To enchant, from the Latin *incant**?**re,* meaning to be put under a spell. I will leave it to the insightful reader to interpret its meaning in the current context.

Needless to say when the authors of the Enhanced Control of Hypertension and Thrombolysis Stroke or ENCHANTED study endeavored to examine the efficacy of tPA in acute ischemic stroke, they selected their trial name appropriately.

Anderson et al examined 3,206 adult patients presenting to the Emergency Department with acute ischemic stroke within 4.5 hours of symptom onset, and randomized them to either standard-dose alteplase (0.9 mg/kg) or a low-dose strategy (0.6 mg/kg)(1).

The authors found that the number of patients who were dead or disabled (a mRS of 2-6) at 90-days was 53.2% and 51.1% in the low-dose and standard-dose groups respectively. This produced an odds ratio of 1.09 for increased risk death or disability in the patients who received the low-dose tPA. The 95%-confidence intervals surrounding this point estimate were 0.95 t0 1.25. Because the upper limit of this confidence interval crossed the predetermined non-inferiority margin of 1.14 the authors were unable to demonstrate non-inferiority. And while these findings initially appear grim for the low-dose strategy, in reality its utility is likely far more promising.

Despite this statistical failure, low-dose tPA performed admirably when compared to its standard-dose comparator. The number of patients alive and independent (mRS of 0,1,or 2) at 90 days was 62.4% and 63% in the low-dose and standard-dose groups respectively. In fact, at both 7 and 90 days, patients randomized to receive standard-dose thrombolytic died more frequently than those who received the low-dose strategy. Death at 7-days occurred in 60 patients (3.6%) in the low-dose cohort compared to 87 (5.3%) in the full-dose group. This 1.7% increase reached statistical significance with a p-value of 0.02. At 90-days the mortality was 8.5% vs 10.3% respectively. Furthermore, patients randomized to the standard-dose group experience higher rates of intracerebral hemorrhage (ICH). Major symptomatic ICH, according to SITS-MOST criteria, occurred in 1.0% of the patients in the low-dose group and in 2.1% of those in the standard-dose group. Fatal ICH occurred in 2.5% and 1.3% of the standard and low-dose groups respectively.

Once again we are faced with a trial examining the efficacy of tPA in acute ischemic stroke where reality and the statistical manner in which it is presented appear distinctly different. Despite the low-dose tPA group outperforming its full-dose comparator in almost every measure, this was a negative trial. This failure is based on how these authors defined success and our continued reliance on hypothesis testing.

Typically, a non-inferiority trial is performed to ensure that a new treatment strategy is *no worse* than the current strategy. This is statistically defined as a threshold below which the novel strategy cannot be considered *non-inferior.*This threshold is typically set by the authors conducting the trial(2). In this case the authors selected their non-inferiority margin at an odds ratio of 1.14. This number is based of the 2014 Cochrane analysis examining the efficacy of IV tPA vs placebo in acute ischemic stroke (3). In which, Wardlaw et al cite an odds ratio of 1.24 for increased death or disability in the patients randomized to receive placebo when compared to those who received IV tPA. Using this number as their guide, Anderson et al estimated that to ensure low -dose tPA retains at least half the efficacy of the standard dosing strategy, a threshold of 1.14 should not be crossed.

Because the upper limit of the 95%-confidence interval crosses the 1.14 threshold, the authors were unable to demonstrate non-inferiority. But failing to demonstrate non-inferiority is not equivalent to proving superiority. In fact, the authors were unable to demonstrate a statistical superiority of either treatment strategy when examining their primary endpoint. The authors found a 2.1% absolute difference in the incidence of death or disability between the two groups in favor of the standard dose tPA. The 95%-CI surrounding this point estimate is -1.4%-5.6%.

Despite its statistical failure, the ENCHANTED trial clearly demonstrates that reduced-dose tPA achieves similar clinical outcomes while decreasing death and ICH. But this exercise distracts us from the more important question, *what is the optimal dosing strategy of IV tPA?* As some of us still believe, and the data would suggest, a further reduction in dosing may be equally potent. A serial dilution until all that remains is the essence of the tPA that once was. This *no-dose strategy*, may further decrease the rates of symptomatic intracerebral hemorrhage and death with no noticeable difference in neurological outcomes. Enchanting indeed…

Sources Cited:

- Anderson CS, Robinson T, Lindley RI, et al. Low-dose versus standard-dose intravenous alteplase in acute ischemic stroke. N Engl J Med. DOI: 10.1056/NEJMoa1515510
- Kaul S, Diamond GA. Good Enough: A Primer on the Analysis and Interpretation of Noninferiority Trials. Ann Intern Med. 2006;145:62-69
- Wardlaw JM, Murray V, Berge E, del Zoppo GJ. Thrombolysis for acute ischae- mic stroke. Cochrane Database Syst Rev 2014;7:CD000213.

### Rory Spiegel

University of Maryland

Resuscitation Fellowship Graduate

Creator

EMNerd.com

#### Latest posts by Rory Spiegel (see all)

- CC Nerd-The Case of the Subjective Truth - October 28, 2018
- EM Nerd-The Case of the Inferior Superiority - October 18, 2018
- EM Nerd-An Addendum to the Case of the Non-inferior Inferiority - September 30, 2018

To summarize then, while the low dose did not meet the non-inferiority threshold, neither did it underperform against the standard dose (neither one is superior).

Based on this trial then, both the low and normal dose may be non-inferior to the upper bounds set by the investigators?

Confusing, but I think thats it, yes?

Brad

Hey Brad,

Thanks for writing! The non-inferiority threshold was based of the efficacy of standard-dose TPA when compared to placebo. As such we are using standard-dose TPA as the reference by which to compare. Based off the data from the ENCHANTED trial, we are unable to “prove” that reduced-dose TPA was non-inferior to standard dose. We are also unable to say that standard dose is superior. This off course is mostly just statistical wordplay as an critical look at the two groups will reveal they essentially had almost identical functional outcomes at 90-days. Hope that helps! Thanks again for writing!

I’m having trouble following the statistics. Could you please explain, “The number of patients alive and independent (mRS of 0,1,or 2) at 90 days was 62.4% and 63% in the low-dose and standard-dose groups respectively.” and “The authors found that the number of patients who were dead or disabled (a mRS of 2-6) at 90-days was 53.2% and 51.1% in the low-dose and standard-dose groups respectively. “. How can 62.4% be alive and well at 90 days and in that same group, 53.2% be disabled or dead? Thanks,

Sean

Paramedic

Uky

Sure Sean thanks for writing

its a essentially how you define “alive and independent” and “dead and disabled.” The most common dichotomous cutoff used to assess functional outcomes with the mRS is alive or independent at 90-days. This means the patient is determined to be a 0, 1 or 2 on the mRS scale. a 2 on the mRS is the highest number you can be given and still live without assistance. If you use this cutoff then at 90 days 62.4% and 63% in the low-dose and standard-dose groups respectively were alive and independent.

The authors used the number of patients who were “dead or disabled” at 90-days as their primary outcome. Not a typical way of examining function at 90-days but reasonable when considering the question they were asking. To be considered dead or disabled you had to have a 90-day mRS of 2,3,4,5, or 6. Because an mRS of 2 is included in both these dichotomous outcomes, they are not just opposite perspectives of the same outcome. As such the number of patients in each group will not total 100%. The technical antipode of “dead and disabled” is a mRS of 0,or 1.

Hope this helps!

[…] Rory Spiegel at EM Nerd: The Case of the Non-Inferior Inferiority Continues […]

[…] [Further reading] EMNerd (EMCrit)The Case of the Non-inferior Inferiority Continues […]

[…] simply, comes down to how easily one interprets “non-inferiority” – a point made nicely by Rory Spiegel in his post – and, further, how one interprets these findings in a Bayesian sense. The prevailing […]

[…] EM Nerd: The Case of the Non-inferior Inferiority Continues […]