Low-dose T1W DCE-MRI for early time points perfusion measurement in patients with intracranial tumors: A pilot study applying the microsphere model to measure absolute cerebral blood flow.

Abstract

Previous studies have measured cerebral blood flow (CBF) with DSC-MRI using an "early time points" (ET) method based on microsphere theory. To develop and assess a new ET method for absolute CBF estimation using low-dose high-temporal (LDHT) T1W-DCE-MRI. Retrospective cohort study. Seven patients with sporadic vestibular schwannoma (VS) who underwent test-retest imaging; one patient with glioblastoma multiforme (GBM) imaged pretreatment; and 12 neurofibromatosis type 2 (NF2) patients undergoing bevacizumab treatment, imaged pre- and 90 days posttreatment. LDHT-DCE-MRI was performed at 1.5 and 3.0T, using 3D spoiled gradient echo with phase cycling. DSC-MRI performed in one patient, using 3D echo-shifted multi-shot echo-planar imaging (PRESTO) at 3T. Through Monte Carlo simulations, CBF estimation using three newly developed average contrast agent concentration (AC) -based methods (ACrPK, ACrMG, ACcomb), was compared against conventional maximum gradient (MG) approaches, at varying Rician noise levels. Reproducibility and applicability of the ACcomb method was assessed in our sporadic-VS/GBM/NF2 patient cohort, respectively. Reproducibility was measured using test-retest coefficient of variation (CoV). Pre- and posttreatment CBF values were compared using paired t-test with Bonferroni correction. Monte Carlo stimulations demonstrated that AC-based methods, particularly ACcomb, offered superior accuracy to conventional MG approaches. Overall test-retest CoV using the ACcomb method was 5.76 in normal-appearing white matter (NAWM). The new ACcomb method produced gray matter/white matter CBF estimates in the NF2 patient cohort of 55.9 ± 13.9/25.8 ± 3.5 on day 0; compared with 155.6 ± 17.2/128.4 ± 29.1 for the classical MG method. There was a moderate (10% using ACcomb and ACrPK) increase in CBF of NAWM 90 days post therapy (P = 0.03 and 0.005). Our new AC-based method of CBF estimation offers excellent reproducibility, and displays more accuracy in both Monte Carlo analysis and clinical data application, than conventional MG-based approaches. 1 Technical Efficacy: Stage 4 J. MAGN. RESON. IMAGING 2018;48:543-557.