Lung Structure and Gas Exchange in Mild Chronic Obstructive Pulmonary Disease 1- 4

JOAN A. BARBERA,5 JOSEP RAMIREZ, JOSEP ROCA, PETER D. WAGNER, JOAN SANCHEZ-LLORET, and ROBERT RODRIGUEZ-ROISIN

Introduction

Patients with chronic obstructive pulmonary disease (COPD) may show a wide spectrum of gas exchange abnormalities. Depending upon the severity of the disease, these include an increase in AaP0 2 alone or moderate to severe hypoxemia with or without hypercapnia. Studies using the multiple inert gas elimination technique (1, 2) in patients under stable clinical conditions (3-5) and in acute respiratory failure (6, 7) have shown that ventilation-perfusion (VA/Q) mismatching accounts for the pulmonary gas exchange inefficiency observed in COPD. In contrast, both intrapulmonary shunting and diffusion limitation appear not to playa significant role (3). The relationships between the mechanisms of gas exchange impairment and the pathologic changes in the lung in COPD have not been as yet fully established. In contrast, correlations between airflow obstruction and lung pathology have been extensively studied (8). Yet, pulmonary gas exchange appears not to be always predictable from maximal expiratory flow rates (9).Although the relationships between arterial P0 2 and Pe0 2 and lung structural abnormalities have also been investigated (to-15), the information provided by such measurements is limited because values of respiratory gases are partly dependent on factors additional to VA/Q inequality. These factors, especially cardiac output, minute ventilation, and oxygen consumption (16), make it difficult to establish clearcut conclusions about the relationships between VA/Q mismatching and pathologic changes. The aim of our study was to investigate the structural basis of VA/Q mismatching in mild COPo. To this end we examined the relationships between preoperative measurements of VA/Q inequality and the morphologic assessment of emphysema and small airways abnormalities in a group of patients with mild

SUMMARY Toinvestigate the influence of pulmonary emphysema and small airwaysabnormalities on ventilation-perfusion (VA/a) mismatching in mild chronic obstructive pulmonary disease (COPO), we studied 23 patients (mean predicted FEV" 76 ± 15%) before lung resection because of a localized neoplasm. Respiratory gas exchange and VA/a distributions were measured while the patients breathed room air and 100% 0,. Breathing room air, the AaPO, was moderately increased (25 ± 12 mm Hg) as was VA/a mismatching, indicated by the dispersion (log SO) of both blood flow (0) and ventilation (V) distributions (log SO Q, 0.78 ± 0.3; and log SO V, 0.66 ± 0.28, respectively) (normal range, 0.3-0.6). AaPO" log SO 0, and log SO V all significantly correlated with the emphy0.57, r 0.62, sema severity assessed morphologically from the resected lung specimens (r and r = 0.45, respectively). Log SO V also significantly correlated with the severity of the inflammatory infiltrate of membranous bronchioles (r = 0.62). During 100% 0, breathing there was an increase In VA/a mismatching (log SO 0 rose to 1.12 ± 0.08, P < 0.001), suggesting release of hypoXic pulmonary vasoconstriction. This Increase in VA/a inequality was not significantly related to the severity of lung pathologic findings. We conclude that, in mild COPO,both pulmonary emphysema and small airways abnormalities contribute to VA/a mismatch, the severity of emphysema being the major morphologic correlate of the Increase in AaPo,. AM REV RESPIR DIS 1990; 141:895-901

=

COPD undergoing resectivelung surgery because of a localized lung neoplasm. Methods Population Patients undergoing surgical resection of a lobe or lung because of a localized lung neoplasm were studied before surgery. Consent was obtained from all the patients according to the requirements of the Clinical Research Committee of the Hospital Clinic, Universitat de Barcelona. All subjects were free of other cardiovascular or systemic disorders and showed localized lesions on chest radiographs. We excluded patients with (1) radiographic evidence of obstructive pneumonitis or atelectasis, (2) lobar or segmental bronchial occlusion noticed by fiberoptic bronchoscopy, or (3) lobar or segmental ventilation or perfusion defects on lung scintigram. Twenty-one men and two women 61 ± 2 (SEM) yr of age were studied. Twenty of the 23 patients were heavy tobacco smokers (50.9 ± 4.5 packyears). As a whole, they had mild airflow obstruction (mean FEV" 76 ± 3070 ofpredicted).

General Protocol The study included a preoperative evaluation of (I) routine pulmonary function tests (PFT) and the single-breath nitrogen washout test, and (2) respiratory gas exchange and VA/Q distributions measured at rest in a semirecumbent position (45 degrees) under steady-state

=

conditions (± 5% variation in heart rate, respiratory frequency, tidal volume, and endtidal 0, and CO,) while the patients breathed room air and also after 30 min of 100% 0, in random order. Gas exchange studies were performed 2.7 ± 0.7 days before thoracotomy. After the surgical procedure, resected specimens were processed for morphologic analysis and routine pathologic examination. Lung speci-

(Received in original form March 22, 1989 and in revised form September 7, 1989) I From the Department of Medicine (Servei de Pneumologia), Pathology and Surgery (Servei de Cirurgia Toracica), Hospital Clinic, Facultat de Medicina, Universitat de Barcelona, Barcelona, Spain, and the Department of Medicine, Section of Physiology, University of California, San Diego, La Jolla, California. , Supported by Grant No. CCA 8309185 from the Joint u.S.-Spain Committee, Grant No. HL17731-11 from the National Heart, Lung, and Blood Institute, Grant No. RG 0766/87 from the NATO Scientific Affairs Division, and by SEPAR/1987. 3 Presented in part at the Annual Meeting of the American Thoracic Society, Las Vegas, May 1988. 4 Correspondence and requests for reprints should be addressed to Robert Rodriguez-Roisin, M.D., Servei de Pneumologia, Hospital Clinic, Villarroel 170, 08036-Barcelona, Spain. 5 Recipient of a Postdoctoral Research Fellowship Award (BII-87/57) from the Fondo de Investigaciones Sanitarias de la Seguridad Social (FISss), Instituto Nacional de la Salud, Spain.

895

896

BARBERA, RAMIREZ, ROCA, WAGNER, SANCHEZ-LLORET, AND RODRIGUEZ-ROISIN

mens with prominent areas of perineoplastic pneumonic consolidation on macroscopic examination were not included in the study (from the original 27 resected specimens, four were rejected because of this criterion).

Pulmonary Function Tests Routine PFT included: (1) forced spirometry and flow-volume curves with bronchodilator response (HP 47804A; Hewlett-Packard, Waltham, MA); (2) single-breath diffusing capacity for carbon monoxide (DLco) (Model A; P. K. Morgan, Chatham, UK); (3) inspiratory capacity (HP 47804A); and (4) plethysmographic thoracic gas volume and airway resistance (Body Pneumo-test; E. Jaeger, Wiirzburg, FRG). The single-breath nitrogen washout test was carried out using the HP 47804A system. Predicted equations for forced spirometry, lung volumes, and DLCO were those of our own laboratory (17, 18), and for the single-breath nitrogen washout test, those of Buist and Ross (19, 20).

Respiratory and Inert Gas Exchange Measurements mood samples were anaerobically collected through a polyethylene catheter (Seldicath; Plastimed, Saint-Leu-La-Foret, France) inserted into the radial artery. Arterial Po, and Pco, and pH were analyzed using polarographic electrodes (IL 1302; Instrumentation Laboratories, Milan, Italy). Hemoglobin concentration was measured using an OSM-2 hemoximeter (Radiometer, Copenhagen, Denmark). A lowdead-space, low-resistance, nonrebreathing valve (No. 1500; Hans Rudolph, Kansas City, MO) connected to a heated metal mixing chamber was used to collect the expired gas. Mixed expired 0, (FEa,) and CO, (FEco,) were measured with a mass spectrometer (Multigas Monitor MS2; BOC-Medishield, London, UK). Minute ventilation and respiratory frequency were measured using a calibrated Wright's spirometer (Respirometer MK8; BOC-Medical, Essex, UK). The AaPo, was calculated from the alveolar gas equation as: AaPo, = [Flo, (PB - 47) - (PAco,/R) Flo, (l - R) (PAco,/R)] - Pao,

+

where Flo, is the 0, inspiratory fraction, PB is barometric pressure, PAcO,is alveolar Pco, assumed to be equal to Paco" and R is the measured respiratory exchange ratio. A threelead EKG and systemic arterial pressures were recorded throughout the study (HP 7830A Monitor and HP 7754B Recorder; HewlettPackard). Cardiac output (QT) was measured using a 5-mg bolus of indocyanine green dye (SERB Laboratoires Pharmaceutiques, Paris, France) injected into a central venous line; dye concentrations sampled from the arterialline were used to compute QT (CO-IO/R; Waters Instruments Inc., Rochester, MN). VAlQdistributions wereestimated by the multiple inert gas elimination technique (1, 2). Particular features of the setup of this technique in our laboratory have been reported

elsewhere (21, 22). Concentration ofthe inert gases in the mixed expired samples and the gas phase of Ns-equilibrated arterial samples were measured by gas chromatography (HP 5880A; Hewlett-Packard, Avondale, PA). Mixed venous concentrations were computed from arterial and mixed expired concentrations and cardiac output using the Fick principle. Ventilation-perfusion distributions were estimated from measured retentions and excretions of the six inert gases using a least squares fit to the data by a multicompartmental model with enforced smoothing (2). The position of the perfusion and ventilation distributions was described by the VA/Q ratio at their mean (first moment) (Q and V, respectively), and their dispersion about the mean by the standard deviation (second moment) on a log scale (log SD Q and log SD V, respectively). The variable DISP R-E*, originally described by Gale and coworkers (23), was used as an index ofthe overall VA/Q heterogeneity. This is the root mean square difference between retention and excretion of the inert gases after correcting for series dead space.

Lung Structure Analysis Removed lung specimens were intrabronchially fixed with 10010 formalin at a constant transpulmonary pressure of 25 em H,O for 48 to 72 h (24). Fixed lung specimens were sliced into sagittal sections 1 em thick, and the midsagittal slice was impregnated with barium sulphate. From the adjacent lateral slices, five (lobes) or eight (lungs) blocks 2 x 2 em were taken by random sampling away from the neoplastic tissue. After paraffin embedding, histologic sections 5 11m thick were stained with hematoxylin-eosin, Masson trichrome, and PAS. The dimensions of each block were measured before and after the histologic processing in order to assess surface and linear shrinkages. The severity of emphysema was scored (emphysema score, ES) on the barium-impregnated midsagittal slice independently by two of the authors (JAB and JR) using the Picture Grading System (25). Results were expressed as the mean value of both observations, and the reproducibility between them was checked at the end of the study. The usefulness of this method in evaluating emphysema on lung lobe slices has been recently demonstrated by Wright and coworkers (26). To microscopically assess emphysema, the average interalveolar wall distance was measured by the mean linear intercept (Lm) method (27). Intercepts were counted on 10 fields (x 40) on each slide using a crossed hairlines (2.43 mm long) ocular graticule. The Lm measurements were corrected for processing shrinkage in all cases. To evaluate small airways abnormalities, all noncartilaginous membranous bronchioles less than 2 mm in internal diameter were examined. Six variables (inflammation, fibrosis, amount of muscle, pigment deposition, goblet cell metaplasia, and squamous cell

metaplasia) weregraded using a reference panel of pictures (28) according to the method of Cosio and coworkers (29). Each pathologic variable was expressed as a percentage of the maximal possible score, and a total pathologic score (TPS) was determined by the sum of the six individual percentages. To check the reproducibility of the bronchiolar measurements, 17slides randomly selected were additionally measured by a second observer (interobserver reproducibility), and the main observer reevaluated 22 slides at the end of the study to assess the intraobserver reproducibility.

Statistical Analysis All data are expressed as mean ± SEM. Measurements at room air and during 100% 0, breathing were compared using Student's paired t tests. The relationships between morphologic and functional variables were assessed using nonparametric correlation tests (Spearman's rank correlation coefficient). These relationships were also examined using a multiple linear regression analysis. Accordingly, gas exchange variables (AaPo" log SD Q, log SD V, and DlSP R-E*) were considered as the dependent variables of the multiple regression, and all the structural measurements were included (stepwise method) as covariates. Both intraobserver and interobserver reproducibility were assessed by Pearson's linear correlation tests. Probability values lower than 0.05 were considered significant in all cases.

Results

The patients and their pulmonary function results are described in table 1. Overall, they showed mild airflow obstruction (FEV h 76 ± 3010 of predicted; FEVtl FVC ratio, 60.6 ± 2.1%). TLC was within normal limits, but RVlTLC was moderately increased. All but two patients showed normal DLco and DLco/VA values (DLco/VA,94.0 ± 4.7% of predicted; range, 44 to 140070). The single-breath N 2 washout test disclosed increased Phase III slope (N, slope, 2.6 ± O.4OJo/L; 187 ± 29% of predicted) and a high closing volume/vital capacity ratio (CVlVC%, 32.1 ± 1.2%; 147 ± 6% of predicted). 'Iaken all together these results characterize a population with mild COPD.

Gas Exchange Measurements while Breathing Room Air Eight patients showed moderate hypoxemia (Pao, < 80 mm Hg), and Paco, was normal in all of them (table 2). Although mean AaPo 2 was moderately increased (25.3 ± 2.4 mm Hg), four patients had AaPo, values within the normal range « 20 mm Hg). Blood flow distributions were unimodal in 15patients, and bimodal (with low VA/Qareas) in eight. In contrast, the ven-

897

WNG STRUCTURE AND GAS EXCHANGE IN MILD COPO TABLE 1 GENERAL CHARACTERISTICS AND PULMONARY FUNCTION DATA OF PATIENTS STUDIED FEV, Age Patient Sex (yr) No.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Mean SEM

M M M M M M M F M

M M M M M M M M M M F

M M M

Pack-years

(L)

55 66 63 62 71 60 69 65 62 53 44 68 65 50 65 53 49 74 54 64 66 52 68

38.0 46.0 87.5 52.5 43.0 67.5 94.0 0.0 55.5 70.0 0.0 52.5 22.0 66.0 42.0 32.0 33.0 37.5 40.0 0.0 52.0 18.5 67.5

2.64 2.75 2.30 2.05 2.48 2.54 2.12 1.92 2.61 3.03 4.28 1.31 1.52 3.27 2.40 2.88 2.28 2.02 3.06 2.02 1.60 2.29 1.78

60.8 1.7

50.8 4.2

2.40 0.14

(% pred)

74 95 74 71 72 84 64 92 64 77 95 44 55 104 86 76 69 77 84 99 57 76 61 76.1 3.1

RVITLC FEV,IFVC FEF2.-7• TLC N 2 Slope DLCO (%) (% pred) (% pred) (% pred) (%) (%/L)

74 64 56 57 62 57 56 59 67 50 72 37 45 69 66 62 62 68 65 71 51 76 47 60.6 2.1

tilation distributions were never bimodal and ~ere devoid of regions of very high VA/Q ratio. The VA/Q ratio at the mean of the blood flow distribution (first

58 44 48 60 43 72 27 75 27

75 106 100 106 90 108 82 117 79 106 115 107 103 119 110 116 116 84 91 93 103 84 102

37 40 43 50 44 37 42 40 45 29 37 52 46 38 47 49 50 49 31 38 55 42 46

45.7 4.0

100.5 2.8

42.9 1.4

74 52 41 33 38 43 27 46 38 28 76 13 16 7~

99 97 88 84 97 53 83 130 92 85 114 56 119 109 143 85 85 93 105 134 93 101 130 98.9 4.7

2.31 1.94 6.88 1.83 2.33 2.21 3.53 0.24 2.04 2.45 1.14 4.86 4.59 1.67 2.41 0.30

1.55 0.33 6.40 1.91 3.47 2.58 0.39

moment) was slightlysubnormal (Q, 0.69 ± 0.03) and the dispersion (second moment) was moderately increased (log SD Q,0.78 ± 0.06) (normal 95% confidence

limits, 0.3 jo p.6) (30). Mean shunt (regions of VA!.Q < 0.005) was very small (1.2 ± 0.2070 9T),. as was perfusion of units with low VAlQ ratios (between0.005 and 0:1) (2.1 ± 0.7% QT; range, zero to 12% QT). The dispersion of the ventilation distribution was slightly increased (log SD V, 0.66 ± 0.06). The amount of ventilation to units with high VA/Q ratios (between 10and 1(0) was slight (q.72 ± 0.31% VE; range, zero to 5%. VEl. Mean dead space (ventilation to VA/Q ratios> 1(0) was not increased at 32.0 ± 2.0% of tidal volume. The overall index of VA/Q heterogeneity (DISP R-E*) was also increased (6.7 ± 0.8) (normal values ~ 3). No significant difference was shown between measured Pao, (83.5 ± 2.1 mm Hg) and Paoz (87.1. ± 2.1 mm Hg) predicted from the VA/Q distributions (p =.0.52), confirming that the observed VA/Q mismatching accounted for the moderate increase in AaPoz.

Effect of 100% O2 Breathing While the patients breathed 100% oxygen, VE significantly increased and QT decreased. Arterial POz rose from 84 ± 2 to 481 ± 11 mm Hg, and Paco z and pH remained unchanged (table 3).

TABLE 2 VENTILATION, CARDIAC OUTPUT, AND GAS EXCHANGE MEASUREMENTS

Patient No.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Mean SEM

Pa02

Pac02

AaP02

VE

v02

QT

Shunt

{mmHg}

(mmHg)

(mmHg)

(Llmin)

(ml/mln)

(Llmin)

(%QT)

Q

LOQ SDa

37.2 34.1 40.5 37.9 39.2 35.4 36.5 39.1 40.9 34.1 40.2 34.7 42.8 34.7 39.2 34.9 38.8 33.5 39.3 40.0 37.1 36.6 38.8

20.2 27.2 12.6 39.7 25.2 49.0 20.7 33.6 26.9 20.5 1.3 41.0 23.0 4.8 8.6 30.5 25.2 22.7 29.6 23.3 40.5 22.9 33.9

7.03 8.14 9.90 8.70 9.75 12.80 8.00 6.50 11.30 12.60 6.50 11.70 5.20 8.60 6.20 10.70 8.50 8.80 9.65 6.50 6.80 7.00 7.80

241 265 280 180 199 289 223 210 275 352 274 270 214 249 291 323 293 254 261 158 154 220 266

7.29 6.01 5.88 5.78 8.45 9.00 4.91 5.64 8.05 6.86 6.90 5.84 7.30 7.70 7.30 5.80 5.50 5.00 5.28 4.57 5.20 6.90 7.35

0.4 0.9 0.5 0.0 2.0 1.2 0.3 2.7 0.1 0.7 0.4 3.4 1.6 3.1 1.9 0.3 0.0 1.4 0.6 3.4 0.8 0.7 1.1

0.54 0.89 0.82 0.46 0.54 0.46 0.73 0.53 0.64 0.82 0.67 0.92 0.54 0.62 0.64 0.85 0.78 0.98 0.90 0.79 0.60 0.74 0.52

0.75 0.93 0.67 1.22 0.59 1.44 0.69 1.26 1.05 0.66 0.34 0.91 0.63 0.37 0.34 0.99 0.82 0.87 0.85 0.34 0.87 0.43 0.90

37.6 0.5

25.3 2.4

8.64 0.44

250 10

6.46 0.25

1.2 0.2

0.69 0.03

0.78 0.06

88 87 85 74 87 64 87 75 78 95 101 71 84 101 93 74 82 91 84 85 75 91 68 83.5 2.1

Dead Space

V

Log SDV

DISP R-E'

29.7 15.3 36.3 43.8 41.6 29.0 41.4 28.5 33.7 40.4 24.1 28.4 6.9 42.4 20.1 41.9 33.4 31.8 38.4 42.2 32.2 18.5 35.5

0.77 1.36 1.39 1.50 1.03 1.83 1.25 1.13 1.17 1.44 0.76 2.25 0.84 0.70 0.72 1.20 1.26 1.38 1.30 0.88 1.31 0.90 0.84

0.51 0.54 0.83 1.27 1.25 1.01 0.77 0.65 0.62 0.84 0.35 0.89 0.70 0.36 0.35 0.42 0.61 0.48 0.50 0.33 0.86 0.43 0.59

4.47 5.46 6.64 12.66 7.91 16.37 7.27 9.71 7.03 7.76 1.78 12.58 6.53 3.39 2.44 3.93 6.33 4.68 4.80 3.10 10.19 3.04 6.22

32.0 2.0

1.18 0.08

0.66 0.06

6.71 0.76

{%

VEl

= oxygenconsumplion;aT = cardiacoutput; Shunt = percentageof bloodflow to unventilatedunits (VAla < 0.005); = log standard deviation of blood flow distribution; Dead Space = percentage 01 ventilation to unperfusedareas (VAla > 100); Ii = mean vAla of ventilation distribution; Log SO V = log standard deviation of ventilationdistribution; olSP R·E· = dispersionof measured retentions and excretionsof inert gases. Definition of abbreviations: vo,

Q = mean VAla of blood flow distribution; Log SO Q

898

BARBERA, RAMIREZ, ROCA, WAGNER, SANCHEZ-LLORET, AND RODRIGUEZ-ROISIN

TABLE 3 COMPARISON OF ROOM AIR AND 100% OXYGEN BREATHING

8.6 6.5 83.5 37.2 7.40 1.2 0.69 0.78 1.18 0.66 6.7

Llmin QT, Llmin Pao 2 , mm Hg Paco 2 , mm Hg pH Shunt, % QT

a

Log SD Q

V Log SD V DISP R-E-

P Value

100% O2

Room Air

\IE,

9.4 6.0 481 37.2 7.41 2.5 0.67 1.12 1.45 0.67 10.2

± 0.4 ± 0.3 ± 2.1 ± 0.5 ± 0.0 ± 0.2 ± 0.03 ± 0.06 ± 0.08 ± 0.06 ± 0.8

± 0.5 ± 0.3 ± 11 ± 0.6 ±0.Q1 ± 0.5 ± 0.05 ± 0.08 ± 0.09 ± 0.05 ± 1.0

.. 0.05 .. 0.001 .. 0.001 NS NS .. 0.01 NS .. 0.001 .. 0.001 NS .. 0.001

Fordefinition of abbreviations, see table 2.

VA/Q mismatching increased (worsened) during 100070 O, breathing, Five patients with a unimodal blood flow distribution while breathing room air displayed a bimodal pattern (containing a low VA/Q mode), and the overall index OfVA/Q mismatching (DISP R-E*) rose from 6.7 ± 0.8 to 10.2 ± 1.0 (p < 0.001). Although mean VA/Q ratio of the blood flow distribution did not change, its dispersion increased significantly (log SD Q,0.78 ± 0.06 while breathing room air and 1.12 ± 0.08 while breathing 100% Os, p < 0.001), with a greater percentage of perfusion to units with low VA/Q ratios (between 0.005 and 0.1) (2.1 ± 0.7% QT while breathing room air and 8.3 ± 1.6% QT while breathing 100% Ol' p