How to Fix the Incentives in Cancer Research

Ned Sharpless is director of the National Cancer Institute.

Ned SHARPLESS: So when President Nixon signed the National Cancer Act in 1971, the thinking was that we would be able to give the president the cure for cancer by the bicentennial, 1976.

Thus began the so-called War on Cancer. It was supposed to be a short war.

SHARPLESS: There had been this really amazing success against pediatric leukemia in the late 1950s and 1960s. And so the thinking was, “Well, if you can cure leukemia like that, we’re just going to do the rest of cancer.” But that didn’t happen.

Why didn’t that happen?

SHARPLESS: Well, it gets to this difference between science and engineering. The engineer sees a problem — “We can build a rocket in 10 years and we’re going to go to the moon.” It was sort of that thinking, that cancer’s a problem and we need to develop a treatment and we’re going to devote a lot of resources to it. But what we had to do really — and are still doing today, frankly — was decades of intricate basic science to really understand the basic biology of cancer. 

Basic science and basic biology perhaps — but overall, cancer has turned out to be incredibly complicated.

SHARPLESS: When I started in this business in the 1990s, we thought of cancer as a couple of diseases. There was breast cancer. There was lung cancer. There was leukemia. The essential paradigm shift in the last few decades has been that cancer really isn’t a small number of entities. It’s hundreds or thousands of diseases that each have their own unique epidemiology and treatment and causes and survivorship challenges. 

Stephen DUBNER: The discovery that the biology was much more complicated than originally thought, what was that like for clinicians — and researchers — who really thought they knew what was what?  

SHARPLESS: I can tell you how it happened for me. I’d been treating patients with cancer. I’d certainly taken care of many women with breast cancer. And then this paper came out that used this new technology — expression profiling — to look at all the R.N.A.’s that a breast cancer makes. And what they showed was that breast cancer was like five diseases. There was this kind of breast cancer that was very common, it may have been 25 percent of cases, that was hiding in plain sight our entire careers that no one had ever really appreciated. And I was like, “Wow, if breast cancer is this complicated, what’s it going to be for lung cancer, or leukemia, or colon cancer?” So each of those diseases requires a different way of thinking. And that paradigm’s been very successful. We’ve started identifying specific kinds of cancer and made a lot of progress against certain types. But other types of cancer are harder to treat. And so we see these marvelous advances benefit some, but not all, patients. And that creates this problem that there’s this unevenness.

These thousands of diseases combine to form the second-leading cause of death in the world, after cardiovascular diseases. In a given year, nearly 10 million people around the world die from cancer. The Covid-19 pandemic will likely result in additional cancer deaths — due to a huge decline in cancer screenings as well as delays in treatment. And those delayed treatments are expected to cost much more, since those cancers will be more advanced once they are detected. On the other hand, the pandemic has disrupted the healthcare system in ways that may benefit cancer treatment in the long run. We’ve gotten more accustomed to telehealth; clinical trials are being done remotely; and there’s a chance that the wildly fast development of Covid vaccines may change the structure, speed, and funding for cancer treatment. Today on Freakonomics Radio: there are other reasons to be optimistic about cancer.

SHARPLESS: There’s this pace of progress like no other period in biomedical research for any disease.

But of course there are still plenty of challenges: 

SHARPLESS: The average American citizen would be surprised by the level of fragmentation in medical data.

And, a new way to approach one of the deadliest cancers:

Diane SIMEONE: “Okay, if clinical trials aren’t working, how do we fix them?”

*      *      *

The National Cancer Institute, founded in 1937, is part of the National Institutes of Health. Because the N.C.I. is the oldest constituent agency and because it deals with such a common and devastating set of diseases, it has strong bipartisan support. It also receives its funding separately from the N.I.H. — about $6 billion a year — which affords a certain amount of self-determination when setting its agenda. The N.C.I.’s mission is to identify, fund, and conduct the most worthwhile cancer research.

SHARPLESS: We certainly have some really great ideas that we are unable to get to. So I spend a lot of time worrying about how we can prioritize funding to make sure we always fund the best science.

Ned Sharpless has been N.C.I. director since 2017. For seven months during that stretch, he was acting commissioner of the Food and Drug Administration. Before his government service, he spent years treating patients and conducting biomedical research; he was particularly interested in the relationship between cancer and aging. This was at the University of North Carolina in Chapel Hill. Sharpless also co-founded a biopharmaceutical firm, and he owns 10 oncology patents. When he took the job at the N.C.I., he took his lab with him. As he said at the time, “It’s very important to me. … It’s nice to understand the problems of a working scientist.”

SHARPLESS: That is correct.

DUBNER: You grew up in Greensboro, yes?

SHARPLESS: Yes, true. North Carolina.

DUBNER: You seem to have no southern accent. What’s that about?

SHARPLESS: So I can do one if needed, first off. Secondly, I did my residency and fellowship in Boston. You say the word “y’all” on rounds exactly one time — rounds stop. Everyone makes fun of you for 20 minutes, and then rounds resume again. And so yeah, it beats it out of you pretty quickly. 

DUBNER: You’re one of the few people who’s had experience as a researcher, an entrepreneur, the head of a large research institution in charge of grants, and head of the regulator for drugs and medical devices. Having worked in all those areas, how do you see things differently than someone less polymathic than that?

SHARPLESS: The thing that has become very clear to me — it’s a nuanced and hard message to tell — but there’s this pace of progress in cancer research that is really exceptional, like no other period in my life — and maybe like no other period in biomedical research for any disease. But it’s hard to talk about because it’s so heterogeneous. You could focus on any specific area and say, “Why is progress slow here or there?” But in aggregate, what’s happening is really amazing. You know, when I was at the F.D.A., about 30 percent of the business in terms of new approvals for devices and drugs was cancer-related. It’s really remarkable.

Over the past three decades or so the death rate from all cancers in the U.S. has declined roughly 30 percent. In 2018 alone, the most recent year for which we have data, the decline was 2.4 percent. That’s the biggest single-year decline ever recorded. This is all good news, but it’s also a long, long way from the Nixon-era hope that cancer was about to be “cured” within five years. There are, however, some things hidden in the cancer data that makes that 30-percent decline even more impressive. For one, the survival rate for younger cancer patients has improved dramatically. And one reason so many older people are still dying from cancer is that they’re not dying from cardiovascular diseases, thanks to a huge drop in mortality there. In other words: many people who in previous generations would have died from heart disease are now living long enough to die instead from cancer.  

SHARPLESS: But, you know, as I said, it’s uneven. There are certainly cancers where we’re making a lot less progress.

Okay, where is less progress being made? From 2012 to 2016, death rates for women decreased for 13 of the 20 most common cancers (including lung, breast, and colorectal) but increased for 5 types (including cancer of the uterus and liver). For men over that same period, death rates decreased for 10 of the 19 most common cancers but increased for six (including liver and non-melanoma skin cancer). 

SHARPLESS: A lot of those data go to 2016, which precedes a lot of the widespread use of immuno-oncology drugs. So as good as those data are, they don’t include a lot of the new therapies that we’ve developed. 

Immuno-oncology — harnessing the body’s own immune system to treat cancer — this has a long history. In the 1970s, for instance, there was a lot of enthusiasm about naturally-occurring proteins called interferons.

SHARPLESS: It was going to jack up your immune system to fight cancer. And this was going to be this universal cure for cancer. And it was really a failure. It didn’t work. And because of that experience and other experiments like that, the cancer-research community — medical oncologists like me — became very skeptical of the idea that the immune system could treat cancer. And in fact, skeptical is probably not strong enough. We thought a lot of these people working on the immune system were literally crazy. You know, we thought they were harming patients and irresponsible. And so it was really a vilified field for many years. But a few great scientists persevered, and started to identify ways to coach the immune system into fighting cancer. 

This is hardly the first time in medical history that the supposedly crazy people turned out to be brilliant. In fact, it happens all the time in medicine. In 2016, the Nobel Prize in Physiology or Medicine went to two researchers, Jim Allison and Tasuku Honjo, for immunotherapy research.

SHARPLESS: And then those therapies started to work, and so that’s really become a successful and very important way to treat cancer that most in the field, myself included, were very skeptical of in the early days.

DUBNER: Let’s talk about lung cancer for a second. Still the most fatal cancer, but decreasing, yes? I guess you could look at it from either side, that there has been progress, or you could say, “Wow, it’s still killing a lot of people.”

SHARPLESS: Yeah, so the first thing to say is, even today, after a lot of progress against lung cancer that reflects various advances, it still kills more people than breast, prostate, and colon combined in the United States every year. So it’s a highly lethal malignancy where we definitely need to make additional advances. And also one of the major things that shaped lung cancer is the use of cigarettes, is tobacco. And so tobacco control over the last few decades is starting to have some success.

And then on top of that, we have some interesting new developments. We have these drugs called kinase inhibitors that target specific subsets of lung cancer. Maybe 15 or 20 percent of lung cancers in the United States are targeted by these pills that are quite effective and not very toxic. And then we’ve also had the introduction of immunotherapy, so these checkpoint inhibitors that are quite active against an even larger fraction of lung cancer and have produced some really marvelous responses. And so now we’re seeing lung cancer mortality decline at the fastest rate in the history that we have kept statistics about lung cancer.  

Once you start to see cancer not as a disease but a massive array of individual diseases, you can appreciate the difficulty in overcoming it. And if you look at it in reverse — from the perspective of a patient who’s already seriously ill — you see even more difficulties. For starters: not all treatments for a given cancer are effective, and they often have brutal side effects. Then, moving backward: not all cancers are able to be detected in time to treat them. And going back even further: we don’t know that much about what causes all these cancers.

I asked a doctor friend of mine for her definition of cancer. “It’s some inflammatory, infectious process that radically alters the activity of healthy cells,” she said. But what triggers that inflammatory, infectious process? There is a long list of risk factors — with an equally long list of caveats — and these factors generally fall into one of three baskets: hereditary, environmental, and behavioral. It’s unclear how many cancers result from inherited genetic variants, and the percentages range among cancer types. As for behavioral and environmental causes: tobacco smoke, asbestos, ultraviolet rays, heavy alcohol consumption? All very likely carcinogenic — although it’s worth noting that 12 percent of all U.S. lung cancer patients have never smoked. What about artificial sweeteners and charred meat? No clear or conclusive evidence.

SHARPLESS: I mean, take a topic like nutrition. You know, the F.D.A. makes a lot of recommendations around the American diet. How much sodium should you eat? How much sugar should you eat? And that’s a complicated research question. You know, we’re very interested for cancer because we think obesity is one of the major drivers of cancer. But the science there is hard. It requires long studies. It’s hard to do randomized trials, if not impossible. And if you want to make regulatory policy on nutrition science, it’s a tough thing to do.

DUBNER: Can you explain the mechanisms by which obesity could be a driver of cancer?

SHARPLESS: Many possible mechanisms. We don’t think it’s a single thing. Obesity is associated with alterations of certain hormone levels that we think might have a carcinogenic effect. Obesity in some situations appears to be associated with chronic inflammation, which we think can be a driver of cancer. And then obesity might be a proxy for other behaviors that confound this analysis and are really the cancer driver. I think the present data are quite strong that as obesity becomes a greater problem in the United States, we’re seeing more of certain kinds of cancer, like breast cancer and gastric cancer and liver cancer. 

DUBNER: Do they tend to be the solid-organ cancers then, that are predominantly driven by obesity, or no?

SHARPLESS: Yeah, we think largely it’s cancers of the upper G.I. tract, the liver and gallbladder and stomach. Certain kinds of breast cancers associated with obesity. And then also maybe pancreas, although probably less strong for that.  

SHARPLESS: Pancreatic cancer is a good example of one of these holdouts where we have these new exciting therapies and these new approaches, but not yet for that disease. Modeling suggests it may be the second most deadly cancer in the United States soon, after lung cancer. 

DUBNER: Yeah, let’s talk about pancreatic cancer. It seems like it’s of particular concern for at least two big reasons. Early detection is hard, and treatment is not very successful, yes?

SHARPLESS: Yeah, I’d say it has the two problems you mentioned and then it has a third problem, which is it’s not going down, like lung cancer, you know. So it’s one that’s going up in incidence and where we’ve made really no meaningful progress in terms of the mortality of that disease.

SIMEONE: If you look over the last decade, the number of cases of pancreatic cancer in the U.S. has increased maybe about 10,000. 

That is Diane Simeone, a pancreatic-cancer surgeon who also runs a research lab at NYU-Langone in New York.

SIMEONE: This year, it will be about 58,000 people in the U.S. that get pancreatic cancer, or maybe half a million people worldwide.

Those numbers were actually for last year; this year they’ll be a bit higher. Still, the incidence, compared to other cancers, isn’t all that high. That’s one reason why pancreatic cancer was historically not well understood.

SIMEONE: It didn’t really attract a lot of attention from funding authorities such as the N.I.H. There wasn’t a big advocacy group. And so it really was neglected. And almost everybody that got it died.

“Almost everybody that got it died.” And now?

SIMEONE: It has still a single-digit survival rate, of 9 percent. 

Some prominent people who’ve died from pancreatic cancer recently: Ruth Bader Ginsberg, Alex Trebek, John Lewis, and Steve Jobs.

SIMEONE: Pancreatic cancer is typically diagnosed late. Unfortunately, the early warning signals are kind of vague and nonspecific. People might have some upper abdominal pain. They may have some unexplained weight loss. Sometimes pancreatic cancer can present with new-onset diabetes, especially if it’s associated with weight loss instead of weight gain. One of the cardinal signs that tips people off is they get jaundice or yellowing of the eyes.

DUBNER: The pancreas is kind of tucked away. Does that play a big part in the difficulty of detection?

SIMEONE: I think so. It’s not easy to feel on physical exam. We do have ways to get a look at it with either C.T. scans or M.R.I.’s or even a kind of fancy endoscopic ultrasound. But those aren’t tests that people routinely have.  

Cancer screening is its own complicated scenario. Ned Sharpless again:

SHARPLESS: One of the big successes in screening has been colonoscopy for colon cancer and pap smears for cervical cancer. The problem with screening is that it can be good at finding cancers that might not actually harm the patient. The classic example here is prostate cancer, which could be a very slow-growing cancer that may not hurt the patient, but having a big surgery or big radiation therapy would hurt the patient. And so balancing overdiagnosis and overtreatment with early detection and getting rid of bad cancers is a tough challenge.

The challenge with pancreatic cancer is that the most effective screenings are fairly expensive, or invasive, and therefore quite rare; also, by the time pancreatic cancer is detected, it’s often fairly advanced.

SIMEONE: Everyone always wonders what is unique about pancreatic cancer that makes it so lethal. And I would say we still have an incomplete understanding. One thing we clearly know is that pancreatic cancer spreads early — or what we call metastasizes — to organs away from the pancreas. We don’t really know why that is. We do think there’s something unique about the micro-environment in which pancreatic cancer arises and grows, but the exact networks that cause it to be so resistant to therapies are still being worked out.

SHARPLESS: Pancreatic cancer is actually one of the more simple. It doesn’t come in as many flavors as say, lung cancer or breast cancer. But the main flavor it comes in turns out to be a really bad disease where we don’t have effective therapies at present, and we don’t have effective screening at present.  

SIMEONE: In an ideal situation, the treatment would be surgical resection, which can be quite effective.

DUBNER: Resection, meaning you remove that part, not the entire pancreas? 

SIMEONE: Right. You remove that part of the pancreas. And it’s interesting — they do a lot of screenings in Japan, and if you look at their data, with surgical resection or removal of pancreatic cancers less than one centimeter, the five-year survival rate is in the order of 60 to 70 percent.  

But, again, since pancreatic cancer often spreads before it’s detected, even surgery isn’t always an option.

SIMEONE: Only 15 percent of patients that come to the clinic have a surgically resectable tumor.

DUBNER: Outside of that 15 percent who are successfully resected, death happens how fast in the case of pancreatic cancer, typically? 

SIMEONE: Months to a year, year-and-a-half. 

DUBNER: You’ve chosen a subspecialty where I imagine you spend a fair amount of time telling people they’re going to die relatively soon. Yes? 

SIMEONE: Well, I always try to create hope. I think it is important for patients to have hope. But we also have to be realistic. And yes, that is the most difficult part of the job, is to tell someone that we don’t have something that can cure them. And I hate it. I’m driven to change that conversation. 

*      *      *

We were speaking with Diane Simeone.

SIMEONE: I’m a surgeon and I’m also a researcher. I’ve been studying pancreatic cancer since the mid-1990s. When we first started working on pancreatic cancer, there were very few researchers in the field. Now I go to scientific conferences and there’ll be six or 700 researchers there, and that makes a new era for pancreatic cancer.

Simeone now practices at NYU-Langone Health, in New York City. Full disclosure: I got to know Simeone through the treatment of a family member. I was impressed not just with her work as a doctor but her zeal for busting open the paradigm for a cancer that is particularly hard to detect, hard to treat, and therefore often fatal. She is one of the ringleaders of a new collaborative platform to change how pancreatic-cancer research is done, on both the diagnostic and therapeutic sides.

SIMEONE: The straw that broke the camel’s back for me is, the year before we started on this, there were new immunotherapies out, and there were, I think, six different centers doing the same exact single-agent trial to see if it worked in pancreatic cancer, and all those trials failed. And of course, no one is talking to each other. 

DUBNER: And they’re not talking to each other because why?

SIMEONE: We all work in our own institutions. Everybody’s got their own grants. The incentive system is for individual achievement and not group, collective effort. Now, I think that is starting to change. But for really complex problems like what we’re talking about for pancreatic cancer, it felt like we just had to rewrite how we work together. And so we embarked upon this effort and built a new clinical-trial ecosystem.  

This ecosystem is called Precision Promise.

SIMEONE: Precision Promise is what’s called an adaptive-platform trial. And you know, it’s funny, even if I ask clinicians if they know what that means, a lot of people don’t, because it’s really a new way to do clinical trials. 

Precision Promise has been rolled out in collaboration with the Pancreatic Cancer Action Network:

SIMEONE: Or PanCAN, as they’re affectionately known, which is a large nonprofit patient-advocacy foundation. And the nice thing is, they could serve as, an honest broker to help bring everyone together.

This new platform is already enrolling patients at more than a dozen different hospitals and research centers, and hopes to expand to 30 or 40 in the next few years.

SIMEONE: You can’t join this effort unless you’re willing to work together and to share data. So data sharing, we thought, was critically important.

SHARPLESS: Yeah, data aggregation is one of my favorite topics — this keeps me up at night.  

That, again, is Ned Sharpless, director of the National Cancer Institute.

SHARPLESS: I think the average American citizen would be surprised by the level of fragmentation of medical data. That it’s not held by any central repository and easily searchable. So the N.C.I. has long been frustrated by our inability to really see what’s going on and has developed a number of data sets that try and aggregate the national cancer story to make it more understandable. One of the most successful, actually, is this thing called SEER, which was created back in the 1970’s. 

SEER stands for Surveillance, Epidemiology, and End Results.

SHARPLESS: SEER is a registry of who gets cancer and who dies of cancer. But you can link it to Medicare, which has claims data about how they got treated, but still you would like more information than that. You’d really like to be able to look at the medical records of patients and understand, you know, what therapies had they had before? And what risk factors do they have for cancer? And to get that out of the medical record, that turns out to be complicated. 

DUBNER: What value would there be in aggregating all that siloed data? What would it actually produce in terms of research understanding and potentially cancer treatment?

SHARPLESS: I personally believe that would be highly valuable because, you know, once you realize that cancer is lots of different diseases, you also realize that large randomized trials are hard to do when you’re talking about lots and lots of uncommon cancers. Then you really need to learn from every patient. You need to learn what happens to each individual with cancer. 

SIMEONE: So computing power is important. 

Diane Simeone again.

SIMEONE: In fact, as part of this effort, the Pancreatic Cancer Action Network has put together a big data team because we realize the power of this is the data that we’re gonna get from every patient. Not only in how they respond to the therapy, but we also have an imaging team looking at the imaging every patient gets, which is going to be standardized across all the sites, and is there information we can glean there? We’re going to be doing blood-based tests to see, is there a blood test that will tell us more quickly if we should switch therapies for patients until waiting down the road? How do we best treat pain? How do we best treat nutrition? In fact, patients are all going to get a Fitbit to see how activity and sensors can help in patient care. 

SHARPLESS: One area where we’ve seen some progress in lung cancer is using artificial intelligence to look at the CAT scans of people who smoke a lot. Because we have thousands of these patients, we can train the A.I. to look at features of that chest C.T. and say, “This person is at increased risk for lung cancer, even though the radiologist would read this as normal.” So I think that what you would need to do that in pancreatic cancer is thousands of patients with thousands of M.R.I.s who’ve been followed for five to 10 years. We don’t have that data set yet in pancreatic cancer.  

SIMEONE: So one of the things I — while there’s all this great science going on, we hadn’t really innovated in the clinical-trial space enough to have clinical trials that could be as impactful as we needed. If you looked at how many patients with pancreatic cancer in this country are in clinical trials, it’s only 4 percent of patients. And you don’t get new therapies without seeing if they work through clinical trials. And so we said, “Okay, if clinical trials aren’t working like we need, how do we fix them? If we wipe the slate clean, can we try to design a clinical-trial system to take all that great science and change the course of this disease?”

That, at least, is the promise of the Precision Promise platform that Simeone has been building out, with collaborators at a variety of big-time cancer hospitals and research centers. Across the network, there are just two control groups of patients, who receive what’s called “standard of care” treatment — in this case, one of two chemotherapy protocols. The rest of the patients are randomized into one of four treatment groups, each receiving a different experimental therapy. So there’s less redundancy than there is in separate clinical trials. And if one therapy fails early on, patients in that treatment arm can switch to another group. In theory, the trial can go on forever this way, with failed treatments swapped out for newer ones. That is why it’s called an “adaptive-platform trial.”

SIMEONE: This is a randomized trial, but many large-scale randomized trials have the same number of patients in the control arm and the experimental arm. And for this trial, 30 percent of the patients are in the control arm and 70 percent of the patients are in the experimental arms. So that’s one area that’s quite different. The ability for patients to get multiple therapies has never been done before in clinical trials, where both of those therapies can count for F.D.A. approval. This is new territory.  

Simeone and her collaborators in fact reached out to the F.D.A. to help build the Precision Promise platform.

SIMEONE: And they were huge advocates. They even assigned a senior member of the F.D.A. to help us. The other missing ingredient was the drug companies. So we had a hard time making advances in pancreatic cancer because a lot of the clinical trials failed. And so pancreatic cancer became a graveyard for clinical trials. But by building this clinical-trial ecosystem that had so many advantages to it, we actually were able to reach out to the pharmaceutical industry and bring them on board, so that now we have about 30 different pharmaceutical companies that are working with us.

DUBNER: So if a treatment emerges that looks potentially successful, how does this change the timing of F.D.A. approval? 

SIMEONE: It should cut it in half. So it’s really a remarkable incentive for the pharmaceutical industry to get on board. We even got the F.D.A. to allow us to do this concept called a lead-in, where a relatively small number of patients — say, 20 to 30 patients — could be tested and the drug company could look at the data and then make a go/no-go decision with a 60-day waiting period. So it de-risks the pharmaceutical industry from investing in new therapies for pancreatic cancer. 

All these changes — creating collective incentives versus competitive ones; cycling in experimental therapies much faster; getting research data out of the silos of multiple institutions — it’s all especially necessary for pancreatic cancer. It’s not a common-enough disease to generate the mountains of data that researchers would like to have, but it is deadly enough to kill most people who get it. I asked Ned Sharpless from the National Cancer Institute about this new approach to clinical trials. He provided a useful history lesson.

SHARPLESS: So when I was starting out as an oncology fellow, most of our trials were heavily influenced by what I call the cardiology paradigm. These are these massive trials that would have like 1,000 patients in arm A and 1,000 patients in arm B. And arm A would get aspirin and heparin and arm B would get heparin only, right? And they’d follow these patients for years. And then they would see, “Oh, the guys who got aspirin did 3 percent better in terms of the risk of heart attack” or unstable angina or something, than the people that didn’t. And that would become the new standard. And then the new trial would use that regimen plus the next drug. And adding, 3 percent here, 5 percent there — cardiologists made a lot of progress.

This is how cardiovascular mortality fell so much, as we mentioned earlier.

SHARPLESS: One trial at a time. Subtle changes. Large randomized trials over and over again. We tried that in cancer. And it breaks down pretty quickly. 

That’s because there’s so much more variation in cancers than in cardiovascular diseases, Sharpless says, as well as more variation in the cause of each cancer.

SHARPLESS: So this sort of fragmentation from the cardiology paradigm, large randomized trials to these smaller, nimble sometimes unrandomized trials, has been a recurring theme in oncology. So you get allocated to therapy based on the mutations, the molecular genetics of your cancer. That’s really different. And that poses a lot of interesting challenges. One of them is for the F.D.A., right? The F.D.A. is used to seeing 800-patient, randomized-controlled, placebo-controlled trials. And now, the developers of these drugs are coming to them, “Here, I got like 50 patients with myeloma. They got this drug. All of them responded. In the historical control, like a third of them would respond. All of them has got to be better than a third. So therefore, you should approve my drug.”

And, as a regulator, that’s a tough challenge. When can you trust this unrandomized design versus when should you insist on the gold standard of purity? The issue that comes up is this one called equipoise, which is how the F.D.A. refers to the state in which you can no longer accept the hypothesis that the drug doesn’t work. At some point, you pass equipoise and then we think that it’s unethical to not provide that drug to the patient.

DUBNER: When there’s a successful treatment that comes on market, what usually makes the headlines — if it makes the headlines — is the cost, right? So this is a source of common public outrage, that treatments, especially for rare cancers, are extraordinarily expensive. What do we, the public, not understand that you know?

SHARPLESS: You know, drug pricing is a very complicated topic. I think it defies simple solution. But then there are lots of things that have this tradeoff between stifling the innovation of new medicines, because they’re expensive to make, versus these immense costs to patients. So I would make a couple of comments. The first one is, it bothers me as a physician that any patient with cancer has to choose between their medicine and their rent. That happens in the United States, unfortunately. I feel tremendous empathy for those patients. And I think that when we are able to do something about that, we should. You know, approving more generic medicines. That is something the F.D.A. wants to do. That will lower drug prices eventually, although it can take longer than you would like.

The second thing to say is it’s very expensive to make new drugs. It’s really hard to do. Most of them fail. And it really does cost the billions of dollars per successful drug that the pharmaceutical industry always says this is why they have to charge so much. It is true. But let me say one other thing that I think is a really important point, which is that having a drug that works for someone, but which is tremendously expensive, is a better problem than having no drug at all, right? And that’s, in cancer, often what we’re talking about. Once you have an expensive drug that works, that becomes an engineering challenge. How do you make that drug cheaper? How do you make another version of it? And we are much more successful tackling that problem than we are at developing totally new, out-of-the-box therapies.

Therapeutics are one major goal of Diane Simeone’s endeavor. But that is just one goal. As with all therapeutics — and especially for something as vicious as pancreatic cancer — they’re more viable when the cancer is detected earlier.

SIMEONE: So clearly, early detection is the holy grail for pancreatic cancer. There’s been a fair amount of effort to try to develop a test. But I will say it’s been relatively small scale. It’s been scattered, small efforts instead of a concentrated, coordinated one. But there are some moves afoot to change that. 

Moves afoot to change that — and the whole ecosystem around pancreatic cancer: from the root causes to early detection to viable therapeutics, all in the hopes of shifting the curve of one of the most fatal cancers.

SIMEONE: One of the goals I’ve put out there for our field is for all of us to say out loud that we’re going to have a 50 percent survival rate for pancreatic cancer in 10 years. And I think we can do it if we’re strategic, both in developing more effective therapies, but clearly and importantly, by figuring out who’s at risk, better testing for susceptibility genes, and putting people in screening programs. If someone were to develop a pancreatic cancer, it’s found when it’s much smaller, and the ability to resect it surgically can approach 90 percent as opposed to the typical 15 percent we’re dealing with now, I think we can get to that 50 percent.

It may be tempting, when you look back at the last several decades of cancer research, to focus on all the failures. There have been a lot of failures, as there usually are when you’re trying to advance science. One big problem in all scientific research is that negative results are often forgotten, even buried. Ned Sharpless of the National Cancer Institute says that’s a big mistake — but that it’s changing.

SHARPLESS: One of the happy things about the internet era and changes in publication practices is that negative data are getting out better than they used to. So in the old days, you do trials, and it didn’t work, and you wouldn’t publish it. And then someone else would do the same trial and it wouldn’t work. That was bad for lots of reasons. Bad for the patients. Waste of resources. But now we try to make negative data available, particularly when it involves human subjects. I mean, negative data is some of the most important data to me as a scientist. You expect this to work and then you try it, and it fails, that can be one of the most informative things that moves biology.

So I can give one nice example of this. We have been trying to treat a disease called neurofibromatosis in kids for decades. They’re at risk for developing these tumors, and those tumors can turn into cancer, but they can also be very, very disabling. And we did lots and lots and lots of clinical trials. None of them worked. And it was really, really frustrating. But now in the last couple of years, we’ve identified a new therapy and that therapy is quite effective. It’s a pill that kids can take every day. It makes their tumors stop growing, and we think even decreases the risk to turning into real, full-blown cancer later. So it’s of immense clinical benefit. And if we didn’t have all that negative data, if we hadn’t known what doesn’t work over and over again, we wouldn’t have really fully had confidence in this positive result. So the positive result was really empowered by the decades of negative data.

Before we go, let me offer one last reason to think that cancer research and treatment may continue to improve. As devastating as the Covid-19 pandemic has been to our lives and livelihoods, it has also spurred one of the most impressive medical responses in history, maybe the most impressive. At least three successful vaccines have already been developed, in roughly one-tenth the time that a vaccine usually takes. The bench science has been formidable; the clinical trials large and diverse; the regulatory approval, as speedy as it gets. Last August, a few months before the Moderna and Pfizer-BioNTech vaccines were given emergency-use authorization by the F.D.A., we interviewed former F.D.A. commissioner Peggy Hamburg. I asked if she thought that the mechanisms developed and the alliances built in pursuit of a Covid-19 vaccine might prove beneficial for the treatment of other diseases.

Peggy HAMBURG: I absolutely do. I think that some of these advances were already underway to some degree, and we are seeing the real value of systematic attention to the infrastructure for clinical trials, as well as the importance of innovation in how clinical trials are done. Testing multiple different drugs against one placebo arm allows you to learn a lot about a lot of drugs as quickly as you can with more people getting access to one of the potential candidates with just the one placebo arm.

Covid-19, I think, marks a hugely important moment in time when the scientific research community came together across disciplines and sectors and borders in order to collaborate. And I hope we won’t lose that spirit of collaboration, because I think it is absolutely essential. But it requires companies to work together in ways that historically they haven’t always been comfortable.

We also interviewed Tal Zaks, the chief medical officer at Moderna. Though they are now famous for their Covid-19 vaccine, Moderna has also been applying their mRNA technology toward cancer and other therapeutics. I asked Zaks what kind of knock-on effects their vaccine success might have for other treatments.

Tal ZAKS: I think from my perch, looking specifically at mRNA technology, it will have done two important things. We will have proven the ability of this technology to scale up manufacturing, and that scale-up of manufacturing will have implications not just for other vaccines, also for other mRNA medicines coming down the pike. And I think the proof that this vaccine works will translate into a much higher degree of confidence and the probability of success overall for mRNA medicines, as well as the ability to scale them up. I think the broader thing that I hope we all take away from this is the strength of collaboration that this pandemic has forced. I think none of us who are in the throes of doing this will ever go back to business as usual when it comes to relationships and interactions between companies, between government agencies, etc., at least I hope not. 

That’s our show for this week; thanks to all the scientists who took the time to teach us today: Tal Zaks, Peggy Hamburg, and especially Ned Sharpless from the National Cancer Institute and Diane Simeone from NYU-Langone and the Precision Promise project.

*      *      *

Freakonomics Radio is produced by Stitcher and Renbud Radio. This episode was produced by Matt Hickey with help from Daphne Chen. Our staff also includes Alison Craiglow, Greg Rippin, Mark McClusky, Mary Diduch, Zack Lapinski, and Emma Tyrrell; we had help this week from Jasmin Klinger. Our theme song is “Mr. Fortune,” by the Hitchhikers; the rest of the music was composed by Luis Guerra. You can subscribe to Freakonomics Radio on Apple Podcasts, Stitcher, or wherever you get your podcasts.