The Minimal Clinically Important Difference for Achievement of Substantial Reperfusion With Endovascular Thrombectomy Devices in Acute Ischemic Stroke Treatment

Once again having the completely wrong endpoint to be measured, reperfusion, rather than 100% recovery.

The Minimal Clinically Important Difference for Achievement of Substantial Reperfusion With Endovascular Thrombectomy Devices in Acute Ischemic Stroke Treatment

Chun-Jen Lin^1* and Jeffrey L. Saver²

• ¹Neurological Institute, Taipei Veterans General Hospital and School of Medicine, National Yang-Ming University, Taipei, Taiwan
• ²Comprehensive Stroke Center and Department of Neurology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, CA, United States

Background and Purpose: Recent noninferiority clinical trials of novel endovascular thrombectomy devices for acute ischemic stroke have used substantial reperfusion as the primary outcome of achievement. Determining the minimal clinically important difference (MCID) is an essential step for the design of noninferiority clinical trials.

Materials and Methods: We surveyed international neuro-interventionalist and noninterventional vascular neurologist investigators. The questionnaire included demographic characteristics, level of clinical experience, and their MCID clinical scenario-based judgment regarding the MCID for the outcome substantial reperfusion (thrombolysis in cerebral infarction score 2b-3) within 3 passes.

Results: Survey responses were received from 58 of 200 experts. Among responders, 75.9% were neuro-interventionalists (most commonly interventional neuroradiologists and interventional neurologists, followed by endovascular neurosurgeons), and 24.1% were noninterventional vascular neurologists; 87.9% had been in practice for more than 5 years, and 67.3% devoted more than half of their practice to stroke care. Responder–nonresponder and continuum of resistance analysis indicated responders were representative of the full expert population. Among experts, the median MCID for substantial reperfusion was 3.1–5% (interquartile range 1.1–3% to 5.1–10%). MCID distributions did not differ among neuro-interventionalists and noninterventional vascular neurologists.

Conclusions: Neuro-interventionl and noninterventional stroke experts judged that the minimal clinically important difference in comparing thrombectomy devices for achieving substantial reperfusion is 3.1–5%. This MCID, lower than noninferiority margins used in several recent clinical trials, can inform trial designs and clinical development.

The appropriate clinical trial designs to assess device treatments for a disease may be divided into four broad types, based on 2 types of endpoint and 2 types of design. The two types of primary endpoints are (1) technical efficacy endpoints, used when the trial is assessing the device's effectiveness as a tool (does it do what it says?), and (2) clinical efficacy endpoints when the trial is assessing the device's effectiveness as a care strategy (does the patient have a better final health state as a result?). The two types of designs are (1) superiority clinical trials, in which a novel device is tested to determine if it yields better technical or clinical outcomes than a standard device or medical therapy alone, and (2) noninferiority clinical trials, in which a novel device is tested to determine if it yields technical or clinical outcomes at least as good as a standard device or medical therapy alone.

Trials using technical, rather than clinical, efficacy outcomes may be of particular importance at early and at final stages in the initial development of a device class and use case. Early in device development, iterative innovation in device design to optimize its technical performance is a leading concern. Once device designs are sufficiently mature, it is desirable to perform superiority trials against older care strategies not using the device class to demonstrate that using the device class does ultimately improve the final clinical outcomes of patients. When a device class has been shown to have clinical utility, further novel, within-class devices may often be adequately assessed again with the use of technical efficacy primary endpoints. At this stage, when the device class has already attained a generally high level of technical performance success, noninferiority trials with technical efficacy endpoints are useful, in which novel device variations are shown to be at least as good as existing, approved devices.

The development of endovascular thrombectomy (EVT) devices for acute ischemic stroke due to large vessel occlusion (AIS-LVO) has recently generally followed this trajectory. A wave of several superiority design trials were first performed and demonstrated the superiority of endovascular thrombectomy over medical therapy alone (1–9). Thereafter, trials assessing successor variations in thrombectomy device design have generally used noninferiority designs, seeking to demonstrate that newer devices are at least as good as existing devices (10–12).

For noninferiority clinical trial design, even more so than for superiority clinical trial design, determining the minimal clinically important difference (MCID) in an outcome is a crucial, and challenging, step that affects trial power. The MCID is the smallest change in a disease outcome that a patient and/or a care provider would identify as worthwhile (13). For superiority trials, seeking to determine if new treatment A is better than conventional treatment B, the MCID establishes the threshold difference at which superiority can be declared. If treatment A yields an incremental increase in favorable outcomes that exceeds the MCID, then superiority is established. For noninferiority and equivalence trials, the MCID establishes the threshold value (“margin”) at which noninferiority or equivalence can be declared. If treatment A does not yield an incremental increase in favorable outcomes that exceeds the MCID, then noninferiority is established. For both superiority and noninferiority/equivalence trials, requisite sample size is inversely related to the amplitude of the MCID. The smaller the MCID for a particular outcome, the larger a trial must be in order to be adequately powered to determine if a novel treatment exceeds or does not exceed that threshold.

However, for EVT treatment of AIS-LVO, the MCID has not been formally established for the most commonly employed technical efficacy outcome: achieving substantial reperfusion (i.e., achieving a thrombolysis in cerebral infarction, TICI, scale score of 2b-3). During the conduct of a recent noninferiority meta-analysis of randomized trials comparing different EVT techniques, a systematic literature search identified two expert survey studies characterizing the MCID for the clinical efficacy outcome of functional independence (modified Rankin Scale 0–2) at 3 months but no expert survey study characterizing the MCID for the technical efficacy outcome of achievement of substantial reperfusion (14). The current study was, therefore, undertaken to provide that desirable additional frame.

Approaches to establishing the MCID for a particular outcome are of 3 broad types: Delphi expert–based approaches, anchor-based approaches, and distribution-based approaches (15). For an acute-onset disease such as stroke, the Delphi expert–based approach is generally preferred (16), especially for a technical efficacy outcome such as reperfusion, which requires technical expertise to appreciate, rather than patient-level subjective experience of a disease state. Therefore, we undertook a Delphi-expert survey study of leading international neuro-interventionalists and noninternventional vascular neurologists to the MCID of achieving substantial reperfusion with endovascular thrombectomy devices in patients with AIS-LVO.