Central Review of PET/TC: First Spanish Experience in a Phase 2 Randomized Trial in Diffuse Large B-Cell Lymphoma (DLBCL) Patients

Background:

Positron emission tomography / computed tomography with fluorodeoxyglucose (FDG-PET/CT) is widely accepted for staging and for end treatment assessment in patients with DLBCL, and has demonstrated to have prognostic impact when used to evaluate early response to chemotherapy (CT) in this group of patients. In the last years, many changes in the way of interpreting PET/CT have been proposed, both for interim and for end of treatment (Barrington SG, J Clin Oncol 2014;32:3048-3058)

We present preliminary data of the first Spanish centralized PET/CT review in a Phase 2 randomized trial in young patients with poor prognosis diffuse large B-cell lymphoma (DLBCL), which compares 6 cycles of RCHOP versus 6 cycles of a modified RCHOP regimen, Bortezomid-R CAP (BRCAP). ClinicalTrials.gov Identifier: NCT01848132.

The main objective is to evaluate concordance between reviewers in interim and final PET, in order to stablish the best criteria for PET/CT assessment. PET2 ability to predict PET4 result is also analyzed.

Methods:

A blinded, prospective, centralized review in real time of PET/CT images was realized by the GELTAMO PET network. For each patient, images of basal (PET0), interim PET2 and PET4 and final PET after completion of chemotherapy (PET 6) were sent to a central platform, and then analysed by the review panel composed of seven expert nuclear medicine physicians. PET2 and PET4 were interpreted visually based on Deauville criteria (considering scores 4 and 5 as positive), and also semiquantitavely (considering a positive PET2 when ΔSUVmax≤66% and a positive PET4 when ΔSUVmax≤70%). PET6 was interpreted following Deauville criteria. Final result of every PET/CT was defined as positive or negative by the central review panel. When discordance between visual and semiquantitative analysis was found in interim PET, semiquantitative method was determinant of final result. A positive PET4 result determined dropped out from trial.

Concordance between all readers was analysed using Cohen's kappa coefficient.

Results:

Of the first 76 patients that underwent PET/CT, 64 patients completed PET0, PET2 and PET4; 34 patients underwent PET0, PET2, PET4 and PET6.

In the central review, 44/64 patients (69%) were PET2(-) and 20/64 (31%) were PET2(+); 44/64 (69%) were PET4(-) and 20/64 (31%) PET4(+); 21/34 (62%) were PET6(-) and 13/34 (38%) were PET6(+).

We found 43 patients with negative PET2 and PET4, 19 patients with positive PET2 and PET4, 1 patient with positive PET2 and negative PET4, and 1 patient with negative PET2 and positive PET4. PET2 result was predictive of PET4 (p<0.001).

Concordance between reviewers for PET2 using visual assessment was good (median kappa=0,74) and very good using semiquantitative analysis (median kappa: 0,83). Concordance between reviewers for PET4 using visual assessment was good (kappa=0,72) and very good when semiquantitative analysis was used (median kappa=0,87). For PET6, concordance between readers was moderate (kappa=0.45).

Conclusion:

In this homogeneus group of patients, a centralized semiquantitative analysis of interim PET after 2 and 4 cycles of chemotherapy improves concordance between readers in comparison with visual analysis. When semiquantitative method is used, PET2 is predictive of PET4 result.

In the same way, visual analysis of PET6 using Deauville criteria seems to have a worse concordance between readers, but more patients need to be analyzed to confirm this result.