Association of cerebral collateral cascade with early neurological deterioration after endovascular thrombectomy in patients with acute ischemic stroke

Introduction

In patients with acute ischemic stroke (AIS) due to large-vessel occlusion (LVO), early neurological deterioration (END) presents a major clinical challenge, often resulting in poor outcomes despite timely recanalization (1,2). The mechanisms underlying END are multifactorial, with poor collateral circulation and impaired cerebral hemodynamics playing a central role (3,4).

Previous studies have shown that poor arterial collaterals are associated with fast infarct growth, unfavorable outcomes, and limited efficacy of reperfusion therapy (5-7). Venous outflow (VO), reflecting the efficiency of cerebral venous drainage, and the hypoperfusion intensity ratio (HIR), representing tissue-level collaterals, have also been linked to accelerated infarct progression and increased susceptibility to reperfusion injury (8,9). These observations highlight the importance of considering collateral circulation as an integrated, multilevel system.

The concept of the cerebral collateral cascade (CCC) involves the integration of arterial, venous, and tissue-level collateral perfusion into a functional hierarchy that reflects the sequential and interdependent maintenance of cerebral blood flow during ischemia and reperfusion (10). Disruption of this cascade can compromise hemodynamic stability. However, the role of the CCC in early neurological status after thrombectomy remains poorly understood. In this study, we examined whether disruption of the CCC is independently associated with END after endovascular thrombectomy (EVT) and evaluated its ability to serve as an early imaging biomarker for risk stratification and individualized post-reperfusion management. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2394/rc).

Methods

Study design and patient selection

We retrospectively analyzed data collected in an internal prospective stroke database at Xuanwu Hospital from January 2020 to December 2022. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments and was approved by Ethics Committee of Xuanwu Hospital. The requirement for individual consent was waived due to retrospective nature of this study. In this study, patients were classified into an END group and a non-END group based on changes in National Institutes of Health Stroke Scale (NIHSS) scores. Clinical and imaging characteristics were compared between the two groups. In addition, cerebral collateral status was evaluated according to the CCC, and patients were further categorized into a CCC− group (poor arterial, venous, and tissue-level collaterals), CCC+ group (good arterial, venous, and tissue-level collaterals), and CCC_mixed (the remainder of the patients). Comparisons of clinical and imaging parameters were then performed across the three CCC categories.

The inclusion criteria were as follows: (I) age ≥18 years; (II) LVO in the anterior circulation, including the terminal portion of the internal carotid artery (ICA) or the M1 or proximal M2 segments of the middle cerebral artery (MCA); (III) a one-stop computed tomography (CT) scan within 24 hours of last known well time, including noncontrast CT, CT perfusion (CTP), and head-neck CT angiography (CTA); (IV) EVT as determined by clinical assessment; and (V) END, defined as an increase of ≥4 points in the NIHSS at 24 hours after EVT compared with baseline.

Meanwhile, the exclusion criteria were (I) a lack of clinical or imaging data; (II) poor image quality due to motion artifacts; and (III) inability to adequately evaluate veins on baseline CTA due to insufficient visualization.

Clinical characteristics

Clinical, treatment, and follow-up information was obtained from electronic medical records, including age, sex, cerebrovascular risk factors, NIHSS, the Modified Rankin Scale (mRS), onset-to-door time, door-to-imaging time, and Alberta Stroke Program Early CT Score (ASPECTS).

Incomplete reperfusion was defined as a Modified Thrombolysis in Cerebral Infarction (mTICI) score of 0–2a. Patients underwent routine follow-up assessments at 90 days after discharge. An mRS score of 0–2 was considered a favorable outcome, whereas a score of 3–6 was considered an unfavorable outcome.

Scan technology and post-processing

Scans were performed with a 256-row helical CT scanner (Revolution CT, GE HealthCare, Chicago, IL, USA). Whole-brain CTP scanning parameters included a tube voltage of 70 kV, a tube current of 100 mAs, and a slice thickness of 5 mm. A high-pressure syringe was used to inject 40 mL of iopromide (370 mg/mL; Bayer, Leverkusen, USA) through the antecubital vein and subsequently 40 mL of isotonic saline, both at a rate of 6 mL/s. Dynamic perfusion scanning was initiated 5 seconds after contrast injection and continued for a total duration of 60 seconds. The acquisition consisted of 30 temporal phases.

For head-neck CTA, a tube voltage of 140 kV was used with automatic tube current modulation. A total of 55 mL of iopromide (370 mg/mL) and 40 mL of saline were injected at a rate of 5 mL/s. The CTA scanning range extended from the aortic arch to the skull base, with a slice thickness of 0.625 mm.

Image processing and collateral evaluation

Two radiologists, each with over 5 years of experience and blinded to clinical information, independently evaluated the imaging. Any disagreements were resolved by a senior radiologist.

CTP data were analyzed with artificial intelligence-powered SK-CTPDoc perfusion software (Shukun Technology, Beijing, China). The infarct core was defined as regions with relative cerebral blood flow (rCBF) <30%, and the total hypoperfused area was defined as regions with a time to maximum of the residue function (Tmax) >6 seconds (11,12). Penumbra volume was obtained by subtracting the infarct core from the hypoperfusion area.

Arterial collaterals and VO were evaluated on CTA, while tissue-level collaterals were derived from CTP. Arterial collaterals were graded via the Tan score through a comparison of pial arterial filling in the affected hemisphere with that of the contralateral hemisphere (13). Consistent with previous studies, Tan scores of 2–3 were considered favorable, whereas scores of 0–1 indicated unfavorable arterial collaterals. VO was evaluated via the Cortical Vein Opacification Score (COVES), which assesses opacification of the vein of Labbé, the superficial middle cerebral vein, and the sphenoparietal sinus (14). Each vein was scored as 0 (not visible), 1 (partial opacification), or 2 (complete opacification), yielding a total score ranging from 0 to 6. A COVES ≥3 was defined as favorable VO, while a score <3 was defined as unfavorable VO.

Tissue-level collaterals were quantified via the HIR, calculated as the volume ratio of Tmax >10 s to Tmax >6 s. An HIR ≤0.4 was considered a favorable tissue-level collateral, whereas an HIR >0.4 indicated an unfavorable one (15).

The CCC assessment included arterial, tissue, and venous level, as per the conceptual framework proposed by Faizy et al. (10). CCC− was defined as unfavorable arterial, venous, and tissue collaterals (Tan score <2, HIR >0.4, and COVES <3). CCC+ was defined as favorable status across all three components (Tan scale ≥2, HIR ≤0.4, and COVES ≥3). Meanwhile, the category of CCC_mixed included all remaining combinations.

Statistical analysis

All statistical analyses were performed with SPSS software version 26.0 (IBM Corp., Armonk, NY, USA). Continuous variables are expressed as median with interquartile range (IQR) and were compared with the Mann-Whitney test or Kruskal-Wallis test, as appropriate. Categorical variables are presented as counts with percentages and were compared with the Chi-squared or Fisher exact test. Interobserver agreement for the Tan score and COVES was assessed via the intraclass correlation coefficient (ICC). ICCs were calculated based on a two-way random-effects model with absolute agreement, and 95% confidence intervals (CIs) are reported. An ICC value >0.75 was considered indicative of good agreement. Spearman correlation analyses were performed to assess the relationships among arterial, venous, and tissue-level collaterals. Correlation coefficients were interpreted as follows: <0.3, weak; 0.3–0.5, moderate; and >0.5, strong.

Variables with a P value <0.05 in the univariate logistic analyses were incorporated into multivariable logistic regression models to identify independent predictors of END. A baseline clinical model was first established. Additional multivariable models were then constructed by separately incorporating the CCC classification, unfavorable Tan score (<2), unfavorable COVES (<3), and unfavorable HIR (>0.4). Model discrimination was compared with the area under the receiver operating characteristic (ROC) curve (AUC). Multicollinearity was tested via the variance inflation factor (VIF), with VIF >5 indicating significant collinearity.

Sensitivity analyses were additionally performed under alternative HIR thresholds (>0.3 and >0.5) and an alternative COVES cutoff (<4), as well as quantitative HIR and COVES measures; analyses were further repeated in patients without hemorrhagic transformation.

All statistical tests were two-tailed, and P<0.05 was considered statistically significant.

Results

Clinical and imaging profile of patients

A total of 293 patients were included in the final analysis, among whom 64 (21.8%) were in the END group and 229 (78.2%) in the non-END group. The median age of the cohort was 64 years (IQR, 54–70 years), with a median onset-to-door time of 330 minutes (IQR, 212–552 minutes), a baseline NIHSS of 14 (IQR, 11–19), and an ASPECTS of 8 (IQR, 7–9).

Patients with END, as compared to those without END, were more likely to have high blood pressure (HBP) (64.1% vs. 42.4%; P=0.002) and coronary heart disease (CHD) (26.6% vs. 15.7%; P=0.046). They also exhibited a higher median baseline NIHSS (END: 16, IQR, 12–20; no END: 13, IQR, 11–18; P=0.009), lower ASPECTS (END: 8, IQR, 5–9; no END: 8, IQR, 7–9; P=0.004), and larger infarct core volume (END: 39.80 mL, IQR, 9.55–94.50 mL; no END: 10.70 mL, IQR, 2.05–30.00 mL; P<0.001).

In terms of treatment and outcome parameters, patients with END, as compared to those without it, experienced longer median puncture-to-recanalization times (END: 45 min, IQR, 30–107 min; no END: IQR, 36 min, 25–50 min; P=0.008) and had a higher median proportion of incomplete reperfusion grade (mTICI 0–2a; 18.7% vs. 7.9%; P=0.011); they were also more prone to hemorrhagic transformation (53.1% vs. 31.9%; P=0.002). At 90 days, functional outcomes were significantly poorer in patients with END (median mRS 4, IQR, 2–6) than in those without END (median mRS 2, IQR, 1–4; P<0.001) (Table 1).

Regarding collateral characteristics, interobserver reliability for collateral assessments was high, with the Tan score yielding an ICC of 0.86 (95% CI: 0.83–0.89) and COVES yielding an ICC of 0.82 (95% CI: 0.78–0.86). Correlation analyses indicated significant associations among the three collateral components. The Tan score was negatively correlated with HIR (r=–0.345; P<0.001), the Tan score was positively correlated with COVES (r=0.197; P=0.001), and HIR was negatively correlated with COVES (r=–0.188; P=0.001).

Patients who experienced END demonstrated poorer collateral profiles compared to those without END; specifically, they had lower median arterial collateral scores (END: 1, IQR, 1–2; no END: 2, IQR, 1–2; P=0.017) and COVES (END: 2, IQR, 1–3; no END: 3, IQR, 2–4; P<0.001) and a higher median HIR (END: 0.49, IQR, 0.32–0.59; no END: 0.34, IQR, 0.15–0.50; P<0.001). When the Tan score, COVES, and HIR were dichotomized, the END group, as compared to the non-END group, had a higher proportion of unfavorable Tan (53.1% vs. 33.2%; P=0.004), unfavorable COVES (71.9% vs. 42.8%; P<0.001), and unfavorable HIR (67.2% vs. 39.3%; P<0.001); the END group also exhibited higher proportions of CCC– (34.4% vs. 10.9%; P<0.001) and CCC_mixed (62.5% vs. 59.4%; P<0.001) (Figure 1).

Figure 1 Comparison of collaterals between the END and non-END groups. The x-axis represents percentage. In the END group, the proportion of unfavorable Tan score (<2) was 53.1%, which was higher than the 33.2% in the non-END group (P=0.004). Unfavorable HIR (>0.4) was observed in 67.2% of patients with END and 39.3% of patients without END (P<0.001), while unfavorable VO (COVES <3) occurred in 71.9% and 42.8%, respectively (P<0.001). CCC– and CCC_mixed were more frequent in the END group (34.4% and 62.5%, respectively) than in the non-END group (10.9% and 59.4%, respectively; P<0.001). CCC, cerebral collateral cascade; COVES, Cortical Vein Opacification Score; END, early neurological deterioration; HIR, hypoperfusion intensity ratio; VO, venous outflow.

Group comparison based on CCC

The cohort included 47 (16.0%) patients in the CCC− group, 176 (60.1%) in the CCC_mixed group, and 70 (23.9%) in the CCC+ group. Table 2 provides the detailed baseline and clinical characteristics across CCC subgroups.

At baseline, patients with CCC− exhibited the most severe neurological deficits and had the highest median NIHSS score of 18 (IQR, 13–20), followed by the CCC_mixed group with 15 (IQR, 12–19) and the CCC+ group with 12 (IQR, 9–16) (P<0.001). The median onset-to-door time was shortest in the CCC− group (259 min; IQR, 181–441 min), followed by the CCC_mixed (325 min; IQR, 184–534 min) and CCC+ (387 min, IQR, 255–670 min) groups (P=0.024).

Regarding imaging characteristics, the median ASPECTS was lowest in the CCC− group (6, IQR, 4–9), intermediate in the CCC_mixed group (8, IQR, 7–9), and highest in the CCC+ group (9, IQR, 7–9; P<0.001). The median baseline infarct core volumes were significantly larger in the CCC− group (81.00 mL, IQR, 18.00–120.40 mL) than in the CCC_mixed (15.70 mL, IQR, 5.00–37.75 mL) and CCC+ (3.40 mL, IQR, 0.00–9.65 mL) groups (P<0.001). A similar pattern was observed for penumbral volume, which was greatest in the CCC− group (151.10 mL, IQR, 91.00–180.20 mL) than in the CCC_mixed (131.40 mL, IQR, 90.00–176.68 mL) and CCC+ (96.65 mL, IQR, 59.90–122.48 mL) groups (P<0.001).

At 24 hours after thrombectomy, the CCC− group had the highest median NIHSS (19, IQR, 14–40) compared with CCC_mixed (14, IQR, 6–20) and CCC+ (7, IQR, 4–12) groups (P<0.001). Significant differences in the proportion of END were observed between the three groups, with rates of 46.8%, 22.7%, and 2.9% in the CCC−, CCC_mixed, and CCC+ groups, respectively (P<0.001). At 90 days, outcomes were poorest in the CCC− group, with a median mRS score of 4 (IQR, 1–6), which was greater than the score of 2 (IQR, 1–4) in the CCC_mixed group and 1 (IQR, 1–3) in the CCC+ group (P<0.001) (Figures 2-4).

Figure 2 A 66-year-old male presented with occlusion of the M1 segment of the left middle cerebral artery. The Tan score was 0 (A). VO assessment showed no visualization of the sphenoparietal sinus (B, the arrow), superficial middle cerebral vein (C, the arrow), or vein of Labbé (D, the arrow), resulting in a COVES score of 0. Perfusion analysis demonstrated regions with differing degrees of perfusion delay. Blue indicates Tmax >4 s, green indicates Tmax >6 s, yellow indicates Tmax >8 s, and red indicates Tmax >10 s (E). The patient had a Tmax >6 s volume of 223.3 mL and a Tmax >10 s volume of 120.2 mL, yielding an HIR of 0.54. This patient was classified into the CCC– group, with a baseline NIHSS score of 19, a 24-hour post-EVT NIHSS score of 31 (END group), and a 90-day mRS score of 6. CCC, cerebral collateral cascade; COVES, Cortical Vein Opacification Score; END, early neurological deterioration; EVT, endovascular thrombectomy; HIR, hypoperfusion intensity ratio; mRS, Modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; Tmax, time to maximum of the residue function; VO, venous outflow.

Figure 3 An 81-year-old male presented with occlusion of the left internal carotid artery terminus. The Tan score was 3 (A). VO assessment showed no visualization of the sphenoparietal sinus (B, the arrow), normal visualization of the superficial middle cerebral vein (C, the arrow), and moderate opacification of the vein of Labbé (D, the arrow), resulting in a COVES score of 3. Perfusion analysis demonstrated regions with differing degrees of perfusion delay. Blue indicates a Tmax >4 s, green indicates a Tmax >6 s, and red indicates a Tmax >10 s (E). The patient had a Tmax >6 s volume of 122.3 mL and a Tmax >10 s volume of 0 mL, yielding an HIR of 0. This patient was classified into the CCC+ group, with a baseline NIHSS of 9, a 24-hour post-EVT NIHSS of 3 (non-END group), and a 90-day mRS score of 2. CCC, cerebral collateral cascade; COVES, Cortical Vein Opacification Score; END, early neurological deterioration; EVT, endovascular thrombectomy; HIR, hypoperfusion intensity ratio; mRS, Modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; Tmax, time to maximum of the residue function; VO, venous outflow.

Figure 4 A 65-year-old male presented with occlusion of the left internal carotid artery terminus. The Tan score was 2 (A). VO assessment showed no visualization of the sphenoparietal sinus (B, the arrow) or superficial middle cerebral vein (C, the arrow) and moderate opacification of the vein of Labbé (D, the arrow), resulting in a COVES score of 1 (B-D). Perfusion analysis demonstrated regions with differing degrees of perfusion delay. Green indicates a Tmax >6 s, and red indicates a Tmax >10 s (E). The patient had a Tmax >6 s volume of 272.70 mL and a Tmax >10 s volume of 136.50 mL, yielding an HIR of 0.50. This patient was classified into the CCC_mixed group, with a baseline NIHSS of 20, a 24-hour post-EVT NIHSS score of 20 (non-END group), and a 90-day mRS score of 5. CCC, cerebral collateral cascade; COVES, Cortical Vein Opacification Score; END, early neurological deterioration; EVT, endovascular thrombectomy; HIR, hypoperfusion intensity ratio; mRS, Modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; Tmax, time to maximum of the residue function; VO, venous outflow.

Multivariate logistic regression

In the multivariate logistic regression analysis, all VIF values were below 2, indicating no significant multicollinearity among the variables. Adjustments were made for HBP, CHD, baseline NIHSS, ASPECTS, infarct core volume, puncture-to-recanalization time, mTICI 0–2a, and hemorrhagic transformation, after which it was found that the independent predictors of END were COVES <3 [adjusted odds ratio (aOR) =2.74; 95% CI: 1.40–5.38; P=0.003], CCC− (aOR =13.46; 95% CI: 2.55–70.98; P=0.002), and CCC_mixed (aOR =8.01; 95% CI: 1.76–36.47; P=0.007). Neither Tan score <2 (aOR =1.17; 95% CI: 0.57–2.40; P=0.661) nor HIR >0.4 (aOR =1.68, 95% CI: 0.80–3.55, P=0.200) remained significant in the adjusted model (Table S1 and Table 3).

ROC curve analysis indicated that the CCC model had the highest predictive accuracy for END (AUC =0.824; 95% CI: 0.766–0.882), outperforming both the VO model (AUC =0.808, 95% CI: 0.746–0.871) and the clinical model (AUC =0.793; 95% CI: 0.727–0.860). The ROC comparisons are presented in Figure 5A.

Figure 5 ROC curves of the three models for predicting END. (A) The CCC model demonstrated the highest predictive performance, with an AUC of 0.824 (95% CI: 0.766–0.882), followed by the VO model (AUC =0.808, 95% CI: 0.746–0.871) and the clinical model (AUC =0.793, 95% CI: 0.727–0.860). (B) Among the patients without hemorrhagic transformation, the CCC model still demonstrated the best predictive performance (AUC =0.806; 95% CI: 0.722–0.889), followed by the VO model (AUC =0.786; 95% CI: 0.700–0.872) and the clinical model (AUC =0.767; 95% CI: 0.668–0.865). AUC, area under the receiver operating characteristic curve; CCC, cerebral collateral cascade; CI, confidence interval; END, early neurological deterioration; ROC, receiver operating characteristic; VO, venous outflow.

Sensitivity analysis

In the sensitivity analyses, it was found that END was not independently associated with HIR >0.3 (aOR =1.32; 95% CI: 0.59–2.97; P=0.499), HIR >0.5 (aOR =1.37; 95% CI: 0.64–2.94; P=0.422), or HIR as a continuous variable (aOR =2.17, 95% CI: 0.33–14.39, P=0.422) after adjustment. In contrast, independent predictors of END were COVES <4 (aOR =5.34; 95% CI: 1.95–14.65; P=0.001) and quantitative COVES (aOR =0.72; 95% CI: 0.58–0.89; P=0.002), with AUCs of 0.822 (95% CI: 0.762–0.882) and 0.805 (95% CI: 0.742–0.869), respectively (Table S2).

A total of 186 patients without hemorrhagic transformation were included in the sensitivity analysis, of whom 30 (16.1%) experienced END. In univariate analyses, it was found that END was significantly associated with unfavorable arterial collaterals (OR =3.00; 95% CI: 1.34–6.70; P=0.007), venous collaterals (OR =3.44; 95% CI: 1.48–8.01; P=0.004), and tissue-level collaterals (OR =2.88; 95% CI: 1.26–6.55; P=0.012), CCC− (OR =33.00; 95% CI: 3.93–277.38; P=0.001), and CCC_mixed (aOR =8.44; 95% CI: 1.09–65.19; P=0.031) (Table S3).

After adjustments were made for ASPECTS, infarct core volume, puncture-to-recanalization time, and mTICI, multivariable logistic regression identified the independent predictors of END to be COVES <3 (aOR =2.53; 95% CI: 1.02–56.26; P=0.045) and CCC− (aOR =14.34; 95% CI: 1.46–141.00; P=0.022) (Table S4). As shown in Figure 5B, the CCC model maintained superior diagnostic performance (AUC =0.806; 95% CI: 0.722–0.889) as compared with the VO model (AUC =0.786; 95% CI: 0.700–0.872) and the clinical model (AUC =0.767; 95% CI: 0.668–0.865).

Discussion

This study systematically investigated the relationship between comprehensive collateral circulation status and END in patients with LVO-AIS following EVT. Using a CCC classification system, we found that patients classified as CCC− exhibited larger baseline infarct cores, greater severity of neurological deficits, higher rates of END after EVT, and poorer 90-day functional outcomes. Multivariate logistic regression analysis identified CCC−, CCC_mixed, and poor COVES as independent predictors of END. Compared with models based solely on clinical variables or a single COVES parameter, the CCC-based integrated model achieved superior predictive performance. Furthermore, sensitivity analyses confirmed that CCC− remained a strong predictor of unexplained END, highlighting the pivotal role of comprehensive collateral assessment in guiding EVT prognostication.

END remains a major clinical challenge, affecting a substantial proportion of patients undergoing EVT and often predicting poor long-term outcomes (16,17). A prospective multicenter study by Liu et al. (18) indicated that END was significantly associated with unfavorable 90-day outcomes in patients with AIS, and our findings further underscore its clinical significance. The underlying mechanisms are multifactorial, involving reperfusion injury, hemorrhagic transformation, microcirculatory perfusion disturbances, and imbalances in energy metabolism (19-21). Despite the growing clinical recognition of END, reliable imaging biomarkers for its early prediction remain lacking. Conventional imaging assessments primarily emphasize arterial collateral circulation, while the critical contributions of veins and tissue-level hemodynamics to stroke evolution have been largely overlooked (22,23).

This study confirmed CCC to be an independent predictor of END. From a pathophysiological perspective, the role of collateral circulation in AIS reflects the integrated capacity of brain tissue to sustain energy metabolism and cellular viability under ischemic stress (24). Insufficient arterial collaterals result in reduced perfusion pressure, limiting the ischemic penumbra’s ability to receive adequate blood flow to sustain metabolic demands (25). As microcirculatory perfusion becomes further impaired, a cascade of metabolic disturbances—including calcium overload and mitochondrial dysfunction—ensues, ultimately accelerating neuronal death (26). In contrast, an intact and well-organized collateral cascade helps preserve cerebral blood flow stability, delays infarct progression, and consequently improves reperfusion outcomes (27,28). Therefore, the CCC can be regarded as a comprehensive indicator of the compensatory capacity of ischemic tissue, providing multidimensional information that enhances the accuracy of predicting early neurological changes after stroke. Previous work by Faizy et al. (10) supported CCC as a valuable predictor of final infarct volume and long-term functional outcomes in patients with AIS. Building on this framework, our study extends the clinical relevance of CCC by demonstrating its value in predicting END, which represents a critical phase in stroke evolution and has been consistently associated with unfavorable long-term outcomes. Therefore, the ability to identify patients at high risk of END based on baseline CCC may enable earlier risk stratification, inform individualized therapeutic decision-making, and potentially improve downstream clinical outcomes. By shifting the prognostic focus from long-term end points to early neurological dynamics, our findings expand the temporal and clinical scope of the CCC concept.

Among the various components of the CCC, unfavorable COVES has been identified as an independent predictor of END. This finding aligns with recent studies indicating that venous drainage reflects both the integrity of the microcirculation and the adequacy of reperfusion (29,30). Impaired VO may signify obstruction in the downstream capillary bed or microvascular stasis, leading to ongoing tissue injury despite successful macrovascular recanalization (31). Furthermore, elevated venous pressure and disruption of the blood-brain barrier can exacerbate cerebral edema and precipitate secondary hemorrhage, thereby contributing to neurological deterioration (32,33).

Further analysis in this study demonstrated that the CCC model outperformed those models based solely on COVES or clinical parameters, exhibiting superior discriminative ability and clinical applicability. These findings further support the notion that END is a multifactorial pathological process, whose occurrence depends not only on the success of large-vessel recanalization but also on the overall efficacy of macroscopic and microscopic collateral circulation. Notably, some patients may experience unexplained END after EVT (34-36). Our sensitivity analysis showed that, after patients with END secondary to hemorrhagic transformation were excluded, the predictive performance of the CCC model remained robust, highlighting its unique value in identifying neurological deterioration unrelated to procedural complications.

In clinical practice, the CCC can complement existing predictive factors such as infarct core volume and onset-to-recanalization time, thereby optimizing patient selection and postoperative management strategies (37). Patients classified as CCC− or CCC_mixed may benefit from more intensive hemodynamic monitoring, enhanced neuroprotective interventions, or individualized reperfusion targets. Importantly, CCC assessment can be performed using conventional CTA/CTP imaging without significantly increasing the examination burden, highlighting its strong clinical feasibility.

This study involved several limitations that should be addressed. First, as we employed a single-center retrospective design, selection bias could have arisen. Second, although a standardized imaging acquisition and interpretation protocol were employed, the assessment of collateral circulation nonetheless involved subjective judgment. Third, there is currently no standardized definition of END; therefore, this study adopted the definition most commonly used in literature, which is based on clinical neurological deterioration rather than quantitative imaging thresholds. Future studies integrating imaging markers are warranted to further elucidate the pathophysiological basis of END. Finally, a substantial proportion of cases were classified as CCC_mixed. Because the CCC framework integrates three distinct components, the CCC_mixed group inherently encompasses a diversity of collateral impairment states, making it difficult to identify the predominant collateral dysfunction driving tissue injury. This thus limits the inference of the relative contribution of underlying injury mechanisms based on the current imaging-based classification, particularly within this subgroup. Future studies should therefore validate CCC in large multicenter cohorts, further decompose CCC_mixed into biologically interpretable subtypes, and integrate advanced imaging, molecular biomarkers, and artificial intelligence methods to clarify the specific injury cascades associated with distinct patterns of collateral dysfunction. This would ultimately contribute to the development of more precise, mechanism-based therapeutic strategies.

Conclusions

This study demonstrated that CCC assessment, comprising information from arterial, venous, and tissue levels, provides a holistic evaluation of cerebral perfusion and effectively predicts END in patients undergoing EVT for AIS. Both poor CCC and impaired VO were identified as independent predictors of END, and the CCC-based model exhibited superior predictive performance compared with traditional clinical indicators and single collateral parameters. Incorporating collateral cascade assessment into clinical practice may enhance patient risk stratification, inform individualized postoperative management, and improve the accuracy of prognostic evaluation in patients with AIS.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2394/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2394/dss

Funding: This study was supported by Beijing Natural Science Foundation (No. 7264403, to R.C.) and Xuanwu Hospital Elite Cultivation Program (No. YC20250116, to R.C.).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-aw-2394/coif). R.C. reports receiving funding from Beijing Natural Science Foundation (No. 7264403) and Xuanwu Hospital Elite Cultivation Program (No. YC20250116). The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by Ethics Committee of Xuanwu Hospital and individual consent for this retrospective analysis was waived.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Zhang M, Xing P, Tang J, Shi L, Yang P, Zhang Y, Zhang L, Peng Y, Liu S, Zhang L, Fu J, Liu J. DIRECT-MT Investigators. Predictors and outcome of early neurological deterioration after endovascular thrombectomy: a secondary analysis of the DIRECT-MT trial. J Neurointerv Surg 2023;15:e9-e16. [Crossref] [PubMed]
2. Sun D, Tong X, Huo X, Jia B. Raynald, Wang A, Ma G, Ma N, Gao F, Mo D, Song L, Sun X, Liu L, Deng Y, Li X, Wang B, Luo G, Wang Y, Miao Z; ANGEL-ACT study group. Unexplained early neurological deterioration after endovascular treatment for acute large vessel occlusion: incidence, predictors, and clinical impact: Data from ANGEL-ACT registry. J Neurointerv Surg 2022;14:875-80. [Crossref] [PubMed]
3. Bosshart SL, Kappelhof M, Stebner A, Fujiwara S, Cimflova P, Schüngel MS, et al. Decision-Making for Endovascular Thrombectomy in Patients With Large Vessel Occlusions and Mild Neurological Deficit: A Consensus Statement. J Stroke 2025;27:338-49. [Crossref] [PubMed]
4. Farooqui A, Albayram MS, Reddy VBN, Nagaraja N. Neurological deterioration and computed tomography perfusion changes with increased time to peak in lacunar stroke. Brain Circ 2022;8:17-23. [Crossref] [PubMed]
5. Sarraj A, Hassan AE, Grotta J, Blackburn S, Day A, Abraham M, et al. Early Infarct Growth Rate Correlation With Endovascular Thrombectomy Clinical Outcomes: Analysis From the SELECT Study. Stroke 2021;52:57-69. [Crossref] [PubMed]
6. Regenhardt RW, González RG, He J, Lev MH, Singhal AB. Symmetric CTA Collaterals Identify Patients with Slow-progressing Stroke Likely to Benefit from Late Thrombectomy. Radiology 2022;302:400-7. [Crossref] [PubMed]
7. Xin Y, Zhou H, Xue MC, Wang Q, Zhou SL, Zhou XY, Ji TJ. Correlation of fluid-attenuated inversion recovery sequence vascular hyperintensity in magnetic resonance with collateral circulation and short-term clinical prognosis in acute ischemic stroke. Quant Imaging Med Surg 2024;14:4123-33. [Crossref] [PubMed]
8. Seners P, Yuen N, Olivot JM, Mlynash M, Heit JJ, Christensen S, et al. Factors Associated With Fast Early Infarct Growth in Patients With Acute Ischemic Stroke With a Large Vessel Occlusion. Neurology 2023;101:e2126-37. [Crossref] [PubMed]
9. Cao R, Lu Y, Li W, Yu F, Hu S, Chen K, Wang G, Sun C, Ma Q, Zhang M, Chen J, Lu J. Dynamic CTA-Based Whole-Brain Arterial-Venous Collateral Assessment for Predicting Futile Recanalization in Acute Ischemic Stroke. Aging Dis 2025; Epub ahead of print. [Crossref]
10. Faizy TD, Mlynash M, Kabiri R, Christensen S, Kuraitis GM, Mader MM, Flottmann F, Broocks G, Lansberg MG, Albers GW, Marks MP, Fiehler J, Wintermark M, Heit JJ. The Cerebral Collateral Cascade: Comprehensive Blood Flow in Ischemic Stroke. Neurology 2022;98:e2296-306. [Crossref] [PubMed]
11. Pisani L, Haussen DC, Mohammaden M, Perry da Camara C, Jillella DV, Rodrigues GM, Bouslama M, Al-Bayati A, Prater A, Liberato B, Frankel MR, Nogueira RG. Comparison of CT Perfusion Software Packages for Thrombectomy Eligibility. Ann Neurol 2023;94:848-55. [Crossref] [PubMed]
12. Schwartzmann Y, Joubran H, Jubeh T, Metanis I, Alpernas A, Jonas-Kimchi T, Sadeh U, Gomori JM, Cohen JE, Hallevi H, Molad J, Leker RR. Association of CT perfusion parameters with outcomes in patients with medium vessel occlusion undergoing endovascular thrombectomy. Ther Adv Neurol Disord 2025;18:17562864251333515. [Crossref] [PubMed]
13. Winkelmeier L, Kniep H, Thomalla G, Bendszus M, Subtil F, Bonekamp S, et al. Arterial Collaterals and Endovascular Treatment Effect in Acute Ischemic Stroke with Large Infarct: A Secondary Analysis of the TENSION Trial. Radiology 2025;314:e242401. [Crossref] [PubMed]
14. Winkelmeier L, Heit JJ, Adusumilli G, Geest V, Guenego A, Broocks G, Prüter J, Gloyer NO, Meyer L, Kniep H, Lansberg MG, Albers GW, Wintermark M, Fiehler J, Faizy TD. Poor venous outflow profiles increase the risk of reperfusion hemorrhage after endovascular treatment. J Cereb Blood Flow Metab 2023;43:72-83. [Crossref] [PubMed]
15. Hamam O, Gudenkauf J, Moustafa R, Cho A, Montes D, Sharara M, Moustafa A, Radmard M, Nabi M, Chen K, Sepehri S, Shin C, Mazumdar I, Kim M, Mohseni A, Malhotra A, Romero J, Yedavalli V. Hypoperfusion Intensity Ratio as an Indirect Imaging Surrogate in Patients With Anterior Circulation Large-Vessel Occlusion and Association of Baseline Characteristics With Poor Collateral Status. J Am Heart Assoc 2024;13:e030897. [Crossref] [PubMed]
16. Li Z, Zhang H, Han J, Chu Z, Zhao S, Yang Q, Huang X, Zhou Z. Time Course and Clinical Relevance of Neurological Deterioration After Endovascular Recanalization Therapy for Anterior Circulation Large Vessel Occlusion Stroke. Front Aging Neurosci 2021;13:651614. [Crossref] [PubMed]
17. Shi HX, Li C, Zhang YQ, Li X, Liu AF, Liu YE, Jiang WJ, Lv J. Predictors of Early Neurological Deterioration Occurring within 24 h in Acute Ischemic Stroke following Reperfusion Therapy: A Systematic Review and Meta-Analysis. J Integr Neurosci 2023;22:52. [Crossref] [PubMed]
18. Liu H, Liu K, Zhang K, Zong C, Yang H, Li Y, Li S, Wang X, Zhao J, Xia Z, Song B, Xu X, Gao Y. Early neurological deterioration in patients with acute ischemic stroke: a prospective multicenter cohort study. Ther Adv Neurol Disord 2023;16:17562864221147743. [Crossref] [PubMed]
19. Katsanos AH, Malhotra K, Ahmed N, Seitidis G, Mistry EA, Mavridis D, et al. Blood Pressure After Endovascular Thrombectomy and Outcomes in Patients With Acute Ischemic Stroke: An Individual Patient Data Meta-analysis. Neurology 2022;98:e291-301. [Crossref] [PubMed]
20. Zhu W, Zhou J, Ma B, Fan C. Predictors of early neurological deterioration in patients with intracerebral hemorrhage: a systematic review and meta-analysis. J Neurol 2024;271:2980-91. [Crossref] [PubMed]
21. Werring DJ, Ozkan H, Doubal F, Dawson J, Freemantle N, Hassan A, Le STN, Mallon D, Mendel R, Markus HS, Minhas JS, Webb AJS. Early neurological deterioration in acute lacunar ischemic stroke: Systematic review of incidence, mechanisms, and prospects for treatment. Int J Stroke 2025;20:7-20. [Crossref] [PubMed]
22. Consoli A, Pizzuto S, Sgreccia A, Di Maria F, Coskun O, Rodesch G, Lapergue B, Felblinger J, Chen B, Bracard S. Angiographic collateral venous phase: a novel landmark for leptomeningeal collaterals evaluation in acute ischemic stroke. J Neurointerv Surg 2023;15:e323-9. [Crossref] [PubMed]
23. Mei J, Salim HA, Lakhani DA, Balar A, Vagal V, Koneru M, et al. Integrating perfusion imaging derived venous outflow and tissue-level collateral parameters in a comprehensive clinical model enhances prognostication in large vessel occlusion stroke. Neurotherapeutics 2025;22:e00632. [Crossref] [PubMed]
24. Faber JE. Collateral blood vessels in stroke and ischemic disease: Formation, physiology, rarefaction, remodeling. J Cereb Blood Flow Metab 2025;45:1007-30. [Crossref] [PubMed]
25. Mangiardi M, Bonura A, Iaccarino G, Alessiani M, Bravi MC, Crupi D, Pezzella FR, Fabiano S, Pampana E, Stilo F, Alfano G, Anticoli S. The Pathophysiology of Collateral Circulation in Acute Ischemic Stroke. Diagnostics (Basel) 2023;13:2425. [Crossref] [PubMed]
26. Uniken Venema SM, Dankbaar JW, van der Lugt A, Dippel DWJ, van der Worp HB. Cerebral Collateral Circulation in the Era of Reperfusion Therapies for Acute Ischemic Stroke. Stroke 2022;53:3222-34. [Crossref] [PubMed]
27. Huang L, Zhang H, Li W, Gong C, Jiang S, Li Z, Yuan J, Xu T, Chen Y, Zhang L, Wang Y. Cerebral collateral cascade associated with infarct growth rate in ischemic stroke patients undergoing endovascular treatment. Eur J Radiol 2025;193:112418. [Crossref] [PubMed]
28. Huang L, Jiang S, Gong C, Wu G, Guo J, Liu J, Yuan J, Wang Y, Xu T, Liu C, Chen S, Chen Y. Favorable Cerebral Collateral Cascades Improve Futile Recanalization by Reducing Ischemic Core Volume in Acute Ischemic Stroke Patients Undergoing Endovascular Treatment. Transl Stroke Res 2025;16:1689-97. [Crossref] [PubMed]
29. Li X, Lin L, Zhang J, Fan Y, Xing S, Jiang L, Yang Z, Zhao J, Li J. Microvascular dysfunction associated with unfavorable venous outflow in acute ischemic stroke patients. J Cereb Blood Flow Metab 2023;43:106-15. [Crossref] [PubMed]
30. Heitkamp C, Winkelmeier L, Heit JJ, Albers GW, Lansberg MG, Kniep H, Broocks G, Stracke CP, Schell M, Guenego A, Paech D, Wintermark M, Fiehler J, Faizy TD. Early neurological deterioration in patients with acute ischemic stroke is linked to unfavorable cerebral venous outflow. Eur Stroke J 2024;9:162-71. [Crossref] [PubMed]
31. Heitkamp C, Winkelmeier L, Heit JJ, Albers GW, Lansberg MG, Wintermark M, Broocks G, van Horn N, Kniep HC, Sporns PB, Zeleňák K, Fiehler J, Faizy TD. Unfavorable cerebral venous outflow is associated with futile recanalization in acute ischemic stroke patients. Eur J Neurol 2023;30:2684-92. [Crossref] [PubMed]
32. van Horn N, Heit JJ, Kabiri R, Broocks G, Christensen S, Mlynash M, Meyer L, Schoenfeld MH, Lansberg MG, Albers GW, Fiehler J, Wintermark M, Faizy TD. Venous outflow profiles are associated with early edema progression in ischemic stroke. Int J Stroke 2022;17:1078-84. [Crossref] [PubMed]
33. Faizy TD, Kabiri R, Christensen S, Mlynash M, Kuraitis G, Meyer L, Marks MP, Broocks G, Flottmann F, Lansberg MG, Albers GW, Fiehler J, Wintermark M, Heit JJ. Venous Outflow Profiles Are Linked to Cerebral Edema Formation at Noncontrast Head CT after Treatment in Acute Ischemic Stroke Regardless of Collateral Vessel Status at CT Angiography. Radiology 2021;299:682-90. [Crossref] [PubMed]
34. Pensato U, Coutts SB, van Adel B, Chapot R, Puetz V, Demchuk A, Goyal M, Hill MD, Ospel JM. Incidence, Associations, and Mechanisms of Unexplained Early Neurologic Deterioration After Thrombectomy in Stroke Patients. Neurology 2025;105:e214261. [Crossref] [PubMed]
35. Bourcier R, Goyal M, Muir KW, Desal H, Dippel DWJ, Majoie CBLM, et al. Risk factors of unexplained early neurological deterioration after treatment for ischemic stroke due to large vessel occlusion: a post hoc analysis of the HERMES study. J Neurointerv Surg 2023;15:221-6. [Crossref] [PubMed]
36. Girot JB, Richard S, Gariel F, Sibon I, Labreuche J, Kyheng M, Gory B, Dargazanli C, Maier B, Consoli A, Daumas-Duport B, Lapergue B, Bourcier R. ETIS Investigators. Predictors of Unexplained Early Neurological Deterioration After Endovascular Treatment for Acute Ischemic Stroke. Stroke 2020;51:2943-50. [Crossref] [PubMed]
37. Wu W, Cheng Y, Ma J, Zuo Q, Zhang L, Fu L, Li Y, Wang X. Development and validation of a prognostic nomogram incorporating automated collateral score and computed tomography perfusion parameters for patients with acute ischemic stroke undergoing endovascular thrombectomy. Quant Imaging Med Surg 2025;15:6738-50. [Crossref] [PubMed]