This study has revealed new aspects of airway remodelling in the large airways in smokers with or without COPD. We have attempted to differentiate effects of smoking from the presence of established disease as defined by the GOLD initiative.
Our main results may be summarised as follows:
1. Rbm thickness was not different between groups.
2. The Rbm was fragmented and had markedly increased splitting in smokers and COPD (Figures and ), and especially in current smoking COPD.
3. The Rbm was hypervascular in smokers but not in ES-COPD.
4. The LP was hypovascular in smokers but not in ES-COPD.
5. Vessel staining for VEGF was increased in smokers and COPD, but especially in current smokers with COPD.
We did not find a significant difference between groups in Rbm thickness. Previous studies have been contradictory. One group found thicker Rbm in COPD compared with controls, [12
] with both COPD and control groups in this study being ex-smokers except for 3 COPD subjects who were never smokers. Others have not found this difference [13
]. We did find the Rbm thickness to be very variable in smokers and in COPD, and because of the fragmentation it was less easy to quantitate accurately.
The main changes in the Rbm in smokers and COPD were marked fragmentation and hyper-vascularisation which are novel findings and not previously published in the COPD literature. Rbm splitting, we propose, could be the result of either new layers being formed by the epithelium or more likely degradation of the Rbm by proteolytic enzymes. Rbm splitting has been reported previously in the glomerular basement membrane and endothelial basement membrane of tubules in kidney transplant rejection [15
]. Cornell et al.
proposed that splitting is the consequence of repeated episodes of injury with new basement membrane layers formed as part of a repair process.
Smoking induces repeated injury to the airway epithelium. As Cornell et al. proposed for kidney rejection, this may induce epithelial repair with formation of a new layer in the Rbm. This is compatible with the correlation of smoking history and length of splitting in our study and also explains the observed nonhomogeneity of the width of the Rbm in smokers.
However, the presence of splitting may well represent a change or degradation in Rbm matrix proteins. We believe the changes are unlikely to be an artifact of processing as this was the same for all groups, and in previous work in asthma, where those changes are not seen. Recently, differences in the components of collagen and other proteins in the Rbm in a study comparing asthma, COPD and controls have been described [12
]. Change in proteinase activity, which has been shown in COPD, [17
] may potentially explain this phenomenon. The correlation of splitting with historical amounts of smoking confirms that it is likely related to cumulative insult to the airway mucosa.
Although COPD subjects were significantly older than the control group, there was no correlation between age and the length of splitting in either COPD group, analysed separately or together, nor in the S-N group. Multivariable analysis showed that age is not a predictor of splitting (p = 0.4) but pack-year smoking history is (p < 0.02) (Table ). The presence of splitting in ES-COPD means we need a longitudinal study to assess whether the Rbm is truly unable to repair itself after smoking cessation, and to relate this to proteinase activity.
Correlation analysis for Rbm splitting for both COPD groups and smokers with normal lung function*†
Current smokers, irrespective of their pulmonary function, had increased vessel numbers in relation to the Rbm. This pathological change may be reversible with smoking cessation, as ES-COPD was not different from H-N but was different from S-N and both current smoker groups taken together. Again a longitudinal smoking cessation is now needed to confirm this and explore the mechanisms involved. We stained a number of matched slides with Factor VIII, which stains endothelium of blood vessels [18
], which confirmed that the structures stained by Collagen IV were indeed vessels.
We found more vessels stained for VEGF in the Rbm of current smokers and COPD, but VEGF staining was most marked in current smoking COPD subjects. VEGF is present in actively proliferating endothelium and is a marker of active angiogenesis [19
]. Therefore, we suggest that angiogenesis appeared to be equally active in COPD subjects who had quit smoking, suggesting that it is not reversible. Again, a properly designed longitudinal smoking cessation study will be necessary to confirm this.
In contrast, we found fewer blood vessels in the LP in current smokers, but not in ES-COPD. There have been few previous studies investigating vascular changes in large airway endobronchial biopsies in COPD, and none to our knowledge that have differentiated between the Rbm and LP. Calabrese et al.
in a study on bronchoscopically-obtained biopsies reported more vessels in the LP of smokers, and concluded that angiogenesis is a part of airway remodelling in smokers. They did not find any relationship between remodelling changes and lung function or clinical manifestations [20
]. Another recent Italian study found larger vascular area in BB from ex-smokers with moderate to severe COPD compared to control subjects. The number of vessels was not different between groups [21
A potential explanation for these previous findings, which appear to contrast with our own, would be the different selection criteria employed. For example, Calabrese et al. recruited smokers with normal lung function or COPD with clinical criteria of chronic bronchitis and they excluded subjects with emphysema. Chronic bronchitis, which at least anecdotally is not as prominent a feature of COPD in Australia as in Europe, was almost completely absent in our S-N subjects without being selected on this basis. We did not exclude subjects with emphysema in our COPD groups (Table ) and tried to include a "typical" local COPD population. Zanini et al. recruited moderate to severe COPD subjects that had quit for more than 10 years and they did not study current smoker COPD subjects. We studied COPD subjects with mild to moderate COPD. In our study current smokers with COPD had the most marked changes. Further, we separately counted vessels in the Rbm and LP. However, if the Rbm- associated vessels were added to vessels in the LP we still found fewer vessels overall in the mucosa in current smokers (data not shown).
There are other studies that examined airway vascularity in COPD but used subjects with peripheral lung cancer to study only smaller airways in lung resection specimens [22
]. Hashimoto et al.
did not find any differences in vessels in medium sized airways (internal diameter 2-5 mm) between COPD and nonsmoking controls, and Kuwano et al.
did not find a significant difference in vessel density in the mucosa of peripheral airways in subjects with mild COPD compared with controls without airway disease.
The reason for hypovascularity of the LP in smokers in our study could not be explained on the basis of the VEGF data produced. Pulmonary VEGF reduction in smokers has been reported [24
]. Hypovascularity of the LP in current smokers may be analogous to the observation that down-regulated VEGF within the lung parenchyma is associated with the development of emphysema [26
]. Our current study did not find reduced VEGF activity either in the Rbm or in the LP in current smoker groups, with the percentage of vessels in the LP staining for VEGF not being significantly different between groups. However, an explanation for this apparent paradox could be that VEGF is functionally unavailable for new vessel formation in the presence of cigarette smoke [28
]. The finding of normal vessels in ES-COPD supports this idea. More studies of the angiopoietic system in the airways in smokers are indicated. Deprivation of other angiogenic factors, such as angiopoietin-1 and/or down-regulation of endothelial VEGF receptors should also be considered and studied. Whatever the mechanism, hypovascularity of the LP is a smoking effect that may be reversible with quitting, but a specific longitudinal study is needed to confirm that.
The strong relationship between Rbm vessel-related VEGF and better FEV1% predicted in S-COPD group is interesting. There is some evidence that some aspects of remodelling may have a protective effect [2
], and potentially angiogenesis in the Rbm could increase airway stiffness and resist dynamic compression which is frequently a physiological problem in COPD. Similarly, the positive correlation between FVC% predicted and the number of vessels in the LP in the S-COPD group, probably reflecting less air trapping with more LP vessels, supports this idea. However, this is likely to be a reflection of the situation in the small airways which were not sampled in our study. These suggestions could be confirmed by direct assessment of airway distensibility in future studies [31
]. Thus, at this stage we can not confirm that the associations between vessel changes and lung function are causative and further investigation is required.
COPD groups in our study were significantly older than H-N and S-N. However, the age range in COPD was wide and detailed uni- and multi-variable analyses did not suggest that age was a factor influencing the main findings.