SUVmax on Initial Post-treatment PET-CT Predicts Local Failure in Patients With Previously Irradiated, Recurrent Head-and-Neck Cancer Treated With SBRT

Abstract

Stereotactic body radiation therapy (SBRT) has seen increasing use as a salvage strategy for patients with recurrent, previously-irradiated squamous cell carcinoma of the head and neck (rSCCHN). It is unknown which PET characteristics correlate best with tumor control. We investigated whether PET SUV metrics on initial post-treatment PET-CT scan could predict subsequent local failure (LF) in patients with initial response or stable disease after SBRT for rSCCHN. A retrospective cohort study of 37 patients with rSCCHN treated with SBRT between 2005 and 2009. Patients were included if they had a pre-treatment PET-CT scan, at least two post-treatment PET-CT scans, and demonstrated either initial complete response (CR), partial response (PR), or stable disease (SD) with no LF on first follow-up PET-CT. Treatment planning scans with planning target volumes (PTVs) were deformably registered to pre and post-treatment PET-CTs using VelocityAI. SUVmax and SUVmean within the PTV was recorded for each PET-CT. Local failures were contoured on the second or later post-treatment PET-CT and defined as any failure whose nearest edge was within 1 cm of the PTV. SUVmax and SUVmean within PTV were compared to development of LF. Median time from completion of SBRT to first post-treatment PET-CT was 2 months (range 1.1-5.3 mo). Eighteen of 37 patients (49%) developed LF. Estimated median LF-free survival was 12.2 months. Higher SUVmax on initial post-treatment PET-CT was associated with significantly shorter time to LF both as a continuous variable (Cox regression p=0.01) and with a cut point of SUVmax > 4.0 (log rank p=0.011). 13/18 patients (72%) with SUVmax > 4.0 on initial post-treatment PET-CT developed subsequent LF vs. 5/19 patients (26%) with SUVmax < 4.0 (chi-square p=0.009). Higher pre-treatment SUVmean was associated with a borderline significant shorter time to LF (p=0.056). There was no significant association between time to LF and pre-treatment SUVmax (p=0.1) or initial post-treatment SUVmean (p=0.23). Analysis of PET SUV metrics may provide valuable early clues to long-term prognosis in patients with rSCCHN treated with SBRT. SUVmax within PTV on first post-treatment PET-CT was found to predict subsequent local failure in patients with initial CR/PR/SD. The threshold SUVmax appeared to be 4.0, above which patients were at a higher risk of LF. These findings may help clinicians predict whether patients who initially respond or have stable disease will eventually fail SBRT.