Study Criteria Applied to Real Life—A Multicenter Analysis of Stroke Patients Undergoing Endovascular Treatment in Clinical Practice

Nonstandard Abbreviations and Acronyms

Randomized controlled clinical trials (RCT) have demonstrated the positive effect of endovascular treatment (EVT) in patients with anterior circulation large vessel occlusion (LVO).^{1, 2, 3, 4, 5} In most of these trials, strict inclusion and exclusion criteria were applied. Based on these results EVT has become the standard of care for patients suffering from LVO. However, current guidelines allow treating patients outside of these criteria to a certain extent. Studies investigating the outcome of stroke patients undergoing EVT for LVO in clinical practice show clinical and procedural outcomes which differ from the one published in the RCT.^{6, 7, 8, 9}

To our knowledge, this study is the first multicentric study analyzing the extent the criteria of the RCT can be applied and the results expected in stroke patients undergoing endovascular treatment for LVO in daily practice. Furthermore, this work investigates the outcome of real‐life patients who underwent endovascular treatment despite not meeting all of the RCT criteria.

To do so, we applied the inclusion and exclusion criteria of the RCT and examined their gross selectiveness as well as patient characteristics, clinical and interventional outcome parameters of stroke patients undergoing EVT in daily routine before and after application of the criteria.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Patients

A total of 2611 patients enrolled in the GSR‐ET were screened for inclusion.^{6, 7} The GSR‐ET is an ongoing, open‐label, prospective, multicenter registry of consecutively collected EVT patients, with 25 participating stroke centers in Germany. The inclusion criteria of the SWIFT‐PRIME (Solitaire with the Intention for Thrombectomy as Primary Endovascular Treatment trial), MR CLEAN (Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands trial), ESCAPE (Endovascular Treatment for Small Core and Anterior Circulation Proximal Occlusion with Emphasis on Minimizing CT to Recanalization Times trial), DAWN (DWI or CTP Assessment with Clinical Mismatch in the Triage of Wake‐Up and Late Presenting Strokes Undergoing Neurointervention with Trevo trial), and DEFUSE‐3 (Endovascular Therapy Following Imaging Evaluation for Ischemic Stroke trial) trials were applied to patients without prior selection (Figure 1A).

• Download figure
• Download PowerPoint

Statistical Analysis

Baseline characteristics were compared performing Fisher Exact test for categorical variables, Mann–Whitney U test (non‐normally distributed data) and the unpaired Student t‐test (normally distributed data) for continuous variables. Statistical analyses were performed using SPSS Version 22. Study protocols and procedures were conducted in compliance with the Declaration of Helsinki and in accordance to ethical guidelines. A P<0.05 was considered statistically significant.

Results

Trial Criteria Applied to Patients of the GSR‐ET Registry

Following selection, it was found that only 35% (MR CLEAN), 14% (ESCAPE), 9% (SWIFT‐PRIME), 4% (DAWN), and 3% (DEFUSE‐3) of all patients of the GSR‐ET could have been included in the respective trials (Figure 1A and 1B). Because the patients of the GSR‐ET were treated before the publication of the DAWN and DEFUSE‐3 trials, and in accordance with guidelines with more stringent time windows, only few patients met the criteria and were therefore excluded from statistical analysis.

Clinical Factors That Deferred From the Trial Criteria and Their Influence on the Clinical and Procedural Outcomes

An analysis of the clinical characteristics of the patients of the GSR‐ET showed differences in a multitude of aspects compared with the trial populations of the RCT (Table 1). With regard to clinical characteristics, the patients of the GSR‐ET were older than the patients of the RCT with a median age of 75 years (interquartile range, 64–82 years). In terms of their medical history, the patients of the GSR‐ET had a lower percentage of individuals with a modified Rankin Scale of 0 or 1 (81%) compared with the MR CLEAN (90%) or the SWIFT‐PRIME trial (98%). As for the occluded vessel, we observed several differences, eg, a lower percentage of patients suffering from an occlusion of the proximal segment (M1) of the middle cerebral artery (54% versus 66% in the MR CLEAN or 77% in the SWIFT‐PRIME study). Additionally, the GSR‐ET trial population includes patients with an occlusion of the anterior cerebral artery (3%) and in 5% of the cases an occlusion in the posterior circulation.

In an additional step, we set out to identify the most selective parameter of the RCTs (Figure 2); in the MR CLEAN trial, the criterion to start treatment within 6 hours from symptom onset was the most selective, and was fulfilled by 51% of all patients of the GSR‐ET. For those that did not fulfill the criterion, the time point of symptom onset was either unknown (43%) or the time from symptom onset to start of treatment exceeded 6 hours (6%). For the SWIFT‐PRIME trial, initiation of i.v. lysis within 4.5 hours from symptom onset was amongst the most selective criteria, because of the fact that 45% of all patients of the GSR‐ET did not receive i.v. lysis. In contrast, the initiation of endovascular treatment within 60 minutes of baseline non‐contrast CT was the most selective criterion of the ESCAPE trial (fulfilled by 47% of all patients of the GSR‐ET).

• Download figure
• Download PowerPoint

Following application of the inclusion criteria, the patients of the GSR‐ET showed an increase of excellent (modified Rankin scale at 90 days of 0–1) and good (modified Rankin Scale at 90 days of 0–2) functional outcome (Table 2). Statistically significant differences were observed if the criteria of the SWIFT‐PRIME (excellent clinical outcome in 26% [unselected GSR] versus 42% [selected GSR]; P=0.031) and ESCAPE (excellent clinical outcome 26% [unselected GSR] versus 35% [selected GSR]; P=0.042) trial were applied (Table S1). The same held true for the percentage of Thrombolysis in Cerebral Infarction Scale 2b and 3 recanalizations (Table 2) which, after application of the trial criteria, increased from 83% (all patients of the GSR‐ET) to 85% (MR CLEAN), 91% (SWIFT‐PRIME) or 88% (ESCAPE) (Table S2).

Clinical and Procedural Outcomes of Patients of the GSR‐ET Inside and Outside the Clinical Trials

Furthermore, the clinical outcomes of the patients of the GSR‐ET who fulfilled the study criteria of the MR CLEAN, SWIFT‐PRIME, and ESCAPE trials were compared with the EVT arm of each of these studies. The percentage of excellent clinical outcome of the patients of the GSR‐ET increased, regardless of the applied study criteria. The results were comparable with the published data of the RCT (Figure 3A). The patients of the GSR‐ET showed a good clinical outcome in 42% if selected according to the MR CLEAN criteria (33% in the EVT study arm of the MR CLEAN trial), 56% if the SWIFT‐PRIME criteria were applied (60% in the EVT study arm of the SWIFT‐PRIME trial), and 46% if the patients of the GSR‐ET were selected according to the criteria of the ESCAPE trial (53% in the EVT study arm of the ESCAPE trial). No statistical significance could be demonstrated.

• Download figure
• Download PowerPoint

In a next step, we investigated the outcome of patients of the GSR‐ET who did not meet the inclusion criteria of the MR CLEAN study (n=1701). We observed a decrease in good clinical outcome from 41% to 34% (Figure 3B) which was still comparable with the results published by the investigators of the MR CLEAN trial (33%, Figure 3A).

Discussion

In this study, <35% of the patients who were treated in clinical routine would have met the inclusion criteria of the recent large RCTs.^{6, 7} Following selection according to the trial criteria, the clinical and procedural outcome parameters of the patients of the GSR‐ET improved significantly. We observed that the clinical and procedural outcomes of the patients from the GSR‐ET were comparable with those of the RCT after selection and that patients who did not meet the criteria of the RCT still had high rates of good clinical outcome.^{1, 2, 3} To the best of our knowledge, our study is the first multicentric study to investigate how and to which extent the results of the large EVT trials can be applied to stroke patients treated on a daily basis in the clinical routine. We show that only a relatively small subset of the patients fulfilled the inclusion criteria of the large EVT trials, a fact that should be kept in mind when interpreting the outcome parameters of patients who are treated during daily practice.

Interestingly, amongst the MR CLEAN, SWIFT‐PRIME, and ESCAPE studies, the degree of patient selection varies greatly.^{1, 2, 3, 4, 5, 10, 11} We demonstrated that, for our cohort, the MR CLEAN trial is the least selective, with 35% of our patients fulfilling the criteria. This result could likely be explained by several deferring factors. For example, the MR CLEAN trial has the fewest, least stringent criteria amongst the RCT.¹ This is in stark contrast to the SWIFT‐PRIME and ESCAPE trials, which have not only more, but also more narrowly defined inclusion criteria, leading to a highly selective patient study population and, subsequently, a higher rate of good and excellent outcomes.^{2, 3} Indeed, <10% of the patients from the GSR‐ET fulfilled the criteria of the ESCAPE and SWIFT‐PRIME study. Amongst the factors that most frequently diverged from the trial criteria and guideline recommendations were patient characteristics and time metrics. One factor are preexisting disabling deficits (modified Rankin Scale >1) which can be found in 19% of the patients of the GSR‐ET, compared with much lower rates in the RCT (10% in the MR CLEAN or 2% in the SWIFT‐PRIME trial).^{1, 3, 6, 7, 9} Other factors are the median patient age (75 years) which is higher in the GSR‐ET group than in the RCT trials (eg, age 66 years in the MR CLEAN or 65 years in the SWIFT‐PRIME study) as well as a high proportion (28%) of patients with an unknown time of symptom onset and the fact that the GSR‐ET registry contains patients with a low ASPECT score.^{6, 7} These factors are known predictors to influence the clinical outcome of stroke patients and are integral parts of international guidelines for the treatment of stroke patients,^{12, 13, 14} thereby providing an explanation for the inferior clinical outcome of unselected stroke patients treated in real life when compared with the results of the RCT.

The importance of the above mentioned time metrics for the daily practice becomes even more obvious, when looking at the large percentage of patients from the GSR‐ET who did not meet the inclusion criteria of RCT because of the fact that the time of symptom onset could not be determined or was too long (Figure 2A). This important aspect was addressed by the DAWN and DEFUSE‐3 trials which changed the stroke treatment guidelines with regard to the time metrics, since these RCT demonstrated that thrombectomy can improve the outcome of stroke patients beyond 6 hours since symptom onset if multimodal imaging was used for patients selection.^{10, 11} Therefore it appears that there are other factors besides time for which we have to adjust for when making treatment decisions for stroke patients in daily practice, which is currently being investigated in ongoing studies.

In addition, there are several other aspects that should be considered when interpreting the outcomes of patients of the GSR‐ET after selection according to the RCT trial criteria. Amongst others this includes the year of patient admission. Treatment of the patients of the GSR‐ET took place between 2015 and 2018, a period in which EVT of stroke patients was already the standard of care for patients with LVO at the participating study centers. In contrast, at the time of the RCTs, endovascular treatment of stroke patients was a novel approach. Thus, the observed differences in clinical and procedural outcome could be partially attributable to the level of experience of the interventionalists.

The RCTs were conducted in large, highly specialized stroke departments whereas the real‐world data show the results from a heterogeneous group of stroke centers, ranging from hospitals with <50 cases to hospitals with several hundred cases a year. Because of this heterogeneity, one could expect inferior clinical and procedural outcome in real‐life patients since several studies have shown the association of annual hospital volume and the outcome of stroke patients, and higher mortality has been described in real world data.^{7, 15, 16} However, the results of our study showed that applying the inclusion criteria of the RCT on real‐world data made the real‐world data comparable with the results of the RCT (Figure 3). This held true not only for clinical outcome but also for procedural parameters. Hence, our data suggest that the procedural quality of MT in real‐life is comparable with that in RCT.

Intriguingly the results of our study suggest that it seems ethical to treat patients outside of the trial qualification criteria, since when looking at the outcome of patients of the GSR‐ET who did not fulfill all of the trial criteria, their clinical outcome was not inferior to the MR CLEAN results,¹ even if these patients had a relatively poor prognosis in the natural course (eg, because of their significantly higher age). Good clinical outcome has been previously reported in patients treated outside top‐tier evidence.^{17, 18, 19} Hence, overly selective treatment criteria could lead to the exclusion of patients that benefit from EVT.¹⁵ Therefore, we need further trials to identify factors beyond the established inclusion and exclusion criteria of the RCTs or recommendations of international guidelines to identify stroke patients who benefit from endovascular stroke treatment.

Furthermore, our study raises the question of the external validity of the large RCT. Based on the results of the RCT, EVT has become the cornerstone of ischemic stroke management but unfortunately it remains unclear to which extend the results of the trials can be reasonably applied to stroke patients with LVO in routine practice. In real life, RCTs cannot be expected to produce results that are directly relevant to all patients and all settings, but they should be designed and reported in a way that allows clinicians to judge to whom they can reasonably be applied. Unfortunately, the applicability varies amongst the RCT for patients with LVO, mostly because of differences in the selection of the patients, setting of the trials as well differences between the trial protocol and routine practice.^{20, 21} This leaves the clinician with uncertainty which patient to treat and which not, if they do not completely fulfill the study criteria of the RCT.

Our study has several limitations. Our registry included patients treated by EVT based on individual treatment decisions by the interventionalists. This may lead to a selection bias towards patients in whom EVT was deemed to likely be successful. As this was not an RCT, we cannot judge efficacy of EVT. However, recanalization rates and clinical outcome of selected patients of the GSR‐ET were comparable or even better than those of the treatment arm of the MR CLEAN trial and therefore a treatment effect is highly probable.

In conclusion, the majority of stroke patients treated by EVT in clinical practice do not meet the stringent criteria of the large EVT guideline defining RCT, despite this observation, patients who did not meet the criteria of the RCT still had considerable rates of good clinical outcome.

Appendix

GSR‐ET Collaborators

Site Investigators: Tobias Boeckh‐Behrens, MD, Klinikum r.d. Isar, Munich, Germany; Silke Wunderlich, MD, Klinikum r.d. Isar, Munich, Germany; Martin Wiesmann, MD, Uniklinik RWTH Aachen, Germany; Ulrike Ernemann, MD, Tübingen University Hospital, Germany; Till‐Karsten Hauser, MD, Tübingen University Hospital, Germany; Eberhard Siebert, MD, Charité—Campus Benjamin Franklin und Campus Charité Mitte, Berlin, Germany; Sarah Zweynert, MD, Charité—Campus Virchow Klinikum, Berlin, Germany; Georg Bohner, MD, Charité—Campus Virchow Klinikum, Berlin, Germany; Alexander Ludolph, MD, Sana Klinikum Offenbach, Germany; Karl‐Heinz Henn, MD, Sana Klinikum Offenbach, Germany; Waltraud Pfeilschifter, MD, Uniklinik Frankfurt/Main, Germany; Marlis Wagner, MD, Uniklinik Frankfurt/Main, Germany; Joachim Röther, MD, Asklepios Klinik Altona, Hamburg, Germany; Bernd Eckert, MD, Asklepios Klinik Altona, Hamburg, Germany; Jörg Berrouschot, MD, Klinikum Altenburger Land, Altenburg, Germany; Albrecht Bormann, MD, Klinikum Altenburger Land, Altenburg, Germany; Christian Gerloff, MD, University Medical Center Hamburg‐Eppendorf, Hamburg, Germany; Elke Hattingen, MD, University Hospital Bonn, Germany; Gabor Petzold, MD, University Hospital Bonn, Germany; Sven Thonke, MD, Klinikum Hanau, Germany; Christopher Bangard, MD, Klinikum Hanau, Germany; Christoffer Kraemer, MD, Klinikum Lüneburg, Germany; Martin Dichgans, MD, Ludwig Maximilian University of Munich, Germany; Frank Wollenwebe, MD, Ludwig Maximilian University of Munich, Germany; Lars Kellert, MD, Ludwig Maxi milian University of Munich, Germany; Franziska Dorn, MD, Ludwig Maximilian University of Munich, Germany; Moriz Herzberg, MD, Ludwig Maximilian University of Munich, Germany; Marios Psychogios, MD, Georg‐August‐Universität Göttingen, Germany; Jan Liman, MD, Georg‐August‐Universität Göttingen, Germany; Martina Petersen, MD, Klinikum Osnabrück, Germany; Florian Stögbauer, MD, Klinikum Osnabrück, Germany; Peter Kraft, MD, University Hospital Würzburg, Germany; Mirko Pham, MD, University Hospital Würzburg, Germany; Michael Braun, MD, Bezirkskrankenhaus Günzburg, Germany; Gerhard FHamann, MD, Bezirkskrankenhaus Günzburg, Germany; Andreas Kastrup, MD, Klinikum Bremen Mitte, Germany; Christian Roth, MD, Klinikum Bremen Mitte, Germany; Klaus Gröschel, MD, University Medical Center Mainz, Germany; Timo Uphaus, MD, University Medical Center Mainz, Germany; Volker Limmroth, MD, Kliniken Köln, Germany.

Sources of Funding

None.

Disclosures

Fiehler is a consultant for Acandis, Boehringer Ingelheim, Codman, Microvention, Sequent, and Stryker and a speaker for Bayer Healthcare, Bracco, Covidien/ev3, Penumbra, Philips, and Siemens. He has received grants from Bundesministeriums für Wirtschaft und Energie, Bundesministerium für Bildung und Forschung, Deutsche Forschungsgemeinschaft, European Union, Covidien, Stryker (Thrombectomy in patients ineligible for iv tPA study), and Microvention (ERic Acute StrokE Recanalization study). Thomalla received personal fees as a consultant or lecturer from Acandis, Bayer, Boehringer Ingelheim, Bristol‐Myers Squibb/Pfizer, Daichi Sankyo, and Stryker and research grants from Bayer, Federal Ministry for Economic Affairs and Energy (Bundesministeriums für Wirtschaft und Energie), Corona‐ Foundation, German Research Foundation (Deutsche Forschungsgemeinschaft), Else Kröner‐ Fresenius Foundation, European Union (Horizon 2020), and German Innovation Fund. Gerloff reports honoraria as speaker and/or scientific advisory board member from Actitor Biotech, Amgen, Bayer Healthcare, Boehringer Ingelheim, Prediction Biosciences, Sanofi Aventis. The remaining authors have no disclosures to report.

Footnotes