Perfusion adds clinical value to CT evaluation of stroke

The National Institute of Neurological Disorders and Stroke (NINDS) study of MRI and CT for emergency stroke assessment1 redefined the roles of the 2 imaging modalities most relevant to the diagnosis and characterization of stroke. It elevated MRI to the status of gold standard for both academic and community hospitals and denigrated CT by quantifying its shortcomings.

The numbers were damning for CT. The 54% accuracy rate of noncontrast CT for diagnosing ischemic stroke was only slightly better than a coin toss, said Steven Warach, MD, the supervising author of the NINDS study and section chief of stroke diagnostics and therapeutics at NINDS. He confirmed that the trial should encourage every hospital that treats stroke to develop protocols making MRI the primary test for stroke.

"Purely from the standpoint of wanting the most accurate diagnosis, MRI should be the preferred neuroimaging choice for evaluating possible stroke, unless there is a contraindication," he said.

The NINDS study findings do not mean that CT is dead as an instrument for stroke analysis, but innovation is needed to keep the modality in the game. CT captures other crucial aspects of hemorrhagic stroke, and this capability will preserve its position on the front line of stroke diagnosis at Massachusetts General Hospital, said R. Gilberto Gonzalez, MD, PhD, director of neuroradiology at MGH.

For ischemic stroke, CT angiography can differentiate between small- and large-vessel occlusion. For hemorrhage, CTA provides the fastest and most reliable way to determine an intracranial hemorrhage's origin, Gonzalez said.

Researchers are developing perfusion CT strategies to match development in MR, but progress is still preliminary, with no agreement yet on the optimal protocol. In fact, studies by Max Wintermark, MD, an assistant professor of radiology at the University of California, Los Angeles, and Michael Lev, MD, an associate professor of radiology at Harvard Medical School and director of emergency neuroradiology at MGH, paint different pictures of the best way to measure penumbra with perfusion CT.

Wintermark relied on receiver operator characteristic analysis to isolate optimal parameters based on evaluation of perfusion CT cerebral blood volume, cerebral blood flow, mean transit time, and time-to-peak measures. He found that relative MTT with a threshold of 145% compared with MTT in corresponding voxels in the contralateral cerebral hemisphere was best suited for isolating the penumbral perfusion perimeter, the region of tissue at risk of infarction. Absolute CT CBV with an optimal threshold of 2 mL x 1000 g-1 was the best technique for identifying a volume equivalent to the ischemic core.2

Lev found fault with Wintermark's approach, noting in an interview at the 2007 International Stroke Conference that it did not account for differences in the perfusion rates of normal white and gray matter. Recent work by Lev's group presented as a scientific poster at the show indicates that a relative MR diffusion-weighted imaging (DWI) threshold of about 50% reduction in the rate of ischemic tissue compared with normal tissue in the contralateral hemisphere provides a better measure for the volume of the ischemic core. Using r2 values, Lev calculated a very strong correlation rate of 0.9 in applications of the 50% diffusion threshold.

Lev's group has found that CTA source images also show promise for defining the infarct core. These images are blood volume-weighted, giving them the capacity to function like DWI, he said. This capability would be a bonus because CTA is often performed anyway to identify possible vessel occlusion in the circle of Willis after a stroke.

REFERENCES

• Chalela JA, Kidwell CS, Nentwich LM, et al. Magnetic resonance imaging and computed tomography in emergency assessment of patients with suspected acute stroke: a prospective comparison. Lancet 2007;369:293-298.

• Wintermark M, Flanders AE, Velthius B, et al. Perfusion-CT assessment of infarct core and penumbra: receiver operating characteristic curve analysis in 130 patients suspected of acute hemispheric stroke. Stroke 2006;37:979-985.