True brain metastatic burden may be underestimated by current diagnostic MRI protocols in the clinic. In radiology, the most accurate method for brain tumor detection is contrast-enhanced T1-weighted (T1w) MRI, acquired before and after intravenous administration of Gadolinium (Gd) contrast agents.¹ In these images, tumors appear as regions of signal enhancement post-Gd because their growth causes damage to the local blood-brain barrier (BBB), which allows Gd to leak out of the vasculature and accumulate. Importantly, it is now understood that some metastases may grow in the brain without disrupting the BBB and instead develop around preexisting brain vasculature to access nutrients. In this case, Gd cannot cross the intact BBB and tumors may go undetected.² Preclinical studies performed in our lab illustrated this phenomenon by comparing tumor detection using FIESTA (fast imaging employing steady-state acquisition) with post-Gd T1w images in a breast cancer brain metastasis model (Figure 1). A significant number of Gd-impermeable tumors (not detected using post-Gd T1w MRI) were visible in FIESTA images and could be detected earlier in metastatic progression.³ Moreover, substantial heterogeneity existed in BBB permeability to Gd in different mouse models of brain metastasis and this affected tumor detection. In one model, post-Gd T1w imaging failed to detect more than 36% of brain metastases that were visible in FIESTA images.⁴ These preclinical findings warrant evaluation in patients. Here we translate our preclinical study to the clinic to evaluate the utility of FIESTA imaging for visualizing additional Gd-impermeable tumors in brain metastatic breast cancer patients. Breast cancer patients with a new diagnosis of brain metastasis by MRI or CT were enrolled in our study (n=10 to date; recruiting 30) at the London Regional Cancer Program under an institution approved ethics protocol. The 3D FIESTA protocol was optimized (resolution: 1x1x1 mm, TR/TE≈5.6/2.2, BW=64, α=45°, NEX=1, time≈6 min) and added to the standard radiation oncology MR simulation (SIM) imaging session, which included T1w pre/post-Gd and T2-weighted protocols (resolution: 1x1x2 mm, 1x1x3mm respectively). Images were acquired on a 1.5 T MR-SIM with a head surface coil and analyzed for tumor incidence, size, and contrast-to-noise ratio (CNR). The reporting radiologist and radiation oncologist were blind to the FIESTA result and followed standard of care for the patient. A reference radiologist (KYT) reported on FIESTA after the patient’s treatment was completed. Follow-up MRI, including FIESTA, will be acquired 3 months post-treatment.

In this cohort, 35 brain metastases were detected and had an average longest axial radius of 1.3 ± 0.9 cm. Tumor incidence ranged from 1-16 brain metastases per person, with two patients presenting with solitary tumors. The FIESTA sequence had the highest resolution with isotropic voxels of 1 mm. The CNR of tumor to contralateral brain was comparable to T2w images, though the FIESTA contrast mechanism is different and relies on T2/T1 relaxation times (Figure 2). The reference radiologist has reported on the FIESTA images for five patients. Of particular interest were several suspicious lesions observed in close proximity to the cerebrospinal fluid (CSF) on FIESTA images, which were not obvious in either T1w or T2w images (Figure 3). In the top row the border of a lesion is clear in FIESTA but obscure in T1w or T2w images. We hypothesize that this is an arachnoid cyst because of the fluid-filled appearance; however, metastatic growth cannot be ruled out at this time. In the bottom row a hyperintense lesion in FIESTA is isointense on T1 and T2w images. This may represent a brain metastasis. The FIESTA sequence may also offer additional information about tumor content; for example a tumor that appeared homogenous on clinical scans appeared heterogeneous with FIESTA (Figure 4). Using the FIESTA sequence alone, the reference radiologist was able to detect 5/7 tumors present on the T1w and T2w images. Both of the tumors that could not be detected using FIESTA were located on the ventricles, and one was also undetectable on T2w images (Figure 5).The FIESTA protocol was successfully implemented at 1.5 T for brain metastasis detection in breast cancer patients. To date, 10/30 patients have been recruited and preliminary results are promising. Initial data demonstrated that FIESTA offers the ability to visualize intra-tumor heterogeneity and additional lesions that are not discernable in standard clinical scans. Gd-enhanced T1w imaging is still necessary for detection of small brain tumors near ventricles as FIESTA fell short in this area. When all patients are enrolled, we will be able to confirm whether FIESTA may improve brain metastasis detection and how it may lead to better patient management and survival.