Jane Knöchel (1,2), Sathej Gopalakrishnan (2,4), Charlotte Kloft (3), Wilhelm Huisinga (1)
(1) Computational Physiology Group, Institute of Mathematics, Universitaet Potsdam, Germany (2) PharMetrX Graduate Research Training Program, Freie Universitaet Berlin and Universitaet Potsdam, Germany (3) Department of Clinical Pharmacy and Biochemistry, Freie Universitaet Berlin, Germany (4) Current address: AbbVie, Ludwigshafen, Germany
Objectives: The emergence of resistant mutants within treated HIV patients has been studied intensively by deterministic mathematical models. These models, however, are not able to predict the observed probability of treatment success/failure in a treated patient population, which hinders comparison to clinical data. For realistic mutational landscapes, the relevance of randomness of mutation dynamics on the appearance of drug resistance has not been studied. Our objectives were therefore to quantify the impact of random effects on the development of resistance and treatment outcome.
Methods: We used a two-stage mechanistic HIV infection model [1] to describe in vivo viral and mutational dynamics. The model includes drug-specific mutation pathways and resistance factors estimated from clinical data. The stochastic model was implemented based on a hybrid model approach [2] to stochastically model the appearance of new mutations. The accumulation of mutations for two drugs: zidovudine (ZDV) and indinavir (IDV) was studied. Simulations were performed in Matlab (R 2014a).
Results: We observed a delayed appearance of mutations in the stochastic model compared to the deterministic model. Whereas the extent of the delay was low for preexisting drug-resistant genotypes, for non-preexisting genotypes that developed under therapy, we found a more pronounced delay (up to twofold). A main reason for the observed difference is the way the viral numbers are represented in the two modelling approaches. While the stochastic model correctly accounts for the discreteness of viral numbers, the deterministic model approximates viral numbers continuously. As a consequence, all mutations are instantaneously present in the deterministic model. Further we observed different dynamics of accumulation of drug-resistant mutants. In the stochastic model some mutants have a transient rise after which they decline again, while this was not observed in the deterministic model.
Conclusions: Randomness and discreteness of viral load impact the emergence of mutations. This finding is expected to have a major impact on the mutation dynamics under antiviral combination therapy.
References:
[1] Gopalakrishnan S, Montazeri H, Menz S, Beerenwinkel N and Huisinga W (2014) Estimating HIV-1 Fitness Characteristics from Cross-Sectional Genotype Data. PLoS Comput Biol 10(11): e1003886. doi:10.1371/journal.pcbi.1003886
[2] von Kleist M, Menz S, Stocker H, Arasteh K, Schütte C and Huisinga W (2011) HIV Quasispecies Dynamics during Pro-Active Treatment Switching: Impact on Multi-Drug Resistance and Resistance Archiving in Latent Reservoirs. PLoS One 6(3)
Reference: PAGE 24 (2015) Abstr 3544 [www.page-meeting.org/?abstract=3544]
Poster: Drug/Disease modeling - Infection