*Article* **Low Hypoperfusion Intensity Ratio Is Associated with a Favorable Outcome Even in Large Ischemic Core and Delayed Recanalization Time**

**Jang-Hyun Baek 1,2, Young Dae Kim 2,3, Ki Jeong Lee 2,4, Jin Kyo Choi 2,5, Minyoul Baik 2, Byung Moon Kim 6, Dong Joon Kim 6, Ji Hoe Heo 2,3 and Hyo Suk Nam 2,3,\***


**Abstract:** In ischemic brain tissue, hypoperfusion severity can be assessed using the hypoperfusion intensity ratio (HIR). We evaluated the link between HIR and clinical outcomes after successful recanalization by endovascular treatment. We retrospectively reviewed 162 consecutive patients who underwent endovascular treatment for intracranial large vessel occlusion. The HIR was calculated using an automated software program, with initial computed tomography perfusion images. The HIR was compared between patients with and without favorable outcomes. To observe the modifying effect of the HIR on the well-known major outcome determinants, regression analyses were performed in the low and high HIR groups. The median HIR value was significantly lower in patients with a favorable outcome, with an optimal cut-off point of 0.54. The HIR was an independent factor for a favorable outcome in a specific multivariable model and was significantly correlated with the Alberta Stroke Program Early Computed Tomography Score (ASPECTS). In contrast to the high HIR group, the low HIR group showed that ASPECTS and onset-to-recanalization time were not independently associated with a favorable outcome. Finally, the low HIR group had a more favorable outcome even in cases with an unfavorable ASPECTS and onset-to-recanalization time. The HIR could be useful in predicting outcomes after successful recanalization.

**Keywords:** hypoperfusion; collaterality; stroke; outcome; thrombectomy

#### **1. Introduction**

Hypoperfusion severity and duration are important factors affecting the clinical outcome of patients with acute ischemic stroke who undergo endovascular treatment (EVT). Collateral status is a commonly used method that reflects hypoperfusion severity. Robust collateral flow is associated with smaller ischemic core lesions and slower progression, which may lead to improved clinical outcomes [1–4]. Moreover, time windows for EVT eligibility can be determined based on the collateral status. Patients with better collateral flow may have a more favorable outcome, even in cases in which recanalization is delayed [5].

Hypoperfusion severity can be assessed directly using the hypoperfusion intensity ratio (HIR) [6]. The HIR reflects the proportion of the critically hypoperfused lesion (Tmax > 10 s) in the whole hypoperfused lesion (e.g., Tmax > 6 s) on perfusion images [7,8].

**Citation:** Baek, J.-H.; Kim, Y.D.; Lee, K.J.; Choi, J.K.; Baik, M.; Kim, B.M.; Kim, D.J.; Heo, J.H.; Nam, H.S. Low Hypoperfusion Intensity Ratio Is Associated with a Favorable Outcome Even in Large Ischemic Core and Delayed Recanalization Time. *J. Clin. Med.* **2021**, *10*, 1869. https:// doi.org/10.3390/jcm10091869

Academic Editor: Peter Sporns

Received: 18 March 2021 Accepted: 22 April 2021 Published: 26 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

The HIR correlates well with the quality of the collateral status and is considered a quantitative marker of the collateral status [7,9,10]. Furthermore, the HIR has been reported to be significantly associated with initial stroke severity, target mismatch profile, and infarct growth [6–8,11]. Based on the collateral nature of the HIR, we assumed that the clinical outcome after EVT may be affected by the HIR, as in the case of the collateral status.

Accordingly, we hypothesized that a low HIR is associated with a favorable outcome, even in patients with a lower Alberta Stroke Program Early Computed Tomography Score (ASPECTS) or longer onset-to-recanalization time in patients who achieved recanalization through EVT.

#### **2. Materials and Methods**

#### *2.1. Participants*

We retrospectively reviewed consecutive patients with acute stroke who underwent EVT for intracranial large artery occlusion in the anterior circulation between January 2016 and April 2020. All patients had been registered to a prospectively maintained registry of a tertiary stroke center. EVT was considered for patients who met the following criteria: (1) Computed tomography (CT) angiography-confirmed, endovascularly accessible intracranial occlusions associated with neurological symptoms; (2) in the earlier study period, within 8 h from stroke onset; patients within 8–12 h were also considered if they had an ASPECTS ≥ 7; (3) the eligibility criteria of the Diffusion and Perfusion Imaging Evaluation for Understanding Stroke Evolution 3 (DEFUSE 3) and DWI or CTP Assessment with Clinical Mismatch in the Triage of Wake-Up and Late Presenting Strokes Undergoing Neurointervention with Trevo (DAWN) trials were applied to patients within 6–24 h from stroke onset; and (4) initial National Institutes of Health Stroke Scale (NIHSS) score ≥ 4. We also preferably performed EVT for patients with a premorbid modified Rankin Scale (mRS) score ≤ 3. Patients eligible for intravenous tissue plasminogen activator treatment were treated with 0.9 mg/kg tissue plasminogen activator. EVT procedures were performed under local anesthesia. In most cases, a stent retriever was used as the front-line EVT modality and a balloon-guiding catheter was routinely used. The detailed procedure is described elsewhere [12].

We included patients with intracranial large vessel occlusion, which was defined as occlusion of the intracranial internal carotid artery or middle cerebral artery M1 or proximal M2 segment, and those in whom the occlusion was successfully recanalized by EVT. Successful recanalization was defined as modified Thrombolysis in Cerebral Infarction grade 2b or 3.

#### *2.2. Imaging and Clinical Data*

All patients underwent CT as soon as they arrived at the hospital. The pretreatment CT scan included non-contrast CT images, CT angiography, and CT perfusion images. To obtain the HIR, hypoperfusion lesion volumes were quantitatively assessed from CT perfusion images using an automated software (RAPID, iSchemaView, Menlo Park, CA, USA). The lesion volume was measured by a stroke neurologist without any manual correction. The HIR was calculated as the ratio of the lesion volume with Tmax > 10 s to Tmax >6s[7].

Data on all variables used in this study were collected from the prospective registry of patients with acute stroke. The functional outcome was assessed based on the mRS score at 3 months after stroke onset and was primarily evaluated by stroke neurologists during the patient's routine clinical follow-up at 3 months ± 2 weeks. If a patient was unable to present to the follow-up appointment, a stroke neurologist or a trained nurse interviewed the patient or their family via telephone to determine the mRS score using a standard questionnaire. A favorable outcome was defined as an mRS score of 0–2. Intracerebral hemorrhage (ICH) was assessed on follow-up CT or magnetic resonance images obtained 24 ± 6 h after EVT. The assessment of ICH was based on the consensus of stroke neurologists, neurointerventionalists, and neuroradiologists during the regular

stroke conference. The determination of ICH was immediately entered into the prospective registry. ICH was regarded as symptomatic if the patient's NIHSS score increased by ≥ 4 according to the European Cooperative Acute Stroke Study III [13].

The ASPECTS was reassessed by two independent neurointerventionalists and stroke neurologists [14], who only had access to the initial non-contrast CT images and who were completely blinded to any endovascular outcome and clinical information, except for the lesion side. The interrater agreement for the ASPECTS was good (intraclass correlation coefficient, 0.657). Discrepancies in the assessment of cases were resolved by consensus.

#### *2.3. Statistical Analysis*

Based on the study hypotheses, we performed the following analyses step by step. First, to evaluate the association between the HIR and clinical outcomes, patients were assigned to the favorable outcome group or the unfavorable outcome group. Then, (1) HIR values were compared among groups along with patient demographics, risk factors for stroke, clinical and radiological severity of stroke, and symptomatic ICH. Student's *t*-test, the Mann–Whitney *U* test, the chi-square test, and Fisher's exact test were used for group comparisons. (2) We calculated the optimal cut-off point of the HIR for a favorable outcome using the Youden index. Through the receiver operating characteristic curve analysis, the area under the curve was also calculated. (3) To quantify the association between the HIR and a favorable outcome, we performed univariable binary logistic regression analyses for a favorable outcome. Multivariable analysis was also performed to identify whether the HIR was an independent variable for a favorable outcome. Variables with a *p*-value of < 0.1 in univariable analyses were entered in the multivariable model.

Second, to evaluate whether the HIR can modify the effect of the well-established outcome determinants (the ASPECTS and onset-to-recanalization time) on clinical outcomes, the HIR was dichotomized into low and high by its optimal cut-off point for a favorable outcome. According to the low HIR group and the high HIR group, the association between the ASPECTS and a favorable outcome was analyzed by logistic regression analyses. The association between onset-to-recanalization time and a favorable outcome was also analyzed in the same way. We also plotted regression curves by combining the HIR, the ASPECTS, onset-to-recanalization time, and a favorable outcome. For this, patients were assigned to one of the four groups according to the HIR and ASPECTS: (1) the low HIR and small core (ASPECTS ≥ 8) group, (2) the low HIR and large core (ASPECTS < 8) group, (3) high HIR and small core group, and (4) the high HIR and large core group. Regression curves of onset-to-recanalization time for a favorable outcome in the respective groups were compared.

A *p*-value of < 0.05 was considered statistically significant for 95% confidence intervals (CIs). All statistical analyses were performed using R software (version 4.0.1, The R Foundation, r-project.org, Vienna, Austria).

#### **3. Results**

Of the 188 patients who underwent successful recanalization of intracranial large vessel occlusion, perfusion images of 26 (13.9%) were not analyzable due to poor quality, including motion artifacts (*n* = 24) and undetectable Tmax > 6 s perfusion lesions (*n* = 2). A total of 162 patients (mean age, 70.7 ± 12.8 years; male, 51.9%) who met the inclusion criteria were analyzed (Figure 1). The median imaging-to-recanalization time was 126.0 min (interquartile range [IQR], 97.0–153.5). The mean lesion volumes of Tmax > 6 s and Tmax > 10 s were 160.8 mL (IQR, 104.2–205.2) and 80.3 mL (IQR, 30.3–136.0), respectively. The median HIR value was 0.51 (IQR, 0.29–0.68).

**Figure 1.** Patients selection flow chart.

#### *3.1. Association between the HIR and Clinical Outcomes*

Of the patients included, 85 (52.5%) patients had a favorable outcome. The median HIR value was significantly lower in patients with a favorable outcome than those with an unfavorable outcome (0.45 [IQR, 0.15–0.54] vs. 0.60 [IQR, 0.44–0.73]; *p* < 0.001; Table 1). The optimal cut-off point of the HIR for a favorable outcome was 0.54 (sensitivity, 63.6%; specificity, 74.1%). The area under the curve of the HIR to predict a favorable outcome was 0.728 (95% CI, 0.651–0.805; *p* < 0.001).

In univariable logistic regression analysis, the HIR was significantly associated with a favorable outcome (odds ratio [OR], 0.69 per 0.1; 95% CI, 0.59–0.81; *p* < 0.001; Figure 2), although it was not an independent factor for a favorable outcome after adjustment (adjusted OR [aOR], 0.84 per 0.1; 95% CI, 0.68–1.03; *p* = 0.094; Model 1 in Table 2). We found that the HIR was significantly correlated with the ASPECTS (correlation coefficient, −0.49; *p* < 0.001; Figure 3). Considering the collinearity, the HIR was independently associated with a favorable outcome in a model without the ASPECTS (aOR, 0.76 per 0.1; 95% CI, 0.62–0.92; *p* = 0.006; Model 2 in Table 2).


**Table 1.** Comparison of clinical variables and hypoperfusion intensity ratio between patients with and without favorable outcome.

Values in parentheses represent the standard deviation (±) or number of patients (%); brackets represent first and third quartiles, respectively. NIHSS, National Institutes of Health Stroke Scale; tPA, tissue plasminogen activator; ASPECTS, Alberta Stroke Program Early Computed Tomography Score; ICH, intracerebral hemorrhage; rCBF, relative cerebral blood flow.



aOR, adjusted odds ratio; CI, confidence interval; NIHSS, National Institutes of Health Stroke Scale; tPA, tissue plasminogen activator; ASPECTS, Alberta Stroke Program Early Computed Tomography Score.

**Figure 2.** Association between the hypoperfusion intensity ratio and a favorable outcome.

**Figure 3.** Association between hypoperfusion intensity ratio and Alberta Stroke Program Early CT Score (ASPECTS).

#### *3.2. Association between the ASPECTS, Onset-To-Recanalization Time, and Clinical Outcomes According to the HIR*

Based on the optimal cut-off point of the HIR for a favorable outcome, 91 (56.2%) patients were assigned to the low HIR group (HIR < 0.54). The effect of the ASPECTS and onset-to-recanalization time on a favorable outcome was different between the low and high HIR groups. In the low HIR group, only the initial NIHSS score was associated with a favorable outcome. The ASPECTS (aOR, 1.20; 95% CI, 0.92–1.56; *p* = 0.178) and onset-torecanalization time (aOR, 0.97 per 30 min; 95% CI, 0.94–1.01; *p* = 0.194) were not significantly associated with a favorable outcome (Table 3; Figure 4A,B). In contrast, in the high HIR group, the ASPECTS was an independent factor for a favorable outcome (aOR, 1.49; 95% CI, 1.09–2.03; *p* = 0.012). Although onset-to-recanalization time was not independently associated with a favorable outcome (OR, 0.92 per 30 min; 95% CI, 0.79–1.07; *p* = 0.283) in the high HIR group, the probability of a favorable outcome decreased sharply when onset-to-recanalization time was relatively short (Figure 4B). As a whole, 3-dimensional regression planes showed that the probability of a favorable outcome was still above 20% even under the lower ASPECTS and longer onset-to-recanalization time in the low HIR group (Figure 5). In contrast, the combined probability of a favorable outcome was sharply decreased as the ASPECTS and onset-to-recanalization time changed in the high HIR group.


**Table 3.** Effects of the Alberta Stroke Program Early Computed Tomography Score (ASPECTS) and onset-to-recanalization time on a favorable outcome in the low hypoperfusion intensity ratio (HIR) group and the high HIR group.

aOR, adjusted odds ratio; CI, confidence interval; NIHSS, National Institutes of Health Stroke Scale; tPA, tissue plasminogen activator; ASPECTS, Alberta Stroke Program Early Computed Tomography Score.

**Figure 4.** Influence of the Alberta Stroke Program Early CT Score (ASPECTS) and onset-to-recanalization time on clinical outcomes based on the hypoperfusion intensity ratio (HIR). (**A**) Although the ASPECTS was not significantly associated with a favorable outcome in the low HIR group (*p* = 0.178), it was an independent factor for a favorable outcome in the high HIR group (*p* = 0.012). (**B**) For patients with a high HIR, the chances of a favorable outcome sharply decrease when the onset-to-recanalization time is relatively short. (**C**) Considering both the HIR and ASPECTS, patients with a low HIR (<0.54) and small core (ASPECTS ≥ 8) have the best clinical outcome in all ranges of onset-to-recanalization time (*p* < 0.05). In contrast, patients with a high HIR (≥ 0.54) and large core (ASPECTS < 8) have the worst outcome in all ranges of onset-to-recanalization time. Patients with a high HIR and small core have a more favorable outcome than those with a low HIR and large core when onset-to-recanalization time is relatively short. However, for patients with a high HIR and small core, the chance of a favorable outcome decreases more drastically with the course of time and then finally reverses from a particular point of onset-to-recanalization time.

**Figure 5.** The probability of a favorable outcome according to Alberta Stroke Program Early CT Score (ASPECTS) and onset-to-recanalization time in the low hypoperfusion intensity ratio (HIR) group (**A**) and the high HIR group (**B**). (**A**) In the low HIR group, the probability of a favorable outcome was still above 20% even under the lower ASPECTS and longer onset-to-recanalization time. (**B**) The probability of a favorable outcome was sharply decreased as the ASPECTS and onset-to-recanalization time changed in the high HIR group.

Based on the HIR and ASPECTS, 55 (34.0%) patients had a low HIR and small core, 12 (7.4%) patients had a high HIR and small core, 36 (22.2%) patients had a low HIR and large core, and 59 (36.4%) patients had a high HIR and large core. The decreasing trend of a favorable outcome was significantly different between the groups (*p* < 0.05). Patients with a low HIR and small core had the highest probability of a favorable outcome throughout the study period, while patients with a high HIR and large core had the lowest probability. Patients with a high HIR and small core had a more favorable outcome than those with a low HIR and large core when the onset-to-recanalization time was relatively short. However, their clinical outcomes were reversed after the onset-to-recanalization time of 493 min since the probability of a favorable outcome decreased substantially in patients with a high HIR and small core (Figure 4C).

#### **4. Discussion**

In this study, we found that clinical outcomes were substantially different based on the HIR value. A low HIR led to a favorable outcome even in cases with a low ASPECTS and longer onset-to-recanalization time. In contrast, in patients with a high HIR, the ASPECTS was still important for clinical outcome. In addition, the chance of a favorable outcome drastically decreased when the onset-to-recanalization time was relatively short in the high HIR group. Although a study simply reported the association between the HIR and the clinical outcome [7], our study showed more specifically that the probability of a favorable outcome based on the onset-to-recanalization time and the ASPECTS was clearly disparate according to the HIR. The HIR might be an ancillary preprocedural marker to predict the clinical outcomes after EVT.

The collateral nature of HIR was more clearly observed in patients with a high HIR and small core. They had a more favorable outcome than patients with a large core in the shorter onset-to-recanalization time, where the clinical outcome was mainly determined by their core size. However, in the high HIR group, clinical outcome was less favorable despite a small core. Patients with good collaterals could have a favorable outcome even with a longer onset-to-recanalization time; however, the probability of a favorable outcome was substantially reduced in patients with poor collaterals [5]. In addition, the effect of rapid recanalization on a favorable outcome was deliberated in patients with good

collaterals, similar to that seen in the low HIR group in our study. In the same manner, it appears that a low HIR may also have a favorable effect on infarct development and its progression. In a previous study, a high HIR was significantly associated with greater infarct growth [7,15–17]. Our study showed that the HIR alone was an independent factor for a favorable outcome in the multivariable model without considering the ASPECTS. After adjusting the ASPECTS, the HIR was not an independent factor for a favorable outcome, which would be mainly because of tight correlation between the HIR and the ASPECTS.

Specific perfusion deficit volumes—for examples, ischemic core volume (relative cerebral blood flow < 30%) and penumbral volume (Tmax > 6 s) can be associated with clinical outcome. In fact, also in this study, ischemic core volume was independently associated with a favorable outcome, as ASPECTS does. Although such a perfusion parameter is still important in determining clinical outcome, HIR seems still valuable in clinical practice. Along with the common perfusion deficit parameters, HIR could be one of the influencing factors for clinical outcome. Based on its collateral nature, HIR might differentially affect clinical outcome even under the similar ischemic core status or onset-to-recanalization time.

Calculation of the HIR can be clinically advantageous compared to the evaluation of the collateral status, a typical marker reflecting hypoperfusion severity. First, although the assessment of collateral status is easy to perform on pretreatment CT angiography images, quantifying the collateral status are inconsistent and may be qualitative or merely categorical. Most importantly, the evaluation of the collateral status is subjective and reviewer dependent. In contrast, the HIR is more objective in nature. The use of an automated software for the calculation of the HIR can minimize the potential interrater variability and is associated with greater generalization capacity. In addition, the HIR is essentially quantitative as a continuous value, which enables quantitative analysis and can reveal subtle differences in hypoperfusion severity. Furthermore, the HIR can be assessed even in a magnetic resonance-based eligibility strategy. Typical CT angiography-based collateral assessment methods cannot be applied to time-of-flight magnetic resonance angiography. However, the HIR can be calculated with perfusion-weighted images taken using the magnetic resonance technique; thus, its use may be more generalizable. Second, based on the specific cut-off point for a favorable outcome, the HIR may constitute a clinical element that can be used to determine EVT eligibility. In a retrospective study, patients eligible for EVT had a more favorable HIR value than those ineligible for EVT [18]. The HIR may also be a dependable eligibility factor in interfacility transfer since the HIR can significantly predict future infarct growth [8]. According to the recent guidelines for EVT, perfusion imaging with lesion volume analysis has been widely performed to determine the EVT eligibility. Unlike the earlier situation where simple neuroimaging is favored, advanced neuroimaging, including software-based lesion volume analysis, has been increasingly used for EVT eligibility since the DEFUSE 3 and DAWN trials. Although lesion volume analysis is not available in all stroke centers, it is becoming increasingly popular.

In previous studies, a cut-off point of the HIR to predict a favorable functional outcome were not calculated; the cut-off points of the HIR for the collateral status and infarct growth were calculated as 0.40 and 0.50, respectively [7–9]. Most studies merely dichotomized the HIR by its median value, with a range of 0.30–0.45, to evaluate the association of the dichotomized HIR value with the fluid-attenuated inversion recovery (FLAIR) hyperintense vessel sign, infarct growth, or first-pass effect in mechanical thrombectomy [7,19–21]. In these studies, a median HIR value of <0.40 was significantly associated with a positive outcome [7]. In our study, the cut-off point of the HIR for a favorable outcome was calculated as 0.54, which was higher than the 0.40 value, although direct comparison is limited. This may be due to the fact that our study included only patients who underwent successful recanalization. Successful recanalization may give a more favorable clinical outcome in patients with higher HIR values than those with lower HIR values.

This study has several limitations. First, this study was performed retrospectively using prospectively collected data on consecutive patients diagnosed with acute ischemic stroke. The treatment eligibility criteria and protocol were revised during the study period according to the guidelines. Furthermore, the study results from a single center can limit generalizability. Although the HIR value was a rather objective finding obtained from an automated software, the cut-off point of the HIR for a favorable outcome and its significance may be limited to the specific study population. Thus, the results of this study should be interpreted with caution. Second, we evaluated the significance of the HIR only in patients who underwent successful recanalization; however, considering its collateral nature, the HIR may also influence the clinical outcomes in cases of EVT failure, because good collaterals were significantly associated with a favorable outcome even for patients in whom recanalization was unsuccessful [22]. Further studies are needed to demonstrate that the HIR could be a marker in case of EVT failure and to validate the current cut-off in that group.

#### **5. Conclusions**

The HIR was associated with clinical outcomes in patients who underwent successful recanalization using EVT. Specifically, a low HIR was associated with a favorable outcome. EVT might need to be considered for patients with a low HIR despite the relatively unfavorable ischemic core status and time profile. Further prospective studies are needed to establish the EVT eligibility criteria based on HIR.

**Author Contributions:** Conceptualization, H.S.N.; data curation, J.-H.B., Y.D.K., K.J.L., J.K.C., M.B., B.M.K., D.J.K., J.H.H., and H.S.N.; formal analysis, J.-H.B. and H.S.N.; funding acquisition, H.S.N.; methodology, J.-H.B. and H.S.N.; writing—original draft, J.-H.B.; writing—review and editing, J.-H.B. and H.S.N. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by a grant from the Korea Health Technology R&D Project through the Korea Health Industry Development Institute, funded by the Ministry of Health & Welfare, Republic of Korea (HI19C0481, HC19C0028).

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board of Severance Hospital (4-2020-1201; 14 December 2020).

**Informed Consent Statement:** Patient consent was waived due to the retrospective nature of the study.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


### *Article* **Hypoperfusion Index Ratio as a Surrogate of Collateral Scoring on CT Angiogram in Large Vessel Stroke**

**Chun-Min Wang, Yu-Ming Chang \*, Pi-Shan Sung \* and Chih-Hung Chen**

Department of Neurology, National Cheng Kung University Hospital, College of Medicine, National Cheng Kung University, Tainan 704, Taiwan; ro2003yes@gmail.com (C.-M.W.); lchih@mail.ncku.edu.tw (C.-H.C.)

**\*** Correspondence: cornworldmirror@hotmail.com (Y.-M.C.); pishansung@gmail.com (P.-S.S.);

Tel.: +886-6-2353535-5481 (P.-S.S.); Fax: +886-6-2374285 (P.-S.S.)

**Abstract:** Background: This study was to evaluate the correlation of the hypoperfusion intensity ratio (HIR) with the collateral score from multiphase computed tomography angiography (mCTA) among patients with large vessel stroke. Method: From February 2019 to May 2020, we retrospectively reviewed the patients with large vessel strokes (intracranial carotid artery or proximal middle cerebral artery occlusion). HIR was defined as a Tmax > 10 s lesion volume divided by a Tmax > 6 s lesion volume, which was calculated by automatic software (Syngo.via, Siemens). The correlation between the HIR and mCTA score was evaluated by Pearson's correlation. The cutoff value predicting the mCTA score was evaluated by receiver operating characteristic analysis. Result: Ninety-four patients were enrolled in the final analysis. The patients with good collaterals had a smaller core volume (37.3 ± 24.7 vs. 116.5 ± 70 mL, *p* < 0.001) and lower HIR (0.51 ± 0.2 vs. 0.73 ± 0.13, *p* < 0.001) than those with poor collaterals. A higher HIR was correlated with a poorer collateral score by Pearson's correlation. (r = −0.64, *p* < 0.001). The receiver operating characteristic (ROC) analysis suggested that the best HIR value for predicting a good collateral score was 0.68 (area under curve: 0.82). Conclusion: HIR is a good surrogate of collateral circulation in patients with acute large artery occlusion.

**Keywords:** hypoperfusion index ratio; collateral circulation; collateral scoring; CTA; CTP; large vessel occlusion

#### **1. Introduction**

Evaluating pial (leptomeningeal) collateral status is of great importance in predicting the evolution of infarction [1], predicting the prognosis of acute ischemic stroke [2], and selecting eligible patients for endovascular thrombectomy (EVT) [3]. Leptomeningeal collateral flow can be assessed by conventional angiography, computed tomography angiography (CTA) (including single-phase CTA, dynamic CTA, and multiphase CTA) and magnetic resonance imaging (MRI) [4]. Different scoring systems have been proposed, and most of them are semiquantitative measures [5]. In clinical practice, multiphase CTA (mCTA) has become one of the most reliable and rapid techniques to visualize collateral circulation [6]. However, there may be potentially an interrater difference in reading the result of mCTA and obtaining collateral scores on mCTA in real-world setting.

Along with mCTA, computed tomography perfusion (CTP) plays a role in decision making regarding the management of acute stroke, especially before EVT [7]. The development of automatic postprocessing software for CTP gives physicians more quantitative and rapid measures to evaluate the infarct core and potentially salvageable tissue. Calculated by automatic software, the hypoperfusion intensity ratio (HIR) was defined as the Tmax > 10 s lesion volume (Tmax10) divided by the Tmax > 6 s lesion volume (Tmax6) [8]. The HIR has been shown to predict the rate of infarct growth and functional outcome at 90 days after stroke in the DEFUSE 2 cohort; thus, it is thought to be a clinical parameter that evaluates the degree of collateral circulation [8]. In another retrospective study, patients who met

**Citation:** Wang, C.-M.; Chang, Y.-M.; Sung, P.-S.; Chen, C.-H. Hypoperfusion Index Ratio as a Surrogate of Collateral Scoring on CT Angiogram in Large Vessel Stroke. *J. Clin. Med.* **2021**, *10*, 1296. https://doi.org/10.3390/jcm10061296

Academic Editor: Hyo Suk Nam

Received: 8 February 2021 Accepted: 18 March 2021 Published: 21 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

the American Heart Association guidelines for thrombectomy were more likely to have a lower HIR [9].

A recent study showed that HIR was correlated with collateral circulation in digital subtraction angiography (DSA), suggesting a cutoff value (HIR < 0.4) as the best prediction for good DSA collaterals [10]. There may be a correlation between collateral scores on mCTA and HIR, but the cutoff value of HIR for the prediction of good collaterals on mCTA may be different from the cutoff value to predict good DSA collaterals.

On the other hand, the studies above all used RAPID software (iSchemaView, Menlo Park, CA, USA) as postprocessing software. Although other software programs have been developed and have shown some degree of agreement with RAPID [11,12], no study has demonstrated that the HIR calculated by other automatic software correlates with collateral status.

In this study, we aimed to establish the association between the HIR calculated by Syngo.via and the collateral score by mCTA and determine the best cutoff value for predicting good collaterals on mCTA.

#### **2. Materials and Methods**

#### *2.1. Patient Inclusion, Population, and Clinical Data*

National Cheng Kung University Hospital (NCKUH) is a 1320-bed tertiary medical center in southern Taiwan that can provide intravenous tissue plasminogen activator injection (IV-tPA) and endovascular thrombectomy (EVT). Between 800 and 900 patients with acute ischemic stroke are admitted to our stroke ward annually. As a participating hospital of the nationwide Taiwan Stroke Registry (TSR) [13], NCKUH has been maintaining prospective stroke registries according to the TSR protocol since 2006. Our comprehensive stroke center prospectively enrolls patients who present to our hospital within 10 days after stroke onset and receive CT and/or MRI for the index stroke. The patients' demographic characteristics and medical history were recorded according to a predefined system.

In this study, we retrospectively identified consecutive patients receiving CTA and CTP scans on arrival at our emergency department for acute stroke management between February 2019 and May 2020. By assessing each patient's CTA data, we enrolled patients with occlusions in either the internal carotid artery (ICA) or the M1 and M2 branches of the middle cerebral artery (MCA). Patients without large vessel occlusion (LVO) or with occlusion in the anterior cerebral artery (ACA), posterior cerebral artery (PCA), posterior circulation, or multiple sites were excluded because the collateral scores were non applicable. The demographic data, last known well time or onset time, initial National Institutes of Health Stroke Scale (NIHSS) score, comorbidities, details of IV-tPA and/or EVT, and modified Rankin scale (mRS) at discharge, were obtained from our registry system.

All patients needed to complete written consents prior to receive brain imaging. This study was approved by the Institutional Review Board of National Cheng Kung University Hospital (B-ER-109192).

#### *2.2. Multiphase Computed Tomography Angiography Collateral Score, Hypoperfusion Intensity Ratio, and the Eligibility of EVT*

The mCTA protocol was described in a previous study [6]. In brief, three phases (peak arterial phase, peak venous phase, and late venous phase) of consecutive scanning with an interval of 8 s were obtained, allowing for time-resolved assessment. The mCTA collateral scores (range from 0 to 5) are defined as follows: Grade 5—no filling delay compared to the asymptomatic contralateral hemisphere, normal pial vessels in the affected hemisphere; Grade 4—a filling delay of one phase in the affected hemisphere, but the extent and prominence of pial vessels is the same; Grade 3—a filling delay of two phases in the affected hemisphere, or a delay of one phase with a significantly reduced number of vessels in the ischemic territory; Grade 2—a filling delay of two phases in the affected hemisphere with a significantly reduced number of vessels in the ischemic territory, or one phase delay showing regions without visible vessels; Grade 1—only a few vessels are visible in the

affected hemisphere in any phase; Grade 0—no vessels visible in the affected hemisphere in any phase.

The mCTA collateral scores were independently assessed by two raters (Wang C-M and Chang Y-M). Those results with different mCTA collateral scores were further discussed at the research conference by these two raters. A final mCTA collateral score was given after discussion and agreement. An mCTA collateral score of 3 or lower indicates poor collateral status [14].

The CTP images were postprocessed by the software Syngo.via CT Neuro Perfusion (version VB30 HF03; Siemens Healthcare, Erlangen, Germany). Tmax is defined as the time to maximum of the residue function obtained by deconvolution [15]. The volume of the ischemic core, penumbra and perfusion mismatch were automatically calculated based on cerebral blood flow (<30%) and Tmax (>6 s) lesion volume. The HIR was defined as the Tmax > 10 s lesion volume divided by the Tmax>6s lesion volume.

The eligibility criteria of EVT at our site are mainly based on the guidelines from the American Heart Association/American Stroke Association (AHA/ASA) [16] and Taiwan Stroke Society [17]. In brief, for patients with LVO within 6 h of stroke onset and an Alberta stroke program early CT score (ASPECTS) ≥ 6, EVT was considered unless patients had poor baseline conditions, such as a pre-mRS score greater than 2, terminal cancer status, unstable vital signs or multiple comorbidities. Perfusion imaging may provide complementary information alongside CTA for neurointerventionists, especially for patients within 6 h to 24 h of last known normal. For those patients, EVT was considered case by case based on the criteria of the DAWN [18] or DEFUSE 3 trial [19].

#### *2.3. Outcome and Statistical Analysis*

The continuous variables (age, NIHSS score, time after stroke onset, HIR, mCTA collateral score, ischemic core, penumbra, and perfusion mismatch volume) are expressed as the means ± standard deviation (SD) or median, quartiles and interquartile range (IQR). The nominal variables (medical history of comorbidity and medication, occlusion sites, IV-tPA and EVT) were summarized as frequency descriptive analyses. The interrater reliability of mCTA score was measured by using Cohen's kappa coefficient.

The subjects were divided into subgroups based on the mCTA collateral score (good collaterals (scores 4–5) vs. poor collaterals (scores 0–3)). Univariate analyses were performed to compare the age, initial stroke severity assessed by initial NIHSS score, ischemic core volume, penumbra volume, perfusion mismatch volume and perfusion ratio between groups by using independent T-test or Mann–Whitney U-test; the sex, comorbidity, medication history, treatment with IV-tPA or EVT, and post-EVT TICI score between groups by using Pearson's chi-squared test. The correlation between the HIR and mCTA collateral score was calculated using Pearson's correlation. Statistical tests are considered significant at a < 0.05 level. Receiver operating characteristic (ROC) curve analysis was performed to determine an HIR threshold for predicting good collaterals, which was defined as mCTA collateral scores of 4–5.

#### **3. Results**

From February 2019 to May 2020, 341 patients with acute ischemic stroke underwent CTA and CTP at NCKUH. After excluding those without LVO (*n* = 153), those with stroke in the posterior circulation and PCA territory (*n* = 56), those in the ACA territory (*n* = 9), those with bilateral or multiple occlusion sites (*n* = 3) and those with poor image quality (e.g., failure to be processed by software, poorly enhanced vessels, severe motion artifacts, etc.) or missing data (*n* = 26), 94 patients were enrolled in the final analysis (male/female: *n* = 59/36) (Figure 1). The mean age was 72 (SD: 12.9, range: 30–94), and the median NIHSS score was 21 (IQR: 14–27). The occlusion sites were at the ICA (*n* = 23), M1 (*n* = 42) and M2 (*n* = 29). The median HIR was 0.65 (IQR: 0.47–0.74), and the median mCTA score was 4 (IQR: 2–4). The mCTA score showed substantial agreement between the two raters with a kappa value of 0.64.

**Figure 1.** The flow chart of enrollment. CTP, computed topography perfusion; mCTA, multiphase computed topographic angiography; ACA anterior cerebral artery; ICA, internal carotid artery; M1/M2, M1, and M2 segments of the middle cerebral artery.

There were no significant differences in age, sex, history of hypertension, diabetes mellitus, coronary artery disease, congestive heart failure, prior antiplatelet or anticoagulant use, or tobacco use between patients with good and poor collaterals (Table 1). The patients with good collaterals had significantly lower stroke severity (median NIHSS = 14, IQR: 10–21) than those with poor collaterals (median NIHSS = 25, IQR: 21–30, *p* < 0.001). There were also more patients with good collaterals receiving IV-tPA (44.2% versus 16.7%, *p* = 0.004). There were no significant differences in the percentage of patients receiving EVT between the two groups.

**Table 1.** Demographic characteristics of patients with poor collaterals (score 0–3) and good collaterals (score 4–5) based on multiphase CT angiography collateral score in acute ischemic stroke.



**Table 1.** *Cont*.

NIHSS, National Institutes of Health Stroke Scale; ICA, internal carotid artery; M1/M2, M1, and M2 segments of the middle cerebral artery; EVT, endovascular thrombectomy; CTP, computed tomography perfusion.

The patients with good collaterals had smaller cores (37.3 ± 24.7 vs. 116.5 ± 70 mL, *p* < 0.001) and Tmax6 (120 ± 64.9 vs. 203.5 ± 88 mL, *p* < 0.001) and Tmax10 volumes (59.2 ± 31.1 vs. 152 ± 82.6 mL, *p* < 0.001) and lower HIRs (0.51 ± 0.2 vs. 0.73 ± 0.13, *p* < 0.001), as well as lower mRS scores (2 (IQR: 1–3.75) vs. 5 (IQR: 2–6), *p* = 0.02) at discharge (Table 2).

**Table 2.** Core volume, Tmax > 6- and 10-s lesion volume, hypoperfusion index ratio (HIR) and modified Rankin scale (mRS) at discharge in patients with poor collaterals (score 0–3) and good collaterals (score 4–5) based on the multiphase CT angiography collateral score.


HIR, hypoperfusion intensity ratio.

A higher HIR was correlated with a poorer collateral score by Pearson's correlation (r = −0.64, *p* < 0.001) (Figure 2). The ROC analysis suggested that the best value for predicting a good collateral score was 0.68, with a sensitivity of 0.76, specificity of 0.81, and area under the curve of 0.82 (Figure 3).

**Figure 2.** Scatter plot of hypoperfusion index (HIR) and multiphase CT angiography (mCTA) collateral score. Pearson's r = –0.64, *p* < 0.001.

**Figure 3.** Receiver operating characteristic (ROC) analysis of the hypoperfusion index (HIR) to predict good collateral by multiphase CT angiography (mCTA) collateral score (4 or 5). The best predicted value of HIR was 0.68, with a sensitivity of 76%, specificity of 81% and area under curve (AUC) of 0.82.

#### **4. Discussion**

Our study found that the HIR is correlated with the mCTA collateral score in patients with acute occlusions at the ICA, M1, or M2 segment of the MCA, with 0.68 being the best value that predicts a good collateral score by Syngo.via CT Neuro Perfusion software.

In clinical practice, we may evaluate collateral status directly via CTA, and the collateral scores were correlated with clinical outcomes, even with reperfusion by IV-tPA and EVT [2,20,21]. However, there are some pitfalls in scoring the collateral status from CTA. It is somehow subjective and rater-dependent, and the raters need to be trained to reduce the interrater variability. Objective automatic software may be helpful in the clinical scenario of managing acute ischemic stroke. HIR, automatically calculated by software, is defined as Tmax10 divided by Tmax6. Tmax6 was shown to predict penumbra well in a previous study [22], while Tmax10 was found to represent the most endangered tissue with extremely delayed perfusion [23]. HIR may be considered a quantitative measure of collateral blood flow to the brain as "tissue-level collaterals". One previous study demonstrated that HIR was correlated with collateral circulation in DSA in patients during EVT [10]. Another analysis of the SWIFT PRIME study also showed that collateral status was correlated with relative blood volume and HIR by using RAPID software [24]. Our findings confirm the concept of using HIR as an indicator for collateral circulation even with a different software package and provide a particular cutoff value of HIR to predict good collateral status by using Syngo.via software.

In the study mentioned above [8], the ROC analysis showed that an HIR > 0.4 had a sensitivity of 0.66 and a specificity of 0.70 for predicting poor collateral flow. Another study [10] also revealed that an HIR < 0.403 best predicted good angiographic collaterals with a sensitivity of 0.79 and specificity of 0.56. Both reports used RAPID software. We found that the cutoff value for predicting good collaterals in mCTA was 0.68 by using Syngo.via software. This may result from the different algorithms of image processing and chosen parameters in different software packages. A study reported that the infarct core predicted by Syngo.via will meet good agreement with RAPID if changing the parameters to CBV < 1.2 mL/100 mL and applying an additional smoothing filter [12], while another study suggested that the predicted volume of the infarct core calculated by Syngo.via could be concordant with RAPID if the relative cerebral blood flow threshold is changed to <20% [25]. In the same study, when analyzed as a subgroup, in patients with LVOs (ICA and M1 occlusion), there was no statistically significant difference between the calculated values for the core and hypoperfusion volumes. To the best of our knowledge, there have been no studies comparing and correlating the Tmax10 or HIR of the two software packages. Further study is warranted to correlate the HIR acquired by different software programs. Despite some difference in predicted volume of core and penumbra, Syngo.via and RAPID software showed high concordance in correctly triaging patients into "go or no-go" for EVT in real-world settings [26].

Potential delay may exist between activating the thrombectomy team and the actual reperfusion time. Better collateral circulation may extend the survival period of the penumbra. A study showed that patients with good collaterals had a smaller infarct core and higher mismatch ratio in ICA and M1/M2 occlusion and within 12 h of stroke onset [27]. Interestingly, in the DEFUSE 3 cohort, good collaterals were associated with reduced ischemic core growth but not neurologic outcome in the late therapeutic window [28]. Another study also showed that in patients with LVO who underwent endovascular intervention, collateral status was strongly associated with MCA territory final infarct volumes but not correlated with favorable outcomes at discharge [24]. The authors' explanation was that good collaterals preserved the watershed area of the ACA/MCA and MCA/PCA but had no influence on certain critical brain regions (such as the precentral gyrus and the posterior limb of the internal capsule), which have larger impacts on functional independence. On the other hand, in the DEFUSE 2 cohort, final infarction volume increased in association with HIR quartiles as well as infarct growth regardless of reperfusion. After adjusting for the factor of early reperfusion, a favorable outcome was still associated with a low HIR [8]. Therefore, by using HIR as a surrogate for "tissue-level collateral assessment", a stroke neurologist may have additional information to accelerate the patient selection of EVT and predict the outcome. If our result is further validated in other independent database, multiphase CTA may not be necessary in most of the cases because the HIR could already represent the collateral status and even correlate prognosis better. Patients could benefit from reduced exposure of radiation and contrast medium, and we may save more time during pre-EVT evaluation.

There were some limitations of our study. First, this was a retrospective observational study in a single medical center, and sampling bias was inevitable, although we collected consecutive acute stroke patients who received CTA/CTP. Second, despite the specific definition, the mCTA score is a subjective scoring system that may have interrater differences, and there are different scoring systems for collaterals in CTA [29], which are not fully comparable. Third, we only included patients with acute stroke and large vessel occlusion patients. Those with occlusion sites other than the ICA or MCA were excluded; thus, the correlation of the mCTA score and HIR may not be reliable at other occlusion sites or in post-acute stage of stroke. Fourth, the proportion of patients undergoing EVT was comparable between good and poor collaterals in our study. This may be due to the higher proportion of M2 occlusion in patients with good collaterals, and EVT in patients with M2 occlusions is optional at our sites. Fifth, the sensitivity and specificity of this threshold by Syngo.via has not been validated in other independent database. Furthermore, the chronic stenosis of large vessels, old stroke and poor cardiac output tremendously affect the image quality of CTP. Therefore, the above conditions may interfere with the relationship between collateral status and HIR. Last but not least, this study was performed in an Eastern Asian population; thus, extrapolating our findings to other ethnicities must be done with caution.

#### **5. Conclusions**

In our study, we found that a lower HIR correlates with a good mCTA collateral core in patients with occlusions in the ICA and M1 and M2 segments of the MCA. The best cutoff value of HIR is 0.68 to predict good collaterals by Syngo.via. The HIR is a good surrogate of tissue-level collateral status, even in different automatic software packages. However, the best HIR cutoff value to predict good collaterals may be adjusted by different software programs.

**Author Contributions:** Conceptualization, C.-M.W., Y.-M.C. and P.-S.S.; methodology: C.-M.W. and Y.-M.C.; software, C.-M.W.: validation: C.-M.W., P.-S.S. and C.-H.C.; formal analysis, C.-M.W. and P.-S.S.; investigation, P.-S.S.; resources, P.-S.S.; data curation, C.-M.W. and Y.-M.C.; writing—original draft preparation, C.-M.W.; writing—review and editing, C.-M.W. and P.-S.S.; visualization, C.-H.C.; supervision, P.-S.S.; project administration, P.-S.S.; funding acquisition, Y.-M.C. and P.-S.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This work received grants from National Cheng Kung University Hospital (Grant numbers: NCKUH-10902039 and NCKUH-11003022) and the Ministry of Science and Technology of Taiwan via contract MOST 108-2321-B-006-024–MY2. The funding source was not involved in any of the study processes or the writing of this manuscript. This research had no relationship with industry and pharmaceutical companies.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the This study was approved by the Institutional Review Board of National Cheng Kung University Hospital (B-ER-109192).

**Informed Consent Statement:** All patients needed to complete written consents prior to receive brain imaging. Informed consent was waived due to retrospective analysis of imaging study and was approved by the institutional review board of National Cheng Kung University Hospital.

**Data Availability Statement:** The data was available upon reasonable email request.

**Acknowledgments:** We thank the staff of the stroke center of National Cheng Kung University Hospital for their support.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


#### *Article*
