Sunday, 3 July 2011

How To Read A Cancer Paper - Part 2

By Pan Pantziarka

Recap
In the first article in this series we looked at interpreting the results of in vitro studies, particularly those that look at the effect of different substances on tumour cells. In this second piece we focus attention on in vivo studies – in other words studies that take place in living tissue rather than in a Petri dish. For the most part these studies use rats and mice, though sometimes you'll find other animals being used. Again we will focus cancer research of the sort that looks at what effect a given substance (particular foods, supplements, vitamins or minerals, drugs etc) has on cancer.

Mice and Rats
The good news is that if you are a rat or a mouse there has never been a better time to have cancer. Again and again we see fantastic results in rodents. Cancers of many different sorts are slowed, stopped, destroyed. It is impossible not to feel excited by some of these results, and they are truly remarkable. Unfortunately however, these fantastic results on lab rats do not translate as well to humans. Why is that? We saw that test tube studies have all kinds of issues, surely these problems disappear once you move from glass dishes to living beings?

Some of the disadvantages of the test tube approach are easily solved in rats, mice and other animals. No longer do you have just a flat layer of cancer cells to bathe in your anti-cancer agent, instead you have three-dimensional tumours embedded in tissue, with a blood supply and tumour microenvironment. And, interestingly enough, we find that some agents that have fantastic results in the test tube are all of a sudden not so powerful against tumours. Instead the same agents have lower efficacy and in some cases seem to have lost all trace of anti-cancer effect. Why? Because the high dosing and guaranteed test tube conditions cannot be transferred to a living being. Doses that killed tumour cells under glass can also be toxic to living animals. In some cases you find that there is what is termed a biphasic response – at the lower dose that can be achieved in animals the agent turns out to be pro-cancerous, and at the higher dose it does kill the tumours but also kills the animals or makes them seriously sick. Not good.

However, there are also cases when the results in animals are good. Tumours that are growing fast are slowed down substantially, or else they stop growing completely or even regress and disappear. Let's be clear, each of these responses is significant. Having an agent that slows growth is important, some cancers are highly aggressive so slowing them down is important. Having a tumour stop growing (which is normally what doctors would class as stable disease), is a major achievement. And having tumours shrink and eventually die is precisely what we want. So, while these positive responses are not as common as those great results you see in test tube studies, they are striking nevertheless and should give us grounds for optimism. It's not hard to find studies that show that curcumin, quercetin, sodium bicarbonate, ascorbic acid (vitamin C) and plenty of other agents can deliver positive results in rats.

Great Results?
Onwards to human trials and similar positive results – right? Unfortunately not. History is littered with stellar pre-clinical results that have not translated well into clinical practice. Anti-angiogenic therapies, to take one well-known example, work by stopping tumours sprouting the blood supply they need to survive. In the test tube it works, in rats it works, in humans...well, it's never worked the way the rat models suggest it should. And why? As with the in vitro results, we need to dig below the surface to uncover the major problems with in vivo results.
The first thing to note is that a mouse is a mouse, a rat is a rat and a human is something else altogether. Mice and rats don’t get human cancers. A mouse will get murine cancer, in the same way that dogs are prone to canine cancers and we are prone to human cancers. And, for the purposes of experimentation you can’t afford to wait around to for you mouse to develop cancer spontaneously. In practice this means chemically inducing cancer or the use of engineered mice.

There are mice that are specially bred to spontaneously develop murine cancers. This ensures a supply of mouse patients that can be used to explore how the disease develops or, more usefully, what works in stopping the mice getting sick in the first place. If your interest is in the process of metastasising, then you can watch how the tumour spontaneously develop in these mice, how it progresses and then metastasises to other parts of the body. This is very useful in terms of understanding how these things happen, and there is an awful lot that we don’t know even at very basic levels. However, these are mice cancers, not human. If you’re testing a new treatment, translating from this type of mouse model to human disease is simply not reliable.

An alternative is to use what are called xenograft models. Here you take a mouse and inject a human cancer into it. This becomes the target of investigation and treatments are assessed against these human cancers and not murine ones. But doing this is not straightforward for two simple reasons. First and most obvious is that the mouse immune system is fairly good at rejecting transplanted human cells, even cancer cells. So it’s not easy to get human cancers to grow in a fully immune-competent mouse.

Secondly, and less obviously, the cells that are transplanted are not fully representative of human cancers either. Again, as was pointed out in the first part of this series, standard cell lines of the type that are supplied by cell libraries represent only one cell type. Tumours contains many different cells types, they are not just a blob of the same cell type. Furthermore, cancer cells mutate and adapt, so cells that have been taken from a patient biopsy twenty or thirty years ago and then kept under glass for generation after generation will have changed in order to survive in the test tube environment.

One way round the first issue, and the one most commonly used in practice is to engineer various types of mice with malfunctioning immune systems. There are different flavours of such mice: the nude mouse, the athymic mouse, SCID (severe combined immunodeficiency) and so on. These different flavours of mice have become the model of choice for much cancer (and other disease) research. The immunodeficient mouse enables scientists to plant human cancers into the mice and then letting the tumours develop. Furthermore, this ease of transplanting tumours means that scientists can take any type of cancer and plant it somewhere in the mouse – very often it’s the flank because tumours there are easy to see and measure using callipers. Treatments can be assessed, basic research into the underlying biology can take place, we can learn a lot.

Missing Immunity
But of course we are not dealing with the real thing, by changing the immune system we have removed a key component of real animals and real people. And, as is becoming increasingly clear as the science advances, in real cancer, the immune system is more than just an innocent bystander. Real tumours work in concert with the local environment, recruiting some types of immune cells, subverting others or just sending out all kinds of cytokines and messages. Taking out the immune system removes this essential interplay.

Some of these problems are in the process of being addressed. For example, there is an increasing use of orthotopic xenograft models, where, for example, breast cancers are implanted in the mammary pads, or osteosarcoma cells implanted in bone etc. These at least have the merit of being implanted in the same kind of cells that they would in a human. It’s a step in the right direction, but we still have the issue that these are being implanted in immunodeficient mice. It is also generally the case that these orthotopic tumours do not necessarily grow in the same way in the mouse than they do in humans.

The next step in the process will be the development of mouse tumours that more closely resemble human tumours. These murine tumours will have been engineered to express similar pathways, and carry similar mutations to human cancers. They will be implanted into fully immunocompetent mice, and develop with the mouse immune system in play. There are already examples of such systems available, though in comparison to immunodeficient mice the cost is considerably higher, as is the skill required to care for them.

What about the treatments?
So far we have just focused on how closely mice with cancer match humans with cancer. But there are other things to look at when reading the results of experiments where mice have been treated with different agents. As before, let’s focus on the kind of substances that we can buy over the counter in supplement form: curcumin, quercetin, modified citrus pectins etc. The key question to ask is how was the substance delivered?

As patients or family, we are often looking for evidence that taking something like resveratrol has powerful anticancer activity. We’ve looked at the test tube results and are impressed. Now we look at the mice and rat experiments and the results look good. But are they?

In many animal experiments the substance being tested is not given orally. As with the test tube experiments, it’s often the pure substance that’s tested and not the metabolites. Most  often the substance is injected intravenously or intraperitoneally (injected into the body cavity), or sometimes intramuscularly or subcutaneously. Often the substance has been dissolved in DMSO or other chemical. From the point of view of the experimenter this makes sense – you can guarantee the dose and control how much gets into the body. For us, this is not much help. We won’t be looking to inject, we want to take it in tablet or powder form.

There are cases, however, where the dose is oral. Normally this is described as by gavage – in other words it is inserted into the stomach via a tube. In some papers you’ll see that the agent is mixed into the diet or the drink and the mouse isn’t force fed. These are the experiments that are most interesting from our point of view. Good results here are normally worth taking note of. There are some things to be aware of, the most obvious being the translation of a mouse dose to a human dose, but there are standard formulae for doing that.

So, have we found our perfect experiment? If mice eating substance X at dose Y show good results against the type of cancer we are interested in, does that mean we should go ahead and start taking substance X too? Unfortunately we have to come back to looking at what type of mice were used…

Even if the results are good, you still need to know whether the mice were immunodeficient or not. You need to know whether the tumours were orthotopic or not. If human cancers were used, what subtype and how closely do the mice models match the human disease? You need to know what else was going on as regards diet and so on.

What's the point?
The point of all this is not to say that looking at mice models of cancer is no good. It’s just that there are so many caveats and so many discrepancies between what goes on in us and what goes on in mice and rats that we need to exercise caution. Time and again promising results in mice turn out to be failures in human trials. We, as patients or carers, need to exercise the same caution as the drug companies should when looking at results.

This also begs the question of how many wrong turns science has taken because of the difficulties of using mouse models. Who knows how many positive results have been missed because the mice were missing key components of the immune system? And how much time has been lost because fantastic results disappear when the immune system does come into play?

Finally, if you do find some great results in your search, then the obvious next place to look is at human trials. And that’s where we’ll look in the next part of this series.

No comments:

Post a comment