Impact of pre- and early per-treatment FDG-PET based dose-escalation on local tumour control in fractionated irradiated FaDu xenograft tumours

OBJECTIVE:To investigate local tumour control after dose-escalation based on [18F]2-fluoro-2-deoxy-d-glucose (FDG) positron emission tomography (PET) obtained before and early during fractionated irradiation. MATERIALS AND METHODS:85 mice bearing FaDu xenografts underwent FDG-PET twice: first immediately prior to the first 2-Gy fraction of irradiation (PET1_0) and second after 18°Gy (PET2_18). After these 9 fractions, animals were randomly allocated to: (1) continuation of 2-Gy fractions (cumulative dose of 60°Gy; n=31), (2) dose-escalation with 3-Gy fractions (cumulative EQD2-dose 86.25°Gy [α/β-value: 10]; n=25), or (3) with 4-Gy fractions (cumulative EQD2-dose 116°Gy; n=29). The effects of SUVmax0°Gy, SUVmax18°Gy, and dose on local tumour control were analysed in two ways. First, the Cox proportional hazards model was used with two covariates: continuous SUVmax values and dose. Second, the Kaplan-Meier method was used, with tumours classified according to SUVmax greater than or less than (1) median maximum standardized uptake value (SUVmax) at PET1_0 and PET2_18, or (2) the cut-off value 2.5. RESULTS:The multivariate Cox analysis revealed a significant negative association between higher SUVmax determined before start of treatment and local control (HR=1.59, [95% CI 1.04, 2.42], p=0.031), whereas higher dose had a significant positive effect (HR=0.95, [0.93, 0.98], p<0.001). In contrast, FDG uptake at 18Gy did not correlate with local control (HR=1.14, [0.53, 2.45], p=0.73). Neither FDG uptake prior to irradiation nor at 18Gy correlated with local control irrespective of the delivered dose (log-rank test) when using the median SUVmax values for stratification (SUVmax0Gy: 60Gy: p=0.25, 86.25Gy: p=0.47, 116Gy: p=0.88 and SUVmax18Gy: 60Gy: p=0.42, 86.25Gy: p=0.34, 116Gy: p=0.99). By contrast, stratifying the animals by the cut-off 2.5 at PET1_0 reveals a significant difference in local control for the 60Gy group (p=0.034), but not for the other dose groups. At PET2_18, no significant effect for any dose group was detected. CONCLUSIONS:The multivariate Cox analysis revealed a significantly higher hazard of recurrence for mice with higher SUVmax determined before start of treatment. These results support the hypothesis that patients with high pre-therapeutic FDG uptake should be considered at increased risk of local failure and therefore as possible candidates for dose escalation strategies.