Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.

Purpose:

To evaluate the predictive value of imaging and clinical and physiological measurements of chronic obstructive pulmonary disease (COPD) in patients monitored for more than 5 years for pulmonary exacerbations that required hospitalization.

Materials and Methods:

Exacerbations requiring hospitalization were monitored over 5 years in 91 subjects who provided written informed consent. Study was local research ethics board and Health Canada approved and HIPAA compliant. Subjects with COPD underwent spirometry, plethysmography, diffusing capacity of carbon monoxide, St George’s Respiratory Questionnaire, 6-minute walk test, and imaging. Computed tomographic (CT) wall area and relative area with attenuation values less than 2950 HU (RA950), helium 3 (3He) magnetic resonance (MR) imaging ventilation defect percentage (VDP), and apparent diffusion coefficient were generated. Zero-inflated Poisson model was used to compare number of hospitalizations with lung function and imaging measurements.

Results:

Twenty-four subjects were hospitalized 58 times and had significantly worse forced expiratory volume in 1 second (FEV1) (P , .0001), CT RA950 (P = .02), and 3He VDP (P , .0001) than values in 67 subjects who were not hospitalized. In mild to moderate COPD, nine hospitalized subjects had significantly worse FEV1 (P = .02) and 3He VDP (P = .02) than values in 52 subjects who were not hospitalized. 3He VDP was quantitatively related to CT airway morphology (r = 0.26, P = .01) and quantitatively (r = 0.61, P , .0001) and spatially related to emphysema; this spatial relationship was significantly greater for hospitalized patients with COPD than unhospitalized patients (P = .0006). For all subjects, number of prior hospitalizations (P , .0001), 6-minute walk test distance (P , .0001), CT RA950 (P = .03), and 3He VDP (P = .002) were significantly related to number of hospitalizations. For 61 subjects with mild to moderate COPD, only 3He VDP was significantly associated with COPD exacerbations (P = .01).

1

 From the Imaging Research Laboratories, Robarts Research Institute, 1151 Richmond St, London, ON, Canada N6A 5B7 (M.K., D.P., G.P.); Department of Medical Biophysics (M.K., D.P., G.P.) and Division of Respirology, Department of Medicine (D.G.M.), University of Western Ontario, London, Ont, Canada; and Department of Radiology and James Hogg Research Centre, University of British Columbia and Vancouver General Hospital, Vancouver, BC, Canada (H.O.C.). Received January 21, 2014; revision requested March 19; revision received March 21; accepted April 22; final version accepted April 29. M.K. acknowledges PhD scholarship support from the Natural Sciences and Engineering Research Council, Canada, as well as postdoctoral fellowship funding from the Canadian Institutes of Health Research Scholarship and the IMPACT CIHR Strategic Training Program. G.P. gratefully acknowledges support from a CIHR New Investigator Award. Ongoing research funding from CIHR Team Grant CIF 97687 (Thoracic Imaging Network of Canada, or TinCAN) is also gratefully acknowledged. Address correspondence to G.P. (e-mail: [email protected]).  RSNA, 2014

q

Radiology: Volume 273: Number 3—December 2014  n  radiology.rsna.org

Conclusion:

3

He MR imaging VDP represents a mixed airways–emphysema phenotype and helps identify subjects with mild to moderate COPD who are at risk for exacerbation that requires hospitalization.  RSNA, 2014

q

887

Imaging

Miranda Kirby, PhD Damien Pike, BSc Harvey O. Coxson, PhD David G. McCormack, MD Grace Parraga, PhD

Original Research  n  Thoracic

Hyperpolarized 3He Ventilation Defects Used to Predict Pulmonary Exacerbations in Mild to Moderate Chronic Obstructive Pulmonary Disease1

THORACIC IMAGING: Prediction of Pulmonary Exacerbations in Chronic Obstructive Pulmonary Disease

O

ne of the greatest burdens for patients with chronic obstructive pulmonary disease (COPD) is frequent and/or recurrent COPD-related exacerbations that require hospital admission. Many patients with COPD who receive hospital-based exacerbation care are readmitted within a few weeks or months (1), and this is related

Advances in Knowledge nn In 91 subjects with chronic obstructive pulmonary disease (COPD) who were monitored for exacerbations that required hospital-based care, there were 58 COPD exacerbation–related hospitalizations in 24 subjects over 5 years; subjects with COPD who required hospital-based exacerbation care had significantly lower forced expiratory volume in 1 second (P , .0001) and worse 6-minute walk test distance (P , .0001), relative area with CT attenuation values less than 2950 HU (RA950) (P = .02), and 3He ventilation defect percentage (VDP) (P , .0001) than values in 67 subjects who did not require hospitalization. nn In a subgroup of 61 subjects with mild to moderate COPD (Global initiative for chronic Obstructive Lung Disease, or GOLD, category I and II), nine patients who required hospitalization had significantly worse 6-minute walk test distance (P = .001) and 3He VDP values (P = .02) than those who did not require hospitalization. nn A multivariable zero-inflated Poisson model showed that hospitalization for COPD exacerbation was associated with prior hospitalization (P , .0001), 6-minute walk test distance (P , .0001), CT RA950 (P = .03), and 3He VDP (P = .002); in a subgroup of subjects with mild to moderate COPD, only 3He VDP (P = .01) was significantly associated with hospitalization. 888

to more rapid rates of disease progression (2), reduced quality of life (3), and mortality (4). Importantly, nearly four out of five patients will die within 9 years after hospital admission for COPD exacerbation (5). Despite the significant effect of exacerbations that require hospitalization, there are few ways to phenotype patients with high risk, particularly in mild COPD. For example, although hospitalization risk is higher in patients with more severe disease (6), patients with mild disease also have frequent exacerbations (6) and, to date, the strongest predictor of exacerbation risk is a prior history of exacerbation (7). Importantly, strategies to reduce hospitalizations for acute COPD exacerbations, such as inhaled steroids or oral macrolides, are not appropriate for all patients (8). More detailed phenotyping of all patients with COPD, including those at high risk of hospitalization according to physiological, biological (9), genetic, and imaging phenotypes (10,11), may provide new therapeutic windows of opportunity for targeted treatment before hospitalization and/or disease worsening becomes inevitable. Imaging phenotypes provide quantitative, regional, and independent measurements of the two underlying disease mechanisms in COPD, obstructive airways disease, and emphysema. For example, using computed tomographic (CT) measurements of the COPD gene cohort, exacerbation frequency was shown to be related to

Implication for Patient Care nn 3He ventilation defects in COPD reflect obstructed or remodeled airways and emphysema, representing a mixed airway–parenchyma phenotype; 3He VDP was related to pulmonary exacerbations that required hospitalbased care for all subjects with COPD (P = .002) and could be used to identify subjects with milder disease who are at risk for hospitalization due to exacerbation (P = .01).

Kirby et al

both emphysema- and airway-predominant phenotypes (11). In another study in which CT visual scoring methods were used (10), a higher number of exacerbations were reported in patients with COPD who had a mixed emphysema and bronchiectasis–peribronchial thickening phenotype than that in patients with only emphysema. While CT has been used in these previous studies to evaluate anatomic and pulmonary structural phenotypes as predictors of COPD exacerbation, hyperpolarized helium 3 (3He) magnetic resonance (MR) imaging also provides both pulmonary structural and functional information. 3 He MR imaging apparent diffusion coefficient (ADC) measurements reflect airspace size and are elevated in smokers and patients with emphysema (12–14) and correlate significantly with age (15), CT emphysema measures (13,16–18), and histologic findings (19). 3He MR imaging regions of signal void or “ventilation defects” have also been observed in patients with COPD, and these correspond to signal voids in xenon 133 (133Xe) ventilation scintigraphy images (20,21). Published online before print 10.1148/radiol.14140161  Content code: Radiology 2014; 273:887–896 Abbreviations: ADC = apparent diffusion coefficient COPD = chronic obstructive pulmonary disease DLco = diffusing capacity of carbon monoxide FEV1 = forced expiratory volume in 1 second FVC = forced vital capacity GOLD = Global initiative for chronic Obstructive Lung Disease RA950 = relative area with attenuation values less than 2950 HU SGRQ = St George’s Respiratory Questionnaire VDP = ventilation defect percentage Author contributions: Guarantor of integrity of entire study, G.P.; study concepts/ study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, M.K., H.O.C., G.P.; clinical studies, M.K., D.G.M., G.P.; experimental studies, D.P.; statistical analysis, M.K., H.O.C., G.P.; and manuscript editing, all authors Conflicts of interest are listed at the end of this article.

radiology.rsna.org  n Radiology: Volume 273: Number 3—December 2014

THORACIC IMAGING: Prediction of Pulmonary Exacerbations in Chronic Obstructive Pulmonary Disease

Importantly, ventilation defects in COPD are highly reproducible (22), resolve in response to bronchodilator therapy (23) and endobronchial stents (24), and show evidence of progressive worsening over time (25). Although the exact cause of 3He ventilation defects is still unknown, investigators in previous asthma studies suggest that 3 He defects reflect airway wall thickening and remodeling (26). We hypothesized that in COPD, 3He MR imaging ventilation defects reflect the effects of both airway wall remodeling and emphysematous destruction. Moreover, as a mixed airway–parenchyma measurement, 3He MR imaging ventilation defects may provide a way to better understand COPD exacerbations (27), particularly in subjects with mild to moderate COPD. We hypothesize that 3He MR imaging ventilation defects are significantly associated with the number of hospitalizations in patients with mild to moderate COPD. Therefore, the purpose of this study was to evaluate the predictive value of imaging and clinical and physiological measurements of patients with COPD who were monitored over 5 years for pulmonary exacerbations that required hospitalization.

Materials and Methods Subjects All patients provided written informed consent to undergo a protocol that was approved by a local research ethics board and Health Canada (approval no. 129463) and that was compliant with the Health Insurance Portability and Accountability Act. The inclusion criteria for subjects enrolled in this study were subjects between 50 and 85 years of age with a diagnosis of COPD and a ratio of postbronchodilator forced expiratory volume in 1 second (FEV1)/forced vital capacity (FVC) less than 0.70, in accordance with the Global initiative for chronic Obstructive Lung Disease (GOLD) criteria (28). Exclusion criteria included a current diagnosis of asthma or other respiratory conditions concomitant with COPD, but we did

not prospectively exclude subjects with other COPD comorbidities. Because of the significant effect of hospitalizations on mortality in COPD (5), we focused on severe exacerbations that required hospitalization. The number of acute exacerbations that required hospitalization or emergency room visits was determined by using patient hospital records (PowerChart; Cerner, London, UK) for 2.5 years prior to and 2.5 years after the initial subject visit, for a total of 5 years of follow-up. The total number of hospitalizations was defined as the total number of hospitalizations that occurred for all subjects for 2.5 years prior to and 2.5 years after the study visit (from time 22.5 to 2.5 years). The number of prior hospitalizations was defined as the total number of hospitalizations between 2.5 years and 5 years prior to the study visit (from time 25 to 22.5 years).

Pulmonary Function Tests Postbronchodilator spirometry and plethysmography were performed approximately 25 minutes after inhalation of 400 mg salbutamol sulfate USP (ApoSalvent CFC Free Inhalation Aerosol; Apotex, Ontario, ON, Canada) via a spacer device. A body plethysmograph (MedGraphics, Saint Paul, Minn) was used to measure FEV1 and FVC, with a minimum of three acceptable spirometry maneuvers according to American Thoracic Society and European Respiratory Society guidelines (29–31); lung volumes (inspiratory capacity, residual volume) and diffusing capacity of carbon monoxide (DLco) were also measured by using the attached gas analyzer. Subjects also completed the St George’s Respiratory Questionnaire (SGRQ) (32) and a standard 6-minute walk test (33). Imaging MR imaging was performed by using a 3.0-T MR750 (GE Health Care, Milwaukee, Wis) system (22) for acquisition of conventional hydrogen 1 (1H), 3 He static ventilation, and 3He diffusion-weighted MR images. Conventional 1H MR imaging was performed

Radiology: Volume 273: Number 3—December 2014  n  radiology.rsna.org

Kirby et al

during 1.0-L breath hold of N2 from a 1.0-L Tedlar bag after achieving functional residual capacity. Prior to 3He static ventilation and 3He diffusionweighted MR imaging, a spin-exchange polarizer system (Polarean, Durham, NC) was used to polarize 3He gas to 30%–40%. 3He static ventilation and diffusion-weighted imaging were performed after inhalation of a 3He/N2 gas mixture (3He dose = 5 mL/kg body weight) from a 1.0-L Tedlar bag after achieving functional residual capacity, as described previously (22). Multidetector CT was performed by using a 64-section system (Lightspeed VCT; GE Healthcare) in breath hold after inhalation of 1.0 L of N2 once functional residual capacity was achieved (18) to match the MR imaging lung volume. The Evaluation of COPD Longitudinally to Identify Predictive Surrogate End points, or ECLIPSE, imaging protocol (34) was adapted and used in this study, with 64 3 0.625-mm collimation, 120 kVp, 100 effective mA, 500-msec tube rotation time, pitch of 1.25, and image reconstruction with a standard convolution kernel to 1.25 mm. We calculated CT radiation dose according to our manufacturer settings by using the Imaging Performance Assessment of CT, or ImPACT, CT patient dosimetry calculator based on software from the Health Protection Agency of the United Kingdom (NRPB-SR250), and the total effective dose for an average adult was 1.8 mSv.

Image Analysis All MR imaging and CT analyses were performed by an expert in quantitative imaging analysis (M.K.) with 5 years of experience in developing and performing semiautomated 3He MR imaging segmentation with custom-built software generated by using MATLAB R2007b (MathWorks, Natick, Mass). The inter- and intrareproducibility of the 3He MR imaging segmentation software has been evaluated previously and was demonstrated to be high in subjects with COPD (35); therefore, MR images were evaluated once by a single expert observer. 3He MR imaging ventilation defect percentage 889

THORACIC IMAGING: Prediction of Pulmonary Exacerbations in Chronic Obstructive Pulmonary Disease

Figure 1

Figure 1:  Graph shows frequency of the total number of hospitalizations for all subjects. Most subjects (74%) were not hospitalized for a COPD exacerbation (67 of 91), 13% were admitted once (12 of 91), 8% were admitted twice (seven of 91), and 5% were admitted on three or more occasions (five of 91).

(VDP) was quantified by registering the 3He static ventilation images to the 1 H MR images to delineate the defect boundary as described previously (36) by using custom-built software (35). 3 He apparent diffusion coefficient (ADC) maps were also generated from 3 He diffusion-weighted images as described previously (37). CT wall area was measured for all fifth-generation airways (38), the relative area with attenuation values less than 2950 HU (RA950) (39), and low-attenuation clusters of connected regions with CT densitometry values below 2950 HU (40) generated by using Pulmonary Workstation 2.0 (VIDA Diagnostics, Coralville, Iowa). We also evaluated the spatial overlap of ventilation defects at MR imaging and emphysema at CT by generating a CT RA950 “density mask” and coregistering this with the 3He MR imaging ventilation defect mask. Registration of the CT density mask and the 3He MR imaging defect mask were performed by using landmark-based registration as described previously (35). Spatial agreement was evaluated by using an overlap coefficient—the intersection of the 3He ventilation defect and CT density mask voxels expressed as a percentage of total number of 3He ventilation defect mask voxels. 890

Statistical Analysis Comparisons were performed by using two-sample t tests with GraphPad Prism 6 (GraphPad Software, San Diego, Calif); a Fisher exact test was performed for categorical variables. A Shapiro-Wilk normality test was performed prior to the two sample t test, and if the normality test was not passed, the nonparametric Mann-Whitney test was used. Linear regression (r2) and Pearson correlation (r) were also performed by using GraphPad Prism. A zero-inflated Poisson model was used to evaluate the relationship between the total number of hospitalizations with the number of prior hospitalizations, FEV1, DLCO, SGRQ total score, 6-minute walk test distance, and imaging measures (VDP, ADC, RA950, and wall area) by using the PROC COUNTREG procedure in SAS 9.2 software (SAS Institute, Cary, NC). In the model, age, sex, height, body mass index, pack-years of smoking, and smoking status were included as covariates. Sex and smoking status were treated as categorical variables. A zeroinflated Poisson model (41) was chosen because the total number of hospitalizations contains a large proportion of zeros. Results In total, we evaluated 156 current or former smokers. Of the 156 subjects, 91 subjects had a diagnosis of COPD and were monitored for a 5-year period for exacerbations and quantitatively evaluated during a single study. Of these, 24 subjects (24 of 91 [26%]) were hospitalized for COPD exacerbation treatment at least once, and 67 patients were not. Figure 1 shows the frequency distribution for the total number of hospitalizations for all subjects. Although most subjects were not hospitalized for a COPD exacerbation (67 of 91 [74%]), 12 subjects were admitted once (12 of 91 [13%]), seven subjects were admitted twice (seven of 91 [8%]), and five subjects were admitted on three or more occasions (five of 91 [5%]) for a total of 58 hospitalizations during the 5-year period. Thirteen subjects had at least one hospitalization

Kirby et al

within 2.5 years prior to the study visit (mean, 64 weeks 6 39; range, 19–152 weeks), and 19 subjects had at least one hospitalization within 2.5 years after the study visit. Table 1 shows subject demographics, pulmonary function, 6-minute walk test distances, SGRQ scores, and CT and 3He MR imaging measurements for all subjects with COPD and for the subgroup with GOLD I-II (mild to moderate) COPD. Subjects with COPD who were hospitalized and those who were not hospitalized were not significantly different with respect to age, sex, height, body mass index, smoking status, and pack-years of smoking. However, for all subjects with COPD who were hospitalized, there was significantly worse FEV1 (P , .0001), residual volume (P = .001), DLCO (P = .01), 6-minute walk test distance (P , .0001), and SGRQ total score (P , .0001). For the subjects with mild to moderate COPD who were hospitalized, there was also significantly worse FEV1 (P = .02), 6-minute walk test distance (P = .001), and SGRQ total score (P = .047). Figure 2 shows the coronal 3He MR imaging ventilation image with the corresponding 3He MR imaging ADC map, CT low-attenuation cluster map, and 3He defect mask registered to the RA950 density mask for a subject with mild to moderate or severe COPD who was not hospitalized and a subject with mild to moderate or severe COPD who was hospitalized for a COPD exacerbation. Subjects with COPD who were hospitalized had greater 3He ventilation defects, hyperintense ADC maps reflective of more advanced emphysema, and larger clusters of CT low-attenuation regions as compared with subjects who were not hospitalized. As shown in Table 1, for all subjects with COPD, those who were hospitalized had significantly worse CT RA950 (P = .02) and 3He MR imaging VDP (P , .0001) than those who were not hospitalized; CT wall area (reflective of airway wall morphology) and 3He MR imaging ADC (reflective of emphysema) were not significantly different. For the mildto-moderate COPD subgroup, those who were hospitalized had significantly

radiology.rsna.org  n Radiology: Volume 273: Number 3—December 2014

THORACIC IMAGING: Prediction of Pulmonary Exacerbations in Chronic Obstructive Pulmonary Disease

Kirby et al

Table 1 Clinical, Functional, and Imaging Measurements in All Subjects with COPD and Those with GOLD I-II (Mild to Moderate) COPD Only without Hospitalization and Those with One or More Hospitalizations All Subjects with COPD Parameter Subject demographics   Age (y)   No. of women   Height (cm)   Body mass index (kg/m2)   No. of pack-years   No. of current smokers Pulmonary function tests  FEV1 (%)pred  FEV1/FVC   Inspiratory capacity (%)pred   Residual volume (%)pred   DLco (%)pred   Six-minute walk test (m)   SGRQ total score Imaging measurements  RA950 (%)   Low-attenuation clusters   Wall area (cm2)   ADC (cm2/sec)   VDP (%)  VDP–RA950 overlap (%)

Mild to Moderate COPD Only

No Hospitalization (n = 67)

One or More Hospitalizations (n = 24)

P Value

No Hospitalization (n = 52)

One or More Hospitalizations (n = 9)

P Value

71 6 9 21 (31) 169 6 8 26.5 6 4.8 46 6 30 10 (15)

71 6 7 10 (42) 165 6 9 26.3 6 3.9 48 6 28 3 (12)

.85 .45 .06 .83 .67 ..99

71 6 9 17 (33) 169 6 8 26.6 6 4.3 43 6 29 8 (15)

73 6 6 4 (44) 167 6 9 26.7 6 3.8 47 6 20 1 (11)

.40 .71 .52 .93 .50 ..99

67 6 21 52 6 12 93 6 23 149 6 43 56 6 20

44 6 20 41 6 12 77 6 21 183 6 51 44 6 18

,.0001 ,.0001 .004 .001 .01

75 6 17 57 6 9 99 6 20 136 6 33 61 6 19

63 6 17 52 6 10 93 6 18 142 6 44 53 6 21

.02 .20 .46 .67 .22

395 6 87 36.0 6 18.1

305 6 69 53.4 6 14.4

,.0001 ,.0001

408 6 75 31.5 6 16.9

331 6 56 43.4 6 9.4

.001 .047

968 21.77 6 0.22 23.8 6 4.2 0.41 6 0.10 16 6 9 34 6 22

17 6 14 21.72 6 0.20 24.2 6 5.9 0.46 6 0.12 25 6 9 54 6 25

.02 .30 .30 .03 ,.0001 .0006

768 21.79 6 0.23 23.6 6 4.0 0.39 6 0.09 13 6 7 28 6 20

10 6 11 21.76 6 0.26 23.7 6 4.6 0.40 6 0.12 20 6 9 42 6 31

.97 .96 .94 .96 .02 .24

Note.—Data are presented as mean values 6standard deviations, unless indicated otherwise. Numbers in parentheses are percentages. (%)pred = percent predicted.

worse 3He MR imaging VDP (P = .02) than those who were not hospitalized. There were no significant differences for RA950, wall area, or ADC. To better understand the origin of 3He ventilation defects, the spatial relationship between 3He ventilation defects and CT-derived emphysema is also shown in Figure 2. Greater spatial overlap between 3He ventilation defects and CT-derived emphysema (shown in yellow) is shown for subjects who were hospitalized and subjects who were not hospitalized during the 5-year follow-up period (P = .0006); 3He VDP was also quantitatively correlated with CT wall area (r = 0.26, P = .01), RA950 (r = 0.61, P , .0001), and emphysematous lesion size (low-attenuation clusters, r = 0.34, P = .001). Table 2 shows the result of a multivariable model that included the

number of prior hospitalizations, along with functional (FEV1, DLco, SGRQ, 6-minute walk test) and imaging (RA950, wall area, VDP, and ADC) measurements for all subjects with COPD, as well as the subgroup of subjects with GOLD I-II (mild to moderate) COPD. After adjusting for covariates, the number of prior hospitalizations was significantly associated with the number of hospitalizations during the 5-year time frame, with a twofold increase in the expected number of hospitalizations per prior hospitalization. The 6-minute walk test distance was also significantly associated with number of hospitalizations (P , .0001); each 1-m decrease in the 6-minute walk test distance resulted in a onefold increase in the expected number of hospitalizations. Both CT RA950 (P = .03) and 3He VDP (P = .03)

Radiology: Volume 273: Number 3—December 2014  n  radiology.rsna.org

were significantly associated with the number of hospitalizations; a 1% increase in VDP and RA950 each resulted in onefold increases in the expected number of hospitalizations. For subjects with mild to moderate COPD, VDP was the only variable significantly associated with the number of hospitalizations (P = .01); each 1% increase in VDP resulted in a onefold increase in the expected number of hospitalizations.

Discussion We monitored COPD exacerbations that required hospitalization in a relatively small group of 91 subjects and observed that (a) 24 of 91 subjects who had COPD-driven hospitalization had significantly lower FEV1 and worse 6-minute walk test distance, CT 891

THORACIC IMAGING: Prediction of Pulmonary Exacerbations in Chronic Obstructive Pulmonary Disease

Kirby et al

Figure 2

Figure 2:  3He MR and CT images for all subjects with COPD and those in the mild-to-moderate COPD subgroup, with and without exacerbations that required hospitalization. Top row: 3He MR images; coronal static ventilation images (blue) are registered to images in the thoracic cavity. Second row: 3He MR imaging coronal ADC map. Third row: CT-derived three-dimensional low-attenuation clusters (LAC) of emphysema, represented as spheres with CT densitometry values below 2950 HU registered to the three-dimensional reconstruction of the airway tree. Bottom row: coronal 3He defect mask registered to the RA950 density mask. The first patient (left) is a 75-year-old man (FEV1 = 42% predicted, FEV1/FVC = 39%, VDP = 27%, ADC = 0.42 cm2/sec, RA950 = 4%). The second patient (middle left) is a 75-year-old man (FEV1 = 30% predicted, FEV1/FVC = 30%, VDP = 34%, ADC = 0.48 cm2/sec, RA950 = 23%). The third patient (middle right) is a 60-year-old woman (FEV1 = 76% predicted, FEV1/FVC = 60%, VDP = 4%, ADC = 0.24 cm2/sec, RA950 = 1%). The fourth patient (right) is a 66-year-old man (FEV1 = 59% predicted, FEV1/FVC = 38%, VDP = 25%, ADC = 0.34 cm2/sec, RA950 = 5%).

RA950, and 3He VDP values than those in subjects who did not require hospitalization; (b) in a subgroup of subjects with mild to moderate COPD, hospitalization was reported in nine of 61 subjects with significantly worse FEV1, 892

6-minute walk test distance, and 3He VDP values than those in subjects who were not hospitalized; (c) significantly greater spatial overlap between 3He ventilation defects and emphysema observed at CT was observed for subjects

that were hospitalized; and (d) a multivariable model demonstrated that the number of prior hospitalizations, 6-minute walk test distance, RA950, and VDP values were significantly associated with the number of hospitalizations for

radiology.rsna.org  n Radiology: Volume 273: Number 3—December 2014

THORACIC IMAGING: Prediction of Pulmonary Exacerbations in Chronic Obstructive Pulmonary Disease

Table 2 Multivariable Regression Model for Number of Hospitalizations in All Subjects with COPD and Those in the Mild-to-Moderate COPD Subgroup Variable* All subjects with COPD   No. of prior hospitalizations  FEV1 (% predicted)   DLco (% predicted)   Six-minute walk test (m)   SGRQ total score   Wall area (mm2)  RA950 (%)   VDP (%)   ADC (cm2/sec) Subjects with mild to moderate COPD   No. of prior hospitalizations†  FEV1 (% predicted)   DLco (% predicted)   Six-minute walk test (m)   SGRQ total score   Wall area (mm2)  RA950 (% predicted)   VDP (%)   ADC (cm2/sec)

Parameter Estimate

Standard Error

0.63 0.00 20.01 20.02 20.02 20.04 20.15 0.11 7.96

0.15 0.02 0.02 0.00 0.02 0.03 0.07 0.04 9.43

.

. 20.05 20.08

0.04 0.04

0.01 0.11 0.18 20.11 0.17 212.17

… 0.06 0.13 0.09 0.07 10.79

Fold Change

P Value

1.87 1.00 0.99 0.98 0.98 0.96 0.86 1.12 2868.46

,.0001 .92 .69 ,.0001 .36 .20 .03 .002 .40

.

. 0.95 0.92

.15 .06

1.11 1.20 0.89 1.19 0.00

… .07 .17 .19 .01 .26

….

* Adjusted for age, sex, height, body mass index, pack-years, and smoking status. †

No prior hospitalizations for mild to moderate COPD.

all subjects with COPD, but only VDP was significantly associated with the number of hospitalizations in subjects with mild to moderate COPD. It is important to note that in contrast with previous work (6,11), we focused only on COPD exacerbations that required hospitalization—a very conservative definition and reflective of the fact that COPD exacerbations that require hospital admission are straightforward to define and monitor and have a large effect on patient prognosis (4), quality of life, and healthcare costs. We think this is important because for clinical translation, the economics of increased surveillance with MR imaging must be balanced with the economic and societal benefits of reducing COPD hospitalizations. The economic burden of COPD hospitalizations is substantial. COPD is responsible for 1 million hospitalizations and 27 million physician visits each year, costing the health care system $21–$25 billion annually (42). In a recent evaluation of

COPD healthcare usage, one in every four hospital beds was estimated to be occupied by patients with COPD, with an average hospital admission lasting 10 days and costing approximately $10 000 per admission (43,44). Moreover, patients with COPD who receive hospitalbased exacerbation care are likely to have another exacerbation and be readmitted within a few weeks or months (1). Although the cost of MR imaging, polarization equipment, and gas is relatively high compared with CT or chest radiography, the cost per MR imaging examination is likely far lower than the cost associated with a single hospital stay. Therefore, such costs may be well justified if MR imaging could be used to prevent frequent hospitalizations and thereby significantly reduce the longterm economic burden on the health care system, as well as patient burden. In previous studies, investigators reported the association between exacerbation frequency and COPD disease severity (6,45), and, therefore, we were

Radiology: Volume 273: Number 3—December 2014  n  radiology.rsna.org

Kirby et al

not surprised that in the current study, subjects with COPD with exacerbations had more severe disease. In contrast to previous findings (45), however, we also observed COPD exacerbations in patients with milder disease. Patients with COPD experience exacerbations regardless of FEV1, underscoring the notion that there is a frequent exacerbation phenotype that cannot be predicted by using spirometry alone (6). Moreover, because there are more patients with COPD who have mild to moderate disease in the general population compared to severe disease, the overall burden of hospitalizations may be higher in mild COPD (6). Therefore, strategies that focus on identifying patients at risk who have milder COPD are important. We also noted that subjects with COPD who required hospitalizations had significantly worse emphysema and 3 He VDP values. This is in agreement with other work (11), although investigators in that study also reported greater CT airway wall area in subjects with COPD with exacerbations. We did not observe this here, perhaps because we were not powered to detect significant CT wall area measurement differences. It is well understood that thoracic CT has a fundamental spatial resolution limit that restricts quantitative measurements to the fifth- or sixthgeneration airways, and such measurements are also necessarily restricted to surviving airways, which may bias results. To overcome some of these limitations, CT measurements derived from inspiratory and expiratory images have been developed (46), but these may not reflect the structure-function relationships in the smaller peripheral airways—the major site of airflow limitation in COPD (47). We also note that although 3He ADC is sensitive to early or mild disease (12,13), no differences between subgroups were observed here. This may be due to the dependence of the ADC measurement on well-ventilated lung, meaning that large emphysematous bullae that fill slowly or not at all during MR imaging acquisition (48) cannot yield ADC values. This generates an inherent bias for ADC values, 893

THORACIC IMAGING: Prediction of Pulmonary Exacerbations in Chronic Obstructive Pulmonary Disease

which, in advanced COPD, is important to consider. The quantitative relationship observed for ventilation defects and emphysema led us to speculate about the origin of ventilation defects in COPD, including a potential spatial relationship with partially or fully occluded airways and to emphysematous bullae. Emphysema alone may result in ventilation defects because of their long time constants for filling and emptying associated with these enlarged parenchymal spaces (48). A significant quantitative relationship and spatial overlap between ventilation defects and emphysema was observed here, and this strongly suggests that in these patients, emphysema is one cause of ventilation defects. Moreover, we think this means that ventilation defects represent both airway and emphysema contributions to COPD and can therefore be considered as a mixed airway–parenchyma measurement. Certainly, future studies should focus on the spatial structure– function relationships of ventilation defects in COPD to better understand the contributions of airway abnormalities that lead to bullae and the presence of ventilation defects where there is no emphysema. In the multivariable modeling we used, it was interesting that a history of prior hospitalizations was a significant predictor of hospitalization frequency (6), but FEV1 was not. We believe that this finding underscores the fact that for all subjects with COPD, and, importantly, for those with milder COPD, there may be an exacerbation susceptibility phenotype that cannot be predicted with FEV1 and that imaging may offer this important information. We also reported that 3He VDP was significantly associated with number of hospitalizations for all subjects with COPD, and it was the only measurement significantly associated with hospitalization in the subgroup of subjects with mild to moderate COPD. We believe VDP is sensitive to thickening of the airway wall and obstruction of the airway lumen, as well as the long time constants for filling in emphysematous 894

lung regions, and may therefore have increased sensitivity to the underlying disease changes that occur in subjects with COPD that experience frequent exacerbations. Importantly, as FEV1 is not useful for predicting exacerbation risk in patients with mild to moderate COPD, 3He VDP may therefore provide an improved understanding of the underlying regional changes or phenotype of patients with high hospitalization risk and help identify these subjects for more extensive monitoring. 3He VDP may also be useful in selecting subgroups of these patients with mild to moderate COPD who are at high hospitalization risk for clinical trials of targeted treatments. We must acknowledge that this study is limited by several factors. First, some of the parameters related to exacerbation risk were not included in the models generated here because we did not record these data. For example, in the ECLIPSE COPD cohort, a history of heartburn and/ or gastroesophageal reflux and white blood cell count was predictive of exacerbation frequency (6). We did not prospectively acquire these measures and therefore cannot determine their role here. We also did not prospectively collect information regarding the subjects’ treatment regimens or how treatment changed during or after exacerbations over the 5-year time period. We acknowledge that monitoring and recording the subjects’ medication is important for our understanding of exacerbation risk. In addition, we did not evaluate the presence of bronchiectasis here, although recent reports suggest that there is a high prevalence of bronchiectasis in patients with moderate to severe COPD (49,50) and that bronchiectasis is related to severe exacerbations and mortality of all causes (49). We must also acknowledge that because hospitalizations prior to and after the study visit were used, we cannot infer a temporal relationship between imaging measurements and hospitalization. In other words, we cannot ascertain whether 3He MR imaging ventilation defects lead to COPD hospitalizations or whether hospitalizations

Kirby et al

lead to greater 3He MR imaging ventilation defects. Clearly, larger studies are required in which subjects with COPD are evaluated immediately prior to and after a COPD exacerbation. Another limitation of this study is that, aside from CT, we did not directly compare our results with those obtained by using other imaging modalities. It would be of interest to determine whether measurements derived from other imaging modalities or other MR imaging techniques, including ultra-short echo time MR imaging (51), Fourier decomposition MR imaging (52,53), and oxygen-enhanced MR imaging (54), also show significant associations with hospitalizations in mild to moderate COPD. We also did not prospectively perform a power or sample size calculation to determine if the multivariable model would have enough power to enable detection of significant relationships between each of the predictor variables with the number of hospitalizations. Finally, we must acknowledge that hyperpolarized 3 He MR imaging is still only performed at a handful of research sites because of the limited global supply of 3He gas and the specialized polarization equipment and physics expertise required, and it is currently not used in routine clinical practice. However, hyperpolarized 129Xe gas is substantially more abundant, and with the recent improvements in commercialized turnkey systems, there is the potential for broader dissemination and adoption of hyperpolarized gas MR imaging for COPD imaging. In summary, 3He VDP was significantly associated with hospitalization for COPD exacerbation in all subjects and was the only measurement significantly associated with hospitalization in mild to moderate COPD, suggesting that functional MR imaging could be used to identify subjects with milder disease who are at risk for exacerbations that require hospital care. Acknowledgments: We thank S. McKay, BSc, D. Buchanan, MSc, and S. Blamires, CCRS, RPT, for subject recruitment, clinical coordination, and clinical database management; A. Wheatley, BSc, for production and dispensing of

radiology.rsna.org  n Radiology: Volume 273: Number 3—December 2014

THORACIC IMAGING: Prediction of Pulmonary Exacerbations in Chronic Obstructive Pulmonary Disease

3

He gas; and T. Szekeres, RTMR, for MR imaging of research volunteers. Disclosures of Conflicts of Interest: M.K. disclosed no relevant relationships. D.P. disclosed no relevant relationships. H.O.C. Activities related to the present article: disclosed no relevant relationships. Activities not related to the present article: author received payment for consultancy with GSK, and institution received grants from GSK and Spiration. Other relationships: disclosed no relevant relationships. D.G.M. disclosed no relevant relationships. G.P. disclosed no relevant relationships.

References 1. Dalal AA, Shah M, D’Souza AO, Rane P. Costs of COPD exacerbations in the emergency department and inpatient setting. Respir Med 2011;105(3):454–460. 2. Donaldson GC, Seemungal TA, Bhowmik A, Wedzicha JA. Relationship between exacerbation frequency and lung function decline in chronic obstructive pulmonary disease. Thorax 2002;57(10):847–852. 3. Seemungal TA, Donaldson GC, Paul EA, Bestall JC, Jeffries DJ, Wedzicha JA. Effect of exacerbation on quality of life in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1998;157(5 Pt 1):1418–1422. 4. Soler-Cataluña JJ, Martínez-García MA, Román Sánchez P, Salcedo E, Navarro M, Ochando R. Severe acute exacerbations and mortality in patients with chronic obstructive pulmonary disease. Thorax 2005; 60(11):925–931. 5. Gudmundsson G, Ulrik CS, Gislason T, et al. Long-term survival in patients hospitalized for chronic obstructive pulmonary disease: a prospective observational study in the Nordic countries. Int J Chron Obstruct Pulmon Dis 2012;7:571–576. 6. Hurst JR, Vestbo J, Anzueto A, et al. Susceptibility to exacerbation in chronic obstructive pulmonary disease. N Engl J Med 2010;363(12):1128–1138. 7. Donaldson GC, Wedzicha JA. COPD exacerbations. 1: Epidemiology. Thorax 2006;61(2):164–168. 8. Wilkinson T, Wedzicha JA. Strategies for improving outcomes of COPD exacerbations. Int J Chron Obstruct Pulmon Dis 2006;1(3):335–342. 9. Burgel PR, Nesme-Meyer P, Chanez P, et al. Cough and sputum production are associated with frequent exacerbations and hospitalizations in COPD subjects. Chest 2009;135(4):975–982.

10. Tulek B, Kivrak AS, Ozbek S, Kanat F, Suerdem M. Phenotyping of chronic obstructive pulmonary disease using the modified Bhalla scoring system for high-resolution computed tomography. Can Respir J 2013;20(2):91– 96. 11. Han MK, Kazerooni EA, Lynch DA, et al. Chronic obstructive pulmonary disease exacerbations in the COPDGene study: associated radiologic phenotypes. Radiology 2011;261(1):274–282. 12. Kirby M, Owrangi A, Svenningsen S, et al. On the role of abnormal DL(CO) in exsmokers without airflow limitation: symptoms, exercise capacity and hyperpolarised helium-3 MRI. Thorax 2013;68(8):752–759. 13. Fain SB, Panth SR, Evans MD, et al. Early emphysematous changes in asymptomatic smokers: detection with 3He MR imaging. Radiology 2006;239(3):875–883. 14. Swift AJ, Wild JM, Fichele S, et al. Emphysematous changes and normal variation in smokers and COPD patients using diffusion 3 He MRI. Eur J Radiol 2005;54(3):352–358. 15. Fain SB, Altes TA, Panth SR, et al. Detection of age-dependent changes in healthy adult lungs with diffusion-weighted 3He MRI. Acad Radiol 2005;12(11):1385–1393. 16. Fain SB, Gonzalez-Fernandez G, Peterson ET, et al. Evaluation of structure-function relationships in asthma using multidetector CT and hyperpolarized He-3 MRI. Acad Radiol 2008;15(6):753–762. 17. Diaz S, Casselbrant I, Piitulainen E, et al. Validity of apparent diffusion coefficient hyperpolarized 3He-MRI using MSCT and pulmonary function tests as references. Eur J Radiol 2009;71(2):257–263. 18. Kirby M, Svenningsen S, Owrangi A, et al. Hyperpolarized 3He and 129Xe MR imaging in healthy volunteers and patients with chronic obstructive pulmonary disease. Radiology 2012;265(2):600–610.

Kirby et al

apparent diffusion coefficients in chronic obstructive pulmonary disease: preliminary results at 3.0 Tesla. Invest Radiol 2007;42(6):384–391. 23. Kirby M, Mathew L, Heydarian M, Ete mad-Rezai R, McCormack DG, Parraga G. Chronic obstructive pulmonary disease: quantification of bronchodilator effects by using hyperpolarized ³He MR imaging. Radiology 2011;261(1):283–292. 24. Mathew L, Kirby M, Farquhar D, et al. Hyperpolarized 3He functional magnetic resonance imaging of bronchoscopic airway bypass in chronic obstructive pulmonary disease. Can Respir J 2012;19(1):41–43. 25. Kirby M, Mathew L, Wheatley A, Santyr GE, McCormack DG, Parraga G. Chronic obstructive pulmonary disease: longitudinal hyperpolarized (3)He MR imaging. Radiology 2010;256(1):280–289. 26. Svenningsen S, Kirby M, Starr D, et al. What are ventilation defects in asthma? Thorax 2014;69(1):63–71. 27. Kirby M, Kanhere N, Etemad-Rezai R, McCormack DG, Parraga G. Hyperpolarized helium-3 magnetic resonance imaging of chronic obstructive pulmonary disease exacerbation. J Magn Reson Imaging 2013;37(5): 1223–1227. 28. Global Initiative for Chronic Lung Disease. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease. Updated 2013. 2013. 29. Miller MR, Hankinson J, Brusasco V, et al. Standardisation of spirometry. Eur Respir J 2005;26(2):319–338. 30. Macintyre N, Crapo RO, Viegi G, et al. Standardisation of the single-breath determination of carbon monoxide uptake in the lung. Eur Respir J 2005;26(4):720–735. 31. Wanger J, Clausen JL, Coates A, et al. Standardisation of the measurement of lung volumes. Eur Respir J 2005;26(3):511–522.

19. Woods JC, Choong CK, Yablonskiy DA, et al. Hyperpolarized 3He diffusion MRI and histology in pulmonary emphysema. Magn Reson Med 2006;56(6):1293–1300.

32. Jones PW, Quirk FH, Baveystock CM. The St George’s Respiratory Questionnaire. Respir Med 1991;85(Suppl B):25–31; discussion 33–37.

20. de Lange EE, Mugler JP 3rd, Brookeman JR, et al. Lung air spaces: MR imaging evaluation with hyperpolarized 3He gas. Radiology 1999;210(3):851–857.

33. Enright PL. The six-minute walk test. Respir Care 2003;48(8):783–785.

21. Altes TA, Rehm PK, Harrell F, Salerno M, Daniel TM, De Lange EE. Ventilation imaging of the lung: comparison of hyperpolarized helium-3 MR imaging with Xe-133 scintigraphy. Acad Radiol 2004;11(7):729–734. 22. Parraga G, Ouriadov A, Evans A, et al. Hyperpolarized 3He ventilation defects and

Radiology: Volume 273: Number 3—December 2014  n  radiology.rsna.org

34. Vestbo J, Anderson W, Coxson HO, et al. Evaluation of COPD Longitudinally to Identify Predictive Surrogate End-points (ECLIPSE). Eur Respir J 2008;31(4):869– 873. 35. Kirby M, Heydarian M, Svenningsen S, et al. Hyperpolarized 3He magnetic resonance functional imaging semiautomated segmentation. Acad Radiol 2012;19(2):141–152.

895

THORACIC IMAGING: Prediction of Pulmonary Exacerbations in Chronic Obstructive Pulmonary Disease

36. Woodhouse N, Wild JM, Paley MN, et al. Combined helium-3/proton magnetic resonance imaging measurement of ventilated lung volumes in smokers compared to never-smokers. J Magn Reson Imaging 2005;21(4):365–369. 37. Kirby M, Heydarian M, Wheatley A, Mc Cormack DG, Parraga G. Evaluating bronchodilator effects in chronic obstructive pulmonary disease using diffusion-weighted hyperpolarized helium-3 magnetic resonance imaging. J Appl Physiol (1985) 2012;112(4):651–657. 38. Nakano Y, Muro S, Sakai H, et al. Com puted tomographic measurements of airway dimensions and emphysema in smokers. Correlation with lung function. Am J Respir Crit Care Med 2000;162(3 Pt 1):1102–1108. 39. Gevenois PA, de Maertelaer V, De Vuyst P, Zanen J, Yernault JC. Comparison of computed density and macroscopic morphometry in pulmonary emphysema. Am J Respir Crit Care Med 1995;152(2):653–657. 40. Mishima M, Hirai T, Itoh H, et al. Complexity of terminal airspace geometry assessed by lung computed tomography in normal subjects and patients with chronic obstructive pulmonary disease. Proc Natl Acad Sci U S A 1999;96(16):8829–8834.

chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2013;187(6):596– 601. 43. Mittmann N, Kuramoto L, Seung SJ, Haddon JM, Bradley-Kennedy C, Fitzgerald JM. The cost of moderate and severe COPD exacerbations to the Canadian healthcare system. Respir Med 2008;102(3): 413–421. 44. Dalal AA, Christensen L, Liu F, Riedel AA. Direct costs of chronic obstructive pulmonary disease among managed care patients. Int J Chron Obstruct Pulmon Dis 2010;5:341–349. 45. Greenberg SB, Allen M, Wilson J, Atmar RL. Respiratory viral infections in adults with and without chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2000;162(1):167–173. 46. Galbán CJ, Han MK, Boes JL, et al. Computed tomography-based biomarker provides unique signature for diagnosis of COPD phenotypes and disease progression. Nat Med 2012;18(11):1711–1715. 47. Hogg JC, Macklem PT, Thurlbeck WM. Site and nature of airway obstruction in chronic obstructive lung disease. N Engl J Med 1968;278(25):1355–1360.

41. Lambert D. Zero-inflated Poisson regression, with an application to defects in manufacturing. Technometrics 1992;34(1):1–14.

48. Marshall H, Deppe MH, Parra-Robles J, et al. Direct visualisation of collateral ventilation in COPD with hyperpolarised gas MRI. Thorax 2012;67(7):613–617.

42. Gershon AS, Guan J, Victor JC, Goldstein R, To T. Quantifying health services use for

49. Martínez-García MA, Soler-Cataluña JJ, Donat Sanz Y, et al. Factors associated with

896

Kirby et al

bronchiectasis in patients with COPD. Chest 2011;140(5):1130–1137. 50. Martínez-García MA, de la Rosa Carrillo D, Soler-Cataluña JJ, et al. Prognostic value of bronchiectasis in patients with moderate-to-severe chronic obstructive pulmonary disease. Am J Respir Crit Care Med 2013;187(8):823–831. 51. Lutterbey G, Gieseke J, von Falkenhausen M, Morakkabati N, Schild H. Lung MRI at 3.0 T: a comparison of helical CT and highfield MRI in the detection of diffuse lung disease. Eur Radiol 2005;15(2):324–328. 52. Deimling M, Jellus V, Geiger B, Chefd’hotel C. Time resolved lung ventilation imaging by Fourier decomposition [abstr]. In: Proceedings of the Sixteenth Meeting of the International Society for Magnetic Resonance in Medicine. Berkeley, Calif: International Society for Magnetic Resonance in Medicine, 2008; 2639. 53. Bauman G, Puderbach M, Deimling M, et al. Non–contrast-enhanced perfusion and ventilation assessment of the human lung by means of Fourier decomposition in proton MRI. Magn Reson Med 2009;62(3):656– 664. 54. Ohno Y, Hatabu H, Takenaka D, Van Cauteren M, Fujii M, Sugimura K. Dynamic oxygen-enhanced MRI reflects diffusing capacity of the lung. Magn Reson Med 2002;47(6): 1139–1144.

radiology.rsna.org  n Radiology: Volume 273: Number 3—December 2014

Hyperpolarized (3)He ventilation defects used to predict pulmonary exacerbations in mild to moderate chronic obstructive pulmonary disease.

To evaluate the predictive value of imaging and clinical and physiological measurements of chronic obstructive pulmonary disease ( COPD chronic obstru...
1012KB Sizes 2 Downloads 5 Views