Perfusion-Weighted Imaging/Diffusion-Weighted Imaging Mismatch on MRI Can Now Be Used to Select Patients for Recombinant Tissue Plasminogen Activator Beyond 3 Hours
My German colleagues claim that MRI can now be used to select patients for tissue plasminogen activator (tPA) therapy beyond the 3-hour time window. I wish that were so. As a clinical investigator, it would make my life much easier. However, I still have important reservations.
The computed tomography (CT) criteria for excluding stroke victims within 3 hours were defined in the National Institutes of Health tPA trial3 as being the presence of an increased density on the image. This is associated with hemorrhage. However, as my colleagues note, “Although never formally assessed, CT is commonly considered the ‘gold standard’ to demonstrate (intracerebral hemorrhage).”2 Nevertheless, the tPA trials showed that excluding patients with this abnormality leads to demonstrable efficacy of treatment for acute stroke patients. The reason we want to exclude patients with hemorrhage is that a thrombolytic will not be efficacious and may increase the hemorrhage rate.1,3
Now the imagers want to substitute MRI for CT. We would like MRI to provide us with 2 types of information: (1) identification of salvageable tissue, and (2) exclusion of hemorrhage. Unfortunately, at present, we are not sure that MRI can do either. At least 3 clinical trials are currently in progress to find out whether perfusion/diffusion mismatch can be used to select patients with salvageable tissue. However, none of these trials have been published, so I cannot review the data. The claim is that the area in between the adequately perfused tissue and the DWI abnormality is the potential salvageable tissue. This is a common-sense idea, but similar armchair speculations in the past have been shown to be inaccurate. The standards for measurement of these differences are not yet well worked out, and there is much work left to be done. A problem with MRI is that we are not sure what the images mean. At a basic level, DWI, semiquantitatively, measures proton movement, and PWI qualitatively measures the presence of contrast material, but that is far from being evidence for potentially salvageable tissue.
As opposed to the claims of Drs Schellinger and Fiebach, CT is not the “gold standard” for these purposes; histopathology is. To the best of my knowledge, neither CT not MRI has ever been calibrated against this true standard. This is a critical missing bit of information, and appropriate animal studies could be done to investigate this issue. The question is not how sensitive MRI is; we know it is reasonably sensitive to detect large (by histological standards) lesions. The question is how specific it is. When a blood vessel ruptures, 2 things happen: (1) a parenchymal hemorrhage is produced, and (2) the tissue supplied by that vessel becomes infracted. Furthermore, even histopathology during the first few hours after stroke onset cannot identify salvageable tissue, so we have no “gold standard” for that.
We all would love to have a method for measurement of tissue viability that is better than simple time measurement from symptom onset. It is possible that empiric criteria will be identified that will be useful for these purposes, but the studies to prove it have not yet been completed, and it may be impossible. The clinical rating scales we use for our outcome measurements are fairly crude, and a more sensitive and specific test would be highly desirable. But there is no evidence that any type of image can do that. The correlation between image size and clinical deficits is poor. A large lesion in a relatively silent area will not produce much clinical damage, whereas a very small lesion in the internal capsule can produce substantial loss of neurological function. In the end, the regulatory agencies of most countries will require behavioral standards to approve new therapies. Images cannot currently supply this evidence, and it is likely that they never will. I hope that the imagers will come up with something we can use, at least for identification of potentially useful drugs in phase II trials. For now, I remain a skeptic.
- Received October 22, 2004.
- Accepted October 22, 2004.
Chapman D, Lyden P, Lapchak PA, Nunez S, Thibodeaux H, and Zivin J. Comparison of TNK with wild-type–t-PA in a rabbit thromboembolic stroke model: thrombolytic efficacy and incidence of cerebral hemorrhage. Stroke. 2001; 32: 748–752.
Fiebach JB, Schellinger PD, Gass A, Kucinski T, Siebler M, Villringer A, Olkers P, Hirsch JG, Heiland S, Wilde P, Jansen O, Rother J, Hacke W, Sartor K. Stroke magnetic resonance imaging is accurate in hyperacute intracerebral hemorrhage. Stroke. 2004; 35: 502–506.