Brain Imaging Struggles for Psychiatric Respect

TALK VERSUS DRUGS

OCD is a disorder in which a person is relentlessly hounded by intense feelings that something is amiss – germ-laden hands, for example – and must be rectified immediately. During the 1990s, PET work by Baxter and colleagues at the University of California, Los Angeles, established that OCD is marked by elevated metabolism in the orbital frontal cortex and right caudate nucleus. Studies also demonstrated that behavioral therapy and SRI drugs can lessen these abnormalities, and correlations were found between treatment responses and different patterns of brain metabolism.

OCD is a relatively well-defined, homogeneous condition. Major depressive disorder, in contrast, is considered a group of illnesses. Not surprisingly, PET images reveal a bewildering array of brain abnormalities. Metabolic signals from imaging, nevertheless, are "pretty well behaved," says Helen S. Mayberg, a psychiatry and neurology professor at Emory University School of Medicine in Atlanta. "They change with treatment, and they change differently when people get better than when they don't."

Courtesy of Helen S. Mayberg

These brain scans, made by positron emission tomography, show post-treatment increases (orange) and decreases (blue) in regional glucose metabolism. The subjects were depressed patients who responded to cognitive behavior therapy (CBT) or paroxetine, a serotonin-reuptake-inhibitor drug. CBT resulted in metabolic decreases in frontal and parietal cortex, and increases in hippocampus; paroxetine had the opposite effects. (From K. Goldapple et al., Arch Gen Psychiatry, 61:34–41, 2004.)

Mayberg and colleagues at the University of Toronto recently published a PET paper detailing how cognitive-behavior therapy (CBT) affected depressed patients. Lasting several months, CBT uses homework assignments and guided practice to identify relationships between a patient's thoughts, feelings, and behaviors. The researchers found that when 14 subjects responded to CBT, glucose metabolism decreased in certain cortical regions and increased in the hippocampus.¹ In an earlier study of 13 other depressed subjects, the SRI drug paroxetine (Paxil) had the opposite effects on those same areas. Several cortical areas affected by CBT, but not by the drug, involve "self-actualization, self-reference, and reappraisal of information," notes Mayberg.

In 2001, a team of scientists, including Baxter, compared paroxetine treatment of depression with interpersonal therapy (IPT), a conversational approach that is less structured than CBT and deals with themes including social deficits, role disputes and transitions, and grief. PET scans showed metabolic dissimilarities between the 10 subjects who chose to take medication and the 14 who opted for IPT,² but these variations were not identical to those detected by Mayberg. "My suspicion would be that, if not different patient populations, it's the different treatments," Baxter muses about the divergent findings. " [IPT and CBT] really do focus on different aspects of depression."

COST-EFFECTIVE APPLICATIONS

Though these studies suggest potential brain markers for effective treatment, imaging's clinical applicability still elicits much skepticism from psychiatrists. One qualm stems from a PET scan's price tag of several thousand dollars. "Suppose you found a pattern of cerebral metabolism that indicated that you really did have a better chance with medication than psychotherapy, or vice-versa," says Michael E. Thase, a psychiatry professor at the University of Pittsburgh School of Medicine. "The cost of that test might actually be more than the cost of treatment." (Psychoanalysis, which can last many years, is not a standard therapy for depression.)

Thase acknowledges, however, that brain imaging might be cost-effective when the treatment options include electroconvulsive therapy (ECT) or psychosurgery. Harold Sackeim, a psychiatry and radiology professor at Columbia University College of Physicians and Surgeons in New York, is now using PET to track cerebral blood flow at patients' second ECT session. His goal is to be able to predict who will eventually benefit from shock treatment.

Massachusetts General Hospital is among a handful of US medical centers performing anterior cingulotomies, operations in which neurosurgeons lesion a cortical area in patients with intractable depression and OCD. But only about half of patients who undergo this operation respond positively. Using PET, Rauch and his colleagues discovered that responders with depression have higher presurgical metabolism in the left subgenual prefrontal cortex and left thalamus.³ Another study found that presurgical metabolism in OCD responders is elevated in a part of the right posterior cingulate cortex.⁴ This area connects to all nodes of the abnormal circuit that is OCD's hallmark.

Rauch hopes to replicate these small studies, but patient accrual is slow because surgeons at Mass General perform only about a dozen cingulotomies each year. "Obviously, as the data set gets larger and the findings persist, then confidence grows" about using PET to decide for or against surgery, Rauch says. Meanwhile, he has begun to collect imaging data on deep-brain stimulation, a newer surgical approach to treating depression and OCD.

What especially concerns Rauch is data reproducibility, which can be elusive when the focus is a heterogeneous syndrome like depression. Wayne C. Drevets, a psychiatrist and senior investigator at the National Institute of Mental Health (NIMH), finds that he can boost reproducibility of PET results by scanning only patients with recurrent or early-onset depression or a family history of the disease. He and his NIMH boss, Robert B. Innis, are starting to image an even more exclusive sample: people with genetic polymorphisms linked to depression. The downside is that Drevets cannot examine the effects of talk-based treatments. "People who have recurrent major depressive episodes that meet our inclusion criteria typically don't respond to psychotherapy," he explains.

NON-PET PROJECTS

No research strategy, of course, applies universally; consider PET, for example. David R. Rosenberg, a professor of child psychiatry at Wayne State University School of Medicine in Detroit, specializes in OCD and depression in children, whose developing nervous systems cannot be exposed to even the scant radioactivity of a PET scan. Instead, his team uses an MRI-based technology called proton magnetic resonance spectroscopy. It allows scientists to monitor changes in both anatomy and neurochemistry.

Courtesy of Lewis R. Baxter Jr.

Positron emission tomography reveals how abnormal patterns of glucose metabolism (blue is low; red is high) in two brain regions decrease after a patient with obsessive-compulsive disorder is treated with paroxetine, a serotonin-reuptake-inhibitor drug. For unclear reasons, the head of the right striatum's caudate nucleus often shows a much greater decrease than the comparable area in the left striatum. (Scans from a patient reported in S. Saxena et al., Arch Gen Psychiatry, 59:250–61, 2002.)

Rosenberg is launching two imaging studies that will compare medication and CBT treatment of 100 children with OCD and 100 with depression. Similarly sized control groups, which will not receive sham treatments, should signal whether any brain-pattern changes observed in patients "are different from the changes that a healthy adolescent would just have over time," he says.

Rosenberg's team has also initiated pilot fMRI projects. "Our hope, on these higher-power MRI machines, which can do things faster, is that we can do all three studies – the functional, the chemical, and the structural – in the same child in the same scanning session," he states. "But we're not quite there yet."

Other investigators also mention current or prospective fMRI projects, but few papers have been published describing the fMRI effects of psychiatric treatments. Unlike PET, fMRI requires subjects to perform some task during the scan, so protocols typically require patients to look at emotion-provoking stimuli. Brain imaging then reveals blood-flow responses in depressed subjects before and after treatment, or compared to nondepressed controls.⁵

Magnetic-resonance research, however, could soon become far more sophisticated. Last September the University of Illinois at Chicago (UIC) inaugurated what it describes as the world's most powerful MRI machine for examining humans. Containing a 50-ton, 9.4 Tesla magnet, the device will facilitate psychiatric studies, predicts Keith R. Thulborn, a professor of radiology, physiology, and biophysics at UIC. (State-of-the-art clinical scanners currently use 3 Tesla magnets.)

"Where brain dysfunction happens is at the level of the biochemistry that supports the function," Thulborn explains. The new machine should allow biochemists to "move beyond looking at just a water or fat signal, which is what conventional MRI is based on." More specifically, scientists should be able to track the fate of injections of two rare nonradioactive isotopes: carbon-13, ultimately incorporated into carbon dioxide during glucose metabolism, and oxygen-17, incorporated into water during oxidative phosphorylation. The machine should also be able to measure sodium and phosphorus concentrations, which can indicate metabolism and tissue death.

SIMPLISTIC INTERPRETATION?

PET technology is also advancing, as more radioligands are developed and spatial resolution shrinks to about four millimeters. But as imager and reagents improve, doubts persist as to what exactly they reveal. Do changes in brain scans during psychiatric treatment show the abatement of symptoms or the elimination of a disorder's causes?

"A lot of the interpretation of that [imaging] work, in my own view, has been somewhat simplistic," asserts Sackeim. The field's biggest problem, he continues, "is that it's largely correlational." To experimentally establish cause-and-effect relationships, Sackeim envisions curing psychiatric disorders by engineering particular brain patterns, not by drugs or talk therapy, but by focal techniques such as transcranial magnetic stimulation.

Mayberg contends that psychiatric brain imaging has moved slightly beyond the correlational stage. Noting that "we have markers that tell us that people who get better on therapy have brain-network states that are different from people who do well on drug," she describes the next generation of studies as randomizing patients to drug, psychotherapy, or placebo, and then seeing "whether or not you properly predicted outcome."

Nevertheless, no imaging study is apparently imminent that will tackle the link between antidepressant drugs and suicide. One difficulty is disentangling preexisting suicidal tendencies from those triggered by drugs. Suicidal patients usually receive medications whose effects on brain scans would swamp the effects of any endogenous abnormality, remarks Drevets. To skirt this confounding factor, he and David A. Brent, at the University of Pittsburgh, are conducting a PET study of unmedicated depressed people. Some of the subjects formerly considered killing themselves; others were never seriously suicidal. Drevets expects that this pilot project, which measures cerebral blood flow and serotonin-receptor concentrations, will be finished in about a year.