The n e w e ng l a n d j o u r na l of m e dic i n e

edi t or i a l s

Progesterone for Traumatic Brain Injury

Resisting the Sirens Song
Lee H. Schwamm, M.D.

The results of two randomized, controlled trials
of the neurosteroid progesterone in patients with
traumatic brain injury (TBI), now published in the
Journal,1,2 showed no benefit with respect to favor-
able functional outcome at 6 months, as assessed
by means of the Extended Glasgow Outcome
Scale (GOS-E), or several prespecified secondary
outcomes. Both trials selected and stratified
patients on the basis of a Glasgow Coma Scale
(GCS) score. The Study of a Neuroprotective
Agent, Progesterone, in Severe Traumatic Brain
Injury (SYNAPSE), conducted by Skolnick et al.,1
completed enrollment of 1195 participants with
severe TBI, whereas the Progesterone for Trau-
matic Brain Injury, Experimental Clinical Treat-
ment (PROTECT III) trial, conducted by Wright
et al.,2 which enrolled participants with moder-
ate-to-severe TBI, was halted for futility after
882 participants had undergone randomization.
By current standards, these phase 3 TBI trials
were exceptionally well designed and conducted.
However, these trials share much in common
with other failed phase 3 trials in clinical neuro-
protection, in which informative preclinical mod-
els and biomarkers are often lacking.3-5 Both
trials were guided by studies in animal models
and by the results of two small, phase 2 trials:
the PROTECT study in the United States,6 and a
study by Xiao et al. in China.7 Given the nega-
tive findings of the two phase 3 trials, the inves-
tigators appropriately call for a comprehensive
review of the TBI translational research strategy.
I would argue that the problem lies deeper and
reflects a fundamental bias in the current prac-
tice of scientific inquiry.
In a widely cited article, Ioannidis8 argued
that many published positive research findings
are probably false, owing to unrecognized bias
and the low odds of a true relationship existing
before the start of the research study. The risk
of false positive findings is greatest when the
effect size is small, the number of studies or
participants is small, the number of relation-
ships tested is large, or there is substantial pro-
fessional or financial investment in a specific
outcome. The efforts needed to secure funding
for clinical trials reinforce these biases.
Both phase 2 trials showed a consistent re-
duction in early mortality but only modest im-
provement in functional outcome. Despite this,
functional outcome was chosen as the end point
in the phase 3 trials. In the phase 2 PROTECT
study, the 25 participants with moderate TBI
had a significant benefit at 30 days after injury,
as assessed by means of the score on the origi-
nal Glasgow Outcome Scale (GOS) or an extend-
ed version, but this was one of four prespecified
outcome measures. When this outcome is reana-
lyzed post hoc by including all 100 participants
in the study, the statistical significance of the
clinical benefit disappears (P=0.42 by Fishers
exact test). In addition, in the trial registry filing
of the PROTECT study9 there was an expecta-
tion that there would be 50% with good out-
come in the placebo group, which sets a pretty
high bar for success in a subsequent phase 3
trial.
The Chinese phase 2 trial did not report
sample-size calculations. In its registration fil-
ings,10 it listed multiple primary efficacy out-
comes (i.e., GCS, GOS, and Functional Indepen-
dence Measure scores at 3 and 6 months),
several of which favored progesterone. But the
benefits in the phase 2 trials were very modest,

2522 n engl j med 371;26nejm.orgdecember 25, 2014

The New England Journal of Medicine

Downloaded from nejm.org on April 24, 2015. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.
 editorials

since significance (by Fishers exact test) disap-
pears if a single patient in the placebo group is
artificially reassigned from an unfavorable out-
come to a favorable outcome on the basis of the
group data provided. In other words, the modest
benefit observed in the sparsely available data
from phase 2 trials of progesterone in patients
with brain injury suggested low prestudy odds
of a true relationship existing. Nevertheless, inves-
As a neuroscience community, we must emu-
late Homers crew of the Odyssey by plugging
our ears with beeswax and steeling ourselves
against the temptations of the Sirens song if we
wish to navigate past the obstacles in our path,
emerge victorious, and proceed onward to new
adventures.
Disclosure forms provided by the author are available with the
full text of this article at NEJM.org.

tigators and funders were convinced enough by
these tantalizing data to initiate phase 3 trials.
With favorable-outcome rates of more than 50%
observed among the participants assigned to re-
ceive placebo, neither phase 3 trial had much
opportunity to show benefit.
Slower-than-expected enrollment, overly opti-
mistic effect sizes, better-than-expected perfor-
mance in the placebo groups, and sample-size
estimates that were based on one of multiple ef-
ficacy outcomes in small safety trials are com-
monly observed patterns in failed trials. The
opportunity cost of these failed studies is high.
If we are to break this cycle, we need a radical
change in the culture of investigation and its
funding. This includes the creation of collabora-
tive research networks such as those recently
implemented by the National Institute of Neuro-
logical Disorders and Stroke, more rigorous re-
porting and pooling of preclinical data, coordi-
nated and sequential exploratory phase 2 trials
that use standardized outcomes to replicate po-
tential findings, and phase 3 trials designed to
test well-vetted hypotheses.8 These changes may
result in more extensive and costlier early-phase
work and fewer reported false positive findings
owing to more early failures of replication, but
eventually greater numbers of successful phase 3
trials. Negative trial results, when they do occur,
will contribute more to medical progress because
they will have definitively answered a question
that was worth asking in the first place.
From the Department of Neurology, TeleStroke and Acute
Stroke Services, and Institute for Heart, Vascular, and Stroke
Care, Massachusetts General Hospital, and the Department of
Neurology, Harvard Medical School both in Boston.
This article was published on December 10, 2014, at NEJM.org.
1. Skolnick BE, Maas AI, Narayan RK, et al. A clinical trial of
progesterone for severe traumatic brain injury. N Engl J Med 2014;
371:2467-76.
2. Wright DW, Yeatts SD, Silbergleit R, et al. Very early admin-
istration of progesterone for acute traumatic brain injury. N Engl
J Med 2014;371:2457-66.
3. Derdeyn CP, Chimowitz MI, Lynn MJ, et al. Aggressive med-
ical treatment with or without stenting in high-risk patients
with intracranial artery stenosis (SAMMPRIS): the final results
of a randomised trial. Lancet 2014;383:333-41.
4. Mohr JP, Parides MK, Stapf C, et al. Medical management
with or without interventional therapy for unruptured brain arte-
riovenous malformations (ARUBA): a multicentre, non-blinded,
randomised trial. Lancet 2014;383:614-21.
5. Shuaib A, Lees KR, Lyden P, et al. NXY-059 for the treatment
of acute ischemic stroke. N Engl J Med 2007;357:562-71.
6. Wright DW, Kellermann AL, Hertzberg VS, et al. ProTECT:
a randomized clinical trial of progesterone for acute traumatic
brain injury. Ann Emerg Med 2007;49:391-402.
7. Xiao G, Wei J, Yan W, Wang W, Lu Z. Improved outcomes
from the administration of progesterone for patients with acute
severe traumatic brain injury: a randomized controlled trial. Crit
Care 2008;12:R61.
8. Ioannidis JP. Why most published research findings are
false. PLoS Med 2005;2(8):e124.
9. ClinicalTrials.gov Registry. Trial ID NCT00048646: pro-
gesterone treatment of blunt traumatic brain injury (http://
clinicaltrials.gov/ct2/show/NCT00048646).
10. Australian New Zealand Clinical Trials Registry. Trial ID
ACTRN12607000545460: evaluating the effects of progester-
one on neurologic outcome of the patients with severe trau-
matic brain injury (http://www.anzctr.org.au/trial/registration/
trialreview.aspx?actrn=12607000545460).
DOI: 10.1056/NEJMe1412951
Copyright 2014 Massachusetts Medical Society.

Clone Wars The Emergence of Neoplastic Blood-Cell Clones

with Aging
Janis L. Abkowitz, M.D.

Blood cells originate from hematopoietic stem cur with DNA replication during cell division,
cells (HSCs). HSCs are infrequent (<2 per 108 bone HSC quiescence maximizes genomic stability and
marrow cells, or about 11,000 to 22,000 per per- thus safety.
son) and rarely divide.1,2 Because mutations oc- Mutations, however, can and do occur. Most

n engl j med 371;26nejm.orgdecember 25, 2014 2523

The New England Journal of Medicine
Downloaded from nejm.org on April 24, 2015. For personal use only. No other uses without permission.
Copyright 2014 Massachusetts Medical Society. All rights reserved.