Making it to Manufacturing

The potential success of broadly neutralizing monoclonal antibodies for HIV prevention, treatment, and possibly even a cure could come at a cost

By Michael Dumiak

As researchers scrutinize the scores of antibodies isolated recently that can neutralize a wide variety of HIV strains, others are thinking about the potential price of success for using these antibodies as a possible means of preventing or treating HIV. Making these antibodies in mass quantities, at going rates, will not be cheap.

There are currently about two dozen bioproduction facilities that are capable of producing monoclonal broadly neutralizing antibodies (bNAbs) in bulk, commercial-sized batches. Whether the metric-ton sized lots can be made for the millions who would benefit from a successful HIV antibody-based therapy or preventive isn’t really the question. The difficulty won’t, in all likelihood, be with capacity or technology. It will be cost. Innovation in protein production methods and new technologies may be able to bring potential prices down. Given this rather serious constraint in antibody manufacturing, right now the lab seems to have a little more momentum than the factory.

However, the Bill & Melinda Gates Foundation and a few other institutions are now analyzing the costs, timelines, and best ways to potentially manufacture these monoclonal antibodies. Should these antibodies be proven successful in treating or preventing HIV infection, perhaps then manufacturing won’t be such an obstacle.

Genetic broth in steel tanks

In 1986, the same year the International Committee on Taxonomy of Viruses ruled that human immunodeficiency virus should be the term for the etiological agent of AIDS, the US Food and Drug Administration approved Muromonab-CD3 (OKT3) as a drug for human use. Marketed by Janssen-Cilag and used until 2010 to fight rejection of transplanted organs, the immunosuppressant OKT3 became the first licensed monoclonal antibody.

The researchers who first created Muromonab-CD3 used genetic material cloned in hybrid cell lines—hybridomas—fused from mouse spleen B cells and myeloma, or cancer cells. The fused hybridoma divides perpetually, which is a property of the myeloma cell, and it produces antibodies, which is something the B cell does. Later OKT3 was produced in the ascites (abdominal fluid) in the peritoneal cavity of mice. Hermann Katinger, an Austrian microbiologist and founder of biopharmaceutical manufacturer Polymun Scientific, recalls that in the early days of antibody production and using animal cell technology that hybridomas, and specifically human-mouse hybridomas, were quite difficult to work with for mass cultures.

The switch to using cells lines from Chinese hamster ovaries (CHO) for cloning and expression of specific heavy and light chain antibody genes by the late 1980s—heavy chains and light chains being types of amino acid sequences that make up the basic units of a protein—solved some of these problems. Katinger’s group went on to make some of the first broadly neutralizing monoclonal antibodies against HIV, including 2G12, 2F5, and 4E10.

From that point on, bulk production of antibodies in bioreactors became more and more standardized. Brian Kelley, now vice president of bioprocess development at Genentech, wrote in 2009 that the move over to CHO as an early stage incubator for bulk protein production allowed producers to take advantage of common technologies already used for recombinant products (mAbs 5, 443, 2009). Over time, using CHO for this purpose also gained the confidence of the drug regulators enforcing good manufacturing practices (GMP), and of the manufacturers themselves, such as Lonza and Boehringer Ingelheim, who built large production plants based on this method.

Looking at antibody production in bulk now, sources say, is something akin to air travel: everything’s changed and nothing’s changed. A whale of an Airbus can take more than 500 people at a time from Singapore to London, but it still relies on velocity and lift.

Contemporary industrial production of monoclonal antibodies—once the arduous early upstream work of isolating the genetic material, cultivating it, and placing it into cell lines is completed—would be broadly recognizable to counterparts doing this work in the early 90s. The cell lines expressing the antibodies are placed into large stainless steel bioreactor tanks, ranging anywhere from 1,000-liter to 20,000-liter capacity for very large batches. This cellular ‘broth’ is then purified, generally using a licensed medium or resin containing Protein A, which binds immunoglobulins and can fish out antibodies from a 15,000-liter harvest to 95% purity in a single step, yielding titers of one to five grams of protein per liter. Further purification comes from filtering, straining, and column chromatography, leading to the end product, which is then put into vials or syringe format for administering to a patient.

During the late ’90s and early aughts, antibody production capacities were barely able to keep pace with demand and analysts predicted manufacturing shortfalls. That worry was unfounded. As bioreactor production capacity increased, so did expression levels and cell densities, with harvest titers increasing as a result. Kelley outlines in his 2009 mAbs paper what a ‘model’ plant would be: it would run six 15,000-liter bioreactors, processing titers of five grams per liter with no purification limitations. Such a plant, running at full capacity, could produce 10 tons of monoclonal antibody a year.

New research, new possibilities

Meanwhile, antibody research in the HIV field is going through a remarkable renaissance, spurred by the 2009 discovery of the bNAbs PG9 and PG16 (Science 326, 285, 2009). Researchers went on to isolate scores of other antibodies (Nature 447, 466, 2011). These potent proteins have unusual shapes and characteristics that make them particularly good at neutralizing many strains of this notoriously difficult to tame virus.

Researchers are pursuing a number of different directions in learning how best to use the bNAbs against HIV. “The research discovery effort was originally aimed at isolating neutralizing antibodies so we can understand how they bind to HIV, to solve crystal structures, and to inform how to design vaccine initiatives,” says John Mascola, director of the National Institute of Allergy and Infectious Disease’s Vaccine Research Center (VRC). This is still a pursuit, though it’s a challenging one.

But when scientists observed the potency of bNAbs, they also started thinking about how to use them in other ways, including clinically in what’s called passive transfer or passive administration, when the antibody is directly injected in an effort to prevent HIV transmission. This could be particularly useful in preventing mother-to-child HIV transmission by administering the bNAbs to pregnant or breastfeeding women (PLOS Med 2014, doi:10.1371/journal.pmed.1001616; see Brief, this issue). Antibody-based prophylaxis could also be a replacement for antiretroviral (ARV)-based pre-exposure prophylaxis (PrEP)—using ARVs to prevent HIV infection. Adherence to daily PrEP drugs has shown in some cases to be inconsistent. Replacing the need for daily pills with a monthly or quarterly injection of bNAbs may in certain settings be a more successful alternative. Studies in non-human primates also suggest that administering the HIV bNAbs along with ARV therapy could potentially lead to a functional cure (PNAS 10, 1073, 2013).

Passive administration of bNAbs is already being tested in the clinic. These antibodies are manufactured at the VRC’s pilot plant, located on the outskirts of Frederick, Maryland. This plant can make vaccines and antibody products under GMP conditions and has small and medium-scale capacities to produce monoclonal antibodies: enough for Phase I and II trials, but not enough for widespread commercial use. This plant is supplying the bNAb VRC01, isolated by researchers at the VRC, for the institute’s two ongoing Phase I clinical trials (VRC 601 and VRC 602) testing the safety and pharmacokinetics of the antibody when administered directly to humans, and can also supply subsequent Phase I and II studies, Mascola says.

The pilot plant uses stable CHO expression lines and stainless steel bioreactors, presumably for less than the US$8 million or so it would have cost, according to a presentation delivered by the VRC’s Vaccine Production Program Chief Richard Schwartz at a 2012 advisory committee meeting, to hire a contract manufacturing organization to produce enough VRC01 monoclonal antibody for a partial Phase I study.

VRC 601 and VRC 602, the first results of which are due in a matter of months, involve groups of 15 to 25 people—one a group of healthy volunteers, the other a group of HIV-infected volunteers. “Both are pretty standard dose escalation studies, starting at a low dose and going up to a standard dose,” Mascola says. In this case, the doses range from one milligram of antibody per kilo of body weight to 40 milligrams per kilo. If an average person weighs 75 kilos, and a standard dose of 20 mg per kilogram of antibody is used, the necessary dose of antibody could be as much as 1,500 milligrams a person. That’s just for a small Phase I trial. Mascola says researchers are already planning ahead for larger and more ambitious trials for HIV therapy, and possibly for prevention, and he expects these could get started as soon as next year.

“One approach would be to treat patients who are on successful antiretroviral therapy and to look with sophisticated assays to see if there’s any beneficial effect in adding antibodies, for example on the viral reservoir or the amount of cells infected,” he says. Another study might determine whether patients who are on ART and doing well—but who need daily treatments—might be able to take a break from the drugs, replacing them with antibody treatments delivered once a month or every two months for a period of several months. This might limit drug toxicity and limit the burden of treatment, at least for a time.

Should any of these approaches pan out, HIV antibody-based therapy would be joining a half dozen or so of existing monoclonal antibodies used to treat a variety of conditions, including cancer and arthritis. All existing monoclonal antibody therapies are backed by big pharma: Johnson & Johnson, Biogen, Roche, Genentech, Pfizer, Bristol-Meyers Squibb, Eli Lilly, and Merck. They are all manufactured using cell-line expression methods that would be similarly employed, broadly speaking, in making HIV bNAbs in bulk batches. These products show that industrial-scale antibody manufacturing is certainly possible, but the question is at what cost.

The cost of grams

Genentech’s Kelley says that production costs for bulk monocolonal antibodies have come down since the late ’90s from about $300 a gram to $100 gram—and could possibly be lowered to $20 a gram, using his model plant. But Steve Hadley, senior program officer for vaccine development at the Bill & Melinda Gates Foundation, says that if bNAbs are to reach their potential against HIV—as a general prophylaxis, as protection against mother-to-child transmission, or as a therapy taken in connection with ART—the price needs to come down to something more like $3 a gram to be feasible for the poorest places in the world where it is most needed. Right now there is no strategy to do this. So Hadley is in the early stages of figuring out what is needed to get bNAbs into the clinic at low cost and what the fully mature costs would be, particularly for passive immunization therapies for HIV.

“Our target profile is a quarterly subcutaneous injection,” he says. “You come into the clinic, you get a subcutaneous injection of antibody, and there’s nothing you need to do other than come back in three months and get another injection. There would be better compliance, as long as people came back in three months to get another injection. And the side effect profile would be much different, hopefully reduced to none.”

He projects production using current manufacturing technologies and steel tank bioreactors at 15,000-20,000 liter scale to make HIV bNAbs will cost $30-$50 per gram at best. “Looking at the large number of people you’d have to dose to have a meaningful impact, we really are a long way from where we need to be from a cost perspective,” he says.

While vaccine researchers are busy trying to identify immunogens that could get the immune system to do the difficult work of making these bNAbs, developing the ability to manufacture these human proteins in large scale could be required should antibodies be proven effective in HIV prevention or treatment. “The reason we need the large volumes is that when you start running the numbers, and figuring the actual metric tonnage of antibody you’d need to passively administer these antibodies in Sub-Saharan Africa…” Hadley says, trailing off. “You start looking at one to five to ten metric tons of antibody per year.”

Appointing Hadley to the case is a sign the Gates Foundation, for one, is taking antibody manufacturing seriously. Hadley’s knowledge of the industry and his background in recombinant protein development and production prompted the Foundation to bring him on last year. His team is considering how to bring together investigators and product development partners, Hadley says, to think through how the Foundation should be in investing in bioprocess research to support antibody development.

Improving manufacturing for bNAbs happens in a limited number of ways. One option is to increase the potency and half-life of the protein and thereby reduce the amount of antibody needed. Hadley argues that so far there has not been a consolidated effort to engineer the antibodies to improve their potencies in order to achieve a lower required dose. The other option is to make a better production line. While the methods haven’t fundamentally changed for manufacturing using steel bioreactors, the equipment and materials have improved.

But that improvement doesn’t necessarily solve the issue, either. Momentum and innovation in bioprocessing seems to be on the side of small- and medium-sized batch makers, and not in the big industrial-capacity production needed in order to consider an operation on the scale of introducing passive transfer of bNAbs through Sub-Saharan Africa.

Smaller batches mean producers have to use less Protein A to purify their antibodies. “That’s a very expensive processing step,” Hadley says. It can cost more than $10,000 a liter and it remains under license, at least in its modified form. There are other options for purification of antibodies, including updated precipitation or recrystallization technologies. While neither method would improve yields or is as powerful as using current protein A purification, they would cut costs.

Replacing CHO with a more optimal cell type is also an option, but the deep comfort regulators (and manufacturers) maintain with the hamster line, and the worry about contamination from bacteria, mycoplasma, or viruses when taking on new lines, is a big hurdle to establishing a new method.

Right now Lonza and Boehringer-Ingelheim continue to rule the roost in terms of large-scale bulk manufacturing of antibodies, but there are a number of small- to mid-size producers—Gallus, KBI Biopharma, Fuji Film Diosynth, CMC Biologics, Catalent, and Rentschler Biotechnologie—that are active in developing and incorporating new and recent upstream processing methods.

These producers are moving away from stainless steel toward single-use reactors that use disposable bags to perform the cell culture, saving on cleanup costs. Other recent developments include using a continuous bioprocessing or perfusion process, which uses a smaller bioreactor, under 1,000 liters or so, and allows for the continuous harvesting of protein. Particularly promising are the possibilities for doing this using simulated moving bed chromatography along with the continuous flow reactors. Simulated moving beds utilize an instrument—from Tarpon, or Novasep, for example—which highly regulates the movement of a column series of the automated feed and exit valves. This technique, also described as multi-column continuous chromatography, switches the flowpath and tightly orchestrates the stream of purified product harvested from the reactor.

“It means you utilize a much smaller quantity of resins, but at the maximum capacity,” says Stefan Schmidt, a vice president responsible for downstream production at Rentschler Biotechnologie near Ulm, Germany. “And this saves you a bit of material costs.”

Whether these alternatives and innovations could ever make a real difference in bringing down costs for industrial production of HIV bNAbs, however, remains an open question.

Beyond bioreactors

Other potential ways to get around the costs associated with industrial-scale manufacturing of bNAbs might be to use plant cell lines to express the protein—or to use a viral vector to introduce the antibody gene and have the body itself take over production.

Yvonne Rosenberg, an Australian-born immunologist and chief executive of PlantVax, in Rockville, Maryland, can produce hundreds of milligrams per kilo expression of VRC01 using transient transfection in tobacco leaves. Plant-cell protein production has a relatively long history, but Rosenberg says she thinks it is finally picking up steam. June marked the first meeting in Berlin of a new group called the International Society for Plant Molecular Farming, drawing more than 100 participants. The Pharma-Planta project and its European partners completed Phase I human trials using 2G12. The trial proved that a monoclonal antibody produced and isolated in tobacco plants could be safe for use in humans. A group at St. George’s University in London, citing the high bulk costs of industrial antibody production, are expressing VRC01 in tobacco plants in an ongoing effort to show it can be done effectively that way (Plant Biotechnol. J. doi: 10.1111/pbi.12137).

“If you make more than a ton a year, I don’t think anything beats CHO right now,” Rosenberg concedes, citing the purity of product, the efficiency of the process, and the comfort of regulators. “But plants have a big advantage because they are adaptable and you can make new antibodies and test them quickly.”

Others wonder about its feasibility. Currently, production plants need to be kept in greenhouses in order to control the environment and this affects cost. Supporters counter that the upfront and production costs for plants can be one-tenth the costs of that for CHO production using steel tanks, even though the downstream costs are higher. Greenhouses are becoming more sophisticated and plants grown in these conditions are more uniform and productive.

Another kind of delivery system is also being pursued by Philip Johnson, chief scientific officer and executive vice president at the Children’s Hospital of Philadelphia. He and his colleagues are testing the idea of inserting the PG9 antibody gene in an adeno-associated virus vector and injecting this into humans. “The idea is that we make a vector that contains the gene that represents the antibody. The gene then gets into the cell—in our application, into muscle cells—the muscle cell then makes the antibody that gets into circulation and continues to make it,” Johnson says.

Bioreactors are still in the picture here, but in this case they are used to make the vector instead of a protein. By only needing one administration per patient, successful vector delivery would brighten the prospect of endless manufacturing costs considerably. Johnson is working with the International AIDS Vaccine Initiative and the University of Surrey in the UK on a Phase I clinical trial investigating whether healthy volunteers injected with the PG9-encoded vector will then produce the antibody and at what levels. The California Institute of Technology’s David Baltimore and his team are trying a similar approach using different antibodies and a different vector. So-called gene therapy, however, remains shadowed by safety concerns.

For every research question answered, basic production questions come closer. The Gates Foundation’s Hadley responds by posing a simple equation. “Our interest is global access,” he says. Hadley says the intent is to create partnerships with large pharmaceutical or biotechnology companies that can, and will, do commercial-scale antibody development. “We’d be thrilled if we get Phase I data that says this bNAb looks great, and a large company comes along and says, ‘We’ll take it from here.’”

Michael Dumiak reports on global science, technology, and public health and is based in Berlin.