Phenotyping of chronic obstructive pulmonary disease using the modified Bhalla scoring system for high-resolution computed tomography

1Department of Chest Diseases; 2 Department of Radiology, Selcuk University Faculty of Medicine, Selcuklu, Konya, Turkey Correspondence: Dr Baykal Tulek, Selcuk Universitesi Tip Fakultesi, Gogus Hastaliklari AD, Selcuklu, Konya, Turkey. Telephone 90-532-7878327, fax 90-332-241-6065, e-mail baykaltulek@yahoo.com Chronic obstructive pulmonary disease (COPD) is a chronic inflammatory disease of the airways and lung parenchyma that leads to substantial morbidity and mortality, and affects approximately 10% of the population >40 years of age (1-3). COPD is highly heterogeneous with regard to clinical presentation, response to treatment, limitation of lung function and survival expectancy. Defining a condition so heterogeneous exclusively by a patient’s forced expiratory volume in 1 s (FEV1) may not always be adequate. The characterization of different phenotypes of COPD to better prognosticate mortality and exacerbation frequency has, therefore, increasingly attracted interest in recent years (4). High-resolution computed tomography (HRCT) is currently the preferred imaging method in the diagnosis of diseases such as bronchiectasis (B) and emphysema (E), which cause morphological changes in the lungs (5). Its usefulness in the differential diagnosis of obstructive lung diseases has been established. The incidental finding of B and E has increased with the use of HRCT. Studies involving COPD patients also show that morphological phenotypes characterized with the help of HRCT may play an important role in the prognosis and management of the disease (6,7). The Bhalla scoring system (8), widely used particularly in pediatric or adult cystic fibrosis patients, is a method demonstrated to correlate with clinical and physiological characteristics (9,10). To date, we are not aware of studies involving COPD patients correlating the Bhalla scoring system with clinical and physiological parameters of inflammation. The objective of the present study was to characterize the original article

All patients were in a stable phase of disease at the time of HRCT imaging and none had experienced an exacerbation within the previous six weeks.Medical records of all patients were reviewed and detailed histories of the patients were obtained.Patients with a previous diagnosis of B or those with clinical evidence of B were not eligible for the study.In addition, patients with sinobronchitis or diffuse panbronchiolitis, asthma, interstitial or inflammatory lung disease, sequelae of tuberculosis, history of cerebrovascular incidents, heart disease or severe esophageal reflux were also excluded from the study.The patients' smoking history, start date of COPD-related complaints, daily expectoration volume and the number of acute attacks in the year preceding HRCT imaging were also recorded from the patients' charts.Results of respiratory function tests performed in the stable phase, C-reactive protein and total albumin levels, and erythrocyte sedimentation rate (ESR) within the period from one before to one month after HRCT imaging were also recorded.
HRCT imaging without contrast medium was performed using a fourdetector computed tomography (CT) scanner (Aquilon 4, Toshiba, Japan).Sequential scanning from the pulmonary apex to the diaphragm at 10 mm intervals with 1 mm collimation was performed at maximal inspiration on supine patients.Other details of image acquisition were as follows: voltage 120 kVp, effective tube current 150 mAs, detector collimation 1 mm × 4, acquisition time 1 ms; a bone algorithm and a 512 × 512 matrix were used for image reconstruction.The lung windows had a width of 1600 Houndsfield units and a level of -600.
All of the examination results were evaluated retrospectively according to the modified Bhalla scoring system by two independent radiologists (ASK and SD) who were unaware of the patients clinical condition.The scores attributed by ASK and SD were compared to calculate interobserver agreement.The scoring system proposed by Bhalla et al (8) for CT scans evaluates the degree and extent of B, mucus plug formation and peribronchial thickening (PBT), and the number of bronchial divisions that are involved.Abscesses, sacculations, bullae, E, areas of collapse and consolidation are also recorded.This scoring system was modified by the addition of an evaluation of mosaic perfusion, and a total score was obtained by adding the scores for each class of abnormality; total scores could range from 0 to 27 (Table 1).Radiological findings were reported according to established, published guidelines.Definitions for the findings were as follows: B -internal diameter of a bronchus larger than that of its companion pulmonary artery with the bronchial lumen failing to taper over ≥2 cm and visualization of bronchi <1 cm from the costal pleura (Figures 1 and 2); PBT -thickness of the bronchial wall >1 mm (Figures 1 and 2); mucus plugging -image of plugs in large bronchi or peripheral tree-in-bud sign; sacculation -cystic appearance of bronchi; bulla -round air space of >1 cm diameter with a thin wall; E -decreased attenuation, disorganization of the vascular structure, absence of recognizable walls (Figure 2); consolidation -intensified opacification of the lung hiding the parenchymal pattern; and mosaic perfusion -hyperlucent, hypovascular zone.Patients with morphological changes were classified into one of three groups according to their HRCT phenotype(s): E only; E+B/PBT; and B/PBT only.In defining these phenotypes, a cut-off Bhalla score ≥1 for any finding was used for each group.

Statistics
For continuous variables, the descriptive summary parameters were the mean and SD; categorical variables were tabulated as absolute counts and percentages.Variables were tested for normality of distribution using the Kolmogorov-Smirnov test.Clinical findings and spirometry and blood values of the phenotype groups were tested for differences using the Kruskal-Wallis test.Paired comparisons using the Mann-Whitney U test were performed in the presence of a significant result of the overall comparisons.A Bonferroni correction was applied in the case of multiple comparisons.Spearman's rank correlation (r) was used to investigate correlations between HRCT scores and other variables.Interobserver agreement for scoring was tested and, for each observation, the alpha coefficient for agreement was calculated and expressed with its 95% confidence limits.The level of statistical significance was set at an alpha value of 5% for type I error.

RESulTS
The mean (± SD) age of the 80 COPD patients was 68±8 years; 76 (96%) of these patients were male.All stages of COPD were represented, although a majority exhibited moderate (46.3%) and severe (30%) disease.Spirometry studies were available for all patients, while CRP level had been obtained in 74, ESR in 71 and total albumin level in 61.All subjects were in the 'current smoker' category (ie, currently smoking or stopped smoking less than one year previously).Patient and disease characteristics are summarized in Table 2.
HRCT evaluations diagnosed E in 47 (58.8%) patients, B in 27 (33%) and PBT in 25 (31.3%).Patients were classified into one of three groups according to HRCT phenotype(s) (ie, E only, E+B/PBT and B/PBT only [the overlap between B and PBT was substantial; hence, the evaluation of B and PBT as a single group]) (Figure 3).
When grouping the patients according to the presence or absence of HRCT findings, no significant differences were found with regard to age, sex, smoking status, COPD duration and albumin level; spirometric measurements were worse in those with HRCT findings, who also had significantly higher CRP and ESR values (Table 3).
On comparing the three phenotypes described as E only, E+B/PBT and B/PBT only, the differences in FVC, FEV 1 , FEV 1 /FVC ratio, CRP level and the number of exacerbations were found to be significant (Table 4).Pairwise comparison using the Mann-Whitney U test between the E only and E+B/PBT groups showed significantly lower FVC, FEV 1 and FEV 1 /FVC ratio, and higher CRP level and number of exacerbations compared with the B/PBT only group.Comparison between the E+B/PBT and the B/PBT patients did not demonstrate a significant difference, while CRP values and number of exacerbations were only apparently higher in the E+B/PBT group.Bonferroni correction was used to evaluate the significance levels of type-I error for pairwise comparisons; values <0.0166 were considered to be statistically significant (Table 5).

dISCuSSIOn
In the present study, which included 80 patients at all stages of COPD from mild to very severe, 26 had no morphological abnormality while 54 had at least one.Patients with no morphological abnormality had mild to moderate COPD according to GOLD stage, and were similar in age, smoking history and duration of COPD symptoms to patients who had HRCT findings.In the latter group, spirometric measurements were worse, and number of exacerbations, CRP and ESR values were higher than those in the first group.These results showed that HRCT examinations may provide direct measurements of pulmonary pathologies and have good correlation with some commonly used clinical measurements in COPD patients, similar to that reported in a limited number of studies in the literature (12).Providing direct measurements of pulmonary pathologies via HRCT may contribute to improved disease management in COPD patients.We divided patients with HRCT findings into one of three groups: E only, E+B/PBT and B/PBT only; comparison of the latter two groups with the former showed significantly worse or lower values for FEV 1 and the FEV 1 /FVC ratio.This same comparison also revealed significantly higher CRP levels and one-year exacerbation counts in the E+B/PBT and B/PBT groups compared with the E only group.The number of exacerbations during the past year and the CRP levels of patients with B/PBT in their HRCT were approximately two-to threefold higher than those with E only.A comparison of B/PBT with or without E did not reveal any significant differences.This result led us to believe that the presence of B/PBT in COPD patients was associated with a more severe clinical course.
Bronchiectasis and COPD have many common clinical and pathological characteristics (13)(14)(15)(16).Publications reporting that an important proportion of COPD patients also have bronchiectasis have been increasing in recent years.Patel et al (17) determined that 50% of the patients in their study had bronchiectasis.Bronchial inflammation and the chronic presence of potentially pathogenic microorganisms (PPM) are more frequent and more extensive in patients with COPD and bronchiectasis, and their infectious exacerbations last longer.Another recent study (18) found a 57.6% frequency of bronchiectasis in COPD patients.Bronchiectasis in these patients was independently associated with severe airflow obstruction, PPM cultured in the sputum and ≥1 exacerbations with hospitalization in the past year.In our study, cases with concomitant bronchiectasis comprised 33.8% of the patient population.We believe that the higher proportion of bronchiectasis in these two studies (17,18) may be due to the fact that those patient populations consisted of cases with moderate to severe COPD.In our study, however, 17.5% of the patients had mild disease and, when these were excluded, the bronchiectasis rate rose to 40%.The scores for the extent and severity of bronchiectasis in our patients were significantly correlated with FEV 1 , CRP and ESR values, as well as the number of exacerbations.Current guidelines continue to consider spirometric measurements, chiefly FEV 1 , to be most important in diagnosing COPD and determining its severity.While they are generally useful, these functional measurements are not, by themselves, sufficient to characterize the heterogeneous pathological processes in COPD.Quantitative evaluations of CT imaging results have been shown to be associated with airway function measurements (19,20), main COPD outcome parameters (20)(21)(22) and systemic inflammatory mediator levels (23).The use of CT imaging as an index for respiratory tract disease remains to be validated (24,25).Our study confirmed the presence of significant correlations calculated using Spearman's rank correlation between HRCT scores and the values of FEV 1 and CRP and the number of exacerbations.It was interesting to note that while the correlations between HRCT scores or spirometric measurements and the exacerbation count were comparable, the correlation between HRCT score and CRP levels was stronger than between HRCT score and spirometry (r=0.785 and r=0.596, respectively).
CRP is a useful marker of systemic inflammation.It is easily measured and widely used in clinical practice.CRP levels have been shown to be elevated in patients with stable COPD (26).Some studies have reported an inverse relationship between CRP levels and FEV 1 ; several reports also suggest that CRP may be a strong long-term outcome predictor for COPD/airway obstruction (27)(28)(29).Our study found a significant correlation between HRCT score and CRP levels; CRP levels were also higher in B/PBT patients compared with the E only group.A direct correlation between bronchial microbial colonization and CRP levels has been reported (30).The high CRP level in our patients with B is likely related to increased bronchial microbial colonization.However, the absence of PPM culture was a limitation of our study.
In a study investigating the relationship among emphysema frequency, severity of COPD and CRP levels in 651 male patients, Omori et al (31) showed that the severity of COPD was variable across patients with comparable degrees of emphysema.CRP levels were not higher in patients with mild or moderate emphysema compared with those with no emphysema in this study.Although substantially elevated CRP levels were detected in patients with severe emphysema, this group only comprised two patients.CRP values were higher in patients with emphysema in our study compared with those without this condition (5.4 mg/L versus 3.5 mg/L); E scores were also significantly correlated with both FEV 1 and CRP measurements.In our opinion, the differences between our study and the study by Omori et al (31) were due to differences in patient populations.While the COPD severity grade of our patients ranged from mild to very severe, participants in the study by Omori et al (31) were relatively healthier.
Morphological changes observed using HRCT may contribute substantially to the management of COPD.Kitaguchi et al (6) reported significant linkage between the presence of bronchial wall thickening, sputum eosinophilia and response to corticosteroid therapy.Martinez-Garcia et al (18) indicated that bronchiectasis is related to pathological bacterial colonization, mainly by Pseudomonas species, and that this may be a predictor of exacerbations.This prompts us to recommend that studies evaluating the efficacy of bronchiectasis treatments, such as inhaled antibiotics, macrolides and airway clearance modalities, should also be performed in patients with overlapping COPD and B (32).Consistent with our study, it has been shown that CT-detected increases in E and airway wall thickness are associated with exacerbation frequency, suggesting that HRCT may help in the identification of patient subgroups with exacerbations for targeted research or phenotype-specific therapy (33).Because CT is widely used in both daily practice and lung cancer screening trials, it may become a useful tool for the detection of new COPD, and also for identification of subclinical COPD cases with normal pulmonary function tests or in which pulmonary function testing is not routinely performed (12).Additionally, HRCT may also help in the selection of patients for lung volume reduction applications and evaluation of response to alpha-1 antitrypsin augmentation therapy (34)(35)(36).
phenotypes of COPD according to HRCT findings, and to assess the correlation of HRCT scores with clinical and physiological indicators of systemic inflammation.

Figure 2 )Figure 3 )
Figure 2) High-resolution computed tomography image of the lower lobes demonstrating bilateral cystic bronchiectasis with bronchial wall thickening (arrows).Predominantly left-sided subpleural distribution of emphysematous spaces are noted (*)

Figure 4 )
Figure 4) The median total high-resolution computed tomography (HRCT) scores according to the Global initiative for Obstructive Lung Disease stages.COPD Chronic obstructive pulmonary diease

Table 1 Modified bhalla scoring system HRCT findings
*Modification to the Bhalla scoring system.BPS Bronchopulmonary segments; HRCT High-resolution computed tomography

Table 3 Clinical, spirometric and laboratory comparisons in patients with demonstrable absence or presence of high- resolution computed tomography scan evidence of bronchiectasis (b), peribronchial thickening (PbT) or emphysema (e) Characteristic None (n=26) b/PbT/e (n=54) P
Data presented as mean ± SD unless otherwise indicated.COPD Chronic obstructive pulmonary disease; ESR Erythrocyte sedimentation rate; FEV 1 Forced expiratory volume in 1 s; FVC Forced vital capacity

Table 5 Pairwise comparisons of the patient groups regarding FVC, FeV 1 , FeV 1 /FVC, CRP, eSR and number of exacerbations
Bonferroni correction was used to evaluate the significance levels of type 1 error for pairwise comparisons and values <0.0166 were considered to be statistically significant.B Bronchiectasis; CRP C-reactive protein; E Emphysema; ESR Erythrocyte sedimentation rate; FEV 1 Forced expiratory volume in 1 s; FVC Forced vital capacity; PBT Peribronchial thickening *

Table 6 Univariate (Spearman's rho) correlations between structural features on HRCT and FVC, FeV 1 , FeV 1 /FVC ratio, CRP level, eSR and number of exacerbations in previous year
CRP C-reactive protein; ESR Erythrocyte sedimentation rate; FEV 1 Forced expiratory volume in 1 s; FVC Forced vital capacity; HRCT High-resolution computed tomography