ABC/2 Method Shows Strong Accuracy in Measuring Infarct Volume
The ability of the ABC/2 method to correctly identify patients with an infarct volume below a pre-defined cut-point proved to be very high for various cut-points.
Robert Fahed, MD
Newly published data demonstrated that Diffusion Weighted Imaging (DWI) infarct volume measurement using ABC/2 method is accurate, reliable, and may be an acceptable alternative to RAPID software for thrombectomy decision-making.
Senior author Robert Fahed, MD, neuroradiology department, Rothschild Foundation Hospital, and colleagues found a >80% accuracy with the ABC/2 method for each rater and each volume cut-point.
Several infarct volume cut-points used in recent thrombectomy trials were used to calculate accuracy and reliability parameters: <21ml versus ≥21ml, <31ml versus ≥31ml, <51ml versus ≥51ml, and >70ml versus ≥70 ml.
The ABC/2 method is a validated tool for the assessment of intracranial hemorrhage volume and has been previously observed to measure infarct volume. The study authors compared the accuracy and reliability of ABC/2 to the automated RAPID software, which they claim has become the main imaging criterion for patient selection for thrombectomy beyond 6 hours from the time last known well (LKW).
READ MORE: SB623 Mesenchymal Stem Cell Implantation Efficacious and Well-Tolerated in TBI
Four physicians were blinded to the RAPID results, and measured DWI infarct volume using the ABC/2 method. Among a cohort of 105 patients (61% male; mean age, 72.4 years [±14.1]) presenting with late or unknown onset of stroke, researchers found no major variation in the accuracy parameters across the different pre-determined volume cut-points.
The group and subgroup validation parameters were calculated using mean sensitivity, specificity, and accuracy. Inter-rater and intra-rater agreement were measured using Fleiss’ kappa statistics with 95% bias-corrected confidence intervals obtained by 1000 bootstrap resampling. Inter-rater agreement was substantial-to-excellent for all cut-points, and the proportion of cases with perfect agreement among all raters varied between 66.7% and 81.9%, depending on the cut-point.
All raters included in the study were clinicians involved in the management of patients with acute ischemic stroke (AIS), routinely using brain magnetic resonance imaging (MRI). Among the 4 raters who participated, 2 of them were vascular neurologists and 2 interventional neuroradiologists.
Intra-rater agreement was substantial for both raters at each cut-point, except for the <70 ml threshold where only 1 rater showed only moderate intrarater agreement. Notably, depending on the cut-point, the raters’ dichotomized infarct volume measurement shifted between both readings from 6.25% to 18.75% of the time.
Fahed and colleagues also explored whether judgments about candidacy for thrombectomy based on ABC/2 would agree with similar judgments based on RAPID calculations of infarct volume. Thrombectomy decisions based on RAPID and raters’ measurements were compared using the DAWN trial criteria.
Using this trial criteria, researchers found that the thrombectomy decision would have led to treatment of 47.6% of patients (50 of 105) with RAPID, and 50.5% of patients (53 of 105) to 62.8% of patients (66 of 105) with the ABC/2 method, depending on the rater.
Also depending on the rater, contradictory decisions occurred in 12.4% (13 of 105) to 15.2% (16 of 105) for a mean of 14.3%. Contradictory decisions were mainly thought to be related to over-treatment because of infarct volume underestimation below the adequate cut-point.
The study authors found that thrombectomy would have been erroneously refused in only 9 of 420 (2.1%) cases with the ABC/2 method.
There was a substantial inter-rater agreement for all raters (K = 0.799; 95% CI 0.721–0.877), with perfect agreement among all 4 raters occurring in 81.9% of cases. Additionally, inter-rater agreement was excellent for both raters (K = 0.808; 95% CI, 0.601–1.000; and 0.811; 95% CI, 0.607–1.000, respectively).
Change in the thrombectomy decision due to rater variations between both readings would have occurred in 9.4% of cases. “More studies are needed to assess the performance of the ABC/2 method on CTP or for the application of other trials’ selection criteria,” the study authors concluded.
