The Cancer Genome Atlas (TCGA)

Bookmark and Share
Published: 21 Apr 2011
Views: 5373
Dr Paul Spellman - National Cancer Institute, Bethesda, USA
Dr Paul Spellman talks about The Cancer Genome Atlas (TCGA) and how this could help further the treatment of cancer. TCGA is a project working to catalogue genetic mutations responsible for cancer. Clinicians are sequencing the genomes of patients with any of 20 different cancers and hope that this could target clinical trials at the specific patient sub-groups that would benefit most.

Dr Spellman explains how an increasing number of laboratories are becoming able to conduct genome sequencing and contribute to the TGCA project, discusses how clinicians could apply the findings in practice to decide on treatment and effect patient outlook and suggests that in future patients may start to request for their genome to be sequenced in order to aid their treatment.

AACR 102nd Annual Meeting, 2—6 April 2011, Orlando, Florida

The Cancer Genome Atlas (TCGA)

Dr Paul Spellman – National Cancer Institute, Bethesda, USA is here at the AACR Annual Meeting. Paul Spellman from the NCI Bethesda, you are doing one of the biggest jobs around looking at the cancer genome atlas, TCGA, appropriately named, and you are looking at twenty different cancers typically, and you are finding out what it is about the genes in those cancers that actually influence the progress of the disease and therefore guide clinicians. What can you tell me that’s new? Because we already have groups of patients that we know need to be treated in a certain way, but can you refine that now that you have this massive atlas available?

We are identifying some sub-markers that are clinically actionable, that might exist in a very small fraction of patients. So work we’ve been doing in ovarian cancer shows the presence of genetic abnormalities that are already clinically targetable, but in 2% of patients. So the idea that you have to target 2% of all patients with high grade serous ovarian cancer would be a very small number, in the US there would be 400 patients a year.

Could you pick out those sub-groups of patients and exclude all other patients from a particular directed therapy?

That would be the implication and I think the challenge for the clinical group is to actually figure out how to run trials where only a very small number of patients would be entry eligible. The other major goal would be to pick out the sets of patients where there might be 2% of patients with this abnormality and 2% with that abnormality, and use informatic analyses to figure out which patients are the same and therefore it is not 2% in this trial and 2% in that trial, it is 4% that have a combination of abnormalities. And that would be a strategy for increasing the size of the population so you didn’t have to go down to these very, very small numbers. But it may be that you have a limited target and that there are patients who will have excellent therapeutic outcomes from very particular targeted therapies.

Looking at the genome atlas, looking at genes in fact, is quite difficult. Is this something that can get into the lab and be done in the average hospital?

It’s getting closer. As the ability to do sequencing drops dramatically it is now becoming routine for the research community to be able to sequence all the genes in the human genome from a regular research lab. That is something that, apart from my job within the cancer genome atlas, I have a research group at Lawrence Berkeley Laboratory and we do that in my group and my group is in the order of 10 people. So it is something that a research lab can do and it something that is moving towards the clinical testing arena where we will have probably, certainly within the next 5 years, in the next 2 years the ability to have complete gene exome sequencing and complete genome sequencing in individual patients’ tumours.

Complete genome sequencing, it sounds formidable, it is quite complex; how do you recommend doctors to get their heads around some of the details and actually apply some of the knowledge?

It will be necessary to work closely with the clinical testing laboratories that would provide the service, I think. As long as it can be done in that sort of way the data should come back and list the aberrations that are present in each individual’s case, but there will be many of them that probably mean something and we won’t know about for a long time. So that it is individual case reports of patients who have specific abnormalities in clinical practice and then how they respond to therapies will be quite valuable from a research point of view, and so I think that publishing those interesting case studies will be quite useful.

If you do identify some of these sub-groups very accurately, how much might you improve the outlook for an individual patient do you think?

It’s a good question, I would use trastuzumab as a prime example of potential outcome. So in the metastatic setting it is a modest increase in overall survival but there are a lot of patients who have HER2 positive disease who are effectively cured in the pre-metastatic setting, in the local lesion setting. So the big benefit of trastuzumab is not that we get a small increase in patient survival for those patients who have metastatic disease; it’s that patients who don’t have metastatic disease can be cured by it. And I think that the same thing will be true for these other lesions, so we will prove that they have therapeutic benefit in the metastatic setting and then we will apply them clinically in a local disease setting.

So that’s an example in breast cancer. You also mentioned about ovarian cancer; do you have any other hot tips?

Well TCJ is pursuing twenty different tumours and in collaboration with the ICGC there are now forty different projects being taken on worldwide to sequence tumour genomes. And so TCJ, for example, is doing pancreas, lung squamous, lung adeno and colon, rectal, so a very long list of the major killers in the United States. And then the ICGC projects are tackling tumours around the world that are major killers around the rest of the world. So those data are being shipped online in more or less real time and the data themselves are quite complex. So directly to clinicians I think there will be slight lag as we try to figure out how migrate those data to clinicians.

You mentioned trastuzumab. That applies to a large number of patients with breast cancer, for example.


But you were saying that you can target very small groups. Are there examples of what you might do for them?

In our ovarian serous carcinoma study, high grade serous ovarian cancers, there is one patient who has a BRAF mutation out of 312 patients. That is a mutation we understand, it’s a target for which there is a drug that in some aspects works; it looks like it’s working in the melanoma setting. So as a clinician if I saw a patient with a BRAF mutation, even in an indication where BRAF is not supposed to be a driver, I would certainly consider it and if the patient has got metastatic disease, which they mostly do in high grade serous, I wouldn’t expect to cure them but I might expect it to help them.

So you might, by knowing what’s in the gene, you might be able to pick out a well known treatment.


That you wouldn’t have thought of otherwise?

Absolutely, that’s right. This has been published before but high grade serous ovarian cancers also have, a couple of per cent of the tumours have, focal amplifications of HER2 so you could use trastuzumab for those patients as well. So there are clinical trials that show that there is no benefit there, so it clearly is not a perfect theory but it is one that I think is worth pursuing.

How would you like to see cancer doctors using the cancer genome atlas, TCGA?

Ideally we would see these translated into clinical trials. Speaking from personal history, my father-in-law was diagnosed with a neuroendocrine carcinoma about ten years ago and he lived about 6 years, and he was getting extremely sick right around the time that the EGFR work came out showing the point mutations in EGFR and the ability for imatinib to help those patients. So we pushed to have his EGFR local sequenced and today I would have pushed to have his whole genome sequenced to be able to look for anything that might have helped him. So I think individual clinicians will start to see this in the clinic; that patients want their genome sequenced because they are going to be able to afford to do it themselves in a lot of cases. Even if the insurance won’t cover it, it’s a few thousand dollars at this point; 5000 dollars for a couple of different companies who will sequence your genome.

So do you recommend people to have their genome sequenced?

I don’t know that I would recommend it, and it will certainly put strain on the infrastructure in the medical community but, as someone who watched someone they loved die, I would have done it. And so that’s not necessarily policy but it is something I would have done.

So the science, the detailed science of the gene, can make important therapeutic strides forwards?

I think it absolutely can and, to me, it would be worth the risk.

Paul Spellman, it’s great having you here with us on

Thank you very much.