A computational model of hemorrhage and dehydration suggests a pathophysiological mechanism: Starling-mediated protein trapping.

We sought to understand the degree to which a single computational cardiovascular model could replicate the typical responses of healthy subjects through a breadth of blood loss patterns and whether such a model could illuminate the cause-effect relationships that underlie the observed responses. The model consisted of compartments for the upper body, lower body, viscera, and kidneys as well as a four-chambered heart and a pulmonary compartment. Transcapillary fluid flux was governed by Starling forces, whereas lymphatic flow was driven by hydrostatic tissue pressure and scaled by a lymphatic activation term. We adjusted parameters based on results from one protocol involving moderate continual blood loss in a canine model. Next, we simulated six additional protocols spanning euvolemic and dehydrated subjects and compared in silico behavior with in vivo hemodynamic responses and fluid shifts. The model was able to replicate group-averaged behavior (i.e., within 1 or 2 SEs) of the rate and quantity of vascular refill and the associated cardiac output during slow, moderate, and rapid ongoing blood losses, the restitution after the cessation of blood loss, and the absence of restitution in dehydrated subjects. The model suggested that the earlier phase of restitution, i.e., transcapillary fluid shifts, was antagonistic to the later phase of restitution, i.e., protein return via lymphatics. This phenomenon was termed "interstitial protein trapping." In conclusion, the model appears valid for a range of blood loss patterns and prehydration states. Further investigation into the in vivo relevance of interstitial protein trapping is justified.