DWI vs. Post-mortem Examination of Patients with Acute Stroke: In Search of a Gold Standard
DWI vs. Post-mortem Examination of Patients with Acute Stroke: In Search of a Gold Standard
Abstract & Commentary
Source: Kelly PJ, et al. Diffusion MRI in ischemic stroke compared to pathologically verified infarction. Neurology. 2001;56:914-920.
Diffusion weighted imaging (DWI) has become the new standard in the diagnosis of acute stroke. Because DWI directly detects tissue ischemia, it has a sensitivity and specificity far superior to standard T2 or FLAIR sequences. As previous Neurology Alert reviews have noted, however, DWI does not correlate directly with stroke in every case. False-negative "misses" and false-positive "mistakes," do occur. Not every stroke is assured to produce a DWI signal and many other entities other than stroke, such as glioma, demyelination, or seizure, may produce DWI abnormalities. In the current report, Kelly and colleagues compare DWI against the most certain marker of stroke, infarction of brain tissue pathologically verified at autopsy.
Kelly et al report on 11 patients studied retrospectively who underwent DWI scanning for evaluation of stroke syndromes and who subsequently died. Each underwent full autopsy examination, including gross and microscopic analysis. A total of 25 strokes were detected at autopsy; 23 of which were seen on DWI. Two strokes detected by DWI were not found at autopsy. The sensitivity, specificity, positive predictive and negative predictive values for detection of stroke using DWI were, therefore (in a small cohort), 88.5%, 96.6%, 85.2%, and 97.4%, respectively.
Three additional cases were studied prospectively, with the neuropathologist blinded to the DWI findings. The correlation between DWI and pathologically proven strokes was virtually 1:1, including tiny cortical emboli. In 1 case, a large left posterior cerebral artery territory infarct, pathologically aged 12-24 hours, was not detected by DWI and almost certainly occurred during the interval between the last MRI scan and the patient’s death. Similarly, Kelly et al describe 2 DWI "misses" in the retrospective cohort, both of which were found in patients with endocarditis who were intubated, sedated, and likely sustained ongoing emboli between the last MRI and death. The 2 possible DWI "mistakes" probably did represent tiny strokes missed on 1 cm autopsy slice specimens. Finally, 2 other patients who were initially thought to have presented with stroke based on clinical exam, were ultimately diagnosed with Todd’s paralysis and metabolic encephalopathy. In these cases, both DWI and pathological examination confirmed the absence of an ischemic lesion.
Commentary
The data presented by Kelly et al represent an important correlation between radiologic data and neuropathology. DWI is at least 90% accurate, and probably more, in diagnosis of acute stroke. But DWI or any imaging procedure must ultimately be interpreted in the context of the clinical history and neurological examination. The majority of the autopsy specimens in Kelly et al’s study were not reviewed blindly. It is, therefore, possible that the pathologist actually used the DWI to assist in localizing areas of infarction at autopsy. Although Kelly argues that, "neuropathologists at our institution do not routinely examine neuroimaging data prior to sectioning the brain," it is likely that pathologists reviewed clinical histories, and ultimately CT, MRI, or other associated data before signing out their final conclusions. Any comparison of DWI to autopsy data is prone to the possibility of interim events occurring between the most recent DWI and the time of death or the chance of the pathological prosector missing one or more tiny infarcts. Unless DWI is performed immediately postmortem or neuropathological methods are 100% diagnostic of neuronal ischemia, the MRI will remain just one element in a comprehensive exercise of clinicopathological correlation. —Alan Z. Segal
Subscribe Now for Access
You have reached your article limit for the month. We hope you found our articles both enjoyable and insightful. For information on new subscriptions, product trials, alternative billing arrangements or group and site discounts please call 800-688-2421. We look forward to having you as a long-term member of the Relias Media community.