EDITOR’S NOTE: Unfortunately, this weekend, I was forced to get my slides together for the upcoming SBM Conference, editing a manuscript for resubmission, working on a manuscript that I should have submitted six months ago, reading over some grants, …well, you get the idea. What this means is that, alas, I didn’t have any time to prepare one of the new, long posts that you’ve come to love (or hate). Fortunately, there are a lot of other things I’ve written out there that can be rapidly adapted to SBM. For instance, what I am about to present now. Since I wrote this, I’ve thought of a couple of things that I should have said the first time (and was kicking myself for not having done so); so publishing an updated version here allows me to rectify those omissions.
A couple of weeks ago, there was a lot of hype about a study that hadn’t been released yet. Indeed, there was a story in Wired entitled and an editorial by the study’s lead author in Nature entitled . Not bad for a study that hadn’t been released yet. Intrepid medical and science blogger that I am, I waited until the actual study was published a week ago the June 1 episode of . It’s a clever study, but the hype over it was a bit overblown. For example:
For all the weapons deployed in the war on cancer, from chemicals to radiation to nanotechnology, the underlying strategy has remained the same: Detect and destroy, with no compromise given to the killer. But Robert Gatenby wants to strike a peace.
A mathematical oncologist at the Moffitt Cancer Center, Gatenby is part of a new generation of researchers who conceive of cancer as a dynamic, evolutionary system. According to his models, trying to wipe cancer out altogether actually makes it stronger by helping drug-resistant cells flourish. Rather than fighting cancer by trying to eradicate its every last cell, he suggests doctors might fare better by intentionally keeping tumors in a long-term stalemate.
Maybe I’m being a bit picky, but what annoys me about the news reports on this study is that the concept of turning cancer into a manageable chronic disease like diabetes or hypertension is not by any means a new idea. Remember, one of my major research interests is the inhibition of tumor angiogenesis. Consequently, I know that the late, great Judah Folkman first proposed the concept of using antiangiogenic therapy to at least as early as the mid-1990’s. The only difference is the strategy that he proposed. The idea had also been floating around for quite a while before that, although I honestly do not know who first came up with it.
But let’s see what Dr. Gatenby proposes. What makes it interesting is that his study actually looks at how scientists have applied evolutionary principles to cancer until recently, argues that we’ve been doing it wrong. He then proposes a way to use the evolutionary dynamics of applied ecology. He may well be on to something. First, here’s :
The German Nobel laureate Paul Ehrlich introduced the concept of ‘magic bullets’ more than 100 years ago: compounds that could be engineered to selectively target and kill tumour cells or disease-causing organisms without affecting the normal cells in the body. The success of antibiotics 50 years later seemed to be a strong validation of Ehrlich’s idea. Indeed, so influential and enduring was medicine’s triumph over bacteria that the ‘war on cancer’ continues to be driven by the implicit assumption that magic bullets will one day be found for the disease.
Yet lessons learned in dealing with exotic species, combined with recent mathematical models of the evolutionary dynamics of tumours, indicate that eradicating most disseminated cancers may be impossible. And, more importantly, trying to do so could make the problem worse.
Traditionally, cytotoxic chemotherapy has been given in either as fixed doses close to the maximum tolerated dose or, as has been common more recently, a regimen known as “dose dense.” Basically, the fixed dose schedules involve giving as much chemotherapy as the patient can tolerate up to what is known as the “maximum tolerated” dose and giving it over as short a period of time as possible, while “dose-dense” therapies try to target chemotherapy doses to the time of maximal tumor growth, when tumors are maximally sensitive to chemotherapy. This strategy is based on what is called the Norton-Simon model. One key assumption behind such therapies is that chemotherapy fails because of the evolution of resistant cells after chemotherapy has begun. The idea behind this strategy is to hit the tumor cells as hard as possible as fast as possible to kill as many cells as possible and minimize the opportunity to develop resistance. Dose-dense chemotherapy has definitely resulted in improvements in survival in multiple tumors but rarely results in cure, at least in the common “solid” malignancies that kill so many, such as breast, prostate, lung, and colon cancer. However, that improvement sometimes comes at a price: Increased toxicity and side effects.
Based on Judah Folkman’s work, around the turn of the century (the 21st century, not the 20th century) Robert Kerbel proposed a new regimen known as metronomic chemotherapy. Metronomic therapy involves giving chemotherapy either continuously or at frequent dosing but at a much lower dose, the idea being that, because blood vessels are lined by genetically stable endothelial cells, they do not evolve resistance, and chemotherapy can be antiangiogenic. The idea was to deliver the same total dose of chemotherapy but without all the toxicity, meanwhile keeping the tumor in check or shrinking it by the effect the chemotherapy has on the tumor blood vessels. The drawback is that long periods of therapy may be required and the cumulative doses may end up being actually higher than more standard therapies. On the other hand, this latter aspect may not be a drawback because metronomic chemotherapy may allow a greater cumulative dose, with a concurrent greater cumulative effect. Metronomic chemotherapy is a promising concept, but thus far clinical trials in humans have been fairly disappointing.
One aspect that is shared among both of these therapy modalities is that they generally both involve fixed schedules and fixed doses. What Dr. Gatenby proposes to get around this is to apply what he calls “adaptive therapy.” This therapy is based on population ecology and the observation that the development of resistance does not come free. Indeed, resistant cells need to expend energy in order to do what cells do to overcome chemotherapy; for exmple, to repair DNA faster, pump the chemotherapy out of the cell, bypass intracellular signaling pathways blocked by new targeted therapies, or crank out enough peptides that induce the ingrowth of new blood vessels in order to overcome therapies that block these factors. In other words resistant cells tend to have a lower fitness under normal conditions. It is only the selective pressure of chemotherapy that allows resistant cells to proliferate faster than normal cells, and, indeed, resistant cells tend to lose their resistance when the selective pressure is removed.
Given this concept, Gatenby likens adaptive therapy to :
Gatenby: How people treat invasive species can provide an analogy for thinking about cancer therapy. In treating a field for a pest, for example, you might treat three-quarters of it with a pesticide, and leave the other quarter untreated. Pesticide-sensitive pests remain there, and they spread out into the field after treatment, preventing pesticide resistance from becoming dominant.
Using pesticides on an entire field is like what we’re doing with cancer now. And we all agree that we’d rather get rid of the pests altogether, but if you can’t do it, if every time you have an infestation you treat it and get resistance, then you try a different strategy. The alternative is to try to reduce the pest population so that it doesn’t damage your crop, and accept the fact that they’re going to be there. That’s what I’m talking about with cancer.
Wired.com: What type of treatment would that involve?
Gatenby: Instead of fixing the dose of the drugs, you fix the size of the tumor. Your whole goal is to keep the tumor stable. You continuously alter the drug, the dose, the timing of the dose, with that goal in mind.
Our models show that in the absence of therapy, cancer cells that haven’t evolved resistance will proliferate at the expense of the less-fit resistant ones. And, when a large number of the sensitive cells are killed, for instance by aggressive therapies, the resistant types are able to proliferate unconstrained. This means that high doses of chemotherapy might actually increase the likelihood of a tumour becoming unresponsive to further therapy.
So, just as the judicious use of pesticides can be used to successfully control invasive species, a therapeutic strategy explicitly designed to maintain a stable, tolerable tumour volume could increase a patient’s survival by allowing sensitive cells to suppress the growth of resistant ones.
It’s a fascinating concept. The idea is to keep from killing off too many of the sensitive cancer cells, so that they can grow to a certain point and keep the resistant cells in check. But can it work?
The paper published a week ago presents evidence that, at least in mouse models, it might be able to. I will admit that a lot of the mathematics in the paper are beyond me. There was a time when I was in college and taking all sorts of calculus and differential equations when these equations wouldn’t make my brain hurt to look at them, but if you don’t use it you lose it, and lost it I have (mostly). Suffice it to say that the model takes into account estimates of variability of fitness in tumor cells making up the population, dosing, differential uptake with tumor size, and other critical parameters. The concept of adaptive therapy requires that chemotherapy doses be adjusted to maintain constant tumor volume, increasing dosage if the tumor grows and decreasing it if the tumor shrinks. First, the mathematical model:
The graphs above represent modeling of dose dense/maximum tolerated dose (MTD) therapy, adaptive therapy, adaptive therapy (ADAP), and three varieties of metrnomic therapy, continuous infusion, high frequency, and low frequency. Four combinations of mixed cell populations were tested, including:
- FR with high free-field fitness and high sensitivity to therapy
- R with lower fitness and low sensitivity to therapy
- S with low fitness and high sensitivity
- ER with high intrinsic sensitivity and fitness but in an environment that restricts proliferation and response.
Combinations that were modeled included: (a) ”FS and R,” (b) ”S and FR,” (c) ”FS and R and ER,” and (d) ”FS and ER.” Strikingly, by day 1,500 of tumor growth (1,100 days after therapy was started), the tumor treated using the MTD strategy had grown to be the largest whereas those treated with metronomic therapy were smallest. When the simulations were run out to many thousands more days, until the tumor burden achieved the lethal threshold, all patients in the MTD and metronomic therapies eventually succumbed to their disease. In this model the tumors treated with adaptive therapy remained stable even after a period exceeding 10,000 days. In other words, tumors treated with MTD had the best initial response rate to therapy but tended to develop resistance rapidly, while tumors treated with metronomic chemotherapy remained stable and did not grow appreciably for much longer but nonetheless eventually developed resistance to the point where the tumor escaped therapy and killed the host. In contrast, tumors treated with adaptive therapy remained stable for a very long time.
Mathematical models are all well and good, but does adaptive chemotherapy work for real? To test that, Gatenby designed an adaptive therapy protocol for a mouse model of ovarian cancer. It was a tricky experiment to do, as his group had to measure the tumor burden every three days and then adjust the chemotherapy dose according to their behavior, decreasing the dose for each mouse if its tumor shrank and increasing the dose if it grew. All of this was done fore each and every mouse, meaning that there could be as many doses of chemotherapy as there were mice in the adaptive therapy group. Here’s the method:
The adaptive group received an initial dose of 50 mg/kg and thereafter the tumors were evaluated every 3 days and the dose was adjusted to maintain a stable tumor volume. The algorithm for dosing basically represented “a shot in the dark” because no prior experience was available to parameterize the models. Drug doses were established in increments of 10 mg/kg starting at the starting dose of 50 mg/kg. A treatment decision was made at the time of each measurement. If the tumor remained stable (defined as the no more than a 10% change from the prior volume using caliper measurements), no drug would be administered. If the tumor diminished in size or remained stable for two or more measurements, the next dose would be decreased by one 10 mg/kg decrement. If the tumor increased in size greater than 10%, the same dose of drug would be administered. If the tumor again increased in size, the dose would be increased to the next higher level.
As you can see, this is a pretty labor intensive regimen. No doubt Gatenby will be able to refine his method and develop a protocol that isn’t in essence a reasonable guess, but for now there isn’t a lot to guide scientists in developing such adaptive protocols.
Here’s the result:
In the mice, the adaptive regimen using carboplatin clearly worked better than the standard carboplatin regimen, suggesting that adaptive therapy can work. As Gatenby puts it:
Our analysis shows that, in the absence of therapy, the fitter, chemosensitive cells actually suppress the growth of the less fit but resistant population. Therapies designed to kill maximum numbers of cancer cells produce an environment in which the resistant cells both survive and are unopposed by the fitter, chemosensitive populations. This permits rapid regrowth of a therapy-resistant cancer. Alternatively, if therapy is limited to allow a significant number of chemosensitive cells to survive, they will, in turn, suppress the growth of the resistant population. We hypothesized that under these circumstances, adaptive therapy should be designed to maintain a normal cohort of surviving sensitive cells.
Another interesting aspect of this study is that it’s been known for some time that using metronomic chemotherapy allows a larger total dose of chemotherapy given over a longer period of time with lower toxicity. It works well in mice, but unfortunately is less stunningly effective in humans (much like antiangiogenic therapy, alas). The larger total dose of chemotherapy that can be delivered is one reason that has been postulated as an explanation for why metronomic chemotherapy can be more effective than dose dense chemotherapy. There’s no reason to think that adaptive chemotherapy wouldn’t behave similarly and allow for a larger total dose. But, in this model at least, it went beyond that. The adaptive chemotherapy group the dose required to maintain tumor stability decreased with time from 50 mg/kg to 10 mg/kg. In the experiment I showed, the individtual doses were 50, 40, 40, 30, 30, 20, 20, 10, 10, 10, 10, 10, 10, 10, 10, 10 mg/kg. This observation is consistent with a stabilization of the tumor cell population consistent with the evolutionary and ecological model used to test the study hypothesis.
The power of evolutionary principles is that they apply to more than just populations of organisms. They can equally apply to populations of cells within an organism, like cancer. In other words, evolution acts at both the organism level and the cellular leve. Tumors, given their genetic instability, enormous heterogeneity, and subpopulations of cells with different fitness and sensitivity to selective pressures are a perfect system to apply the principles of evolutionary ecology to. What’s fascinating about this study is that it appears that using evolutionary principles in a savvier way than we have in the past can work. In theory and in at least one animal model, it can produce more effective chemotherapeutic regimens. Indeed, one fascinating observation is that, the longer the tumors were treated with adaptive therapy, the less chemotherapy was required and the longer the intervals between doses that were needed to maintain a constant volume. Like the concept of antiangiogenic therapy proposed by Judah Folkman, however, applying evolution to cancer may require a rethinking of how we deal with cancer.
Unfortunately, I don’t see an obvious or immediate application of adaptive chemtherapy in humans. The reason is that it would be very cumbersome, labor-intensive, and expensive. Tumor measurements far more frequent than what we routinely do now would be required, as would frequent adjustments in chemotherapy dosing. As a strictly practical matter, it would be very hard to implement. Indeed, this model was very simplistic in that it tested adaptive chemotherapy using one drug. In reality, very few chemotherapy regimens in common use involve only one drug, and any truly adaptive therapy would have to adjust multiple drugs, with a concomitant exponential increase in complexity administrating it. Also, from a strictly clinical standpoint, tumors that are large or advanced would need to be shrunk because their size causes serious symptoms. How that would be integrated into an adaptive regimen remains to be seen.
Another potential problem is one that has been seen by scientists trying to use a similar approach to control HIV infection: compensatory mutations. In antibacterial therapy, the long term removal of antibiotics has not thus far resulted in the disappearance of resistant strains, and this is due to compensatory mutations that can restore the fitness of these strains. This phenomenon has only been described in viruses and bacteria, but it would not be surprising if they also occurred in cancer cells.
Even so, there is one potential use that I can envision for this sort of ecological approach to produce adaptive chemotherapy. This would be as a means of treating tumors that have well-validated serum tumor markers that correlate well with tumor burden in individual patients; for example, colorectal cancer (tumor marker: carcinoembryonic antigen, or CEA) or prostate cancer (tumor marker: prostate-specific antigen, or PSA). One can imagine an implantable pump that could measure the levels of these tumor markers and then, according to algorithms developed based on ecological and evolutionary principles, continuously adjust the dose of metronomic chemotherapy to keep a patient’s tumors in check.
Finally, as Gatenby himself points out, these sorts of approaches will not render the search for cures unnecessary. After all, consider other chronic diseases. Diabetes, for instance, can be managed quite well on a chronic basis, but what patient with type I diabetes wouldn’t want to be cured and thus able to throw away his insulin syringes? In the case of cancer, cures remain preferable, but, like the case of diabetes, sometimes settling for chronic management is the best we can do.
Gatenby, R., Silva, A., Gillies, R., & Frieden, B. (2009). Adaptive Therapy Cancer Research, 69 (11), 4894-4903 DOI: