Management of radiological cerebral vasospasm after coil embolization for ruptured intracranial aneurysms: A noninferiority bicentric cohort study

Abstract

OBJECTIVE:

Whether radiological cerebral vasospasm (CVS) should be proactively treated following subarachnoid hemorrhage to reduce the development of delayed cerebral ischemia (DCI) and brain infarction remains unclear. In this bicentric retrospective study, we sought to determine if a first-line intravenous therapy (IVT) to early detect and treat CVS could be noninferior on neurological outcomes to a first-line endovascular therapy (ET) requiring intra-arterial procedures for confirmed CVS as the reference approach.

METHODS:

Two cohorts of unselected patients exposed to the risk of CVS were studied following coil embolization for ruptured intracranial aneurysms. In the IVT group, CVS was actively detected and treated using a continuous intravenous infusion of milrinone for at least 7 days. In the ET group, selective intra-arterial infusions of nimodipine and milrinone were performed to treat confirmed CVS, and repeated if necessary. The primary endpoint was the proportion of patients with a good functional neurological outcome at 12 months using the modified Rankin Scale 0–2, with a noninferiority margin of 15%.

RESULTS:

The good outcome of patients exposed to the risk of CVS was noninferior between the IVT and ET strategies: 108/119 patients (91%; 95% confidence interval [CI] 84% to 95%) versus 77/97 patients (79%; 95% CI: 70% to 87%), respectively. The IVT strategy led to the diagnosis of CVS in more patients in a shorter delay, and a reduced proportion of patients with DCI and brain infarction. There was no difference in the drug-related side effects between the two groups.

CONCLUSION:

A proactive intravenous treatment with milrinone to treat CVS had noninferior functional outcome compared to endovascular treatment using the selective intra-arterial infusion of vasodilator drugs.

Introduction

While the incidence of nontraumatic aneurysmal subarachnoid hemorrhage (SAH) is decreasing, poor-grade SAH continues to exhibit a significant association with elevated morbidity and mortality rates.[1] Common neurological complications postrepair of ruptured aneurysms encompass delayed cerebral ischemia (DCI) and are linked to microvascular dysfunction and radiological cerebral vasospasm (CVS), which is a major predictor of morbidity and mortality. Effective management of DCI, entailing the maintenance of euvolemia and induced arterial hypertension, is imperative following SAH to enhance neurological outcomes.[2,3,4] Additional procedures, such as endovascular infusion of intra-arterial vasodilators (nimodipine and milrinone) and mechanical angioplasty, are proposed as supplementary interventions but none of them are clearly recommended.

The intricate relationship between CVS and DCI cannot be straightforwardly elucidated by arterial narrowing alone,[5,6] thereby casting doubt on the notion of CVS as the primary contributor to DCI post-SAH. Approximately 70% of SAH patients present some degree of CVS within 2 weeks of bleeding, whereas only 30% develop DCI.[7] Clinical trials have failed to demonstrate clinical improvement by the targeted reversal of CVS using vasodilator drugs such as clazosentan or prostacyclin.[8,9] Percutaneous balloon angioplasty as a first-line treatment showed no improvement in neurological outcomes compared to a noninterventional strategy.[10] In contrast, cilostazol, a selective phosphodiesterase 3 inhibitor with vasodilator properties, was effective in preventing CVS and DCI development and improved neurological outcomes compared to placebo.[11] In observational studies, milrinone, which is another selective phosphodiesterase 3 inhibitor, was also associated with CVS reversal and a low rate of DCI.[12,13] Intra-arterial milrinone infusion was found effective to reverse CVS with DCI refractory to standard intravenous milrinone therapy.[14] Based on six observational studies, a review proposed a standardized protocol for intravenous milrinone infusion in DCI treatment.[15]

However, whether early and proactive intravenous milrinone use for CVS reversal can prevent DCI development remains unknown. This study aimed to compare the effectiveness of two center-driven strategies for managing SAH patients exposed to the risk of CVS with two different approaches: a proactive strategy to detect CVS and initiate an early first-line intravenous therapy (IVT) with milrinone versus a conservative strategy with a first-line endovascular therapy (ET) requiring selective intra-arterial infusions of nimodipine and milrinone or mechanical angioplasty in confirmed CVS. We hypothesized that patients managed with IVT would have noninferior functional outcomes compared to those managed with ET.

Methods

The study was a retrospective comparison of the medical records of two cohorts of unselected, consecutive adult patients who had coil embolization for SAH within 72 h of admission between January 2018 and December 2020. All patients were inherently exposed to the risk of CVS. They were managed in the neurological intensive care units (NICU) of two high-volume teaching hospitals with a multidisciplinary team, i.e., Center A and Center B. The French data protection authority (National Commission on Informatics and Liberty) granted electronic database access on August 14, 2018 (Re#2205066 v0). No ethical approval or patient consent was, therefore, sought in this retrospective study. Patients or their relatives received written information about the research and could refuse to participate according to French law.[16] Exclusion criteria were nonaneurysmal SAH, surgical clipping, preexisting of acute cerebrovascular disease (vascular, traumatic, tumoral, or infectious), or death before aneurysm exclusion.

Both centers had similar SAH care protocols in line with international guidelines, with special attention to the maintenance of euvolemia, normothermia, and electrolyte/metabolic balance.[2,3,17] Patients with altered consciousness were sedated and mechanically ventilated, before treating any hydrocephalus and/or intraventricular hemorrhage by external ventricular drainage (EVD). Intracranial pressure (ICP) was monitored to maintain cerebral perfusion pressure above 70 mmHg through vasoactive support with norepinephrine and plasma volume expansion with crystalloids if needed. Brain tissue oxygen pressure (PbtO2) monitoring was left to the discretion of the incharge physician. Oral nimodipine (60 mg × 6/day) was administered for 21 days postaneurysm rupture. In conscious patients with no ICP monitoring, Mean arterial pressure (MAP) was maintained at baseline values. Hyponatremia with high natriuresis was treated with sodium supplementation and, if needed, fludrocortisone or urea therapy. Transcranial Doppler (TCD) assessments of the two middle cerebral arteries were performed at least once a day for at least the initial 10 days in the two centers.

Management of cerebral vasospasm

Following coil embolization, patients were managed according to whether they developed CVS or not, using clinical criteria and TCD velocities. Protocols differed slightly, notably in TCD alert thresholds and diagnostic imaging methods: with computed tomography angiography (CTA) in Center A, and CTA or digital subtraction arteriography (DSA) in Center B. CVS was defined as a focal arterial narrowing of ≥ 50% relative to the inlet diameter. In both centers, patients with CVS received treatment including elevation of arterial blood pressure and maintenance of euvolemia. However, the two centers differed from their approach to treat CVS. Center A has an intravenous proactive approach while Center B has an in situ intra-arterial approach.

In Center A, the management of CVS was previously described.[13] Any new clinical events or TCD changes, such as mean blood flow velocities (FVm) of more than 120 cm/s in the middle cerebral artery or 100 cm/s in the anterior cerebral artery, and/or FVm alterations of more than 50 cm/s within 24 h, were investigated using CTA. Asymptomatic patients had systematic CTA on days 6–10. If CVS was confirmed, a continuous intravenous infusion of milrinone (Corotrope^® Sanofi-Aventis, Gentilly, France) was initiated without delay at a rate of 1 μg/kg/min, by central venous catheter under invasive blood pressure control. The infusion rate was then reduced by 50% in patients with poor hemodynamic tolerance. IVT started with a milrinone intravenous bolus (8 mg over 30 min) if the patient presented a focal deficit. Milrinone infusion was maintained from 7 to 14 days depending on clinical, TCD, and CTA response, before being tapered off over 48 h following a satisfactory clinical and radiological improvement. Mechanical angioplasty and/or intra-arterial milrinone administration were considered second-line procedures in the absence of clinical or radiological improvement.

In Center B, the procedure was initiated in the presence of major and minor indicators of CVS. Major indicators included new focal deficits, FVm of more than 200 cm/s in one middle cerebral artery, FVm changes of more than 50 cm/s within 24 h, three-fold increase in FVm compared to baseline at admission, a Lindegaard ratio (LR) of more than 6, or PbtO2 values below 10 mmHg without other cause of brain hypoxia. The occurrence of one major indicator led to performing a DSA. Minor indicators were clinical symptoms other than focal deficits, FVm of more than 100 cm/s in one middle cerebral artery, LR between 3 and 6, two-fold increase in FVm compared to baseline at admission, or PbtO2 values between 10 and 22 mmHg without other cause of brain hypoxia. CTA was used to investigate one or more minor indicators. In the presence of CVS, an individual risk–benefit assessment of ET was discussed between the neuroradiologist and intensivist. If the procedure was confirmed, the patient was installed in the neuroradiology suite for treatment under local anesthesia or general anesthesia if presenting altered consciousness. Selective infusion of nimodipine (2 mg over 5 min) and milrinone (5 mg over 30 min) was administered in the main artery supplying the vasospastic segment. Chemical angioplasty was performed daily until improvement was found according to clinical, TCD, and CTA response. Two to five treatment procedures per patient were usually needed. Mechanical angioplasty and/or intravenous milrinone infusion were considered second-line interventions in the absence of clinical or radiological improvement. A central venous catheter and an intra-arterial catheter were not systematically inserted in nonventilated patients.

Endpoints

The primary endpoint was the proportion of SAH patients in each cohort with a good functional neurological outcome 12 months after coil embolization, as measured using the modified Rankin Scale (mRS) during a medical consultation. The mRS ranges from 0 (no symptoms) to 6 (death). The neuroradiologist and intensivist in charge from each center used mRS assessment on a routine basis. All patients, irrespective of subsequent diagnosis of CVS or DCI, were included in the primary analysis due to the inherent risk of neurological deterioration. A good outcome was defined by mRS scores of 0 (no symptoms), 1 (no significant disability), or 2 (slight disability), where patients are either asymptomatic or have minimal disability that does not significantly impact daily activities.

Secondary endpoints included (i) the management of patients with CVS, including the delay to initiate treatment from CVS diagnosis, the proportion of patients with second-line procedures for persistent CVS, and their neurological outcome at 1 year; (ii) the proportion of patients with DCI, defined as a decrease in the Glasgow Coma Scale (GCS) score of 2 points or more compared to admission or focal neurological impairment for more than 1 h, that could not be due to other causes such as hydrocephalus, electrolyte abnormalities, medication effect, rebleeding, and new-onset infection;[18] (iii) the proportion of cerebral infarction identified through CT-scan or magnetic resonance imaging within 6 weeks after the initial bleeding, excluding coiling-related infarctions;[18] (iv) the proportion of side effects in relation to the CVS treatment, including hemodynamic instability, cardiac arrhythmia, hypokalemia, and endovascular complications.

Statistical analysis

The sample size calculation assumed that a good outcome with mRS 0–2 at 1 year after coil embolization would be found in a proportion of 77% of patients, as reported in previous studies.[19,20] We analyze our primary endpoint using binary logistic regression, with a one-sided two-sample comparison of proportions test using a normal approximation for noninferiority with a margin of 15%.[21] The 15% noninferiority margin was selected based on the methodology employed in the COMPASS trial,[21] which assessed noninferiority in the context of neuroradiological therapeutics. A statistical power of 80% and an alpha risk of 0.05 indicated that 98 patients per group would be required to conclude that IVT was noninferior to ET. This size calculation included a 10% loss to follow-up to 108 patients per group. The noninferiority test was to be followed by a superiority test if the noninferiority test was positive. A post hoc equivalence test based on the primary outcome was computed using the two one-sided test with the same margins as the noninferiority study, i.e., −15% to +15%. The equivalence was declared if each group had a confidence interval (CI) within the noninferiority interval. In addition, a multivariate logistic regression analysis was performed on the primary outcome after adjustment on imbalanced variables including time to aneurysm exclusion, CVS rates, mechanical ventilation (MV) in CVS patients, and ICU length of stay.

The proportion of patients with DCI, cerebral infarction, and CVS with a good outcome was compared between the two groups using the exact Fisher test. Quantitative data on outcomes were compared using a Student’s t-test or a Mann–Whitney U-test as appropriate. Post hoc comparison tests were applied for other comparisons. Data were presented as the median and interquartile range (25^th to 75^th percentiles) for continuous variables unless otherwise stated. Ninety-five percent CIs were used for estimated proportions. Statistical significance was declared when P < 0.05. Analyses were conducted using R version 4.0.3 and Stata version 15.0 (Stata Corp., College Station, TX, USA). Results were reported following the STROBE guidelines for retrospective studies.

Clinical trial registry

No clinical trials were involved due to the retrospective study.

Results

During the study period, 125 patients from Center A (IVT) and 107 patients from Center B (ET) who underwent coil embolization following aneurysmal SAH were included in the analysis [Figure 1]. Of these, 6 patients (Center A) and 10 patients (Center B) were lost to follow-up at 12 months. Patient characteristics were similar between the two centers [Table 1]. Both groups had a high proportion of patients with a modified Fisher scale of 3–4 (> 80%) at baseline. Patients in Center A had a shorter time from brain imaging to coil embolization, a higher incidence of CVS, and a longer ICU stay compared to Center B. Fewer patients in Center A required MV for over 48 h compared to Center B, though the duration of MV was comparable between groups (19 days [10–25] vs. 18 days; [14–30] P = 0.9).

Figure 1.

Flowchart of subarachnoid haemorrhage patients according to their management following coil embolization: Intra-venous therapy in Center A or endovascular therapy in Center B. SAH: Subarachnoid haemorrhage

Table 1.

Characteristics of the two cohorts of subarachnoid hemorrhage patients who were managed after coiling embolization with either intra-venous therapy in center A or endovascular therapy in center B

Variables	IVT center A (125 patients), n (%)	ET center B (107 patients), n (%)	P
Age, (years)	54 (46–62)	51 (46–62)	0.7
Male	46 (37)	39 (36)	1.0
Medical history
Arterial hypertension	39 (31)	47 (44)	0.06
Diabetes	0	3 (3)	0.1
Active smoking	61 (49)	40 (37)	0.1
Prior antithrombotic treatment	6 (5)	7 (7)	0.6
WFNS Grade III–V	21 (16)	25 (23)	0.3
Modified Fisher Scale 3 or 4	103 (82)	87 (81)	0.9
Time to aneurysm exclusion (h)
From clinical symptoms	50 (17–50)	53 (18–69)	0.2
From brain imaging	16 (11–20)	26 (11–44)	< 0.001
EVD before aneurysm exclusion	60 (48)	46 (43)	0.5
Surgical hematoma evacuation	4 (3)	2 (2)	0.7
CVS	79 (63)	48 (45)	0.005
Mechanical ventilation > 48 h
All patients	20 (16)	28 (26)	0.09
Patients with CVS	15 (19)	18 (37)	0.04
Length of stay in the ICU (days)	18 (11–23)	14 (9–21)	0.05

Data are expressed as the median (25^th–75^th percentiles) unless otherwise specified. WFNS: World federation of neurological surgeons, EVD: External ventricular drain, CVS: Radiological cerebral vasospasm, ICU: Intensive care unit, IVT: Intra-venous therapy, ET: Endovascular therapy

Primary endpoint

A good outcome (mRS 0–2) at 1 year was achieved in 91% (108/119) of patients in Center A (95% CI: 84% to 95%) and 79% (77/97) in Center B (95% CI: 70% to 87%), resulting in a significant absolute difference of 11% (95% CI: 3% to 19%; P = 0.02). The distribution of mRS scores is shown in Figure 2. While the IVT strategy was not inferior to ET, it did not meet equivalence criteria, as the upper margin of the difference (19%) exceeded the noninferiority margin of 15%.

Figure 2.

Distribution of the modified Rankin Scale scores at 1 year in the 2 cohorts of patients who were managed via either intravenous therapy or endovascular therapy after coil embolization. mRS: Modified Rankin Scale

Multivariate logistic regression adjusting for imbalanced variables (CVS, time from diagnosis to aneurysm exclusion, and MV) showed that the IVT strategy had an odds ratio of 2.72 (95% CI: 1.18–6.5) for achieving a good outcome compared to the ET strategy. This post hoc analysis confirmed the noninferiority of the two strategies, as the lower limit of the odds ratio (1.18) exceeded the estimated lower limit (0.49) based on the 15% noninferiority margin. These sensitive analyses were conducted using narrower noninferiority margins of 10% and 5%. The noninferiority was validated for both narrower margins in univariate and multivariate analysis.

Secondary endpoints

In total, 79 patients from Center A and 48 patients from Center B were diagnosed with CVS. The IVT strategy was associated with a significantly lower proportion of patients who were untreated for CVS, fewer cases of DCI or cerebral infarction, a shorter time to diagnosis of CVS, and a reduced number of second-line procedures compared to ET [Table 2]. Within the IVT group, one patient received both intraarterial and intravenous milrinone due to CVS and DCI during the coiling procedure, and two patients required mechanical angioplasty. In contrast, the ET group had 93 selective intraarterial infusions of nimodipine and milrinone in 32 patients, averaging 2.9 procedures per patient. In addition, eight ET patients underwent second-line procedures: intravenous milrinone (n = 3), mechanical angioplasty (n = 4), or both (n = 1). No significant differences in treatment-related side effects were observed between the two groups [Table 3], and there were no endovascular complications in either cohort.

Table 2.

Endpoints of patients with cerebral vasospasm according to their specific management of cerebral vasospasm intra-venous therapy or endovascular therapy

	IVT (79 patients), n (%)	ET (48 patients), n (%)	P
Patients not treated for CVS	5 (6)	16 (33)	< 0.001
Onset of treatment from CVS diagnosis (h)	16 (11–20)	19 (10–44)	0.037
Rescue procedures	3 (4)	8 (19)
DCI	6 (8)	17 (35)	< 0.001
Cerebral infarction	4 (5)	20 (42)	< 0.001

Data are expressed as the median (25^th–75^th percentiles) unless otherwise specified. CVS: Cerebral vasospasm, IVT: Intra-venous therapy, ET: Endovascular therapy, DCI: Delayed cerebral ischemia

Table 3.

Side effects of treatments in two cohorts of subarachnoid hemorrhage patients who were treated for cerebral vasospasm with either intravenous therapy or endovascular therapy

	IVT (79 patients), n (%)	ET (48 patients), n (%)	P
Hemodynamical instability leading to
50% reduction or discontinue milrinone	3 (4)	1 (2)	1.0
Increasing dose of norepinephrine	13 (17)	4 (8)	0.3
Arrhythmia	3 (4)	0	0.3
Hypokalemia	2 (3)	0	0.5

CVS: Cerebral vasospasm, IVT: Intravenous therapy, ET: Endovascular therapy, SAH: Subarachnoid hemorrhage

Of the 16 patients lost to follow-up, 10 (4 from the IVT group and 6 from the ET group) with CVS could not be assessed using the mRS at 1 year. Among those remaining, 89% (67/75) of IVT patients and 76% (32/42) of ET patients had a good outcome (mRS 0–2) at 1 year after SAH repair, yielding an absolute difference of 13% (95% CI: −3% to 30%), which was not statistically significant (P = 0.11).

Discussion

The optimal management of CVS to reduce DCI and prevent brain infarction, thereby improving neurological outcomes, remains a subject of active investigation. The treatment of vasospasm and DCI in SAH is a yet evolving challenge, as highlighted by numerous recent studies emphasizing its critical impact on patient prognosis. To our knowledge, this study is the first to directly compare these two therapeutic approaches for vasospasm management. Beyond the established use of oral nimodipine, induced hypertension, and euvolemia maintenance,[2,3,4] no pharmacological or mechanical intervention tested in clinical trials has demonstrated superior efficacy in improving outcomes for CVS patients. A recent network meta-analysis of randomized trials identified nimodipine and cilostazol as the most effective treatments for preventing DCI and improving functional neurological outcomes in SAH patients.[22] These findings underscore the importance of early therapeutic intervention to address microvascular mechanisms, such as endothelial damage, microvascular thrombosis, tissue inflammation, oxidative stress, and cortical spreading depolarizations, all of which contribute to the development of DCI.[5,6]

Milrinone, a selective phosphodiesterase 3 inhibitor, has emerged as a promising agent for CVS management due to its dual mechanisms of action. By inhibiting phosphodiesterase type 3, milrinone increases intracellular cAMP levels, inducing vascular smooth muscle relaxation and enhancing cardiac contractility.[12,23,24] These effects improve both cerebral blood flow and systemic perfusion.[13,15,25,26] Moreover, IVT with milrinone could synergize with ET; IV milrinone administered post-ET may sustain vasodilation, counter residual vasospasm, and optimize cerebral perfusion.[12,24,27] In addition to its vasodilatory effects, milrinone possesses anti-inflammatory properties, modulating inflammatory cytokines (e.g., interleukin-6, tumor necrosis factor-α) and reducing oxidative stress.[15,17,25,26] These actions enhance endothelial function and mitigate the neuroinflammatory cascade, critical contributors to CVS pathophysiology. Furthermore, milrinone’s modulation of mitochondrial metabolism and attenuation of reactive oxygen species production align with therapeutic strategies targeting neurovascular integrity and metabolic stabilization.[13,15] Together, these mechanisms position milrinone as a key component in the therapeutic arsenal against CVS.

In this bicentric cohort study, patients were managed in the NICU according to international guidelines, and care bundles were comparable. Since all patients received care management following coil embolization and were exposed to the same risk of neurological degradation due to the high proportion of patients with a modified Fisher scale of 3–4 in the two groups, we decided to analyze all consecutive, unselected patients during the study period. Because we found no difference in outcomes between intravenous and intra-arterial milrinone infusion in CVS patients in a single center,[13] we conducted a noninferiority study testing two different approaches in two centers. We favored a clinical and pragmatic approach to collect functional outcomes for the primary endpoint by comparing mRS at 1 year. The mRS score is the most widely used score for assessing patient functional outcomes in neurological studies, including SAH.[10,12,13,14,20,26,28] Molyneux et al.[20] have shown that mRS is a reliable predictor of functional recovery, with lower scores correlating with better outcomes in SAH management and reduced disability from CVS and DCI.

Although the two cohorts were comparable in terms of predictors of DCI such as age, female sex, active smoking, preexisting diabetes, WFNS and modified Fisher scale scorings, rates of EVD,[1] the IVT cohort had reduced rates of DCI and cerebral infarction. Interestingly, the IVT approach resulted in more and earlier diagnosis of CVS than the ET approach. IVT has been associated with a shorter initiation time, which is important in time-sensitive settings, and with fewer complications requiring MV. No more milrinone-related side effects were, however, found with the IVT strategy.

By contrast, the ET approach requires access to the neuroradiology suite, the availability of human resources, and the decision to treat taken by the neurointensivist and the neuroradiologist in charge. This situation might have hindered a timely intra-arterial procedure even in a site with high-volume activity. In addition, in our study, the ET strategy required more often MV due to the repeated procedures under general anesthesia in this group. Although not found in our study, ET carries more acute, procedure-related risks.[28] Vascular injury, including dissection or perforation, is a concern due to the invasive nature of catheterization, with the potential for hemorrhagic complications. Endovascular access complications, such as hematoma or pseudoaneurysm, are also possible. In addition, contrast-induced nephropathy and thromboembolic events are common complications due to the use of contrast agents and catheter manipulation. The cost of ET can be high due to the need for specialized procedures, personnel, and equipment. In accordance with the benefits of a proactive approach, reduced rates of DCI and better functional outcomes were found in a cohort of patients with CVS who were treated earlier with endovascular intervention compared to those with a delayed procedure.[22] Although IVT is not superior to ET, it can provide similar outcomes without increased risk, making it a good alternative in specific patient populations. IVT tends to be more cost-effective, requiring fewer resources and fewer second-line procedures. Overall, IVT is less invasive and involves systemic effects, which can be managed with careful monitoring, while ET carries more immediate procedural risks but is localized to the targeted vessels. This makes IVT an economically viable option, particularly for hospitals with budget constraints or those in high-volume settings.

Our results align with recent studies. Regarding endovascular treatment, a preventive, frequent, and early strategy seems to reduce the risk of DCI particularly.[29] Regarding the use of milrinone, a pooled systematic review[23] assessing the efficacy of milrinone in treating DCI highlighted a potential signal of efficacy, but it was based on small-scale studies with methodological limitations. Nonetheless, this review emphasized that milrinone is a safe and feasible treatment option. Moreover, the Milrispasme study[26] showed that milrinone, combined with hypertension, had a favorable effect on 6-month functional and radiological outcomes and reduced the need for endovascular interventions for vasospasm compared to hypertension treatment alone. In this study, milrinone was associated with a higher incidence of metabolic disorders, though arrhythmias, myocardial ischemia, and thrombocytopenia were rare. Finally, a 2021 review[15] suggested that early and aggressive use of milrinone may offer benefits in treating DCI, with few adverse effects. Our study challenges recent literature with more favorable outcomes where recent studies report lower values of the mRS (e.g., 73% for IVT[23]). Variations in mRS assessment (12 months in our study vs. 6 months in these studies), a center effect, or a favorable selection bias (e.g., exclusion of surgically clipped aneurysms) may interfere with the interpretation of our results.

Our study has several limitations that warrant discussion. First, its retrospective and bicentric design precludes establishing a causal relationship between treatment modalities and outcomes. Nevertheless, the inclusion of unselected, consecutively admitted patients during the study period likely minimized attrition bias. In addition, a post hoc multivariate logistic regression analysis was conducted to address potential confounding factors between the two centers. The primary endpoint analysis accounted for the key variables of time to aneurysm exclusion, rates of CVS, the need for MV among CVS patients, and ICU length of stay. These post hoc analyses showed no influence of these parameters on the primary endpoint. Regrettably, we were unable to perform Kaplan–Meier analysis on the primary endpoint and the mortality due to the lack of accurate data collection.

Regarding biases in patient management, the relatively homogeneous approach to SAH treatment across French neurocritical care units a result of longstanding teaching and academic activities within a common scholarly society—likely reduced variability. This study was designed with the understanding that both centers share a similar culture of SAH management, with the primary difference being their approach to CVS. However, the loss to follow-up patients may have impacted the study’s outcomes, limiting the generalizability of the findings. Future prospective randomized trials should aim to address these limitations by harmonizing overall management, defining standardized treatment protocols (e.g., timing of aneurysm embolization, management of metabolic complications, and intracranial hypertension), and clearly specifying inclusion criteria such as clinical and TCD detection of vasospasm confirmed by imaging. Crossover designs and intention-to-treat analyses could mitigate the lack of blinding between the IVT and ET treatment groups.

The retrospective nature of the study limited the sample size to 108 patients, which included an estimated 10% loss to follow-up. This number was calculated based on a noninferiority margin of 15%,[29] chosen at the time of the study design for its clinical relevance and alignment with prior literature. Although more recent studies[30,31] on SAH and acute ischemic stroke have utilized narrower margins (respectively 12.5% and 7%), these were published after our study’s initiation. Nonetheless, post hoc sensitivity analyses using narrower noninferiority margins of 10% and 5% confirmed our findings in both univariate and multivariate analyses. We performed this binary logistic regression analysis rather than the proportional odds assumption which is relevant to ordinal logistic regression models, typically used for ordered outcomes (e.g., Likert scales).

Differences in patient selection for aneurysm repair were observed between the two centers. In Center A, the most severe SAH cases were reassessed after EVD placement before proceeding with coil embolization; only patients with a GCS motor score of 4–6 underwent the procedure. In contrast, Center B performed coiling in nearly all patients regardless of initial presentation or neurological reassessment. Despite these differences, excluding patients who died after aneurysm repair (n = 17) did not alter the noninferiority results, underscoring the robustness of the analysis.

We tried to improve internal validity by minimizing confounding factors and improving cohorts’ comparability with the inclusions and exclusions criteria. However, these restrictive criteria limited the generalizability of the findings. Exclusions related to delayed treatment, preexisting cerebral pathologies, and surgical clipping may have biased the analysis toward favorable outcomes. The exclusion of patients with prior brain injuries or nonaneurysmal SAH further narrows the relevance of the findings to broader clinical populations. Moreover, the exclusion of patients who were unable to benefit from aneurysm securing within the recommended timeframe (< 72 h)[4] may have introduced selection bias, favoring patients with optimal treatment timelines and potentially overestimating the efficacy of early treatment. Nevertheless, we believe that these inclusion and exclusion criteria were fully in line with the standards of a SAH study.

In summary, while this study provides valuable insights into CVS management, its limitations highlight the need for future prospective randomized trials with standardized protocols, broader sample sizes, and strategies to minimize inter-institutional variability.

Conclusion

An early and proactive approach to CVS management, including initial IV milrinone treatment, demonstrated noninferior functional outcomes compared to reference selective ET. This strategy was associated with reduced rates of DCI and cerebral infarction, offering a practical and cost-effective first-line treatment option. Future prospective randomized trials should standardize protocols, harmonize SAH management practices, and refine inclusion criteria to further validate these findings.

Author contributions

Study design/planning: OV, MM, NE and JFP; Study conduct: OV, AP, MM and NE; Data analysis: OV, AP, MM, CL, NE and JFP; Writing the paper: OV, NE and JFP; Revising the paper: all authors.

Ethical policy and institutional review board statement

The study was performed in accordance with the Declaration of Helsinki. No ethical approval or patient consent was sought due to the retrospective study.

Data availability statement

The datasets used and/or analyzed in the current study are available from the corresponding author on reasonable request.

Conflicts of interest

Dr. Payen declares having received personal fees from Sedana Medical and IDD-CDM Lavoisier, outside the submitted manuscript. The other authors declare having no conflicts of interest.

Funding Statement

Nil.