The relationship between exposure to tobacco smoke and chronic obstructive pulmonary disease (COPD) is well described.1 Two manifestations of COPD include emphysematous destruction of the lung parenchyma and elevated measures of total lung capacity.2 However, there is increasing awareness that smoking may also result in areas of increased lung density — termed interstitial lung abnormalities — on high-resolution computed tomography (HRCT).3,4 The extent to which interstitial lung abnormalities may be associated with a lesser amount of emphysema and lower measures of total lung capacity than anticipated on the basis of known smoking exposure is unclear. We determined the relationship between radiographic interstitial lung abnormalities and HRCT measures of total lung capacity and emphysema in a cohort of non-Hispanic white and black smokers who had been recruited for the COPDGene Study on the basis of a self-reported history of more than 10 pack-years of smoking. Since we oversampled participants on the basis of COPD status,5 we evaluated whether the associations between interstitial lung abnormalities and both total lung capacity and emphysema were modified by COPD status.
Methods
Study Design
From November 2007 through April 2010, a total of 2508 smokers (1867 [74%] non-Hispanic white and 641 [26%] black) between the ages of 45 and 80 years with a history of at least 10 pack-years were enrolled at 21 clinical centers under the auspices of the COPDGene Study, which is ongoing and has been described previously.4,5 Participants with a history of any active lung disease other than asthma, emphysema, or COPD were excluded from the study. Spirometry was performed in accordance with the recommendations of the American Thoracic Society and the European Respiratory Society.6 HRCT was performed at full inspiration and at relaxed exhalation. Quantitative measures of total lung capacity and emphysema were performed with the Airway Inspector (a free, open-source tool used for CT-based image analysis; available at http://www.airwayinspector.org/). The COPDGene Study was approved by the institutional review boards of all participating centers, and all participants provided written informed consent. A detailed description of the study methods appears in the Supplementary Appendix, available with the full text of this article at NEJM.org.
Visual HRCT Analysis
We divided the visual HRCT analysis into two stages. In stage 1, HRCT scans were evaluated by three readers (two chest radiologists and one pulmonologist) with the use of a sequential reading method, as previously described.4 Interstitial lung abnormalities were defined as nondependent changes affecting more than 5% of any lung zone and included nondependent ground-glass or reticular abnormalities, diffuse centrilobular nodularity, nonemphysematous cysts, honeycombing, and traction bronchiectasis (Figure 1Figure 1Four Major Radiographic Subtypes of Interstitial Lung Abnormalities.).7,8 Focal or unilateral ground-glass attenuation, focal or unilateral reticulation, and patchy ground-glass abnormalities (present in <5% of the lung) were considered to be indeterminate findings. The fraction of the lung that met the criteria for radiographic emphysema was not included in the estimation of interstitial lung abnormalities. In stage 2 of the visual HRCT analysis, we divided the participants with interstitial lung abnormalities into four major radiographic subtypes: predominant centrilobular or peribronchial ground-glass opacities sparing the peripheral lung parenchyma (Figure 1A); reticular, nodular, or ground-glass opacities in a predominantly subpleural distribution (Figure 1B); mixed centrilobular and subpleural abnormalities (Figure 1C); and extensive radiographic changes consistent with firm radiographic evidence of interstitial lung disease according to the guidelines of the American Thoracic and European Respiratory Societies9 (Figure 1D). Participants with interstitial lung abnormalities were divided into these four radiographic groups on the basis of the consensus opinion of the three readers, who were unaware of each participant's clinical characteristics.
Statistical Analysis
Total lung capacity was evaluated as both a continuous variable (in liters and as the percent of the predicted value10) and as a binary variable (<80% or 80% of the predicted value11). Lung volume at relaxed exhalation was evaluated as a continuous variable (in liters). The percentage of the lung that was emphysematous was evaluated as a continuous variable (defined by a threshold of both −950 Hounsfield units12 and −910 Hounsfield units13). We defined COPD as a binary variable in accordance with the criteria of the Global Initiative for Chronic Obstructive Lung Disease (GOLD) for disease at stage 2 or higher.14 Bivariate analyses were conducted with Fisher's exact test (for categorical variables) and two-tailed t-tests or the Wilcoxon rank-sum test (for continuous variables) as appropriate. Linear regression models were used for continuous variables and logistic-regression models for binary variables in multivariate analyses. All of the final multivariate models included the variables for age, sex, race, body-mass index (BMI), pack-years of smoking, smoking status (former vs. current smoker), diagnosis of COPD (defined as GOLD stage 2 or higher14), and additional covariates as described below. We oversampled participants on the basis of COPD status5 and performed interaction tests to evaluate whether COPD status modified the associations between interstitial lung abnormalities and both total lung capacity and emphysema. P values of less than 0.05 were considered to indicate statistical significance. All analyses were performed with the use of SAS software, version 9.1 (SAS Institute).


Results
Characteristics of the Study Participants
Of the 2508 participants originally recruited, 2416 (96%) had an HRCT scan available and were included in this analysis. Of these 2416 participants, 1171 (48%) were women, 613 (25%) were black, 1060 (44%) were active smokers, and 1002 (41%) met the GOLD criteria for COPD. Of the 2416 HRCT scans evaluated, 194 (8%) showed interstitial lung abnormalities, 861 (36%) were indeterminate, and 1361 (56%) did not show interstitial lung abnormalities (Table 1Table 1Baseline Characteristics of the Study Participants. and Figure 2AFigure 2Study Enrollment and Findings.). Of the 1421 HRCT scans scored by at least two readers (in stage 1 of the visual HRCT analysis), 899 (63%) had concordant scores. Among the 522 scans for which the evaluations were not concordant, a majority (510 [98%]) involved one indeterminate reading; discrepancies in the interpretation of HRCT scans with respect to the presence or absence of interstitial lung abnormalities were less common (12 scans [2%]). Baseline characteristics of study participants in whom interstitial lung abnormalities were detected, those in whom interstitial lung abnormalities were not detected, and those for whom the diagnosis was indeterminate are shown in Table 1 and in Table E1 in the Supplementary Appendix. As compared with participants without interstitial lung abnormalities, those with interstitial lung abnormalities were significantly older, had a higher BMI, and had a greater amount of exposure to tobacco smoke. Both HRCT measurements of total lung capacity and lung volume at relaxed exhalation were lower in participants with interstitial lung abnormalities. In addition, participants with interstitial lung abnormalities were less likely to have COPD, were more likely to have spirometric measurements that could not be classified according to the GOLD criteria for COPD (forced expiratory volume in 1 second [FEV1], ≤80% of the predicted value; ratio of FEV1 to forced vital capacity [FVC], >0.7), and were more likely to have a lower percentage of emphysema. Although there was an increased frequency of spirometric restriction in participants with interstitial lung abnormalities, the absence of an association with other baseline spirometric measures and the broad distribution of participants with interstitial lung abnormalities on a plot of FEV1 (as a percentage of the predicted value) against the ratio of FEV1 to FVC suggest that spirometry alone is not helpful in classifying interstitial lung abnormalities (Figure 2B, and Figure E1 and Table E2 in the Supplementary Appendix).
Total Lung Capacity and Restrictive Lung Deficit
In adjusted models, the HRCT scans for participants with interstitial lung abnormalities, as compared with the scans for participants without such abnormalities, revealed evidence of reduced total lung capacity (Table 2Table 2Univariate and Multivariate Analyses of the Association between Interstitial Lung Abnormalities and Metrics of Restrictive and Obstructive Lung Disease. and Figure 2C). Similarly, participants with interstitial lung abnormalities had reductions in lung volume at relaxed exhalation (−0.293 liters; 95% confidence interval [CI], −0.430 to −0.156; P<0.001). The odds of a restrictive deficit in participants with interstitial abnormalities were 2.3 times the odds in participants without such abnormalities (Table 2).
Emphysema and COPD
Interstitial lung abnormalities were associated with a lower percentage of emphysema (at −950 and −910 Hounsfield units) in adjusted models (Table 2). Participants with interstitial lung abnormalities had a 47% decrease in their odds of having COPD (Table 2); the strength of this association was influenced by GOLD stage (P for the analysis of variance between GOLD stages 2 through 4 and interstitial lung abnormalities <0.001).
Effects of COPD

In analyses stratified according to COPD status, interstitial lung abnormalities were associated with reduced total lung capacity in participants with COPD (−12% of the predicted value; 95% CI, −17 to −8; P<0.001) and in those without COPD (−7% of the predicted value; 95% CI, −10 to −4; P<0.001); the magnitude of the reduction in total lung capacity was greater in participants with COPD (P=0.01 for the interaction between COPD and interstitial lung abnormalities). Findings were similar for emphysema — the magnitude of the reduction in emphysema was greater in participants with COPD (at −950 Hounsfield units, −7%; 95% CI, −10 to −4, P<0.001) than in those without COPD (defined by a threshold of −950 Hounsfield units, −0.6%; 95% CI, −1.3 to 0.1; P=0.08; P<0.001 for the interaction between COPD and interstitial lung abnormalities). After adjustment for the extent of emphysema, the reductions in total lung capacity were similar between participants with COPD (−7% of the predicted value; 95% CI, −11 to −4; P<0.001) and those without COPD (−6% of the predicted value; 95% CI, −9 to −3; P<0.001). This suggests that in participants with COPD, interstitial lung abnormalities are associated with a reduction in total lung capacity that can be explained by the contributions of both a restrictive lung deficit and an additional reduction in the physiological burden of emphysema (e.g., reduced gas trapping leading to lower lung volumes).
Major Subtypes of Interstitial Lung Abnormalities
Of the 194 participants with interstitial lung abnormalities, 37 (19%) could be classified as having centrilobular abnormalities (Figure 1A), 107 (55%) as having subpleural abnormalities (Figure 1B), 38 (20%) as having centrilobular and subpleural (or mixed) abnormalities (Figure 1C), and 12 (6%) as having radiographic interstitial lung disease (Figure 1D) (see also Tables E3 and E4 of the Supplementary Appendix). Table 3Table 3Multivariate Analyses of the Association between Subtypes of Interstitial Lung Abnormalities and Metrics of Restrictive and Obstructive Lung Disease. shows the association between the subtypes of interstitial lung abnormalities and the measures of restrictive and obstructive lung disease. As compared with participants who did not have interstitial lung abnormalities, those with radiographic interstitial lung disease had the greatest reduction in lung volumes, followed by intermediate reductions in the subpleural and mixed subtypes, with the centrilobular subtype having the smallest reduction in lung volumes (P=0.02 for analysis of variance between subtypes) (Table 3). Emphysema was reduced by a similar magnitude in all subtypes of interstitial lung abnormalities (Table 3).
Exposure to Tobacco Smoke
As reported previously,3 both the extent of exposure to tobacco smoke and smoking status were associated with the odds of having interstitial lung abnormalities, in adjusted models (see the Supplementary Appendix). The specific interstitial abnormality most strongly associated with current smoking status was the presence of centrilobular nodules (odds ratio, 4.82; 95% CI, 2.47 to 9.44; P<0.001). For findings on the association between interstitial lung abnormalities and variables other than smoking, see Table E5 and elsewhere in the Supplementary Appendix.


Discussion
Our analysis of HRCT scans from this large cohort shows that interstitial lung abnormalities are present in approximately 8% of smokers. The findings also show that interstitial lung abnormalities are associated with both reduced total lung capacity and a lesser amount of emphysema in smokers, and the magnitude of these reductions is greatest among those with COPD. We found that smokers with interstitial lung abnormalities have reduced total lung capacity (the extent of which varies according to the subtype of interstitial lung abnormality) and are at an increased risk for a restrictive lung deficit. Although reductions in total lung capacity are expected in established clinical interstitial lung disease,17 our data provide a quantitative estimate of the degree to which interstitial lung abnormalities are associated with reductions in total lung capacity. A major finding of our analyses is the inverse association between interstitial lung abnormalities and the severity of COPD or of emphysema (particularly among participants with COPD). We considered the possibility that interstitial lung abnormalities would result in an erroneous underestimation of the amount of emphysema by increasing the overall lung density defined by Hounsfield-unit thresholds. Several lines of evidence suggest that a density shift in the distribution of Hounsfield units is unlikely to explain our findings. First, the associations we found between emphysema and interstitial lung abnormalities were not most prominent in the lower lobes, where more interstitial abnormalities are expected (see the Supplementary Appendix). Second, the reductions in emphysema noted in participants with interstitial lung abnormalities were paired with the physiological consequences of reduced emphysema (e.g., additional reductions in total lung capacity). Third, we noted inverse associations between the presence of interstitial lung abnormalities and clinically diagnosable COPD, a variable that is independent of the measurement of emphysema with the use of HRCT. Our findings are consistent with, and add weight to, previous studies showing that cigarette smoking is associated with both spirometric restriction18 and areas of high attenuation on HRCT.3 Since emphysema and interstitial lung abnormalities have opposing effects on lung volume, our findings suggest that HRCT may provide important diagnostic information in smokers whose total lung capacity is unexpectedly "normal.” We speculate that this could be clinically important to physicians who may think that a patient who does not have symptoms or characteristic abnormalities on lung-function tests is disease-free, when in fact the patient could be affected by two of the consequences of smoking — emphysema and interstitial lung abnormalities. It is possible that a number of smokers with interstitial lung abnormalities have clinically diagnosable respiratory bronchiolitis, a well-described interstitial lung disease that is related to smoking and associated with ground-glass opacities and centrilobular nodules,19,20 or smoking-related interstitial fibrosis,21 a less well-defined entity with features overlapping those of usual interstitial pneumonia and emphysema.22 However, among the participants in the COPDGene Study with interstitial lung abnormalities, 81% (157) had specific radiographic features and reductions in lung volumes that are not typical of respiratory bronchiolitis.9 As mentioned previously, expected reductions in lung volumes among patients with smoking-related interstitial lung abnormalities could be masked by concomitant emphysema. Our study has several limitations. First, we recognize that congestive heart failure, compression artifacts from bullous emphysema, and atelectasis could mimic the changes on chest HRCT that we have defined as interstitial lung abnormalities. However, in a prior study that excluded participants with heart failure (defined by a physician's diagnosis), a similar association between interstitial lung abnormalities and cigarette smoking was noted.3 In addition, a strong inverse association between emphysema and interstitial lung abnormalities suggests that compression artifacts associated with bullous emphysema are an unlikely explanation for our findings. Moreover, our study shows similar associations with total lung capacity and lung volume at relaxed exhalation, which suggests that our findings are probably not the result of differences in inspiratory effort (or atelectasis). Second, although our measurements of total lung capacity were obtained by means of HRCT, not body plethysmography, previous studies have consistently reported very high degrees of correlation between these measurements (r2 approximately 0.9),23,24 and recent data suggest that plethysmography may be a less accurate measurement of total lung capacity than radiographic measurement in patients with COPD.25 Third, since our population includes smokers with an oversampling of participants with COPD, caution should be exercised in extrapolating our findings to general population samples.
We have found that as compared with smokers without interstitial lung abnormalities, smokers with interstitial abnormalities on HRCT, particularly smokers with COPD, have a reduced total lung capacity and a lesser amount of emphysema. Longitudinal follow-up studies of persons with interstitial lung abnormalities will be required to determine whether these radiographic abnormalities, and the associated reductions in lung volumes, are transient or stable, or whether they will progress to clinically significant disease.


Supported by grants from the National Institutes of Health (U01 HL089897 and U01 HL089856, to COPDGene; K23 HL089353, to Dr. Washko; K08 HL092222, to Dr. Hunninghake; K25 HL104085, to Dr. Estépar; T32 HL07427, to Dr. Brehm; 5R21CA116271-2, to Dr. Hatabu; and K23 HL087030, to Dr. Rosas) and by an award from the Parker B. Francis Foundation (to Dr. Washko).
Dr. Washko reports receiving consulting fees from Medimmune; Dr. Lynch, grant support from Siemens and consulting or board membership fees from Actelion, Gilead, Intermune, Novartis, and Perceptive Imaging; Dr. Khorasani, grant support from GE Medical Systems; Dr. Silverman, grant support from the COPD Foundation and GlaxoSmithKline and consulting fees from GlaxoSmithKline and AstraZeneca; and Dr. Hatabu, grant support from Toshiba Medical and AZE. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.
No other potential conflict of interest relevant to this article was reported.
Drs. Washko and Hunninghake contributed equally to this article. Source Information

From the Pulmonary and Critical Care Division, Brigham and Women's Hospital and Harvard Medical School (G.R.W., G.M.H., I.E.F., J.M.B., A.A.D., K.D., E.K.S., I.O.R.), and the Channing Laboratory (G.M.H., J.C.R., J.M.B., K.D., E.K.S.), the Department of Radiology (M.N., Y.O., T.Y., R.S.J.E., K.P.A., R.K., H.H.), the Center for Pulmonary Functional Imaging (M.N., Y.O., T.Y., H.H.), and the Surgical Planning Laboratory, Department of Radiology (J.C.R., R.S.J.E.), Brigham and Women's Hospital — all in Boston; the Department of Radiology, National Jewish Medical and Research Center, Denver (D.A.L.); the Department of Pulmonary Diseases, Pontificia Universidad Católica de Chile, Santiago, Chile (A.A.D.); and the Division of Pulmonary and Critical Care Medicine, University of Pittsburgh, Pittsburgh (F.C.S.). Address reprint requests to Dr. Rosas at the Pulmonary and Critical Care Division, Department of Medicine, Brigham and Women's Hospital, 75 Francis St., Boston, MA 02115, or at irosas@rics.bwh.harvard.edu; or to Dr. Hatabu at the Center for Pulmonary Functional Imaging, Department of Radiology, Brigham and Women's Hospital, 75 Francis St., Boston, MA 02115, or at hhatabu@partners.org.