Abstract
Abstract 320
Positron emision tomography with [18F]fluorodeoxyglucose (PET) performed after one to four cycles of chemotherapy could predict treatment outcome in diffuse large B cell lymphoma (DLBCL). However, suitable criteria to interpret interim PET remain to be established. Visual analysis was related to reproducibility concerns and false positive results in previous reports, and standardized uptake value (SUV) was proposed to improve interim PET accuracy. To compare the clinical usefulness of both methods, we analysed sequential interim PET according to visual and SUV criteria in a prospective multicenter phase II trial.
From 2007 to April 2009, 113 patients with aaIPI 2–3 factors DLBCL were randomly assigned to receive an induction treatment with four cycles of either R-ACVBP or R-CHOP14 in the LNH2007-3B GELA trial. PET was done at baseline (PET0) and after 2 (PET2) and 4 cycles (PET4) of immuno-chemotherapy. All PET scans were centrally reviewed and interpreted using visual criteria. SUVmax reduction values between PET0 and PET2 (DSUVmaxPET0-2) or PET4 (DSUVmaxPET0-4) were available for 85 out of 113 patients. After using the receiver operating characteristics approach, patients with a DSUVmaxPET0-2 >66% were considered as good responders after 2 cycles and those with a DSUVmaxPET0-4 >70% were considered as good responder at end of induction treatment. Progression-free survival (PFS) was analyzed according to PET results based on visual criteria and DSUVmax.
Using visual analysis, respectively 34% and 49% of patients achieved a negative PET2 and PET4. An agreement between on-site and review panel was observed in 89% and 92% of cases for PET2 and PET4 leading to a kappa coefficient of 0.769 and 0.836 respectively. Seventy (82%) patients had a DSUVmaxPET0-2 > 66% and 74 (88%) had a DSUVmaxPET0-4 > 70%. Using DSUVmax analysis, 45 (78%) of the 58 PET2 positive patients and 33 (80%) of the 41 PET4 positive patients could be reclassified as good responder after 2 and 4 cycles of immuno-chemotherapy, respectively. PET2 and PET4 results based on visual criteria had no impact on PFS (p=0.99 and p=.077, respectively). By constrast, PET results based on DSUVmax better predicted PFS: patients with a DSUVmaxPET0-2 >66% had a better 1-year PFS than patients with a lower DSUVmaxPET0-2 (89% v 56%; p=0.0031). Similarly, patients with a DSUVmaxPET0-4 >70% reached a 90% 1-year PFS, while patients with a lower DSUVmaxPET0-4 only had 23% 1-year PFS (p<0.0001).
Real-time central PET review is feasible in a multicenter setting. Visual analysis had a good reproducibility but was associated to an excess of positive results as well for PET2 as for PET4. SUVmax reduction between baseline and interim PET was more accurate than visual analysis to predict early outcome of patients treated for high risk DLBCL. It also appears to be the best method so far to assess PET response either at mid or end of induction treatment.
No relevant conflicts of interest to declare.
This icon denotes an abstract that is clinically relevant.
