| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 All authors: Department of Abdominal Imaging and Intervention, Massachusetts General Hospital and Harvard Medical School, 55 Fruit St., White 270, Boston, MA 02114.
Received August 15, 2004;
accepted after revision January 12, 2005.
This study was partly sponsored by Advanced Magnetics Inc., Cambridge, MA,
which provided materials for the study.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Seventy-seven patients (58 men, 19 women) with proven primary cancer (bladder [n = 20], breast [n = 10], endometrial [n = 1], renal [n = 3], penile [n = 4], prostate [n = 31], rectal [n = 1], testicular [n = 5], and ureteral [n = 2]) who were scheduled for surgical lymph node dissection were enrolled in the study. In these patients, 169 lymph nodes (mean size, 11.2 mm) were evaluated on T2*-weighted gradient-refocused echo MRI at l.5 T both before and 24-36 hr after the IV administration of ferumoxtran-10 (2.6 mg Fe/kg). Two blinded reviewers with differing levels of interpreting experience separately performed qualitative image evaluation. A 6-point scale was used to characterize lymph nodes on contrast-enhanced images alone and on combined unenhanced and contrast-enhanced images. Receiver operating characteristic (ROC) analysis was performed separately for both reviewers.
RESULTS. Of the 169 lymph nodes evaluated, 55 were benign and 114 malignant by histopathologic analysis. The results of the ROC analysis comparing contrast-enhanced images ([Az = area under ROC curve] reviewer 1, Az = 0.92; reviewer 2, Az = 0.94) alone with combined unenhanced and contrast-enhanced images (reviewer 1, Az = 0.94; reviewer 2, Az = 0.93) showed a statistically significant difference (p = 0.01) for reviewer 1 but no difference for reviewer 2 (p = 0.88). Reviewer 2 was more experienced in interpreting ferumoxtran-10-enhanced images than reviewer 1.
CONCLUSION. On ferumoxtran-10-enhanced MR lymphangiography, contrast-enhanced images alone may suffice for lymph node characterization. However, a certain level of interpretation experience may be required before contrast-enhanced images can be used alone.
Keywords: contrast media • ferumoxtran-10 • lymph nodes • MR lymphangiography • oncologic imaging
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
However, obtaining the reported high level of accuracy with this new technique of MR lymphangiography requires the standard practice of performing two separate MRI examinations—unenhanced and contrast enhanced—with the latter performed 24 hr after the administration of ferumoxtran-10. This technique is noninvasive and accurate, but adhering to such a two-part study may affect patient compliance because the patient must return for contrast-enhanced MRI and also incurs more cost than with a single MRI examination. To date, to our knowledge, no study has compared the relative accuracy of the contrast-enhanced study alone with that of the combined unenhanced and contrast-enhanced study. The purpose of this study was to determine if a single MRI examination performed 24 hr after ferumoxtran-10 administration can be used for accurate lymph node characterization.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Patients
The patients included in this study were part of a larger extended phase 3
clinical trial evaluating the use of ferumoxtran-10 for lymph node staging in
patients with various primary cancers. The present study included 77 patients
(58 men, 19 women; mean age, 60.3 years; range, 28-84 years) with proven
primary cancer (bladder [n = 20], breast [n = 10],
endometrial [n = 1], renal [n = 3], penile [n = 4],
prostate [n = 31], rectal [n = 1], testicular [n =
5], and ureteral [n = 2]) who were scheduled for either surgical
lymph node dissection or imaging-guided lymph node biopsy. Patient inclusion
criteria for the larger phase 3 trial included age older than 18 years, a
confirmed primary cancer, and the possibility of nodal metastases.
Ferumoxtran-10 Administration
Ferumoxtran-10 was provided as a lyophilized powder consisting of
ultrasmall superparamagnetic iron oxide nanoparticles covered with
low-molecular-weight dextran. The contrast material was reconstituted using 10
mL of normal saline, after which a weight-adjusted dose (2.6 mg Fe per
kilogram of body weight) was withdrawn and diluted with 50 mL of saline and
infused through a 5-µm filter at a rate of 4 mL/min.
MRI
MRI was performed at 1.5 T (System 9X, GE Healthcare) with region-specific
phased-array coils. Identical sequences were obtained before and 24-36 hr
after the administration of ferumoxtran-10. The pulse sequences performed
included T2-weighted fast spin-echo (TR/TE, 4,500/80; flip angle, 90°;
field of view, 24-28 cm; slice thickness, 3 mm; matrix, 256 x 256;
number of excitations, 3; average acquisition time, 4.2 min),
T2*-weighted gradient-echo (TR/TE, 2,100/24; flip angle, 70°;
field of view, 26-28 cm; slice thickness, 3 mm; matrix, 160 x 256;
number of excitations, 2; average acquisition time, 6.4 min), and T1-weighted
gradient-refocused echo (GRE) sequences obtained in different anatomic planes
(TR/TE, 175/1.8; flip angle, 80°; field of view, 22-30 cm; slice
thickness, 4 mm; matrix, 128 x 256; number of excitations, 1; average
acquisition time, 22 sec).
Care was taken to ensure that all relevant locoregional nodal drainage areas were included in the imaging. All sequences were performed in the standard axial plane, which allowed consistent inclusion of all local and regional lymph nodes. For T1-weighted GRE sequences, additional coronal and oblique images were also obtained to localize the nodes in relation to vascular anatomic landmarks.
These listed imaging sequences and parameters were optimized to reduce motion artifacts, maximize signal-to-noise ratio, and provide diagnostically useful images of the pelvis, abdomen, and chest within clinically acceptable time limits. In particular, when scanning the abdomen and lower chest, respiratory triggering was used for the T2-weighted fast spin-echo sequences, respiratory gating for the T2*-weighted GRE sequences, and breath-holding for the T1-weighted GRE sequences. To further reduce respiratory artifacts from anterior abdominal wall motion, anterior saturation bands were placed on the subcutaneous fat.
Image Analysis
The images were analyzed by two radiologists who were blinded to the
histopathology results and were not involved in patient enrollment, scanning,
or data gathering. The images were selected and organized by a radiologist who
was not involved in the image analysis. The first step in image selection was
the identification of pathologically correlated local and regional lymph nodes
in each patient. Once the pathologically correlated nodes were identified,
three axial T2*-weighted GRE images for each node were selected on
the unenhanced and contrast-enhanced scans. Three images of the node were
selected to avoid partial volume artifacts. Reviewer 1 had recently completed
a fellowship in abdominal imaging that included occasional discussion of
ferumoxtran-10-enhanced MRI. Reviewer 2 had worked with iron oxide MRI
contrast agents for 18 years and had 6 years of experience interpreting
ferumoxtran-10-enhanced MR images of lymph nodes. The reviewers assessed the
images independently. All images were analyzed on an Advantage Windows
workstation (version 4.0, GE Healthcare). The images were reviewed in two
sessions. To limit learning bias, the interval between sessions was at least 8
weeks, and the images were randomly assigned to each reviewer and to each
session.
In the first session, the reviewers interpreted only the ferumoxtran-10-enhanced T2*-weighted GRE images. The reviewers were provided three axial images for each lymph node, one at the level of the node and one above and one below the actual image. Each reviewer assigned each node a confidence rating based on a 6-point scale as follows: 1, definitely benign; 2, most likely benign; 3, probably benign; 4, probably malignant; 5, most likely malignant; and 6, definitely malignant. A 6-point scale was used so that the reviewer evaluations could all be classified as either benign or malignant. Published criteria for lymph nodes enhanced with ferumoxtran-10 (Table 1) were used for differentiating benign from malignant nodes [1].
|
In the second session, the reviewers interpreted the combined unenhanced and the ferumoxtran-10-enhanced T2*-weighted GRE images. Again, each node was scored by the reviewers on the basis of the 6-point confidence scale.
Statistical Analysis
With the use of receiver operating characteristic (ROC) evaluation software
(ROCKIT, version 0.9B; Charles Metz), the diagnostic accuracy of the
contrast-enhanced images alone versus the combined unenhanced and
contrast-enhanced images was determined for each reviewer by calculating the
area under the ROC curve (Az). Differences between ROC
curves were tested for significance (p 
0.05) using the
two-tailed area test for paired data. According to the ROCKIT Users'
Guide [12], the program
performs a univariate z-score test of the difference between the
areas under the two ROC curves.
In addition, because the reviewers used a 6-point scale, the primary efficacy parameters of sensitivity and specificity were also calculated for each reviewer for the contrast-enhanced images alone and for the combined data set by collapsing the 6-point interpretations into the binary categories benign or malignant.
For both the contrast-enhanced interpretation and the combined unenhanced
and contrast-enhanced interpretation, responses of the two reviewers were
compared using kappa analysis for ordinal data. Kappa values were used as
measures of agreement between each participant and the consensus beyond that
expected by chance [13,
14]. Interobserver agreement
was rated as follows: slight, 
< 0.21; fair, = 0.21-0.40;
moderate, = 0.41-0.60; substantial, = 0.61-0.80; and almost
perfect, = 0.81-1.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The ROC curves formed on the basis of recorded data of the two reviewers are shown in Figures 1A and 1B. For reviewer 1, with fewer years of experience in interpretation of ferumoxtran-10-enhanced MRI, the area under the curve for the contrast-enhanced images alone (Az = 0.92) was statistically significantly different (p = 0.01) from the area under the curve for combined unenhanced and contrast-enhanced images (Az = 0.94). For reviewer 2, with more years of experience in interpreting ferumoxtran-10-enhanced MRI, the area under the curve for the contrast-enhanced images alone (Az = 0.94) was not statistically significantly different (p = 0.88) from the area under the curve for the combined unenhanced and contrast-enhanced images (Az = 0.93).
|
|
The calculated sensitivity, specificity, and positive and negative predictive values for each reviewer are shown in Table 2.
|
Evaluating contrast-enhanced images alone, reviewers 1 and 2 gave the same score in 152 lymph nodes and disagreed by a difference of only one ordinal in 17 lymph nodes. This produced a kappa value of 0.42 (fair agreement). In evaluating the combined unenhanced and contrast-enhanced images, reviewer 1 agreed with reviewer 2 in 155 lymph nodes, and they disagreed by one ordinal in 13 lymph nodes. This produced a kappa value of 0.48, fair but improved agreement.
In two patients, minor back pain was seen after the start of ferumoxtran-10 infusion that disappeared immediately after the infusion was stopped. When infusion was restarted a few minutes later, the symptoms did not reappear. No medical treatment was needed.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our results show that an experienced reviewer can achieve high accuracy for nodal characterization without the addition of unenhanced images, whereas an inexperienced reviewer benefits slightly by having the unenhanced images (Figs. 2A and 2B). However, even without them, both reviewers are highly accurate, and specific reviewer training might ameliorate this difference. We propose, therefore, that ferumoxtran-10-enhanced lymph node imaging can be performed efficaciously without unenhanced images. The contrast material can be injected in the patient at a site remote from the MR suite, and the patient can return the next day for a single MRI examination. That examination should definitely include T1-weighted, T2-weighted, and T2*-weighted GRE images. The T1-weighted images provide high spatial resolution and are important for identifying lymph nodes in relation to enhanced vessels. The T2-weighted images also allow node identification. The T2*-weighted GRE images display the contrast effect, with uniformly dark nodes implying ferumoxtran-10 uptake and benignity [15]. Nodes that have not turned dark, or that contain islands of nodal tissue that have not turned dark, indicate node replacement by tumor.
|
|
The process of lymph node diagnosis for staging in cancer patients involves two separate but related steps, node identification and node evaluation. Lymph nodes must first be differentiated on cross-sectional images from other structures having an oval cross-section before they can be evaluated as benign or malignant. These other structures predominantly include vessels, strands of seminal vesicles, sigmoid diverticula, hematomas, and other features that could be mistaken for lymph nodes. This differentiation process is facilitated by multiple MRI sequences that have different native contrast characteristics. On unenhanced T2-weighted images, most lymph nodes are hypointense to surrounding fat. Although T1-weighted images are often promoted as providing visualization of anatomy, we have found that T2-weighted images are as useful as T1-weighted images for the detection of lymph nodes. Recognition of some lymph nodes may be impaired when only contrast-enhanced images are available. However, the nodes that might go unrecognized likely would be those that have become hypointense after ferumoxtran-10 administration. Because these nodes are benign, failing to detect them would not lead to serious errors in interpretation that would affect patient staging or therapeutic planning. We did not directly test this hypothesis because the reviewers were presented lymph nodes clearly identified as such for their determination of benignity or malignancy.
Nodal infiltration by tumors is not an all-or-nothing phenomenon. Some nodes present with only part of the node replaced by tumor, resulting in focal defects that can be detected as hyperintense only after the residual normal nodal tissue has taken up ferumoxtran-10 and dropped in signal intensity. Because many nodes contain a fatty central hilum that will also remain hyperintense, lymph node evaluation with ferumoxtran-10 requires careful scrutiny of T1-weighted images to identify a fatty central hilum. Heavily T1-weighted images remain unaffected by ferumoxtran-10, so these images can be obtained after administration of ferumoxtran-10 and do not require a separate unenhanced imaging session.
Our study is limited by the fact that the reviewers were presented with images that had the nodes clearly identified. Thus, the true difference in accuracy that takes into account node detection before characterization was not assessed. Our present study was a preliminary analysis to establish the usefulness of obtaining unenhanced images in addition to contrast-enhanced images, and because our results encourage the use of contrast-enhanced imaging alone, we are undertaking a larger study in which the reviewers will be presented with entire data sets to assess the true difference in accuracy.
In conclusion, contrast-enhanced images alone after ferumoxtran-10 enhancement may suffice for lymph node characterization. However, a certain level of interpretation experience may be required before contrast-enhanced images can be used independently of the unenhanced images. Ferumoxtran-10 is currently not approved in the United States and is under evaluation by the U.S. Food and Drug Administration.
Acknowledgments
We thank Elkan Halpern for statistical advice and Peter Mueller, Jack
Wittenberg, and Paula Jacobs for reviewing the manuscript and providing useful
suggestions.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  X. M. Zhang, H. L. Zhang, D. Yu, Y. Dai, D. Bi, M. R. Prince, and C. Li 3-T MRI of Rectal Carcinoma: Preoperative Diagnosis, Staging, and Planning of Sphincter-Sparing Surgery Am. J. Roentgenol., May 1, 2008; 190(5): 1271 - 1278. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F E Moron and J Szklaruk Learning the nodal stations in the abdomen Br. J. Radiol., October 1, 2007; 80(958): 841 - 848. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. L. Speight and M. Roach III Advances in the Treatment of Localized Prostate Cancer: The Role of Anatomic and Functional Imaging in Men Managed With Radiotherapy J. Clin. Oncol., March 10, 2007; 25(8): 987 - 995. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Saksena, M. Harisinghani, P. Hahn, J. Kim, A. Saokar, B. King, and R. Weissleder Comparison of lymphotropic nanoparticle-enhanced MRI sequences in patients with various primary cancers. Am. J. Roentgenol., December 1, 2006; 187(6): W582 - W588. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
