Assessing the Pulmonary Vascular Responsiveness to Oxygen with Proton MRI.

Ventilation-perfusion matching occurs passively and is also actively regulated through hypoxic pulmonary vasoconstriction (HPV). The extent of HPV activity in humans, particularly normal subjects is uncertain. Current evaluation of HPV assesses changes in ventilation-perfusion relationships/pulmonary vascular resistance with hypoxia and are invasive, or unsuitable for patients because of safety concerns. We used a non-invasive imaging-based approach to quantify the pulmonary vascular response to oxygen as a metric of HPV by measuring perfusion changes between breathing 21% and 30%O2 using arterial spin labeling (ASL) MRI. We hypothesized that the differences between 21% and 30%O2 images reflecting HPV release would be 1) significantly greater than differences without FI O2 changes (e.g.,21-21% and 30-30%O2 ) and 2) negatively associated with ventilation-perfusion mismatch. Perfusion was quantified in the right lung in normoxia (baseline), after 15min of 30%O2 breathing (hyperoxia) and 15min normoxic recovery (recovery) in healthy subjects (7M,7F;Age=41.4±19.6y). Normalized, smoothed, and registered pairs of perfusion images were subtracted and the mean square difference (MSD) calculated. Separately, regional alveolar ventilation and perfusion were quantified from specific ventilation, proton density, and ASL imaging; the spatial variance of ventilation-perfusion (s2 V̇A /Q̇) distributions were calculated. The O2 -responsive MSD was reproducible (R2 =0.94,P<0.0001) and greater (0.16±0.06,P<0.0001) than from subtracted images collected under the same FI O2 (baseline=0.09±0.04,hyperoxia=0.08±0.04,recovery=0.08±0.03), which were not different from one another (P=0.2). The O2 -responsive MSD was correlated with s2 V̇A /Q̇ These data suggest that active HPV optimizes ventilation-perfusion matching in normal subjects. This noninvasive approach could be applied to patients with different disease phenotypes to assess HPV and ventilation-perfusion mismatch.