Clinical MRI cutoff points for predicting lacunar stroke may not exist: need for a grading rather than a dichotomizing system.

BACKGROUND

Lacunar stroke has been defined as an infarct <15 mm in diameter in the presence of symptoms of lacunar syndromes. We investigated a new approach in predicting whether a deep infarct is caused by small arterial occlusion.

METHODS

A total of 319 consecutive patients with acute symptomatic infarcts within the striatocapsular territory underwent diffusion-weighted imaging (DWI) and diagnostic workups, including vascular and cardiological studies. Predictors for nonlacunar mechanisms were evaluated by logistic regression analysis, with the size of infarct (1-mm increase) and stroke syndrome (traditional vs. atypical lacunar syndrome vs. cortical syndrome) graded rather than dichotomized.

RESULTS

Amongst the 171 patients who did not meet the established criteria for lacunar stroke, that is, deep infarct of >or=15 mm or presenting symptoms of nontraditional lacunar syndrome, a documented etiology could not be determined in 97 (56.7%) patients. In contrast, amongst the 148 patients who met the criteria, 27 (18.2%) had nonlacunar mechanisms. Logistic regression analysis identified the variables that predicted nonlacunar stroke mechanisms with statistical significance as nontraditional lacunar syndromes (OR 2.19 for atypical lacunar syndrome, and OR 6.72 for cortical syndrome), infarct size on DWI (OR 1.05 per 1-mm increase), and unilateral multiple deep infarcts on DWI (OR 2.22, p < 0.05 in all cases). Receiver operating characteristic curves showed that discrimination power of the model derived from logistic regression analysis (grading system) was better than that of the previously established dichotomizing criterion in predicting nonlacunar mechanisms (p = 0.004).

CONCLUSIONS

A clinically significant proportion of clinical MRI lacunae are associated with underlying nonlacunar mechanisms. Decisions regarding the extent of diagnostic procedures in patients with subcortical infarcts can be guided by the point value in terms of the stroke syndrome and infarct patterns, as well as the size of infarct.