| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Nuclear Medicine |
1 From the Department of Radiology, Duke University Medical Center, Durham, NC. From the 2000 RSNA scientific assembly. Received May 14, 2001; revision requested June 27; final revision received May 6, 2002; accepted May 14. Address correspondence to E.M.R., Department of Nuclear Medicine, Charlton 1N, Mayo Clinic, 200 First St SW, Rochester, MN 55905 (e-mail: Rohren.Eric@mayo.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Forty patients underwent brain PET and contrast material–enhanced brain MR imaging, with a maximum of 30 days between examinations. PET and MR images were each retrospectively reviewed by two independent readers who were blinded to the clinical history and results of the other technique. Presence of metastatic disease was recorded for each modality. Sensitivity and specificity of FDG PET were determined with MR imaging as the standard. Statistical analysis was performed with the Fisher exact test and the logistic regression model.
RESULTS: Sixteen patients had cerebral metastases at MR imaging, and in 12 of these, PET scans were interpreted as showing metastatic disease (in four, scans were false-negative). Twenty-four patients had no cerebral metastases at MR imaging, and 20 of these had PET scans interpreted as normal (in four, scans were false-positive). For identification of patients with cerebral metastases, FDG PET had a sensitivity of 75% (12 of 16) and a specificity of 83% (20 of 24). Thirty-eight metastatic lesions were seen at MR imaging; 23 (61%) of these were identified at PET. Size was a statistically significant factor that influenced lesion detection at PET (P < .001).
CONCLUSION: Only 61% of metastatic lesions in the brain were identified at PET. In particular, detection of small lesions was difficult.
© RSNA, 2002
Index terms: Brain neoplasms, metastases, 10.38 • Brain neoplasms, MR, 10.121411, 10.12143, 10.38 • Brain neoplasms, PET, 10.12163, 10.1217 • Fluorine, radioactive, 10.12163, 10.1217 • Positron emission tomography (PET), comparative studies
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Tumor localization and staging with FDG PET for a variety of malignancies is accurate and cost effective (1). However, the role of brain imaging with PET in patients undergoing staging of non-CNS malignancy has been questioned. In an early report, Griffeth et al (6) found that with FDG PET, only 68% of discrete metastatic lesions in the brain were identified; in these studies, computed tomography (CT) or magnetic resonance (MR) imaging was used as a reference standard. In a small number of patients with non–small cell lung cancer, Marom et al (7) reported that FDG PET brain images were positive in three of five patients with cerebral metastases.
In addition to evaluation of lesion detection with PET, assessment of whether identification of these lesions changes clinical treatment is also important. It has been found that when scanning of the brain is incorporated into an imaging protocol for patients with non-CNS malignancies, a change in treatment occurs in fewer than 1% of patients (8). When PET is used as a screening tool for cerebral metastases, the number of false-positive and false-negative scans can equal or exceed the number of new true diagnoses (9).
We believe the limitation of these previous reports to be the lack of a blinded, retrospective review of the imaging examinations to determine the sensitivity and specificity of PET for cerebral metastases. Also, lesion size has been suggested as a factor in detection at PET, but it has not been shown to indicate a statistically significant difference (6) because of an insufficient number of lesions. To assess the appropriateness of FDG PET screening for cerebral metastases in patients with non-CNS malignancies, further information about the accuracy of limitations of PET is needed (10). The purpose of our study was to compare FDG PET with the current standard, MR imaging, by using a blinded retrospective review, to determine the sensitivity and specificity of FDG PET for detection of cerebral metastases, and to determine the factors that may affect lesion conspicuity.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Patient Demographics
Forty patients, including 27 men and 13 women, who ranged in age from 17 to 84 years (mean age, 57.6 years) met inclusion criteria. The average interval between PET and MR imaging was 9 days (range, 0–29 days). In 17 patients, PET was performed prior to MR imaging; in 17 patients, MR imaging was performed prior to PET; in six patients, PET and MR imaging were performed on the same day. PET was performed for the following indications: non–small cell lung cancer (n = 10), melanoma (n = 10), non-Hodgkin lymphoma (n = 4), breast carcinoma (n = 4), head and neck cancer (n = 4), soft-tissue sarcoma (n = 3), colorectal carcinoma (n = 2), small cell lung cancer (n = 1), and solitary pulmonary nodule (n = 1). PET was also performed in one patient who was suspected of having an occult malignancy (n = 1).
The medical records of patients were reviewed by one of us (E.M.R.) to determine whether patients demonstrated neurologic symptoms at the time of PET and MR imaging. Eighteen of 40 patients had neurologic signs or symptoms at the time of PET and MR imaging, and the symptoms included headache (n = 6), altered mentation (n = 6), cranial nerve impairment (n = 5), peripheral nerve impairment (n = 4), altered gait (n = 1), and seizures (n = 1). Five patients had more than one symptom at presentation. All imaging examinations were performed as part of the clinical work-up of the patients and not as part of a research protocol.
FDG PET Examinations
18F was produced at an in-house cyclotron (CS-30; Cyclotron, Berkeley, Calif), and FDG was made by using an automated FDG synthesis device (PETtrace; GE Medical Systems, Milwaukee, Wis). Patients fasted for a minimum of 4 hours prior to injection; 145 µCi (5.4 MBq) per kilogram of body weight of 18F FDG was administered intravenously to a maximum dose of 20 mCi (740 MBq). After injection, patients were allowed to rest quietly in a dimly lit room for at least 30 minutes during the uptake phase. PET was performed with a dedicated full-ring PET scanner (Advance PET; GE Medical Systems). Brain imaging was performed by using an additional table position, with a 25.6-cm-diameter scanning field and 4.25-mm transverse sections.
One table position (14.4 cm) was used to image the brain. After a 1-minute positioning scan was obtained, a three-dimensional acquisition was performed. For count rates greater than 50,000 counts per second, a 6-minute acquisition time was used. For count rates of less than 50,000 counts per second, imaging time in minutes was determined by the formula of 300,000 divided by the measured counts per second, to a maximum imaging time of 10 minutes. The acquisition data were reconstructed by using the reprojection method of Kinahan and Rogers and filtered with a Hann filter. With this method, the full width half maximum of brain images at our institution was 6 mm.
MR Imaging
MR imaging examinations were performed with a 1.5-T unit (Signa; GE Medical Systems). Imaging sequences performed in all patients included transverse spin-echo unenhanced and enhanced T1-weighted MR imaging (repetition time msec/echo time msec, 500/20), spin-echo proton density–weighted (2,500/30) and T2-weighted (2,500/80) MR imaging, and coronal spin-echo enhanced T1-weighted MR imaging (500/20). The MR contrast agent was gadopentetate dimeglumine (Magnevist; Berlex Imaging, Wayne, NJ), which was intravenously administered in a dose of 0.1 mmol/kg.
Image Interpretation
Image interpretation was performed by four readers, two for MR imaging and two for PET. Hard copies of MR images were reviewed retrospectively by two independent readers with 11 years of experience each. MR images were scored according to presence of metastases and number and location of lesions, and three-plane measurements were recorded for all metastases. Readers of MR images also noted any other intracranial abnormalities, such as cerebral infarction.
Raw data from FDG PET images were loaded onto a workstation (Advance; GE Medical Systems) and displayed with multiplanar reconstruction, as is typically performed during clinical interpretation of these images. Images were typically viewed in gray scale on a black background, but multiple color schemes could be used, and these schemes included linear and logarithmic scales. Images were reviewed by two independent readers with 3 and 20 years of experience, respectively, and abnormalities that were suggestive of metastatic disease were recorded; these abnormalities included number, location, estimation of size, and FDG uptake compared with that of normal cerebral cortex.
Discrepant interpretations between readers of each group were resolved by consensus. All readers were blinded to the clinical history of the patients and to the results of the other imaging examination.
Statistical Analysis
The results of PET and MR imaging were correlated, and sensitivity and specificity were calculated by using MR imaging as the reference standard. In the case of false-positive PET scans, MR images were reviewed to ascertain whether any nonmetastatic structural abnormality could account for the interpretation of the PET scans. The correlation between lesion detection with FDG PET and tumor type was analyzed with the two-tailed Fisher exact test. The relationship between lesion size and detection rate with PET was analyzed by using the logistic regression model with a generalized estimating equation adjustment for clustering of lesions within subjects.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
In two of the false-positive PET scans, a hypometabolic region was identified, and it was thought to be a hypometabolic metastasis. However, MR imaging showed the abnormality in both of these cases to be caused by prior cerebral infarction rather than by malignancy. With respect to the other two false-positive PET scans, no corresponding abnormality was seen at MR imaging, and in retrospect, the activity was believed to be caused by regional variations in gray matter activity (one cortical and one thalamic) that mimicked a focal hypermetabolic lesion.
The results of the retrospective image review were correlated with the information in the patients’ medical records to assess any relationship between patient signs and symptoms and imaging results. Of the 40 patients examined, 18 had neurologic signs or symptoms at the time of examination, whereas the remaining 22 patients were neurologically asymptomatic. Correlated with imaging results, 10 (56%) of 18 of symptomatic patients proved to have metastatic disease, compared with six (38%) of 16 asymptomatic patients. Of the 12 patients with disease detected at both MR imaging and PET (true-positive PET scans), eight (67%) were symptomatic at the time of examination. Of the 12 patients with no metastatic disease detected at either MR imaging or PET (true-negative PET scans), seven (58%) were symptomatic. Of the four patients in whom PET scans were negative but MR images were positive (false-negative PET scans), two (50%) were symptomatic, and of the four patients in whom PET scans were positive but MR images were negative (false-positive PET scans), one (25%) was symptomatic.
Detection of individual metastatic lesions with FDG PET was evaluated next. In the 16 patients with cerebral metastases, 38 individual metastatic lesions were identified at MR imaging. The number of lesions in a patient ranged from one to eight. Twenty-three of 38 lesions were identified at PET, and the overall lesion detection rate was 61%. There was variability in the appearance of the metastatic lesions detected at FDG PET. Fourteen of 38 lesions were identified as discrete, hypermetabolic foci, with FDG activity qualitatively greater than that in gray matter (Fig 1). In two additional patients, the metastatic disease was also hypermetabolic, but it manifested as leptomeningeal spread of tumor rather than as an intraparenchymal abnormality (Fig 2). The remaining seven metastatic lesions had FDG activity substantially less than that of gray matter (Fig 3).
|
|
|
|
|
|
|
|
|
|
Undetected lesions occurred with a variety of malignancies (Fig 5). There was no statistical significance in a correlation of lesion detection at PET with tumor type (P = .237) when all tumors were considered. When the statistical analysis was limited to the three tumor types that accounted for the majority of the metastatic lesions, namely, melanoma, lung carcinoma, and lymphoma, there was a trend toward a relationship between tumor type and lesion detection. However, this also was not considered to indicate a statistically significant difference (P = .085). Average lesion diameter in the transverse plane was determined for each tumor type on the basis of MR imaging measurements as follows: melanoma, 1.4 cm; lung carcinoma, 1.5 cm; soft-tissue sarcoma, 2.2 cm; colon carcinoma, 2.3 cm; lymphoma, 2.4 cm; and breast carcinoma, 7.0 cm.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
To address these issues, we used a blinded retrospective review of FDG PET scans of patients with known non-CNS malignancy or with findings suggestive of it who were selected only because they had also undergone MR imaging of the brain within 30 days of PET. We chose contrast-enhanced MR imaging as the sole reference standard in this study, because histologic proof was not obtained for any of the patients in this study. Although the precise accuracy of contrast-enhanced MR imaging for cerebral metastases is unknown, it is currently the best available imaging test for detection of cerebral metastases (13), and a positive MR image is usually considered diagnostic at our institution, given the invasiveness of a cerebral biopsy. Although the presence of neurologic symptoms can be an indicator of cerebral metastases, when we retrospectively examined the patients in our study, the presence or absence of neurologic signs or symptoms was not a strong predictor of metastatic disease to the brain.
As a screening test for the detection of cerebral metastases, FDG PET is insensitive for the detection of disease, and small lesions in particular may not be detected with PET. By using FDG PET to determine the presence or absence of metastatic disease to the brain, we found the sensitivity and specificity of PET to be 75% and 83%, respectively. When FDG PET was used to delineate the number of metastatic lesions in a particular patient, we found that only 61% of lesions were prospectively identified. In three cases, patients with undetected lesions had other brain metastases that were identified at PET, and these patients were, therefore, correctly identified as having cerebral metastatic disease. However, in four patients, the omission of lesions led to the inaccurate interpretation of the PET scan as normal.
Size was a significant factor in lesion conspicuity at FDG PET (P < .001). Probability curves can be used to predict that the likelihood of detecting a metastatic lesion 1 cm in size is only approximately 40% and that a lesion must be approximately 1.8 cm before the mean detection rate is 90%. The dependence of PET imaging on lesion size is caused by several factors: Small volumes of tumor may not accumulate sufficient FDG to generate a discernable signal, the spatial resolution of PET is intrinsically limited by the nature of positron annihilation events and PET instrumentation factors, and small metastatic foci in or adjacent to cerebral cortex may be inconspicuous against the background of normal gray matter.
Although we found no statistically significant link between lesion detection and tumor type, we did observe a trend toward undetected lesions in patients with melanoma. We suspect that this is due to the fact that the metastatic lesions in patients with melanoma tend to be smaller than those in patients with other types of malignancies, but other factors may play a role as well. Larger numbers of patients are needed to demonstrate a relationship between tumor type and detection at FDG PET that is considered statistically significant.
A potential limitation of our study is bias introduced by our selection criteria. Patients included in this study underwent both FDG PET and brain MR imaging, with a maximum of 30 days between examinations. This selection process resulted in a prevalence of cerebral metastases in our group of 40%, which is likely higher than would be expected in the general population of patients undergoing whole-body FDG PET examination for malignancy. This bias did not influence the generation of our data, because findings of both modalities were interpreted without knowledge of the clinical findings or results of the other imaging test. However, PET imaging in a group with lower disease prevalence would be expected to result in a higher number of false-positive scans and a corresponding decrease in the specificity of PET.
The question of whether a screening brain scan should be included as part of a whole-body FDG PET staging protocol in patients with malignancy must ultimately be decided by the individual PET facility and its referring physicians. In this study, we found that even with a full-ring dedicated PET scanner, only approximately two-thirds of brain metastases were detected by using PET, and one-quarter of patients with cerebral metastases had a negative PET scan of the brain. These limitations could lead to inaccurate disease staging and a false sense of security on the basis of a negative PET scan of the brain. Scanning time is also a consideration in a busy PET facility, and the added time of scanning the brain can decrease the total number of patients undergoing scanning per day.
Because of the findings in this study, we no longer routinely perform scanning of the brain in patients undergoing whole-body FDG PET for staging of non-CNS malignancy. The omission of the brain scanning decreases the scanning time per patient by approximately 10 minutes, and this reduction improves patient throughput in our department. The time saved on an average day allows for an additional one or two examinations per day with our PET scanner, which in turn decreases the waiting time on the PET schedule. At times, scanning of the brain is specifically requested by the referring physician because features in a patient’s history or physical examination findings are suggestive of cerebral metastases. We will accommodate these requests, but we also educate the physician regarding the limitations of PET to avoid errors in treatment due to potentially misleading PET results. Although there is no consensus about the value of screening MR imaging of the brain in patients with non-CNS malignancy, MR imaging shows more lesions than PET, and PET of the brain should not supplant the judicious use of anatomic imaging in these patients.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, E.M.R.; study concepts and design, all authors; literature research, E.M.R., R.E.C.; clinical studies, all authors; data acquisition, all authors; data analysis/interpretation, E.M.R.; statistical analysis, E.M.R.; manuscript preparation and definition of intellectual content, all authors; manuscript editing, E.M.R.; manuscript revision/review and final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  N. C. Nguyen, E. I. Akduman, M. H. Sayed, D. Oliver, and M. M. Osman Increased 18F-FDG Uptake in the Posterior Ocular Bulb Is Associated with Brain Metastasis: A Retrospective Study Am. J. Roentgenol., December 1, 2008; 191(6): W268 - W274. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Ohno, H. Koyama, Y. Onishi, D. Takenaka, M. Nogami, T. Yoshikawa, S. Matsumoto, Y. Kotani, and K. Sugimura Non-Small Cell Lung Cancer: Whole-Body MR Examination for M-Stage Assessment--Utility for Whole-Body Diffusion-weighted Imaging Compared with Integrated FDG PET/CT Radiology, August 1, 2008; 248(2): 643 - 654. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. E. Posther, L. M. McCall, D. H. Harpole Jr., C. E. Reed, J. B. Putnam Jr., V. W. Rusch, and B. A. Siegel Yield of Brain 18F-FDG PET in Evaluating Patients with Potentially Operable Non-Small Cell Lung Cancer J. Nucl. Med., October 1, 2006; 47(10): 1607 - 1611. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. F. Bruzzi, M. T. Truong, E. M. Marom, O. Mawlawi, D. A. Podoloff, H. A. Macapinlac, and R. F. Munden Incidental Findings on Integrated PET/CT That Do Not Accumulate 18F-FDG Am. J. Roentgenol., October 1, 2006; 187(4): 1116 - 1123. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. B. Lobrano Partnerships in Oncology and Radiology: The Role of Radiology in the Detection, Staging, and Follow-up of Lung Cancer Oncologist, July 1, 2006; 11(7): 774 - 779. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. U. Lin, J. R. Bellon, and E. P. Winer CNS Metastases in Breast Cancer J. Clin. Oncol., September 1, 2004; 22(17): 3608 - 3617. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. M. Rohren, T. G. Turkington, and R. E. Coleman Clinical Applications of PET in Oncology Radiology, May 1, 2004; 231(2): 305 - 332. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. M. Osman Whole-Body Imaging for Cancer Staging JAMA, March 17, 2004; 291(11): 1320 - 1320. [Full Text] [PDF]
|
|
|
|
|
|
E. M. Kamel, D. Zwahlen, M. T. Wyss, K. D. Stumpe, G. K. von Schulthess, and H. C. Steinert Whole-Body 18F-FDG PET Improves the Management of Patients with Small Cell Lung Cancer J. Nucl. Med., December 1, 2003; 44(12): 1911 - 1917. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
