Analysis of lung multiple indicator data with an effective diffusivity model of capillary exchange.

In this study, we have compared an effective diffusivity model with several models previously used to describe lung microvascular exchange of small molecules (multiple indicator curves): the Crone integral extraction model, the Sangren-Sheppard model, and the Rowlett-Harris model. The parameters of each model were adjusted to achieve a best fit of multiple-tracer data from six isolated dog lung preparations perfused under zone III conditions at three flows and from five awake sheep before and after histamine infusion. The effective diffusivity model was the best at matching the data and the unique falling extraction pattern observed in the lung. When the isolated lung data was analyzed, the permeability-surface area products (PS) of three models were all significantly lower at lower flows. However, the effective diffusivity model parameter (D 1/2 S, an effective diffusivity-surface area product) was not significantly affected by flow reduction. Lung lymph flow and lymph-to-plasma protein ratios from the awake sheep indicated that PS and D 1/2 S should have increased, and except for the Sangren-Sheppard model, all models predicted significantly increased values. By use of sensitivity analysis, the parameters of the effective diffusivity model were found to be more independently determined than those of the Sangren-Sheppard model, but data collected at longer times were necessary to reliably identify the extravascular volume parameter (LS, diffusing distance-surface area product) of the effective diffusivity model. We conclude that the effective diffusivity model is a better descriptor of multiple tracer data from the lung, that its parameters are more reliably and independently determined, and that it more reliably describes the effects of flow and histamine infusion on capillary-tissue exchange.