Development of Feed Forward and Conditional Models Based on Rodent Imaging Data to Predict Effects of Nitrogen Mustard on Pulmonary Function

Abstract

Nitrogen mustard (NM; (bis(2-chloroethyl)methylamine) is a cytotoxic vesicant known to cause pulmonary damage. In these studies, we developed mathematical models that allow prediction of NM-induced functional deficits in the lung based on structural alterations identified by magnetic resonance imaging (MRI) and computed tomography (CT). Male Wistar rats were exposed to PBS (CTL) or NM (0.125 mg/kg) via i.t. instillation. CT and MRI scans were performed prior to exposure and 28 d post exposure. Pulmonary function was assessed using a SCIREQ flexiVent at a positive end expiratory pressure (PEEP) of 3 cm H2O. Because of accuracy and minimal artifacts, MRI data was used to calculate total lung volume (W). As CT is capable of differentiating between regions of air and water within the lung, it was used as a direct measure of fluid density within the tissue. Relative lung volumes calculated from CT data were used to summarize hyperinflation (I), normal lung, and consolidated lung tissue (B) using voxel density. Using these elements, a feed forward model was constructed to predict respiratory impedance (Z), a measure of airflow, with the equation: z = (A(η-j))/ωα where A = B/(I+W). Calculated values of A from imaging data confirmed alterations in lung structure following NM exposure (1.38 ± 0.08, n=4, CTL vs. 0.99 ± 0.16, n=3, NM) as a result of hyperinflation (I). The predicted Z spectra derived from imaging data was compared to the measured Z spectra obtained from the flexiVent data. A strong correlation was observed between predicted and measured Z spectra at 28 d post NM (R2=0.997 ± 0.01, n=7). Predicted and measured Z spectra were consistent with previous findings that NM caused a loss of ventilatory function at high frequencies. To further evaluate lung function, components of the Z spectra were used to conditionally model resistance, RL = (a+bf)/(c+f) and elastance, EL = E0 + ΔE(1 – е-βf). This model allowed us to differentiate functional contributions of tissue and airway components at low and high frequencies in imaging and flexiVent data sets. Together, our results demonstrate that structural imaging data can be used to assess lung function as a toxicological endpoint following exposure to mustard vesicants. The use of this model will aid in the rapid evaluation of potential countermeasures to mitigate lung toxicity.