INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Based on the multiple inert gas elimination technique (MIGET), extensive studies of impaired gas exchange in patients with chronic lung diseases, including chronic obstructive pulmonary disease (COPD) and interstitial lung diseases, have been performed by several investigators (1). The main conclusion derived from these studies is that the principal abnormality impeding pulmonary gas exchange is ventilation-perfusion (A/) inequality. This held true irrespective of the underlying disease, suggesting that the reduced arterial P O₂ in various chronic lung diseases can be accounted for primarily by A/ mismatch, with only a minor contribution from diffusion-limited O ₂ exchange. Recently, however, applying the MIGET to patients with interstitial lung diseases, Eklund and colleagues (6) and Agusti and coworkers (7) have shown that abnormalities of gas exchange in interstitial lung diseases are distorted by a diffusion-limited process during exercise as well as at rest, highlighting the necessity of reevaluating the significance of impaired gas exchange caused by diffusion limitation in patients with interstitial diseases. On the other hand, recent studies done by Torres and colleagues (8), Barbera and coworkers (9), and Rossi and colleagues (10) have provided support for the earlier reports on COPD (3), all of which demonstrated that diffusion limitations contribute minimally to the arterial hypoxemia in patients with COPD. All of the aforementioned studies estimated the significance of diffusion-limited O₂ exchange by comparing measured Pa_O2 values with those predicted from MIGET analysis (1). Although this procedure has been found to be practically useful for evaluating the overall contribution of diffusion-limited gas exchange (2, 3, 6), it may not allow precise assessment of diffusion abnormalities. Determination of the distribution of diffusion properties, represented by diffusing capacity (D), appears to be important, along with determining the A/ distribution, in clarifying the relationship between functional and morphologic abnormalities occurring in diseased lungs. We recently devised a method permitting simultaneous estimation of the representative distribution of A/ and the diffusing capacity to perfusion (D/ ) in the lung using nine gases as indicators (11). Applying this method to patients with a variety of chronic lung diseases, we have attempted to clarify the distributions of A/ and of D/ simultaneously and the significance of a diffusion limitation on impeded gas exchange in patients with either idiopathic pulmonary fibrosis (IPF) or COPD. By means of the A/ and D/ distributions thus determined, we have endeavored to shed light on the potential relation between pathologic alterations and the resulting functional abnormalities in patients afflicted with IPF or COPD.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection

The protocols were approved by the institutional review board for human studies, and informed consent was obtained from the subjects. We studied 52 patients with chronic lung disease suffering from either IPF or COPD. In twenty-seven of them, IPF had been diagnosed, and 25 had various manifestations of COPD. The diagnosis of IPF was established according to a modified version of the criteria reported by Fulmer and colleagues (14). High-resolution computed tomography (HRCT) was performed with the patient in the supine position, holding a breath at deep inspiration, without contrast medium. Images were obtained at 10-mm intervals from the apex to the diaphragm (2.0-mm section thickness). Each image was photographed at window levels of 500 Hounsfield units (HU) and window widths of 1,500 HU. Patients suspected of having IPF were examined by transbronchial lung biopsy (TBLB) to assess the pathologic abnormalities in their lungs. TBLB specimens were generally obtained from the right lower lung field. In addition, isotope examinations with gallium (Ga) were performed in 21 patients. Because precise determination of the IPF disease stage from TBLB specimens is rather difficult because of limited sampling capacity, we classified the lungs of patients with IPF mainly on the basis of HRCT findings. Recent studies done by Muller and colleagues (15) as well as by Nishimura and coworkers (16) have shown that HRCT appearance predicts with considerable accuracy lung histologic features obtained on open lung biopsy, i.e., a ground-glass pattern correlates well with cellular histopathology (alveolitis and/or acinitis), whereas a reticular pattern suggests extensive fibrosis. Therefore, we categorized the patients with IPF into three groups based on HRCT findings (17). Group I (n = 6): ground-glass with no obvious reticular pattern (predominantly inflammatory alveolitis/ acinitis with little fibrosis); Group II (n = 9): reticular pattern associated with honeycombing cysts but without distinct ground-glass opacities (predominantly fibrosis with little alveolitis/acinitis); Group III (n = 12): a mixture of ground-glass and reticular patterns (coexistence of alveolitis/acinitis with fibrosis). Classification of the patients with IPF from HRCT findings was independently made by two of the investigators (one is a physician and the other is a radiologist) who were blind to the details of gas exchange and pulmonary function studies. Only the patients categorized into the same group by the two observers were used for the further analysis.

Subjects with FEV₁/FVC (FEV₁%) below the 95% confidence limit of the values predicted from the equation constructed from the data obtained in nonsmoking Japanese volunteers, with no pulmonary disorders (18), were assumed to have clinically significant chronic airflow obstruction (CAO). The classification of COPD conformed to that proposed by the American Thoracic Society (19), i.e., chronic pulmonary emphysema (CPE), chronic bronchitis (CB), and peripheral airway disease (PAD). Emphysematous lesions were examined by means of HRCT. CT images (2.0-mm section thickness) obtained at the upper (midportion of the intrathoracic trachea) and lower lung field (1 cm above the diaphragm) were photographed at levels of 500 HU and window widths of 1,500 HU. The extent of emphysema in each section was quantified by calculating the relative area of low CT density (%LDA) using the attenuation mask program (20). According to Muller and colleagues (20) and Kinsella and coworkers (21), LDA was defined as an area with a CT density below 910 HU. An average value of the %LDA in the upper and lower lung zones was used as an objective measure of the extent of emphysema. Preliminary examinations of the %LDA of normal volunteers (nonsmokers 28 to 73 yr of age, n = 28) indicated that the average values and 95% confidence limits of %LDA were 6.0 and 19.8%, respectively. In addition, %LDA values in normal subjects were independent of age. Therefore, we assumed that subjects with an average %LDA of more than 20% would have pathologically significant emphysematous changes in their lungs. On the basis of these findings, we classified the subjects with significant CAO and pathologic %LDA but with no other remarkable abnormalities on chest HRCT and minimal productive cough as CPE. Although it is very difficult to precisely classify subjects with chronic bronchitis (CB), we defined such subjects as having persistent cough and expectoration of at least a 3-mo duration for two successive years with sustained CAO (19) but without apparent chest radiographic, including HRCT, abnormalities. PAD was considered to exist when CAO with no emphysematous changes but with bronchiolar abnormalities on HRCT was evident. Bronchiolar abnormalities were based on the recognition of either centrilobular diffuse fine nodules on HRCT, corresponding to bronchiolitis in the acini, or thickened bronchiolar walls with a diameter of less than 3 mm, corresponding to abnormalities of membraneous bronchioles (22). Among the 13 patients manifesting a CPE component, seven and six had CPE only and CPE associated with CB, respectively, There were five subjects who had predominantly PAD, and seven had both PAD and CB components. Among the subjects studied, there were no patients with sustained CAO clearly attributable to CB alone. Recognition of emphysema and bronchiolar abnormalities on HRCT were again made by the two observers without allowing them to know the results of gas exchange studies and pulmonary function tests.

Protocol

As ventilatory indicators, we measured VC, FVC, and the expiratory flow-volume curve by means of an electronic spirometer (MFR-8200; Nihon Kohden, Tokyo). Respiratory impedance was examined by the oscillation method (MZR-4000; Nihon Kohden). FRC, RV, and TLC were measured by the helium-dilution method (Chestac-55V; Chest, Tokyo). Pulmonary function tests were performed within 2-wk before or after right heart catheterization. Steady-state diffusing capacity (DL_CO) and AaPO₂ were estimated from the data obtained during right heart catheterization.

Subjects were placed in a supine position, and a balloon-tipped Swan-Ganz catheter (7F) was introduced into the femoral vein and advanced into the pulmonary artery. An arterial cannula was inserted into the femoral artery. The patients were given a mixture of 21% O₂ and 0.1% CO in N₂ as the inspired gas, and saline containing a small quantity of six foreign inert gases, including sulfur hexafluoride (SF6), ethane, cyclopropane, halothane, diethyl ether, and acetone was infused via the antecubital vein at a rate of 2 ml/min. After a steady state had been established, samples of expired gas and arterial and mixed venous blood were taken simultaneously. We measured the concentrations of O₂ and CO₂ in the expired gas with a mass spectrometer (WSMR-1400; Western, Chiba), and PO₂, PCO₂, and pH in the blood sample were determined with electrodes (Model 1306; Instrumentation Laboratories, Lexington, MA). Inert gases, with the exception of SF6, were analyzed with gas chromatograph equipped with a flame ionization detector (FID) (G.C. 163; Hitachi, Tokyo). SF6 measurements were made with a gas chromatograph with an electron capture detector (G.C. 163; Hitachi). We determined the CO concentration in the sample with FID after converting CO into methane using nickel as a catalyst (MT-20; Gasukuro Kogyo, Tokyo).

A/Q· and D/Q· Analysis

The data obtained during right heart catheterization were analyzed by a recently developed numerical method (11). Briefly, this analytical method is designed to minimize the sum of squares of deviation (SSQ) between the measured arterial concentrations of the nine indicator gases and the values predicted from a lung model in which gas exchange efficiency is taken to be dominated by A/ and D/ distributions. To minimize the SSQ, we modified the enforced smoothing technique elaborated by Wagner and Evans (23, 24). Twenty lung units were assumed along the A/ and along the D/ axis. The A/ ratio is dimensionless, but D/ is expressed in milliliters ( STPD)/ (ml · mm Hg). Although indicator gases have their own D values, they are mutually related via Krogh diffusion constants. Therefore, the D values of indicator gases were normalized to that of O₂ in respective lung units. Our previous study (12) demonstrated that the combination of six inert gases, O₂, CO₂, and CO allowed characterization, with considerable reliability, of the behavior of gas exchange attributed to lung units having A/ ranging from 0.005 to 100 and D/ ranging f rom 0.00001 to 1.0. Furthermore, our previous results (11, 12) suggested that diffusion-limited O₂ exchange occurs in lung units with D/ less than 0.001 regardless of their A/ values. Therefore, the regions with D/ of less than 0.001 were defined as low D/ regions. The regions with A/ less than 0.1 were taken to be low A/ regions in which the units with shunted blood were included. The regions with A/ exceeding 10 were taken as high A/ areas. These regions included dead space. The final result was expressed as a virtually continuous distribution of in the A/-D/ field. The amount of V·A in each lung unit was calculated from the and A/ values.

Pa_O2 and AaPO₂ Predicted from MIGET Analysis

To practically confirm the accuracy of the current method in estimating impaired gas exchange caused by the diffusion limitation contribution to overall arterial hypoxemia, we additionally calculated the predicted Pa_O2 using the well-established MIGET analysis (1, 23, 24), assuming that alveolar PO₂ in each gas exchange unit is in equilibrium with that in end-capillary blood. From the predicted Pa_O2, we estimated AaPO₂ (predicted AaPO₂) and compared this value with that obtained from the actually measured Pa_O2 (measured AaPO₂). Mean alveolar PO₂ (PA_O2) was defined as the ventilation-weighted value of alveolar PO₂ in respective gas exchange units, and used for calculating both measured and predicted AaPO₂. Predicted AaPO₂ was taken to be the value governed by the A/ mismatch alone, whereas measured AaPO₂ was assumed to be determined by diffusion-limited gas exchange as well as by A/ mismatch. Difference between measured and predicted AaPO₂ divided by measured AaPO₂ was used as the indicator exhibiting the relative contribution of diffusion limitation to overall gas exchange impairment.

Statistical Analysis

Whenever possible, we compared the data obtained from IPF and those from COPD using ANOVA followed by multiple comparison analysis with the Scheffe examination. When the data did not show an equal variance, however, we applied the Kruskal-Wallis test to assess statistical significance. Differences in measured AaPO₂ and predicted AaPO₂ from the MIGET analysis were estimated by the paired t test or Wilcoxon's test. A p value less than 0.05 was considered to be significant. Values are presented as means ± SD.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

HRCT Findings, Ventilatory Functions, and Blood Gas Determinations in IPF

TBLB specimens obtained from Group I of the patients with IPF showed active inflammation in the acini, which was pathologically consistent with desquamative interstitial pneumonia with marked cellular accumulation in distal air spaces with minimal interstitial thickening. The subjects exhibited augmented pulmonary Ga uptake and marked ground-glass opacities with minimal honeycombing cysts on HRCT. Group II patients had an extensive reticular pattern associated with honeycombing cysts but showed minimal ground-glass opacities on HRCT. TBLB specimens revealed findings compatible with interstitial fibrosis with honeycombing, in which alveolar collapse, thickened air-space walls, and minimal interstitial cellular proliferation were observed. Pulmonary Ga uptake was indistinct in these patients. HRCT obtained in the Group III patients demonstrated an extensive reticular pattern with significant ground-glass opacities. TBLB specimens obtained from lesions showing the reticular pattern with honeycombing and from an area of ground-glass opacities revealed, respectively, interstitial fibrosis with few inflammatory cells and cellular infiltration of the alveolar walls in addition to desquamation within the alveolar spaces. Pulmonary Ga uptake in Group III patients showed patchy augmentation.

Lung volumes, including TLC and FRC, were significantly reduced in the Group II and Group III patients compared with those in Group I; %VC values in Groups II and III were significantly smaller than those in Group I (Table 1). Other ventilatory parameters did not differ among the three IPF subgroups. FEV₁% values, peak flow (PFR), and airflow rates at 50% of FVC (₅₀) in subjects with IPF were well preserved (Table 1).

Arterial oxygenation was significantly reduced in Groups II and III, resulting in widened AaPO₂ in these subjects compared with Group I (Table 2). Pa_CO2, blood gas determinations in mixed venous blood, and steady-state diffusing capacity (DL_CO) did not differ among the three groups. Pulmonary hemodynamic data, including cardiac output, pulmonary arterial pressure, and vascular resistance, were statistically the same in the three groups (Table 2).

A/ and D/ Distributions in IPF

The patients with IPF in Group I, who had distinct ground-glass opacities with no obvious honeycombing cysts on HRCT, exhibited modest hypoxemia (Table 2). distributions along the D/ axis in these subjects were markedly inhomogeneous, i.e., bimodal, and D/ regions lower than 0.001 did exist, receiving, on the average, 7% of total cardiac output (Figure 1 and Table 3). On the other hand, distributions along the A/ axis in these subjects were fairly homogeneous, i.e., only 3% of cardiac output was distributed to remarkably low A/ regions, including the shunt compartment. High A/ regions received 39% of total ventilation, most of which was distributed to the dead space compartment (Figure 1 and Table 3).

View larger version (30K): [in this window] [in a new window]   Figure 1.   Representative distributions of in a A/-D/ field estimated in IPF patients. X axis: D/ (ml/ (ml · mm Hg)); Y axis: A/ (dimensionless); Z axis: fractional perfusion. ( A) Group I patient: bimodal D/ distribution with a modest abnormality in the A/ distribution. ( B) Another patient with Group I: A/ and D/ distributions qualitatively the same as those of (A). ( C ) Patient from Group II: bimodal abnormality in the A/ distribution with low A/ regions. The D/ distribution was unimodal, but with recognizable low D/ regions. ( D) Patient with Group III features: low and high A/ regions associated with low D/ regions.

Group II subjects, with an extensive reticular pattern associated with honeycombing cysts but minimal ground-glass opacities on HRCT, showed more severe hypoxemia than that seen in Group I (Table 2). The distributions along the D/ axis in Group II were unimodal, but broad and relative perfusion of low D/ regions averaged 10%, whereas their A/ distributions were markedly abnormal, and low and high A/ regions received, respectively, 15% of cardiac output and 51% of total ventilation, both values being significantly greater than those obtained in the Group I subjects (Figure 1 and Table 3).

Group III patients with IPF, demonstrating a mixture of the reticular pattern and ground-glass opacities on HRCT, showed broad but unimodal distributions along the D/ axis. The regions with D/ of less than 0.001 received 16% of cardiac output. The relative perfusion of low D/ regions was significantly larger in Group III than in Group II, though the extent of hypoxemia differed minimally between the two groups (Table 2). distributions along the A/ axis in Group III were qualitatively similar to those observed in Group II (Figure 1). Twelve percent of cardiac output was distributed to low A/ regions, whereas 56% of total ventilation went to high A/ regions (Table 3). These values were essentially the same as those obtained in Group II but significantly larger than those in Group I.

HRCT Findings, Ventilatory Functions, and Blood Gas Determinations in COPD

The %LDA values in patients in whom PAD dominated were le ss, whereas those in the patients with CPE were distinctly greater, more than 20% corresponding to the upper limit of %LDA obtained for nonsmoker volunteers (Table 1). Although %LDA values in the patients with CPE were significantly greater than in those with either PAD or PAD associated with CB (PAD+CB), there were no differences in %LDA values between CPE and CPE associated with CB (CPE+CB); %LDA values were comparable in the patients with PAD and those with PAD+CB (Table 1).

There were no differences in any ventilatory parameters between the CPE and CPE+CB groups. Nor did ventilatory indices differ between the PAD and the PAD+CB groups. TLC and FRC in the patients with CPE were clearly increased compared with those in patients with PAD, whereas RV/TLC and %VC values did not differ between these groups of patients. Although FEV₁, PFR, and ₅₀ deteriorated in all patients with COPD, they were significantly greater in the group with PAD than in the group with CPE.

With the exception of Pa_O2, AaPO₂, and DL_CO, there were no differences in blood gas determinations or pulmonary hemodynamics between the CPE and the PAD groups (Table 2). In addition, these parameters did not differ either between CPE and CPE+CB or between PAD and PAD+CB. Arterial hypoxemia was more severe, whereas reduced DL_CO was less evident, in patients with PAD than in those with CPE (Table 2).

A/ and D/ Distributions in COPD

There were no statistically significant differences in any values reflecting D/ and A/ distributions between patients with CPE and those with CPE +CB. Therefore, quantitative indices of D/ and A/ distributions were combined in these patients. The same held true for the D/ and A/ distributions in patients with PAD and for those in patients with PAD +CB (Figure 2).

View larger version (28K): [in this window] [in a new window]   Figure 2.   A/ and D/ distributions in COPD. ( A) Patient with CPE only; bimodal A/ distribution with high A/ regions. There were no low D/ regions. ( B) Patient with both CPE and CB components; inhomogeneity along the A/ axis, but no low D/ regions. ( C  ) Patient with PAD only; bimodal A/ distribution in combination with low A/. There were no regions with low D/ ratios. ( D) Patient with both PAD and CB features; bimodal abnormality in the A/ distribution with low A/ areas. No low D/ regions were found.

Patients with COPD and CPE (i.e., CPE or CPE+CB) exhibited little perfusion of either low D/ or low A/ areas, but they had significant ventilation to high A/ areas in proximity to the dead space, resulting in minimal abnormality along the D/ axis but a bimodal abnormality along the A/ axis. Relative perfusion of low D/ regions was less than 0.1%, whereas low A/ regions received 4% of cardiac output. The sum of relative ventilation of high A/ regions, including the dead space, amounted to 57% of total ventilation (Table 3).

Although there was no perfusion entering low D regions in patients with COPD and prominent PAD features (i.e., PAD and PAD+CB), low A/ regions in these patients received 18% of cardiac output, the value being significantly greater than that obtained in patients with primarily CPE (Table 3). Relative ventilation distributed to high A/ regions in patients with PAD was not augmented and was found to account for 37% of total ventilation, most of which was made up of dead space, such that D/ distributions were essentially normal, whereas A/ patterns were bimodal, representing low A/ regions in patients with PAD (Figure 2).

Comparison between Measured and Predicted AaPO₂ in MIGET Analysis

Our patients with IPF showed consistently lower predicted than measured AaPO₂, whereas no significant differences between these two variables were seen in patients with COPD (Figure 3). Measured and predicted AaPO₂ indicated that 34 and 66% of the AaPO₂ in the Group I patients with IPF were, respectively, due to diffusion-limited O₂ exchange and A/ mismatch. In Groups II and III patients with IPF, 15 and 21% of the AaPO₂, respectively, were caused by diffusion-limited O₂ exchange. The relative contribution of diffusion limitation to overall AaPO₂ was the greatest in Group I patients with IPF, but it was small in Groups II and III. The AaPO₂ in COPD was primarily attributable to A/ mismatch with a minimal contribution from diffusion-limited O ₂ exchange.

View larger version (32K): [in this window] [in a new window]   Figure 3.   Effect of diffusion-limited gas exchange on AaPO2 in IPF. Group I: n = 6; Group II: n = 9; Group III: n = 12. Values presented are means. Closed circles indicate individual data points. Asterisks indicate significantly smaller than the values obtained for Group I patients.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Critique of Methods

Using O₂, CO₂, CO, and six foreign inert gases, we attempted to predict a representative distribution of in the A/-D/ field in terms of the newly developed analytical method (11- 13). Minimization of the SSQ was made by applying the enforced smoothing technique (23), by which a stable solution can be obtained even when the number of unknowns exceeds that of equations. Thus, we found that the newly developed analytical procedure allows the essential features of D/ and A/ distributions with a unimodal or a bimodal pattern, along either the D/ or the A/ axis, to be defined (12). Separation of distribution along the A/ axis was quantitatively coincident with that of the MIGET analysis. distribution along the D/ axis was discernible as long as peaks were situated at a distance of a decade in the A/-D/ field. Although the detectable ranges as well as the limits of A/ and D/ distributions were theoretically examined in our previous report (12), the crucial issue is inclusion of respiratory gases into the estimation of the distributions, leading to the difficulty in predicting the accuracy of the method. Because O₂ and CO₂ were indispensable to determining the A/ and D/ distributions and were not withheld from the calculations, they could not be used for looking into the validity of the numerical approach. Therefore, we examined the applicability and limits of our method in practical terms, i.e., we compared distributions in the A/-D/ field obtained from patients with differences between the measured AaPO₂ and that predicted from the classic MIGET analysis (Figure 3). In patients with IPF, measured AaPO₂ values were significantly greater than those predicted from the MIGET, indicating the existence of diffusion-limited O₂ exchange in IPF. This finding is highly consistent with the evidence that most patients with IPF exhibit significant perfusion of low D/ regions (Figure 1 and Table 3). On the other hand, patients with COPD did not show significant differences between measured and predicted AaPO₂, suggesting that there may be little or no diffusion-limited O₂ exchange in COPD. Again, this finding supports the results of A/ and D/ distributions observed in COPD, i.e., no significant perfusion of low D/ regions (Figure 2 and Table 3). On the basis of these observations, we believe that our numerical approach does provide a representative description of the abnormal gas exchange caused by A/ and D/ inhomogeneities, though it certainly has the limitations.

To calculate the alveolar gas tension of an indicator gas in a given lung unit, we assumed the absence of stratified inhomogeneities caused by impaired diffusive gas mixing in acinar airways. Therefore, the current method would presumably be valuable for estimating diffusive resistance imposed by the alveolar-capillary membrane but not for detecting increases in gas phase diffusive resistance occurring within the airways in diseased lungs (26).

Limitation of O₂ Exchange by Diffusion in IPF and COPD

We found that diffusion-limited O₂ exchange causes arterial hypoxemia in patients with IPF but not in those with COPD (Figures 1-3 and Table 3). The observation of an O₂ diffusion limitation in patients with IPF at rest is at variance with earlier reports (2), which included patients with a wide variety of interstitial pneumonias, but it is in accordance with the recent studies done in carefully selected patients (6, 7). Eklund and colleagues (6) studied the importance of O₂ diffusion- limited exchange in patients with sarcoidosis, whereas Agusti and coworkers (7) used patients with IPF. These investigators (6, 7) confirmed that, as in the present study, O₂ diffusion-limited gas exchange is one of the factors causing arterial hypoxemia in patients with interstitial diseases, even in the resting state. On the other hand, we found that diffusion limitation contributes minimally to Pa_O2 reduction in patients with COPD, which is highly consistent with earlier reports (3, 5, 8).

Relation between Morphologic Alterations and A/ and D/ Inequalities in IPF

The patients with remarkable ground-glass opacities on HRCT (Group I) exhibited bimodal D/ distributions, with D/ areas lower than 0.001 but only modest abnormalities along the A/ axis (Figure 1 and Table 3). These findings suggest that ground-glass opacities, corresponding to active alveolitis/acinitis, produce primarily low D/ regions that significantly restrict gas exchange via a diffusion limitation. As reported by Gottlieb and Snider (30), active alveolitis may allow cellular and noncellular exudate to accumulate in both alveolar spaces and the interstitium. This may increase the diffusion path in aqueous media in which the diffusivity of gases falls by a factor of 10⁴ in comparison with that in gaseous media (31). Modest A/ abnormalities in the Group I patients may be due to incomplete collapse and distortion of air spaces, probably caused by exudate in alveolar spaces. Although the relative contribution of diffusion-limited O ₂ exchange to impairment of overall gas exchange was much higher in Group I than in Groups II and III (Figure 3), arterial hypoxemia and widening of AaPO₂ were less marked in Group I (Table 2), indicating that impaired O₂ exchange may be highly sensitive to A/ inhomogeneities rather than to a diffusion limitation.

Patients with IPF who were allocated to Group II, which was characterized by an extensive reticular pattern but minimal ground-glass opacities on HRCT, demonstrated moderate arterial hypoxemia associated with both low A/ and low D/ regions (Tables 2 and 3 and Figure 1), such that the hypoxemia in these patients may be attributable to inhomogeneous distribution of A/ as well as D/. These findings lend support to the idea that the reticular pattern on HRCT, corresponding to fibrotic alterations, causes distinct abnormalities along both the A/ and the D/ axis. Cassan and colleagues (32), in a morphometric study of patients with IPF and distinct fibrosis, found that the harmonic mean thickness of the alveolar-capillary membrane increased by 70%. They also found that mean alveolar surface area was reduced to 41% and capillary surface area to 39% of the values in normal control subjects, suggesting that diffusion properties were markedly distorted in IPF with fibrosis. An increase in the diffusion path caused by enlarged interstitial spaces and a decrease in the effective surface area result in low D/ regions. Honeycombing cysts, i.e., remodeling of the lung architecture, may create poorly ventilated air spaces, producing low A/ regions (30). Augmented high A/ regions in Group II (Figure 1 and Table 3) may be attributable to redistribution of pulmonary blood flow from severely affected portions in lower lung fields to those maintaining relatively normal structures in upper lung fields (30).

Patients in Group III, who had features of both Group I and Group II, demonstrated inhomogeneities along both the A/ and the D/ axis (Figure 1 and Table 3). Augmentation of low D/ regions in Group III may be attributable to combined impairments in diffusion properties, including the accumulation of exudate within air spaces (a possible cause in Group I) and increased thickening of alveolar septa as well as diminished surface areas for gas exchange (a possible cause in Group II). Arterial hypoxemia and relative perfusion of low A/ regions did not differ significantly between Groups II and III, though perfusion of low D/ regions in Group III significantly exceeded that in Group II (Figure 1 and Table 3). These findings again indicate that A/ inhomogeneities are more important than diffusion-limited gas exchange in determining a decline in Pa_O2when significant abnormalities in A/ and D/ distributions coexist.

Relation between Morphological Alterations and A/ and D/ inequalities in COPD

General characteristics of A/ and D/ distributions in patients with COPD can be summarized as follows (Figure 2): (1) irrespective of the COPD subtype, an essential abnormality is present along the A/ axis, whereas there is virtually no abnormality along the D/ axis; (2) patients with CPE have fairly broad distributions of along the A/ axis accompanied by regions with A/ values exceeding 10 but no regions with A/ values below 0.1; ( 3) patients with PAD have low A/ but no high A/ regions.

Although we did not have an opportunity to analyze A/ and D/ distributions in patients with only CB, the results obtained in patients with CB in combination with either PAD or CPE demonstrated that the persistent cough and hypersecretion associated with CB may not be essential for inducing an inhomogeneous A/ distribution (Figure 2 and Table 3). The subjects who met the criteria for PAD and CB were found to have low A/ areas only, whereas those who had features of CPE and CB solely showed high A/ regions. These findings indicate that the most important pathologic factor producing A/ distribution abnormalities in COPD is either emphysematous changes (high A/ areas) or peripheral airway involvement (low A/ areas), whereas CB per se produces minimal inhomogeneity in A/ distributions.

Although emphysematous regions would be expected to decrease the effective surface area for gas exchange (33), we found no evidence of the existence of low D/ areas in subjects with CPE. Thurlbeck and colleagues (33) analyzed alveolar surface areas in patients with emphysema of various degrees and showed that the average alveolar surface area in the patients with pathologically significant emphysematous lesions was 77% of the predicted value, indicating that the extent of alveolar surface area diminution in emphysema might be much smaller than that in interstitial pneumonia. Furthermore, emphysematous alterations may not expand the diffusion path. These observations may explain why the subjects with CPE in the current study did not show significant regions with low D/ values.

In conclusion, IPF primarily causes significant abnormalities in D/ distributions associated with the formation of low A/ regions. Active alveolitis and/or acinitis appear to be of major importance in producing low D/ regions, whereas fibrotic alterations appear to cause both low D/ and low A/ regions. The main abnormality in COPD is an inhomogeneous distribution of along the A/ axis but not along the D/ axis. Emphysematous lesions yield high A/ regions, whereas peripheral airway involvement produces low A/ regions.