Application of 4D Flow MRI to Predict Clinical Outcomes after Stereotactic Radiosurgery (SRS) for Cerebral Arteriovenous Malformations (AVMs)

Abstract

<h3>Purpose/Objective(s)</h3> SRS is standard of care for unresectable cerebral AVMs. Nidus obliteration is the standard clinical metric of treatment success but can take years to occur. Hemorrhage risk is uncertain during this latency period. We hypothesized that early changes in blood flow through feeding arteries after SRS would predict future nidus obliteration and hemorrhage risk. We used 4D flow MRI to prospectively measure blood flow in feeding arteries prior to and 6 months after AVM SRS and correlated the early blood flow response to SRS to later clinical outcomes. <h3>Materials/Methods</h3> <i>Patients:</i> Patients undergoing SRS for AVM were enrolled prospectively. 4D flow MRI was performed prior to SRS and 6 months post-treatment. <i>Imaging:</i> Imaging was performed on a 3T scanner with dedicated head coil. 4D flow sequences were based on a 3D spoiled gradient echo sequence. Three directional velocity encoding gradients were used with Venc 100-200 cm/s. Additional MRI sequences were obtained to identify the nidus and feeding arteries. <i>Flow Analysis:</i> In each case the dominant feeding artery was identified. Mean blood flow over the cardiac cycle was measured in a proximal segment of the feeding artery. Blood flow was measured in the same location both before and 6 months after SRS. Blood flow was measured using software with FDA clearance for clinical 4D flow post-processing. Response was defined as fractional arterial blood flow after treatment vs. before treatment. <i>Statistical Analysis:</i> We defined the cut-off between ‘responders' and ‘non-responders' as the median blood flow response. Survival analysis was performed to assess the relationship between binarized blood flow response and nidus obliteration, with bleeding or death prior to obliteration treated as competing risks. Gray's test was used to assess for equality in the cumulative incidence functions for obliteration or bleeding/death between responders and non-responders. <h3>Results</h3> 18 subjects have completed pre- and post-treatment imaging. Median time to post-treatment imaging was 179 Days (IQR 169-202 Days). Median response was 0.52 (IQR 0.44-0.84). Median follow up is 755 Days (IQR 438-1188 Days). There have been no deaths. Amongst responders, there have been 4 obliteration events and 0 bleeding events. Amongst non-responders there have been 0 obliteration events and 2 bleeding events. There is a trend toward an increase in cumulative incidence of obliteration amongst responders (p=0.08) and in bleeding amongst non-responders (p=0.11). <h3>Conclusion</h3> Preliminary results suggest 4D flow MRI may be an early predictor of clinical outcomes after AVM SRS. Further follow up with additional events will be required to confirm a predictive relationship and to determine an optimal cutoff for distinguishing responders from non-responders.