- This study externally validated the HERMES-24 score for predicting functional outcomes after endovascular therapy in late-window large vessel occlusion.
- Researchers analyzed 2,117 German Stroke Registry patients with large vessel occlusion treated beyond six hours from symptom onset.
- The HERMES-24 score achieved an AUC of 0.876 (95% CI 0.859-0.889) for mRS ≤2 and 0.856 (95% CI 0.837-0.875) for mRS ≤3.
- The authors concluded the HERMES-24 score shows good discriminative performance, but with lower accuracy than in trial populations.
- Clinicians should apply the HERMES-24 score cautiously, especially in patients with lower baseline NIH Stroke Scale scores (AUC 0.850).
Navigating Prognosis After Late-Window Stroke Intervention
Endovascular therapy (EVT) has reshaped the management of acute ischemic stroke from large vessel occlusion (LVO), improving functional outcomes for many patients [1, 2]. Initially confined to a narrow time frame, the indications for EVT now extend up to 24 hours from symptom onset for patients selected with advanced imaging [3, 4]. This expansion, while beneficial, complicates outcome prediction due to diverse clinical presentations and comorbidities. As clinicians manage more late-presenting LVO cases, the need for reliable prognostic tools to inform patient discussions and post-intervention care has become critical [5]. A recent meta-analysis underscored this, showing EVT in patients with large core infarcts decreased mortality (odds ratio [OR]: 0.67, 95% confidence interval [CI]: 0.51-0.87) and improved outcomes (modified Rankin Scale [mRS] 0-2; OR: 3.54, 95% CI: 1.96-6.4) despite a higher risk of intracranial hemorrhage [6]. A new validation study now assesses the real-world utility of a specific prognostic score designed for this challenging patient population.
Validating a Predictive Tool in Clinical Practice
The HERMES-24 score, a prognostic tool developed within the controlled environment of randomized clinical trials, has previously shown accuracy in predicting outcomes for late-window LVO patients. The present study sought to externally validate this score in a large, real-world cohort to determine its performance in routine clinical practice. Researchers drew data from the German Stroke Registry, focusing on patients with LVO who underwent endovascular therapy more than 6 hours after symptom onset or last seen well. Through a complete case analysis, they carefully curated the cohort by excluding patients who were already functionally dependent before their stroke. Specifically, patients with a premorbid modified Rankin Scale (mRS) score above 2 were excluded from the analysis predicting a 90-day mRS of 2 or less, ensuring the outcome being measured was a return to functional independence. The score's ability to predict 90-day functional status was then evaluated using the c-statistic, a value equivalent to the area under the curve (AUC) that measures a model's ability to correctly distinguish between patients with different outcomes.
Patient Characteristics in the Validation Cohort
The validation was conducted on a substantial cohort of 2,117 patients, providing a robust sample of individuals treated in a real-world clinical setting. The population had a mean age of 74 ± 13.3 years, and 55.4% were female. These demographics reflect a typical, older population undergoing late-window stroke interventions. On admission, the median National Institutes of Health Stroke Scale (NIHSS) score was 14, with an interquartile range of 9 to 18. An NIHSS score in this range signifies a moderate to severe neurological deficit, which is characteristic of patients presenting with LVO who are candidates for thrombectomy. This confirms the study cohort is representative of the target population for whom such a prognostic tool would be clinically applied.
HERMES-24 Score Performance in Real-World Settings
When tested in this large, real-world cohort, the HERMES-24 score demonstrated strong predictive capabilities. The primary measure of performance was its discriminative ability, or how well it could separate patients likely to have a favorable outcome from those who would not. For predicting functional independence at 90 days, defined as a modified Rankin Scale (mRS) score of 2 or less, the score achieved an area under the curve (AUC) of 0.876 (95% CI 0.859-0.889). An AUC of 1.0 represents perfect prediction, while 0.5 represents no better than chance, placing this result in the category of good to excellent discrimination. For the broader outcome of an mRS score of 3 or less, which includes patients with slight disability who do not require daily assistance, the score's performance remained robust, with an AUC of 0.856 (95% CI 0.837-0.875). These findings confirm the score's utility outside the strict confines of a clinical trial.
Nuances in Predictive Accuracy
Deeper analysis of the score's performance revealed an important clinical nuance. While the HERMES-24 score performed well overall, its accuracy was attenuated in a specific subgroup of patients. The researchers found lower predictive performance in patients presenting with less severe initial strokes, defined as an admission National Institutes of Health Stroke Scale (NIHSS) score below 18. In this particular subgroup, the area under the curve (AUC) for predicting a 90-day modified Rankin Scale (mRS) score of 2 or less was 0.850 (95% CI 0.832-0.870). Although still a respectable level of performance, this is notably lower than the AUC of 0.876 observed for the entire cohort. This finding suggests that for patients with milder initial deficits, the score may be a less precise instrument for prognosis.
Clinical Implications and Considerations
This external validation provides valuable support for the cautious clinical application of the HERMES-24 score. The study confirms that the tool maintains good discriminative performance for 90-day functional outcomes in a large, real-world population of late-window LVO patients undergoing thrombectomy. It can serve as a useful adjunct for clinicians when counseling patients and families about expected recovery. However, two key considerations must temper its use. First, the score's overall performance in this real-world cohort was lower than that reported in the original, more selective trial populations. Second, and perhaps more critically for daily practice, its predictive power is reduced in patients with lower baseline NIHSS scores. This suggests that while the HERMES-24 score is a helpful tool for risk stratification, it should not replace clinical judgment, particularly when evaluating patients who present with less severe neurological deficits.
References
1. Berkhemer OA, Fransen P, Beumer D, et al. A Randomized Trial of Intraarterial Treatment for Acute Ischemic Stroke. New England Journal of Medicine. 2014. doi:10.1056/nejmoa1411587
2. Uchida K, Rinkel LA, Ospel JM, Diprose WK, Goyal M. A systematic review of randomized controlled trials of endovascular therapy for stroke on mortality and disability.. Journal of the neurological sciences. 2024. doi:10.1016/j.jns.2024.122991
3. Albers GW, Marks MP, Kemp S, et al. Thrombectomy for Stroke at 6 to 16 Hours with Selection by Perfusion Imaging. New England Journal of Medicine. 2018. doi:10.1056/nejmoa1713973
4. Nogueira RG, Jadhav AP, Haussen DC, et al. Thrombectomy 6 to 24 Hours after Stroke with a Mismatch between Deficit and Infarct. New England Journal of Medicine. 2017. doi:10.1056/nejmoa1706442
5. Tanaka K, Brown SC, Goyal M, et al. Abstract WP248: Developing and Validating Post-Treatment HERMES Score to Predict Outcome From Anterior Circulation Large Vessel Occlusion Stroke: A Meta-Analysis of Individual Data From 7 Randomized Clinical Trials. Stroke. 2024. doi:10.1161/str.55.suppl_1.wp248
6. Abdollahifard S, Taherifard E, Sadeghi A, Kiadeh PRH, Yousefi O, Mowla A. Endovascular therapy for acute stroke with a large infarct core: A systematic review and meta-analysis.. Journal of stroke and cerebrovascular diseases : the official journal of National Stroke Association. 2023. doi:10.1016/j.jstrokecerebrovasdis.2023.107427