Article Text
Abstract
Background and purpose Optimal thresholds for ischemic penumbra detected by CT perfusion (CTP) in patients with acute ischemic stroke (AIS) have not been elucidated. In this study we investigated optimal thresholds for salvageable ischemic penumbra and the risk of hemorrhagic transformation (HT).
Methods A total of 156 consecutive patients with AIS treated with mechanical thrombectomy (MT) at our hospital were enrolled. Absolute (a) and relative (r) CTP parameters including cerebral blood flow (aCBF and rCBF), cerebral blood volume (aCBV and rCBV), and mean transit time (aMTT and rMTT) were evaluated for their value in detecting ischemic penumbra in each of seven arbitrary regions of interest defined by the major supplying blood vessel. Optimal thresholds were calculated by performing receiver operating characteristic curve analysis in 47 patients who achieved Thrombolysis In Cerebral Infarction (TICI) grade 3 recanalization. The risk of HT after MT was evaluated in 101 patients who achieved TICI grade 2b–3 recanalization.
Results Absolute CTP parameters for distinguishing ischemic penumbra from ischemic core were as follows: aCBF, 27.8 mL/100 g/min (area under the curve 0.82); aCBV, 2.1 mL/100 g (0.75); and aMTT, 7.30 s (0.70). Relative CTP parameters were as follows: rCBF, 0.62 (0.81); rCBV, 0.83 (0.87); and rMTT, 1.61 (0.73). CBF was significantly lower in areas of HT than in areas of infarction (aCBF, p<0.01; rCBF, p<0.001).
Conclusions CTP may be able to predict treatable ischemic penumbra and the risk of HT after MT in patients with AIS.
- Blood Flow
- CT perfusion
- Hemorrhage
- Stroke
- Thrombectomy
Statistics from Altmetric.com
Introduction
According to recent reports of rapid and efficient endovascular treatment techniques, rates of successful therapeutic reperfusion are gradually increasing.1–7 Reperfusion therapy for patients with acute ischemic stroke (AIS) serves to rescue the hypoxic, yet potentially salvageable, ischemic penumbra.8 CT perfusion (CTP) has the potential to delineate areas of salvageable penumbra from ischemic core based on cerebral blood flow (CBF), cerebral blood volume (CBV), and mean transit time (MTT).9–11 Ischemic penumbra is tissue with low CBF but high or preserved CBV (CBF/CBV mismatch), whereas ischemic core is defined as tissue with reductions in both CBF and CBV (CBF/CBV match).11 However, optimal thresholds for these parameters are not yet known because data have been derived from patients with intravenous thrombolysis therapy, and timing for complete recanalization has been unclear. Moreover, CTP varies depending on the model and program,12 and fully automated software such as RAPID is not available in Japan.13 14 As such, it is necessary to understand the optimal thresholds for each CTP model in order to demonstrate the validity of the therapeutic indication.
In addition, treatment of acute-phase cerebral ischemia is accompanied by a risk of hemorrhagic complications such as intracerebral hemorrhage. Predicting the risk of hemorrhagic transformation (HT) by performing imaging studies before mechanical thrombectomy (MT) is important in determining treatment indications and strategies.15–17 In this study of a group of patients who underwent MT at our hospital we investigated optimal thresholds for CTP parameters that discriminate ischemic penumbra from ischemic core. We also examined the relationships between CTP parameters and risk of HT.
Methods
Ethics statement
This study was approved by the Research Ethics Committee of Baba Memorial Hospital; the procedures followed were in accordance with institutional guidelines. Written informed consent was obtained on admission from each patient or the patient’s legal representative.
Patients
One-hundred and fifty-six consecutive patients with AIS were treated with MT at the hospital from November 2011 to December 2014. Conventional non-contrast CT (NCCT) was performed within 6 hours of symptom onset to rule out intracranial hemorrhage before MT. CT angiography (CTA) and four slices (levels) of CTP were performed immediately after NCCT. Patients were administered intravenous recombinant tissue plasminogen activator (rtPA) before MT if time from symptom onset was within 4.5 hours and no contraindications were present. At our hospital we use the Penumbra system (Penumbra, Oakland, California, USA) as the first choice, but also use other devices such as stent retriever Solitaire FR(ev3 Endovascular, Plymouth, Minnesota, USA) or Trevo ProVue (Stryker Neurovascular, Kalamazoo Michigan, USA), depending on the situation, with the aim of complete reperfusion. Thrombolysis In Cerebral Infarction (TICI) grade, time from onset to recanalization, and time from CTP to recanalization were recorded after MT. Of these consecutive patients with AIS treated with MT, those who achieved TICI 2b–3 reperfusion had strictly controlled systolic blood pressure to <130 mm Hg immediately after reperfusion for at least 48 hours. Inclusion criteria for the present study were as follows: (1) age >20 years; (2) intake within 8 hours from symptom onset (if symptom onset was uncertain, we defined the last time of asymptomatic status as symptom onset); (3) anterior circulation or posterior cerebral artery occlusion on admission; and (4) National Institutes of Health Stroke Scale (NIHSS) score >4. Exclusion criteria were as follows: (1) use of variant algorithms for CTP analysis in early cases (13 patients); (2) history of large cerebral infarction (3 patients); (3) basilar or vertebral artery occlusion (18 patients); and (4) TICI grade 0–2a after recanalization (24 patients). Clinical data recorded included age, sex, and preoperative NIHSS score. Baseline characteristics of the study population are given in table 1.
Imaging protocol
CT was performed using a 64-slice CT scanner (Aquilion CX, TSX-101A/NA; Toshiba, Tokyo, Japan). CTA was performed using 60 mL of contrast medium (Oypalomin 300; Konica Minolta, Tokyo, Japan) infused at a rate of 3–5 mL/s, with scanning started manually once the contrast medium reached the common carotid artery. Following this, CTP was performed at a rate of 5 mL/s via automated antecubital venous injection using the following parameters: 120 kV tube voltage, 50 mA tube current, and 1.0 s rotation. Arterial input was obtained from the anterior cerebral artery, while venous output was obtained by selecting the posterior area of the superior sagittal sinus. Images obtained by CTP were processed using the delay-sensitive Box-modulation transfer function (Box-MTF) method. Perfusion maps showing average CBF, CBV, and MTT were reconstructed. CTP imaging and perfusion map reconstruction required an average of 6 min. The slice selected for CTP was at the level of the third ventricle and basal ganglia. According to blood supply, seven arbitrary regions of interest (ROIs) were set on the affected hemisphere. Then, absolute CTP parameters (aCBF, aCBV, and aMTT) were recorded for each ROI in the ipsilateral (affected) hemisphere. Because absolute values are expected to vary due to the administration method of the contrast medium and circulation dynamics, relative values were also evaluated. Thus, values for the mirror-imaged ROIs in the contralateral (normal) hemisphere were recorded to permit calculation of relative CTP parameters (rCBF, rCBV, and rMTT). Relative CTP parameters are presumed to evaluate corrected values between individual differences compared with absolute values. These were calculated as the ipsilateral value divided by the contralateral value (figure 1).
Data analysis
In order to define the ischemic area, two experienced neurosurgeons with over 10 years of stroke experience defined both ischemic and non-ischemic areas in each ROI based on a 10% CBF decrease or 10% MTT increase, compared with the contralateral side, using CTP images obtained on admission(figure 2). These ischemic areas were reconfirmed by verifying occluded vessels on CTA or angiography. After MT, patients were stratified according to TICI grade. Moreover, each ROI in an ischemic area was subdivided into infarction and non-infarction areas using NCCT or MRI follow-up results within 7 days from symptom onset (figure 2). NCCT images obtained 24–36 hours after intervention were used to assess for areas of HT.16 Intracranial hemorrhage was classified as hemorrhagic infarction (HI) or parenchymal hemorrhage (PH) according to the Safe Implementation of Thrombolysis in Stroke-Monitoring Study definition.18
Statistical analysis
Statistical analysis was performed using JMP V.11.2.0 (SAS Institute, Cary, North Carolina, USA). Mean CTP parameters in non-ischemic, infarction, and non-infarction areas were analysed by performing one-way analysis of variance followed by multiple comparisons using the Tukey–Kramer method (figure 2). CTP parameter histograms were created for the ROIs pooled across all patients in each group. Receiver operating characteristic (ROC) curves were calculated for the TICI grade 3 group using Youden’s method for post-therapeutic cerebral infarction to determine optimal thresholds for predicting the presence of infarction at follow-up (figure 2). Mean CTP parameters in infarction and hemorrhagic areas were analysed using Welch’s t-test (figure 3).
Results
Ischemic core and ischemic penumbra
Of 156 consecutive patients treated with MT, 47 achieved TICI grade 3 recanalization and were included for analysis. Of 329 areas, 140 were non-ischemic, and 189 were ischemic (figure 2). CTP parameters in the ischemic area featured lower a/rCBF in both non-infarction and infarction areas compared with those of the non-ischemic area (aCBF, p<0.001; rCBF, p<0.001). CTP parameters in the ischemic area also featured higher a/rMTT in both non-infarction and infarction areas compared with those of the non-ischemic area (aMTT, p<0.001; rMTT, p<0.001). However, both a/rCBF and a/rMTT showed no significant differences between non-infarction and infarction areas after MT. Therefore, neither CBF nor MTT could differentiate between ischemic core and ischemic penumbra before MT in the present cohort. On the other hand, both aCBV and rCBV in infarction areas were lower than those in non-infarction areas (aCBV, p<0.001; rCBV, p<0.001), although these parameters showed no significant differences when comparing non-infarction with non-ischemic areas. Thus, these results reflect the fact that ischemic penumbra was defined as the area in which CBV was preserved but CBF was reduced or MTT was elevated. According to ROC analysis of the TICI 3 recanalization group, absolute CTP parameters for distinguishing ischemic penumbra from ischemic core were as follows: aCBF, 27.8 mL/100 g/min (area under the curve (AUC) 0.82); aCBV, 2.1 mL/100 g (0.75); and aMTT, 7.3 s (0.70). Relative CTP parameters were as follows: rCBF, 0.62 (0.81); rCBV, 0.83 (0.87); and rMTT, 1.61 (0.73) (figure 2).
HT
Among the 101 patients who achieved TICI grade 2b–3 recanalization, HT was found in 15 areas in 15 patients (14.9%). Of these areas, 13 were evaluated (excluding 2 patients with HT in areas other than a ROI), which revealed HI grade 1 (2 patients), HI grade 2 (5 patients), PH grade 1 (2 patients), and PH grade 2 (4 patients). Symptomatic intracerebral hemorrhage, defined as a decrease in NIHSS score of at least four points due to intracerebral hemorrhage, was observed in four patients. There were significant decreases in aCBF (p=0.0079) and rCBF (p=0.0002) in areas where HT occurred compared with areas where cerebral infarction occurred. On the other hand, there were no significant differences between infarction and HT areas in both a/rCBF and a/rMTT (figure 3).
Discussion
In this study we estimated the CTP parameters on admission and the corresponding tissue fate, such as cerebral infarction or HT, based on data of patients who achieved reperfusion by MT. Our findings showed that, in areas where CBF was reduced but CBV was preserved, cerebral infarction was avoided by complete reperfusion. On the other hand, in areas where both CBF and CBV were reduced, cerebral infarction was completed even after reperfusion was achieved. Therefore, as reported previously,11 we confirmed that the combination of CBF and CBV could define ischemic penumbra and ischemic core. We also determined the optimal parameter thresholds for CTP parameters for discriminating ischemic penumbra from ischemic core. Furthermore, preoperative CBF in areas of HT after MT was found to be lower than that in areas of cerebral infarction.
In previous reports the ROIs were set in high-diffusion-weighted intensity areas after thrombolytic therapy, while optimal thresholds for distinguishing ischemic core were obtained by comparing CTP data with the final infarction area.10–12 On the other hand, we set the ROIs according to blood supply before MT. Previous research of 25 patients treated with thrombolytic therapy proposed that aCBV of 2.0 mL/100 g was the optimal parameter to determine infarction core.10 From our data of 47 patients treated with MT, the optimal threshold for aCBV in ischemic core was 2.1 mL/100 g, which is equivalent to the previous report. Campbell et al pointed out, from their study of 54 consecutive patients, that CBF also can be used as an optimal parameter.19 Similarly, from our data, a/rCBF showed a high area under the curve in ROC analysis, suggesting that CBF could be a useful indicator of ischemic penumbra. Values of both aCBF (27.8 mL/100 g/min) and rCBF (0.62) in our data are higher than those reported previously.19 This is mainly due to the degree of smoothing and the specific deconvolution algorithm used, and the reliability of the threshold is not impaired. From these results, in setting the ROIs according to blood supply, it was shown that the optimal CTP threshold for ischemic penumbra can be useful to determine a treatment strategy for MT. Semiquantitative evaluation is a more universal indicator considering differences such as degenerative lesions in white matter (leukoaraiosis).19 Our data also showed that rCBV was a reliable parameter to distinguish ischemic penumbra in AIS. However, the optimal threshold for rCBV for ischemic penumbra from our data (0.83) was higher than that in a previous report (about 0.60).19 Therefore, it is suggested that the therapeutic indications obtained from rCBV in patients with cerebral artery occlusion in AIS may have a narrow range. Separate evaluation of optimal thresholds in the perforating area (figure 1: T, C) and cortical and subcortical areas (figure 1: A, M1–3, P) revealed that the perforating area showed slightly higher thresholds (aCBV, 2.0 mL/100 g (AUC 0.75); rCBV, 0.86 (0.85)) for ischemic core compared with cortical and subcortical areas (aCBV, 1.60 mL/100 g (AUC 0.72); rCBV, 0.81 (0.86)). Therefore, the basal ganglia and thalamus were expected to have lower ischemic tolerance compared with cortical and subcortical areas. However, we were unable to identify the reason for the difference in optimal threshold for rCBV compared with previous reports. This difference is presumed to be caused by differences in the patient population, especially the inclusion of elderly patients in the present cohort. We identified optimal thresholds for distinguishing ischemic penumbra from ischemic core, but there was a region where reperfusion could avoid cerebral infarction even in areas below the optimal threshold. Therefore, if the risk of thrombectomy is not high, the cut-off CBV value for therapeutic indication may be less than the optimal threshold.
As mentioned previously, CTP imaging obtained by administering a contrast agent has a problem of variance between commercial software.12 According to data from Kudo et al, the variations in rCBV between commercial software are smaller than the variations in rCBF and rMTT.12 The Box-MTF deconvolution algorithm is sensitive to tracer delay. In particular, it has been pointed out that the ischemic region is overestimated. However, if an appropriate threshold is selected, delay-sensitive and delay-insensitive deconvolutions should provide similar ischemic core and ischemic penumbra volumes.9 Therefore, the Box-MTF method can be useful for therapeutic decision making for patients with AIS.
Tissue prognosis after reperfusion in one ischemic region has diversity, such as infarction, non-infarction, and bleeding. Therefore, it is difficult to accurately evaluate the relationships between CTP parameters and post-therapeutic HT in one ischemic region. Since our ROIs were decided along with the circulation regions, it is advantageous to accurately examine the relationships between tissue prognosis and CTP parameters. There are reports that HT has occurred in areas with CBF less than 13 mL/100 g/min.20 In our study of 13 cases, HT areas, compared with infarction areas, demonstrated lower average a/rCBF. Hemorrhagic complications are expected to occur in reperfusion regions where collateral circulation is not developed, such as where CBF is strongly decreased or in regions where infarction is completed due to ischemia. It is thought that HT may be caused by disruption of the blood–brain barrier due to severe CBF reduction. Importantly, it appears that the degree of anatomical and physiological disruption of the blood–brain barrier is highly dependent on the duration of ischemia.21 22 However, a recent meta-analysis of patients with large vessel ischemic stroke, symptomatic hemorrhage, and major PH showed no interactions of time in the MT group.23 Although there is a difference in degree between HT and major PH or symptomatic hemorrhage, the degree of ischemia (ie, the extent of collateral circulation) may have a stronger influence on HT after MT than time from onset to reperfusion in patients with AIS.
The incidence of asymptomatic and symptomatic intracerebral hemorrhage has been reported to vary from 13.4% to 40.9% and from 2.0% to 11.2%, respectively, in patients with AIS treated with MT.17 24–27 The 14.9% incidence of HT and 3% incidence of symptomatic intracerebral hemorrhage observed in our study fall within these ranges. In our hospital, patients with an extensive area of CBF/CBV match in the ischemic area are excluded from mechanical thrombus retrieval therapy. We believe that patient selection using CTP may be one of the reasons for the reduced incidence of symptomatic intracerebral hemorrhage in this cohort. Regarding the MT therapeutic strategy, evidence relating time and tissue fate is scarce.28 Some studies have shown that endovascular treatment of AIS might be performed safely with no time limit in appropriately selected patients, based on the presence of perfusion parameter mismatch as revealed by CTP or MRI.28–30 Thus, accurate information on the relationships between CTP parameter thresholds and recanalization time may provide us with a clinical indication for MT recanalization in cases of unknown-onset stroke. A need exists to clarify the relationship between optimal threshold of a CTP parameter and therapeutic time.
Identification of ischemic penumbra based on CBF/CBV or MTT/CBV mismatch using CTP on admission and prediction of HT played an important role in selection of our short-term appropriate treatment strategy and judgment of patient selection for MT at our hospital. However, our single-centre retrospective study has several limitations. First, because we evaluated CTP parameters from one brain slice through the basal ganglia, we could not evaluate total volume of ischemic penumbra or ischemic core accurately. Second, we excluded patients from undergoing MT who showed extensive early CT signs or a decrease in CBV in the entire ischemic area. Therefore, this study admittedly has some potential patient selection bias. Third, it is important to bear in mind that we did not evaluate the relationships between CTP parameters and patients’ clinical outcomes; a single ROI was the unit of analysis, not a whole patient. To overcome these limitations, we could derive the relationship between estimated tissue fate in the whole brain and time from imaging to reperfusion in a future investigation using CTP threshold information from the present ROI-based CTP analysis.
Conclusions
We evaluated CTP data of patients with AIS treated with MT and assessed the usefulness of CTP in patient selection for MT. We found that CTP has implications for predicting treatable areas and the risk of HT for MT. These data provide us with indications and contraindications for the therapeutic strategy of MT in these patients.
Acknowledgments
We would like to thank Editage (www.editage.jp) for English language editing.
References
Footnotes
Contributors KK: Substantial contributions to the conception or design of the work; the acquisition, analysis, or interpretation of data; drafting of the manuscript or revising it critically for important intellectual content; final approval of the version to be published; agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. JU: Substantial contributions to the acquisition, analysis, or interpretation of data; editing of the manuscript; final approval of the version to be published. RO, RN, SN, KM, YI: Substantial contributions to the acquisition of data, final approval of the version to be published. HG: Supervised the project; agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Competing interests None declared.
Ethics approval Research Ethics Committee of Baba Memorial Hospital.
Provenance and peer review Not commissioned; externally peer reviewed.