This systematic review of the controlled studies that reported the effect of pre-treatment DHEA on IVF outcome in women with diminished ovarian reserve suggests that DHEA does not improve the quantitative ovarian response and pregnancy outcome. While the ovarian response as defined by the number of oocytes retrieved was significantly lower, the clinical pregnancy rate was marginally superior with a relative risk of 1.87 (95% CI 0.96-3.64; P=0.07) in the DHEA group. The miscarriage rate was similar between the DHEA and control groups on meta-analysis of the two reported controlled studies. This finding, however, is based on few data as there was only one study  which reported live birth rate, which was similar between the DHEA and control groups when only one cycle per participant was considered for analysis.
While there are several self-controlled case series on reported significant benefits in terms of ovarian response and pregnancy rates with the use of pre-treatment DHEA adjuvant during IVF, this systematic review of controlled studies failed to demonstrate such a benefit. However, while noting the trend of a positive effect of DHEA on the pregnancy rate in this review, the lack of a significant difference may be because of a small sample size with the overall number of participants that are included in the meta-analysis is only 198. In the study of Barad et al. 2007 included in this review, the authors have also reported spontaneous pregnancy (n=6/16) and pregnancy following IUI (6/9), which occurred during the three to four months waiting time of pre-IVF DHEA adjuvant treatment . When these data were included in the meta-analysis, there was a significantly increased pregnancy rates in the DHEA arm over the controls (RR 2.46 95% CI 1.35, 4.48; P=0.003). Since our objective was to investigate the effect of DHEA in IVF cycle, we have included only the IVF population in the primary analysis.
At present, there is only a single small randomised controlled trial by Wiser et al. 2010 reported in literature . The small sample size in this study resulted in only a minimal effect toward the result of the meta-analysis (Figures 1 and 2). Furthermore, many limitations and weaknesses of Wiser’s study have been criticised. First, there was no priori sample size estimation in the study. The authors included two cycles with varying duration of DHEA adjuvant treatment (7 – 18 weeks) and the authors continued the trial until a significant difference in cumulative live birth rate between the study and control groups was achieved. There was no difference in the live birth rate between the study and control groups following their first cycle of IVF with the mean duration of DHEA therapy in the study group was for only about 8.5 weeks. Secondly, both patients and health care providers were not blinded in this study therefore bias could occur, or patients in the control group might have sought over-the-counter medication for themselves. And finally, one letter to the editor expressed concern over the statistical analysis in this study suggesting that because the author had included two consecutive IVF cycle data from each group, Kaplan-Meier survival analysis should have been used rather than Fisher’s exact test, which was originally reported .
Most data on the favourable effect of DHEA adjuvant supplementation currently come from the study group led by Barad and Gleicher at the Centre of Human Reproduction, New York. They have published a series of self-controlled studies and retrospective case control analyses in which the benefits of DHEA are summarized as follows; i) increased oocyte yield  ii) higher fertilization rate  iii) improved embryo morphological grading  iv) increased pregnancy rate , v) lower miscarriage rate when compared to the national IVF statistics , and vi) lower aneuploidy rate . We have included two of their suitable controlled studies in our meta-analysis (Table 1). The authors also suggested that the benefit of DHEA treatment would be most effective if it is supplemented for at least 3-4 months, which is equal to the time needed for the early growing follicles to reach the gonadotropin-responsive stage . Therefore, they believed that DHEA acts in both ovarian recruitment and early folliculogenesis. If this hypothesis is true, the duration of DHEA treatment is possibly a key factor in effecting a favourable change in ovarian response and pregnancy rates following IVF.
The data from the self-controlled studies should be interpreted with caution because of potential bias. In one review by Urman and Yakin, the authors retrospectively analysed IVF outcome in the second cycle of 801 patients who have history of poor ovarian response (production of ≤ 4 oocytes) in the first IVF cycle treatment in their centre. It was found that almost 40% of this cohort developed better response yielding more than 4 oocytes in their next cycle . These data also confirm that poor response cannot be predicted only by history. Using other predictive tools to determine ovarian reserve, for instance antral follicle count or serum AMH, or both, is also critical to identify patients who will mostly benefits from DHEA in the clinical trial .
This systematic review did not include any self controlled studies, but included only the studies that had a group of contemporaneous controls. Our results in this review indicate that DHEA decreases the ovarian response as indicated by reduced number of oocytes retrieved at egg collection in the study arm. While this finding is surprising and it is difficult to come up with a scientific explanation, this systematic review is limited by a small number of treatment cycles included in the meta-analysis and by the heterogeneity of the included studies. On the contrary to the above finding, as indicated by trends of improving clinical pregnancy and of reducing miscarriage albeit no statistical significance, DHEA may have a positive effect on improving oocyte and embryo quality. It is already established that poor oocyte quality, which leads to producing poorer quality embryos, represents one of the clinical presentations of ovarian ageing . The finding of potential effects of DHEA on oocyte quality and ovarian response warrant well-designed randomised controlled clinical trials of an adequate sample size using well-defined uniform inclusion criteria before recommending the use of DHEA in standard assisted reproduction treatment. In addition, further in-vivo and in-vitro embryological and endocrinological research to elucidate the mechanism of action of DHEA on ovarian folliculogenesis and on oocyte/embryo quality are also required.
Recent meta-analysis which evaluated the effect of adjuvant androgens (DHEA or Testosterone) or androgen-modulating agents (Letrozole, aromatase inhibitor) in previous poor responders has failed to demonstrate any significant difference in the ongoing pregnancy rate, live birth rates, and numbers of oocytes retrieved when compared with the control group, who have had no adjuvant therapy . Another systematic review by Bosdou et al. 2012 reports a significant increase in clinical pregnancy and live birth rates in poor responder women who are pre-treated with transdermal testosterone, but not in other androgen modulating agents including DHEA, when compared to controls . The review by Bosdou et al. on DHEA included only one study with a small sample size (n=33) in contrast to this review in which we included two other eligible controlled studies with larger sample sizes. However, it is important to note that all the reviews including ours are still limited by small sample sizes and heterogeneity between the studies included. In addition, there is a wide methodological variation in different studies in terms of different preparations that are used as adjuvants, and duration and timing, in relation to down regulation and ovarian stimulation during IVF, of adjuvant treatments. This conflicting data regarding the use of androgen and androgen modulating agents require further investigation by well-designed randomised controlled trials.