Local anesthetic systemic toxicity and animal models for rescue paradigms: can pigs fly?

Mauch and colleagues (1) provide in a recent issue of Pediatric Anesthesia an intriguing study of treatment for local anesthetic systemic toxicity (LAST) in piglets receiving sevoflurane anesthesia. Their key finding was that epinephrine administration was the common factor among all the survivors in each of the four treatment groups comprising epinephrine alone, intravenous lipid emulsion (ILE) alone, ILE plus epinephrine, and ILE plus vasopressin. There was no difference in overall survival among the groups. However, only animals receiving epinephrine had initial return of spontaneous circulation (ROSC) and those with delayed ROSC after an initial treatment failure returned only after a second dose of epinephrine. They also noted that among the ILE-treated animals achieving ROSC, lipid appeared to have a ‘stabilizing effect’ as they required less adrenergic support to maintain circulation. Nevertheless, they concluded that for LAST, ‘.... epinephrine should be the first-line rescue drug’. The study by Mauch et al. provides an excellent opportunity to share a secret known only to those who have performed such experiments: it is exceedingly difficult to establish an animal model of LAST. Readers looking only for the bottom line of any such study are unlikely to appreciate the gargantuan effort that went into their experiments, starting with formulating a model system that can answer tough questions and provide clinically meaningful insights into a complex biological phenomenon. Our own experience can serve as a canon. We decided several years ago to set up a model of bupivacaine toxicity to compare ILE with other resuscitation methods – much as Mauch has done. We soon realized that every element of the process was fraught with problems. A wrong choice would introduce potential experimental confounders that could limit the relevance of the results or muddy their interpretation. It took the full effort of our laboratory over 10 months and nearly 100 experiments to finalize a system that was highly reproducible and could test hypotheses related to treatment of LAST – all this before collecting any data that would be published (2). These are some of the key questions to answer first, the ‘front load’ to such experiments: choice of animal, anesthetic, local anesthetic, its dose, route and rate of administration, treatment protocol(s) and sample size, metrics, endpoints, how to define recovery (e.g., ROSC), and statistical interpretation. Each question might appear uncomplicated, but answering them is rarely straightforward and never simple. Moreover, while simultaneously retaining focus on testing the underlying hypotheses (this is scientific method, after all), one must continually question the clinical relevance of the experiments – that is, ‘Will the results influence practice’? One can make an educated guess about each of these elements, but in all likelihood as the pilot experiments progress, the model will be revised many times for practical (e.g., financial), technical, and scientific reasons. Even apparently straightforward questions become very challenging but we will focus on four in this editorial. First, ‘How should we dose the local anesthetic’? If we assume the use of intravenous injection (does this really model clinical LAST?) should a single, defined challenge be administered to all animals or, alternately, should one use an infusion to a chosen physiological endpoint? Which option offers the best chance to test a hypothesis? The first would likely give the most reproducible plasma concentration for animals of a given size range but does not account for interindividual variation in susceptibility to LAST, a very real clinical problem. The later will potentially control for such variation but leaves the experiments open to substantial variation in dose. We tried both methods and opted for the fixed-dose and chose a strain of highly inbred, congenic rats as a means of limiting interindividual variation to the local anesthetic. Mauch et al. chose the fixed endpoint. However, the sizable range of doses required (5.1–20.2 mgÆkg in one group) confirmed substantial interindividual variation in pharmacokinetics or sensitivity to bupivacaine. This becomes more problematic when there are intergroup differences as appears to be the case for groups 1 (epi only) and 2 (lipid only), which received mean doses of 7.7 and 9.0 mgÆkg, respectively. Overall differences among the several groups might have eluded statistical detection, but these doses look different enough to concern me. Were these groups challenged equally? This issue could be particularly problematic since the median plasma concentrations appear different (70 versus 88 microM, groups 1 and 2 respectively). Was this really a fair comparison if the ‘winning’ treatment was given to the group with the smaller local anesthetic challenge? Moreover, the median plasma bupivacaine levels among surviving animals in these groups (68 and 101 lM, for groups 1 and 2, respectively) also seem sufficiently different to suggest that in these animals, ILE exerted a sink effect, conferred added resistance to bupivacaine, or both. It appears that in avoiding inter-