| Patrick O. Brown |
 |
|
|
 |
 |
|
[Slide 8]
[Slide 9]
[Slide 10]
[Slide 11]
|
For example, by looking at the levels of expression of this set of genes that are expressed in mitochondria and involved in aerobic respiration, we can predict that these tissues with bright yellow color, heart, skeletal muscle, kidney, brain, have the highest oxygen demand.
By looking at closely at these genes that are involved in progression of the cell division cycle, we can identify which of these cells and tissues have particularly rapid proliferation rates.
And by looking closely at these genes that are expressed only in males, we can tell which of these cells and tissues came from men or boys, and which came from women or girls.
Of course, there are hundreds of other features that you can see in these gene expression patterns that give us similar information about the molecular and biological properties of the samples, so that we can read the gene expression patterns using the map like this as a kind of picture of the biological characteristics of the specimens.
[Slide 8]
Well, as Steve has already told you, many scientists around the world have been using DNA microarrays to study gene expression patterns in human diseases, particularly cancer.
What you see here is a map in a very similar format to the one I just showed, that summarizes gene expression patterns in a set of several hundreds different cancer samples from several hundreds different patients representing many different cancer diagnoses.
In this map, one of the features you can see from this tree on the top that represents the similarities in the expression patterns of tumors, which I've color-coded according to the classical diagnosis of the cancer, is that the gene expression patterns readily distinguish breast cancers from prostate cancers from brain tumors, kidney cancers, and so forth. The different kinds of cancers have correspondingly very different gene expression patterns that give rise to the differences in the cell's properties.
And the global gene expression patterns of cancers are also readily distinguishable from the normal tissues of the body.
You can see from these patterns a lot of information about physiology and molecular regulation in each of these cancers. We can also see directly which genes are selectively expressed in each tumor compared to normal tissues, and which potential molecular targets are available for a targeted drug therapy. And with all the new molecularly targeted drugs that are under development and starting to see clinical use, this kind of analysis is going to be a very important component of the individualized therapy for these patients.
By identifying the molecular patterns that distinguish each of these cancers from the normal tissues of the body, these patterns provide an excellent starting point for new screening and diagnostic tests for early detection and diagnosis of cancer.
One of the most important contributions of DNA microarray analysis to studies of cancer has been in the development of a new approach to classification and diagnosis of cancer based on gene expression programs. Let's look at a couple of examples.
Even though you can easily tell apart all the breast cancers from any of these other different kinds of cancer, if you look more closely, you can see that not all of these breast cancers are alike and in fact, every single patient's tumor is unique in its gene expression pattern. This is exciting because we've known for a very long time that two patients who have a breast cancer that looks identical under the microscope, that we would classify identically, can have very different clinical courses and different responses to therapy. And now we can start to see the molecular differences that we suspect are responsible for differences in behavior. And indeed, we are finding again and again that the molecular subtypes we define on the basis of gene expression programs can predict differences in therapeutic response and clinical course in cancers that were previously indistinguishable.
[Slide 9]
I am going to give a little example of how we use this approach to study a particular kind of cancer. I am going to focus a stomach cancer, which is a very important medical problem here in Japan, as you all know, and it is the second leading cause of death from cancer in the world.
[Slide 10]
A post-doc of my lab, Xin Chen, and these collaborators: S.Y. Leung, S.T. Yuen and Sam So, used DNA microarrays to profile gene expressions in more than 100 tumor samples from patients with stomach cancer, and as you can see from all the variation in this map, all the variations in their gene expression patterns, no two cancers from different patients were alike. In fact, we found that they were incredibly diverse.
There's a ton of useful information about the molecular characteristics of these tumors in our dataset like this. But with so much variation, how do we focus our attention? How can we choose particular genes that are most likely to be informative to the study?
[Slide 11]
One way is to deliberately look for the genes, whose expression is especially strongly associated with important biological and clinical characteristics of this cancer - for example, patient's survival - and we did that. When we looked for genes whose expression patterns paralleled patient's survival, we found that one gene - PLA2 group 2A, a phospholipase - was strongly associated with the survival of the patient after the tumor was removed.
When we divided all the gastric cancer patients into two groups, one group with very high expression of this gene and another group with low expression - (the range of expression between the highest and lowest differed by more than 2,000 )- what we found was that the patients who expressed a lot of this enzyme were more than 3 times more likely to survive for 5 years or longer than the patients who expressed this gene at a low level. In fact, by 4 years after surgery 80 percent of the patients with low level of expression had died from the disease, whereas 2/3 of the patients with high expression of this enzyme were still alive after 6 years.
We were particularly excited about this gene, because the equivalent gene in a mouse had been shown several years ago to reduce the severity of a particular kind of intestinal tumor in mice. That suggested that this gene might be a potential tumor suppressor gene.
It's even more exciting because of things that we know about the biochemistry of this enzyme. The results suggest some possible ways that we might be able to help patients with stomach cancer.
First, trivially, just by measuring the expression of this gene, we can already recognize one group of patients with once high level of expression, who have a very good chance of being cured, and the second group who are likely to need much more intensive therapy.
|
|
|
