Quora is a question-and-answer site. You can view all my contributions here; selected highlights are listed below. I encourage you to check out the other answers submitted for each question, too!
Category Archives: mutation
[Originally posted on Occam’s Corner at Guardian Science, in February 2014]
In the 2000 M Night Shyamalan film Unbreakable, Samuel L Jackson’s character – a man born with a severe form of brittle bone disease – asks Bruce Willis’s character, ‘if there is someone like me in the world, and I am at one end of the spectrum, couldn’t there be someone else, the opposite of me at the other end? Someone who doesn’t get sick, who doesn’t get hurt like the rest of us?’ Some cancer researchers are trying to answer the same question; however, in real life, it’s just not that easy to unmask a superhero
In June 2013, I described how sequencing the highly abnormal genomes of cancer cells can identify some of the mutations that drive the progress of the disease (and how that’s only the beginning of the story). In the discussion that ensued, reader BlueSky3 commented:
Hope more likely rests in examining the control systems/defence mechanisms of those carrying a hereditary cancer fault which has perhaps persisted in evolutionary time. Not all mutation carriers succumb. Why not?
This comment was so interesting that it’s been percolating at the back of my brain for months. We know that women who inherit a faulty copy of the BRCA1 or BRCA2 gene have a highly elevated risk of developing breast or ovarian cancer – but cancer is not an inevitability. We also know that smokers have a highly elevated risk of developing lung, throat, or oral cancer – but some of them don’t. Why?
The answer lies in the complexity of cancer. The first mutation that starts the first abnormal cell down its path to malignancy can be caused by any number of factors: genetic predisposition, radiation, chemical agents, viruses. Similarly, any number of factors can influence the direction the disease travels thereafter. The first mutated cell has to escape everything the body can throw at it – DNA repair, the shutdown of cell division, programmed cell death, the immune system – before it can become truly dangerous. This complexity creates a number of possible points of intervention. Many, especially those related to the health of the immune system, are at least partially related to lifestyle factors, but in this article I’m going to focus on natural-born superheroes only – that is, those who inherit genetic factors that protect them from cancer.
The suggestion to study people with known cancer predisposition mutations who don’t go on to develop cancer is a great one, but not an easy one. If someone is unaware they have such a mutation, and they remain healthy, doctors and researchers have no way to identify them as a subject of interest. Additionally, many people who do know they have such a mutation can now take preventive measures (such as Angelina Jolie’s recent pre-emptive double mastectomy upon learning her BRCA gene mutation status), and we have no way of knowing whether they would have gone on to develop cancer if these measures had not been taken. All this adds up to a very small sample size to study, which makes the identification of subtle genetic correlations extremely difficult. It is possible, however, to search for “superhero” genes among the much larger general population, and to relate some of the findings back to more specialised populations such as those with inherited mutations in BRCA and other cancer susceptibility genes.
Many years ago, I attended a seminar by local researcher Michael Hayden about using very rare genetic disorders on one end of a spectrum to find new ways to fight very common disorders at the other end of the same spectrum. For instance, Hayden learned of a family with an inherited inability to feel pain, and was able to identify the faulty protein responsible; his lab recently published the results of a preliminary trial of a drug that targets the same protein in people without the disorder, and that may represent an entirely new class of painkillers.)
Unfortunately, finding people with genetic protection from cancer isn’t this straightforward. A person with no ability to feel pain will come to the attention of the medical profession early in life, when they walk for a week on a broken bone or show some other outward sign of their mutation. However, someone with an unusual degree of genetic protection from cancer is unlikely to present in the same way, making it harder to identify the relevant gene variants and to extrapolate from this knowledge to find a way to help prevent cancer in others.
Scientists are a resourceful bunch, though, and we’re starting to make progress despite these limitations. One approach is to look for protective gene variants in the general population, by comparing the gene sequences of people with cancer to those of healthy controls of comparable age and with similar risk factors. For example, in 2004 Angela Cox’s group at the University of Sheffield looked for correlations between breast cancer and the sequences of genes involved in programmed cell death. (This process, also known as apoptosis, is one of the body’s defence mechanisms that a cancer cell must evade if it is to go on to form a tumour. Apoptosis can be triggered by a number of different signalling pathways, each with multiple components; see the diagram for part of the picture.)
Cox’s team found that women who’d inherited a variant called D302H in the apoptosis-related CASP8 gene were less likely to develop breast cancer. This variant has since been shown to correlate with a reduced risk of prostate and other cancers, and in 2010 the Group for Assessment of Hereditary Cancer of Valencia Community reported that “CASP8 D302H polymorphism delays the age of onset of breast cancer in BRCA1 and BRCA2 carriers” – making its carriers not unbreakable, but definitely less fragile than the rest of us.
Like so many others, recent technology advances mean that this field of research is now dominated by large-scale whole-genome studies. In 2013, a major European cancer genetics consortium called COGS(Collaborative Oncological Gene-environment Study) published a series of papers describing the results of a massive genome-wide association study of 100,000 cancer patients and 100,000 healthy controls. The study was designed to identify genetic variants that affect the risk of developing hormonally mediated (ie breast, ovarian and prostate) cancers. As expected, most genetic variants were found to increase the risk of cancer, but a few protective variants were also identified. For example, a variant in a component of the telomerase enzyme, which repairs the protective cap structures at the end of chromosomes, correlated with longer caps and reduced risk of some forms of breast cancer, including BRCA-related breast cancer.
The power of whole genome sequencing is also being applied to the study of people of advanced age who’ve avoided the most common causes of death, including cancer. There are a number of “super-ager” studies of this kind under way, including one at my organisation (I’m not involved with the project in any way, but I hear about it in meetings and in conversations at the pub after work). Dr Angela Brooks-Wilson leads the study, which involves sequencing the genomes of people aged 85 or older who are in good health, and who’ve never been diagnosed with cancer, heart disease, stroke, pulmonary disease, diabetes, or Alzheimer’s disease. It’s early days still, but hopes are high.
Back in June, reader BlueSky3 continued:
We are spending millions on dissecting the ‘cancer genome’ in minute detail and on genome wide association studies, shame a bit of the research money cannot be diverted to genetically dissecting the differences between mutation carriers living into their nineties and their less fortunate relatives who succumb to cancer in their thirties
I hope this article has demonstrated that we are in fact making some progress in this direction. We haven’t found our superhero, and I have no last-minute plot twist up my sleeve – but labs full of everyday heroes are on the case, and this story is bound to have many sequels over the years.
[Originally posted on Occam’s Corner at Guardian Science, in June 2013]
The first cancer to have every “letter” of its genome sequenced and every mutation recorded was a leukemia, in a study that was published in the journal Nature in 2008. Genomes of other cancers – lung, breast, melanoma – quickly followed, all also published in top-tier journals and heralded in the media as major breakthroughs. Now, just a few years later, the era of cancer genomics research is well established and single-genome studies are already old hat – it takes much larger studies these days, involving the analysis of dozens of cases, to attract the same kind of attention as the early studies. With the first wave of research behind us, several centres around the world are now starting to study how to incorporate genomics into clinical cancer diagnostics and treatment.
The power and the promise of genomics is that, given enough money, we can start to personalise the treatment given to each patient. For instance, imagine a hypothetical mutation already known to be present in 70% of, say, bone cancers. A targeted drug is developed that works well in that 70% of patients, but does nothing for the other 30%, and whose effects (or lack thereof) take weeks or even months to detect. Sequencing newly diagnosed bone tumours before choosing a treatment lets you give the drug to those who will benefit from it, and find another option for the other 30% without having to put them through weeks or months of futile treatment, complete with nasty side-effects. If you also routinely sequence other types of cancer, you might find that 5% of, say, liver tumours contain the same mutation, and can be successfully treated with a bone cancer drug that might not otherwise be offered to liver cancer patients.
If the history of cancer research and treatment has taught us one thing, however, it’s that things are never quite that simple.
Take the example of a mutation called BRAF(V600E), which is found in a number of cancers, including melanoma. A drug called vemurafenib that targets this mutation has been developed and works well against melanoma, a notoriously aggressive and hard-to-treat cancer. However, when the same drug was given to patients whose colon cancers also contained the BRAF(V600E) mutation, it didn’t work. This puzzle was solved last year by a team who discovered that colon cancer cells contain high levels of a protein called epidermal growth factor receptor that protects them from the effects of vemurafenib; melanoma cells don’t contain much of this protein, which explains the difference in response between these two tumour types.
Chalk this one up as a learning experience for a young field; we now know to look at mutations in the context of the other genes and proteins that are active in the whole cell, not as single entities.
There’s a lot of useful information still to be gleaned from cancer genomes, and – no doubt – a lot of other learning experiences in our future. But with lives on the line, can we find a way to learn these lessons sooner rather than later?
One intriguing option is to pair cancer genomics with a technique called xenografting, which involves inserting a small piece of a patient’s tumour into a mouse. The idea is that the patient’s tumour can be sequenced, promising-looking mutations identified, and candidate drugs (and combinations of drugs) tested against that patient’s tumour in a number of “avatar mice”. This approach can help doctors choose the right treatment for each patient much faster, and with less risk of subjecting them to potentially futile treatments and side-effects; it can also give us early warning of the kind of interplay between a gene mutation and its cellular context seen in the case of BRAF(V600E). As an added bonus, this kind of study – and it is very much in the early research phase at the moment, not part of standard clinical care – can also feed information and tissue samples back to research labs, to help with their work on drug resistance mechanisms and other aspects of the cancer genome in context.
It’s early days for the avatar mouse, a model that is not without its problems. From what I understand of xenografting, it’s as much art as science; some tumour types refuse to “take”, while others start growing immediately. It’s also highly likely that the mix of different cell types within the original human tumour changes during the process of implantation into a mouse, meaning that the transplanted tumour might not respond in the same way as the original. But work is under way, and it is going to teach us a lot.
We will need to explore more than one avenue of investigation to counter the manoeuvres of an ever-evolving enemy. Genomics is a powerful tool that is already helping us to make small advances. Considering the genome in context will take us even further.
“Some people think that science is just all this technology around, but NO it’s something much deeper than that. Science, scientific thinking, scientific method is for me the only philosophical construct that the human race has developed to determine what is reliably true”
— Sir Harry Kroto, Nobel Laureate in Chemistry, 2010.
I don’t usually write about the nitty gritty details of new scientific discoveries. Why? Well, because (a) there are so few readers who are interested in them, and (b) I get enough of that kind of thing at work. Last week, though, I read a paper that’s just too good not to share.
My delight with the paper stems not just from the actual findings — although they are very cool — but also with the flow of the piece of work, the “story”. It’s just such a neat and satisfying illustration of how science is done, and why it’s so cool. I intend to focus on this general concept, since so few of you are likely to be interested in the nitty gritty details; if you are interested, the more technical aspects of the post are in italics, for ease of identification (or skipping, depending on reader preference).
Science is awesome. Image courtesy of Eva Amsen.
The paper is titled “Massive Genomic Rearrangement Acquired in a Single Catastrophic Event during Cancer Development”, and it was published in the journal Cell on 7 January 2011. I’ve included the full citation and a link to the paper at the end of this post.
Here’s how you do science:
Play with a shiny new toy…
Not every scientist is a gadget geek who gravitates to the latest sexy technology, but many are. Some people will always make accusations of bandwagon jumping, of playing for the sake of playing; but other people will always be attracted by the possibility that exciting new technology is the path to exciting new discoveries. If the Next Big Thing really does let you do things that weren’t possible before… well, if you can find the right way to apply it, and ask the right questions, you can find some very interesting answers.
One of the coolest new toys in my field is called next-generation sequencing, although I’m not sure why; it’s been around for a few years and should surely be called current-generation sequencing by now. (It’s also — more rarely — known as massively parallel sequencing. Let’s go with that name).
Many scientists (including some of my colleagues, although they weren’t involved with the current study) have started using massively parallel sequencing to explore cancer genomes. This technology allows us, for the first time, to identify every single gene mutation and gene rearrangement present within a tumour. Finding these changes is the first step toward developing new drugs, as well as new genetic tests that help physicians direct the right drugs to the right patients. The first tumour genome sequence was published a couple of years ago, and many more have followed. The authors of this paper are part of this massive sequencing endeavour.
…and use it to discover something new.
You have to be good to be lucky, and you have to be lucky to be good. If you’re using the right techniques in the right way, with the right controls, and if you happen to pick the right samples to analyse, you might just strike gold. In this case the researchers analysed samples taken from ten cancer patients, and uncovered a previously unknown phenomenon that’s only present in around 3% of all tumours.
The authors were sequencing DNA from ten patients with chronic lymphocytic leukemia. In most cases, they found what you’d expect: a limited number of genetic rearrangements, spread out across the whole genome. But in one patient they found a massive number of rearrangements in a small segment of the genome; one arm of one chromosome had gone completely haywire, and contained 42 separate rearrangements.
The next step was to check that this first case wasn’t a one-off. The team scanned hundreds of other tumour cells, from various types of cancer, and found that 2-3% of the tumours contained similar massive rearrangements of either single chromosomes or discrete chromosome segments.
Form a hypothesis and make some predictions…
The novel phenomenon described in this paper doesn’t fit the conventional view of how cancer develops. The authors proposed a different mechanism that would better explain their observations; this mechanism (or model) was their central hypothesis. They then designed experiments that would give one set of results if their hypothesis was correct, and a different set of results if the new phenomenon they’d observed was caused by the conventional mechanism of cancer development.
This process is absolutely key to how science is done. It’s a shame that it’s so rarely laid out explicitly in writing.
The conventional view is that tumours accumulate a series of independent genetic rearrangements over time. However, the rearrangements the team had found looked like the aftermath of a single, dramatic event — a chromosome shattering into many pieces, and being stitched haphazardly back together by the cell’s DNA repair machinery. An event like this would result in distinct types and patterns of genetic rearrangements compared to the patterns caused by a series of changes happening over time.
…and then test them.
Again, this is a crucial part of the scientific method. If your predictions pan out — if your results support your hypothesis — you think up new experiments to test your hypothesis that describes the phenomenon, and keep testing away until you’re either fairly confident in your hypothesis, or until it fails a test. If you experience a failure, you re-think your hypothesis based on the new evidence, and subject it to a new round of tests. Repeat until your results are publishable.
The authors laid out, point by point, why the rearrangements they’d observed fit the single event mechanism better than the progressive change model. They also ran computer simulations of both mechanisms, and found that the results generated by the single event model fit their observations better than did the results generated by the progressive change model.
The hypothesis also predicts that some rearrangements will result in genetic disruptions that cause cancer. The team went back to the cells in which they’d found massively rearranged chromosome segments, and found examples of deletions, amplifications, and other changes to known cancer genes — changes that one would indeed expect to contribute to the development of the disease. Overall, they came up with some very nice evidence to support their hypothesis.
Give your discovery a name…
Naming your discovery is a very rare privilege in science, one I’ve never experienced — I’ve only ever worked on known phenomena involving known genes in known species. I did have some fun with one gene I worked on, called SPAM1 (from SPerm Adhesion Molecule 1), inserting Monty Python references into my research presentations and such, but nothing more creative than that. ( Other scientists have had much more fun with their discoveries).
Image used in one of my research presentations, circa 2004
Having bagged the first published description of catastrophic rearrangements of discrete parts of the human genome, the authors got to name the phenomenon. They chose the word chromothripsis, from the Greek chromos (for chromosome) and thripsis (for shattering into pieces).
…and then tell everyone about it.
This paper isn’t just a description of a novel and exciting finding that happens to neatly illustrate the scientific method; it’s also very nicely written. Scientific papers are subject to certain constraints and conventions, but the standard formats do leave some room for differences in the quality of writing, and for expressions of individual style.
For example, my PhD supervisor (who’s a great scientist, a good bloke, and a Guardian reader. Hi, Dave, if you’re reading!) loves the word “remarkably”. Every time I spot a new paper from his lab, I start scanning the abstract (summary) for a sentence that begins “Remarkably, we found”, and I’m rarely disappointed. This paper goes one better: there’s one instance of “Here we describe a quite remarkable phenomenon”, and also a sentence that begins with “Astoundingly,” which I don’t think I’ve ever seen before.
I suppose if you’ve discovered something novel enough to deserve a new name, you’re entitled to use the word astoundingly.
I also enthusiastically underlined a section that reads as follows:
“the DNA machinery is pasting [hundreds of shards of DNA] together in a helter-skelter tumult of activity”.
Phrases like “helter-skelter tumult of activity” are not part of your standard scientific writing style, but maybe they should be!
The publication of this paper is not the end of the chromothripsis story. It’s barely even the beginning. There’s just enough room for doubt that the hypothesis will need to be subjected to further tests by the original team and by others who’ve read their paper. (This is normal and is part of the scientific process). There are new predictions to be made and tested. There are correlations to be made between this new phenomenon and how the cancer progresses in patients.
Part of the doubt stems from our incomplete knowledge of how chromosomes are broken and repaired in cancer cells; as more pieces of this puzzle are filled in, the new knowledge can be used to design new tests of the chromothripsis hypothesis. And scientists will need to find more evidence for cancer-causing changes induced by chromothripsis before the hypothesis is widely accepted.
However, these caveats needn’t impede the progress of this story. The authors have made a couple of predictions about what causes chromothripsis — ionising radiation, perhaps, or loss of the protective telomere structures found at the ends of chromosomes — and have proposed experiments that would test these predictions (given how long it can take to write a paper and get it published, these experiments are almost certainly already underway, if not complete).
I’d also like to see someone look for correlations between which tumours display evidence of chromothripsis and how the tumour responded to treatment, or whether it came back or spread to another part of the body after initial relapse. These are important clinical questions, and I hope to see some answers appearing in the scientific literature over the next few years.
And so the boundaries of human knowledge expand, slowly but surely.
I left the active research phase of my career behind almost five years ago, but my own miniscule contributions remain a source of immense pride to me. Whatever else I may do in life, the papers in which I described my novel results will still exist as part of the permanent record of human curiosity.
And luckily, I find science to be just as exciting as a spectator sport as it was when I was still working in a lab!
Stephens PJ, Greenman CD, Fu B, Yang F, Bignell GR, Mudie LJ, Pleasance ED, Lau KW, Beare D, Stebbings LA, McLaren S, Lin ML, McBride DJ, Varela I, Nik-Zainal S, Leroy C, Jia M, Menzies A, Butler AP, Teague JW, Quail MA, Burton J, Swerdlow H, Carter NP, Morsberger LA, Iacobuzio-Donahue C, Follows GA, Green AR, Flanagan AM, Stratton MR, Futreal PA, Campbell PJ. Massive Genomic Rearrangement Acquired in a Single Catastrophic Event during Cancer Development. Cell(2011) 144(1):27-40.