Abstract
354
Objectives We have previously observed that tumor regression in LARC and relapse free and overall survival can be predicted by early changes in FDG uptake during preoperative radiochemotherapy. We are now evaluating whether the same holds true in a larger series of patients treated with an experimental bevacizumab (BEV) containing radiochemotherapy schedule.
Methods Sixty two patients with MRI defined high risk LARC received 3 biweekly courses of oxaliplatin (100 mg/m2)/ raltitrexed (2.5 mg/m2) on day 1, and fluorouracil (800 mg/m2)/ folinic acid (250 mg/m2) on day 2 during pelvic RT (45 Gy). For the first 16 patients, BEV (5 mg/kg) was administered on day 1 for the first 2 courses (schedule A), whereas it was administered on day -4 for the first two courses in the remaining 44 patients (schedule B). FDG PET-CT was performed at baseline, on day 12 and prior to surgery. VOI analysis was utilized to measure mean and maximum SUV, volume and total lesion glycolisis (TLG, mean SUV x lesion volume) of the lesion in each study. Pathologic response was defined by using a 5-degree tumor regression grade (TRG) scale according to Mandard’s modified classification. ROC analysis was utilized to determine whether PET parameters or their % change over time correlated to response and ability to predict outcome.
Results Complete data was available in 60 patients. Three of 16 patients who were treated under schedule A (19 %) and 22/44 patients under schedule B (50%) were complete pathologic responders (TRG1, overall 25/60 patients, 42%). Baseline and presurgical PET parameters showed no correlation to complete pathologic response. However, PET data obtained on day 12 showed that the early TLG % change relative to baseline was significantly higher in TRG1 patients (median 74% range -31 to 86%) compared to incomplete responders (TRG2-4, median 42%, range -128 to 90%, p < 0.0005). ROC analysis showed an optimal cutoff value of 59.5% for identifying TRG1 responders (AUC 0.76, sensitivity 71%, specificity 80%, p<0.005). Median followup was 59 months (range 11-101). The treatment was very effective with prolonged relapse free (RFS, median 83 months, 95% CI 74-92 months) and overall (OS, median 88 months, 95% CI 82-95 months) survival for the entire population. However, irrespective of pathologic response, patients showing early % TLG change above the 59.5% cutoff value showed longer RFS (median 92 months, 95% CI 82-102 vs 76 months, 95% CI 62-89) and OS (median 97 months, 95% CI 91-102 vs 82 months, 95% CI 71-93) compared to the remaining patients, with both values showing a definite trend toward statistical significance (p=0.067 and 0.056, respectively).
Conclusions Under this schedule FDG-PET is a very sensitive early surrogate marker of treatment efficacy as it can predict complete pathologic response. This early assessment shows a trend for statistical correlation with relapse free and overall survival. FDG-PET may aid in tailoring therapeutic regimens through early identification of nonresponsive patients who may benefit from more aggressive treatment prior to surgical resection.