DOI: 10.1007/s00259-017-3776-3Pages: 2034-2041

Predictive value of 18F-FDG PET/CT in adults with T-cell lymphoblastic lymphoma: post hoc analysis of results from the GRAALL-LYSA LLO3 trial

1. Centre Henri-Becquerel, Department of Nuclear Medicine

2. University of Rouen, QuantIF–LITIS (EA [Equipe d’Accueil] 4108), Faculty of Medicine

3. Rouen University Hospital, Department of Biostatistics

4. Hôpital Tenon, AP-HP, Department of Nuclear Medicine

5. Université Paris Diderot, Department of Hematology, Hôpital Saint-Louis, AP-HP

6. Centre Henri Becquerel and Normandie Univ UNIROUEN, Inserm U1245 and Department of Hematology

Correspondence to:
Stéphanie Becker
Tel: +33




We examined whether FDG PET can be used to predict outcome in patients with lymphoblastic lymphoma (LL).


This was a retrospective post hoc analysis of data from the GRAAL-LYSA LL03 trial, in which the treatment of LL using an adapted paediatric-like acute lymphoblastic leukaemia protocol was evaluated. PET data acquired at baseline and after induction were analysed. Maximum standardized uptake values (SUVmax), total metabolic tumour volume and total lesion glycolysis were measured at baseline. The relative changes in SUVmax from baseline (ΔSUVmax) and the Deauville score were determined after induction.


The population analysed comprised 36 patients with T-type LL. SUVmax using a cut-off value of ≤8.76 vs. >8.76 was predictive of 3-year event-free survival (31.6% vs. 80.4%; p = 0.013) and overall survival (35.0% vs. 83.7%; p = 0.028). ΔSUVmax using a cut-off value of ≤80% vs. >80% tended also to be predictive of 3-year event-free survival (40.0% vs. 76.0%; p = 0.054) and overall survival (49.2% vs. 85.6%; p = 0.085). Total metabolic tumour volume, baseline total lesion glycolysis and response according to the Deauville score were not predictive of outcome.


A low initial SUVmax was predictive of worse outcomes in our series of patients with T-type LL. Although relatively few patients were included, the study also suggested that ΔSUVmax may be useful for predicting therapeutic efficacy.

To access the full text, please Sign in

If you have institutional access, please click here

  • Accepted: Jul 6, 2017
  • Online: Jul 21, 2017

Article Tools