Estimating viral titres in solutions with low viral loads.

An important consideration in the manufacture of products derived from animal or human sources is the virus reduction capacity of the manufacturing process as estimated using validated bench-scale models of relevant manufacturing steps. In these studies, manufacturing process intermediates are spiked with virus and processed using the bench-scale model and the resulting viral titres of input and output samples are typically determined using cell-based infectivity assays. In these assays, the Spearman-Kärber (SK) method is commonly used to estimate titres when there is one or more positive observation (i.e., the presence of any viral cytopathic effect). The SK method is most accurate when the proportion of positive observations ranges from <0.1 to >0.9 across dilutions but can be biased otherwise. Maximum likelihood (ML) based on a single-hit Poisson model is an alternative widely used estimation method. We compared SK with ML and found the methods to have similar properties except for situations in which the concentration of virus is low but measurable. In this case, the SK method produces upwardly biased estimates of titres. Based on our results, we recommend the use of either ML or SK at most virus concentrations; however, at low virus concentrations ML is preferred.