Apparently, Diffusion Coefficient Value and Stroke Treatment Remains Mysterious

Diffusion-weighted imaging (DWI) has become a well-recognized tool in the diagnostic armamentarium of neuroradiologists. The ability of DWI to depict acutely ischemic tissue in the appropriate clinical setting with high sensitivity and high specificity has made DWI the diagnostic method of choice whenever the question of brain ischemia is raised. We know that the high signal intensity on a DWI of an infarct is due to the restriction of water mobility in a voxel of tissue; this appearance represents tissue that keeps much of its baseline signal intensity while all the normal tissue loses its signal intensity as the diffusion-encoding gradients are applied. In short, the apparent diffusion coefficient (ADC), the number that describes water mobility, decreases in ischemic tissue compared with normal values; this change indicates decreased microscopic dephasing in ischemic tissue.

Despite the widespread acceptance of this technique, many fundamental questions about DWI remain. Chief among these is, “Why does ADC decrease?” The short answer is this: We do not know for sure. Fortunately, a tremendous amount of work has been done in animal models to investigate the answer to this question (recently and masterfully reviewed in reference 1), and, therefore, we do know some things. Current evidence indicates, for example, that the ADC decrease is associated with energy failure and cytotoxic edema or the swelling of the cells related to a depolarization-induced water influx from the extracellular compartment. Why this water influx causes a decrease in the ADC value is still unclear, because current data suggest that the water mobility decreases with acute ischemia both inside and outside the cell.

A lack of precise understanding about the biophysical mechanisms by which the ADC decrease occurs does not preclude us from using DWI as a diagnostic tool in the clinic, but an air of mystery around the meaning of DWI does persist. After all, in experimental models, a number of nonischemic events (eg, administration of an excitotoxin such as N-methyl-d-aspartate [NMDA]) can also cause a decrease in the ADC and cause cell swelling despite normal adenosine triphosphate (ATP) levels. This observation may suggest that a simple direct relationship, such as one in which the ADC decrease is directly correlated to tissue demise, may not be present. Nevertheless, human experience particularly shows that, in the absence of therapy for suspected acute cerebral ischemia, tissue with a low ADC value almost always progresses to infarction; therefore, the ADC value provides us with a powerful diagnostic tool.

The mystery deepens when we begin to consider how ADC changes with intervention. Our understanding here is most complete in animal models, in part because of the difficulty of examining humans with their widely varying pathologic conditions and in part because of the lack of effective therapies for most patients with stroke. We do know that, with early reperfusion, this decreased ADC value can normalize; this change reflects a normalization of the tissue energy status. Although, this normalization of ADC was initially thought to mean that the tissue was salvaged by reperfusion, both human and animal data have shown that this is not the case; infarcts can still often occur in tissue despite early reperfusion and normalization of diffusion and perfusion images. The phenomenon of infarction despite apparent normalization early after reperfusion is termed delayed injury or secondary injury. We knew that tissue with high signal intensity at DWI progressed to infarction, but now, the converse is clearly not always true; tissue that with a bright appearance at DWI that then becomes normal may or may not be salvaged from infarction. Because of this possibility, delayed or secondary injury is increasing scrutinized, and we are beginning to understand more about it with recent experimental results. For example, we know now that, although ADC values might normalize with acute reperfusion, metabolic imaging studies show that cerebral protein synthesis remains suppressed and abnormal. In short, this secondary injury is thought to involve failure of mitochondrial function, perhaps including apoptosis.

In this issue of AJNR, more evidence about secondary injury is presented (pages 180–188). Li et al document that, in a rat model of ischemia, 30 minutes of occlusion in the middle cerebral artery followed by reperfusion resulted in the normalization of ADC values, but histologic changes were persistent and severe; this finding confirms that of similar work done in animal models of hypoxia-ischemia. This result is important because it indicates that, although decreases in ADC values are associated with cellular swelling, the (pseudo)normalization of the ADC does not indicate normalization of cellular swelling and other forms of damage. Again, this observation confirms that the ischemic cascade may be a more difficult foe than we had hoped, and it suggests that, while decreases in the ADC are extremely valuable in the diagnosis of acute cerebral ischemia, the use of DWI to monitor therapeutic interventions such as thrombolysis may have some pitfalls. Because so few patients are currently receiving thrombolysis, the day-to-day effect of this point on most of our practices is, unfortunately, limited. However, it has important implications for the design of novel interventions, because early reperfusion and even normalization of energy balances clearly may not be enough to successfully treat acute ischemic stroke.

One message that these recent findings highlight—one that goes beyond the specific topic of delayed injury in brain ischemia—is the important interplay between results in animal models and human findings. Certainly, the tremendous amount of work done in animal models, by groups all around the world, has vastly improved our understanding of acute cerebral ischemia. However, those conducting animal experiments have also learned from the human experience, and perhaps nowhere else are the shortcomings of seemingly good animal models so apparent as in the modeling of acute cerebral ischemia and its treatment. The best illustration may be the fact that more than 100 novel medications have been shown to be highly successful in the treatment of ischemia in animal models, and yet, each has failed to work in humans (2). Certainly. such experience highlights that, while animal models are essential, such models do not preclude careful clinical investigation–something to which we in neuroradiology are in an excellent position to contribute. Indeed, the observation of delayed injury after thrombolysis in humans seems to have sparked renewed interest in this area of study in animal models; an example is the work of Li et al in this issue.

In short, animal models have a lot to teach us, but our observations in the clinic remain extremely valuable. Such interplay between the clinic and the laboratory should encourage all of us interested in the care of patients with stroke to continue our communications with each other, because we each contribute our own piece of information to solving the mystery of how to treat acute cerebral ischemia.

• Copyright © American Society of Neuroradiology