PET-CT in Orthopaedic Oncology

The topic notes that the brain is intensely FDG-avid, that brown fat causes false positives, and that the protocol requires fasting and "avoiding talking/chewing" — all of which are consequences of where FDG normally goes. Reading a PET means judging pathology against this normal background, so it is worth setting out.

View larger

SUV is quoted throughout this topic (a "2.5 cut-off," SUVmax, SUVmean, and PERCIST's "SUL peak"), and a whole controversy turns on there being "no universal cut-off" — so it is worth defining what the number actually is.

Definition. The standardised uptake value is a semi-quantitative, dimensionless ratio that normalises the FDG activity concentration measured in a region to the injected dose and the patient's size: SUV = tissue activity concentration (e.g. MBq/mL) divided by [injected dose (MBq) / body weight (g)]. In effect it asks "how much brighter is this lesion than it would be if the tracer were spread evenly through the body?" It is a stand-in for metabolic rate, not a direct measurement of malignancy.

The common variants:

Why a single threshold cannot be transferred. Absolute SUV is not comparable between centres or time points because it depends on blood glucose (high glucose competes with FDG and lowers tumour SUV), uptake time (SUV keeps rising for an hour or more, so the scan delay must be fixed), body habitus / lean mass (fat dilutes weight-based SUV — hence SUL), scanner model and reconstruction algorithm, and the partial-volume effect (lesions below roughly twice the scanner resolution falsely read low). This is exactly why a number like 2.5 cannot reliably separate benign from malignant across all settings, and why PERCIST mandates tightly controlled, identical conditions and uses the normal liver (or blood pool) as an internal reference for response. (The general physics of PET acquisition is developed in the nuclear-medicine-orthopaedics topic.)