Pulmonary Vascular Volume by Quantitative CT in Dyspneic Smokers with Minor Emphysema

Abstract Reduced lung diffusing capacity for carbon monoxide (DLCO) at rest and increased ventilation (⩒E)-carbon dioxide output (⩒CO2) during exercise are frequent findings in dyspneic smokers with largely preserved FEV1. It remains unclear whether low DLCO and high ⩒E-⩒CO2 are mere reflections of alveolar destruction (i.e. emphysema) or impaired pulmonary perfusion in non-emphysematous tissue contributes to these functional abnormalities. Sixty-four smokers (41 males, FEV1= 84 ± 13%predicted) underwent pulmonary function tests, an incremental exercise test, and quantitative chest computed tomography. Total pulmonary vascular volume (TPVV) was calculated for the entire segmented vascular tree (VIDA Vision™). Using the median % low attenuation area (-950 HU), participants were dichotomized into “Trace” or “Mild” emphysema (E), each group classified into preserved versus reduced DLCO. Within each emphysema subgroup, participants with abnormally low DLCO showed lower TPVV, higher ⩒E-⩒CO2, and exertional dyspnea than those with preserved DLCO (p < 0.05). TPVV (r = 0.34; p = 0.01), but not emphysema (r = −0.05; p = 0.67), correlated with lower DLCO after adjusting for age and height. Despite lower emphysema burden, Trace-E participants with reduced DLCO had lower TPVV, higher dyspnea, and lower peak work rate than the Mild-E with preserved DLCO (p < 0.05). Interestingly, TPVV (but not emphysema) correlated inversely with both dyspnea-work rate (r = −0.36, p = 0.004) and dyspnea-⩒E slopes (r = −0.40, p = 0.001). Reduced pulmonary vascular volume adjusted by emphysema extent is associated with low DLCO and heightened exertional ventilation in dyspneic smokers with minor emphysema. Impaired perfusion of non-emphysematous regions of the lungs has greater functional and clinical consequences than hitherto assumed in these subjects.


Introduction
There is growing recognition that low lung diffusing capacity for carbon monoxide (DL CO ) and increased exertional ventilation relative to metabolic demand (high minute ventilation (⩒ E )-carbon dioxide output (⩒CO 2 ) are consistently found in dyspneic smokers and patients with mild chronic obstructive pulmonary disease (COPD) [1][2][3][4][5].Since these abnormalities may reflect increased wasted ventilation in the physiological dead space [3,[6][7][8], they have been largely ascribed to reduced surface area for gas transfer due to emphysematous destruction of the alveoli [9,10].
It is noteworthy, however, that there is substantial variability in the relationship between DL CO and ⩒ E -⩒CO 2 with the extent of emphysema measured by computed tomography (CT) [11][12][13][14].Moreover, a sizable fraction of smokers with no or trace emphysema, as qualitatively assessed by CT [15], may have a DL CO below the lower limit of normal (LLN) and a high ⩒ E -⩒CO 2 [3,4].These intriguing findings suggest that mechanisms beyond emphysema may contribute to these functional abnormalities.In this context, impaired pulmonary perfusion in non-emphysematous regions of the lungs may be associated with higher absolute (i.e.whole-lung) ventilation relative to perfusion and areas of high ventilation/perfusion ratio [10], both conspiring to decrease DL CO [16,17] and increase ⩒ E -⩒CO 2 [6].The latter, in particular, may further increase secondary to heightened afferent ventilatory stimuli due to the "upstream" (increased pulmonary vascular pressures, reduced pulmonary vascular distensibility, right ventricle-pulmonary arterial uncoupling) [18] and the "downstream" (reduced left ventricle pre-load and low cardiac output) [19], correlates of impaired lung perfusion [20,21].
Currently, there are only a few options to estimate perfusion using standard, non-contrasted CT scans of the chest [22].More experience has been accrued with measurements of the total pulmonary vascular volume (TPVV), representing the combined volume of the intraparenchymal pulmonary arteries and veins and the aggregate small blood vessel volume for the whole lung [23][24][25].Interestingly, lower TPVV was associated with lower DL CO and greater exertional dyspnea in large population-based studies [19,24,26].Although TPVV is unlikely to faithfully reflect the capillary volume available for gas exchange, a low TPVV has been associated with reduced left-ventricular filling and left atrium volume, i.e. it provides a useful index of "bulk" pulmonary blood flow [19,27].Thus, relating emphysema-adjusted TPVV to DL CO , ⩒ E -⩒CO 2 , and exertional dyspnea in smokers may shed new light on the putative contribution of impaired perfusion of non-emphysematous parenchyma to these physiological and sensory abnormalities.
The present study, therefore, was undertaken to answer the following question: does impaired lung perfusion -as suggested by a low TPVV -contribute to decrease DL CO and increase ⩒ E -⩒CO 2 beyond the expected by emphysema extension in dyspneic smokers with largely preserved forced expiratory volume in 1 sec (FEV 1 )?We hypothesized that for a similar emphysema burden, smokers showing an abnormally low DL CO would have lower TPVV, higher ⩒ E -⩒CO 2 , and higher exertional dyspnea than those with preserved DL CO [10,21].

Participants
This retrospective study included 64 smokers who were recruited from general respiratory clinics and were screened for potential COPD in the Respiratory Investigation Unit at the Kingston Health Sciences Center in Kingston, Ontario, Canada between 2011 and 2014.Detailed inclusion and exclusion criteria for smokers included in the Canadian Respiratory Research Network database have been published elsewhere [5,28].Participants avoided smoking, caffeine, heavy meals, alcohol, and vigorous exercise for ≥4 h before visits.Although some participants were included in previously published studies [5,28], the current data set is different; thus, there is no overlap with any of these studies.The study received ethical approval from Queen's University's Health Sciences Research Ethics Board (DMED-1674-14).

Procedures
Pulmonary function tests (spirometry, body plethysmography, and DL CO ) were performed using SensorMedics system (Vmax229d and Autobox V62J; SensorMedics, Yorba Linda, CA).Using the LLN (5 th percentile) based on the Global Lung Function Initiative reference equations, participants were separated into preserved or reduced DL CO [29].On the same day, incremental cardiopulmonary exercise tests (CPETs) with dynamic measurements of inspiratory capacity (IC) [30] were conducted on an electronically braked cycle-ergometer (Ergoline800s; SensorMedics) using the SensorMedics-Vmax229d system™.Dyspnea and leg discomfort intensities were assessed by the modified 10-point Borg scale at rest, each workload, and at peak exercise [31].The work rate (WR) of 60 W represented the highest equivalent work rate (HEWR) achieved by all participants.
Chest quantitative CT image analysis was performed on data acquired within 3 weeks of the CPET visit using VIDA Vision Software (VIDA Diagnostics Inc., Coralville, Iowa) for measurements of the relative area of the lung with attenuation <-950 HU (%LAA -950 ) and pulmonary vascular measures as previously described [32,33].For pulmonary vessel segmentation, the approach was based on tube enhancement using a cylindrical shape model that employed an eigenvalue of the Hessian matrix which served as a filter to extract vessels.At junctions, vessel branch points were identified from noise by applying a thinning method which then allowed for the selection of objects with many branch points.TPVV was calculated for the entire segmented vascular tree [25].Participants were dichotomized based on emphysema (E) extent using the median percent (2.2%) of %LAA -950 into Trace-E or Mild-E, being further dichotomized into those with preserved or reduced DL CO .

Statistical analysis
The sample initially included 66 smokers in total, but two were excluded due to missing data.One-way analysis of variance was used for between-group comparisons with post hoc tests (Dunn's multiple comparison test, Tukey Test, or Holm-Sidak method).The association between DL CO and TPVV was tested using a multiple linear regression model while accounting for age, height, and %LAA -950 .Linear regression models were also used to test the association between TPVV, and relevant parameters acquired during CPET after accounting for age and %LAA -950 .Pearson's correlation was used to assess other linear associations.Data are reported as mean ± SD unless otherwise specified.Statistical significance was set at p < 0.05.
Participants with "preserved" versus "reduced" DL CO were well matched for age, height, body mass index, smoking history, spirometry, and lung volumes within each emphysema group (all p > 0.05; Table 1).Despite similar %LAA -950 within a given emphysema group (p > 0.05; Figure 1(a)), participants with reduced DL CO showed systematically lower TPVV than those with preserved DL CO (p < 0.05; Figure 1(b)).TPVV (Figure 1(d)), but not %LAA -950 (Figure 1(c)), was significantly related to DLCO in the whole sample after adjusting for age and height (Table 2).In addition, TPVV was lower in Trace-E showing reduced DL CO versus Mild-E but preserved DL CO (p < 0.05; Figure 1(b)).

Relationship with physiological responses to incremental CPET
Within each emphysema group, there were no between-group differences in maximal exercise capacity (peak oxygen uptake (⩒O 2 )); of note, however, participants with Trace-E and reduced DL CO had significantly lower peak work rate than those with Mild-E but preserved DL CO (Table 3).Irrespective of emphysema severity, ⩒ E /⩒CO 2 at its nadir (Table 3) and at the HEWR (Figure 2(a)) were higher, and end-tidal CO 2 (P ET CO 2 ) was lower (Table 3 and Figure 2(b)) -in participants with reduced versus preserved DL CO (p < 0.05).Higher ⩒ E at the HEWR in participants with reduced DLCO (Table 3) was achieved via faster breathing frequency (Fb) rather than larger tidal volume (V T ) (Table 3), i.e. they showed higher Fb/V T ratios (p < 0.05 in the Trace-E group; Figure 2(c)).As shown in Figure 2(d) and Table 3, there were no between-group differences in V T /IC (or other indexes of volume constraints) (p > 0.05); of note, V T /IC was typically below the threshold for critical inspiratory constraints (0.7) [30].

Relationship with sensory responses to exercise
Regardless of emphysema severity, participants showing reduced DL CO consistently reported higher dyspnea scores than those with preserved DL CO (p < 0.05; Figure 2(e)).For instance, whereas ~ 70% of participants (22/32) with preserved DL CO reported "none-to-very, very light" dyspnea at the HEWR (Borg scores 0-1), more than half of those with reduced DL CO reported at least "mild" dyspnea (Borg scores ≥ 2) (p < 0.01).Moreover, participants showing reduced DL CO either with Trace-E or Mild-E reported higher dyspnea scores compared to those showing preserved DL CO with Mild-E or Trace-E, respectively (p < 0.05; Table 3).Additionally, higher dyspnea-WR and dyspnea-⩒ E slopes were found in those with reduced versus preserved DL CO within the Trace-E group (p < 0.05; Table 3).The intensity of leg discomfort, however, did not differ between the groups (Figure 2(f)).TPVV (but not emphysema) correlated inversely with both dyspnea/WR slope (r=-0.36,p = 0.004) and dyspnea/⩒ E slope (r=-0.40,p = 0.001).Furthermore, in a linear regression model, TPVV was the only predictor of both dyspnea/WR slope (Rsqr = 0.20, p = 0.01), and dyspnea/⩒ E slope (Rsqr = 0.23, p = 0.002) after adjusting for %LAA -950 , Table 2. Similar results were obtained when DL CO  was considered in a multiple regression analysis (data not shown), i.e.only TPVV predicted exertional dyspnea (p < 0.01).

Discussion
The present study investigated the potential association between CT measurement of whole-lung vascular volume (TPVV) with key functional predictors of exertional dyspnea (low DL CO and high ⩒ E -⩒CO 2 ) in smokers showing only trace/mild emphysema.After controlling for indices of ventilatory dysfunction and the extent of emphysema, we found lower TPVV in subjects with reduced DL CO .Consistent with our main hypothesis, they showed higher ⩒ E -⩒CO 2 and reported higher dyspnea scores than their counterparts with preserved DL CO .Thus, subjects exposed to cigarette smoke showing pulmonary vascular attenuation exhibit higher ventilatory requirements on exertion which can importantly contribute to unpleasant respiratory sensations, independent of coexistent respiratory mechanical abnormalities [4][5][6][7][8].

Smoking and pulmonary vascular abnormalities
Potential mechanisms leading to low pulmonary vascular volume among smokers with preserved right ventricle function include a) emphysematous destruction, b) vascular compression by adjacent pockets of gas trapping, c) hypoxic vasoconstriction, and d) microvascular inflammatory damage/destruction [34][35][36][37], and e) reduced whole-lung perfusion.Mechanisms a) to c) are unlikely to contribute decisively to a low pulmonary vascular volume since TPVV varied markedly for a given emphysema burden (Figure 1), the groups were well-matched for the severity of airflow obstruction/gas trapping (Table 1), and none of the participants showed evidence of (resting or exertional) hypoxemia.
Given that TPVV provides a global index of pulmonary blood volume [23][24][25], no inferences can be drawn concerning the status of capillary blood flow (d).Thus, it is conceivable that lower TPVV in subjects with reduced DL CO reflected impaired overall lung perfusion (e) [19,27], likely associated with increased heterogeneity in distal blood flow relative to ventilation.For instance, microvascular abnormalities previously described in smokers [34][35][36]38] could also be present in larger vessels.Moreover, diminished pulmonary arterial distensibility and pulmonary artery-right atrium uncoupling are known causes of reduced overall pulmonary blood flow and increased exertional ventilation [39].Although the decrease in small vessel volume on CT has been associated with histological loss of vascular cross-sectional area in smokers [40], more research is required to clarify the relative contribution of large versus small pulmonary vessels and bronchial versus pulmonary circulation to TPVV [22].

The role of quantitative vascular CT in phenotyping dyspneic smokers
There is growing recognition that CT-based pulmonary vascular imaging is related to clinically-relevant outcomes in smokers [23,26], and in patients in the initial stages of COPD [19].In the population-based MESA study, lower TPVV on CT was associated with higher levels of activity-related dyspnea among ever-smokers, including those without COPD [19,26].In the current study, we extend on MESA's results by showing that this remains the case even in subjects with trivial emphysema; moreover, we objectively measured dyspnea intensity as a function of exercise intensity and ventilation (Table 3).We found that TPVV was associated with reduced DL CO and dyspnea intensity during exercise after adjusting for age and the degree of emphysema severity (Table 2).It is conceivable, however, that the association between TPVV and emphysema extension strengthens as the emphysema worsens in patients with moderate-to-severe COPD [11].Given the putative association between pulmonary vasculopathy and emphysema development [37], it remains to be demonstrated whether the vascular abnormalities seen in the sub-set of smokers with lower TPVV would predispose them to eventual alveolar destruction [36,41].

Linking a low TPVV and DL CO to high V . E -V . CO 2 and exertional dyspnea in smokers
An abnormally low DL CO among smokers might be a consequence of a) maldistribution of ventilation reducing the "accessible" alveolar volume (V A ), b) diminished alveolar surface area for gas exchange, and c) reduced pulmonary blood flow [16].Mechanisms a) and b) are unlikely to be major contributors in the present study since there were no between-group differences in the V A /TLC ratio (Table 1) and emphysema extent did not correlate with DL CO (Figure 1(c)).Impaired whole-lung perfusion exposed by a low TPVV [19,27] is poised to increase overall ventilation relative to perfusion and decrease perfusion across relatively well-ventilated areas of the lungs (c) [10].These abnormalities decrease the efficiency of the lungs as gas exchangers, reducing DL CO and increasing ⩒ E -⩒CO 2 [6,10,16,21].In this context, the low P ET CO 2 seen in patients with reduced DL CO and high ⩒ E -⩒CO 2 may indicate the diluting effects of poorly perfused areas on the expired CO 2 and/or alveolar hyperventilation [6].A high physiological dead space is thought to increase neurochemical afferent stimulation leading to a faster breathing pattern [8,42]; in fact, high ⩒ E in those with low DL CO was reached via a higher Fb (Table 3 and Figure 2(c)) despite the availability of mechanical inspiratory reserves to further increase V T (Figure 2(d)) [43].Since an exaggerated ventilatory response to exercise signals increased inspiratory neural drive [5,44,45], those abnormalities were translated into exertional dyspnea in smokers showing lower TPVV (Table 2).

Practical implications: current and future
The prevailing view that any decrease in DL CO reflects alveolar destruction should be viewed with caution in smokers.As depicted in Figure 1(c), participants showing % LAA of 2-3% similar to those reported in never-smokers, elderly subjects [46], have DL CO values as low as 20% predicted.Exertional dyspnea is frequently ascribed to unfitness in smokers with preserved FEV 1 : a reduced DL CO in these subjects should raise the suspicion of negative (largely vascular) consequences of smoking which were not captured by other pulmonary function tests or emphysema extent on CT.A high ⩒ E /⩒CO 2 on CPET might provide another piece of evidence supporting pulmonary vascular abnormalities -exposed by a low TPVV -even when CT is qualitatively and quantitatively normal with regards to parenchymal thresholds for emphysema.The best strategies are to combine quantitative CT metrics of pulmonary vascular volume with functional measurements at rest (DL CO ) and during exercise (⩒ E -⩒CO 2 ) in individual subjects, however, this deserves larger prospective investigations.Similarly, if future studies show a link between low TPVV and DL CO with negative circulatory outcomes, smokers with low DL CO might benefit from a more proactive approach to cardiovascular risk reduction [47].For the same reasons, these subjects might be particularly prone to derive benefit from any novel therapeutical option to mitigate the negative effects of cigarette smoking on pulmonary vasculature [34].In this perspective, preliminary data showed that vasodilators can reduce ⩒ E /⩒CO 2 and dyspnea without changing lung mechanics [48].

Study limitations
TPVV is a static, global measure of lung vascular volume, likely reflecting a simultaneous reduction in arterial, venous, and microvascular volumes, thus, it does not provide an accurate estimation of blood flow across the gas-exchanging units.For instance, narrowing and loss of the small vessels may coexist with proximal dilation and engorgement of the large vessels in pulmonary arterial hypertension [49], the net impact on TPVV likely being highly variable in individual subjects [50].We did not have objective data about central hemodynamics which may have contributed to a low TPVV [19]; however, those with known cardiac comorbid conditions were excluded and O 2 pulse (a CPET variable influenced by stroke volume) did not differ between subjects with preserved versus reduced DL CO (Table 3).DL CO was not corrected for differences in circulating hemoglobin, but participants had no history of anemia.Although active smokers were instructed to refrain from smoking for ≥4 h before visits, carboxy-hemoglobin concentration was not measured.
In a posthoc analysis (data not shown), we found that females had less emphysema but similar DL CO (% predicted) compared to males; interestingly, the former group showed lower TPVV but higher dyspnea burden.The interaction between sex-specific attributes and the TPVV-emphysema relationship deserves further scrutiny.

Conclusion
Among smokers with only minor emphysema, reduced DL CO was associated with low pulmonary blood volume -but not emphysema burden -as measured by quantitative CT.These vascular abnormalities were associated with higher ventilatory demands on exertion, prompting greater activity-related dyspnea.Thus, impaired perfusion in non-emphysematous regions of the lungs has greater functional and clinical consequences than hitherto assumed in smokers with largely preserved FEV 1 .The emergence of protective treatments to halt the negative consequences of smoking on pulmonary vessels before progression to extensive emphysema might prove useful to address two potentially treatable traits of mild-moderate COPD: gas exchange inefficiency and exertional dyspnea.

Figure 1 .
Figure 1.selected measurements of chest quantitative computed tomography: (a) percentage of emphysema as assessed by low attenuation area (laa) <-950 hu (laa <-950hu ), and (b) total pulmonary vascular volume (tPvv) in smokers separated by emphysema burden and Dl CO .boxes depict the first to third quartiles; central lines denote the median.Whiskers range from the 5 th to the 95 th percentiles.Panels (c) and (d) show Pearson's correlations between Dl CO and both laa <-950hu (panel c) and tPvv (panel d) where dotted lines represent the 95% confidence interval (Ci) for the slope of the regression line.Abbreviations: Dl CO , single breath diffusing capacity of the lung for carbon monoxide.

Figure 2 .
Figure 2. selected ventilatory and sensory parameters during incremental cardiopulmonary exercise test at the highest equivalent work rate of 60 watts in smokers separated by emphysema burden and Dl CO .(a) ventilatory equivalent for carbon dioxide output (V .e /V .CO 2 ), (b) end-tidal CO 2 (P et CO 2 ), (c) breathing frequency/tidal volume ratio (Fb/v t ), (d) v t /inspiratory capacity (iC), (e) dyspnea borg ratings and (f ) leg discomfort borg ratings.boxes depict the first to third quartiles; central lines denote the median.Whiskers range from the 5 th to the 95 th percentiles.Abbreviations: Dl CO = single breath diffusing capacity of the lung for carbon monoxide.

Table 1 .
subjects general and resting functional characteristics.
a p < 0.05 trace-e preserved Dl CO vs. trace-e reduced Dl CO .b p < 0.05 trace-e preserved Dl CO vs. mild-e reduced Dl CO .c p < 0.05 mild-e preserved Dl CO vs. mild-e reduced Dl CO .d p < 0.05 trace-e reduced Dl CO vs. mild-e preserved Dl CO .abbreviations: bmi, body mass index; Cat, COPD assessment test; Dl CO , single breath diffusing capacity of the lung for carbon monoxide; f, female; fev 1 , forced expired volume in 1 s; frC, functional residual capacity; fvC, forced vital capacity; iC, inspiratory capacity; laa, low attenuation area of the lung below -950 hu as a percentage of the total lung area; lln, lower limit of normal; m, male; mild-e, mild emphysema; mrC, medical research Council; ns, non significant; rv, residual volume; taC, total airway count; tlC, total lung capacity; tPvv, total pulmonary vascular volume; trace-e, trace emphysema; v a , alveolar volume measured by single breath gas dilution; aW, airway.

Table 2 .
results of the multiple linear regression models predicting Dl CO and exertional dyspnea.

Table 3 .
selected physiological and sensory responses at peak and submaximal exercise in smokers separated by emphysema burden and Dl CO .
a p < 0.05 trace-e preserved Dl CO vs. trace-e reduced Dl CO .b p < 0.05 trace-e preserved Dl CO vs. mild-e reduced Dl CO .c p < 0.05 mild-e preserved Dl CO vs. mild-e reduced Dl CO .d p < 0.05 trace-e reduced Dl CO vs. mild-e preserved Dl CO .abbreviations: Dl CO , single-breath diffusing capacity of the lung for carbon monoxide; Fb, breathing frequency; heWr, highest equivalent work rate; hr, heart rate; iC, inspiratory capacity; irv, inspiratory reserve volume; mild-e, mild emphysema; O 2 pulse, oxygen pulse; P et CO 2 , partial pressure of end-tidal carbon dioxide; ns, non significant; rer, respiratory exchange ratio; spO 2 , arterial oxygen saturation measured by pulse oximetry; tlC, total lung capacity; trace-e, trace emphysema; V .e , minute ventilation; V .e /V .CO 2, ventilatory equivalent for carbon dioxide output; V .O 2 , oxygen uptake; v t tidal volume; Wr, work rate.