My first cancer patient
Curing incurable cancer with Darwinian principles, mathematical modeling, and teamwork
Between November 4 and November 8, I've participated in the 9th IMO Workshop, organized by Moffitt Cancer Center, a week-long competitive science event. Five teams were competing for 50.000 dollars of research money. Each team was given a terminally ill cancer patient and one objective: try to come up with a course of action that would save the patient. We had five days to achieve what medical science couldn't achieve in 50 years. In one week I worked more than I normally do in a month, and experienced more scientifically and emotionally than I normally do in a year. This post will be about that crazy week.
This piece is intended to be suitable for a general audience.
The IMO Workshop
The workshop is an annual event of the Integrated Mathematical Oncology (IMO) department of Moffitt Cancer Institute. Experts from all around the world come to Tampa, Florida and work in teams on a project about that year's theme. The prize for the two winning teams is 50.000 dollars of research money. Most attendees of the workshop are biologists, applied mathematicians, computer scientists, and medical doctors. This time there was also an economist. The workshop themes can relate to therapy, e.g. Immunotherapy in 2015, or the environment of cancer like Stroma in 2017, or a specific aspect, like Metastasis in 2012, or Resistance Mechanisms in 2016. This year's theme was Evolutionary Tumor Board. This year, the second time in IMO history, the teams included actual oncologists, and for the first time in IMO history, real patient cases.
The posters of IMO Workshops. Ours is the one on the right.
A tumor board is a group of oncologists and surgeons who convene to review a cancer patient's case. The board's function is to advise the treating physician and come up with a treatment plan. Imagine Dr House and his team but only for cancer patients. An evolutionary tumor board, IMO's own creation, also includes evolutionary biologists for reasons that will become clear shortly.
During the workshop, the teams played the role of such an evolutionary tumor board. Every team had a patient, who could be simulated or a retrospective one (i.e. a former patient who had since died). The rag-tag tumor board's job was to come up with a treatment plan taking into account the evolutinary dynamics of their patient's disease.
Our team, Team Orange was the only team with an actual, living, dying patient. Our patient, let's call him Jim, was at stage 4 of his colon cancer. The disease was metastatic, meaning that cancer has spread through his body, he had tumors in his liver and in his lungs. At this stage the patient is considered incurable and therapy is limited to palliative care. In other words, the doctors aren't trying to cure him anymore, instead they are doing their best to help him and his family make the most of his remaining life. In expectation, that is no longer than 5 months. Jim, however, wanted none of that. In his letter to his oncologist he had written: "I want a Cure".
Team Orange, was led by this very oncologist, Dr Dawn Lemanne. She brought us our patient and we had four days to come up with a viable treatment plan for his incurable disease.
SOC and MTD
In order to understand the novelties of evolutionary therapy it is useful to understand the current standard of care (SOC). Cancer is different from most other forms of illness in that the stuff making you ill are the cells of your body. These cells have undergone mutations that make them, at first, useless for your body, before becoming a steadily growing burden. At some point you simply have too much of them. They form tumors which will eventually become big enough to shut down your organs. If that doesn't happen, they eat away all the nutrients in your body until you starve to death. If your tumors are inoperable and targeted therapy isn't an option, then you most likely get some sort of chemotherapy: drugs that attack rapidly dividing cells. Since cancer divides more rapidly than anything, this seems just like what you need. The problem is that chemotherapy, while it is more than capable of destroying cancer cells, causes toxicity to the body by killing your rapidly dividing normal cells as well. For this reason the current SOC is to give the patient the Maximum Tolerated Dose (MTD) of the drug, the maximum amount that doesn't kill you. It's logical, right? You need large dose to kill as many cancer cells as possible, and the largest that you can go without dying is the MTD.
This strategy works in roughly 60% of all cancer cases. Once the disease is metastatic, however, that number drops down to nearly zero. Let me hammer this in one more time. Once the disease becomes metastatic, the therapy that your doctor is most likely to prescribe will fail, and the doctor knows it.
The SOC of breast cancer: All roads lead to "continue therapy until progression or unacceptable toxicity".
What usually happens, what we've seen happening over and over and over again since the 1990s is that the oncologist starts the patient on an MTD dose of the least toxic form of chemotherapy. It seems to work for a while, then it stops working. The disease becomes resistant - it no longer responds to the drug. Then the patient is put on another type of drug with the same outcome. Then another. Every successive line of drug is worse than the previous ones. If the patient is still alive, he or she is not looking well at this point. This continues until the patient dies. Likely as not, they die of an infection their immune system is not able to handle due to the toxicity. They may very well decide to go without treatment to maintain a more fulfilling lifestyle, in which case it's the tumor burden that kills them. The point is, in the overwhelming majority of cases, unless the patient finds a way to die a non-cancer related death during the months or years of therapy, the metastatic disease is fatal.
If Jim had not been adamant that he wants something other than the SOC, if he hadn't found Dawn this would have awaited him as well.
Evolutionary therapy 1: Adaptive therapy
The idea behind evolutionary therapy is to treat cancer like an adaptive organism which evolves with its environment. If the environment is riddled with chemo, then, at the cost of billions of dead cells, cancer will find a way to adapt to it. By introducing a new drug and changing the environment only when the tumor has evolved resistance to the previous one, the physician, unwittingly, makes its job that much easier.
The main insight of evolutionary therapy, therefore, is to change the environment as often as possible to prevent or prolong cancer's adaptation. This is where game theory enters the picture. Therapy-sensitive cancer cells, resistant cells, and the body's normal cells are all populations in the many evolutionary games that surround cancer. The success of one population is dependent on the populations it interacts with. Thus, to anticipate and steer cancer's response to therapy, one must understand the underlying game.
The first insight in this direction was the discovery that three types of prostate cancer cells, if left alone, are engaged in a spatial version of the rock-paper-scissors game: each of the three types promote another type's growth, while inhibiting that of the third in a cyclic fashion.
Two of the three types, T+ and TP, respond to the SOC, a combination of two MTDs of hormone therapy, while the third type, T-, doesn't. In every case of metastatic prostate cancer, these two drugs are deployed to wipe out T+ and TP, giving the patient an extra 12 months of life. Then, the T- type simply takes over, proliferates unhindered, and kills the patient. This therapy was clearly designed without knowledge of the game, and is known to fail every time. Every. Single. Time. The question is, can we do better by exploiting the game?
The first clinical trial of adaptive therapy as of early 2018 (I didn't find the more recent data). Metastatic prostate cancer patients on adaptive therapy (left panel) get less treatment and survive longer than those receiving SOC (right panel). Progression means therapy failure and, in this case, the patient's death.
By the knowledge of the game, the solution seems obvious: Instead of trying to exterminate the whole tumor by MTD, try to manage the tumor. Keep the composition as stable as you can by using the drugs to control T+ and TP and using them to control T-. By applying one of the drugs in lower doses, progression-free survival is at least doubled by using less than half of the (very expensive and toxic) drug!
Evolutonary therapy 2: First strike-second strike
With the clinical trials of adaptive therapy of prostate cancer still ongoing it is unclear how much it can improve prostate cancer survival. One thing is for sure: if it continues to succeed, it will change how we view and treat cancer dramatically. If immunotherapy was worth a Nobel Prize in Medicine in 2017, I'm willing to bet this one will be as well. Mark my words that Moffitt's own Robert Gatenby will receive it within the next ten years.
The main disadvantage of adaptive therapies of all kinds is that it doesn't actually cure cancer. It simply gives a patient a longer time to figure out how to die of something else while providing a better quality of life. While that is already a huge achievement, our patient, let's call him Jim, wanted none of that. Jim wanted a Cure, which can only be achieved by extinction therapy.
Extinction therapy is MTD without any palliative concerns. The physician is firing on all cylinders until cancer, or, much more likely, the patient, goes down. Trying to cure a patient this way is like typing in a password by banging your head against the keyboard. Once again, success has been modest.
The prevailing paradigm that hinders progress is that of "no harm". As long as a line of therapy appears to be working, the physician cannot and dare not withdraw it. In any other kind of disease this would be a sensible strategy, but with cancer, this leads back to the MTD trap: cancer eventually adapts to the environment and becomes resistant, making the drug that was working so well useless in the future.
The evolutionary insight in this case is to switch therapies before resistance sets in. The first line of therapy - the first strike - is meant to weaken the tumor, but rather than continuing until cancer is seemingly wiped away, the physician withdraws it. Instead, the patient gets a second line of therapy. If the two therapies induce different resistance mechanisms, then the cancer cells' resistance to the first strike is useless against the second strike.
Again, a frequent, radical change in the tumor's ecology is enough to delay or even prevent resistance. This time, the objective is not to keep the tumor alive at a manageable level but to get the most out of each of the strikes and kick cancer while it's down until the cells are so few in number that they cannot sustain their population. Our approach against cancer is best summarized by the immortal words of a 19th century Corsican oncologist:
You must not fight too often with one enemy [by using the same therapeutic strategy], or you will teach him all your art of war.
That is our plan.
Team Orange vs Jim's colon cancer
On Monday, November 4th, after the introductions were completed, Dawn presented Jim's case to the team. He is a middle aged male from another continent and we weren't allowed to know much more about him at that point. He was first diagnosed with colon cancer in 2017. At that point he was stage 3, meaning that his cancer was disseminated but has not appeared in any of his distant organs. He received chemotherapy and his colon was resected, getting rid of the primary tumor. His cancer was clinically undetectable at this point. Unfortunately, next year his condition worsened. His cancer, completely absent from the colon now, has presented in his liver, and later, in his lungs. He was now stage 4 and deemed incurable.
Team Orange propaganda piece. The leadership is in the inner circle (from the top, clockwise): Medhi, Joel, Dawn. The team members are in the outer circle (from the top, clockwise): Mascot, Peter, Sasha, Anna, Carolin, Audrey, Gergely, Savieay, Mariah, Kiran, Monica, Michal, Bryce.
Joel Brown, a professor at Moffitt, co-leader of the team, and my host, helped us formulate our objective which was to provide eco-evolutionary insight for Dawn in order to help treat Jim. We read up on the literature, allowed the thought to sink in, then went out for the workshop's social event.
On Tuesday we actually got to work. Our team - 15 people and one taxidermied squirrel - broke up into smaller, more efficient task forces. We had people reading up on the therapies and how they work, two-three groups building mathematical models of various aspects of Jim's disease (one of which included me), one group setting up an experiment on how different diets might affect the efficacies of drugs, and one group collecting material for the competitive element, Friday's final presentation. After our brains stopped working we had some beers and chicken in the office and called it a day.
Wednesday was a bit of a less energetic day. Our enthusiasm wasn't fueling us anymore, we were getting tired, and didn't have a clear picture of what we wanted to achieve. We had to make a lot of hard modeling choices to make which are unrewarding, cost a lot of energy, and take you down a lot of 'what-if' rabbit holes. Nevertheless, they had to be made and we made them. It didn't feel like it, but we were making progress.
By Thursday I'm pretty sure we were exhausted but we were nowhere near submission. Just before lunch, Joel had us all meet in a conference room for a group picture, which should have seemed odd but knowing Joel for the extrovert that he is it didn't. Once we gathered, our patient walked in and OH MY GOD.
It had nothing to do with his appearance, in fact, other than his terminal cancer, Jim was in perfect shape. It was that I'm a fucking economist. To all colleagues of mine who had dealt with impostor syndrome before: you have no idea. We had a little chat with Jim and he told us how moved he feels that all these smart people from various backgrounds (nearly choked on my tea) are all there to help him. In any case, us being tired no longer seemed to matter.
I'm sure that meeting Jim has put the 50k out of everyone's minds. Our singular goal was to be useful to him and Dawn. Even I was able to become useful, hell, if I might say so, I was very useful! We only stopped working for the very short gala dinner the whole day and stayed until the 2:00 am. Joel shooed those who were speaking next day to bed but my modeling partner Gergely still made changes to the files at 4:00 am. We were back in the office by 6:30 to put together the models and rehearse the presentation. We did not get to rehearse the presentation (this will become important later).
Friday morning, the teams convened for the presentation session. We went last of all the teams. Two of them were good but beatable, the other two were pretty impressive in their presentations, but only one of them had come up with some real good stuff content-wise. We all went on stage with a feeling that we had a winning hand, but at the very least a guaranteed second place.
By the time I went on stage we must have been in, like, minute 34 of the supposedly 25-minute presentation.
We blew it. We blew it hard. That presentation we didn't rehearse took about 45 minutes rather than the supposed 25. We thus de facto disqualified ourselves from the competition. We got chided by the chief organizer and were destroyed in the popular vote by our very tired and hungry electorate (we went well into lunchtime). Team Blue won, very deservedly, while Team Green were the runner-ups. Yet there are rumors about some in the jury putting up a little fight to make us winners despite us basically ignoring the only rule of the competition.
For us, it didn't matter much (or so we keep saying). We made history: our team was the first of all IMO workshops to conduct actual experiments during the workshop, for the workshop. More importantly, we believe that what we did may actually help save Jim.
What we did
The main strategy that Dawn and Joel envisioned is the first strike-second strike. It's clear that SOC is not the way to go, while Jim is adamantly against adaptive therapy. His attitude is not born out of ignorance either, he's a very well educated man in a high mental skills job, he's simply tired of dealing with this shit and he doesn't want to do it forever. He's already beaten an unrelated cancer of a different type a couple of decades back so you might say he has reasons to be optimistic about his chances.
So, first strike-second strike, how do we go about it? For that to work you need drugs which induce different resistance mechanisms so when cancer adapts to the first, it will be susceptible to the second. Ideally, you need more, but two is the least we were hoping for. The team put together a list of drugs and an order in which to use them to achieve maximum effect. Some drugs are for colon cancer. Some are used against other types of cancer. You might call that throwing in the kitchen sink (as did one of the jury members) and you're not fully mistaken. The truth is, we need the variety. Orthodox or not, as long as we can maintain a change in the cancer's environment and attack it from multiple angles, we'll take it.
The next thing we needed was a timetable of when to switch the drugs. Based on case studies and hunches, Dawn wanted to switch every two months but we all knew that's very little more than a guess. There aren't many useful statistics around when you're trying to cure incurable cancers. So we put together a model based on real scan data on how quickly Jim's cancer has grown, we estimated treatment efficacy by how long he would survive on SOC, and we did our best to come up with what happens to resistance upon a switch of the drug.
All of the three above graphs measure tumor burden (vertical axis) as a function of simulated days (horizontal axis). Our model predicted that if you switch every 60 days (left figure), Jim's cancer will end up killing him even if we had 10 different strikes. You'd be switching at a time when the number of cancer cells (purple) is already on the rise and thus resistant cells (red) are cannot be driven extinct. If you switch every 40 days (center figure), you switch too quickly and not using the full potential of your strikes, for the same outcome: Jim dies. If you time your switches to be anywhere between 45 or 52 days (right figure), then, according to the simulation, the tumor goes extinct.
When I showed this on Friday early morning to a very sleep-deprived Dawn, she, I kid you not, literally jumped out of her seat and exclaimed in joy. She adjusted her strategy to switch every 50 days in accordance with our model. When I learned of this my blood froze to ice. Not because our model isn't well put-together: in fact, I believe it's as accurate as it can be given our (lack of) data. But still, I didn't expect to influence the decision making of an oncologist with 30+ years of medical experience with a person's life on the line.
Before you become worried let me reassure you. The timing of the strikes will be decided by a team of actual oncologists, and will mostly depend on Jim's response to the first strike, not on what a mathematical model is saying, especially one calibrated by What's-his-face from Wherever, Eastern Europe, PhD in Something Not Relevant At All. Everything our team came up with will be filtered through an actual tumor board and Jim himself. Nevertheless, if that's not cool, I don't know what is.
There's one more thing our team did but I haven't talked about.
Diet
I bring this topic up very reluctantly. Not because it's not important, but because the general public is terrible at handling it and I'm a big proponent of epistemic responsibility. But here it goes: a good diet can help cure Jim's cancer.
No, no, no, no, no, no. Diet won't cure cancer by itself. Nor is there a diet by which you can avoid getting cancer. It's that cancer likes glucose (sugar) a lot. One way to lower the glucose level in your body is to stop eating. Unfortunately, your normal cells also like glucose so you can't just starve out cancer like you would a besieged fortress. But. Cancer cells use glucose much less efficiently than normal cells, known as the Warburg effect. They use it to reproduce, repair themselves, and move around but they're using orders of magnitude more than normal cells without getting much out of it. At this point, nobody knows exactly why, they most likely do that to starve off their neighboring non-cancer cells. Crucially, once they're facing an adverse, chemo-ridden environment, they use glucose to build and maintain pumps on their surface to get rid of the toxic drug as well as to repair the DNA damage they sustained from the chemo.
The cool part is this: If Jim performs a 24-hour fasting diet just before he gets therapy, he deprives the cancer cells in his body from nutrients just when they need them the most. Without enough energy to operate those chemo pumps, cancer's resistance collapses and the therapy will be more effective. Our team had a group of modelers analyze this effect and another group doing in vitro experiments on actual colon cancer cells in high glucose and low glucose environments.
Artist's illustration of cancer cells developing resistance to chemotherapy.
Let me say again, diet is not a cure. By itself, fasting is next to useless against cancer. But, by a very marginal effort, the patient can give therapy a much-needed boost. For Jim, this may very well mean extending the one-week window we predicted for the therapy switch to work effectively. If our model turns out to be correct but inaccurately calibrated, which is highly possible, this may mean the difference between life and death.
Another good thing about the fasting diet is that it induces a drastic, one-time change in cancer's environment. This lines up with the philosophy of our strategy perfectly which is to never let cancer adapt. I don't know of any other diet that is circulating the Internet as a Complementary and Alternative Medicine (CAM) against cancer which has this feature.
Conclusion
I honestly don't know what the conclusion is at this point. Our competition is over but Jim's fight has just begun. I'll post an update once we have more on how he's doing.
It is clear to me that if an economist, completely untrained in cancer biology, is not only tolerated but very much welcome in an actual high-stakes competition with money and lives on the line, then something must be wrong. Either IMO is mad, or the status quo of cancer treatment is mad, and I'm not qualified to pass judgement on which one it is. But my impression is that IMO is not mad. My impression is that IMO is made out of dedicated and creative professionals who are (unlike the hack that I am) highly trained when it comes to cancer, and who are not afraid to think outside the box. I really believe that the people in this department will change the way we view and treat cancer for the better. I get the impression that
their work may make a greater impact on the survivability of
metastatic cancer than has been made in the last 30 years.
As an extra, they're overall really nice people who are a privilege to be around.
With that said, I know where my true calling is, and it is in economics departments researching game theory (and networks). Game theory used to be a true force of nature, it changed the way we think of social sciences. It embedded itself into economics so deeply that even those economists who profess to dislike it must use it in their day-to-day (which gives me more schadenfreude than it should). Now, it seems game theory is in a bit of a dry season, the ideas no longer groundbreaking, the results not quite as spectacular as they used to be, while social sciences seemed to have moved on to other, in my opinion, less cerebral and more boring tools to use. New applications like cancer (and networks) might just be what the field needs to rediscover and revitalize itself into a second Golden Age of Game Theory. Amazing people and exciting discoveries aside, I will continue to work with IMO just to tap onto the golden age of their field and their department and maybe channel some of it through to mine.
That, and to see more alligators.
November 24, 2019.
Footnotes:
Realistically, the workshop was more about learning and less about the treatment plan we come up with. Evolutionary tumor boards are still a brand new idea and these types of workshops are part of a greater outreach program.
Normally, a tumor board meeting takes hours rather than days and there isn't much modeling involved.
In case that needed to be said, Jim's real name isn't Jim. All others in the post are real people and I encourage you to look them up.
Links for further reading:
Dissemination on (future Nobel Prize winner) Bob Gatenby and adaptive therapy:
https://www.wired.com/story/cancer-treatment-darwin-evolution/
Dr Dawn Lemanne's talk on the evolutionary principles of cancer therapy (31 minutes):
https://www.youtube.com/watch?v=1tteYZfMat4
IMO Director Alexander (Sandy) Anderson on evolutionary therapy (17 minutes):
https://www.youtube.com/watch?v=YD6XVabU-k8&feature=emb_title
News piece of Sandy and IMO (3 minutes):
https://www.youtube.com/watch?v=6V3r7014sc4
Cool mathematical oncology blog (technical):
http://blog.mathematical-oncology.org/
Full breast cancer treatment guidelines of 2019:
https://www2.tri-kobe.org/nccn/guideline/breast/english/breast.pdf
Comments and questions should be addressed to peter.bayer7@gmail.com.