Validation as New Imaging Biomarker

Background In order to select patients most likely to benefit for thrombolysis and to predict patient outcome in acute ischemic stroke, the volumetric assessment of the infarcted tissue is used. However, infarct volume estimation on Diffusion weighted imaging (DWI) has moderate interrater variability despite the excellent contrast between ischemic lesion and healthy tissue. In this study, we compared volumetric measurements of DWI hyperintensity to a simple maximum orthogonal diameter approach to identify thresholds indicating infarct size >70 ml and >100 ml. Methods Patients presenting with ischemic stroke with an NIHSS of ≥ 8 were examined with stroke MRI within 24 h after symptom onset. For assessment of the orthogonal DWI lesion diameters (od- values) the image with the largest lesion appearance was chosen. The maximal diameter of the lesion was determined and a second diameter was measured perpendicular. Both diameters were multiplied. Od-values were compared to volumetric measurement and od-value thresholds identifying a lesion size of > 70 ml and > 100 ml were determined. In a selected dataset with an even distribution of lesion sizes we compared the results of the od value thresholds with results of the ABC/2 and estimations of lesion volumes made by two resident physicians. Results For 108 included patients (53 female, mean age 71.36 years) with a median infarct volume of 13.4 ml we found an excellent correlation between volumetric measures and od-values (r2 = 0.951). Infarct volume >100 ml corresponds to an od-value cut off of 42; > 70 ml corresponds to an od-value of 32. In the compiled dataset (n = 50) od-value thresholds identified infarcts > 100 ml / > 70 ml with a sensitivity of 90%/ 93% and with a specificity of 98%/ 89%. The od-value offered a higher accuracy in identifying large infarctions compared to both visual estimations and the ABC/2 method. Conclusion The simple od-value enables identification of large DWI lesions in acute stroke. The cutoff of 42 is useful to identify large infarctions with volume larger than 100 ml. Further studies can analyze the therapeutic utility of this new method