Nonuniformity of diffusing capacity from small alveolar gas samples is increased in smokers.

BACKGROUND
Although centrilobular emphysema, and small airway, interstitial and alveoli inflammation can be detected pathologically in the lungs of smokers with relatively well preserved lung function, these changes are difficult to assess using available physiological tests. Because submaximal single breath washout (SBWSM) manoeuvres improve the detection of abnormalities in ventilation inhomogeneity in the lung periphery in smokers compared with traditional vital capacity manoeuvres, SBWSM manoeuvres were used in this study to measure temporal differences in diffusing capacity using a rapid response carbon monoxide analyzer.


OBJECTIVE
To determine whether abnormalities in the lung periphery can be detected in smokers with normal forced expired volumes in 1 s using the three-equation diffusing capacity (DLcoSB-3EQ) among small alveolar gas samples and whether the abnormalities correlate with increases in peripheral ventilation inhomogeneity.


PARTICIPANTS AND DESIGN
Cross-sectional study in 21 smokers and 21 nonsmokers all with normal forced exhaled flow rates.


METHODS
Both smokers and nonsmokers performed SBWSM manoeuvres consisting of slow inhalation of test gas from functional residual capacity to one-half inspiratory capacity with either 0 or 10 s of breath holding and slow exhalation to residual volume (RV). They also performed conventional vital capacity single breath (SBWVC) manoeuvres consisting of slow inhalation of test gas from RV to total lung capacity and, without breath holding, slow exhalation to RV. DLcoSB-3EQ was calculated from the total alveolar gas sample. DLcoSB-3EQ was also calculated from four equal sequential, simulated aliquots of the total alveolar gas sample. DLcoSB-3EQ values from the four alveolar samples were normalized by expressing each as a percentge of DLcoSB-3EQ from the entire alveolar gas sample. An index of variation (DI) among the small-sample DLcoSB-3EQ values was correlated with the normalized phase III helium slope (Sn) and the mixing efficiency (Emix).


RESULTS
For SBWSM, DI was increased in smokers at 0 s of breath holding compared with nonsmokers, and correlated with age, smoking pack-years and Sn. The decrease in DI with breath holding was greater in smokers and correlated with the change in Sn with breath holding. For SBWVC manoeuvres, there were no differences due to smoking in Sn or Emix, but DI was increased in smokers and correlated with age and smoking pack-years, but not with Sn.


CONCLUSIONS
For SBWSM manoeuvres the increase in DI in smokers correlated with breath hold time-dependent increases in Sn, suggesting that the changes in DI reflected the same structural alterations that caused increases in peripheral ventilation inhomogeneity. For SBWVC manoeuvres, the increase in DI in smokers was not associated with changes in ventilation inhomogeneity, suggesting that the effect of smoking on DI during this manoeuvre was due to smoke-related changes in alveolar capillary diffusion, rather than due solely to alterations in the distribution of ventilation.


DJ Cotton, JT Mink, BL Graham. Nonuniformity of diffusing capacity from small alveolar gas samples is increased in smokers. Can Respir J 1998;5(2):101-108.
BACKGROUND: Although centrilobular emphysema, and small airway, interstitial and alveoli inflammation can be detected pathologically in the lungs of smokers with relatively well preserved lung function, these changes are difficult to assess using available physiological tests. Because submaximal single breath washout (SBW SM ) manoeuvres improve the detection of abnormalities in ventilation inhomogeneity in the lung periphery in smokers compared with traditional vital capacity manoeuvres, SBW SM manoeuvres were used in this study to measure temporal differences in diffusing capacity using a rapid response carbon monoxide analyzer. OBJECTIVE: To determine whether abnormalities in the lung periphery can be detected in smokers with normal forced expired volumes in 1 s using the three-equation diffusing capacity (DLco SB -3EQ) among small alveolar gas samples and whether the abnormalities correlate with increases in peripheral ventilation inhomogeneity. PARTICIPANTS AND DESIGN: Cross-sectional study in 21 smokers and 21 nonsmokers all with normal forced exhaled flow rates. METHODS: Both smokers and nonsmokers performed SBW SM manoeuvres consisting of slow inhalation of test gas from functional residual capacity to one-half inspiratory capacity with either 0 or 10 s of breath holding and slow exhalation to residual volume (RV). They also performed conventional vital capacity single breath (SBW VC ) manoeuvres consisting of slow inhalation of test gas from RV to total lung capacity and, without breath holding, slow exhalation to RV. DLco SB -3EQ was calculated from the total alveolar gas sample. DLco SB -3EQ was also calculated from four equal sequential, simulated aliquots of the total alveolar gas sample. DLco SB -3EQ values from the four alveolar samples were normalized by expressing each as a percentge of DLco SB -3EQ from the entire alveolar gas sample. An index of variation (D I ) among the small-sample DLco SB -3EQ values was correlated with the normalized phase III helium slope (S n ) and the mixing efficiency (E mix ). RESULTS: For SBW SM , D I was increased in smokers at 0 s of breath holding compared with nonsmokers, and correlated with age, smoking pack-years and S n . The decrease in D I with breath holding was greater in smokers and correlated with the change in S n with breath holding. For SBW VC manoeuvres, there were no differences due to smoking in S n or E mix , but D I was increased in smokers and correlated with age and smoking pack-years, but not with S n . CONCLUSIONS: For SBW SM manoeuvres the increase in D I in smokers correlated with breath hold time-dependent increases in S n , suggesting that the changes in D I reflected the same structural alterations that caused increases in peripheral ventilation inhomogeneity. For SBW VC manoeuvres, the increase in D I in smokers was not associated with changes in ventilation inhomogeneity, suggesting that the effect of smoking on D I during this manoeuvre was due to smoke-related changes in alveolar capillary diffusion, rather than due solely to alterations in the distribution of ventilation.
S moking is associated with inflammation and fibrosis in small airways and alveoli (1)(2)(3), and vascular deficiency emphysema (4,5). Using conventional vital capacity single breath washout (SBW VC ) manoeuvres without breath holding, abnormalities in the phase III slope of nitrogen were found in healthy smokers which correlated with pathological changes in the small airways (3). However, prospective studies showed that the abnormalities in SBW VC maneouvres did not specifically identify those at risk of developing chronic airflow limitation (6). Others have proposed the use of more complex physiological techniques that highlight abnormalities in ventilation inhomogeneity in the lung periphery (7) using multibreath washout techniques (8) and gases of differing diffusivities (9), but these approaches are more demanding to implement as screening tests in large populations.
A refinement of SBW VC manoeuvres with potential application to epidemiological investigation was proposed because it might provide more information about peripheral ventilation inhomogeneity (10,11). This refinement (SBW SM ) consisted of reducing the inspired volume of the SBW manoeuvre to one-half inspiratory capacity (IC) and initiating the SBW manoeuvre at a preinspiratory lung volume of functional residual capacity (FRC). In smokers with normal forced exhaled flow rates, abnormalities in the normalized phase III slope for helium (S n ) were detected using SBW SM that were not evident using SBW VC manoeuvres (11).
The difference between the two types of manoeuvres was thought to be caused by differences in the relative contributions of several types of ventilation inhomogeneity. For SBW VC manoeuvres, a dominant mechanism of nonuniform gas distribution was convective dependent, topographical ventilation inhomogeneity due to asynchronous emptying of large regions with unequal specific ventilations (12)(13)(14). In contrast with SBW SM manoeuvre, the major origin of the ventilation inhomogeneity was in the lung periphery due to either intraregional convective dependent gradients among closely adjacent regions or the interaction of diffusion and convection at peripheral branch points within the acinus (12,15). A characteristic feature of inhomogeneity due to the interaction of convection and diffusion at peripheral branch points, and to a lesser extent to inhomogeneity due to intraregional convective dependent gradients, was that the tracer gas concentration gradients caused by this type of inhomogeneity were markedly diminished with short periods of breath holding. This is as a result of cardiogenic and diffusive gas mixing within the lung periphery (7,12). Hence, the steeper S n in smokers for SBW SM manoeuvres at 0 s of breath holding and the greater change in S n in smokers over short periods of breath holding indicated a superior ability of the SBW SM manoeuvre to identify an abnormality in peripheral ventilation inhomogeneity in smokers compared with the SBW VC manoeuvre (11).
In this study, we extended the analysis of SBW SM manoeuvre in smokers by evaluating the simultaneously measured disappearance curve of carbon monoxide. The three-equation diffusing capacity for carbon monoxide (DLCO SB -3EQ), measured from four equal and sequential aliquots of the total alveolar gas sample, was also affected by HISTORIQUE : Bien que l'emphysème centrolobulaire et l'inflammation des petites voies aériennes, l'inflammation de l'interstitium et des alvéoles, puissent être décelés à l'aide d'un examen pathologique dans les poumons des fumeurs présentant par ailleurs une fonction pulmonaire relativement bien conservée, ces changements sont difficiles à évaluer au moyen des tests physiologiques disponibles. Parce que le test de rinçage en respiration unique sous-maximale (SBW SM ) permet de mieux déceler les anomalies dans l'hétérogénéité de la ventilation dans la périphérie du poumon chez les fumeurs comparativement à l'épreuve classique se basant sur la capacité vitale, le test de SBW SM a été utilisé dans la présente étude pour mesurer les différences temporelles dans la capacité de diffusion à l'aide d'un analyseur rapide de monoxyde de carbone. OBJECTIF : Déterminer si les anomalies de la ventilation dans la périphérie du poumon peuvent être décelées chez les fumeurs dont le volume expiratoire maximum/seconde (VEMS) est normal au moyen de la méthode de la capacité de diffusion à trois équations (DL co SB -3EQ) dans des petits échantillons de gaz alvéolaire, et si les anomalies corrèlent avec les augmentations de l'hétérogénéité de la ventilation périphérique. PARTICIPANTS ET MODÈLE : Étude transversale auprès de 21 fumeurs et de 21 non-fumeurs présentant tous des débits expiratoires normaux. MÉTHODES : Les fumeurs et les non-fumeurs ont été soumis au test de SBW SM consistant en une inhalation lente du gaz test à partir de la capacité résiduelle fonctionnelle jusqu'à la première moitié de la capacité inspiratoire, et après 0 s ou 10 s en apnée, en expirant lentement jusqu'au volume résiduel (VR). Ils ont aussi été soumis au test classique de rinçage basé sur la capacité vitale (SBW VC ) consistant en une inhalation lente du gaz test à partir du VR jusqu'à la capacité pulmonaire totale et, sans apnée, en expirant lentement jusqu'au VR. La DL co SB -3EQ a été calculée à partir de l'échantillon complet de gaz alvéolaire. La DL co SB -3EQ a aussi été calculée à partir de quatre parties aliquotes égales, simulées et en séquence de la totalité de l'échantillon de gaz alvéolaire. Les valeurs de la DL co SB -3EQ provenant des quatre échantillons alvéolaires ont été standardisées en exprimant chacune d'entre elles comme un pourcentage de la DL co SB -3EQ obtenue à partir de la totalité de l'échantillon de gaz alvéolaire. Un index de variation (D I ) parmi les valeurs de la DL co SB -3EQ provenant des petits échantillons a été corrélé avec la pente de la phase III normalisée de l'hélium (S n ) et l'efficacité de la mixique (E mix ). RÉSULTATS : Dans le cas du SBW SM , le D I augmentait chez les fumeurs à 0 s d'apnée comparativement aux non-fumeurs, et corrélait avec l'âge, le nombre de paquets/année et la S n . La diminution de D I en apnée était plus importante chez les fumeurs et corrélait avec le changement dans la S n en apnée. Pour ce qui est des épreuves du SBW VC , il n'y avait aucune différence attribuable au fait de fumer dans la S n ou dans le E mix , cependant, le D I augmentait chez les fumeurs et corrélait avec l'âge et le nombre de paquets/année, mais pas avec la S n . CONCLUSIONS : En ce qui concerne les épreuves de SBW SM , l'augmentation de D I chez les fumeurs corrélait avec les augmentations de la S n dépendante du temps d'apnée ce qui laisse croire que les changements dans D I reflétaient les mêmes altérations structurales qui causaient une augmentation de l'hétérogénéité de la ventilation périphérique. Dans le cas des épreuves de SBW VC , l'augmentation de D I chez les fumeurs n'était pas associée avec les changements dans l'hétérogénéité de la ventilation, ce qui permet de croire que l'effet de fumer sur D I pendant cette épreuve était dû aux changements induits par la fumée sur la diffusion capillaire alvéolaire plutôt qu'aux seules altérations de la distribution de la ventilation. manoeuvres that altered the distribution of ventilation in the lung periphery in normal nonsmoking subjects (16). Compared with the predicted carbon monoxide washout for a uniform lung model, the carbon monoxide concentration was higher than expected early in exhalation and lower than expected later in exhalation in normal subjects for SBW SM manouvre without breath holding. This resulted in a reduced DLCO SB -3EQ when calculated from the first alveolar sample early in exhalation but an increased DLCO SB -3EQ when measured from the fourth sample at near residual volume (RV) in normal subjects. These differences in DLCO SB -3EQ among small alveolar gas samples were attributed to timedependent carbon monoxide concentration gradients in the lung periphery because the effect largely disappeared with short periods of breath holding (16).
We wished to determine whether the changes in DLCO SB -3EQ among small alveolar gas samples were abnormal in smokers with otherwise relatively normal lung function. We hypothesized that smoking would magnify the degree of nonuniformity (D I ) of DLCO SB -3EQ measured from small alveolar gas samples at 0 s of breath holding in smokers, particularly for SBW SM manoeuvres. As well we thought that D I would correlate with increases in peripheral ventilation inhomogeneity in smokers using the normalized phase III helium slope (11), but would disappear with short periods of breath holding (16).

PATIENTS AND METHODS
Study group: Twenty-one smokers and 21 nonsmokers were recruited from the community and hospital personnel. In both groups as previously reported (11), subjects denied recent respiratory symptoms. Spirometric testing revealed that the forced vital capacity (FVC), the forced expired volume in 1 s (FEV 1 ) and the maximal mid-expiratory flows (FEF  were within the 95% confidence limits of normal using previously reported regressions (17). Smokers had a current cigarette consumption of at least 10 cigarettes per day and a cumulative exposure of at least four pack-years. Nonsmokers had no current exposure and had smoked less than 100 cigarettes in their lifetime. Apparatus and equipment: Using equipment previously described (11,18), seated subjects, at rest, breathed test gas containing 0.3% carbon monoxide, 10% helium, 21% oxygen and the balance nitrogen through a low dead space twoway valve. Flow and volume were measured continuously with a pneumotach mounted in the wall of a bag-in-box system, helium concentration with a mass spectrometer and carbon monoxide concentration with a rapidly responding infrared analyzer. Throughout each SBW, including two to three preceding tidal breaths and a standardized deep breath, flow, volume, carbon monoxide and helium signals were stored digitally (50 Hz) for later computer analysis. Protocol: The protocol used in this study has been previously reported (11). All SBWs were preceded by a standardized deep breath of room air (19) and were performed singly in random order on one study day. An individualized volume versus time 'template' of the prescribed manoeuvre (18) was displayed on a monitor to assist subjects in performing the following SBWs: SBW SM manoeuvres consisting of slow inhalation from FRC to one-half IC, with either 0 s or 10 s of breath holding and slow exhalation to RV; and conventional SBW VC manoeuvres consisting of slow inhalation from RV to total lung capacity, and without breath holding, slow exhalation to RV. All inspired and expired flows for the preceding deep breath and for all SBW manoeuvres were maintained at 0.5 L/s. Analysis: For each SBW, the inspired and expired volumes, inspiratory and expiratory times, and the breath hold time were measured (11). RV was calculated by measuring the mass of helium inhaled and exhaled to determine the mass of helium remaining in the lung, as previously described (18,20,21). It was assumed that the mean helium concentration in the lung at RV was equal to the measured helium concentration at end expiration.
Because SBW manoeuvres were analyzed after storage of the flow, carbon monoxide and helium signals in digital form, DLco SB -3EQ from any one or a number of simulated alveolar samples from the same manoeuvre could be analyzed (18). In this analysis the three-equation diffusing capacity using the simulated entire alveolar gas sample was computed (13,18,20). DLCO SB -3EQ from the mean carbon monoxide concentration in each of four simulated sequential and equal alveolar gas samples (16) was calculated, and each DLCO SB -3EQ value was normalized by expressing it as a percentage of DLCO SB -3EQ calculated from the entire alveolar  gas sample (18). For all DLco SB -3EQ calculations, the mean carbon monoxide concentration in the exhaled alveolar gas of the deep breath of room air immediately preceding the single breath manoeuvre was used as the estimate of the carbon monoxide back pressure for the subsequent SBW manoeuvre (22). D I described the degree to which the four DLco SB -3EQ values measured from the four alveolar gas samples deviated from the single DLco SB -3EQ value measured from the entire alveolar gas sample by calculating the root-mean-square difference (18) of the normalized DLco SB -3EQ values ( Figure  1). D I quantified the degree of nonuniformity of DLco SB -3EQ measured from the four small samples, each normalized by DLco SB -3EQ measured from the total sample.
As previously reported (11) the phase III helium slope (DHe/L) between 33% and 67% of the exhaled volume was measured, and the phase III slope (S n ) was normalized by dividing DHe/L by the predicted ideal, mean end-inspired alveolar helium concentration ([He] A PRED ) (13,18). The mixing efficiency (E mix ) was calculated using a previously described computer algorithm (18). Statistics: Group mean data for smokers and nonsmokers were compared using unpaired Student's t tests. Differences due to the SBW manoeuvre and to breath hold time were assessed using paired Student's t tests in both smokers and nonsmokers. Correlations were performed using linear regression analysis.

RESULTS
The characteristics of the smoking and nonsmoking groups, which have been previously reported (11), were similar in terms of sex distribution (smokers: 12 women and nine men; nonsmokers: 10 women and 11 men), age (smokers 39±8 years; nonsmokers 38±7 years, mean ± SD) and height (smokers 172±10 cm; nonsmokers 174±10 cm). There were no significant differences in FVC (smokers 105±10% predicted; nonsmokers 101±8% predicted) or FEV 1 (smokers 100±9% predicted; nonsmokers 102±8% predicted). The FEF 25-75 was smaller (P<0.05) in smokers (90±24% predicted) than in nonsmokers (106±21% predicted). The smokers had a mean of 21±12 pack-years of cigarette smoking (range of four to 55 pack-years). There were no differences in the expired volume, maximum end-inspired volume, pre-inspiratory lung volume or breath hold time for comparable manoeuvres between smokers and nonsmokers ( Table 1). SBW SM at 0 s of breath holding: DLCO SB -3EQ from the total alveolar sample was slightly reduced in smokers (Table 1; P<0.05). DLCO SB -3EQ measured from small alveolar gas samples was reduced in the first alveolar sample, measured at high lung volumes just after alveolar dead space clearance, but increased progressively in subsequent samples (Figure 2, left graph). D I was therefore greater in smokers (42±29%) than in nonsmokers (18±7%) ( Table 1; P<0.001). D I correlated with age in smokers (Figure 3; r 2 =0.23; P=0.026) and pack-years of smoking ( Figure 3; r 2 =0.36; P=0.004). S n , but not E mix , was also markedly increased in smokers (Table 1; P<0.001). D I at 0 s of breath holding correlated significantly with S n at 0 s of breath holding in both smokers (r 2 =0.63; P<0.001) and nonsmokers (r 2 =0.38; P<0.01) (Figure 4). The day-to-day coefficient of variation of D I for this manoeuvre was 8.0±4.6% in 10 normal subjects whose value were repeated on five successive days (22). Effect of 10 s of breath holding on SBW SM : DLco SB -3EQ from the total alveolar gas sample was similar between smokers and nonsmokers (Table 1). For DLco SB -3EQ measured from small alveolar gas samples there was markedly less deviation of all four samples from DLco SB -3EQ measured from the total sample ( Figure 2). However, the pattern of change remained similar to that observed at 0 s of breath holding with DLco SB -3EQ being lower for sample 1 but higher for sample 4. Whereas D I was markedly reduced, compared with those measured at 0 s of breath holding, it remained slightly, but significantly higher in smokers (5±3%) than nonsmokers (3±1%) ( Table 1) and correlated with pack-years of smoking (r 2 =0.51; P<0.001) and with S n (r 2 =0.22; P<0.05).

Change in D I with breath holding for SBW SM :
The decrease in D I from 0 s to 10 s of breath holding correlated with S n measured at 0 s of breath holding in both nonsmokers (r 2 =0.37; P<0.01) and smokers (r 2 =0.61; P<0.01), but did not correlate with S n at 10 s of breath holding in either group. In smokers the decrease in D I with breath holding time (DD I ) was significantly greater in smokers than nonsmokers (Table 1) and correlated with the corresponding change in S n (DS n ) ( Figure 5, r 2 =0.51; P<0.001). In nonsmokers DD I and DS n were both less than in smokers, but the correlation between them was also significant (r 2 =0.31; P<0.01). SBW VC : DLCO SB -3EQ measured from the whole sample was slightly lower in smokers (Table 1). D I was significantly increased in smokers (Table 1; P<0.05) and correlated with age in smokers (r 2 =0.48; P<0.001) but not in nonsmokers, and with pack-years of smoking (r 2 =0.55; P<.001) ( Figure  6). However, there were no differences between smokers and nonsmokers for either S n or E mix (  smokers or nonsmokers. The day-to-day coefficient of variation of D I for this manoeuvre was 1.6±5.9% in 10 normal subjects who repeated the manoeuvre on five successive days (22).

DISCUSSION
The most remarkable finding in this study was that D I was significantly higher in smokers for SBW SM at 0 s of breath holding and correlated with cumulative pack-years of smoking. The fact that, for these manoeuvres, D I correlated with both age and pack-years of smoking in smokers, but was not affected by age in nonsmokers (Figure 3), implied a specific effect of smoking on D I that could not be explained by ageing. Although the precise pathological changes responsible for the increase in D I in smokers for SBW SM manoeuvres at 0 s of breath holding must await future structure-function correlations, the effects of breath holding on both D I and simultaneous measurements of ventilation inhomogeneity allow us to draw some inferences about the possible mechanisms for the effects of smoking on D I .
Smoking produces a number of pathological lesions in the lung (1)(2)(3)23,24). Recent observations using computed tomography (CT) revealed surprisingly common abnormalities in smokers with relatively normal lung function. The abnormalities consisted of ground glass attenuation, micronodules and diffuse emphysematous changes in smokers, but these changes were not found in nonsmokers (24). Pathologically, ground glass attenuation correlated with alveolitis and interstitial inflammation; parenchymal micronodules correlated with bronchiolectasis and peribronchial fibrosis, similar to those previously found in smokers (3); and emphysema exceeded the extent observed from CT analysis (1).
For SBW SM at 0 s of breath holding the increase in D I with smoking pack-years could be related to any one or a combination of these abnormalities. Macroscopic emphysema was unlikely to have caused the increase in D I in this group of smokers. Previous postmortem studies in smokers found no relationship between either emphysema grade, or the destructive index, and the single breath phase III slope (25) in the present study. Therefore, inflammation and fibrosis in small airways most likely accounted for the increase in D I in these healthy smokers in the present study. This was supported by the correlation of D I with S n in smokers ( Figure 4) for SBW SM manoeuvres at 0 s of breath holding. Moreover, breath holding dramatically decreased D I for SBW SM manoeuvres (Figure 2), and the decrease in D I with breath holding correlated with the change in S n with breath holding (Figure 4). This observation further supported the concept that the effects of smoking on D I occurred primarily in the pulmonary parenchyma distal to the membranous bronchioles (26).
Two possible mechanisms could explain this observation. First, the decrease in D I and the shallower S n with breath holding for SBW SM may both be explained by the common effect of small airway inflammation and/or regional loss of elastic recoil on intraregional convective dependent inhomogeneity, the interaction of convection and diffusion at peripheral branch points, or both phenomena (12). Second, the breath holding effects on D I and S n may not necessarily have been causally linked to alterations in ventilation inhomogeneity as previously suggested to explain the effects of breath holding on D I in normal subjects (16). Rather, both may have been altered by time, but for different reasons. S n may have been steeper in smokers because of increased peripheral ventilation inhomogeneity. D I may have been increased, at least in part, by an additional effect of an exaggerated serial gradient in alveolar-capillary diffusion in smokers, such that diffusion was preferentially reduced in the proximal versus distal region of the terminal respiratory unit. Without breath holding, such a gradient in diffusion within the acinus would have created a higher carbon monoxide concentration in proximal regions that emptied early in exhalation, accounting for the higher than expected carbon monoxide concentration early in exhalation (sample 1) and the lower than expected carbon monoxide concentration later in exhalation (16). However, with breath holding this serial gradient in carbon monoxide concentration within the acinus would have rapidly decreased over time because of cardiogenic and diffusive gas mixing within the air phase of the terminal respiratory unit. Early centrilobular emphysema (1,24) could have caused such a preferential decrease in diffusion in the proximal (alveolar ducts), compared with the distal (alveole), regions of the terminal respiratory units, thus exaggerating the effects of breath holding on D I seen in normal subjects (16).
Abnormalities in the phase III nitrogen slope in a group of healthy smokers with otherwise normal lung function have been previously reported for SBW VC , but only when the smokers forcibly exhaled to RV immediately before performing the SBW VC manoeuvre (27). The proposed mechanism for this effect of smoking was that the rapid forced exhalation just preceding the SBW VC manoeuvre delayed regional emptying in peripheral regions distal to small airway inflammation and narrowing in the lungs of smokers. This reveals the presence of this disease process by transiently altering the spatial distribution ventilation by increasing specific ventilation in regions distal to the airway narrowing. In the present study, SBW VC manoeuvres were preceded by a deep breath, consisting of slow exhalation to RV, so that we would not have expected changes in ventilation inhomogeneity in otherwise healthy smokers due to the former mechanism (27); indeed, S n and E mix for SBW VC manoeuvres in smokers were normal. However, D I was unexpectedly increased and did not correlate with changes in S n in smokers suggesting that an additional factor, other than alterations in ventilation inhomogeneity, was involved. The increase in D I due to smoking for these SBW VC manoeuvres may have been due to reductions in gas diffusion in macroscopic regions of emphysema that remained ventilated, but that emptied preferentially early in expiration. This appeared unlikely because macroscopic emphysema was thought to be ventilated poorly and largely by collateral channels (28). Alternatively, and more likely, the effect of smoking on D I for these SBW VC manoeuvres may have been due to the same mechanism proposed for SBW SM manoeuvres as described above. That is, centrilobular emphysema could have caused proximal destruction of terminal respiratory units, resulting in an exaggeration of the serial gradients in diffusion from the proximal (reduced diffusion) to the distal (increased diffusion) segments of terminal respiratory units.
Previous reports found that the conventional single breath diffusing capacity for carbon monoxide (DLCO SB ) (29) was reduced in smokers (5,30) and correlated with macroscopic disease, as detected by CT scan techniques (31). However, in seated patients at rest DLCO SB was insensitive to the interregional nonuniformity in the distribution of emphysema that often preferentially affects the apex of the lung in smokers (5). Conventional DLCO SB testing, as distinct from the DLCO SB -3EQ (20), was also less useful as a screening tool because the method was spuriously affected by the variations in the way the single breath test was performed (32). Interpretation of DLCO SB in smokers has also been potentially confounded by uncertainties about precise corrections of DLCO SB for carbon monoxide back pressure, carboxyhemoglobin and the binding affinity of hemoglobin (29). In the present study we corrected all measurements of DLCO SB -3EQ for the carbon monoxide back pressure by measuring the carbon monoxide in the exhaled gas during the exhalation phase of the deep breath of room air preceding each DLCO SB -3EQ manoeuvre (18). The infrared carbon monoxide analyzer was sufficiently sensitive to allow measurement of background carbon monoxide in nonsmokers at levels of 0 to 70 ppm (18). The carbon monoxide analyzer was linear in this range and was checked daily (18).
Although we did not measure carboxy-hemoglobin (29) in the smokers in this study, most of whom reported smoking in the 2 h before testing, it would likely have been elevated, accounting for most of the apparent reduction in DLCO SB -3EQ measured from the total sample in smokers compared with nonsmokers (22). DLCO SB -3EQ measured from the total sample in smokers would have been affected by both carbon monoxide back pressure and carboxy-hemoglobin. However, we found that D I required no corrections because it was not affected by increases in carboxy-hemoglobin in normal subjects (22).
The present method of measuring DLCO SB -3EQ from four sequential alveolar gas samples has a number of potential limitations. An analyzer with appropriate instrumentation and software is not currently commercially available. We employed a carbon monoxide analyzer that was modified to decrease its response time to 110 ms (18). The use of mass spectrometry for analysis of helium, while providing excellent signal characteristics, is too expensive for routine implementation. The manoeuvres also required instantaneous biofeedback to ensure reproducibility. This is feasible using available computer technology, and the software has been developed. In our experience, naive subjects are able to complete these manoeuvres with minimal prior coaching, but the reproducibility is less than in trained subjects (33). Finally, although this report demonstrates clear changes in smokers, the effect of smoking on D I is not necesarily specific. Other pathological processes that lead to small airway obstruction (34) such as bronchiolitis obliterans following lung transplant, could have similar effects.

CONCLUSIONS
We have shown that D I was increased in smokers in whom FEV 1 was normal. The effect, which was much greater for SBW SM manoeuvres, largely disappeared with breath holding and correlated with a greater change in S n with breath holding, indicating that the increase in D I was caused by events occurring in the lung periphery. However, the precise structural correlates require definition. Because D I was also increased in smokers for SBW VC manoeuvres, in the absence of smoking-induced alterations in ventilation inhomogeneity, the effects of smoking on D I may have been due to intraregional changes in diffusion across the air-blood barrier. D I may be a more sensitive indicator of changes in the lung periphery due to smoking.