Diffusion Tensor Imaging and Mild Traumatic Brain Injury in Soldiers: Abnormal Findings, Uncertain Implications

Each year in the United States, approximately 1–2 million people sustain a traumatic brain injury (TBI), making this the second most common neurological disorder after headaches (1, 2). Approximately 80% are mild TBIs, which are those injuries where the Glasgow Coma Scale score is greater than 13, loss of consciousness is less than 30 minutes, posttraumatic amnesia is less than 24 hours, or even a brief alteration of mental state (“dazed and confused”) occurs (3, 4). Over the past 5–10 years, interest in mild TBI has been increasing because of sports concussions and injuries among U.S. military and civilian personnel serving in Iraq and Afghanistan. An estimated 15%–20% of soldiers sustain mild TBIs during their deployment in these theaters (1, 5).

Routine neuroimaging techniques (CT and MRI) do not usually reveal abnormalities after mild TBI, so extensive research is being conducted to find methods with greater sensitivity and specificity. White matter tracts such as the corpus callosum, internal capsule, and corona radiata appear to be the most vulnerable to injury. Diffusion tensor imaging (DTI) is able to detect damage to axonal tracts by using a measure of directional water diffusion (fractional anisotropy). Both positive and negative findings have been reported (2, 6). Some studies analyzed whole brain areas while others focused on specific regions of interest. Unfortunately, there has been no methodological uniformity among studies in defining abnormality. Since abnormalities are often calculated based on differences from healthy comparison subjects, the question arises as to what is the most appropriate control. As individuals with mild TBI may also have posttraumatic stress disorder (PTSD), depression, anxiety, headaches, sleep disturbance, and other symptoms, researchers need to compare individuals with TBI with those who have psychiatric symptoms but no injury. A recent study found DTI abnormalities in combat-exposed soldiers, which normalized after 1.5 years, but the soldiers had neither PTSD nor TBI (7).

In a recent meta-analysis, Aoki et al. (8) analyzed 13 studies of mild TBI that met specific criteria. They found significant reductions in fractional anisotropy in the posterior portions of the corpus callosum (the splenium). As a comparison, Lipton et al. (9) performed DTI 2 weeks, 3 months, and 6 months after injury in 34 patients with mild TBI and 30 matched healthy comparison subjects. They found varied patterns of white matter abnormalities in the TBI patients over time. At 2 weeks, areas of both low and high fractional anisotropy were found in 32 patients in the corona radiata, the precentral white matter, the internal capsule, and the deep and subcortical white matter. These patients had low fractional anisotropy in the splenium of the corpus callosum, but high fractional anisotropy in the anterior corpus callosum (genu and body). Only six had high fractional anisotropy in the splenium. At 3 months, most patients had more areas of high fractional anisotropy and fewer low areas. By 6 months, 10/10 patients had areas of high fractional anisotropy, while 7/10 had areas of low fractional anisotropy (a pattern of decline of low areas over time). This demonstrates that patterns of abnormalities differ among patients and over time. This finding complicates the analyses of studies that “average” abnormal areas and meta-analyses of multiple studies.

The study by Jorge et al. (10) in this issue provides an additional method of analyzing DTI data and examines whether specific regions of interest or the number of abnormal areas is the most sensitive method. The authors hypothesized that lesions may occur in different areas, depending on the injury and the individual, and thus that restricting analysis to specific regions may overlook significant pathology. They studied returning veterans of the wars in Iraq and Afghanistan and examined 72 individuals with mild TBI from blast exposure (32 of these veterans also had acceleration and rotational forces), 21 veterans without TBI, and 14 civilians with TBI who did not have any psychopathology. They analyzed DTI with a voxel-based analysis and a method to detect spatially heterogeneous areas of decreased fractional anisotropy (“potholes”). MRI was obtained within 90 days of trauma for civilians and approximately 50 months for the military groups. Individuals were assessed for psychiatric diagnosis (including ratings for PTSD and depression) and for executive function and memory. Average voxel-based fractional anisotropy failed to reveal differences among these groups; however, there was a difference in the number of potholes detected in veterans with TBI relative to those without. While the mild TBI veteran group had more potholes than the veterans without TBI, the civilian TBI group had more potholes than the military TBI groups. The authors did not examine whether there were fewer potholes in the veterans who had only blast injury. The number of potholes was associated with TBI severity, and a negative correlation was observed between the number of potholes in the corpus callosum and executive functioning. Psychiatric disorders and symptoms were not related to the number of potholes.

These findings are intriguing and promising, and they need replication. Jorge et al. are to be congratulated for including an appropriate comparison group that had similar psychiatric problems but did not experience mild TBI, in addition to a civilian group with mild TBI. The study’s strength is that it examines DTI data in a different way (which emphasizes the fact that we have no accepted consensus as to the best way to do this). The weakness is that this study cannot be viewed as a study of blast injury, but rather a study of abnormalities in soldiers who have had mild TBI and emotional problems. At this time, we need to view the potholes as an unproven possible pathophysiological signature that may be an epiphenomenon of what we are calling a brain injury. We do not know the prognostic significance of an abnormality, and it appears that the abnormalities may change over time.

While the authors did not advocate the use of potholes as a diagnostic biomarker for TBI, clinicians must be cautious and not indiscriminately apply an invaluable research technique to the many individuals who have suffered a concussion, especially in forensic settings. Before using DTI in patients with suspected mild TBI, consider this scenario: You had a concussion 6 months ago, and still don’t feel right. You are slower to process things and have trouble multitasking. You tire easily and are irritable. Routine brain imaging results have been normal. Your doctor orders a 3-T MRI with DTI that reveals white matter lesions consistent with a TBI. Does this finding make you feel better or worse? Does the belief that a brain injury has been confirmed provide relief (a cause has been found, and it’s not “in my head”), or distress (there is brain damage, and I won’t get better)? Does this help or interfere with recovery?

The vast majority of individuals with a single concussion fully recover. The evidence indicates that a positive expectation of recovery significantly predicts prognosis (11), and many unexplored factors may influence symptoms and recovery (12). Will the finding of an abnormality on DTI—which has no bearing on treatment decisions or interventions, since those are symptom based—affect recovery? Studies such as the one by Jorge et al. are important steps forward in trying to understand possible pathological changes, because they control for common coexisting problems. Unfortunately, these studies do not yet provide us with a gold-standard clinical diagnostic measure for TBI.