Of Mice and Men…
Why we have to be careful with soundbite “cancer cure” headlines.
So it’s been a while since I last wrote on something topical and answers to a few questions posed to me about a couple of recent papers prompted me to think it might be useful to collect these thoughts in a more structured way, so here goes.
Much of the drive comes from this science news piece. The right diet can boost potency of cancer drugs. Consuming the right foods can tweak a tumour’s metabolism, making chemotherapy more effective.
I was brought to this because there’s actually nothing new in the statement, although the extrapolation outlined in the article title is quite misleading as it makes the usual jump of assuming something shown in lab models and mice is wholly predictive of activity in human beings. “Tweaking” with diet also more commonly tends to work in a counter productive manner. So let me make a few observations which might contextualise my statement above.
First point, kinda obvious, a mouse isn’t a human.
Now we know this empirically cause one is a small fluffy white ball of fur and the other is orders of magnitude larger, not furred and has no tail. (in these days of fake news and pseudoscience, I have to just check we’re all on the same page ! ) That’s not the only difference of course and mice, especially laboratory mice, have many practical, anatomical, physiological, immunological, metabolic and pharmacological differences from humans. From the point of view of cancer research it’s probably best to think of them as cute, complicated 3 dimensional petri dishes. Perhaps the two biggest differences are as follows. Firstly, in many cases the cancers grown in mice originated from humans. Cells intrinsically don’t like growing in hosts of a different species. Indeed most cells, unless they become cancerous, don’t even like growing outside their originating tissue (hence why blood cells don’t grown in the skin and skin cells don’t grow in the lung (unless they metastasise from skin cancer, for example). So growing human tumours in mice means you have to use animals with virtually no immune system. Their growth pattern is also often very different and much more rapid than it is in a real human cancer. Secondly, lab mice are incredibly homogenous (or similar to each other). This is vital for laboratory work as we scientists like to be able to reproduce things multiple times so our stats looks good and if we find something important, our buddies or competitors elsewhere should be able to replicate it. Human cancers and their hosts vary enormously and the breath of that variability isn’t really modelled.
To cut a much longer discussion right down. The lab models we have, mice, dishes etc don’t have that great a track record of predicting successful therapeutics. Depending on exactly how you break it down somewhere between 1/10 of the agents that look promising in the first phase of human studies (when the main lab stuff is finished) actually go on to make a useful and marketable cancer therapeutic. Roughly 9/10 of the “promising cancer studies” fall by the wayside as they are tested in real human cancer. Real human cancer is complicated and multidimentional. There isn’t a big queue of other methods/techniques so it’s the best we have for now.
“Roughly 9/10 of the “promising cancer studies” fall by the wayside as they are tested in real human cancer.”
So any paper discussing a new cancer development that is just based on cells and mice (as both of the cited papers are) should really only ever be “interesting” because it’s got the odds stacked against it until someone has invested a few billion $/€ and run through years of clinical trials in real people to show that it’s meaningful.
So we all agree then that mice aren’t humans and we really shouldn’t claim cures for disease based on a single set of results from them? Excellent we are now ready to move on to the complicated stuff.
Second point -redundant biochemistry
Humans survive on earth because of redundancy. Redundancy means that our component systems and biochemistry has backup built in. So, for example: despite a continuous need for a constant level of the sugar glucose in my blood stream (which I usually get from my diet) there are fail safes and backups which mean that even if you drop me in an isolated location with no food, I’ll survive for several weeks and my blood sugar will be kept at a fairly constant level. Mission critical body chemistry, like sugar, the level of acidity of blood etc. have complex levels of backups and this gives rise to constant maintained states called homeostasis. My body can attack my pancreas and I probably won’t notice until its functionality drops below 1/10th of its original (I would then be a type I diabetic). I can survive on one kidney or lung yada, yada yada, you get what I mean. One way of thinking about life is that most of us get 80–100 years until the redundancy falls over and a key system necessary for life irrevocably fails and we die. This redundancy extends from the most basic biochemical pathways making and breaking down chemicals in our body right up to organs. Some things we don’t have much redundancy for of course. Take away oxygen and I’ll be irrevocably harmed in typically 4–6 minutes.
“One way of thinking about life is that most of us get 80–100 years until the redundancy falls over and a key system necessary for life irrevocably fails and we die”
Homeostasis (the fancy term for these redundancies) and inbuilt redundancy means that it can be very difficult to impact key systems in a useful way in a human being. So I could starve a cancer by giving it no food, the problem is I’ve no way to do that without starving myself and the more complex systems in my body are more sensitive to the disruptions caused by no food than the simpler more adaptive ones in the cancer. So, I’ll die of starvation (thereby killing the cancer of course) before the tumour does.
So messing with key pathways in cells and mice is relatively easy, doing so in humans is much more complex and incredibly difficult to predict its outcome in populations of patients.
Third point — did we really already know that diet can impact cancer drug potency?
Yep, we did and diet is already known to have major impacts on cancer treatment on several different levels. You learn this kind of stuff when you do specialist qualifications, go to a professional college for many years and sip from the fountain of knowledge passed down by clever predecessors rather than a Dr. Google self-certificate from Dunning Kruger university summarising the confirmational bias of your browser history!
The complex biochemistry that our lives depend on requires energy and some basic chemicals to drive the reactions with which we think, move and protect ourselves from nasty bugs that would love to use us a food source. At the highest level if our diet during cancer treatment doesn’t provide those nutrients, we get very sick and the cancer drugs can make us much sicker to the point of imminent crisis and even risk of death. This may mean that treatment has to stop altogether and oncology specialist dietitians (AKA people went to a real college specialising in human dietetics and had to prove their vast learnings to recognised experts; not nutritional therapists, journalists, celebrities chefs, Netflix junkies etc) spend many years learning the special practical scientific and practical skills needed to provide the patient nutrition needed to help give cancer drug treatment best chance to work.
“if our diet during cancer treatment doesn’t provide those nutrients, we get very sick and the cancer drugs can make us much sicker to the point of imminent crisis and even risk of death”
At the next level down, we know that certain food stuffs can greatly impact the absorption of medicines. Many of the newer cancer drugs taken by mouth are chemically said to be “lipophilic”. This fancy word basically means these medicines hate being in water and much prefer to be in oils and fats. The net result is that the fat content of a meal can have an enormous impact on the levels of these drugs that get out of the gut and circulates in the body. This can be a huge issue and for some of these agents, you can more than double the amount of drug that gets into the body from the same dose by taking the agent with fatty meals. That might sound great and boost potency etc. but actually, on the whole, it tends to be bad as usually you just see more side effects. Hence, the companies that make the drugs will have a label telling the patient when they should take the medicine relative to their meal habits, thereby helping better ensure predictable absorption and effect.
And at the next level down, again, some of the ingredients in food can hugely impact cancer drug potency. The classic example that I talk about when I teach health care professionals about this topic is grapefruit. Now everyone knows that grapefruits are fruit and fruit is “healthy” so that should be good, right? Unfortunately it’s more complex (it usually is when us scientist start looking at stuff, spoil sports!). Grapefruit (and indeed several other citrus and some other fruits) naturally contain some interesting chemicals. A few of these are metabolised by enzymes in the liver that are also very important for breaking down cancer drugs. So taking grapefruit can stop the action of these enzymes. The result being that the liver isn’t able to metabolise the cancer drug as much as it should be, the levels build up and the cancer medicine may make the person much sicker than normal. Medicines that are affected by this phenomenon (and there are a lot of them) will have a warning somewhere in the patient instructions.
“grapefruit can stop the action of these enzymes. The result being that the liver isn’t able to metabolise the cancer drug”
So, you can hopefully see why I say that science has known about diet medicine interactions for a long time.
Now back to the stuff folks were asking me about.
The first paper mentioned a lab finding that increasing the amino acid histidine intake may potentiate the effect of methotrexate . Interesting but histidine is involved in quite a lot of important cellular events and it’s not clear what the human ramifications of modifying histidine levels might be on the whole system.
More of the article focussed on this paper. One of the supervisory authors is a well-known oncologist and writer in the field. I’ll give a shameless plug for his book “the Emperor of all Maladies” as it should be compulsory reading for everyone. The paper is really excellent calibre. The research and, potentially more impactfully, the reports of the research, mention the “keto diet” in the context of treatment. Folks may know that I’ve dissected the inadequacies, weaknesses, lies and dangers of this diet in general cancer treatment a little in the past. Unfortunately, some commentators and self appointed gurus who appear to neither have read the article itself, and wouldn’t have the training to understand it even if they did, seemed to jump on this research as if it was Moses returning with the 10 commandments. This is despite the fact that the paper doesn’t show the keto diet having a general beneficial effect in cancer. How come I’m pretty sure that’s the case? Well because there are a few reasons and the paper actually says it in black and white when you read it.
1) Of course to start with it’s a lab paper (cell and animal models — limitations touched on in the preceding monologue) so hopefully you agree that while we can say lab stuff is promising and interesting we can’t go calling it a treatment or cure (reference first point of this article).
2) In the treatment groups where they gave the keto diet to animals it didn’t have any benefit, indeed in some models of blood cancer it made things worse. Now that’s not compellingly negative as it’s still only a lab finding but there are a number of cancers where we know the mutations make them very dependent on fats for growth and which can be stimulated by fat levels so it’s not surprising that a lab model might see such cells grow more rapidly and kill the test animals quicker. Here’s a direct quote from the paper “It is important to note that treatment with the ketogenic diet alone had variable effects in different tumor models indicating that the dietary changes themselves were insufficient to cause the tumor responses observed across the murine models. In some instances, such as the AML model, the ketogenic diet alone accelerated disease progression suggesting that this diet may be detrimental for some cancer patients when used in isolation”
“direct quote from the paper “It is important to note that treatment with the ketogenic diet alone had variable effects in different tumor models indicating that the dietary changes themselves were insufficient to cause the tumor responses observed across the murine models”
Is there any other reason we should be cautious in our interpretation ?
3) The paper focusses on insulin signalling in cancer cells being treated with a particular new group of putative cancer drugs (some of which have regulatory approval) called PI3K inhibitors. There was a lot of hope that these agents would be very useful, treat lots of cancer and make lots of money. The mutation occurs fairly frequently in quite a few caners, although its not clear what percentage of cancers are actually driven to grow by it. Hundreds of patients and billions of dollars later clinical trials of these agents have been underwhelming. Used on their own, any responses tend to be short lived and patients often become diabetic because the drugs affects insulin signalling. The authors cleverly wondered if stopping the induced insulin signalling aberration with targeted medicines or a keto diet (which can reduce insulin signalling in these mice) might restore sensitivity to this underperforming medicine. They showed in their models that if you get cells that are PI3K-dependent, it does indeed appear that playing with insulin signalling at the same time can make the drug work better. However, there are many who suspect that these drugs don’t work very well not just because of insulin signalling amplification but because tumours are a mixed bag of different mutations and just because a mutation is present doesn’t mean that it’s a dominant mutation (a driver mutation) that the cells need to survive and prosper. Basically, just because a mutation is present doesn’t mean its important. There are toxicity worries with these drugs too (which could also theoretically be made worse as well).
4) Finally, let’s be optimistic and hopeful- the lab work extrapolates exactly over to humans- what would we get? We might envisage some sort of test for cancers that are dependent on PI3K mutant signalling. In that subset of patients with PI3-dependent cancers they might get one of the drugs and simultaneously be put on a keto diet to best ensure activity of the PI3K inhibitor.
Keto diets are tough going, really, really tough. I’d say that of course but actually to be fair even the more moderate cultist supporters of the diet in cancer admit it and the clever ones try to provide workarounds to overcome the major problems it causes. There have been quite a few trials of this approach and none have moved forward because the diet is really hard for many patients to keep going with, especially if they are already sick with cancer and aside from anecdote there are no trials showing statistically that it has any benefit. It’s a restrictive diet so many common food stuffs are out and the diet tends to make patients much more susceptible to the negative effects of cancer treatment in the short term. In the longer term though the nutritional imbalance associated with such a dramatic change requires specialist support to ensure adequate balances of other nutrients are provided. Despite false claims to the contrary, professional cancer dietitians can and will support those on a keto diet and their knowledge and skills will be vital if the patient is to be given the support needed to take best opportunity from the putative advance of combining one of these medicines with keto diet being suggested. This may be a bad thing for keto cultists because there are no scenarios where doctors are going to be letting internet- educated amateurs anywhere near patients whose lives they’ll be trying to help with the trials. Assuming trials in humans commence rapidly, it’s likely to be a minimum of 3–4 years and probably longer before there is any evidence of this approach working. Fingers crossed though that is one of the one of the research findings that does correlate to patients.
“Assuming trials in humans commence rapidly, it’s likely to be a minimum of 3–4 years and probably longer before there is any evidence of this approach working”
If you’ve lasted this far in the text you might at least now agree that my musings suggests that things might not be as simple as was first presented. That’s because these things are really, really complicated and lives are at stake. Hence, the best way that people can help stack the odds in their favour is to listen to their own doctor and get expert advice and not listen to the rumours and soundbites from a cabal of people trying to get you to click on their webpage (that they often get advertising revenue for), buy their diet book or training course or reinforce their egotistical view that they alone know what’s best to cure the world of all its ills.
So to summarise
1. We’ve known for a long time that diet can modify cancer treatment.
2. Referring to the main paper discussed, the research pertains only to lab models of a specific mutation in cancer - PI3K (not all cancers, no test in humans conducted).
3. The research found that even in these cancers the keto diet didn’t do much and in some cases made things worse.
4. They found that an induced keto state did make the PI3K inhibitors work better in the cell (lab) models expressing this mutation.
5. Trials in actual humans will take time to show if this effect is real and useful and there’s a healthy scepticism that PI3K may not be as important as was once though.
Neither paper is actually really supporting the miraculous soundbite headline of the claim therein. In five years we may know a lot more and we should keep fingers crossed until then.
