Pregnenolone treatment reduces severity of negative symptoms in recent-onset schizophrenia: An 8-week, double-blind, randomized add-on two-center trial




Management of recent-onset schizophrenia (SZ) and schizoaffective disorder (SA) is challenging owing to frequent insufficient response to antipsychotic agents. This study aimed to test the efficacy and safety of the neurosteroid pregnenolone in patients with recent-onset SZ/SA.


Sixty out- and inpatients who met DSM-IV criteria for SZ/SA, with suboptimal response to antipsychotics were recruited for an 8-week, double-blind, randomized, placebo-controlled, two-center add-on trial, that was conducted between 2008 and 2011. Participants were randomized to receive either pregnenolone (50 mg/day) or placebo added on to antipsychotic medications. The primary outcome measures were the Positive and Negative Symptoms Scale and the Assessment of Negative Symptoms scores. Secondary outcomes included assessments of functioning, and side-effects.


Analysis was by linear mixed model. Fifty-two participants (86.7%) completed the trial. Compared to placebo, adjunctive pregnenolone significantly reduced Positive and Negative Symptoms Scale negative symptom scores with moderate effect sizes (d = 0.79). Significant improvement was observed in weeks 6 and 8 of pregnenolone therapy among patients who were not treated with concomitant mood stabilizers (arms × visit × mood stabilizers; P = 0.010). Likewise, pregnenolone significantly reduced Assessment of Negative Symptoms scores compared to placebo (d = 0.57), especially on blunted affect, avolition and anhedonia domain scores. Other symptoms, functioning, and side-effects were not significantly affected by adjunctive pregnenolone. Antipsychotic agents, benzodiazepines and sex did not associate with pregnenolone augmentation. Pregnenolone was well tolerated.


Thus, add-on pregnenolone reduces the severity of negative symptoms in recent-onset schizophrenia and schizoaffective disorder, especially among patients who are not treated with concomitant mood stabilizers. Further studies are warranted.

Although antipsychotics have proven effective in ameliorating psychotic symptoms, in a substantial proportion of recent-onset schizophrenia (SZ) and schizoaffective (SA) disorder patients, the clinical response remains insufficient. When response to antipsychotics is inadequate, augmentation strategies are often implemented.

Pregnenolone (PREG), pregnenolone sulfate (PREGS), dehydroepiandrosterone (DHEA) and its sulfate ester (DHEAS), progesterone and androstenedione are called ‘neuroactive steroids’ and ‘neurosteroids’ as they are produced in situ in the brain, adrenal glands and gonads. PREG and other neurosteroids act as potential signaling molecules for neocortical organization during brain development and regulate neuronal function by affecting neuronal excitability through prominent modulatory effects on the γ-aminobutyric acid type A (GABAA), N-methyl-D-aspartate (NMDA), sigma-1,[1, 2] cholinergic[3] and dopamine systems.[4] They regulate the growth of neurons and cerebral brain-derived neurotrophic factor levels, enhance myelination and synaptogenesis, and exhibit neuroprotective properties.[5]

There is evidence that PREG and its metabolites may be involved in the pathophysiology of schizophrenia, mood disorders, dementia and substance abuse.[6-8] Clinical studies demonstrated low circulating levels of PREG in the elderly, including those with dementia,[6] in individuals with major depression,[7] anxiety disorder,[9] and in chronic medicated schizophrenia patients.[10]

Two clinical trials with add-on PREG in chronic SZ/SA were published. Marx et al.[11] claimed that nine patients receiving PREG (fixed escalating doses to 500 mg/day) added on to second-generation antipsychotics (SGA) demonstrated significantly (P = 0.048) greater improvement in the Scale for the Assessment of Negative Symptoms (SANS) scores compared with nine patients who received placebo. This pilot study was based on an extremely small sample and had other limitations.[12]

Another study investigated adjunctive ‘low-dose’ and ‘high-dose’ PREG (30 mg/day and 200 mg/day, respectively) added to ongoing antipsychotics in the treatment of chronic SZ/SA in an 8-week, controlled, double-blind, randomized, parallel-group trial.[13] Fifty-eight patients were randomized, 44 patients completed the trial. Compared with placebo, PREG-30 administration was associated with significant reduction in positive symptom scores, extrapyramidal symptoms, improvement in attention, and working memory performance, whereas subjects treated with PREG-200 did not differ on outcome variable scores. PREG was well tolerated. Circulating PREG was found to be significantly higher among the patients treated by PREG compared to the placebo group. More detailed reviews of PREG's metabolism, mode of action, and these pilot trials were recently published.[12, 14]

We conducted a randomized, double-blind, placebo-controlled study among a sample of 60 patients with recent-onset SZ/SA. Add-on PREG (50 mg/day) for 8 weeks demonstrated significant amelioration of the visual attention deficit measured with the Matching to Sample Visual Search task compared to the placebo group in recent-onset SZ/SA.[15] This report aimed to test the efficacy and safety of the neurosteroid PREG in patients with DSM-IV recent-onset SZ/SA. We hypothesized that add-on PREG compared to placebo would diminish persistent clinical symptoms in patients with recent-onset psychotic disorders.


Study design

We undertook an 8-week, randomized, double-blind, placebo-controlled study that was initiated by the investigators and was conducted between February 2008 and January 2011 independent of any commercial entities. Our objective was to determine whether PREG ameliorates severity of symptoms among patients with SZ/SA. Prior to starting the study, all subjects provided written informed consent after receiving a full explanation regarding the nature of the study, and its potential risks and benefits. The Institutional Review Boards of the two participating centers and the national Ministry of Health Ethical Review Board approved the study.


Participants were recruited from the inpatient and outpatient services of two large state referral hospitals: Sha'ar Menashe Mental Health Center and Tirat Carmel Mental Health Center, which are affiliated to the Rappaport Faculty of Medicine, Technion – Israel Institute of Technology (Haifa). Recruitment was initiated in February 2008, ended in November 2010, and the last patient completed treatment in January 2011. For inclusion in the study, patients were required to be between the ages of 18 and 40 years (inclusive) and meet criteria of the DSM-IV for a diagnosis of paranoid schizophrenia or schizoaffective disorder and suboptimal response to previous treatment, which was defined by two criteria used in the clinical trials:[16] (i) persistent positive symptoms (hallucinations, delusions, or marked thought disorder) after at least 6 contiguous weeks of antipsychotic treatment; and (ii) a poor level of functioning over the past 2 years, defined by the lack of competitive employment or enrollment in an academic or vocational program. We use the term ‘suboptimal response to treatment’ to highlight that our criteria are different from ‘treatment resistance’.[17] In addition, patients were required to have: (i) duration of illness less than 5 years since onset of first psychotic episode, which was established based on the first visit to a psychiatrist; (ii) a score of at least 3 on the Clinical Global Impression Scale (CGI-S)[18] at entrance to the study; (iii) at least 2 weeks of ongoing treatment with current antipsychotic agents before the pre-treatment stabilization period; (iv) stable symptoms throughout the 2-week pre-treatment stabilization period with no more than a 20% change in the Positive and Negative Syndrome Scale (PANSS)[19] total score; (v) no change in anticholinergic, benzodiazepine, or mood stabilizer medications during the pre-treatment stabilization period, as well as no change in antipsychotics, anticholinergics, benzodiazepines, or mood stabilizers during the 8-week duration of the study; and (vi) ability to participate fully in the informed consent process, or have a legal guardian able to participate in the informed consent process.

Major exclusion criteria included: an unstable medical condition, any significant medical or neurological illnesses, pregnancy, and treatment with any steroid or hormonal supplement (e.g. estrogen). The absence of medical or neurologic illnesses was verified by means of a routine laboratory investigation that included blood cell count with differential, liver function tests, glucose and cholesterol levels, physical and neurologic examinations, reports of the patient's family physician, and medical records. It was forbidden to add any other psychoactive medication before entry or during the entire study period.

All participants had received an antipsychotic medication for at least 3 months. At baseline, 24 patients were treated with first-generation antipsychotics (FGA; chlorpromazine, haloperidol, haloperidol decanoate, perphenazine, zuclopenthixol, zuclopenthixol decanoate, fluphenazine decanoate); 23 patients were treated with SGA (risperidone, olanzapine, quetiapine, ziprasidone, clozapine); and 13 patients received both types of antipsychotic medications (combined therapy [COMB]). Chlorpromazine-equivalent (CPZ) doses were calculated based on published data.[20, 21] The mean CPZ (±SD) in the FGA group was 443 (±101) mg/day, in the SGA group, 392 (±84) mg/day, and in the COMB therapy group 493 (±109) mg/day. Besides the antipsychotic medications, the patients continued to take mood stabilizers (n = 14; valproate, carbamazepine, lamotrigine), benzodiazepines (n = 19), anti-Parkinson agents (n = 33), and antidepressants (n = 3) that they received prior to study recruitment.

Study procedure and treatment

At the initial screening visit, a thorough clinical and psychiatric examination was performed on patients who met entry criteria, and severity of the disorder and symptoms were evaluated with the CGI-S and the Positive and Negative Syndrome Scale (PANSS).[19] Blood was collected for routine laboratory tests. Senior psychiatrists (M.S.R., and A.K.) at each site enrolled and established patients' diagnoses according to DSM-IV criteria.

Patients who were clinically stable for the 2 weeks of the lead-in phase (with no more than a 20% change in PANSS total score), were randomized into two groups: patients who received PREG and patients who received a placebo, each for 8 weeks in a double-blind manner. The randomization procedure was performed using Random Allocation Software (Version 1.0, May 2004; available at: We used a low-PREG dose (50 mg/day; Biosynergy, Boise, ID, USA), taking into account findings from a previous study of the use of PREG (30 mg/day) in schizophrenia, which showed safety and efficacy at a low dose of PREG (12). Follow-up visits for psychiatric and safety assessments were conducted at weeks 2, 4, 6, and 8.

Outcome measures

Psychiatrists who were blind to the patients' medication performed all outcome measures. The CGI-S score was used for screening at visit assessments. The primary efficacy variables were the change from baseline of the PANSS and the SANS22 scores to the end of the clinical trial. Other measures were the Global Assessment of Functioning (GAF)[23] and the Extrapyramidal Symptom Rating Scale.[24]

Safety evaluations included solicited adverse event reporting, tests of neurological status, routine laboratory tests, electrocardiogram monitoring, and recording of vital signs and bodyweight. These evaluations were repeated every 2 weeks for the duration of the study in addition to the regular above-described clinical ratings. Raters were trained before the study to produce acceptable levels of inter-rater reliability, estimated by intraclass correlation coefficient (ICC), for the primary diagnosis, CGI-S, PANSS, SANS, GAF, and Extrapyramidal Symptom Rating Scale (ESRS) (ICC = 0.92, 0.83, 0.89, 0.85, and 0.86, respectively).

Statistical analysis

Statistical analysis was performed using Number Cruncher Statistical Systems (Kaysville, UT, USA),[25] and by fitting a linear mixed model, also known as a hierarchical linear model.[26] The linear mixed model we used included the two treatment arm effects (PREG vs placebo), time (five visits), and DSM-IV diagnosis (295.3; 295.7). The visit was included as a categorical variable (weeks 0, 2, 4, 6, 8), because it reflects the changes along time better than a linear trend. In addition, a possible effect of the covariates, such as concomitant treatment (mood stabilizers, benzodiazepines, anti-Parkinson agents, and antidepressants), antipsychotic treatment (FGA, SGA, and COMB), side-effects, and sex on outcome variables over time, was also analyzed.

Post-hoc analysis was carried out in cases of significant outcomes, using the Tukey–Kramer method. An effect size (Cohen's d) was calculated for raw scores of outcome variables at the last examination (week 8) for between-group comparisons by the estimated pooled standard deviation (SD). A small effect size was defined as d ≥ 0.2, a moderate effect size as d ≥ 0.5, and a large effect size as d ≥ 0.8.[27] The Bonferroni correction was also applied. The data were expressed as the mean ± SD or ± SE. Continuous variables were compared using the two-tailed t-test, or the Wilcoxon signed-rank test (z) for assessing the difference in medians. Differences in the frequency of categorical variables were examined with the χ2-test. For all analyses, the level of statistical significance was defined as an alpha less than 0.05.


Characteristics of the experimental groups

Of 78 screened subjects with ongoing residual symptoms, 18 patients did not enter the study. Two subjects were excluded due to organic brain damage, five patients due to comorbidity with substance abuse, four patients had serious medical illnesses, and seven patients did not agree to participate.

Sixty patients with a suboptimal response to antipsychotic agents were randomized. Baseline characteristics of the population sampled and patients completing the clinical trial are presented in Table 1.

Table 1. Baseline characteristics of the population being sampled and patients completing the clinical trial
Baseline values of variablesRandomized sample (n = 60)Patients completing the clinical trial
Pregnenolone (n = 25)Placebo (n = 27)Significance
MeanSDMeanSDMeanSDt (z)P
  1. aFGA: chlorpromazine, haloperidol, haloperidol decanoate, perphenazine, zuclopenthixol, zuclopenthixol decanoate, fluphenazine decanoate; SGA: risperidone, olanzapine, quetiapine, ziprasidone, clozapine; COMB: FGA and SGA.
  2. bOf 29 patients treated with pregnenolone, eight patients were treated with mood stabilizers (valproate, carbamazepine, lamotrigine), 10 patients with benzodiazepines, 17 patients with anti-Parkinson agents and one patient with antidepressants; of 31 patients treated with placebo, six, nine, 16, and two patients were treated with mood stabilizers, benzodiazepines, anti-Parkinson agents and antidepressants, respectively.
  3. CGI-S, Clinical Global Impression Scale; COMB, combined therapy; ESRS, Extrapyramidal Symptom Rating Scale; FGA, first-generation antipsychotic agents; GAF, Global Assessment of Functioning; PANSS, Positive and Negative Syndrome Scale; SANS, Scale for the Assessment of Negative Symptoms; SGA, second-generation antipsychotic agents.
Age (years)27.25.426.
Education (years)
Age of onset (years)
Number of admissions2.
Duration of illness (years)
CGI-S score3.
PANSS total score59.711.358.211.963.710.50.80.45
Negative scale score15.43.515.63.316.
Positive scale score13.
General scale score30.811.130.26.731.
SANS total score34.311.
GAF score60.66.059.312.558.
ESRS score5.
Sex (male/female)52/822/323/4χ2 = 0.08, d.f. = 1, P = 0.77
Married13.3%12.0%14.8%χ2 = 1.3, d.f. = 1, P = 0.76
Diagnosis (DSM-IV):       
Schizophrenia, paranoid type3863.3%1560.0%1970.4%χ2 = 0.62, d.f. = 1, P = 0.43
Schizoaffective disorders2236.7%1040.0%829.6%
Antipsychotic drugs: FGA/SGA/COMBab24/23/1310/8/79/13/5χ2 = 1.5, d.f. = 2, P = 0.47

Comparisons between PREG (n = 29) and placebo (n = 31) groups for the 2-week lead-in phase and at baseline assessments of PANSS and GAF scale scores did not reach significant levels of difference between the two treatment arms (all P's >0.05, data not shown).

Of the eight subjects who dropped out, four patients received PREG, and four patients received placebo. More specifically, three patients from the PREG group and one patient from the placebo group dropped out during the first 2 weeks of the study. Another four patients (one patient from the PREG group and three patients from the placebo group) dropped out between the 4th and 6th weeks. Reasons for dropping out were not related to PREG or placebo administration: three patients withdrew consent, two patients had a change in antipsychotic drugs, one patient dropped out because of clinical deterioration, one patient because of an extrapyramidal adverse event, and one more due to noncompliance. There were no notable imbalances between the two treatment groups (PREG = 25, and placebo = 27) in baseline characteristics, nor between the 52 completers and the eight non-completers (Table 1).

Pregnenolone reduces negative symptom scores

As shown in Table 2, among the PREG group, a significant reduction in scores was found in the PANSS negative scale (P = 0.0017) in comparison to placebo. Mean reduction on the PANSS negative subscale was −4.6 scores among PREG subjects compared with −2.5 scores among placebo subjects with about medium effect size (Cohen's d = 0.79) (Fig. 1). When between-group differences on PANSS negative scale scores were adjusted for ESRS and PANSS positive scale scores they remained significant (F = 9.4, P = 0.002; and F = 5.7, P = 0.017, respectively). In the PANSS item analysis, patients randomized to PREG demonstrated significantly greater improvements compared to placebo group in emotional withdrawal (N2; F1,288 = 7.8, P = 0.005), poor rapport (N3; F1,288 = 9.8, P = 0.002, P < 0.001), passive/apathetic social withdrawal (N4; F1,288 = 12.8, P < 0.001, P < 0.001), and stereotyped thinking (N7; F1,288 = 17.1, P < 0.001, P = 0.002) scores.

Figure 1.

Mean Positive and Negative Syndrome Scale scores for (image) placebo and (image) pregnenolone over time.

Table 2. Changes in outcome measures between baseline and end-point
VariablesPregnenolone (n = 29)Placebo (n = 31)Significance
Changes between baseline and end-point95%CI for mean differencesChanges between baseline and end-point95%CI for mean differencesTreatment conditiona (d.f. = 1 288)Timeb (d.f. = 4 288)
  1. aPregnenolone versus placebo.
  2. bWeek 0, 2, 4, 6, and 8.
  3. cThe Bonferroni correction for the six outcome measures was 0.008 (P = 0.05/6), thus between-group differences on PANSS negative scale scores (P = 0.0017) remained significant.
  4. CI, confidence interval; GAF, Global Assessment of Functioning; PANSS, Positive and Negative Syndrome Scale; SANS, Scale for the Assessment of Negative Symptoms.
PANSS total score−16.26.8−19.1−13.4−12.38.2−20.5−<0.0001
Negative subscale−4.62.1−5.8−2.9−2.53.2−3.8−2.310.10.0017c7.9<0.0001
Positive subscale−4.32.7−5.4−3.1−4.42.7−5.6−<0.0001
General subscale−8.74.2−9.9−6.4−9.05.0−11.0−<0.0001
SANS, total score−12.211.2−17.0−7.4−9.111.6−13.7−<0.001
GAF score8.<0.001

Patients randomized to PREG demonstrated significant reduction in SANS total scores (−12.2 ± 11.2) compared with patients who received placebo (−9.1 ± 11.6; P = 0.003) with moderate effect sizes (Cohen's d = 0.57; Table 2). PREG group patients significantly improved on three SANS domain scores compared to the placebo group: blunted affect (F1,288 = 5.6, P = 0.019), avolition (F1,288 = 4.7, P = 0.032), anhedonia (F1,288 = 13.7, P < 0.001), but not on alogia (F1,288 = 3.9, P = 0.051) or inattention (F1,288 = 2.8, P = 0.095) scores.

Other outcome assessments (PANSS total, positive scale, general psychopathology scale, and GAF scores) did not show significant differences following treatment with PREG compared with placebo. Table 2 also shows that both PREG and placebo augmentation resulted in statistically significant decreases from baseline visit to end-point of the study on all outcome measures (‘time’, all P-values < 0.001). An interaction of arms × time for PANSS and GAF scores did not reach a significant level (all P-values > 0.05).


Table 3 presents analysis of interaction between the reduction in the PANSS negative scale scores and the patient's treatment with mood stabilizers (arms × time × mood stabilizers; F = 3.35, P = 0.010). For 14 patients of both groups who received mood stabilizers there was no significant difference between the two treatment arms, however there was significant improvement in the PREG group compared to the placebo group that was not concomitantly treated with mood stabilizers.

Table 3. Mixed model estimates of PANSS negative scale scores and effects of the concomitant treatment with mood stabilizers
Effectd.f.Den d.f.F valuePr > F
  1. aWeek 2 (4, 6, and 8) means week 0 (baseline) minus week 2 (4, 6, and 8).
  2. Adj P, standard multiplicity-adjusted P-values; Den d.f., number of degrees of freedom associated with model errors; Pr > F, P-value associated with the F statistic of a given effect and test statistic; Pr > |t|, P-value associated with the t statistic of a given effect and test statistic.
Arms (pregnenolone vs placebo)162.453.200.078
Time (visit: week 0, 2, 4, 6, and 8)4196.319.57<0.0001
Treatment arms × Time (visit)4196.41.310.267
Mood stabilizers1169.20.100.752
Treatment arms × Mood stabilizers1168.72.730.100
Time × Mood stabilizers4196.43.310.011
Treatment arms × Time × Mood stabilizers4196.43.350.010
Adjustment for Multiplicity: Simulated
Treatment armsTime (week)aEstimateErrord.f.t-valuePr > |t|Adj P
Patients not treated with mood stabilizers
Pregnenolone versus placebo2−1.161.30198.5−0.890.3730.950
Patients treated with mood stabilizers
Pregnenolone versus placebo21.360.76198.81.810.0720.382

Accordingly, the change versus baseline in weeks 6 and 8 was different between arms for patients not treated with mood stabilizers (P = 0.005, adj. P = 0.038, and P = 0.008, adj. P = 0.050, respectively). No significant main effect of the DSM-IV diagnosis, sex, concomitant treatment with benzodiazepines, and anti-Parkinson agents, type of antipsychotics, and interactions, such as arms by type of antipsychotics and by visits on PANSS subscales, SANS, GAF, and ESRS ratings, was observed (all P-values > 0.05).

Tolerability and safety

Before starting this trial, 17 patients in the PREG group and 16 patients in the placebo group were treated with anti-Parkinson agents. One patient (treated with haloperidol decanoate and PREG) was dropped from the study because of an extrapyramidal adverse event. No other new treatment-related adverse events occurred in either group. No differences between the two treatment arms were noted on ESRS scores (F1,260 = 0.66, P = 0.42). Both PREG and placebo augmentation resulted in statistically significant decreases from baseline to end-point of the study on the ESRS (F4,260 = 15.1, P < 0.001), without significant interaction: treatment arms × time (F4,260 = 0.64, P = 0.63). There were no clinically significant changes in vital signs, electrocardiograms, or clinical laboratory variables associated with treatment. Thus, the administration of PREG was well tolerated.


This is the first randomized, double-blind, placebo-controlled add-on trial of PREG 50 mg/day to ongoing antipsychotic treatment in subjects suffering from recent-onset SZ/SA with suboptimal response to antipsychotics. Several comments and possible explanations for the obtained findings need to be considered.

First, add-on PREG 50 mg/day for 8 weeks significantly reduced PANSS negative scale and SANS total scores in comparison to placebo with moderate effect sizes (Cohen d = 0.79 and d = 0.57, respectively). The responding symptoms included PANSS emotional and apathetic social withdrawal, poor rapport, and stereotyped thinking (N2, N3, N4, and N7) and SANS blunted affect, avolition and anhedonia domain scores. Given the limited effects of PANSS positive scale and ESRS scores on the reduction of the negative scale scores in PREG-treated patients, it is possible to hypothesize that adjunctive PREG might reduce primary negative symptoms.

Second, adjunctive PREG reduced the severity of PANSS negative symptoms in weeks 6 and 8 of therapy, especially among patients who were not concomitantly treated with mood stabilizers. While somewhat speculative, it may be suggested that PREG augmentation has an interaction with mood stabilizer agents that are administered in both SZ, and SA disorder. This assumption seems to be supported by preclinical findings.[28, 29] Furthermore, neurosteroids and antiepileptic agents may exert an effect via GABA and glutamate. For instance, PREG modulates inhibitory synaptic transmission by enhancing GABAA receptor desensitization.[30] Much more research is needed to test these assumptions.

Third, antipsychotic agents, DSM-IV diagnosis, side-effects, and sex did not associate significantly with effect of PREG augmentation. General functioning and extrapyramidal side-effects were not affected by adjunctive PREG in our sample. The observed differences in the findings between our previous 8-week trial with PREG (30 mg/day[12]) and the present sample, particularly with regard to negative symptoms, may be explained by significant differences in background characteristics (age, predominance of male subjects, and illness duration: 15.1 vs 2.5 years), sample sizes (14 vs 25), and severity of clinical condition (at baseline, PANSS subscale scores were substantially lower in the present sample compared to our previous trial). There is clear evidence of age-related changes in circulating neurosteroids, such as DHEA, and DHEAS, PREG, and PREGS.[31, 32] The present study suggests that PREG did not induce amelioration of extrapyramidal side-effects, measured with ESRS, among recent-onset SZ/SA disorder patients contrary to findings from a study of chronic SZ/SA.[12] This discrepancy may be explained, at least in part, by differences in the populations studied.

Fourth, although one new extrapyramidal adverse effect was observed, PREG treatment was generally well tolerated, confirming previous trials with PREG.[11, 12]

Although it remains poorly understood at this stage, the modulatory effect of PREG and its sulfate on GABAA, NMDA, sigma-1, cholinergic, and dopamine systems[1-4] may account for PREG's influence on clinical improvement. Another possibility is that PREG regulates the growth of neurons and cerebral brain-derived neurotrophic factor (BDNF) levels, enhances the myelination and synaptogenesis in the CNS, and demonstrates neuroprotective properties.[33, 34] The potential effect of adjunctive PREG on negative symptoms may be of considerable clinical interest, and may also expand our knowledge of the mechanisms involved in the development of negative symptoms. Further testing of this hypothesis is warranted.

The present findings should be accepted cautiously. The relatively modest sample size of this trial should be carefully examined before generalizing the findings to other groups. Limitations of this study also include the relatively short duration of the study. Long-term, large-scale studies are required to obtain greater statistical significance and more confident clinical generalization. In addition, it would be important in larger sample sizes to investigate whether any interaction exists between PREG and any specific medication and to compare responses to PREG in those receiving typical and atypical antipsychotics within the context of a standardized medication regimen.

On the basis of these results, we conclude that adjunctive pregnenolone significantly reduces PANSS negative scale and SANS domain scores in comparison to placebo among patients with recent-onset SZ and SA disorders. As we have some concerns about clinical versus statistical significance, larger studies should be conducted to establish the potential utility of the treatment.


This study was supported by a generous grant from The Stanley Foundation (#08TGF-1189). The authors are grateful to Dr Tatiana Umansky from the Statistics Laboratory of the Technion, headed by Professor Ayala Cohen, and to Rena Kurs for editing this manuscript. None of the authors have any conflicts of interest to report. The funding source did not influence the choice of journal for publication of the results of the study, and was not involved in analysis of results or writing the manuscript. Trial Registration: Identifier: NCT00847600.