| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Diagnostic Radiology, The University of Texas M. D. Anderson
Cancer Center, 1515 Holcombe Blvd., Unit 1350, Houston, TX 77030-4009.
2 Department of Nuclear Medicine and PET, Hong Kong Sanatorium and Hospital,
Happy Valley, Hong Kong.
3 Department of Nuclear Medicine, The University of Texas M. D. Anderson Cancer
Center, Houston, TX.
Received November 20, 2007;
accepted after revision June 3, 2008.
Address correspondence to R. Vikram
(rvikram{at}mdanderson.org).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. We reviewed the radiology reports and images of patients with known malignant disease who had undergone PET/CT for cancer staging or surveillance and who had adrenal nodules at least 1 cm in diameter. We identified 112 adrenal nodules in 96 patients. Two-dimensional PET had been performed 1 hour after administration of 18F-FDG. Unenhanced CT was performed for attenuation correction, to determine lesion size, and for coregistration with PET data. Adrenal nodules were considered to have a positive PET result if the average standardized uptake value was greater than that of the liver. Follow-up data and biopsy reports were used to determine the pathologic status of the adrenal nodules.
RESULTS. Thirty adrenal lesions were malignant. Twenty-five of the 30 malignant nodules had positive PET results. Twelve of 82 benign nodules were PET positive with a sensitivity of 83.3% and specificity of 85.4%. Patients with four of five malignant nodules with negative PET results had received previous therapy. The positive predictive value for detection of malignant lesions was 67%, and the negative predictive value was 93%.
CONCLUSION. Adrenal masses that are not FDG avid are likely to be benign with a high negative predictive value. Especially in patients undergoing therapy, however, there is a small but statistically significant false-negative rate. A considerable proportion of benign nodules have increased FDG activity.
Keywords: adrenal gland • malignancy • metastasis • PET/CT
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Numerous studies [11–19] have been conducted to evaluate the use of PET for characterizing adrenal masses. Some of these studies, however, had selected patient populations and were biased toward a particular primary tumor [13, 15, 17, 20]. Many studies also have been focused on the use of 18F-FDG PET alone rather than in combination with CT [15, 20]. PET/CT has the advantage of providing both anatomic and functional data, and coregistration of the data makes it easier to localize and analyze areas of increased FDG activity. The increasing use of PET/CT in the evaluation of cancer has brought back to center stage the question of the use of this technique in cancer staging [18, 21]. As data and experience with PET/CT accumulate, it is being recognized that a small percentage of benign adrenal lesions exhibit FDG activity, mimicking metastatic lesions [3]. The purpose of this retrospective study was to determine the utility of PET/CT in differentiating benign from malignant adrenal nodules in a population of patients with cancer.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Patients
We performed an electronic records search for radiology reports of CT and
PET/CT performed from January 1, 2004, through December 31, 2005, that
contained any of the phrases "adrenal mass," "adrenal
metastasis," "adrenal nodule," and "adrenal
adenoma." From among those records, we selected the cases of patients
who had PET/CT findings of adrenal nodules at least 1 cm in diameter and
underwent imaging follow-up for at least 6 months after PET/CT. Demographic
details, including patient age and sex and location of the primary tumor, were
recorded for all patients. Treatment history also was recorded for patients
who had false-negative findings of adrenal lesions. A total of 112 adrenal
nodules in 96 consecutively imaged patients met the criteria for the study.
All relevant previous and subsequent cross-sectional images were reviewed to
determine the stability of lesions over time. In the six cases in which it was
available, the histopathologic diagnosis was used to determine the nature of
the nodule.
|
|
|
|
|
Unenhanced CT from the base of the skull to the upper thigh was performed for attenuation correction and diagnosis (300 mA; tube rotation time, 0.5 second; 120 kVp; table speed, 13.5 mm/rotation; beam collimation, 8 x 1.25 mm). Axial CT images were reconstructed with a soft reconstruction kernel with a slice thickness of 3.75 mm and an interval of 3.27 mm to match the PET images.
PET scans were acquired in the 2D mode for 3 minutes per bed position, and the images were reconstructed with standard vendor-provided reconstruction algorithms incorporating ordered subset expectation maximization. Attenuation correction of PET images was performed with attenuation data from the CT component of the examination. The manufacturer's software was used to correct emission data for scatter, random events, and dead-time losses.
Imaging Data Analysis
The existing CT and PET components of the study were reviewed on
workstations (Advantage Windows 4.2 software, GE Healthcare). The PET and
fused images were analyzed in both axial and coronal planes. Two radiologists
and two nuclear medicine physicians reviewed the images, and decisions were
reached in consensus.
We used the CT component of the study to measure the size of the adrenal gland and the PET component of the study to measure the average standardized uptake value (SUV) over a region of interest (ROI) placed on the liver and adrenal nodule. The PET/CT images reconstructed in the coronal and axial planes were used to confirm accurate placement of the ROI on the adrenal gland.
By consensus, two patients were not included in the study because of considerable misregistration of the PET and CT images. If there was minor discrepancy in coregistration, the ROI was placed after inspection of the PET scans in both the coronal and axial planes with renal and collecting system activity as a guide to the location of the adrenal gland.
Quantitative analysis of FDG uptake by the adrenal lesions was performed with a 3D ROI on the fused axial and coronal images (Fig. 1A, 1B, 1C, 1D, 1E). The ROI included at least two thirds of the adrenal lesion on the PET/CT fused images. Care was taken to avoid the periphery of the lesion and thereby minimize the partial volume effect.
Quantitative analysis of FDG uptake by the liver was performed with a 2D ROI placed on a relatively homogeneous part of the right lobe of the liver that was free of visible metastatic disease (Fig. 1A, 1B, 1C, 1D, 1E). Average and maximum SUVs were automatically generated by the software with the equation SUV = Ctis / Dinj / body weight, in which SUV is normalized for body weight in kilograms, Ctis is tissue concentration in megabecquerels per milliliter, and Dinj is the injected dose in megabecquerels.
|
|
|
|
|
All relevant clinical data, including pathology reports when available, were reviewed. All relevant previous and follow-up imaging data were reviewed to determine the nature of the adrenal nodule. Adrenal nodules that had either an interval increase in size or a more than 20% decrease in size after appropriate therapy were considered malignant. Nodules that remained stable for at least 6 months of follow-up were considered benign.
We entered the recorded data into a spreadsheet (Excel, Microsoft). The differences between the average SUV of the adrenal nodules and those of the liver were computed, and the lesions were classified as PET positive or PET negative. These findings were correlated with the final diagnosis of the lesion as malignant or benign. Malignant lesions that were PET positive were considered true-positive, and those that were PET negative were considered false-negative. Likewise, benign lesions that were PET negative were considered true-negative and those that were PET positive were considered false-positive.
We also calculated sensitivity, specificity, and positive and negative predictive values using the formulas shown in Table 1. We used the Microsoft Excel statistical function to calculate the means, medians, and 25th and 75th percentiles of the av erage SUV for the benign and malignant nodules.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Thirty-seven of the lesions were considered PET positive, and 75 were considered PET negative. Of the 37 PET-positive lesions, 25 were malignant (i.e., true-positive) and 12 were benign (i.e., false-positive). Of the 75 PET-negative lesions, 70 were benign (i.e., true-negative), and five were malignant (i.e., false-negative).
Only adrenal nodules at least 1 cm in diameter were included in this study. The mean size of all 112 adrenal nodules was 2.4 cm (range, 1–13 cm). The mean size of the 82 benign lesions was 1.8 cm (range, 1–4.3 cm), and that of the 30 malignant lesions was 4.03 cm (range, 1–13 cm). The mean size of the 75 PET-negative nodules was 1.81 cm (range, 1–4.3 cm), and that of the 37 PET-positive nodules was 3.7 cm (range, 1–13 cm).
The mean size of the 12 PET-positive benign nodules (false-positive) was 1.9 cm (range, 1–2.7 cm). These lesions had a mean adrenal-to-liver average SUV ratio of 1.32 (range, 1.0–2.40). Six of the patients with these lesions had non–small cell lung cancer; three had lymphoma; and one each had gastric carcinoma, melanoma, and squamous cell carcinoma of the head and neck. The mean size of five PET-negative malignant adrenal nodules (false-negative) was 1.3 cm (1, 1.3, 1.3, 1.4, and 1.5 cm). These lesions had a mean adrenal-to-liver average SUV ratio of 0.61 (range, 0.48–0.71). Among the patients with these lesions, two had malignant melanoma, and one each had lymphoma, invasive thymoma, and squamous cell cancer of the larynx. Four of the five patients had previously undergone chemotherapy. The sensitivity of PET/CT in determination of malignancy was 83.3%, and the specificity was 85.4%. The positive and negative predictive values were 67% and 93%, respectively. The mean average SUV of malignant adrenal nodules was 7.2 (range, 1–39.9), and that of the benign nodules was 1.9 (range, 0.9–4.8) (Table 3, Fig. 3). The mean adrenal-to-liver average SUV ratio of the 82 benign nodules was 0.79 (range, 0.38–2.4). The mean adrenal-to-liver average SUV ratio of the 30 malignant lesions was 3.56 (range, 0.46–21) (Fig. 4).
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
It is well established that CT attenuation values depend on a number of factors, including the use of conventional versus helical technique, peak kilovoltage, calibration, scanner geometry, and the body habitus of patients. Most scans are obtained with helical scanners, and the absolute attenuation suggested in the literature should be used with caution because helical data are manipulated to form cross-sectional images. Since the introduction of helical CT scanners, absolute attenuation does not have a universal value and is more prone to error, particularly if careful calibrations are not performed [23–26].
Chemical shift MRI relies on differences in resonance frequency rates of protons in fat and water molecules to identify lipid-rich adenomas. Because of this difference, signals between lipid and water protons within a voxel are canceled on out-of-phase images. The decrease in signal intensity is proportional to the amount of lipid in the tissue [8, 27, 28]. The use of chemical shift MRI has been found to have sensitivity and specificity in the range of 81–100% and 94–100%. Chemical shift MRI is currently considered the most sensitive noninvasive test for differentiating adenoma from metastasis [6] and is generally used when CT findings are indeterminate. Despite overall success with unenhanced CT and chemical shift MRI, a small but important proportion of adrenal lesions are not accurately characterized with either technique, and biopsy is needed for characterization of the lesions [10]. Kim et al. [9] found that four of 33 adrenal lesions were wrongly categorized as benign on the basis of CT and MRI findings.
Unlike CT and MRI, FDG PET is a functional study in which glucose metabolism is detected. Boland et al. [11], in one of the first studies, reported a sensitivity and specificity of 100%. The use of coregistered PET/CT data enables confident localization of adrenal nodules. This technique is particularly useful in cases in which the adrenal nodule has metabolic activity equal to or less than the background activity. Bagheri et al. [29] found that 68% of 20 normal adrenal glands had increased FDG activity compared with the background and that identification of adrenal glands was difficult with PET alone.
Metser et al. [30] suggested using a maximum SUV of 3.1 for detecting malignant adrenal nodules. Our study, however, revealed considerable overlap with benign adrenal nodules that had low-grade uptake (Fig. 4), as also was found in another study [18]. Thirty-three of 82 benign nodules in our study had a maximum SUV of more than 3.1 (range, 1.2–7), and five of the 82 benign nodules had an average SUV greater than 3.1. This finding is hardly surprising given the nature of SUVs. Measurement of SUV depends on a number of factors, such as recording of dose, noise, resolution of the scanner, time between injection of FDG and acquisition of the scan, patient body weight, and individual variations in patient metabolism. In our study, applying the suggested maximum SUV cutoff value of 3.1 would have substantially decreased the accuracy of the test and would have increased the number of false-positive results.
The use of an internal control for comparison should correct for some of the variables that affect SUV calculations. Both background activity and liver activity have been used as internal controls in previous studies. We chose to use liver activity over background activity because a few previous studies [16–18, 31] showed that an adrenal nodule is more accurately characterized with this method. Because the liver normally has heterogeneous FDG uptake, we used average SUV measured with a 2D ROI placed on an appropriate slice of the right lobe of the liver. We then used a 3D ROI to measure the average SUV of the adrenal nodule by placing the ROI over at least two thirds of the nodule. We used a combination of coronal and axial images to make fine adjustments to the ROI box. The average SUV thus recorded, although perhaps difficult to reproduce, is probably more representative of visual perception but allows more objective comparison than visual analysis alone. Our inclusion of only lesions at least 1 cm in diameter allowed us to place the 3D ROI carefully without much difficulty.
Although at the time of the study, we recorded the attenuation values of all adrenal nodules, we found that several of our patients, particularly those with melanoma, cutaneous lymphoma, and comorbid con ditions, under went scanning with their arms at their sides. This position made the at tenuation values unreliable owing to the streak artifact introduced. Of the 12 false-positive nodules, only two had an attenuation value less than 10 HU and hence arguably would have been correctly classified as benign. However, the other 10 nodules would still have been indeterminate or would have been wrongly classified; 46 of the 82 benign nodules had an attenuation value greater than 10 HU. All malignant nodules measured more than 10 HU, including the five false-negative nodules. With this finding in mind, we decided not to use the additional information offered by unenhanced CT attenuation values to further characterize the adrenal nodules. Moreover, the utility and limitations of the attenuation values of adrenal glands on unenhanced CT have been reported extensively in the literature.
The prevalence of malignant nodules in our study population was 26.8%, a rate comparable with those in other published reports, such as a study in which the pre valence was 27% in postmortem examinations of patients with malignant disease [3]. The mean size of the benign lesions was 1.8 cm, and the mean size of the malignant lesions was 4.03 cm. This finding is in keeping with published findings that benign nodules are significantly smaller than malignant nodules. Likewise, we found that the mean size of PET-negative nodules was less than the mean size of PET-positive nodules (1.81 vs 3.7 cm). This finding occurred because there was a substantially higher proportion of malignant nodules in the PET-positive group. It is interesting to note, however, that the five PET-negative malignant nodules were significantly smaller (mean diameter, 1.3 cm; range, 1–1.5 cm) than the average size of malignant lesions. This finding is important and should be interpreted in light of the resolution of the PET scanner. It has been estimated that the intrinsic resolution of a Discovery DST PET scanner (GE Healthcare) is approximately 6.5 mm [32]. For objects less than twice this size (1.3 cm), we should expect some partial volume effect and underestimation of SUV. However, four of the five false-negative lesions were 1.3 cm in diameter or larger, large enough to be accurately characterized at PET/CT.
It would seem logical to argue that because nodules smaller than 1 cm in diameter were excluded in our study, the sensitivity and specificity of 83.3% and 85.4% should be greater than those reported in previous studies. This difference may have a few explanations. First, unlike some studies of PET/CT including only patients with lung cancer [13, 17], our study was performed with a heterogeneous group of patients with all types of malignant diseases. It can be argued that a group of patients with lung cancer is more likely to have adrenal metastasis than is a group of patients with different cancers, potentially introducing interpreter bias. However, 50% of our patients with a false-positive adrenal nodule had a bronchogenic primary tumor (Table 2). Second, the patients with four of five PET-negative malignant nodules had undergone previous therapy, and the nodules were probably rendered metabolically inactive. Third, we can argue that any low-grade metabolic activity of benign nodules was better recorded in our study group owing to the larger size of the nodules that we included. Unlike other investigators [17, 18], we did not include any nodule that was smaller than 1 cm in diameter. We used this criterion because the spatial resolution of most PET scanners is closer to 1 cm and because in theory inclusion of nodules smaller than 1 cm can render sensitivity and specificity that may not accurately represent the capacity of the instrument, increasing the number of false-negative lesions. Kumar et al. [17] reported that three of five false-negative lesions were smaller than 11 mm in diameter.
We found that nearly 15% of benign adrenal nodules had low-grade metabolic activity equal to or greater than the metabolic activity of the liver (average SUV range, 1.8–4.8). Several other investigators have reported increased metabolic activity in benign adrenal nodules. Erasmus et al. [13], in a study of 33 adrenal nodules in patients with bronchogenic carcinoma, found that two adenomas had increased metabolic activity, bringing the specificity down to 80%. Maurea et al. [14] reported a specificity of 93% due to benign pheochromocytoma that exhibited FDG uptake. In a study of adrenal lesions in patients with lung cancer, Kumar et al. [17] found four benign lesions that had metabolic activity equal to or greater than that of the liver and reported a specificity of 90%. Finally, a study by Blake et al. [18] showed that according to PET criteria, two benign nodules were indeter minate com pared with liver activity. In our study, the slightly higher proportion of false-positive results and the lower specificity (85.4%) than in other studies can be explained by our size criterion.
Of the five malignant nodules in our study that were PET negative, four were found in patients who had undergone therapy before they presented for baseline PET/CT at our institution. It is important to consider this factor in interpretation of PET/CT scans. Although we did not account for this factor, our negative predictive value was very high (93.3%). Had we excluded previously treated patients from our cohort, the negative predictive value would have been even higher.
Our study had a few important limitations. First, like many similar studies, it was conducted retrospectively. Second, we used follow-up data to determine the nature of the adrenal nodules because we had histologic proof of only six of the 112 nodules. This situation is reflective of current practice. Most adrenal nodules are characterized with either follow-up data or imaging findings, and biopsy is reserved for a select few indeterminate lesions [33]. Moreover, the presence of other metastatic lesions makes biopsy irrelevant in a considerable proportion of our patients. Third, we recognize that previous therapy is an important factor to be considered in interpretation of PET/CT scans. Although a history of therapy was probed for patients with false-negative adrenal lesions, we are unable to express the accuracy for patients who underwent previous chemotherapy at another institution. This question is an important and interesting one to be answered in the future. Finally, we cannot apply our results to nodules smaller than 1 cm in diameter because such small nodules were excluded from our study.
On the basis of all our findings, we conclude that PET/CT is useful for characterizing adrenal nodules. The technique has a high negative predictive value, par ticularly for lesions that are larger than 1.3 cm when the findings are interpreted in light of previous chemotherapy. It is important to be aware, however, that a small but important number of benign nodules exhibit low-grade FDG avidity. In the cases of patients for whom the accuracy of the diagnosis is critical, percutaneous biopsy continues to have a role, and PET/CT cannot assume the role of a reference standard in determining the nature of adrenal nodules.
Acknowledgments
We thank Karen F. Philips for help in reviewing the manuscript and Osama
Mawlawi for advice.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  K. TAMURA, K. YOSHIKAWA, H. ISHIKAWA, M. HASEBE, H. TSUJI, T. YANAGI, K. SUZUKI, A. KUBO, and H. TSUJII Carbon-11-Methionine PET Imaging of Choroidal Melanoma and the Time Course after Carbon Ion Beam Radiotherapy Anticancer Res, May 1, 2009; 29(5): 1507 - 1514. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
