So, in my previous post, Damn statistics, I explained how Epidemiology is used to identify risk factors, by finding correlations between different activities, such as smoking or meat-eating and cases of cancer. In this post, I’m going to talk about Drug Development, because the way that this type of work gets started uses much of the same techniques I talked about before.
How do researchers go about deciding how to develop a new drug? Well, obviously, they will want a drug that targets some process or factor that is important for tumour growth, so if you can disrupt the process or factor with the drug, the tumour stops growing. But how do the researchers decide which processes or factors ARE important for tumour growth or, in contrast, which ones are not that important?
Well, the techniques that I spoke about in Damn Statistics are also used here, in order to identify new targets for Drug Development. But, rather than looking at differences in lifestyle or environment between different patients, researchers will look for differences between the make up of the tumours themselves.
So, let’s say that there are some breast cancer researchers, who are interested in a specific growth factor, which we’ll call “Bill” (why? Why not!) The researchers know that “Bill” is commonly found in breast tissue, so they suspect that it might also be involved in the development of breast cancer. In other words, in some people, the researchers think that too much “Bill” is made in the breast tissue, which makes the cells start to grow and divide too much and too fast. If they are right, then the constant “Bill” would make the cell growth run out of control, and….Bingo! The person gets breast cancer.
If this hypothesis is true, then this would make “Bill” a good target to try and develop a new drug against. So, the researchers will look to see if the hypothesis is right. They will do this by having a look at the tumours of previous and current breast cancer patients.
Every patient who is suspected of having cancer will have a biopsy taken – basically, the doctors cut off a bit of the suspect tissue and then run tests to see if it is cancerous or not. Every patient is asked if the biopsy sample can be kept to help in future research and if they agree, the biopsy tissue is preserved in special tissue banks. This means that there is a large number of samples from different patients. As the medical history of all of the patients is known to the researchers, they will know which ones survived and which ones didn’t. And of the ones that didn’t, they will know whether they succumbed after a few months, or whether they managed to to survive for 1 year, or 2 years, or 5 years…..or maybe even longer.
So, the researchers can use all of this information to test their hypothesis that “Bill” is linked to breast cancer. What they will do, is take wafer-thin slices of the stored biopsy tissue and stain them with dyes which only stick to “Bill” (it’s a bit more complex than that, obviously, but you get the general idea – I’ll explain in more detail another time). The upshot is that if “Bill” is there, the tissue will be stained. If “Bill” is not there, it won’t. Also, the more “Bill” there is, the stronger the stain will be. If you do this for, say, 100 samples, you could end up with a range of results. Some of the samples could have little staining (ie. low levels of “Bill”), some may be heavily stained (ie. high levels of “Bill”) and some might be unstained (ie. no “Bill” at all). The researchers can then separate out the samples into groups (called “stratification”) with low, high and negative “Bill” levels.
The researchers can then check the records, to see what happened to the patients the biopsies came from. Did they have a low stage disease (ie. relatively benign tumour that was easily treated), or a high stage disease (ie. aggressive, malignant tumour which needed intense treatment)? Did the patient survive & recover? If not, how long did they survive? Months? Years? And, do the levels of “Bill” in the biopsy tissues correlate with a particular outcome?
This is the point where the researchers can start to test their hypothesis and see if “Bill” is linked to cancer or not. So, they might discover that all of the biopsy samples with no “Bill” came from patients with benign, easily treated tumour, while those samples that did have “Bill” came from patients that had aggressive diseases. Or, they might find that ALL the biopsies had “Bill”, but those with low levels came from patients who lived longer or survived, while those with high “Bill” came from patients who succumbed quickly. Either of these results would mean that “Bill” correlated with outcome, meaning that the research supported the original hypothesis.
This would give the researchers the confidence to start suggesting that “Bill” could be a good target for drug development, ie. that it would be a good idea to try and find chemicals that stuck to “Bill” and stop it working.
Here’s an example of how it works. The figure is taken from this study, which looked at patients with osteosarcoma (bone cancer). This type of graph is called a Kaplan-Meier plot and shows the effects of a “Bill” called Beta-Catenin. This particular “Bill” can switch on certain genes and is involved in sticking cells together inside tissues. But, cancer cells can make too much Beta-Catenin and the Kaplan-Meier plot shows how the survival of patients correlates with the amount of Beta-Catenin their cancers make.
The way you read it is simple. Look at each “staircase”. Each time the line “steps down”, it means that a patient has died. So, the more rapidly the patients die, the steeper the “staircase” looks. Conversely, the more slowly the patients die, the shallower the “staircase” looks. So, from this plot it is obvious that the patients with tumours which make a lot of Beta-Catenin (ie. “positive” in the figure), are dying more quickly than the patients with tumours which lack Beta-Catenin (ie. “negative” in the figure). Now, this would suggest that Beta-Catenin would be a good target for Drug Development and, indeed, a lot of time and effort has gone into making a drug which targets it.
In the same way, many other chemotherapy drugs, both those being used now, and those being designed, started with studies like these: by looking at different “Bills” to see if a correlation could be found that could make them an attractive target. The next step is to try and find molecules which stick to the “Bill” and then try and modify the molecule to make it inhibit the “Bill” more effectively and then tweak it enough, so that it can be made as a drug that can be given to patients….
…. But that’s another story…
Lu, Y., Guan, G., Chen, J., Hu, B., Sun, C., Ma, Q., Wen, Y., Qiu, X., & Zhou, Y. (2015). Aberrant CXCR4 and β-catenin expression in osteosarcoma correlates with patient survival Oncology Letters DOI: 10.3892/ol.2015.3535