Correlation between high-resolution computed tomography features and patients' characteristics in chronic obstructive pulmonary disease

BACKGROUND: During the last few decades, high-resolution computed tomography (HRCT) has come up as a new diagnostic modality to diagnose emphysematous and chronic bronchitis components of chronic obstructive pulmonary disease (COPD). The present study was undertaken to evaluate for various quantitative and qualitative HRCT features in patients with COPD, and to detect patients' characteristics that correlate with these HRCT features. MATERIALS AND METHODS: Forty male patients with COPD attending the COPD clinic at a tertiary referral hospital and postgraduate medical institute were included in the study. Various HRCT features, including tracheal index, thoracic cage ratio, sterno-aortic distance, thoracic cross-sectional area, vascular attenuation, vascular distortion, mosaic attenuation pattern, and directly visible small airways, were evaluated and correlated with patients' characteristics, including age, duration of illness, quantum of smoking, dyspnea score, quality-of-life index, and various spirometric indices. RESULTS: We found significant correlations of various quantitative and qualitative HRCT features with age, duration of illness, quantum of smoking, quality-of-life index, and the spirometric indices showing the extent of airways obstruction. CONCLUSIONS: Various quantitative and qualitative HRCT features were found to correlate with patients' characteristics, spirometric indices, and health-related quality-of-life score, suggesting that HRCT is useful not only in detecting emphysema and its various subtypes but also in predicting the extent and severity of COPD.

C hronic obstructive pulmonary disease (COPD) is a leading cause of morbidity and mortality worldwide.The economic and social burden due to it is both substantial and projected to increase in the coming decades due to continued exposure to COPD risk factors and the changing age profile of the world's population.COPD mortality trends generally track several decades behind smoking trends.In the US in 2000, more women than men died of COPD or its related complications. [1]PD consists of a heterogeneous group of disorders, including emphysema, peripheral airways disease, and chronic bronchitis.The diagnosis of COPD is conventionally based upon spirometry; [1] But during the early stage of the disease, conventional spirometry may reveal no abnormality as the earliest changes in COPD affect the alveolar walls and small airways. [2]Also, defi nition of emphysema is a pathologic one, and it requires histopathological examination to authenticate the disease, which is not usually carried out and not often consented to by the patient.
During the last few decades, with the advent of HRCT as a new diagnostic modality, there has been an increased interest of diagnosing emphysematous and chronic bronchitis components of COPD using HRCT.Few isolated studies have suggested the role of HRCT in diagnosing different entities grouped under COPD.In the present study, apart from evaluating for the HRCT features of COPD, we have tried to detect patients' characteristics, including age, duration of illness, quantum of smoking, dyspnea score, quality-of-life index, and various spirometric indices that are correlated with the various established quantitative and qualitative HRCT features of COPD.

Materials and Methods
The present study was conducted in the Departments of Respiratory  Obstructive Lung Disease (GOLD) 2004 update'. [3]All included patients had a smoking history of more than 20 pack years and had airflow limitation that was not fully reversible (postbronchodilator FEV 1 less than 80% of the predicted value in combination with FEV 1 /FVC not more than 70%).
They had an increase in FEV 1 less than 200 mL or less than 12% of baseline value 20 minutes after 2 puffs of inhaled salbutamol that was given by a metered dose inhaler through a spacer.The patients were not included in the study if they had any evidence of coexisting bronchiectasis, cystic fi brosis, tuberculosis, bronchial asthma, interstitial lung disease, bronchogenic carcinoma, previous lung surgery, or coronary artery disease.They were excluded if they had any concomitant disorders like diabetes mellitus, chronic alcoholism, uremia, or sarcoidosis.
The study protocol required each patient to undergo routine clinical and laboratory evaluation, including complete blood examination, urine examination, chest radiograph, electrocardiography, echocardiography, and spirometry.Degree of breathlessness was measured by using Medical Research Council (MRC) dyspnea scale. [4]Smoking pack years were based on the mode of smoking (bidi, cigarette, or hookka), daily consumption, and total years smoked.One pack year was 20 cigarettes smoked/day for 1 year. [5]For bidi, cigarette equivalents were calculated by multiplying with 0.5 to the quantum of bidis; [6] and for hookka, 12.5 g of loose tobacco was equivalent to one packet of 20 cigarettes. [7]e spirometry was carried out on Transfer Test Model 'C', P. K. Morgan, Chatham, Kent, UK.All patients were required to withhold inhaled short-acting bronchodilators 6 hours before test, long-acting β-2 agonists 12 hours before test, and sustained-release theophyline 24 hours before test.Spirometric indices were calculated using the best out of 3 technically satisfactory performances as per recommendations of American Thoracic Society. [8]The following parameters were recorded: PEFR (peak expiratory fl ow rate) in liters/ min, FEV 1 (forced expiratory volume in the fi rst second) in liters, FVC (forced vital capacity) in liters, FEV 1 /FVC% (forced expiratory volume in fi rst second/forced vital capacity %), SVC (slow vital capacity) in liters, and FEV 1 /SVC% (forced expiratory volume in the fi rst second/slow vital capacity %).A questionnaire, Airways Questionnaire 30, validated by previous studies, [9,10]  of view large enough to encompass the patient.Images were obtained at maximum inspiration by using 1 mm collimation at 120 kV (kilo volt) (p) and 90 mA with a 0.75-second acquisition time.Scans were taken at 10-mm intervals with patient in the supine position.Images were reconstructed using a high-spatial frequency algorithm and a 512 × 512 matrix.No contrast was used.Following features were evaluated using HRCT: 1. Tracheal index: A ratio of transverse to anteroposterior diameter at a plane 1 cm above aortic arch.Saber-sheath trachea was described as when the tracheal index was less than 2/3.

Results
The present study included 40 male patients with COPD.The patients' characteristics were as shown in Table 1.The mean age of the COPD patients was 58.55 ± 7.99 years, and the mean duration of illness was 12.63 ± 6.76 years.The smoking pack years ranged from 20 to 74 years.Their mean dyspnea score was 3.65 ± 0.92.The detailed spirometric values are shown in Table 1.The FEV1/FVC ratio varied from 29% to 68%, and the oxygen saturation was 88.5% to 98.6%.Thirty-four of the 40 patients had airway questionnaire 30 (AQ30) score between 15 and 25, signifying that these patients were having a quality of life that ranged from poor to average.The patients' characteristics suggest that the patients represented a fairly large spectrum of COPD illness when seen from the severity perspective.
Table 1 also shows the group data of the quantitative HRCT features.The mean tracheal index was 0.67.The mean thoracic cage ratio at carina and 5 cm below the carina was 69.69% and 73.27% respectively.The sterno-aortic distance at carinal level was 3.01 ± 0.86 cm, and the thoracic cross-sectional area 1 cm below the top of aortic arch was 87.57± 10.47 cm 2 /m 2 .
The HRCT features in individual patients were as shown in Table 2. Saber-sheath trachea described as when tracheal index was less than 0.67 was found in 14 patients.Abnormal thoracic cage ratio was more common at 5 cm below carina (11 patients) than at carina (5 patients).Sterno-aortic distance over 4 cm was found in 5 patients.Large ratio of thoracic crosssectional area/ht 2 was present in 28 patients.Among subjective fi ndings, presence of directly visible small airways was the most common fi nding, followed by vascular attenuation and mosaic attenuation pattern [Figure 1].Vascular distortion was present only in 8 patients.Vascular distortion and mosaic attenuation pattern were coexisting with vascular attenuation.Among the three types of emphysema, centriacinar emphysema was observed most frequently (16 patients, 40%), followed by distal acinar (13 patients, 32.5%) and panacinar emphysema (11 patients, 27.5%) [Figure 2].Twenty-fi ve (62.5%) patients had at least one type of emphysema.
Various HRCT features suggestive of hyperinflation of thoracic cage were evaluated for their possible correlation with patients' characteristics, and the results in detail are displayed in Table 3. Tracheal index was observed to have a highly signifi cant negative correlation with age, duration of illness, pack years, dyspnea scale, and AQ30 score.It had a highly positive correlation with FEV 1 , FEV 1 /FVC ratio, and FEV 1 /SVC ratio; the correlation with PEFR was also signifi cant.Thoracic cage ratio at carina and the thoracic cage ratio at 5 cm below carina had a highly positive correlation with age, duration of illness, pack years, dyspnea scale, and AQ30 score; they had highly negative correlation with FEV 1 , FEV 1 /FVC ratio, and FEV 1 /SVC ratio.Similarly, sterno-aortic distance and thoracic cross-sectional area had a highly positive correlation with age, duration of illness, pack years, dyspnea scale, and AQ30 score; they had highly negative correlation with FEV 1 , FEV 1 /FVC ratio, and FEV 1 /SVC ratio.

Discussion
Chronic obstructive pulmonary disease (COPD) is a preventable and treatable disease, with some signifi cant extrapulmonary effects that may contribute to the severity in individual patients.Its pulmonary component is characterized by airfl ow limitation that is not fully reversible.It is a disease state that has seen signifi cant changes in defi ning and excluding criteria over the past 50 years [1] ; and as a result, some of the earlier studies are not expected to have inclusion criteria of COPD patients as per recent guidelines.There have been lots of variations in the patients' characteristics included in various previous studies.Many studies had selected patients with emphysema only. [11,12]Some studies [13][14][15][16][17] did not have reversibility criteria in accordance with Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines.In the present study, all subjects included were smokers and had irreversible or partially reversible airfl ow limitation, as required for the diagnosis of COPD.The patients having poorly reversible airfl ow limitation observed in bronchiectasis, cystic fi brosis, tuberculosis, or asthma were not included.This might have led to some variation in the outcome of our study, but the present study has clinical relevance as the diagnosis of COPD was based on the clinical and the spirometric criteria that are invariably used in clinical practice.
We studied the HRCT features which comprised of quantitative and qualitative parameters.Various studies had evaluated quantitative parameters, including tracheal index, sterno-aortic distance, thoracic cage ratio, and thoracic cross-sectional area, that are suggestive of hyperinfl ated lungs and rib cage seen in COPD.We observed that all these quantitative parameters were correlated signifi cantly with age, duration of illness, dyspnea score, smoking pack years, PEFR, FEV 1 , FEV 1 / FVC ratio, FEV 1 /SVC ratio, and health-related quality-oflife (AQ30) score.No signifi cant correlation was noted with oxygen saturation, which probably was due to the fact that the patients included were having stable course of COPD; 34/40 patients had normal oxygen saturation.Trigaux and co-workers [18] found signifi cant correlation of tracheal index with FEV 1 , FRC, and sterno-spine distance.In a different study, saber-sheath trachea was established as a specifi c radiographic diagnostic parameter in diagnosis of COPD (specifi city, 92.9%), although the sensitivity (39.1%) was low. [14]We found saber-sheath trachea in 14/40 patients (sensitivity 35%).In a study by Arakawa, tracheal index, thoracic cage ratio, and sterno-aortic distance were signifi cantly correlated with FEV 1 /FVC ratio in 74 patients (44 with normal lung function, 30 with COPD). [19]One study [20] showed a good correlation between thoracic cross-sectional area and total lung capacity, functional residual capacity, residual volume, as well as Fletcher's fi ve-degree scale of dyspnea.
In our study, presence of vascular attenuation correlated signifi cantly with duration of illness, smoking pack years, dyspnea score, FEV 1 , FEV 1 /FVC ratio, and FEV 1 /SVC ratio.In a study by Kuwano and co-workers, CT score was assessed by examining low attenuation areas, vascular attenuation, and distortion; [21] it correlated signifi cantly with DL CO , FEV 1 / VC ratio, maximum mid-expiratory fl ow rate, and residual volume/total lung capacity.In another study by MacNee et al., vascular attenuation and distortion on CT scans of 32 patients correlated significantly with DL CO , FEV 1 % predicted, and FEV 1 /FVC%. [22] the present study, no signifi cant correlation was observed between vascular distortion and various spirometric indices; however, vascular distortion was absent in all 20 patients who had FEV 1 /FVC > 50%.Absence of signifi cant correlation was possibly because we had not carried out grading of vascular distortion.Other studies have found good correlation of grading of vascular distortion with pulmonary function tests. [21,22]It appears from fi ndings of our study and other studies that vascular distortion is present in very advanced cases of emphysema and in those with severely deranged pulmonary functions.
As per study protocol, we studied for mosaic attenuation pattern at end-inspiration.Mosaic attenuation is more pronounced on scans obtained at end-exhalation compared to that found at end-inspiration as in the more conventional end-inspiration technique. [16] attenuation pattern in association with other HRCT features is observed to be helpful in distinguishing different disease entities grouped under COPD. [17] our study, directly visible small airways were observed as the most prevalent HRCT feature (36/40 patients).Hogg et al. showed that despite a decreased radius of small airways, airfl ow resistance is not signifi cantly increased because of the large number of these peripheral bronchioles. [23]Hence noteworthy disease must be present at the level of peripheral airways, to be detected on conventional pulmonary function tests. [24]Here, HRCT can play an infallible role in a diagnosis of involvement of small airways.
Low attenuation areas due to emphysema were seen by us in 25/40 (62.5%) patients.The presence of this feature correlated signifi cantly with age of patient, duration of illness, smoking pack years, dyspnea score, and spirometric indices.The emphysematous features had upper lobe predominance, explained by the fact that all patients were moderate-toheavy smokers.In a study by Miniati et al,, the HRCT score based on the extent of emphysema correlated with FEV 1 , FVC, forced mid-expiratory fl ows, maximum fl ows at 50% and 75% of FVC, residual volume, total lung capacity, and diffusion constant. [25]Klein et al. [12] observed that centriacinar emphysema was the dominant parenchymal abnormality and was strongly correlated with DL CO .Similarly, emphysema on CT was observed to be correlated with FEV 1 and K CO in 20 patients. [26] could fi nd only one previous study which explored correlation between HRCT features of emphysema and HRQL. [27]This study used St. George's Respiratory Questionnaire scores and the SF-36 to measure HRQL in patients with alpha 1 -antitrypsin defi ciency and found clear relationship between emphysema measured by lung density and HRQL.Our study measured HRQL by using AQ30 in COPD patients and found signifi cant correlation with all measured parameters of lung hyperinfl ation.In comparison to time-consuming and complicated instruments to measure HRQL, AQ30 included in the present study took signifi cantly less time to complete and score.
To conclude, in the present study, we were able to detect signifi cant correlations between various quantitative HRCT features and patients' characteristics, spirometric indices, as well as health-related quality-of-life score.Moreover, significant correlations were observed using qualitative HRCT features.This suggests that HRCT is helpful not only in detecting emphysema and its various subtypes but also in predicting the extent and severity of the COPD.

Figure 1 :
Figure 1: HRCT features in COPD: (A) Vascular attenuation, (B) vascular distortion and increased branching angle, (C) directly visible small airways as ring shadows (white arrowheads), and (D) mosaic attenuation pattern Medicine and Radiodiagnosis at the Postgraduate Institute of Medical Sciences [PGIMS], Rohtak, India.The study was approved by the Institutional Board of Studies and by the Ethical Committee.The patients were enrolled from the COPD Clinic run at the Department of Respiratory Medicine, PGIMS, Rohtak.Eligible male patients with COPD, aged 40 years or more, and who provided explicit written consent, attending the clinic during the period from February 2006 to December 2006 were included.The diagnosis of COPD was based on the criteria defi ned by 'Global Initiative for Chronic

Box 1: Health-related quality of life questionnaire (AQ 30)
Does your chest trouble affect you at any time other than when you are having an attack?(25) Do you do all the things you want to regardless of the effects on your chest trouble?(26) Because of your chest trouble, do you often feel helpless?(27) Do you work badly when your chest trouble is bad? (28) Because of your chest trouble, do you have diffi culty doing housework?(29) Is your sex life affected by your chest trouble?(30) Do you suffer from discomfort when you cough?Gupta et al.: HRCT and COPD patients' characteristics 2. Thoracic cage ratio: A ratio of anteroposterior to transverse diameter.It was evaluated at two planes: [a] tracheal carina, [b] 5 cm below carina.3. Sterno-aortic distance: Distance from posterior surface of sternum to anterior margin of aorta at carinal level.4. Thoracic cross-sectional area: Thoracic cross-sectional area (TCSA) was measured on HRCT images made 1 cm below the top of aortic arch.The ratio of TCSA to the square of height (TCSA/height 2 ) was calculated for each patient.

Table 4 .
The vascular attenuation had a signifi cant correlation with age of the patients, duration of illness, dyspnea scale, smoking pack years, FEV 1 , FEV 1 /FVC ratio, and FEV 1 /SVC ratio.Vascular distortion had no signifi cant correlation with any parameter.The mosaic attenuation pattern correlated signifi cantly with duration of illness, dyspnea scale, smoking pack years, FEV 1 , FEV 1 /FVC ratio, and FEV 1 /SVC ratio.Correlation of presence of directly visible small airways was statistically signifi cant only with dyspnea scale.Centriacinar emphysema had statistically signifi cant correlation with PEFR, FEV 1 , FVC, FEV 1 /FVC ratio, and FEV 1 /SVC ratio.Correlation of presence of any type of emphysema with age, duration, dyspnea scale, pack years, PEFR, FEV 1 , FVC, FEV 1 / FVC ratio, and FEV 1 /SVC ratio was signifi cant.

Table 4 : Statistical signifi cance [P value] of correlation between qualitative HRCT features and study variables (signifi cant correlations are shown in bold)
This is the probable reason for the detection of mosaic attenuation pattern in only 16 (40%) out of 40 patients in the present study.All these 16 patients had associated features of vascular attenuation and directly visible small airways.Presence of this feature correlated signifi cantly with duration of illness, smoking pack years, dyspnea score, PEFR, FEV 1 , FEV 1 /FVC ratio, and FEV 1 /SVC ratio.Mosaic