Cracking the Cancer Code

December 17, 2009 9:59 AM

Cracking the Cancer Code: We already know that all cancers are caused by DNA mutations acquired during a person's lifetime. But what mutations actually cause cancer? We may be one step closer to finding out. International research teams led by the Cancer Genome Project at the Wellcome Trust Sanger Institute have now mapped the entire genetic code of two of the most common human cancers: lung and skin (malignant melanoma). Their findings have the potential to revolutionize preventative and treatment therapies as well as pave the way for new early detection tests. More.

Additional research is now underway to map the mutations of cancerous breast, hepatic, oral, stomach, ovarian, pancreatic and brain cells.

posted by zarq (36 comments total)

From The Guardian:

"The researchers took diseased cells from a 45-year-old man with a type of skin cancer called malignant melanoma, and from a 55-year-old man with small cell lung cancer. They then used advanced genetic sequencing machines to read the full genomes of both the cancer cells and healthy tissues taken from the same patients.

By comparing the genetic makeup of the diseased and healthy cells, the scientists created catalogues of all the mutations found only in the cancerous tissues. Most of these genetic glitches are harmless, but every once in a while a mutation causes major damage that pushes a cell closer to becoming cancerous.

The scientists focused on skin and lung cancer because the environmental causes are well known. Most melanomas are triggered by overexposure to ultraviolet rays in sunlight as a child, while almost all small cell lung cancer is caused by smoking."

posted by zarq at 9:59 AM on December 17, 2009

Wow. Great post. I wonder though why they are confident that this one man's cancer is typical. If cancer is caused by mutations, wouldn't we expect wild variations between individual cancers?
posted by msalt at 10:02 AM on December 17, 2009

Researchers also gained insights into how more than 60 carcinogens associated with cigarette smoke bind to and chemically modify human DNA, ultimately leading to cancer-causing genetic mutations.
posted by zarq at 10:03 AM on December 17, 2009

I do so hate having to rely on poor secondhand reporting of new research.

A comprehensive catalogue of somatic mutations from a human cancer genome

A small-cell lung cancer genome with complex signatures of tobacco exposure
posted by killdevil at 10:24 AM on December 17, 2009 [3 favorites]

So we could have a DNA backup and just revert to last save every year?
posted by TwelveTwo at 10:25 AM on December 17, 2009 [2 favorites]

So does nothing this make it easier to cure cancer, or to discriminate against people who are going to fall victim to it?
posted by jefficator at 10:30 AM on December 17, 2009

knowing*
posted by jefficator at 10:33 AM on December 17, 2009

I do so hate having to rely on poor secondhand reporting of new research.

Thanks for these links. Google Scholar turned up nothing, and all I could find at Nature were pay portals.

So what did the media get wrong? (Assuming of course, that the "poor reporting" you're referring to wasn't my own in this FPP.)
posted by zarq at 10:33 AM on December 17, 2009

What good timing!
posted by Pastabagel at 10:38 AM on December 17, 2009

Oops, I meant to link to this comment I made earlier this morning.
posted by Pastabagel at 10:39 AM on December 17, 2009

So does nothing this make it easier to cure cancer, or to discriminate against people who are going to fall victim to it?

They're mapping the mutations that actually trigger the cancer, not ones that indicate a genetic predisposition to it--if you've got these mutations, you already have the cancer.
posted by Sidhedevil at 10:42 AM on December 17, 2009

If cancer is caused by mutations, wouldn't we expect wild variations between individual cancers?

Good question. I'm honestly not sure, and hope someone who is more knowledgeable will weigh in with an answer.

I do know that some diseases (Cystic Fibrosis comes to mind) can be caused by many different mutations, but not all of them are required for the disease to manifest.
posted by zarq at 10:47 AM on December 17, 2009

If cancer is caused by mutations, wouldn't we expect wild variations between individual cancers?

Absolutely. This is one of the reasons that we still haven't found reliable treatments for cancers. While tumors all share certain characteristics, each tumor has a unique mutation profile. Some of these mutations are inherited, others are accumulated throughout a person's life, and they are scattered throughout the genome. Furthermore, there are different forms of mutations - some are single base-pair changes, others are large sections of DNA that can be moved around, duplicated, or deleted.

One of the ways that we get around this is by looking at genomic data from *lots* of tumors. By comparing them, we can begin to tease out which mutations are rare, and probably random, and which are seen in lots of tumors, which suggests that they're being selected for.

We also do this by thinking beyond single genes and remembering that many genes act together in pathways. It may be the case that if gene A and gene B work together to slow down cell growth, a mutation in either of them may be sufficient to allow the tumor to grow rapidly. These kinds of pattern only emerges, though, when you look at lots of different tumors.

That's where projects like The Cancer Genome Atlas come in. They're sequencing hundreds of tumors from lots of different cancer types. The first rough data from glioblastoma was released about a year ago. They're now working on lung and ovarian cancers and with more recent funding from ARRA, they're ramping up to do about two dozen more types of cancer.

Moving beyond research, all of this variation also means that patient A's tumor may need a different treatment plan than patient B's tumor. This is where the idea of personal genomics comes in. We're getting to the point where genome sequencing is cheap, and it may soon be routine for your oncologists to sequence your tumor, determine which cancer-driving mutations you have, and give you a personalized cocktail of drugs and therapy that is most likely to attack your tumor's specific problems.

All of this is still ten or twenty years down the road, but make no mistake: This will change medicine in very fundamental ways.


(disclaimer: The glioblastoma data is sort of a self-link, since I'm on the (~200 person) author list of that paper. I'm tangentally involved in the TCGA project and use the data in my research).
posted by chrisamiller at 10:53 AM on December 17, 2009 [12 favorites]

FWIW, I was accusing not your post but those sciencedaily articles articles of "poor reporting."

Thanks for these links. Google Scholar turned up nothing, and all I could find at Nature were pay portals.

Hooray for publicly-funded science being reported in journals that are largely inaccessible to the public!
posted by killdevil at 10:57 AM on December 17, 2009 [4 favorites]

Interesting results. It'll get more interesting when they have 30-40 genomes of each cancer type sequenced so they can sift out the irrelevant mutations statistically and associate them all with existing tumor types.

Wonder when they'll have the actual sequence data browsable at Ensembl or UCSC.
posted by benzenedream at 11:03 AM on December 17, 2009

FWIW, I was accusing not your post but those sciencedaily articles articles of "poor reporting."

I figured. I hope I didn't come off as too defensive. Realized after I hit "post comment" that I had forgotten to add in a smiley face. :)

Hooray for publicly-funded science being reported in journals that are largely inaccessible to the public!

Exactly!

When Google Scholar was launched I was convinced it would change the playing field and increase free public access. Unfortunately, its only accomplishment seems to be as an index to article abstracts.
posted by zarq at 11:12 AM on December 17, 2009

In lung cancer the damage is almost entirely caused by smoking and in skin cancer or malignant melanoma by ultra violent sunlight.

Nice proofreading, Telegraph.
posted by RogerB at 11:28 AM on December 17, 2009 [3 favorites]

malignant melanoma by ultra violent sunlight

What we were after now was the old surprise visit to the Korova Tanning Salon. That was a real kick and good for laughs and lashings of the old ultra violent.
posted by benzenedream at 11:45 AM on December 17, 2009 [5 favorites]

malignant melanoma by ultra violent sunlight

As someone who has been diagnosed with melanoma (so far localized, removed, etc. etc.) I can attest to the ultra violent nature of sunlight.
posted by Danf at 11:48 AM on December 17, 2009 [1 favorite]

There's kind of two sequencing efforts going on in cancer circles right now. Things like the above mentioned study, where they sequence EVERYTHING but only for a few samples. This is useful for finding mutations in novel genes that maybe we didn't realize were altered in cancer.

The other tactic is the above-mentioned study where scientists take a list of 601 genes they're interested in and sequence them in 91 tumors. Although the Nature paper is not publicly available, some of the raw data is! We use this kind of data for 1) discovering novel mutations in genes we already knew were altered in other cancers. 2) finding out if certain mutations are likely to occur together.

As sequencing prices come down the two tactics will eventually merge, but that's a few years down the road yet. And sequencing is only one part of the puzzle. Cells will not just have mutations in their DNA, they'll gain and lose huge chunks of DNA, and methylate and demethylate genes, both of which can contribute to carcinogenesis.

What's the goal of all of this? One of the newish things in cancer treatment is the idea of targeted cancer therapy. Instead of using chemotherapy or radiation which will attack all rapidly dividing cells in a patient (the cancer, but also skin, hair, stomach lining, etc) if we knew exactly what was altered in a patient's tumor we could give them inhibitors of those specific alterations. The poster child for this therapy is Imatinib/Gleevec. With a single pill 90% of CML patients can have a durable remission (but then the leukemia mutates around the treatment).
posted by Endure You Are Not Alone at 12:27 PM on December 17, 2009 [1 favorite]

Hooray for publicly-funded science being reported in journals that are largely inaccessible to the public


All manuscripts from NIH-funded work must be submitted for public access.

http://publicaccess.nih.gov/
posted by rxrfrx at 12:30 PM on December 17, 2009 [2 favorites]

There is a disclaimer at the very end of this article:

Author information: Reprints and permissions information is available at www.nature.com/reprints. This paper is distributed under the terms of the Creative Commons Attribution Non-Commercial-Share-Alike licence, and is freely available to all readers at www.nature.com/nature.

(Emphasis mine)

Except, it isn't. It seems as if one either needs a paid subscription to Nature or to pay for permission to access this single article.

Is it possible that their embargo entitles Nature to limit access for a fee, temporarily?
posted by zarq at 12:41 PM on December 17, 2009

They're mapping the mutations that actually trigger the cancer, not ones that indicate a genetic predisposition to it

Well, yes and no. In order to tell which mutations are tumor-specific, many studies use matched normals, so they sequence chunks of DNA from healthy cells too. This can give some information on whether the person was predisposed to cancer or not.

As one example of how predisposition might work, remember that each person has two copies of each gene. So if one of them is defective, the other one may compensate, and often times you can keep chugging along with no problems. It's problematic, though because your cells no longer have that built-in redundancy. If you get a mutation in the other one, there's no backup, and it may push you down the slope towards tumorigenesis.

Long term, we're absolutely interested in understanding the risks associated with different mutations throughout your genome. Not so we can institute some Gattaca-like discrimination program, but so we can prevent diseases before they start. If you know that you're at high-risk for cancer, you may be sure to put on extra sunscreen. If you know you're susceptible to high blood pressure, there might be drugs you can take or dietary changes you can make that help prevent you from ever getting it.
posted by chrisamiller at 12:47 PM on December 17, 2009 [1 favorite]

We already know that all cancers are caused by DNA mutations acquired during a person's lifetime. But what mutations actually cause cancer?

I don't study cancer, but has this hypothesis been shown to be well-posed? That is, is it clear that there are strong genetic determinants associated with a given tumor type?

I'm just slightly skeptical because I know people have done large-scale sequencing studies in an attempt to find the genetic determinants of height. This is a strongly inherited trait, and is obviously quantitative, yet I believe all of the identified factors together explain less than 10% of its variability. I thus wonder if cancer is going to turn out to be more of a death-by-a-thousand cuts phenomenon, where many small effects combine to throw everything out of whack.
posted by dubitoergosum at 12:47 PM on December 17, 2009

Come back in six months....

From the NIH Public Access page "To help advance science and improve human health, the Policy requires that these papers are accessible to the public on PubMed Central no later than 12 months after publication."

And from Nature:
Authors are able to opt-in to the service as part of the journal’s online submission process. If authors opt-in:
• they are asked to complete a simple web form to:
- select a funder from a list
- provide further information needed by PMC and UKPMC
• On acceptance, NPG deposits the accepted version of the author’s manuscript, setting a public release date of 6-months post-publication
• PMC or UKPMC then asks the author to confirm the deposition and check the manuscript
• Six months after the article is published, it is freely accessible

posted by Endure You Are Not Alone at 12:50 PM on December 17, 2009 [1 favorite]

zarq, the NIH public access mandate (linked above) states:

To help advance science and improve human health, the Policy requires that these papers are accessible to the public on PubMed Central no later than 12 months after publication.

So yeah, the publishers still get first dibs, even though large chunks of the scientific community are working to make open access more broadly applied.

As for that CC license, it's covered here: http://www.nature.com/nature/journal/v450/n7171/full/450762b.html

For many years, a more generous arrangement has been made for papers reporting full genome sequences... These papers are freely accessible on NPG's website from the moment of publication. This recognizes a consistent character of 'genome' papers: they represent the completion of a key and fundamental research resource, describing and reflecting on what has been revealed but not usually providing insights into mechani

In short, Nature plays nice with genome papers, and makes them accessible. The reason you can't see this particular paper is because it hasn't actually been published yet. It's still an advance online preprint.

The distinction, in this age of the web, is somewhat silly. (posting on the internet IS publishing!) The PLoS journals in particular are starting to realize this and are really pushing the envelope in terms of online publishing and incorporating feedback and review mechanisms.
posted by chrisamiller at 12:59 PM on December 17, 2009

As sequencing prices come down the two tactics will eventually merge, but that's a few years down the road yet. And sequencing is only one part of the puzzle. Cells will not just have mutations in their DNA, they'll gain and lose huge chunks of DNA, and methylate and demethylate genes, both of which can contribute to carcinogenesis.

It seems likely that as sequencing gets cheaper, it will replace array CGH approaches as well - with enough sequence, you not only get all the mutations but you can look at the amount of sequence you get across each bit of the genome, and use it to get the copy number and all the rearrangements of the genome. (Which they do in the two papers in the FPP, and find some interesting rearrangements, including some which fuse two genes together.)
posted by penguinliz at 1:53 PM on December 17, 2009

Wow. Great post. I wonder though why they are confident that this one man's cancer is typical. If cancer is caused by mutations, wouldn't we expect wild variations between individual cancers?

They aren't, a large part of what makes this so exciting is that this is a really powerful way to figure out how typical the mutations which generate cancers are.

There will never be a cure for cancer, if only because there are a near infinite number of possible cancers, but in five years this data will likely be giving other researchers awesome directions to go in for treatments that will likely be available in another five to fifteen years after that.
posted by Blasdelb at 2:50 PM on December 17, 2009

while almost all small cell lung cancer is caused by smoking.

Minor derail: once the cancer has been initiated, nicotine "could enhance new blood vessel growth" as well as increase the rate of cell division. So after the many compounds in tobacco initiate a cancer, the nicotine enhances its ability to grow.
posted by dragonsi55 at 2:50 PM on December 17, 2009

If cancer is caused by mutations, wouldn't we expect wild variations between individual cancers?

Even within a tumor you can have a variety of mutation types.

Some researchers are looking into cancer in terms of population biology.

Cancer cells within a single tumor can accumulate different mutations over time. Those mutations can give them a reproductive advantage over other cancerous cells within the same tumor!

Some mutations are beneficial for certain cancer cells and deleterious for others, since the tumor gets access to a finite amount of nutrition from the host, and all the cancer cells are, in a sense, "fighting" over the same resources. So that complication creates trouble for the kind of differential analysis reported in this post.

Even then, assuming you find a drug that targets a cancer cell with one particular mutation, once you kill off those cells, other tumor cells with different genotypes can "take over" and flourish.

The variety of mutations within a single tumor alone can be problematic for cancer treatment and research, let alone between different patients.

In addition to the lab bench research outlined here, computer simulations of populations of cells in a simulated tissue are also used to gauge how a tumor might evolve and behave, given different initial assumptions and treatments.
posted by Blazecock Pileon at 4:00 PM on December 17, 2009 [1 favorite]

So we could have a DNA backup and just revert to last save every year?

You would need to replace the nucleus in every cell in your body.
posted by delmoi at 5:30 PM on December 17, 2009

Kind of a self-link, as it's written by one of my colleagues, but this is a Q&A piece from earlier this year with two Washington University researchers who are doing work exactly along these lines. Definitely interesting stuff.
posted by limeonaire at 6:41 PM on December 17, 2009

Hooray for publicly-funded science being reported in journals that are largely inaccessible to the public!

Personally, I'm ok with subtle or questionable hypotheses being kicked around in the backrooms of science for peer review and further study before it hits the 6 o'clock news.

If cancer is caused by mutations, wouldn't we expect wild variations between individual cancers?

If I understand the process correctly, the vast majority of cell mutations simply result in cell death, and most mutations quickly dies out. Only a very few of the possible mutations result in a monster cancer cell that multiplies without control. So I would expect that the genetic signatures of the more common cancerous mutations will soon be known. Their first use would likely be for detection, and later, treatments can hopefully be engineered to recognize and attack the cancerous cells.

Not much fun watching my parents go through cancer treatments, but one is currently cancer-free and the other just completed treatment with a good prognosis. so...yay science...
posted by Artful Codger at 8:36 PM on December 17, 2009

My daughter's cancer was caused by a mutation on, I believe, chromosome 22 that took place at the very beginning, creating a germline level INI1 a/k/a hSNF5 a/k/a SMARCB1 mutation. This is essentially a tumor-suppressor gene; her body could not stop cancer from growing, and new tumors from appearing. To put it in non-scientific words, the genetic mutation that 'caused' her cancer did so by making her body unable to stop the cell division of mutated cells, something the rest of us do naturally. The point: not every instance of gene mutation causes a cancer to grow, some make the body unable to stop them.

Poor kid was perfect in every way, even though she couldn't stop cancer from growing. Thanks for this post, I always appreciate learning more about this stuff.
posted by bunnycup at 7:35 PM on December 18, 2009 [1 favorite]

PS, the presence or absence of this mutation at the germline level (like my daughter) or on a somatic level (localized to specific mutated cells resulting in tumors) is known now - as opposed to June 2008 when she was diagnosed - to affect prognosis. Studies she participated in are looking at whether to change treatment recommendations based on the genetic findings.
posted by bunnycup at 7:37 PM on December 18, 2009

You would need to replace the nucleus in every cell in your body.

Can't we just use a massive dosage of some programmed agent, brought to you by FUTURETECH™?
posted by TwelveTwo at 7:36 PM on December 19, 2009