Safety and Efficacy of Nerinetide at Year 1 in Participants Enrolled in ESCAPE‐NEXT: A Multicenter, Double‐Blind, Randomized Controlled Trial

Abstract

Background

Nerinetide is a neuroprotective agent recently evaluated in the ESCAPE‐NEXT (Efficacy and Safety of Nerinetide in Participants With Acute Ischemic Stroke Undergoing Endovascular Thrombectomy Excluding Thrombolysis) trial (NCT04462536), which was terminated after failing to meet its Day 90 primary end point; however, by that time, Year 1 follow‐up outcomes were already available for 513 participants.

Methods

The primary end point at Year 1 was functional independence, defined as modified Rankin Scale score 0 to 2, analyzed by logistic regression adjusted for treatment and baseline covariates. In a post hoc analysis, the interaction between early (<3 hours) versus late (3–12 hours) enrollment window and treatment effect was also tested and the results reported separately by enrollment window.

Results

A total of 513 of 850 participants had documented Year 1 outcomes before study termination, of whom 442 reached their scheduled Year 1 visit. In the nerinetide group, 110 (48.0%) of 229 participants achieved functional independence at Year 1 compared with 102 (47.9%) of 213 in the placebo group (adjusted odds ratio [aOR], 1.12 [95% CI, 0.74–1.71]; P=0.593). There was treatment effect modification by enrollment window (early versus late; P _interaction=0.044). Among early window participants (n=163), 51 (52.6%) in the nerinetide group and 30 (45.5%) in placebo achieved functional independence (aOR, 2.80 [95% CI, 1.18–6.66], P=0.019) at Year 1. Additionally, the nerinetide group exhibited improved survival (aOR, 2.61 [95% CI, 1.17–5.83], P=0.019). Conversely, no significant clinical benefit of nerinetide at Year 1 was observed among late window participants. Analyses on all 513 participants with documented Year 1 outcomes provided similar results.

Conclusions

Long‐term benefits of early administration of neuroprotection may emerge up to 1 year after stroke.

Nonstandard Abbreviations and Acronyms

mRS

modified Rankin Scale

Clinical Perspective.

What Is New?

• Consistent with a recent patient‐level meta‐analysis demonstrating the benefit of nerinetide treatment at Day 90 among early window participants across three separate trials in acute ischemic stroke, participants in the ESCAPE‐NEXT (Efficacy and Safety of Nerinetide in Participants With Acute Ischemic Stroke Undergoing Endovascular Thrombectomy Excluding Thrombolysis) cohort treated within 3 hours demonstrated significant gains in functional independence and survival at Year 1, despite only nominal and more modest benefit at Day 90.

What Are the Clinical Implications?

• Early window patients likely overrepresent “fast progressors,” a biologically distinct group in whom neuroprotection can meaningfully slow infarct progression before reperfusion.

• Routine reliance on Day 90 end points may underestimate the therapeutic effect of neuroprotective agents, strengthening the case for including Year 1 functional and survival outcomes in future trial designs.

Nerinetide slows ischemic brain injury by targeting PSD‐95 (postsynaptic density 95 protein) dependent signaling pathways, thereby uncoupling GluN2B‐containing NMDA receptors from neuronal nitric oxide synthase and attenuating downstream neurotoxic cascades. ¹ , ² In so doing, it acts in neurons within the ischemic penumbra, the target of neuroprotection. The safety and efficacy of nerinetide have recently been examined in 2 large trials conducted in participants with acute ischemic stroke who were selected for endovascular thrombectomy. These comprised the Phase 3 ESCAPE‐NA1 (Safety and Efficacy of NA‐1 [Nerinetide] in Subjects Undergoing Endovascular Thrombectomy for Stroke) trial, ³ and the subsequent ESCAPE‐NEXT (Efficacy and Safety of Nerinetide in Participants With Acute Ischemic Stroke Undergoing Endovascular Thrombectomy Excluding Thrombolysis) trial, ⁴ which was intended to replicate the effects of nerinetide seen in ESCAPE‐NA1 in participants not previously treated with thrombolytics. The primary end point in both trials was functional independence (modified Rankin Scale score 0–2) as evaluated Day 90 after enrollment.

ESCAPE‐NA1 enrolled 1105 participants within 12 hours of symptom onset using imaging to identify patients with small infarct cores and salvageable penumbra to ensure study drug was administered before stroke progression was complete. Enrollment was stratified by whether participants received prior intravenous alteplase treatment. The trial overall failed to meet its primary end point but revealed a significant interaction between prior alteplase treatment and treatment effect. In participants who were not treated with alteplase, nerinetide treatment resulted in clinically meaningful improvements in functional independence, reduced mortality, and reduced infarction volumes as compared with placebo. ⁴ , ⁵ , ⁶ In the stratum of participants who received alteplase, nerinetide treatment was ineffective. The likely reason for this was cleavage of nerinetide by plasmin following its activation by alteplase, ⁷ a process corroborated in the trial by reduced nerinetide levels in pharmacokinetic samples in patients administered alteplase.

To confirm the encouraging results of the no‐alteplase stratum of the ESCAPE‐NA1 trial, ESCAPE‐NEXT enrolled 850 participants under a similar design including a 12‐hour enrollment window, similar patient selection criteria and treatment with endovascular thrombectomy without prior thrombolysis. Unlike ESCAPE‐NA1, the ESCAPE‐NEXT study protocol also prespecified a Year 1 follow up in which the primary and key secondary end points were evaluated. After unmasking of the main trial results at Day 90, the trial was terminated for failing to meet its primary end point. By then, 513 of the 850 participants enrolled had documented Year 1 outcomes. Here we report the Year 1 follow‐up results for these participants as prespecified in the statistical analysis plan.

ESCAPE‐NEXT and ESCAPE‐NA1 (collectively, the “ESCAPE” trials) were conducted contemporaneously with the FRONTIER (Efficacy and Safety of Intravenous Nerinetide Initiated by Paramedics in the Field for Acute Cerebral Ischemia Within 3 h of Symptom Onset) trial, ⁸ which pursued a different strategy. Whereas the ESCAPE trials relied on in‐hospital imaging to screen for participants with salvageable ischemic penumbra up to 12 hours post stroke onset, FRONTIER relied on administering nerinetide or placebo to participants with suspected stroke as quickly as possible. FRONTIER enrolled participants with a suspected stroke in the ambulance within 3 hours of their last known well time. The intent was that, even in the absence of imaging, most of the participants ultimately diagnosed with acute ischemic stroke would have salvageable brain in this early window. Additionally, because nerinetide was administered before hospital arrival, none of the participants were treated with alteplase before nerinetide, eliminating the possibility of the cleavage of nerinetide by alteplase. Nerinetide was not effective in all participants with suspected stroke, especially because hemorrhagic strokes, stroke mimics, transient ischemic attacks, or strokes not treated with reperfusion therapies would not benefit from treatment with an anti‐ischemic agent. However, the trial demonstrated that in participants with a confirmed diagnosis of acute ischemic stroke and selected for reperfusion with thrombolysis, endovascular thrombectomy, or both, nerinetide showed a clinically significant benefit over placebo.

In all 3 trials, participants with ischemic stroke randomized in the first 3 hours after symptom onset and selected for reperfusion appeared to benefit from nerinetide. A patient‐level meta‐analysis of these trials showed a clinically significant treatment benefit for nerinetide across several outcome measures, including modified Rankin Scale (mRS) score, the incidence of worsening of stroke, and infarction volumes. ⁹ The findings of the meta‐analysis reconcile the demonstrated efficacy of nerinetide in animal models when administered within 3 hours of ischemia onset followed by reperfusion ¹ , ¹⁰ , ¹¹ , ¹² , ¹³ with the clinical findings across all 3 nerinetide trials. This supported the notion that participants enrolled early after stroke onset comprise a distinct population, likely one that includes more rapidly progressing strokes that may benefit from neuroprotection. Therefore, in addition to examining the effects of nerinetide in ESCAPE‐NEXT at Year 1, we also hypothesized that the benefits of nerinetide could be most evident in participants enrolled within the first 3 hours (“early window”) as compared with 3 to 12 hours (“late window”) after symptom onset.

METHODS

Data Availability

The study authors plan to make the ESCAPE‐NEXT data available to the public at a future date still to be determined. At that time, the deidentified data will be provided along with a data dictionary, the study protocol, the statistical analysis plan, and the informed consent template. Rules for data access will be determined in the future.

Ethics Approval

The trial was approved by the ethics board, primarily in Calgary (REB21–1258), as well as at each site and by the responsible regulatory authorities and monitored by an independent data monitoring committee. Written informed consent was obtained from the patients or a surrogate decision maker or where approved using an emergency consent process; in most sites this involved consent from the investigator after consultation with an independent physician who was not otherwise participating in the trial followed by a regained capacity consent process for ongoing participation in the trial occurring after treatment was given.

Study Design

The Year 1 follow‐up presented here leverages the previously completed ESCAPE‐NEXT trial whose full design details are reported elsewhere. ⁴ In brief, ESCAPE‐NEXT enrolled patients with anterior‐circulation large‐vessel occlusion selected for endovascular thrombectomy without thrombolysis up to 12 hours since last seen well. Across 77 sites in 9 countries, participants were randomized 1:1 to a single intravenous dose of nerinetide or matching saline placebo.

The trial included 2 preplanned database locks, the first for completion of Day 90 outcomes for the main trial and the second lock at the completion of the Year 1 outcomes. The original trial ethics approvals and participants' consents permitted collection of both Day 90 and Year 1 outcomes. The trial was stopped early, before the collection of Year 1 outcomes for all enrolled participants, after analyses after the first database lock failed to meet the primary end point. The study was registered (NCT04462536) and the study protocol and statistical analysis plan are available in Data S1.

Randomization and Masking

ESCAPE‐NEXT participants were randomized using a real‐time minimized randomization procedure ¹⁴ , ¹⁵ , ¹⁶ stratified by time from symptom onset (≤4.5 versus >4.5 hours) and balanced on key prognostic variables, including age, sex, baseline National Institutes of Health Stroke Scale score, ¹⁷ occlusion location, imaging‐to‐randomization time, baseline Alberta Stroke Program Early Computed Tomography Score category ¹⁸ and clinical site. Treatment allocation was fully masked to all trial staff using visually identical study drug vials. ⁴

Participants

Adult participants who had previously been living independently, as indicated by a Barthel Index score >90, were eligible if presenting with a disabling ischemic stroke, defined by a baseline National Institutes of Health Stroke Scale score >5. Participants were required to have a small to moderate ischemic core with imaging evidence of potentially salvageable brain tissue, along with moderate to good collateral circulation as assessed by available computer tomography‐based vascular or perfusion imaging. Patients were excluded if the treating physician determined, based on current treatment guidelines and available medical evidence, that therapy with a tissue plasminogen activator was indicated.

For complete inclusion and exclusion criteria, please see the protocol provided in Data S1.

Procedures

After qualifying imaging, participants rapidly received endovascular thrombectomy according to standard of care. Participants also received study drug, either placebo or 2.6 mg/kg up to 270 mg over 10±1 minutes through a dedicated intravenous line. The study drug was administered as soon as possible after randomization. Sites were asked to ensure that the study drug was administered before the end of the endovascular thrombectomy procedure defined as arterial access closure. Centralized monitoring of imaging and treatment speed, with feedback to sites, ensured adherence to patient eligibility criteria and to workflow metrics. Sites were expected to adhere to national guidelines for stroke unit care, stroke rehabilitation, and secondary stroke prevention.

Participants were contacted by telemedicine or telephone for their Year 1 visit, defined as day 365 (±30 days) post enrollment. The Year 1 visit was conducted by trained assessors blinded to treatment allocation and to the participant's 90‐day outcome.

Clinical Assessments and Outcomes

The primary outcome of the Year 1 follow‐up was the proportion of participants with independent functioning on the mRS (range, 0 [no symptoms] to 6 [death]), as defined by a score of 0 to 2 at Year 1 post randomization, scored by staff trained and certified in the use of the mRS. The secondary end points included mortality rate (defined as the number of deaths observed divided by the number of subjects observed over the Year 1 follow‐up period), the proportion of subjects with independent function on activities of daily living (defined as a score of ≥95 on the modified Barthel Index [range, 0 to 100, with higher scores indicating a greater ability to complete activities of daily living]) and health‐related quality of life (as measured by the EuroQol visual analog scale). Site staff conducting the Year 1 assessments remained blinded to treatment group allocation and to the Day 90 outcomes until after the Year 1 database lock.

Handling of Missing Data

As prespecified in the statistical analysis plan, no data were imputed for the Year 1 primary and secondary efficacy analyses. However, the statistical analysis plan did prespecify a sensitivity analysis where multiple imputation was used to generate any missing Year 1 outcomes across the trial population of 850 participants. In further post hoc sensitivity analyses, we also imputed missing outcomes to the last observation carried forward and, separately, using the most conservative approach of imputing to the worst possible outcome of death.

Statistical Analysis

The primary analysis population comprised all participants with a known mRS score at Day 90 and Year 1 enrolled at least 11 months before trial termination, meaning that they had reached the protocol‐defined Year 1 outcome window of 365 days (±30 days) post enrollment. To be clear, all participants who enrolled at least 11 months before trial termination and who died at any point during the trial are included in the primary analysis population. As illustrated by the survival curves, the majority of deaths occurred before the Day 90 end point. Including these deaths but excluding participants who would have survived to their Year 1 assessment had the trial not been terminated would artificially inflate the observed mortality rate. Therefore, we focused in the main paper on the 442 participants who all had the opportunity to reach their scheduled Year 1 assessment to most accurately represent the clinical population. This approach avoids overestimation of mortality due to early termination of the trial as deaths before the scheduled Year 1 follow‐up would still be documented as such at Year 1 and nonfatal outcomes could be assessed at each participant's scheduled Year 1 visit. This approach also allows direct comparison of outcomes at Day 90 versus Year 1 in the same patient population.

The early window population comprised those participants enrolled in the trial within 3 hours of symptom onset, whereas the late window comprised participants enrolled between 3 and 12 hours.

As a sensitivity analysis, all analyses performed on the primary analysis population were also conducted as prespecified in the statistical analysis plan, without imputation, on all participants for whom the Year 1 outcome was known (n=513 for the primary end point), a population that includes the mortalities that occurred within 11 months of trial termination. These results are presented in Data S1.

Statistical analyses for the Year 1 follow‐up were performed according to the protocol and prespecified statistical analysis plan employing the same statistical models as the Day 90 outcomes. Day 90 and Year 1 outcomes for mRS score, mortality and Barthel Index were evaluated using a logistic regression model to estimate an adjusted odds ratio (aOR) of a treatment effect by including treatment (2‐level factor variable nerinetide/placebo) and 8 baseline covariates thought at the time the trial was designed to be prognostic of stroke outcomes: age (as a continuous integer in years), sex (as a 2‐level factor variable female/male), baseline National Institutes of Health Stroke Scale score (as a continuous integer from 0 to 42), arterial occlusion location (as a 2‐level factor variable internal carotid artery/middle cerebral artery), time from qualifying imaging to randomization (as a continuous integer in minutes), baseline Alberta Stroke Program Early Computed Tomography Score (3‐level factor including 0–4 [poor], 5–7 [moderate], and 8–10 [good]), pooled clinical site (as a 3‐level factor variable for Europe/North America/rest of world) and time from stroke onset to randomization (≤4.5 versus >4.5 hours). EuroQoL‐Visual Analog Scale data were fit to an ANCOVA model using the same covariates as the primary model. The analyses were conducted at a 2‐sided type 1 error rate of 0.05.

In a post hoc analysis, we tested for treatment effect modification by enrollment window (early [0–3 hours] versus late [3–12 hours]) using the same logistic regression model as specified for the primary end point, adding an interaction term of treatment effect by window. We report the results separately by time from symptom onset to randomization for early window (0–3 hours) and late window (3–12 hours) participants. These subgroup analyses were also conducted using the same prespecified statistical models.

As per the statistical analysis plan, missing outcomes were not imputed except for the sensitivity analyses. Additionally, EuroQoL‐Visual Analog Scale was evaluated only for participants who were alive. Multiple imputation by chained equations were performed for the primary end point and mortality where missing outcomes were imputed using a binary logistic regression model to generate 1000 randomized data sets. Each data set was then analyzed, the results pooled and then reported. When imputing to worst outcome, all missing Year 1 data were imputed to death (mRS score=6). When imputing to last observation carried forward, missing Year 1 data were imputed in the following order of priority: (1) Day 90 outcome, if available, (2) Day 30 outcome, if available, (3) death (mRS score=6).

The cumulative probability of survival was displayed using Kaplan–Meier plots. Differences between treatment groups were analyzed with an unadjusted Cox proportional hazards model containing only treatment group. For the adjusted survival analysis, the Cox proportional hazards model was adjusted using the same prespecified covariates as the primary end point.

Alluvial plots are used to visually represent changes in functional outcomes over time at an individual participant level. mRS is represented as a vertical column with the height of each section proportional to the number of participants in the given category. The width of the horizontal ribbons (alluvia) is proportional to the number of participants changing between two specific categories from Day 90 to Year 1.

Statistical analyses and plots were generated with the R statistical package (v4.4.1). This study is reported in accordance with the Consolidated Standards of Reporting Trials 2025 guidelines.

RESULTS

Between December 6, 2020 and January 31, 2023, the trial enrolled to completion with 850 participants, 454 to nerinetide and 396 to placebo. On July 20, 2023, the Year 1 follow‐up of participants was discontinued after the main study failed to meet its primary end point at Day 90. At the time of discontinuation, 513 participants (nerinetide=270, placebo=243) had documented Year 1 outcomes. Of these, 442 participants (nerinetide=229, placebo=213) comprising the primary analysis population had reached their scheduled Year 1 follow‐up (see Methods, Figure 1, and Figure S1) and the results are presented here in the main text. A sensitivity analysis, which repeats the findings presented in the main text using all 513 participants with known Year 1 outcomes, is presented in Figures S2 through S5 and Tables S1 through S5.

Figure 1. Trial profile at Year 1 follow‐up.

Primary analysis population (n=442) is presented in the main paper. Analysis of all known Year 1 outcomes (n=513 for the primary end point) as prespecified in the statistical analysis plan is presented in Data S1.

Of the 442 enrolled participants in the primary analysis population, 229 (52.1%) were allocated to the nerinetide and 213 (47.9%) to the placebo group. In the early window, 97 (59.5%) were allocated to nerinetide and 66 (40.5%) to placebo versus late window where 132 (47.3%) were allocated to nerinetide and 147 (52.7%) to placebo. All participants underwent attempted endovascular thrombectomy; no participants deviated from the protocol by receiving alteplase or tenecteplase intravenously or intra‐arterially before or concurrent with the study drug.

Baseline characteristics (Table 1) between nerinetide and placebo were similar for the most important prognostic variables that were included in the primary analysis, including age and baselines stroke severity by the National Institutes of Health Stroke Scale. The median time and interquartile range (IQR) from stroke onset to the start of study drug infusion was shorter in the nerinetide group than the placebo group (222 [IQR, 135–404] versus 285 [IQR, 153–525] minutes) in the overall analysis population but similar in the early window (127 [IQR, 109–151] versus 128 [IQR, 97–151] minutes) and late window (IQR, 376 [273–532] versus 404 [IQR, 275–593] minutes) populations.

Table 1.

Baseline Characteristics of Primary Analysis Population

	All participants		Early window participants		Late window participants
	Nerinetide (n=229)	Placebo (n=213)	Nerinetide (n=97)	Placebo (n=66)	Nerinetide (n=132)	Placebo (n=147)
Age, y	76 (67 to 83)	75 (65 to 84)	77 (67 to 83)	74.5 (59 to 85)	76 (67 to 84)	75 (67 to 83)
Sex
Female	113 (49.3%)	98 (46.0%)	44 (45.4%)	27 (40.9%)	69 (52.3%)	71 (48.3%)
Male	116 (50.7%)	115 (54.0%)	53 (54.6%)	39 (59.1%)	63 (47.7%)	76 (51.7%)
Region
North America	125 (54.6%)	122 (57.3%)	48 (49.5%)	32 (48.5%)	77 (58.3%)	90 (61.2%)
Europe	83 (36.2%)	68 (31.9%)	42 (43.3%)	24 (36.4%)	41 (31.1%)	44 (29.9%)
Rest of world	21 (9.2%)	23 (10.8%)	7 (7.2%)	10 (15.2%)	14 (10.6%)	13 (8.8%)
Medical history
Artery disease	70 (30.6%)	46 (21.6%)	32 (33%)	20 (30.3%)	38 (28.8%)	26 (17.7%)
Congestive heart failure	41 (17.9%)	28 (13.1%)	18 (18.6%)	8 (12.1%)	23 (17.4%)	20 (13.6%)
Ischemic stroke/transient ischemic attack	48 (21%)	48 (22.5%)	18 (18.6%)	19 (28.8%)	30 (22.7%)	29 (19.7%)
Intracerebral hemorrhage	5 (2.2%)	7 (3.3%)	3 (3.1%)	4 (6.1%)	2 (1.5%)	3 (2%)
Major surgery	22 (9.6%)	11 (5.2%)	13 (13.4%)	4 (6.1%)	9 (6.8%)	7 (4.8%)
Vascular disease	13 (5.7%)	13 (6.1%)	7 (7.2%)	4 (6.1%)	6 (4.5%)	9 (6.1%)
Renal failure	35 (15.3%)	19 (8.9%)	18 (18.6%)	9 (13.6%)	17 (12.9%)	10 (6.8%)
Hyperlipidemia	120 (52.4%)	107 (50.2%)	52 (53.6%)	37 (56.1%)	68 (51.5%)	70 (47.6%)
Atrial fibrillation	111 (48.5%)	106 (49.8%)	59 (60.8%)	39 (59.1%)	52 (39.4%)	67 (45.6%)
Anticoagulants	85 (37.1%)	77 (36.2%)	45 (46.4%)	32 (48.5%)	40 (30.3%)	45 (30.6%)
Hypertension	179 (78.2%)	163 (76.5%)	79 (81.4%)	53 (80.3%)	100 (75.8%)	110 (74.8%)
Diabetes	63 (27.5%)	52 (24.4%)	27 (27.8%)	14 (21.2%)	36 (27.3%)	38 (25.9%)
Smoking (current/past)	92 (40.2%)	103 (48.4%)	38 (39.2%)	30 (45.5%)	54 (40.9%)	73 (49.7%)
Pulmonary disease	22 (9.6%)	15 (7%)	8 (8.2%)	3 (4.5%)	14 (10.6%)	12 (8.2%)
COVID‐19 (current/past)	22 (9.6%)	28 (13.1%)	8 (8.2%)	15 (22.7%)	14 (10.6%)	13 (8.8%)
Clinical factors
National Institutes of Health Stroke Scale score baseline	16 (12 to 19)	16 (12 to 19)	17 (12 to 20)	18 (13 to 21)	15 (11 to 19)	15 (12 to 19)
Alberta Stroke Program Early Computed Tomography Score baseline
5–7	71 (31.0%)	65 (30.5%)	15 (15.5%)	10 (15.2%)	56 (42.4%)	55 (37.4%)
8–10	158 (69.0%)	148 (69.5%)	82 (84.5%)	56 (84.8%)	76 (57.6%)	92 (62.6%)
Occlusion location
Internal carotid artery	51 (22.3%)	41 (19.2%)	27 (27.8%)	17 (25.8%)	24 (18.2%)	24 (16.3%)
Middle cerebral artery	178 (77.7%)	172 (80.8%)	70 (72.2%)	49 (74.2%)	108 (81.8%)	123 (83.7%)
Treatment and workflow
Time strata
≤4.5 h	128 (55.9%)	100 (46.9%)	180 (100.0%)	132 (100.0%)	101 (76.5%)	113 (76.9%)
>4.5 h	101 (44.1%)	113 (53.1%)	0 (0.0%)	0 (0.0%)	132 (100.0%)	147 (100.0%)
<3 h	97 (42.4%)	66 (31.0%)	97 (100.0%)	66 (100.0%)	0 (0.0%)	0 (0.0%)
≥3 h	132 (57.6%)	147 (69.0%)	0 (0.0%)	0 (0.0%)	132 (100.0%)	147 (100.0%)
Onset to randomization, min	222 (135 to 404)	285 (153 to 525)	127 (109 to 151)	128 (97 to 151)	376 (273 to 532)	404 (275 to 593)
Imaging to randomization, min	38 (23 to 63)	39 (22 to 65)	38 (22 to 60)	43 (21 to 66)	37.5 (23 to 64)	38 (23 to 64)
Imaging to study drug, min	52 (35 to 78)	55 (35 to 79)	54 (36 to 78)	59 (33 to 83)	51 (35 to 78)	54 (36 to 77)
Drug to groin puncture, min	0 (−10 to 9)	0 (−14 to 10)	−1 (−10 to 8)	0 (−14 to 17)	0 (−9 to 9)	−1 (−14 to 9)
Study drug to reperfusion, min	43 (21 to 68)	41 (20 to 76)	41 (21 to 70)	36 (19 to 70)	49 (20 to 65)	41 (23 to 76)
Expanded thrombolysis in cerebral ischemia
2b‐2c‐3	198 (86.5%)	187 (87.8%)	84 (86.6%)	56 (84.8%)	114 (86.4%)	131 (89.1%)
2c‐3	137 (59.8%)	125 (58.7%)	58 (59.8%)	38 (57.6%)	79 (59.8%)	87 (59.2%)

Data are n (%) or median (interquartile range).

The nerinetide group had higher rates of coronary artery disease, congestive heart failure, major surgery, renal failure, and diabetes but lower rates of smoking and history of COVID‐19 infection (Table 1). The early window population exhibited a similar pattern of medical history, with the nerinetide group also having a reduced history of ischemic stroke or transient ischemic attack as well as higher median age 77 [IQR, 67–83] versus placebo 74.5 [IQR, 59–85].

The overall workflow (onset to randomization, imaging to study drug, and study drug to reperfusion) and quality of reperfusion (on the expanded Thrombolysis in Cerebral Ischemia scale) were similar in both groups (Table 1).

The primary outcome of the proportion of participants achieving an mRS score 0 to 2 at Year 1 was 48.0% (110 of 229) in the nerinetide group and 47.9% (102 of 213) (Table 2) in the placebo group (aOR 1.129[5% CI, 0.74–1.71], P = 0.593; Figures 2A and 3A).

Table 2.

Overall Outcomes at Year 1 and at Day 90 in the Primary Analysis Population

	Adjusted outcomes (prespecified primary analysis), odds ratio (95% CI)	Unadjusted effect size		Odds ratio (95% CI)
		Nerinetide (n=229)	Placebo (n=213)
1 y
Primary outcome
mRS score 0–2	1.12 (0.74–1.71)	110 (48.0%)	102 (47.9%)	1.01 (0.69–1.46)
Secondary outcomes
Mortality reduction	1.42 (0.89–2.25)	58 (25.3%)	63 (29.6%)	1.24 (0.81–1.88)
Barthel Index 95–100	0.96 (0.63–1.47)	103 (45.2%)	102 (48.1%)	0.89 (0.61–1.29)
EQ‐VAS score	1.12 (0.66–1.91)	75 (60–86)	75 (60–85)	1.05 (0.61–1.79)
D 90
Primary outcome
mRS score 0–2	1.05 (0.69–1.59)	108 (47.2%)	101 (47.4%)	0.99 (0.68–1.44)
Secondary outcomes
Mortality reduction	1.26 (0.74–2.13)	44 (19.2%)	43 (20.2%)	1.06 (0.67–1.70)
Barthel Index 95–100	0.93 (0.61–1.42)	104 (45.6%)	103 (48.6%)	0.90 (0.62–1.31)
EQ‐VAS score	1.1 (0.67–1.8)	75 (60–85)	75 (60–90)	1.08 (0.65–1.80)

Data are in n (%) or median (interquartile range).

EQ‐VAS indicates EuroQoL Visual Analog Scale; and mRS, modified Rankin Scale.

Figure 2. Distribution of modified Rankin Scale scores at Day 90 and Year 1.

Stacked bars are labeled with percentages within each mRS category for (A) the entire primary analysis population (n=442); (B) the subgroup of participants randomized within 3 hours of stroke onset (n=163); and (C) the subgroup of participants randomized within 3 to 12 hours from stroke onset (n=279). mRS indicates modified Rankin Scale.

Figure 3. Overall outcomes and sensitivity analyses for the primary analysis population (n=442).

Forest plot of efficacy outcomes in the (A) overall, (B) early window, and (C) late window participants presented as adjusted odds ratios that were adjusted per statistical analysis plan for the prerandomization variables of participant age, baseline National Institutes of Health Stroke Scale score, baseline Alberta Stroke Program Early CT Score, time from CT to randomization, sex, geographic region (North America, Europe, rest of world), occlusion location (internal carotid artery/middle cerebral artery), and enrollment stratum. Sensitivity cases trend in the same direction as the main efficacy analysis for the (D) early window participants despite imputing data to worst outcome or last observation carried forward (LOCF). No significant trend was observed in the (E) late window sensitivity cases. aOR indicates adjusted odds ratio; CT, computed tomography; EQ, EuroQoL; and LOCF, last observation carried forward.

There was evidence of treatment effect modification by enrollment window for the primary end point at Year 1. This was observed in all participants with a known Year 1 outcome (n=513, P _interaction=0.032) and in the primary analysis population (n=442, P _interaction=0.044). There was no heterogeneity of treatment effect by sex in the overall population (P _interaction=0.30). We did not analyze efficacy by race and ethnicity because of the small sample sizes.

The primary outcome of the proportion of participants achieving functional independence, defined as an mRS score 0 to 2, at Year 1 in the early window population was 52.6% (51 of 97) in the nerinetide group and 45.5% (30 of 66) (Table 3) in the placebo group (aOR, 2.8 [95% CI, 1.18–6.66], P=0.019), an increased benefit of nerinetide from that seen at Day 90 (aOR, 1.13 [95% CI, 0.54–2.35], P=0.750; Figures 2B and 3B). Conversely, no significant clinical benefit of nerinetide in achieving functional independence at Year 1 was observed in the late window participants (44.7%% (59 of 132) nerinetide, 49.0% (72 of 147) placebo, Table 3), aOR, 0.81 [95% CI, 0.48–1.36], P=0.417; Figures 2C and 3C.

Table 3.

Outcomes by Early or Late Window Enrollment at Year 1 and Day 90 in the Primary Analysis Population

	Adjusted outcomes (prespecified primary analysis)		Unadjusted effect size
	Early window (n=163) OR (95% CI)	Late window (n=279) OR (95% CI)	Early window (n=163)			Late window (n=279)
			Nerinetide (n=97)	Placebo (n=66)	OR (95% CI)	Nerinetide (n=132)	Placebo (n=147)	OR (95% CI)
1 y
Primary outcome
mRS score 0–2	2.80 (1.18–6.6)	0.81 (0.48–1.36)	51 (52.6%)	30 (45.5%)	1.33 (0.71–2.49)	59 (44.7%)	72 (49.0%)	0.84 (0.53–1.35)
Secondary outcomes
Mortality reduction	2.61 (1.17–5.83)	1.09 (0.61–1.97)	25 (25.8%)	24 (36.4%)	1.65 (0.84–3.24)	33 (25.0%)	39 (26.5%)	1.08 (0.63–1.86)
Barthel Index 95–100	1.80 (0.80–4.05)	0.76 (0.45–1.28)	45 (46.4%)	30 (45.5%)	1.04 (0.55–1.95)	58 (43.9%)	72 (49.0%)	0.82 (0.51–1.31)
EQ‐VAS score	1.78 (0.78–4.03)	1.09 (0.53–2.25)	75 (61–87)	75 (58–86)	1.29 (0.58–2.83)	75 (60–85)	75 (60–85)	0.91 (0.44–1.88)
D 90
Primary outcome
mRS score 0–2	1.13 (0.54–2.35)	1.01 (0.60–1.69)	48 (49.5%)	35 (53.0%)	0.87 (0.46–1.62)	60 (45.5%)	66 (44.9%)	1.02 (0.64–1.64)
Secondary outcomes
Mortality reduction	1.45 (0.62–3.39)	1.21 (0.60–2.43)	23 (23.7%)	16 (24.2%)	1.03 (0.50–2.14)	21 (15.9%)	27 (18.4%)	1.19 (0.64–2.22)
Barthel Index 95–100	1.30 (0.62–2.84)	0.82 (0.49–1.39)	45 (46.4%)	32 (48.5%)	0.92 (0.49–1.72)	59 (44.7%)	71 (48.3%)	0.89 (0.55–1.43)
EQ‐VAS score	1.98 (0.94–4.20)	0.80 (0.41–1.55)	75 (65–85)	75 (58–86)	1.9 (0.90–4.04)	71 (59–85)	75 (60–85)	0.75 (0.38–1.47)

Data are in n (%) or median (interquartile range).

EQ‐VAS indicates EuroQoL‐Visual Analog Scale; mRS, modified Rankin Scale; and OR, odds ratio.

In the early window participants, treatment with nerinetide was also associated with an absolute mortality reduction of 10.6% at Year 1. Deaths occurred in 25.8% (25 of 97) in the nerinetide group and 36.4% (24 of 66) in the placebo group (aOR for mortality reduction, 2.61 [95% CI, 1.17–5.83], P=0.019), an increased mortality benefit for nerinetide from that seen at Day 90 (aOR, 1.45 [95% CI, 0.62–3.39], P=0.394). The remaining secondary end points were not significant but showed similar trends in treatment effect at Year 1 for activities of daily living (Barthel Index responder rate, aOR, 1.8 [95% CI, 0.8–4.05], P=0.158) and survivor‐rated health status (EuroQol visual analog scale, aOR, 1.78 [95% CI, 0.78–4.03], P=0.173). Conversely, in the late window participants, no significant clinical benefit was observed at Year 1 in those treated with nerinetide versus placebo in achieving functional independence (n=279, aOR, 0.81 [95% CI, 0.48–1.36], P=0.417) or reducing mortality (aOR, 1.09 [95% CI, 0.61–1.97], P=0.772, Figure 3C).

Sensitivity analyses on the primary outcome and mortality were conducted in the early versus late window participants to test the robustness of our findings. Termination of the study before completion of the Year 1 follow‐up resulted in missing outcomes for 337 of 850 (39.6%) participants (122 of 312 (39.1%) early window and 215 of 538 (40.0%) late window participants). For early window participants (n=312) with missing data imputed according to the prespecified multiple imputation method, the results at Year 1 remained significant for both functional independence (mRS score=0–2) (aOR, 2.52 [95% CI, 1.21–5.24], P=0.013) and mortality (aOR, 2.19 [95% CI, 1.1–4.37], P=0.027, Figure 3D). For the same early window participants using the post hoc approach of imputing to last observation carried forward, the primary outcome (mRS score=0–2) indicated a benefit from nerinetide (aOR, 1.95 [95% CI, 1.14–3.34], P=0.014) and trended toward a reduction of mortality (aOR, 1.5 [95% CI, 0.84–2.67], P=0.166). Similar trends to nerinetide benefit were observed in early window participants when imputing to worst outcome (mRS score=6 or death) for both functional independence (nerinetide imputed responder rate 28.9% versus 22.7% for placebo, aOR, 1.56 [95% CI, 0.87–2.8], P=0.134) and reduction of mortality (nerinetide imputed mortality rate 58.3% versus 66.7% for placebo, aOR, 1.47 [95% CI, 0.9–2.41], P=0.124). For the late window participants (n=538), no significant treatment effects were observed when imputing missing outcomes in any analysis (Figure 3E).

To further explore Year 1 mortality, we performed survival analysis, both unadjusted using Kaplan–Meier analysis and adjusted via a Cox proportional hazards regression as prespecified in the statistical analysis plan. In the overall primary analysis population, there was no significant difference between treatment groups in the unadjusted analysis (P=0.329) (Figure 4A) or adjusted analysis (P=0.153) (Figure 4B). In the early window participants, the unadjusted survival curves are similar between treatment groups through Day 90 then diverge toward Year 1 (P=0.202, Figure 4C). In the adjusted Cox proportional hazards model of survival, the treatment group curves begin to diverge before Day 90 and remain separated through Year 1 (P=0.031) (Figure 4D). The early window placebo participants fared significantly worse, with an adjusted Year 1 survival probability of 61.3%. By contrast, the adjusted Year 1 probability of survival for the early window nerinetide participants was 75.7%. No significant differences by treatment group were seen in either the unadjusted or adjusted models for late window participants (Figure 4E and 4F).

Figure 4. Unadjusted and adjusted survival curves.

Unadjusted Kaplan–Meier curves for the (A) overall, (C) early window, and (E) late window participants with P values obtained from an unadjusted Cox proportional hazards model. Cox proportional hazards curve for the (B) overall, (D) early window, and (F) late window participants are adjusted per statistical analysis plan for the same prerandomization variables as the primary end point: participant age, baseline National Institutes of Health Stroke Scale score, baseline Alberta Stroke Program Early CT Score, time from CT to randomization, sex, geographic region (North America, Europe, rest of world), occlusion location (internal carotid artery/middle cerebral artery), and enrollment stratum. CT indicates computed tomography.

The results presented thus far indicate greater benefit with nerinetide treatment than with placebo at Year 1 than Day 90. To investigate which participant outcomes were driving this effect, we next examined post hoc how the categorical mRS data changed between Day 90 and Year 1 for each individual participant (Figure 5). Even when the proportions of functional independence were stable in the overall population between Day 90 and Year 1, there were clinically important changes among individuals. The mRS outcomes for 39.8% of participants changed between Day 90 and Year 1. This was most notable in the rise in mortality observed between Day 90 and Year 1 in all groups except the early window nerinetide participants. In terms of functional independence, 4 of 97 (4.1%) participants treated with nerinetide improved from nonresponders (mRS score≥3) at Day 90 to responders (mRS score=0–2) at Year 1 whereas only 1 of 97 (1.0%) declined from responder to nonresponder over the same period. The opposite pattern was seen in the overall functional decline of early window placebo participants where 1 of 66 (1.5%) participants became a responder during the Year 1 follow‐up period but 6 of 66 (9.1%) declined from responder at Day 90 to nonresponder at Year 1.

Figure 5. Alluvial plot of evolution in mRS outcomes from Day 90 to Year 1.

Stacked bars represent the proportion of participants with the given mRS outcome and the bands between Day 90 and Year 1 outcomes are proportional to the number of participants retaining or changing mRS outcome. aOR indicates adjusted odds ratio; and mRS, modified Rankin Scale.

No safety data were collected during the follow‐up period between the Day 90 visit and the Year 1 visit. In the Year 1 primary analysis population, serious adverse events as collected during the main trial up to Day 90 occurred in 38.4% (88 of 229) in the nerinetide and 40.4% (86 of 213) in the placebo group (Table S4). In the early window population, the rate of serious adverse events was lower in the nerinetide group at 39.2% (38 of 97) compared with placebo at 48.5% (32 of 66). The 2 most common serious adverse events showed a similar pattern: stroke in evolution (nerinetide 6.2% (6 of 97) versus placebo 10.6% (7 of 66)) and ischemic stroke (nerinetide 3.1% (3 of 97) versus placebo 6.1% (4 of 66)).

DISCUSSION

Nerinetide was not effective at the Year 1 follow‐up in improving functional outcome in all participants randomized within 12 hours of stroke onset. However, in participants enrolled in the first 3 hours from stroke onset, nerinetide treatment was associated with higher rates of functional independence and reduced mortality at Year 1 compared with placebo.

These results that nerinetide may be of benefit to patients with ischemic stroke selected for reperfusion if given within 3 hours of stoke onset are supported by the patient‐level meta‐analysis of the efficacy of nerinetide at Day 90 across the FRONTIER, ESCAPE‐NA1, and ESCAPE‐NEXT trials. ⁹ That analysis demonstrated that in each trial individually, the maximal clinical benefits of nerinetide were observed within 3 hours of symptom onset. The current analysis extends these findings by showing that the clinical benefits of neuroprotection may continue to accrue up to Year 1.

The lack of benefit of nerinetide in ESCAPE‐NEXT participants enrolled 3 to 12 hours from stroke onset raises the possibility that these are clinically distinct patients from those enrolled within 3 hours of stroke onset. ¹⁹ , ²⁰ Participants were enrolled into the study when imaging demonstrated or inferred a small ischemic core implying a salvageable penumbra. This biased late window enrollment toward participants whose strokes progressed slowly enough to qualify up to 12 hours after symptom onset. Although these patients could still benefit from endovascular thrombectomy to resolve the ischemia, ²¹ , ²² , ²³ they were not necessarily candidates for benefit from nerinetide. Because nerinetide is intended to slow stroke progression, administering it to those whose strokes are already progressing slowly may not be as helpful as treating patients whose strokes are progressing rapidly. The challenge is compounded when neuroprotection is not administered until shortly before thrombectomy, as occurred in ESCAPE‐NEXT, in which the study drug was given close to the time of groin puncture (Table 1), allowing little opportunity to prevent stroke progression or to demonstrate an incremental benefit on top of high‐quality reperfusion. ⁴ , ⁹ This was potentially a limitation of the study design, as the investigational drug could be administered only at the same hospital in which rapid access to endovascular thrombectomy was available.

An important characteristic of the clinical benefit of nerinetide at Year 1 in the early window participants is that it was not discernible in the same participants at Day 90 (Figures 3B and 4C, Table 3). This observation stands in contrast to the published comparisons of outcomes at Day 90 and Year 1 in other trials, including the National Institute of Neurological Disorders and Stroke tissue plasminogen activator trials, ²⁴ ALIAS (Albumin in Acute Ischemic Stroke) ²⁵ and IMS‐III (Interventional Management of Stroke III). ²⁶ In those studies, functional independence post stroke was substantially similar between Day 90 and Year 1; however, more than a third of participants changed mRS score over that time period. ²⁷ Thus, the benefits of nerinetide might continue to accrue beyond Day 90.

The mechanism by which nerinetide might impart the observed late benefit in early window participants is unknown and ESCAPE‐NEXT was the only nerinetide trial to investigate such effects beyond Day 90. One hypothesis consistent with the findings presented here is that nerinetide stabilized the functional independence and survival of participants treated early in the progression of their stroke whereas similar participants treated with placebo continue to decline up to Year 1. This possibility is consistent with nerinetide's action of slowing stroke progression ⁵ , ⁶ and is supported by the survival analysis where early window participants (which would include fast progressors) receiving nerinetide had an adjusted survival rate indistinguishable from late window participants (including predominantly slow progressors) at 75.7% whereas the early window participants treated with placebo declined significantly over time to an adjusted survival rate of 61.3%, an absolute effect size on survival of 14.4% (Figure 4C and 4D).

Our findings are also consistent with the possibility that nerinetide sets patients on a different recovery trajectory or accelerates the recovery that most patients already experience following reperfusion therapy. Analysis of the individual patient trajectories (Figure 5) supports this view in that early window participants treated with nerinetide were more likely gain functional independence between Day 90 and Year 1 than lose functional independence (Figure 5). Conversely, early window participants treated with placebo were more likely to lose functional independence between Day 90 and Year 1 than gain functional independence. These findings again would be consistent with nerinetide's action of slowing stroke progression thereby sparing infarct growth until reperfusion was achieved.

Finally, it is also possible that the benefits of neuroprotection take longer than 90 days to manifest. However, this may be less likely given that a similar population— patients with ischemic stroke selected for reperfusion in the FRONTIER trial, ⁸ appeared to benefit by Day 90. Nonetheless, it is possible that these benefits could have continued but were not observed in the absence of further late follow‐up.

Our analyses were conducted in strict adherence to the prespecified statistical analysis plan using the planned statistical models. However, additional analyses, specifically those of the early versus late window participants, are limited in that they were conducted on a post hoc stratification informed by the Day 90 results of ESCAPE‐NA1, ESCAPE‐NEXT, and FRONTIER. As such, randomization was not minimized for early and late windows and there were more nerinetide participants than placebo participants in the early window. Although the median age for early window nerinetide participants is 2.5 years older than placebo (77 versus 74.5), the demographics otherwise appear balanced and Year 1 outcomes for early window participants, both unadjusted and adjusted, point in the same direction of nerinetide benefit. The greatest limitation of our Year 1 follow‐up is that the study was discontinued before all participants were contacted, excluding 337 participants. However, both the prespecified and additional sensitivity analyses suggest that the findings in our incomplete cohort are likely to be robust. The data used for analysis were from a well‐conducted, randomized placebo‐controlled trial designed to minimize bias.

The hypothesis that neuroprotection is most effective when delivered within 3 hours of stroke onset is currently being tested in the ACT‐42 (NoNO‐42 Trial in Acute Ischemic Stroke Patients Selected for Thrombolysis With or Without Endovascular Thrombectomy) trial (NCT06403267) with a next generation version of nerinetide. This trial is planned to include Year 1 follow‐up.

CONCLUSIONS

In summary, whereas nerinetide did not benefit all patients with stroke selected for endovascular thrombectomy within 12 hours on stroke onset, ESCAPE‐NEXT participants presenting within 3 hours of symptom onset, who likely include the fast progressors most at risk of poor outcomes, appear to benefit the most. The treatment benefit was more apparent at Year 1 than Day 90. Late window participants may represent slow progressors who are least likely to benefit from neuroprotectant treatments, which are intended to slow stroke progression. Our study promotes the hypothesis that long‐term benefits of neuroprotection may emerge up to a year after stroke suggesting that Year 1 outcomes could be useful in stroke neuroprotection trials.

Sources of Funding

ESCAPE‐NEXT was funded by NoNO Inc. and the public granting agency Canadian Institutes of Health Research funded ESCAPE‐NEXT.

Disclosures

Corey Adams reports stock, stock options, and executive role in NoNO Inc. Kathy Heard reports stock, stock options, and executive role in NoNO Inc. Yatika Kohli reports stock options and executive role in NoNO Inc. Mario Muto reports consulting fees from Stryker Neurovascular. Mirko Pham reports grants from German Research Foundation and honoraria from Merck. Richard H. Swartz reports grants from the National Institutes of Health, Canadian Institutes of Health Research, and Ontario Brain Institute, Data and Safety Monitoring Committee membership for Hoffman La Roche, and stock or stock options in FollowMD. Mayank Goyal reports grants Medtronic and Cerenovus; consulting fees from Microvention, Stryker Neurovascular, Medronic, Mentice, Phillips, and Penumbra; and stock ownership in Circle Neurovascular. Michael D. Hill reports grants from NoNO and Canadian Institutes of Health Research, which directly supported the ESCAPE‐NEXT study, past grants from NoNO, Medtronic, Boehringer‐Ingelheim, Canadian Institutes of Health Research, and Heart & Stroke Foundation; consulting fees from Sun Pharma and Brainsgate, patents licensed to Circle Neurovascular, Data and Safety Monitoring Committee chairmanship for the DUMAS and Oncover Hiltonel trials, Data and Safety Monitoring Committee membership for the ARTESIA, BRAIN‐AF, and LAAOS‐4 trials, stock ownership in Circle Neurovascular, and is the current president of the Canadian Neurological Sciences Federation (nonprofit sector). Michael Tymianski reports stock, stock options and chief executive officer role in NoNO Inc. The remaining authors have no disclosures to report.

Supporting information

Data S1: ESCAPE‐NEXT Investigators

Supplemental Methods

Tables S1–S5

Figures S1–S5