PEM development picks up pace

Although FDG-PET has demonstrated its capability to identify and characterize breast malignancies, whole-body PET may not be the most efficient technology to image breast cancer. The scanners are expensive and access to the modality is limited, when compared with other breast imaging tools such as CT, MR, x-ray mammography, and ultrasound.

In addition, a recent study in Radiology by researchers from Duke University in Durham, NC, found that whole-body PET scanners have a limited ability to depict small lesions, and breast abnormalities can be difficult to localize anatomically. During the past few years, however, positron emission mammography (PEM) units have been developed to overcome the restrictions of whole-body PET for staging breast cancer.

"Dedicated PET mammography units that can image positron-emitting tracers in the breast have several potential benefits over whole-body tomography, including high sensitivity for the emitted radiation, improved spatial resolution, substantially reduced attenuation, and reduced cost," wrote the authors of the pilot study (Radiology, February 2005, Vol. 234, pp. 527-534).

The Duke researchers collaborated with a team from the Thomas Jefferson National Accelerator Facility in Newport News, VA, to develop a dedicated PET mammography unit. Their PEM device consists of two 15 x 20-cm planar detectors and 3 x 3 x 10-mm lutetium gadolinium oxyorthosilicate scintillator elements.

The detectors can be positioned vertically or rotated for oblique views, and are typically separated by 6-9 cm, according to the researchers. The group used a non-3D back-projection method with 1.5-mm image pixels, allowing a measured transverse spatial resolution of 4.1 mm.

The PEM device, used with a conventional mammography unit, utilized a compression paddle for mild compression against the lower PEM detector. The researchers noted that substantially less compression was used with the PEM unit than with conventional mammography. A total of 23 patients, in whom diagnostic mammography or ultrasound had demonstrated lesions suggestive of malignancy, were imaged for the study, according to the authors.

Patients were asked to fast four hours prior to radiopharmaceutical administration, and received an FDG dose of between 2.0-2.5 mCi, depending on FDG decay and availability. An hour after FDG injection, the patients underwent a five-minute image acquisition of the affected breast in the craniocaudal projection. During the image-acquisition phase, breast imaging specialist Dr. Eric Rosen reviewed the raw images to optimize positioning.

The images were processed and reconstructed in the craniocaudal and coronal planes. Each lesion underwent image-guided core-needle biopsy using conventional imaging technology immediately after the PEM imaging. The reconstructed PEM images were reviewed by Dr. Rosen to determine if mammographic abnormalities were demonstrated by PEM.

"For this study, only visually conspicuous foci that had increased FDG accumulation compared with that of the background at the site of the known lesion were considered a positive result; nonfocal or patchy regions of FDG activity were disregarded," the authors wrote.

PEM demonstrated 20 focal abnormalities, of which 18 were malignant and two were benign. The benign lesions represented areas of fat necrosis, one at the site of a prior core-needle biopsy and the other near an excisional breast biopsy, the team reported.

They also noted that three of the 20 malignant lesions were demonstrated at conventional mammography, but not at PEM. Two of the false-negatives were masses, and the other was microcalcifications. Histologic examination proved the masses to be invasive ductal carcinoma, while the microcalcifications represented ductal carcinoma in situ with microinvasion. The researchers noted that all three of these lesions were located in the posterior third of the breast.

"This finding is explained by both the smaller range of acceptable angles of coincidence for a lesion near the edge (versus the center) of the detector and the physical limitations imposed by the detector plates, which exclude far posterior lesions from the field-of-view," the researchers stated.

The overall sensitivity of the device for malignancy was 86%, with a positive predictive value of 90%. The authors reported a calculated specificity of 33% and a negative predictive value of 25%. The design of the study did not allow the accurate determination of specificity, the authors noted.

The team suggested that additional research will be required to establish the sensitivity and specificity of PEM for demonstrating malignancy for nonpalpable lesions, not just category 5 lesions. They are, however, confident in the design of their PEM device.

"The main advantage of our PET mammography unit when compared with other units is the large field-of-view, which permits the entire breast to be imaged in a single acquisition," they wrote. "Our unit has a useful 15 x 20-cm field-of-view compared with the 6.5 x 5.5-cm and 5.6 x 5.6-cm field-of-view of previously reported PET mammography units."

In related news, the PEM field grew slightly more crowded with the publication of a proposed PEM camera by a team from the University of Toronto in Ontario. Researchers at the institution base their compact dual-head PEM camera on an amorphous selenium (a-Se) avalanche photodetector and the scintillator lutetium oxyorthosilicate (LSO) (Technology in Cancer Research and Treatment, February 2005, Vol. 4:1, pp. 61-68).

The group believes that the camera will deliver high collection efficiency by combining the fast scintillation light decay and high light yield of LSO with the quantum efficiency, large avalanche gain, and rapid response time of a-Se.

Established PEM vendor Naviscan PET Systems of Rockville, MD, also received press attention (TCRT, February 2005, Vol. 4:1, pp. 55-60) this month with the publication of data featuring the use of its technology.

An international team of researchers using Naviscan's full-breast PEM device reported a sensitivity of 93% and a specificity of 83% for characterizing lesions identified as suspicious on the basis of conventional imaging or physical examination. In addition, the group wrote that a sensitivity of 91% was preserved for intraductal cancers, which represent more than 30% of the reported breast cancers.