3.1 Mammalian cell lines are widely used in the production of biological therapeutics, such as monoclonal antibodies and other recombinant proteins. Some of these cell lines, like rodent cell lines, are known to contain genes encoding endogenous retroviral-like particles or produce endogenous retrovirus, but there is no evidence of an association between rodent retrovirus and disease in humans. Adventitious viruses can be introduced into a drug substance manufacturing process from other sources, and contamination of human therapeutics is a safety concern (3).

3.2 Virus filtration, an orthogonal technology in a virus clearance platform to such steps as low pH or surfactant inactivation, has traditionally been accepted as a robust method for virus clearance when well designed. Size exclusion has been shown to be the primary mechanism of virus removal by virus retentive filtration, that is, larger viruses are more easily retained than smaller viruses such as parvoviruses (4, 5). Large virus retention has also been shown to be insensitive to process fluid characteristics such as protein type, protein concentration, pH, and ionic strength (4, 6, 7, 8, 9, 10). In contrast, for small viruses, aspects like flow pausing and/or flux decay can impact clearance (4, 6, 11).

3.3 Large virus retentive filters, or retrovirus filters, are tested for removal of larger enveloped viruses like retrovirus or MuLV (80?nm to 100 nm) and have undetectable levels of the large bacteriophage PR772 (64?nm to 82?nm) (1). Small virus retentive filters, or parvovirus filters, are designed to remove parvovirus, like MMV (18?nm to 26?nm) (1). Since size exclusion has been demonstrated as the mechanism of virus retention, retroviruses, which are three to four times larger than parvoviruses, should be large enough to be completely retained, with undetectable levels of retrovirus in the filtrate, by all small virus retentive filters designed to remove parvovirus.

3.4 Numerous published studies and reviews encompassing data from the last 20 years have shown both large and small virus retentive filters are effective and consistent for removal of retrovirus. In published reviews of regulatory submissions from 1990 through 2010, rare occurrences of retrovirus breakthrough did occur across both large and small virus retentive filters. These anomalies, however, were not resolved and could be attributed to study design, experimental artifacts, or limitations of the meta-analyses performed on the regulatory submission (12). In a summary of 89 submissions to Paul Ehrlich Institute (PEI), processes using either large or small virus retentive filters showed no detection of any infectious particles from large viruses (12). A collection of viral filtration results across eight biopharmaceutical companies showed no large virus breakthrough across any small virus retentive filter for all 198 experiments reviewed (7). Additionally, a recent review of 20 plus years of small virus retentive filter experiments from two viral clearance testing houses showed only 0.61?% (14 out of 2311 experiments) viral filtration studies performed with larger viruses had detectable, replicated virus (10). This manuscript further suggests that all positive results were not due to viral breakthrough of integral small virus retentive filters, but rather to other causes that included virus detection assay, aerosolization of virus during filtration, splashes, spills, or potential use of non-integral laboratory scale filters.

3.5 The level of justifiable LRV for a modular claim from a censored dataset, that is, all observations are below a certain LRV level, can be difficult to estimate (13), and is almost always an underestimation of an actual value. LRV approximations from censored data are influenced by viral spike volume, viral spike titer, load volume, and assay sensitivities to product matrices, in addition to small virus retentive filter performance. In the Mattila review, where no large virus was detected in 198 experiments, small virus retentive filters showed an average clearance >5.86 LRV ±0.91 for Reo3, a medium-sized virus (60?nm to 80?nm), suggesting at least equal or better clearance for the larger retrovirus, MuLV (7). For creation of modular claim from censored values, Stuckey, et al. proposed that the highest LRV in set of censored data could be used for a modular claim (6). In these experiments, load material spiked with both parvovirus and retrovirus was used to challenge small virus retentive filters. Parvovirus breakthrough was observed but no retrovirus breakthrough was detected, further supporting both a size exclusion-based mechanism of virus retention and the robustness of retrovirus retention using small virus retentive filters (6). Additionally, even large virus retentive filters, designed to retain larger viruses and having on average larger pore sizes than small virus retentive filters, are classified as having the ability to retain >6 logs of PR772, a 64?nm to 82?nm bacteriophage often used as a surrogate for retrovirus, in virus filtration studies (1, 14). These large virus filters have been shown to clear >8 logs of PR772 (15), and by definition clear more than 6 log₁₀ of retrovirus (12). These published data, collectively, support a modular claim for small virus retentive filters of >6.0 LRV for retrovirus (MuLV).

3.6 Implementing parameters of small virus retentive filtration established by this practice can provide robust retrovirus removal and can be used as a modular retrovirus validation of the virus filtration step. In conjunction with other clearance unit operations (for example, chromatography and inactivation by pH or surfactants), sufficient overall retrovirus clearance can be achieved (3).