Abstract
Objective
The objective of this study was to determine whether baseline computed tomography (CT) markers of cerebral small vessel disease (cSVD), as a surrogate of brain frailty, explain the association between age and functional outcome, potentially accounting for the reduced apparent benefit of mechanical thrombectomy (MT) in elderly patients in the RESILIENT trial.
Methods
RESILIENT was a multicenter, prospective, randomized, open‐label trial with blinded outcome assessment conducted in Brazil. Patients with anterior circulation large‐vessel occlusion stroke were randomized to MT plus guideline‐based care or guideline‐based care alone, including intravenous alteplase when eligible. A vascular neurologist blinded to clinical data evaluated baseline CT scans for cSVD markers (leukoaraiosis, lacunes, and atrophy) to derive a composite cSVD score. Multivariable logistic regression identified independent predictors of good functional outcome (modified Rankin Scale [mRS] = 0–2 at 90 days). Treatment effects were assessed across subgroups defined by age (<70 or ≥70 years) and cSVD score (0–1 vs 2–3). Mediation analysis quantified the indirect effect of age on outcome through cSVD.
Results
Patients with good outcomes were younger, had lower National Institutes of Health Stroke Scale (NIHSS) scores, better collaterals, lower glucose levels, lower cSVD burden, and were more frequently treated with MT. Independent predictors of good outcome included lower NIHSS, MT, lower cSVD score, and lower glucose levels. MT benefit was restricted to patients <70 years with low cSVD burden (odds ratio [OR] = 4.16, 95% confidence interval [CI] = 1.6–10.4, p < 0.01). No benefit was observed in older patients or in those with cSVD > 1. Mediation analysis showed that cSVD significantly mediated the association between age and outcome (average causal mediation effect [ACME] = –0.003, 95% CI = –0.005 to 0.00, p = 0.010).
Interpretation
Baseline CT markers of cSVD were independently associated with poorer outcomes and mediated the association between age and functional outcome in the RESILIENT trial, potentially explaining the lack of MT efficacy in older patients. ANN NEUROL 2026;100:361–369
Since the initial trials demonstrating the overwhelming benefit of mechanical thrombectomy (MT) in acute large vessel occlusion strokes (LVOS) back in 2015, 1 there has been a continuous expansion on its indications with several additional studies reinforcing its robust efficacy in many different clinical scenarios and populations. 2 , 3 , 4 In 2020, the RESILIENT trial, a study conducted across 12 public hospitals within the Brazilian universal public health care system, established the safety and efficacy of MT in low‐resource settings. 2 However, the elderly trial participants (≥70 years old) did not seem to benefit from endovascular treatment. These findings had not been observed in any of the previous MT randomized trials, which all derived from high‐income countries and have invariably demonstrated the preservation of treatment effect across all age groups. 4
We hypothesize that baseline CT markers of cerebral small vessel disease (cSVD), as a surrogate of brain frailty in the socioeconomically vulnerable population treated within the Brazilian public health care system, may partly account for the association between age and poorer functional outcomes, and help explain the observed attenuation of treatment benefit among elderly patients. 5 , 6 , 7
Methods
This is a post hoc analysis of the RESILIENT trial population. RESILIENT was a multicenter, prospective, randomized, open‐label, controlled trial with a blinded central outcome evaluation conducted in the Brazilian public health care system. Patients were assigned in a 1:1 ratio to be treated with MT and guideline‐based care (thrombectomy group) or with guideline‐based care alone (control group), including intravenous alteplase for eligible patients within 4.5 hours after the onset of symptoms. Patients were eligible for inclusion in the trial if they were 18 years of age or older; had an occlusion involving the intracranial internal carotid artery, the first segment of the middle cerebral artery (M1), or both that could be treated within 8 hours after symptom onset (defined as the time when the patient was last seen in a normal state of health); had a pre‐stroke score of 0 or 1 on the modified Rankin scale (mRS; range = 0 [no symptoms] to 6 [death]); and had a score of 8 or higher on the National Institutes of Health Stroke Scale (NIHSS; range = 0 to 42, with higher values indicating more severe deficit) at presentation. Details on the RESILIENT trial protocol, statistical analysis, baseline characteristics, and main results have been published previously. 2
The National Regulatory Agency in Research and the institutional review board of each participating center approved the trial protocol (RESILIENT ClinicalTrials.gov number, NCT02216643). Enrolled patients or their surrogates provided written informed consent.
Imaging and Outcome
Baseline computed tomography (CT) scans were analyzed by a single trained vascular neurologist blinded to clinical details and randomization allocation. According to the methods reported by Appleton et al, 6 the cSVD burden was assessed based on a composite score for the presence, location, and size of baseline imaging changes compared to a standard template as follows: (1) severe atrophy (present in cortical and/or central regions), (2) severe leukoaraiosis (present in anterior and/or posterior regions), and (3) presence of old lacunar infarcts, to a maximum score of 3. The primary end point was a good outcome at 90 days, defined by an mRS between 0 and 2, denotating functional independence. All randomized patients had baseline CT imaging available and were included in the present analysis.
Statistical Analysis
Continuous variables were reported as mean ± SD or median ± interquartile range. Categorical variables were reported as numbers and proportions. Statistical significance for intergroup differences was assessed using the χ 2 test for categorical variables and the Mann Whitney U test and Student t test for continuous variables, applying the Shapiro–Wilk test for normality. For multivariable analyses, we conducted an exploratory, clinically grounded multivariable penalized model (Firth's penalized logistic regression) including age, baseline NIHSS, collateral score, glucose, ASPECTS, thrombectomy allocation, and cSVD score. To minimize small‐sample bias, we also performed a sensitivity analysis penalized logistic regression with a reduced 6‐variable model (age, NIHSS, collaterals, glucose, thrombectomy, and cSVD score; Supplementary Material). Variance inflation factors (VIFs) were calculated to assess multicollinearity. Exploratory interaction terms among continuous age, cSVD score, and thrombectomy allocation were also tested. Correlation between age and cSVD score (rho) was assessed using Spearman's rank correlation coefficient.
A forest plot was generated as an exploratory analysis to examine the effect of MT in subgroups defined by age (<70 vs ≥70 years) and cSVD score (0–1 vs >1). Adjustment for multiple testing was performed using the Holm procedure. In addition, a mediation analysis was conducted to evaluate the relationship among age, cSVD score, and functional outcome.
Causal mediation analysis was conducted using the mediation package in R software. We specified a logistic regression model including age, cSVD score, thrombectomy, NIHSS, ASPECTS, glucose, and collateral status. The average causal mediation effect (ACME), average direct effect (ADE), total effect, and proportion mediated were estimated. To account for the non‐normal sampling distribution of the indirect effect, we used nonparametric bootstrap resampling with 1,000 simulations and reported percentile‐based 95% confidence intervals (CIs).
Analyses were performed using the R software version 4.0.4, the JASP software version 0.16.1, Python (version 3.x) with pandas for data manipulation, stats models for statistical modeling, and matplotlib and networks for visualizations.
Results
All 221 randomized patients from the RESILIENT trial were included in this analysis. One hundred four (47%) were women, 103 (46.6%) were White, and approximately 70% received intravenous thrombolysis. Hypertension was the most common cardiovascular risk factor, present in over 60% of patients. Cardioembolism was the most prevalent stroke etiology, responsible for 42.1% of the events. Almost 80% of the patients had MCA M1 occlusions (Table 1).
TABLE 1.
Demographic, Clinical Measures, and Imaging Findings Between the Two Groups and the Overall Participants
| Variables | All (n = 221) | mRS 0–2 (n = 61) | mRS 3–6 (n = 160) | p |
|---|---|---|---|---|
| Age (median, IQR) | 67 (54–76) | 61 (46–71) | 68 (57.7–76) | < 0.01 |
| F sex (%) | 104 (47) | 27 (44.3) | 77 (48.1) | 0.71 |
| White race (%) | 118 (53.4) | 29 (47.5) | 89 (55.6) | 0.35 |
| NIHSS score (median, IQR) | 17.8 (14–21) | 15 (11–19) | 19 (15.7–22) | < 0.01 |
| Hypertension (%) | 145 (65.6) | 38 (62.3) | 107 (66.9) | 0.63 |
| History of stroke or TIA (%) | 25 (24.7) | 9 (14.7) | 16 (10) | 0.44 |
| Atrial fibrillation (%) | 27 (12.2) | 8 (13) | 19 (12) | 0.98 |
| Diabetes mellitus (%) | 60 (27.1) | 10 (16.4) | 50 (13.2) | 0.04 |
| Current or past tobacco use (%) | 39 (17.6) | 8 (13.1) | 31 (19.4) | 0.33 |
| Dyslipidemia (%) | 55 (24.9) | 15 (24.6) | 40 (25) | 1.00 |
| Heart failure (%) | 43 (19.4) | 8 (13.1) | 35 (21.9) | 0.20 |
| Previous mRS (median, IQR) | 0 (0–0) | 0 (0–0) | 0 (0–0) | 0.66 |
| Median systolic blood pressure (IQR) | 141 (130–160) | 140 (129–155) | 145.5 (130–166.2) | 0.03 |
| Median glucose level at arrival (IQR) nmol/liter | 124 (107–148) | 114 (94–131) | 126 (108–154) | < 0.01 |
| Treatment with intravenous alteplase (%) | 155 (70.1) | 43 (70.5) | 112 (70) | 1.00 |
| Treatment with thrombectomy (%) | 111 (50.2) | 39 (64) | 72 (45) | 0.02 |
| Baseline ASPECTS (median, IQR) | 8 (7–9) | 8 (7–8) | 8 (7–9) | 0.88 |
| Baseline collateral score (median, IQR) | 2 (2–3) | 2 (2–3) | 2 (2–2) | < 0.01 |
| TOAST classification | ‐ | ‐ | ‐ | 0.37 |
| Cardioembolism | 93 (42.1) | 24 (39.3) | 69 (43) | |
| Large artery atherosclerosis | 28 (12.6) | 6 (9.8) | 22 (13.7) | |
| Small vessel disease | 0 (0) | 0 (0) | 0 (0) | |
| Stroke of other determined etiology | 5 (2.2) | 3 (4.9) | 2 (1.2) | |
| Stroke of undetermined etiology | 90 (40.7) | 25 (41) | 65 (40.6) | |
| Recanalization (TICI 2b/3) (%) | 90 (40.7) | 34 (55.7) | 56 (35) | 0.31 |
| Extracranial ICA occlusion (%) | 12 (5.4) | 3 (4.9) | 9 (5.6) | 1.00 |
| Intracranial arterial occlusion (%) | ‐ | ‐ | ‐ | 0.06 |
| ICA | 42 (19) | 6 (9.8) | 36 (22.5) | |
| M1 | 176 (79.6) | 54 (88.5) | 122 (76.2) | |
| M2 | 3 (1.3) | 1 (1.6) | 2 (1.2) | |
| Median time from stroke onset to rtPA (IQR) min | 168 (125–215) | 174 (126–215) | 166 (125–211) | 0.65 |
| Median time from stroke onset to groin puncture (IQR) min | 265 (221–333) | 261.5 (223.7–326.2) | 280 (220–334) | 0.45 |
| cSVD score (median, IQR) | 1 (0–2) | 0 (0–1) | 1 (0–2) | < 0.01 |
cSVD = cerebral small vessel disease; ICA = internal carotid artery; IQR = interquartile range; M1 and M2 = segments of middle cerebral artery; mRS = modified Rankin scale; NIHSS = National Institute of Health Score Scale; TIA = transient ischemic stroke; TICI = thrombolysis in cerebral infarction grading system; TOAST = Trial of ORG 10172 in Acute Stroke Treatment.
Note: The p values in bold represent statistical significance.
Patients with good outcomes (mRS = 0–2) at 90 days were younger, had lower baseline NIHSS scores, glucose levels at admission, better collaterals, higher rates of MT, and lower cSVD scores (see Table 1). Among the 89 patients ≥70 years, 42 (47.2%) had a cSVD score higher than 1, with a modest positive correlation between the score and age (rho = 0.47, p < 0.01; Fig 1).
FIGURE 1.
Association between age and cSVD score. cSVD = cerebral small vessel disease.
After adjusting for the aforementioned variables, only NIHSS (odds ratio [OR] = 0.87; 95% CI = 0.81–0.93, p < 0.01); treatment with MT (OR = 2.24, 95% CI = 1.15–4.46, p = 0.02); cSVD score (OR = 0.57, 95% CI = 0.36–0.88, p = 0.01); and baseline glucose levels (OR = 0.99, 95% CI = 0.98–0.99, p = 0.02) remained associated with good outcome at 90 days (Table 2 and Appendix S1). Notably, age was not independently associated with good outcome (OR = 0.99, 95% CI = 0.96–1.01, p = 0.28). Variance inflation factors were <2.5 for all predictors, suggesting no significant multicollinearity. We additionally tested for interaction among continuous age, cSVD score, and thrombectomy allocation. No statistically significant interaction was observed. However, this analysis was exploratory and limited by the modest sample size. [Correction added on 17 July 2026, after first online publication: Appendix 1 has been removed from the article and included as Supporting Information in this version.]
TABLE 2.
Multivariate Penalized Logistic Regression for Good Outcome (mRS = 0–2) RESILIENT Participants
| Variables | OR | 95% CI | p |
|---|---|---|---|
| NIHSS | 0.87 | 0.81–0.93 | < 0.01 |
| Mechanical thrombectomy | 2.24 | 1.15–4.46 | 0.02 |
| cSVD Score | 0.57 | 0.36–0.88 | 0.01 |
| Glucose levels at arrival | 0.99 | 0.98–0.99 | 0.02 |
Note: Adjusted for age, systolic blood pressure at arrival, and current or past tobacco use.
CI = confidence interval; cSVD = cerebral small vessel disease; mRS = modified Rankin scale; NIHSS = National Institute of Health Score Scale; OR = odds ratio.
Note: The p values in bold represent statistical significance.
After stratifying by age and cSVD scores, treatment with MT was significantly associated with good outcome in patients aged <70 years with a cSVD score of 0 to 1 (n = 112 patients, OR = 4.16, 95% CI = 1.6–10.4, p < 0.01, adjusted p < 0.01) but significance was not reached in those aged <70 years with a cSVD score of > 1 (n = 20 patients, OR = 2.95, 95% CI = 0.20–43.7, p = 0.43, adjusted p = 1.00). No significant association between MT and good outcomes was found in patients ≥70 years old (cSVD score = 0–1: 47 patients; OR = 0.84, 95% CI = 0.20–3.53, p = 0.81, adjusted p = 1.00 vs >1: 42 patients; OR = 0.76, 95% CI = 0.12–4.63, p = 0.76, adjusted p = 1.00; Fig 2).
FIGURE 2.
Effect of mechanical thrombectomy across age and cSVD subgroups. cSVD = cerebral small vessel disease. [Color figure can be viewed at www.annalsofneurology.org]
The effect of age on functional independence was mediated by cSVD burden. The analysis demonstrated a significant indirect effect: approximately 53% of the detrimental effect of age on good outcome was mediated by cSVD (ACME = −0.003, 95% CI = –0.005 to 0.00, p = 0.010). The direct effect of age, independent of cSVD, was not statistically significant (ADE p = 0.278). The total effect of age on outcome remained significant (p = 0.018; Fig 3).
FIGURE 3.
The cSVD score as a mediator of the association between age and good outcome. cSVD = cerebral small vessel disease.
Discussion
This post hoc analysis of the RESILIENT trial provides unique insights into the associations among age, cSVD burden, and the benefit of endovascular thrombectomy. Specifically, we have demonstrated that cSVD burden rather than age alone seems to account for the lack of benefit of MT in the elderly population of the RESILIENT trial. We used a composite score of cSVD imaging markers on the baseline CT as a surrogate for frailty. We investigated its influence on the outcomes of RESILIENT patients, highlighting an independent relationship between cSVD burden and good outcome, and suggesting a significant and robust effect of cSVD burden as a key player in the mediation between age and functional outcome.
It is important to distinguish the influence of age as a prognostic factor from its role as a potential treatment effect modifier in mechanical thrombectomy. Elderly patients undergoing thrombectomy typically have worse functional outcomes as compared with their younger counterparts; however, in the absence of thrombectomy, the elderly fare disproportionally worse relatively to the young, and thus the gradient between treatment and non‐treatment (eg, treatment benefit) is typically maintained across age groups. In other words, whereas age has been consistently identified as a strong predictor of outcome in thrombectomy, age has not been a treatment effect modifier, at least across high‐income country populations. 8 Indeed, randomized trials have demonstrated broad benefit of thrombectomy without evidence of treatment effect modification by age, 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 with patient‐level meta‐analyses from HERMES (early window) and AURORA (late window) even showing numerically greater, although not statistically significant, benefit among older compared with younger patients. 1 , 17 In contrast, findings from the RESILIENT trial suggest that age may act as a treatment effect modifier in lower‐resource settings. Specifically, patients younger than 70 years experienced one of the largest treatment effects reported across thrombectomy trials (90‐day mRS = 0–2: OR = 4.95, 95% CI = 2.10–11.89), whereas the benefit was significantly attenuated or lost in older patients (≥70 years, OR = 1.08, 95% CI = 0.30–28.93, p interaction = 0.03 and ≥80 years, OR = 0.42, 95% CI = 0.03–6.31, p interaction = 0.02). 2 This novel observation may reflect the unique socioeconomic characteristics of our trial population and motivated the present investigation into potential underlying biological mechanisms.
What could explain the lack of benefit from MT in the elderly population of RESILIENT? The robust benefit observed among patients younger than 70 years suggests that this finding was not attributable to procedural results or health care infrastructure. Rather, the most plausible explanation lies in the characteristics of the study population. Our analysis supports this interpretation, demonstrating that cSVD burden on baseline imaging appeared to mediate much of the detrimental effect of age on functional outcomes. The cSVD score, derived from a composite assessment of cortical and/or central atrophy, leukoaraiosis, and old lacunar infarcts, reflects key components commonly used in the evaluation of brain frailty.
Brain frailty refers to a state of diminished neurocognitive reserve and greater vulnerability due to structural and functional changes in the brain, including neuronal loss, reductional synaptic density, and decreased neuroplasticity, predisposing patients to poor outcomes. 5 Brain frailty occurs as a response to different stressors, including but not limited to aging. Indeed, it is well‐known that the more vulnerable stroke population from low‐ and middle‐income countries have worse outcomes and a higher case fatality than those from high‐income countries. 18 These differences in outcomes are likely driven by socioeconomic disparities 19 , resulting in a higher burden of poorly controlled comorbidities, worse nutrition status, and ultimately greater brain vulnerability. Several studies have proposed various methods to quantify brain frailty through the neuroimaging assessment based mainly on the burden of cSVD markers, with strong association with long‐term cognitive outcomes. 5 , 20 , 21
The cumulative evidence supports adopting a biological rather than purely chronological approach to age in treatment selection and prognostication for patients with stroke. This perspective was highlighted in a recent secondary analysis of the ESCAPE‐NA1 trial, where investigators demonstrated that the association between age and 90‐day outcomes after MT was predominantly mediated by neuroimaging markers of frailty—such as brain atrophy and small vessel disease—rather than chronological age alone. 22
In line with these findings, our study showed that younger patients with a lower cSVD burden experienced the greatest benefit from MT. Indeed, mediation analysis revealed a modest direct effect of age on functional outcomes, whereas the cSVD score exhibited a strong negative association with good outcomes. Although no formal evidence of effect modification by cSVD was observed, the consistency of findings across multivariable and mediation analyses supports its role as a key biological determinant of outcome regarding age.
Furthermore, the association between cSVD and poor functional outcomes likely reflects a state of Systemic Microvasculature Frailty. In this framework, visible markers of cerebral microangiopathy serve as a surrogate for global endothelial dysfunction that extends beyond the neurovascular unit to other low‐resistance organs, such as the kidneys and heart. 23 , 24 Recent evidence suggests that a high brain frailty score is an independent predictor of contrast‐associated acute kidney injury following thrombectomy, 23 , 25 significantly outperforming individual imaging markers. This systemic vulnerability may narrow the therapeutic window in elderly patients. For instance, the same microvascular alterations associated with cSVD may predispose to both impaired vascular integrity and increased susceptibility to hemorrhagic transformation. 26 These risks may be further amplified by the frequent need for anticoagulation in conditions such as atrial fibrillation, 27 as well as by the presence of prior structural brain injury, 28 potentially contributing to the attenuated treatment benefit observed in the RESILIENT trial, and possibly in similar settings.
Beyond individual clinical outcomes, the identification of brain frailty has profound socioeconomic and systemic implications, particularly in resource‐limited health care settings. Systemic complications, such as contrast‐associated acute kidney injury and hemorrhagic transformation, 23 are major drivers of increased hospitalization costs and prolonged length of stay, highlighting the importance of optimizing the cost‐effectiveness of mechanical thrombectomy.
Limitations
This study has important limitations. First, this was a post hoc analysis and the overall sample size was modest (221 patients in total, including only 89 aged ≥70 years), making the results susceptible to chance. The relatively small number of elderly participants and of favorable outcomes (n = 61 overall) also limits the stability of multivariable models and raises concern for overfitting. To address this, we used a clinically grounded multivariable model and performed a sensitivity analysis using Firth penalized logistic regression; both consistently identified the cSVD score, rather than age, as an independent predictor. Nonetheless, the limited number of events still warrants caution in interpretation.
Second, subgroup and interaction analyses should be regarded as exploratory, given the small sample sizes, wide CIs, and limited statistical power. These analyses are hypothesis‐generating, and adequately powered prospective studies will be required to determine the influence of age more definitively, cSVD burden, and other markers of frailty on the benefit of thrombectomy.
Third, baseline imaging in the RESILIENT trial relied primarily on CT, and larger chronic non‐lacunar infarcts were not specifically assessed in the current analysis, limiting a more comprehensive evaluation of brain frailty. Our imaging analysis was specifically focused on CT markers of cerebral small vessel disease. Because chronic infarcts may be clinically silent, their potential contribution cannot be excluded and warrants further investigation. Despite the blinded neuroimaging assessment, there is some subjectivity in the current methods to evaluate cSVD imaging markers, which could potentially result in interpretation bias. Fourth, imaging markers were assessed by a single rater, introducing the potential for rater bias. Artificial intelligence algorithms have been recently developed to assess brain frailty and may provide greater objectivity. 29 , 30 , 31 , 32 , 33 , 34 Incorporating measures of physical frailty may also have refined our understanding of vulnerability in this patient population.
Although the RESILIENT trial was completed prior to the coronavirus disease 2019 (COVID‐19) pandemic, this post hoc analysis was conducted during that period, and external systemic factors may have indirectly influenced healthcare delivery and outcomes. 35 , 36 , 37 , 38 In addition, despite being conducted in academic centers, disparities in neurocritical care across Brazil may have impacted post‐stroke management and recovery, potentially affecting functional outcomes. 39
Finally, our analysis focused on good outcomes (90‐day mRS = 0–2) rather than overall functional shift, so it remains possible that patients with brain frailty may still experience meaningful reductions in disability even if they do not achieve full independence. Future analyses using the combined datasets from the RESILIENT and RESILIENT‐Extend trials are planned to further explore these questions.
Conclusions
Baseline CT markers of cSVD were independently associated with poorer outcomes in patients with LVO and may underlie the lack of MT benefit observed among older patients in the RESILIENT trial. Further studies are warranted to clarify the interplay among cSVD, brain frailty, and treatment effect of thrombectomy to validate this hypothesis.
Author Contributions
T.O.G. and R.K.M.F. contributed to the conception and design of the study. T.O.G., L.A.C., O.P.N., F.O.L., L.C.R., R.K.M.F., M.B.F., F.M., B.S.M.P., M.R.C., D.B., G.S.S., V.F.Z., J.M.O.F., D.C.B., D.S.L., S.C.O.M., and R.G.N. contributed to the acquisition and analysis of data. R.K.M.F. and R.G.N. contributed to drafting the text or preparing the figures.
Potential Conflicts of Interest
R.K.M.F. reports receiving consulting fees from Allm; lecture fees from Bayer, Medtronic, and Pfizer; and advisory board and lecture fees from Boehringer Ingelheim. J.M.M.A. reports receiving consulting fees from Balt USA, Medtronic, MicroVention, Penumbra, and Stryker. D.S. reports consulting fees from Boehringer Ingelheim and lecture fees from Pfizer. F.O.L. reports lecture fees from Boehringer Ingelheim. M.B.F. reports grant support from Johnson & Johnson Medical Devices and Diagnostics Group–Latin America, and lecture fees and travel support from Medtronic Vascular. V.F.Z. reports lecture fees from Boehringer Ingelheim and travel support from Pfizer. D.B. reports travel support from Medtronic. D.C. reports consulting fees from Allm, lecture fees from Boehringer Ingelheim, and travel support from iSchemaView. D.L. reports consulting fees and travel support from Cerenovus, consulting fees from Genentech and Medtronic, and consulting and lecture fees from Stryker. O.P.N. reports lecture fees from Boehringer Ingelheim, Medtronic, and Pfizer. R.G.N. reports consulting fees from Anaconda Biomed, Genentech, Imperative Care, Phenox, Prolong Pharmaceuticals, and Stryker; advisory board fees from Biogen and Cerenovus; consulting fees from and stock ownership in Brainomix, Corindus, Perfuze, and Viz.ai; and advisory board fees and stock ownership in Ceretrieve and Vesalio. The remaining authors have nothing to report.
Supporting information
Data S1: Supporting Information.
Appendix S1: Sensitivity Analysis With Six‐Variable Model for Good Outcome (mRS 0–2) (Firth Regression).
Acknowledgment
Supported by the Ministry of Health of Brazil, Departamento de Ciência e Tecnologia, and the Conselho Nacional de Desenvolvimento Científico e Tecnológico (grant numbers 402401/2013‐1; 401821/2015‐3; 402388/2013‐5; and 443861/2018‐8), with unrestricted device or software donations from Medtronic, Penumbra, iSchemaView, Brainomix, and Allm. This work is part of the doctoral thesis of T.O.G., supported by CAPES (Finance Code 001).
Data Availability
All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.
References
- 1. Goyal M, Menon BK, Van Zwam WH, et al. Endovascular thrombectomy after large‐vessel ischaemic stroke: a meta‐analysis of individual patient data from five randomised trials. Lancet 2016;387:1723–1731. [DOI] [PubMed] [Google Scholar]
- 2. Martins SO, Mont'Alverne F, Rebello LC, et al. Thrombectomy for stroke in the public health care system of Brazil. N Engl J Med 2020. Jun;382:2316–2326. [DOI] [PubMed] [Google Scholar]
- 3. McDonough RV, Ospel JM, Campbell BCV, et al. Functional outcomes of patients ≥85 years with acute ischemic stroke following EVT: a HERMES substudy. Stroke 2022;53:2220–2226. [DOI] [PubMed] [Google Scholar]
- 4. Nogueira RG, Lima FO, Pontes‐Neto OM, et al. Randomization of endovascular treatment with stent‐retriever and/or thromboaspiration versus best medical therapy in acute ischemic stroke due to large vessel occlusion trial: rationale and design. Int J Stroke 2021;16:100–109. [DOI] [PubMed] [Google Scholar]
- 5. Appleton JP, Woodhouse LJ, Adami A, et al. Imaging markers of small vessel disease and brain frailty, and outcomes in acute stroke. Neurol Int 2020;94:e439–e452. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Yassi N, Flicker L, Campbell BCV. Brain frailty and small vessel disease for stroke outcome prediction: are we there yet? Neurology 2020;94:191–192. [DOI] [PubMed] [Google Scholar]
- 7. Evans NR, Wall J, To B, et al. Clinical frailty independently predicts early mortality after ischaemic stroke. Age Ageing 2020;1–4:588–591. [DOI] [PubMed] [Google Scholar]
- 8. Nogueira RG, Ribó M. Endovascular treatment of acute stroke. Stroke 2019;50:2612–2618. [DOI] [PubMed] [Google Scholar]
- 9. Nogueira RG, Jadhav AP, Haussen DC, et al. Thrombectomy 6 to 24 hours after stroke with a mismatch between deficit and infarct. N Engl J Med 2018;378:11–21. [DOI] [PubMed] [Google Scholar]
- 10. Albers GW, Marks MP, Kemp S, et al. Thrombectomy for stroke at 6 to 16 hours with selection by perfusion imaging. N Engl J Med 2018;378:708–718. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Tao C, Nogueira RG, Zhu Y, et al. Trial of endovascular treatment of acute basilar‐artery occlusion. N Engl J Med 2022;387:1361–1372. [DOI] [PubMed] [Google Scholar]
- 12. Jovin TG, Li C, Wu L, et al. Trial of thrombectomy 6 to 24 hours after stroke due to basilar‐artery occlusion. N Engl J Med 2022;387:1373–1384. [DOI] [PubMed] [Google Scholar]
- 13. Sarraj A, Hassan AE, Abraham MG, et al. Trial of endovascular thrombectomy for large ischemic strokes. N Engl J Med 2023;388:1259–1271. 10.1056/NEJMoa2214403. [DOI] [PubMed] [Google Scholar]
- 14. Yoshimura S, Sakai N, Yamagami H, et al. Endovascular therapy for acute stroke with a large ischemic region. N Engl J Med 2022;386:1303–1313. 10.1056/NEJMoa2118191. [DOI] [PubMed] [Google Scholar]
- 15. Huo X, Ma G, Tong X, et al. Trial of endovascular therapy for acute ischemic stroke with large infarct. N Engl J Med 2023;388:1272–1283. 10.1056/NEJMoa2213379. [DOI] [PubMed] [Google Scholar]
- 16. Bendszus M, Bonekamp S, Herweh C, et al. Endovascular thrombectomy for acute ischaemic stroke with established large infarct: multicentre, open‐label, randomised trial. Lancet (London, England) 2023;402:1753–1763. [DOI] [PubMed] [Google Scholar]
- 17. Jovin TG, Nogueira RG, Lansberg MG, et al. Thrombectomy for anterior circulation stroke beyond 6 h from time last known well (AURORA): a systematic review and individual patient data meta‐analysis. Lancet 2021;398:139–147. [DOI] [PubMed] [Google Scholar]
- 18. Yusuf S, Rangarajan S, Teo K, et al. Cardiovascular risk and events in 17 low‐, middle‐, and high‐income countries. N Engl J Med 2014;371:818–827. [DOI] [PubMed] [Google Scholar]
- 19. Ouriques Martins SC, Pontes‐Neto OM, Pille A, et al. Reperfusion therapy for acute ischemic stroke: where are we in 2023? Arquivos de Neuro‐Psiquiatria Arq Neuropsiquiatr 2023;81:1030–1039. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Bu B, Li J, Chen S, et al. Imaging markers of brain frailty and outcome in patients with acute ischemic stroke. Stroke 2021;52:1004–1011. 10.1161/STROKEAHA.120.029841. [DOI] [PubMed] [Google Scholar]
- 21. Debette S, Markus HS. The clinical importance of white matter hyperintensities on brain magnetic resonance imaging: systematic review and meta‐analysis. BMJ 2010;341:c3666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Benali F, Singh N, Fladt J, et al. Mediation of age and thrombectomy outcome by neuroimaging markers of frailty in patients with stroke. JAMA Netw Open 2023;6:e2349628. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Goulart TO, Esaki AS, de Andrade JBC, et al. Systemic microvasculature frailty: brain frailty score predicts contrast‐associated acute kidney injury after thrombectomy for stroke. Clin Neurol Neurosurg 2026;263:109323. 10.1016/j.clineuro.2026.109323. [DOI] [PubMed] [Google Scholar]
- 24. Zhang Y, Hughes T, Zhang W, et al. Abstract FR502: cerebral perfusion in the heart‐brain Axis: cardiovascular determinants and association with subclinical brain injury. Hypertension 2025;82. 10.1161/hyp.82.suppl_1.fr502. [DOI] [Google Scholar]
- 25. Goulart TO, Schirmer MD, Bonkhoff AK, et al. Abstract 414: association of atrophy, lacunes, and white matter hyperintensities with outcomes after thrombectomy: a Bayesian analysis. Stroke Vasc Interv Neurol 2025;5:e70000_414. 10.1161/svi270000_414. [DOI] [Google Scholar]
- 26. Goulart TO, Schirmer MD, Bonkhoff AK, et al. Abstract 415: pre‐treatment hemorrhagic infarct is associated with worse outcome after mechanical thrombectomy: a multicenter Bayesian analysis. Stroke Vasc Interv Neurol 2025;5:e70000_415. 10.1161/svi270000_415. [DOI] [Google Scholar]
- 27. Clares de Andrade JB, Pisa I, de Oliveira NS, et al. Predicting incident atrial fibrillation after stroke: a scoping review of clinical scores, biomarkers, and AI‐enhanced strategies. Cerebrovasc Dis Extra 2025;16:1–27. 10.1159/000550097. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Goulart TO, Wang C, Fuse K, et al. Intracranial hemorrhage with direct Oral anticoagulants vs low‐molecular‐weight heparin in brain tumors. Neurology 2025;105:e214140. 10.1212/wnl.0000000000214140. [DOI] [PubMed] [Google Scholar]
- 29. Jiang Q, Li C, Zhang W, et al. Application of artificial intelligence‐based magnetic resonance imaging in cerebral small vessel disease: a review. CNS Neurosci Ther 2022;30:e14841. 10.1111/cns.14841. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Goulart TO. Algoritmos de inteligência artificial para avaliação da doença cerebral de pequenos vasos e sua associação com desfechos clínicos em pacientes com AVC isquêmico submetidos à trombectomia mecânica (Doctoral dissertation, Universidade de São Paulo), 2024.
- 31. Goulart T, Junior JN, Miyaji RO, et al. Abstract WP192: differentiation between contrast extravasation and hemorrhagic transformation in tomography of acute ischemic stroke after thrombectomy with artificial intelligence. Stroke 2025;56:AWP288–AWP192. 10.1161/str.56.suppl_1.wp192. [DOI] [Google Scholar]
- 32. Goulart TO, Miyaji RO, Nather Júnior JC, et al. Deep learning and machine learning for differentiation between contrast extravasation and hemorrhagic transformation in post‐thrombectomy stroke CT. J Neuroradiol 2026;53:101530. 10.1016/j.neurad.2026.101530. [DOI] [PubMed] [Google Scholar]
- 33. Zhang F, Wang S, Hu Y, et al. Fully automatic classification of brain atrophy on NCCT images in cerebral small vessel disease. Front Neurol 2022;13:846348. 10.3389/fneur.2022.846348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Khan A, Ahmed M, Zhao W, Li F. Computer‐aided extraction of select MRI markers of cerebral small vessel disease: a systematic review. ArXiv, 2022. Available at: https://arxiv.org/abs/2204.01411. [DOI] [PubMed]
- 35. Goulart TO. Análise epidemiológica de internações, óbitos, custos, tratamentos e procedimentos relacionados a doenças cerebrovasculares no Brasil, nos períodos pré, intra e pós‐pandemia de COVID‐19, 2023.
- 36. Goulart TO, Marazzi TBM, Monteiro RA, et al. Nationwide epidemiological study of subarachnoid hemorrhage: trends in admissions, mortality, seasonality, costs, clipping, embolization, and the impact of COVID‐19. Front Neurol 2025;16:1630224. 10.3389/fneur.2025.1630224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Goulart TO, Marazzi TBM, Monteiro RA, et al. Spontaneous intracranial hemorrhage in Brazil: trends, seasonality, pandemic disruption, and cost burden. World Neurosurg 2026;207:124793. 10.1016/j.wneu.2026.124793. [DOI] [PubMed] [Google Scholar]
- 38. Goulart TO, Perdoná GS, Monteiro RA, et al. Abstract WP288: epidemiological analysis of incidence, treatment, mortality and costs of cerebrovascular diseases in Brazil between 2017 and 2022. Stroke 2024;55:AWP288–AWP288. 10.1161/str.55.suppl_1.wp288. [DOI] [Google Scholar]
- 39. Marazzi TBM, Camilo MR, Goulart TO, et al. Is it time to develop neurocritical care education in Brazil? Analysis of the Brazilian national survey on education in neurocritical care. Arq Neuropsiquiatr 2026;84:001–008. 10.1055/s-0046-1816040. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data S1: Supporting Information.
Appendix S1: Sensitivity Analysis With Six‐Variable Model for Good Outcome (mRS 0–2) (Firth Regression).
Data Availability Statement
All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.