key: cord-0005039-dtobju6z
authors: Koplove, H. M.
title: Cell culture and purification process — purity and safety testing
date: 1994
journal: Ann Hematol
DOI: 10.1007/bf01774524
sha: 82ca37a96a762c097861f14dcaf780d9bd32d88e
doc_id: 5039
cord_uid: dtobju6z

nan

To perform the cell culture process on a large scale, we have to grow enough cells to inoculate our 2500-1 bioreactors. With mammalian cell culture, generating sufficient inoculum is a relatively slow process; we have to grow the cells slowly, because they are very fastidious regarding cell density in a culture (Fig. 2) . During inoculum development (Fig.3 ) we use spinner flasks; we start at approximately 250ml or less and go up ultimately to 8-1 spinner flasks.

In the previous paper our bioreactors are described. Figure4 shows our cell-culture area. Three 2500-1 bioreactors are in the factor VIII producing suite, in addition there are harvest tanks. We use a batch re-feed process (Fig.5) , in which we start with a bioreactor with a full volume of medium; we remove most of the bioreactor volume, which is then used for further purification. The cell culture remaining in the bioreactor is essentially the inoculum for the next medium feeding. This process continues for approximately 15 cycles, a typical run. From one inoculum we normally generate about 15 separate large-scale bioreactor batches.

We typically combine three bioreactor batches, ending up with approximately 70001 of conditioned medium. The medium at this point contains the C H O cells, which of course we no longer want, but we do want the product. We would like to remove the other S16 Fig. 3 Inoculum preparation Fig. 4 Cell-culture area, with 2500-1 bioreactors and harvest tanks components from the system, i.e., rAHF, rvWF, medium components, and CHO (host cell) residues, so that we have a pure product. From the 70001 our goal is to come up with just 11 of a purified, concentrated, safe, and effective drug substance.

We would like to leave factor VIII as integral as possible; we would like to remove all of the residues so that we have a very safe product; and we would like to do this at a cost of goods that is acceptable to the patients. To achieve these goals, we have developed a purification process as shown in Fig. 6 . The first step is filtration, in which we remove the cells by physical means with a a filter that is small enough to allow factor VIII to pass through. The next step is immunoaffinity chromatography, using the same monoclonal antibody that is used for purification of the plasma-derived factor VIII preparation Hemofil M. The basis for the separation is reactivity: the monoclonal antibody identifies a specific region, a specific epitope, on the factor VIII molecule, and achieves a very high degree of purification. Most of the residues flow through the immunoaffinity column (Fig.7) . Factor VIII is then eluted off. Two different ion-exchange steps follow. By changing the p H o f the column load, we change the charge on factor VIII. The residues, having a different charge, pass through the column. The most important advantage of recombinant factor VIII is the absence of viral contaminants. Clearly, viral safety is a very important aspect of factor VIII. Of course viruses that are not present in the human source should not be introduced into recombinant factor VIII. We believe that we have eliminated problems that have occurred in the past as a result of human-source and raw materials, but we do not want to add any more problems to the situation. We therefore have worked out a five-point viral safety program.

The first point of this program is the choice of raw materials. We use several animal-sourced raw materials (but no serum, as already mentioned in the previous paper), and we have sourced our animal-based raw materials from certified herds. These three animal-sourced raw materials are subjected to solvent treatment steps and subsequent chromatography separation and are quite pure. It should again be emphasized that we do not use serum, so the general "gemisch" that one would get from serum, from fetal calf serum or whatever, is not present. All of our animal sources are highly purified and are derived only from cattle in Canada or the USA. One example of these animal-based raw materials is bovine insulin, which is exposed to low p H a n d high ethanol concentration and is purified via precipitation, gel filtration, etc (Fig. 8) . The insulin is also tested for typical viruses. All other animal-sourced raw materials are also purified extensively, further assuring safety.

The next point in this program is the choice of cell line. Very few human viruses actually infect Chinese hamster ovary cells. Table 1 shows the results of a study by Dr. Lubiniecki and colleagues [1] . Human pathogenic viruses that were demonstrated to be nn were used to infect CHO cells. The viruses in fact were infectious to the reference cell, but most were not infectious to the CHO cells. Several viruses were infectious to the CHO cells, some in the parainfluenza line, mumps, and reo viruses. We used these virus lines as a base for our process validation studies.

We also make sure that the cells, both before and after cultivation, are free of any endogenous or adventitious viruses (Fig. 9 ) by means of a very extensive series of tests for microbial contaminants and various viral contaminants. With regard to retroviral particles, we determine whether the particles are infectious or defective, either before or after the campaign.

The fourth point in our program is "aseptic p r a ctices". The bioreactors are run in a very aseptic manner (Fig. 10) . Therefore, human employees, who really are the worst source of viruses, are not able to infect the product or the CHO cells.

The fifth and probably most important point is "process validation". Even if we were able to get a virus into this system, our purification process has a very good capacity to remove it. This has been proven by validating the processes: we have chosen viruses from the various classes that could potentially infect the CHO system (Fig. 11) ; we scaled down the purification process to laboratory scale, we added these viruses to the scaled-down purification process, and we analyzed the removal of the virus through the system -just in case. We demonstrated a decrease of viruses of 7-13 logs. Therefore, we feel comfortable in claiming that we have a process that would not allow viruses in, would test against viruses, and, in the extremely rare chance that a virus actually infected the cells, would remove it. It can be assumed that it is a virus-flee product. Figure 12 shows the monoclonal antibody columns, two large columns acting as a single column in the cold room. The volumes get much smaller as you go through the process. In the end there is 11 of drug-substance (Fig. 13) , sufficient for thousands of patients, stored in a freezer at our site until it is fully tested and released for further processing. Figure 14 is an overview of our testing program. The consistency of the process is demonstrated by quite a bit of testing. We test the working cell bank mentioned above; we assay the inoculum and the conditioned medium for microbial purity and endotoxin levels; we calculate the productivity of the bioreactor; we do postproduction cell-line testing; we make certain that the chromatography buffers are contaminant free; we ana- lyze the eluates from the columns to make certain that the proper amount of factor VIII is in the system, and that the columns perform properly as in-process controls. Before we release the product to Hyland for fur-ther processing we have to go through a very rigorous set of release tests as well. Finally, the utilities in our production process are thoroughly analyzed to make certain that we are not adding any microbial contami-S19 9 Microbial contaminants -Mycoplasma by both direct and indirect tests -Sterility for bacteria and fungi 9 Adventitious viral contaminants -In vitro assay using 6 cell lines (human, bovine, primate and hamster) 9 Retroviral contaminants -Transmission EM -Infectivity tets-S +L-for xenotropic viruses -Reverse transcriptase assay Fig. 9 Puritytesting of cells prior to and after cultivation Fig. 13 End-product -11, sufficient for 2000 patients

Model nants or viruses to the general environment. When we finish our testing program, Baxter begins. They test our incoming product to make certain it is acceptable; they have their own sterility and endotoxin program for the final filled product. The F D A tests every lot of product before its release to the US market. Recombinate is an extremely well-tested product by the time it is received by the physician or the patient. As a matter of fact, Genetics Institute alone runs over 400 laboratory tests on every lot of factor V I I I produced. Figure 15 shows a gel electrophoresis of Recombinate concentrate before and Fig. 16 after thrombin digestion. The outside lanes are markers. The lanes show several lots of the recombinant factor V I I I produced between 1986 (left lane) and today (right lane). It can be seen that the various protein bands show a very good consistency from one lot to the other, which is really the goal of a biological production process. A similar situation can be seen when the product is digested with thrombin and stained. One batch looks virtually identical to the next. Therefore, we feel comfortable in claiming that it is a controlled and consistent produc- tion process, that the product is virally safe, that it is extensively tested, and we are producing a high-quality, safe, and efficacious product.

A multifaceted approach to assure that recombinant T-PA is free of adventitious virus