Predicting the effects of combining broadly neutralizing antibodies (bNabs) binding to different HIV viral epitopes
Objectives: Promising findings demonstrate that broadly neutralizing antibodies (bNabs) significantly reduce viral loads in people living with HIV—giving hope for its eventual use in treatment and cure strategies. The objective here was to model the inhibition caused by a range of bNAbs across a panel of 199 clade-C HIV-1 virus strains, using nonlinear mixed effects modelling, and to predict the effects of combining bNAbs binding to different viral epitopes in order to support programmatic decision-making.
Methods: A panel of bNabs targeting different epitopes were tested in a highly quantitative pseudovirus neutralization assay using TZM-bl target cells  on a panel of 199 viral clones. Neutralization curves of 9 different bNabs, binding to envelope protein, top of the trimer, or to same site as the human HIV receptor CD4, in the 0.000128 to 10 ug/mL concentration range, were available for analysis. Sigmoidal Emax models were fitted to the data from each bNAb, including between-strain variability on E0, IC50, and the Hill factor. For combinations of bNAbs, the effects were predicted using the principles of Loewe interaction and Bliss independence  and were compared to the observed inhibition in assays using these combinations. From the model, a good estimation of inhibition effects- the concentration giving at least 80% inhibition in 80% of viruses types, the IC80(80), could be estimated. The FOCE method and NONMEM 7.2 was used to fit the data.
Results: For each bNAb, the inhibition curves varied widely between the virus strains. Notably, some bNabs do not always achieve 100% neutralization even at high concentrations. Despite this challenge, data were well fitted using a sigmoidal Emax model which included a mixture model for IC50. Even so, the variability in IC50 within each mixture subset remained large.
IC50s generally correlated across strains for bNAbs binding to the same epitope. Predictions using the Bliss independence method agreed well with the observed effect of combinations.
Conclusion: In vitro data of bNAb effects in neutralizing HIV-1 virus showed a highly variable potency between viral strains. Data could be well fitted using sigmoidal Emax models with a mixture model on IC50. When assuming Bliss independence, the model proved predictive for the effects of combinations of bNAbs binding to different viral epitopes. The bNab combinatorics predictive platform is firmly at the core of data-driven decision-making for the bNab
 Kong, R., Louder, M. K., Wagh, K., Bailer, R. T., deCamp, A., Greene, K., … Mascola, J. R. (2015). Improving Neutralization Potency and Breadth by Combining Broadly Reactive HIV-1 Antibodies Targeting Major Neutralization Epitopes. Journal of Virology, 89(5), 2659–2671.
 Berenbaum, M. C. (1989). What is synergy? Pharmacol Rev, 41, 93–141 program.
Full Reference: Olsson Gisleskog, P., Seaman, M., Anklesaria, P. , Jumbe, S. Predicting the effects of combining broadly neutralizing antibodies (bNabs) binding to different HIV viral epitopes. PAGE 25 (2016) Abstr 5983 [www.page-meeting.org/?abstract=5983]