By: Dr. Johannes Kaesmacher, Inselspital, Bern, Switzerland
twitter: @CheesemakerMD
Most trials comparing intravenous thrombolysis (IVT) plus mechanical thrombectomy (MT) with direct mechanical thrombectomy alone (MT) were designed as non-inferiority trials. Non-inferiority trials have been used for different reasons and with different goals. Hence, their results cannot be interpreted without acknowledging the non-inferiority margin and the framework of the trial. For some specific scenarios and drug classes preservation of e.g. 50% of the treatment effect will suffice for demonstrating a specific kind of non-inferiority (i.e. demonstrating indirect superiority over placebo) while other margins are used for different, usually more conservative purposes. The chosen margin is the central aspect of the trial, will depend on what the investigators want to demonstrate, how the trial will be interpreted and how much financial and organisational resources have to be bound.
The highest level of non-inferiority is usually claimed if data provides evidence that treatments are clinically indistinguishable. This implies that the chosen non-inferiority margin is smaller than the minimal clinically important difference, which is the smallest difference meaningful to patients. Other levels of non-inferiority may include reasonable comparability acknowledging a distinct level of uncertainty, or e.g. preservation of a substantial fraction of the treatment-effect considering other advantages (e.g. lower cost or substantially fewer side effects).
Considering the stated aim and chosen absolute risk difference of non-inferiority margins in trials comparing IVT + MT versus direct MT, it seems reasonable to conclude that the trials fall into the category of aiming to demonstrate reasonable comparability or probably less likely, clinical acceptability considering other advantages.
In three of these trials, which used absolute risk differences, the margins were defined so that 60% or 51% of the combined IVT+MT treatment effect (versus IVT alone) in SWIFT Prime is preserved. These are by all means generous margins, considering the framework of demonstrating reasonable comparability. In light of these margins and associated sample sizes, it becomes clear that each trial alone and even a meta-analysis of all trials will not be powered to demonstrate clinical indistinguishability (considering that the underlying true absolute risk difference is zero). While one can test non-inferiority regarding this margin for guideline purposes and to formally demonstrate that this most conservative margin is not met, this is methodologically an expected result. It therefore seems reasonable to put forward the argument that the trials should be tested within the non-inferiority framework they were designed in. This leads us to the questions at what level of uncertainty reasonable comparability should be claimed and at what level physicians would change their management. Should one consider skipping IVT if one can be sufficiently sure that no more than ten people out of 100 will not regain functional independence due to skipping IVT? Should it be one in 100? Should it be one in 1000?
Likely, the answers to this question differ according to the physician asked, their training background, their priors regarding IVT or MT and possibly prior participation in trials.
Currently we are faced with an array of published/presented trials, with most showing an inconclusive result (neither non-inferiority, nor inferiority or superiority is demonstrated). It is my perception that after the presentation/publication of the recent trials, the opinions seems to have shifted in favor of IVT before MT. However, one could also change the perspective and argue that probably no guideline would currently have recommended IVT before MT if MT had been the active comparator and IVT + MT was tested as the experimental arm in a superiority trial design.
It is easy to conclude that – in contrast to developments in acute myocardial infarction – thrombolysis before the intervention is far away from biting the dust, but the dust from inconclusive results is far away from being settled. One leap forward may be a deliberate discussion about what physicians and patients would consider an acceptable uncertainty, acknowledging the trial frame-work/margins the trials were designed in, chose the margins for pooled analyses accordingly and consider clinical meaningful and likely causal secondary endpoints, for which the power to demonstrate reasonable comparability or even clinical indistinguishability may be higher.
It might be that there will be a role for direct MT, but maybe the choirs of stent-retrievers and distal aspiration catheters are already singing along, kindly asking IVT to stay:
How do you think I’m going to get along
Without you when you’re gone
Reference included in the title and at the end of the text by Queen – Another One Bites the Dust
Contributions regarding non-inferiority trials and their framework by Jeffrey L. Saver (UCLA, Los Angeles, USA).