^{1}

^{2}

^{*}

^{1}

^{3}

^{*}

^{2}

^{1}

^{2}

^{4}

^{5}

^{1}

^{2}

^{6}

^{7}

^{2}

^{8}

^{9}

^{10}

^{11}

^{12}

^{13}

^{14}

^{15}

^{1}

^{2}

^{1}

^{2}

These authors contributed equally as first author.

Infarct volume and other imaging markers are increasingly used as surrogate measures for clinical outcome in acute ischemic stroke research, but how improvements in these imaging surrogates translate into better clinical outcomes is currently unclear. We investigated how changes in infarct volume at 24 hours alter the probability of achieving good clinical outcome (modified Rankin Scale [mRS] 0–2).

Data are from endovascular thrombectomy patients from the randomized controlled ESCAPE-NA1 (Efficacy and Safety of Nerinetide for the Treatment of Acute Ischaemic Stroke) trial. Infarct volume at 24 hours was manually segmented on non-contrast computed tomography or diffusion-weighted magnetic resonance imaging. Probabilities of achieving good outcome based on infarct volume were obtained from a multivariable logistic regression model. The probability of good outcome was plotted against infarct volume using linear spline regression.

A total of 1,099 patients were included in the analysis (median final infarct volume 24.9 mL [interquartile range: 6.6–92.2]). The relationship between total infarct volume and good outcome probability was nearly linear for infarct volumes between 0 mL and 250 mL. In this range, a 10% increase in the probability of achieving mRS 0–2 required a decrease in infarct volume of approximately 34.0 mL (95% confidence interval: -32.5 to -35.6). At infarct volumes above 250 mL, the probability of achieving mRS 0–2 probability was near zero. The relationships of tissue-specific infarct volumes and parenchymal hemorrhage volume generally showed similar patterns, although variability was high.

There seems to be a near-linear association between total infarct volume and probability of achieving good outcome for infarcts up to 250 mL, whereas patients with infarct volumes greater than 250 mL are highly unlikely to have a favorable outcome.

Acute ischemic stroke (AIS) is often caused by a thrombus blocking blood flow in an intracranial artery. The result is ischemia and subsequent infarction of brain tissue, with resulting permanent disability, or, in severe cases, death [

We used a randomized trial sample of AIS patients who underwent EVT to investigate the association of quantitative imaging markers, namely total infarct volume, grey matter infarct volume, white matter infarct volume, and parenchymal hemorrhage volume on 24-h follow-up imaging with clinical outcomes at 90 days.

This study was a

All imaging data (baseline NCCT, multiphase CTA, catheter angiography, and follow-up NCCT and DWI-MRI) were reviewed in consensus readings (i.e., two-physician consensus readings were performed by default for all assessments) by an independent core lab (JO, BM, MJ, CZ, MA, MG, LR) (

The primary outcome in this study was the adjusted probability of achieving good functional outcome, defined as mRS 0–2 at 90 days. This probability was derived from a binary logistic regression analysis (see below).

Patient baseline characteristics in the entire patient sample and patients with and without good outcome were assessed using descriptive statistics and compared using Wilcoxon rank-sum test, Kruskal-Wallis test, or Fisher’s exact test, as appropriate to the type of data. To investigate the relationship between each quantitative imaging marker and good outcome, we obtained adjusted probabilities of achieving good outcome from multivariable logistic regression analysis. For each imaging marker, a separate regression model was built (

Predicted probabilities of good outcome with respective 95% confidence intervals (CIs) were plotted against the range of quantitative imaging markers. We then determined the change in total infarct volume that translates into a 1%, 5%, and 10% increase in the probability of mRS 0–2 by using a spline regression, whereby knots were placed based on visual assessment of the data distribution. Improvement of model fit with spline regression over and beyond a linear regression model was tested using a likelihood ratio test. The above-mentioned percentages were chosen based on a landmark study by Cranston et al. [

We performed a sensitivity analysis in which excellent clinical outcome (mRS 0–1 at 90 days) rather than good outcome (mRS 0–2 at 90 days) was used as the dependent variable. Although only 1 reader performed infarct volume measurements, there may be slight differences between measurements when different readers segment infarct volumes. Therefore, inter-rater agreement for tissue-specific infarct volume measurements was assessed in 25 randomly selected patients using Bland-Altman plots. All analyses were performed using Stata version 17 (StataCorp., College Station, TX, USA).

Of the 1,105 patients included in ESCAPE-NA1, follow-up imaging for total infarct volume assessment was available for 1,099 patients (NCCT: n=652, DWI-MRI: n=447) that were included in the analysis. Of the 447 patients with follow-up MRI, diffusion-weighted sequences could be postprocessed and tissue-specific infarct volumes obtained in 358 patients. Appropriate sequences for parenchymal hemorrhage volumetry were available in 1,054 patients. Baseline characteristics, quantitative tissue variables, and clinical outcomes of the study sample are shown in

Adjusted probabilities of achieving mRS 0–2 across the range of total infarct volume with respective 95% CIs are depicted in

Adjusted probabilities of achieving mRS 0–2 across the range of grey matter infarct volumes, white matter infarct volumes, and hemorrhage volumes with respective 95% CIs are depicted in

When using mRS 0–1 as the dependent variable, associations of overall infarct volume, tissue-specific infarct volumes, and hemorrhage volume were similar to the main analysis. Adjusted probabilities of achieving mRS 0–1 across the range of grey matter infarct volumes, white matter infarct volumes, and hemorrhage volumes with respective 95% CIs are depicted in

In this

Infarct volume reflects the extent of ischemic tissue damage, which is ultimately responsible for the disability AIS patients suffer from. Infarct volume has been shown to be a strong predictor of 90-day clinical outcome, independent of patient age, NIHSS, and other baseline characteristics. Some infarct patterns such as corticospinal tract infarction have a near-deterministic relationship with poor outcomes [

However, in order to establish quantitative imaging biomarkers of ischemic tissue damage as valid surrogate outcome markers, a strong and predictable association with clinical outcomes has to be proven. Furthermore, once such a relationship has been established, and if a treatment shows benefit on one of those imaging surrogate outcomes, it should only be considered relevant if this change in surrogate outcomes also results in a change in clinical outcomes that is meaningful to patients. For mRS 0–2, the minimally clinically important difference—i.e., the smallest change that is considered meaningful by patients—ranges between 1% and 10% [

We previously showed that the strength of the association between infarct volume and clinical outcome depends on infarct size [

There are several new pharmacological stroke treatments under investigation, all of which tackle different tissue damage mechanisms. Their effects may be reflected by different imaging markers: for example, drugs preventing primarily neuronal damage would be expected to reduce grey matter infarct volume more than white matter infarct volume, and drugs stabilizing the blood-brain barrier should reduce parenchymal hemorrhage volume. We, therefore, assessed the association between adjusted probability of good outcome and other quantitative tissue imaging markers, namely grey matter infarct volume, white matter infarct volume, and parenchymal hemorrhage volume. The patterns we observed resembled the pattern of total infarct volume, with a linear-appearing association with good outcome probability at small to moderate volumes, and near-zero probability of good outcome at large volumes. We further observed a steeper decrease in good outcome probability per mL white matter infarct volume increase compared to grey matter infarct volume, which confirms that white matter damage may be more detrimental than grey matter damage [

Our study has several limitations. First, infarct volumes were measured at 24 hours; however, it is known that infarcts continue to enlarge beyond 24 hours [

In conclusion, we observed a near-linear association between total infarct volume and adjusted probability of achieving good outcome in patients with infarcts up to 250 mL, while the probability in patients with larger infarcts was near zero. In the former subgroup, a decrease in infarct volume of approximately 3.4 mL, 17 mL, and 34 mL was roughly associated with a 1%, 5%, and 10% increase in the probability of achieving a good outcome respectively, although there is some uncertainty surrounding these estimates.

Supplementary materials related to this article can be found online at

Imaging core lab members

Tissue imaging markers used in this analysis

Variables included in the binary logistic regression models

Infarct segmentation on non-contrast head CT using the open-source software ITK-SNAP. (A) A segmented infarct area (total infarct volume) on a non-contrast head CT slice. (B) The same slice without the segmentation volume. Manual segmentation was repeated for each slice. Segmentations of grey matter infarct volumes, basal ganglia infarct volumes, and hemorrhage volumes was performed in an identical manner.

Example of a grey matter segmentation. (A) The 24-h diffusion-weighted MRI without segmentation. (B) The segmentation volume on an exemplary slice is highlighted in red. White matter infarcts (white arrows) are not included in the segmentation volume.

Inter-rater agreement for total 24-h infarct volume measurements for reader 1 (JO) and reader 2 (LR). Bland-Altman plot shows the average of reader 1 and reader 2 infarct volume measurements on the x-axis and the difference between both volume measurements on the y-axis. Each dot represents an individual patient. Mean difference between readers for total infarct volume was -0.30 mL (standard deviation 5.80 mL). Mean difference between readers for grey matter infarct volume was -2.34 mL (standard deviation 3.74 mL). Mean difference between readers for white matter infarct volume was 0.66 mL (standard deviation 3.05 mL).

Histogram illustrating the distribution of 24-h (A) white matter infarct volumes and (B) grey matter infarct volumes in the patient sample.

Adjusted probabilities of achieving mRS 0–1 across the range of grey matter infarct volumes, white matter infarct volumes, and hemorrhage volumes with respective 95% confidence intervals. (A) shows the association with whate matter infarct volume, (B) with grey matter infarct volume, (C) with total infarct volume, and (D) with hemorrhage volume. Note that the x-axes were truncated at 300 mL in (A) and (B), at 400 mL in (C) and at 500 mL in (D) for illustrative purposes. Black lines indicate adjusted probability of achieving mRS 0–2 and blue shaded areas indicate 95% CIs. Adjusted mRS probabilities were obtained from a binary logistic regression model. mRS, modified Rankin Scale; CI, confidence interval.

The ESCAPE-NA1 trial was funded by Alberta Innovates, Canadian Institutes of Health Research, NoNO Inc.

Johanna Ospel is a consultant for Nicolab. Mayank Goyal reports personal fees from Mentice, personal fees from Medtronic, personal fees from MicroVention, and personal fees from Stryker outside the submitted work; in addition, Dr Goyal has a patent to Systems of acute stroke diagnosis issued and licensed. The remaining authors have nothing to disclose.

Conceptualization: JMO. Study design: JMO. Methodology: JMO, LR. Data collection: all authors. Investigation: all authors. Statistical analysis: JMO. Writing—original draft: JMO. Writing—review & editing: all authors. Funding acquisition: MG, MH. Approval of final manuscript: all authors.

We acknowledge the ESCAPE-NA1 investigators and all the patients and their families for their contribution.

Probability of mRS 0–2 at 90 days as a function of total 24-hour infarct volume. (A) Adjusted probabilities of good clinical outcome (mRS 0–2 at 90 days) across the range of total 24-h infarct volumes from 0 mL to 500 mL (blue line) with corresponding 95% CIs (blue shaded area). Adjusted mRS probabilities were obtained from a binary logistic regression model with adjustment for patient age, sex, and baseline NIHSS. (B) Linear spline function modeling the association between total infarct volume and adjusted probability of achieving 90-day mRS 0–2. Knot at 250 mL was placed based on visual assessment and model fit parameters. Individual dots represent volumes and predicted probabilities of achieving mRS 0–2 for individual patients. Note that the x-axis in (A) and (B) was truncated at 500 mL for illustrative purposes. mRS, modified Rankin Scale; CI, confidence interval; NIHSS, National Institutes of Health Stroke Scale.

Adjusted probabilities for mRS 0–2 and respective confidence intervals and tissue-specific infarct and hemorrhage volumes. Adjusted probabilities of good clinical outcome (mRS 0–2 at 90 days) across the range of 24-h (A) grey matter infarct volumes, (B) white matter infarct volumes, and (C) parenchymal hemorrhage volumes. Note that the x-axes were truncated at 500 mL in (A) and 300 mL in (B) and (C) for illustrative purposes. Blue lines indicate adjusted probability of achieving mRS 0–2 and blue shaded areas indicate 95% CIs. Adjusted mRS probabilities were obtained from a binary logistic regression model (Supplementary Table 3). (D) A histogram of total infarct volumes in the patient sample. mRS, modified Rankin Scale; CI, confidence interval.

Baseline characteristics, clinical outcomes, quantitative tissue variables in the study sample (n=1,099)

Characteristic | Entire patient sample (n=1,099) | Patients with 90-day mRS 0–2 (n=665) | Patients with 90-day mRS >2 (n=434) | |
---|---|---|---|---|

Baseline characteristics | ||||

Age (yr) | 70.8 (60.7–79.8), n=1,099 | 66.7 (58.4–75.9), n=665 | 76.6 (66.9–83.3), n=434 | <0.001 |

Female sex | 546/1,099 (49.7) | 312/665 (46.9) | 234/434 (53.9) | 0.026 |

Baseline NIHSS | 17 (12–21), n=1,099 | 16 (11–20), n=665 | 18 (15–22), n=434 | <0.001 |

Baseline ASPECTS | 8 (7–9), n=1,099 | 8 (7–9), n=665 | 8 (7–9), n=434 | 0.024 |

ASPECTS | 0.068 | |||

0–6 | 192/1,099 (17.5) | 102/665 (15.3) | 90/434 (20.7) | |

7–8 | 587/1,099 (53.4) | 362/665 (54.4) | 225/434 (51.8) | |

9–10 | 320/1,099 (29.1) | 201/665 (30.2) | 119/434 (27.4) | |

Collaterals | <0.001 | |||

Poor | 46/1,087 (4.2) | 15/658 (2.3) | 31/429 (7.2) | |

Intermediate | 851/1,087 (78.3) | 524/658 (79.6) | 327/429 (76.2) | |

Good | 190/1,087 (17.5) | 119/658 (18.1) | 71/429 (16.6) | |

Treatment, procedural outcomes, and workflow times | ||||

Successful reperfusion (eTICI 2b-3) | 954/1,093 (87.3) | 618/661 (93.5) | 336/432 (77.8) | <0.001 |

Near-complete reperfusion (eTICI 2c-3) | 505/1,093 (46.2) | 349/661 (52.8) | 156/432 (36.1) | <0.001 |

Time from onset to randomization (min) | 187 (121–309), n=1,099 | 170 (114–275), n=665 | 217 (135–350), n=434 | <0.001 |

Time from study drug to reperfusion (min) | 22 (8–41), n=966 | 21 (8–37), n=627 | 26 (9–52), n=339 | 0.001 |

Time from qualifying CT to groin puncture (min) | 45 (30–65), n=1,096 | 44 (30–62), n=663 | 48 (31–70), n=433 | 0.059 |

General anesthesia | 190/1,093 (17.4) | 86/662 (13.0) | 104/421 (24.1) | <0.001 |

Intravenous alteplase | 656/1,099 (59.7) | 423/665 (63.6) | 233/434 (53.7) | 0.001 |

Intravenous nerinetide | 547/1,099 (49.8) | 336/665 (50.5) | 211/434 (48.6) | 0.538 |

Quantitative tissue imaging variables | ||||

Total infarct volume (mL) | 24.9 (6.6–92.2), n=1,099 | 13.0 (3.9–36.3), n=665 | 87.1 (26.8–203.2), n=434 | <0.001 |

Grey matter infarct volume (mL)^{*} |
18.8 (8.5–37.3), n=358 | 13.6 (6.3–26.9), n=245 | 35.2 (17.9–60.9), n=113 | <0.001 |

White matter infarct volume (mL)^{*} |
0 (0–8.3), n=358 | 0 (0–1.4), n=245 | 8.1 (0–52.6), n=113 | <0.001 |

Superficial grey matter infarct volume (mL)^{*} |
6.3 (0–29.3), n=358 | 2.0 (0–17.5), n=245 | 25.3 (6.1–51.9), n=113 | <0.001 |

Deep grey matter infarct volume (mL)^{*} |
7.0 (2.1–13.5), n=358 | 6.5 (2.1–12.9), n=245 | 8.8 (2.1–15.7), n=113 | 0.205 |

Parenchymal hemorrhage | 351/1,099 (31.9) | 197/665 (29.6) | 197/434 (45.4) | <0.001 |

Parenchymal hemorrhage volume (mL)^{†} |
0 (0–1.1), n=1,054 | 0 (0–0), n=648 | 0 (0–8.9), n=406 | <0.001 |

Clinical outcomes | ||||

mRS 0–2 at 90 days | 665/1,099 (60.5) | - | - | - |

Mortality at 90 days | 142/1,099 (12.9) | - | 142/434 (32.7) | - |

Ordinal mRS at 90 days | 2 (1–4), n=1,099 | 1 (0–2), n=665 | 5 (3–6), n=434 | - |

Symptomatic intracranial hemorrhage | 31/1,096 (2.8) | 4/664 (0.6) | 27/432 (6.3) | <0.001 |

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

mRS, modified Rankin Scale; NIHSS, National Institutes of Health Stroke Scale; ASPECTS, Alberta Stroke Program Early Computed Tomography Score; eTICI, expanded Treatment in Cerebral Infarction Score; CT, computed tomography; DWI, diffusion-weighted imaging; MRI, magnetic resonance imaging.

Only assessed in patients with available 24-h follow-up DWI-MRI (n=358);

Only assessed in patients with available appropriate 24-h imaging (n=1,054).