From molecule to medicine, with Ross Rheingans-Yoo

I'm joined this week by my friend Ross Rheingans-Yoo, who has had an eclectic career, most recently in investment in pharmaceutical R&D. We chatted about what the lifecycle of drugs is, how trials are regulated by the FDA in the U.S. (and what the history is there), and how processes with many well-credentialed PhDs, lives hanging in the balance, and billions of dollars on the line are sometimes surprisingly unoptimized.

[Patrick notes: I add inline observations to the transcript after the recording of the conversation, set out in this fashion. As always, transcripts at Complex Systems are not stenographed for literally reproducing the conversation, but are edited (including by an AI this time) with an eye to producing the best readable version of a spoken conversation.]

Sponsors

Manifold is my favorite prediction market platform, making it easy to bet on anything from elections to crypto drama. Get started with a bonus at https://manifold.markets/complexsystems when you spend $20. Play money markets open to anyone; cash prizes only available to U.S. residents of at least 18 years of age. Terms and conditions apply.

Check is the leading payroll infrastructure provider and pioneer of embedded payroll. Check makes it easy for any SaaS platform to build a payroll business, and already powers 60+ popular platforms. Head to checkhq.com/complex and tell them patio11 sent you.

Timestamps

(00:00) Intro
(02:28) Ross’ career transition to drug development
(03:12) The drug development process
(06:22) Clinical trials and FDA approval
(11:48) Challenges in clinical trials
(14:50) Case study: COVID-19 trials
(18:00) Sponsors: Manifold Markets | Check
(19:55) Pharmaceutical economics
(38:13) Rare diseases and regulatory strategies
(45:18) Advanced market commitments explained
(45:54) Operation warp speed and its impact
(47:45) How to get accelerated approvals
(52:49) The thalidomide tragedy and its legacy
(01:03:17) Modern regulatory challenges and patient advocacy
(01:07:14) Reviving abandoned drugs
(01:12:06) Innovative approaches to drug trials
(01:24:26) The future of pharmaceutical development
(01:26:34) Wrap

Transcript

Patrick McKenzie: Hideho everybody, my name is Patrick McKenzie, better known as Patio11 on the Internet, and I'm here with my buddy Ross Rheingans-Yoo.

Ross Rheingans-Yoo: Hey Patrick, good to be here.

Patrick McKenzie: Great to have you. we're going to talk a little bit about clinical trials and drug development today. 

Patrick McKenzie: So, like myself and many guests on this podcast, you've had a bit of a winding career. I first met you when you were at Jane Street. You brought me in to be a guest speaker, if I remember right. And then you ended up in drug development. What happened in the middle?

[Patrick notes: The live conversation said drug "discovery" rather than development. Ross has explained to me that, in the industry, discovery is the more scientific arm of the process; development is the financial, logistical, and commercial work of turning that science into something which scalably improves human lives.]

Ross Rheingans-Yoo: Sure. I spent five and a half years at Jane Street, first in New York and then in Hong Kong, which is when we had you as a virtual guest speaker. I have a lot of positive things to say about the place, but those belong in a different podcast.

[Patrick notes: This makes Ross by my count the third former Jane Street employee we’ve had on Complex Systems, which is partially a reflection of my skewed personal and professional networks, and partly downstream of very interesting things they do with regards to identifying and attracting early-career talent.] 

Ross Rheingans-Yoo: At the beginning of 2022, I was invited to take a position at a philanthropic foundation funded by someone who had recently made a significant amount of money in crypto. I thought this would provide an opportunity to work on more exciting and timely projects than I had been involved with recently. So, I went to work as a program officer for the FTX Foundation.

I held that position until about November of that year when I unfortunately learned that one of the challenges facing a philanthropic foundation can be issues related to the sources of its donors' funds. I had several legal interactions with the bankruptcy estate following those events. However, two things I’m glad to have come from that experience are: first, the estate withdrew several unfortunate allegations they had made against me personally.

Secondly, in the final resolution of the settlements, the estate agreed that while the potential returns from some equity investments I had directed would be recovered for the benefit of creditors, they would not seek to pull money out of the companies that were continuing to actively develop drugs and push those programs forward. I’m pleased that the companies I supported were able to continue their flight paths without much interruption after those settlements.

Ross’ career transition to drug development

But we’re here to talk about drug development more generally. So, what do you want to know? Where do we start?

Patrick McKenzie: Yeah, I’m sorry you had that rough experience, but I’m glad it worked out in the end. 

[Patrick notes: I will note, not for the first time, that the FTX caper had many, many layers to it. One of them was an affinity fraud against people broadly affiliated with effective altruism, and another layer was a privilege escalation attack on the institutions of the American professional-managerial class. FTX’s senior management has few people who will admit to being friends with them prior to November 2022, for understandable reasons, but I think it’s instructive to reflect on how it fooled so many people for as long as it did.

To not put words in anyone else’s mouth, I was surprised by the FTX implosion, and I shouldn’t have been, as a notoriously crypto-skeptical individual who had enough of the pieces to put the story together. Partly this was because I thought I understood the crime: FTX/Alameda was Tether’s money launderer, and I assumed they were mostly competently executed on their side businesses while making their serious money from gatekeeping Tether’s access to the U.S. financial system. I did not have a very developed point of view on their degree of understanding that this was actually their business; one can deliver that fee for services as arbitrage profits, and the only party willing to crime will keep reaping the profits. SBF et al were quite brazen in saying that this was the plan; few financial services CEOs get on Bloomberg and say, in as many words, that most compliance officers would look at their founding narrative and diagnose it as money laundering.

But the other reason was, in looking at FTX management, I saw people who looked like me, talked like me, had clearly spent a lot of their lives reading the same websites I did, and struck me as having an ethical system I didn’t quite share but could appreciate. They were going to rapaciously harvest profits by enabling degenerate gamblers and committing process crimes, achieve regulatory capture to get those crimes forgiven, and then spend the money on charitable causes. And they were clearly achieving incipient success on harvesting profits and incipient success on regulatory capture.

And at no point prior to the collapse did I ask the fateful question: “The story being told here is one designed to work on me. Am I the mark, or sufficiently similar to the marks that a strategy designed to exploit them also exploits me?”]

Patrick McKenzie: For the benefit of an audience who might not have spent a significant portion of their career in this field, there's a process that can feel a bit opaque, and then, at  some point, you go to your doctor with a complaint. Your doctor diagnoses you with a condition and prescribes a medication for it. You take the medication, and your insurance company covers at least a portion of it, in the typical case.

That vague blob of uncertainty is where a lot of drugs go to die, as I understand it. So let’s start at the beginning: where does a drug begin its life?

The drug development process

Ross Rheingans-Yoo: The development of a drug is, in my experience, more honored in the exception than in the rule. However, there is a fairly standard process that we can use as a jumping-off point for discussing the exceptions. The prototypical drug development story begins in a university lab, where a scientific team, often headed by professors of medicine, investigates the effect of a particular molecule on a specific biological system, hoping that this effect will eventually help resolve a disease.

This work typically takes place within the university setting until it reaches a point where the research transitions from exploratory studies to a more structured plan for bringing the drug to market. At this stage, the drug goes through the university's tech transfer office and then ends up in a startup company that owns the intellectual property. The university may retain a small equity stake, and that startup is then poised to seek venture capital funding to advance the development.

At this point, the drug is moving from investigative studies in petri dishes and animal models to a more concerted and planned roadmap, first in animals. After those trials, it will begin human trials through several conventional stages, eventually seeking final approval from the FDA and a quasi-approval from the insurance industry, which involves agreeing to cover the drug.

This is the lifecycle of drug discovery which happens in private industry, as opposed to in an academic lab.

So, after the company has spun out from the university, it often continues conducting trials in non-human animals. These trials aim to compile enough data to make an initial application to the FDA to study the drug in human. That is known as an IND, or investigational new drug application. When the FDA grants that stamp of approval, it represents the first of several green lights that allow the drug to be administered to humans for the first time.

Patrick McKenzie: My understanding is that many of these startup companies have essentially one asset: the drug or set of drugs they are attempting to bring to market. After they receive FDA approval, these companies are often acquired by larger drug manufacturers, allowing the startups to leverage their unique core competencies in drug development while the majors handle marketing, manufacturing, and distribution. Is that roughly accurate?

Clinical trials and FDA approval

Ross Rheingans-Yoo: Yes, you've correctly identified that the company that spins out of the university is rarely the one selling you the drug a decade later.

I want to highlight one detail: the company that begins the development often doesn't hand off the project immediately after IND approval. They typically conduct some studies in humans to further de-risk the prospect for potential acquirers.

In a typical scenario, the spinoff company would complete its non-human testing, obtain investigational new drug approval, and then carry out phases of safety testing in healthy volunteers to identify any obvious risks. Following that, they would conduct proof of efficacy and proof of concept testing. This is less for the regulatory apparatus but more for the commercial apparatus, like insurers. Gathering this evidence helps convince investors or potential acquirers that this drug should continue to advance.

It's generally considered necessary for any acquisition transaction. Plenty of companies have drugs they believe should work and can argue that they should work, but very few drugs get acquired before there's evidence of positive outcomes in humans.

Patrick McKenzie: So, we’re looking at a funnel here, and the pass rates at different stages probably vary. We start with thousands of molecules to find one candidate that looks interesting enough to move from the petri dish to animal testing. What’s the drop-off rate for moving to full human studies?

Ross Rheingans-Yoo: Great question! The numbers might not be what people expect.

In broad generalities—and while it’s possible to get arbitrarily specific about particular indications, types of drugs, types of molecules, and situations—if I speak generally, you might see that about two-thirds of drugs that begin phase one safety testing in healthy volunteers proceed to the next phase. Industry-wide studies don’t always delve into the specifics, like why a trial might have been stopped, whether due to adverse reactions, changing corporate priorities, or funding issues. So, if we abstract those details away, the rough statistic is that around two-thirds of drugs that start phase one will move on to phase two.

In phase two, we focus on efficacy testing—essentially proof of concept in patients who actually have the condition we’re trying to treat. At this stage, you might see success rates between 40% and 50%.

Patrick McKenzie: One of the reasons this process feels interminably long to patients, who often wonder why their disease hasn't been cured yet, is that it takes many shots on goal to start entering the funnel before one eventually makes it through. The latency on this process is likely several years, right? If I’m a bright postdoc with a promising molecule today, and things proceed at the usual pace, when could that realistically be consumable by a patient outside of a trial?

Ross Rheingans-Yoo: If you’re a bright postdoc working on a promising molecule in your lab, dreaming of starting a company to bring it to patients, you’d be lucky to finish that journey in much less than a decade. The process involves multiple shots on goal, and the timeline can be extended significantly if things go wrong. A straightforward path through this process typically takes about seven years for many common conditions, assuming nothing goes awry.

We’ll talk later about what could make this different, and what procedural levers could make it dramatically faster. 

To summarize the funnel: we see a two-thirds success rate in moving from safety trials to initial efficacy testing, and then 40% to 50% from efficacy testing to pivotal trials. A phase three trial has parameters agreed upon with the FDA from the start, meaning if it succeeds, they won’t have further questions or requests before making their final determination. About two-thirds of drugs that pass their safety and efficacy trials successfully complete their pivotal trials. Of those, approximately 85%—or five-sixths—receive FDA approval.

When you multiply those pass rates, it translates to between one in six and one in ten drugs that are first given to humans actually becoming approved medicines. As I mentioned, this entire process typically takes seven to ten years in a prototypical case.

[Patrick notes: As it is always difficult presenting complex multi-stage funnels from memory, would point out that interested readers can compare Ross’ gloss here to academic research on this subject. Here’s a table:

] 

Challenges in clinical trials

Patrick McKenzie: One thing I’ve heard from others... well, stepping back for a moment: 

To what extent are we constrained by physics or pharmaceutical reality and to what extent are we constrained by the systems that we decided to institute?

Ross Rheingans-Yoo: When we think about the process of drug development, particularly the clinical campaign, we’re simultaneously trying to understand what is true about the world and convince the regulatory apparatus of those truths. I like to view this as a combination of a statistical problem multiplied by a logistical problem.

From my experience as a quantitative trader, I see this as looking through a small keyhole to view the world. We want to know what is true, but what we observe through that keyhole is often noisy and composed of numerous granular anecdotes. Statistics tells us we need to gather a certain amount of data—observations from various patients—before we can arrive at a clear understanding of what’s happening.

There’s also the aspect of how we choose to interact with each patient: in what setting, for how long, and what metrics we will measure during that process. These design parameters can multiply the cost per patient and the time required to recruit the patients we want to study.

Broadly speaking, each of these choices multiplies the fundamental statistical problem we need to solve.

Patrick McKenzie: Then there’s the medical reality of diseases proceeding across various time scales, and those timelines don’t necessarily align with our desire to hit milestones or “get things done by Q3.” 

Recruiting patients for trials—I have an interesting anecdote there for later. But for context, there’s been a specialization of labor in the clinical trial market, right? Who actually runs these trials nowadays?

Ross Rheingans-Yoo: That's right. The key term you’re looking for here is "clinical research organization" or CRO—these are specialized firms that handle the actual running of clinical trials. So, if we consider our startup that spun out of a university lab, it might only have half a dozen team members, and their main action item will be contracting with a CRO to run their trials.

When you ask about step-to-step efficiency or who’s optimizing the process, keep in mind that the front-line personnel are not typically doctors with a vision of saving patients in a decade. Instead, they’re specialists managing the trials professionally.

Case study: COVID-19 trials

Patrick McKenzie: So, my fun anecdote about recruiting for a clinical trial. I happened to graduate from Washington University in St. Louis, and by coincidence, during my stint with the U.S. vaccine location information architecture project, WashU was running a trial called Stop COVID. They were studying the efficacy of fluvoxamine, an antidepressant that might have off-label potential as an acute COVID treatment. The idea was to offer patients a range of effective medications to reduce severe clinical outcomes from their exposure to COVID.

I got an outsider’s look at their efforts.

One of the unique circumstances for the WashU team was that, since fluvoxamine was already FDA-approved, they didn’t need the full regulatory gauntlet but just needed data showing improved outcomes if patients took this readily available drug. They still had to run a study, recruit thousands—or ultimately, a few hundred—patients, and do so within a very short window. You need to recruit each patient between contracting COVID and (days later!) either recovering totally or experiencing a severe clinical outcome.

So, they went with an Internet marketing campaign to reach people who’d recently gotten COVID. Naturally, they encountered some issues because, as it turns out, med school doesn’t exactly cover Internet marketing.

One skilled individual who happened to know HTML on the team wrote up a 200-question screener to assess eligibility for the trial. They sent potential patients to this screener, and, as many who work online could predict, many people dropped off before completing all 200 questions. When I connected with the team, I assumed there must be a solid regulatory or procedure-based constraint for the length. I asked, “Is this an institutional review board requirement? Medical ethics? Something specific to drug development?” The answer? “We have 200 questions because that's what the grad student implemented.”

So, I asked, “How many of these questions do you actually need?” They replied, “Four.” At that point, I offered the help of someone skilled in conversion rate optimization to design a simple landing page with just those four essential questions to screen prospective patients and capture contact info for trial enrollment. Surprisingly, that approach worked. So, there are definitely opportunities for upskilling across the clinical trial supply chain. [Patrick notes: A full discussion of the clinical trial is beyond my ken, but I’m as-far-as-I-know accurately summarizing VaccinateCA’s small engagement with it, which was a) staffing my good friend Keith Perhac on their patient intake form and b) attempting to assist with patient recruitment via e.g. Facebook ads which we funded.] 

And two, this was a clear example of how asking, “This thing that’s broken—why is it broken? Can it be improved?” often yields value. It reinforced my belief that, contrary to assumptions, just because many PhDs were involved in a project doesn’t mean that all possible alpha has been optimized out.

[Patrick notes: Eliezer Yudkowsky has a book-length treatment, Inadequate Equilibria, exploring why our intuition that “systems generally run about as efficiently as they can be made to run” functions well for certain systems (e.g. public equities markets, which are practically the base case for the Efficient Markets Hypothesis) and grossly mispredicts other systems. Medicine and drug development are both examined as miniature case studies within it.] 

Pharmaceutical economics

Ross Rheingans-Yoo: That’s right. One of the most surprising adjustments I had to make moving from public equity markets to clinical trials was realizing just how much untapped potential exists. There’s a surprising amount of low-hanging fruit, and there are practices that, if reconsidered, could lead to better medical and commercial outcomes. Often these simply haven’t been considered.

The fact that a 200-question survey could be reduced to four if someone simply asked, “How many questions do we really need?” doesn’t surprise me at all.

The Stop COVID—or StopCOVID-2—trial from WashU is interesting because the investigators were collaborating on another trial I know well.

[Patrick notes: Does the world ever seem preposterously small to anyone else?]

Ross Rheingans-Yoo: For some specifics, I believe the StopCOVID trial at WashU was recruiting for several months, averaging about 100 patients a month. [Patrick notes: I obviously wanted that to be higher, but 2021 was a bit busy for VaccinateCA and myself, and we were limited in how much we could throw at their project.] They stopped at around 500 participants, realizing that, at the recruitment rate and with the criteria for measuring outcomes, they wouldn’t achieve statistically significant results anytime soon. So, they closed it out once they understood it wasn’t on track to deliver meaningful data.

Patrick McKenzie: It’s counterintuitive, but there are plenty of ways to lose while winning in drug development.

You can end a trial early because it becomes unethical to continue—like if preliminary data suggests the drug’s effectiveness is so strong you should start offering it immediately to everyone in the study. "Immediately" might be a bit of a stretch administratively, but it applies to the patients currently enrolled.

Or, you might stop a trial for the opposite reason—if it becomes clear the treatment isn’t effective.

In the case of StopCOVID, the trial was partially overtaken by events: the rollout of COVID vaccines started reducing severe cases, making it harder to recruit patients for countermeasure studies.

[Patrick notes: The U.S. then had a rough experience with Omicron, for people wondering why the graph didn’t stay improved.] 

Ross Rheingans-Yoo: Sure, at least in that period and in that locale. 

The interesting aspect of the StopCOVID-2 trial, which ran in early 2021, was that it was happening around the same time as the TOGETHER trial in Brazil. The TOGETHER trial studied multiple drugs, including fluvoxamine, and their recruitment overlapped with StopCOVID-2’s fluvoxamine recruitment.

They were recruiting about three times as many patients in roughly the same timeframe—1,500 patients instead of 500. Crucially, around 15% of the TOGETHER trial patients experienced clinically significant events, whereas only about 5% of StopCOVID-2’s patients did. Those numbers matter because, statistically, the study’s aim is to set a threshold for worsening health outcomes and determine if we can prevent patients from reaching that threshold. If I give the drug to someone who doesn’t get worse—like the 85% of people who naturally improve—they don’t contribute meaningfully to the study’s statistical analysis. What we want to observe is whether the percentage needing additional care decreases, say from 15% to 11%, or from 5% to 4.2%. To detect shifts within smaller percentages, though, we need a larger sample size. So, when I say TOGETHER recruited three times as fast, it’s also about the trial progressing towards statistically meaningful results faster.

Patrick McKenzie: This is the classic “statistics 102” topic on statistical power that often gets overlooked—even by people in marketing. A common mistake in A/B testing, for example, is looking for statistically significant results without enough sample size to detect a real effect at the levels one expects conversions to happen at.

[Patrick notes: As an illustration of this, a software company which currently has a trial-to-paid conversion rate of 3% and wants to run A/B tests which might generate 10% lifts (i.e. 3.3%) needs to burn ~60k trials per iteration to achieve 95% confidence that it will detect those lifts. This is difficult for companies which don’t get 60k trials in a 6 month period.

Awareness of statistical power in marketing departments is below awareness of statistical significance, and marketing departments understand statistical significance better than poorly educated sectors of society like e.g. doctors, who mostly are not given the lecture on why Bayes’ Rule matters for ordering tests.

(A test with a 1% false positive rate and a 1% false negative rate applied to a population with a 1% true incidence rate of Disease X will, contingent on flagging a patient as having X, have a what percent chance of being accurate? Most physicians answer ~99% when asked, not ~50%.)] 

Ross Rheingans-Yoo: Exactly. TOGETHER’s recruitment rate for fluvoxamine patients didn’t just outpace StopCOVID-2 by sheer numbers; statistically, they were achieving useful data far faster on a per-patient basis. In practical terms, they were progressing toward their goal about nine times faster than StopCOVID-2. This difference highlights why a trial in the U.S.—with the vaccine rollout impacting COVID cases—proceeded very differently from one in Brazil, where TOGETHER was unaffected by widespread vaccination and operating under different conditions.

Ultimately, TOGETHER’s fluvoxamine study showed a roughly 30% reduction in patients needing hospitalization or intensive care, which was a significant finding. That meant a reduction from about 16% of patients needing further care down to 11%.

Frankly, if I could push a button to reduce my hospitalization odds by 30%, I’d push it immediately. By comparison, StopCOVID-2, with around 500 patients prior to cutting their fluvoxamine arm, saw a reduction in severe outcomes from about 5.5% to 4.5%. While it was still a positive result, it leaned closer to a 10% reduction rather than 30%, possibly due to the baseline context of medical care and vaccination rates in the U.S. versus Brazil.

Patrick McKenzie: I guess a headline takeaway here is that not every well-executed medical trial will deliver groundbreaking results. There were a lot of negatives in society’s COVID response, but one clear positive was how we mobilized quickly. We took multiple shots on goal, trialing various vaccine and treatment candidates. We ramped up medical research funding and launched trials at unprecedented speed, leading to effective vaccines and treatments in under two years.

Of course, if 100% of treatments studied had achieved 80% efficacy, we’d probably be taking too few risks and should aim for higher-hanging fruit. But with trillions of dollars and millions of lives at stake, we also needed to take some longer shots.

This whole discussion is a “stats 102” reminder: a disease with a short incubation period—an acute illness you might catch only once or a few times—creates logistical hurdles for running effective trials. And that ties back into the economics of a drug, its appeal for funding and acquisition, and ultimately its commercialization potential. Could you elaborate on how, from the perspective of Seeing Like A Pharmaceutical Company, certain types of drugs are prioritized for funding, and the complex factors around different conditions that make this likely?

Ross Rheingans-Yoo: Sure. I think you’ve pinpointed the core of this discussion by distinguishing between acute and chronic conditions. I’d add that there’s substantial concern—legitimate concern—about the long-term consequences of COVID infection, particularly severe cases. It’s telling that while there was a rapid and successful effort to develop acute countermeasures, the study of COVID’s chronic effects has been a protracted, almost stalled effort. We certainly don’t hear weekly updates on that front. So, why is this? Viewing it through a pharmaceutical company’s lens offers some insight.

From a startup’s perspective, and from that of its venture capital backers, it’s about navigating a process where, ultimately, only 10% to 15% of drugs are approved. The question then becomes: what are the revenue prospects if the drug succeeds? Because once approved, nearly 0% of the drug’s cost is in production. Manufacturing authorized molecules is inexpensive; costs from there are largely marketing and operations, resulting in very high margins. However, you still need to recoup the hundreds of millions spent on R&D—not to mention covering the costs of the nine drugs that didn’t make it, which your VCs also funded.

Patrick McKenzie: And on top of that, you’re racing against the clock. Patent exclusivity lasts only about 20 years, after which generics can compete directly with you, replicating the drug’s compositions and effects.

[Patrick notes: In the live conversation I erroneously estimated 10-15 years for the patent window. Mea maxima culpa.]

Ross Rheingans-Yoo: Exactly. The patent essentially serves as a social contract: the government enforces your exclusive rights to profit from your innovation for 20 years in exchange for revealing how it’s made. After that, anyone can produce the same molecule, perhaps in a pill of a slightly different color or shape. But those 20 years start ticking from the date of the patent application, not when the drug hits the market. Given the time it takes to get through development—often close to a decade, as discussed previously—you may only profit from your exclusivity for 10 years, starting long after your initial investors’ capital was deployed. So, in essence, you’re looking at a limited exclusivity window to recoup your high costs of capital.

When we add up success rates across the pipeline, a typical clinical campaign requires the drug to generate around $2 billion in net present value at the point of approval—effectively, $200 million annually over a decade.

Patrick McKenzie: So drug development becomes a hits-driven business where a blockbuster like Ozempic ends up footing the bill for numerous trials.

Ross Rheingans-Yoo: Precisely. For a “hit” drug to cover its own costs, plus those of nine dead siblings and associated capital costs, it needs to bring in about $200 million per year over a 10-year span. This limits the class of drugs that could get studied to ones which could credibly produce this outcome.

In practice, this often means targeting drugs that can be sold to insured Americans and taken for many years for each patient. It’s difficult to generate revenue selling a two-week course for $20,000 to a limited number of patients; political and socioeconomic factors typically prevent that. So the path to $200 million in annual revenue for a decade often involves medications for chronic conditions, taken by insured patients for years and billed in the five figures annually.

[Patrick notes: In a bit of art imitates life, can I reproduce my Steam review of the video game Big Pharma. The review was written for humor value and the game frequently is a polemic (in addition to being a fun puzzle game), but I’m amused that a toy simulation of the industry helps one develop useful intuitions here.]

]

Ross Rheingans-Yoo: This narrows our focus to diseases that are prevalent among insured Americans, particularly elderly Americans who are developing a set of age-related conditions. And because this is the population you’re treating, it’s also the population you’ll end up studying. If your goal is to prevent cardiac issues in Americans over 55, you’ll design studies around Americans over 55 who are at elevated risk of cardiac events.

This is a very difficult place to begin drug development, from a statistical and logistical sense!

First, delivering healthcare in the U.S. is costly. It will be much more expensive than delivering broadly similar healthcare in, to use our earlier example, Brazil. By a multiple.

Second, with older patients, many already have multiple health conditions. Unlike in COVID studies, where you can often find patients with only one primary condition, patients in cardiac or cancer trials frequently have other pre-existing issues. Your typical patient is going to have something else wrong with them that was wrong with them before they got whatever condition you're trying to treat. Even if your treatment is successful, they will still have that other condition wrong with them after you treat them.

This adds “noise” to your measurements, requiring a larger sample to isolate the drug’s effect. These issues present over multiple years and so we’ll need to study the effects of treatment, the durability of those effects, and the progression of disease over multiple years. You’re likely looking at least a 12-month study period.

And we’re probably not catching the condition close to onset. In expectation, the patient will likely have developed this years before entering our trial. And this makes it harder to measure and predict whether drugs will be helpful to future patients who are offered them early in their experience of the disease. (For example, we just brought up the difficulty of finding COVID patients within days of them developing COVID.) 

For cardiovascular issues, neurodegenerative conditions, and obesity-linked illnesses, the patient will often present for treatment (and enroll in your study) years after the issue has started impacting their life, initially in relatively minor ways.

And so you’re attempting to move the needle a bit against a lot of statistical background noise, in a medical context where care is extremely expensive. These combine to drive the high cost of trials.

Rare diseases and regulatory strategies

Patrick McKenzie: Right. These factors interact with each other.

Backsolving from some numbers you gave earlier:

And if you’re aiming to recruit, say, 20,000 patients whose insurance will reimburse at $10,000 per year, there’s a larger funnel at play. Many more than 20,000 people need to have this condition to yield a manageable patient pool, since only some will seek medical help.

This will also screen out potential drug development for issues which are real but are below some salience floor for patients. If I go to the doctor, for any reason, it tends to cost me $400 for five minutes of a physician’s attention. [Patrick notes: Personal experience; your results may vary. Also in Japan it was $40, in part because Japanese doctors who are not practice owners earn far less than American doctors.] And so you’re restricted in looking at patient populations which will largely be downstream of “What impacts to my quality of life are worth many thousands of dollars to me?”

And thus we have funnels upon funnels: one in patients, where some percentage see a doctor, some get a successful diagnosis, and some are potentially available for clinical trials. And among conditions, where some fall out of the funnel pretty quickly.

An example: diseases which don’t have a potential patient population in the hundreds of thousands or millions, so-called rare diseases, likely never make it into the funnel at all, because you can’t backsolve to 20,000 patients on the de facto subscription plan that you need to make the numbers work.

[Patrick notes: Presumably a medical researcher would see drug pipeline in my SaaS company and I see a SaaS company in his drug pipeline. Be that as it may.] 

Ross Rheingans-Yoo: You’re right, but there’s an interesting nuance with rare diseases. 

You can hold constant $10,000 per patient-year as an insurance company’s willingness to pay. Hwoever, this is subject to social and political factors acting on insurance companies. Insurance companies may pay far more per patient for conditions with no other effective treatments.

So, one viable strategy is targeting a rare disease where you charge $100,000 annually but only need to treat 2,000 patients—a concentrated market with no alternatives. On the other hand, you could aim to treat a $100 condition for 2 million patients, though this brings its own logistical challenges. 

[Patrick notes: A prominent example which came to market in my own lifetime: you can take a pill for seasonal allergies now. It’s ~$100 during allergy season and lets you opt out of what was generally an inescapable low-to-medium level of multi-month misery that affected a large swathe of the population. (Which, fascinatingly, started a new chapter in marketing drugs to potential patients directly versus primarily marketing drugs to doctors, in part because seasonal allergies were not something a physician was likely to bring up as a chief complaint requiring treatment.)]

Ross Rheingans-Yoo: Broadly, the path that seems more viable to companies is selling a high-priced drug to a small, high-need group rather than aiming for broad, low-cost access. With rare diseases, where patient numbers are small but needs are critical, regulatory approvals can sometimes be easier, and competition is often limited. 

This has made rare diseases an increasingly appealing niche in commercial circles due to the financial and regulatory dynamics at play. A concentrated population and one with few other options makes the political economy of getting regulatory approval easier. This strategy is increasingly popular; you didn’t come up with the phrase “rare diseases.” This is an increasingly interesting area of study for commercial reasons.

[Patrick notes: A deserved nose tweak.]

Patrick McKenzie: One of the interesting policy developments in recent years is that while we have regulatory constraints that might ideally be shorter.

We've also started essentially selling priority access to the regulatory process.

To put it simply, there’s now a system to receive expedited treatment from regulators. There are various prioritization mechanisms, but one we’ve settled on involves issuing vouchers as rewards for beneficial services to society. These vouchers are transferable and are often awarded to firms that develop treatments for rare diseases or for conditions prevalent in underserved populations overseas. Essentially, the reward is a voucher for faster FDA review next time, which can be worth tens or even hundreds of millions of dollars—a significant incentive if you manage to secure one.

Ross Rheingans-Yoo: Yes, the term here is "priority review voucher" (PRV). It’s a fascinating piece of bureaucratic system design. The original statute awards PRVs as incentives for developing treatments for pediatric diseases or rare conditions with limited options affecting small patient populations (tens of thousands).

A company that gains approval for such a treatment receives a voucher, which places its next drug application in the FDA’s priority review category. There are other routes to get into this category, like targeting rare pediatric diseases, or life-threatening rare diseases. But, if you're simply developing a new version of an existing blockbuster drug, the FDA requires a PRV to move it up the queue.

And so simply by being willing to move applications between priority queues, the FDA was able to conjure up an instrument worth about $100 million each, which it could use to incentivize drug development in policy priority areas.

[Patrick notes: Per GAO, actual recorded sales have been $67 million to $350 million. The GAO scores this program as self-funding because the FDA charges you a few million dollars in administrative costs to transfer one of these vouchers.] 

Ross Rheingans-Yoo: For context, recent PRVs have often been used for anti-obesity drugs—a highly profitable area. So, while developing the 10th anti-obesity drug might not be a public health priority, the PRV awarded for a rare or critical treatment can then fast-track any subsequent application.

Over the years, the eligible disease categories have changed based on congressional statute—rare pediatric and so-called tropical diseases are examples, with the latter term referring to conditions that may only cause minor issues in the U.S. but are devastating in lower-income countries. Pandemic threats were also included temporarily. Programs are generally authorized in five-year increments, subject to renewal if policy interest persists. (These are tacked onto appropriations bills.)

Patrick McKenzie: There’s an impedance mismatch here, unfortunately. If a postdoc needs to lash themselves to the mast for 10 years to bring a treatment to market, but the eligibility windows for priority incentives open and close at five-year intervals, those timelines don’t always align.

Advanced market commitments explained

 Another notable development has been the concept of advanced market commitments (AMCs), primarily for vaccines, though they may apply to other drugs as well. Can you give some color on what an AMC entails?

Ross Rheingans-Yoo: Certainly. AMCs are less about regulation and more about securing the commercial viability of treatments. They’ve gained traction as policy interventions, particularly since COVID. 

Operation warp speed and its impact

[Patrick notes: The first notable one was back in 2007 on pneumococcal vaccines. My former colleague Nan Ransohoff, who runs the project I’m about to mention, has written about the history of AMCs.] 

Early in 2020, with the urgent need for countermeasures, Operation Warp Speed in the U.S. incorporated AMCs. [Patrick notes: If you want a book, here are two. The GAO report is also good reading.] Essentially, the government committed to buying a treatment in massive quantities—worth hundreds of millions or even billions of dollars—if it met specific efficacy and accessibility criteria.

For example, we could specify that we’re looking for treatments that reduces COVID hospitalizations by 50% or more and can be administered by a nurse outside a hospital setting. If you produce one of these, we’ll spend a billion dollars on doses. This greatly simplifies the financial path to hitting that $2 billion target to unlock funding, by taking something that was in the national interest anyway and making it a bankable commitment earlier in the project.

Patrick McKenzie: Interestingly, Stripe used a similar approach with their Frontier project. For years, carbon removal initiatives struggled with the question of who would fund them. Stripe pulled together a coalition of large companies and online businesses, committing to buy carbon removal at above-market rates—or effectively, without a current market rate. We will buy this outcome and we will buy it in quantity.

This model might even merit a separate podcast episode.

[Patrick notes: Frontier is a public benefit LLC backed by a consortium of partners; Stripe’s product which helps to fund it is Stripe Climate. Obligatory disclaimer: I previously worked for Stripe, am currently an advisor there, and did (a relatively tiny amount of) work on this.]

How to get accelerated approvals

Ross Rheingans-Yoo: The third major component in the regulatory landscape is what's known as "accelerated approval." This approach originated from the AIDS pandemic through patient advocacy. [Patrick notes: Ross recommends this history to those interested.] In certain cases where the condition is serious and in the U.S. policy interest to expedite a new treatment, the FDA allows a drug to be sold after proving it’s safe and effective, even if all follow-up studies aren’t complete.

Under normal circumstances, after a drug completes one pivotal study, the FDA would require a second study to ensure replicability. They want to know you can hit the baseball twice. This catches out a lot of drugs that pass a single study at p=0.05 but can’t consistently replicate results. However, with accelerated approval, companies can start selling the drug while they conduct a second study.

Accelerated approval began out of the AIDS pandemic. This exceptional treatment has become the de facto standard pathway for life-threatening conditions without an existing treatment standard-of-care.

For example, in cancer treatment, approval might be based on tumor shrinkage rather than waiting to see if patients live longer over five years. This surrogate endpoint” is a proxy, but a useful proxy, particularly if you have no other drug available for the cancer you have. 

In such cases, the FDA lets companies sell the drug. It will mandate follow-up studies that track patient outcomes, but this buys the companies selling the drug 4-5 years shaved off of their time-to-market. This pathway is highly attractive for companies as it allows revenue generation sooner, which can be crucial to making the financial case for a drug.

The pharma strategy world is intensely focused on what endpoints can be measured early for accelerated approval, but, as expected, less than 100% of drugs approved through a single, surrogate endpoint study actually demonstrate a real benefit in long-term outcomes. Consequently, we see drugs on the market for four years, only to later find they lack true efficacy based on follow-up studies. We should expect to continue to see this as long as we have this program in place.

As you often say, Patrick, the "right" amount of failure in a system like this is not zero. 

[Patrick notes: This has become one of my most-quoted lines in my corners of the Internet, and frustratingly, I think I mostly inadvertently lifted it from Dan Davies ("It is highly unlikely that the optimal level of fraud is zero.") Though I suppose that the right number of phrases to lift is non-zero.]

Ross Rheingans-Yoo: Setting a zero-tolerance policy could lead to outcomes on the margin that we wouldn’t want.

So, while it may stir ongoing debate within the medical and commercial communities, I believe the right failure rate for accelerated approvals isn't zero. Discussions on whether the FDA demands appropriate endpoints, or if companies are gaming the system by selecting favorable metrics, will likely continue as long as the program exists—until or unless it’s entirely eliminated, though I think zero would be the wrong choice.

The thalidomide tragedy and its legacy

Patrick McKenzie: Exactly. Every institution has its own path dependencies. In the FDA’s case, one of these is the thalidomide episode, which shaped its current form, though perhaps not with the agility we'd design from scratch.

Ross Rheingans-Yoo: Absolutely. I’m happy to recount that story. It actually begins before thalidomide and ties into a 30-year cycle in the FDA's evolution. 

[Patrick notes: I successfully nerdsniped Ross into writing about thalidomide. Success!]

Ross Rheingans-Yoo: We’ll link this story in the show notes, but the concept of government oversight for product safety originated around the turn of the 20th century, with events like Upton Sinclair’s The Jungle in 1906. Sinclair aimed to highlight the appalling conditions faced by meatpacking workers, losing fingers to industrial machines, but readers fixated on the shocking revelation that human fingers sometimes ended up in sausages. This unintended reaction created a public uproar. The Jungle wasn’t the but-for cause but a symptom of broader issues within the system. Soon after, the Pure Food and Drug Act was enacted, marking the government’s first steps into regulating food and drug purity.

This initial regulatory stance focused on purity: ensuring that products contained only what they claimed and nothing undisclosed. However, relying on purity alone had its limitations, as we’ll see.

In 1937, the S.E. Massengill Company introduced a revolutionary new product—elixir sulfanilamide, a liquid form of the popular antibiotic sulfanilamide. The elixir consisted of pure sulfanilamide dissolved in pure diethylene glycol with pure raspberry flavoring for palatability. Sulfanilamide was a widely used antibiotic and effective by the standards of the time.

The problem is that diethylene glycol. It was extremely effective at dissolving things and widely used as an industrial solvent. Unfortunately, it is also a deadly poison. The result was disastrous: this seemingly "pure" product led to hundreds of deaths. The FDA mobilized its entire staff to locate, impound, and destroy all bottles to prevent further poisoning. Massengill faced fines totaling $20,000 and civil suits amounting to $150,000—penalties that were insufficient given several hundred deaths. This was unacceptable by the policy standards of the U.S.

The elixir sulfanilamide disaster exposed a major gap in policy: the focus on purity failed to address product safety. Consequently, one year later, in 1938, the Federal Food, Drug, and Cosmetic Act was passed, establishing safety testing requirements before drugs could be marketed. Now, companies had to prove that products didn’t harm people—not just that they were pure. The act introduced the investigational new drug application process, requiring that companies disclose studies to the government and restricting drug distribution to qualified investigators. (No more free samples.)

[Patrick notes: Ross wrote, after seeing the transcript:

Unfortunately I think I got this slightly wrong on air and while the 1938 FFDCA did require companies to limit their samples to qualified investigators, they didn't have a requirement to register with the US (i.e., IND) – that was only introduced by the 1961 Kefauver-Harris Amendment instead. That regime of "limit samples, but without registration" was the one in which 17 American children were born with thalidomide-induced birth defects, and – the level of medical accidents is a policy choice – was effectively tightened by some amount by the imposition of the registration requirement.

]

Fast forward twenty years: a German company develops a mild sedative—a blend of thalidomide, quinine, vitamin C, and aspirin. This compound was marketed to relieve nausea, anxiety, and promote sleep, especially for pregnant women experiencing morning sickness. Thalidomide was licensed in much of Europe as a remedy for insomnia, colds, and headaches.

In 1960, the drug was submitted to the FDA for approval. The application landed on the desk of Dr. Frances Kelsey, who had previously consulted for the FDA (23 years prior) on elixir of sulfanilamide’s post-mortem investigation. Now, after a career at the University of Chicago, Dr. Kelsey joined the FDA, and her first assignment was to review Richardson-Merrell’s application to sell Kevadon—the brand name for thalidomide—in the U.S.

Frances Kelsey would later write that while thalidomide appeared non-toxic in animal and human trials, there wasn’t sufficient information on its absorption, distribution in the body, or the potential for concentration buildup. So she sent the application back to the company with a “very nice; try again” note. The company pushed back, essentially saying, “What do you mean? This isn’t the standard applied to other drugs.” They expected the typical approach: submit studies showing it was administered to people who remained safe after 90 days. Blood concentration data was, in their view, above and beyond.

But Kelsey’s academic background had included work on anti-malarial drugs, which metabolized differently across species. In her research, she’d studied a drug that did stop malaria in rabbits, but pregnant rabbits metabolized it dramatically more slowly, causing toxic buildup.

With this in mind, she held firm, insisting that she was the bureaucrat whose approval was required and that she required drug concentration studies from the company. After a couple of months, the company resubmitted—not with a new study, certainly not only 2 months later, but additional testimonials. Once again, she rejected it, repeating her requirement for blood concentration data.

This back-and-forth continued for the better part of a year. Meanwhile, studies in Europe began showing adverse effects in thalidomide patients. The British Medical Journal reported cases of peripheral neuropathy—nerve damage—in elderly patients taking thalidomide. Eventually, the European medical establishment discovered that the drug was causing birth defects among pregnant users in Germany.

(As an interesting aside: The German regulatory state’s ability to See birth defects was impaired at the time. Germany had the recent history of committing monstrous crimes against perceived impaired people and so twenty years later institutionally discouraged anyone from looking into birth defects.)

As a result of this confluence of factors, thalidomide had been administered to thousands in Europe, but relatively few in the U.S. While some investigational samples had leaked out domestically, Frances Kelsey’s insistence on stricter testing effectively kept it off the U.S. market. Tragically, the severe impacts—birth defects, miscarriages—were most prevalent in Europe, where the drug was commercially available. This case struck a powerful public chord: something clearly had to be done to prevent a similar disaster in the U.S., just as the sulfanilamide crisis had catalyzed change 25 years prior.

Kelsey’s persistence on requiring concentration data ultimately paid off, especially since thalidomide was found to build up in pregnant women specifically. Had the policy response maximized for conciseness, the FDA might have formalized her approach by making such scrutiny mandatory. But the political climate in the U.S. at the time was uniquely shaped by Senator Estes Kefauver. In his final Senate term, Kefauver was on a campaign to reign in pharmaceutical profits, alleging that companies were overstating drug efficacy and profiting excessively. This sentiment wasn’t new; anti-Big Pharma rhetoric had been simmering since the late 1950s.

Against this backdrop, thalidomide’s tragedy across the Atlantic provided Kefauver with the political momentum he needed. The bill that his staff had ready required pharmaceutical companies to prove not only the safety but also the efficacy of new drugs through repeated studies. 

[Patrick notes: Pace my conversations with Dave Kasten, it is enormously useful, if you perceive there to be an upcoming window for legislation, to have draft copies of the legislation already written when that window opens. Write the binder or someone will write it for you, and their version will be the one adopted, because legislation has a lead time associated with it and “you” will be unable to produce a compelling alternative from a standing start if you wait until the window is open. (I’d note that Dave et al have worked on an AI regulation white paper for that reason, among others.) ]

Modern regulatory challenges and patient advocacy

Ross Rheingans-Yoo: The bill passed quickly after the thalidomide scandal but was remarkably disconnected with what went right and wrong with thalidomide in the U.S. It was more a culmination of ongoing frustrations about drug pricing and efficacy claims.

The FDA later enshrined much of Kelsey’s approach in their review protocols. Kelsey herself became a prominent figure at the FDA, eventually ending up Director of Scientific Investigations. She has extraordinary influence in the practice of the FDA’s regulation.

Meanwhile, the Kefauver amendments mandated rigorous efficacy testing, a move aimed at tempering Big Pharma’s influence. This emphasis on efficacy persisted through the 1990s until the AIDS crisis, when activists argued that lengthy trials delayed access to potentially life-saving treatments. Their advocacy ushered in the accelerated approval pathway, allowing drugs with preliminary efficacy indications to reach patients faster.

This advocacy model proved effective, and soon, patient advocacy groups became an almost required part of the approval process.

Over time, these grassroots movements gained backing from pharmaceutical companies, leading to a symbiotic relationship between patient advocacy and industry interests. This leads to a for-profit/non-profit industrial complex due to mutual incentive compatibility. Now, for many drugs, securing support from an advocacy group has become almost essential to gain FDA attention, marking a shift in the political economy of drug approval.

This started from a true and noble place, but there is an element of potential astroturfing which happens. As more and more applications before the FDA are attached to well-funded patient advocacy groups, the FDA is increasingly disincentivized to give the time of day to any application which is not advocated for, and so spinning up an advocacy group becomes a cost-of-doing-business for a drug development startup.

[Patrick notes: I have to say explicitly here, though I didn’t want to interrupt Ross in flow: We have choices as to the political economy of drug regulation. Through historically contingent pathways, we have a certain amount of considered-to-be-necessary inefficiency that you can circumvent with a waiver, and we have a prioritization-through-recruiting-compelling-patient-advocates system which weights towards socially esteemed patients who are easy to organize and who photograph well. We could choose other systems.]

Ross Rheingans-Yoo: The COVID pandemic, 30 years after the AIDS crisis, has sparked another wave of regulatory reconsideration. Inside the FDA and broader health establishment, there’s a growing push to streamline drug access further. A noteworthy development is the acceptance of data from international trials. We were chastised, justifiably, when most successful COVID trials for therapeutics happened outside the U.S. We know we cannot accept “only those trials that happen in the city of Boston.” 

However, this shift has yet to permeate fully through the regulatory structure and consulting circles. We’re still seeing history play out.

And so you brought up thalidomide from the 1960s, contextualized it with the history of the 1930s, saw in the 1990s the reaction to the thalidomide cycle, and might now be in the early years of the post-COVID cycle.

Reviving abandoned drugs

Patrick McKenzie: Thank you for the history lesson and the insight into how we got to the current regulatory landscape. Hopefully, things will improve with focused efforts.

You’ve mentioned that there’s significant untapped potential in the system.

And, since I invested in your initiative, I know you’ve found a drug that had essentially been abandoned. How does something like that happen?

[Patrick notes: It is my general practice to disclose angel investments where relevant, though depending on the company they don’t necessarily make the public list until they exit stealth mode. I don’t do much biotech investing, for very predictable reasons.]

Ross Rheingans-Yoo: Absolutely. This is a great example of how the pharmaceutical field isn’t as meticulously scrutinized as we might assume. Sometimes promising treatments fall through the cracks. In this case, it’s a story of how something stuck in the system can get “unstuck.”

Zooming out for context: we discussed the Stop COVID trial out of WashU in the U.S., and the TOGETHER trial in Brazil, which tested multiple drugs in different arms. Among them was fluvoxamine, a generic, off-patent generic drug that, as a side effect, reduces inflammation. The trial showed it reduced hospitalizations by around 30% in unvaccinated, untreated patients.

Another drug tested in the TOGETHER trial was a new, on-patent biologically derived a modified immune-signaling hormone designed to activate the immune response with reduced inflammation. The goal was to bolster the immune system against viruses that survive by suppressing this pathway.

This drug’s Brazilian trial showed it reduced hospitalizations by over 70% in a largely vaccinated population. This was around the time I was working with the Foundation, and we approached the company, exploring what support might be needed to develop this potentially crucial drug. Even if Paxlovid/fluvoxamine is a solution for many, there are patients for whom it’s unsuitable, or where supply is limited. Having another effective tool in the arsenal seemed essential.

Without going into excessive detail, the political climate around emergency use authorizations had shifted by then. The FDA advised the company that an emergency approval was unlikely in light of available treatments. They suggested the company seek full approval, which would require a new study, including a partially U.S.-based cohort and meeting additional criteria.

At that point, the company, amid internal strategic turmoil, decided to pivot away from this drug—originally a passion project by one of its directors—to pursue another development path.

Patrick McKenzie: Just to call back to a previous episode [Patrick notes: probably best discussed with Dwarkesh on his podcast actually], shifting political winds impact government systems and other institutions alike. Back in January and February of 2021, there was significant enthusiasm among funders to support VaccinateCA. One key driver was that many staff at funders (and their ultimate money sources, who are not the same literal humans who sign the checks) were anxiously awaiting their own COVID vaccinations.

Fast forward just three months: people in our social circles, who had prioritized this issue, had received their shots for themselves and their loved ones. Though millions were still unvaccinated, funder urgency dropped from “this is my top priority” to “COVID vaccines are a solved problem, right?” Interest shifted toward other nations, other diseases, and other causes.

So, tangent aside, we’re often stuck with decision-making frameworks that aren’t ideal, but they’re what we’ve got.

Innovative approaches to drug trials

Ross Rheingans-Yoo: This story, though, has a somewhat belatedly hopeful twist. Earlier this year, in 2024—around two years after the drug’s initial success in Brazil—the company found itself at a crossroads. After attempting a different commercial strategy that fell short, they entered Chapter 11 bankruptcy. Their plan was to sell off assets, repay creditors, and potentially dissolve. Fortunately, they ended up recovering substantially more than needed, enabling them not only to cover creditor obligations but also to issue a significant dividend to shareholders, which exceeded their pre-bankruptcy valuation.

In the process, they auctioned off several active and dormant programs. Third and fourth on the list this immune-signaling drug and another treatment for hepatitis D.

Hepatitis D is a rare, chronic liver disease with around 100,000 patients in the U.S. and 200,000 in Europe. Since pricing for on-patent drugs in Europe is typically about 50% lower than in the U.S., the two markets end up fairly balanced.

The drug we’re discussing was at the finish line! It ran into issues with the dosing regimen used in its pivotal study. While it initially resolved the virus in some patients, a portion saw the virus return over a 12-month continuous dosing period. This caused the average efficacy curve to drop, which led the company to abandon the program. They sold it alongside the COVID drug we discussed earlier.

Interestingly, a scientific group bought both drugs. One of their board members, whose lab at Stanford originally developed the hepatitis D treatment, believed it had potential if administered in pulses rather than continuously. They also saw potential in the COVID drug for broader applications like flu, which has few highly effective treatments.

[Patrick notes: One thing which the COVID experience reminds us of: the flu, while a nuisance and considered mostly unavoidable by most people, kills many people every year. It is considered not an earthshattering emergency because most people it kills are older and because the flu is long established in our society. But, on the merits, we should urgently prefer ending the flu, and the right size for investment in that goal by society is easily in the tens of billions of dollars.] 

Ross Rheingans-Yoo: With the goal of demonstrating this potential, the team raised funds to buy these programs out of bankruptcy. Thanks Patrick for being a small investor in that effort. We began pursuing additional funding for the remaining $40 million required to advance these drugs. This figure seems modest compared to industry norms, where some would say, “No, you’re missing a zero if you want to get two unapproved drugs to market.”

What excites me about this field is that, even in the final approval stages, there are so many strategic levers to pull—from targeting the right patient subgroups to optimizing the locations and methods for conducting trials. For instance, we could choose to give a cancer drug only to patients meeting specific conditions, yielding a stronger effect, or aim for a broader group, where efficacy might be diluted but patient recruitment is faster and easier.

Additionally, practical questions like where trials should be conducted to reduce per-patient costs, and how rigorously we retain participants from the very first interaction, are often overlooked. Yet these small, cumulative decisions impact everything, from funding requirements to trial timelines. However, many in the industry prefer the safer option: replicating past successes, even if it means spending five times more. Why look to Brazil when you could run trials at a top-tier hospital in Boston? There are compelling answers to that, but it’s an ongoing balancing act between risk and opportunity.

If you conduct a trial at the world’s top hospital in Boston, you’re competing with countless other studies for participants. For that hospital, your study might not be the week’s top priority. But if you shift to a medical institution in, say, Minas Gerais in Brazil, or hospitals in Central Asia, Poland, or Romania, your trial can become the most significant event happening there. You’re likely to get higher-quality data, more attentive study oversight, and lower medical care costs. However, going this route involves explaining your unconventional approach to investors and navigating perceptions about risk.

There’s an enormous untapped potential in making choices better aligned with each drug’s unique profile. 

Take COVID drug trials, for instance: studying them with an oncology-style approach—overly specific patient filters and limited recruitment—would be a misstep. For COVID, the ideal setup is a primary care clinic where anyone presenting with a cough can receive the drug, with initial safety confirmed, and results returned from the lab a day later. You’d then follow these patients for symptom resolution and long-term side effects. Done this way, you could recruit hundreds of patients daily and wrap up the study in months.

Contrast this with oncology trials, which typically involve narrowing the patient pool through countless filters—prior treatment history, unique biomarkers, and more—until you’re targeting a tiny population. For example, to get 100 patients out of a possible 3,000 nationwide who meet specific criteria, you might need a recruiter stationed in one hospital’s lobby, hoping to catch the one single qualified patient that you know receives care at that hospital. Due to medical privacy laws, you can’t directly contact them, so your best chance is intercepting them in person.

Patrick McKenzie: Just to clarify, this isn’t hypothetical. This actually happens in the world we live in.

Ross Rheingans-Yoo: Yes, absolutely—this is exactly how things operate in our world. Patient recruitment is a job for specialists. If you’re a startup aiming to bring a new drug to market, you might hire a recruitment firm specifically to find trial participants. These firms will charge a five-figure sum per patient, which helps cover the recruiter’s time, sitting in a hospital lobby in hopes of intercepting the right patient. Soon enough, you’re budgeting five figures per patient, calculating how many you’ll need, and maybe even cutting corners on trial size due to costs. Starting from that mindset, you might assume a patient recruitment firm is simply a given.

Consider a COVID trial, where costs initially seem lower than for an oncology drug, which might run into hundreds of millions. If your baseline is $100,000 per patient, aiming to reduce costs becomes a stretch exercise—perhaps $200,000 down to $90,000 per patient. Yet even $90,000 per COVID patient is astronomical when you could recruit them directly at a primary care clinic.

Ethically, you must deliver high-quality medical care, but that should cost thousands per patient, not close to six figures. Meanwhile, financial reports from major pharmaceutical companies show that developing a leading COVID treatment required trials costing well into nine figures for a 2,000-patient study.

Innovative approaches to drug trials

Patrick McKenzie: Yeah, it’s a bit of a paradox—we have a highly advanced industry with brilliant people, yet some of the “free energy” available involves understanding elementary statistics.

There’s a recurring criticism that tech has taken the brightest minds of our generation to optimize ad clicks. But it turns out that understanding conversion funnels—even for ads—can be directly applied to patient care, as we saw with our minor contribution to Stop COVID, where Keith Perhac optimized our online funnels. [Patrick notes: Keith is a long-time friend of mine, and before moving to Portland was the other American software entrepreneur in Ogaki, Japan. We staffed him specifically on the WashU COVID trial landing page. And thus the bingo card guy and an analytics entrepreneur ended up with some small contribution to cutting-edge medical research.]

Patrick McKenzie: Hopefully, in a few years, we’ll be able to check back and see if there’s been real patient-facing impact from taking this shelved drug and pushing it through new trials.

Ross Rheingans-Yoo: We’ll see, and I’m hopeful. There’s a lot of potential here, even if it’s not always clear how to systematically pursue it. My three years since pivoting into this space have shown me that meaningful progress can be made by simply looking at the actual bottlenecks. Instead of treating each challenge as just another generic problem, giving focused, interdisciplinary attention—incorporating insights from both tech and medical fields—can sometimes unlock solutions. There’s untapped potential in blending these perspectives, and it has the power to save lives.

Patrick McKenzie: So, on that optimistic note, Ross, where can people follow your future adventures online?

Ross Rheingans-Yoo: Sure. I'm on Twitter at _RossRy. (I did not join Twitter early enough.) And I blog online at blog.rossry.net. I write a bit about pharma and intend to write more. [Patrick notes: If you want to hear more from Ross, and help me nerdsnipe him into writing more about these subjects, sign up for his newsletter.] 

Patrick McKenzie: All right, Ross, thanks very much for the conversation. Everybody else, I'll see you next week.

Ross Rheingans-Yoo: Thanks Patrick, it's been a blast.