Efficacy of progesterone for moderate to severe traumatic brain injury: a meta-analysis of randomized clinical trials

Progesterone has been shown to have neuroprotective effects in multiple animal models of brain injury, whereas the efficacy and safety in patients with traumatic brain injury (TBI) remains contentious. Here, a total of seven randomized controlled trials (RCTs) with 2492 participants were included to perform this meta-analysis. Compared with placebo, there was no significant decrease to be found in the rate of death or vegetative state for patients with acute TBI (RR = 0.88, 95%CI = 0.70, 1.09, p = 0.24). Furthermore, progesterone was not associated with good recovery in comparison with placebo (RR = 1.00, 95%CI = 0.88, 1.14, p = 0.95). Together, our study suggested that progesterone did not improve outcomes over placebo in the treatment of acute TBI.

Scientific RepoRts | 5:13442 | DOi: 10.1038/srep13442 moderate-to-severe TBI (GCS ≤ 12) 13,14 . All female patients were excluded in one included study 12 . The primary characteristics and quality assessments of the included RCTs are summarized in Table 1 and 2, respectively.
Meta-analysis outcomes. Death or vegetative state. The meta-analysis of seven RCTs with a random-effects model demonstrated that progesterone did not significantly reduce the rate of death or vegetative state in patients with acute TBI between the two groups (RR = 0.88, 95%CI = 0.70, 1.09, p = 0.24, I 2 = 45%) ( Fig. 2A). Subgroup meta-analysis was performed (Table 3). Due to the limited number of available studies, meta-regression was not pursued further. A similar result was observed in patients with severe TBI (RR = 0.86, 95%CI = 0.68, 1.09, p = 0.20, I 2 = 48%) (Fig. 2B). The sensitivity analysis was performed to examine the influence of different models on the pooled estimates. There were  Good recovery. In total, five of the included RCTs had an assessment of good recovery (GOS = 5) at the end of follow-up. No significant heterogeneity was observed in TBI (I 2 = 0%). Compared with placebo, the combined data using a fixed-effects model did not show that progesterone significantly increased the rate of good recovery (RR = 1.00, 95%CI = 0.88, 1.14, p = 0.95) (Fig. 3A). There was also no evidence to indicate that progesterone could improve the outcome for a good recovery in severe TBI (RR = 1.04, 95%CI = 0.91, 1.19, p = 0.54, I 2 = 0%) (Fig. 3B).

Discussion
The pathophysiology of acute TBI is a complex, interwoven and multifactorial process, which includes primary and secondary injury 15,16 . TBI-induced secondary injury has been considered to be a potential  Table 2. Risk of bias of the articles included in the meta-analysis. target for therapeutic intervention involving reduction and prevention of inflammation, calcium flux, oxidative stress, necrosis, and apoptosis 17,18 . Based on the efficacy and safety in animal models, progesterone has been regarded to be a potent candidate for the treatment of TBI [19][20][21][22] . However, the relevant clinical trials of progesterone came to inconsistent conclusions 10,12,14 . The previous review of progesterone for the treatment of TBI included only three small-scale and low-quality studies 23 . In this current study, we selected 7 relevant RCTs including 2492 patients (progesterone: 1276 cases, placebo: 1216 cases) hospitalized for acute TBI to assess the efficacy of progesterone therapy on the Glasgow Outcome Scale (GOS) score and for adverse events. Some previous clinical studies demonstrated that progesterone was a neuroprotective agent and improved outcomes for patients with acute severe TBI 8,10 . However, we found no significant difference between the progesterone-treated group and the placebo group in the rate of death or vegetative state. Moreover, our results showed that progesterone was not associated with good recovery at the end of the follow-up period. To date, various drugs have been investigated in clinical trials, yet none has been   proven to reduce mortality significantly at the confirmatory stage [24][25][26][27] . The trauma of individual patients could not be controlled well in comparison with the animal model. The heterogeneity and variability of TBI may be one of the important reasons 14,28 . This classification scheme of patients may be relatively insensitive using the Glasgow Coma Scale (GCS) or the Glasgow Outcome Scale-Extended (GOSE) 28,29 . Some limitations must be noted in this present study. First, one included study excluded female patients as a result of side effects on the menstrual cycle 12 . Second, due to the lack of available data, we did not analyze other clinical outcomes except mortality and good recovery. It was unknown whether progesterone promoted the recovery of motor and sensory skills. Finally, the follow-up was short-term and varied across the studies. Thus, an appropriate dosage and a long-term follow-up may be necessary to further investigate the efficacy of progesterone in the treatment of acute TBI.
In conclusion, the pooled data did not support the idea that progesterone was superior to placebo in the treatment of acute TBI. Progesterone may be not effective in lowering the incidence of death or vegetative state in patients with acute TBI.

Methods
Search strategy. Our electronic search was conducted in PubMed, Embase, and the Cochrane Library databases until May 10, 2015. The core terms included "progesterone" and "head injury, " "traumatic brain injury, " "TBI, " "random, " and "random*". There was no language limitation. We also searched Google Scholar and checked the reference lists of the included studies to identify any additional eligible articles.
Inclusion criteria. Studies were included if they met the following criteria: (1) adults (older than 18 years) with a diagnosis of acute TBI, (2) progesterone compared with placebo (or no progesterone), and (3) randomized controlled trials. Duplicate articles, reviews, case reports, and studies without extractable data were excluded.
Data extraction and outcome measures. Two authors (CL and HQH) independently extracted the following data from each included study in the standard form: (1) study characteristics (author's name, date of publication, study design, sample size), (2) characteristics of participants (age and gender), (3) interventions (administration, duration, and dosage), and (4) outcomes (GOS and adverse events). Any discrepancies were discussed and resolved by the research team when necessary. The efficacy outcome was assessed with death or vegetative state (GOS = 1 or 2) and good recovery (GOS = 5) at the 6 months after TBI or end of the follow-up period. Adverse events included pneumonia and sepsis. Quality assessment. The eligible studies were evaluated according to the Cochrane Collaboration's tool 30 . The domains were as follow: selection bias (random method and allocation concealment), performance and detection bias (blinding of participants, personnel and outcome assessment), attrition bias (incomplete outcome data), and reporting bias (selective reporting).

Statistical analysis.
The data were analyzed with the Cochrane Review Manager 5.3 and STATA 11.0 software according to the preferred reporting items for systematic reviews and meta-analysis (PRISMA) statement 31,32 . Risk ratios (RR) were calculated and pooled with a 95% confidence interval (CI) for dichotomous variables. The heterogeneity was estimated using the I 2 test, which was considered to be low heterogeneity when I 2 ≤ 25%. A fixed-effects random effects model was used if the I 2 was ≤ 25%.