Published online before print December 4, 2007, 10.1148/radiol.2461061907
(Radiology 2007;246:596.)
A more recent version of this article appeared on December 1, 2007
© RSNA, 2007
Nodal Metastasis in Non–Small Cell Lung Cancer: Accuracy of 3.0-T MR Imaging1
Ha Young Kim, MD,
Chin A Yi, MD,
Kyung Soo Lee, MD,
Myung Jin Chung, MD,
Yoon Kyung Kim, MD,
Bong Keun Choi, PhD,
Hojoong Kim, MD, and
O Jung Kwon, MD
1 From the Department of Radiology and Center for Imaging Science (H.Y.K., C.A.Y., K.S.L., M.J.C., Y.K.K.) and Division of Pulmonary and Critical Care Medicine, Department of Medicine (H.K., O.J.K.), Samsung Medical Center, Sungkyunkwan University School of Medicine, 50, Ilwon-Dong, Kangnam-gu, Seoul 135-710, Korea; and Department of Biostatistics, Kyunghee University College of Medicine, Seoul, Korea (B.K.C.). Received November 7, 2006; revision requested January 12, 2007; revision received January 17; accepted February 28; final version accepted June 1. Supported by Samsung Medical Center Clinical Research Development Program grants (CRDP CRS105-22-1 and CRS 106-11-1).
Address correspondence to K.S.L. (e-mail: kyungs.lee{at}samsung.com).
 |
ABSTRACT
|
|---|
Purpose: To prospectively evaluate the diagnostic accuracy of 3.0-T magnetic resonance (MR) imaging in the detection of non–small cell lung cancer nodal metastasis, with histopathologic analysis as the reference standard.
Materials and Methods: Institutional review board approval and informed consent were obtained. From July 2005 to May 2006, 113 patients (91 men, 22 women; age range, 34–82 years; mean age, 61 years) with non–small cell lung cancer underwent thoracic 3.0-T MR imaging followed by surgery or mediastinoscopy. The lymph node–to-tumor ratios (LTRs) of signal intensity and nodal morphologic characteristics (such as eccentric cortical thickening or obliteration of the fatty hilum) were assessed on T2-weighted triple-inversion black-blood fast spin-echo images. Nodal short-axis diameter was assessed on T1-weighted three-dimensional fast field-echo images. Receiver operating characteristic and multivariate logistic regression analyses were used for statistical evaluation.
Results: The cutoff value (LTR > 0.84) proved to be most appropriate (area under the receiver operating characteristic curve = 0.735, P < .001) in the detection of a nodal metastasis. Of the various parameters examined, morphologic characteristics appeared to be the most significant (P < .001) parameters for depicting a malignant node (multivariate logistic regression analyses; odds ratio, 7.5). Nodal morphology was analyzed, and diagnostic sensitivity, specificity, and accuracy were 53% (39 of 74 nodal stations), 91% (453 of 496 nodal stations), and 86% (492 of 570 nodal stations), respectively.
Conclusion: Morphologic details of lymph nodes on T2-weighted triple-inversion black-blood fast spin-echo MR images are significant for detection of mediastinal or hilar nodal metastasis at 3.0-T MR imaging.
© RSNA, 2007
|
INTRODUCTION
|
|---|
Non–small cell lung cancer accounts for 75%–80% of all lung cancers and is the leading cause of death from cancer worldwide (1). Preoperative tumor staging in patients with non–small cell lung cancer is important when selecting those patients with localized disease who are most likely to benefit from surgical resection. Although computed tomography (CT) is widely used in the preoperative evaluation of tumor size and adjacent structure invasion, the results of many studies have shown that CT has limited accuracy in the evaluation of nodal status because it yields only presumptive evidence of metastatic disease that is based on size criteria (2–6). Overall, CT has a sensitivity of 41%–68% and a specificity of 43%–97% in the detection of mediastinal nodal metastasis, neither of which is optimal for clinical decision making (2,4,5,7,8). Study results demonstrate that positron emission tomography (PET) and integrated PET/CT are consistently more accurate than CT in the detection and exclusion of nodal disease, with a mean sensitivity of approximately 80% and a mean specificity of 90% (9,10).
Reports have indicated the ability of magnetic resonance (MR) imaging to reveal mediastinal invasion by a tumor and to help identify hilar and mediastinal nodal metastases (1,11–17). Signal-to-noise ratio in MR imaging is directly related to static magnetic field strength. An increased signal-to-noise ratio achieved by means of adopting a high (3.0-T) magnetic field strength will increase spatial resolution, shorten examination time, and reduce section thickness (18). However, high magnetic field strength is associated with increased susceptibility artifacts and magnetic field distortions.
The fast spin-echo sequence (also called the echo-train spin-echo sequence) is one of many techniques that have been proposed to reduce susceptibility artifacts and increase signal-to-noise ratio (19). This sequence is based on rapid repetitive rephasing as a result of the train of multiple 180° refocusing pulses and gradients, particularly with short echo spacing. Short echo spacing makes the pulse sequence relatively resistant to inhomogeneous magnetic susceptibility because of the reduction in T2* effects. Thus, fast spin-echo sequences appear to be suitable for thoracic MR imaging. Several techniques that involve the use of these fast spin-echo sequences have been developed and used effectively for thoracic imaging. These techniques include the (a) short inversion time inversion-recovery sequence for fat suppression with an enhanced conspicuity of focal lung lesions (20–23), (b) half-Fourier single-shot fast spin-echo sequence for breath-hold cardiac-gated T2-weighted imaging (24), and (c) T2-weighted fast spin-echo and short inversion time inversion-recovery sequences with use of black-blood techniques to produce images that are insensitive to flow artifacts (25). However, to our knowledge, these refined pulse sequences based on fast spin-echo applications have not been used under 3.0-T MR conditions. Thus, the purpose of our study was to prospectively evaluate the diagnostic accuracy of 3.0-T MR imaging in the detection of non–small cell lung cancer nodal metastasis, with histopathologic analysis as the reference standard.
|
MATERIALS AND METHODS
|
|---|
This study was performed in full accordance with the guidelines of the institutional review board at Samsung Medical Center. Written informed consent was obtained from all patients. The patients understood that the surgeons would be aware of the routine preoperative imaging study findings but not the 3.0-T MR findings that were being investigated in this study.
Patients
From July 2005 to May 2006, 132 consecutive patients with histopathologically proved non–small cell lung cancer who were not suspected of having nonnodal extrathoracic metastatic lesions at chest CT (with intravenous injection of 100 mL of iopamidol [Iomeron 300; Bracco, Milan, Italy] from the thoracic inlet to the middle poles of kidneys) underwent thoracic MR imaging with a 3.0-T system (Achieva; Philips Medical Systems, Best, the Netherlands). The histopathologic cell types of the primary tumors in these 132 patients were confirmed with thoracotomy (n = 100), percutaneous needle biopsy (n = 23), or bronchoscopic biopsy (n = 9). Of these 132 patients, 19 were excluded because nodal staging was not performed for the following reasons: 11 patients were suspected of having extrathoracic metastasis at conventional staging (contrast material–enhanced CT of the chest and upper abdomen, isotope bone scanning, and brain imaging in selected patients with tumors that had a high tumor or nodal stage) or whole-body PET/CT, and eight underwent previous chemotherapy (n = 1) or chemotherapy and radiation therapy (n = 7). Thus, nodal stages were determined in 113 patients (91 men, 22 women; age range, 34–82 years; mean age, 61 years; weight range, 49–98 kg; mean weight, 66 kg ± 8 [standard deviation]) by means of thoracotomy (n = 100) or mediastinoscopy (n = 13) (Fig 1).
MR Imaging
The mean interval between MR imaging and surgical mediastinal nodal staging was 6.1 days (range, 1–22 days; median, 6 days). All MR examinations were performed with the aforementioned 3.0-T MR imager equipped with a gradient system that was capable of operating at an 80 mT/m maximum gradient amplitude, a 0.2-msec rise time, and a 200 T/m/sec slew rate. A sensitivity-encoding cardiac coil (Philips Medical Systems) with a six-coil element was used to obtain dedicated transverse thoracic images. No contrast agent was administered. The average MR examination time was less than 15 minutes.
A breath-hold T2-weighted triple-inversion black-blood fast spin-echo sequence was used with cardiac gating and the following parameters: repetition time msec/echo time msec, 1200–1800/60 (effective); two R-R intervals; echo train length, 21; 400-mm field of view; acquisition matrix size, 256 x 256 reconstructed to 512 x 512; 5-mm section thickness; and 1-mm intersection gap with transverse orientation. A double-inversion blood-nulling preparation pulse was applied at the R-wave trigger to suppress blood signals with an inversion delay of 697.6 msec followed by a third inversion pulse for fat suppression (spectral presaturation by means of inversion recovery). Two sections were acquired per breath hold, and 40–44 sections were needed to cover the entirety of the lungs. MR images were acquired with breath holding at end expiration; thus, 20–22 breath holds were of a similar lung volume, and this enabled us to avoid image overlap or skip between sections. The imaging time per breath hold of 16 seconds resulted in a total acquisition time of between 5 minutes 20 seconds and 5 minutes 52 seconds depending on the total number of sections. A sensitivity-encoding factor of two was applied in the phase-encoding direction to increase the acquisition speed.
A T1-weighted three-dimensional multishot fast field-echo sequence was used during breath holding. The sequence was optimized by using a fast field-echo factor of 50, 24 fast field-echo shots, and a radial direction of k-space filling with the following parameters: 3.00/1.49, 100° flip angle, two signals acquired, 400-mm field of view, and acquisition matrix size of 224 x 224 reconstructed to 512 x 512. Fat suppression was achieved by using a spectral presaturation attenuated by inversion recovery method with a 90-msec inversion delay. Four consecutive slabs without any gap were obtained in the foot-to-head direction, covering the entirety of the lungs during one breath hold per slab. The slabs were 66 mm thick, and each slab was divided into 11 partitions, resulting in 11 6-mm-thick transverse images. The imaging time per breath hold was 13.5 seconds, resulting in a total acquisition time of 54 seconds. A sensitivity-encoding factor of two was applied in the phase-encoding direction to increase the acquisition speed.
Image Analysis for Nodal Metastasis
All image analyses and measurements were performed prospectively by two chest radiologists (H.Y.K., C.A.Y.; each with 2 years of thoracic MR interpretation experience, including lung cancer staging, prior to the start of this study). They read the MR images together, and decisions were made prospectively in consensus before surgery.
Nodal metastasis was determined on a per–nodal station basis by assessing the signal intensity ratios of lymph nodes to primary tumors (lymph node–to-tumor ratio [LTR], a modification of methods used by Ohno et al [17], who measured lymph node–to-saline ratio), morphologic characteristics of the lymph nodes, and nodal short-axis diameter on T2-weighted triple-inversion black-blood fast spin-echo and T1-weighted three-dimensional fast field-echo images. First, on the T2-weighted triple-inversion black-blood fast spin-echo images, the signal intensity of the lymph nodes was measured from the circular or oval regions of interest (ROIs) drawn over the cortex of the lymph nodes, excluding the fatty hilum. The ROI covered nearly the full thickness of the nodal cortex. When the nodes had a thickened cortex with peripherally displaced or obliterated fatty hilum, the ROI covered more than two-thirds of the thickest cortex. The signal intensities of the primary tumors were recorded by using ROIs that covered between one-half and two-thirds of tumors away from areas of necrosis. The regions of postobstructive pneumonia were specifically avoided when measuring signal intensities by using ROIs. Two signal intensity measurements were obtained for each node and tumor, and the mean signal intensity values were used to calculate the LTR. LTR was calculated by dividing the signal intensity of the lymph node by the signal intensity of the tumor.
Second, to perform qualitative analysis, the morphologic features of lymph nodes were evaluated on T2-weighted triple-inversion black-blood fast spin-echo images. Because the cortex of normal lymph nodes had a uniform thickness that was less than half the diameter of the fatty hilum, eccentric (asymmetric) cortical thickening or obliteration of the fatty hilum was interpreted as being positive for metastasis (26,27). Third, the lymph nodes were evaluated in terms of their short-axis diameter on the T1-weighted three-dimensional fast field-echo images. Lymph nodes with a short-axis diameter of at least 10 mm were considered positive for metastasis. In addition, nodal staging was performed with these parameters on a per-patient basis.
Histopathologic Staging and Image Correlation
Mediastinal and hilar nodal metastasis status was determined in 100 patients in whom thoracotomy had been performed (lobectomy, n = 90; pneumonectomy, n = 10). In 13 additional patients, nodal metastases were detected with mediastinoscopy.
Tumor resection and extensive mediastinal lymph node dissection (thoracotomy) and mediastinoscopy were performed by one of two experienced thoracic surgeons (with 13–18 years of thoracic surgery experience) after they had considered the results of preoperative CT or PET/CT. However, the surgeons were unaware of the MR findings. During mediastinoscopy, the surgeons attempted to harvest entire nodes, and American Thoracic Society lymph node map areas of 2R, 4R, 2L, 4L, and 7 were routinely sampled. At thoracotomy, the surgeons sampled all visible and palpable lymph nodes that were accessible in the hilum and mediastinum according to our surgical protocol. Namely, all encountered lymph nodes were removed from American Thoracic Society lymph node map areas of 10R, 9, 8, 7, 4R, 3, and 2R for right lung tumors and from areas 10L, 9, 8, 7, 6, 5, and 4L for left lung tumors. When necessary, especially when imaging results suggested possible nodal metastasis in nodal stations 1 (highest mediastinal) or 2L (when tumor was located in the left lung), these nodes were also evaluated during mediastinoscopy or thoracotomy.
A lung pathologist with 11 years of experience evaluated the tumors (ie, histopathologic class, size, involvement of surrounding organ, necrosis, and distance from the resection margin) and lymph nodes (ie, location and number). The surgeons had labeled dissected lymph nodes by numbering them (describing the nodal station) on the basis of the lymph node map definitions for lung cancer staging proposed by Mountain and Dresler (28). The pathologist evaluated the nodes as numbered in the surgical field and recorded the presence or absence of tumor in the nodes. Specimens were stained with hematoxylin-eosin and examined with light microscopy. The histopathologic stage was then determined and recorded in each patient.
One radiologist (K.S.L., with 17 years of experience in the interpretation of thoracic images) who did not participate in MR image analyses compared histopathologic findings with MR imaging results. If any node in a nodal station was positive for malignancy, that station was deemed positive at MR imaging or histopathologic analysis. For MR imaging, the positive diagnostic criteria for malignant nodes in terms of the LTR (with a determined cutoff value > 0.84), morphologic characteristics (eccentric cortical thickening or obliteration of fatty hilum), and short-axis diameter (10 mm or greater) of the identified nodal stations were applied for each nodal station.
Statistical Analysis
The optimal cutoff point for LTR in terms of differentiating metastatic from benign lymph nodes was calculated by using receiver operating characteristic curve analysis (29).
Sensitivity, specificity, and accuracy for the detection of lymph node metastasis were calculated on a per–nodal station basis by using each of the three previously mentioned parameters (LTR, morphologic characteristics, and short-axis diameter) for the identified nodal stations and were compared between the three parameters by using the McNemar test. Univariate analysis was performed to identify the parameters that were used to differentiate malignant and benign nodes. Multivariate analysis was performed by using the multiple logistic regression method (the model involves the use of backward stepwise elimination with variables eliminated at P < .1) to determine if these parameters were useful independently of each other in distinguishing malignant from benign lymph nodes. Within the framework of the multiple logistic regression method, the likelihood ratio test was used to assess the significance of predictors (30). Odds ratios and 95% confidence intervals were determined.
The McNemar test was used to evaluate accuracy and to over- and understage rates of the nodal stages (with three parameters mentioned previously) on a per-patient basis. For all statistical analyses, P < .05 was considered to indicate a significant difference. All clinical and histopathologic data were collected from a database and analyzed with SPSS statistical software (version 10.1.4; SPSS, Chicago, Ill).
|
RESULTS
|
|---|
Tumor Cell Types and Distribution of Nodal Stages
Histopathologic analyses (n = 113) revealed adenocarcinoma (n = 58), squamous cell carcinoma (n = 41), large cell carcinoma (n = 3), adenosquamous cell carcinoma (n = 3), pleomorphic carcinoma (n = 3), sarcomatoid carcinoma (n = 2), unspecified non–small cell lung cancer (n = 2), and atypical carcinoid tumor (n = 1). Sixty-two (55%) patients had N0 stage disease; 23 (20%), N1 stage disease; 24 (21%), N2 stage disease; and four (4%), N3 stage disease.
Metastasis on a Per-Nodal-Station Basis
A total of 570 nodal stations were evaluated in 113 patients. Seventy-four (13%) nodal stations were metastatic. Findings were positive for malignancy in the hilar nodes only, mediastinal nodes only, and both hilar and mediastinal nodes in 23, 16, and 12 patients, respectively. Thirty-four (34%) of the 101 hilar nodal stations evaluated were positive for malignancy, whereas 40 (9%) of 469 mediastinal nodal stations were positive for malignancy (Table 1).
View this table:
[in this window]
[in a new window]
|
Table 1. MR Morphologic Classification of Lymph Node Stations Shown to be Histologically Positive for Malignancy at Biopsy or Resection
|
|
We used receiver operating characteristic curve analysis to select a cutoff value (LTR > 0.84) (area under the receiver operating characteristic curve, 0.735; P = .001) as the threshold for positive LTR test results (Fig 2).
View larger version (22K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2: Graph shows results of receiver operating characteristic analysis of LTR on a per–nodal station basis with an optimal cutoff value (LTR > 0.84), with 65% sensitivity and 74% specificity.
|
|
Univariate analysis revealed nodal morphology (odds ratio, 11.7; P < .001) of eccentric cortical thickening or obliterated fatty hilum (Figs 3–5), node size (short-axis diameter > 10 mm; odds ratio, 6.0; P < .001), and LTR with an optimal cutoff value of more than 0.84 (odds ratio, 5.4; P < .001) as significant parameters for depicting malignant nodes. However, multivariate logistic regression analysis revealed morphologic characteristics (odds ratio, 7.5; P < .001) and LTR (odds ratio, 2.7; P = .001) as significant parameters for depicting malignant nodes (Table 2).
View larger version (91K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3a: Mediastinal nodes show typical malignant morphology in a 66-year-old woman with T2 squamous cell lung carcinoma. (a) Transverse contrast-enhanced CT image (5.0-mm section thickness) obtained at the level of the aortic arch shows two lymph nodes (arrows) in the right lower paratracheal region (nodal station 4R). (b) Transverse T2-weighted triple-inversion black-blood fast spin-echo MR image (1714/60, 90° flip angle) obtained at a level similar to that of a shows a 19-mm lymph node (arrows) with eccentric cortical thickening and a 10-mm lymph node (arrowhead) with obliterated fatty hilum in the same region; LTRs of the two nodes were 0.84 and 0.87, respectively. Both nodes contained metastatic cells. Also note obstructive pneumonia with high signal intensity in the posterior segment of the right upper lobe.
|
|
View larger version (121K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3b: Mediastinal nodes show typical malignant morphology in a 66-year-old woman with T2 squamous cell lung carcinoma. (a) Transverse contrast-enhanced CT image (5.0-mm section thickness) obtained at the level of the aortic arch shows two lymph nodes (arrows) in the right lower paratracheal region (nodal station 4R). (b) Transverse T2-weighted triple-inversion black-blood fast spin-echo MR image (1714/60, 90° flip angle) obtained at a level similar to that of a shows a 19-mm lymph node (arrows) with eccentric cortical thickening and a 10-mm lymph node (arrowhead) with obliterated fatty hilum in the same region; LTRs of the two nodes were 0.84 and 0.87, respectively. Both nodes contained metastatic cells. Also note obstructive pneumonia with high signal intensity in the posterior segment of the right upper lobe.
|
|
View larger version (101K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4a: Mediastinal node shows typical malignant morphology in a 70-year-old man with T3 lung adenocarcinoma. (a) Transverse contrast-enhanced CT image (5.0-mm section thickness) obtained at the level of the main bronchi shows an 8-mm oval lymph node (arrow) in the aortopulmonary window region (nodal station 5). (b) Transverse T2-weighted triple-inversion black-blood fast spin-echo MR image (1500/60, 90° flip angle) obtained at a level similar to that of a shows the same node (arrow) with high signal intensity and eccentric cortical thickening. LTR of this node was 0.92. This was confirmed as a metastatic node.
|
|
View larger version (123K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 4b: Mediastinal node shows typical malignant morphology in a 70-year-old man with T3 lung adenocarcinoma. (a) Transverse contrast-enhanced CT image (5.0-mm section thickness) obtained at the level of the main bronchi shows an 8-mm oval lymph node (arrow) in the aortopulmonary window region (nodal station 5). (b) Transverse T2-weighted triple-inversion black-blood fast spin-echo MR image (1500/60, 90° flip angle) obtained at a level similar to that of a shows the same node (arrow) with high signal intensity and eccentric cortical thickening. LTR of this node was 0.92. This was confirmed as a metastatic node.
|
|
View larger version (79K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5a: Morphologically benign reactive right hilar lymph node enlargement in a 73-year-old man with T4N0 adenosquamous cell carcinoma in the right upper lobe. (a) Transverse contrast-enhanced CT image (5.0-mm section thickness) obtained at the level of the right bronchus intermedius shows a 13-mm lymph node (arrow) in the right hilum. Also note the lung nodule (arrowhead) in the right upper lobe. Calcified subcarinal node and calcified granuloma are present in the left lung. (b) Transverse T2-weighted triple-inversion black-blood fast spin-echo MR image (2000/60, 90° flip angle) obtained at a level similar to that of a shows the same node (arrow) has fatty hilum and smooth cortical thickening. This node proved to be hyperplastic and reactive. LTR of this node was 0.38. Also note the small pneumothorax (arrowhead) anteriorly in the right pleural space caused by previous percutaneous needle biopsy.
|
|
View larger version (117K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5b: Morphologically benign reactive right hilar lymph node enlargement in a 73-year-old man with T4N0 adenosquamous cell carcinoma in the right upper lobe. (a) Transverse contrast-enhanced CT image (5.0-mm section thickness) obtained at the level of the right bronchus intermedius shows a 13-mm lymph node (arrow) in the right hilum. Also note the lung nodule (arrowhead) in the right upper lobe. Calcified subcarinal node and calcified granuloma are present in the left lung. (b) Transverse T2-weighted triple-inversion black-blood fast spin-echo MR image (2000/60, 90° flip angle) obtained at a level similar to that of a shows the same node (arrow) has fatty hilum and smooth cortical thickening. This node proved to be hyperplastic and reactive. LTR of this node was 0.38. Also note the small pneumothorax (arrowhead) anteriorly in the right pleural space caused by previous percutaneous needle biopsy.
|
|
View this table:
[in this window]
[in a new window]
|
Table 2. Uni- and Multivariate Analyses of Parameters for Nodal Metastasis Detection on a Per-Nodal-Station Basis
|
|
When an LTR of more than 0.84 (Table 3) was used as the threshold for detecting a nodal metastasis, sensitivity, specificity, and accuracy were 65% (48 of 74 nodal stations), 74% (369 of 496 nodal stations), and 73% (417 of 570 nodal stations), respectively. When the morphologic characteristics were analyzed, sensitivity, specificity, and accuracy were 53% (39 of 74 nodal stations), 91% (453 of 496 nodal stations), and 86% (492 of 570 nodal stations), respectively.
View this table:
[in this window]
[in a new window]
|
Table 3. Diagnostic Accuracy of 3.0-T MR Imaging for Nodal Staging with Various Parameters on a Per-Nodal-Station Basis
|
|
Metastasis on a Per-Patient Basis
Accuracy of nodal metastasis detection (Table 4) with use of the morphologic features was 68% (77 of 113 patients). Nodal metastases were overstaged in 20 (18%) patients (Fig 6) and understaged in 16 (14%) patients with use of morphologic nodal evaluation.
View larger version (106K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6a: False-positive reactive lymph node enlargement in a 53-year-old man with T2 squamous cell lung carcinoma. (a) Transverse T2-weighted triple-inversion black-blood fast spin-echo MR image (1714/60, 90° flip angle) obtained at the left atrial level shows a 28-mm high-signal-intensity nodule (arrows) in the right lung leading to obstructive pneumonia in the right lower lobe. (b) Transverse triple-inversion black-blood MR image obtained at the level of the right bronchus intermedius shows an 18-mm lymph node (arrows) in the subcarinal area with high signal intensity and obliterated fatty hilum. LTR of this node was 1.32. This node was hyperplastic and reactive.
|
|
View larger version (95K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6b: False-positive reactive lymph node enlargement in a 53-year-old man with T2 squamous cell lung carcinoma. (a) Transverse T2-weighted triple-inversion black-blood fast spin-echo MR image (1714/60, 90° flip angle) obtained at the left atrial level shows a 28-mm high-signal-intensity nodule (arrows) in the right lung leading to obstructive pneumonia in the right lower lobe. (b) Transverse triple-inversion black-blood MR image obtained at the level of the right bronchus intermedius shows an 18-mm lymph node (arrows) in the subcarinal area with high signal intensity and obliterated fatty hilum. LTR of this node was 1.32. This node was hyperplastic and reactive.
|
|
|
DISCUSSION
|
|---|
Multivariate logistic regression analysis revealed the morphologic features of eccentric cortical thickening or an absence of fatty hilum in the nodes (odds ratio, 7.5) and LTR (odds ratio, 2.7) as significant parameters for detection of metastasis. With use of the morphologic features (having the higher odds ratio of the two parameters), sensitivity, specificity, and accuracy were 53% (39 of 74 nodal stations), 91% (453 of 496 nodal stations), and 86% (492 of 570 nodal stations), respectively, on a per–nodal station basis. On a per-patient basis, the nodal stages were correctly established in 68% of patients.
In our study, 74 (13%) of 570 nodal stations appeared to be metastatic. With only a 13% prevalence of metastatic nodal stations, the specificity and accuracy of any test should be good as long as there are few false-positive findings. We found a specificity of 91% and an accuracy of 86%. However, a sensitivity of only 53% in the face of such a low prevalence of disease might result in understaging and lead to unnecessary surgery.
Most soft-tissue masses and tumors of various causes have high signal intensity on images obtained with T2-weighted sequences (31). In addition, some investigators have reported significant differences between malignant and benign nodes in terms of their T1 and T2 relaxation times (31–33). Malignant lymph nodes have been reported to have higher signal intensity than benign lymph nodes on T1- and T2-weighted images (16,32,33). Thus, it is believed that lymph nodes with metastasis would have a long T2 relaxation time similar to that of the primary tumor, and measuring the LTR would be useful for predicting the benign versus malignant origin of identifiable lymph nodes. Our results supported this idea. Multivariate analysis revealed the LTR to be an independent parameter for distinguishing between malignant and benign nodes. However, when LTR was used as the sole parameter for nodal staging, it had lower specificity (74%) and accuracy (73%) on a per–nodal station basis and lower accuracy (46%) on a per-patient basis than when morphology was used as the sole parameter. Thus, evaluation of nodal morphologic characteristics on MR images results in better accuracy than does calculation of LTRs.
Ohno et al (17) reported that a short inversion time inversion-recovery sequence performed under 1.5-T MR conditions enabled the detection of metastatic lymph nodes in patients with non–small cell lung cancer. According to Ohno et al (17), sensitivity, specificity, and accuracy of 93% (37 of 40 patients), 87% (61 of 70 patients), and 89% (98 of 110 patients), respectively, could be achieved on a per-patient basis with use of the lymph node–to-saline ratio (ratio of signal intensity of the lymph node to signal intensity of the saline) as a parameter. These results, particularly the high sensitivity, are quite different from our results. (The highest accuracy value was 68% on a per-patient basis with analysis of the morphologic features of nodes.) This discrepancy might be caused by differences in patient population (the presence or absence of granulomatous disease endemicity and different T stages of non–small cell lung cancer), MR sequence (with or without the black-blood technique), MR unit (1.5 T vs 3.0 T), and/or diagnostic parameters expressing the increased signal intensity of metastatic lymph nodes (lymph node–saline ratio vs LTR).
MR imaging has two main disadvantages. First, calcification within the lymph nodes might be overlooked with MR imaging. Therefore, an enlarged lymph node from a prior granulomatous infection may be misdiagnosed as a metastasis. Second, because the spatial resolution of MR imaging is poorer than that of CT, a group of discrete adjacent normal-sized nodes may occasionally blur together on MR images and appear as a single large nodal mass, which may be diagnosed erroneously as a metastasis (34,35). However, the main advantage of 3.0-T MR imaging is the signal-to-noise ratio gain that scales linearly with increasing field strength (36,37). Thus, high-magnetic-field-strength MR imaging could be used flexibly to reduce acquisition time, improve spatial resolution, or both (37–39). The routine use of electrocardiographically gated MR imaging to view the mediastinum may also alleviate the problem of low spatial resolution (12).
In conclusion, the morphologic details of lymph nodes on T2-weighted triple-inversion black-blood fast spin-echo MR images appear to be the most important for mediastinal or hilar nodal metastasis detection with 3.0-T MR imaging, an analysis of which can yield high specificity (91%) and accuracy (86%).
|
ADVANCES IN KNOWLEDGE
|
|---|
- Nodal morphology on T2-weighted triple-inversion black-blood fast spin-echo 3.0-T MR images has sensitivity, specificity, and accuracy of 53% (39 of 74 nodal stations), 91% (453 of 496 nodal stations), and 86% (492 of 570 nodal stations), respectively, for detection of metastasis.
- Multivariate logistic regression analysis revealed morphologic characteristics and lymph node–to-tumor ratio (LTR) to be significant parameters for detection of metastasis.
- When using the LTR as a diagnostic parameter, a cutoff value of more than 0.84 is most appropriate for detecting nodal metastasis.
|
IMPLICATION FOR PATIENT CARE
|
|---|
- MR imaging at 3.0 T may become a helpful adjunctive technique for lung cancer staging because it can yield high specificity (91%) and accuracy (86%) in the detection of hilar and mediastinal nodal metastases.
|
FOOTNOTES
|
|---|
Abbreviations: LTR = lymph node–to-tumor ratio
Guarantor of integrity of entire study, K.S.L.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, H.Y.K., C.A.Y., K.S.L., M.J.C., Y.K.K.; clinical studies, H.Y.K., C.A.Y., K.S.L., Y.K.K., H.K., O.J.K.; statistical analysis, H.Y.K., C.A.Y., K.S.L., B.K.C.; and manuscript editing, all authors
Authors stated no financial relationship to disclose.
|
References
|
|---|
- Ohno Y, Sugimura K, Hatabu H. MR imaging of lung cancer. Eur J Radiol 2002;44:172–181. [CrossRef][Medline]
- Libshitz HI, McKenna RJ Jr, Haynie TP, McMurtrey MJ, Mountain CT. Mediastinal evaluation in lung cancer. Radiology 1984;151:295–299. [Abstract/Free Full Text]
- Musset D, Grenier P, Carette MF, et al. Primary lung cancer staging: prospective comparative study of MR imaging with CT. Radiology 1986;160:607–611. [Abstract/Free Full Text]
- McLoud TC, Bourgouin PM, Greenberg RW, et al. Bronchogenic carcinoma: analysis of staging in the mediastinum with CT by correlative lymph node mapping and sampling. Radiology 1992;182:319–323. [Abstract/Free Full Text]
- Seely JM, Mayo JR, Miller RR, Muller NL. T1 lung cancer: prevalence of mediastinal nodal metastases and diagnostic accuracy of CT. Radiology 1993;186:129–132. [Abstract/Free Full Text]
- Scott WJ, Gobar LS, Terry JD, Dewan NA, Sunderland JJ. Mediastinal lymph node staging of non-small-cell lung cancer: a prospective comparison of computed tomography and positron emission tomography. J Thorac Cardiovasc Surg 1996;111:642–648. [Abstract/Free Full Text]
- Webb WR, Gatsonis C, Zerhouni EA, et al. CT and MR imaging in staging non-small cell bronchogenic carcinoma: report of the Radiologic Diagnostic Oncology Group. Radiology 1991;178:705–713. [Abstract/Free Full Text]
- Gupta NC, Tamim WJ, Graeber GG, Bishop HA, Hobbs GR. Mediastinal lymph node sampling following positron emission tomography with fluorodeoxyglucose imaging in lung cancer staging. Chest 2001;120:521–527. [CrossRef][Medline]
- Shim SS, Lee KS, Kim BT, et al. Non-small cell lung cancer: prospective comparison of integrated FDG PET/CT and CT alone for preoperative staging. Radiology 2005;236:1011–1019. [Abstract/Free Full Text]
- Dwamena BA, Sonnad SS, Angobaldo JO, Wahl RL. Metastases from non-small cell lung cancer: mediastinal staging in the 1990s—meta-analytic comparison of PET and CT. Radiology 1999;213:530–536. [Abstract/Free Full Text]
- Erasmus JJ, Truong MT, Munden RF. CT, MR, and PET imaging in staging of non-small-cell lung cancer. Semin Roentgenol 2005;40:126–142. [CrossRef][Medline]
- Webb WR. MR imaging in the evaluation and staging of lung cancer. Semin Ultrasound CT MR 1988;9:53–66. [Medline]
- Boiselle PM, Patz EF Jr, Vining DJ, Weissleder R, Shepard JA, McLoud TC. Imaging of mediastinal lymph nodes: CT, MR, and FDG PET. RadioGraphics 1998;18:1061–1069. [Abstract]
- Kauczor HU, Kreitner KF. Contrast-enhanced MRI of the lung. Eur J Radiol 2000;34:196–207. [CrossRef][Medline]
- Templeton PA, Caskey CI, Zerhouni EA. Current uses of CT and MR imaging in the staging of lung cancer. Radiol Clin North Am 1990;28:631–646. [Medline]
- Glazer GM, Orringer MB, Chenevert TL, et al. Mediastinal lymph nodes: relaxation time/pathologic correlation and implications in staging of lung cancer with MR imaging. Radiology 1988;168:429–431. [Abstract/Free Full Text]
- Ohno Y, Hatabu H, Takenaka D, et al. Metastases in mediastinal and hilar lymph nodes in patients with non-small cell lung cancer: quantitative and qualitative assessment with STIR turbo spin-echo MR imaging. Radiology 2004;231:872–879. [Abstract/Free Full Text]
- Takahashi M, Uematsu H, Hatabu H. MR imaging at high magnetic fields. Eur J Radiol 2003;46:45–52. [CrossRef][Medline]
- Brown MA, Semelka RC. MR imaging abbreviations, definitions, and descriptions: a review. Radiology 1999;213:647–662. [Free Full Text]
- Walker R, Kessar P, Blanchard R, et al. Turbo STIR magnetic resonance imaging as a whole-body screening tool for metastases in patients with breast carcinoma: preliminary clinical experience. J Magn Reson Imaging 2000;11:343–350. [CrossRef][Medline]
- Viano AM, Gronemeyer SA, Haliloglu M, Hoffer FA. Improved MR imaging for patients with metallic implants. Magn Reson Imaging 2000;18:287–295. [CrossRef][Medline]
- Eustace S, Tello R, DeCarvalho V, Carey J, Melhem E, Yucel EK. Whole body turbo STIR MRI in unknown primary tumor detection. J Magn Reson Imaging 1998;8:751–753. [Medline]
- Takenaka D, Ohno Y, Hatabu H, et al. Differentiation of metastatic versus non-metastatic mediastinal lymph nodes in patients with non-small cell lung cancer using respiratory-triggered short inversion time inversion recovery (STIR) turbo spin-echo MR imaging. Eur J Radiol 2002;44:216–224. [CrossRef][Medline]
- Hatabu H, Gaa J, Tadamura E, et al. MR imaging of pulmonary parenchyma with a half-Fourier single-shot turbo spin-echo (HASTE) sequence. Eur J Radiol 1999;29:152–159. [CrossRef][Medline]
- Yamashita Y, Yokoyama T, Tomiguchi S, Takahashi M, Ando M. MR imaging of focal lung lesions: elimination of flow and motion artifact by breath-hold ECG-gated and black-blood techniques on T2-weighted turbo SE and STIR sequences. J Magn Reson Imaging 1999;9:691–698. [CrossRef][Medline]
- Esen G. Ultrasound of superficial lymph nodes. Eur J Radiol 2006;58:345–359. [CrossRef][Medline]
- Krestan C, Herneth AM, Formanek M, Czerny C. Modern imaging lymph node staging of the head and neck region. Eur J Radiol 2006;58:360–366. [CrossRef][Medline]
- Mountain CF, Dresler CM. Regional lymph node classification for lung cancer staging. Chest 1997;111:1718–1723. [CrossRef][Medline]
- Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29–36. [Abstract/Free Full Text]
- Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic research: principles and quantitative methods. New York, NY: Wiley, 1982; 447–456.
- Sundaram M, McGuire MH, Schajowicz F. Soft-tissue masses: histologic basis for decreased signal (short T2) on T2-weighted MR images. AJR Am J Roentgenol 1987;148:1247–1250. [Abstract/Free Full Text]
- Chakraborti KL, Jena A. MR evaluation of the mediastinal lymph nodes. Indian J Chest Dis Allied Sci 1997;39:19–25. [Medline]
- Wiener JI, Chako AC, Merten CW, Gross S, Coffey EL, Stein HL. Breast and axillary tissue MR imaging: correlation of signal intensities and relaxation times with pathologic findings. Radiology 1986;160:299–305. [Abstract/Free Full Text]
- Boiselle PM. MR imaging of thoracic lymph nodes: a comparison of computed tomography and positron emission tomography. Magn Reson Imaging Clin N Am 2000;8:33–41. [Medline]
- Quint LE, Francis IR, Wahl RL, Gross BH, Glazer GM. Preoperative staging of non-small-cell carcinoma of the lung: imaging methods. AJR Am J Roentgenol 1995;164:1349–1359. [Abstract/Free Full Text]
- Willinek WA, Born M, Simon B, et al. Time-of-flight MR angiography: comparison of 3.0-T imaging and 1.5-T imaging—initial experience. Radiology 2003;229:913–920. [Abstract/Free Full Text]
- Campeau NG, Huston J 3rd, Bernstein MA, Lin C, Gibbs GF. Magnetic resonance angiography at 3.0 Tesla: initial clinical experience. Top Magn Reson Imaging 2001;12:183–204. [CrossRef][Medline]
- Robitaille PM, Abduljalil AM, Kangarlu A. Ultra high resolution imaging of the human head at 8 tesla: 2K x 2K for Y2K. J Comput Assist Tomogr 2000;24:2–8. [CrossRef][Medline]
- Thulborn KR. Clinical rationale for very-high-field (3.0 Tesla) functional magnetic resonance imaging. Top Magn Reson Imaging 1999;10:37–50. [CrossRef][Medline]
TOP
ABSTRACT
INTRODUCTION
