Precision medicine has the potential to change fundamentally how health care is practiced, but requires a health care workforce that understands the complexities of this field. One important component of Precision Medicine is the use of an individual’s genomic information to offer targeted treatment, tailored to the individual. Our course aims to provide participants with some baseline knowledge of genomics, an overview of the clinical applications of genomic medicine, the skills to evaluate the clinical validity and utility of new tests, and an appreciation of the associated ethical and social issues inherent in this field. The course is geared toward practicing health care providers, although it should be accessible to anyone with a background in the biological sciences and a basic understanding of genetics. It is designed to be succinct and clinically-focused, offering both conceptual and practical information about real-world applications of genomics. Two lessons offer a basic primer on molecular genomics relevant to the individual patient as well as to patient populations. The remaining five lessons focus on five applications of genomics and present the material as case studies, highlighting the strengths, limitations, and issues that arise in the use of each test.
Module 3: Tumor Genetic Landscape
Loading...
From the course by University of California, San Francisco
Genomic and Precision Medicine
75 ratings
University of California, San Francisco
Genomic and Precision Medicine
75 ratings
From the lesson
Week 7
Our final lecture explores tumor profiling for targeting cancer treatment and the use of blood-based gene expression profiles in cancer prognosis.
Meet the Instructors
Jeanette McCarthy, MPH, PhDAdjunct Associate Professor
UCSF Department of Medicine
Robert L. Nussbaum, MDChief and Professor of Medicine
Division of Genomic Medicine
Next we're going to explore what the genetic landscape of a tumor looks like. Tumor cells harbor a large number of mutations and these mutations affect the gene expression and ultimately, the proteins that are expressed on the cell surface and also within the tumor within, within the nucleus of the cell. So what I wanted to talk about next is how we actually measure tumor gene expression. And one of the things that we've been doing for a number of years is to measure single markers things like HER2, estrogen receptor, and progesterone receptor. These are three common markers that have been evaluated on the surface of breast tumors for a number of years. But now, with some of the newer genomic technologies, we're able to look at genome wide measures of gene expression. And, we do this using expression arrays or other methods, such as RNA-seq. I just wanted to describe these very briefly, as you may encounter them in, in your clinical practice. So, expression arrays, basically, what an expression micro array is, is it's a slide for example, a glass slide, to which DNA probes have been added. These probes are short segments of DNA that represent a gene. And we can put tens of thousands, thousands to tens of thousands of these on a slide and evaluate them all at, at one go. So we could put on markers for all, you know, 20 some thousand genes in the genome.
And they're single-stranded DNA molecules. What we do next is we take a tumor sample and we isolate the RNA. What does RNA represent? It represents all of the genes that are expressed in that tumor. So we take the RNA from the tumor and we label it with fluorescent markers and then we mix it with this fixed probe on the slide, and we see if it sticks or not, if it anneals. It will anneal or bind if it's complimentary to this, to the DNA that's on the slide. If it's fully complimentary, we'll get a fluorescence signal. And that signal will tell you number one, is that specific gene that this probe represents being expressed in that tumor sample, and the intensity of that signal will tell you how much that gene is being expressed. So it's both a qualitative and quantitative assay. If the probe does not anneal or bind to this complimentary DNA, you won't get any signal. So this is a nice way to evaluate across the genome which genes are expressed and how much they are expressed, relative to each other in one experiment.
Another approach to do essentially the same thing, but using next generation sequencing, is called RNA sequencing or RNA-seq. In this case, what you're doing is you're taking the tumor sample and you're again isolating the RNA, and you've got your little pieces of RNA here, which represent different genes that are expressed in that tumor. And then we sequence each of these. And in sequencing these you'll recall from a couple lectures back, we end up aligning these sequences. And we have what's called a read depth, which tells you how many times you've observed these different fragments of sequence. And if you basically count those, that will give you a good indication of the quantitative measure of gene expression in this tumor. So either using expression micro-arrays or RNA-seq, we're able to quantify the amount of gene expression in each of these tumors and identify which genes are turned on and which genes are turned off.
Another thing that we might want to be able to do is actually measure, at the level of DNA, the somatic mutations that occur in a tumor. And so if you think about a tissue and the blue represents the tumor and the yellow represents the surrounding normal tissue. What we can do is we can sequence the DNA from both the tumor and the normal and compare the two. Why would we want to do this? Well if we want to know which of these mutations are somatic, in other words, which of these are the acquired mutations that may be driving this cancer. We need to compare it with the normal tissue to ensure that these weren't some inherited mutation that, that was found in the tissue normally. So, we take the DNA from the tumor, and we sequence it. And we take the DNA from the normal surrounding tissue and we sequence that. The normal tissue will just give you germ line variance.
The tumor DNA will give you germ line variance and somatic changes. And by comparing these two, you can essentially subtract out and, and, and end up with what these somatic changes are in that tumor.
A number of different groups are carrying out, carrying out large-scale efforts to characterize tumors at the molecular level. We have the Cancer Genome Atlas in the US and the International Cancer Genome Consortium. Both of these groups have under taken projects to sequence hundreds, if not thousands, of tumors across a number of different tumor types, and deposit that information into the public domain, so that any researcher can evaluate it.
One of the databases that houses information about somatic variation in tumors is the COSMIC database. In this database, you can access information, again about how many genes are mutated, what the specific mutations are. It's a very rich resource for any researcher who wants to learn more about the genetic landscape of tumors.
Okay, so how many genes are actually mutated in the average solid tumor? Some work was done much of this based on the Cancer Genome Atlas project to look at how many mutations were found across different tumor types. And this was published in Volgastein et al, from 2013 and Science.
What they report is that on average 33 to 66 genes are mutated in a solid tumor. And this number varies depending on the cancer type and most of these variants are actually single nucleotide variants which is what we would expect. It mirrors what is going on in, in the mutational landscape of a, of a normal genome.
So why do these mutations, why do these tumor types vary in the number of mutated genes in cancer? Well, if you look at the diagram here, it shows you what the different cancers are and how many genes are mutated. And, and it begins to make a little sense. You look at something like the pediatric cancers down here. They're characterized by, by a very low mutational burden. Why is that? Well, they're pediatric patients, and we know that cancer accumulates over a lifetime. The older a patient is, the more likely that, that their tumor will have acquired additional mutations. Then, you have examples where the cancer is due to an environmental carcinogen, lung cancers and melanoma. They have a relatively high burden, because we have several mechanisms that are contributing to the mutational burden. And then on the far left here, on the extreme off the charts, we have colorectal cancer. And this makes sense. As I mentioned, Lynch syndrome, one of the familial colorectal cancer syndromes, is actually due to mutations in DNA repair enzymes. So it makes sense that a tumor cell, a colorectal cancer tumor cell of this nature would have acquired a lot more mutations. [BLANK_AUDIO]. In the cancer world, we talk about drivers and passengers. Drivers are mutations that provide a selective growth advantage. Passenger mutations are just random mutations that have occurred but they're really not driving this neo-plastic growth. They're just benign, just along for the ride. And so how do we actually identify what the driver mutations are and, and how do they differ from the passenger mutations? So to do this again, researchers have looked at hundreds of tumors across any given tumor type and tried to look at a couple of different features of them. The first feature is the frequency, and if you look at the, the diagram here for colorectal cancer, looking across hundreds of color rectal cancer samples, you can see a plot of the frequency, with which different genes are mutated in these tumors. And, you can begin to see patterns. There are some genes that are mutated at a relatively low frequency, these are hills, there are some genes that are not mutated at all, those don't even show up here. And there are some genes where, when you look across different tumor samples from different patients and the same kind of cancer, you see these genes are mutated quite often. Things like APC, which is one of the colon cancer genes that we know about. KRAS, TP53 and others all show high levels of mutation in colorectal cancer. So the idea is that maybe these are actually drivers. They're the ones propelling the disease forward. So it's important to recognize them in that way.
Another way to identify driver genes, is by look at different patterns of mutation. Not just the frequency but the patterns. Something like what was found in the IDH1 gene in brain tumors. For this tumor, what we find is not only was the IDH1 gene mutated quite often, but it was essentially the same mutation every time. So this is telling us that not only is this gene probably a driver of cancer, but this specific mutation is probably a driver as well.
One of the questions that we've been trying to answer is, how many driver genes exist out there, and some of the research based on sequencing of nearly 3,300 tumors reveal that there are about 125 mutational mutation driver genes that were identified. 71 of them, tumor suppressor genes, 54 oncogenes, so. Now we have a ballpark of how many different genes that we're, we're dealing with in terms of driving cancer forward.
One of the things with cancer, the thing that usually kills you, is a metastasis. That is when the cancer invades the blood stream and is able to blood or lymph system, and is able to circulate and arrive at different places in the body and lodge itself and begin to grow there. The most common sites for metastasis include the brain, lungs, liver, and lymph nodes. So one of the questions that researchers have been trying to answer is whether there are specific genes that become mutated which then lead to metastasis. Unfortunately, research to date has not been able to identify any such genes controlling metastasis in that way. And the recent thinking is that Metastatic events are really just a stochastic process. Once a tumor cell acquires a certain number of mutations, it just becomes metastatic and, and, and moves on. Metastatic disease accounts for about 90% of cancer deaths, so this is an important area to focus on.
Coursera provides universal access to the world’s best education, partnering with top universities and organizations to offer courses online.
© 2017 Coursera Inc. All rights reserved.
