Hopkins Medical News
COVER STORY Annals Profile Research

MRI scan of a human brain

Gathered in a luminous basement, some of the world’s most creative MRI specialists can now view a stroke under way. In the process, they're redefining treatment.

As the stocking feet of the elderly patient protrude from the base of the great MRI machine, MRI imagea picture of the man's brain flashes onto the monitor. Within minutes, Norm Beauchamp is using the technique known as diffusion imaging to pinpoint dead tissue and find out if this man has had a stroke.

By reviewing the perfusion imaging, Peter van Zijl learns that no remaining tissue is at risk... The stroke is already irreversible. and because the patient lingered too long MRI scan
before coming to the hospital, this man will not be a candidate for a new clot-dissolving drug.

he darkened room adjoining the basement of the Victorian Houck building has an enclosed, submarine feel. What light there is comes mostly from three video monitors and illumined control panels. It reflects on the faces of the 10 men and women working in the room—a young crew, like something from a Navy recruitment poster. They confer and tap instructions on the panels. A constant beep sounds like sonar. An image flashes onto one monitor: in lovely detail, a cross-section of a patient’s brain. Then, abruptly, the terrain changes. It’s less detailed, mostly gray. But as deeper slices of the brain go by, an undeniable dark splotch appears in a suspect area.

What’s becoming clear to those around the console is that the 76-year-old man lying in the hull of the machine is in the midst of a stroke. A clot blocks an artery in his brain, and the darkened region—the area the blood vessel normally serves—testifies silently: This tissue has begun to die.

The Software Makes the Difference

elcome to MRI at Hopkins, where last year’s fantasy becomes radiological reality. Here, the most advanced technology in the field is allowing these intense, young radiologists to observe with unprecedented clarity what’s happening inside a patient’s head. And what they learn from their images of his tissues could alter his life as never before. If the pictures show this elderly man is a good candidate for a recently approved clot-dissolving drug and he’s treated quickly, he could well be spared loss of language and mobility once his blood resumes its normal flow to the brain.

MRI, or magnetic resonance imaging, has come a long way since scientists developed the basic concept some 25 years ago as a sort of X-ray for soft tissues. Complex and expensive, the technology is dependent on huge magnets that pull the body’s hydrogen protons into alignment—like the needles on so many compasses. When the protons slip back into magnetized order, the energy they release translates from a signal into an image on a monitor. Because the strength of the signal can vary depending on the tug of nearby atoms, tissues with densely packed atoms, like fat, show up differently from less dense ones like bone.

When MRI first moved into the clinic, in the early ’80s, physicians had never seen anything like it. There, on the screen, were high-resolution images obtained noninvasively and without the use of potentially harmful radiation. Today, there’s hardly a health center around that doesn’t offer patients access to MRI. And in fact, Hopkins’ basic machine, which is providing such precise insight into this aging stroke patient’s brain, is the same gargantuan, $2 million magnetic resonance imaging device you might find in a community hospital. What’s different about the Hopkins setup lies in the innovative software driving the scanner that’s dreamed up and tested here.

“MRI mutates with the software on the average of every year and a half,” explains David Bluemke, M.D., Ph.D., the radiologist who heads Hopkins’ clinical MR program. And thanks to the Department of Radiology’s dogged recruitment of some of the world’s most creative MR radiologists, physicists and engineers, the software now taking shape in this subterranean place is putting new spins on familiar uses of the machine.

New studies push the limits of the innovative software every day. One now under way, for example, compares mammograms with images obtained by using a new MRI-detectable contrast medium as a way to screen for breast cancer in women with denser-than-usual breasts. The technique also can reveal malignancies at the margins of lumpectomy surgery. Another trial explores MRI’s ability to distinguish normal and abnormal lymph nodes in patients with kidney cancer and to pinpoint metastatic liver masses in patients with colon cancer. Ultimately, advances like these could eliminate unnecessary exploratory surgery.

At this moment, however, the team around the monitor is working on MRI techniques that could prove to have the most far-reaching implications: observing a stroke in progress and determining on the spot the extent of injured tissue.

Watching Dying Brain Tissue

For this elderly man whose stocking feet now protrude from the base of the great machine, yesterday had begun normally. But shortly after breakfast, he felt lightheaded and puzzled by a sudden numbing weakness in one leg. It wasn’t the first time he’d experienced this combination of unsettling symptoms. Two years before, these same telltale signs had signaled the onset of a stroke. He waited most of the day to see whether the sensation would pass. When it didn’t, he called an ambulance and headed for Hopkins.

“When someone comes to you with what looks like a stroke, like brain tissue is dying,” says neuroradiologist Norman Beauchamp Jr., M.D., “you have a multitude of questions to answer quickly. Before you can do anything, you first want to weed out diagnoses like epilepsy or blood infection that can prompt the same symptoms, and you want to make sure you’re not seeing a person who’s experienced the beginnings of a stroke that cleared itself, leaving a temporarily stunned brain.” If all signs still appear to show a stroke in progress, the physician’s next challenge, Beauchamp explains, is to find out quickly whether it is, in fact, ischemic—from a blocked brain vessel—or less commonly, from a burst blood vessel. If it’s ischemic, then you want to know why, exactly. Is a clot obstructing a vessel? Or has the clot dissolved, but left irreversibly injured brain tissue? Treatment will depend on that diagnosis. The second challenge will be to determine how much of the adjoining area is at risk of irreparable damage.

These questions could be answered conventionally, says Beauchamp, a Hopkins-trained neuroradiologist who came on the faculty in 1995 to focus on the clinical end of the booming technology. Blood tests, CT scans, angiography, ultrasound and nuclear tracers will do the trick. But while some things, like blood work or spotting a brain hemorrhageBehind the software with CT scanning, go fast, others are maddeningly slow. It’s not unusual to wait six to 12 hours before a CT scan can pick up an ischemic stroke unequivocally. “And that’s not acceptable anymore,” Beauchamp says, “if you want to be able to use tPA beyond three hours after the incident.”

tPA stands for tissue plasminogen activator. Last year, the FDA approved this clot-dissolving drug, long used to treat heart attacks, as a therapy for ischemic strokes. But there’s a caveat. For use in the brain, tPA comes with a time limit: It’s safe only within three hours of a stroke’s onset. Injecting the drug beyond that time, when some brain tissue has died, significantly raises the risk of brain hemorrhage. And because it’s so difficult to tell quickly whether brain tissue is dead or merely stunned, clinicians err on the safe side. They scrupulously restrict tPA to patients who arrive at a hospital within the window. Today, perhaps only 4 percent of those with an ischemic stroke get tPA.

This is tremendously frustrating for Beauchamp, an impassioned and articulate one-man chamber of commerce for MRI. “I’d like to see that number change,” he says, “and extend tPA’s therapeutic window. To do that, we need to know with certainty how much of a patient’s brain tissue is still functioning. And,” he drops his voice to a hush, “that’s where MRI is so valuable.” MRI, he says, can answer most stroke questions in less time than existing methods, and can do it noninvasively—and at one sitting. “Otherwise,” he shrugs, “what’re you going to do? Ferry an acutely ill patient from place to place for all these tests?”

Each Patient's Revealing Signature

What Beauchamp, physicist Peter C.M. van Zijl, Ph.D., from the Division of MRI Research, and a team of others have done is unique. They’ve pioneered a single-session, multiple-MRI scan technique for stroke patients. That’s like developing a long-armed machine to pick out selected chocolates from a neighbor’s Whitman’s Sampler. You program it to pick out the jellies, then the truffles, etc. And when you reassemble them on your own side of the fence, you’ve got a box tailored to your needs. The Hopkins researchers can program the MR scanner to home in, in sequence, on four or five different properties of a person’s brain tissue, each of which can vary. Then they combine the results to form a patient’s revealing “signature.”

One type of MR scan, for instance, picks up hemosiderin, an iron-based product of deteriorating blood. If the substance accumulates in the brain, it’s likely the patient has suffered a hemorrhage. Another technique, an MR version of angiography, looks immediately for blockage in large brain vessels. A third, called a T2 scan, shows if brain tissue is dead.

But it’s with a new type of scan, known as diffusion weighted imaging, that the Hopkins team is making its mark internationally. For years, physicians who treat stroke have recognized the value in tracking, minute-by-minute, how water moves around brain cells. Flow out of a cell is a sign the cell has died and deteriorated. Flow in means the cell membrane is injured but potentially alive. Diffusion imaging in combination with a blood flow scan called perfusion imaging provides the potential of telling at once which type is salvageable and which is not. Physicians should then know how to tailor treatment.

Peter Barker looking for molecules that will highlight metabolic changes in the patient's brain

MR experts were split, however, over exactly how to carry out diffusion imaging—whether they should study the motion of water over the brain in a specific direction and how exactly they should interpret the different scans. All the while, patients whose lives might have been changed with thrombolytic drugs were slipping away.

Enter van Zijl, a man caught up in the idea of diffusion scanning from the start of his career at the National Institutes of Health (which continues to fund his work) until he came to Hopkins as a high-powered recruit in late 1992. Secure in his mathematics, the Dutch-born professor of radiology and biophysics took on the diffusion controversy. Over time, his equations showed in no uncertain terms that images should include readings in at least three directions, simultaneously. Then, with Beauchamp and a group of others, van Zijl demonstrated that multidimensional diffusion images—added to the signature—produced the most accurate picture of a stroke under way.

This is Hot Stuff

ow, as word of the turnaround spreads worldwide, the prospect of treating certain strokes is changing. “This is hot stuff,” attests neurologist Robert Wityk, M.D., who heads Hopkins’ acute stroke team, a recently formed group aiming to establish for strokes the kind of quick-treatment protocols that exist for heart attack.

For today’s patient—this aging man who’s already had one stroke—who, in his 45-minutes under the scanner is having nearly 300 images taken of his brain, the improved technology will clearly show the difference between dead tissue and tissue that may be saved.

“What we hope the technology will allow us to do ultimately,” Wityk says, “is to know what state the brain is in at the time we see the patient. If there are nerves alive, aggressive treatment might be able to save functions like speech and mobility. If the stroke has completed its course, we’d know to take a more conservative approach.”

Peter van Zijl reviewing perfusion imageNow, add to the armamentarium yet another technique: MR spectroscopy. It’s an elite advance, evolving at Hopkins and just a handful of medical centers across the country. Like other MR techniques, it relies on a sensitivity to hydrogen atoms. The difference is that this time the hydrogen isn’t in water; it’s in molecules that fluctuate with the disease process. Understanding pathology through this indirect method is like learning someone has died by coming across the funeral bill. A few patient studies show, however, that the technique holds promise for revealing areas of brain-cell injury or death in suspected strokes. The next step in stroke treatment could lie here.

Hopkins’ Peter Barker, Ph.D., a world leader in MR spectroscopy, and his team of researchers and clinicians have found that they can pinpoint several molecules that highlight metabolic changes in stroke. Barker’s spectroscopy software senses, for example, reductions in N-acetyl aspartate (NAA), a molecule that plummets in dead or dying nerve cells. Alternately, he can detect a rise in choline, a cell membrane molecule released as cells deteriorate. And a buildup of the metabolic by-product lactate gives a quick idea of whether cells are functioning or at death’s door. Like Beauchamp’s signature studies, spectroscopy may soon guide tPA use and help test a new class of stroke drugs known as cytoprotective agents.

But for the moment, as important as cutting-edge technology, Beauchamp notes, will be getting patients to the hospital fast so the high-tech advances can help while the tPA clock is still ticking. As the team completes its review of the scans of the elderly man, Beauchamp points to the lacunar area of the brain, a mass of “little doo-dad vessels in the middle of the brain” that supply key motor areas. On the T2 scan: “That bright spot there verifies dead tissue.” Tap, tap, the screen splits, now adding the man’s diffusion imaging of the same area. But now there’s the dark spot: “That means water’s not moving—probably edema.” Add a perfusion scan, which comes out bright in the same lacunar area, registering no blood flow: “All this means he definitely had a stroke,” says Beauchamp, “and it’s a new one.” But this patient—who lingered all day with his symptoms and had a large area of dead tissue—will not be a candidate for tPA.

A year or two from now, Beauchamp speculates, that should change—provided that patients learn to recognize signs of a stroke and get to the hospital right away. Improvements in the procedure, he thinks, will whittle scan time to a half-hour. And by then, the Hopkins team also should have completed the critical research that shows how effectively MRI signatures can advise on use of tPA. Once that's done, says Beauchamp, MR scanning could become routine for strokes. And that could open up the drug to thousands of patients who otherwise would reach the emergency room too late.