MISTIE Intracerebral Hemorrhage Intervention Shows Promise in New Statistical Analysis
Oct. 19, 2023 - Eden McCleskeyIn a second look at a high-profile trial that failed to meet primary endpoints, Houston Methodist researchers have found that patients who received a promising intracerebral hemorrhage treatment were substantially more likely to achieve at least moderate functional independence than those who didn't.
Using a Bayesian reanalysis, the Houston Methodist team showed that minimally-invasive surgery with thrombolysis for intracerebral hemorrhage (MISTIE) has a 68%-87% probability of increasing the proportion of patients who benefit from the intervention. No proven effective therapy currently exists following this devastating form of stroke.
"While intracerebral hemorrhage certainly remains a therapeutic challenge, these numbers represent a much more positive potential outlook than current statistics for the condition suggest," said Dr. Abdulaziz Bako, a post-doctoral fellow and first author of the analysis study. "We believe the new analysis has significant implications for clinicians and patient families, offering a valuable framework for shared decision-making where conventional treatment options are limited."
The new analysis, recently published in Neurology, provides more evidence that the therapy can help standardize care against intracerebral hemorrhage, which has the highest mortality and morbidity rate of all strokes. Original findings from the 2019 Phase III MISTIE trial reported a more modest benefit from the therapy.
"Spontaneous intracerebral hemorrhage, or a blood clot that travels to the brain, constitutes 10% of all strokes and 50% of stroke mortality," said Dr. Gavin Britz, director of the Houston Methodist Neurological Institute and a co-author of the analysis. "Roughly 40%-50% of those who experience it will die within 30 days, and less than a third will survive with any form of autonomy in their daily life."
MISTIE uses a catheter to decrease the clot size to 15 mL or less in the hopes of leaving patients with a modified Rankin Scale score of 0-3 — a positive functional outcome indicative of increased mobility, autonomy and quality of life.
"The initial trial used a frequentist approach which can only tell you if there's a less than 5% chance that the outcome was due to random chance," Dr. Bako said. "While this is useful information — and our analysis doesn't alter that particular result — we felt like a newer statistical methodology, such as the Bayesian model, might give patients and family members a better idea of the procedure's probability of success, rather than simply telling them that there are no options for them at this point."
The Bayesian statistical approach requires significant computational power, which is easier to come by today than in years past. It provides probabilities of potential treatment effects, in contrast to the binary cut-off of frequentist methods, allowing for a more nuanced understanding of the potential benefits of the MISTIE intervention.
The Bayesian method uses various prior probability distributions, or assumptions about the accuracy of the data set, to predict a range of estimated effects.
In the current study, the range of probability of the MISTIE intervention leading to a good 365-day modified Rankin score was 68% using a skeptical prior, 70% using a minimally informative prior, 73% using a data-derived prior and 87% using an enthusiastic prior.
The study represents a potential paradigm shift for researchers who feel like potential innovation is stymied by too heavy a reliance on the frequentist statistical method and P-values less than .05.
"One important message to come out of this is that neurology and neurosurgery researchers should really consider using a Bayesian approach for future trials, because very few of us are currently using it even though it offers a lot of benefits when it comes to measuring the continuous probability of observing a specific treatment effect," Dr. Bako said.