Demonstrate the use of magnetic resonance imaging (MRI) and spectroscopy (MRS) for characterizing the relationship between cerebrovascular reserve and brain temperature measurements in patients with chronic steno-occlusive disease.Five patients undergoing treatment for chronic steno-occlusive disease underwent a uniform, two-day MRI scan protocol with acetazolamide (ACZ) challenge (1 g slow IV infusion over 3-5 minutes) outlined in Figure 1. All scanning was performed on a Siemens TIM Trio 3 T clinical scanner using a 32-channel head array coil. A T1 MPRAGE anatomical image was used to guide multi-voxel MRS (MRSI) grid placement and to plan the arterial spin labeling (ASL) and blood oxygen level dependent imaging (BOLD) scans (TR=1900 ms; T1=900 ms; TE=3.5 ms; flip angle=9°; 1 mm isotropic resolution). A multi-inversion time (TI) 3D-ASL sequence with background suppression (10 distinct TI, 300-3000 ms, at 300 ms increments) was acquired both pre- and post-ACZ. BOLD data included five minutes of baseline acquisition, followed by ACZ injection and data collection for 20 minutes post-ACZ. ASL and BOLD data were acquired as previously reported.¹ MRSI scans for thermometry were acquired pre- and post-ACZ injection with the semiLASER spectroscopy sequence using an 8x8 voxel region of interest (TR=1700; TE=35 ms; averages=3; 10x10x15 mm³ voxels). Thermometry data was processed using LCModel and custom software written in Matlab (Mathworks, 2015a). Absolute temperature was calculated using the difference between water and N­-acetylaspartate proton resonance frequencies and the relationship 0.01 ppm/°C. Temperature changes (ΔT) were calculated as a difference between post- and pre-ACZ absolute temperatures. Average cerebrovascular reserve values from both ASL and BOLD (defined as percent signal change during continuous pre- and post-ACZ monitoring) and ΔT were calculated voxel-wise to determine the relationship between temperature and cerebrovascular reserve. Correlations between ΔT with cerebrovascular reserve for ipsilateral and contralateral hemispheres were determined by fitting the data to polynomial regressions and significance was determined with ANOVA.Representative cerebrovascular reserve and ΔT maps are shown in Figure 2. Significant global voxel-wise correlations were observed between ΔT and ASL cerebrovascular reserve (r = 0.31, p < .001). For the ipsilateral voxels, significant correlations between ΔT and ASL cerebrovascular reserve (r = 0.52, p < .001) and BOLD cerebrovascular reserve (r = -0.31, p = .003) were observed. The contralateral voxels also showed a correlation between ΔT and ASL cerebrovascular reserve (r = 0.43, p < .001). Interestingly, the correlation between the ipsilateral ΔT and ASL cerebrovascular reserve followed a quadratic relationship whereas the contralateral relationship for the same parameters was linear (Figure 3) suggesting disparate mechanisms relating perfusion and temperature in normal tissues as compared with regions of hemodynamic impairment.These findings support that brain temperature is a measurable and dynamic neuroimaging biomarker potentially relating cerebrovascular flow and metabolism as predicted in physiologic models of metabolic heat production and dissipation. As the first reported use of MRI and MRS to correlate cerebrovascular reserve with brain temperature fluctuations, the results compel further study into the mechanistic nature of cerebral temperature regulation and the potential use of non-invasive MR thermometry in diagnosis, prognostication, and treatment selection in chronic steno-occlusive disease.We have shown the feasibility of absolute brain thermometry measurements in a small patient population, as well as a dynamic temperature response that is significantly correlated with both cerebrovascular reserve and pathology. These results will be used to inform future studies exploring the intersection of blood flow, cerebral metabolism, and brain temperature.