Determinants of Chronic Hypercapnia in Japanese Men With Obstructive Sleep Apnea Syndrome

Determinants of Chronic Hypercapnia in Japanese Men With Obstructive Sleep Apnea Syndrome

Determinants of Chronic Hypercapnia in Japanese Men With Obstructive Sleep Apnea Syndrome* Tsuneto Akashiba, MD; Seiji Kawahara, MD; Naoko Kosaka, MD;...

111KB Sizes 5 Downloads 45 Views

Determinants of Chronic Hypercapnia in Japanese Men With Obstructive Sleep Apnea Syndrome* Tsuneto Akashiba, MD; Seiji Kawahara, MD; Naoko Kosaka, MD; Daisuke Ito, MD; Osamu Saito, MD; Tohru Majima, MD; and Takashi Horie, MD, FCCP

Study objective: To identify the determinants of chronic hypercapnia (ie, PaCO2, > 45 mm Hg) in men with obstructive sleep apnea syndrome (OSAS) without airflow obstruction. Design: An analysis was conducted of 143 male patients with OSAS, which had been diagnosed by polysomnography (PSG), who had been referred to a university hospital. Patients were classified as hypercapnic (ie, PaCO2, > 45 mm Hg) and normocapnic (ie, PaCO2, < 45 mm Hg), and obese (ie, body mass index [BMI], > 30 kg/m2) or nonobese (ie, BMI, < 30 kg/m2). Patients with airflow obstruction (ie, FEV1/FVC ratio, < 70%) were excluded from the study. Baseline clinical characteristics, pulmonary function, PSG data, and blood gas data were compared between hypercapnic and normocapnic patients. Correlations between PaCO2 and several anthropometric, respiratory, and polysomnographic variables were determined by stepwise multiple regression analysis. Results: Fifty-five patients (38%) were hypercapnic. Hypercapnic patients were younger and heavier, and had more abnormalities on pulmonary and PSG testing. Stepwise multiple regression analysis revealed that the PaCO2 level was influenced significantly by the mean level of arterial oxygen saturation (SaO2) during sleep and by the percent of vital capacity (%VC) (R2 ⴝ 0.430; p < 0.0001), indicating that 43% of the total variance in the PaCO2 could be explained by the mean SaO2 and %VC in hypercapnic patients. In contrast, only 13% of the total variance in the PaCO2 was accounted for by the mean SaO2 and BMI in normocapnic patients (R2 ⴝ 0.134; p ⴝ 0.0034). The mean SaO2, %VC, and PaO2 were selected as independent variables for predicting the PaCO2 in obese patients. These variables explained 41% of the total variance in the PaCO2 (R2 ⴝ 0.407; p < 0.0001), whereas the mean SaO2 only accounted for 13% of the total variance in PaCO2 levels in nonobese patients (R2 ⴝ 0.134; p ⴝ 0.0064). Conclusion: Nocturnal desaturation and restrictive pulmonary impairment play major roles in determining the PaCO2 in hypercapnic and obese OSAS patients without airflow obstruction. (CHEST 2002; 121:415– 421) Key words: hypercapnia; obstructive sleep apnea syndrome; oxygen desaturation; restrictive pulmonary impairment Abbreviations: AHI ⫽ apnea-hypopnea index; BMI ⫽ body mass index; ERV ⫽ expiratory reserve volume; FRC ⫽ functional residual capacity; ODI ⫽ oxygen desaturation index; OHS ⫽ obesity-hypoventilation syndrome; OSAS ⫽ obstructive sleep apnea syndrome; PSG ⫽ polysomnography; RV ⫽ residual volume; Sao2 ⫽ arterial oxygen saturation; %VC ⫽ percent of vital capacity

hypercapnia occurs in a subset of patients C hronic with obstructive sleep apnea syndrome (OSAS).1–10 However, why some patients with OSAS have hypercapnia while others do not is not wellunderstood. Chronic hypercapnia is associated with adverse pathophysiologic consequences and is a pre*First Department of Internal Medicine, Nihon University School of Medicine, Tokyo, Japan. Manuscript received June 21, 2000; revision accepted August 6, 2001. Correspondence to: Tsuneto Akashiba, MD, 30 –1 Oyaguchi Kamimachi, Itabashi-Ku, Tokyo, Japan 173-8610

dictor of poor survival rate in patients with COPD.11 The prognosis of patients with OSAS and chronic hypercapnia has not been investigated. Obesity-hypoventilation syndrome (OHS) is characterized by obesity, hypersomnolence, and hypercapnia, and it occurs in the setting of severe OSAS. For editorial comment see page 320 In previous studies,1–9 chronic airway obstruction, obesity, gender, nocturnal respiratory abnormalities, abnormal ventilatory responses, and alcohol intake CHEST / 121 / 2 / FEBRUARY, 2002

415

have been shown to contribute to daytime hypercapnia in patients with OSAS. Furthermore, it has been suggested that chronic airway obstruction plays a major role in the development of chronic hypercapnia in patients with OSAS.4 It is strange that chronic airway obstruction is more associated with the development of hypercapnia than are respiratory abnormalities during sleep in patients with OSAS. A 1994 study10 also has shown that obstructive and restrictive pulmonary disturbances and impaired ventilatory responses contributed to daytime hypercapnia in patients with OSAS. However, these studies included women and patients with COPD. Furthermore, the numbers of patients were relatively small, except for one study.7 Because COPD12–14 and gender15–17 strongly influence the ventilatory response, the mechanisms responsible for chronic hypercapnia in patients with OSAS may be affected by these factors. Therefore, in the present study, we examined the determinants of chronic hypercapnia in a large number of male patients with OSAS but without chronic airway obstruction.

Materials and Methods The study cohort consisted of 146 male patients who had received a diagnosis of OSAS based on the results of polysomnography (PSG). Patients with airway obstruction (ie, FEV1/FVC ratio, ⬍ 70%) were excluded from this study. All patients gave informed consent for participation. No patients had chronic lung disease or were receiving bronchodilator therapy. A full night of PSG with continuous recordings of the EEG, electrooculogram, submental electromyogram, ECG, airflow at the nose and mouth (by thermistor recording), movement of the rib cage and abdomen (by inductance plethysmography), and arterial oxygen saturation (Sao2) was performed in all of the patients. The analysis and interpretation of the PSG data were performed using standard techniques.18 An apnea was defined as a cessation of airflow at the nose and mouth lasting at least 10 s. A hypopnea was defined as a decrease in airflow, rib cage excursions, or abdominal excursions by ⬎ 50% associated with an oxygen desaturation of at least 4% below the preceding baseline value.19 The apnea-hypopnea index (AHI) was calculated as the number of apnea and hypopnea episodes per hour of sleep. The patients with AHI scores of ⬎ 10 received a diagnosis of OSAS and were included in this study. The oxygen desaturation index (ODI) was calculated by dividing the total number of episodes of oxygen desaturation by the total sleep time, with desaturation defined as a ⱖ 10-s reduction in the Sao2 (ie, ⱖ 4% of baseline at the nadia), independent of airflow or thoracoabdominal movements.20 The mean and minimum Sao2 during sleep also were calculated from the PSG data. Conventional spirometry and static lung volumes were measured by autospirometry (Chestak; Chest Co; Tokyo, Japan). The lung volumes were measured by the single-breath helium dilution technique. The percent predicted values were obtained from the literature.21 Arterial blood samples were drawn from a radial artery with the patients in the supine position and awake. Arterial blood samples were analyzed by an autoanalyzer (model ABL3000; Radiometer Co; Tokyo, Japan). Hypercapnia was defined as a Paco2 of ⱖ 45 mm Hg. Normocapnia was defined as 416

a Paco2 of ⬍ 45 mm Hg. Obesity was defined as a body mass index (BMl) of ⱖ 30 kg/m2, and nonobesity was defined as a BMI of ⬍ 30 kg/m2. Statistical Analysis Results are presented as the mean ⫾ SD. Group differences were assessed with an unpaired t test. We also determined Pearson linear correlations between certain variables. Correlations between Paco2 and anthropometric, respiratory, and polysomnographic variables were determined by stepwise multiple regression analysis using a statistical program (StatView, version 4; SAS Institute; Cary, NC). Each of the variables was entered into the multiple regression analysis if its F value was ⬎ 4. A p value ⬍ 0.05 was considered to be statistically significant.

Results The mean age was 48.3 ⫾ 11.3 years, and the mean BMI was 29.6 ⫾ 5.7 kg/m2. The results of pulmonary function tests were normal in all but six of the patients who had restrictive ventilatory abnormalities (ie, percent vital capacity [%VC], ⬍ 80%). No patients had evidence of obstructive airway disease (FEV1/FVC ratio, ⬍ 70%). The average daytime Pao2 and Paco2 levels were 80.7 ⫾ 9.6 and 44.5 ⫾ 5.3 mm Hg, respectively. The comparisons between hypercapnic and normocapnic patients on baseline characteristics, PSG data, pulmonary function data, and blood gas levels are shown in Table 1. Fifty-five of the 143 patients (38%) were hypercapnic. Hypercapnic patients were younger and heavier, and had more severe OSAS than did the normocapnic patients. The mean %VC in hypercapnic patients was significantly lower than that in normocapnic patients, although the VC, expiratory reserve volume (ERV), functional residual capacity (FRC), percent of FRC, residual volume (RV), and percent of RV were not significantly different between hypercapnic and normocapnic patients. The mean Pao2 in hypercapnic patients was significantly lower than that in normocapnic patients. In hypercapnic patients, the Paco2 correlated with BMI, mean Sao2, VC, and %VC (Table 2). The mean Sao2 most closely correlated with Paco2 (r ⫽ ⫺0.509; p ⬍ 0.0001). Stepwise multiple regression analysis was performed to identify the factors contributing to the increase in Paco2 in the hypercapnic patients. The results showed that the level of Paco2 was significantly influenced by the mean Sao2 and %VC. When both variables were incorporated in the model, 43% of the total variance of Paco2 could be explained in hypercapnic patients (R2 ⫽ 0.430; p ⬍ 0.0001; Fig 1). Although the mean Sao2 and BMI correlated with Paco2 based on stepwise multiple regression analysis, only 13% of the total variance in the Paco2 was accounted for by these variables in normocapnic patients (R2 ⫽ 0.134; p ⫽ 0.0034; Fig 2). Clinical Investigations

Table 1—Clinical Characteristics of the Hypercapnic and Normocapnic Patients* Characteristics

Hypercapnic (n ⫽ 55)

Normocapnic (n ⫽ 88)

p Value

Age, yr BMI, kg/m2 AHI, events/h Mean Sao2, % Lowest Sao2, % ODI, events/h VC mL % FEV1/FVC, % ERV, mL FRC mL % RV mL % Pao2, mm Hg Paco2, mm Hg

45.4 ⫾ 10.2 (26–68) 32.0 ⫾ 7.0 (17.2–56.4) 60.7 ⫾ 21.0 (12.4–101.0) 83.3 ⫾ 8.7 (54.3–96.0) 62.0 ⫾ 9.4 (40.0–80.0) 62.2 ⫾ 18.1 (16.2–112.4)

50.2 ⫾ 11.7 (31–72) 28.1 ⫾ 4.1 (21.1–39.1) 47.6 ⫾ 25.6 (10.4–141.7) 89.6 ⫾ 7.0 (71.0–99.0) 69.1 ⫾ 10.5 (50.0–92.0) 53.4 ⫾ 22.3 (15.8–136.8)

0.026 ⬍ 0.0001 0.0029 ⬍ 0.0001 ⬍ 0.0001 0.0044

3,968.7 ⫾ 676.6 (2,165–5,660) 103.8 ⫾ 15.4 (51.0–134.9) 82.1 ⫾ 5.6 (71.2–98.0) 1,034.0 ⫾ 413.4 (170–1,870)

4,169.8 ⫾ 740.0 (2,580–6,490) 112.2 ⫾ 16.1 (74.8–152.7) 79.2 ⫾ 5.0 (70.1–94.5) 1,124.2 ⫾ 440.1 (290–2,090)

0.1060 0.0026 0.0612 0.2339

2,454.2 ⫾ 599.9 (1,500–4,410) 94.2 ⫾ 24.2 (46.4–164.9)

2,622.3 ⫾ 779.8 (970–5,570) 95.2 ⫾ 25.1 (46.6–178.9)

0.1829 0.1450

1,452.2 ⫾ 499.3 (540–2,830) 96.4 ⫾ 31.9 (39.8–175.0) 74.0 ⫾ 9.2 (46.4–101.8) 49.0 ⫾ 5.6 (45.0–75.0)

1,555.7 ⫾ 496.0 (490–3,480) 101.2 ⫾ 31.4 (34.1–214.2) 80.7 ⫾ 9.6 (58.9–105.2) 41.6 ⫾ 2.3 (36.4–44.9)

0.2825 0.4656 ⬍ 0.0001 ⬍ 0.0001

*Values given as mean ⫾ SD (range), unless otherwise indicated.

In order to determine the influences of obesity on hypercapnia, we grouped patients according to BMI. Fifty-seven of the patients were obese, and all of the physiologic variables, except for RV, VC, and percent of RV, in obese patients were statistically different from nonobese patients (Table 3). The mean Paco2 in obese patients was higher than that in nonobese patients, and the mean Pao2 in obese patients was lower than that in nonobese patients. The mean Sao2, %VC, and Pao2 were selected as independent factors contributing to the increase in the Paco2 based on stepwise multiple regression analysis in obese patients. The model that included these vari-

Table 2—Correlations Between PaCO2 and Other Variables in Hypercapnic Patients Variables Age BMI AHI Sao2 Mean Lowest ODI VC %VC FEV1/FVC, % ERV FRC %FRC RV %RV Pao2

Regression Coefficient

ables showed a relatively good correlation with Paco2 (R2 ⫽ 0.407; p ⬍ 0.0001). In contrast, only the mean Sao2 correlated weakly with Paco2 in the nonobese patients (R2 ⫽ 0.134; p ⫽ 0.0064). Discussion We found a relatively high incidence of daytime hypercapnia in patients with severe OSAS (38%). In previous studies,2– 4,7–10 the prevalence of hypercapnia in patients with OSAS varied from 14 to 57%. However, the number of patients in these previous studies was small. In the largest study, Leech et al7 examined 111 patients with OSAS and found that 41 of 111 patients (36%) had chronic hypercapnia.

p Value

⫺ 0.187 0.342 0.067

0.1715 0.0107 0.6265

⫺ 0.509 ⫺ 0.202 ⫺ 0.142 0.304 ⫺ 0.409 ⫺ 0.155 ⫺ 0.256 ⫺ 0.200 ⫺ 0.138 0.007 ⫺ 0.018 ⫺ 0.264

⬍ 0.0001 0.1391 0.2361 0.0240 0.0019 0.2590 0.0623 0.1513 0.3541 0.9622 0.9063 0.0517

Figure 1. Relationship between the observed Paco2 and predicted Paco2 based on multiple regression analysis of data from hypercapnic patients. CHEST / 121 / 2 / FEBRUARY, 2002

417

Figure 2. Relationship between the observed Paco2 and predicted Paco2 based on multiple regression analysis of data from normocapnic patients.

Although alveolar hypoventilation, hypercapnia, and hypoxemia occur in some patients with chronic obstructive or restrictive pulmonary diseases, it is generally only found in patients with advanced disease. In the setting of COPD, hypercapnia is usually not observed unless the FEV1 is ⬍ 1 L or is 35% of the predicted value,22 and is a predictor of poor survival rates.11 OHS23,24 is also characterized by chronic hypercapnia and is often associated with severe OSAS. Therefore, it is important to identify factors that predict chronic hypercapnia in patients with OSAS. We found that hypercapnic patients were younger, heavier, and had more abnormal pulmonary function

and polysomnographic data than did normocapnic patients. Furthermore, there were significant correlations between Paco2 and BMI, mean Sao2, and %VC in hypercapnic patients. The mean Sao2 had the strongest correlation with Paco2 in hypercapnic. Additionally, stepwise multiple regression analysis showed that the Paco2 was significantly influenced by the mean Sao2 and %VC. When these variables were included in the model, 43% of the total variance in the Paco2 level in hypercapnic patients could be explained by the severity of nocturnal desaturation and the severity of restrictive pulmonary impairment. These results suggest that the mean Sao2 plays the most important role in determining the Paco2 in patients with OSAS. Onal et al25 also demonstrated that younger patients had more severe OSAS than did older patients and that age correlated significantly with AHI. However, they did not determine the relationship between Paco2 and age or AHI. Because neither age nor AHI correlated directly with Paco2 in hypercapnic patients in the present study, it is likely that age and AHI did not contribute to the development of daytime hypercapnia. In previous studies, a number of factors, including chronic airway obstruction,4,7–9 obesity,3,4,7,8 gender,7,17 nocturnal respiratory abnormalities,7 depressed ventilatory responses,1,2,9,10 and alcohol intake,8 have been suggested as determinants of hypercapnia in patients with OSAS. Bradley et al4 emphasized that chronic airway obstruction plays a major role in the development of hypercapnia in

Table 3—Comparisons Between Obese and Nonobese Patients* Characteristics Age, yr BMI, kg/m2 AHI, events/h Sao2, % Mean Lowest ODI VC mL % FEV1/FVC, % ERV, mL FRC mL % RV mL % Pao2, mm Hg Paco2, mm Hg

Obese Patients (n ⫽ 57)

Nonobese Patients (n ⫽ 86)

p Value

44.3 ⫾ 10.0 (26–67) 34.9 ⫾ 5.1 (30.0–56.4) 64.2 ⫾ 19.1 (30.4–101.0)

50.8 ⫾ 13.1 (28–72) 26.3 ⫾ 2.3 (17.2–29.8) 52.0 ⫾ 22.9 (10.4–141.7)

0.0007 ⬍ 0.0001 0.0019

80.0 ⫾ 7.7 (54.3–94.0) 61.0 ⫾ 8.3 (40.0–79.0) 66.1 ⫾ 20.6 (37.1–136.8)

90.3 ⫾ 5.3 (69.8–99.0) 67.2 ⫾ 9.6 (50.0–92.0) 52.3 ⫾ 18.9 (15.8–112.4)

⬍ 0.0001 0.0003 0.0008

4,034.1 ⫾ 733.8 (2,150.0–5,880.0) 104.4 ⫾ 15.6 (51.0–134.9) 81.8 ⫾ 5.8 (70.6–98.0) 908.5 ⫾ 408.9 (170–1,870)

4,129.1 ⫾ 712.9 (2,580.0–6,420.0) 112.5 ⫾ 16.5 (74.0–152.7) 79.6 ⫾ 4.9 (74.0–94.5) 1,187.3 ⫾ 385.5 (290–2090)

0.4451 0.0006 0.0245 0.0002

2,330.8 ⫾ 612.3 (970.0–3,440.0) 89.7 ⫾ 39.4 (46.4–162.3)

2,713.9 ⫾ 701.5 (1,180.0–5,570.0) 103.3 ⫾ 35.4 (46.6–178.9)

0.0019 ⬍ 0.0001

1,456.0 ⫾ 474.7 (540.0–2,480.0) 97.2 ⫾ 31.2 (39.8–135.4) 73.8 ⫾ 8.7 (46.4–91.9) 46.2 ⫾ 6.9 (37.8–75.0)

1,547.5 ⫾ 524.2 (490.0–3,480.0) 100.9 ⫾ 31.9 (34.4–214.2) 80.5 ⫾ 9.7 (66.6–102.0) 43.5 ⫾ 3.4 (36.5–51.0)

0.3048 0.5493 ⬍ 0.0001 0.0014

*Values given as mean ⫾ SD (range), unless otherwise indicated. 418

Clinical Investigations

patients with OSAS. Leech et al7 also found that daytime hypoxemia, mechanical impairment of respiration due to obesity or obstructive airway disease, and an increased AHI contribute to increases in Paco2 in patients with OSAS. However, these studies included patients with COPD and women. Because the ventilatory response to increased Paco2 is affected by the presence of chronic airway obstruction12–14 and gender,14 –16 it is possible that the hypercapnia observed in their studies was not directly due to OSAS. Therefore, we selected only male patients with OSAS but without chronic airway obstruction. Although we cannot exclude the presence of mild COPD when the values of the FEV1/ FVC ratio are between 70% and 80%, we think that mild COPD (ie, FEV1/FVC ratio, ⬎ 70%) has little effect on daytime arterial blood gas levels. Therefore, we found that the severity of desaturation during sleep, as expressed by the mean Sao2, and restrictive impairment, as expressed by the %VC, were major factors in the development of daytime hypercapnia. Respiratory disturbances during sleep have not appeared to be important factors for daytime hypercapnia in previous studies. Only Leech et al7 found that an increased AHI contributed to the development of hypercapnia. Although Jones et al3 showed that hypoxemia during sleep is more severe in patients with OHS than in patients without OHS, they did not determine the relationship between nocturnal hypoxemia and daytime hypercapnia. We found that the mean Sao2 is the strongest predictor of hypercapnia, whereas AHI was not related to the Paco2 in patients with OSAS. Although the severity of OSAS usually is expressed by the AHI or by the degree of desaturation during sleep, the severity of nocturnal desaturation appears to be more important for the development of cardiopulmonary complications of OSAS than AHI. Tsai et al20 showed that the ODI can reliably assess the severity of hypoxemia during sleep in patients with OSAS. However, in the present study, ODI did not correlate significantly with Paco2, although the ODI was greater in hypercapnic patients than in normocapnic patients and correlated significantly with AHI. The reason for the discrepancy between the present results and those of Tsai et al20 is unclear. Further investigations of the best method to evaluate oxygen desaturation during sleep are necessary. Chronic hypoventilation affects the pulmonary circulation and induces pulmonary hypertension, which is the main cause for a poorer prognosis in patients with various lung diseases.26 –28 Therefore, if chronic hypercapnia is associated with the severity of oxygen desaturation during sleep, as shown in our study, avoiding nocturnal desaturation

may improve pulmonary circulation and, therefore, the prognosis in hypercapnic patients with OSAS. Although obesity is considered to be an important contributing factor for the development of OSAS, the relationship between obesity and chronic hypercapnia is unclear.1–3,5,7,10,23,24 Although hypercapnic patients are more obese than normocapnic patients with OSAS,3,4,7,8 a direct correlation has not previously been shown between obesity and Paco2. Leech et al7 found a correlation between Paco2 and the weight/height ratio, but multiple regression analysis did not identify the weight/height ratio as an independent factor predicting Paco2. Another study10 also showed that hypercapnic patients were heavier and had greater body surface areas than did normocapnic patients, although the BMI was not significantly different between the two groups. Although the correlation between Paco2 and obesity was not statistically significant in that study, we found that hypercapnic patients were more obese than normocapnic patients and that the BMI correlated with Paco2. However, the BMI was not an independent factor for predicting Paco2 on multiple regression analysis. Instead, the %VC, in addition to the mean Sao2, was identified by stepwise multiple regression analysis as a predictor of Paco2. The %VC was also significantly different between hypercapnic and normocapnic patients and correlated with Paco2 in the hypercapnic patients. Because the BMI also correlated with %VC (r ⫽ ⫺0.300; p ⬍ 0.01), it is likely that the effects of BMI on Paco2 were predominantly due to changes in the %VC. Therefore, obesity may affect the Paco2 through its effects on pulmonary volumes. Stepwise multiple regression analysis selected three variables (ie, mean Sao2, %VC, and Pao2) as independent factors for predicting Paco2 in obese patients, explaining 41% of the total variance in the Paco2, whereas the mean Sao2 only explained 13% of the variance in the Paco2 in nonobese patients. The mean Sao2 and %VC were also independent factors for determining the Paco2 in hypercapnic patients. These results suggest that the restrictive impairment of pulmonary volumes caused by obesity had an additive effect on the increase in Paco2, which is consistent with previous results.7,10 Although we hypothesized that the reduction in VC was due to obesity because the %VC correlated significantly with BMI, this change may reflect poor patient effort or neuromuscular dysfunction. However, the pulmonary function tests were performed repeatedly by trained technicians, and we took medical histories to rule out neuromuscular disease. Although we found that desaturation during sleep and restrictive pulmonary disturbances were the most significant determinants of Paco2 in hypercapCHEST / 121 / 2 / FEBRUARY, 2002

419

nic and obese patients with OSAS, these models accounted for less than half of the variance in the Paco2. A number of previous studies1,2,5,6,9,10,29 –32 have examined the relationship between hypercapnia and ventilatory responses in patients with OSAS. Although the results were not always consistent in these studies, several studiesl,2,29,30 demonstrated that hypercapnic or hypoxic ventilatory responses are diminished in hypercapnic patients with OSAS or OHS and that defects in the ventilatory response play an important role in the development of chronic hypercapnia. Because we did not measure ventilatory responses in this study, we cannot comment on how ventilatory responses contribute to hypercapnia in patients with OSAS. It is possible that diminished ventilatory responses to hypercapnia or hypoxia may account for the remaining variability in the Paco2 in this study. Although we found that oxygen desaturation (ie, mean Sao2) during sleep plays a major role in the development of daytime hypercapnia in patients with OSAS, the mechanisms responsible for the relationship between nocturnal hypoxemia and daytime hypercapnia are still unknown. Chronic hypoxemia and sleep fragmentation may affect respiratory control. Decreased lung volumes cause an increase in the alveolar-arterial oxygen tension gradient and hypoxemia. A lower daytime Pao2 is associated with greater nocturnal desaturation and, thus, is more likely to affect respiratory control. Therefore, respiratory control could determine the relationship between daytime hypercapnia and nocturnal hypoxemia. However, the mechanical restrictive impairment, as assessed by the VC measurement in this study, may have additive effects with these mechanisms. When the gas exchange abnormality caused by upper airway collapse becomes severe, the normalization of Pao2 and Paco2 may be incomplete during the postapneic period. The normalization of gas exchange abnormalities is dependent on the magnitude of the increase in ventilation, which is determined by chest wall mechanics33 and chemosensitivity.34 We found that restrictive pulmonary impairment is another independent determinant of daytime hypercapnia, although we did not measure chemosensitivity. This mechanical restrictive disorder may attenuate the increase in ventilation and cause greater nocturnal hypoxemia and hypercapnia during sleep, thereby affecting daytime gas exchange. Javaheri et al10 also found that the patients with hypercapnic OSAS and a diminished ventilatory response have a lower maximum voluntary ventilation than did normocapnic OSAS patients. Finally, the findings of the present study are consistent with the findings of another large population study.7 Leech et al7 have identified the daytime 420

Pao2 and gender as risk factors for the development of hypercapnia and also have shown that the daytime Pao2, AHI, and FVC contribute to the severity of hypercapnia. Although their study population was heterogeneous, including female patients and patients with COPD, the patients in our study were relatively homogeneous (male patients with OSAS) and the findings were similar. Specifically, nocturnal oxygen desaturation and pulmonary restrictive impairment were found to be major risk factors for the development of daytime hypercapnia in patients with OSAS in both studies. References 1 Zwillich CW, Sutton FD, Pierson DJ, et al. Decreased hypoxic ventilatory drive in the obesity-hypoventilation syndrome. Am J Med l975; 59:343–348 2 Garay SM, Rapoport D, Sorkin B, et al. Regulation of ventilation in the obstructive sleep apnea syndrome. Am Rev Respir Dis 1981; 124:451– 457 3 Jones JB, Wilhoit SC, Findley LJ, et al. Oxyhemoglobin saturation during sleep in subjects with and without obesityhypoventilation syndrome. Chest l985; 88:9 –15 4 Bradley TD, Rutherford R, Lue F, et al. Role of diffuse airway obstruction in the hypercapnia of obstructive sleep apnea. Am Rev Respir Dis 1986; 134:920 –924 5 Rapoport DM, Garay SM, Epstein H, et al. Hypercapnia in the obstructive sleep apnea syndrome: a reevaluation of the Pickwickian syndrome. Chest 1986; 89:627– 635 6 Borthon-Jones M, Sullivan CE. Time course of change in ventilatory response to CO2 with long-term CPAP therapy for obstructive sleep apnea. Am Rev Respir Dis 1987; 135:144 – 147 7 Leech JA, Onal E, Baer P, et al. Determinants of hypercapnia in occlusive sleep apnea syndrome. Chest 1987; 92:807– 813 8 Chan SC, Grunstein PR, Bye PTP, et al. Obstructive sleep apnea with severe chronic airflow limitation. Am Rev Respir Dis 1989; 140:1274 –1278 9 Javaheri S, Colangelo G, Corser B, et al. Familial respiratory chemosensitivity does not predict hypercapnia of patients with sleep apnea-hypopnea syndrome. Am Rev Respir Dis 1992; 145:837– 840 10 Javaheri S, Colangelo G, Lacy W, et al. Chronic hypercapnia in obstructive sleep apnea-hypopnea syndrome. Sleep 1994; 17:416 – 423 11 Burrows B, Earl RH. Course and prognosis of chronic obstructive lung disease. N Engl J Med 1969; 280:379 – 404 12 Altose MD, McCauley WC, Kelsen SG, et al. Effects of hypercapnia and inspiratory flow-resistive loading on respiratory activity in chronic airway obstruction. J Clin Invest 1977; 59:500 –507 13 Rebuck AS, Slutsky AS. Control of breathing in disease of the respiratory tract and lungs. In: Cherniack NS, Widdicombe JG, eds. Handbook of physiology: control of breathing (vol 2). Bethesda, MD: American Physiological Society, 1986; 771– 792 14 Oliven A, Kelsen SG, Deal EC, et al. Mechanisms underlying CO2 retention during flow-resistive loading in patients with chronic obstructive pulmonary disease. J Clin Invest 1983; 71:1442–1449 15 Leech JA, Onal E, Dulberg C, et al. A comparison of men and women with occlusive sleep apnea syndrome. Chest 1988; 94:983–988 Clinical Investigations

16 Kunitomo F, Kimura H, Tatsumi K, et al. Sex differences in awake ventilation drive and abnormal breathing during sleep in eucapnic obesity. Chest 1988; 93:968 –976 17 Sin DD, Jones RL, Man GC. Hypercapnic ventilatory response in patients with and without obstructive sleep apnea: do age, gender, obesity, and daytime Pco2 matter? Chest 2000; 117:454 – 459 18 Rechtschaffen A, Kales A. A manual of standardized terminology, techniques and scoring systems for sleep stages of human subjects. Washington, DC: National Institute of Health, 1968; publication No. 204 19 Gould GA, Whyte KF, Rhind GB, et al. The sleep hypopnea syndrome. Am Rev Respir Dis 1988; 137:895– 898 20 Tsai WH, Flemons WW, Whitelaw WA, et al. A comparison of apnea-hypopnea indices derived from different definitions of hypopnea. Am J Respir Crit Care Med 1999; 159:43– 48 21 Goldman HI, Becklake MR. Respiratory function test: normal values at median altitudes and the prediction of normal results. Am Rev Respir Dis 1959; 79:457– 467 22 Krachman S, Criner GJ. Hypoventilation syndrome. Clin Chest Med 1998; 19:l39 –155 23 Burwell CS, Robin ED, Whaley RD, et al. Extreme obesity associated with alveolar hypoventilation: a Pickwickian syndrome. Am J Med 1956; 21:811– 818 24 Rochester DF, Enson Y. Current concepts in the pathogenesis of the obesity-hypoventilation syndrome. Am J Med 1974; 57:402– 420 25 Onal E, Leech JA, Lopata M. Relationship between pulmonary function and sleep-induced respiratory abnormalities. Chest 1985; 87:437– 441

26 Weizenblum E, Loiseau A, Hirth C, et al. Course of pulmonary hemodynamics in patients with chronic obstructive pulmonary disease. Chest 1979; 75:659 – 662 27 Weizenblum E, Sautegeau A, Ehrhart M, et al. Long term course of pulmonary arterial pressure in chronic obstructive pulmonary disease. Am Rev Respir Dis 1984; 130;993–998 28 Fletcher EC, Donner CF, Midgren B, et al. Survival in COPD patients with daytime Pao2 ⬎ 60 mm Hg with and without nocturnal oxygen desaturation. Chest 1992; 101:649 – 655 29 Rajagopal KR, Abbrecht PH, Tellis CJ. Control of breathing in obstructive sleep apnea. Chest 1984; 85:174 –180 30 Lopata M, Onal E. Mass loading, sleep apnea, and the pathogenesis of obesity hypoventilation. Am Rev Respir Dis 1982; 126:640 – 645 31 Benlloch E, Cordero P, Morales P, et al. Ventilatory pattern at rest and response to hypercapnic stimulation in patients with obstructive sleep apnea syndrome. Respiration 1995; 62:4 –9 32 Verbraecker J, De Backer W, Willemen M, et al. Chronic CO2 drive in patients with obstructive sleep apnea and effect of CPAP. Respir Physiol 1995; 101:279 –287 33 Cherniack RM, Snidal DP. The effect of obstruction to breathing on the ventilatory response to CO2. J Clin Invest 1956; 35:1286 –1290 34 Satoh M, Hida W, Chonan T, et al. Role of hypoxic drive in regulation of postapneic ventilation during sleep in patients with obstructive sleep apnea. Am Rev Respir Dis 1991; 143:481– 485

CHEST / 121 / 2 / FEBRUARY, 2002

421