The Pill That Could

Chapter 7: Approval

On a Thursday afternoon in late February 2001, in offices in East Hanover, New Jersey, a team at Novartis Pharmaceuticals Corporation finished assembling the boxes.

There were dozens of them. Each box contained a slab of paper several inches thick, bound in the kind of three-ring binder that pharmaceutical regulatory affairs departments keep in industrial quantities. The boxes contained, between them, the New Drug Application for STI-571 — the formal request to the United States Food and Drug Administration for permission to sell the drug as a treatment for chronic myeloid leukemia.

The application ran to roughly seventy thousand pages. This is, by the standards of New Drug Applications, neither unusually small nor unusually large. The pages contained the chemistry of the molecule, the manufacturing process, the analytical methods used to verify the purity and potency of each batch, the toxicology results from rats and dogs, the pharmacokinetic data showing how the drug was absorbed and distributed and metabolized in animals and humans, the full statistical analysis of three pivotal phase 2 trials, the clinical case reports for every serious adverse event observed in any patient on the drug, and the proposed labeling — the document that would become the official prescribing information if the drug was approved. There were also, in supplementary appendices, summaries of the phase 1 data, the preclinical pharmacology, and the rationale for the proposed indication.

The application was submitted on a rolling basis through February 2001, with the final submission completed by the twenty-seventh of the month. This was an unusual procedure. Most NDAs are submitted as a single complete package. Rolling submission — in which sections of the application are sent to the FDA as they are completed, rather than waiting until everything is finished — is reserved for drugs the agency has already agreed to evaluate quickly, on the grounds that the indication addresses a serious unmet medical need.

The FDA had agreed. STI-571 had received both fast track designation and priority review status, the latter shortening the agency's stated review clock from the standard ten months to six. Novartis was hoping for faster than that, and so was the FDA, and so were the people the drug was for.

On May 10, 2001, the FDA approved the drug.

Two and a half months had passed between the completion of the application and the approval. This was, at the time, the fastest review the FDA had ever conducted of a cancer drug. It would remain, for some years, near the top of the all-time list.

* * *

It is worth saying something about how drug approval normally works in the United States, because the speed of the STI-571 review is only striking if you know what speeds are usual.

Drug approval in the United States is governed by the Food, Drug, and Cosmetic Act, the framework for which dates to 1938 and the substantive elements of which date to 1962. The 1962 amendments — passed in the wake of the thalidomide tragedy in Europe, which had not, by accident more than design, reached the American market because of the skepticism of an FDA reviewer named Frances Kelsey — require that any new drug, before it can be marketed, demonstrate both safety and efficacy. Safety is the older requirement, going back to the original 1938 Act. Efficacy was the new one. The combination has, since 1962, defined what a pharmaceutical company has to prove to bring a drug to market in the United States, and the system that has grown up around the requirement is, by some measures, the most thorough drug-evaluation regime in the world.

The standard path runs in stages. After a drug candidate has been characterized in cells and animals, the company files an Investigational New Drug application — an IND — which gives the FDA the chance to weigh in before any humans are exposed. If the IND clears, phase 1 trials in humans begin: small studies, usually in patients with advanced disease, designed mostly to identify dose-limiting toxicity. Phase 2 follows: larger studies, typically in the range of dozens to a few hundred patients, designed to demonstrate that the drug has biological activity in the disease of interest. Phase 3 follows: large randomized controlled trials, typically several hundred to several thousand patients, comparing the new drug against either placebo or the existing standard of care, designed to demonstrate that the drug actually improves outcomes that matter — survival, time to progression, freedom from symptoms, quality of life. Only after phase 3 does the company file the New Drug Application that asks the FDA to allow marketing.

The whole process, end to end, has historically taken something like ten to fifteen years from the first lab work to the first prescription. The clinical portion alone, from the start of phase 1 to the filing of the NDA, has averaged six to eight years. Once the NDA is filed, the FDA's standard review clock is ten months for a normal application, six months for a priority application. The clock is not the entire story — the agency can stop the clock to ask for additional information, and for most applications the clock does in fact stop and start — but it is the framework within which the review happens.

Inside this framework, two and a half months from filing to approval is essentially the floor. It happens only when the data are unambiguous and the disease is serious and the agency has been working closely with the sponsor through the development program.

All 13 chapters of The Pill That Could · Library · All courses