(416a) Inter-Individual Variability in Physiological Response to Lipid Infusions Predicts Considerable Heterogeneity in Outcomes for Lipid Resuscitation: A Physiologically Based Pharmacokinetic-Pharmacodynamic Study in a Virtual Population

Local anesthetic systemic toxicity (LAST) is a life-threatening event in which the introduction of anesthetic to the bloodstream induces cardiac arrest. Anesthetic-induced cardiac arrest is refractory to standard life support measures and is thus often fatal. A bid to understand the etiology of hypersensitivity to local anesthetics led to the discovery of a potential antidote for LAST: intravenous lipid emulsions. These suspensions, composed primarily of vegetable oil, egg phospholipids, and water, are freely available in clinical settings. Their off-label use as antidotes to anesthetic toxicity first yielded a successful clinical resuscitation in 2006. Two years later, a lipid suspension was used to revive a patient suffering from >70 minutes of cardiac arrest that was unresponsive to all other resuscitation efforts. Intriguingly, the offending toxin in this case was not an anesthetic, but a combination of 2 other pharmaceutical compounds. An ensuing series of small animal studies and anecdotal case reports hints at the possibility of lipid emulsions being broad-spectrum antidotes to drug toxicity.

Unfortunately, our understanding of when and how lipid interventions prove beneficial is still limited to clinical anecdotes and small animal experiments. Both sources of information are plagued by heterogeneous outcomes and limited exploration of the multivariate parameter space characterizing cardiac life support. Hence, even the underlying mechanism of therapeutic efficacy has long been the subject of speculation and heuristic reasoning with limited possibility for experimental validation. What we now know about the multifunctional mechanism of action has been obtained through our group’s integrative studies incorporating mechanistic mathematical modeling.

In our most recent work, we sought to further demonstrate how mathematical analysis can be used to gain insight into the likely success of lipid-mediated resuscitation. Importantly, we embraced the inherent variability between pre-clinical subjects and established how estimation of relevant physiological parameters on an individual basis can inform treatment protocols by permitting the analysis of outcome heterogeneity over simulated virtual populations. Using data previously collected in a study involving rats (N=7), we employed uncertainty quantification approaches to identify a subset of 2 key parameters that could be estimated for each subject. One parameter was determined to be correlated with the magnitude of cardiac response to lipid infusion, while the other played a role in determining the smoothness of the response over time. To account for the dynamic nature of these underlying physiological processes, we adjusted our existing pharmacokinetic-pharmacodynamic model to allow time variation of the 2 selected parameters; these were permitted to vary according to four linear splines of equal length. Model comparison tests were employed to ascertain that the choice of 4 splines was most appropriate for our data set.

From our individualized estimates, we constructed population distributions for each parameter. We then sampled from these distributions to create a virtual population (N=10,000) undergoing emulsion therapy after a simulated overdose of the anesthetic bupivacaine. The virtual population exhibited a wide range of expected outcomes, with only a subset of the population expected to recover baseline cardiac function upon treatment. We thereby demonstrate the limitations of using mean responses from small preclinical studies to heuristically posit guidelines for lipid-mediated cardiac resuscitation.