Article Text
PDF
Abstract
Introduction Hemorrhagic transformation (HT) is a frequent complication after mechanical thrombectomy (MT), causing acute neurologic deterioration and poorer long-term functional outcomes. Classically, CT-based grading systems like the Heidelberg grade have been used to assess HT severity. No prior studies have applied Heidelberg grading to the single-shot gradient-echo echo-planar images acquired for dynamic susceptibility contrast-enhanced (DSC) MR perfusion. Given that DSC MR perfusion images are routinely acquired for assessment of ischemic penumbra, evaluation of HT on unprocessed DSC images may preclude the need for additional gradient-echo (GRE) or susceptibility-weighted imaging (SWI) sequences. This study assessed the reliability and performance of DSC MR perfusion images to detect HT using the Heidelberg classification in comparison to SWI, GRE, and CT.
Methods This study retrospectively reviewed patients with LVO undergoing MT at our institution from January 2020-August 2022. Only patients who received MRI or CT after undergoing MT were included. Two independent observers were then blinded to clinical information and asked to evaluate degree of HT on post-MT DSC MR perfusion and SWI or GRE sequences using Heidelberg. This was also done for post-MT CT, if available. Symptomatic ICH was independently assessed according to joint commission guidelines of hemorrhage on imaging and worsening NIHSS.
Results The reliability of reader ratings (Shrout-Fleiss Intraclass correlation coefficient) were similar across all DSC-MR perfusion (0.539), SWI/GRE (0.489), and CT (0.557). When pooling ratings and excluding cases with CT contrast staining, they differed between the 3 acquisitions (P<0.001). Pairwise comparisons showed the proportion with HT differed between SWI/GRE (136/213, 63.9%) and CT (111/213, 52.1%, P<0.001), SWI/GRE and DSC-MR perfusion (101/213, 47.4%, P<0.001), but not between DSC-MR perfusion and CT (P=0.332). Dichotomizing Heidelberg ratings as more likely to be clinically significant (≥grade 2) vs non-significant, proportions did not differ between SWI/GRE (93/213, 43.7%) and CT (79/213, 37.1%, P=0.093) and between CT and DSC MR perfusion (81/213, 38%, P=0.904). Grouping Heidelberg ratings into 0, 1+2, and >2, the distributions significantly differed between SWI/GRE and CT, SWI/GRE and MR perfusion, and MR perfusion and CT (P<0.001).
Conclusion Among the 3 acquisitions, HT was identified more often on SWI/GRE than on CT or MR perfusion. For HT more likely to be clinically significant (Heidelberg grade 2 or higher), both SWI/GRE and DSC-MR perfusion performed similarly to CT. Thus, when obtaining DSC MR perfusion images for ischemic stroke imaging, SWI/GRE may not be necessary for evaluating HT.
Disclosures C. Reynolds: None. B. Renshaw: None. S. Vedantham: None. S. Rogers: None.