| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Nuclear Medicine |
1 From the Departments of Radiology (B.F.B., T.M.B., S.R., C.C.M.), Otolaryngology (B.F.B., L.A.Z., C.H.S., J.T.J.), Neurology (C.C.M.), and Psychiatry (C.C.M.), University of Pittsburgh School of Medicine, 200 Lothrop St, PUH Rm D-132, Pittsburgh, PA 15213. Received January 23, 2004; revision requested April 2; final revision received October 4; accepted October 12. Address correspondence to B.F.B. (e-mail: bfb1@pitt.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Study was approved by the institutional review board, and patient informed consent was waived. Sixty-five consecutive patients (42 men, 23 women; age range, 43–83 years) known to have or suspected of having head and neck cancer were examined with combined PET/CT. CT was performed with intravenous administration of a contrast agent, and the CT data were used for attenuation correction. Each examination was interpreted in three ways: PET images in the absence of CT data, CT images in the absence of PET data, and fused PET/CT images. Probability of malignancy of each lesion was assigned a score by using a five-point scale. Receiver operating characteristic (ROC) analyses were performed by using biopsy, imaging, or clinical follow-up as the reference standard. The minimum follow-up was 6 months (range, 6–12 months). The results were additionally analyzed to assess the degree of radiologist confidence.
RESULTS: Follow-up was available for 64 (98%) of 65 patients. ROC analyses demonstrated that PET/CT is significantly superior to PET or CT alone for depiction of malignancy in the head and neck (P < .05). In this series, PET/CT had a sensitivity of 98%, a specificity of 92%, and an accuracy of 94%. Radiologist confidence was substantially higher with the combined modality.
CONCLUSION: Combined PET/CT is more accurate than PET or CT alone for the depiction of malignancy in the head and neck. Radiologist confidence is improved with the combined modality.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Traditional radiographic modalities are insensitive in depicting small metastases or early posttreatment recurrence because they rely on morphologic changes that can be slow to progress (2). Functional imaging modalities, such as dynamic magnetic resonance (MR) imaging (3), Doppler sonography (4), and positron emission tomography (PET) (5) seek to address this deficiency by interrogating the physiologic properties of the tissues. Advanced MR techniques such as magnetization transfer, T1
-weighted imaging, and tissue-specific contrast agents have also been used to identify cancer of the head and neck (6–8). However, none of these techniques have definitively emerged as a superior option for detection of malignancy.
The effectiveness of using PET with fluorine 18 fluorodeoxyglucose (FDG) has been documented in many different types of cancer (9–11), but the poor spatial resolution of PET is particularly limiting within the intricate anatomy of the head and neck. Lack of anatomic landmarks on PET scans can further hinder localization of FDG uptake. Also, the variable physiologic uptake of FDG in the head and neck can decrease the specificity of PET interpretations (12). Some tumors in the head and neck, particularly those of glandular origin, have intrinsically variable FDG uptake, making PET less reliable for the detection of malignancy. Combined PET/CT scanners overcome some of these limitations by fusing the anatomic data of CT with the functional data of PET (13,14). The benefits of fused PET/CT have been established in several regions of the body (10,11,15,16), and PET/CT has been shown to be superior to whole-body MR imaging for staging of malignancies (17), but there has been no large prospective study in which the accuracy of combined PET/CT in the head and neck was addressed. Combined PET/CT may be particularly useful in the head and neck because of the intricate anatomy, variable tumor physiology, and unpredictable normal FDG uptake in head and neck tumors.
Thus, the purpose of our study was to prospectively determine whether combined PET/CT is more accurate than either PET or CT alone in depicting malignant lesions in the head and neck.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Imaging
PET and CT imaging were performed with two commercial combined PET/CT scanners. The first scanner combines a septa-less three-dimensional PET scanner (EXACT HR+; CPS Innovations, Knoxville, Tenn) with a helical CT scanner (Somatom Emotion; Siemens Medical Solutions, Forscheim, Germany). The second scanner (Reveal; CTI Medical Systems, Knoxville, Tenn) combines a dual-channel CT with a PET scanner (LSO Allegra; CTI Medical Systems). PET/CT imaging from the skull base to the abdomen was performed approximately 1 hour after an intravenous injection of 8–15 mCi (296–555 MBq) of FDG.
Patients were instructed to fast for 4–6 hours prior to scanning. Whole-blood glucose levels were measured prior to injection. In accordance with our clinical protocol, patients with a moderately elevated blood glucose level (200–250 mg/dL [11.1–13.9 mmol/L]) intravenously received insulin (2–4 U) prior to FDG injection; patients with a markedly elevated blood glucose level (>250 mg/dL [>13.9 mmol/L]) were rescheduled.
Helical CT (pitch of 1.0, 120–140 mAs, 130 kVp) was performed immediately preceding the acquisition of PET emission data. A contrast material (120 mL of ioversol, Optiray-350; Mallinckrodt, St Louis, Mo) was intravenously injected in all patients unless it was contraindicated because of a severe allergy to the contrast material.
PET images (5 minutes per bed position, four to five bed positions per subject) were reconstructed with and without attenuation correction. Attenuation correction was based on the CT attenuation coefficients, which were computed by using iterative algorithms (18). Helical CT scans were reconstructed with a section thickness of 2.4 mm and a matrix of 512 x 512 to match the parameters of the PET scans.
Image Interpretation
Transverse CT images were reviewed on a picture archiving and communication system workstation (Stentor, Brisbane, Calif). PET images; fused PET/CT images; and reconstructions of PET, CT, and PET/CT into sagittal and coronal planes were reviewed on a workstation that features Syngo Fusion software (Siemens Medical Solutions).
Each examination was subjected to three analyses. A board-certified neuroradiologist (B.F.B.) with 2 years of head and neck experience after dedicated head and neck fellowship reviewed CT data without access to PET data. A board-certified neuroradiologist (C.C.M.), also board-certified in nuclear medicine, with 16 years of PET experience after dedicated PET fellowship reviewed PET images without access to CT data. Each reviewer had access to multiplanar reconstructions and to all prior clinical and radiographic data, including prior PET/CT scans. Clinical data were supplied by means of our electronic medical records, as would be available for clinical readings. On completion of the independent PET and CT ratings, the two reviewers rendered a consensus opinion and had complete access to the fused PET/CT images.
A lesion was defined as a radiologic abnormality if a tumor was suspected or could not be reasonably excluded. Each lesion was assigned a score according to a five-point scale: score of 1, definitely benign; score of 2, probably benign; score of 3, equivocal; score of 4, probably malignant; and score of 5, definitely malignant. Each examination was also assigned a score of an overall likelihood of tumor presence, defined as the maximum value of the lesions seen at that examination, and was graded with the same five-point scale. If a lesion was not appreciated with one of the modalities but needed to be included in the analysis because it was seen with another modality, the unappreciated lesion was assigned to a "not visible" category. For the purpose of statistical analysis, "not visible" lesions were combined with "definitely benign" lesions in category 1. Subcentimeter lung lesions were excluded from the analysis because the large number of such lesions would have biased our results. No such lesions had appreciable FDG uptake.
Reference Standard
Patients were followed up for a minimum of 6 months. The average follow-up was 8.9 months (range, 6–12 months). Biopsy, imaging follow-up (contrast material–enhanced CT or PET/CT), and clinical follow-up were all used as reference standards for determining the presence or absence of a tumor. Histologic appearance of the tumor was determined at biopsy in all patients.
Statistical Analysis
Receiver operating characteristic (ROC) curves were generated to determine the differences in accuracy between the three modalities (CT vs PET vs PET/CT). One set of ROC curves was generated for the likelihood of malignancy in each lesion, and another set of ROC curves was generated for the likelihood of malignancy in each patient. Both of the two ROC analyses were necessary to account for statistical clustering of lesions (eg, a patient with an identifiable metastasis is more likely to have other identifiable metastases). The ROC curves were compared with an area test (Rockit 0.9B; C. E. Metz, University of Chicago, 1998) (19). The data were considered ordinal and were correlated. A two-tailed P value was used to compare PET with CT, while a one-tailed P value was used to compare PET with PET/CT and CT with PET/CT. This decision was based on a prospective determination that a combined modality should be equal or superior to its components. P value of .05 or less was considered to indicate a statistically significant difference. Although ROC curves are a powerful statistical technique, it is useful to calculate sensitivity and specificity values for comparison with other diagnostic tests. Sensitivity, specificity, and accuracy were calculated with equivocal cases considered positive, reflecting the probable clinical use of equivocal results.
A separate analysis was conducted to establish whether radiologist confidence was altered by the combined modality. Indefinite ratings (categories 2–4) for CT and PET were evaluated to determine whether the fused images allowed these lesions to be placed in a definitive category (categories 1 and 5).
An unpaired Student t test was used to determine whether there were significant differences in age between male and female patients.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
A total of 125 lesions were identified in 58 (91%) of the 64 patients. Of the 125 lesions, 46 (37%) were identified with CT but not with PET, 26 (21%) were identified with PET but not with CT, and 50 (40%) were identified with both modalities; three (2%) lesions were identified only with fused PET/CT. Twenty lesions were evaluated at biopsy, 15 were evaluated at imaging, and 90 were evaluated clinically. Clinical and imaging follow-up revealed no additional lesions that had been missed on the PET/CT scans.
The ROC curves are shown in Figure 1. The statistical analysis is summarized in Table 1. In the assessment of lesions, areas under the ROC curves at CT, PET, and PET/CT were 0.75, 0.92, and 0.99, respectively. In the assessment of patients, areas under the ROC curves at CT, PET, and PET/CT were 0.87, 0.96, and 0.98, respectively. In the assessment of individual lesions, PET/CT was statistically superior to PET (P < .05) and CT (P < .05). In the assessment of the overall tumor presence in a patient, PET/CT was statistically superior to CT (P < .05). Although PET/CT performed better than did PET in the assessment of the overall tumor presence, the difference did not reach statistical significance in our population (P = .14).
|
|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Combined PET/CT scanners have become commercially available only in the past few years. Institutions with PET scanners may be reluctant to purchase a combined scanner without data that demonstrate an improvement in patient care. The results of this study indicate that combined scanners provide a substantial improvement in the detection and evaluation of head and neck malignancies.
PET had a particular utility in the immediate posttreatment interval, when CT often cannot reliably help distinguish residual tumor from scar or radiation change. Half of the studies in which PET depicted malignancy that was overlooked at CT occurred within 4 weeks of therapy. There were fewer examples of lesions seen at CT that were missed at PET. These were typically tumors with low FDG avidity, such as tumors of glandular origin. The major advantage of CT was its ability to correctly help characterize physiologic uptake that would otherwise have been mistaken for tumor at PET alone. CT was also useful for localization of PET abnormalities, particularly in patients with surgically-altered anatomy.
There were several potential sources of bias in our study. Although the CT component of our combined scanner was of diagnostic quality, the images were not as detailed as those acquired with a dedicated CT scanner. Acquisition field of view cannot be tailored to smaller body parts such as the neck while still accommodating the entire abdomen. Extremity position also cannot be tailored to a body part; the thorax is imaged with the arms at the patient’s side to reduce artifact in the head and neck. However, the difference in accuracy between CT and PET/CT suggests that these minor issues of CT quality would not change the results of our study. Furthermore, the analysis of the errors encountered at CT suggests that these errors would not have been prevented with a dedicated CT.
For PET/CT scanners, CT attenuation maps are used to correct for PET attenuation rather than an additional transmission scan. This introduces a variety of artifacts not seen on transmission-corrected PET scans, so that our PET scans are not completely comparable to PET scans obtained without CT (20–22). The use of intravenous CT contrast material further complicates the analysis by introducing additional attenuation-correction artifacts (23). However, the interpreting nuclear medicine physician had access to uncorrected images throughout our study. In our experience, a radiologist familiar with the artifacts seen on CT-corrected PET scans will not lose clinical accuracy, compared with artifacts seen on transmission-corrected PET scans (24).
Having two reviewers come to a consensus regarding the PET/CT images may have provided additional bias. In many clinical settings, only one radiologist reviews a fused PET/CT examination. The addition of a second set of eyes in the research environment could have improved the overall accuracy of PET/CT in our study. In fact, however, we would recommend that whenever practical, all clinical interpretations of fused PET/CT examinations of the head and neck be reviewed by radiologists with expertise in nuclear medicine and neuroradiology, as is done at our institution. We intentionally provided the reviewers in this study with all the pertinent clinical and prior radiographic information, because we believed that this would better reflect the actual clinical practice.
Another potential criticism of this study is that we did not stratify our results to compare different histopathologic conditions. Although squamous cell carcinoma has a consistently high FDG avidity, glandular tumors have variable FDG uptake and thus may have different imaging requirements. Because the majority of our patients had squamous cell carcinoma, and squamous cell carcinoma is by far the most frequent tumor of the head and neck, we chose our follow-up interval on the basis of the clinical behavior of squamous cell carcinoma. Glandular tumors may require longer follow-up. Future studies are needed to evaluate the accuracy of PET/CT for different histopathologic conditions in the head and neck. We believe that the results of this study reflect the true prevalence of different tumors in our patient population.
The results of this study are in general agreement with those of previous studies of the head and neck in which PET was compared with PET/CT (25). Our study adds further quantitation, with ROC analysis applied to a similar clinical situation. The prospective nature of our study also adds to the reliability of these results.
Although the comparison of PET with PET/CT did not reach statistical significance for the overall presence of a tumor in a patient, we believe that the difference may reach statistical significance with the inclusion of more patients. Even in the absence of a significant improvement in accuracy, however, the improvement in radiologist confidence is sufficient, we believe, to justify the use of combined PET/CT.
There are several advantages of PET/CT that are not easily quantified. We did not attempt to account for the improved localization with PET/CT over that of PET alone; lesions seen at PET were not penalized for improper localization. We also did not specifically analyze the ability to accurately evaluate the extent of a tumor. Because we did not attempt to quantify these advantages, our results may, in fact, underestimate the value of combined PET/CT.
ROC curves are a powerful statistical technique for comparing the accuracy of examinations without relying on the arbitrary cutoffs needed for sensitivity and specificity measurements. However, sensitivity and specificity are useful for comparing modalities from different trials. With a sensitivity of 98% and a specificity of 92%, PET/CT was an extremely accurate test for head and neck malignancy in this study.
Post hoc fusion of CT and PET data is another method for combining functional and anatomic data. To our knowledge, there have been no studies in which post hoc fusion was directly compared with combined PET/CT imaging. However, we believe that the artifacts introduced by repositioning the patient for a second examination negate many of the advantages of the fused images, particularly in the head and neck where slight misregistration can be critically important.
In conclusion, combined PET/CT is potentially more accurate than PET or CT alone for the evaluation of malignancy in the head and neck. Radiologist confidence is improved with the combined modality. More research is needed regarding the effectiveness of PET/CT in unusual (nonsquamous cell carcinoma) tumors of the head and neck.
|
FOOTNOTES |
|---|
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, B.F.B.; study concepts and design, B.F.B., C.C.M.; literature research, B.F.B., C.C.M, T.M.B.; clinical studies, B.F.B., C.C.M.; data acquisition, all authors; data analysis/interpretation, C.C.M., B.F.B.; statistical analysis, B.F.B.; manuscript preparation, B.F.B.; manuscript definition of intellectual content, B.F.B., C.C.M.; manuscript editing, revision/review, and final version approval, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Comoretto, L. Balestreri, E. Borsatti, M. Cimitan, G. Franchin, and M. Lise Detection and Restaging of Residual and/or Recurrent Nasopharyngeal Carcinoma after Chemotherapy and Radiation Therapy: Comparison of MR Imaging and FDG PET/CT Radiology, August 18, 2008; (2008) 2491071753. [Abstract] [Full Text]
|
|
|
|
 |
|
 Y-H CHEN, J J-M JIAN, K-Y CHAN, S Y TSAI, S H CHENG, K-C L YEN, and J C-H CHENG Definitive chemoirradiation for resectable head and neck cancer: treatment outcome and prognostic significance of MRI findings Br. J. Radiol., June 1, 2008; 81(966): 490 - 498. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Boughanim, S. Leboulleux, A. Rey, C. Tuan Pham, Y. Zafrani, P. Duvillard, J. Lumbroso, C. Haie-Meder, M. Schlumberger, and P. Morice Histologic Results of Para-Aortic Lymphadenectomy in Patients Treated for Stage IB2/II Cervical Cancer With Negative [18F]Fluorodeoxyglucose Positron Emission Tomography Scans in the Para-Aortic Area J. Clin. Oncol., May 20, 2008; 26(15): 2558 - 2561. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. C. Ong, H. Schoder, N. Y. Lee, S. G. Patel, D. Carlson, M. Fury, D. G. Pfister, J. P. Shah, S. M. Larson, and D. H. Kraus Clinical Utility of 18F-FDG PET/CT in Assessing the Neck After Concurrent Chemoradiotherapy for Locoregional Advanced Head and Neck Cancer J. Nucl. Med., April 1, 2008; 49(4): 532 - 540. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. W. Fletcher, B. Djulbegovic, H. P. Soares, B. A. Siegel, V. J. Lowe, G. H. Lyman, R. E. Coleman, R. Wahl, J. C. Paschold, N. Avril, et al. Recommendations on the Use of 18F-FDG PET in Oncology J. Nucl. Med., March 1, 2008; 49(3): 480 - 508. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. N. Shah and B. F. Branstetter IV Oncocytoma of the Parotid Gland: A Potential False-Positive Finding on 18F-FDG PET Am. J. Roentgenol., October 1, 2007; 189(4): W212 - W214. [Full Text] [PDF]
|
|
|
|
|
|
A. Shammas, B. Degirmenci, J. M. Mountz, B. M. McCook, B. Branstetter, B. B. Bencherif, J. M. Joyce, S. E. Carty, H. A. Kuffner, and N. Avril 18F-FDG PET/CT in Patients with Suspected Recurrent or Metastatic Well-Differentiated Thyroid Cancer J. Nucl. Med., February 1, 2007; 48(2): 221 - 226. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. M. Blodgett, C. C. Meltzer, and D. W. Townsend PET/CT: Form and Function Radiology, February 1, 2007; 242(2): 360 - 385. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Czernin, M. Allen-Auerbach, and H. R. Schelbert Improvements in Cancer Staging with PET/CT: Literature-Based Evidence as of September 2006 J. Nucl. Med., January 1, 2007; 48(1_suppl): 78S - 88S. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. D. Berry and G. J. R. Cook Positron emission tomography in oncology Br. Med. Bull., December 14, 2006; (2006) ldl013v1. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. F. Branstetter IV Physiologic 18F-FDG uptake in the serratus anterior muscle: a potential false-positive on PET resolved with PET/CT. Am. J. Roentgenol., August 1, 2006; 187(2): W235 - W235. [Full Text] [PDF]
|
|
|
|
 |
|
 S. K. Mukherji and C. R. Bradford Controversies: Is There a Role for Positron-Emission Tomographic CT in the Initial Staging of Head and Neck Squamous Cell Carcinoma? AJNR Am. J. Neuroradiol., February 1, 2006; 27(2): 243 - 245. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
