Smoke exposure as a determinant of autoantibody titre in α₁-antitrypsin deficiency and COPD

RESULTS

Anti-elastin antibody levels and associations

Anti-elastin antibodies were detected in all 257 AATD samples, all 22 healthy subjects and 113 out of 116 COPD subjects. The mean level was significantly higher in healthy subjects than those with COPD (p = 0.00001), and those with AATD (p = 0.008). Levels in AATD were higher than in usual COPD (p<0.00001), illustrated in figure 1. Levels in AATD were higher in those within this group who did not have COPD (p = 0.032; fig. 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1–

Mean anti-elastin antibody levels stratified by group. The chart shows the mean±se of circulating antibody concentration in each group. Healthy subjects exhibited significantly higher concentrations than those with chronic obstructive pulmonary disease (COPD) (p = 0.00001), and α₁-antitrypsin deficiency (AATD) (p = 0.008). AATD subjects also exhibited higher concentrations than usual COPD subjects (p<0.00001). The values shown are unadjusted for smoke exposure.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2–

Anti-elastin antibody concentrations are higher in those with normal lung function in α₁-antitrypsin deficiency (AATD). The chart shows the relationship between forced expiratory volume in 1 s (FEV₁) and anti-elastin levels in AATD which is significant; those with FEV₁ >80% predicted and no evidence of airflow obstruction had higher levels (p = 0.032).

Within AATD anti-elastin antibodies showed a negative correlation with pack-yrs smoked (r = -0.199, p = 0.001), although visually the relationship did not appear linear (fig. 3). We therefore stratified into categories by pack-yrs smoked, and showed a clear threshold at 10 pack-yrs, above which antibodies were lower (fig. 3, p = 0.004). Consistent with this, the mean level in never smokers was significantly higher than the smoke exposed (51.14±13.05 versus 27.04±5.21; p = 0.04). A concomitant positive correlation with FEV₁ was seen (r = 0.189, p = 0.005) which was lost after adjustment for pack-yrs smoked (r = 0.084, p = 0.251). A trend towards higher levels in those without emphysema (51.12±17.33 versus 29.30±6.06; p = 0.074) was also lost after similar adjustments (p = 0.622). The unadjusted correlations in AATD appeared to be driven by never smokers, and indeed in usual COPD, where all individuals had smoked, no relationship between antibody level and pack-yrs or FEV₁ was seen (both p>0.25). Emphysema analyses were not performed in usual COPD because of the low proportion of CT scans available.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3–

Anti-elastin antibody concentrations relate to smoke exposure in α₁-antitrypsin deficiency. a) The scatterplot shows the relationship between smoke exposure and anti-elastin antibody titre. Whilst it is statistically significant as a linear relationship (r = -0.199, p = 0.001), visually it appeared to be driven by low level smoking. b) Smoke exposure was divided into categories according to pack-yrs smoked. Anti-elastin antibodies then appeared to exhibit a threshold level at 10 pack-yrs. c) Comparisons of antibody levels were made using 10 pack-yrs to stratify the population. Antibody levels were higher in those who have smoked <10 pack-yrs (p = 0.004).

Anti-elastin antibody levels were then compared between groups after adjustment for pack-yrs smoked; the difference between AATD and usual COPD was lost (p = 0.258). This lack of relationship remained when including only AATD individuals who had spirometrically defined COPD (p = 0.131). All other between group comparisons of antibody levels became nonsignificant after adjustment for smoke exposure (all p>0.6). Never smokers with AATD and no evidence of COPD (n = 24) were then compared to healthy never smokers without AATD (n = 22) in order to ascertain if AATD itself were a significant determinant of antibody levels after control for smoke exposure and the presence of COPD; there was no difference in their anti-elastin antibody levels (p = 0.295).

Current versus past smoking did not show a significant difference in antibody levels in either group (both p>0.5). Years since last cigarette exposure did not relate to antibody levels, nor did age at commencement of smoking (both p>0.5). No sex or age relationships to antibody levels were seen in AATD or usual COPD, and sub-stratification by sex did not change the relationship with pack-yrs seen (r = -0.208, p = 0.014 in males and r = -0.255, p = 0.018 in females).

ACPA associations

Anti-CCP antibodies were higher in usual COPD than AATD (p = 0.002, fig. 4), and exceeded the conventional cut-off for a positive level in eight AATD subjects and six of those with usual COPD. Anti-MCV antibodies were positive in 34 AATD subjects and 20 with usual COPD; the level did not differ between groups (p = 0.876, fig. 4). There was no difference in either ACPA between AATD and controls (p = 0.412). In AATD, neither ACPA showed any relationship to FEV₁ (anti-CCP: p = 0.608 for high levels, p = 0.868 as a continuous variable; anti-MCV: p = 0.710 and p = 0.864, respectively). Equally, in usual COPD, neither ACPA was independently associated with FEV₁ (anti-CCP: p = 0.424 and 0.969; anti-MCV: p = 0.957 and 0.406). No age or sex associations were seen in either group (all p>0.5). In sex sub-stratified analyses the lack of relationship with FEV₁ persisted (all p>0.6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4–

Anti-citrullinated protein antibody (ACPA) concentrations stratified by group. The chart shows the mean anti-cyclic citrullinated peptide (anti-CCP) and anti-mutated citrullinated vimentin (MCV) concentration in the three groups studied. Anti-CCP titre was significantly higher in usual chronic obstructive pulmonary disease (COPD) than α₁-antitrypsin deficiency (AATD) (p = 0.002) but no other significant differences were seen. ▪: anti-CCP; ▒: anti-MCV.

Neither anti-CCP nor anti-MCV antibodies differed between those who had never smoked and those who had been smoke exposed in AATD (p = 0.903), a test which could not be performed in usual COPD since all subjects had smoked. No linear dose–response relationship (pack-yrs versus ACPA) existed in AATD (anti-CCP: r = 0.002, p = 0.979; anti-MCV: r = -0.069, p = 0.301) or usual COPD (anti-CCP: r = -0.158, p = 0.161; anti-MCV: r = -0.024, p = 0.864). However, when split into pack-yr categories in AATD, as before, a threshold was apparent at 10 pack-yrs, although less clear than that seen for anti-elastin antibodies (fig. 5). However, the difference in titre between those either side of the 10 pack-yr threshold was not significant (anti-CCP, p = 0.661; anti-MCV, p = 0.110), and the incidence of positive ACPA did not differ above and below this threshold (p = 0.761 and 0.503, respectively). Given that no relationship with smoke was observed further inter-group comparisons after adjustment for smoke exposure were not performed.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5–

Anti-citrullinated protein antibody (ACPA) relationships to smoke exposure in α₁-antitrypsin deficiency. a and b show ACPA levels stratified by smoke exposure categories. Although a threshold was suggested visually at 10 pack-yrs this was not significant (anti-cyclic citrullinated peptide (anti-CCP), p = 0.661 and anti-mutated citrullinated vimentin (anti-MCV), p = 0.110).

Since ACPA are diagnostic of rheumatoid arthritis, the incidence of this, and other autoimmune diseases, was determined in all those with a positive ACPA. Of the 36 AATD subjects with at least one positive ACPA, none had rheumatoid arthritis either at baseline or during follow-up (mean duration of follow-up since serum collection was 9 yrs). However, four had autoimmune thyroid disease at baseline, one had ulcerative colitis at baseline and a further individual developed ulcerative colitis during follow-up. Amongst the 22 ACPA positive individuals with COPD three had rheumatoid arthritis. To date, none has completed 1 yr of follow-up.

DISCUSSION

The data presented here show a clear relationship between circulating anti-elastin antibody levels and smoke exposure, and no relationship between these antibodies and clinical phenotype after smoking is taken into account. Conversely ACPA, which are theoretically generated after smoke exposure, did not show any significant relationship to smoke exposure in our cohorts, and less marked differences between groups.

One study has previously reported high anti-elastin antibody levels in COPD 9, which we were unable to confirm. This study concerned a group with emphysema, thus it could be argued that our analyses should have been restricted to those with CT-proven emphysema if we wished to confirm their findings. We did not exclude subjects who had not been scanned or who did not have emphysema, as our primary objective was to compare levels between groups whose elastin breakdown might be expected to differ 6, and against smoke exposure, not to reproduce the work of Lee et al. 9. However, secondary analyses comparing antibody titre between individuals with and without emphysema in AATD did not reveal a significant difference. It should be noted that differences in titre seen by Lee et al. 9 between groups were small, and that this observation was a small part of their study, which examined anti-elastin autoimmunity more specifically, using lung tissue to assess anti-elastin antibody complexes, and T-cell responses to elastin peptides. Our results, in direct contrast to theirs, showed anti-elastin antibodies were higher in those with better lung function and in healthy individuals. This implies that the presence and level of circulating anti-elastin antibody is not a key feature of COPD, an observation recently reported in another study 11, and may even fall as disease progresses. We hypothesise that this might be due to ongoing, or increasing, antibody consumption by binding in the lung. This is consistent with the increased antibodies seen in lung tissue by Lee et al. 9, although longitudinal studies involving both lung tissue and circulating samples would be required to definitively answer this question. Antibody levels do not rise with cessation of smoke exposure, consistent with the perpetuation of inflammation and lung damage in COPD seen after this time, with measureable disease progression in many subjects 20.

None of the studies of anti-elastin antibodies to date has taken effects of smoking into consideration in the analyses pertaining to antibody levels, despite the theory that anti-elastin antibody generation occurs secondary to smoke-induced elastin liberation in the lung 9. The data presented here show a clear difference in anti-elastin antibody levels between smokers and those who have never smoked in AATD, and a relationship between pack-yrs smoked and antibody levels. There was also a threshold level of smoke exposure apparent, a feature identified in another large cohort of AATD subjects recently, which contributed to their predictive model for FEV₁ 21. Smoke exposure was common in AATD, higher in usual COPD, and absent in controls, thus between-group comparisons of antibody levels, which appeared highly significant initially, required adjustment for this exposure. After such adjustment, not only was the relationship with FEV₁ lost within AATD, but also all inter-group comparisons ceased to be significant. This, together with the recent suggestion that smoking may be a risk factor for autoimmunity 12, is important to note in future studies of the prevalence of other autoantibodies in COPD and control populations. Subjects with AATD may be more sensitive to other environmental influences on lung function, as shown by relationships between outdoor air pollution and lung function in this group 22. We surmised that differences in elastin breakdown and subsequent anti-elastin generation might still occur between nonsmokers with and without AATD for this reason. This was not confirmed, perhaps because the effect sizes of agents other than cigarette smoke are small in both groups 22, 23.

Although differences in ACPA titre were seen between COPD, AATD and controls, the differences were less marked than with anti-elastin antibodies and less clearly related to smoke exposure. This is the first study, to our knowledge, of these antibodies in lung disease, and our findings will require further validation; nevertheless, it is interesting to consider their significance. The pathogenic role of ACPA in rheumatoid arthritis is thought to involve complement activation 24 and alteration of the T-cell repertoire in the joint 25, with reaction to self as a result. Our data would suggest that this may be organ specific, occurring just in the joint despite the fact that any protein containing citrulline could be a target for ACPA. Alternatively the effect on synovium could depend upon a further genetic or environmental influence, given that ACPA positive individuals here did not always exhibit rheumatoid arthritis (or indeed other autoimmune diseases). Shared epitope alleles, contained within class II of the human leukocyte antigen system (HLA-DRB1*04, *01 and *10), are important in determining risk of rheumatoid arthritis 26 and ACPA status 27. Recent data suggest that smoking is a key feature of determining ACPA-positive disease risk in the presence of these alleles 27, 28, consistent with this hypothesis. ACPA may occur up to 10 yrs before clinical features of rheumatoid arthritis appear 29, the titre relating inversely to the time until diagnosis 30, such that highly positive individuals are likely to be diagnosed soon. It will therefore be of importance to follow up positive individuals in the cohort, particularly since the incidence of positive anti-CCP antibodies exceeded that reported in healthy controls 29 five-fold in AATD and eight-fold in usual COPD. The majority of AATD subjects have, however, been followed up for >8 yrs since their baseline assessment and blood sampling, making it unlikely that many new cases will be found in this group. In the usual COPD cohort, high ACPA levels were more common than in AATD, and more common in both groups than in the general population. Intuitively, the role of smoking in determining pulmonary citrullination 14, and subsequent antibody formation, might have been thought responsible. Again, however, our data suggested that in AATD ACPA levels were lower in heavier smokers, albeit not statistically different. Since there was no significant dose–response effect adjustments of ACPA level for lifetime lack of smoke exposure, analyses could not be carried out to determine if differences between AATD, COPD and healthy individuals persisted after adjusting for this feature. Larger cohorts may be required for further clarification of this issue. Further study of citrullination in smokers is also warranted to clarify the mechanism of antibody generation and relationship to ACPA suggested here.

Autoimmunity is a current topic of debate in COPD pathogenesis, since some characteristic features of autoimmunity are present. First, a lymphocytic infiltrate in the lung occurs, with increasing pulmonary CD4+ and CD8+ T-cell numbers relating to worsening disease 31, 32. Autoantibodies to pulmonary epithelium have also been demonstrated 33. Furthermore, oligoclonal T-cell populations have been demonstrated in the lung, suggesting accumulation due to antigenic stimulation 34. Evidence from animal models also supports a role for autoimmunity in COPD. Specifically, the transfer of pathogenic CD4+ T-cells from rats with emphysema and anti-endothelial cell antibodies results in the development of emphysema in healthy animals 35, and mice injected with elastin peptides develop both anti-elastin antibodies and a bronchoalveolar lavage fluid profile similar to that seen in COPD 36, 37. The initial report of elevated anti-elastin antibodies by Lee et al. 9 has now been refuted by two small studies (in addition to the data presented here), the first in a mixed-disease cohort 10, and the second in pure-disease cohorts, but with small numbers in each 11. Indeed, in the latter study the authors noted that antibody levels were higher in healthy subjects, consistent with the data we report here. Autoimmune processes are generally thought to be more common in female subjects, hence we undertook some sex-specific analyses but again no differences were seen. Taken together, the results suggest that anti-elastin antibodies are unlikely to be a useful disease marker in COPD and, if autoimmunity exists directed against elastin, circulating anti-elastin antibody titres may be of less importance than local levels in lung tissue.

Our study is weakened by the absence of a healthy smoking group for comparison to the usual COPD group, although comparisons within AATD of healthy individuals and those with lung disease, controlled for smoking, act as a suitable surrogate for this. We also have not tested for IgM, IgA and other antibody subtypes for anti-elastin. This feature was investigated by a previous group, whose findings concurred with ours in terms of between-group comparisons, such that we felt it was unnecessary to repeat their experiments 11.

In summary, our data show a clear relationship between smoke exposure and anti-elastin antibody levels, and no difference between AATD, COPD and healthy individuals after adjustment for this feature. This suggests that circulating anti-elastin levels cannot be used as a disease marker and that future study of the autoimmunity hypothesis may need to be directed to antibody complexes in lung tissue rather than circulating antibodies.