The clinical significance of diffusion-weighted MR imaging in infratentorial strokes

Objective: To study the association between diffusion-weighted imaging (DWI) characteristics and stroke etiology, stroke severity, and functional outcome in patients with infratentorial strokes. Methods: The authors prospectively studied 22 consecutive patients with acute infratentorial strokes. They used a blinded comparison of DWI features (number, distribution, and volume of lesions) with clinical characteristics, namely, stroke etiology (Trial of ORG 10172 in Acute Stroke Treatment [TOAST] classification), severity (NIH Stroke Scale [NIHSS]), length of stay (LOS), and functional 3-month outcome using modified Rankin Scale, Barthel Index, and a dichotomized outcome status (living at home vs institutionalization or death). Results: Acute infratentorial DWI lesions were detected in 95% (21/22) of the patients. The number (p = 0.01) and the distribution (p < 0.001) of DWI lesions were correlated with stroke etiology. Patients with cardioembolic strokes (n = 5) had more DWI lesions (8.0 +/- 6.0) than those with other stroke etiologies (n = 17; 1.3 +/- 0.9; P < 0.001). Their lesion distribution differed from that of patients with noncardioembolic strokes (p < 0.001). Clinically silent, acute DWI lesions in the anterior circulation in addition to their infratentorial lesions were visualized in 3 of 5 patients with cardioembolic stroke and in none of 17 patients without sources of cardioembolism (p < 0.001). Pure infratentorial lesions were present in 15 of 17 patients with noncardioembolic strokes and in none of 5 cardioembolic stroke patients (p < 0.001). DWI lesion volume was not correlated with NIHSS score, LOS, outcome scores, or outcome status. Conclusion: In infratentorial strokes, multiple DWI lesions and a distribution of subsidiary, clinically silent DWI lesions in the anterior circulation suggest a cardioembolic stroke etiology. However, DWI lesion volume did not correlate with the NIHSS score and was no predictor of outcome.