点击显示 收起
1 From the Center for In-vivo Hyperpolarized Gas MR Imaging, Departments of Radiology (M.S., E.E.d.L., T.A.A., J.R.B., J.P.M.), Biomedical Engineering (M.S., T.A.A., J.R.B., J.P.M.), and Internal Medicine (J.D.T.), University of Virginia School of Medicine, 1000 Lee St, Box 800170, Charlottesville, VA 22908. From the 2000 RSNA scientific assembly. Received November 24, 2000; revision requested January 22, 2001; revision received April 23; accepted May 15. Supported in part by National Institutes of Health grant R44-HL059022; the University of Virginia Pratt Fund; the Dean of the Medical School, Robert M. Carey, MD; and Siemens Medical Solutions, Iselin, NJ.
| ABSTRACT |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
MATERIALS AND METHODS: Hyperpolarized 3He diffusion MR imaging was performed in 16 healthy volunteers and 11 patients. Coronal diffusion-sensitized MR images were obtained during suspended respiration after inhalation of laser-polarized 3He gas, and images of the ADC were calculated. Spirometry was performed immediately before imaging. The mean and SD of the ADCs were compared between subject groups and were correlated with spirometric indexes.
RESULTS: ADC images were homogeneous in volunteers, but demonstrated regional variations in patients. The mean and SD of the ADCs for patients were significantly larger (P < .002) than those for volunteers. The mean ADCs for all subjects correlated with the percentage of predicted forced expiratory volume in 1 second, or FEV1, (r = -0.797, P < .001) and the ratio of FEV1 to forced vital capacity, or FVC, (r = -0.930, P < .001). ADC images in patients demonstrated a significant increase (P < .001) in the ADCs in the upper regions compared with the lower regions of the lung.
CONCLUSION: Hyperpolarized 3He diffusion MR imaging demonstrated potential for use in evaluating the global and regional severity of emphysema.
Index terms: Alpha-1–antitrypsin deficiency, 60.7511 • Emphysema, pulmonary, 60.751 • Helium • Lung, MR, 60.12143 • Magnetic resonance (MR), contrast enhancement, 60.121412, 61.12143 • Magnetic resonance (MR), nuclei other than H, 60.121412, 60.12147
| INTRODUCTION |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
Hyperpolarized helium 3 (3He) magnetic resonance (MR) imaging is a recently developed technique that is capable of producing high-spatial-resolution MR images of the lung air spaces after inhalation of laser-polarized helium gas (5–8). Hyperpolarized 3He ventilation MR imaging shows promise for differentiating healthy lungs from those with diseases, such as chronic obstructive pulmonary disease (9,10), asthma (11), and cystic fibrosis (12). The contrast in these MR images is determined by the regional spin density of the helium and reflects the gas distribution in the ventilated portions of the lung. However, these MR images do not provide information about the integrity of the lung microstructure in the ventilated regions.
The application of diffusion MR imaging methods (13,14) to hyperpolarized 3He imaging may present the opportunity to probe the lung microstructure. Helium has a high self-diffusion coefficient (2 cm2/sec), and helium atoms in an unrestricted space experience displacements of as much as several millimeters during the echo time of gradient-echo pulse sequences typically used for hyperpolarized 3He MR imaging. However, when helium is confined to spaces, such as the distal airway structures of the lung, its motion is restricted, which results in smaller displacements and a decrease in the apparent diffusion coefficient (ADC) as measured with MR imaging (15,16). The ADC of hyperpolarized 3He has been measured in animals with elastase-induced emphysema (15), in small numbers of healthy subjects (16–20), and in patients with lung disease (18–20). In these preliminary studies, the ADCs in emphysematous lungs were increased relative to those in healthy lungs, which suggests that the ADC of helium can indeed provide quantitative information on the status of the pulmonary microstructure. The reproducibility of ADC measurements in healthy human subjects has also been demonstrated (17).
The purpose of our investigation was to quantitatively evaluate hyperpolarized 3He diffusion MR imaging of the lung in healthy volunteers and patients with emphysema and to determine whether ADCs measured with MR imaging correlate with spirometric values.
| MATERIALS AND METHODS |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
The patient group, comprising 11 subjects (eight men and three women; age range, 54–76 years; mean age, 66 years), was chosen from patients who were referred from our institutional pulmonary clinic and who had a clinical diagnosis of smoking-related chronic obstructive pulmonary disease (n = 10) or -1–antitrypsin deficiency (n = 1). Additional criteria for inclusion were that the subject did not have any other pathologic lung conditions, including lung cancer (verified with chest radiography or CT), or prior pulmonary surgery. In addition, only subjects with an abnormal FEV1 of less than 1.5 L and an FEV1/FVC ratio of less than 0.70 were included.
In 10 of the 11 subjects, chest radiographs depicted hyperinflation, reduction in the number and size of pulmonary vessels, and obvious bullous changes, all of which are indicative of emphysema. Four of the 10 patients also had CT scans, which showed changes caused by emphysema that manifested as heterogeneity of the lung tissues with areas of low attenuation, bullae, and paucity of the vessels. Three patients also underwent nuclear ventilation-perfusion MR imaging, which demonstrated diffuse matched defects and air trapping as typically seen in cases of emphysema. Hyperinflation was not seen at chest radiography in the patient with -1–antitrypsin deficiency. At nuclear ventilation-perfusion MR imaging in this patient, however, extensive matched areas of decreased ventilation and perfusion involved both lower lung zones. These findings were assumed to be caused by emphysema, considering the underlying clinical disease.
All experiments were performed on the basis of a physician’s Investigational New Drug application (IND 57,866) for imaging with hyperpolarized 3He with a protocol approved by our institutional review board. All subjects gave written informed consent before they participated in the study. Throughout each study, the subject’s heart rate and oxygen saturation level were monitored (Omni-Track Vital Signs Monitoring System, model 3100; Invivo Research, Orlando, Fla), and all studies were supervised by a radiologist (T.A.A., E.E.d.L.) or pulmonologist (J.D.T.).
3He Polarization and Delivery
The 3He gas was polarized by means of collisional spin exchange with an optically pumped rubidium vapor by using a commercial system (model 9600 Helium Polarizer; Nycomed Amersham Imaging, Durham, NC). After polarization, the gas was transferred to a plastic bag (Tedlar; Jensen Inert Products, Coral Springs, Fla), transported to the imager, and inhaled by the subject. The 3He polarization levels ranged from 15% to 35%.
MR Imaging
The MR studies were performed with a commercial 1.5-T whole-body imager (Magnetom Vision; Siemens Medical Systems, Iselin, NJ) modified to operate at the 3He resonant frequency of 48 MHz by adding a broadband radio-frequency amplifier and a flexible 3He chest radio-frequency coil (IGC Medical Advances, Milwaukee, Wis). Two sets of 3He MR images were collected in each subject. Each image set was acquired during suspended respiration, immediately after inhalation of 0.35 L of polarized 3He diluted to a total volume of 1.0 L with a filler gas of N2 or helium 4 (4He). The maximum breath-hold duration was less than 20 seconds. During the first breath hold, a set of coronal 3He ventilation MR images was obtained that covered the entire lung volume by using a gradient-echo fast low-angle shot, or FLASH, pulse sequence with the following typical parameters: repetition time msec /echo time msec, 9/3.7; flip angle, 10°–15°; matrix, 100 x 256; field of view, 38 x 50 cm; section thickness, 10 mm. With use of these MR images as a guide, a subset of sections for ADC measurements was selected by two of the authors (M.S., J.P.M.) in consensus.
When multiple ADC images were acquired, they were distributed approximately evenly across the lung volume. During the second breath hold, the MR images required to calculate an ADC image for each selected section were acquired with a diffusion-sensitized fast low-angle shot pulse sequence with the following parameters: 16/6; flip angle, 9°; matrix, 80–100 x 256; field of view, 38–45 x 50–60 cm; section thickness, 10–20 mm. Diffusion sensitization in the fast low-angle shot pulse sequence was achieved by adding gradients in the readout (superior-to-inferior) direction. The duration of a comfortable breath hold was determined for each subject before the imaging session. The number of coronal ADC images to be acquired was calculated by dividing this breath-hold duration by the acquisition time per image (range, 2.6–3.2 or 5.1–6.4 seconds per section, corresponding to two or four b values, respectively), which yielded between one and seven MR images for our subjects.
In the first 12 studies, four b values (0.4, 0.8, 1.2, and 1.6 sec/cm2) were used to generate each ADC image. However, the relatively long time required to acquire each section (range, 5.1–6.4 seconds) permitted only a small number of sections (three or fewer) to be acquired in these subjects. With the aim of reducing the acquisition time per section, we compared the ADC parameters calculated with two and four b values, wherein the lowest and highest b values were used for the calculations with two b values. Because there were only minor differences between the parameters with two and four b values (Results) for the first 12 subjects, all subsequent studies were performed with two b values (0.4 and 1.6 sec/cm2; range, 2.6–3.2 seconds for each section) to permit greater anatomic coverage for a given breath-hold duration.
The filler gas used for the inhalation was 4He for the first 10 volunteers and seven patients; N2 was used for the remaining subjects. The change in filler gas was made by the manufacturer of the polarization system in conjunction with an equipment upgrade. Because a different filler gas could potentially affect the measured ADC of 3He, we performed a comparison study in two of the 16 volunteers. In these volunteers, ventilation and ADC images were obtained as described previously, with N2 as the filler gas. In the same imaging session, a second set of ADC images were acquired with 4He as the filler gas. To further investigate the effect of different filler gases on the ADC measurement, the mean ADCs for the subset of the 16 volunteers who were aged 45 years or younger and received 4He (six subjects, four men and two women; age range, 23–42 years; mean age, 33 years) were compared with those for the subset of age-matched volunteers who received N2 (eight subjects, five men and three women; age range, 21–41 years; mean age, 27 years).
MR Imaging Data Analysis
The image data were transferred from the MR imager to a separate computer workstation (Ultra 10; Sun Microsystems, Palo Alto, Calif) for analysis with postprocessing software (MATLAB; MathWorks, Natick, Mass). By using the diffusion-sensitized 3He MR images that corresponded to each anatomic section, ADC images were calculated by means of linear least squares fitting of the natural log of the signal intensity versus the b value on a pixel-by-pixel basis. To eliminate bias from noise on the magnitude-reconstructed diffusion-sensitized MR images, any pixel location whose signal intensity was less than 2.5 times the SD of the background noise, on any of the MR images, was set to zero. The signal intensities for all remaining pixels on the MR images were corrected for the noise bias with the following formula: SIC = (SI2 - 2), where SIC is the corrected signal intensity, SI is the original signal intensity, and is the SD of the background noise (21).
To eliminate bias from the ADC parameters calculated in the large airways (which demonstrated ADCs substantially larger than those for the lung parenchyma), one of the authors (M.S.) manually segmented the trachea and main bronchi from the diffusion-sensitized MR images before analysis. For each calculated ADC image, histograms of the ADCs were created, and the mean and SD of the ADCs were calculated. The mean values for the individual subjects (I) and the SD values for the individual subjects (SDI) were determined by averaging the corresponding values from all sections. Similarly, the mean values for all subjects in a group (G) and the SD values for all subjects in a group (SDG) for the patient and volunteer groups were calculated by averaging the corresponding values for all subjects in each group. The G and SDG values are reported as the mean of the corresponding subject values plus or minus 1 SD of this mean value.
To assess regional variations in ADCs between the upper and lower portions of the lung, the coronal ADC images were divided into a superior and an inferior half, and the mean ADC was calculated for each half. The images were divided into superior and inferior halves as follows: In the anterior and posterior sections, the MR images were divided exactly in half, whereas in the central sections the MR images were divided at the distal termination of the main stem bronchi. To calculate the group mean difference between the upper and lower portions of the lung (UL) for the patient and volunteer groups, the mean of the inferior half was subtracted from that of the superior half for each ADC image, after which these differences between halves were averaged over all subjects within each group.
Spirometric Evaluation
Spirometry (model PB100; Puritan Bennett, Lenexa, Kan) was performed in all subjects, with use of the Knudson 1983 reference tables for predicted normal limits (22), immediately before imaging. The functional indexes that were recorded for correlation analysis included the percentage of predicted FEV1 and FEV1/FVC.
Statistical Analysis
For the comparison of ADC parameters calculated with two or four b values in the 12 subjects who underwent imaging with four b values, I values calculated from two b values were compared with those calculated from four b values with a paired two-tailed t test, and the corresponding SDI values were compared with an F test. The I values for the subset of volunteers who were aged 45 years or younger and received 4He were compared with those for the subset of age-matched volunteers who received N2 with an unpaired two-tailed t test. The G value for the volunteers was compared with that for the patients by using an unpaired one-tailed t test, and the corresponding SDG values were compared by using an F test. The UL value for volunteers was compared with that for patients by using an unpaired one-tailed t test. Pearson correlation coefficients were calculated between the I and SDI values and the corresponding percentage of predicted FEV1 and FEV1/FVC values. These coefficients were calculated both for the volunteer and patient groups individually and for the two groups combined. For the volunteer and patient groups combined, linear least squares regression lines with 95% CIs were calculated for the I and SDI values and the corresponding percentage of predicted FEV1 and FEV1/FVC values.
| RESULTS |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
|
In the two volunteers imaged with both N2 and 4He as a filler gas, the mean ADCs for N2 were 5% less than those for 4He. Furthermore, the I values for the subset of volunteers who were aged 45 years or younger and received 4He were not statistically different from those for the subset of age-matched volunteers who received N2. These results suggest that the change of filler gas was not likely to affect the results of this study.
In volunteers, the 3He gas was uniformly distributed throughout the lung parenchyma on the ventilation MR images (Fig 1, A). The ADC images were also homogeneous, with values that typically ranged between 0.16 and 0.30 cm2/sec (Fig 1, B). The ADC histograms (Fig 1, C) demonstrated mean values that were low compared with those for the patients (Fig 2) and were narrow, corresponding to low SDI values, presumably owing to the presence of many small evenly sized air spaces in healthy lung.
|
|
|
|
|
|
|
|
| DISCUSSION |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
The ADCs probably depend on the underlying length scales, surface-to-volume ratio, and tortuosity of the microstructure of the distal airway structures in a very complex manner. The tissue destruction that occurs in emphysema should cause an increase in the length scales of the airways and a decrease in both their surface-to-volume ratio and tortuosity. Each of these changes should, in theory, cause an increase in the ADCs in emphysematous regions of the lung. Nonetheless, owing to the complexities of these structures, the dependence of the ADCs on the underlying microstructure is of a statistical nature and is thus not easily ascertained with direct measurements or by using any current mathematic models. Furthermore, the ADCs depend on the timing parameters of the pulse sequence for the range of diffusion times that can be probed. This dependence is complex and poorly understood for diffusion imaging with hyperpolarized gases in complex structures.
We found in our study that the mean ADC for the group of volunteers was comparatively low at 0.23 cm2/sec. This value is in the same range as that obtained by Chen et al in guinea pigs (23) and by Saam et al in healthy humans (18,20). The ADC images obtained in the volunteers showed generally uniform values, as reflected by the comparatively narrow ADC histograms, and there was not a significant difference between the ADCs in the upper and lower regions of the lung.
The ADCs for patients were significantly larger than those for volunteers, which is probably due to the increased size and altered morphology of the distal air spaces of the lung. Saam et al (18,20) found that the mean ADC in patients with severe emphysema was increased by a factor of 2.5. In our study, which involved patients with disease of varying severity, the overall mean ADC was a factor of 2.0 greater than that for volunteers. Furthermore, the increase in the variance of the ADCs with emphysema was found to be significant in our study, consistent with regional variations in ADCs that were evident during visual inspection of the ADC images.
The heterogeneity in ADCs for the individual patients most likely resulted from the variations in distribution of the underlying disease, and in all likelihood the variability in the ADCs among the patients reflected the differences in disease severity. The destruction from centrilobular emphysema, which is the predominant form of emphysema in smokers, is generally greater in the apices of the lung than in the bases. We found, in the patients with smoking-related emphysema, a significant increase in the mean ADC for the upper portion of the lung compared with that for the lower portion; this finding suggests enlargement and morphologic alteration of the distal lung structures (ie, terminal bronchi and alveoli) of this region. Some of the patients had more heterogeneous regions of increased ADCs; nonetheless, the largest ADCs were in the upper portions of the lung. This pattern is consistent with the progression of centrilobular emphysema, which affects the lung more globally as the disease increases in severity. Panlobular emphysema, which is characteristic of emphysema in -1–antitrypsin deficiency, affects the whole lung without a predominance of disease in the upper lobes.
The correlations that were demonstrated between ADC parameters and spirometric indexes support our conjecture that the variability in the ADCs may reflect differences in the severity of emphysema. FEV1/FVC showed the greatest correlation with the ADC parameters, although the percentage of predicted FEV1 was also highly correlated. FEV1/FVC is believed to be more sensitive than percentage of predicted FEV1 for the early detection of emphysema, but the latter is thought to be a better indicator of disease severity. To our knowledge, this is the first study performed to compare 3He ADC results with another measure of pulmonary function. Because spirometry is used clinically to assess the severity of emphysema, it is an important finding that the ADCs showed similar trends. Spirometry is not very sensitive to changes in small airway resistance and, thus, to emphysema that is not yet clinically evident. Since ADCs provide an indirect measurement of the size and morphology of the distal airways, they may be more sensitive than spirometry for detecting changes in these airway structures. Further study is needed in individuals who potentially have emphysema but who have normal spirometric findings. In addition, ADC images provide a regional assessment of parenchymal disease, which is not available from spirometric measurements.
An advantage of ADC imaging over ventilation imaging is that the former is not biased by regional differences in the spin density, T1, or coil sensitivity, and does not depend on the absolute signal intensity. This is because the ADC is calculated on the basis of a ratio of signal intensities from MR images that differ only in their diffusion weighting. On the other hand, a potential limitation of ADC imaging is that information can be obtained from only regions of the lung that are ventilated with sufficient hyperpolarized 3He; in regions that are very poorly ventilated due to severe air-flow obstruction, there may not be sufficient signal to determine the ADC. This situation could introduce a bias in the mean or SD of the ADCs. If sufficient hyperpolarized 3He does not flow into regions with substantial parenchymal damage, the calculated ADC parameters may underestimate the degree of disease. Nonetheless, if the regions that are sufficiently ventilated for ADC measurements correlate with the functional portions of the lung, then ADC parameters may correlate better with lung function than does thin-section CT, which assesses both ventilated and nonventilated parenchyma. In addition, although thin-section CT can provide structural information about areas that are not ventilated, abnormal ventilation can only be inferred from the CT images, on the basis of findings such as atelectasis or air trapping.
Owing to breath-hold limitations in the patients with severe emphysema, only a relatively small number of ADC image sections (one to four) could be acquired with the current pulse sequence. We are developing more rapid pulse sequences that will enable measurement of the ADC in contiguous 1-cm sections that cover the whole lung in a single breath hold shorter than 10 seconds. Such a technique would permit evaluation of the entire lung. This would be advantageous for determining the overall involvement, particularly since the severity of emphysema can vary considerably; thus, the information obtained from only a limited number of sections may not be representative of the overall disease extent.
Another limitation of this study is the lack of a standard of reference for the diagnosis of emphysema. One of the problems in evaluating emphysema is that its definition is based on histologic findings; thus, histologic examination is the only true standard of reference. CT investigations have shown that thin-section CT provides an assessment of emphysema that agrees well with findings at histologic examination. However, thin-section CT is not used in routine clinical evaluation of patients with chronic obstructive pulmonary disease. In future studies, findings with 3He ADC imaging should be compared prospectively with those with both thin-section CT and spirometry.
The difference in the age distribution for the volunteers compared with that for the patients is also a limitation of this preliminary study. It is reasonable to expect that measurable changes in the ADC parameters may occur with age, although we anticipate that these changes will be small compared with those seen in clinically apparent emphysema. Nonetheless, in future studies, it will be important to quantify any age-related trends in the ADCs so that these changes can be distinguished from those that may occur in early emphysema.
Hyperpolarized 3He diffusion MR imaging appears to be capable of depicting structural changes in the distal airways, which would allow assessment of the extent and severity of emphysema. The ADC parameters for our group of volunteers were significantly different from those for our group of patients; the ADC parameters also correlated with findings at spirometry. This technique may provide insight into the early development and progression of changes in the lung that occur in emphysema. In addition, the regional specificity provided by this technique might be important for presurgical planning or for assessing drug regimens that are targeted at reversing microstructural damage.
| ACKNOWLEDGMENTS |
|---|
| REFERENCES |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|