Investigating the pharmacokinetics of snake venom – a cocktail of toxins

Background: Snake venom is a mixture of protein-toxins, which can cause a range of biological effects. The protein components in snake venom have molecular sizes ranging from 4 kDa to 150 kDa and are expected to exhibit different pharmacokinetics (PK) profiles. Therefore, the phenomenological appearance of plasma snake venom concentrations is governed by the combination of disposition processes inherent to each protein component. Understanding the influence of the PK of each molecular weight fraction on the overall venom PK can improve our understanding of the individual toxin PK profile and help identify how venom PK can be influenced by the composition of the dominating proteins.

Aims: This study aims (1) to determine the effects of different molecular weight proteins on the time course of the snake venom, and (2) to determine the ability to identify toxin profiles based on their integral only.

Methods: The relationships between proteins of variable molecular weights and their clearance (CL) and volume of distribution (V) were investigated and found to follow a simple log-linear relationship. Both aims were addressed using a stochastic simulation estimation (SSE) study using MATLAB for simulation and NONMEM for estimation. Sixteen variations of venom comprising two to nine toxins of variable molecular weights were investigated. Each venom variation was evaluated in a SSE study involving 100 virtual patients with a rich sampling scheme. The prior population values of CL and Vfor each molecular weight toxin were generated from a distribution based on their molecular weight. Individual values of CL and V were simulated from the population values assuming an exponential between subject variability model and a combined residual error model was used to generate the data. The venom data were modelled as the sum of each toxin data under three scenarios: (i) an intravenous (IV) bolus 1-compartment model using the population parameter estimates without uncertainty in the prior [perfect case], (ii) an IV bolus 1-compartment model using population parameter values generated from the prior including uncertainty [best field case], and (iii) a first-order absorption 1-compartment model with population parameter values generated from the regression model with uncertainty [likely field case]. The venom concentration-time course was modelled using 1- to 9- compartment models. Akaike information criterion (AIC) was used as a basis for model selection.

Results: The concentration-time data of sixteen venoms were best described by 2- to 3- compartment model in all three scenarios. Data from venoms comprised of two compounds favoured a 2-compartmental fit, over a 1-compartmental fit. Data from venoms comprising more than 4 components seldom preferred more than a 3-compartment model. In scenario (i), data of all venoms that comprised of three or more compounds were best described by a 3-compartment model. In scenarios (ii and iii), where uncertainty was incorporated, data of venoms with three or more components were best described by 2- or 3- compartment models. Despite venoms comprising more than 3 toxins of various molecular sizes and PK characteristics, we can observe no more than 3-compartmental behaviour.

Conclusion: A population pharmacokinetic analysis of venom data does not support the identification of more than a three-compartment profile. This indicates that it is not possible to determine the toxin profile of venoms based on measuring whole venom only, except in circumstances where there are a minor number of highly expressed determinants.