Objectives: Human challenge studies (HCS) are used to evaluate new antiviral treatments and establish clinical proof of concept. Typically, a mean viral load (VL) profile is presented as a key figure in antiviral therapy publications. In contrast to fixed dosing times relative to the viral challenge (inoculation), the study design for RSV (respiratory syncytial virus) challenge models includes triggered dosing times (i.e., variable dosing time based on qualitative PCR positivity post-challenge) which adds a layer of complexity to the information encoded. We sought to illustrate common errors in interpreting the typical VL profile and propose an alternate figure that might reduce biases in interpretation.
Methods: We used viral dynamics modelling to fit clinical data from a published RSV HCS [1, 2]. Placebo and treatment arms were simulated in a triggered dosing scheme. We used simulations to illustrate differences in individual VL dynamics and summary VL in terms of features (VL peak, time), and explored the impact of imputation rules for values below the lower limit of quantification (LLOQ).
Results: An alternate key mean VL figure is proposed which includes stratification by dosing time relative to time of challenge, number of individuals in each stratification, and clear indication of the LLOQ. Simulated examples are used to demonstrate that the antiviral treatment effect observed via reduction in VL may be confounded by variable dosing times. Specifically, later dosing times (relative to viral challenge) may lessen the observed reduction in VL.
Conclusions: An alternative presentation of the mean VL figure should be considered to provide a more unbiased illustration of antiviral effect in the setting of challenge studies with a triggered dosing design.
Citations: [1] DeVincenzo, J.P., et al., Oral GS-5806 activity in a respiratory syncytial virus challenge study. N Engl J Med, 2014. 371(8): p. 711-22. [2] Kelly, G., et al., Use of qualitative integrative cycler PCR (qicPCR) to identify optimal therapeutic dosing time-points in a Respiratory Syncytial Virus Human Viral Challenge Model (hVCM). J Virol Methods, 2015. 224: p. 83-90.
