Idiopathic pulmonary fibrosis (IPF) is a fatal condition in which healthy lung is replaced by fibrotic tissue. The disease is characterized by disruption of the alveolar-capillary barrier and subsequent leakage of plasma into airspaces. We aim to develop a method for quantification of capillary permeability to albumin given strong evidence that vascular leak promotes fibrosis and correlates inversely with outcome^1,2. We focus on pre-clinical applications as IPF pathogenesis remains poorly understood and animal models are key for studying the disease and for assessment of potential therapies. The goal of this study is to assess whether pulmonary leak could be quantified using dynamic MRI with an intravascular tracer.C57Bl/6 mice were administered one intratracheal dose of bleomycin at 1.0 U/kg to induce fibrosis. One cohort was imaged on days 3 (n=3) and 7 (n=2) after injury, while a second group was studied on days 5 (n=2) and 13 (n=3). Four healthy animals were included as controls. Imaging was performed on a 4.7T Bruker system. A 3D respiratory-gated UTE sequence (FOV 40.8 mm3, matrix 136x136x136, voxel 0.3x0.3x0.3 mm, TE 20 μsec, TR 8 ms, 57836 radial spokes, 1 average) was used to obtain images of the lung pre and post injection of Gd-DTPA-BSA (75 μmol Gd/kg). Two UTE acquisitions with FA 10° and 60° were performed at baseline. Six images with FA 60° were acquired after injection at 30-minute intervals. Regions of interest were drawn in the lung and adjacent muscle on baseline UTE images with FA = 10°. Normalized lung signal intensity (nSI = SI lung/SI muscle) was calculated for all cohorts and used as an indicator of fibrotic burden. Dynamic analysis was based on images with FA = 60°. Signal intensities of the lung, SI_Lung, and the inferior vena cava (IVC), SI_Blood, were obtained for each imaging time point. Pre-contrast signal intensities were subtracted from post-contrast values to obtain the temporal signal response of the lung, ΔSI_Lung (t), and the IVC, ΔSI_Blood (t). Our experimental parameters were selected such that ‘delta’ signal intensity in a given tissue can be assumed to be proportional to the concentration of gadolinium in that tissue. Data was fit using a two-compartment model^3,4 to calculate pulmonary blood volume (BV) and fractional leak rate (FLR) of Gd-DTPA-BSA out and into the lungs (Fig. 1). Finally, permeability surface area product (PS) was derived as PS = BV*FLR_out*(1-Hct)⁴. Lung tissue changes were not evident at day 3 post insult corroborated by no significant change in baseline nSI of this cohort compared to controls. Hyperintense regions indicative of tissue injury were observed at all later time points with nSI increasing continuously from day 5 to day 13, suggestive of progression of pathology (Fig. 2). Capillary permeability in bleomycin-treated animals rose sharply on day 3 after insult, returned to baseline by day 5 and increased moderately from day 5 to day 13 (Fig. 3).

Fibrosis: Lung SI on unenhanced UTE has been shown to increase with disease progression and to correlate with histological grading of fibrosis⁵. Given the longitudinal nature of this preliminary study histological scoring of fibrosis was not possible at all time points. However, our baseline nSI measurements are indicative of the anticipated disease progression (increased matrix production and edema) with early signs of pathology evident at day 5 and a maximum disease burden observed at day 13.

Vascular leak: We hypothesize that there are two stages to the time course of albumin leak in our disease model. Initially, acute inflammation causes a temporary increase in capillary endothelial permeability exemplified by a rapid rate of albumin leakage at day 3 and a return to baseline levels by day 5. This observation is in agreement with measurements of pro-inflammatory cytokine levels (e.g. interlukin-1, interferon-γ) that have been found to increase rapidly by day 3 and to drop to baseline at later time points⁶. During the second stage (days 5-13), albumin leakage rises in parallel with the increase in fibrotic burden. We assume that this is caused by an increase in capillary hydrostatic pressure resultant from rising interstitial space density with progression of fibrosis. To our knowledge the time course of vascular leak in pulmonary fibrosis has not been previously reported. Additional imaging experiments and histological studies are underway to reproduce and elucidate the biophysical origin of our preliminary observations.

We demonstrated that it is feasible to non-invasively quantify pulmonary leak with MRI. This method could be useful for studying the role of vascular leak in fibrosis development as well as for monitoring of disease progression and response to treatment.