Performance of serial CT ASPECTS for predicting stroke outcomes in patients with thrombolyzed acute ischemic stroke

INTRODUCTION

Stroke ranks the second-most widespread contributor to mortality and impairment globally, and its incidence continues to increase within the aging demographic.^[1] In Thailand, the death rate from acute ischemic stroke (AIS) is 10%, and 50% of patients suffer from disabilities.^[2] The availability of endovascular treatment remains restricted in many Thai hospitals. The predominant approach for treating AIS remains the intravenous recombinant tissue plasminogen activator (rt-PA).^[3,4] Nevertheless, a comprehensive meta-analysis of observational studies and randomized trials identified a mortality rate of up to 17.5% among AIS patients treated with rt-PA.^[5]

The Alberta Stroke Program Early Computed Tomography Score (ASPECTS) is a commonly employed scoring system that utilizes CT or magnetic resonance imaging to assess early ischemic changes (EICs).^[6] It demonstrates high sensitivity, specificity, and correlation with functional outcomes as symptomatic intracranial hemorrhage.^[7] A study found a significant negative correlation coefficient of −0.680 (P < 0.001) between the ASPECTS and the National Institutes of Health Stroke Score (NIHSS).^[8] Moreover, the 24- h NIHSS outperformed the baseline NIHSS in predicting long-term stroke outcomes. With substantial supporting data, this instrument has proven its reliability and validity as an early surrogate for clinical outcomes in AIS patients.^[9] Therefore, analyzing EICs through a serial ASPECTS evaluation can help predict mortality in thrombolyzed AIS patients.

In resource-limited settings like Thailand, budget constraints hinder the routine use of advanced neuroimaging for diagnosing all AIS patients. The predictive ability of serial ASPECTS assessment using non-contrast CT (NCCT) for mortality following thrombolysis in Thailand is insufficient. Therefore, this retrospective study evaluated the predictive accuracy of serial ASPECTS evaluations for inhospital mortality (IHM) and unfavorable outcome in individuals with acute anterior circulation ischemic stroke (AACIS) who received thrombolytic treatment.

MATERIALS AND METHODS

RESULTS

Data from 345 individuals with thrombolyzed AACIS were included in this retrospective study; 42 patients were excluded, as detailed in [Figure 1]. The median NIHSS score was 11 (IQR: 8–17), the mean age was 61.8 ± 15.2 years, 53.4% were male, and the average follow-up duration was 8.8 ± 10.7 days. [Table 1] compares serial ASPECTS assessment findings between non-survivors and survivors and between those with unfavorable and favorable outcome. The rate of IHM was 18.4% (64/345), and the unfavorable outcome was 61.5% (213/345). The non-survivors, compared to the survivors, were more likely to have been aged 70 years or older and had a higher prevalence of comorbidities such as atrial fibrillation, congestive heart failure, chronic kidney disease, a history of malignancy, swallowing dysfunction, alteration of consciousness, aphasia, neglect, cranial nerve disorder, gaze paresis, pre-stroke functional status (mRS of ≥2), higher systolic and diastolic blood pressure, NIHSS ≥16, large artery atherosclerosis, cardioembolic stroke, and prior antihypertensive treatment before rt-PA administration. Patients who experienced an unfavorable outcome demonstrated a greater incidence of predisposing factors and medical conditions, including a history of atrial fibrillation, dysarthria, swallowing dysfunction, altered consciousness, aphasia, neglect, gaze paresis, elevated blood pressure on admission, NIHSS score ≥16, large artery atherosclerosis, cardioembolic stroke, and antihypertensive treatment before rt-PA administration, in comparison to patients with a favorable outcome. The IHM and unfavorable outcome predictability between the laboratory findings and serial ASPECTS assessment of thrombolyzed AIS patients who survived and those who experienced IHM were evaluated. Non-survivors had higher neutrophil-to-lymphocyte count ratio, lower hemoglobin and hematocrit levels, higher blood glucose levels at admission, lower baseline ASPECTS, lower 24-h ASPECTS, and greater change in ASPECTS compared to survivors. Similarly, patients with an unfavorable outcome exhibited lower hematocrit levels, higher blood glucose levels at admission, lower baseline ASPECTS, lower 24-h ASPECTS, and greater change in ASPECTS than those with a favorable outcome. [Figures 2 and 3] present the distributions of the baseline ASPECTS, 24-h ASPECTS, and change in ASPECTS for patients classified by IHM or survival and unfavorable or favorable outcome. A significant difference in the distribution of serial ASPECTS assessment was observed between patients with IHM and unfavorable outcome [P < 0.001, Supplemental Table 1]. No deaths occurred among patients with 24-h ASPECTS scores of 7–10, while the mortality rates were 18.75% and 81.25% for scores of 5–6 and 0–4, respectively. Patients with baseline ASPECTS ≤7, 24-h ASPECTS ≤4, and change in ASPECTS ≥3 had mortality rates of 62.5% (40/64), 81.25% (52/64), and 86.15% (56/64), respectively.

Close

Figure 1:

The patient flow chart. ASPECTS: Alberta Stroke Program Early CT Score, NCCT: Non-contrast computed tomography, rt-PA: Recombinant tissue plasminogen activator.

Export to PPT

Close

Figure 2:

Bar graph depicting the proportion of patients with varying serial ASPECTS from baseline to 24-h imaging, grouped by IHM or survival status. (a) Bar graph of the baseline ASPECTS. (b) Bar graph of the 24-h ASPECTS. (c) Bar graph of the change in ASPECTS. ASPECTS: Alberta Stroke Program Early CT Score, IHM: Inhospital mortality.

Export to PPT

Close

Figure 3:

Bar graph depicting the proportion of patients with varying serial ASPECTS from baseline to 24-h imaging, grouped by favorable or unfavorable outcome. (a) Bar graph of the baseline ASPECTS. (b) Bar graph of the 24-h ASPECTS. (c) Bar graph of the change in ASPECTS. ASPECTS: Alberta Stroke Program Early CT Score.

Export to PPT

[Table 2] shows the performance and cutoff scores of the baseline ASPECTS, 24-h ASPECTS, and changes in ASPECTS in predicting IHM and unfavorable outcome. Change in ASPECTS showed the highest sensitivity but the lowest specificity, whereas the baseline ASPECTS exhibited the lowest sensitivity. The optimal cutoff values for predicting IHM were ≤7 for baseline ASPECTS, ≤4 for 24-h ASPECTS, and ≥3 for change in ASPECTS, with corresponding Youden’s indices of 0.571, 0.832, and 0.726, respectively. In predicting unfavorable outcome, the optimal cutoff values were ≤8 for baseline ASPECTS, ≤6 for 24-h ASPECTS, and ≥1 for change in ASPECTS, yielding Youden’s indices of 0.440, 0.535, and 0.459, respectively – [Figure 4] presents the ROC curves for the serial ASPECTS assessment. The AuROC of the baseline ASPECTS, 24-h ASPECTS, and change in ASPECTS for predicting IHM was 0.823, 0.955, and 0.920, respectively. The AuROC of the baseline ASPECTS, 24-h ASPECTS, and change in ASPECTS for predicting unfavorable outcome was 0.744, 0.853, and 0.800, respectively. The 24-h ASPECTS and change in ASPECTS outperformed the baseline ASPECTS in predicting IHM and unfavorable outcome. These three scales exhibited a significant association with IHM and unfavorable outcome in univariate and multivariate analyses. Furthermore, the 24-h ASPECTS and change in ASPECTS emerged as robust predictors for IHM (AOR: 14.38, 95% CI: 1.69–122.57, P = 0.015 and AOR: 16.70, 95% CI: 4.36–64.01, P < 0.001), as well as unfavorable outcome (AOR: 5.58, 95% CI: 1.83– 17.01, P = 0.002 and AOR: 4.85, 95% CI: 2.45–9.60, P < 0.001), according to the findings presented in [Supplemental Table 2].

Close

Figure 4:

(a) ROC curve and AuROC for IHM of the three scales among thrombolyzed stroke patients. (b) ROC curve and AuROC for unfavorable outcome of the three scales at the time of hospital discharge among thrombolyzed stroke patients. ASPECTS: Alberta Stroke Program Early CT Score, AuROC: Area under the receiver operating characteristic curve, IHM: Inhospital mortality, ROC: Receiver operating characteristic.

Export to PPT

DISCUSSION

Our study found that the 24-h ASPECTS and change in ASPECTS had a greater predictive accuracy for IHM and unfavorable outcome than the baseline ASPECTS. In line with recent research, 24-h ASPECTS on NCCT demonstrated superior predictive accuracy in predicting unfavorable outcome at 3 months compared to the baseline ASPECTS and change in ASPECTS (AuROC = 0.78).^[17,18]

Our findings were consistent with those of the SWIFT study, which examined the efficacy of 24-h ASPECTS in predicting the therapeutic benefits of reperfusion. Participants in the study were randomly assigned to receive either the Merci device or the Solitaire stent retriever for arterial recanalization. In endovascular therapy, the 24-h ASPECTS emerged as the most robust predictor for predicting clinical outcomes at the 3-month follow-up. However, our study differed from the SWIFT trial regarding demographics and treatment approach, as we specifically administered rt-PA. Second, our study differed from the SWIFT trial regarding the patient population, as we included multiple stroke etiologies rather than focusing solely on large vessel occlusion. The SWIFT study reported a higher median NIHSS score (18 [IQR 8–28] vs. 11 [IQR 8–17]), indicating more severe stroke severity.^[19]

Our research revealed that the change in ASPECTS and the 24-h ASPECTS had relatively similar predictive abilities for IHM and unfavorable outcome. The change in ASPECTS might be correlated with the occurrence of reperfusion syndrome. Complications of reperfusion injury, including penumbral damage, ischemic expansion, hemorrhagic transformation, seizures, malignant cerebral edema, and herniation, were all found to be associated with poorer outcomes in AIS patients. Accurately assessing severity and criticality is crucial to determining enhanced monitoring or treatment requirements. Serial changes in tissue damage on routine NCCT can offer valuable insights into how brain tissues respond to varying degrees of reperfusion.^[20] The findings indicated that the 24-h ASPECTS and change in ASPECTS outperformed the baseline ASPECTS in predicting IHM after intravenous rt-PA administration. Furthermore, the change in ASPECTS exhibited the most incredible sensitivity in predicting unfavorable outcome. These measures demonstrated higher AuROC and sensitivity, suggesting their potential to identify individuals at risk of IHM.

The baseline ASPECTS had a lower sensitivity than the 24-h ASPECTS and change in ASPECTS for predicting IHM and unfavorable outcome; this suggested a higher likelihood of incorrectly identifying non-high-risk patients as having a risk of mortality or unfavorable outcome. However, considering the critical importance of predicting IHM, caution should be exercised when using the baseline ASPECTS. Our study contrasted with the report of Esmael et al.^[21] regarding the predictive value of the baseline ASPECTS for unfavorable outcome. Specifically, we observed a sensitivity of 73% and a specificity of 81% when using a baseline ASPECTS cutoff score of ≤7. Our study revealed different cutoff scores; however, a baseline ASPECTS of ≤8 was consistently associated with an unfavorable outcome, showing a lower sensitivity (45.5%) than previous studies. The duration of the follow-up period employed to assess adverse events across various studies contributes to variation in study results. Our research examined outcomes at the point of hospital discharge, whereas other studies measured outcomes during a 3-month follow-up period.

The superior predictive ability of 24-h ASPECTS and change in ASPECTS, compared to the baseline ASPECTS, can be attributed to the following reasons. First, there was a significant error rate when using the brain NCCT to identify EICs. The challenge lies in precisely interpreting the indistinct appearance of EICs on the baseline brain NCCT, which demands considerable expertise for accurate interpretation. Early brain NCCT might not have reliably detected ASPECTS changes, which could have indicated ischemic changes or associated vasogenic and cytotoxic edema due to inherent technical limitations.^[22] Second, NCCT ASPECTS has a lower sensitivity for detecting EICs than DWI ASPECTS.^[23] Third, it is crucial to consider the methodological limitations associated with the ASPECTS cutoff scores, which stem from the unequal weighting given to regions affected by EICs. Ischemic lesions with identical ASPECTS ratings can differ in size depending on their location.^[8]

This study has several limitations. First, the study was carried out at a single center using a retrospective cohort design, encompassing a relatively small sample size. Further validation through larger, multicenter studies was necessary. Second, it is important to acknowledge that the absence of endovascular treatment at our center could impact the generalizability of the findings. While a meta-analysis has demonstrated the significant advantages of endovascular thrombectomy compared to standard medical care, our study included only patients who received rt-PA without endovascular treatment. This limitation poses challenges in evaluating the overall treatment effect.^[24] Further research should explore the impact of serial ASPECTS on mortality in various stroke treatment approaches, including endovascular thrombectomy. In addition, investigating the influence of 24-h ASPECTS and change in ASPECTS on other complications, such as malignant cerebral edema or stroke-associated pneumonia following thrombolytic treatment, warrants further investigation. Finally, it is essential to mention that our study did not include patients with posterior circulation ischemic stroke, as the ASPECTS score cannot reliably predict their prognosis.

CONCLUSION

Our study demonstrated that 24-h ASPECTS and change in ASPECTS on NCCT were more effective in predicting IHM and unfavorable outcome in patients with thrombolyzed AACIS treated with rt-PA. These findings will highlight the importance of incorporating 24-h ASPECTS and change in ASPECTS to assist physicians in risk stratification, make early decisions in stroke care, prioritize immediate care, and provide intensive neurological monitoring. However, further external validation research will be necessary to corroborate these findings.