About this Author
DBL%20Hendrix%20small.png College chemistry, 1983

Derek Lowe The 2002 Model

Dbl%20new%20portrait%20B%26W.png After 10 years of blogging. . .

Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases. To contact Derek email him directly: Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
Not Voodoo

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
A New Merck, Reviewed
Liberal Arts Chemistry
Electron Pusher
All Things Metathesis
C&E News Blogs
Chemiotics II
Chemical Space
Noel O'Blog
In Vivo Blog
Terra Sigilatta
BBSRC/Douglas Kell
Realizations in Biostatistics
ChemSpider Blog
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Eye on FDA
Chemical Forums
Symyx Blog
Sceptical Chymist
Lamentations on Chemistry
Computational Organic Chemistry
Mining Drugs
Henry Rzepa

Science Blogs and News:
Bad Science
The Loom
Uncertain Principles
Fierce Biotech
Blogs for Industry
Omics! Omics!
Young Female Scientist
Notional Slurry
Nobel Intent
SciTech Daily
Science Blog
Gene Expression (I)
Gene Expression (II)
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net

Medical Blogs
DB's Medical Rants
Science-Based Medicine
Respectful Insolence
Diabetes Mine

Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem

Politics / Current Events
Virginia Postrel
Belmont Club
Mickey Kaus

Belles Lettres
Uncouth Reflections
Arts and Letters Daily
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline

In the Pipeline

« In Which I Reminisce About the Prins Reaction, Chemical Abstracts, and John Keats | Main | Westphal Leaves GSK »

April 5, 2011

So, You Thought Breast Cancer Was Complicated?

Email This Entry

Posted by Derek

You may have detected, here and there, a certain amount of skepticism on this blog about the direct application of genomic information to complex human diseases. And several times I've beaten the drum for the position that there is no such disease as "cancer" - just a lot of conditions that all result in the phenotype of uncontrolled cellular growth.

Well, here's some pretty dramatic evidence in favor of both of those positions. A new study, one of those things that could only be done with modern sequencing techniques, has given us the hardest data yet on the genomic basis of cancerous cells. This massive effort completely sequenced the tumors from 50 different breast cancer patients, along with nearby healthy cells as controls for each case.

Over 1700 mutations were found - but only three of them showed up in as many as 10% of the patients. The great majority were unique to each patient, and they were all over the place: deletions, frame shifts, translocations, what have you. The lead author of the study told Nature News that the results were "complex and somewhat alarming", and I second that, only pausing to drop the "somewhat". I add that qualification because these patients were already more homogeneous than the normal run of breast cancer cases - they were all estrogen-receptor positive, picked for trials of an aromatase inhibitor.

Half the tumors were estrogen-sensitive, and half weren't, and one of the goals of the study was to see if any genetic signatures could be found that would distinguish these patients. There was an association with the MAP3K1 gene, but hardly a powerfully predictive one, since that one only showed up in 10% of the samples to start with. (Mind you, that still makes it one of the top three mutations).

The Nature piece contains some brave-face material about how this study has uncovered a whole list of new therapeutic targets, but sheesh. What are the odds that any of these will prove to be crucial, even for the low percentage of women who turn out to have them? No, instead of making me yearn for ever-more-personalized targeted therapies, this makes me think that early detection and powerful, walloping chemotherapy (and surgery) must be the way to go for now. I mean, this was still only fifty patients, and uncovered this much complexity: how tangled must the real world be?

We'll get a chance to start finding out - the same team is now moving ahead to expand this effort to 1,000 patients. These are also, I believe, from clinical trials, so we'll be able to correlate outcomes with exact genetic sequences. If there are any correlations that we can understand, that is. . .that's the next thing that I'm really looking forward to seeing. If the whole personalized-medicine idea is ever to work, this is just the sort of thing that's going to have to be done. But we shouldn't be surprised if the results, for some time to come, are that the whole era of personalized medicine is a lot further away than we might have thought.

Comments (33) + TrackBacks (0) | Category: Cancer


1. Rick Willis on April 5, 2011 1:03 PM writes...

If you could treat 10% of breast cancer patients with a MAP3K1 inhibitor, that wouldn't be too shabby. No imatinib, but not bad.

The 1700 unique mutations is scary, but I'd guess most of them appear sometime after the beginning of cancer. The problem is sorting out the causative mutations from the ones that crop up after P53 loses control of things. Also, how often will inhibition of the causative mutant still be effective after all the other mutations appear.

Permalink to Comment

2. wuhwoo on April 5, 2011 1:09 PM writes...

Maybe the NIH's new pharma enterprise should consist of a 5000 year plan to serially drug every tractable target in the human genome. This is a very scary result.

Permalink to Comment

3. Anonymous on April 5, 2011 1:16 PM writes...

I had heard that cancer was at least 80+ discrete diseases more than five years ago, and this implies even more. However, presence certainly does not imply unique diseases, let alone causalities. Perhaps some of these MIGHT have a common upstream trigger, while most others are the "whims of chaos"?

Permalink to Comment

4. soydegales on April 5, 2011 1:19 PM writes...

apparently, we really are not that smart. the risk to patients is that a big chunk of the industry is betting the farm on "personalized" approaches to cancer therapy. keeps a lot of people very busy but data like these argue that the whole edifice might be built on somewhat shaky foundations. given lead times to approval are so long, one hopes this might at least prompt a bit of objectivity and diversification of strategy. we should show some humility, cancer is an incredibly tough problem - some of the successful oncology drug discovery efforts from the the last century might be able teach us a thing or two.

Permalink to Comment

5. luysii on April 5, 2011 1:44 PM writes...

Well, we know that cancer cells are genetically unstable. The hope (looking slimmer all the time) is that the crucial mutations are 'driver' mutations, common to each type of tumor, pushing the cell into malignancy. The fact that so few mutations were common to many tumors (and that no mutation was common to all) argues strongly against this. Now people will be looking at pathways (such as that from extracellular receptor kinase (aka ERKs) to the nucleus, which might be a commonality among the tumors.

We've had intimations of this before -- from the Cancer Genome Atlas study (criticized by some as a boondoggle). [ Cell vol. 136 p. 834 '09 ] "Large scale sequencing of multiple cancers has so far failed to identify new, high-frequency mutation targets in addition to those previously identified." Translation -- it hasn't turned up anything we didn't already know. Rather every tumor contains many mutations (an average of 63 in pancreatic cancer and 60 in glioblastoma) of low frequency.

Sad, but challenging. But, we have no alternative but to keep plugging away. It's unlikely that a new improved MBA will cut the Gordian knot.

Permalink to Comment

6. LA on April 5, 2011 1:50 PM writes...

While I agree that personalized medicine for everyone isn't as close as we like, that doesn't mean we should throw in the towel. In breast cancer alone, we now test for ER, PR, and Her-2 nue. The results of these three tests dramatically affect treatment. Without them , we could give potentially toxic treatment to someone who won't benefit at all. This is why we do research- to find the things that do matter. One of those three mutations mentioned may mean less toxicity or a better response. May not matter to everyone, but it sure matters to that 10%.

Permalink to Comment

7. Carl Lumma on April 5, 2011 1:51 PM writes...

Once a metastatic cell line appears, you're gonna need a certain degree of computational horsepower at the nanoscale. Antibody-based targeting is one try, either with alpha emitters or perhaps something like Mullis' Altermune approach.

Stopping mutations from accumulating to dangerous levels is perhaps a more prudent approach. We know antioxidants don't do a damn thing (as a class). Radiation hormesis is promising. And I suspect the last 50 years of work on low-dose radiation will eventually be recognized as a tragic misapplication of the precautionary principle.

Permalink to Comment

8. Lester Freamon on April 5, 2011 1:51 PM writes...

Re: MAP3K1: it's a tumor suppressor, so an inhibitor isn't what you're looking for. Most of these studies seem to reveal tons of tumor suppressors with inactivating mutations, but that doesn't really give you anything in terms of a drug target.
Honestly to agree with #2 I think it's gotten to the point where making inhibitors of every enzyme and receptor in the genome would get us clinical results faster. Call it blindly targeted therapy.

Permalink to Comment

9. chris on April 5, 2011 2:04 PM writes...

"...reveal that these cancers' genetic fingerprints are highly diverse; of the 1,700 gene mutations they found in total, most were unique to individual patients' tumours, and only three occurred in 10% or more."

Whilst the mutations might be unique I'm not clear if there were many mutations in a limited number of specific genes?

Permalink to Comment

10. ech on April 5, 2011 2:19 PM writes...

Would it make sense to preserve cell samples that can be sequenced from the subjects in drug trials to look for correlations with side-effects, which might be done later when sequencing falls further in price? Or are they already set to do this?

Permalink to Comment

11. Deanne on April 5, 2011 2:22 PM writes...

@Chris -

Doubtful. If there were, then they would've played up the fact that there were genes in common among the cancer samples sequenced. That isn't uncommon in other diseases (disease-causing mutations found throughout a gene, collectively speaking, even if there are hotspots) - cystic fibrosis comes to mind.

Permalink to Comment

12. pete on April 5, 2011 2:49 PM writes...

Presumably a number of these rare mutational combos will be confirmed by experiment to be 'cancer-drivers' -- I'm not clear that they have been, yet.

But assuming it's true, then this definitely changes the landscape. IMHO, it underscores the importance of more widely classifying cancers according to their comparative degree of disregulation among major cell signaling networks. Think PieChart displays showing the relative biasing of activity among major pathways in Tumor X vs Tumor Y.

Permalink to Comment

13. David on April 5, 2011 2:54 PM writes...

What I find amusiing is the recent tendency to label some breast cancer as "triple negative" (negative estrogen receptor, progesterone receptor and no HER-2 over-expressing.) You read all about "triple negative" breast cancer in the journals and you hear about it at the conferences. Every oncologist investigator talks about "triple negative" breast cancer as though it was one disease. One disease? Give me a break! Nothing could be further from the truth.


Permalink to Comment

14. johnnyboy on April 5, 2011 3:00 PM writes...

But I think something like this was already known from previous cancer sequencing efforts (albeit on smaller numbers of tumors, and with a smaller total number of mutations). The idea was to move away from looking at mutations individually and grouping them into their signalling pathways, which are necessarily much fewer. Therapeutics efforts should strive to inhibit the implicated pathways, rather than focussing on single mutations. That was the idea anyway, but with such a whopping number of mutations identified, I don't know if that really translates into that many fewer pathways. Also to consider is that the authors probably don't know at this stage how many of these mutations actually translate into a proliferative effect, i'd imagine that the large majority have no significant effect.

This said, I totally agree with your statement on the more appropriate approach to cancer treatment. Trying to treat advanced cancer is like shutting the barn door after the horse has bolted, met a nice mare, and produced several equine generations. I wish that the same amount of scientific inquiry and money that is put into cancer therapeutics was also put into early detection technology; the more one understands how intractable cancer is, the more early detection seems the only rational approach to significantly prolong life.

Permalink to Comment

15. Brian on April 5, 2011 3:01 PM writes...

Luysii is right, CGA results anticipated this.

Early detection plus aggressive treatment does seem the way to go. But early detection of what, exactly? If there are no common markers what do you look for? Most of these mutations alone are probably not indicative of anything. Even as a group they may not be indicative in all conditions.

Permalink to Comment

16. johnnboy on April 5, 2011 3:47 PM writes...

Every cancer would likely require a different approach - some with imaging, some with biomarkers. For the tumor sites less prone to imaging, seems to me that proteomics approaches in body fluids would be the first step to look for tumor markers. Obviously a difficult and long process to find good markers, and finding a panel of markers is much more likely than any single one, but again, efforts in that direction seem to me more useful than multi-billion dollars spent on developing drugs that provide 4 month survival benefits...

Permalink to Comment

17. Anonymous on April 5, 2011 4:46 PM writes...

Sorry but these are genetic mutations, right? What proteins out of these are actually being transcribed, and are functional? Functional meaning, working for or against the cell?

Permalink to Comment

18. Henning Makholm on April 5, 2011 5:18 PM writes...

As a baseline for the "1700 mutations" figure, if you sequence DNA from two random somatic cells from the same healthy test subject, how many differences does one typically find?

Ten? One? Zero?

Permalink to Comment

19. barry on April 5, 2011 6:03 PM writes...

Cancer cells tend to accumulate mutations. That's not news. We learned with CML that a SINGLE genetic event (forming the "philadelphia chromosome") is enough to give rise to a cancer which will eventually have hundreds or thousands of mutations. Still, treatment with imatinib EARLY, when the single mutation is driving the disease is astonishingly efficacious.
As long as we run our first clinical trial only in old, refractory cancers that have accumulated hundreds of mutations, the chance of seeing any signal of efficacy is lousy. We need to learn from Gleevec. The population in which you can see efficacy (and the population you can save) is the population of very early cancers which are still driven by one or two mutations.

Permalink to Comment

20. RKN on April 5, 2011 6:21 PM writes...

Very similar to what was done years ago in breast & colorectal cancer ("The consensus coding sequences of human breast and colorectal cancers.", Sjoblom et al., Science, 2006)

Few driver genes found to be common to all tumors. Check. Disparate driver genes collapse to fewer pathways. Check.

Still, without a more thorough understanding what's going on at the level of the proteome (molecular networks), more full genome screens alone will move us no closer to understanding the common (if any) dys-functions of cancer cells.

1000 more samples w/matched controls will only confirm what we already suspect is true - "Yes, cancer is complicated" - and leave us scratching our collective heads, I'm afraid. We need integrative systems biology approaches to have a fighting chance at understanding *functionally* what's going on to trigger and sustain these tumors.

Permalink to Comment

21. cliffintokyo on April 5, 2011 6:54 PM writes...

Perhaps it's time to focus research more on stimulating each patient's own immune system to recognize and fight off that patient's own unique mutated cancerous lesions. Patient-specific vaccines anybody?

Permalink to Comment

22. anonymouse on April 6, 2011 8:31 AM writes...

So how many hundreds of years away these days is it until we have nanotech? Seems like this study shows that "cancer" is the generic result of transcription errors, however and wherever they occur, no? We need better error checking in our cells.

Permalink to Comment

23. Rick on April 6, 2011 9:48 AM writes...

I just love this new trend of using the word "complex" to refer to things that we used to call just plain "sloppy", as in "Handling of mortgage backed securities, such as collateralized debt obligations is very complex."

Permalink to Comment

24. Celbio on April 6, 2011 9:48 AM writes...

My current gig is aiming to utilize genetics of patients with inflammatory disease, and yes Derek, it is tough, very tough, so your skepticism is warranted. However, I have felt for some time that our chance for success is greater than in cancer as we are looking for stable genetic profiles rather than the huge challenge of wandering genetics of cancer cells.

Some very good academic work in the field has identified targets, which went into drug development and the clinic only to fail, but I am not sure these early examples are the best we can do in PM. Why did this class of drugs fail? Perhaps a number of reasons, but the high p value of association is driven by a difference from case vs control allele frequency, but if one looks in the patients, you see that the risk allele is in less than 5% of patients. Genetic evidence, yes, solid reason to examine broader efficacy, yes, but not compelling evidence to expect efficacy of the agent in all-comers. The trial results are in line with genetic evidence that points to multiple mechanisms of pathophysiology.

Multiple mechanisms of disease, differences by population, clinical phenotypes, sub-phenotypes.... all these are the sharp knives of Personalized Medicine that cleave population sizes rapidly down to irrelevance. The key will be, as mentioned by posters above, to find points of intervention that reassemble these small groups, so that the medicines we make are not truly 'personalized'. For that, I still favor empirical pharmacology, informed by genetics, but looking a lot like old school drug discovery.

Permalink to Comment

25. daen on April 6, 2011 12:27 PM writes...

These cancers have a distinct time progression, and studies like these are tantamount to looking at a china shop after a bomb has gone off - there's little trace of the bomb and a vast amount of broken crockery, a lot of which has actually been broken by flying shards of other cups and saucers rather than by the original bomb.

It would be interesting to see groupings for those 1700 mutations, both across function and time. Isn't it going to be the case that initial mutations are be in a fairly small set of functional groupings - DNA repair, apoptosis etc - and only later accrue in other areas?

Permalink to Comment

26. Rick on April 6, 2011 1:42 PM writes...

Cellbio, #24,
You "still favor empirical pharmacology, informed by genetics, but looking a lot like old school drug discovery." Dude, that's sooo 20th century! Unless you want people to think you're a dried up old fossil you HAVE to update your vocabulary. Nowadays all that stuff is called "Chemical Biology". "Pharmacology" makes today's cool, happening people think of Walgreens, which, ironically, is where more and more drug discovery does NOT lead these days.

Permalink to Comment

27. Commissar on April 6, 2011 3:05 PM writes...


The party officials are way ahead of you. Even Orwell would be trembling:

" The focus of healthcare will shift over the
next ten years from disease to wellness. We are
developing metrics for assessing an individual’s
wellness. There will be a wellness industry that
in time could dwarf the healthcare industry.
Medicine will also be focused entirely on the
individual in the future. We will all have the
equivalent of iPods that will be recording enor-
mous amounts of personal data and transmit-
ting it to servers for analyses that will monitor
your wellness status and report developments
that are a cause for alarm by sending you a sig-
nal, such as “Slow down on eating.”

Permalink to Comment

28. Vader on April 6, 2011 4:10 PM writes...

Tolstoy on breast cancer:

"Happy cells are all alike; every unhappy cell is unhappy in its own way."

Permalink to Comment

29. Henry on April 6, 2011 6:13 PM writes...

I agree with you that there's no such disease as cancer. Considering all the myths that we've heard about the disease, such as antiperspirants or deodorants causes breast cancer, or that microwaving plastic containers releases harmful, cancer-causing substances into our food. Health institutions should make certain information publicly available to the public that backs the theory of cancer not being a disease. At least people can be well informed, rather than following myths.

Permalink to Comment

30. Cellbio on April 6, 2011 11:26 PM writes...


How non-adaptive of me. I should have referred to Integrative systems Biology resulting in seamless integration so efficient we cut the cost to develop blockbuster molecular targeted nano medicines from 66 to only 33 million while filing INDs in less than 140 characters.

Permalink to Comment

31. pete on April 7, 2011 12:49 AM writes...

@30 Cellbio
..don't forget the leveraging of synergies

Permalink to Comment

32. Still Scared of Dinosaurs on April 7, 2011 4:10 AM writes...


Love it, how about this one?

Kafka on expanding the study to 1000 patients:

"We will run this experiment so that you feel you have left nothing untried."

Permalink to Comment

33. Vanessa on October 18, 2011 2:36 AM writes...

I realize this post is a few months old. But it has recently come to light that the human body is more microbe than human cells. This fact together with the lack of common mutations seems to indicate that cancer might actually be a fungus, bacteria or other microbe. This isn't too surprising. H. Pylori is linked to MALT lymphoma and aflatoxins are a potent hepatic carcinogen.

Permalink to Comment


Remember Me?


Email this entry to:

Your email address:

Message (optional):

The Last Post
The GSK Layoffs Continue, By Proxy
The Move is Nigh
Another Alzheimer's IPO
Cutbacks at C&E News
Sanofi Pays to Get Back Into Oncology
An Irresponsible Statement About Curing Cancer
Oliver Sacks on Turning Back to Chemistry