Rate of progression of lung function impairment in α₁-antitrypsin deficiency

Progression of emphysema in α₁-antitrypsin (α₁-AT) deficiency (α₁-ATD) is known to occur at an accelerated rate compared with usual chronic obstructive pulmonary disease (COPD) 1, 2. At present, there is uncertainty regarding which patients show the greatest rate of progression, and, therefore, may show the clearest signal for α₁-AT augmentation trials or response to future treatments. This reflects a lack of knowledge of the natural history of the disease and completion of effective clinical trials of treatment.

Logically, effective preventative therapy should be introduced early in order to prevent subsequent deterioration. However, it is not currently known whether all patients with α₁-ATD deteriorate and at what rate. Index patients identified by presentation to healthcare services exhibit worse lung function than matched non-index siblings 3. Lifelong nonsmokers show less progression and lower mortality 1, 4, but a significant number develop airflow obstruction in middle age 5. Nevertheless, many subjects remain unidentified because either the diagnosis has not been considered or they remain clinically well. In order to identify all patients, extensive screening would be necessary, with long-term follow-up, such as in the Swedish cohort study 6.

The variation in progression rate in α₁-ATD patients has also hampered clinical therapy trials since large numbers of subjects need to be studied over a long period of time in order to determine efficacy 7. However, targeting only patients who are rapidly progressing for such trials would reduce the numbers needed and decrease the necessary duration of the study. Indirect data from the US National Institutes of Health (NIH) registry provide supporting evidence for this concept. The only patients demonstrating a possible benefit of augmentation therapy were those in the moderately affected group with a rapidly declining forced expiratory volume in 1 s (FEV₁) 1. Furthermore, this is supported by the observation that rapid decliners in the German sequential study showed a subsequent slowing of progression following augmentation 8.

The present study was designed to answer several questions. First, it was designed to assess the progression of airflow obstruction and a more specific measure of emphysema (the carbon monoxide transfer factor) in patients with a broad spectrum of physiological impairment; secondly to identify factors that are associated with the decline in lung function, and; finally, to determine factors that are associated with the most rapid decline in order to identify the most appropriate patients for clinical trials and those most likely to benefit from effective interventions.

METHODS

The UK Antitrypsin Deficiency Assessment and Programme for Treatment (ADAPT) programme (funded by a noncomercial grant from Talecris Biotherapeutics Inc., Research Triangle Park, NC, USA) has been collecting data on α₁-ATD patients prospectively since 1996, in order to gain understanding of the natural history of the condition and form a basis for future treatments. None received α₁-AT augmentation therapy, since it is not yet licensed in the UK. At the time of analysis, all patients who had been followed for ≥3 yrs were identified; 40 patients were excluded because they had less than four consecutive annual lung function measurements including that at baseline. An additional three were excluded because they had received lung transplants. These 43 excluded patients were of milder severity, on average, than those included in the analysis (mean baseline FEV₁ 70.5% of the predicted value in those excluded compared with 54.3% pred in those included). There were thus 101 patients with a PiZ phenotype on the registry who had had lung function recorded annually over a 3-yr period. Using regression equations, the mean decline in FEV₁ and carbon monoxide transfer corrected for lung volume (K_CO) was calculated over the 3 yrs (four measurements) for each patient. The patients were then divided into groups according to baseline percentage predicted FEV₁, equivalent to American Thoracic Society (ATS)/European Respiratory Society (ERS) severity groups for COPD 9. The mean decline in FEV₁ and K_CO over 3 yrs was then determined for each group. Factors associated with the decline were identified from baseline characteristics by univariate analysis. All correlates were then entered into a linear regression analysis in order to identify independent factors that predicted overall decline. This compared FEV₁ and K_CO decline as continuous variables against the factors, adjusting for age, sex, cumulative smoking exposure and baseline lung function.

In order to identify factors characteristic of rapid decline, the 101 patients as a whole were then separately divided into tertiles according to speed of FEV₁ decline. The fast-decline tertile and slow-decline tertile were compared, using univariate and multivariate analyses, for differences in the following parameters, assessed at baseline: sex; body mass index (BMI); acute reversibility to bronchodilator (BDR; defined by ATS criteria; ≥200 mL change in FEV₁ and 12% change from baseline FEV₁ after 400 μg inhaled salbutamol 10); smoking status; chronic bronchitis (UK Medical Research Council criteria 11); age; health status scores, from the 36-item short-form health survey (SF-36) physical and mental component scores, and St George’s Respiratory Questionnaire total score; exacerbation rates characterised as type 1 and 2 as described by Anthonisen 12, derived from self-reported retrospective recall on an annual questionnaire; baseline FEV₁ (% pred); baseline K_CO (% pred); and extent of emphysema on computed tomography (CT) scan (inspiratory and expiratory films, lower and upper zones) using the voxel index (-910 HU) as described previously 13.

The 95 out of the 101 patients who had complete K_CO data were also divided into tertiles according to their rate of K_CO decline, and univariate and multivariate analyses were performed comparing the fast- and slow-decline tertiles for the same parameters as used in the decline in FEV₁ analyses described above.

The lung function equipment used was the MasterScreen PFT (Jaeger, Hoechberg, Germany), and quality control of equipment and technician input was according to ATS/ERS standards 14–16.

High-resolution CT scans were performed, using a GE ProSpeed Scanner (General Electric Medical Systems, Milwaukee, WI, USA) to obtain 1-mm slices. The scanner was calibrated weekly using water and air. A full scan was performed at maximal inspiration (10-mm intervals) and a limited scan on expiration (30-mm intervals). Two slices were chosen for analysis, the level of the aortic arch (upper zone) and the level of the inferior pulmonary vein/right atrial confluence (lower zone). The data were subjected to density-mask analysis, which highlighted lung voxels with a density of <-910 HU. The voxel index is the percentage of highlighted voxels with a density below this threshold, reflecting the proportion of emphysematous tissue.

The exacerbation data were obtained from annual questionnaires based on retrospective recall. The questions were as follows. 1) “Have you had any episodes of increased sputum volume or purulence since the last visit? If yes…How many? Which months?” 2) “Have you had any episodes of increased breathlessness since the last visit? If yes…How many? In which months?”. Where the answer was yes to both questions, the number of occasions on which the identified months matched was the number of Antonisen type 1 (all three symptoms) and type 2 exacerbations (two of the three symptoms) during that year 17.

Ethical approval was granted by the Local Research Ethics Committee (University Hospital Birmingham, Birmingham, UK), and all patients gave informed consent for the investigations.

RESULTS

For the patients as a whole, the mean decline in FEV₁ was 49.9±7.4 mL·yr⁻¹. When divided into severity groups according to baseline percentage predicted FEV₁, the fastest mean decline in FEV₁ occurred in the moderate severity group (FEV₁ 50–80% pred) at 90.1±19.7 mL·yr⁻¹. The speed of decline was also faster than average in the severe group (FEV₁ 30–50% pred) at 51.9±7.6 mL·yr⁻¹, but lower than average in the mild group (FEV₁ >80% pred) at 31.6±19.3 mL·yr⁻¹ and in the very severe group (FEV₁ <30% pred) at 8.1±9.6 mL·yr⁻¹. These results are summarised in figure 1⇓.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1—

Forced expiratory volume in 1 s (FEV₁) decline according to severity group based on percentage predicted FEV₁ (mild: >80% predicted (31.6±19.3 mL·yr⁻¹; n = 18); moderate: 50–80% pred (90.1±19.7 mL·yr⁻¹; n = 26); severe: 30–50% pred (51.9±7.6 mL·yr⁻¹; n = 38); very severe: <30% pred (8.1±9.6 mL·yr⁻¹; n = 19)). Data are presented as mean±sem FEV₁ at the start and end of the 3-yr follow-up. The greatest decline occurred in the moderate severity group. The overall mean decline in FEV₁ was 49.9±7.4 mL·yr⁻¹.

However, the results for K_CO decline differed from those for FEV₁. The mean K_CO decline for the whole group was 0.015±0.004 mmol·min⁻¹·kPa⁻¹·L⁻¹·yr⁻¹. When divided into severity groups according to baseline percentage predicted FEV₁, there was a faster decline in K_CO in the severe (0.030±0.006 mmol·min⁻¹·kPa⁻¹·L⁻¹·yr⁻¹) and very severe groups (0.025±0.008 mmol·min⁻¹·kPa⁻¹·L⁻¹·yr⁻¹) than in the moderate (0.004±0.007 mmol·min⁻¹·kPa⁻¹·L⁻¹·yr⁻¹) and mild groups (-0.0122±0.012 mmol·min⁻¹·kPa⁻¹·L⁻¹·yr⁻¹) (fig. 2⇓).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2—

Transfer coefficient of the lung for carbon monoxide (K_CO) decline according to severity group based on percentage predicted FEV₁ (mild: >80% predicted (-0.0122±0.012 mmol·min⁻¹·kPa⁻¹·L⁻¹·yr⁻¹; n = 16); moderate: 50–80% pred (0.004±0.007 mmol·min⁻¹·kPa⁻¹·L⁻¹·yr⁻¹; n = 26); severe: 30–50% pred (0.030±0.006 mmol·min⁻¹·kPa⁻¹·L⁻¹·yr⁻¹; n = 36); very severe: <30% pred (0.025±0.008 mmol·min⁻¹·kPa⁻¹·L⁻¹·yr⁻¹; n = 17)). Data are presented as mean±sem K_CO at the start and end of the 3-yr follow-up. The greatest decline occurred in the severe and very severe groups. The overall mean decline in K_CO was 0.015±0.004 mmol·min⁻¹·kPa⁻¹·L⁻¹·yr⁻¹.

Multiple linear regression of FEV₁ decline as a continuous variable on the factors listed in table 1⇓, adjusting for age, sex, cumulative smoking exposure and baseline FEV₁, showed baseline K_CO, upper zone inspiratory CT scan voxel index and BMI were significantly associated with fast decline. BMI was most strongly associated with FEV₁ decline (p = 0.008), and, once this was entered into the model, none of the other possible explanatory variables were significant.

Table 1⇑ shows the results of univariate analysis of parameters that may be associated with FEV₁ decline (with p-values) for differences between the fast decline (n = 33) and slow decline tertile (n = 34). In the fast-decline group, there were more patients with BDR (73 versus 41%; p = 0.010), more males (79 versus 56%; p = 0.048) and a lower BMI (24.0 versus 26.1; p = 0.042). Multivariate analyses comparing the fast- and slow-decline tertiles indicated that the features that were independently predictive of fast decline in FEV₁ were BDR, low BMI, high exacerbation rate and a high SF-36 component score (table 2⇓).

Multiple linear regression of K_CO decline as a continuous variable on the factors listed in table 1⇑, adjusting for age, sex, cumulative smoking exposure and baseline K_CO, showed that baseline FEV₁ and the four CT scan voxel indices were significantly associated with fast decline. The lower zone expiratory CT scan voxel index showed the strongest association with K_CO decline (p = 0.002), and, once this was entered into the model, none of the other possible explanatory variables were significant.

Table 3⇓ shows the results of univariate analysis for parameters potentially associated with K_CO decline, when comparing the fast- and slow-decline tertiles. FEV₁ (41.8 versus 60.2% pred; p = 0.002) and mean emphysema voxel index scores on lower zone expiratory scan (47.4 versus 33.1%; p = 0.010) and upper zone expiratory scan (24.2 versus 16.6%; p = 0.042) were significantly different between the two groups. When multivariate analyses were performed comparing the fast- and slow-decline tertiles (table 4⇓), the only parameter that was independently predictive of fast decline in K_CO was FEV₁.

DISCUSSION

The UK database provides a unique opportunity for studying multiple factors in a cohort of highly characterised α₁-ATD patients not receiving augmentation therapy. Those with consecutive annual lung function measurements showed a mean decline in FEV₁ determined by summary statistics over a 3-yr period of 49.9 mL·yr⁻¹. There have been few such studies reported in the literature, although the patients in the placebo group (n = 28) in the Dutch/Danish pilot study of α₁-AT augmentation therapy 7 showed a mean decline in FEV₁ of 59.1 mL·yr⁻¹ over 3 yrs. In a comparative study between Danish patients not receiving α₁-AT augmentation (n = 97) and German patients receiving augmentation 18, the Danish group exhibited a mean decline in FEV₁ of 75.0 mL·yr⁻¹. In a German study before and after α₁-AT augmentation treatment 8, the pre-treatment group (n = 96) showed a decline in FEV₁ of 49.2 mL·yr⁻¹. Finally, in a US α₁-AT registry study 1, the mean decline in FEV₁ was 56 mL·yr⁻¹ in those never receiving α₁-AT augmentation therapy. Thus, with the exception of the Danish/German comparative group 18, data from all of these studies are comparable, despite the wide range of initial FEV₁ in the present patients.

The decline is dependent upon several factors. First, it relates to the initial FEV₁, and the present data show that the greatest change (90.1 mL·yr⁻¹) occurs in those with an initial moderate FEV₁ impairment (50–80% pred), which is comparable with results from the US registry of 81.2 mL·yr⁻¹ in those not receiving augmentation therapy 1. The lack of decline (mean 8.1 mL·yr⁻¹) in the most severe group probably reflects a survivor effect 12, since, by study design, data could only be obtained from patients who survived ≥3 yrs. Since mortality reflects FEV₁ 19–24 it is likely that any rapid decliners in this group would have died during the study period. Why this observation is at variance with data from the NIH report for a similar group (mean decline of 46.5 mL·yr⁻¹ in those with an FEV₁ of <35% pred not receiving augmentation therapy) remains unknown, especially since the median follow-up time was longer (52 months) in the NIH study.

When FEV₁ decline was compared as a continuous variable, correcting for various confounding factors, BMI was found to show the best association in this more general analysis. In order to identify a specific subset at risk of rapid decline, comparison was made between the two extreme tertiles of decline. This has implications for both selection of patients for clinical trials of potential interventions and early introduction of effective therapies. Many factors were found to be associated with more-rapid decline in these analyses. The finding that FEV₁ decline was greater in patients with BDR and in males is in agreement with data from the US registry 1. Lower BMI has been linked with greater progression of disease and mortality in α₁-ATD 25 and usual COPD 26. In the logistic multivariate analysis, BDR, low BMI and exacerbation frequency were found to be independent predictors of decline in FEV₁.

Exacerbation frequency is known to relate to a speedier decline in lung function in α₁-ATD 27 and usual COPD 26. However, the relationship with better physical health status may at first seem counter-intuitive. The most severely restricted patients, however, are those with the lowest FEV₁, and the reduced FEV₁ decline in this group probably explains the association. Nevertheless, with all of these confounding factors, differences in any may explain the greater rate of progression seen in the untreated group in the Danish/German comparative study, as well as possibly the range of initial impairment 18.

The data differed for K_CO decline, which was greatest in patients with severe disease, as defined by baseline percentage predicted FEV₁. This would suggest that rapid decline in gas transfer is a late phenomenon in disease progression. Unlike FEV₁ decline, which largely reflects bronchial disease, K_CO decline reflects alveolar destruction alone. The analyses confirmed that only factors associated with disease severity (baseline FEV₁ and CT voxel indices) were significantly associated with K_CO decline. Recent studies have shown that emphysema distribution relates differentially to FEV₁ and K_CO 28, 29. Emphysema in α₁-ATD tends to dominate in the lower zones and spread to the upper zones as disease progresses. Lower-zone emphysema has been shown to affect FEV₁ more than K_CO, and upper-zone emphysema has the opposite effect. Therefore, it might be expected that K_CO decline would become more pronounced in more severe disease as emphysema progresses from the bases to involve the upper zones, as found here.

These data provide information central to the identification of fast decliners. For FEV₁, the decline is greatest in moderate-to-severe disease, and, in this group, BDR, low BMI and increased exacerbation frequency independently predict the rate. Thus, if FEV₁ decline is the primary outcome, patients with these characteristics would be best recruited for the testing of interventional strategies and instigation of effective preventative therapy.

Although K_CO is a more specific measure of emphysema, it progresses most rapidly in the most severe groups. At this point, physiological impairment is well established and it is unlikely that gas transfer would be an effective marker for identifying rapid decliners early enough in the disease to be effective or provide a robust group for long-term studies for determining the efficacy of new treatments.

In the current study, CT scans were not available for all patients over the 3 yrs. However, other studies have shown that this parameter alone shows progression independent of disease stage 30. This reinforces its use as a primary outcome measure, especially since it is the best indirect measure of pathological emphysema. If the efficacy of specific interventions is confirmed using CT scores as an outcome, it is also likely to become the measure of choice in determining rapid progression before physiological tests become adversely affected.

The current study had some limitations. The analysis was performed only on those patients for whom four consecutive annual pulmonary function test results were available, in order to obtain the most accurate regression data. Therefore, patients were excluded who did not have consecutive lung function tests performed because of missed appointments, withdrawal from the programme or death. Exclusion of this latter group, in particular, could modify the associations with declining lung function towards factors that influence survival (the healthy survivor effect). The results of the logistic regression analyses compared the fast and slow tertiles of lung function decline, with the aim of identifying differences between the two extreme groups, but, when a separate linear regression analysis was undertaken assessing lung function decline as a continuous variable, the results were slightly different. Most data were determined objectively, but the exacerbation data relied upon subjective recall, and, since the patients were visiting the centre from all parts of the country, independent verification of exacerbations and hospitalisations from health records was impossible. Nevertheless, when diary card identification and primary care records have been assessed, such recall has proven reasonably reliable 27, suggesting that the associations found here are likely to be valid.

In summary, it has been shown that, in a group of PiZ phenotype α₁-ATD patients, FEV₁ decline was greatest in those with moderately severe disease, and this showed associations with BDR, BMI, male sex and (in a multiple regression analysis) exacerbation rate. K_CO decline, conversely, was greatest in severe disease, and was only associated with other measures of disease severity (FEV₁ and CT densitometry). These findings have implications for the subgroups of patients to target in future clinical trials, and the stage at which effective therapy should be targeted.

Statement of interest

A statement of interest for R.A. Stockley can be found at www.erj.ersjournals.com/misc/statements.dtl