For a long time, cancer was believed to be one disease. But as our understanding has grown over the last 20 years, so has our awareness of its complexity.
We now know that there are hundreds of different types of cancer – and that each one differs in its biology, the genes that cause it and how it grows. This is why patients receive different treatments, and why survival varies between cancer types.
And to further complicate things, research has shown that one ‘type’ of cancer can be further broken down into distinct sub-types, each of which has its own molecular fingerprint and unique weak spots. For example, our METABRIC study showed that ‘breast cancer’ is in fact 10 distinct diseases all of which harbour diverse genetic mistakes and respond differently to treatment.
While it might seem disheartening to learn that cancer is such a complex enemy, it’s actually hugely beneficial to realise that this is the case.
Armed with this knowledge, scientists can improve treatments and develop new drugs that target each cancer and the various sub-groups. It can also help doctors offer more informed, accurate advice to patients about what treatment could work best for them.
That’s why we’re excited about new research published today in the journal EBioMedicine, by a team of researchers led by Dr Alastair Lamb, who works at both our Cambridge Institute, and the University of Cambridge’s Academic Urology Group.
Dr Lamb’s team has shown, for the first time, that prostate cancer can be divided into five distinct groups, each of which has a unique molecular signature that appears to predict how well a patient will do after surgery.
So what exactly did the researchers do, and what do their findings mean for prostate cancer patients?
Spot the difference
First, a little background.
To divide (or ‘stratify’) patients into sub-groups, researchers look for differences between them.
Traditionally, this has been done by looking at how the cells of a tumour appear down a microscope, or whether the tumour manufactures certain molecules.
But increasingly, this can be done by analysing the DNA in a patients’ tumour. And there are now many different ways of doing this.
For example, researchers can look to see if certain areas of DNA are deleted or repeated – so-called copy-number alterations. Patients with similar deletion or repetition patterns are then grouped together, and can be followed to see if they share other characteristics too.
Alternatively, by measuring the levels of hundreds, or sometimes thousands of molecules called messenger RNAs (mRNAs), researchers can stratify patients based on the activity levels of different genes in their tumour.
Each of these methods on its own is good at finding differences between patients and identifying distinct cancer sub-types.
But Dr Lamb’s team decided to use a combined approach which looked at both copy-number alterations and changes in mRNA levels.
“If we find a copy-number alteration that results in a change at the mRNA level, the logical deduction is that this would have more of an impact,” he told us.
“So by combining these techniques, we could potentially identify an important group of genes that are dramatically altered in prostate cancer patients.”
100 is the magic number
First, the researchers looked at copy-number alterations in tumour samples from 156 men who had undergone surgery in Cambridge to remove their prostate after being diagnosed with cancer. They then analysed the levels of thousands of mRNAs in these same samples.
Next, they compared the two sets of results. Could they find regions of genetic disruption that, by altering gene activity – the level at which they are switched on or off – could be driving the growth of tumours?
The answer was yes – they spotted around 1,000 genes where changes in the copy-number – that is, they were deleted or repeated – also changed gene activity.
The researchers then homed in on 100 genes with the greatest change compared with healthy samples and which seemed to be crucial to prostate cancer growth, and looked to see how their activities varied across the 156 men.
They found that by using this group of 100 genes – or gene signature – prostate cancer patients who’d had surgery to remove their prostate could be divided into five distinct sub-groups.
- One group had lots of DNA deletions and consequently low activity of certain genes
- Another had high amounts of DNA repetition which resulted in increased activity of specific genes.
- Two more groups had very few copy-number alterations, or changes in activity.
- The fifth and final group had some – but not too many – copy-number alterations.
To confirm their findings, the researchers performed the same analysis on another group of 103 men, who’d had surgery in Sweden.
Confirming their findings, they identified the same set of 100 genes in these men, and confirmed that they could also be divided into five groups based on this gene signature.
These are exciting results – it’s the first time anyone had shown that prostate cancer could be grouped in this way.
But do men with different gene signatures actually have different forms of prostate cancer? To find out, the researchers decided to see how the men fared after treatment, and whether that was related to which of the five groups they belonged to.
Specifically, they wanted to see if the cancer was more likely to come back in one group of men compared with another.
Again, the answer was yes.
Among the Cambridge men, when the researchers compared the number in each group who’s cancer came backafter surgery, they found that men in groups with greater numbers of gene changes were more likely to relapse, compared with patients who had fewer changes.
The researchers checked their findings in the Swedish group – those with higher levels of genetic alterations were also more likely to relapse after surgery.
This, says Lamb, could one day help doctors tailor treatments for men with prostate cancer.
“The idea is that these gene signatures could be used alongside other clinical tests like the Gleason score and PSA test to give doctors a more accurate idea of how men will do after an operation to remove their prostate,” Lamb says.
“By combining the molecular information provided by this gene signature with existing clinical information, doctors might be able to identify those most at risk of relapse and treat them accordingly. For example, if a man had a low Gleason score but high levels of genetic alteration, he would be considered to be at a moderate-to-high risk of relapse.
“Doctors could use this extra molecular information to get a more accurate idea of what’s happening in a patient’s tumour and provide them with more tailored treatment advice.”
There’s always a ‘but’
These are exciting results. But like most research, there are caveats.
This research was carried out on samples from men who had already had surgery to remove their prostate. So, because these weren’t samples taken at the time they were initially diagnosed, there’s work to do to see whether this gene signature can be used in practice, to divide prostate cancer patients into different groups based on diagnostic biopsies, and whether it should be used to determine treatment from diagnosis.
We still need to do more research to see if this technique can be used routinely by doctors in the hospital
– Dr Alastair Lamb
It’s something Lamb is aware of, and something he hopes to address in the future.
“At the moment there are a few obstacles to overcome before our approach can be used on diagnostic samples. For example, can we get enough material from a diagnostic biopsy to be able to do the analysis?”
“We need to be sure that the sample taken at biopsy represents the majority of cancer cells in the tumour.”
A relatively new technique called targeted biopsying may be able to overcome this problem, he says.
“Targeted biopsying uses an MRI scan in combination with standard histopathology to identify the best area of the tumour to take a sample from. The idea is that the biggest area – which generally represents the bulk of the tumour – will show up on the MRI. So using these techniques together should give doctors confidence that they are biopsying the most important part of the tumour.
“Once we overcome these challenges we can apply what we have learned from our latest study to try and predict from an earlier stage which men are most at risk of relapse.”
Lamb also cautions that there is still work to do before classifying patients based on this signature is possible.
“We still need to do more research to see if this technique can be used routinely by doctors in the hospital. We also need to confirm that it can change how we manage patients’ treatment for the better, and reduce the number of men relapsing after prostate surgery.”
In the future, these findings could help make sure that prostate cancer patients are given the most appropriate treatment and that those most at risk of the cancer returning, or with a more aggressive form of the disease, are offered the most appropriate treatment that would increase their chances of survival.
That’s something to be excited about.
Ross-Adams et al. Integration of copy number and transcriptomics provides risk stratification in prostate cancer: a discovery and validation cohort study. EBioMedicine. DOI: 10.1016/j.ebiom.2015.07.017.