Patients with pulmonary sarcoidosis may variously

Transcription

1 Membrane and Capillary Blood Components of Diffusion Capacity of the Lung for Carbon Monoxide in Pulmonary Sarcoidosis* Relation to Exercise Gas Exchange Christine Lamberto, MD; Hilario Nunes, MD; Philippe Le Toumelin, MD; Florence Duperron, MD; Dominique Valeyre, MD; and Christine Clerici, MD, PhD Background: Resting pulmonary diffusing capacity of the lung for carbon monoxide (DLCO) is known to be the best predictor of arterial desaturation during exercise in patients with sarcoidosis. However, the relative contribution of each of the two components of DLCO alveolar membrane diffusing capacity (Dm) and pulmonary capillary blood volume (Vc) remains unclear. Study objectives: To evaluate which component is responsible for the decrease of resting DLCO in patients with sarcoidosis, and to determine which resting pulmonary function test, including Dm and Vc, is the best predictor of gas exchange abnormalities during submaximal exercise. Design: Prospective analysis of patients referred to our department of respiratory medicine. Patients: Twenty four patients with pulmonary sarcoidosis were separated into two groups according to chest radiographic findings: group 1, stages 2 and 3 (n 15); group 2, stage 4 (n 9). All the patients completed pulmonary function tests (flows, volumes, single-breath DLCO, transfer coefficient [Ka], Dm, Vc) and submaximal exercise (two steady-state levels of mild and moderate exercise corresponding respectively to a target oxygen consumption of approximately 10 to 15 ml/min/kg). Results: DLCO was reduced in the two groups (group 1, 63 16% of predicted; group 2, 64 16% of predicted). Dm was severely decreased (group 1, 58 24% of predicted; group 2, 51 15% of predicted), whereas Vc was unchanged or only mildly decreased (group 1, 81 18% of predicted; group 2, 85 28% of predicted). Whatever the group of patients and the exercise level, Dm and DLCO were the strongest predictors (p < 0.001) of gas exchange abnormalities. Ka or volumes were weak predictors, and Vc or flows were not related with exercise gas exchange. Conclusions: This study demonstrates that a decrease in Dm mostly accounts for resting DLCO reduction, and that Dm as well as DLCO are highly predictive of gas exchange abnormalities at exercise in patients with sarcoidosis. (CHEST 2004; 125: ) Key words: exercise testing; membrane diffusing capacity; pulmonary capillary blood volume; pulmonary diffusing capacity; pulmonary gas exchange; pulmonary sarcoidosis Abbreviations: Dlco lung diffusing capacity for carbon monoxide; Dm membrane diffusing capacity; E1 mild exercise; E2 moderate exercise; Fio 2 inspiratory oxygen fraction; Ka carbon monoxide transfer coefficient; MMEF maximum middle expiratory flow between 25% and 75% of FVC; P(A-a)O 2 alveolar-arterial oxygen pressure difference; P(a-et)CO 2 arterial-end-tidal carbon dioxide difference; Pao 2 alveolar oxygen pressure; rate of carbon monoxide uptake of whole blood; TLC total lung capacity; Vc pulmonary capillary blood volume; VC slow vital capacity; Vd/Vt dead space to tidal volume ratio; V o 2 oxygen consumption; Vt tidal volume Patients with pulmonary sarcoidosis may variously combine a decrease in lung volumes, a loss of diffusion capacity of the lung for carbon monoxide (Dlco), and gas exchange abnormalities during exercise, including decreased Pao 2 and increased alveolar-arterial oxygen pressure difference (P[Aa]O 2 ). 1 8 Dlco and arterial desaturation at exercise have been reported to be the strongest functional parameters that correlate with the extent and the severity of sarcoidosis, assessed by either pathologic scores 9,10 or high-resolution CT. 2 Interestingly, Dlco has also been found to be fairly correlated with gas exchange abnormalities at exercise, and particularly to be the best predictive and sensitive index of a fall in Pao ,11 Alteration of Dlco may result from changes in gas CHEST / 125 / 6/ JUNE,

2 exchange area, barrier thickness, and ventilationperfusion-diffusion mismatching of the lung. Insofar as the pulmonary capillary endothelium accounts for a part of the gas exchange area and membrane thickness, abnormalities in Dlco may also reflect involvement of pulmonary capillaries. Vascular involvement is common in pulmonary sarcoidosis since it is observed between 42% and 100% of cases on histologic examination. 9,12,13 Moreover, Emirgil et al 14 observed a good correlation between pulmonary vascular resistance and Dlco, supporting the contribution of vascular involvement in the decrease in Dlco. Using the classic method of Roughton and Forster, 15 it is possible to partition Dlco into its two components: alveolar capillary membrane diffusing capacity (Dm) and pulmonary capillary blood volume (Vc). To our knowledge, only two studies 16,17 have evaluated which component of Dlco is altered in pulmonary sarcoidosis. However, to date, no study has examined the relationship between Dm and Vc, and gas exchange at exercise in these patients. The main purposes of the study were as follows: (1) to evaluate the mechanisms of loss in Dlco in patients with pulmonary sarcoidosis, by measuring its two components, Dm and Vc; and (2) to determine within resting pulmonary function tests including Dm and Vc, which is the best predictor of gas exchange abnormalities during submaximal exercise. Patients Materials and Methods Twenty-four patients with pulmonary sarcoidosis histologically confirmed were prospectively studied in our department of physiology between February 1999 and June All the patients had pulmonary infiltration shown on chest radiography and were classified according to American Thoracic Society staging. 18 Group 1 (n 15) included patients with radiographic stages 2 or 3 (no evidence of pulmonary fibrosis), and group 2 (n 9) included patients with radiographic stage 4 (pulmonary fibrosis). Patients were not included if they had another respiratory disease, cardiovascular disease, cardiac or muscular involvement of sarcoidosis, if they were treated with -adrenergic antagonists, or if they had a contraindication to exercise testing. *From the Departments of Physiology (Drs. Lamberto, Duperron, and Clerici), Respiratory Medicine (Drs. Nunes and Valeyre), and Anesthesiology (Dr. Le Toumelin), Assistance Publique/ Hôpitaux de Paris, Hôpital Avicenne, Bobigny, France. Manuscript received July 15, 2003; revision accepted January 6, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( permissions@chestnet.org). Correspondence to: Christine Lamberto, MD, Department of Physiology, Hôpital Avicenne, 125 route de Stalingrad Bobigny, France; christine.lamberto@avc.ap-hop-paris.fr We have not included radiographic stage 1 patients because they usually do not have exercise desaturation. All patients provided informed consent. The overall population consisted of 11 men and 13 women with a mean age of years ( SD). Fifteen of the 24 patients had at least one extrathoracic localization of sarcoidosis. Eighteen patients were nonsmokers, 5 were ex-smokers, and 1 patient of group 1 was a current smoker (8 pack-years). At the moment of inclusion, 10 patients received oral corticosteroids (4 of 15 patients vs 6 of 9 patients, respectively, for groups 1 and 2). The two groups of patients were similar, except for age (40 11 years vs years, respectively, for groups 1 and 2; p 0.05). Pulmonary Function Tests Slow vital capacity (VC) and total lung capacity (TLC) were recorded using a body plethysmograph. FEV 1 /VC and maximum middle expiratory flow between 25% and 75% of FVC (MMEF ) were obtained from flow-volume curves. Dlco and carbon monoxide transfer coefficient (Ka) were measured by the single-breath method, according to the recommendations of the European Respiratory Society. 19 Results were corrected for hemoglobin and carboxyhemoglobin concentration according to Cotes. 20 Dm and Vc were determined using the method of Roughton and Forster, 15 modified by Cotes. 20 As oxygen and carbon monoxide compete directly for the available hemoglobin sites, the rate of carbon monoxide uptake of whole blood ( ) is inversely proportional to alveolar oxygen pressure (Pao 2 ) [and 1/ proportional to Pao 2 ]. Then, Dlco decrease as Pao 2 increases, following the following equation: 1 Dlco 1 Dm 1 Vc where 1/Dlco is the total resistance to diffusion, 1/Dm is the membrane resistance, and 1/ Vc is the pulmonary capillary blood fixation resistance. Dm and Vc can be estimated from Dlco measured at two levels of inspired oxygen fraction (Fio 2 ): normal and high. For each Dlco measurement, we considered the mean of at least two reproducible sets (within 10%) with an interval of 5 min between each set. The sequence of measurements was in the following order: standard Dlco (at normal Fio 2 ) was measured first (inspiratory mixture: 9.48% helium, 0.264% carbon monoxide, and 20% oxygen), with the subject breathing room air between each set. Second, the subject was allowed to breath pure oxygen for at least 5 min to complete alveolar nitrogen washout. Then, high-fio 2 Dlco was measured (inspiratory mixture: 7.74% helium, 0.264% carbon monoxide, and 92% oxygen) without discontinuing to breathe pure oxygen between each set. Pao 2 was monitored during washout and analyzed in the alveolar gas sample used for Dlco measurement. 1/ was calculated according to Cotes, 21 and took into account the measured Pao 2. For predicted values, we used European Coal and Steel Community 22 for volumes and Dlco, Cotes 20 for Dm and Vc, and Knudson et al 23 for flow-volume curve. The individual values of pulmonary function test are expressed as percentage of predicted values. Exercise Testing Studies were performed using a treadmill exercise protocol allowing steady-state alveolar-arterial measurement for mild and moderate exercise intensity. The exercise test was preceded by a practice trial in order to familiarize the patient with the treadmill and to determine the more comfortable walking speed (from 3 to 5 kilometers per hour). After 3 min of stable resting values, the 2062 Clinical Investigations

3 patient underwent two standardized steps from 3 to 5 min each, corresponding to a target oxygen consumption (V o 2 ) of approximately 10 ml/min/kg (mild exercise [E1]) and 15 ml/min/kg (moderate exercise [E2]). The first step corresponded to slow walking (from 2 to 3 kilometers per hour), and the second step corresponded to normal walking (3 to 5 kilometers per hour) with or without a small grade. Exercise was stopped before the E2 level when pulse oximetry saturation fell below 80%. At rest and throughout exercise, a 12-lead ECG was monitored continuously. Heart rate, tidal volume (Vt), respiratory rate, minute ventilation, mouth oxygen and carbon dioxide concentration, pulse oximetry, and end-tidal CO 2 were monitored on line (Oxycon ; Jaeger; Hoechberg, Germany; and Biox 3740; Datex- Ohmeda; Louisville, CO). For each steady-state step, blood gases (Pao 2 and Paco 2 ) [ABL 725; Radiometer; Copenhagen, Denmark] were measured from arterialized capillary blood sample at prewarmed ear lobe (capsaicin cream). In all patients, the quality of the arterialized capillary blood Po 2 recorded at rest before exercise was attested by the lack of difference (paired t test), with the radial Pao 2 recorded for the pulmonary function test within the same week ( for the arterialized sample vs mm Hg for the radial sample). From these measured values, we calculated V o 2, carbon dioxide production, and respiratory quotient. Pao 2 was determined using alveolar air equation with arterialized Paco 2 and measured respiratory quotient. It allowed us to calculate P(A-a)O 2 [Pao 2 Pao 2 ]. Physiologic dead space to Vt ratio (Vd/Vt) was calculated according to the Bohr formula using arterialized Paco 2 : Vd/Vt (Paco 2 Peco 2 ) Vds Paco 2 Vt where Peco 2 is the mixed expired Pco 2, and Vds is the system dead space. Arterial-end-tidal CO 2 pressure difference (P[a-et]CO 2 ) was calculated as the difference between arterialized and end-tidal carbon dioxide pressure. Predicted values for gas exchange at rest or exercise were taken from Wasserman et al. 24 Statistical Methods Statistical analyses were carried out using SAS version 8.2 (SAS Institute; Cary, NC). Values are expressed as mean SD. Comparison of measured values of pulmonary function tests with predicted values were made using a Student paired t test. Comparison between the two groups were made with a Student t test after a Kurtosis and Bartlett test was used to ascertain that samples had equal variances. The relationships between resting and exercise parameters were studied with linear regression. In order to determine the best predictor test for gas exchange abnormalities at exercise, multivariate analysis with a general linear model (GLM Procedure; SPSS; Chicago, IL) was used. For this model, the variables described were Pao 2, P(A-a)O 2, and Vd/Vt at rest and at each exercise level. Three explanatory parameters were chosen: an exchange parameter, Dm, Vc, or Dlco; a volume parameter, VC or TLC; and a flow parameter, MMEF or FEV 1 /VC. The robustness of the model was tested by the method of adding or withdrawal the extreme values; p 0.05 was considered significant. Results Pulmonary Function Results of function tests are summarized in Table 1. Both groups demonstrated a moderate reduction Table 1 Resting Pulmonary Function* Variables Group 1 (n 15) Group 2 (n 9) Overall (n 24) VC, % predicted TLC, % predicted Actual FEV 1 /VC, % FEV 1 /VC, % predicted MMEF 25 75, % predicted Dlco, % predicted Ka, % predicted Dm, % predicted Vc, % predicted *Data are presented as mean SD. Measured values are compared with predicted values using paired t test. p p p The only significant difference between groups was found for actual value of FEV 1 /VC, using t test, p in lung volumes and airflows. Dlco was severely reduced in the two groups (63 17% of predicted in group 1, p 0.05; 64 16% of predicted in group 2, p 0.05). The actual value of FEV 1 /VC was smaller in group 2 than in group 1, but this difference disappeared when FEV 1 /VC was expressed in percentage of predicted value. Since FEV 1 /VC decreases with age, this finding likely reflects the age difference between the two groups. Analysis of the two components of Dlco showed that the reduction of Dlco was mainly due to a dramatic decrease of Dm in the two groups (58 24% of predicted in group 1, p 0.05; 51 15% of predicted in group 2, p 0.05) whereas Vc was either unchanged in group 2 or only moderately reduced in group 1 (Table 1, Fig 1). Gas exchange data at rest are shown on Table 2. Pao 2 was normal with a slight increase in P(A-a)O 2. There was a slight widening in P(a-et)CO 2, but Vd/Vt was normal. None of the Figure 1. Individual values of Dlco, Dm, and Vc are represented in both groups of patients. Group 1: Radiographic stage 2 or 3. Group 2: Radiographic stage 4. Mean values in each group are indicated with a dark line. CHEST / 125 / 6/ JUNE,

4 Table 2 Parameters of Gas Exchange at Rest and Exercise* Variables resting parameters was significantly different between the two groups (except for actual value of FEV 1 /VC). Exercise Test Group 1 (n 15) Group 2 (n 9) Overall Population (n 24) Pao 2 at rest, Pao 2 E1, Pao 2 E2, P(A-a)O 2 at rest, P(A-a)O 2 E1, P(A-a)O 2 E2, P(a-et)CO 2 at rest, P(a-et)CO 2 E1, P(a-et)CO 2 E2, Vd/Vt at rest, % Vd/Vt E1, % Vd/Vt E2, % *Data are presented as mean SD. Measurements were performed at rest and at two levels of exercise: E1 (10 ml/min/kg of V o 2 ) and E2 (15 ml/min/kg of V o 2 ). Gas exchange data at exercise are reported in Table 2 and Figure 2. An important drop in Pao 2 was observed as a function of exercise intensity concomitantly with a large widening of P(A-a)O 2 (Fig 2). P(a-et)CO 2 and Vd/Vt were increased at exercise. For the two steady-state steps of exercise (10 ml/ min/kg and 15 ml/min/kg V o 2 ), gas exchange parameters were not significantly different between the two groups. Relations Between Resting Lung Function and Exercise Gas Exchange Values in the Overall Population Since there was no significant difference in pulmonary function and gas exchange characteristics between the two groups, linear regressions and the GLM Procedure were performed in the overall population. Using linear regression, the strongest relations between pulmonary function parameters and exercise Pao 2 or P(A-a)O 2 were evidenced with Dm and Dlco (p 0.001), but the best r values were observed with Dm for each step of exercise (Table 3, Fig 3). Also, Dm and Dlco were correlated with exercise Vd/Vt particularly at the E2 level (Table 3). Ka relations were far weaker than those found with Dlco or Dm. By contrast, neither Vc nor volumes or expiratory flows were related with exercise gas exchange parameters (Table 3). Figure 3 illustrates the main results for the E2 level. Multivariate analysis was performed to determine the best resting functional predictor of exercise gas exchange abnormalities. Table 4 illustrates the results obtained with VC, MMEF 25 75, and Dm (these parameters were those with the best coefficients on linear regression). Dm and Dlco were very strong predictors of exercise Pao 2 and P(A-a)O 2 (p 0.001). Yet, when Dm and Dlco were included in the same analysis (linked parameters), Dlco failed to reach any significance. By contrast, volumes, flows or Vc were never significant predic- Figure 2. Evolution of Pao 2 and P(A-a)O 2 with exercise in the two groups of patients. Pao 2 (left, A) and P(A-a)O 2 (right, B) are plotted against V o 2. Group 1: Radiographic stage 2 or 3. Group 2: Radiographic stage 4. Values are the mean SD in each group Clinical Investigations

5 Table 3 Correlation Between Resting Pulmonary Function Tests and Gas Exchange Parameters at Exercise in the Overall Population* Variables Pao 2 E1 Pao 2 E2 P(A-a)O 2 E1 P(A-a)O 2 E2 Vd/Vt E1 Vd/Vt E2 Dlco, % predicted 0.71 ( ) 0.74 ( ) 0.75 ( ) 0.79 ( ) ( ) Ka, % predicted ( ) 0.51 ( ) 0.51 ( ) Dm, % predicted 0.87 ( ) 0.82 ( ) 0.84 ( ) 0.81 ( ) 0.50 ( ) 0.65 ( ) Vc, % predicted VC, % predicted 0.47 ( ) TLC, % predicted FEV 1 /VC, % predicted MMEF 25/75, % predicted *Analysis was performed in the overall population at both levels of exercise E1 and E2, using linear regressions. Values are regression coefficients. positive slope of the relation; negative slope of the relation. p p p tors. Results were not modified when using TLC in place of VC or FEV 1 /VC in place of MMEF A threshold inducing a marked widening of P(A-a)O 2 at the E2 level was manually assessed for Dlco or Dm on Figure 3: Dlco 65% and Dm 60%. Figure 4 shows the evolution of P(A-a)O 2 with V o 2 for each subject as a function of Dm or Dlco threshold. This threshold permits the separation of patients who demonstrate a large increase of P(A-a)O 2 even for a mild exercise from those with only moderate change in P(A-a)O 2. Discussion The present study provides the first comprehensive analysis of the relation between the two compo- Figure 3. Relation between exercise P(A-a)O 2 and resting Dlco, Dm, Vc, or Ka. Analysis was performed in overall population at the E2 level of exercise using linear regression between P(A-a)O 2 and Dlco (top left, A), Ka (top right, B), Dm (bottom left, C), or Vc (bottom right, D); r is the regression coefficient. Group 1: Radiographic stage 2 or 3. Group 2: Radiographic stage 4. Pred predicted. CHEST / 125 / 6/ JUNE,

6 Table 4 Resting Prediction of Gas Exchange Abnormalities at Exercise* Variables Dm VC MMEF Pao 2 E Pao 2 E P(A-a)O 2 E P(A-a)O 2 E Vd/Vt E Vd/Vt E *Multivariate analysis was a general linear model. Shown are the p values obtained when analysis was performed with Dm, VC, and MMEF as explanatory variables; p 0.05 was considered significant. nents of Dlco Dm and Vc and exercise gas exchange in pulmonary sarcoidosis. It clearly shows that the reduction of Dm mostly accounts for the decrease of resting Dlco, and that this parameter is the best resting predictor of gas exchange abnormalities during exercise. In sarcoidosis, alteration of Dlco may reflect changes in gas exchange area, barrier thickness, and ventilation-perfusion-diffusion mismatching of the lung. The assessment of the two components of Dlco Dm and Vc may be useful to understand the underlying mechanisms of its alteration. In our study, resting Dlco was impaired at rest in the two groups of patients without (radiographic stage 2 3) or with pulmonary fibrosis (radiographic stage 4). We demonstrate that in both groups the decrease of Dm is the major determinant of impaired Dlco, whereas changes in Vc are relatively minor. To date, only two studies have evaluated membrane and capillary components of Dlco in sarcoidosis. In early stages of sarcoidosis without pulmonary infiltration (radiographic stage 0 and 1), Dujic et al 17 reported an increase of Dm while Vc was normal. Similar to our findings, Saumon et al 16 showed that the reduced Dlco in pulmonary sarcoidosis was mainly related to a decrease of Dm with a moderate fall in Vc in patients with fibrosis. The mild impairment of Vc may seem especially intriguing regarding two points previously described in pulmonary sarcoidosis: first, vascular involvement is known to be highly frequent, estimated between 42% and 100% of cases 9,12,13 ; second, Emirgil et al 14 evidenced a good correlation between pulmonary vascular resistance and Dlco, suggesting a restriction of the pulmonary vascular bed in this disorder. Yet, the measure of Vc in resting conditions may not be a sensitive indicator of lung vascular involvement, because of the variability of cardiac output and of vascular recruitment at rest, particularly in sarcoidosis where the distribution of pulmonary lesions is heterogeneous. In fact, a recent review by Hsia 25 pointed out that in several pulmonary diseases there is a vascular recruitment at rest, which results in diverting the blood flow from the affected alveolar capillary units to the remaining unaffected zones. 25 This functional compensating mechanism leads to a Dlco recruitment at rest with the persistence of low diffusionperfusion ratio. Our results support this mechanism. First, vascular recruitment at rest may explain the fact that Vc is mildly altered and that Figure 4. Evolution of P(A-a)O 2 with exercise for each patient as a function of their Dlco or Dm values. Patients were separated according to their Dlco value (left, A) or Dm value (right, B). The thresholds of 65% of the predicted value for Dlco or 60% for Dm were manually determined on Figure 3. See Figure 3 legend for expansion of abbreviation Clinical Investigations

7 Dm is less affected than Dlco. Second, the Vd/Vt being more abnormal at exercise than at rest agrees with a redistribution of pulmonary flow at rest. Resting Dlco is clearly known to be a good predictor of abnormal exercise gas exchange in pulmonary sarcoidosis. 1 4 Karetsky and McDonough 3 showed that Dlco 55% had a high sensitivity (85%) and specificity (91%) in predicting the fall in Pao 2 at exercise. Risk et al 1 showed that patients with sarcoidosis who had a Dlco 50% also evidenced a fall in Pao 2 at exercise. Our results indicate that, whatever the group, only patients with Dlco 65% or Dm 60% have a marked increase in P(A-a)O 2 even for mild exercise (intensity of 10 ml/min/kg of V o 2 ), and that both Dlco and Dm are related to exercise-induced fall in Pao 2 and increase in P(A-a)O 2. The relation of Dm with these exercise parameters is stronger than Dlco, and Ka was a far weaker predictor than Dm or Dlco (Table 3, Fig 3). Dm is the only independent resting predictor of Pao 2 and P(A-a)O 2 at exercise (Table 4). In addition, Vc was not significantly related to gas abnormalities at exercise. The severity of exercise desaturation is known to depend on overall pulmonary involvement, which may be better approached by Dm than Dlco or Vc, as a consequence of vascular recruitment or variable cardiac output at rest. We did not assess pulmonary arterial pressure and cardiac output in this study, and this point will require further investigation. A previous study 5 has already demonstrated that Dlco failed to increase with exercise, but the measure of its two components at exercise would be very interesting to accurately understand the mechanisms of exercise-induced gas exchange abnormalities. In our study, the physiologic features at rest of the two groups without or with pulmonary fibrosis are not different. Two reasons can be evoked: first, the group of patients without fibrosis is more severe than expected in terms of pulmonary function, 1 3,7 probably because our respiratory department is a reference center for sarcoidosis; second, the severity of sarcoidosis varies widely in each group, resulting in a large SD. In addition to Dlco and Dm, exercise behavior is similar between the two groups. Then, neither resting Dm or Dlco nor exercise desaturation can discriminate inflammatory and fibrotic changes in pulmonary sarcoidosis, as previously proved with Dlco in fibrosing alveolitis. 26,27 Conclusion In conclusion, this study demonstrates that in sarcoidosis without or with pulmonary fibrosis, the decrease of Dlco is mainly related to its Dm component and that Dm is the best predictor of abnormal gas exchange at exercise. Vc is only mildly altered, probably because of vascular recruitment at rest. Vascular involvement in patients with sarcoidosis might be more accurately approached by analyzing Dlco and its two components at exercise. References 1 Risk C, Epler GR, Gaensler EA. Exercise alveolar-arterial oxygen pressure difference in interstitial lung disease. Chest 1984; 85: Medinger AE, Khouri S, Rohatgi PK. Sarcoidosis: the value of exercise testing. Chest 2001; 120: Karetzky M, McDonough M. Exercise and resting pulmonary function in sarcoidosis. Sarcoidosis Vasc Diffuse Lung Dis 1996; 13: Sietsema KE, Kraft M, Ginzton L, et al. Abnormal oxygen uptake responses to exercise in patients with mild pulmonary sarcoidosis. Chest 1992; 102: Ingram CG, Reid PC, Johnston RN. Exercise testing in pulmonary sarcoidosis. Thorax 1982; 37: Miller A, Brown LK, Sloane MF, et al. Cardiorespiratory responses to incremental exercise in sarcoidosis patients with normal spirometry. Chest 1995; 107: Spiro SG, Dowdeswell IR, Clark TJ. An analysis of submaximal exercise responses in patients with sarcoidosis and fibrosing alveolitis. Br J Dis Chest 1981; 75: Matthews JI, Hooper RG. Exercise testing in pulmonary sarcoidosis. Chest 1983; 83: Carrington C. Structure and function in sarcoidosis. Ann N Y Acad Sci 1976; 278: Young RL, Lordon RE, Krumholz RA, et al. Pulmonary sarcoidosis: 1. Pathophysiologic correlations. Am Rev Respir Dis 1968; 97: Sue DY, Oren A, Hansen JE, et al. Diffusing capacity for carbon monoxide as a predictor of gas exchange during exercise. N Engl J Med 1987; 316: Rosen Y, Moon S, Huang CT, et al. Granulomatous pulmonary angiitis in sarcoidosis. Arch Pathol Lab Med 1977; 101: Takemura T, Matsui Y, Saiki S, et al. Pulmonary vascular involvement in sarcoidosis: a report of 40 autopsy cases. Hum Pathol 1992; 23: Emirgil C, Sobol BJ, Herbert WH, et al. The lesser circulation in pulmonary fibrosis secondary to sarcoidosis and its relationship to respiratory function. Chest 1971; 60: Roughton F, Forster R. Relative importance of diffusion and chemical reaction rates in determining rate of exchange of gases in the human lung with special reference to true diffusing capacity of pulmonary membrane and volume of blood in the lung capillaries. J Appl Physiol 1957; 11: Saumon G, Georges R, Loiseau A, et al. Membrane diffusing capacity and pulmonary capillary blood volume in pulmonary sarcoidosis. Ann N Y Acad Sci 1976; 278: Dujic Z, Tocilj J, Eterovic D. Increase of lung transfer factor in early sarcoidosis. Respir Med 1995; 89: Statement of sarcoidosis: joint statement of the American Thoracic Society (ATS), the European Respiratory Society (ERS) and the World Association of Sarcoidosis and Other Granulomatous Disorders (WASOG) adopted by the ATS Board of Directors and by the ERS Executive Committee, February Am J Respir Crit Care Med 1999; 160: CHEST / 125 / 6/ JUNE,

8 19 Cotes JE, Chinn DJ, Quanjer PH, et al. Standardization of the measurement of transfer factor (diffusing capacity): Report Working Party Standardization of Lung Function Tests, European Community for Coal and Steel; official position of the European Respiratory Society. Eur Respir J Suppl 1993; 16: Cotes J. Reference values for adults of European descent: lung function. Boston, MA: Blackwell Scientific Publications, 1979; Cotes JE. Lung function assessment and application in medicine. 5th ed. Boston, MA: Blackwell Scientific Publications, European Community for Steel and Coal. Standardized lung function testing. Bull Eur Physiol Respir 1983; 19: Knudson RJ, Lebowitz MD, Holberg CJ, et al. Changes in the normal maximal expiratory flow-volume curve with growth and aging. Am Rev Respir Dis 1983; 127: Wasserman K, Jansen J, Sue DY, et al. Normal values. In: Principles of exercise testing and interpretation. Philadelphia, PA: Lea & Febriger, 1987; Hsia CC. Recruitment of lung diffusing capacity: update of concept and application. Chest 2002; 122: Chinet T, Jaubert F, Dusser D, et al. Effects of inflammation and fibrosis on pulmonary function in diffuse lung fibrosis. Thorax 1990; 45: Fulmer JD, Roberts WC, von Gal ER, et al. Morphologicphysiologic correlates of the severity of fibrosis and degree of cellularity in idiopathic pulmonary fibrosis. J Clin Invest 1979; 63: Clinical Investigations