Getting personal with cancer

Recent work on developing targeted treatments for pancreatic cancer is paving the way for new approaches in personalised medicine.

To appropriate Tolstoy: healthy cells are all alike; every unhealthy cell is unhealthy in its own way. This is particularly the case when it comes to pancreatic cancer. While it is only the twelfth most common cancer in men and ninth most common in women, pancreatic cancer is the fourth leading cause of cancer death for both sexes and is projected to be the second leading cause of cancer death by 2020.

Pancreatic cancer continues to rise up the lethality charts in part because no made major inroads have been made into treating it in over 50 years of concerted endeavour. And this is due to its aggressive and invasive nature, along with the staggering diversity found in pancreatic cancers.

Where healthy pancreas cells have their manifold genes, functions and regulatory systems intact, pancreatic cancer cells break the mould in innumerable ways, making them difficult to target with conventional chemotherapy treatments.

A typical clinical trial of a new chemotherapy drug with a few hundred pancreatic cancer patients might only see a handful respond positively to the treatment. Often that number is below statistical significance and insufficient to demonstrate efficacy, and the drug is shelved.

However, new research is showing how the very genetic diversity of pancreatic cancer could prove to be its Achilles heel. Where conventional therapies fail due to the diversity of the cancers being targeted, personalised therapies can hone in on those very individual differences and target them in a more refined way, and this could pave the way for new personalised treatments of other cancers.

Molecular heterogeneity

The molecular diversity of pancreatic cancer has been highlighted in recent work by the International Cancer Genome Consortium (ICGC) and its subproject, the Australian Pancreatic Cancer Genome Initiative (APGI), co-led by Professor Andrew Biankin at the Garvan and Professor Sean Grimmond at the Institute of Molecular Bioscience at the University of Queensland and their colleagues.

The APGI is already well ahead of its five-year goal set in 2009 of gathering sequence data from 350 pancreatic cancer patients. It has samples of normal and tumour tissue collected from over 500 patients, and over 400 of these have already been genomically characterised by Grimmond and his colleagues at the Queensland Centre for Medical Genomics using the latest in high-throughput sequencing technology.

The findings have already been illuminating. As Biankin points out, one of the key discoveries is that pancreatic cancer is better understood not as a single disease but as a constellation of multiple diseases characterised by a tremendous diversity of molecular aberrations.

In a paper published in February this year in Current Opinion in Genetics & Development, Biankin, Grimmond and Dr David Chang from the University of Glasgow highlight that even histologically indistinguishable cancers can have radically different molecular aberrations. And while there are some common mutations among tumours, there is a ‘long tail’ of mutations with a frequency of less than 5%, and it is unclear how many of these might be driver versus passenger mutations.

Adding to the complexity of this picture is the observation that there is often further intratumoural heterogeneity among metastases in a single patient.

Matching treatment to patient

Attempting to treat pancreatic cancer as a single disease and develop drugs that target it in a conventional way is clearly going to run into some serious hurdles.

“Previously we’ve been asking: does this new drug work in a third or more of this particular group of cancer patients? And if it didn’t work, we abandoned it because the system couldn’t deal with how to advance them to the clinic,” said Biankin.

However, in many of these trials there have been a few exceptional responders: patients who see a genuine benefit from the drug in question. So some of the drugs do appear to work against some of the cancers, but the trick is in figuring out which drugs work with which patients.

“The challenge is to identify those patients beforehand and, to use one of Sean’s phrases, take the guesswork out of chemotherapy, so you’re matching the right treatment to the right patient,” says Biankin.

However, using conventional clinical trial practices to apply the multitude of drugs to the manifold cohorts of patients in order to find clear statistical patterns is simply untenable: we’d be at it for centuries before we had the data we need to prescribe the therapies appropriately. And pancreatic cancer patients often don’t even have months to wait for the outcome.

Making it personal

The very promise of personalised medicine is that it will take a more sophisticated approach to targeting treatments at highly heterogeneous diseases such as pancreatic cancer.

The volumes of sequencing data, along with greater insight into the functional role various mutations play and how drugs can intervene, are yielding a new approach to cancer treatment. One possibility is identifying similarities between certain subtypes of pancreatic cancer and other cancers.

“When we look at pancreatic cancer we see a few genes that are very common, then a whole slew of other genes that are involved in other cancers, but occurring at low percentages,” said Grimmond. “That can be depressing from a traditional point of view, but from a personalised medicine point of view, it’s actually a great opportunity.”

This raises the prospect that drugs that are currently used to treat other cancers such as breast or gastric might be repurposed for particular subtypes of pancreatic cancer. If those cancers have the same biotype as a certain pancreatic cancer - meaning they exhibit the same types of genetic damage - then drugs that treat one might treat the other.

As Biankin points out, this approach involves flipping some conventional wisdom about cancer on its head.

“What we do now is group cancers by organ and split them on their molecular differences,” he explained. “The reverse approach may be more practical for low prevalence subtypes.”

New drugs can also be targeted at specific mutations, and can be tested in clinical trials consisting only of patients with cancers that are believed to exhibit those mutations. This is more like treating pancreatic cancer as the many diseases it is rather than taking a more conventional monolithic approach.

The technique could pave the way for personalised treatments of other cancers as well, says Grimmond. “The reason I started working on pancreatic cancer is I think it’s the exemplar for going forward with personalised medicine,” he says. “This is because we have a disease here that traditional methods are clearly failing, and if we want to try to prove clinical utility for personalised medicine, it’s the best cancer to start with.”

Personal challenges

The promise is clear, although neither Biankin nor Grimmond underestimates the hurdles that are yet to be overcome in bringing personalised cancer therapies to the clinic.

Interestingly, of all the challenges, technology isn’t one of them. As Grimmond points out, next-generation sequencing has the uncanny ability to undergo a tectonic shift every couple of years, delivering massive improvements in the speed, depth and cost of sequencing.

“I envisage that we’ll see a continued scaling up of the sequencing technologies currently being used,” he said. “And inevitably the cost of sequencing will continue to drop. When we get down to that $1000 for a cancer genome at an appropriate depth, rather than just $1000 for a germ-line genome, we’ll see sequencing much more readily adopted.”

And when that reaches $100 for a cancer genome, the potential for personalised treatment will truly explode. However, the major difficulties in advancing personalised cancer treatments lie elsewhere.

“The challenges are not in the technology or the science, or doing the screening,” Biankin said. “The challenges are logistical.”

For example, simply getting the tumour sample from the pathology department and processing it for sequencing on a broad scale is difficult in itself.

“You’d think it’d be easy, but that’s one of the hardest bits,” Biankin reported.

Another challenge is keeping patients in the clinical trials themselves. If a patient is entered into a trial where they might receive a new therapy or the existing standard of care, why would they risk the latter if they know it has a historically dreadful track record?

As Biankin states, the infrastructure also simply doesn’t exist to run widespread personalised treatments. The existing infrastructure - and culture of treatment - is predicated on the monolithic model, and is likely to be difficult to adapt to a personalised model, where each patient may end up being treated as a cohort of one.

According to Biankin, the system needs a rethink to be able to adapt to the personalised approach.

“We need a structural change in Australia,” he said. “That’s something we’ve been working towards for a number of years, and there are a few people in government who are pushing for a framework for a stratified medicines program in cancer.”

Other countries are also facing similar challenges but are getting on the front foot, such as in the United Kingdom.

“Cancer Research UK has a program set up where you have biospecimen collection hubs, technical hubs to do the assays and who facilitate the clinical trials,” Biankin said, adding that Australia needs to follow a similar path.

There are some who are pioneering a whole-of-system approach to personalised medicine, such as Professor David Thomas at the Peter MacCallum Cancer Institute (soon to be at Garvan), who is heading up Cancer2015. This initiative aims to weave genomics into clinical care, and Thomas is planning to bring the program from Victoria to New South Wales.

Deep IMPaCT

Biankin and colleagues are also running a pilot trial called IMPaCT (Individualised Molecular Pancreatic Cancer Therapy), which seeks to use personalised methods to treat pancreatic cancer and develop new processes that can be applied to other personalised treatments more generally.

The trial itself started when Biankin was pouring over the data that was emerging from the sequencing initiatives.

“There was information coming through that, if I was a patient, I’d want to know,” he recalled. He wanted to find ways to return that information to patients, and hopefully use it to direct their treatments.

Under IMPaCT, patients are screened for a molecular phenotype that might be treatable with existing therapies and are then randomised into the existing standard of care, receiving gemcitabine or the personalised treatment.

The trial is currently targeting a few particular phenotypes, including HER2/neu amplification, to be treated with gemcitabine and trastuzumab; BRCA1, BRCA2 and PALB2 mutations, treated with 5-FU and mitomycin C; and KRAS wild type, treated with gemcitabine and erlotinib.

“The IMPaCT model is transferable to other cancers where we’re seeing a failure by the limits of standard medicine to give any improvement in outcome,” said Grimmond. “And if you look at the plots of the cancer types that have very poor five-year survival, they are excellent candidates straight away, such as oesophageal cancer or cancer of unknown primary. This is where there’s an immediate application.”

The hope is that the IMPaCT trial will reveal new systems that can smooth the process of identifying candidate patients and getting personalised treatments to them as quickly and efficiently as possible. Although Biankin and Grimmond acknowledge that it will likely take some time to train a new generation of researchers and clinicians and encourage them to work more closely together and develop processes to make personalised medicine a reality.

Professor Andrew Biankin is Head of Pancreatic Cancer Research at the Garvan Institute of Medical Research in Sydney and is a conjoint Professor at the University of New South Wales. He is also Director of the Wolfson Wohl Cancer Research Centre and Regius Chair of Surgery at the University of Glasgow in the UK. He is a surgeon scientist who specialises in pancreatic diseases, particularly focusing on pancreatic cancer and its precursor lesions. He completed a medical degree and a PhD into the molecular pathology of pancreatic cancer at the University of New South Wales. Following that he was NHMRC Neil Hamilton Fairley Postdoctoral Fellow at Johns Hopkins University in the US. Biankin is the Chairman of the NSW Pancreatic Cancer Network, and is clinical lead of the Australian Pancreatic Cancer Genome Initiative with Professor Sean Grimmond.

Professor Sean Grimmond is a Laboratory Head of the Genomics of Development and Disease Division at the University of Queensland’s Institute for Molecular Bioscience (IMB) and Director of IMB’s Queensland Centre for Medical Genomics. He is also Chair of Medical Genomics at the Translational Research Centre at the University of Glasgow in the UK. He completed a PhD in genetics at the University of Queensland and postdoctoral studies at the Queensland Institute of Medical Research, and the MRC Mammalian Genetics Unit in the United Kingdom. His research over the last decade has focused on defining the molecular networks controlling biological processes and pathological states through genome-wide surveying of sequence content, transcriptome complexity and epigenomic signatures. Grimmond is the co-lead of the Australian Pancreatic Cancer Genome Initiative with Professor Andrew Biankin.

Image credit: ©reineg/Dollar Photo Club