Background An arbitrary perfusion imaging “mismatch” ratio (total perfusion defect : irreversibly damaged “core” volumes) of >1.2 has been widely utilised in research to select patients for reperfusion therapies in acute ischaemic stroke. MRI studies suggest that a higher ratio may define a greater differential treatment response. We evaluated whether CTP mismatch ratio provided additional prognostic information compared to clinical examination and appearances on non–contrast CT (NCCT).

Methods Patients recruited to two sub–6 hour prospective observational studies of ischaemic stroke who received intravenous thrombolysis (rtPA) and who had a deficit on acute CTP imaging were included for analysis. CTp measured perfusion deficit was defined by delay time >2s, and core defined by delay time >2s plus relative cerebral blood flow <40% (Bivard et al, Brain, 2011). Post–processing was blind to follow up imaging and clinical variables. Volumetric mismatch was calculated as the ratio of total volume of perfusion defect:core volume. Clinico–radiological features were assessed for their predictive value for major neurological improvement (reduction between admission and 24h National Institutes of Health Stroke Scale [NIHSS] of ≥8 or 24 hour score of ≤1) and poor outcome (30 day modified Rankin Scale of 5–6). The following predictors were compared using univariate logistic regression: admission NIHSS, NCCT Alberta Stroke Program Early CT Score (ASPECTS), core volume, penumbral volume (total perfusion deficit minus core volume), and CTP mismatch ratio. Significant predictors were explored with receiver operating characteristic curves and dichotomised by their optimal cut points for multivariate regression. Recursive partitioning was used to explore the prognostic utility of considering >1 predictor where multiple predictors were significant on regression. Analysis was repeated for the subset of subjects who recanalised.

Results 61 subjects were included: median admission NIHSS=13, median onset–to–CT time=160 minutes. Twenty subjects had confirmed recanalisation on 24–72h CT angiography. No predictor was significantly associated with major neurological improvement, with CTP mismatch performing best (p=0.078). On expansion of this outcome to include subjects with a 24 hour NIHSS score of ≤2 or 8 or more points improvement, only CTP mismatch ratio was a significant predictor (Optimal cut point: >3.4; Positive Predictive Value [PPV]: 45; Negative Predictive Value [NPV]: 88.5). Mismatch ratio (Optimal cut point: <3.05; PPV: 41.7, NPV: 94.1) and admission NIHSS (Optimal cut point: >12; PPV: 33.3; NPV:92.9) predicted poor outcome and were independently significant on multivariate regression. The use of NIHSS and mismatch ratio thresholds together substantially increased PPV (PPV: 60; NPV: 93). No factor was a significant predictor of major neurological improvement among recanalised subjects, but mismatch ratio performed best (p=0.063) on univariate regression and strongly on ROC analysis (AUC=0.823, cut point>3.4) [Figure 1].

Conclusions Acute CTP mismatch ratio predicted early neurological improvement after IV rtPA treatment, with large mismatch ratios yielding better prediction of favourable early response. The combination of NIHSS and CTP mismatch may improve prediction of poor outcome. Since all patients received rtPA, no conclusion can be made about the interaction between imaging variables and rtPA.