PET CT Criteria for “Positive” Result in Skull Base Imaging Are Problematic: Case Series and Systematic Review of the Literature

Background and Objectives: Positron Emission Scan superimposed on Computerized Tomography (PET CT) are frequently utilized in initial diagnosis and follow-up of skull base lesions. However, the standardized uptake value (SUV) that best optimizes high sensitivity and specificity for characterizing skull base lesions is not well established. Our objective is to highlight diagnostic conundrums arising from current SUV cutoff criteria. We also perform a systematic review of the literature to propose SUV cutoff criteria and other associated strategies to optimize diagnosis.

Study Design: Case series and systematic review of the literature.

Methods: A retrospective review of electronic charts of patients who underwent surgical resection or biopsy for skull base lesions at our institution (between 2010–9/2013) was performed. Patients with a PET CT scan and tissue biopsy were included. A systematic review of the literature (2000 - August, 2013) was conducted using relevant search terms on Pubmed and Embase databases.

Results: Twenty-four PET CT scans from 15 patients with lesions primarily within the skull base, or with significant extension into it were identified. Of these, 11 were performed for primary imaging and 13 for post-treatment follow-up. Indications for PET CT imaging were: nasopharyngeal carcinoma (n = 2), sinonasal squamous cell carcinoma (n = 3), basal cell carcinoma of the nasal tip (n = 1), sinonasal undifferentiated carcinoma (n = 2), metastatic renal cell carcinoma (n = 2), sinonasal lymphoma (n = 1), sinonasal adenocarcinoma (n = 1), cutaneous and sinonasal melanoma (n = 3), and osteomyelitis (n = 1).

When using a SUV cutoff of 2.5 to denote positivity for malignancy, 18 of the 24 PET CT had “positive” results. Of these18, 12 were confirmed to be true positives on biopsy. Etiologies of the other positive results (false positives) at the skull base included osteomyelitis, radiation necrosis, and post-therapy inflammation. Although the negative predictive value (NPV) of PET CT was 100%, the positive predictive value (PPV) was 66%. It was noted to have 100% sensitivity and 50% specificity for identification of skull base malignancy. When the SUV cutoff was raised to 3, sensitivity decreased to 91.7% and specificity increased marginally to 58.3% with a PPV of 68.8%.

On systematic review of the literature, we did not find a well defined and reliable cut off SUV value for identifying malignancy in the skull base. Most authors advised using an SUV of 2.5 -3.

Conclusions: PET CT scans have high sensitivity, low specificity and sub-optimal PPV in defining malignant skull base lesions at current SUV cutoff criteria. The sensitivity of PET CT at 2.5 SUV is 100%, with a low specificity of 50%. When using an SUV cutoff of 3, specificity only marginally increased to 58.3% but sensitivity fell to 91.7%. Therefore, increasing SUV cut-off levels above 2.5 is undesirable as a declining ability to identifying 8% of malignant skull base lesions would be problematic. We propose continued use of the 2.5 SUV cutoff. However, due to the low PPV of the test (66%) at this cutoff, all “positive” lesions should undergo further evaluation by biopsy or additional diagnostic testing (e.g., MRI, SPECT CT).