The route to ‘precision’ medicine – four key challenges we need to meet

We’ve often written about the revolution in ‘precision’, or ‘personalised’, cancer medicine.

This is the aspiration that, armed with both a detailed readout of the molecular faults in a patient’s tumour, and new generation of drugs that precisely target them, doctors will be able to select the best, most effective treatments for their patients and then accurately monitor their success.

And with new ‘targeted’ drugs – like erlotinib and cetuximab – now available for certain patients whose tumours bear particular genetic faults, this is slowly becoming a reality.

In a sense, there’s nothing that revolutionary about the concept: doctors have long been able to offer patients a variety of tests, and use the results to guide how they treat their patients.

But as anyone who’s been affected by cancer knows, the tests and treatments in use today are far from perfect. Even as average survival rates continue to edge steadily upwards, doctors often can’t say for sure whether a particular patient will respond to a given drug, or whether their cancer’s been completely cured. There’s still too much ‘hit and miss’, and too much ‘wait and see’.

So ‘precision’ medicine is perhaps more of a refinement than a revolution:  a move towards incrementally more effective, targeted drugs; less uncertainty, more accuracy – and, ultimately, better care for people with the disease.

But how do we get from here to there? We’ve written before about the ‘slow dawn’ of this new era, and the scientific and practical hurdles that researchers and healthcare systems are striving to overcome.

In October last year, in an attempt to begin to clear these hurdles, the first ever ‘Molecular Analysis for Personalised Therapy’ conference took place in Paris.

Unlike many other research conferences, its organisers – Cancer Research UK’s Professor Charlie Swanton, and Professor Fabrice André and Professor Jean-Charles Soria from France’s Gustave Roussy Institute – wanted a meeting specifically to discuss the practical issues doctors and researchers face as they try to make precision medicine a reality for patients.

And over two intense days – and scores of presentations, discussions and debate – four key challenges emerged.

1. Which molecules should we test for?

Analysis of data from countless thousands of patients has shown that tumours contain a multitude of different genetic aberrations, known as mutations. But just a handful of these actually fuel the disease’s growth: so-called ‘driver’ mutations. The rest seem to be collateral, accumulating as the tumour grows – ‘passengers’ along for the ride.

So a key challenge is to work out which of the thousands of mutations in a patient’s tumour are drivers – a challenge made harder when you realise one tumour’s genetic driver may be another’s passenger. ‘Context’ matters as much as ‘content’.

So a hot topic at the conference was the race to develop software that accurately sorts the drivers from the passengers, on a patient by patient basis. This will be no mean feat, especially given that, as Cancer Research UK’s Dr Nick McGranahan pointed out, research suggests that “we probably haven’t identified all the possible cancer genes yet.”

Nicky Mcgranahan discussing cancer driver genes with relevance to cancer therapy at MAP conference pic.twitter.com/emMMGuHoDQ

— Charles Swanton (@CharlesSwanton) October 23, 2015

But it’s a challenge that needs to be met, and soon: there are now hundreds of experimental drugs in trials, designed to target particular faulty driver mutations. Finding out whether these drugs work in practice will require a lot of clinical trials – and accurately and ethically recruiting patients to these trials means working out whether they’re likely to benefit in the first place.

2. Will experimental treatments actually work in a given patient?

This discussion led to a second hot topic, around the level of confidence doctors can have that a given experimental treatment will work in a given patient.

Consider the drug vemurafenib, designed to target cancers driven by a faulty version of the BRAF gene. Success in the lab led to trials showing that the drug can be highly effective in patients with advanced melanoma. It’s now routinely available on the NHS.

But what about in other cancer types? BRAF-mutant bowel cancers also seemed sensitive to the drug in the lab – but when this was tested in patients with BRAF-mutant bowel cancer, it only worked for a minority. Further research revealed the reason why (and how to overcome this) – but the story highlights that there are different levels of certainty behind the rationale for targeting different mutations.

Professor Andre suggested there needs to be an internationally agreed categorisation system, where – for a given gene mutation – doctors and patients can understand the level of evidence that a particular experimental treatment is likely to work (and thus whether this is sensible to trial).

Fabrice Andre- druggable mutations, access to cancer therapies and levels of evidence for actionable mutations pic.twitter.com/baR5e9XZ7h

— Charles Swanton (@CharlesSwanton) October 23, 2015

“What we really need is to assign different mutations a category, based on the level of evidence that they’re linked to a particular cancer,” he told the conference. This, he says, would not only help doctors make better decisions, it would also allow for better design and interpretation of clinical trials.

The take-home message was that the clinical research community needs to get a lot smarter in how it categorises mutations, and what’s known about targeting them. This will allow for better designed trials, and more transparent discussions as patients consent to take part.

3. Which tests should we use?

It’s now quicker and cheaper than ever before to plot out a cancer’s entire DNA genome. Other forms of analysis, collectively known as ‘-omics’ technologies, can look at the levels of molecules like RNA or proteins in samples of a patient’s tumour. But despite the falling price and speed of these tests, they can still result in terabytes of complex information, needing specialist, and thus time-consuming, interpretation.

In the lab, these technologies have transformed researchers’ ability to tease out cancer’s secrets. But when it comes to the real-world urgency of treating patients, there isn’t time to wait months or even years for results of these complex, careful analyses.

So the third key challenge is how to develop reliable, efficient, accurate molecular tests that are also affordable and quick enough for routine use. And this means knowing exactly what you’re measuring, and why.

Professor Elaine Mardis, of Washington University in the US, gave a whistle-stop tour of the promises and problems with ‘next generation’ DNA analysis, and the challenges of bringing it into routine clinical use – particularly in terms of the expertise for reliable interpretation.

Elaine Mardis at MAP Unicancer CRUK ESMO meeting discussing cancer actionable driver events pic.twitter.com/nYGcPS5FOk

— Charles Swanton (@CharlesSwanton) October 23, 2015

Another concept that cropped up several times was the ability to spot traces of cancer’s DNA (known as ‘circulating free DNA’, or cfDNA) in a patient’s blood, allowing a non-invasive monitoring and analysis of their cancer.

This, said Cancer Research UK’s Dr Nitzan Rosenfeld, probably won’t ever completely replace current methods – scans, biopsies etc – but it will improve the way doctors can follow how a patient’s tumour evolves over the course of treatment.

Mardis predicted the technique would enter routine use “in the next five to 10 years”. Rosenfeld was even more optimistic: “Whatever timescale you think, it’ll be quicker – that’s how fast things are moving at the moment,” he told us.

4. Fragmentation, fragmentation, fragmentation

The final thematic challenge that emerged from the conference was in making sense of the huge diversity of approaches developing around the world. And several researchers raised concerns about the sheer number of different platforms for sequencing genes.

As discussed at a lunchtime briefing session, different centres in the US are developing their own ‘panel tests’ for specific cancer-linked mutations, making standardisation and monitoring very difficult. As healthcare provision becomes more fragmented, it becomes ever harder to keep things joined up, and to monitor ‘what works’.

As Prof Jordi Rodon of the Vall d’Hebron Institute of Oncology said, to implement genomic medicine, “you need a number of groups working together, so there’s a challenge in getting everyone to pull in the same direction.”

Hope for the future

As the delegates headed back to their institutes, the overwhelming feeling the conference left us with was one of cautious optimism. Science has delivered an understanding of cancer that was almost unimaginable just a decade ago. But bringing these advances to patients routinely is proving challenging. We can see the size and shape of the prize but it remains, for many, tantalisingly out of reach.

On the one hand, we need more and better drugs: among the hundreds of patients recruited to one French gene-sequencing trial, IMPACT, just 20 or so turned out to be suitable for an existing targeted drug. The rest were offered ‘standard’ treatment.

On the other, as these drugs emerge from trials, there needs to be radical change to the way healthcare is organised and paid for, so that patients can benefit from them. Pathology labs need to adopt and properly use expensive new technology, and information needs to be carefully collected and shared across complex healthcare systems. Worrlyingly, the NHS in England and Wales lacks a nationwide molecular diagnostics service, despite Government commitments to provide one. As the new era dawns, it’s needed ever more urgently.

The most important thing is that conferences like MAP are bringing together experts to thrash out these issues – and there’s a clear sense of momentum to get them sorted. The optimism was perhaps best summed up by Cancer Research UK’s Professor Richard Marais:

“I’m very optimistic – at least now we know what we’re facing. In 1960s, we launched the War on Cancer, and tried to put man on the moon.

“But putting man on the moon was just a technical challenge – we knew where the moon was, and how to get there. With cancer, it’s taken longer – in the 60s we knew almost nothing about it. In 2015, we’ve worked out where ‘the moon’ is – now it’s a technical exercise to get us there.”

– Henry

• The next Molecular Analysis for Personalised Therapy conference will take place in London on 23rd-24th September 2016. Details will be posted on http://www.map-onco.net/