Holiday Reading (2) – Poking the Blancmange

An evolutionary hiccup

It’s well known that tracing our family tree back 400 million years reveals a fishy past. This history is enshrined in our DNA in the pattern of nerves that control breathing. From time to time that control throws a wobbly in the form of involuntary spasms of the diaphragm manifested as a fit of hiccups – what the medics call singultus, which in Latin means sobbing – readily brought on by contemplating a comprehensive map of intracellular signalling pathways. Hiccups, however, are caused, as Neil Shubin, in his wonderful book Your Inner Fish has explained, by a mis-firing neuron in our brain stems that produces the type of electric signals that control the regular motion of amphibian gills. A genetic recipe hoarded in the nuclear loft is inadvertently recalled. For the most part this result of DIY evolution is no more than mildly embarrassing, although the poor fellow who made the Guinness Book of Records by hiccupping for 68 years may have used a stronger term.

As we’re really talking about cancer, we should mention that persistent spasms of hiccups and difficultly swallowing may be indicators of esophageal cancer, where a tumour in the gullet (the tube connecting the back of the mouth to the stomach) grows into the trachea and flips the hiccup switch by mechanical pressure.

Je pense, donc je suis un blanc-manger

Whilst the key feature of all these pathways is that they connect the outside world to the nucleus of a cell, it’s become clear that each pathway does not exist in isolation. Individual pathways can talk to each other – sometimes called cross-talk – individual domino runs intersecting, if you like. So evolution has cooked up thousands of proteins floating around in our cells that can be mapped into discrete signal pathways but, in the molecular jostle of the cell, each may affect any of the others – if not directly then via just a few intermediates. To avoid the Tokyo subway syndrome it’s easiest to think of the cell as a blancmange: poke it anywhere and the whole thing wobbles.


The complex network of signalling pathways in cells.

Left: the dots represent proteins that inter-communicate (lines) – best thought of as a blancmange.

Why is grasping this picture of what seems like a molecular madhouse important? Well, one thing we should bear in mind is that the set-up may look chaotic to us but our cells somehow make perfect sense of it all because they take clear decisions as to what to do. But the reason for grappling with it at all, other than to be humbled by our ignorance, is that these signal systems are a major target for anti-cancer drugs. To be more precise, it’s disruptions in these proliferation-controlling pathways, caused by mutations, at which we take aim with the contents of our drug cocktail cabinet.

What goes wrong in cancer?

If you want a three word definition of cancer ‘cells behaving badly’ will do fine. If you insist on being scientific ‘abnormal cell proliferation’ covers it nicely, meaning that control of cell replication has been overcome to the extent that cells reproduce more rapidly than they should or at an inappropriate time or in the wrong place. Underlying this abnormal behaviour is damage to DNA, that is, mutations. This remains true even if the initial cause does not directly affect DNA. It’s estimated that about 20% of the global cancer burden comes from infections, mainly in contaminated drinking water. These can cause chronic inflammation that eventually leads to mutations and thence to cancer. Other factors, for example, tobacco smoke and radiation, can directly damage DNA and about 10% of cancers are set off by what you might call a taxing inheritance – mutations already present in DNA at birth.

The capacity for high-throughput sequencing of complete human genomes has spawned ambitious projects that include Genomics England’s sequencing of 100,000 genomes by 2017 and The Cancer Genome Atlas that aims to provide a mutation data base for all the major cancers. One of the most mind-boggling facts that has already emerged from this revolution is the extent of disruption that can occur in the genomes of cancer cells: as many as one hundred thousand mutations within one cell. For the sake of completeness we should note that, cancer being cancer, the mutational spectrum is astonishing and, at the other end of the scale, there’s a childhood leukemia that results from just one change to DNA and there’s a type of central nervous system tumour that appears to develop without any mutations at all. For the most part, however, cancer cells carry a mind-boggling number of mutations and the assumption, nay hope, is that the vast majority of these changes are ‘passenger’ mutations that do not affect cellular behaviour: they’re a by-product of the genetic mayhem characterizing cancer cells. The ones that count are ‘driver’ mutations that can arise in any of several hundred of our 20,000 or so genes, changing the activity of the proteins they encode to contribute to cancer development. Only a small number (half a dozen or so), of these drivers, acting together, is required for cancer to emerge. Thus, although only a relatively small group of ‘drivers’ is needed, almost limitless combinations can arise.

The accumulation of mutations takes time, which is why cancers are largely diseases of old age: two thirds of them only appear in people over the age of sixty. The estimate is that if we lived to 140 everyone would get cancer but, pending that happy day, when or whether the disease manifests itself in an individual is indeed a matter of genetic roulette – genetic evolution within cancer cells. So wonderful has the technology become we can now inspect individual cells in tumours to reveal that driver mutations occur in single cells that can expand to form groups of cells, called clones. These multiple clones can modulate their mutational profile independently and, as a result, proliferate at different rates. So you can picture tumours as a complex patchwork of genetically related, competing clones. In other words, as we’ve suspected all along, cancers are a form of dynamic Darwinism.

The critical point is that key mutations drive cancer and they do so by upsetting the normal working of signal pathways that control whether cells proliferate or not. You could say it’s Nature poking the blancmange but these are delicately selected pokes – the product of the evolution of a cancer’s genetic signature – that just tweak signalling mechanisms enough to make cells a little more likely to multiply. In coming up with drugs that target specific mutations we’re giving the blancmange another poke – the aim being, of course, to prod it back to normality.

An obvious question

Having mentioned that, albeit very rarely, cancers emerge that don’t seem to be driven by changes in the sequence of DNA – how do they do that? The answer lies in epigenetic modifications – any modification of DNA, other than in the sequence of bases, that affects how an organism develops or functions. They’re brought about by tacking small chemical groups either on to some of the bases in DNA itself or on to the proteins (histones) that act like cotton reels around which DNA wraps itself. In effect this makes the DNA more difficult to get at for the molecular machines that turn the information in genes into proteins. So these small chemical additions act as a kind of ‘super switch’ that can, for example, block genes that act as brakes on cell proliferation – hence promoting cancer.


Neil Shubin Your Inner Fish, Random House, 2008.


The Blink of an Eye

You might not have thought of it in quite this way but cancer biology is a bit like having kids. It seems you only have to turn your back and things have changed, not so as to be unrecognizable but enough to have you blinking in surprise, shock or horror. In the cancer field it’s true that, especially over the 12 years since human DNA was first completely sequenced, a fair bit of the jaw-dropping has been due to astonishing technical advances. Thus human genomes (i.e. their DNA sequence) can be laid bare in 24 hours – The International Cancer Genome Consortium now has over 10,000 cancer genomes in its database – and the power of the panoply of ’omics methods to probe ever deeper into the mind-boggling complexity of tumours is quite staggering (we risked a quick peep at just how tricky it is to disentangle a picture of the biology from the vast amounts of data in A Word From The Nerds).

Cancer’s simple

These revelations often leave us gasping at the variety and adaptability of nature and how that shows up time and again in the microworld of cancers. Of course, we’re used to the world being ever-changing but we like to think there are some things that are fixed. The Earth still rolled round the Sun even after the aeroplane was invented. When it comes to cancer the simple but fairly firm idea is that cells pick up changes in their genetic material (i.e. mutations in DNA) and if these affect an appropriate set of genes (i.e. encoded proteins) a cell starts misbehaving – multiplying when it shouldn’t or faster than normal. And that’s cancer. Of the twenty-odd thousand genes that make human beings, several hundred have this ability to be trouble-makers – and a handful at any one time (perhaps five to ten) is all it takes. Like any team, there are some high profile players: genes that crop up time and again in mutant form driving all sorts of different tumours. There’s maybe a dozen of these. The rest are bit part players: actors who can steal the show with a cameo role. In others words they’re low frequency cancer drivers, perfectly capable of doing the job but generally keeping a low profile.

All of which is fine: we can hang on to what we thought we knew. Cancers are caused by cumulative mutations – things are just complicated a bit because of the more or less infinite subtlety that the different combinations can cause. So cancer’s really pretty simple.

Oh no it’s not!

However, just once in a while – mercifully, or we’d all go potty – something comes along that has us, if not standing on our heads, at least wondering which way is up. Welcome Iñigo Martincorena, Peter Campbell and pals from The Sanger Institute in Cambridge – a regular source of wide-eyed wonder in genomics.

They’ve just done something that, on the face of it, was very odd. They carried out a thorough sequence analysis of samples of normal human skin, the skin in question being from eyelids. The plan was to try to get a picture of how cancers develop and eyelid skin is a good place to look because it gets a relatively high exposure to sun. Moreover, it’s easier to get hold of than you might think: there’s an age-related condition in which the skin loses its elasticity causing the eyelid to droop – which can be treated by surgery, i.e. cutting out some of the skin.

Fasten your seat belts: here comes the shaker. In 234 eyelid samples (biopsies) from four people the number of mutations was similar to that in many cancers! Yet more amazing, the mutated genes included most of the key ‘drivers’ of one of the major forms of skin cancer.

Putting numbers on it, they found about 140 driver mutations per square centimeter of skin.

The type of DNA damage was characteristic of the effect of ultraviolet light (e.g., changing C to T – i.e. the base cytosine is mutated to thymine) – so at least that wasn’t a surprise.

1 sq cm

Groups of mutant cells (clones) in a 1 square centimeter of normal eyelid skin.

The circles represent samples of skin that were sequenced. Their sizes and the representation of nested clones are based on the sequences obtained. The outermost layers of normal skin can therefore be viewed as “effectively a battlefield of hundreds of competing mutant clones in every square centimeter of skin.” (from Martincorena et al. 2015).

As Iñigo & Co put it ‘aged sun-exposed skin is a patchwork of thousands of evolving clones with over a quarter of cells carrying cancer-causing mutations.’ Notably, there were clones carrying two or three driver mutations – and yet the tissue showed no sign of cancer and functioned quite normally (apart from its wonky elastic).

Close your eyes: time for a re-think

So, there are thousands of mutations in each skin cell with hundreds of evolving clones per square centimeter and the profile of driver mutations varies between individuals. The obvious question, therefore, is ‘why isn’t this tissue cancerous?’ We don’t know but, given that key ‘drivers’ are present, it seems that these cells either have a kind of master ‘off switch’ that suppresses potent driver combinations or they need a further ‘on switch.’ There’s no evidence for either of these, nor is it clear whether other cell types can show this kind of restraint.

And there’s one more troubling point. Many cancer drugs are designed to target driver mutations and thus to kill the carrier cells. But if these mutations can crop up in normal cells, any such ‘cancer specific’ drugs might cause a good deal of what the military term collateral damage.

As ever in science, an exciting new finding raises yet more questions. Answers will be forthcoming at some point. Just don’t blink!


Martincorena, I. et al. (2015). High burden and pervasive positive selection of somatic mutations in normal human skin. Science 348, 880-886.

Gentlemen! For goodness’ sake …

I reckon there should be a 21st century addition to the family etiquette handbook banning laptops at the breakfast table. It’s anti-social and indeed downright rude: at best you get to your emails ten minutes quicker but it’s also really stupid because computers do not thrive on a diet of milkdrops, cornflake fragments and bits of toast. I never appear without mine – and with it I bring another potential, disgraceful side-effect, manifested in our household on the second day of the New Year when, a few minutes after I’d sat down, booted up and started munching, the air gradually began to turn blue. “Oh dear” muttered youngest son: “he’s on to the science pages of the broadsheets: fingers in ears.” How shrewd. And what good advice.

Rattling my cage

So what was it that so wound me up when I was looking forward to a rather non-sciency, tranquil opening to the year? “Most cancers are caused by bad luck not genes or lifestyle, say scientists”, a headline trumpeted by The Telegraph was a great start, backed-up by much the same parroted in The Independent and The Guardian. The only good news was that, try as I did, I could find no equivalent coverage in The New York Times or The Sydney Morning Herald. Let’s hear it for the colonials – or at least their science editors!

What’s my problem?

Why is it that this sort of journalism so annoys and certainly did so on further reading of those new year contributions? Well, partly because it’s headline-driven rather than a thoughtful effort to inform the public. And then because what’s propagated isn’t totally wrong – that would be easy to deal with – but rather it’s a confused mish-mash of half-truths guaranteed to confuse utterly anyone who doesn’t have an assured grip on their molecular wits.

Let’s get things clear

First let’s get the basic picture clear, then see what “the scientists” really said in this new piece of work and finally illustrate how the Gentlemen (and Gentlewomen) of the Press get me so incensed.

Asked to sketch a current cancer portrait one might say: Cancers are caused by damage to DNA, i.e. mutations. Of our 20,000 or so genes several hundred can acquire mutations that change the activity of the proteins they encode to contribute to cancer development. Only a small number (half a dozen or so) of these ‘driver’ mutations, acting together, are required for cancer to emerge. Thus almost limitless combinations of drivers can arise. The effect of these cancer ‘drivers’ is to make cells proliferate (i.e. divide to make more cells) either at a faster rate than normal, or at the wrong time or in an abnormal place. Environmental factors (e.g., smoking) can increase the mutation rate and hence the chance that cancers will evolve. Most mutations accumulate during the lifetime of the individual (hence most cancers are ‘diseases of old age’). However, about 10% of cancers are started by inherited mutations (that the patient is born with), with further mutations being acquired after birth.

We should also bear in mind that collectively cancer comprise about 200 distinct diseases and that at the level of DNA sequence every tumour is unique.

Pancreatic cancer cells


Cancer cells dividing. Photograph: Visuals Unlimited, Inc./Dr. Stanley Flegler.




What’s new?

The work that the journalists caught on to didn’t describe any new experiments but instead looked at the long-standing puzzle of why cancers, although able to arise anywhere in the body, have a strong tissue bias. For example, tumours are twenty times more common in the large intestine than in the small intestine.

Noting that within many tissues most cells are short-lived and don’t give rise to progeny (and so are unlikely to initiate a tumour), the authors focused on the cells that can self-renew and are therefore responsible for the continued existence and repopulation of the tissue (often called stem cells). Searching the literature, they found 31 tissue types for which it was possible to work out how many stem cell divisions occur in an average human lifetime. Lo and behold, it turned out that the number of divisions correlated quite well with the lifetime risk for cancer in that tissue type i.e. the more replications of stem cells that a tissue requires over its lifetime to sustain its functional, the greater the risk of a tumour emerging in that tissue.

An interpretation of this is that the majority of cancers arise (i.e. are started) as a result of random mutations occurring during DNA replication in normal, non-cancerous cells. The underlying point here is that every time one cell makes two it must first duplicate its genetic material (i.e. replicate its DNA). This process is amazingly efficient but it’s not perfect (cells make a mistake once for every one thousand million coding units (i.e. bases) incorporated into new DNA). In the abstract of their paper the authors describe cancers initiated by these naturally occurring mutations as “bad luck” – unfortunately in my view, as the expression was a sure-fire red rag to the press bulls.

A really irritating example

From The Telegraph: “For years health experts have warned that tumours are driven by a bad diet, lack of exercise, or gene errors passed down from parents… But now a study has shown that most cancers are primarily caused by bad luck rather than poor lifestyle choices or defective DNA.”

NO IT HASN’T. Do you not read what you’ve written and consider how it might come across to readers who think they’ve grasped the basic picture, as summarized above under Let’s get things clear?

What the study confirms is that the major force behind cancers is the accumulation of mutations (defective DNA if you wish) as cells replicate during the lifetime of the individual. To the risk of getting cancers posed by this background to life may be added environmental factors that promote DNA damage and inherited variants in DNA (see A Taxing Inheritance for more about parental contributions).

Is this really anything new?

Well, it’s marginal and certainly not enough to merit the above headlines. The new work doesn’t alter in any way our summary. However, it’s interesting in that it offers an explanation for the wide variation in cancer incidence across different tissues and makes the point, for instance, that the relatively high rate of cell renewal in the lung makes this organ particularly susceptible to the mutagenic effects of cigarette smoke.

So, what about luck?

First we remain as we were: cancers are a fact of life – they’re hard-wired into the biology of life and they’ll come to all of us if we live long enough.

It is certainly true that there are many cancer patients who have had bad luck. They may have always eaten healthily, kept active and physically fit and been teetotal since birth and yet be stricken by, for example, a brain tumour or pancreatic cancer for which there are no known environmental risk factors that we can do anything about. They may have never smoked but nonetheless develop lung cancer (think of Roy Castle).

But it remains the case that for many cancers, it isn’t just about luck, it’s about choices, both for society and for individuals. Mention of environmental factors reminds us that mankind really isn’t doing very well on the self-help front. Eliminating smoking would reduce the global cancer burden (14 million new cases, over 8 million deaths per year) by about 22%. Infections, for example from contaminated drinking water, start about 20% of all cancers whilst alcohol consumption has a hand in about 4% and in the UK over 20% of bowel cancers are linked to eating red and processed meat.

Calm down!

I know that for all the effect my wittering about the quality of science journalism will have I might as well get on to the sports pages. I actually have some sympathy with the Gentlemen of the Press: writing about science is difficult – perhaps we should rejoice that there’s any national coverage. But there is a recurrent problem in the British press (see Not another ‘Great Cancer Breakthrough’!!!) that can easily be avoided. Just report evolving science stories as precisely and clearly as possible. They’re often sensational tales in their own right, so leave the sensationalism to the other pages and tell it as it is.

Rant over. Happy new year. Now, where’s the marmalade?


Tomasetti, C. and Vogelstein, B. (2015). Variation in cancer risk among tissues can be explained by the number of stem cell divisions. Science 347, 78-81.

Molecular Mosaics

Piazza Armerina is a little spot more or less in the centre of Sicily. It is in other words, with due respect to our Sicilian readers, pretty well the middle of nowhere. But you really should go there, off the beaten track though it is, because in the 4th century the local Roman landowner built himself a house. Unfortunately it was later buried by a landslide and remained thus until the 20th century when an excavation revealed that the villa contained the most extensive set of mosaics anywhere in the world. I should tell you that it takes a fair bit to get me excited about archaeological remains but I defy anyone not to be reduced to dropped jaws and softly exhaled ‘wows’ by the Villa Romana del Casale. The mosaics are vast and the scenes depicted span many aspects of the life of the times, from hunting to circuses and, although they won’t tell you who invented the bikini (and of course you’ve wondered!), the images of the ten ‘Bikini Girls’ reveal, so to speak, that such a garment, or lack of it, enlivened the Sicilian summer back in the days when Constantine & Co. were running the show.

Bikini girls

Bikini girls

How big is ‘vast’?

Mosaics seem to me unique among human creative endeavours because you may readily recognize the technical craft and bathe in the history or beauty displayed, as with other art forms, but after a bit you can’t ignore the feeling of sore knees and a twinge of back ache as your body goes into sympathetic spasms with the poor blighters who toiled away for decades to make the things. And how! In the Villa Romana there are 37 million colored tiles – each unique if examined closely enough, each set in position with loving care, each making its minute contributing to the whole.

Catching up – and overtaking – the Romans

It’s taken us seventeen centuries to come up with anything to match the complexity of such creations but the astonishing advances in sequencing our genetic code are now laying bare the mosaicism of tumours. The basic idea is simple to grasp: chip out little bits from different regions of a primary tumour and from secondary growths (metastases) that have seeded from the primary. Then throw the full power of DNA sequencing at them.

The picture that emerges is also simple to describe. Different regions of a tumour – as well as any secondary growths (formed when cells migrate through the circulation to lodge in other sites around the body, called metastases) – differ from each other. They have different ‘DNA signatures’. In other words, although the whole thing starts from a common ancestor (a starter cell if you like), tumours diverge as they grow. So you can think of them like evolving species – and draw family trees just like the ones that show how long it was since you were a chimp.

Kidney tumor muts.006Kidney Tumor tree

Dissecting a tumour (left) and its evolutionary tree.  R1-6: regions of a primary tumour; M: a secondary metastasis. The length of the lines between the kinks is proportional to the number of mutations picked up – thousands of them – on that stretch of the journey.

So tumours are a mosaic …

Indeed. Things are actually even more complicated than that because the sequence obtained for each bit is an average – it’s the predominant mutation pattern. That’s because even a small piece of tissue, say a millimetre in diameter, contains billions of cells (yes, that’s a thousand million) and, if you looked closely enough, you’d find that each individual cell has quirks in its DNA sequence that are all its own. Each is unique – like the tiles of the Villa Romana.

So what?

The molecular complexity, even within one tumour, is utterly mind boggling, but there are two good reasons why we should know what is being unearthed and make an effort to come to terms with it.

The first is that, although cancers are an aberration, the cellular variety they embody is a wonder to behold and the adaptability – mutability if you like – that they reveal gives us a new vantage point from which to contemplate the breathtaking variety of the natural world.

The second is more prosaic: if we’re going to reduce cancers to readily treatable conditions, we must understand the nature of the challenge – what makes a cancer cell tick. However stupefying the picture, that is what we are tackling and the more we know the more rational we can be in dealing with it.

So that’s good news?

Indeed. We’re approaching the point where it would be possible to offer this sort of screen to all new cancer patients and one benefit would be that we’d stop giving people drugs that won’t do any good – because they won’t hit the mutations carried by that individual. Such a programme would make a sizeable hole in the NHS budget but let’s follow the governmental example and not worry about money because we’ve got enough problems with the science. Although the huge number of mutations is daunting, a cheering point is that we can ignore most of them because we know that what’s important is a relatively small set of ‘driver’ mutations – the ones that force the abnormal cell proliferation that is the key feature of cancers – the rest are ‘passengers’, collateral DNA damage if you like, that surf along for the ride.

So we aren’t looking for drugs that can reverse all mutational effects – just enough to hit the drivers.

And the bad news?

We don’t have them. That is, there’s a huge number of drugs that are used against cancers but hardly any of them are ‘specific’ – actually target a molecular defect. Most are like a kind of shotgun that may affect cancer cells but does a lot of collateral damage. Cancers are driven by combinations of driver mutations, of which there are perhaps 20 or so that can come into play, and it’s the specific combinations of those in a given tumour that we need to target.

All of which means that there are two huge challenges: (1) to produce new drugs with good specificity that can hit the major drivers and (2) to be able to deploy them in combinations that give us a chance of outwitting the defences of cancer cells – namely mutating to use other drivers and getting rid of drugs by pumping them out. So, the biology is stunning but, despite the outrageous items that occasionally appear in our newspapers, it is difficult to be optimistic on anything other than a very long time-scale.

If, back in 330 AD, that Roman gent had said to his builder “So here’s the deal, chief: we’re aiming to decorate this place with about 40 million tiles” there would for sure have followed a good deal of stylus sucking as a prelude to “Can’t be done, Magister. Not with all the slaves in Rome” But it was. It just took a very long time and a lot of effort.


Gerlinger, M. et al., (2012). Intratumor Heterogeneity and Branched Evolution Revealed by Multiregion Sequencing. The New England Journal of Medicine 366;10 March 8.