18F-FDG PET image-derived tumor characterization to improve prediction of response to neoadjuvant chemotherapy on locally advanced breast cancer - IMT Atlantique Accéder directement au contenu
Communication Dans Un Congrès SNMMI 2013 : Society of nuclear medicine and molecular imaging annual meeting Année : 2013

18F-FDG PET image-derived tumor characterization to improve prediction of response to neoadjuvant chemotherapy on locally advanced breast cancer

Résumé

Objectives: Early prediction of no response to neoadjuvant chemotherapy (NAC) in breast cancer may help in identifying patients that would benefit from alternative therapeutic strategies. The objective of this study was to assess the potential predictive value of tumor characterization from 18F-FDG PET images including metabolically active tumor volume (MATV) and uptake heterogeneity characterization using texture analysis. Methods: 89 patients were included who underwent scans at baseline (PET1) and before the 3rd cycle of NAC (PET2). Pathological response was assessed after surgery using the Sataloff scale. Several image-derived parameters were extracted from delineated MATVs, including SUV max, peak and mean; total lesion glycolysis (TLG); local/regional textural features. The predictive value of these parameters at PET1 or PET2, and their evolution between PET1 and PET2 ({Delta}param, %) was assessed through receiver operating characteristic (ROC) area under the curve (AUC) analysis. Results: There were 45 responders and 44 nonresponders. There was no response differentiation using PET1 absolute values only for any of the parameters considered (AUCs <0.6). On the contrary, an AUC of 0.82, 0.81, 0.74, 0.73 and 0.67 was obtained for {Delta}TLG, {Delta}MATV, {Delta}SUVmax, {Delta}SUVpeak and {Delta}SUVmean respectively. Several uptake heterogeneity features, such as entropy and variance, could also predict pathological response with an AUC of 0.78 and 0.69 respectively. Conclusions: Significantly better prediction of response to NAC was achieved with {Delta}TLG (specificity 89%, sensitivity 73%) over SUVmax (specificity 58%, sensitivity 84%) (p=0.02). Although uptake heterogeneity characterization by itself did not perform better than standard image-derived indices, they may still provide complementary information that could help in improving prediction, which will be assessed in future studies.
Fichier non déposé

Dates et versions

hal-00853363 , version 1 (22-08-2013)

Identifiants

  • HAL Id : hal-00853363 , version 1

Citer

Mathieu Hatt, David Groheux, Antoine Martineau, Marc Espie, Elif Hindie, et al.. 18F-FDG PET image-derived tumor characterization to improve prediction of response to neoadjuvant chemotherapy on locally advanced breast cancer. SNMMI 2013 : Society of nuclear medicine and molecular imaging annual meeting, Jun 2013, Vancouver, Canada. ⟨hal-00853363⟩
32 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More