Download Cancer Genomes for Cancer Care - The Royal Society of Edinburgh

The Royal Society of Edinburgh
Supported by the Scottish Cruden Foundation
and the Scottish Cancer Foundation
Cancer Genomes for Cancer Care
Professor Andrew Biankin, Director, Wolfson Wohl Cancer Research Centre
and Regius Chair of Surgery, University of Glasgow; Lab Head in Pancreatic
Cancer Research, Garvan Institute of Medical Research, Sydney, Australia
Monday 29 September 2014
Report by Jennifer Trueland
Professor Andrew Biankin made a plea for a new approach to cancer treatment, using personalised
medicine, based on genomics, to ensure patients get the right treatment, first time. This would mean
a radical change in the way we do clinical trials – and require a change of mindset in regulators –
but the potential benefits are huge.
The scene was set by RSE President Sir John Arbuthnott, who said that there is beginning to be a
different philosophy about the pattern of development of diseases, including cancer, and the way in
which individuals react; this can be very significantly affected by someone’s genetic make-up – and
treatments have to be personalised. Advances in molecular biology are making it possible to find
out more about a person’s genetic make-up, and the University of Glasgow is a pioneer in ‘stratified
medicine’, which involves tailoring treatments to people’s personal circumstances.
Professor Biankin began by talking about the history of personalised care. At first, we made
decisions in the clinic based on our observations, and then we had a microscope (a 300-year-old
technology). Now, he said, we have technologies that allow us to peer into what makes a disease a
particular type of disease.
That allows us to break down what we initially thought was one thing – and to find out why one
person will respond to treatment and another, apparently with the same disease, will not. Currently,
our public health systems group all these things together, but a change of mindset is needed. This
is particularly the case with cancer, because cancer is a disease of genes, and we have new
technologies that can discern these genomic differences, he added.
Professor Biankin spoke a little of his personal motivations for spending his working life trying to
make improvements in the treatment of pancreatic cancer, a particularly intractable and aggressive
disease. One of the first patients he treated as a consultant was a 39-year-old patient with
pancreatic cancer. She was otherwise fit and healthy, she was young, and had a ‘tiny’ tumour and
she ‘sailed through’ surgery. Her prospects should have been good, but ten months down the line,
the cancer came back. We really don’t know what we are dealing with; we have a rough idea, and a
few ideas about how to take things forward, but we’re really almost stuck in the way we manage
disease because we’re trying to treat what might be fundamentally different diseases in a similar, if
not identical, way.
We’ve known for almost a thousand years (since Avicenna’s 1025AD The Canon of Medicine) that
we should make sure we’re testing treatments for uniform conditions – we wouldn’t dream of
treating cancer all as ‘just cancer’, as a single entity – we treat it differently depending on where the
cancer is located, he said. Now we’re taking the next step forward, in looking at the differences
between people and the diseases they develop. A lot needs to change, including the way that
regulatory authorities look at approving treatments for disease, and how public health programmes
promote the idea of personal risk.
Professor Biankin then described cancer as “cells growing out of control”. But it’s not just that, he
added; it’s also about cells evolving. Cells are trying to escape the confines of their environment and
live as individuals. Understanding the process of how cells break down the barriers is important; it
can happen with a slow accumulation of change, or with a sudden catastrophe. What we’re trying to
understand are the different defects in the genome of a cancer, some of which we’re born with, and
others that we develop over time due to different environmental exposures or other processes.
Research in this area is being driven by advances in technology – genomics is advancing
“exponentially”. We can now start to map the genetic damage that occurs and use this to help
develop our knowledge about cancer. We started out just understanding the structure of genomes,
then we moved to understanding the biology; now we’re moving to understanding disease in the
clinic.
As an example of how quickly advances are happening, he described how the human genome was
sequenced in 15 years at a cost of $3 billion (for one person). That was in 2001; now we can
sequence a genome for under a thousand dollars in about a day. He spoke a little about the work of
the International Cancer Genome Consortium – 24 countries coming together to understand cancer
better, in the context of available new technologies. Various agreements have been made amongst
the participating countries about how these cancers are mapped out – 500 of each cancer type.
His involvement is in pancreatic cancer, the fourth leading cause of cancer death; less than 5%
survive five years, and 90% die within a year. “It’s a terrible disease,” he said. Surgery works
occasionally and oncologists use treatments that are “modestly effective”, with only incremental
benefit. Nothing has changed in 50 years. Pancreatic cancer gives the opportunity to understand if
and how a personalised approach to cancer can work.
The study that Professor Biankin did in partnership with Professor Sean Grimmond, also now in
Glasgow, involved recruiting 500 people with pancreatic cancer and tracking them as though it was
a clinical trial. When they were operated on, the researchers took a sample and sequenced it,
looking at all genetic abnormalities. They saw four genetic abnormalities that had been known about
for a long time, but they also saw remarkable complexity and heterogeneity, so they thought it was
little wonder that different patients have different cancers that act differently and respond differently
to treatment.
They identified sub-group of cancers that are potentially more likely to respond to a particular drug;
some of these drugs had been tested in the past, but because only four or five patients out of a
hundred were likely to respond, it was below the level required to detect the change and to take it
forward. They also used other techniques to try to determine the important genetic event that might
have led to the disease. Sometimes, genetic events contribute to the cancer, but sometimes they
might be ‘passengers’ so you don’t want to waste time targeting something that’s incidental. They
used animal models to try to determine what is and isn’t important.
That brought them to particular pathways or processes which could potentially be targets for drug
therapy. They have been looking at 14 pathways in detail, where they can map where the genetic
defects are, and which are driving the cancer. When they see groups of defects emerging,
researchers can start to think of where they can target drugs.
Looking deeply into the genomes of cancer, they have started to see patterns or signatures. These
tell us about the core biochemical processes that cause the genomic damage and, as a
consequence, cause the cancer to develop; for example, the exposure to smoking. Smoking leaves
a particular signature in the DNA, as does older age and in inherited mutations such as BRCA1 and
BRCA2 (associated with breast cancer). There’s also a signature associated with UV light damage,
associated with skin cancer.
These signatures have potential future importance in screening, considering risk, early detection
and prevention. Making sense of these events and having the computational power to analyse them
is significant, and is improving as technology becomes more powerful.
He then talked in more detail about precision therapy, particularly in pancreatic cancer. Because we
can start to see differences at a molecular level, we can begin to see different sub-groups within the
main cancer types (e.g., breast, colorectal). The sub-groups have clinically-relevant information and
allow us to select the particular patient for a particular drug. This has economic implications, as well
as being therapeutically important, as you’re not wasting money – or time – on drugs that won’t
work for a particular patient. “Giving 95% of patients the wrong drug isn’t helpful,” he said.
In pancreatic cancer, most patients don’t get drug treatment; some get a toxic combination of drugs
which works in a small percentage of patients. What we should do is give the drugs to the patients
who are likely to respond, and give other patients the opportunity of novel therapies. Unfortunately,
the traditional clinical trial systems that are in place, based on large numbers of patients, poses
challenges when you are trying to develop personalised therapies.
When you look deeply into cancer genomes, you can see drug opportunities with drugs you already
have. For example, the drug Herceptin is used in breast cancer, but the same kinds of genetic
mutations are seen in other cancers, but only in small numbers – 2–3% – of patients. How do we
pick these patients ahead of time?
The researchers could look back into their data and see if by chance the patient got the right
treatment. When it was right by accident, they saw patients having dramatic and exceptional
responses. This provided valuable information for considering potential treatment pathways for
people with similar genetic mutations. There are different patterns with different subgroups – in
pancreatic cancer, it normally spreads to the liver, but in one subgroup it occurs in the lung. These
subgroups are likely to need different clinical approaches.
It’s possible to do this for small groups, but it’s hugely difficult to test it in large numbers of patients.
If you tested everybody, you’d need 7,540 patients just to detect if 10% benefited from a treatment.
The largest pancreatic cancer study in advanced disease is around 800 patients.
We need to do clinical trials differently, identify patients with the right sub-type and give them
treatments accordingly – essentially providing a trial for the patient rather than finding patients for a
trial. People can be tested to see if they fit a particular sub-group, but there are ethical, financial and
other challenges to overcome before they can get the drug that might help. There’s a need for a
move to a parallel model to find a trial for the patient, he said. That would involve testing everyone’s
cancer, identifying the sub-group and matching them to the treatment.
This is something that’s building in the UK – with the idea of giving everyone the optimal treatment
upfront. Precision PANC is an approach that involves transforming the treatment paradigms for
people with pancreatic cancer. There’s an opportunity to do this in pancreatic cancer because there
is no standard treatment and nothing has changed in 50 years – potentially because we haven’t
understood the diversity. Researchers really like being part of such “translational research” efforts
because they can see that what they are doing in the lab – identifying the right drug – has a
potential real impact on patients now, and can make a difference, not in 30 years but in a matter of
months.
A big challenge is to change the current trajectory of the healthcare community, Professor Biankin
said. In the future, we might see that a clinician would get a genomic report and work out the best
therapeutic option for a patient based on evidence acquired through different mechanisms, and not
necessarily the unselected clinical trial. It’s more possible in the UK with its unified healthcare
system. It’s perhaps time we started thinking about a ‘knowledge bank’ approach to disease, with a
rapidly learning health system, and looking to see what might work, based on a growing body of
experience. “It completely freaks the regulators out,” he said, but we have to get past that. The key
element is accumulating data and bringing it together. With appropriate safeguards, new
technologies – such as Google Glass and biosensors – will help clinicians and researchers, he
added.
Questions
The discussion covered a variety of topics, including how practical it is to expect clinicians to read,
understand and act on a molecular phenotype, and how the pharmaceutical industry is reacting to
the changing environment.
On the former, Professor Biankin said that it was possible to do it for individuals, but the challenge is
doing it at scale. He suggested including genomics and informatics experts in the multi-disciplinary
team looking after patients with cancer.
Asked about the human impact of gathering data, Professor Biankin said that investment would be
needed, but that much of the process could be automated.
Asked about borderline cases where it is not certain if something is cancer, let alone which subtype, Professor Biankin said the shades of grey will always be there, but we will refine these
definitions over time.
The pharmaceutical industry is interested in the developments, particularly those working in
translational medicine, he added, but the regulators are still saying that randomised controlled trials
are the only way to go. This is, of course, appropriate where the disease is common; but as we can
better refine our classifications, in many instances we are already at the point where we just
wouldn’t have enough patients who could be practically accessed to test therapies appropriately.
We need to think about how we interpret data and evidence of efficacy in a way that is more precise
and delicate, rather than the current broad brushstroke approach.
RSE President Sir John Arbuthnott asked about how you teach this new way of understanding
treatment and cancer. Professor Biankin said he didn’t think it is that much of a great leap forward in
thinking, but we are trapped in the systems we have developed. We have to move research and
clinical care closer together; it’s about integrating scientific endeavour with clinical training, and with
clinical service delivery, so that all stakeholders are linked and align to a common goal.
Vote of Thanks
The Vote of Thanks was delivered by Professor Robert Steele, Chair of the Scottish Cancer
Foundation, which supported the event. As a surgeon himself, he said he appreciated Professor
Biankin’s acknowledgment that his motivation comes from patients. Cancer treatment is becoming
more complex, he said, and trials need to shift from a “blunderbuss” approach, to become more
nuanced.
Opinions expressed here do not necessarily represent the views of the RSE, nor of its Fellows
The Royal Society of Edinburgh, Scotland’s National Academy, is Scottish Charity No. SC000470