Decoding Health Care

Part 10: The FDA

The FDA was created by the Food and Drugs Act of 1906 in order to prevent the “manufacture, sale, or transportation of adulterated or misbranded or poisonous or deleterious food, drugs, medications, and liquors.” For most of the next 50 years, the scope of FDA authority was defined (beyond agriculture) and there was a nod towards documenting the safety of both pharmaceutical as well as OTC products. One important piece of legislation during this period was the Food, Drug, and Cosmetic Act of 1938. Sulfanilamide was an exciting antibiotic in the mid 1930’s used to treat Strep infections. There was a call for an oral liquid formulation, and the manufacturer unwisely chose diethylene glycol as a solvent. This resulted in over 100 deaths in the US.  The 1938 legislation brought cosmetics and medical devices under the agency’s control, required that drugs be labeled with adequate directions for use, and mandated the pre-market approval of all new drugs.   

But the modern FDA came about as a result of the thalidomide fiasco. Thalidomide was developed as a sedative and anxiolytic specifically for pregnant women, as well as a remedy for morning sickness. It was extensively marketed in Europe and would have been marketed in the US if not for the efforts of Dr. Frances Kelsey, an FDA medical officer who was troubled by the lack of safety data and early reports of the teratogenic effects of the agent. As it turns out, the FDA never approved thalidomide. Dr. Kelsey saved the US from the devastating teratogenic effects of the drug. The photos of children affected by phocomelia, the limb malformations caused by in utero exposure, are heartbreaking. As a consequence, the Kefauver-Harris Drug Amendments passed in 1962 established the current FDA requirement that new drugs be safe and effective.

The FDA is a huge agency, and it touches every American every single day. It has authority over food safety, tobacco, caffeine, dietary supplements, prescription and OTC pharmaceuticals, vaccines, therapeutic biological agents, blood transfusions, medical devices, cosmetics, animal foods and veterinary products. It has 18,000 employees and is under the umbrella of the Dept of Health and Human Services. Most of those employees aren’t political employees, but the head of the agency is. This is important because just like with CMS, the FDA shouldn’t be political but most certainly is political. In fact, when President Trump began his term in 2017, he floated some very unconventional names as candidates for commissioner. He ultimately landed on a more traditional candidate, Scott Gottlieb, who proved to be a competent and effective head of the agency. But the FDA often reflects the political leanings of the party in office.

The FDA came under a lot of scrutiny during the Covid pandemic. Their performance was mixed. They have generally been criticized for their ineptitude in developing and disseminating a reliable Covid test. But they also performed admirably in their application of emergency use authority (EUA) to permit rapid deployment of Covid vaccines which have been generally acknowledged as the agents most responsible for turning the pandemic around.

For our purposes, however, the FDA is interesting because of its role in making cancer therapies available to Americans. The focus of the FDA on safe and effective has proven to be a somewhat laborious, expensive, and slow process. The traditional FDA approval process can take seven years or so. Over the course of my career, patients with cancer, AIDS, and now Alzheimer’s disease have argued passionately that it takes too long to get promising new therapies to desperate patients. The FDA response has been to develop several mechanisms for timely approval: fast track, accelerated approval, priority review, and breakthrough designation. All of these approval paths aim to address unmet medical needs, i.e. to provide a treatment option that is clearly superior to what currently exists. In order to fulfill this goal, the FDA allows studies to depend on ”surrogate” endpoints to prove the efficacy of a novel therapy.

Traditionally, cancer research trials had as their primary endpoint cancer survival: did the new therapy improve the survival of the patient with cancer?  But this can be a difficult endpoint. First of all, people can live with cancer for a long time so it can take a while to show a survival advantage. Second, when a patient can (and usually does) receive several therapies in sequence both before and after the experimental drug it can be difficult to attribute a survival benefit to the novel therapy. Third, to be certain that the survival benefit is due to the therapy a trial design called a randomized clinical trial is required. Under that design, a patient is randomly  assigned (in a blinded fashion, if possible) to the new therapy or to another therapy (or even better, a placebo). In some cases that design is impossible because the disease is too rare; in others, it is viewed as unethical because the new treatment is so much better than the old treatment; in others yet, either patients or physicians are unhappy with randomization. Finally, it can be reasonably argued that survival isn’t the only thing that matters. What if the treatment reduces symptoms?

In order to “accelerate” the trial process, the FDA agreed to accept other markers of clinical benefit that might reasonably predict a survival benefit down the line. The most common is PFS, progression free survival. A good way to think of PFS is how long the cancer is kept in check. Another marker of benefit is response rate. Did the therapy cause the tumor to shrink? Usually a response rate of 50% reduction in tumor size is required. Yet another surrogate might be an improvement in quality of life, though that surrogate is rarely invoked.

Using these surrogate endpoints allows studies to be completed much faster and cheaper, and by extension gets these novel therapies to patients much faster. And they have done exactly that. The vast majority of new cancer drugs are approved by an accelerated pathway. And the vast majority of drugs in the accelerated pathway program are cancer drugs, although recently Alzheimer drugs have come to the market via this route. Seems like the program is working the way it ought to. Not really.

The first problem with the accelerated approval process is that most of the surrogates used in the trials aren’t very good surrogates. By that I mean that they do not predict whether a drug will ultimately prolong or improve life. The second problem is that once a drug is FDA approved in many cases commercial health insurance, Medicare or Medicaid will be compelled to pay for it. And the price will be whatever the pharmaceutical company says it should be. Although there may not be an ABSOLUTE requirement for coverage, in truth those drugs are almost always covered. And for cancer drugs it is essentially absolute. The FDA is explicitly forbidden to consider cost in its deliberations regarding safety and efficacy. This is a problem.

There is an immediate solution. As part of FDA’s accelerated approval process, pharmaceutical sponsors are required to submit post marketing evidence that the approval was appropriate and that the preliminary results reported as surrogate endpoints did in fact translate into real benefit. Unfortunately, withdrawal of an FDA approval due to lack of efficacy data post approval has not happened nearly enough (https://ascopost.com/issues/february-10-2022/oncology-drugs-with-accelerated-approval-is-it-time-for-a-reset/). As a consequence, there are a number of treatments used for cancer that are of unknown real benefit. But the problem is actually much larger. When the FDA approves a therapy, it creates a “label” that summarizes the evidence supporting the approval as well as the known toxicities. This label defines how the drug can be marketed including marketing material sent to physicians.

But physicians can prescribe that FDA approved treatment in other situations where they think it might be helpful. For example, if a drug is approved in lung cancer it can also be used in breast cancer. The limiting factor in that case is whether the insurance company (or Medicare or Medicaid) agrees there is enough evidence to pay for it. The “appropriateness” is determined by an evidence review called a compendium. And the most widely used and respected compendium is compiled by the NCCN, the National Comprehensive Cancer Network. The NCCN has a well-established committee structure of academic physicians (and at least one patient advocate per committee), experts in the specific cancer in question, that weigh the evidence and render an opinion as to whether the treatment passes muster. This has become the primary way a new FDA approved therapy makes its way into common usage. There is good evidence that most oncology prescribing is “off-label”. Filing an FDA application for every new indication is really expensive and time consuming and frankly just not worth it for the pharmaceutical manufacturer.

Now you might say: So what? The response is that these treatments cost money and have side effects. Nobody wants to receive ineffective therapy. So with widespread off label prescribing and the lack of post marketing data to confirm or refute those preliminary results, we get locked into a futile cycle that is difficult to break. And we are about to enter an era in medicine with gene therapy where the stakes are going to be very high.

The reason that gene therapy is such an important topic is that gene therapy has the potential to actually cure patients. Maybe. These will undoubtedly be approved based on surrogate endpoints. We definitely won’t know whether the responses are durable. And the cost per patient will be in the millions of dollars. Post approval data is absolutely mandatory in the gene therapy space. But to move gene therapy forward we will need more than that.

Although the FDA doesn’t consider cost in the safe and effective paradigm, post approval data will be critical to judge what a fair price is. Are we actually paying for cures? The mechanism to accomplish this is outcomes-based reimbursement. In the world of outcomes-based reimbursement, the payer pays in installments based on performance. At various time points, clinical measures will determine whether payments should continue. For example, let’s take gene therapy for sickle cell disease.  There will be an initial payment for the treatment then installments over time if there is continued clinical benefit, i.e. no painful crises, freedom from transfusion, unmeasurable Hb S, for example. This very paradigm was proposed by the CEO of Bluebird several years ago. To be certain there are some details to work out. But I think it makes pretty good sense.

Let’s take a step back. Given all of the challenges with surrogate endpoints, why don’t we adopt a similar payment paradigm for all of these new therapies with big price tags approved via an accelerated pathway. We can determine a fair starting price at approval followed by a period of data collection (already the FDA’s responsibility) followed by the determination of whether we should keep the new drug and at what price. Parenthetically, this is a slight variation on how Germany manages drugs new to market.

I can promise you not everyone likes this approach. Maybe it’s too logical. But we have a BIG problem with accelerated approval and the cost of new therapies. The recent experience with Aduhelm is instructive. Alzheimer’s disease is a horrible way to die and there has been little progress in its treatment. The current leading hypothesis for its etiology is the accumulation of amyloid plaques in the brain. Aduhelm attacks these plaques. Clinical trials showed Aduhelm provided modest reduction in the rate of neurological decline, but at a cost of increased risk of brain bleeding and swelling. Despite intense controversy including profound disagreement of FDA drug advisory committee (in fact that advisory committee voted unanimously against approval), the FDA approved the drug in 2021. The initial price tag was $50 K per patient per year.

Under normal circumstances CMS would just go along. But CMS balked. They felt the evidence for efficacy and safety did not reach their reasonable and necessary threshold. CMS mandated coverage with evidence development and only agreed to cover the drug on a clinical trial. As you might expect, this was fatal to Aduhelm and it was subsequently removed from the market. But this therapeutic approach still has life. A second-generation plaque targeting therapy has been approved by the FDA.

What went wrong? The surrogate endpoint was weak. The clinical impact was small. And most importantly the FDA and CMS were not on the same page. Wouldn’t a collaborative approach with an outcomes-based contract been wiser? Instead, both the FDA and CMS were embarrassed and the entire accelerated pathway in Alzheimer’s disease was brought into question.

There is so much more to discuss with the FDA. We haven’t touched on tobacco regulation. In the Trump administration, Scott Gottlieb made a valiant effort to reduce cigarette consumption by substituting vaping as the nicotine delivery device. His rationale was that nicotine administration via vaping must be safer than inhaled cigarette toxins. Preliminary evidence suggests he may have been correct (https://www.nejm.org/doi/full/10.1056/NEJMe2314977). But he was undone by unscrupulous behavior by certain parties in the vape industry (specifically by launching an aggressive marketing campaign aimed at teens). Although cigarette smoking has decreased dramatically in the US it remains a public health hazard.

We also have not discussed the role of the FDA in medical device regulation. The FDA has broad authority in device regulation. Almost every device used in medicine is subject to review. But most aren’t. The approval process for devices is very different than that for drugs. Ninety percent of devices are either exempt or take the 510K clearance path.  In 510K clearance, the new device just needs to be similar to a device currently in use. That includes saline breast implants and a long list of devices used in the cardiac cath lab. To normal people this might sound nutty. It is.

And we haven’t discussed the role of the FDA in regulating laboratory tests (which has been historically almost non-existent). Imagine you need a very complicated cancer diagnostic test to guide your therapy; it is more likely than not that the performance of that test hasn’t been looked at by the FDA. Most lab test developers don’t really want to have the FDA look at their tests. It is time-consuming and expensive. There are several pieces of legislation now being debated that would strengthen the regulation of lab tests. This is long overdue.

 We also need to consider how the FDA is going to regulate artificial intelligence. Stay tuned. This is going to be a very big deal. It is likely to get extremely political because lots of money is involved. And it is going to be dealt with by the medical device division which already has a lot on its plate.

Finally, we have not discussed the potential conflicts related to the relationships between the regulated sectors and the FDA. This was spotlighted in the Aduhelm controversy where the manufacturer and the FDA seemed a little too chummy. A lot of pre-approval meetings happened behind closed doors and there have been questions asked about the “advice” the FDA offered to the pharmaceutical sponsor, ultimately leading to a congressional investigation (https://www.nytimes.com/2021/07/19/health/alzheimers-drug-aduhelm-fda.html).

There are other potential conflicts of interest. A big chunk of the FDA budget comes from manufacturers who pay fees to the FDA to get their products reviewed (so called PDUFA fees). And we must  acknowledge that the FDA is often a lucrative steppingstone to a plum job in pharma. For an egregious example, refer to the FDA approvals of MS Contin and Oxycontin as discussed in Patrick Radden Keefe’s excellent book, “Empire of Pain: The Secret History of the Sackler Dynasty”.

But I digress. America is better because of the FDA. But the FDA could do more for Americans     (for a list of suggestions see https://sites.google.com/georgetown.edu/pharmedout). The FDA needs to enforce post approval data submission and act promptly if confirmatory data is not confirmatory. We should think about how the FDA might collaborate with other agencies (like CMS or CMMI) with an eye towards outcomes-based contracts that allow rapid access to promising treatments but with ultimate accountability. These contracts could facilitate our transition to value-based care payment models. We will tackle these value-based care models in our next post.