B a c k g ro u n d : The upper limit for the size of a lacuna has been established at a diameter of 15 mm. We aimed to assess infarct size and develop a model for predicting whether a deep infarct is caused by small arterial occlusion. M e t h o d s : We prospectively studied 304 patients with acute, symptomatic, subcortical infarcts. All the patients showed relevant lesions within the striatocapsular territory on diffusion-weighted imaging, and those with concomitant cortical lesions were excluded from this study. We divided the patients into two groups: (a) those with neither occlusion of the relevant vessels nor potential sources of cardioembolism (cryptogenic group, n=206), (b) patients with large artery atherosclerosis, cardioembolism or other documented etiology (non-lacunar group, n=98). We used Receiver Operating Characteristic (ROC) curves to validate the models in predicting non-lacunar mechanisms. R e s u l t s : The sex ratio and risk factor profile did not differ between the groups. Lacunar syndromes were more frequently found in the cryptogenic group (85.4%) than in the non-lacunar group (57.1%) (P<0.001). The larg e s t diameter in patients in the cryptogenic group (14.5±6.8 mm) was smaller than that in patients with other documented etiologies (20.0±11.4 mm) (P<0.001). Discrimination was better when both infarct size and syndrome were considered than when infarct size alone was considered (P=0.031) ; the area under the ROC curve was 0.644 for the latter and 0.700 for the former. In calculating the non-lacunar mechanisms risk index score, the value assigned was one point for each millimeter infarct size, and 17 points for the presence of non-lacunar syndrome. A score of ≥21 points forecast a high probability of non-lacunar mechanisms, yielding a sensitivity of 61% (95% CI, 51-71%) and specificity of 75% (95% CI, 68-81%); positive and negative predictive values were 54% and 80%, r e s p e c t i v e l y. C o n c l u s i o n s : A clinically significant proportion of clinicoradiologic lacunae may be associated with underlying non-lacunar mechanisms. Decisions regarding the extent of diagnostic procedures in patients with subcortical infarcts may be guided by the point value in terms of the stroke syndrome and infarct size. (Korean Journal of Stroke 2005;7(2): 133~143)

B a c k g ro u n d : The upper limit for the size of a lacuna has been established at a diameter of 15 mm. We aimed to assess infarct size and develop a model for predicting whether a deep infarct is caused by small arterial occlusion. M e t h o d s : We prospectively studied 304 patients with acute, symptomatic, subcortical infarcts. All the patients showed relevant lesions within the striatocapsular territory on diffusion-weighted imaging, and those with concomitant cortical lesions were excluded from this study. We divided the patients into two groups: (a) those with neither occlusion of the relevant vessels nor potential sources of cardioembolism (cryptogenic group, n=206), (b) patients with large artery atherosclerosis, cardioembolism or other documented etiology (non-lacunar group, n=98). We used Receiver Operating Characteristic (ROC) curves to validate the models in predicting non-lacunar mechanisms. R e s u l t s : The sex ratio and risk factor profile did not differ between the groups. Lacunar syndromes were more frequently found in the cryptogenic group (85.4%) than in the non-lacunar group (57.1%) (P<0.001). The larg e s t diameter in patients in the cryptogenic group (14.5±6.8 mm) was smaller than that in patients with other documented etiologies (20.0±11.4 mm) (P<0.001). Discrimination was better when both infarct size and syndrome were considered than when infarct size alone was considered (P=0.031) ; the area under the ROC curve was 0.644 for the latter and 0.700 for the former. In calculating the non-lacunar mechanisms risk index score, the value assigned was one point for each millimeter infarct size, and 17 points for the presence of non-lacunar syndrome. A score of ≥21 points forecast a high probability of non-lacunar mechanisms, yielding a sensitivity of 61% (95% CI, 51-71%) and specificity of 75% (95% CI, 68-81%); positive and negative predictive values were 54% and 80%, r e s p e c t i v e l y. C o n c l u s i o n s : A clinically significant proportion of clinicoradiologic lacunae may be associated with underlying non-lacunar mechanisms. Decisions regarding the extent of diagnostic procedures in patients with subcortical infarcts may be guided by the point value in terms of the stroke syndrome and infarct size. (Korean Journal of Stroke 2005;7(2): 133~143)