient at rest in the 30-degree left lateral decubitus position with a 20-degree upper body tilt. The transducer was positioned in the midleft parasternal border for the right ventricular outflow signal and in the apical position for the left ventricular outflow and mitral signals. Standard two-dimensional parasternal short axis and apical five-chamber views were used, respectively. It was necessary to show clearly visible Doppler flow envelopes of the left and right ventricular outflows in each patient before enrolling him or her in the study.

An electrocardiographic signal with 0.04-s marks was displayed with the Doppler signals for event timing purposes. Tracings were recorded on videotape and on a strip-chart recorder at a sweep speed of 100 mm/s with subjects breathholding at end-expiration for at least four cardiac cycles. All measurements were made from the outer borders of the darkest portion of the Doppler flow profiles. Systolic and mean Ppa and cardiac output (CO) were estimated as described below.

Doppler estimation of Ppa. The technique described by Morera and colleagues (15) was used to obtain estimates of Ppa. In brief, at least four beats, preferably consecutive, were analyzed from each interrogated site, and average values were calculated for the following parameters measured from the right and left ventricular outflow tracings: preejection period (PEP), ejection time (ET), and mean acceleration to peak velocity (ACCm; ACCm = peak velocity [pV]/acceleration time [AT]). The empirically derived index "F" was used to compare pressure-related right- and left-sided flow velocity (waveform) characteristics, using the measurements of PEP, ET, and ACCm in terms of their proportionality to pressure: F = PEP × ACCm/ET. "F" was calculated from Doppler trace measurements as: F = (PEP × pV)/(ET × AT).

The "F" index for the right ventricular outflow (FRVO) is proportional to pressure in the pulmonary artery and is described by the equation FRVO = k (Ppa). Similarly the "F" index for the left ventricular outflow (FLVO) is proportional to aortic (systemic) blood pressure (Pao) and is described by the equation FLVO = k (Pao). Therefore, FRVO/FLVO = Ppa/Pao, which was rearranged for the calculation of the Ppa: Ppa = FRVO/FLVO × Pao.

Blood pressure of the right arm at rest was taken at the beginning and the end of each Doppler study with appropriately sized arm cuffs and calibrated mercury sphygomanometer. Average values from three separate measurements of systolic and diastolic blood pressure were used for calculations. Mean systemic blood pressure (SBP) was calculated as one-third systolic plus two-thirds diastolic BP.

Doppler estimation of stroke volume and cardiac output. Stroke volume and CO were estimated by the "nongeometric" technique described by Spodick and Koito (16). The basis of this technique is the relationship: stroke volume = ejection time × ejection rate. The left ventricular ejection time can be precisely measured from Doppler aortic flow traces. The ejection rate can be determined indirectly from the regression equation of Spodick and Koito: mean left ventricular ejection rate (ml/s) = 494 × Doppler mean aortic flow velocity (mV [m/s])  66. Cardiac output was derived by multiplying stoke volume by heart rate recorded at the time of Doppler measurement: CO = (494 × mV  66) × ET × HR.

Our own validation studies (17) show that the Doppler methods described above provide reliable estimates of catheter-derived Ppa and thermodilution cardiac output, with high reproducibility. Using our regression equations for Doppler versus catheter values of Ppa, we defined PH as being present if the Doppler-estimated mean Ppa was  20 mm Hg.

Calculations and derived indices. Total pulmonary vascular resistance (TPVR) was calculated by dividing mean Ppa by CO, and systemic vascular resistance (SVR) was calculated by dividing mean SBP by CO. Cardiac index (CI) was calculated by dividing CO by body surface area (BSA). Body surface area was calculated from the formula: BSA (m²) = weight (kg)^0.425 × height (cm)^0.725 × 0.007184.

Left ventricular diastolic function. Transmitral flow measurements of integrated early passive (E) and late atrial (A) filling velocities were recorded during different experimental conditions. The ratio of integrated velocity-time intervals of A and E waves (A/E ratio) was used as an index of left ventricular diastolic function (18, 19).

Response of the Pulmonary Circulation to Changes in Inspired Oxygen Concentration and Pulmonary Blood Flow

All patients underwent Doppler hemodynamic measurements as described above within 1 wk of lung function measurements and before any treatment of their sleep apnea. Measurements were obtained under baseline conditions (i.e., breathing room air), sustained isocapnic hypoxia (FI_O2 = 0.11, Sa_O2 ~ 80%) and hyperoxia (FI_O2 = 0.50, Sa_O2 ~ 99%) in that order. A dobutamine infusion was then started while patients breathed 50% O₂ (see below). Mask CO₂ and O₂ concentrations were continuously sampled (POET II 602-3 monitor; Criticare Systems) to provide PET_CO2 and PET_O2. Pulse oximetry (finger probe) was recorded continuously using the POET II 602-3 monitor. The minimum arterial Sa_O2 for each sampled period is reported. All physiologic signals were recorded using a computerized data acquisition system (WINDAQ, version 1.15; Dataq Instruments Inc., Akron, OH) with digital sampling speed of 50 Hz.

Sustained isocapnic hypoxia. This was achieved by switching the patient from room air to breathing a hypoxic gas mixture (11% O₂:89% N₂) from a premixed reservoir bag for 20 min. The patient breathed from a tightly fitting Downs full-face CPAP mask (dead space, 75 to 100 ml depending on facial configuration), with inbuilt unidirectional valves. The resistance of the inspiratory circuit was less than 1 cm H₂O/L/s. Isocapnic conditions (± 1 mm Hg) were maintained during hypoxia by a variable manual bleed of CO₂ into the inspiratory side of the breathing circuit, and oxygen was added to the inspiratory circuit to maintain Sa_O2 of 80% if Sa_O2 fell below that value. Doppler measurements were commenced after 10 min of isocapnic hypoxia and completed by 20 min.

Hyperoxic measurements. Doppler hemodynamic measurements were commenced after 10 min of 50% O₂ breathing (air entrainment mask: Multi-vent; Hudson RCI, Temecula, CA) in an attempt to determine the contribution of any hypoxic pulmonary vasoconstriction to resting pulmonary vascular resistance and Ppa. The rationale was that 50% oxygen breathing would eliminate any regional hypoxia. In this part of the experiment no attempt was made to control PET_CO2.

Dobutamine protocol. After 20 to 30 min of 50% O₂ breathing a dobutamine infusion (20, 21) was started at 5 µg/kg/min for 1 min and then increased to 10 µg/kg/min for 8 min. The dobutamine infusion was then increased to 20 µg/kg/min for 8 min and, if tolerated, a further increment to 30 µg/kg/min was administered for 8 min. Doppler measurements were performed during the last 4 min of the each 10, 20, and 30 µg/kg/min dobutamine infusion periods.

Data Analysis and Statistics

Patients were divided into two groups: those with PH (mean Ppa  20 mm Hg while breathing room air at rest, Group I) and those without PH (Group II). Student's t test was used to compare indices of sleep apnea severity and sleep hypoxemia; lung function measurements (including the derived value FRC-CC) and awake arterial blood gases between these two groups. The chi-square test was used to compare categorical variables. Pearson's correlation test was used to assess the relationships between FRC-CC and AaPO₂ and FRC-CC and Ppa. Changes in Ppa within and between the two groups at different levels of Sa_O2 and after incremental increases in pulmonary blood flow (dobutamine infusion) were assessed using ANOVA for repeated measures. Where the F statistic reached statistical significance, post hoc comparisons were performed using the Neumann-Keuls method. A p value of < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

By design, all routine lung function test results of patients were within the normal range: FEV₁ (mean ± SE: 102.5 ± 1.7% predicted, range: 84 to 124% predicted); FVC (96.8 ± 1.7% predicted, range: 81 to 120% predicted); DL_CO (99.6 ± 2.1% predicted, range: 76 to 122% predicted); TLC (96.1 ± 1.8% predicted, range: 81 to 116% predicted). Patients had a wide range of sleep apnea severity (AI: 17.2 ± 2.9/h, range: 0 to 58/h; AHI: 46.2 ± 3.9/h, range: 13 to 102/h; minimum Sa_O2: 78.5 ± 1.7%, range: 51 to 91%; time spent with Sa_O2 < 90%: 35.2 ± 9.1 min, range: 0 to 218 min). Doppler echocardiography and 12-lead ECG measurements during the experiments did not bring to light any unsuspected left ventricular or valvular heart disease.

Eleven patients (10 male, Group I) were pulmonary hypertensive and 21 patients (20 male, Group II) were pulmonary normotensive. Their anthropometric and hemodynamic data are shown in Table 1. The groups were not different with respect to age, body mass index, or smoking history. By definition, mean Ppa was different between the two groups; systolic Ppa and TPVR were also greater in Group I patients. Ppa elevations in Group I patients were only mild (mean Ppa ranged from 20 to 31 mm Hg and systolic Ppa from 26 to 43 mm Hg).

Lung function and gas exchange measurements for Groups I and II are compared in Table 2. There were no differences in spirometric values, gas transfer, or total lung capacity. There was a small (nonsignificant) increase in AaPO₂ in Group I patients compared with Group II patients. FRC-CC was significantly reduced in Group I patients compared with Group II patients. The value for FRC-CC was negative (indicating small airways closure during tidal breathing) in seven of nine (78%) Group I patients, but only three of 18 (17%) Group II patients (²; p < 0.05). It was not possible to obtain single-breath nitrogen measurements in two patients from Group I and three patients from Group II.

For the study population as a whole there were inverse relationships between AaPO₂ and FRC-CC and between mean Ppa and FRC-CC (Figure 1). There was no correlation between Ppa and other indices of lung function. The severity of sleep-disordered breathing was not different between Group I and Group II patients (Table 3), but pulmonary hypertensive patients showed evidence of reduced sleep quality (reduced sleep efficiency and slow-wave sleep) compared with nonhypertensive patients. Sa_O2 measured in the supine position during sleep studies while subjects were awake was nearly identical in Groups I and II (94.3 ± 0.6% and 93.8 ± 1.0%, respectively) (Table 3) and was only marginally different from awake sitting values (95.4 ± 0.23% and 95.2 ± 0.43%) (Table 2).

View larger version (13K): [in this window] [in a new window]   Figure 1.   Correlations between FRC-CC (a measure of small airway closure during tidal breathing) and mean pulmonary artery pressure (Ppa) and ventilation perfusion inequality (P(A-a)O2). FRC = functional residual capacity. CC = closing capacity (measured by single-breath nitrogen test).

The change in Ppa with incremental increases in pulmonary blood flow induced by dobutamine infusion (during 50% O₂ breathing) is shown for Groups I and II in Figure 2. Group II patients demonstrated an increase in mean Ppa at the lowest dobutamine infusion concentration (10 µg/kg/min) compared with baseline Ppa, but thereafter Ppa remained constant at 20 and 30 µg/kg/min infusion rates despite substantial increases in pulmonary blood flow. In contrast, in Group I patients, mean Ppa rose progressively with increasing dobutamine infusion concentrations.

View larger version (11K): [in this window] [in a new window]   Figure 2.   Pulmonary hemodynamic responses to increasing dobutamine infusion concentrations in Group I (Ppa  20 mm Hg) and Group II (Ppa < 20 mm Hg) patients. Points on the curves correspond to different dobutamine infusion rates: from left to right 0, 10, 20, and 30 µg/kg/min, respectively. Patients breathed 50% oxygen to abolish hypoxic pulmonary vasoconstriction. Two patients in Group I and one patient in Group II were intolerant of the highest dobutamine infusion rate, and data were not obtained at that infusion rate. All Ppa and cardiac output values obtained are shown (mean ± SEM), but statistical comparisons were performed only in patients with complete data sets. Note progressive rise in mean Ppa with increasing dobutamine infusion rates in Group I. In contrast, Group II patients showed an increase in Ppa with the first increment in cardiac output, but thereafter Ppa did not increase despite marked increases in cardiac output. *p < 0.05 cf Ppa at 0 µg/kg/min dobutamine. p < 0.05 cf Ppa at 10 µg/kg/min dobutamine.

The effects of different levels of inspired oxygen concentration on mean Ppa in Group I and Group II patients are shown in Figure 3. At all levels of FI_O2 mean Ppa was higher in Group I than in Group II patients. Sa_O2 was the same for both groups at each FI_O2 (FI_O2 0.5: Grp I, 99 ± 0.0%; Grp II, 99 ± 0.1%; FI_O2 0.21: Grp I, 96 ± 0.3%; Grp II, 96 ± 0.2%; FI_O2 0.11: Grp I, 81 ± 0.8; Grp II, 81 ± 0.5%). Both groups demonstrated a reduction in mean Ppa during 50% oxygen breathing compared with room-air breathing. Group I patients demonstrated a greater hypoxic constrictor response than did Group II patients: the change in mean Ppa from a FI_O2 of 0.5 (no hypoxic pulmonary vasoconstrictor stimulus) to a FI_O2 of 0.11 (marked hypoxic pulmonary vasoconstrictor stimulus) was 16.4 ± 1.93 mm Hg and 6.4 ± 0.77 mm Hg, respectively (p < 0.01).

View larger version (15K): [in this window] [in a new window]   Figure 3.   Mean Ppa (mean ± SEM) for Group I (Ppa breathing room air  20 mm Hg) and Group II (Ppa breathing room air < 20 mm Hg) patients at different fractions of inspired oxygen. Isocapnia was maintained during 21% oxygen breathing. SaO2 values were equivalent for Group I and Group II patients at each of the inspired oxygen concentrations (see text). Mean Ppa was greater in Group I than in Group II patients at all levels of inspired oxygen, including during maximal vasodilatation (50% oxygen breathing). The hypoxic pressor response measured as the difference in Ppa between 50 and 11% oxygen breathing was greater in Group I than in Group II patients (see text). *p < 0.05 cf Group II. p < 0.05 cf 50% oxygen breathing. p < 0.05 cf 21% oxygen breathing.

Doppler echocardiographic mitral flow A/E ratio and heart during manipulations of inspired oxygen and dobutamine infusion are shown in Table 4. A/E ratio and heart rate did not differ between Groups I and II under any of the experimental conditions. A/E ratio and heart rate increased incrementally during dobutamine infusions.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of this study indicate that a proportion of patients with OSA without significant lung disease have altered pulmonary hemodynamics, confirming our previous findings (9) and the results of others (1, 3, 4, 8). In the present study we identified a group of patients with OSA in whom mean Ppa was slightly elevated at rest while breathing room air, awake. Compared with patients in whom mean Ppa was normal, this group demonstrated a progressive and marked rise in Ppa when pulmonary blood flow was increased by dobutamine infusion during oxygen breathing. This suggests that these patients had developed structural narrowing of the pulmonary circulation (pulmonary vascular remodeling). The cause for this possible structural remodeling is not immediately clear, but two other observations are relevant. First, it was found that patients with OSA with mild PH had an increased pulmonary vascular pressor response to hypoxia compared with patients without PH. Indices of sleep apnea and sleep hypoxia were not different between patients with and without PH. However, it is possible that an increased pulmonary vascular pressor response to hypoxemic episodes in sleep among patients with OSA found to be mildly pulmonary hypertensive awake may have contributed to the development of structural remodeling of the pulmonary circulation. Second, we found that patients with OSA with PH had evidence of more small airways closure during tidal breathing, and a trend to more ventilation-perfusion inequality compared with patients without PH. The reason for this different behavior of small airways was not established, but to the extent that it may cause regional lung hypoxemia it could also contribute to pulmonary hypertension and pulmonary vascular remodeling.

Right heart catheterization is the gold standard method of assessing pulmonary hemodynamics. A major methodologic consideration in this study, therefore, is whether the Doppler echocardiography methods (15, 16) provided a sufficiently accurate and reliable alternative means of measuring pulmonary hemodynamics. We used a single, skilled echocardiographer to acquire the echocardiographic images and a single trained scorer to make the measurements from hard copy tracings, and not video screen images, of flow profiles. Also, care was taken to exclude patients in whom clear Doppler flow profiles could not be obtained from both pulmonary and aortic outflow tracts. Employing these measurement principles we previously found these Doppler methods provided accurate and highly repeatable estimates of Ppa and CO compared with catheter measurements (17). Correlation coefficients of 0.96 and 0.97 were obtained for Doppler compared with catheter measurements of Ppa and systolic Ppa, respectively. The correlation between Doppler and thermodilution CO was also high (r = 0.97). Repeatability measurements showed intraobserver variability of 5 to 6% for Doppler estimates of Ppa and CO. It is acknowledged that our previous validation studies (17) were conducted under resting conditions. In the present studies some of the measurements were made during conditions of increased cardiac output. However, this is no a priori reason to believe that the relationship between arterial pressure and the various timing intervals of the right and left outflow tracings would be different under conditions of increased flow.

A shortcoming of this report is that we did not have a measure of pulmonary capillary wedge pressure (Ppcw), although others have shown (10) that it is difficult to obtain in patients with OSA by catheter (60% success at rest and only 22% during exercise). Both hypoxia and increased heart rate are capable of reducing left ventricular diastolic function (22, 23) and may therefore have contributed to the increases observed in Ppa during hypoxia and dobutamine infusions in our experiments. Indeed, during dobutamine infusion there was an increase in heart rate and a corresponding, predictable increase in A/E ratio (23). However, it is unlikely that the differences we observed in Ppa between PH and non-PH groups could be explained by differences in left ventricular diastolic function since A/E ratios and heart rate were not different between the groups under any condition.

Approximately one third of our study population had mild PH (Doppler-estimated mean Ppa of 20 mm Hg or more; range, 20 to 31 mm Hg). This result is similar to our earlier study in which we used the same Doppler method for estimating Ppa and found 11 of 27 (41%) of patients with OSA (without lung disease) had evidence of mild PH. A slightly different definition of PH was used at that time: both mean Ppa  20 mm Hg and systolic Ppa  30 mm Hg were required to fulfil the definition. Using this more stringent definition of PH, only five of 32 patients (16%) in the present study would have been classified as pulmonary hypertensive. Previous studies have reported prevalence rates for PH in patients with OSA ranging from approximately 20 to 80% (1, 24), with most studies reporting values in the 20 to 40% range. All studies, apart from the present study and our earlier study, included patients with hypoxemic lung disease. The percentage of patients with obstructive lung disease in these studies varied from approximately 20% in three studies (5, 6, 10) through to 79% (4). It was concluded that the factors most strongly correlated with PH among patients with OSA were lung function impairment and/or daytime hypoxemia. As in our present and previous studies, other investigators have found awake PH to be unrelated or only weakly related to the severity of sleep apnea/sleep hypoxia (3, 5, 10). This has led one commentator (11) to suggest that awake pulmonary hemodynamics are unaffected in OSA unless there is coexisting hypoxemic lung disease. However, we believe such a conclusion cannot be justified given the results of our studies and those of previous reports (25).

We are not the only group to report PH in OSA in the absence of hypoxemic lung disease. Fletcher and colleagues (4) found PH in two of five severely affected patients with OSA who had normal lung function. One of the two pulmonary hypertensive patients had awake hypoxemia. In the other patient awake hypoxemia was trivial and unlikely to have caused hypoxic vasoconstriction. In another study of 65 patients with OSA, Podszus and colleagues (3) reported awake resting PH in 13 patients, 10 of whom had no evidence of cardiac and pulmonary disease. Laaban and colleagues (24) recently reported that 36% of massively obese patients with sleep apnea without lung disease had PH versus only 7% in nonapneic obese patients. Body mass index and lung function were the same in sleep apneic and nonapneic patients, but apneic patients were mildly hypoxemic (Pa_O2 = 77 mm Hg) compared with nonapneic patients (Pa_O2 = 87 mm Hg). Other groups have reported awake PH in OSA in the presence of normal or near normal awake Pa_O2 values. Tilkian and colleagues (1) performed right heart catheterization in 12 patients with OSA while awake or asleep and found awake PH in eight patients, five of whom had awake Pa_O2 values > 75 mm Hg. Also, in a study of 100 consecutively diagnosed patients with OSA, Laks and colleagues (8) found that six of 42 pulmonary hypertensive patients had Pa_O2 levels > 80 mm Hg. Individual patient results from another recent study (10) showed pulmonary hypertensive patients with OSA who had minimal (or absent) awake hypoxemia, lung disease, or both. Finally, three studies (3, 5, 10) have shown that approximately 40% or more of patients with OSA with normal Ppa at rest have abnormally high mean Ppa ( 30 mm Hg) during steady state submaximal exercise, indicating the presence of latent PH. A significant proportion of these patients had no evidence of lung disease. Therefore, although there is little doubt that the coexistence of hypoxemic lung disease strongly promotes the development of PH in OSA, it is also clear that PH can occur in patients with OSA without evidence of significant lung disease.

Lung Function Results

The study was designed to exclude patients with clinical or lung function evidence of lung disease. As expected, lung function parameters, including indices of small airways function (e.g., MMEF) and small airways closure (i.e., closing volume/VC ratio) were within normal limits and were not different between Group I and Group II patients. An unexpected finding was that CC exceeded FRC (implying small airways closure within the tidal range of breathing) in approximately 80% of Group I patients, but in less than 20% of Group II patients. There was also a trend to more ventilation-perfusion inequality among pulmonary hypertensive patients with OSA. For the study population as a whole there was, as expected (26), a relationship between FRC-CC and AaPO₂: the more negative the value for FRC-CC (and therefore the more small airways closure during tidal breathing) the greater the ventilation-perfusion inequality. A relationship was also found between FRC-CC and Ppa. Regional lung hypoxia causing regional pulmonary vasoconstriction may contribute to this relationship, and the reduction in Ppa observed during oxygen breathing supports this hypothesis. Small airways closure during tidal breathing may become more extensive at night when supine (26). However, our results show that the decrease in Sa_O2 between awake sitting and awake supine was very small (approximately 1%) and was not different between the two groups.

We do not have a satisfactory explanation for why small airways closure during tidal breathing was more common among Group I than among Group II patients. One possibility is that Group I patients had an intrinsic abnormality of small airways function that predisposed to premature airway closure. Group I patients did not have lower MMEF and higher closing volume/VC ratios than Group II patients. Although, these measurements are said to reflect small airways function the negative findings do not entirely exclude the presence of small airways abnormalities. One intriguing possibility is that vascular remodeling may alter small airways function. In this model, we hypothesize that repetitive hypoxic pulmonary vasoconstriction in patients with OSA with heightened pulmonary vascular reactivity leads to pulmonary vascular remodeling. The molecular mechanisms underlying the pulmonary vascular changes may also contribute to structural changes in adjacent airways or small airways closure may be due to interdependence between "stiffened" small blood vessels and their adjacent airways within the bronchovascular bundle. A reduction in FRC related to obesity is another potential cause of small airways closure during tidal breathing (27), but BMI was not different between Groups I and II. There was a trend toward lower FRC in Group I patients, with FRC being < 80% predicted in six of 11 (55%) Group I patients compared with seven of 21 (33%) Group II patients (²; p = NS).

Changes in Ppa during Increased Pulmonary Blood Flow

To examine the effects of raised pulmonary blood flow on Ppa we made measurements of Ppa during incremental increases in a dobutamine infusion. Dubutamine is a synthetic catecholamine that increases cardiac contractility, stroke volume, heart rate, and cardiac output (20, 21). Studies have shown no change or slight reductions in pulmonary wedge or left ventricular filling pressures during dobutamine infusions in patients with cardiac disease (21), but the effect of dobutamine in subjects without cardiac disease has, to our knowledge, not been reported. In our study, patients breathed 50% oxygen during dobutamine measurements in an effort to control for possible confounding effects of variable levels of pulmonary vascular tone between patients because of hypoxic vasoconstriction. With increasing dobutamine infusion and pulmonary blood flow, patients with PH showed a progressive increase in Ppa. In contrast, patients without PH showed an increase in Ppa with the first dobutamine increment, but thereafter Ppa plateaued even though the final mean cardiac output was greater than in patients with PH. It is unlikely that these differences could have been due to differences in left ventricular filling and Ppcw since heart rate and A/E time-velocity integral ratio were not different between groups. These results are consistent with our hypothesis that patients with OSA who are mildly pulmonary hypertensive have undergone remodeling of the pulmonary circulation such that there is less pulmonary vascular recruitment or vascular distension during conditions of increased pulmonary blood flow.

Our results with dobutamine infusions are similar to results from earlier studies among patients with OSA in which light exercise was used to increase cardiac output (3, 10). In these earlier studies patients with OSA with PH at rest developed marked rises in Ppa during exercise, but also, approximately 40 to 50% of patients with OSA who had no evidence of PH at rest showed excessive exercise-induced elevations in Ppa, suggesting the presence of latent PH (3, 10). Chaouat and colleagues (10) argued that in some instances the excessive rise in Ppa may have been caused by left ventricular dysfunction during exercise, although in patients with OSA who had normal Ppa at rest, the mean Ppcw, when it could be measured, was normal during exercise. The change in mean Ppa that we observed during dobutamine-induced increases in pulmonary blood flow in patients with OSA without PH are similar to those observed previously in normal subjects subjected to similar incremental increases in cardiac output with exercise (28).

Pulmonary Vasopressor Responses to Hypoxia

In our earlier study (9) we found that during room-air breathing patients with OSA with PH were slightly hypoxemic compared with patients without PH. We considered, therefore, that to assess the vasopressor response to hypoxia by measuring the change in Ppa between room-air breathing and breathing a hypoxic mixture might cause the hypoxic pressor response to be underestimated in PH patients. In the present study, we assessed the hypoxic vasoconstrictor response by measuring the difference in Ppa between 50% oxygen breathing and 11% inspired oxygen concentration in an attempt to standardize the measurement conditions among patients. We found that the pulmonary pressor response to isocapnic hypoxia was increased in patients who had evidence of PH at rest breathing room air.

The mechanisms for the increased pressor response to hypoxia observed in patients with PH are uncertain. One possibility is that such patients may have a genetic predisposition to more vasoconstriction in response to alveolar hypoxia. Such genetic differences have been reported previously in other species (29, 30) and human populations living at high altitude (31). An alternate explanation is that the larger increase in Ppa observed in response to hypoxia was due to differences in pulmonary vascular geometry. We do in fact have some indirect evidence (see dobutamine data above) for vascular remodeling in the patients with OSA with PH. If baseline arteriolar narrowing was present in patients with PH compared with those without PH, the same smooth muscle shortening in response to hypoxia would result in greater increases in vascular resistance and Ppa. Regardless of the mechanism involved, the heightened vasopressor response to hypoxia observed in patients with OSA with PH raises the possibility that such patients may be at risk of further pulmonary vascular remodeling and progressive PH as a result of repetitive sleep apneas.

The only other study of the effects of acute hypoxia on pulmonary hemodynamics in patients with OSA found no difference in the Ppa response to eucapnic hypoxia between 20 patients with OSA and nine normal subjects (32). In this study the hypoxic challenge was induced rapidly (2 to 3 min) with a final Sa_O2 of approximately 70%. Only three of the patients with OSA had PH, making meaningful statistical comparisons of the vasopressor response between PH and non-PH patients with OSA impossible. However, these three patients with OSA did not appear to show an excessive Ppa response to eucapnic or hypercapnic hypoxia. The differences between these findings and the present study may be due to methodologic differences. Laks and colleagues (32) used a very brief hypoxic challenge (2 to 3 min) compared with the 20 min of isocapnic hypoxia in the present study. Although such a brief challenge more closely mimics gas exchanges during an obstructive apnea, measurements made during such a challenge may not adequately reflect the cumulative effects on the pulmonary circulation of several repeated hypoxic episodes in patients with OSA. Laks and colleagues (32) also compared the Ppa change between room-air breathing and the end of the acute hypoxic challenge. Several of their patients with OSA were hypoxemic when room-air breathing raising the possibility that, in some patients at least, the true extent of the pulmonary vascular pressor response to hypoxia may have been underestimated.

Sleep Results

An unexpected finding was that patients with OSA with PH had reduced sleep quality (decreased sleep efficiency, less slow-wave sleep, and proportionately more Stage 2 NREM sleep) compared with patients with OSA without PH, although there was no difference in the frequency of disordered breathing events or sleep arousals between the groups. We have no direct evidence for the cause of poorer sleep quality in patients with OSA with PH. One possible explanation is that the excessive vasopressor response to hypoxia observed among PH patients may have led to a greater increase in right transmural atrial pressures during sleep apneas, causing more nocturnal release of atrial natriuretic peptide. Obstructive sleep apnea has previously been associated with atrial naturietic peptide release and nocturia (33). Atrial natriuretic peptide could conceivably lead to disturbed sleep by increasing urine flow or by inducing intravascular to extravascular fluid shifts (33).

Other Methodologic Considerations

A further methodologic consideration in this study is that lung function and arterial blood gas measurements were performed in the sitting position, whereas hemodynamic and sleep measurements were made supine. FRC is decreased in the supine compared with the sitting position, particularly in obese subjects. The measurements of FRC-CC that we obtained in the sitting position might therefore have underestimated the extent of small airways closure during tidal breathing and gas exchange disturbance in patients supine (26). However, the underestimation of small airways closure and gas exchange impairment was probably small (Sa_O2 decrease between awake sitting and supine approximately 1%), and there was no difference in body mass or the supine Sa_O2 between Groups I and II.

In conclusion, this study has confirmed that mild PH can develop in patients with OSA in the absence of significant lung and cardiovascular disease and in the absence of significant daytime hypoxemia. Patients with OSA who have PH demonstrate a greater increase in Ppa during incremental increases in pulmonary blood flow than do patients without PH, suggesting that once daytime PH is established there is some remodeling (anatomic narrowing) of the pulmonary vasculature. It is difficult to propose a single unifying hypothesis that ties together all our findings in patients with OSA with PH (i.e., pulmonary vascular remodeling, increased small airways closure, and increased pressor responses to hypoxia). One possibility is that some patients develop, principally as a result of their obesity, a tendency towards small airways closure during normal breathing, regional ventilation-perfusion inequality, and alveolar hypoxia. If a subset of these patients also have increased vascular reactivity to alveolar hypoxia, they may develop pulmonary vascular remodeling, independent of the presence of sleep apnea. We believe this to be an unlikely explanation. First, PH and pulmonary vascular remodeling in humans rarely develops without severe alveolar hypoxia and Pa_O2 < 60 mm Hg. This is so for residents at high altitude (31) and patients with hypoxemic lung disease (34). The hypoxemia in our patients with OSA with PH was very mild by comparison. We believe therefore that repetitive sleep hypoxia played a significant role in the development of PH and vascular remodeling in our patients with OSA. However, we acknowledge that without making measurements of small airways function and pulmonary hemodynamics in a BMI matched control group without OSA we cannot exclude the above hypothesis.

An alternative explanation is that patients with OSA who develop PH have a genetically determined increased pressor response to alveolar hypoxia, and that sleep hypoxia in these patients leads to pulmonary vascular remodeling. The difficulty with this hypothesis is explaining the coexistence of small airways closure during tidal breathing in PH patients. We hypothesize that functional changes may occur in bronchioles adjacent to pulmonary arterioles as they undergo remodeling. We know of no experimental evidence showing that hypoxia-induced pulmonary hypertension leads to small airways dysfunction, although an independent association between OSA and small airways obstruction has recently been reported (35).