Integrated PET-CT: Evidence-Based Review of Oncology Indications
Integrated PET-CT: Evidence-Based Review of Oncology Indications
Over the past few years, positron-emission tomography (PET) has been increasingly used in oncology. More recently, the fusion of the functional imaging provided by PET with the anatomic information of, computed tomography (CT) has resulted in a powerful new imaging tool. In practice, the PET-CT image fusion can be done visually, with the help of special software, or through hardware incorporating both PET and CT. We performed a review of disease-specific studies using integrated PET-CT or PET-CT with visual or software fusion and found that PET-CT offers a significant benefit in improving the diagnosis, staging, and therapeutic monitoring of patients with cancer. Integrated PET-CT improves characterization of equivocal lesions and decreases intraobserver variability; it also has the potential to significantly affect treatment planning by guiding biopsies and surgical interventions, defining target volumes for radiation therapy fields, and monitoring response to treatment. More promising clinical applications will likely be possible with other fluorine-18-based isotopes.
Combined-modality positronemission tomography (PET)- computed tomography (CT) is becoming the imaging method of choice for an increasing number of oncology indications. The goal of this paper is to review the evidence-based literature justifying PET-CT fusion. The best evidence comes from prospective studies of integrated PETCT scans compared to other methods of acquiring images, with histopathologic confirmation of disease presence or absence. Unfortunately, very few studies provide this kind of data. Retrospective studies with similar comparisons can be used to provide evidence favoring the use of integrated PET-CT scans in specific clinical situations. Also, inferential conclusions can be drawn from studies where clinical rather than pathologic data are used to establish disease presence or absence. PET is a diagnostic examination from the acquisition of physiologic images based on the detection of positrons emitted by a glucose analog, fluorine-18 (18F)-fluorodeoxyglucose (FDG). Due to increased glucose transport and increased glycolysis in the malignant cells, 18F-FDG is accumulated and subsequently trapped within malignant cells to a greater extent than normal cells. However, FDG is not tumor-specific, as it can also accumulate in benign and inflammatory lesions. Integrated PET-CT PET imaging is based on the physiologic distribution of the tracer and, therefore, has poor spatial localization compared to conventional x-rays, CT, or MRI. As an alternative to PET scans, fusion of PET with CT images has been developed in order to improve the accuracy of PET and provide both functional and anatomic information. Image fusion can be achieved as a visual, software-based, or hardware-based process. In visual image fusion, PET and CT images obtained separately are viewed next to each other and the visual fusion occurs in the reader's mind. In software fusion, PET and CT images obtained separately are coregistered with the help of special software, but this is difficult and timeconsuming. Software fusion works well for brain studies, as brain is covered by rigid skull, which provides ample landmarks for fusion, and there is no organ motion during acquisition of functional and anatomic studies. In contrast, such software fusion is less successful in body imaging, because cardiac and respiratory motions can secondarily affect the accuracy of anatomic and functional studies of various organs in the chest and abdomen. Recently, integrated PET-CT scanners (hardware fusion) have been developed, consisting of a CT scanner and a PET scanner assembled together in one machine. Potential advantages of integrated PET-CT scanners are shorter image acquisition time (20- 40 vs 60-90 min), better anatomic localization than achieved with PET alone, and rapid staging of a malignancy in a single study. In preparation for an integrated PET-CT study, the patient must be fasting for at least 6 hours. Diabetics should have a glucose level less than 200 mg/dL, as a high glucose level may potentially produce false-negative results. Patients need to lie still for a total of 30 minutes-approximately 5 minutes for the CT scan followed by 20 to 30 minutes for the PET scan. A dose of 10 to 20 mCi FDG is injected through a peripheral intravenous (IV) line (not a central line, as retained activity in the central line may cause reconstruction streak artifacts). Use of Contrast Agents
The use of oral and IV CT contrast agents for integrated PET-CT scans is controversial. Usually, it is recommended that the CT portion of PETCT be performed without contrast, as there is concern that contrast could yield incorrect attenuation values in the attenuation correction of PET images. However, it is known that contrast- enhanced CT scans are superior to CT alone, especially for visualizing perivascular structures, lymph nodes, hepatic lesions, and so forth. Oral contrast has been used with integrated PET-CT scans with minimal effect on the CT attenuation correction of PET images (slight increase of FDG uptake in the sigmoid colon) and with higher-quality CT images. Antoch et al used IV and oral contrast in the CT scanning of integrated PETCT and reported a significant improvement of CT reading without compromising PET quality. We administer IV contrast to all patients, unless there is a contraindication such as previous allergic reaction or impaired renal function. We have found that IV contrast does not significantly affect the PET images. The major artifact on the PET images is found in the venous vessels carrying the undiluted contrast to the heart. The non-attenuation-corrected images can be used if there is a question of a contrast artifact, as the contrast artifact will only appear on the CT-based attenuation-corrected images and not the non-attenuation-corrected images. We use oral contrast for all patients with suspected malignancies in the abdomen and pelvis and also found that PET images are not significantly affected. As of May 2004, Medicare approved reimbursements for PET in the evaluation of non-small-cell lung, colorectal, head and neck (including thyroid), melanoma, lymphoma, esophageal, and breast cancers. Data Reviewed
Data from disease-specific studies using integrated PET-CT or PET-CT with visual or software fusion will be reviewed for initial diagnosis, staging, detection of recurrence, radiation treatment planning, and response to treatment. When available, comparative data will be reviewed for sensitivity, specificity, and accuracy. The greater the sensitivity of the imaging study, the greater the likelihood that the image will be positive where there is disease present (ie, the less likely there will be a falsely negative scan). The greater the specificity of the imaging study, the greater the likelihood that the image will be negative when there is no disease present (ie, the lower the chance that there will be a falsely positive scan). The greater the accuracy of the scan, the higher the probability of both true-positive and true-negative images. Table 1 summarizes the sensitivity, specificity, and accuracy of PET-CT in different malignancies as determined from our literature search. Lung Cancer The literature has more evidence supporting the use of PET in lung cancer than in any other malignancy. Integrated PET-CT significantly increases the number of patients with correctly staged non-small-cell lung cancer (NSCLC) and guides treatment. In a landmark prospective study, Lardinois et al compared the accuracies of integrated PET-CT, CT alone, PET alone, and visual correlation of PET and CT scans in the staging of 50 patients with NSCLC. As a reference, the investigators used histopathologic assessment or at least one other imaging method. Compared with visual correlation, integrated PET-CT provided 24 items of additional information in 41% of patients. Integrated PET-CT was found to be more accurate than PET alone in nodal staging. Moreover, tumor staging was significantly more accurate with integrated PET-CT than with CT alone, PET alone, or visual correlation of PET and CT. A subsequent study by Lardinois et al found that of the 300 NSCLC patients staged with integrated PET-CT, approximately one-third had unsuspected extrathoracic lesions, and only 63% of these lesions were NSCLC metastases, suggesting the caveat that PET-positive extrathoracic lesions may mimic metastases of NSCLC. Buck et al prospectively evaluated 128 patients with lung lesions (100 malignant and 28 benign tumors). Integrated PET-CT compared to PET for tumor staging had similar sensitivity (99% vs 98%), greater specificity (75% vs 46%), and better accuracy (94% vs 87%). Integrated PET-CT was similar to PET for nodal involvement (74% sensitivity, 92% specificity, and 86% accuracy). For differentiation of operable (N0-N2) vs inoperable (N3) patients, the accuracy of PETCT was 96%, compared with 91% for PET. In another study of 27 NSCLC patients, Antoch et al found integrated PET-CT to be more accurate in overall tumor staging than PET or CT alone. Primary tumor stage was correctly determined in more patients by integrated PET-CT than by PET or CT alone. In detecting metastasis to lymph nodes, PET-CT was similar to PET and better than CT, with a sensitivity of 89% vs 89% vs 70%, specificity of 94% vs 89% vs 59%, and accuracy of 93% vs 89% vs 63%. Also, more distant metastases were detected with integrated PET-CT scan, than with PET or CT alone. Schaffler et al studied 92 NSCLC patients with pleural abnormalities on CT. Their findings suggest that a negative PET scan for indeterminate pleural abnormalities on CT indicates a benign character, whereas positive findings on PET scan are sensitive for malignancy. Visual fusion PETCT was similar to PET in detection of pleural malignancies. Other than the role in lung cancer staging, coregistered PET-CT has a very important role in radiotherapy planning in these patients. The addition of PET to CT imaging produced changes of 22% to 64% in the treatment volume in 22% to 100% of patients studied.[7,8] Studies have also shown that PET is a useful test for staging small-cell lung cancer (SCLC), potentially modifying both the stage and management of these patients. In our literature search, we could not find any studies of integrated PET-CT in staging SCLC. Breast Cancer Initial Staging
At present, PET is not indicated for breast cancer screening and diagnosis of primary tumor. PET sensitivity is low for breast tumors less than 1 cm in diameter and for certain tumor types such as lobular carcinoma and in situ carcinoma, which are less FDG-avid. In addition, PET is not sufficiently accurate in axillary lymph node staging, as small axillary metastases are frequently missed by the procedure. Therefore, mammography remains the imaging modality of choice for screening of breast cancer and evaluation of breast lesions, and lymph node dissection or sentinel node biopsy the most reliable techniques in staging the axilla. Wang et al used PET-CT for initial staging of 15 patients with breast lesions ranging from 3 to 8 cm. Sensitivity, specificity, and accuracy were 93%, 91%, and 100% for the diagnosis of the primary tumor and 80%, 90%, and 87% for the detection of lymph node metastases, suggesting that PET-CT has higher accuracy than mammography, ultrasound, or PET alone in the initial staging of patients with breast tumors larger than 3 cm. Restaging and Follow-up
The main applications of PET in breast cancer are in restaging and treatment monitoring, as PET has a high sensitivity, specificity, and accuracy in diagnosing recurrent or metastatic breast cancer (Figure 1). A recent study compared PET, PET-CT hardware fusion, and PETCT software fusion in restaging 56 patients with breast cancer. Integrated PET-CT and PET had similar sensitivity (71% vs 66%), specificity (68% vs 75%), and accuracy (70% vs 70%) in detecting residual/recurrent breast cancer. Software fusion was successful in almost 100% of patients and yielded a similar accuracy when compared to hardware fusion. The lower-than-expected accuracy of both PET and PET-CT was explained by the high prevalence of lobular carcinoma (16% of patients) in the study population. Another retrospective study used integrated PET-CT in 82 patients suspected of having breast cancer recurrence because of rising serum tumor markers and/or equivocal imaging findings. Sensitivity, specificity, and accuracy of PET-CT were respectively 90%, 66%, and 85%. The study revealed an impact on management in 85% of patients. Buck et al found that integrated PET-CT changed the management in 36% of 78 breast cancer patients with rising tumor markers. Similarly, Tatsumi et al reported that integrated PET-CT added incremental value to PET alone in 40% of the 60 breast cancer patients studied retrospectively for initial tumor staging, recurrence, and follow-up. Gastrointestinal Malignancies Esophageal and Gastric Cancers
PET scanning detects unrecognized metastatic disease and also predicts response early in the course of therapy for patients with esophageal cancer. Very early studies revealed that combining PET with CT improves the diagnostic accuracy of CT. One prospective study of 26 patients with esophageal or gastroesophageal junction cancers found that the accuracy in determining resectability was 65% for CT vs 88% for PET, and 92% for PET-CT. PET and CT together would have decreased unnecessary surgery by 90%. In a recent retrospective study of 43 patients with newly diagnosed esophageal cancer, staging accuracy improved from 83% with PET to 93% with integrated PETCT.[ 15] Bar-Shalom et al studied 18 esophageal cancer patients and found that PET correctly diagnosed the status of malignancy in 78% of patients. Combined PET-CT improved detection and characterization of suspicious sites on PET or CT in 89% of patients and affected management in 22%. Gastrointestinal Stromal Tumors
PET-CT has an important role in the assessment of gastrointestinal stromal tumor (GIST) response to imatinib (Gleevec), as CT alone may not reveal a response until several months after the start of treatment. Antoch et al studied the response to imatinib in 20 GIST patients, by performing PETCT before and 1, 3, and 6 months after starting treatment. PET-CT detected more lesions than PET or CT alone. Assessment of tumor response at 1 month was accurate in 95% of patients by integrated PET-CT, 90% by visual fusion PET-CT, 85% by PET, and 44% by CT. Integrated PETCT, visual fusion PET-CT, and PET alone accurately diagnosed tumor response in 100% of patients at 3 and 6 months, whereas CT was found to be accurate in only 60% at 3 months and 57% at 6 months. Thus, integrated PET-CT is the imaging test of choice for evaluation of GIST response to imatinib. Colorectal Cancer
PET has a definite role in the early detection of recurrent colorectal cancer and a limited role in primary staging of this cancer. Studies have shown that PET has a higher accuracy than CT in the diagnosis of recurrent colorectal cancer (Figure 2). Newer studies have assessed the role of PET-CT in colorectal cancer. Burger et al studied 65 patients with recurrent colorectal carcinoma. PET-CT was better than PET for diagnosing local recurrence (sensitivity of 96% vs 77% and specificity of 97% vs 89%). PET-CT was also better than PET for the detection of distant metastases (sensitivity of 95% vs 66% and specificity of 98% vs 79%). PETCT also improved interobserver agreement. In a retrospective study of 45 colorectal cancer patients, staging and restaging accuracy improved from 78% with PET to 89% with integrated PET-CT. Another retrospective study comparing integrated PET-CT to PET in the initial staging of 35 colorectal cancer patients found that PET-CT changed the stage in 20% of patients and revealed additional findings not seen on other imaging modalities in 17%. PET-CT also increased the reader's confidence in localizing and characterizing lesions. Integrated PET-CT is also an accurate technique for the detection of pelvic recurrence after surgical removal of rectal cancer, as shown in a study of 62 patients. Integrated PET-CT was better than PET for differentiating malignant from benign FDG uptake in the pelvis, with a sensitivity of 98% vs 82%, specificity of 96% vs 65%, and accuracy of 93% vs 74%. Pancreatic Cancer
Initial studies showed that PET has a higher sensitivity, specificity, and accuracy than CT in diagnosing pancreatic carcinomas. PET is also more accurate than CT in identifying malignant pancreatic cystic lesions. Newer studies have compared PET-CT with PET in pancreatic cancers. Lemke et al studied 104 patients with suspected pancreatic lesions. PET-CT software fusion had a higher sensitivity for malignancy detection than PET or CT (89% vs 84% vs 77%), but did not improve specificity (64%). All image modalities failed to stage lymph node involvement. Another study of 28 patients with pancreatic cancer found that integrated PET-CT improved diagnostic certainty on 20% of sites in 42% of patients with positive PET findings.