Skip to Main Content

EL SEGUNDO, Calif. — CAR-T therapy is almost ready for prime time.

Both Kite Pharma and Novartis plan to file for approval to bring their CAR-T cancer therapies to market in the first quarter of next year. So we took a peek at Kite’s new CAR-T plant here in this industrial city south of the Los Angeles airport to see what they’ve got in the works.


First, a quick refresher: CAR-T is the ultimate precision medicine. It involves taking blood out of a cancer patient’s arm, reengineering the immune cells, then feeding them back into the patient to attack the tumors. That means the therapy is manufactured in very small batches. A CAR-T factory takes up less space (and is orders of magnitude less expensive to get up and running) than a typical plant which churns out big batches of biologic drugs. But it requires an exacting attention to detail.

“When we went to design the manufacturing plant, we had no point of reference — because nobody had done it before,” said Dr. Arie Belldegrun, Kite’s CEO.

Though small, the $30 million factory has style: It’s retrofitted into an industrial building designed by renowned architect Frank Gehry — otherwise known for building edifices like the Guggenheim museum in Spain or the Walt Disney Concert Hall in downtown L.A.


No makeup, no hairspray, no street clothes

The nuances of CAR-T production differ from lab to lab, and company to company. But each follows this general outline:

Once the plasma’s been collected from a patient, Kite scientists place the T cells in media and activate them. From there, they use genetically modified viruses to insert a gene that can make the cells express a receptor that targets an antigen  — that is, a protein expressed on the surface of cancer cells. Basically, the cells are programmed to home in on cancer.

From there, the cells are left in incubators to replicate, in the presence of a number of different growth factors. It takes Kite about a week to grow the T cells, and then another several days for safety testing and quality control. The cells are flash-frozen and packaged.

Because sterility is of utmost importance, workers must strip to their skivvies and don a uniform of blue scrubs, shoe covers, and a bouffant cap before even entering the antechamber outside the CAR-T factory floor. No makeup or hairspray is permitted — any particulate matter might contaminate the cell samples.

The antechamber encircles a smaller set of rooms where the real CAR-T magic takes place. To get in here, workers must wear an all-white gown, mask, and cap, and are double-gloved to handle the samples.

The place, in short, is spotless.

It’s a Henry Ford dream, but miniaturized: Lines of machines that process, incubate, and preserve the CAR-Ts occupy different sections of the plant. In many cases, a workstation handles just a single patient’s cells.

Supply runners, clad in blue, wait in the antechamber to ferry any materials that the workers in white may need. Every disposable cartridge, chemical, or container is in an individual bag, marked with the patient’s name, to try to prevent any cross-contamination.

“The key thing for personalized medicine is really the chain of custody — because when you’re producing drugs for one patient at a time, there just can’t be any mix-ups,” said Tim Sirichoke, vice president of manufacturing at Kite.

The CAR-T production line at Kite's new factory in El Segundo, Calif.
The CAR-T production line at Kite’s new factory in El Segundo, Calif. Kite Pharma

A race to winnow ‘arm to arm’ time

Kite is aiming for an “arm to arm” time of a little over two weeks. (The phrase refers to the time between when the patient’s blood is first drawn and when he receives his first infusion of engineered cells.)

“These patients have very aggressive lymphomas,” said Dr. David Chang, Kite’s chief medical officer. “If they don’t get the product in a short period of time, they lose the window of opportunity to treat these patients.”

Juno Therapeutics, which is also working on CAR-T, is aiming eventually for even faster turnaround — perhaps just two days, CEO Hans Bishop told STAT. He said the company hopes to start with whole blood from the cancer patient, rather than separating out the plasma, as Kite and other companies routinely do.

“We’re working on it, and think it’s a feasible challenge,” Bishop said.

Novartis, for its part, has been manufacturing all of its CAR-T cells in Morris Plains, N.J. — at plant once owned by early cell therapy player Dendreon. Dr. David Lebwohl, who leads the CAR-T work at Novartis, said the minimum “arm to arm” time will be about three weeks — with about nine days of cell growth, and another nine for quality assurance.

Since CAR-T therapy is in its early stages, researchers are still experimenting with all sorts of variables. Most crucial, of course, is balancing efficacy and toxicity. But “arm to arm” time matters, too — cells that spend longer in an incubator may yield a more effective therapy, but the longer the grow time, the more costly it can be for the company.

No inventory, and no room for error

Optimizing the modular nature of a CAR-T plant has been critical for Kite — because everything is made in-house. They do buy tools, reagents, and cellular media from vendors, but not much more.

“The supply chain with cell therapy is totally different than with traditional pharmaceuticals,” said Tim Moore, executive vice president of technical operations at Kite. “Traditional pharma has many handoffs, and you aren’t that close to the patient.”

The reality of bespoke therapy is that there’s no inventory: no prolonged storage, no need to create large batches of media or medicine. But for patients who need a CAR-T therapy, it’s absolutely critical to get the science right — every single time.

“We don’t have any inventory, which helps on a financial level,” Sirichoke said. “But one of our biggest challenges: There’s no room for error for us.”

Of course, errors can occur. For example, sometimes a patient’s T cells are taxed from all the treatments she’s tried, and don’t proliferate in a lab setting as they ought to, Chang said.

Kite’s El Segundo plant can make 4,000 to 5,000 CAR-T therapies each year — a figure that the company wagers should meet patient demand, if its first drug gets approval from the Food and Drug Administration to treat patients with aggressive non-Hodgkin lymphoma.

The plant is expected to go online late next year — though Belldegrun said it’s ready to start production immediately.

  • Does anyone else think it odd that an article on CAR-T never actually defines “CAR-T”? I am sure this is common knowledge to a practicing oncologist but, despite a PhD in molecular biology (Univ Colorado, 1994) and having worked on informatics for TCGA project at NCI, I don’t know what it stands for. Acronym Finder is no help. I finally got the answer (Chimeric Antigen Receptor, T-cell) from
    Giving a parenthetical definition of “CAR” at the beginning of the article would make it much more accessible and save readers the trouble of looking it up. Or, to put it in terms the author’s editor would understand, that also means that readers spend more time on _your_ web page, rather than risk losing them when they search for clarification.

    • Agree, some of the STAT articles are all over the place without much introduction at all for non-science or non-specialist readers. More logical flow to the writing style would help.

  • It seems rather rude to make us click on a story that is interesting only to discover that the story is not available without a subscription. Don’t you think it would be courteous to mark StatPlus articles as such in your emails? Of course, if you are trying to drive away those of us who won’t pay $300/year for a subscription to something we have never seen an article from, you are doing a great job at that. I wonder how many of the editors at Stat have bought cars, homes, etc., without ever seeing them. Even the Boston Globe allows you to see the quality of their articles online before Mr. Henry makes you pay for a subscription.

  • Your UX does not permit a subscription or permissions to read your articles. Why? Are there no articles? Really bad dev and programming. Or no substance? Please let me know.

Comments are closed.