Print

Massa, Christopher Barry

Medical Laboratory Science Program

2008

M.S.

Recruitment and derecruitment (R/D) of airways is known to significantly influence mechanical properties of the respiratory system during artificial ventilation, particularly in states of lung injury. The prevailing view of this phenomenon treats airway R/D as a static function of pressure. Recent experimental and clinical data suggests that this is not the case, but rather that RID is an inherently dynamic process. In order to quantitatively assess the dynamics of lung recruitment during mechanical ventilation we extended a mathematical model by Bates and Irvin (9) for the purpose of fitting experimental data. The model of the lung consists of a parallel network of flow pathways with identical resistive and elastic elements. Each pathway is allowed to be either open, whereby it accumulates flow and decreases overall lung stiffness, or closed, increasing lung elastance and not participating in ventilation. The pathways are characterized by unique critical closing and opening pressures, and opening and closing velocities, each chosen from probability distribution functions. The rate of transition between an open and closed state depends on the magnitude difference between the pressure in the respiratory system and each unit's critical pressure times the airway's opening or closing velocity constant. Since the exact form of the pressure dependence governing recruitment and derecruitment remains unknown we explored four model variants to predict how opening or closing behavior is altered in injury. The lung model was coupled with a computational model of a mechanical ventilator in order to simulate elastance changes following deep inflation (DI) at three levels of Positive End Expiratory Pressure (PEEP). Elastance measurements came from healthy or lung injured mice at 4, 14, 24 or 48 hours following intratracheal instillation of saline (control) or hydrochloric acid (injury). The Nelder and Mead simplex optimization method was used to minimize error between model variants and average experimental elastance for each condition. By comparing the residual error of the fits for each model, we have demonstrated that'only one variant was able to recreate both the transient response to deep inflations and the response to static PEEP. In fitting the best model to data from individual mice we obtained estimates for parameters governing opening and closing behavior. Statistics and model sensitivity were determined for each parameter in every experimental condition. Comparison of parameter values between groups revealed a significant increase in closing and opening pressures from health to injury, which worsened with increasing injury severity. The progressive increase in critical pressures as injury worsens implicates surfactant deactivation as the likely cause of increased propensity for airway closing during acute lung injury.