Multiparametric PET/MR (PET and MR-IVIM) for the evaluation of early treatment response and prediction of tumor recurrence in patients with locally advanced cervical cancer

Abstract

To assess the value of 18F-FDG PET and MR-IVIM parameters before and during concurrent chemoradiotherapy (CCRT) for evaluating early treatment response and predicting tumor recurrence in patients with locally advanced cervical cancer (LACC) using a hybrid PET/MR scanner. Fifty-one patients with LACC underwent pelvic PET/MR scans with an IVIM sequence at two time-points (pretreatment [pre] and midtreatment [mid]). Pre- and mid-PET parameters (SUVmax, MTV, TLG) and IVIM parameters (D, F, D*) and their percentage changes (Δ%SUVmax, Δ%MTV, Δ%TLG, Δ%D, Δ%F, Δ%D*) were calculated. We selected independent imaging parameters and built a combined prediction model incorporating imaging parameters and clinicopathological risk factors. The performance of the combinative evaluation for tumor early shrinkage rates (TESR) and the prediction model for tumor recurrence was assessed. Thirty-two patients were classified into the good response (GR) group with TESR ≥ 50%, and 19 patients were categorized into the poor response (PR) group with TESR < 50%. Δ%D (p = 0.013) and Δ%F (p = 0.006) are independently related to TESR with superior combined diagnostic ability (AUC = 0.901). Pre-TLG, Δ%D, and suspicious lymph node metastasis (SLNM) were selected for the construction of the combined prediction model. The model for identifying the patients with high risk of tumor recurrence reached a moderate predictive ability and good stability with c-index of 0.764 (95% CI, 0.672-0.855). The combined prediction model based on pretreatment PET metabolic parameter (pre-TLG), IVIM-D percentage changes, and LNs status provides great potential to identify the LACC patients with high risk of recurrence at early stage of CCRT. • PET/MR plus IVIM offers various complementary information for LACC. • IVIM-D and IVIM-F percentage changes are independently related to tumor early shrinkage rates. • The combined prediction model can help identify the LACC patients with high risk of tumor recurrence.