Introduction: Several studies have shown the value of PET in diagnosing occult tumours in patients with paraneoplastic neurological syndromes (PNS).

Objective: To audit our experience with PET in the diagnosis of occult tumours in PNS.

Methods: A retrospective analysis of all PET and PET/CT scans done for PNS in South Australia from the time the PET/CT was installed (2002) till September 2008. Results of antibody tests, imaging, final diagnosis and outcome were obtained with a mean follow up of 819 days.

Results: 24 patients (15 women), mean age 62 (range 36-80) were included. The mean interval between symptom onset and PET was 19 days (range 3-29). There were a variety of PNS including subacute sensory neuropathy, cerebellar syndrome, encephalitis, Lambert Eaton myasthenic syndrome, myopathy, transverese myelitis and subacute global neurological deterioration. Abnormal FDG uptake was seen in eight but malignancy was only confirmed in 2 patients. One patient died shortly after PET/CT likely because of lung malignancy. There were 5 false positives. At follow up 14 had no formal diagnosis, 4 had autoimmune illness and in 3 the diagnosis of PNS was revised. The sensitivity was 100%, specificity 76%, positive predictive value 37.5% and negative predictive value 100%.

Conclusion: PET was positive in only 12.5% of these patients. When the 3 patients without PNS are excluded the diagnostic yield of PET is 43%. PET is a useful tool in PNS but patient selection is important.