We Need New Cures
Improving our habits and behavior is undoubtedly a massive opportunity to proactively improve human health, for pennies on the dollar. At the same time, it's not a silver bullet. Many diseases aren't preventable with a change in behavior, and for those that are, we are still far away from "solving" the behavior change problem. It's hard. Some people might respond really well to notifications from their Fitbit to get out and walk. Others might be unable to find time to exercise between working two jobs and raising kids. We need to improve both behavior and biology; this post is about the latter.
The biopharma industry has a bad reputation, and there have been egregious abuses, with Purdue Pharma providing a clear example of what can go wrong. However, at a fundamental level, it's an industry that uses rigorous evidence to discover new therapies that mitigate or cure disease. From vaccines and antibiotics to insulin and chemo, the benefit to humankind has been immense. Plus, it's a lot easier for the average person to take a pill than go for a run, and even then, there's a whole industry built around medication adherence to ensure people don't forget. Drugs can be expensive, but once the patent expires and the drug goes generic, they're (usually) very cheap and provide benefits for generations to come.
In healthcare, incentives are everything, and in this case they're actually pretty aligned. If you can get more effective drugs to market faster, you'll improve the bottom line of pharma companies. But they have a big problem. With patents expiring every 10-15 years, the clock is always ticking to discover new "blockbuster" drugs, and over the last couple decades there have been fewer and fewer new drugs making it to market in spite of hundreds of billions invested in R&D. This isn't just a result of the administrative bloat and rising costs that the rest of the healthcare industry has been suffering from. The low-hanging fruit for new drugs has been plucked. Penicillin was discovered by chance, from fungus floating in through a laboratory window, but now virtually every bioactive compound has been screened against every drug target known to humanity, and it has become an increasingly uphill battle.
Sir Alexander Fleming, the scientist who discovered penicillin in 1928. The bioactive compound was in a fungus floating through the air. It’s unlikely the next game changing drug will be found in a similar way. More here.
We're far from curing human disease. Cancer, Alzheimer's, heart disease, diabetes doggedly continue to be top killers. Aging, the ultimate risk factor for disease and death, has continued to resist our best efforts. Clearly there is much work to be done.
Drug discovery and development is a convoluted pathway, but we need to understand the problem if we're going to improve the situation.
Why Is Making New Drugs so Hard?
It's an understatement to say that biology is incredibly complex. Check out this map of a subset of human metabolism. Humans have invented some things with extremely high complexity, like silicon chips (10B+ transistors) or computer operating systems (10M+ lines of code). Those are systems that we can build and understand from the ground up. The human body is many, many times more complex, and we aren't even able to rationally design a simple, single cell from the ground up, never mind an entire organism.
An extremely cool composite image of a cell. Good luck trying to find a single molecule that impacts this system in exactly the way you’d like. Reference
We've had some great successes over the last century, but it looks like we’ll need to level up our ability to understand and wrangle this complexity before we can see another big wave of progress. Luckily, some are in the works, which I'll dive into in a future post.
How It Works
The long road to FDA approval - one approved drug takes roughly $1.5-2B, and up to 15 years from the early stages of research. Reference
Step 0: Basic Research
This is where we discover those fundamental leaps forward. People sometimes accuse this basic research of being a frivolous waste of public funds, but fundamental discoveries are rarely the product of directly, pragmatically trying to attack a problem. Relativity was discovered from a physicist's thought experiment about turning on a flashlight while going the speed of light. CRISPR was discovered by a biologist investigating how bacteria ward off invading viruses. Through the disinterested application of the scientific method, we uncover new knowledge that can transform our world. This applies to biology as well. We need to discover that the immune system exists, and how it works, before we can develop fancy CAR-T modifications to cure cancer.
Step 1: Drug Discovery
This is academic work that's more pragmatic than the previous step. Diseases are markets, and you create value by creating significant benefit for a large number of people. Of course, it should be a basic human right to be able to afford these cures, and how to pay for drugs as a society can be the topic of a future post. Once you've chosen a disease, you generally focus on a specific protein that's in the critical pathway of the disease, and try to inactivate it with a new drug. You might try to figure out the structure of that target protein to rationally design a molecule that can tightly bind to it, like a key in a lock. You might just throw known bioactive compounds at it and see what sticks. If you find something that binds, you have a great new drug candidate. Congratulations, this might be enough of a starting point for a new biotech company!
Step 2: Preclinical Research
Big deal, you have a molecule that binds to a protein. Does it actually cure the disease? Does it have undesirable side effects? Since biology is so complex, you just have to try it and see. You can develop a cell culture to model the disease of interest, and apply your drug to see if it cures the cells. You can take a mouse, nuke its immune system, graft on human tissue, and give those tissues the disease you're studying, and then administer your drug to see what happens. Was the mouse cured? Did it die or have bad side effects? If these data look good, congratulations! Submit your IND application to the FDA, and if it's approved, your drug is ready for clinical trials. This can actually be enough to go public or get acquired by an established pharmaceutical company - there's still a lot of uncertainty ahead, but this is now a promising and valuable drug candidate.
Step 3: Clinical Trials
Now we need to see what happens in humans. First, give the drug to a small group of people to ensure it's safe (Phase 1), and then give the drug to larger and larger pools of people to ensure it works (while still monitoring for safety). The trial should be randomized and controlled, where one group gets a placebo (control) and another gets the drug (treatment), so that any changes you're observing are actually attributable to the drug. Then you need data on whether the drug is working, and are patients having any negative side effects, and compare the distributions between the treatment and control. If there's a statistically significant effect, and no major downsides (relative to the benefits of the drug), then you're well on your way to approval.
This actually sounds pretty simple, right? If I wanted to run a study on whether melatonin impacted sleep quality, maybe I could take out facebook ads to find 1000 people with Oura Rings, mail them a melatonin pill or a placebo, get them to upload their sleep quality data, and pay them $50 for their time. It might take one month to recruit, one month to gather the data, and a week to analyze the data. Maybe each user takes $30 to acquire with ads, plus the $50 compensation and $1 for the pill, so you'd need $81000, plus the time for a team of 4 to manage the logistics, engagement, and data analysis, so call it $100k-$200k all-in to run such a trial.
This is healthcare, so we're going to need to add some zeros. Phase 3 trials with 1000s of patients regularly cost hundreds of millions of dollars (yes, that works out to $100k+ per patient). How is this possible? Finding appropriate patients for trials is extremely expensive. Running trials can require many hospital visits and expensive tests. Dealing the with FDA is a lengthy and expensive process. You need to compensate patients for taking risks with their health to further your research. Finally, since so much is riding on each trial, ironically there's an impulse to over-spend, gathering extra data just-in-case, controlling every variable in case it reduces your data quality, and paying top dollar for services that could be provided more cheaply. There's a lot that can be done to bring down these costs, and today this is a ripe area of innovation (that I've been diving into as well).
Step 4: Approval
If you get through Phase 3 trials, and can say with confidence that your drug works and doesn't cause excess harm relative to the benefit), the FDA will give their blessing, and you can start selling your drug!
Dealing with the FDA might be a long and painful process, but let's keep in mind that it's a very good thing that we as a society have such a high standard for safety and ethics for products that could cause significant harm. The FDA was first established after the tragedy caused by the S. E. Messengill company in 1937, when they sold a drug to people without even testing in animals first, causing over 100 deaths (including the suicide of the chemist who was responsible). The owner infamously commented that, "We have been supplying a legitimate professional demand and not once could have foreseen the unlooked-for results. I do not feel that there was any responsibility on our part." Counter-balances can be a very good thing. With that said, we certainly should try to make these regulatory processes as efficient as we can.
You’re a Survivor
If you've made it here, congratulations! Only about 10% of drugs entering Phase 1 clinical trials will actually make it to FDA approval, and getting that one approval generally requires $1.5-2 billion of investment over the course of 10-15 years. These are some serious numbers. This builds some empathy for why drug companies sell drugs at a significant markup - it costs pennies to manufacture the drug, but they also need to recoup these massive R&D costs while funding research for the next generation of drugs. The net margin for many pharma companies is 15-20%, profitable but not egregious by any means.
The most important thing we can do is to improve our rate of learning about biology, so we can discover more effective cures for more diseases more quickly. This means increased investment, better tools, and bold new approaches. New tools like gene and cell therapy are already providing promising and fundamentally new approaches. It's also becoming increasingly clear that we will need to lean on software and machine learning to understand the complexity of biology and do some of the heavy lifting for specific problems. As always, better datasets, better algorithms, and more compute will be crucial.
The next step will be to make the validation process faster and more efficient. In the short term this means providing better access to trials, administering more of the trial at home, and using modernizing a very paper- and people-driven process. In the future, moving more and more of the trial process to simulation will be inevitable, making the approval process nearly instantaneous.
I’ve been diving into the world of clinical trials, and am excited to share more in future posts about how they currently work and the many ways we can bring them into the modern era.