Adding body-coil image acquisition to PET/MRI does not entirely match the diagnostic accuracy of standard low-dose PET/CT, and may only serve as a backup scanning option in a small selection of oncology patients, according to a Swiss study.
The prospective study from University Hospital Zurich also found that more time is needed on the MRI to accommodate breath-holds and additional surface coil sequences to match with the standard low-dose PET/CT (Insights into Imaging, 15 May 2013). Lead author Dr. Philippe Appenzeller is from the division of diagnostic and interventional radiology.
Over the last several years, clinical interest in PET/MRI for oncological imaging has increased, given MRI's ability to image soft tissue and concerns about radiation exposure to patients through PET/CT. However, before PET/MRI can replace PET/CT for some clinical indications, Appenzeller and colleagues noted that MRI is "routinely done with surface coils to achieve sufficient resolution as well as signal density."
Even when routine clinical body-coil imaging alone is not adequate to achieve sufficient image quality, there "might be a clinical situation in multimodality imaging where it might be desirable to have a quick whole-body PET/MRI," the authors explained. So, the purpose of their study was to determine the diagnostic utility of a PET/MRI with only an axial T1-weighted fast-gradient-echo sequence using a body coil, which is needed for MRI-based attenuation correction, compared with standard low-dose PET/CT.
Patient cohort
A total of 63 patients with a mean age of 58 (ranging from 19 to 86 years; 23 women, 40 men) referred for either staging or restaging/follow-up of various malignant tumors participated in this prospective study. Sequential PET/CT and MR imaging were performed on a trimodality PET/CT and MRI configuration (Discovery PET/CT 690 and Discovery MR 750, GE Healthcare). The researchers first evaluated PET/CT results and the PET/MRI scans for the presence of PET-positive and PET-negative lesions. Lesions were considered PET-positive if standardized uptake value (SUV) was distinctively greater than the surrounding background activity.
The study used a four-point scale to assess lesion conspicuity. The rating of 1 indicated more than 25% of a lesion's borders could be defined; 2 was used for 25% to 50% of borders depicted; 3 for 50% to 75 % as definable; and 4 indicated more than 75% of a lesion's borders were defined. Additionally, images were divided into three categories: tumor mass, lymph nodes, and lesions. Differences in overall lesion detectability and conspicuity in PET/CT and PET/MRI were investigated, as well as differences in detectability based on the localization and lesion type.
The researchers also noted corresponding anatomical structure for each lesion based on the anatomical information of the coregistered PET/CT and PET/MRI image sections.
Common cancers
Of the 63 subjects, the most common ailment was malignant melanoma in 11 individuals (17%), lung cancer in nine cases (14%), breast cancer in seven patients (11%), and Hodgkin's lymphoma among six people (9%). The study was forced to exclude 17 patients from the final analysis, because there were no suspicious tumorous lesions. The remaining 46 participants were examined with sequential PET/CT and MRI and were found to have at least one lesion.
The final analysis included 126 lesions, with 101 lesions (80%) visible and measurable on both PET/CT and PET/MRI. There were 17 lesions (17%) visible only on PET/CT, six lesions (6%) seen only on PET/MRI, and two lesions (2%) not visualized on either CT or MRI.
In addition, 37 (29%) of the 126 PET-positive findings were characterized as tumor mass, 57 (45%) were determined to be lymph nodes, and 17 (13%) were lesions. Overall, 111 PET-positive findings were evaluated because not all PET-positive findings were seen on PET/CT and/or PET/MRI. Of the remaining 15 lesions, FDG-uptake without morphological correlation did not fit in the defined categories.
Statistical significance
The Zurich group found no statistically significant superiority of PET/CT over PET/MRI or vice versa in terms of lesion conspicuity (p = 0.095). There was, however, a statistically significant superiority regarding conspicuity of PET/CT over PET/MRI in pulmonary lesions (p = 0.016), and a statistically significant advantage of PET/CT over PET/MRI in lymph nodes with lesion conspicuity (p = 0.033).
"A higher mean score concerning bone lesions was found for PET/CT compared with PET/MRI," the authors noted. "However, these differences did not achieve statistical significance."
Based on the numbers, Appenzeller and colleagues concluded PET/MRI does not "entirely match the diagnostic accuracy of standard low-dose PET/CT" with body-coil image acquisition.
"One task for the future of PET/MRI, therefore, might be the evaluation of how many MRI sequences and how much information from the MRI is needed (matrix, breath-holds, different weightings) to match up to a standard low-dose PET/CT, ideally within the same acquisition time," they added