Cautionary Tale Associated in Trial With Rare Error in Cox Regression
InflaRx CEO Niels Riedemann, MD, discusses the issue that arose and cost the company some of its data for its investigational therapy being studied for SARS-CoV-2-induced acute respiratory distress syndrome (ARDS).
The German company InflaRx has developed an investigational therapy, vilobelimab, which is being studied for SARS-CoV-2-induced acute respiratory distress syndrome (ARDS). Vilobelimab is a first-in-class monoclonal antihuman complement factor C5a antibody. In one of its trials, the FDA recommended that the company use a site stratification adjustment.
“We had the FDA recommending us to do that, even though we had run the trial, started the trial and planned the trial without that site stratification adjustment,” InflaRx CEO Niels Riedemann, MD, said. “So we stratified per site, but we didn't adjust for that in the Cox regression model.”
Doing it this way created an error that led to the loss of almost 17% of the outcome data, affecting the trial's statistical integrity. Study sites that contained only 1 enrolled patient or no deaths were discounted entirely.
Read more: Monoclonal Antibody Reduces Mortality for SARS-CoV-2-Induced Acute Respiratory Distress Syndrome
In commentary published late last year in Trials, InflaRx’s experience is used as a case study for what can happen. Riedemann says companies just need to be more cautious when using this methodology for such trials.
“When you take that into consideration in your trial, meaning Cox regression, so you adjust for it, and it's called side stratified Cox regression, then that problem can occur,” he said. “If you still want to use it, you would need to up your trial planning. You need to do your statistical analysis on different planning, on a different power, and add more patients.”
