Paper III: Systematic review on inositols and ovarian reserve

18 studies were selected for the review. Seven studies assessed changes in either AMH or AFC or both. Twelve studies evaluated outcomes following IVF/ICSI treatment. One study evaluated both categories of outcomes.

5.3.1 Included studies:

Of the studies assessing changes in AMH and /or AFC, only two of the seven studies were randomised controlled trials and five were non-randomised studies. All studies were single centre studies. A total of 415 women were recruited to these seven studies. A majority of the studies, five of the seven, were conducted in Italy. Although all participants in the included studies were diagnosed with PCOS, there were varying restrictions in the inclusion criteria based on age, body mass index and insulin resistance in some studies. We have included a study (Pkhaladze 2017(131)) which recruited participants between 13-19 years of age based

on a consensus opinion from the international paediatric subspecialty societies. There was a wide variation in the preparation of inositol used (MI and DCI), doses administered (ranging from 1 gm to 4 gm daily) and duration of administration (between 12 weeks to 6 months) with no explanations for these variations. Four studies assessed the effect on serum AMH and three assessed its effect on AFC.

10 of the 12 studies assessing reproductive outcomes after IVF/ICSI were RCTs and only two were non-randomised studies. All studies were single centre studies. Again, most of the studies were conducted in Italy (10 of the 12). A total of 1225 women, all with a diagnosis of PCOS were included in the 12 trials. Some studies restricted inclusion of participants based on age, body mass index and insulin resistance. Heterogeneity in the preparation of inositols used, dose (1-4 gm /300-1200 mg daily) and duration of treatment (2 weeks to ongoing treatment) were observed. Details are described in the manuscript.

5.3.2 Risk of bias and quality of evidence in included studies

Of the randomised trials, 75% of trials were at low risk for selection bias for random sequence generation but only 17% were considered low risk for allocation concealment. Similarly, only 25% of studies were considered at low risk of performance bias due to blinding. The risk of attrition bias due to incomplete outcome data was low in 42% of studies. None of the trials were considered at low risk of reporting bias.

All the non-randomised studies were considered at a high risk of selection bias due to the presence of confounders or inclusion of participants into the study. 83% were low risk for bias in classification of interventions and 67% were at low risk of bias due to missing data. 50% of studies did not provide information on nonconformities in the planned interventions and all of the studies were at high risk of reporting bias.

The quality of evidence contributing to the review was considered to be very low across most domains. These included inconsistency, indirectness, imprecision and a high risk of bias.

5.3.3 Synthesis of the results:

5.3.3.1 Primary Outcomes:

Anti-Mullerian Hormone

The studies assessing change in serum AMH before and after treatment with inositols showed

include these results for a meta-analysis. Five studies with 172 participants were included in this section of the review. The results for a standardised mean difference in levels of serum AMH before and after treatment for the five individual studies showed was no consistent direction or size of effect. This is graphically represented in the published manuscript.

Antral follicle count

We did not perform a meta-analysis for the studies which assessed changes in the antral follicle counts after treatment with inositols due to clinical and methodological heterogeneity of the included studies. Four studies with 143 participants were included in this section of the review.

The SMD show a high heterogeneity in effect size for AFC. This is graphically represented in the published manuscript.

Overall, the review is unable to demonstrate a clear change in AMH or AFC values following treatment with inositol.

5.3.3.2 Secondary Outcomes:

Number of retrieved oocytes

Eleven studies reported the number of oocytes retrieved. Seven RCTs comprising 722 participants were included in the meta-analysis. We were unable to find a statistically significant difference between the intervention and control arms (MD -0.39, 95% CI -1.11, 0.33). Similar results were seen for a subgroup analysis for MI (MD -0.76 95% CI -2.04, 0.52) and DCI (MD -0.18 95% CI -1.11,0.74).

Data from two studies (Ciotta 2011(132) and Lesoine 2016(133))could not be included in the meta-analysis as data were presented in an unsuitable format suitable for meta-analysis. These studies reported contradictory results. Ciotta 2011 reported a significantly higher number of oocytes following treatment with MI as compared to control (p<0.05) in contrast to Lesoine 2016 who reported a higher number of oocytes in the control group as compared to the MI group. Alviggi 2016 (134) in a non-randomised study reported no significant difference between the MI and control groups (p=0.23) similar to Unfer 2011 (135) who compared treatment with MI to DCI.

Number of metaphase II oocytes

10 studies reported on the number of metaphase II oocytes retrieved. Three RCTs with 207 participants were included in the meta-analysis. No statistically significant difference was seen between the intervention and control arms (MD 0.29, 95% CI -0.83, 1.40). Similar results were seen for a subgroup analysis for MI (MD -0.32 95%CI -1.49, 0.86). Seven other trials could not be included in the meta-analysis as data from these studies were in an unsuitable format for inclusion in the meta-analysis. They reported contradictory results with no consistent direction of effect. Piomboni 2014 (136) reported significantly higher number of MII oocytes, with a moderate effect size, following treatment with DCI as compared to control (MD 1.30 95%CI 0.15, 2.45) in contrast to Isabella 2012 (137) who reported a significant decrease in the number of MII oocytes following treatment with DCI as compared to control. Artini 2013 (138) reported a significant increase in the percentage of top quality oocytes in the group treated with MI as compared to control (p<0.05). Pacchiarotti 2016 (139) and Ciotta 2011 reported no significant differences between the groups. The results from Lesoine 2016 are unclear. Alviggi 2016 in a non-randomised study reported a significant increase in the number of MII oocytes following treatment with MI as compared to control (p=0.03). Unfer 2011 compared treatment with MI to DCI and reported a significant increase in the number of MII oocytes following treatment with MI as compared to DCI (p<0.05).

Number of top-grade embryos

Four RCTs comprising 957 participants and reporting number of top-grade embryos were included in the meta-analysis. We were unable to find a significant difference between the groups (RR 1.02, 95% CI 0.93-1.12). Four studies reporting this outcome had data in an unsuitable format for inclusion in the meta-analysis. These studies did not show a consistent direction of effect. Pacchiarotti (2016) reported no significant difference in the number of top-grade embryos between the MI and control groups. Isabella (2012) reported a significant decrease in the number of top-grade embryos following treatment with DCI (p<0.01) in contrast to Lesoine (2016) who reported a significantly higher number of top-grade embryos following treatment with MI as compared to control (p<0.05). Unfer (2011) reported a significantly higher number of top-grade embryos following treatment with MI as compared to treatment with DCI (p<0.01).

Clinical pregnancy rate

Three RCTs comprising 488 participants and reporting clinical pregnancy rates were included

1.16 95% CI 0.87, 1.53). Three studies (Schillaci 2012 (140), Alviggi 2016 and Wdowiak 2016 (141)) did not detail if the pregnancies were biochemical or clinical and were hence not included in the meta-analysis. These studies reported no significant differences between the groups. When comparing treatment with MI to DCI Unfer (2011) reported no significant difference in clinical pregnancy rates between the two groups.

Live birth rate

Only one trial (Artini 2013) reported live birth rate. This reported a significant improvement following treatment with MI compared to placebo (p<0.05).

Risk of cycle cancellation due to the risk of OHSS

Six studies reporting cycle cancellations due to the risk of ovarian hyperstimulation syndrome were included in the meta-analysis. We were unable to demonstrate a significant difference between groups (RR 0.73, 95% CI 0.39, 1.37). Similar results were seen for a subgroup analysis for MI (RR 0.70 95% CI 0.34, 1.42) and DCI (RR 0.85, 95% CI 0.22-3.29).