Is Progesterone Useful in Traumatic Brain Injury?

Authors: Skolnick BE et al., N Engl J Med 2014 Dec 10;

Two well-conducted studies show that it is not.

Traumatic brain injury (TBI) is a major cause of mortality and morbidity with large indirect and direct costs to society. TBI was the first neurological disease for which progesterone administration was studied. Many animal models have been used to elucidate the mechanism by which progesterone offers neuroprotection following TBI. Several of the studies demonstrated that progesterone reduces post-injury cerebral edema.

Now, two randomized, controlled trials have examined the effect of progesterone in patients with moderate-to-severe TBI. Both trials used a Glasgow Coma Scale (GCS) for the selection and stratification of patients.
Skolnick and colleagues enrolled 1195 participants (age range, 16–70) with severe TBI (GCS score less than 8). Patients were randomized to receive intravenous progesterone or placebo. GCS scores at 6 months (the primary outcome) were not significantly different between the groups (odds ratio, 0.96).

Wright and colleagues enrolled patients with moderate-to-severe TBI. The median age of the patients was 35 years and the mean injury severity score was 24.4 on a scale of 0 to 75. Patients were randomly assigned to progesterone or placebo. The primary outcome was improvement on the Extended GCS at 6 months, defined by initial GCS score. After randomization of 882 participants, the trial was halted for futility.

Comment
These very well designed and conducted phase III trials definitively show that progesterone is of no benefit with respect to favorable functional outcome at 6 months. These findings contrast with other, small trials which showed possible signs of benefit with the use of progesterone. There is increasing concern that in modern research, initial findings of a treatment’s benefit may prove to be false. This should not be surprising, as it can be shown that many claimed research findings are false. Large studies are needed. In addition, meta-analyses with low bias probably can help (PLoS Medicine 2005; 2:e124).