Dichloroacetate (DCA) and cancer: Magical thinking versus Tumor Biology 101

Late yesterday afternoon, I was lazily checking my referral logs to see who might be linking to Respectful Insolence™, as most bloggers like to do from time to time (and any blogger who claims otherwise is probably feeding you a line), when I noticed a fairly large number of visits coming from one location, namely here. I was wondering when this would happen, but it looks as though the regulars at The DCA Site have finally noticed some of my writing. Surprisingly, what they say about me is not that bad, although that’s probably because they seem to have found the least–shall we say?–vigorous post that I’ve done on the topic, in which I describe a paper that suggests a mechanism by which DCA might be ineffective against certain brain metastases.

If there was evidence about why self-experimentation is a bad idea for patients to do and unlikely to produce much, if any useful information, this discussion thread provides it. I’ve written about this before; so I won’t belabor previously made points. (No cracks about how that is what I do best, please.)

The discussion thread begins with a report from a man going by the ‘nym of “squareb” with metastatic sarcoma who reports after 24 days on DCA:

The PET/CT preliminary result shows that the primary sarcoma tumor in the thigh is smaller and has less activity on PET. I did have chemotherapy and radiation to the site in January. This may be a delayed effect of the treatment.

The metastatic tumors grossly have not increased in number, but they have increased in size and their activity has increased on PET. The largest tumor that was 6mm two months ago, 14mm last month is now 17 mm in diameter. You can argue that the rate of growth is decreased, but I do not see the dramatic decrease in tumor size as in the rat studies after 3 weeks of DCA.

I do not think DCA does nothing. On the other hand it is not the magic bullet everyone had hoped for. Of course it may not act on all cancers the same way. It will probably be an useful adjunct to use with other chemotherapies or radiation.

You see, that’s the problem. DCA may do something against sarcoma. It may not. We don’t know, and, as I mentioned before, if it doesn’t do anything in squareb’s case, that would not necessarily mean that it was useless against sarcoma. Almost certainly the shrinkage of his primary tumor and its decreased glucose uptake was due primarily to the radiation and chemotherapy. After all, the metastases are still growing. The hard part about writing about this is that I don’t want be too critical of a patient with metastatic cancer or risk destroying any hope he has. On the other hand, the magical thinking going on in this group after squareb’s report that fuels the desire of other cancer patients to seek out DCA outside the context of a clinical trial has to be addressed. For example, someone going by the name of DaveS (DaveScot, I wonder? The answer appears to be yes.) replied:

“The largest tumor that was 6mm two months ago, 14mm last month is now 17mm in diameter.”
diameter = volume (perfect sphere)

6 mm = 0.11cc
14 mm = 1.43cc
17 mm = 2.57cc

The volume of the largest metastatic tumor increased by 13x in the first month. That’s an increase of 1.90x each week. Before starting DCA the tumor got one more week to grow which at that rate would have put it at 2.77cc. The argument might be that growth in the metastatic tumor didn’t just slow down but rather was halted and possibly reduced in volume by 10%. My math should be right but I’m guessing about the exact number of days and it might not be anywhere near a perfect sphere. No additional tumors is certainly good news. This might be very encouraging results.

This is the perfect opportunity to explain to DaveScot (and, hopefully, to everyone else reading) a little bit about terminology used in cancer chemotherapy trials and to take a page out of Tumor Biology 101; first, the cancer trial terminaology. There are generally four possible results that can be observed in a trial of a new chemotherapeutic agent:

  • Complete response: The tumors shrink away to the point that they are no longer detectable by physical examination, imaging studies (MRI, CT scan, etc.), or tumor markers. Obviously, this is the best possible result. Sadly, it is not seen that often in clinical trials.
  • Partial response: This is usually defined to mean that the tumors shrink by more than 50% (or, in the case where tumor volume cannot be measured easily, tumor markers fall by more than 50%) in response to therapy but remain detectable. More recent definitions have at times loosened this criteria to include tumors that shrink by 30% or more.
  • Stable disease: The tumors either shrink by less than 50% or remain the same size. In some trials, this definition may be broadened to include tumors that increase in size during therapy by less than, depending on the trial, 0-25%, although I’ve pesonally always been suspicious of calling any detectable growth above random variation in imaging measurements “stable.”
  • Progressive disease: Tumors increase in size on therapy and/or new metastatic tumors appear.

Depressingly, by any formal definition you want to use, squareb has demonstrated progressive disease on DCA, particularly when coupled with his additional news (see below). Even he seems to realize it, as he states that he was quite surprised that his metastases hadn’t shrunk, given that his alkaline phosphatase levels had fallen. (Once again, this is yet another example of how unreliable most tumor markers are; alkaline phosphatase is quite nonspecific.) He also shows another insight that is notably lacking among others on the DCA Site discussion boards, particularly DaveScot, when he points out that, although he thinks DCA is doing something, he realizes that it isn’t enough by itself. Indeed, he’s discovered one sad fact about cancer that oncologists and surgeons have learned through long and painful experience over many decades (although, I hasten to point out, not as hard as actually having to suffer through cancer themselves): single agent chemotherapy (which is what squareb is trying with DCA as the single agent) usually produces pretty crappy results in solid metastatic tumors. It’s not at all uncommon in the setting of metastatic disease for single agent chemotherapy to result in response rates of 10-20%–or even lower. In addition, the rapid evolution of resistance is quite common when single agents are used, and there’s no reason to expect that DCA will be any different.

This is core of the problem with all this “do-it-yourself” medical experimentation. The people doing it don’t know the very basics of cancer biology, chemotherapy, or how response rates are measured. They do not even understand the basic terminology. Without that background, it’s like trying to design an airplane with no knowledge of aerodynamics and engineering. Most people wouldn’t even think of doing that, but here we have a number of patients (a couple of whom appear to be physicians, even) trying an unproven drug with only a poor understanding of cancer biology. They’ve done a lot of reading, but they appear not to have the basic background knowledge and practical understanding of oncology to put that information into context, truly a case of a little knowledge being a dangerous thing. There’s a reason that chemotherapy trials need to be rigorously designed and standardized, with controls and careful randomization and matching of patients. There’s a reason for all those specialized doctors, nurses, and infrastructure to follow patients on clinical trials to see if they are responding and to monitor them for adverse reactions and complications. Finally, some may point out that it has been observed that all too often there isn’t a good correlation between initial tumor shrinkage by chemotherapy and prolongation of survival. That is indeed true. Some chemotherapeutic regimens that can produce impressive initial tumor shrinkage result in only marginal increases in overall survival. But what is also true is that if tumors continue to grow rapidly through treatment with a chemotherapeutic agent, that agent will almost certainly not produce a prolongation in survival.

Depressingly, one woman named Sandra chimed in with a misguided attempt to claim that maybe DCA is working for squareb after all:

Because we are dealing with cell division, I think logs are required. Based on the formula:
Final tumor volume = starting tumor volume, times 2 to exponent n
“n” being the time required for one cell to divide.
Isolate and take the log of both sides.

Once we know n, we can solve for what the size of the tumor should have been, had it’s growth not been slowed by DCA.

I end up, assuming doubling of cells with no intervention of the immune system and no dying of cells, to get an expected volume of 18.58cc and a diameter of 2.4cm. If I am correct, DCA has been much more effective than initially thought!

Another commenter named Jeff added later:

There are some positives to be taken from this, specifically a decline in the RATE of growth. At this rate of decline, one would expect the tumor size to decrease in size by your next scan. That doesn’t mean it will necessarily turn out like that, of course.

No, no, no.

I’m very sympathetic to patients like sqaureb, but DaveScot, Jeff, and Sandra (and others like nventr and a scientist called Danny–who obviously does not understand basic tumor growth kinetics–in other threads) are making a very basic mistake that would have been obviated by reading a basic oncology textbook, even more so than DaveScot’s ignorant blather about evolution could be obviated by his actually taking the time to understand what evolutionary theory says. (This really is Tumor Biology 101 material we’re talking about here, second lecture, the one after the basics of what cancer is are explained.) They’re all assuming that the doubling time of a tumor remains constant as it grows. It almost always does no such thing. Indeed, doubling times for tumors nearly always tend to become longer as they get larger, a characteristic known as Gompertzian growth, a term that describes tumor growth that is exponential for a time but then levels off and progressively approaches an asymptote as the doubling time progressively increases. Sometimes, due to contraction of the necrotic center, some tumors will even on occasion appear actually to shrink slightly at different times in their history, even if no therapy is given.

Here’s why Gompertzian growth occurs: Tumor doubling time depends on a number of factors, which include primarily: (1) how long it takes on average for a typical tumor cell to divide; (2) the fraction of tumor cells that are actually proliferating; (3) the fraction of tumor cells that are dying (usually by undergoing apoptosis). The first factor is usually fairly constant, but the last two factors can vary considerably as a tumor progresses. When very small, tumors start out basically doubling in size as fast as their metabolism and blood supply allow because a high fraction of the cells are proliferating and few are dying. However, as a tumor gets larger, it tends to outgrow its blood supply, and tumor angiogenesis is not enough to reach the center. When this happens, the center of the tumor tends to die, which is why the tumors we as surgeons resect often have necrotic (dead) centers. Indeed, they not infrequently contain a cheesy, pasty glob of necrotic tumor in the middle. At the very least, the tumor cells in the center tend to become quiescent or dormant (not dividing) because of lack of adequate nutrients and oxygen, leaving a central tumor mass containing dormant and/or necrotic cells, surrounded by a rim of viable proliferating cells that are solely responsible for continued tumor growth and invasion. Indeed, histological sections of a well-established tumor stained for markers of actively dividing cells will frequently show that only anywhere from 0.5% to 20% of the cells are proliferating, with 20% representing a very rapidly proliferating tumor indeed. The end result of this is that, as a tumor grows, the proportion of cells in its total mass that are actively dividing becomes smaller, leading to a longer apparent doubling time for the tumor mass as a whole. More recent kinetic descriptions may have suggested that Gompertzian growth is an oversimplification (what model isn’t in biology?), but, if anything, these newer data suggest that tumor growth kinetics may be even more variable than we had previously thought. Be that as it may, the bottom line is that the tumor doubling time most definitely is not constant for a given tumor and it most definitely does tend to increase as a tumor enlarges, producing a growth curve that usually looks something like this idealized figure, even in the complete absence of any treatment:


Add to that the uncertainties and inaccuracies in measuring tumors when they are only a few millimeters in diameter, where an error of a millimeter or two can change the apparent volume of the tumor by a large percentage, coupled with Gompertzian tumor growth, and it’s impossible to know where squareb’s tumor was on the curve above when he started DCA, to determine how fast his tumor would have grown without DCA, or to assess whether DCA even slowed its growth down in the least. It’s certainly possible that it has and that more treatment might finally put the brakes on (and for squareb’s sake I hope this is true; this is one time where it would be a wonderful thing for me to be wrong); however, from his anecdote at least, I must, with heavy heart, conclude that it’s highly likely that DCA did not touch his tumor. The same problems hold true for determining response in any one patient. If a tumor shrinks impressively, then one can reasonably conclude that the drug is having an effect; otherwise, because of Gompertzian growth and biological variability in tumors among patients, without doing a randomized trial with experimental groups containing a enough patients to provide adequate statistical power to detect a treatment effect, it’s impossible to tell if a new chemotherapeutic just doesn’t work or whether it actually does slow down tumor growth.

Worse, when combining DCA with chemotherapy, if side effects occur, it can be devilishly difficult to figure out if they’re due to the chemotherapy, the DCA, or an interaction between the two. Once again, that is why clinical trials are needed and why self-experimentation such as this is not only dangerous, but highly unlikely to provide much in the way of useful scientific or clinical information. I can understand why he and Sandra might wish to believe that the decrease in tumor growth was due to DCA rather than Gompertzian growth, but the reality is that DCA probably had no effect, at least for squareb at the dose used. Unless his next scan shows a partial or complete response, we are unlikely to know, although stable disease might be suggestive of some response. Worse yet, in a later comment in the thread, squareb gives some bad news:

On the final reading of the PET/CT there are numerous new nodules. The nodule increase of the largest nodule is from 1.5 to 2.1 cm. Not a good result.

That’s roughly a 2.7-fold increase in the volume of the tumor over one month.

As much as I might sincerely wish it were otherwise and hope that squareb has better results on his next scan, the combination of the rapidly increasing size of the main tumor deposit and the new tumors on PET scan clearly indicate that almost certainly DCA is not working in this patient at this dose. If this were a clinical trial, depending on the specifics of the trial design and the intended treatment time, DCA might be continued for up to a total of two or three months to finish out the protocol, but in the case of rapidly progressive disease while under treatment with the experimental drug strong consideration would probably be made to withdrawing the patient from the trial and trying a different agent or enrolling him on a different trial.

I realize that there are depressingly few options for chemotherapy for metastatic sarcoma, none of them particularly effective, but options do exist. I sincerely hope that squareb abandons his present course, which is clearly not working out, and pursues them. Similarly, I urge everyone self-medicating with DCA to think again about it. squareb’s negative experience does not necessarily mean that the drug doesn’t work, but it does serve as a reminder why unorganized, unsupervised self-experimentation of this sort has the potential to endanger the patients involved, none of whom have any real idea of the risks and the extremely low likelihood of benefit to them, and contributes nothing to medical research. At the end of this, what has been accomplished? squareb and apparently a significant number of other cancer patients have exposed themselves to a real risk, garnered no benefit, and, equally bad, in the process produced no real useful information to guide future use of DCA, as a clinical trial would. Some are likely to be harmed. As much as anyone who doesn’t have metastatic cancer can, I understand the desperation that drove him to do it, but that desperation of cancer patients is what the DCA hype is feeding–and will likely ultimately disappoint. The only information that we’ve gleaned from self-experimentation thus far is that DCA is almost certainly not the miracle cure for cancer that the hype implied.

But, then, we knew that was almost certainly true already before all this reckless and unguided self-experimentation.

ADDENDUM: Walnut has posted his critique on Daily Kos as well.

All Orac posts on DCA:

  1. In which my words will be misinterpreted as “proof” that I am a “pharma shill”
  2. Will donations fund dichloroacetate (DCA) clinical trials?
  3. Too fast to label others as “conspiracy-mongers”?
  4. Dichloroacetate: One more time…
  5. Laying the cluestick on DaveScot over dichloroacetate (DCA) and cancer
  6. A couple of more cluesticks on dichloroacetate (DCA) and cancer
  7. Where to buy dichloroacetate (DCA)? Dichloroacetate suppliers, even?
  8. An uninformative “experiment” on dichloroacetate
  9. Slumming around The DCA Site (TheDCASite.com), appalled at what I’m finding
  10. Slumming around The DCA Site (TheDCASite.com), the finale (for now)
  11. It’s nice to be noticed
  12. The deadly deviousness of the cancer cell, or how dichloroacetate (DCA) might fail
  13. The dichloroacetate (DCA) self-medication phenomenon hits the mainstream media
  14. Dichloroacetate (DCA) and cancer: Magical thinking versus Tumor Biology 101
  15. Checking in with The DCA Site
  16. Dichloroacetate and The DCA Site: A low bar for “success”
  17. Dichloroacetate (DCA): A scientist’s worst nightmare?
  18. Dichloroacetate and The DCA Site: A low bar for “success” (part 2)
  19. “Clinical research” on dichloroacetate by TheDCASite.com: A travesty of science
  20. A family practitioner and epidemiologist are prescribing dichloracetate (DCA) in Canada
  21. An “arrogant medico” makes one last comment on dichloroacetate (DCA)

Posts by fellow ScienceBlogger Abel Pharmboy:

  1. The dichloroacetate (DCA) cancer kerfuffle
  2. Where to buy dichloroacetate…
  3. Local look at dichloroacetate (DCA) hysteria
  4. Edmonton pharmacist asked to stop selling dichloroacetate (DCA)
  5. Four days, four dichloroacetate (DCA) newspaper articles
  6. Perversion of good science
  7. CBC’s ‘The Current’ on dichloroacetate (DCA)