The fearsome complexity of classifying brain tumours

The human brain, and the rest of our central nervous system (CNS), is arguably the most complex and awe-inspiring biological entity that evolution has ever produced. It houses our senses, our emotions, our personalities and – some would say – our very souls.

And while cancer in general is consistently ranked as the disease the public fears the most, cancers that start in the brain are frequently listed at the top of the tree.

This is not without justification. Although things have improved in the last 40 years, brain tumour survival rates are poor compared with many other cancers.

But tumours that begin in our brain, or elsewhere in our CNS or skull (which from this point, for convenience, we’ll collectively call “brain tumours”) are a phenomenally diverse set of diseases. There are more than 130 different types, originating in the different types of cell within nerve tissue, the tissues that surround it and the cells from which it develops.

Among them are some of the fastest-growing, most aggressive and devastating forms of cancer – and some that are extremely slow-growing and less dangerous.

Indeed, as is the case with other parts of the body, people often talk about brain tumours as either ‘benign’ (non-cancerous) or ‘malignant’ (cancer).

This is an oversimplification – the simple benign/malignant dichotomy masks a more nuanced, medical classification system of four ‘grades’, which relate to how fast the tumour is growing and how likely it is to spread (with Grade I tumours being the slowest-growing or ‘low-grade’, and Grade IV the most aggressive or ‘high grade’).

But given advances in medical knowledge, experts now think that the way we classify, report and communicate information about brain tumours needs an overhaul.

In this post we’ll look at what brain tumours are, how this knowledge is evolving, and how it could help change the picture in the UK.

But first, a quick note about what we mean when we mean by “brain tumours”.

So what are brain tumours?

The term “brain tumour” generally refers to tumours that affect the tissues and cells encased by a bony box – the skull and spinal column that normally protects our central nervous system.

Any tumour arising inside this, no matter what grade it is – will, at the very least – press on our central nervous system tissues, displacing and damaging them.

But inside this ‘bony box’ there are several organs in which tumours can begin:

• The brain itself – including the brainstem and a small gland called the pineal gland
• The spinal cord – the part of the central nervous system which reaches out of the skull and down the spinal column, connecting the brain to the rest of your nervous system.

And there are also other non-nervous system tissues in the ‘bony box’ in which tumours can arise:

• These include the pituitary gland,
• a small structure called the craniopharyngeal duct,
• and the membranes that cover the brain and spinal cord (called the meninges).

The pituitary gland lies in its own little bony space at the base of the skull (with the craniopharyngeal duct nearby) and tumours in this organ are sometimes regarded along with other tumours of the hormone (endocrine) system.

But because all these tumours grow inside the skull or spinal column, they can ultimately damage the brain and/or the spinal cord – regardless of their grade.

Given the sheer variety of physical locations, tissues and cells in which “brain tumours” can arise – and their (often wildly) different behaviours – how on earth do we make sense of all this complexity, and talk sensibly and consistently about them, or assess how aggressive they are?

Appearances don’t always matter

No matter where it begins, assessing the likely aggressiveness of a patient’s brain tumour (i.e. its grade) is a crucial step in deciding how to treat it. Pathologists generally assess how a tumour is likely to behave by looking at a tissue sample down a microscope. This is no mean feat, and takes years – sometimes decades – of training.

Until relatively recently, this was the only way to grade a brain tumour. But it’s far from a perfect system, explains Professor Peter Collins, a brain tumour expert based in Cambridge.

“In the past, the terms ‘benign’ and ‘malignant’ were initially applied to brain tumours by simply comparing how brain tumours looked compared to other types of non-brain cancer,” he told us.

The advent of new, molecular and genetic techniques has started to highlight the deficiencies in this system, Collins says. “Pathologists and neuro-oncologists now consider this division, between ‘benign’ and ‘malignant’, obsolete.”

“Some ‘benign’, relatively slow-growing tumours have genetic abnormalities that can form the basis of a very malignant tumour,” he explains. “This means that they can progress to a more malignant state,” despite looking down the microscope, to all intents and purposes, ‘benign’.

On the other hand, we now know that there are slow-growing tumours which are, because of their genetic make-up, much lower risk. “These tumours simply cannot develop into a highly malignant tumour,” he says.

So how do pathologists and other health professionals keep tabs on all this complexity?

To explain, we need to take a look at the international standard system for classifying diseases, known as ICD.

Classifying diseases – why bother?

To accurately keep tabs on different diseases across the world, the World Health Organisation developed the International Classification of Diseases (ICD) system.

Classifying diseases like this has a number of benefits:

• It allows countries to track disease rates over time
• It allows comparisons between countries
• It allows governments and other organisations to allocate resources to different needs, and to adjust these as situations change

The ICD system assigns a three- or four-character code to every medical condition, from measles to melanoma. You can think of these like postcodes: each ICD code is made up of a letter followed by 2 or 3 numbers – and the more numbers there are, the more detailed the information about the disease, just like the letters and numbers in postcodes narrow down to a specific street.

So as well as saying what the disease actually is, the code allows doctors to specify different things about a disease – for example, where it is in the body, and how severe it is.

The ICD system allows tumours to be divided into four broad categories:

• malignant (cancer)
• benign (not cancer)
• in situ (cancer where the cancerous cells have stayed in their normal place in the body – this generally doesn’t apply to brain tumours)
• uncertain or unknown behaviour (where it’s not possible to foretell how the tumour will behave  – often because the tumour’s growing in a place where obtaining samples for testing would be too invasive or dangerous, so only scans have been done to diagnose the tumour).

But to make the most of this system, doctors need to be able to fully and accurately record all cases of the disease – and in the past, this hasn’t always been the case.

Improving things for patients

Improving how we classify – and thus treat – brain tumours is an urgent priority, says Dr Colin Watts, another Cambridge-based brain tumour expert, whose work is funded by Cancer Research UK.

The only way to fully understand what’s going wrong, says Watts, is “to understand the nature of the beast” – in other words, take serious steps to improve the way we collect data on, and classify, brain tumours.

“Accurate statistics relating to incidence, prognosis and treatment will help scientists, clinicians and policy makers take informed decisions about how to improve care,” says Watts.

Without this, he adds, attempts to develop and use molecular markers to classify and investigate brain tumours’ underlying biology will be compromised.

“Good results within clinical trials may be diluted by the inappropriate inclusion of patients with a type of tumour that does not respond to treatment. And prioritisation of funding may inadvertently target resources to the wrong cancer if the behaviour of that cancer cannot be properly defined and measured,” he says.

In short, says Watts, “the NHS cannot manage or treat what it cannot measure, and the accuracy of that measurement is critical.”

So what can be done?

Moving things forward

We’ve blogged several times before about the dramatic improvements in the pipeline for data collection in the NHS, and about how collecting and sharing this data improves things for patients.

One clear way that better data might improve things for brain tumour patients can be seen when you compare the UK’s survival rates with the rest of Europe.

“In the European Union, brain tumours are a significant economic burden, accounting for €5.1 billion in 2010 – that’s about €22,000 [£19,000] per patient,” he says.

“But in the UK this money does not appear to be well-spent compared to other European countries – because outcomes here are poorer than elsewhere.”

Watts points to recent research estimating that, among UK patients diagnosed with malignant brain tumours between 1995 and 1999, there were around 300 avoidable deaths during the following five years. Deaths, in other words, that would not have happened if UK patients had the same survival rates as the European average.

“And if we compare UK survival with the best in Europe – rather than the average – the estimated number of avoidable deaths rises to around 1,100 over five years.”

Asked why this might be the case, Watts is sanguine: “Wherever you are in Europe, for brain tumours, the number of treatment options is very limited compared to many other cancers. There are also far fewer possibilities to join trials.” So patients across the continent probably have access to the same treatments.

“Therefore the UK’s poorer survival rates clearly reflect something to do with the process of care, rather than access to treatments,” suggests Watts.

But exactly which steps in the process – from diagnosis to grading to treatment choice – are substandard in the UK is still a mystery. Improving the way data on brain tumours are collected would shine a spotlight on this complex problem, and identify where improvements can be made.

Painting a better picture

But at Cancer Research UK we’ve also been looking at how we can improve what we do to help make a difference. That’s why, in collaboration with colleagues at the National Cancer Registration Service (Eastern) in Cambridge, and the National Cancer Intelligence Network (NCIN), we’ve decided to change the way we publish brain tumour statistics.

For many years, we’ve collated UK statistics for dozens of different types of cancer – including tumours in the brain and other parts of the CNS – and published them on our website.

But until last year, we only published detailed statistics on malignant (cancerous) brain and other CNS tumours.

This was partly because – as a cancer charity – we’re usually concerned with cancers. But it was also because data collection on other types of brain and other CNS tumours – the benign and uncertain/unknown behaviour ones – wasn’t sufficiently complete or accurate, and it would have been misleading to use it.

So when we updated our statistics last year, we were giving the figure of around 5,000 malignant brain and other CNS tumours in 2009 in the UK (since 2009 was the most recent year for which data were available).

But this year, we’re reporting that there were around 9,150 brain, other CNS and intracranial tumours diagnosed in 2010 in the UK.

We’re able to do this now because, thankfully, information on benign and uncertain/unknown behaviour tumours is getting more reliable all the time. We’re also including pineal, pituitary and craniopharyngeal duct tumours, because their location in the skull means they can be fatal. So we’re now able publish combined data on malignant, benign, and uncertain/unknown behaviour tumours starting in the brain, other CNS and intracranial region (along with details of the precise ICD codes we’re using).

The figure of 9,150 cases is a much better reflection of the total number of people living with these tumours, and being cared for by specialist neuro-oncology teams, and of the impact of this disease on people in the UK.

What about secondary brain cancers – why don’t we include them?

Secondary brain cancers are those that start elsewhere in the body and which spread to the brain, other CNS or intracranial region.

There are several reasons why we don’t count these with the above total.

First, they are biologically very different from primary brain cancers.

Second, there aren’t yet reliable UK data on secondary brain cancers – but we’re keeping a close eye on this and when reliable figures are available, we’ll report them.

But also, including them would mean double-counting – these cancers will have already been counted in the site where they started – and if we do that for brain we’d have to do it for all the other body parts to which cancer can spread, which altogether would mean seriously overestimating the incidence of cancer as a whole.

But this doesn’t mean that we think secondary brain cancers are unimportant – we recognise that they can be as devastating as primary tumours, and we know that they affect a large number of people.

But going back to primary brain tumours – in due course, we’ll be updating our mortality statistics based on the same group of tumours we’re using for our incidence statistics. And the corresponding survival statistics will be updated too, to reflect the diversity in prognosis for the different types of tumour.

We’re keen that other organisations report statistics in line with ours, so that the data available to professionals and the public is as consistent, accurate and relevant as possible.

And we hope that this will ultimately accelerate the development of urgently-needed improvements in research and care of brain tumours – so that one day, they won’t be such a uniquely feared disease.

Henry

• Read about our progress in brain tumour research, including how we discovered the drug temozolomide
• Images via Wikimedia Commons: MRI image; oligodendroma; PET scan.