1 Hemorrhagic Transformation of Ischemic Stroke: Perfusion CT-Based PredictionRichard Aviv, Christopher d’Esterre, Blake Murphy, et al (especially TY Lee) Sunnybrook Health Sciences Centre, Toronto, Ontario Lawson & Robarts Research Institutes, London, Ontario Radiology. 2009;250: 1

2 BACKGROUND: HEMORRHAGE INTO INFARCTIONhemorrhagic transformation (HT): hemorrhagic infarction (HI), petechiae, parenchymal hematoma (PH) complication of thrombolysis of embolic stroke long ischemia affects BBB, vascular permeability permeability surface (PS) product gives contrast leakage rate; vascular lumen to interstitial space1 permeability techniques with CT-P in animal tumor models and limited human cases2 1 Lee TY, AJNR 21: (2000) Lin K, AJNR 28: (2007) 2

3 BACKGROUND OBJECTIVES BENEFITshow whether acute stroke patients who suffer HT have CTP- derived PS (permeability surface) product compared to those without HT BENEFIT help decision for lysis treatment beyond time windows by increasing HT predictability Verify if acute stroke patients who suffer hemorrhagic transformation have increased CT-P derived Permeability Surface Product, compared to those who do not bleed. Benefit could be to extend lysis treatment beyond accepted time windows by predicting hemorrhagic transformation. 3

4 METHODS prospective study exclusion criteria74 patients inclusion criteria - assessed < 3 hrs - IV tPA, no tPA exclusion criteria sent back before full CT protocol (26pts) contraindication to contrast or MRI (1pt) motion between phases (6pts) Included irrespective of thrombolysis Remove inclusion + exclusion??? hemorrhagic (n = 23) hemorrhage on delayed scan types: PH1, PH2, HI1, HI2 non–hemorrhagic (n = 18) infarct only on delayed scan 4

5 METHODS IMAGING CTA - 0.7ml/kg contrast; arch to vertex; 120kVp, 270mAadmission – hyperacute CT stroke series; VCT, GE NCCT -120 kVp, 340 mA, 5mm axial slices, 5mm gap CTA ml/kg contrast; arch to vertex; 120kVp, 270mA 2-phase CTP - 0.5ml/kg contrast; basal ganglia to ventricles; 80kVp, 190mA - 1st phase: 45s, 0.5s/90 images/8 slices. - 2nd phase: 90s/same 8 slices 15s/ 6s, reliable) 5 – 7 days post - follow up NCCT +/- MR for infarct size, potential HT Upon arrival at the ER, patients have CT, CTA, and CT-P. CT-P is prolonged for additional permeability slices and a longer time. Further imaging is done to determine infarct size and bleeding. 5

6 METHODS IMAGE ANALYSIS CT Perfusion 4 (GE) software, function’l maps2nd phase CT Perfusion 4 (GE) software, function’l maps maps of deconvolution, arterial & venous inputs, anterior cerebral artery & sup. sagittal sinus PS product calculated data, acquired during 2nd phase, based on Johnson & Wilson model for capillary exchange 1 Q(t) Ca(t) time DECONVOLUTION Standard CTP maps are done plus permeability surface product map. High Low 1 Am J Physiol 210: , 1966 CBF CBV PS 6

7 METHODS bidirectional model.Johnson and Wilson Model1 bidirectional model. 2nd phase acquisition gives adequate time for diffusion of tracer to extravascular space. allows multiple, simultaneous calculations of parametric values (CBF, CBV, MTT, PS,). Extravascular space What is the other method?? Validated with microspheres (gold standard) Tracer concentration (Cb) is graded throughout vessel lumen and determined by both space (X) and time (t). Tracer free to diffuse back into lumen based on its inter-compartment concentration gradient. 3. Concentration (Ce) in EVS depends on time (t) only. Patlak -assumes tracer does not diffuse back into intravascular PS FCa(t) FCv(t) Intravascular space X 1 Am J Physiol 210: , 1966 7

8 METHODS IMAGE ANALYSIS blinded PS analysis, reduced biasCBF≤ 25 ml-1min-1(100g)-1 threshold for perfusion maps CBF ROI superimposed on co-registered first CBV, PS functional maps using IDL software, RSI Inc. ROIs reflected about midline CBF, CBV vessel thresholds applied: CBF < 100ml/min-1 100g-1 , CBV <8ml/100g 1 3 ROIs done registering PS map with CT-P maps. Avoided lateral ventricles and choroid plexi. 1 TY Lee, Stroke Jul;37(7): PS 8

9 RESULTS NIHSS (Median, IQR) 19 (13-20) 10 (7-15) 0.005Hemorrhage N=23 No Hemorrhage N=18 P Value Female Gender number (%) 10 (43%) 5 (28%) 0.3 Age (median, IQR) 74 (58-83) 72 (64-81) 0.67 MAP >120 number (%) 7 (30%) 2 (11%) 0.25 INR (mean ± SD) 1.04 (0.10) 1.02 (0.07) 0.99 NIHSS (Median, IQR) 19 (13-20) 10 (7-15) 0.005 ASPECTS (Median, IQR) 6 (6-9) 10 (8-10) 0.02 Time to scan (mins) (mean ± SD) 124±765 125±118 0.37 CBF volume cm3 (Mean ± SD) 63±31 52±21 0.17 CBV volume cm3 (Mean ± SD) 48±24 31±13 0.11 TPA given number (%) 17 (74%) Time to TPA (mins) (mean ± SD) 143 (27) 135 (31) 0.65 TPA dose (mg) (mean ± SD) 62±13 57±10 0.43 90 day mRS (Median, IQR) 4 (2-6) 1 (0-2) 0.001 There was some correlation depending on high NIHSS, lower ASPECT scores, whether lysis was done, and outcome Rankin score. There was no difference of time to treatment or tPa dose. 9

10 RESULTS perfusion defects in HT vs no HT patients hemorrhagic patientsCBF hemorrhagic patients non-hemorrhagic patients CBV no signif. diff. mean CBF, CBV, MTT of hemorr’gic regions in patients with hemorrhage vs. no hemorrhage No significant difference comparing patients with bleeding and those without, although there is a trend of less prolonged MMT for the non-bleeders. MTT data shown as mean ± standard error 10

11 ischemic regions in hemorrhage patients vs non hemorrhage patientssignif. diff. (p < 0.01) of mean PS of bleeding regions in those with hemorrhage vs. ischemic regions without hemorrhage Difference of Permeability Surface Product for ischemic regions in patient’s with and without hemorrhage. data as median, 95% C.I. and Q1-3 0.49±0.3 vs 0.09  0.19 ml-1min-1(100g)-1 11

12 RESULTS petechiae vs. hematomano signif. diff. between mean PS for hematoma & petechiae patients For this study, frank hematoma were combined as hemorrhagic transformation, and are shown separately here. data as median, 95% C.I. and Q1-3 12

13 RESULTS univariate analysis for hemorrhageparameter p-value odds ratio lower 95% confidence limit for odds ratio upper 95% confidence PS 0.001 1.01 1.02 Aspects Score 0.47 0.30 0.75 Initial NIHSS at p 1.15 1.03 1.29 TPA given 0.007 6.80 1.68 27.52 Here are results for main factors of Permeability Score, Aspects Score, initial NINSS, and whether treated with by lysis. stepwise multi-variate analysis PS (OR 3.5; 95% CI ; p=0.0007) ASPECTS (OR 0.4, 95% CI ; p=0.002) 13

14 67 M, NIHSS 16 ASPECTS 6 69mgTPA@143mins PS= 0.05150 67 M, NIHSS 16 ASPECTS 6 PS= 0.05 CDID 1068, MRN 14

15 57 F, NIHSS 11 ASPECTS 9 48mgTPA @120mins PS= 0.3 15

16 RESULTS receiver operator curve analysis of PS for HT predictionPS≥0.23 ml-1min -1(100g)-1 AUC 0.92±0.05 (CI ) sensitivity 77% (CI 55-92%) specificity 94% (CI %) PPV 94% (CI %) NPV 75% (CI 60-90%) accuracy: 87% ( %) 16

17 CONCLUSION in hyperacute stroke, microvascular permeability is detected using 2-phase CTP. permeability is common in cerebral infarcts progressing to HT. further prospective investigations needed to explore ability to predict HT in different hemorrhage cohorts 17