Using optical coherence tomography for imaging brain tumours could make neurosurgery safer and more effective, improving patient survival rates, according to new research carried out at Johns Hopkins University. Elly Earls meets Dr Alfredo Quinones-Hinojosa, the clinical leader of the research team behind the discovery, to find out more.

Applying imaging technology usually used for ophthalmology to neurosurgery could allow surgeons to distinguish between cancer and non-cancerous surrounding tissue during surgery with much higher sensitivity and specificity than they can now using MRI or ultrasound. And not only does optical coherence tomography (OCT) perform better than the current methods used to image the brain, as it can pick out minute, real-time details – OCT is to ultrasound or MRI what zoomed-in Google Earth is to zoomed-out Google Maps when it comes to the brain – it’s also cheaper and safer.

After more than half a decade of research, scientists at Johns Hopkins University in Baltimore have managed to decipher how OCT, a technique developed in the 1990s for imaging the retina that produces high-resolution images using light waves without delivering ionising radiation to patients, can be applied to brain surgery, a breakthrough that could eventually enable surgeons of varying levels of expertise to know exactly where to cut.

Currently, even the world’s most accomplished surgeons find it incredibly challenging to get brain surgery exactly right, too often removing either too little of a tumour or too much healthy brain tissue.

"The more cancer you take, the longer a patient survives; that’s been done over and over, and my own work over the past ten years has shown this," says Dr Alfredo Quinones-Hinojosa, a professor of neurosurgery, neuroscience and oncology at the Johns Hopkins University School of Medicine, and the clinical leader of the research team.

With our device, hopefully we will equilibrate the field in such a way that any surgeon around the world will have a machine they can use to rapidly distinguish between tumours and healthy tissue. And it will be affordable.

"However, you also have to make sure you take out real cancer and not normal, functioning brain, because when you take normal brain, those patients actually do even worse. You walk a fine line."

Even Quinones-Hinojosa, a world expert in neurosurgery, doesn’t get it right every time. "When I close a case, I get a post-operative MRI and, in about 20% of those patients, you see cancer, obvious cancer, left behind. And that’s even with someone like me, who’s considered a world expert in taking these tumours out. Imagine people with less expertise; they will most likely leave even more cancer behind."

Potentially not for much longer, though. "What we’re trying to do is very simple: we want to equilibrate the tools that surgeons can use, because the expertise that I, for example, have developed over the years is not the same expertise as a surgeon in the middle of the US or in other countries around the world," Quinones-Hinojosa explains.

"With our device, hopefully we will equilibrate the field in such a way that any surgeon around the world will have a machine they can use to rapidly distinguish between tumours and healthy tissue. And it will be affordable."

Code crackers
So where did the idea of using OCT for brain imaging originally come from? As Quinones-Hinojosa recalls, it all started when an MD-PhD student, Carmen Kut, rotating through his lab, which is one of the few neurosurgeon-led National Institutes of Health federally funded labs in the country, overheard him talking about the difficulties of distinguishing between cancer and normal tissue during surgery.

Kut then spent time in the lab of another world-renowned scientist, Xingde Li, a professor of biomedical engineering, who was leading a research group working to develop OCT technology for various clinical applications beyond ophthalmology, who, at the time, had successfully translated the technique into use for orthopaedic and other procedures.

With guidance from both mentors, she essentially put two and two together. "Why couldn’t we use that light on brain cancer and see if we could crack the code?" she wondered. And that’s exactly what the team set about doing around six years ago.

"The idea came from exploring and hearing people talking and discussing. And like everything, when you cross-fertilise different fields, the best things come up," Quinones-Hinojosa notes, adding that OCT basically works like ultrasound, but with light instead of sound, and that because light travels much faster than sound, OCT is able to resolve small but critical features of brain cancer.

"When you do an ultrasound, you use sound, the sound bounces back and a computer device converts that sound into imaging. With OCT, it’s the same principle but with light; you send light in, the light bounces back, and then we figured out a way to create computer algorithms that use that information to tell you what is cancer and what is normal brain. OCT is already used for other disciplines, for example, to look at the retina, but we cracked the code for brain cancer."

Once the team had optimised what’s called the ‘optical property mapping algorithm’, the system could generate a colour-coded tissue map almost instantly: red indicating cancer and green for healthy tissue, while yellow signifies the middle ground or ‘penumbra’, when there is some cancer and some healthy tissue.

It sounds pretty simple, but the process of getting there certainly wasn’t. "It’s taken five years to crack the incredibly complex code of OCT specifically for brain cancer," Quinones-Hinojosa says. "You see movies where people are trying to crack codes and I always identify with them because the first three years, we thought we would never crack it."

Right on point
The breakthrough came when the team started looking at ‘intermediate grades’ of cancer, as well as the most aggressive ones. "Once we started collecting that information, we realised that brain cancer is highly cellular and it disrupts the myelin that exists in the human brain," Quinones-Hinojosa explains.

Prior to this, the team had been building their technique on the principle that cancerous tissue is denser than non-cancerous tissue – yet density simply didn’t have a big enough effect on OCT readings to be a feasible measure.

When it came to myelin, though, OCT can tell the difference between the healthy cells, which are coated with myelin sheaths, and the cancerous ones, which aren’t, to an extraordinary level of accuracy.

"It’s unbelievable," Quinones-Hinojosa grins. "My ability to tell what is cancer and what isn’t cancer at the edge of a tumour is about 50%; I can flip a coin. But if you ask the machine, the sensitivity and specificity is 80-100%. It tells us where to cut with an incredible level of sensitivity and specificity."

In fact, the only comparison Quinones-Hinojosa feels really does the technique justice is the difference between zoomed-out Google Maps and zoomed-in Google Earth. "If you look at the continent from way up in the distance, you can see the water and you can see the earth. That’s what you can see with an MRI or ultrasound," he elaborates. "But, if you want to see my house and what the backyard and the roof looks like, then you begin to zoom in with Google Earth. That’s exactly what you get to do with OCT.
"Actually, in some ways, OCT is even better than a zoomed-in Google Earth map, since it can also detect real-time changes [equivalent to seeing live car traffic and pedestrians walking].

"You can zoom in to those levels of really, really defining the anatomy at a fine level of what is cancer and what is not cancer. It’s so good that you can see fingers of cancer penetrate into the normal brain and detect the presence of blood flow."

On top of being accurate, OCT is also affordable – compared with several million dollars for an MRI scanner, the researchers predict that an OCT machine would be in the hundreds of thousands field; offers a continuous picture of how surgery is progressing, whereas MRI scanners require an extra hour of operating room time to obtain a single image; and delivers no ionising radiation to patients. Moreover, the technology collects images without touching the brain, therefore reducing infection concerns, and does not require the use of injected contrast agents.

Next steps
So far, the research team has tested the technology on fresh human brain tissue removed during surgeries on around 50 patients as well as in surgery to remove brain tumours from live mice, and the next step is a human clinical trial in November.

"We’re about to launch a clinical trial in Mexico that has been approved by the review board and where we will collect the first human data. Hopefully, early next year, we will be collecting our first human data here in the US," Quinones-Hinojosa notes. "Then, in the next three to five years, we hope to be conducting phase-II and III trials in this technology."

Later, if it is proven to work for brain cancer, the team also hopes to be able to adapt it to work for other cancers in the brain, such as lung and breast metastasis, or in other parts of the body.

"We started with brain cancer because we felt it was the most devastating cancer, but we’re also planning to use this for other types of metastatic cancers, too. There are so many other cancers that you can use it for," Quinones-Hinojosa says.

In the meantime, the potential for improving survival rates for patients with varying levels of brain cancer is significant; the paper published by Quinones-Hinojosa and his team in Science Translational Medicine showed 100% specificity and 80% sensitivity for diagnosing high-grade brain cancer, and 80% specificity and 100% sensitivity for low-grade.

"My ability with my own hands or the tools we have currently to take out these kinds of cancer is suboptimal compared with how good this new technology is," Quinones-Hinojosa concludes.