Experiences with HEK293: A Human Cell Line

Viral Safety Precautions for HEK293

In 2001 Eli Lilly and Company received approval for a recombinant human protein produced in a human cell line, HEK293. This paper covers the strategy that we implemented to assure approval of the product, as well as some of our experiences with issues that arose in obtaining approval for a human therapeutic produced in this cell line.

The strategy for successful approval of any product produced in a eukaryotic cell substrate revolves around properly assembling a viral safety strategy including the following main points. First, an understanding of the viral contamination risk associated with the cell line is necessary. The potential for viral contamination varies between cell types. A search of the scientific literature can identify viruses that can replicate in various cell lines. In all cases the cell line to be used should be tested for the presence of classical viruses. A cell line positive for a viral contaminant has only in rare circumstances, if ever, been allowed to be used to produce a recombinant therapeutic. The presence of latent viruses in the cell line should also be assessed. Many cell lines have retroviral-like particles that can be produced by the cell line when stressed. Second, potential sources of viral contamination need to be identified. As more viral contamination events occur, we are beginning to better understand sources of viral contaminants and routes by which viruses can gain access to the process stream. Animal sourced materials have for some time been identified as a known source of viral contaminants. All human- or animal-sourced raw materials, if they must be used, should have very rigorous control strategies associated with them in order to assure they are free from infectious viruses. Thirdly, the manufacturing process used (i.e., batch vs perfusion) has the potential to affect the risk of viral contamination. Perfusion bioreactors, which undergo media changes daily, provide more opportunities for a viral contaminant to enter the process stream. The fourth point is unrelated to the cell line. However, an understanding of global regulatory agencies is also important. Although many companies focus on the U.S., Europe, and maybe Japan, there are significant growing markets in Central and South America, Australia, and Canada, all of which have sophisticated regulatory agencies. The final point that is critical to a viral safety strategy is an understanding of the viral clearance capability of the purification process for final drug product.

The remainder of this paper expands on the preceding points. Firstly, as to understanding the risks of the cell line, an understanding early in the development process is extremely important. In most cases a cell line devoid of replicating classical viruses is important. Cells such as Chinese hamster ovary (CHO), which express retroviral-like particles, have been used to produce many approved products. Acknowledging that a cell line has risks associated with it and building in safety systems is critical, as are discussions with regulatory agencies early in the development process, such as we had with regard to our product produced in the HEK293 cell line. As detailed viral clearance studies may take a couple years from planning to final data, it is necessary to reach agreement with the regulatory agencies early in the development process. For us, this was the first time that HEK293 had ever been submitted as a recombinant cell line to any agency and, in fact, no human cell line had ever been submitted as the production substrate for a recombinant product.

The risks associated with HEK293 were high, as this cell line can replicate many viruses (Table I). In fact, HEK293 is the cell line of choice in viral clinical laboratories for the isolation of many human viruses and is capable of producing some human viruses at very high titer. One positive aspect of the HEK293 cell line is that it does not have retroviral-like particles or any other latent viruses.

Secondly, potential sources of viral contamination must be identified. Since it was already known that HEK293 cells are highly susceptible to many viruses, it was critical for us to build a robust system of testing and quality control. Development of this system was confounded by the use of two animal-sourced materials in the bioreactor. As a result of this, we employed strict controls on the source of these animal-derived materials in addition to subjecting them to viral inactivation treatments. We also performed additional studies in which we exposed HEK293 cells to bovine viruses to determine if the cell line was susceptible to these viruses. Knowledge of the susceptibility of a cell line to viruses is vital, as it can be key to the design of the viral testing scheme at the end of the bioreactor run, a central component of the viral safety strategy. Regarding sources of virus contamination, we learned a lesson on the potential for human viruses to gain access to the process stream, as we experienced a human adenovirus contamination in a HEK293 bioreactor run. Since the viral contaminant was human adenovirus, we realized that human viruses could somehow gain entrance into the process stream, most likely through the media.

The third point in designing a viral safety strategy is to understand the potential risk to the patient. The International Conference on Harmonisation guideline ICH Q5A discusses this point. Details such as use of the therapeutic, the age of the target population, and whether it is for an acute or chronic condition can be taken into consideration. These items realistically have very little impact on the overall design and the amount of viral clearance achieved.

The fourth point is an understanding of the regulatory environment and an awareness of current viral safety issues. This can be achieved by attending viral safety conferences, understanding the national (e.g., U.S. Food and Drug Administration's Points to Consider) and international regulatory documents (e.g., ICH Q5A). The difficulty is being aware of the “unwritten” regulations that some countries may impose. Therefore, it is important to identify early in the development process in which countries the product will be marketed. Many new developing markets have highly educated regulatory staff and some difficult questions will be raised. In particular, many questions will be asked about the history of the cell line development, so keeping a history of the handling of the cell line through all aspects of development is crucial. For our HEK293 cell line, a question was received about the horse serum that was used more than 10 years prior, during the early development of the cell line. The Japanese regulatory authorities also had concerns regarding the HEK293 master cell bank freezing solution and a bovine lipid derived from adult U.S. cattle that were used in the bioreactor. On another note, regulators from one smaller market asked us to test the master cell bank for Ebola, an example of a question that will require experts to say no or that the data is not available.

The fifth point is an understanding of the viral clearance of the process. Clearance of viruses or virus-like particles must be robust for cell lines that express these agents. For our HEK293 cell line, we conducted extensive viral clearance studies. Eventually, we evaluated the clearance of seven different viruses (Table II presents clearance results for six viruses). Clearance studies for the seventh virus were performed post-approval as requested by one regulatory agency. The seventh virus was bovine polyomavirus. To model clearance of bovine polyomavirus, we used simian virus 40 (SV40) as the model virus for the clearance studies and attained robust clearance. The model viruses used in our studies represented a broad spectrum of virus morphologies and types. Some of the viruses are actual potential contaminants of the animal-sourced materials used in our production process. In our clearance studies, we evaluated almost all of our process steps, but not for each virus, as we knew some viruses would not be affected by specific clearance steps. For example, we knew that the viral inactivation step and the mild heat step would not affect minute virus of mice (MVM) (a parvovirus). There are two dedicated viral clearance steps in our downstream process. They are a proprietary inactivation step and a nanofilter. We also evaluated two chromatography steps and a mild heat step. The viral clearance studies were conducted at a biological safety contract laboratory with extensive experience in conducting the properly designed studies. It took two to three years for the contract laboratory to scale down the process steps, perform the necessary preliminary studies, and, finally, the viral clearance studies. During the course of this work we adhered to the appropriate regulatory guidelines, but it should be noted that the details for properly performing a viral clearance study are not provided in these documents and must be obtained from other sources.

In addition to the points discussed above, there are other safety factors that we found to be important, especially for producing a product made from a human cell line. A sensitive viral detection assay, to test bioreactor material at the end of the bioreactor run, is critical. However, in many cases the material is processed forward by the time the viral detection assay results are available. The in vitro viral detection assays may take up to a month to complete. During our manufacturing bioreactor runs, we collected samples daily from the bioreactor and stored them, but only tested the sample taken the last day of the run. These archived samples proved to be important for us during our investigation of the viral contamination event. By testing the retained samples we were able to identify when the contamination event occurred. This provided us with a better understanding of how a viral contaminant entered the process. The other critical safety factor that we identified was the control of raw materials. Much has been written and discussed regarding the importance of controlling viral contamination in animal-sourced materials, but we determined the human viral contaminant in our event came from a non-animal-sourced material. The paradigm shift for our understanding was that human viruses can enter a process through routine handling of raw materials that are part of the media formulation. With a cell line as sensitive as HEK293 to infection by human viruses, even a minor contamination of a raw material (10 viral particles or more) is sufficient to result in a major contamination event.