Microscope image of pancreatic cancer cells

Pancreatic cancer cells - image courtesy of the London Research Institute EM Unit Pancreatic cancer cells

Although overall cancer survival has doubled over the last 40 years, progress hasn’t been uniform. For some types of cancer, such as pancreatic cancer, improvements have been few and far between.

Despite decades of research, there has been little significant improvement in survival for patients with this hard to treat cancer. Fewer than three in every 100 people diagnosed with pancreatic cancer survive for five years or more – a figure that’s barely budged in decades.

Something has to change.

Professor Andrew Biankin

Professor Andrew Biankin

One person trying hard to improve the outlook for pancreatic cancer patients is Professor Andrew Biankin, who moved from Australia to Scotland to work at Cancer Research UK’s Glasgow Centre in 2013.

As he told us in a recent blog post, Professor Biankin is determined to make progress against the disease on all fronts – understanding it, diagnosing it and treating it.

While he worked in Australia, he helped set up the Australian Pancreatic Cancer Genome Initiative (APGI) – part of the International Cancer Genome Consortium (ICGC) – a global project which aims to read, analyse and understand the genetic make-up of different cancers.

As part of this work, his team studied the genetic make-up – or genome – of tumours from 100 pancreatic patients who agreed to take part in the APGI.

In fascinating findings – published yesterday in the journal Nature – Professor Biankin and his colleagues at Cancer Research UK’s Glasgow Centre, the Garvan Institute of Medical Research and the University of Queensland in Australia have discovered that – based on analysing tumour DNA – there seem to be at least four different types of pancreatic cancer.

And tantalisingly, one of them – known as the ‘unstable’ type – might be extremely responsive to a particular type of chemotherapy not normally used to treat the disease.

Structure underlying chaos

When scientists search tumours for cancer-causing mistakes, they often look for specific errors in the chemical code of a tumour’s DNA. These so-called ‘point mutations’ have been thoroughly catalogued across many types of cancer, and often occur in important genes like p53 or RAS.

But these mistakes aren’t the only ones that cause cancer. Errors can also occur when large bits of a cell’s DNA get moved around, copied, deleted or inserted somewhere else in the genome.

This genetic shuffling is called structural variation. It can occur in different forms: genes get deleted, repeated, turned on or off, or even stitched together to make new, mutant forms – all of which can lead to cancer.

So as well as trying to catalogue the individual ‘spelling mistakes’ in pancreatic cancer, Biankin and Grimmond’s team wanted to know the degree to which genetic shuffling is involved in the disease, and whether different types are related to how cancer cells behave in different patients.

So they turned to a technique called whole-genome sequencing, which allowed them to take a detailed look at all the DNA in tumours from 100 pancreatic cancer patients. They also had access to the patients’ medical history, so they could relate the genetic changes they found to how the patients subsequently fared.

The Usual Suspects

Professor Biankin and his team are looking for the genetic mistakes in pancreatic cancer's DNA code.

Professor Biankin and his team are looking for the genetic mistakes in pancreatic cancer’s DNA code.

Firstly, the researchers looked to see if they could identify any new point mutations in a lot of the pancreatic tumour samples – often the starting point in the search for new targeted cancer drugs.

Several large studies have previously spotted mutations in genes such as KRAS, TP53 and CDKN2A in pancreatic cancers. Normally these genes are involved in controlling how cells grow and multiply, and they’re kept tightly in check so new cells are only made when and where they’re needed. But if they’re faulty, the controls are lost and cells can grow out of control, leading to cancer. Unsurprisingly, the team found that these genes were commonly damaged or faulty in their group of patients.

But, frustratingly, no other genes were found to be altered nearly as frequently across their samples: instead they found a whole host of different mutations in each cancer.

If the team were going to find new genetic culprits for pancreatic cancer – or targets for new drugs – they would have to keep looking.

So next, they decided to look at structural changes in the DNA from patients’ tumours.

The Unknown

The genetic shuffling in the tumours revealed something completely new: the samples could be divided into four distinct groups, based on the level and type of structural variation in each tumour’s DNA.

  • About a fifth (20 per cent) of the tumours were stable, with relatively low levels of DNA shuffling.
  • A third were what they called locally rearranged (30 per cent), meaning shuffling only happened in a particular area of the genome
  • Another third was described as scattered (36 per cent) – they displayed quite a bit of genetic shuffling in lots of areas of the genome
  • And – most intriguingly –about 14 of their hundred samples were unstable tumours, with incredibly chaotic DNA.

Fingering the culprits

Watch an animation exploring 10 things you might not know about pancreatic cancer.

Watch an animation exploring 10 things you might not know about pancreatic cancer.

Having discovered that there are at least four types of pancreatic cancer, The Pancreatic Team decided to look at tumours with unstable genomes in more detail.

Back in 2013, researchers from the Sanger Institute in Cambridge made an important discovery. The identified the molecular fingerprints of at least twenty different damaging processes at work in our DNA (we blogged about it in detail at the time).

One of these fingerprints was frequently seen in breast, ovarian and pancreatic tumours – often, though not always, in those that also had a mutation in either the BRCA1 or BRCA2 gene, which are involved in repairing damaged DNA. When either of these genes is damaged or missing, cells can’t repair their DNA properly and mistakes quickly build up, driving the development of cancer. These BRCA-related tumours tend to have a high level of genetic chaos, just like the unstable pancreatic tumours in the new study.

To find out if similar processes were at work, the team decided to see whether the unstable pancreatic tumours also had the chaotic genetic hallmarks of faulty BRCA genes, as well as checking for faults in the BRCA genes themselves.

They found that nine of the 11 tumours with unstable genomes had the unmistakeable fingerprints caused by faulty BRCA. And most of these also had a mistake in either BRCA1 or BRCA2.

This suggested that the three phenomena were related: having an unstable genome, having lots of BRCA-related DNA damage, and having a faulty BRCA gene.

This opens up an extremely interesting possibility.

Pushing tumours over the edge

Different classes of chemotherapy drug work in different ways. Most pancreatic cancer patients who have chemotherapy are given a drug called gemcitabine – which mimics the structure of the building blocks of our DNA and interferes with how it’s copied inside cancer cells.

Platinum-based drugs like cisplatin form another class of treatment. These form random chemical bridges between DNA molecules, causing them to break up when cancer cells divide. The cells can’t survive with such fragmented DNA, so they die.

Over the years, researchers have discovered that platinum drugs appear to be highly effective in cancers that already have unstable DNA – they seem to ‘push them over the edge’ and kill them. They’re not normally used to treat pancreatic cancer. But under some circumstances patients can and do receive them. In some cases they help but in others they don’t, suggesting there might be a subset of these patients who can benefit from them.

Professor Biankin’s team decided to explore whether the most unstable tumours in their study were the sort of cancers that could be successfully treated with platinum drugs. Of the hundred patients in the study, eight had received platinum drugs – five of whom had unstable tumours, and three who didn’t

Amazingly, two of the ’unstable’ patients had had exceptional responses. Their cancers had completely gone. Two more had robust partial responses, meaning most – but not quite all – of their cancer disappeared.

This is amazing.

Let’s take it outside

Then the team did something even more ingenious. As part of their study, they managed to grow some of the patients’ tumours in groups of mice – known as ‘patient derived xenografts’. These animals were given different drugs to see how they responded.

Where the tumours had come from a patient with an unstable tumour, two out of three tumours responded to platinum drugs and shrank. But none of tumours from other patients responded to the platinum-based chemotherapy.

This is an extremely promising finding – and suggests that a tumour’s ‘unstableness’ – if it can be reliably measured – could be a promising marker for how to treat patients with pancreatic cancer.

What Next?

The team’s results suggest that if a patient’s pancreatic tumour has an unstable genome, they might benefit from platinum based chemotherapy drugs – or even, the researchers speculate, newer ‘PARP inhibitor’ drugs like olaparib. These are therapies that interfere with a DNA repair protein called PARP, and are known to work particularly well in cancers driven by BRCA gene faults.

“Because of technical limitations we’re not quite at the point where we can perform whole genome sequencing in a clinical trial to identify patients with unstable genomes” – Professor Biankin

This would be a big step forward for patients. But it’s still early days.

The only way to find out if an unstable genome is a good way to identify platinum-sensitive pancreatic cancer patients is to develop a robust test that accurately selects these people and assesses how well the drugs work for them in clinical trials.

This is something Professor Biankin is keen to do, although he is aware that the dream is not yet a reality.

“Because of technical limitations we’re not quite at the point where we can perform whole genome sequencing in a clinical trial to identify patients with unstable genomes. But what we can do is look for mistakes in genes we saw were mutated in unstable tumours and use these – along with mutations in the BRCA genes – to select patients for platinum therapy,” he says.

“We can then go back and perform whole genome sequencing on the tumour of those patients that received platinum based therapy in the trial to see if they had unstable genomes, or identify other clues as to why some respond and others don’t.”

A final word

This is what a stretch of DNA code can look like on a computer - it's complicated

This is what a stretch of DNA code can look like on a computer – it’s complicated

This paper gives us new information about pancreatic cancer and offers a potential way of improving treatment and survival for a subset of patients whose cancers have unstable genomes.

It also offers a potential new way to identify these patients, through DNA analysis.

Professor Biankin and his colleagues think that by reading the entire genome of a patient’s cancer, doctors will have a clearer picture of what is going on inside a tumour, compared with only looking at a handful of genes.

But turning this kind of whole genome analysis into a routine, clinical test is not going to be easy. Before using this technique can become a reality there are several practical hurdles that need to be overcome, such as being able to use the kind of samples that are usually provided by clinics, and getting reliable results quickly (and cheaply) enough to be useful for guiding treatment.

As Professor Biankin has told us before, there is no time to lose when it comes to starting therapy for people with this disease. None of these are insurmountable, and further research should provide solutions.

We know that much more needs to be done to improve survival for pancreatic cancer, which is why it’s been chosen as one of our main research areas over the coming years.

These new findings offer hope that we are starting to uncover the weaknesses in this terrible disease, opening doors to desperately-needed treatments that will make a real difference for patients.



  • Waddell, N., et al. (2015). Whole genomes redefine the mutational landscape of pancreatic cancer Nature, 518 (7540), 495-501 DOI: 10.1038/nature14169


DNA code image from Flickr, under CC BY-SA 2.0

Sequencing image from Flickr, under CC BY-NC-ND 2.0