[ACNP 31st Annual Meeting] Areas of controversy of the use of FDG-PET in lung cancer include the use of SUV's in determining whether or not a solitary pulmonary nodule is benign or malignant; what is the role of image fusion; software vs. hardware fusion; pulmonary nodules < 10 mm. At Cedars-Sinai, they perform about 5 thoracic surgeries a day for lung cancer. Dr. Waxman in this talk discussed his experience with lung cancer and PET.


Lots of be benign lesions have high SUV's, and lots of cancers have low SUV's. There is a lot of overlap, and therefore a specific cutoff value should not be used. There is a trend to higher SUV's (and a statistically significant difference) but in individual patients a cutoff value isn't that useful.

If you use any visual focus as defining a lung cancer, the sens=99% but spec = 32%.

Using a 2.5 SUV(max) cutoff, sens = 76% and spec = 43%


FOR N ANALYSIS
=============
Looking at SUV ratios of tumor to background, a cutoff of 1.4 had an accuracy of about 86%


IMAGE FUSION
============
Software fusion is rapidly advancing; the Mirada software appears to work well. If you have control over the CT and the PET scanner, and you can make sure that the patient is positioned the same on both scans, the image fusion SOFTWARE works great. "Fusion, by any means" works great according to Dr. Waxman.

CONCLUSION
==========

SUV's should be viewed as a probability assessment - do not use cutoffs

Both software and hardware fusion can have registration errors.

Fusion (by any means) helps especially for the hilum and mediastinum.

PET/CT helps the radiologists improve their reading of the CT, as well as help the nuclear physician improve their reading of the PET scan.