In the fall of 2020, the president of the United States was diagnosed with COVID-19. At the time he was 74 years old, and by many accounts was not in the best shape of his life. At that point researchers had yet to discover any promising treatment for the disease – the WHO’s Solidarity trial and the UK’s RECOVERY trial hadn’t yielded any useful insights yet.
Donald Trump’s COVID-19 was treated with a massive 8 gram dose of REGEN-COV, an experimental monoclonal antibody medication. No kill like overkill, I suppose. The logic behind mAb therapy is straightforward: rather than waiting for the adaptive immune system to recognize the virus and design its own antibodies, hand them directly to the T-cells on a silver platter. This is the premise underlying drugs like adalimumab and nivolumab, although most monoclonal antibodies are designed to sic the immune system on molecules they wouldn’t otherwise target.
Regeneron’s recombinant antibody cocktail is painfully expensive to produce. In the earliest days of the pandemic tens of thousands of COVID-19 patients other than the president of the United States were treated using a much cheaper source of antibodies: convalescent plasma from other COVID-19 patients. The FDA issued guidelines in August 2020 permitting doctors to use convalescent plasma if they wanted. Many, many people received plasma infusions to fight the disease. Did it help?
We’re not really sure.
Alzheimer’s disease affects around 10% of people over the age of 65, but despite that the causes are poorly understood. One prominent hypothesis concerns amyloid β, a protein that can radically change its own tertiary structure in high enoungh concentration. At these higher concentrations, the theory goes, amyloid β becomes cytotoxic and starts impeding and killing neurons, leading to dementia. Some amyloid β plaque is in nearly all brains as a consequence of ageing, but finding plaques everywhere is a hallmark of Alzheimer’s disease. A great deal of time, effort and money have been spent trying to use these facts to treat Alzheimer’s.
The first breakthrough came in 2021 with Biogen’s aducanumab, another monoclonal antibody drug that binds to amyloid β fibres and slows down plaque formation. It received formal approval from the FDA, and informal disapproval from just about everyone else. The evidence that aducanumab actually improved outcomes was conflicted and unconvincing at best, placing it squarely among the ranks of amyloid β drugs that do not fucking cure dementia. Furthermore, because the drug was given accelerated approval, the lower evidence of efficacy and negative side effects (mostly cerebral edema) were given lesser weighting in the final analysis, on the grounds that something was better than nothing.
To the surprise of only those with low expectations, the administrating agency for Medicare and Medicaid (CMS) declined to pay for this drug. Aducanumab’s approval was so unwarranted that several members of the approval advisory committee resigned in protest. Even the researchers in favour of approval admit that the evidence aducanumab improves functioning in Alzheimer’s patients is weak.
This story echoes the approval of golodirsen, a nifty little gene therapy for Duchenne muscular dystrophy. Golodirsen was inititally rejected by the FDA, with a scathing CRL posted on August 2019 that underscored the weakness of the evidence for golodirsen’s efficacy and highlighted toxicity issues. Mystefyingly, the FDA later approved golodirsen in December 2019. The efficacy and safety of golodirsen, and whether it passes any sort of cost-benefic analysis, are the subject of an ongoing investigation.
Crucially, that investigation is ongoing while Sarepta Therapeutics sells golodirsen to DMD patients for hundreds of thousands of dollars annually.
The FDA was not originally charged with ensuring the efficacy of drugs seeking approval. That addition came in 1962, and was largely the work of Senator Estes Kefauver. After holding senate hearings on the pharmaceutical industry for years, Kefauver became convinced that the science behind the pharmaceutical industry didn’t hold up to real scrutiny, and that doctors and their patients were being systematically duped.
The most famous pharmaceutical story of that era, the thalidomide tragedy, was more or less averted in the United States on the say-so of one woman. FDA reviewer Frances Kelsey happened to know that some drugs could cross the placenta into an unborn fetus thanks to her malaria research experience as a postdoc. When she learned that a sedative marketed for morning sickness in pregnant women had never been tested on pregnant animals, she blocked its approval.
The publicity of the scandal and the audacity of that particular detail gave Kefauver the public support he needed for reform. Although not all of his proposed changes made the final cut, the Kefauver-Harris amendment charged the FDA with ensuring drugs were both safe and effective. This has the unusual consequence of preventing the sale of drugs known to be safe to patients who want to buy them, on the grounds that those drugs have not been demonstrated to be effective. For this reason, people of a libertarian bent might say, the Kefauver-Harris amendment ought to be repealed, and a different mechanism used to evaluate drug efficacy.
This other mechanism doesn’t exist. The government is all we’ve got.
Burden of the Odds
The FDA’s Center for Drug Evaluation and Research offers four different pathways for expedited drug development: Fast Track, Priority Review, Accelerated Approval, and Breakthrough Therapy. Two of those reduce the burden of proof needed for approval; all four save the pharmaceutical industry time and money. More than half of all new drugs take advantage of these options. Various congressional acts have given the FDA these tools (and others) to fill unmet needs faster; in doing so, companies are often held to a lower standard of evidence that their product is good for what it’s indicated for.
Big Pharma™ does not want to reevaluate the merits of a drug that’s already on the market. That takes money, which the sponsoring company would much rather plough into R&D or distribute to their shareholders. Rescinding authorization of Accelerated Approval drugs that are confirmed ineffective is possible but takes years to complete; in worse scenarios, companies have more time to manipulate the data to get the conclusions they want.
Even in the best case for expedited approval procedures, treating infectious diseases, lowering regulatory barriers is no guarantee of a faster or cheaper product. In writing about COVID-19, Derek Lowe points out that clinical trials are inherently expensive and complicated, that cutting corners introduces ambiguities in efficacy data, and that not all COVID-19 research was high quality. The world was fortunate to see tozinameran as early as November 2020, the way that HIV patients were to see HAART as early as 1996.
Zidovudine, which isn’t effective against HIV on its own, was approved in 1987.
In the world where the FDA approves drugs solely on their toxicity data — well, even that’s relative. Drugs like melarsoprol and amphotericin B are only slightly less awful than the diseases they’re prescribed for. But the run-of-the-mill phase I trial is smaller and cheaper than a phase II or phase II trial with thousands of patients enrolled internationally. As noted, most clinical trials even today either successfully determine that the drug is useless or are underpowered, a problem that would balloon to unmanageable proportions under a return to the safety-only regime.
There’s a stable equilibrium there, where CMS takes over the efficacy portion of the FDA’s mandate and other insurance companies follow their lead, but more likely it would fall to every entity responsible for purchasing healthcare to independently determine whether they want to pay for any particular drug. This is a problem: apart from medical interventions that save money in expectation down the line (like semaglutide) insurance companies do not want to pay for healthcare. It might not even be reasonable to expect that, if the market is flooded with drugs that are highly unlikely to work. Pharmaceutical companies could respond by pricing drugs lower, so patients can afford them out of pocket, at the expense of the revenue they need to develop drugs that actually work.
Proven to Our Satisfaction
All the incentives point towards snake oil salesmanship, which is exactly the regulatory environment the FDA is meant to avert. But, even positing some trusted record of efficacy (perhaps administered by the FDA) that insurance companies use to agree on coverage policies, there’s a second reason why restricting the sale of unproven drugs is a public good.
The gold standard of evidence for interventions is the randomized double blind control trial. The control group might get the current standard of care rather than a placebo (cancer patients in clinical trials usually compare new chemotherapy adjuvants to chemotherapy without the adjuvant), but otherwise the requirements are fairly important. An unfortunate fact about demonstrating that a drug works is that sick people need to have the medication withheld.
This sucks. It especially sucks if you are one of the sick people. Uncertainty about whether your injections are an experimental drug or room-temperature saline solution is an indispensible part of the process, and understandably no one likes it. Absent Kefauver-Harris, patients can and will buy the medication on their own. Buyers clubs for medication still exist; the most recent drug I know of with a buyers club is Vertex’s ivacaftor, approved in 2012 for a subset of cystic fibrosis cases with certain CFTR point mutations. Ivacaftor was a Breathrough Therapy, although the evidence of it working was convincing. Most drugs are far less impressive in the clinic. Absent legal barriers to their purchase people will acquire drugs that are approved on safety and not on efficacy, and consequently it will become extremely difficult to establish confidence in Phase III efficacy trials.
Again, there are stable states of drug approval that the pharmaceutical industry could find itself in that produce working medication. Unfortunately, most published results are false, and making it easier for bad science to pass from laboratories to hospitals is incredibly unlikely to be net positive for consumers of healthcare. Proponents of repealing Kefauver-Harris have an admittedly strong point when they say that restricting the sale of unproven drugs violates personal liberties, but when the alternative is allowing the sale of drugs that overwhelmingly don’t help and undermining the integrity of clinical trials, the new plan needs more than a little explication.
The government is in a position right now to solve a collective action problem. It’s a thorny situation, but that’s because of the way reality is, not the government. Many such cases.