The Enigma of Neuroimaging in ADHD

In this issue, results from the attention deficit hyperactivity disorder (ADHD) section of the ENIGMA (Enhanced Neuroimaging Genetics Through Meta-Analysis) consortium are reported for cortical measures (1), complementing an earlier report of intracranial volume and subcortical structures (2). Combining data from 36 sites, these reports represent the largest-sample ADHD neuroimaging studies to date (2,246 individuals with ADHD and 1,934 control subjects).

The ENIGMA consortium and related efforts are part of the broader quest for biomarkers (measurable indicators of some biological state or condition) in neuropsychiatric disorders. Motivations to pursue biomarkers may include deepening our understanding of the molecular, anatomical, and physiological mechanisms of mental illnesses; identifying risk and protective factors; guiding development of interventions and monitoring their progress; and sometimes reifying the biological basis of behavioral illness for the general public.

For ADHD, this last motivation is particularly poignant, because for a large proportion of the public, the question of whether ADHD is “real” is not rhetorical. Despite an enormous amount of evidence from decades of research and clinical trials, many people still believe that the diagnosis of ADHD is just an excuse for people being lazy or undisciplined—or, even more cynically, that it is a ploy by psychiatrists to drum up business or by pharmaceutical companies to make money by “drugging” children and getting them addicted to their products.

ADHD is by definition real in the narrow sense that it is present when people meet certain agreed-upon diagnostic criteria; the criteria outlined in DSM-5 are those most commonly used currently in the United States. However, when contemplating whether ADHD is real, people more often mean in the sense of having identifiable biological differences that can be reliably measured and that can serve as a definitive test of whether someone has or does not have the disorder.

Brain imaging has been a prominently used tool in seeking to establish such markers for ADHD. Spurred on by a widely publicized 1990 study by Zametkin et al. (3) using positron emission tomography (PET) to establish glucose utilization differences in adults with ADHD, many investigators sought to use brain imaging to further characterize the neurobiological basis of the disorder.

Because of concerns regarding the use of ionizing radiation in PET scans, especially for children, most of the subsequent studies employed MRI. MRI allows exquisitely accurate pictures of anatomy, and functional MRI (fMRI) of the physiology of the brain. It does so without ionizing radiation, so it can not only be safely used to scan children but to scan them repeatedly over the course of development. This launched a new era of pediatric neuroscience, with studies of ADHD, the most commonly treated neuropsychiatric disorder of childhood, as a frequent target of investigation.

Anatomical MRI studies of ADHD have undergone four overlapping phases since the late 1980s. The first was to look for clear pathological findings in affected people. The “hole in the brain” approach was not fruitful, as people with ADHD do not have a higher rate of such MRI findings. MRI brain scans of those with ADHD are no more likely than those of healthy control subjects to come back with a clinical radiology report of an abnormality.

The next phase was to compare the size or shape of various substructures in the brain. Although the scans were read clinically as normal, group average size differences were reported for several components of the brain—most notably in the frontal lobes, basal ganglia, and cerebellum. These findings are consistent with lesion studies of nonhuman primates and with neuropsychological findings of brain damage from accidents or strokes.

The third phase was to use longitudinal studies to look at developmental trajectories of the various brain components—to look at a movie instead of a snapshot. This approach proved to provide better discrimination between individuals with and without ADHD, but such studies are more difficult to perform.

The fourth phase has been to use multimodal imaging and graph theoretical approaches not only to examine the size and shape of brain subcomponents but also to characterize how they are related to each other over the course of development. Thus far, this approach has shown the most robust discriminatory power.

The studies strongly support brain differences in a group of people with ADHD and further support the notion that the symptomatology of ADHD and the cortical surface alterations reported lie along a continuum. However, as is currently the case for all other psychiatric disorders, none of these approaches allows for discrimination strong enough to have routine clinical or diagnostic utility in ADHD. Likewise, none has clearly demonstrated to the public that ADHD is “real” in the sense that one can reliably point to an area on an MRI scan and show, without the use of nuanced statistics, where the brain is different.

As with the previous ADHD ENIGMA study of subcortical structures (2), the interpretation of this study is likely to be polarized. Some will note that this study, with the largest-ever sample size, conducted by the leading experts in the world, establishes statistically robust brain differences in individuals with ADHD compared to those without. Others will note the extremely small effect sizes (despite statistical significance), the lack of strong a priori hypotheses regarding the structures that differ between the ADHD and control groups, and the minimal impact of the data in characterizing meaningful subtypes of ADHD, in suggesting novel mechanisms of etiology or psychopathology, in developing new pharmacological or behavioral interventions, or in guiding future directions for research.

The latter interpretation by vast numbers of people would be unfortunate. As large and as well designed and conducted as the ENIGMA studies are, they should not be seen as the definitive answer as to whether there are useful anatomical biomarkers in ADHD—or, to put it more vernacularly, as to whether ADHD is real. Anatomical MRI, even on such a large scale, is but one approach to searching for brain differences in ADHD. Even within the realm of brain anatomy, many differences could go undetected with our current approaches. Meta- and mega-analytic approaches combining data across sites with different methodologies—even when conducted by best current methodology—have inherent limitations. The anatomical resolution of approximately 1 mm means that the smallest unit of measurement still contains millions of neurons and trillions of synaptic connections. Longitudinal data, fMRI, other imaging modalities, or advanced statistical approaches to characterize connectivity may uncover the long-sought biomarkers.

ADHD is a real, impairing, treatable biological condition affecting tens of millions of people. Scientists will continue to search for actionable biomarkers and to improve detection, treatment, and outcome for those afflicted. Brain imaging studies will continue be part of these ongoing efforts. However, care should be taken to communicate findings to the public in accurate and accessible ways so as to decrease the stigma of mental illness and increase support for psychiatric research.