Interpret With Care: Cortical Thinning in Schizophrenia
23 November 2011. Of all the parts of the brain, researchers are seizing upon one of the skinniest to understand neural dysfunction in schizophrenia. Only a few millimeters thick, the cortical layer contains the neurons and connections that comprise much of the brain’s workforce, and appears abnormally thin in multiple magnetic resonance imaging (MRI) studies in schizophrenia. And it grows thinner with time, according to the first longitudinal study of adults with schizophrenia, led by René Kahn of University Center Utrecht in The Netherlands and published in the September issue of the Archives of General Psychiatry.
As a heritable quality (Panizzon et al., 2009) linked to cognition (Ehrlich et al., 2011), cortical thickness offers the tantalizing prospect of insight into the brain circuitry affected in schizophrenia. Though cortical thickness itself is below the resolution of typical MRI, image analysis algorithms can now infer thickness across the entire cortical sheet as it wends its way throughout the brain. Frontal and temporal regions consistently come up thinner in schizophrenia (e.g., Kuperberg et al., 2003)—so consistently that cortical thickness has recently been used as one of several measures in an attempt to automatically distinguish brain scans of individuals with schizophrenia from those of controls (Takayanagi et al., 2011).
But just what to make of the submillimeter deficits detected by these studies is unclear. On a micro-level, researchers are uncertain about the type of brain change giving rise to the thinner MRI measurements. On a macro-level, they are still debating how, if at all, thinning relates to schizophrenia. The new study by Kahn's group raises the specter of progressive tissue loss, with a rate of thinning five times that found in controls. But whether the thinning marks a worrisome driver of pathology, a harmless byproduct of mental illness, or something else entirely will be a challenge to disentangle.
“I totally accept that the MRI findings are real,” said Daniel Weinberger of the Lieber Institute for Brain Development in Baltimore, Maryland. “The only question here is whether the findings inform us about something fundamental to the brain in schizophrenia.”
At the very least, cortical thinning may be part of the wholesale reductions in cortical volume observed time and again in schizophrenia (see SRF related news story). Deconstructing cortex volume into its component parts of thickness and area may provide a more mechanistic view of the brain changes associated with schizophrenia.
“We think that cortical thickness and cortical area come online at different stages, that they might have different trajectories in time, and that they are determined genetically but in different ways,” Ingrid Agartz of the University of Oslo, Norway, told SRF. Agartz was not involved in the new study, but she and colleagues recently found a genetic association with cortical thickness (see SRF related news story), which might influence disease progression in schizophrenia.
“By separating out this thing called cortical volume into area and thickness, we can increase the accuracy of what we are really trying to find, which is an etiological and pathophysiological understanding of the disease,” Agartz said.
The longitudinal view
The challenge, however, lies in sorting out the specifics of the cortical thinning-schizophrenia link. Even just narrowing in on when cortical thinning arises eludes researchers, with cross-sectional studies comparing different groups at a single time point coming up with mixed conclusions. Some find thinning early on: the anterior cingulate cortex appears thinner in people who eventually become psychotic compared to those who don’t (Fornito et al., 2008), and a recent first-episode study found that cortical thinning was not influenced by age, gender, age of onset, duration of illness, or clinical symptoms (Crespo-Facorro et al., 2011). Another recent study finds that thinning distinguishes a subtype of schizophrenia with cognitive impairment, and so could be involved in its particular etiology (Cobia et al., 2011). These sorts of findings suggest that thinning reflects an aberration in early brain development, and contributes to schizophrenia liability.
Yet other studies find thinning comes later, either with illness onset or beyond. For example, first-episode patients imaged soon after psychosis onset exhibit significant cortical thinning compared to controls (Narr et al., 2005), and a study early this year of mainly chronic patients links thinning to illness onset (Kubota et al., 2011). Other studies invoke disease progression, with more pronounced thinning found in older people with schizophrenia (Wiegand et al., 2004) that was also independent of antipsychotic medication (Nesvåg et al., 2008).
“You have the disease itself, you have lifestyle, you have medication procedures that change over time, you have the fact that there is a clinical heterogeneity within schizophrenia, and there's also genetic variation,” Agartz said. “All these different effects are difficult to disentangle, but longitudinal studies might be able to address some of these actions,” she said.
Longitudinal studies also touch on the nature of schizophrenia itself. Does the expression of the illness unfold as the brain tries to cope with a fixed neurological insult that happened early on, or is it actively driven throughout life by some brain-damaging process? By tracking brain features in the same people over time, longitudinal studies can address whether a particular brain anomaly is fixed or changing. Though these views are sometimes pitted against each other as neurodevelopment versus neurodegeneration (Weinberger and McClure, 2002 and Mathalon et al., 2003), they are not necessarily mutually exclusive. Multiple pathological processes that together include both categories may coexist, or a single brain change may arise first early in development, but then worsen later (Pantelis et al., 2005).
Cortical thinning might fit the there-early-but-progresses-later notion: on top of the abnormally thin cortex apparent early on, and the correlation between thinning and age, longitudinal studies are starting to detect progressive thinning with time. For example, a longitudinal study of adolescents with rare, childhood-onset schizophrenia reported cortical thinning that progressed throughout the brain over five years (Thompson et al., 2001). And now, in the new longitudinal study of more typical, chronically ill patients, excessive thinning is also found over five years. While these results don’t rule out additional effects of neurodevelopmental or illness-onset brain changes, they highlight the dynamic nature of the brain in disease, and provocatively suggest that finding ways to stall thinning might improve outcomes.
“There might be multiple waves of cortical thinning throughout life—we really have no clue how these processes work,” says Neeltje van Haren, first author of the recent longitudinal study in adults. “Our study shows there is excessive thinning in the cortex in chronically ill stages, not just early in the illness.”
Using samples examined in a previous study of brain volume (van Haren et al., 2008), van Haren and colleagues compared MRI scans from 96 individuals with schizophrenia and 113 healthy controls. At baseline, the average cortical thickness across the entire brain was about 3 mm and did not differ between the two groups, but regionally, the schizophrenia group exhibited a thinner cortex in the frontal and temporal lobes. Five years later, the schizophrenia group lost an average of 0.05 mm of cortex thickness, while healthy controls only lost 0.01 mm. Even more profound thinning ranging from 0.05 to 0.19 mm could be detected within the frontal and temporal regions.
These hair-thin decrements in cortical thickness were associated with medication intake and functional outcome. A higher amount of typical antipsychotic use during the five years between scans was associated with larger declines in cortical thickness in several brain regions compared to atypical antipsychotics. Though the strength of the effect for atypicals versus typicals varies across studies, the finding fits with antipsychotic-induced changes in brain structure found for brain volume (Ho et al., 2011) and in animal studies (e.g., Vernon et al., 2010).
Van Haren noted that the influence of antipsychotics on brain structure is a hot topic these days (see, e.g., recent comment by Lieberman), but antipsychotics didn’t explain everything in her study. Extensive thinning in the left superior temporal cortex was linked to poor functional outcome, as assessed by a combination of functional and symptomatic measures, in a way that was independent of medication. This dissociation helps rule out the possibility that the most ill had most thinning because they took the most medication.
But distinguishing cause from consequence is tricky. Does thinning drive poor outcome, or does it reflect consequences of a poor outcome, with psychosis or other aspects of the disease literally sculpting the brain? Though current circumstantial evidence for early thinning suggests thinning comes first and drives illness, van Haren said it is far from definitive.
“We cannot give direction to the associations that we find. Based on these kinds of data, it could go either way,” van Haren said. In an ideal world, population-based, longitudinal studies involving frequent brain scans and comprehensive phenotypical data might capture whether this sort of brain change comes before or after illness.
“We really have to scan people before any problems are visible,” she said. “Otherwise I don't see how you can ever distinguish between the chicken and the egg.”
Less intact or artifact?
Others worry that cortical thinning is not related to disease at all, but is merely an epiphenomenon reflecting the different lifestyles of mentally ill people. Weinberger emphasizes that lifestyle variables such as exercise, smoking, and weight gain can push around MRI measures of cortex.
“There are so many things that change the dimensions of these measurements on an MRI scan that are about normal, everyday living characteristics that it's very hard to know what changes in a group of patients with schizophrenia followed for five years represent,” Weinberger told SRF.
Though smoking, and alcohol and drug use were ruled out in van Haren’s study, she said other variables like exercise, cannabis use (Rais et al., 2010), and IQ were being examined.
“I do agree that this is going to be one of the main issues in the next few years, and we haven't even begun to look at how the different confounders interact,” van Haren said. “But I would be surprised if cortical thinning could all be accounted for by lifestyle.”
Weinberger also questions whether MRI results reflect a true structural change in the brain, noting a lack of a postmortem correlate for tissue loss. Without such a correlate, lifestyle-induced physiological variations in the brain, such as changes in blood perfusion, fat content, and water relaxation times, may be more important. “These have nothing to do with the brain becoming less intact,” Weinberger said.
Others point out that MRI-detected thinning could reflect a decrease in the connections made between neurons. Though widespread neuron die-off characteristic of neurodegenerative disorders like Alzheimer's disease has never been detected in postmortem studies of schizophrenia, decreases in the neuropil—the thicket of dendritic trees and connecting axonal branches—have been reported (Selemon et al., 1998; Fornito et al., 2009), and could conceivably contribute to the MRI-detected thinning.
Though neuropil reductions did not always result in a thinner cortex, it may be that MRI studies are outpacing the postmortem ones. Typically conducted on small samples in discrete cortical regions, postmortem studies probably do not have the power to detect the cortical thinning found in MRI studies that look at the entire brain in many more individuals, said Karoly Mirnics of Vanderbilt University in Nashville, Tennessee, who was not involved in the new longitudinal study.
“I believe that the shrinkage is small, but real,” Mirnics told SRF. “But whether it is of clinical consequence is unclear.”
It would be optimal to examine postmortem brain samples from people who participated in MRI studies, Agartz said. “Ideally, one would be able to study someone from the cradle to the grave.”
Sorting out the clinical relevance, if any, of these subtle changes in cortical thickness will be helped by clinical measures sensitive enough to track changes in a person’s symptoms and function over time. Is progressive cortical thinning always associated with clinical deterioration? What happens in those who remain relatively stable?
Longitudinal studies might skim over this issue, because study participants who are available for multiple scanning sessions over long periods of time tend to be clinically stable. To address this, Agartz cites the need to standardize MRI protocols as a step toward sharing data. “Sometimes you wonder if your results are methods-based or cohort-based, so this kind of exchange could be of value,” she said.
Scanning individuals who do not have schizophrenia may also help fill in the meaning of cortical thinning. Last year, Agartz and colleagues found a partially overlapping pattern of thinning between the bipolar disorder subtype that comes with mania and schizophrenia (Rimol et al., 2010). The thinning was not related to medication, and could reflect shared aspects of the two disorders in terms of genetic liability or symptoms. Clues to the genetic influence on cortical thinning also come from studies including healthy relatives, with unaffected offspring of individuals with schizophrenia showing decreased cortical thickness (Bhojraj et al., 2011), but only trends for thinning in unaffected siblings (Goldman et al., 2009). Earlier this year, a single nucleotide polymorphism on 15q12 was associated with cortical thickness, but not schizophrenia (Bakken et al., 2011). These findings suggest that cortical thickness interacts with other risk factors to influence the liability or disease course of schizophrenia.
Despite the evidence for genetic influence on cortical thickness, understanding just how malleable it is may give rise to new treatment strategies. If thinning reflects a true structural change in the brain related to disease, is it permanent or reversible? If reversible, what drives it? Does cortical thickness fluctuate as a result of experience-dependent plasticity, in which aspects of the disease, like psychosis, leave their mark on the brain? Or does thinning reflect impaired plasticity, with the brain unable to respond appropriately to environmental inputs—a notion in line with the underactive glutamate system hypothesized for schizophrenia (see SRF hypothesis)? If so, finding ways to restore plasticity, or normalize cortical thickness, might improve outcomes.
“How plastic is the brain? Can we reverse some of this thinning by affecting, for instance, synaptic function?” Agartz asked. “There need to be more studies.”—Michele Solis.
van Haren NE, Schnack HG, Cahn W, van den Heuvel MP, Lepage C, Collins L, Evans AC, Hulshoff Pol HE, Kahn RS. Changes in cortical thickness during the course of illness in schizophrenia. Arch Gen Psychiatry. 2011 Sep;68(9):871-80. Abstract