Combining Computed Tomography Perfusion and Baseline National Institutes of Health Stroke Scale to Assess the Clinical Penumbra in Ischemic Stroke

Article information

J Stroke. 2025;27(2):270-274

Publication date (electronic) : 2025 May 31

doi : https://doi.org/10.5853/jos.2024.03720

Umberto Pensato^1,2,3, Alexander Stebner³, Salome Bosshart³, Ruchir Shah⁴, Axel Rohr⁵, Ricardo Hanel⁶, Michael E. Kelly⁷, Aditya Bharatha⁸, Michael D. Hill^3,9, Mayank Goyal⁹, Andrew M. Demchuk³, Johanna M. Ospel^,9

¹IRCCS Humanitas Research Hospital, Milan, Italy

²Department of Biomedical Sciences, Humanitas University, Milan, Italy

³Calgary Stroke Program, Departments of Clinical Neurosciences and Radiology, Hotchkiss Brain Institute, Cumming School of Medicine, University of Calgary, Calgary, AB, Canada

⁴Erlanger Hospital, Chattanooga, TN, USA

⁵University of British Columbia, Vancouver, BC, Canada

⁶Lyerly Neurosurgery, Baptist Neurological Institute, Jacksonville, FL, USA

⁷Department of Surgery, University of Saskatchewan, Saskatoon, SK, Canada

⁸St. Michael’s Hospital, University of Toronto, Toronto, ON, Canada

⁹Department of Radiology, University of Calgary Cumming School of Medicine, Calgary, AB, Canada

Correspondence: Johanna M. Ospel Department of Radiology, University of Calgary Cumming School of Medicine, 2500 University Drive NW, Calgary, AB T2N 1N4, Canada Tel: +1-825-3653948 E-mail: johannaospel@gmail.com

Received 2024 September 11; Revised 2025 January 26; Accepted 2025 February 19.

Dear Sir:

In acute ischemic stroke, some neurological deficits may be reversible (corresponding topographically to the ischemic penumbra), while others are not (ischemic core). Computed tomography perfusion (CTP) can quantify the ischemic penumbra volume [1]. However, the clinical eloquence of different brain regions varies considerably, often resulting in a clinical-radiological mismatch: small ischemic areas might cause substantial deficits, whereas large volumes may produce minimal symptoms [2]. Therefore, clinical and radiological information should be viewed as complementary, and combining their insights might provide a more accurate assessment of the clinical penumbra, namely, the potentially reversible clinical deficits of stroke patients caused by the ischemic penumbra. We aim to evaluate how CTP-derived measures, weighted by baseline National Institutes of Health Stroke Scale (NIHSS) scores, correlate with NIHSS improvement after complete reperfusion—the clinical penumbra.

We included patients from the randomized ESCAPE-NA1 (Efficacy and Safety of Nerinetide for the Treatment of Acute Ischaemic Stroke) trial [3] (1) with available baseline CTP, (2) who achieved near-complete reperfusion, and (3) without parenchymal hematoma (PH) after endovascular thrombectomy (EVT). These inclusion criteria were meant to exclusively select patients for whom the ischemic penumbra could be fully saved. Reperfusion status was assessed through final intracranial catheter angiography using the expanded Thrombolysis in Cerebral Infarction (eTICI) scale. Near-complete reperfusion was defined as eTICI 2c/3. The ischemic core was defined as the CTP volume with <30% relative cerebral blood flow (rCBF), the hypoperfused tissue was defined as Tmax >6 seconds volume, and the ischemic penumbra was defined as the difference between hypoperfused tissue and ischemic core (Tmax >6 seconds volume – rCBF <30% volume). “CTP-estimated salvageable NIHSS” was calculated as the proportion of ischemic penumbra on the total hypoperfused tissue, multiplied by the baseline NIHSS (Figure 1): “CTP-estimated salvageable NIHSS”=[(Ischemic penumbra)/(Ischemic penumbra+Ischemic core)] × Baseline NIHSS.

Figure 1.

Exemplary case of “CTP-estimated salvageable NIHSS.” A patient presented with a baseline NIHSS of 21 due to an occlusion of the M1 segment of the right middle cerebral artery. (A) CTP maps estimated a core volume of 128 mL and a penumbra volume of 65 mL. “CTP-estimated salvageable NIHSS” was 7. Near-complete recanalization (eTICI=2c) was achieved 71 minutes after CT acquisition. (B) At 24 hours, the patient’s NIHSS was 13, corresponding to a “truly saved NIHSS” of 8, with extensive infarct seen on diffusion-weighted MRI. CTP, computed tomography perfusion; NIHSS, National Institutes of Health Stroke Scale; eTICI, expanded Thrombolysis in Cerebral Infarction; CT, computed tomography; MRI, magnetic resonance imaging; CBF, cerebral blood flow.

The primary outcome was the “truly saved NIHSS” at 24 hours, i.e., the difference between baseline and 24-hour NIHSS. Secondary outcomes were “truly saved NIHSS” at 48 hours, 5 days, 30 days, and 90 days compared to baseline. Additional information on ethical approval, study design, and imaging acquisition is available in Supplementary Methods.

The correlation between “CTP-estimated salvageable NIHSS” and clinical outcomes was evaluated with Spearman’s rank correlation coefficient (rs), and limits of agreement were visualized in a Bland-Altman plot. For sub-analyses, the patient sample was dichotomized into patients with <15 mL and ≥15 mL baseline ischemic core volume. We chose the higher core volume threshold that included at least one-third of the study sample.

Of the 1,105 patients recruited in the ESCAPE-NA1 study, 169 (15.2%) patients were included in the study sample (Supplementary Figure 1), and their baseline characteristics, treatment workflow metrics, and outcomes are summarized in Table 1. The median baseline NIHSS was 17 (interquartile range [IQR]=12–21), the median “CTP-estimated salvageable NIHSS” was 14 (IQR=11–19), and the median “truly saved NIHSS” measured at different time points was 10 (IQR=6–16) at 24 hours, 11 (IQR=11–16) at 48 hours, 12 (IQR=8–17) at 5 days, 13 (IQR=10–18) at 30 days, and 14 (IQR=10–19) at 90 days. There was a statistically significant and nominally strong positive correlation between “CTP-estimated salvageable NIHSS” and outcomes: at 24 hours (r_s= 0.522, P<0.0001), 48 hours (r_s=0.539, P<0.0001), 5 days (r_s=0.624, P<0.0001), 30 days (r_s=0.689, P<0.0001), and 90 days (r_s=0.714, P<0.0001). The relative mean difference between “CTP-estimated salvageable NIHSS” and outcomes was +4.2 (limits of agreement -8.7 and +17.0) at 24 hours and +0.2 (limits of agreement -8.6 and +9.0) at 90 days (Figure 2). “CTP-estimated salvageable NIHSS” had a higher correlation coefficient with 24-hour “truly saved NIHSS” in comparison to baseline NIHSS (0.522 vs. 0.470) and a better agreement (mean difference +4.2 vs. +6.2).

Table 1.

Epidemiological and baseline clinical features, treatment workflow, and outcomes in the study sample

Figure 2.

Scatter plots with a trendline to illustrate the relationship between “CTP-estimated salvageable NIHSS” and “truly saved NIHSS” at (A) 24 hours and (B) 90 days in the entire cohort. Note that the negative values on the x-axis are due to few instances of unexplained neurological deterioration. Bland-Altman plots illustrate agreement between “CTP-estimated salvageable NIHSS” and “truly saved NIHSS” at (C) 24 hours and (D) 90 days. The mean between “Truly saved NIHSS” and “CTP-estimated salvageable NIHSS” is shown on the x-axis and their difference on the y-axis. Red solid horizontal lines in (C) and (D) show the bias, grey solid horizontal lines indicate the zero-bias line, and red dashed horizontal lines indicate the limits of agreement. Note that the negative value in the plot (C) is explained by a patient with a baseline NIHSS of 7 who experienced an unexplained neurological deterioration resulting in a 24-hour NIHSS of 18, corresponding to a negative “truly saved NIHSS” of -11. The resulting mean (“CTP-salvageable NIHSS”+“truly saved NIHSS”/2) was: [7+(-11)]/2=-2.” CTP, computed tomography perfusion; NIHSS, National Institutes of Health Stroke Scale; r_s, Spearman’s rank correlation coefficient.

In the subgroup with core volume <15 mL, “CTP-estimated salvageable NIHSS” had a higher agreement and a higher correlation coefficient with “truly saved NIHSS” at 24 hours and 90 days compared to baseline NIHSS. In the subgroup with core volume ≥15 mL, “CTP-estimated salvageable NIHSS” had a higher agreement, but a marginally lower correlation coefficient with “truly saved NIHSS” at 24 hours compared to baseline NIHSS (Supplementary Results and Supplementary Figure 2).

Our findings indicate that “CTP-estimated salvageable NIHSS” predicts “truly saved NIHSS” across different time points in acute ischemic stroke patients with large vessel occlusion who have undergone complete reperfusion treatment. However, it performed less well in patients with ≥15 mL of ischemic core. The correlation and agreement between “CTP-estimated salvageable NIHSS and “truly saved NIHSS” was strong but not perfect. The clinical penumbra may have been underestimated due to multiple factors. First, CTP might overestimate the core volume, particularly with rapid reperfusion [4]. Accordingly, CTP findings should probably not play a prominent role in thrombectomy decision-making anymore, especially in the early therapeutic window. The recently published large core EVT trials provide additional evidence to support this approach [5]. Second, current acute neuroimaging struggles to discriminate between partial and complete infarction [6]. Accordingly, evidence suggests that EVT may still be beneficial in the ischemic core tissue by promoting partial rather than complete infarction [6]. In other words, reperfusion of what we call the “ischemic core” (which may, in fact, contain islands of viable tissue), in addition to the ischemic penumbra, might have contributed to improving the clinical outcomes in our sample. Conversely, the clinical penumbra might be overestimated in cases of early neurological deterioration or when CTP underestimates the core volume—a phenomenon referred to as “perfusion scotoma.” [7] Additionally, other factors could lead to inaccuracy in both directions. The relationship between NIHSS and cerebral tissue damaged extension is unclear and might be non-linear, given the remarkably heterogeneous clinical eloquence of cerebral tissue. Finally, the topography of the infarct (i.e., the clinical eloquence) and the brain frailty (i.e., the resilience of the cerebral tissue to any injury) are critical factors that might need to be considered when predicting the clinical penumbra [2,8].

Our study has limitations. First, most patients in our sample exhibited a substantial ischemic penumbra (approximately 95%) and a relatively small median CTP core (7 mL), limiting the generalizability of our findings. Second, we excluded patients with significant hemorrhagic transformation, a complication recurrently associated with reperfusion of a severely damaged ischemic core, potentially selecting a population with a less severely compromised ischemic core. Third, the study sample size was relatively small. Finally, we use the NIHSS to assess the clinical outcome instead of the 3-month modified Rankin Scale (mRS). However, 24-hour NIHSS emerged as the most robust predictor of the 90-day mRS [9], while being arguably less influenced by post-acute confounders such as post-stroke rehabilitation and unrelated diseases. Moreover, our metric correlated even better with outcomes at 90 days.

The “CTP-estimated salvageable NIHSS” represents a simple numeric estimate of potential clinical improvement achievable through complete reperfusion. Our analysis is a preliminary attempt to quantify the clinical penumbra. Combining the topography and eloquence of the hypoperfused cerebral tissue with volumetric measures could enhance our understanding of the clinical penumbra, leading to a more accurate imaging-based prediction of reversible deficits [2,10]. Training an artificial intelligence model to recognize the 24-hour NIHSS scores associated with segmented infarct volume at 24 hours could lead to the development of an automated system able to estimate the contribution of the ischemic core to a patient’s baseline NIHSS. This would provide physicians with an accurate and automated prediction of the clinical penumbra in the acute phase, enhancing prognosis and decision-making.