Session 1.1: Viral Clearance Using Traditional, Well-Understood Unit Operations: Low pH and Detergent Viral Inactivation

Background

The previous viral clearance symposia (2009, 2011, and 2013) presented extensive data from both regulators and industry on viral inactivation using low pH and detergents including Triton X-100 and alternative eco-friendly and cost-effective detergents. The understanding of critical process parameters (CPPs) has been evolving over time for the viral inactivation unit operation. Three presentations at the 2015 symposium discussed low-pH viral inactivation, Triton X-100 viral inactivation, and lauryldimethylamine N-oxide (LDAO) viral inactivation.

A Historical Review of Low-pH Inactivation Data at Biogen (Lynn Conley and Brad Stanley; Biogen)

A historical review of low-pH virus inactivation (VI) data collected by Biogen was performed. Data from 35 studies performed with 17 different monoclonal antibodies (mAbs) were compiled and analyzed. The compilation included data from two viruses, several different buffer systems, two temperatures, and a range of protein concentration. No unexpected trends emerged. The data demonstrate slower kinetics at higher pH and lower temperature, which has been reported. These data have been used as supplementary, company-specific data in support of a modular claim for low-pH virus inactivation.

The data presented below are for studies performed with the common low-pH inactivation model retrovirus xenotropic murine leukemia virus (XMuLV) and the less common Suid herpes virus 1 (SuHV-1), which is an enveloped DNA virus. Data have been separated by virus and process temperature and are sorted in each table from low to high pH.

Xenotropic Murine Leukemia Virus (XMuLV) Inactivation

Table 1-1 shows the inactivation data for XMuLV at ambient temperature (defined as 15–26 °C) across a pH range of 3.70–3.97. Samples were typically analyzed at 0, 5, 15, 30, 60, and 120 min. In all cases, infective virus was not detectable after 30 min. The two cases in which it took until the 30 min time point to observe complete inactivation were both at the high end of the pH range.

The XMuLV inactivation studies performed at 2–8 °C across a pH range of 3.48–4.00 are summarized in Table 1-2. Overall, these data demonstrate slower inactivation kinetics and, unlike the ambient temperature data, include cases where there was detectable virus after a 120 min hold (Table 1-2).

Suid Herpes Virus 1 (SuHV-1) Inactivation

In addition to XMuLV, data collected at ambient temperature for the enveloped virus SuHV-1 are summarized in Table 1-3. While the data are limited to a few data points at the high end of what has been observed with XMuLV, the data demonstrate a similar amount of time until virus is no longer detected. For both XMuLV and SuHV-1, it took between 5 and 30 min to achieve no detectable virus from pH 3.80 to 3.90.

Data collected at 2–8 °C for SuHV-1 are summarized in Table 1-4. Again, the data are similar to what was observed for XMuLV with respect to inactivation kinetics: Cold low pH inactivation kinetics is slower and incomplete inactivation is likely to occur as the pH approaches 4.0.

Generic Conditions for Triton X-100 Viral Inactivation (Julia Bach; Amgen)

An ASTM standard was recently created for the low-pH viral inactivation process used by many pharmaceutical manufacturers. The standard includes acceptable set points for factors such as pH, hold time, composition, and temperature. An ASTM standard for detergent viral inactivation would also be a valuable addition. However, while low-pH and detergent inactivation have both been used for many years, there is not as much data available on the operational parameters that may affect the outcome of a detergent viral inactivation performed during the production of mAbs or other biotech products. We report the results of Triton X-100 XMuLV detergent viral inactivation operations from two mAb viral clearance studies, as well as some additional data exploring the effects of sample matrix and Triton X-100 percentage on the performance of a detergent viral inactivation.

In the two viral clearance studies, the XMuLV detergent viral inactivation steps were performed in clarified mAb harvest at a chilled (5 °C) temperature. Table 2-1 shows the results of the two studies, which were performed with duplicate runs (A and B). Samples were taken at 0.5, 10, and 30 min time points for clarified harvest treated with 0.25% Triton X-100. These samples were diluted to eliminate cytotoxicity and interference as previously established. Subsequently, samples were serially diluted and added to 96-well plate containing PG4 indicator cells. A large-volume sample was taken at the 30 min time point to increase assay sensitivity. The plates were incubated for several days and then observed for virus-induced cytopathic effect (CPE). For each study, residual infectious virus was detected in the large-volume assay at the 30 min time point in one out of the two duplicate runs. However, the overall log reduction value (LRV) for detergent viral inactivation in each study was robust, at >5 LRV. These data indicates that for chilled harvests, 0.25% Triton X-100 may not be sufficient to consistently achieve complete inactivation of XMuLV.

Another study looked at the effect of Triton X-100 percentage and different mAb concentrations on virus inactivation (Table 2-2). Harvest samples with different concentrations of a mAb were inactivated at 15 °C with 0.1% and 0.3% Triton X-100. Large-volume samples (4 mL per condition) were taken at 0.5 and 3 min and plated on PG4 indicator cells. Titrations were not performed on these samples. The data shows that 0.1% Triton X-100 may not be sufficient to completely inactivate XMuLV, particularly at higher mAb concentrations. However, 0.3% Triton X-100 caused complete clearance at the same mAb concentrations. These conditions were not tested in duplicate; therefore, assay variability should not be ruled out. All of the conditions are providing robust detergent inactivation.

In a different set of experiments, two mAbs were subjected to detergent viral inactivation (0.3% Triton X-100, 5% virus spike, 15 °C) using two different sample matrices. For each antibody, a clarified harvest sample was tested, as well as a lysed cell supernatant sample, which would contain a high concentration of cell debris, DNA, lipids, and other cell culture process byproducts. Samples were analyzed at 0.5 and 3 min time points. Table 2-3 shows that for each antibody, although robust virus inactivation of LRV above 4 logs was observed, residual infectious virus was detected in one out of two lysed cell supernatant samples. Complete virus inactivation was observed for the clarified samples. These measurements were performed in singlicate and most of the results are within assay variability. However, publications have described studies that evaluated the effect of specific components, such as lipids and DNA, on detergent virus inactivation and found little effect on overall inactivation (1). These data largely agree with that conclusion, as the inactivation was still robust. Nevertheless, there may be other components in the lysed cell supernatant that could interfere with complete viral inactivation.

Lauryldimethylamine N-oxide (LDAO) Viral Inactivation Robustness Studies (Lynn Conley, Alexis Henry, Edward Koepf, Brad Stanley, Marisa Labanca; Biogen)

In large-scale protein purification processes, treatment of the product stream with a detergent such as Triton X-100 provides a simple, reliable, and effective approach for the inactivation of enveloped viruses. However, in some countries Triton X-100 is classified as an ecotoxic reagent that requires special waste disposal (2). In Denmark, where Biogen operates a large-scale manufacturing facility, Triton X-100 is classified as a Class A waste stream according to the Danish ABC waste water classification system, which means its discharge is regulated and substitution of the reagent is recommended. An alternative detergent for viral inactivation that may be discharged without concentration limits (Class C waste stream) is the amine oxide surfactant LDAO, a molecule that rapidly degrades in the environment. LDAO is a zwitterionic, amphipathic molecule that disrupts lipid–lipid interactions in the virus membrane and solubilizes membrane proteins, both of which serve to inactivate enveloped viruses. These properties allow it to be an effective reagent for viral inactivation in cell culture protein production processes.

Data presented at the 2013 Viral Clearance Symposium showed inactivation of the enveloped viruses XMuLV and SuHV-1 with 0.14% LDAO to be fast and robust (3). For XMuLV, ≥3.9 to ≥4.6 LRF was achieved within 5 min of incubation, while ≥2.3 to ≥5.1 LRF was achieved for SuHV-1. To date, four clinical products have been validated and Investigational New Drug applications have been filed using 0.14% LDAO with 2 h of incubation at 5 °C.

To determine the lowest concentration of LDAO capable of inactivating enveloped viruses, 0.14% LDAO (control) and four decreasing concentrations (A–D) were tested with SuHV-1, vesicular stomatitis virus (VSV), and bovine viral diarrhea virus (BVDV) (Table 3-1). Complete inactivation (≥5.20 LRF) was achieved with SuHV-1 at 5 min with the control and concentrations A–C, while concentration D only achieved 2.60 LRF, even when extended to 120 min of incubation. For VSV and BVDV, complete inactivation was observed at 15 min for all concentrations tested (0.14%, A, and B). For VSV ≥5.91 LRF was achieved with 0.14% LDAO, while the LRV for concentrations A and B were ≥6.40 and ≥6.41, respectively. For BVDV ≥4.22 LRF was achieved with 0.14% LDAO, while ≥4.71 LRF was observed with both concentrations A and B.

To further explore LDAO inactivation robustness and to gain a better understanding of parameters which may influence XMuLV inactivation, additional experiments were carried out that investigated several experimental factors including the nature of the protein feed stream, protein concentration, LDAO concentration, conductivity, and pH (Table 3-2).

In a typical Biogen protein purification process, cell culture fluid (CCF) post-flocculation is clarified by centrifugation, depth and polish filtration, and adjusted to neutral pH, generating the harvested cell culture fluid (HCCF) intermediate. Addition of LDAO to the HCCF intermediate and incubation for 120 min at 5 °C constitutes the viral inactivation step. To determine if higher levels of cell-derived impurities (lipids, host cell proteins, and DNA) affect the ability of LDAO to inactivate XMuLV, clarified cell culture fluid (CCF, no flocculation or depth filtration, only centrifugation and polishing filtration) was examined as a secondary feed stream. The lipid and host cell protein concentration in the clarified CCF was 2× higher than in the HCCF, while the DNA concentration was 1000× higher. Both samples were tested at native pH and conductivity (Table 3-2). The inactivation kinetics with A% LDAO and both feed streams were indistinguishable and complete at 5 min of incubation, indicating that higher concentrations of host cell impurities do not interfere with the ability of LDAO to inactivate XMuLV at 5 °C at this specific LDAO concentration.

As second study investigated the effect of protein concentration (9, 20, and 70 mg/mL) in the feed stream, with samples prepared by spiking highly concentrated, purified protein into HCCF at native conductivity (∼15 mS/cm) and pH (∼7). The 70 mg/mL sample mimics the protein concentration in a process where the HCCF is concentrated several fold (single pass tangential flow filtration) for volume reduction prior to detergent addition, while the 20 mg/mL concentration reflects a very high titer cell culture process. For the control experiment the protein concentration of the material was 9 mg/mL. The inactivation kinetics of XMuLV at 5 °C with all three protein concentrations were the same with complete inactivation achieved by 5 min, demonstrating that protein concentration also did not affect the ability of LDAO at A% concentration to inactivate XMuLV.

Based upon results summarized in Table 3-1, the parameter suspected to influence the inactivation kinetics of XMuLV the most was LDAO concentration. In this experiment four concentrations of LDAO were evaluated (A–D), with A representing the control process. The two highest concentrations (A and B) achieved complete inactivation at 5 min of incubation at 5 °C, whereas the other two samples (C and D) at lower LDAO concentrations did not achieve substantial viral inactivation, even when the incubation was extended out to 120 min.

The ionic strength of a sample can affect the properties of a detergent by influencing its critical micellar concentration and the size of the micelles formed. The effect of conductivity (5, 15, and 25 mS/cm) on viral inactivation was evaluated by spiking NaCl into HCCF (25 mS/cm) or diluting the sample with water (5 mS/cm). Our typical cell culture process generates HCCF with a conductivity of approximately 15 mS/cm. The 25 mS/cm represents a sample that is outside the upper range of what is typically observed at cell culture harvest, whereas the 5 mS/cm sample represents a sample that is diluted as a consequence of depth filter flushing. The results of the conductivity screening experiments showed that LDAO (A% concentration and 5 °C) was capable of achieving complete inactivation of XMuLV at 5 min of incubation at all three conductivities tested.

In processes that utilize Protein A affinity chromatography for capture, the HCCF is typically adjusted to near neutral pH, if necessary. In this specific study, LDAO at A% and 5 °C, was evaluated in HCCF within the pH range of 6.0 to 8.0. The inactivation kinetics at all pHs tested were identical and complete by 5 min, further demonstrating the robustness of this detergent for XMuLV inactivation.

Summary

The data presented by Biogen from 35 studies performed with 17 different mAbs confirmed that low-pH viral inactivation is a reproducible, well understood step. The expected reduction in inactivation kinetics with increased pH and cold temperature is aligned with data presented at the previous viral clearance symposia. The data for XMuLV inactivation at ambient temperature provide strong support for the current ASTM standard, as all studies were performed above the minimum pH specified in ASTM E2888-12 and still indicate no detectable virus after 30 min. In addition, the limited SuHV-1 data align well with the trends observed for XMuLV.

Triton X-100 viral inactivation data presented by Amgen demonstrated that robust viral clearance (≥4 LRV) can be achieved if performed at >0.3% Triton X-100 for 30 min at ambient temperature in clarified harvest fluid. The data also indicate that for certain process conditions, such as a chilled harvest fluid or a complex sample matrix, a higher concentration of Triton X-100 may be needed to achieve complete inactivation of enveloped viruses. Further study is needed to understand what components in the complex sample matrix could interfere with complete viral inactivation.

Biogen's alternative, eco-friendly detergent (LDAO) data indicate that LDAO is an effective and robust detergent for the inactivation of enveloped viruses. Of all the parameters investigated during the study, only the detergent concentration had an impact on XMuLV inactivation. The impurity load of the feed stream, protein concentration, conductivity, and pH did not affect the inactivation kinetics. LDAO appears to be a viable option for viral inactivation in situations where waste disposal for commonly used Triton X-100 may be of concern. Additional studies would be required to support a modular claim for the use of LDAO for viral inactivation.