RESULTS

Section:

ChooseTop of pageAbstractMETHODSRESULTS <<DISCUSSIONReferencesCITING ARTICLES

All subjects were white, with a mean age of 52 years (range 23 to 74 years). Almost all (110) were index cases, and males outnumbered females (70 to 49). The majority had been cigarette smokers (75) or were currently smoking at the time of analysis (17), and 62 patients described symptoms of chronic bronchitis (32). The characteristics of the entire group are shown in Table 1

Emphysema Distribution and Relationship of Densitometry to Physiology

Preliminary assessment of the relationship between physiology and CT densitometry indices in the whole group confirmed that Dl_CO/Va correlated better with the upper zone than the lower zone indices but that FEV₁ correlated better with the lower zone indices (Figure 3)

Figure 3. Preliminary assessment of the relationship between physiology and emphysema distribution correlating computed tomography (CT) densitometry indices derived from upper zone images (at the level of the aortic arch) and lower zone images (at the level of the inferior pulmonary veins) with airflow obstruction and gas transfer. (A) FEV₁ correlates better with lower zone VI (r = −0.741, p < 0.001) than (B) upper zone VI (r = −0.511, p < 0.001), whereas diffusing capacity of the lung for carbon monoxide (Dl_CO/V_A) correlates better with (C) upper zone VI (r = −0.625, p < 0.001) than (D) lower zone VI (r = −0.563, p < 0.001).

Download Figure (Part 1) | Download Figure (Part 2) | Download Figure (Part 3) | Download Figure (Part 4) | Download Powerpoint

[More] [Minimize]

, supporting the findings of our previous limited study (9). The degree of relative hypocapnia correlated with whole lung VI (r = −0.535, p < 0.001, n = 119) (Figure 4)

Figure 4. Correlation between total VI at a threshold of −950 HU and relative hypocapnia as defined by Burrows and colleagues (38) (n = 119, r = −0.53, p < 0.001).

Download Figure | Download Powerpoint

[More] [Minimize]

and the 15th percentile point (r = 0.526, p < 0.001).

Emphysema distribution was assessed primarily using VI, although the correlation between VI and 15th percentile point was good (r = 0.994, p < 0.001), suggesting that either method could have been used for this purpose. The VI profiles in 17 subjects, all with a WLVI of less than 3%, demonstrated small random peaks that were considered to be inconsistent with significant emphysema. These subjects were excluded from further analysis, and their characteristics are shown in Table 2

TABLE 2. Characteristics of subjects with no demonstrable emphysema

	Number	Mean	SD	Mean % Predicted
Age	17	44	10.5	N/A
Pack-years	17	7.4	11.5	N/A
FEV1, L	17	3.1	0.8	100
VC, L	17	4.1	1.1	110
RV, L	17	1.8	0.6	100
TLC, L	17	6.1	1.4	108
DLCO/VA	17	1.5	0.3	87
Whole lung VI, %	17	1.0	0.7	N/A
Upper zone VI , %	17	0.7	1.2	N/A
Lower zone VI , %	17	0.8	0.7	N/A
Total Perc15	17	−889.9	19.34	N/A
Upper zone Perc15	17	−888.2	21.1	N/A
Lower zone Perc15	17	−890.3	19.9	N/A
SGRQ, total score	15	26.6	9.5	N/A
SF-36, physical	17	45.9	8.2	N/A
SF-36, mental	17	57.01	4.9	N/A

For definition of abbreviations see Table 1.

The patient characteristics are given together with the number of subjects from whom data have been obtained. The average and SD for the group are shown together with the average values expressed as a percentage predicted. All lung function measurements were performed after dual bronchodilation with inhaled nebulized salbutamol (2.5 mg) and ipratropium (250 μg).

, indicating that they have essentially normal lung function, although the average Dl_CO/Va less than 100% predicted suggests a mild degree of impaired gas transfer. The majority of the remaining 102 patients (65 patients) had a VI profile consistent with basal emphysema (basal group), and 37 had emphysema extending into the upper lung region (apical group) (Figure 2).

Patients with emphysema were matched primarily for overall emphysema severity (WLVI) and secondarily for age by selecting 37 subjects from the 65 in the basal group to provide the best available match with the 37 patients in the apical group. The characteristics of the two groups and the results of the matched pairs analysis are shown in Table 3

TABLE 3. Matched pairs comparison of subjects with “basal” and “apical” emphysema distribution

	Basal	Apical
	Mean (SD)	Mean (SD)	Mean Difference (95% CI)	p Value
Age	53 (8.8)	55 (9.4)	−2.7 (−6.2 to 0.9)	0.133
Smoking, pack-years	18.8 (11.6)	18.3 (17.1)	0.6 (−7.0 to 8.1)	0.882
Whole lung VI, %	19.2 (10.1)	19.0 (9.9)	0.2 (−0.1 to 0.6)	0.219
Upper zone VI, %	8.4 (7.6)	17.7 (12.8)	−9.3 (−12.3 to −6.3)	< 0.001
Lower zone VI, %	24.0 (13.5)	19.9 (1.0)	4.2 (1.7 to 6.7)	0.001
Whole lung Perc15	−953.8 (17.6)	−952.9 (18.4)	−0.4 (−1.4 to 0.6)	0.436
Upper zone Perc15	−931.9 (21.3)	−950.2 (23.7)	16.1 (9.8 to 22.5)	< 0.001
Lower zone Perc15	−958.8 (17.2)	−956.2 (18.0)	−3.07 (−0.2 to −6.0)	0.039
FEV1, % predicted	41.8 (16.3)	51.8 (18.5)	−9.9 (−16.0 to −3.8)	0.002
VC, % predicted	101.1 (25)	111.7 (21.6)	−10.6 (−20.6 to −7.1)	0.036
DLCO/VA, % predicted	59.4 (15.6)	53.9 (14.5)	6.1 (0.2 to 12.1)	0.044
VA, % predicted TLC	93.0 (13.5)	99.5 (14.4)	−6.5 (−12.6 to −0.4)	0.037
PaO2, kPa	9.0 (0.9)	8.5 (1.0)	0.5 (0.03 to 1.0)	0.016
A-a DO2	6.1(1.1)	6.8 (0.9)	−0.7 (−1.2 to −0.2)	0.007
PaCO2, kPa	4.7 (0.5)	4.5 (0.6)	0.1 (−0.1 to 0.4)	0.280
Relative hypocapnia	−0.7 (0.7)	−0.6 (0.7)	−0.1 (−0.4 to 0.2)	0.365

Definition of abbreviations: A-a DO₂ = alveolar–arterial oxygen gradient; CI = confidence interval; DL_CO/V_A = diffusing capacity of the lung for carbon monoxide (mmol/min/kPa/L); Perc15 = 15th percentile point; VI = voxel index.

Pairs were successfully matched for total amount of emphysema and age with no significant differences between groups. Data are presented as mean with SD in parentheses. The mean difference (“basal” minus “apical”) and 95% CIs are shown, together with the significance of any differences between groups (p).

, with the key findings illustrated in Figure 5

Figure 5. Key results of the matched-pairs analysis showing the mean difference between groups (basal minus apical) with the 95% confidence intervals. The significance of any differences are given (p). (A) Differences for whole lung VI (WLVI) and age are nonsignificant, indicating that the groups were well matched. Significant differences in upper zone VI (UZVI) and lower zone VI (LZVI) between groups demonstrates that the VI profile method used for grouping subjects successfully identified differences in the distribution of emphysema between the upper and lower lung regions. (B) Subjects with basal distribution have greater impairment of FEV₁ and VC than subjects with apical distribution but less impairment of gas exchange as shown by a higher Dl_CO/V_A, and (C) higher Pa_O2 and lower alveolar-arterial oxygen gradient.

Download Figure (Part 1) | Download Figure (Part 2) | Download Figure (Part 3) | Download Powerpoint

[More] [Minimize]

. Assessment of the matching process demonstrated no significant difference between the groups in overall emphysema severity (WLVI) or age, but evaluation of the distribution of emphysema in the two groups confirmed that there was a significant difference in the lower zone and upper zone VI (Figure 5A). The basal distribution pattern was associated with a lower FEV_1, VC, Va, and alveolar–arterial oxygen gradient and a higher Dl_CO/Va and Pa_O2 than the apical pattern (Table 3). Of importance, the mean difference (95% confidence intervals) in FEV₁ was −9.9% (−16.0 to −3.8) and in Dl_CO/Va was 6.1% (0.2 to 12.1), as summarized in Figure 5B. The difference between the two groups could not be explained by differences in sex, index status, smoking history, health status indices, or prevalence of chronic bronchitis (data not shown).

Relationship between PD and Emphysema Distribution

Discordance between FEV₁ and Dl_CO/Va (Dl_CO/Va percentage predicted − FEV₁ percentage predicted) was significantly related to the relative distribution of emphysema described by the difference between upper and lower zone CT indices (lower zone 15th percentile point − upper zone 15th percentile point, r = −0.409, p < 0.001; and lower zone VI − upper zone VI, r = 0.429, p < 0.001). The individual data for PD and distribution (Δ CT) determined by the 15th percentile point are shown in Figure 6

Figure 6. Relationship between physiologic discordance (Dl_CO/V_A percentage predicted − FEV_1% predicted) on the vertical axis and relative distribution of emphysema (lower zone Perc15 − upper zone Perc15) on the horizontal axis; Spearman's correlation coefficient = 0.409, p < 0.001. Relative preservation of Dl_CO/V_A is associated with predominantly basal emphysema, whereas relative preservation of FEV₁ (“nonobstructive emphysema”) is associated with apical predominance.

Download Figure | Download Powerpoint

[More] [Minimize]

. The association of airflow impairment with relative preservation of diffusing capacity related to a predominantly basal pattern of emphysema, whereas the reverse pattern of discordance was related to predominantly apical emphysema.

DISCUSSION

Section:

ChooseTop of pageAbstractMETHODSRESULTSDISCUSSION <<ReferencesCITING ARTICLES

We have shown that the relative degree of impairment in measures of airflow obstruction and gas exchange differs between age-matched subjects with the same overall amount of emphysema in association with the pattern of emphysema distribution. Emphysema that is predominantly located in basal zones is associated with a greater degree of airflow obstruction but a lesser impairment of gas exchange than emphysema extending to the upper zones. These physiologic differences were consistently identified in further analyses that used alternative matching criteria. In the first of these analyses, subjects were matched using WLVI as described in Methods; however, smoking history (pack-years) replaced age as the secondary matching criterion, and a further analysis matched subjects for overall emphysema severity using the 15th percentile point. The results of these two analyses are tabulated in the online supplement.

These findings have important clinical implications. The use of spirometry is recommended in the updated Global Initiative for Chronic Obstructive Lung Disease guidelines for the management of COPD (33) for confirming the diagnosis earlier in “at-risk” patients and for severity staging. It is likely that this will be increasingly performed in a primary care setting as the only physiologic measure. Consequently, it should be widely recognized that emphysema may occur without demonstrable airflow obstruction (namely, Global Initiative for Chronic Obstructive Lung Disease stage 0) but, nevertheless, with coexisting abnormalities in gas exchange. Our findings would suggest that this physiologic pattern is more likely to occur in subjects with usual COPD, in whom the emphysema is more commonly apical, than in AATD-associated COPD. Furthermore, the sole use of FEV₁ for the classification of disease stage in COPD may lead to the relative overestimation of severity in subjects with predominantly basal emphysema compared with subjects with apical disease. It is also of importance that recognition is given to the pattern of discordance described previously here in association with predominantly basal disease (namely, impairment of FEV₁ with relative preservation of Dl_CO/Va) because the diagnosis of AATD is not infrequently delayed (34), with symptoms ascribed to asthma rather than emphysema. The adherence to World Health Organization guidelines for AATD testing in all subjects with COPD or late-onset asthma should, however, prevent this from occurring (35).

In addition, our findings indicate that relatively greater involvement of the upper zones with emphysema may contribute an additive risk for development of pulmonary hypertension secondary to hypoxemia. FEV₁ has been an important predictor of mortality in usual COPD and AATD-associated emphysema, but Dawkins and colleagues using a limited slice protocol have shown recently that CT densitometry applied to upper zone images is a better predictor of mortality in subjects with AATD (36). Mortality in this population is likely related to emphysema severity and consequently to the degree of CT abnormality, which is a more specific measure of emphysema than FEV₁. Our study suggests the possibility that this also has a physiologic explanation, namely, that measurement of the extent of upper zone emphysema may allow identification of a subgroup of patients at greater risk from hypoxemia and possible subsequent pulmonary hypertension. This effect could constitute an increased mortality risk, but further studies are needed to determine the validity of this hypothesis.

Previous studies have inferred regional differences in function from the strength of correlations between physiology and objective CT parameters acquired using limited sampling protocols (1, 16, 18, 37). The scanning protocol employed in this study enabled a comprehensive characterization of disease distribution, and by matching subject pairs for overall emphysema severity, we have provided the first direct evidence of different functional impairment between emphysematous involvement of the upper and lower lung regions. It is likely that the graphical display of lung density is an accurate reflection of both emphysema severity and distribution. Relative hypocapnia has previously been shown to predict emphysema severity (38), and the good correlation with lung density (Figure 4) in this study provides further evidence that CT densitometry relates well to physiologic measures of emphysema. Furthermore, the significant differences between pairs in upper zone and lower zone VI and the 15th percentile point confirms that the method of grouping into basal and apical identified differences in emphysema distribution using either method of calculation.

Emphysema in subjects with AATD is classically described as predominantly basal, but we have shown objectively using CT densitometry profiles that the pattern of emphysema distribution is more heterogeneous than recognized previously (8–10). The matched-pairs analysis and an additional analysis of all 102 patients (data not shown) did not identify any significant differences in age, index status, or smoking history to account for these different patterns of emphysema distribution. Our findings are of importance because guidelines published recently advise antitrypsin testing of subjects with basal emphysema with the implication that other patterns of distribution are not associated with AATD and that testing in these cases is unnecessary (15). The current study in this group of patients suggests that this advice may be misleading in that 36% of patients with emphysema did not have basal predominance. Furthermore, there was sufficient heterogeneity to demonstrate an association between PD of FEV₁ and Dl_CO/Va and emphysema distribution (see Figure 6). It is likely that the cases of discordance reported previously (20, 21) may also relate to the pattern of emphysema distribution as described in this study, particularly because we have similarly demonstrated a relationship between discordance and distribution in a study of subjects with usual COPD (22). However, the mechanism of this effect remains unclear.

The suggestion by Wilson and Galvin (21) that predominantly lower lobe disease may allow preservation of ventilation–perfusion matching so that diffusing capacity is relatively insensitive to the loss of surface area for gas exchange is likely to be only a partial explanation. It is possible that real physiologic differences do exist between the upper and lower lung regions and that these may be responsible for the differences in static lung function parameters and gas exchange that we have identified. The gravitational influences on diffusing capacity and FEV₁ measurements made in the sitting posture are likely to be affected by the pattern of emphysema distribution. In health, the ventilation perfusion ratio varies between approximately 0.7 at the lung bases to 3 at the apex as a result of the changing relationship between ventilation and perfusion (26). The gravitational influences on the perfusion differential between lung apex and base would allow subjects with basal emphysema to maintain Dl_CO/Va through the recruitment of underperfused disease-free lung units in the upper regions. This recruitment would result from the increase in pulmonary perfusion pressure arising secondary to emphysematous change. In contrast, subjects with apical emphysema would be unlikely to maintain gas exchange by similar recruitment of inferior disease-free lung units because these would already be well perfused. Impairment of FEV₁ when measured in a sitting position is likely to be greater in basal than in apical emphysema. The distending forces acting across airway walls reflect intrapleural pressure, and because this becomes less negative toward the lung bases, the distending airway forces and airway patency will therefore be less at the bases than at the apices (39). Furthermore, dynamic airway collapse resulting from reduced parenchymal tethering is likely to be greater in basal emphysema, leading to increased impairment of FEV₁.

In conclusion, variability in emphysema distribution pattern and its relationship with physiology have implications for the use of a single physiologic parameter in screening and in monitoring emphysema progression or response to treatment. Although it is accepted that optimal clinical practice indicates that full lung function testing should be performed in subjects with obstructive pulmonary disease (15, 40), this study demonstrates how the radiologist can obtain additional information derived from CT imaging to facilitate interpretation of physiologic status. The identification of subgroups within AATD is likely to be important in understanding the heterogeneity that exists in the natural history of disease progression in subjects with AATD, and this may also be applicable to usual COPD (41).