Endometrial injection of embryo culture supernatant for subfertile women in assisted reproduction

Charalampos S Siristatidis; Eleni Sertedaki; Vasilios Karageorgiou; Dennis Vaidakis

doi:10.1002/14651858.CD013063.pub2

. 2020 Aug 14;2020(8):CD013063. doi: 10.1002/14651858.CD013063.pub2

Endometrial injection of embryo culture supernatant for subfertile women in assisted reproduction

Charalampos S Siristatidis ^1,^✉, Eleni Sertedaki ², Vasilios Karageorgiou ³, Dennis Vaidakis ⁴

Editor: Cochrane Gynaecology and Fertility Group

PMCID: PMC8094452 PMID: 32797689

Abstract

Background

Despite substantial improvements in the success of assisted reproduction techniques (ART), live birth rates may remain consistently low, and practitioners may look for innovative treatments to improve the outcomes. The injection of embryo culture supernatant in the endometrial cavity can be undertaken at various time intervals before embryo transfer. It provides an altered endometrial environment through the secretion of factors considered to facilitate implantation. It is proposed that injection of the supernatant into the endometrial cavity prior to embryo transfer will stimulate the endometrium and provide better conditions for implantation to take place. An increased implantation rate would subsequently increase rates of clinical pregnancy and live birth, but current robust evidence on the efficacy of injected embryo culture supernatant is lacking.

Objectives

To evaluate the effectiveness and safety of endometrial injection of embryo culture supernatant before embryo transfer in women undergoing ART.

Search methods

Our search strategies were designed with the help of the Cochrane Gynaecology and Fertility Group Information Specialist. We sought to identify all published and unpublished randomised controlled trials (RCTs) meeting inclusion criteria. Searches were performed on 2 December 2019.

We searched the Cochrane Gynaecology and Fertility Group Specialised Register of controlled trials, CENTRAL, MEDLINE, Embase, CINAHL, trials registries and grey literature. We made further searches in the UK National Institute for Health and Care Excellence (NICE) fertility assessment and treatment guidelines. We handsearched reference lists of relevant systematic reviews and RCTs, together with searches of PubMed and Google for any recent trials that have not yet been indexed in the major databases. We had no language or location restrictions.

Selection criteria

We included RCTs testing the use of endometrial injection of embryo culture supernatant before embryo transfer during an ART cycle, compared with the non‐use of this intervention, the use of placebo or the use of any other similar drug.

Data collection and analysis

Two review authors independently selected studies, assessed risk of bias, extracted data from studies and attempted to contact the authors where data were missing. We pooled studies using a fixed‐effect model. Our primary outcomes were live birth/ongoing pregnancy and miscarriage. We performed statistical analysis using Review Manager 5. We assessed evidence quality using GRADE methods.

Main results

We found five RCTs suitable for inclusion in the review (526 women analysed). We made two comparisons: embryo culture supernatant use versus standard care or no intervention; and embryo culture supernatant use versus culture medium.

All studies were published as full‐text articles. Data derived from the reports or through direct communication with investigators were available for the final meta‐analysis performed. The GRADE evidence quality of studies ranged from very low‐quality to moderate‐quality. Factors reducing evidence quality included high risk of bias due to lack of blinding, unclear risk of publication bias and selective outcome reporting, serious inconsistency among study outcomes, and serious imprecision due to wide confidence intervals (CIs) and low numbers of events.

Comparison 1. Endometrial injection of embryo culture supernatant before embryo transfer versus standard care or no intervention:

One study reported live birth only and two reported the composite outcome live birth and ongoing pregnancy. We are uncertain whether endometrial injection of embryo culture supernatant before embryo transfer during an ART cycle improves live birth/ongoing pregnancy rates compared to no intervention (odds ratio (OR) 1.11, 95% CI 0.73 to 1.70; 3 RCTs; n = 340, I² = 84%; very low‐quality evidence). Results suggest that if the chance of live birth/ongoing pregnancy following placebo or no treatment is assumed to be 42%, the chance following the endometrial injection of embryo culture supernatant before embryo transfer would vary between 22% and 81%.

We are also uncertain whether the endometrial injection of embryo culture supernatant could decrease miscarriage rates, compared to no intervention (OR 0.89, 95% CI 0.44 to 1.78, 4 RCTs, n = 430, I² = 58%, very low‐quality evidence). Results suggest that if the chance of miscarriage following placebo or no treatment is assumed to be 9%, the chance following injection of embryo culture supernatant would vary between 3% and 30%.

Concerning the secondary outcomes, we are uncertain whether the injection of embryo culture supernatant prior to embryo transfer could increase clinical pregnancy rates (OR 1.13, 95% CI 0.80 to 1.61; 5 RCTs; n = 526, I² = 0%; very low‐quality evidence), decrease ectopic pregnancy rates (OR 0.32, 95% CI 0.01 to 8.24; n = 250; 2 RCTs; I² = 41%; very low‐quality evidence), decrease multiple pregnancy rates (OR 0.70, 95% CI 0.26 to 1.83; 2 RCTs; n = 150; I² = 63%; very low‐quality evidence), or decrease preterm delivery rates (OR 0.63, 95% CI 0.17 to 2.42; 1 RCT; n = 90; I² = 0%; very low‐quality evidence), compared to no intervention. Finally, there may have been little or no difference in foetal abnormality rates between the two groups (OR 3.10, 95% CI 0.12 to 79.23; 1 RCT; n = 60; I² = 0%; low‐quality evidence).

Comparison 2. Endometrial injection of embryo culture supernatant versus endometrial injection of culture medium before embryo transfer

We are uncertain whether the use of embryo culture supernatant improves clinical pregnancy rates, compared to the use of culture medium (OR 1.09, 95% CI 0.48 to 2.46; n = 96; 1 RCT; very low‐quality evidence). No study reported live birth/ongoing pregnancy, miscarriage, ectopic or multiple pregnancy, preterm delivery or foetal abnormalities.

Authors' conclusions

We are uncertain whether the addition of endometrial injection of embryo culture supernatant before embryo transfer as a routine method for the treatment of women undergoing ART can improve pregnancy outcomes. This conclusion is based on current available data from five RCTs, with evidence quality ranging from very low to moderate across studies. Further large well‐designed RCTs reporting on live births and adverse clinical outcomes are still required to clarify the exact role of endometrial injection of embryo culture supernatant before embryo transfer.

Plain language summary

Endometrial injection of embryo culture supernatant for subfertile women in assisted reproduction

Review question:

Researchers in the Cochrane Collaboration reviewed the evidence about the effectiveness and safety of endometrial injection of embryo culture supernatant before embryo transfer in women undergoing assisted reproduction.

Background:

Assisted reproduction techniques (ART) includes techniques used for treating subfertility, and in vitro fertilisation (IVF) or intracytoplasmic sperm injection (ICSI) are the most common. Despite both clinical and laboratory efforts and improvements in the success of these treatments, pregnancy rates remain relatively low. In IVF, eggs are retrieved from a woman's ovaries and placed in a dish with sperm from her partner or from a donor, in a liquid called embryo culture medium. In ICSI, a single sperm cell is injected into a woman's egg, and then the egg goes into the culture medium.

Various factors allow a communication of embryo with maternal tissues. These could affect the receptivity of the lining of the womb (endometrium). Researchers have proposed that injecting the upper layer of the embryo culture's liquid, called supernatant, into the womb before embryo transfer, might stimulate the lining of the womb and facilitate the embryo's attachment to it. This could improve live birth rates and other ART outcomes.

Although injection before embryo transfer seems to be an encouraging procedure, its effectiveness and safety remain controversial, as there is not much available evidence about outcomes. In this Cochrane Review, we summarised the relevant evidence. We aimed for our conclusions to be as robust as possible, and also to identify any limitations to the evidence.

Study characteristics:

In consultation with the Cochrane Gynaecology and Fertility Group Information Specialist, we performed a comprehensive literature search of the standard medical databases, from each databases' earliest records to December 2019. We searched for all randomised controlled trials (RCTs) (studies in which participants are assigned to a treatment group using a random method) that investigated the effectiveness of endometrial injection of embryo culture supernatant before embryo transfer, during a cycle with IVF or ICSI, versus any other intervention or no intervention (usual care). IVF and ICSI are the two major modalities of ART. We searched for studies irrespective of language and country of origin. Two review authors independently selected and evaluated studies, extracted data, and attempted to contact the authors of studies for which data were missing. Five studies comprising 526 women met our review's inclusion criteria. We identified no ongoing studies.

Key results:

We are uncertain whether the routine use of endometrial injection of embryo culture supernatant before embryo transfer has a positive effect in women undergoing assisted reproduction, compared to usual care, for improving both live birth or ongoing pregnancy and miscarriage rates. We found that the live birth or ongoing pregnancy rate with usual care was 42%, but it varied between 35% and 55% when supernatant was injected. The risk of miscarriage was found to be 9% for usual care, and between 4% and 15% with the injection. We found similar conclusions for the rates of clinical pregnancy, multiple pregnancy and ectopic pregnancy, as well as for preterm delivery and foetal abnormalities. No single outcome measure in our review's included RCTs demonstrated a clear benefit with its use. There is also insufficient evidence to support the routine use of endometrial injection of embryo culture supernatant compared to culture media before embryo transfer.

Quality of the evidence:

The quality of the evidence was very low for nearly all outcomes. Evidence quality was low for foetal abnormalities. The main limitations of the included RCTs were poorly reported study methods, wide variations in the characteristics of the included studies and statistical imprecision due to the small study numbers and few numbers of events reported.

Summary of findings

Background

Description of the condition

Subfertility is a major problem among all populations and affects 10% to 15% of couples of reproductive age (Gnoth 2005). The average age of women who give birth to their first child is rising. Overall, it is almost 29 years of age in Europe, and in Mediterranean countries such as Greece, Italy, Malta and Spain it is one year older (Eurostat 2015; Eurostat 2019). The proportion of women who give birth to their first child between the ages of 30 and 39 is 59.4% in Spain and 51.9% in Greece. Similarly, in the USA, from 2000 to 2014, for all birth orders, the mean age of mothers has increased. Age at first birth has had the largest increase, from 24.9 years in 2000 to 26.3 years in 2014 (CDC 2016). A similar pattern is observed across Europe (Rendall 2010).

Assisted reproduction techniques (ART) include the techniques used for treating subfertility, and is defined as "all interventions that include the in vitro handling of both human oocytes and sperm or of embryos for the purpose of reproduction" (Zegers‐Hochschild 2017). It is estimated that the number of births worldwide as a result of in vitro fertilisation (IVF) has already exceeded eight million births (ESHRE 2018). However, pregnancy rates after IVF have remained stable in the past decade, at around 30% for cases undergoing intracytoplasmic sperm injection (ICSI) (Van Steirteghem 1993; Kuczyński 2001; Ben Rhouma 2003; Motteram 2015; EIM 2017). Advancing maternal age at primigravidity increases the proportion of couples who need assisted reproduction. Bearing in mind the significant impact of subfertility on a couple's quality of life, every effort should be made to increase their chance of live birth. Many interventions have been investigated to overcome this situation (Farquhar 2015), but with conflicting results and no firm consensus (Carney 2012; Lensen 2018; Siristatidis 2018). Some modalities that have been assessed for improving ART outcomes are endometrial injury (Nastri 2015), administration of granulocyte‐colony stimulating growth factor (Kamath 2020), aspirin (Siristatidis 2016), hyaluronic acid (Bontekoe 2014), antioxidants (Showell 2017) and pre‐treatment with androgens (Nagels 2015). In this context, current research focuses on enhanced understanding of cellular and molecular pathways involved in the process of implantation and the potential for targeted interventions to improve implantation rates.

Although there have been continuous efforts to improve understanding of human implantation in order to achieve a balance between regulation and dysregulation of endometrial function, and to facilitate the transfer of high‐quality embryos in subfertile women undergoing assisted reproduction, implantation rates remain limited. There is a lack of useful tests to assess the function of the healthy endometrium and in vivo peri‐implantation events remain in a “black box” (Macklon 2017). Moreover, there are reports pointing out that implantation failure, and especially repeated failure, is linked to failure to properly synchronize the embryo with the woman's "personalized window of implantation" (Valdes 2017).

Description of the intervention

Successful implantation is a result of a complex sequence of physiological events that must be synchronised in order for the zygote to travel through the salpinx and reach the endometrial cavity at a specific time (the 'implantation window'). This synchronisation necessitates a timely cross‐talk between the zygote and the endometrium (Lopata 1996).

For this purpose, an embryo culture supernatant is derived from the patient’s own embryos' culture media and injected into the uterine cavity in an IVF cycle with cleavage stage embryos or blastocysts. Embryos are taken off and put in a separate dish with fresh media, and then the supernatant (culture medium) is aspirated with an embryo transfer catheter. The supernatant injection (approximately 20 μL) is performed transcervically into the endometrial cavity, close to the fundus, using a separate IVF catheter (Goto 2007). The intervention is a straightforward, feasible technique that is easily accomplished before embryo transfer.

How the intervention might work

The rationale for the intervention is that it provides an altered and optimum endometrial environment that might mimic in vivo conditions, through the secretion of embryonic factors, which are considered to facilitate implantation.

Studies suggest the human embryo secretes various factors during its growth and prior to implantation, which seem to contribute to cross‐talk with the maternal tissues, thus modulating endometrial receptivity (Teh 2016). Specifically, reports suggest that the human pre‐implantation embryo produces various factor that modulate implantation potential, including immunosuppressive factors and vascular endothelial growth factor (VEGF), human leukocyte antigen G (HLA‐G), interleukins (including IL‐1 and IL‐8), leukaemia inhibitory factor (LIF), monocyte chemoattractant protein 1 (MCP‐1) and Regulated on Activation, Normal T Cell Expressed and Secreted (RANTES) (Dinarello 1994; Giudice 1995; Tazuke 1996; Krüssel 2000; Spandorfer 2000; Caballero‐Campo 2002; Achache 2006; Desai 2006). Moreover, it downregulates the human mucin gene 1 (MUC‐1), which naturally creates a barrier to the endometrial‐embryo attachment (Meseguer 2001). At the same time, the endometrium regulates other receptors, e.g. for oestrogen and progesterone, in an effort to improve the endometrial receptivity for the process of implantation (Tazuke 1996; Tehraninejad 2012; Teh 2016).

There is evidence to associate a negative impact of ovarian hyperstimulation on embryo‐endometrial synchrony and subsequent pregnancy outcome (Maheshwari 2012; Teh 2016). It is proposed that injection of the supernatant into the endometrial cavity prior to embryo transfer would stimulate the endometrium and provide better conditions (e.g. transfer of some cytokines released in the culture media in the uterine cavity) for implantation to take place. An increased implantation rate would subsequently increase rates of clinical pregnancy and live birth.

Why it is important to do this review

Injection of embryo culture supernatant prior to embryo transfer seems to be a promising procedure, but there is uncertainty as to its effectiveness and safety. In an initial report, the endometrial culture medium was cryopreserved along with blastocysts and transferred warmed two days before the frozen embryo transfer; authors reported significant improvements in implantation and pregnancy rates (Goto 2007). Further studies conducted using both cleavage and blastocyst stage fresh and frozen embryo transfer cycles reported conflicting results (Goto 2009; Zhu 2010; Prapas 2012; Kamath 2015; Hamdi 2018).

There is thus an emerging need to summarise current evidence and provide a clear view on the effectiveness of this practice in order to encourage or disprove its clinical application. In this Cochrane Review, authors systematically reviewed and synthesised the relevant evidence and identified any gaps or limitations in our current understanding. Assessment of the methodological quality of existing and ongoing trials may encourage the conduct of more studies on this topic.

Objectives

To evaluate the effectiveness and safety of endometrial injection of embryo culture supernatant before embryo transfer in women undergoing ART.

Methods

Criteria for considering studies for this review

Types of studies

We included published and unpublished randomised controlled trials (RCTs) that assessed the effectiveness and safety of endometrial injection of embryo culture supernatant before embryo transfer in women undergoing IVF or ICSI. Quasi‐randomised and crossover trials were excluded, unless first phase results were present. We had no limitations in terms of language or study setting.

Types of participants

Women and couples undergoing IVF or ICSI cycles (both fresh and frozen). Oocyte donation cycles were excluded, in order to ensure that the quality of the oocyte could not affect the final outcome due to the intervention examined.

Types of interventions

Endometrial injection of embryo culture supernatant before embryo transfer, during an assisted reproductive cycle with IVF or ICSI (fresh or frozen), versus any other intervention or no intervention (usual care).

Types of outcome measures

Primary outcomes

Effectiveness

Live birth or (in studies not reporting live birth) ongoing pregnancy per woman or couple randomised.

Live birth is defined as the delivery of a live foetus after 20 completed weeks of gestational age. Ongoing pregnancy is defined as the presence of a foetal heart on ultrasound scan after 12 weeks of gestation per woman or couple randomised. cumulative live birth was also to be reported, if data were available.

Safety

Miscarriage rates per woman or couple randomised.

Miscarriage is defined as the loss of pregnancy before 20 completed weeks of gestational age.

Secondary outcomes

Effectiveness

Clinical pregnancy rate per woman or couple randomised;

Clinical pregnancy is defined as the presence of a foetal heart on ultrasound scan at seven weeks of gestation.

Safety

Adverse events per woman or couple randomised:
- multiple and ectopic pregnancy rates;
- foetal growth restriction;
- preterm delivery (< 37 weeks of gestation);
- foetal abnormality rate (chromosomal, congenital and anatomical).

Search methods for identification of studies

We searched for published and unpublished RCTs that assess the impact of endometrial injection of embryo culture supernatant before the embryo transfer, during a cycle with IVF or ICSI, in consultation with the Cochrane Gynaecology and Fertility Group's Information Specialist.

Electronic searches

We searched the following databases:

the Cochrane Gynaecology and Fertility Group Specialised Register; PROCITE platform, searched 2 December 2019 (Appendix 1)
CENTRAL; OVID platform, searched 2 December 2019 (Issue October 2019) (Appendix 2)
MEDLINE; OVID platform, searched from 1946 to 2 December 2019 (Appendix 3),
Embase; OVID platform, searched from 1980 to 2 December 2019 (Appendix 4)
CINAHL Plus; EBSCO platform, searched from 1961 to 2 December 2019 (Appendix 5).

All searches were carried out without any language, setting or date restriction.

We combined the MEDLINE search with the Cochrane Highly Sensitive Search Strategy for identifying RCTs that appears in the Cochrane Handbook of Systematic Reviews of Interventions (Chapter 6, 6.4.11; Lefebvre 2011). We combined the Embase and CINAHL searches with trial filters developed by the Scottish Intercollegiate Guidelines Network (SIGN) (www.sign.ac.uk/methodology/filters.html#random).

We searched the World Health Organization's International Clinical Trials Registry Platform (http://apps.who.int/trialsearch/Default.aspx) and the ClinicalTrials.gov registry for ongoing and registered trials (Appendix 6). We also searched OpenGrey (www.opengrey.eu/) for grey literature. We consulted experienced clinicians to learn of any ongoing or existing studies that we had not identified.

Searching other resources

We examined the references lists of all studies (included and excluded) and relevant reviews in order to identify further relevant articles. We also contacted experts in the field to find any additional studies.

Data collection and analysis

We entered data into Review Manager 5 (RevMan 5) (RevMan 2014). We conducted statistical analysis in accordance with the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).

Selection of studies

Two review authors (DV and ES) independently screened the titles and abstracts of the publications identified by the literature search strategy. We excluded studies that did not meet the inclusion criteria and retrieved the full‐text articles of the remaining publications. We evaluated these independently to identify RCTs eligible for inclusion. Any potential disagreements related to study eligibility were resolved by discussion with another review author (CSS). All excluded studies were listed after full‐text assessment in the ‘Characteristics of excluded studies' tables, and we documented the study selection process in a PRISMA flow chart (Figure 1).

Data extraction and management

Two review authors (DV and ES) independently extracted study characteristics and outcome data from the included studies using a pre‐designed data extraction form. All authors searched for detailed information on participants, interventions, comparators, outcomes, study design, funding sources and declarations of interest for the primary investigators. For studies with multiple publications, we used the main RCT report as the reference and we supplemented it with additional data from the secondary publications.

With the aim of retrieving additional data or methodological details where necessary, we contacted authors of the included studies via email. We sent a reminder if needed (a second email 15 days after the first communication, if we received insufficient data or no reply). Requested information included missing data, potential subgroups that were not reported in the published manuscript and elaboration on methods for the quality assessment. We resolved any potential disagreements through consensus involving one of the other review authors (CSS or VK). One review author imported data into RevMan 5, and a second review author validated the imported values against the data extraction form.

Assessment of risk of bias in included studies

Two review authors (DV and ES) independently assessed risk of bias in the included studies using the Cochrane ‘Risk of bias’ assessment tool for selection bias (random sequence generation and allocation concealment), performance bias (blinding of participants and personnel), detection bias (blinding of outcome assessment), attrition bias (incomplete outcome data), reporting bias and other types of bias (Higgins 2011). A priori, we considered that a degree of bias might inevitably exist in the studies, not only due to the difficulty in blinding clinical staff and embryologists, but also due to the unavoidable interaction between clinicians and patients, given the nature of clinical processes required for the best possible treatment outcomes. Considering these factors, we proceeded with a meticulous evaluation of the methods of each study, and requested further specifications from study authors in order to clarify our evaluations. We also examined the possibility of selective outcome reporting through comparison of outcomes reported in study protocols or in their articles' methods sections with the actual reported outcomes. We expected that pre‐specified primary outcomes would be appropriately described and that adverse events would be reported clearly. Our intention was to group multiple outcomes, if necessary. We resolved any disagreements through discussion with another review author (CSS). We explicitly reported our judgements of risk of bias in the ‘Risk of bias in included studies’ table in the ‘Characteristics of included studies’ section, with relevant information supporting our assessments.

Measures of treatment effect

All defined outcomes were binary (dichotomous), and we used the numbers of events in the control and intervention groups of each study to calculate Mantel‐Haenszel odds ratios (ORs). We used Peto ORs for outcomes with low event rates, when needed, as described in the ‘Data synthesis' section. We reversed the direction of effect of individual studies, if required, to ensure consistency across trials. We presented 95% CIs for the ORs.

Unit of analysis issues

We expected that all studies would have the woman (or couple) as the unit of randomisation. When data were not reported per woman (or couple), e.g. if studies reported data "per cycle," we made every effort either to extract the data from the text or retrieve them through correspondence with the study authors, or both. If we were unsuccessful, our intention was to summarise data in narrative analysis and in additional tables, as necessary.

We counted multiple live births (e.g. twins, triplets) as single live birth events.

Dealing with missing data

We evaluated included studies to determine whether missing data were randomly distributed. Where data were missing, we contacted trial authors to retrieve as much information as possible. Where this was unobtainable, we undertook imputation of individual values for our primary effectiveness outcome live birth/ongoing pregnancy only. Live birth/ongoing pregnancy was assumed not to have occurred in participants not reporting this. For other outcomes, we analysed the available data.

Assessment of heterogeneity

We initially considered whether the clinical and methodological characteristics of the included studies are consistent enough to provide a clinically meaningful results through data pooling in a meta‐analysis.

We assessed statistical heterogeneity with the I² statistic. We considered an I² statistic value of 30% to 60% to suggest moderate heterogeneity, and a value of 60% to 90% to suggest substantial heterogeneity across studies (Higgins 2011). In case of substantial heterogeneity with important clinical impact for a specific outcome, we explored possible explanations through subgroup and sensitivity analyses (where data were available).

Assessment of reporting biases

We aimed to minimise the potential impact of publication bias and other reporting biases by means of a thorough search for published and unpublished data, We planned to use a funnel plot to explore publication bias if we had combined data from 10 or more RCTs in a given meta‐analysis.

Data synthesis

All outcomes were dichotomous (binary). We combined data from similar RCTs, using a fixed‐effect Mantel‐Haenszel model. We reported the pooled ORs with their 95% CIs. Where events were rare, and if all relevant criteria were fulfilled, we considered the Peto method for pooling the data. An increase in the odds of the outcome is displayed graphically in the meta‐analyses to the right of the centre‐line, and a decrease in the odds of an outcome to the left of the centre‐line.

We planned to carry out pooled analyses for the following comparisons: women or couples randomised to receive endometrial injection of embryo culture supernatant before the embryo transfer, during a cycle with IVF or ICSI, versus women/couples randomised to receive one of the following comparators:

any other intervention during a cycle of IVF or ICSI (e.g. endometrial injury or plasma infusion, or medical adjuncts given prior to embryo transfer);
sham or placebo‐type intervention; or
usual care (no additional intervention) during a cycle of IVF or ICSI.

Subgroup analysis and investigation of heterogeneity

Clinical heterogeneity, given the diversity in interventions, may affect the results. Differences in the cross‐talk between endometrium and embryo could affect the outcome of the endometrium culture injection at different times before embryo transfer. In addition, the quality of the embryo on day 3 or day 5 may be different; the blastocyst is considered to be more viable with higher rates of successful implantation and the number of previous ART cycles reflects the potential of each woman to conceive (more unsuccessful attempts reduce the possibilities). Where there was substantial heterogeneity (I² > 60%), we determined effects for the primary and the most clinically important secondary outcomes within the following subgroups, if data were sufficient for any meaningful analyses:

age of the woman (≤ 37 years, 38 to 41 years, ≥ 42 years);
day of embryo transfer (early cleavage or blastocyst);
type of cycle (frozen or fresh); and
time of endometrium culture injection before embryo transfer (e.g. ≥ 1 day and ≥ 1 hour before the transfer).

Sensitivity analysis

We conducted sensitivity analyses for the primary outcomes of live birth/ongoing pregnancy and miscarriage, as well as for the most clinically important secondary outcomes of clinical pregnancy, ectopic pregnancy and multiple pregnancy, in order to determine whether the conclusions are robust to arbitrary decisions made regarding the eligibility and analysis. These analyses included consideration of whether the review conclusions would have differed if:

1. a random‐effects model had been used;

2. the summary effect measure had been risk ratio (RR) rather than OR;

3. studies at high or unclear risk of selection bias (random sequence generation and allocation concealment) or reporting bias and studies at high risk of bias in any other domain had been excluded;

4. the type of studies (full text or abstract) had been considered;

5. the type of denominator in miscarriage (per pregnancy or per woman) had been considered; and

6. the primary outcome had been restricted to live birth or included both live birth and ongoing pregnancy rate.

Overall quality of the evidence: ‘Summary of findings' table

We prepared two ‘Summary of findings' tables using the browser‐based version of GRADEpro (GRADEpro GDT 2015). This table evaluates the overall quality of the body of evidence for the main review comparison (endometrial injection of embryo culture supernatant versus no intervention, or versus any other intervention) on the primary and most important secondary outcomes (live birth/ongoing pregnancy, miscarriage, clinical pregnancy and adverse events) using GRADE criteria on study limitations, consistency of effect, imprecision, indirectness and publication bias. We justified, documented and incorporated judgements about evidence quality into reporting of results for each outcome. Two review authors (ES and DV) independently assessed the quality of the evidence, and resolved any disagreements by consulting a third review author (CSS). We prepared two different ‘Summary of findings' tables according to the comparisons (intervention versus no intervention or usual care; and intervention versus stimulation of the endometrium with culture media).

We extracted study data, formatted our comparisons in data tables and prepared the ‘Summary of findings' tables before writing the results and conclusions of our review.

Results

Description of studies

Results of the search

Through the initial database search and other sources, we identified 8325 articles. After removing duplicate records, and screening 1110 titles and the remaining 800 abstracts, we identified 44 studies as possibly eligible, and retrieved the full‐text report. From these, we excluded 39 studies, with reasons for exclusion. We identified and selected five studies for analysis. We found no studies awaiting further assessment nor any ongoing studies. We have displayed the summary details in the relevant figure (Figure 1).

Included studies

Five RCTs including 526 women in total met the inclusion criteria for this review (Goto 2009; Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015). We selected these for qualitative and quantitative analysis.

We sent emails to authors of all included studies (with appropriate reminders) to obtain additional data.

In accordance with our pre‐specified criteria and the available data, we were able to include all five included studies in our meta‐analysis (Goto 2009; Zhu 2010; Prapas 2012;Tehraninejad 2012; Kamath 2015).

Study design and setting

All five studies were set as RCTs, conducted in IVF centres (private or university‐based) in China, Greece, India, Iran and Japan. Sample size was small in three of the five included studies (Zhu 2010,Tehraninejad 2012; Kamath 2015), while two studies were relatively larger: Goto 2009 randomised 144 women and Prapas 2012 randomised 400 women. All studies were conducted in single centres. All studies were published as full articles.

Participants

Baseline characteristics of the participants did not differ significantly across studies or intervention groups, Four studies included subfertile women, younger than 38 years old, who were eligible for IVF or ICSI treatment, and excluded women with risk factors for negative clinical outcomes, such as hydrosalpinx or endometriosis. One study (Prapas 2012) had a higher mean (standard deviation (SD)) age of included women (37.2 (5.2) for experimental, 35.8 (5.8) for control group). Three studies (Zhu 2010; Tehraninejad 2012; Kamath 2015) reported the mean (SD) duration of couple infertility, which ranged from 5.1 (2.8) years (Zhu 2010) to 8.07 (3.53 years) (Kamath 2015). Three studies reported basal FSH levels, which ranged from 5.1 (2.3) mIU/mL (Zhu 2010) to 6.7 (2.3) mIU/mL (Tehraninejad 2012).

Type of infertility was reported by two studies (Zhu 2010; Kamath 2015). Infertility was primary in 55% and 57% and secondary in 45% and 33% of the included women, respectively. In Prapas 2012, the 200 oocyte donor cycles were evenly distributed across both groups and the study was included. After communication with study authors, we obtained and only included the 200 cycles for whom non‐donor oocytes were used. Concerning the employment of fresh or frozen strategies, one study (Goto 2009) described frozen‐thaw embryo transfer for the purposes of the study.

Data availability

In all five studies, data on all randomised women were available for analysis. All studies reported no losses after randomisation and thorough follow‐up.

Interventions

In the intervention groups of all studies, before the embryo transfer, the embryo culture supernatant was injected into the uterine cavity, using an embryo transfer catheter placed just beyond the internal os, prior to embryo transfer.

In all studies, the comparison was between injection of embryo culture supernatant before embryo transfer, versus no injection prior to transfer (Goto 2009; Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015). One study had an extra comparison, examining injection of embryo culture supernatant before embryo transfer versus injection of culture medium before embryo transfer (Goto 2009).

In two studies, embryos were transferred either on day 3 or day 5 (Prapas 2012, Zhu 2010), while in the remaining studies embryos were transferred on day 5 (Goto 2009; Tehraninejad 2012; Kamath 2015).

In one study (Goto 2009) authors used frozen thawed blastocysts and embryo culture medium and in another (Kamath 2015) authors used vitrified warmed blastocysts.

Outcomes

Rates of live birth or ongoing pregnancy were reported in three studies (340 women analysed; Prapas 2012; Tehraninejad 2012; Kamath 2015). One study reported on live birth rates only (Kamath 2015), while two studies reported on live birth or ongoing pregnancy rates as a combined outcome (Prapas 2012; Tehraninejad 2012). Miscarriage rates were reported in four studies (430 women; Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015).

Regarding the secondary outcomes, clinical pregnancy was reported in all five studies. Ectopic pregnancy was reported in one study (190 women; Prapas 2012). Multiple pregnancy was reported in two studies (150 women; Zhu 2010; Kamath 2015).

The preterm delivery rate was reported in one study (90 women; Tehraninejad 2012). Foetal abnormality was reported in one study (60 women; Kamath 2015).

Excluded studies

Of the 800 records we identified after removal of duplicates, we excluded 766 studies on the basis of the abstract (Figure 1). Of the remaining 44 papers, all full texts were retrieved. Of these, we subsequently excluded 38 as non‐RCTs, or because they did not explore the outcomes of interest pre‐specified in this review. We excluded one quasi‐randomised RCT (Goto 2007) (see Characteristics of excluded studies).

Studies awaiting assessment

No studies are awaiting further classification.

Ongoing studies

We found no ongoing studies related to the objective of this review.

Risk of bias in included studies

The risk of bias is described in detail in the 'Risk of bias' table in Characteristics of included studies, and presented in Figure 2 and Figure 3. The decision has been made after sending emails to the study authors in an attempt to retrieve any further data.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

One study had a published registered protocol in international databases (Kamath 2015), and one study had an ethical approval obtained, reported as a specific number in the text (Tehraninejad 2012). The domains for whom high or unclear risk was noted were blinding of assessment (Goto 2009; Zhu 2010; Tehraninejad 2012; Kamath 2015) and blinding of participants (Goto 2009; Zhu 2010; Kamath 2015).

Allocation

Random sequence generation

All five studies were introduced as RCTs. We assessed all five of them to be at low risk of selection bias for random sequence generation, as the investigators used computer generated randomization sequences for the selection of the women or, in one case, coloured‐marble lots drawn by a blinded technician.

Allocation concealment

We considered four studies (Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015) to be at low risk of bias in this domain, having used sealed, consecutively numbered opaque envelopes or by blinded personnel, so that patients and investigators could not foresee the interventions. Concerning Goto 2009, in the absence of the relevant data in the published report, we tried to contact the investigators, but we did not receive any additional information. We judged this domain to be at unclear risk of bias (Goto 2009).

Blinding

Two studies were considered to be at low risk of performance bias because they described clearly blinding of patients, physicians and laboratory personnel (Prapas 2012; Tehraninejad 2012). The remaining three studies were at high risk of performance bias, due to lack of blinding of participants and personnel or because in the absence of details, we deemed that the method used was likely to have affected the results (Goto 2009; Zhu 2010; Kamath 2015).

We assessed two studies to be at low risk of detection bias (Goto 2009; Prapas 2012), according to information provided by the investigators through direct communication. Two studies were at unclear risk of detection bias, due to lack of sufficient information to reach a conclusion (Tehraninejad 2012; Kamath 2015). We judged one study to be at high risk of detection bias due to the complete absence of details concerning blinding of outcome assessment. We deemed that the method used was likely to have affected the results (Zhu 2010).

Incomplete outcome data

All five studies were assessed to be at low risk of attrition bias, as no losses through follow‐up were included and all subjects initially randomised were included in the statistical analysis of the studies' selected outcomes.

Selective reporting

We assessed one study to be at low risk of reporting bias (Kamath 2015) due to the adequate presentation of favourable and adverse clinical outcomes, and to the existence of a registered protocol in the Indian trial registry (CTRI/2013/01/003280), published before the study was conducted. We assessed three studies (Zhu 2010; Prapas 2012; Tehraninejad 2012) to be at unclear risk of reporting bias, as all pre‐specified outcomes of the study were adequately reported, including reporting of adverse events. However, these studies did not have published protocols pre‐specifying the study objectives. We assessed one study to be at high risk of reporting bias, as the investigators did not report the primary adverse event (miscarriage) (Goto 2009). This would be expected from a study that focused on embryological outcomes, avoiding a longer follow‐up that could reveal clinically important results, such as live births or ongoing pregnancies (Goto 2009).

Other potential sources of bias

We assessed one study to be at unclear risk of other bias because of a lack of sufficient information to determine otherwise (Goto 2009).

All other four studies were judged at low risk of other bias, because we could not detect further methodological gaps and we had no rationale to question another source inducing bias in the results (Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015)

Effects of interventions

See: Table 1; Table 2

Summary of findings 1. Embryo culture supernatant injection compared to no intervention for subfertile women in assisted reproduction.

Embryo culture supernatant injection compared to no intervention for subfertile women in assisted reproduction
Patient or population: Subfertile women undergoing assisted reproduction Setting: IVF Units Intervention: Embryo culture supernatant injection before embryo transfer Comparison: No intervention
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Quality of the evidence (GRADE)	Comments
Outcomes	Risk with no intervention	Risk with embryo culture supernatant	Relative effect (95% CI)	№ of participants (studies)	Quality of the evidence (GRADE)	Comments
Live birth/ongoing pregnancy	419 per 1,000	419 per 1,000 (218 to 808)	OR 1.11 (0.73 to 1.7)	340 (3 RCTs)	⊕⊝⊝⊝ VERY LOW ^1,2,3
Miscarriage	88 per 1,000	102 per 1,000 (29 to 303)	OR 0.89 (0.44 to 1.78)	430 (4 RCTs)	⊕⊝⊝⊝ VERY LOW ^1,2,3
Clinical pregnancy	513 per 1,000	544 per 1,000 (462 to 636)	OR 1.13 (0.80 to 1.61)	526 (5 RCTs)	⊕⊝⊝⊝ VERY LOW^1,2,3
Ectopic pregnancy	8 per 1,000	3 per 1,000 (0 to 61)	OR 0.32 (0.01 to 8.24)	250 (2 RCTs)	⊕⊝⊝⊝ VERY LOW^1,2,3
Multiple pregnancy	147 per 1,000	107 per 1,000 (43 to 239)	OR 0.70 (0.26 to 1.83)	150 (2 RCTs)	⊕⊝⊝⊝ VERY LOW^1,2,3
Preterm delivery	133 per 1,000	88 per 1,000 (25 to 271)	OR 0.63 (0.17 to 2.42)	90 (1 RCT)	⊕⊝⊝⊝ VERY LOW^2,4
Foetal abnormalities	0 per 1,000	0 per 1,000 (0 to 0)	OR 3.10 (0.12 to 79.23)	60 (1 RCT)	⊕⊕⊕⊝ LOW⁴
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  CI: Confidence interval; OR: Odds ratio
GRADE Working Group grades of evidence High quality: further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: we are very uncertain about the estimate.

Open in a new tab

¹Downgraded one level for imprecision; wide CIs and low number of events. ² Downgraded one level due to studies at high risk of bias and unclear risk of publication bias ³ Downgraded one level due to high heterogeneity ⁴Downgraded two levels for imprecision; wide CIs and low number of events.

Summary of findings 2. Embryo culture supernatant compared to culture medium injection for subfertile women in assisted reproduction.

Embryo culture supernatant compared to culture medium injection for subfertile women in assisted reproduction
Patient or population: Subfertile women undergoing assisted reproduction Setting: IVF Units Intervention: Embryo culture supernatant injection before embryo transfer Comparison: Culture medium injection
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Quality of the evidence (GRADE)	Comments
Outcomes	Risk with culture medium injection	Risk with embryo culture supernatant	Relative effect (95% CI)	№ of participants (studies)	Quality of the evidence (GRADE)	Comments
Live birth/ongoing pregnancy	Not reported in any study in this comparison
Miscarriage	Not reported in any study in this comparison
Clinical pregnancy	583 per 1,000	604 per 1,000 (402 to 775)	OR 1.09 (0.48 to 2.46)	96 (1 RCT)	⊕⊝⊝⊝ VERY LOW^1,2
Ectopic pregnancy	Not reported in any study in this comparison
Multiple pregnancy	Not reported in any study in this comparison
Preterm delivery	Not reported in any study in this comparison
Foetal abnormalities	Not reported in any study in this comparison
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  CI: Confidence interval; RR: Risk ratio; OR: Odds ratio;
GRADE Working Group grades of evidence High quality: further research is very unlikely to change our confidence in the estimate of effect. Moderate quality: further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate. Low quality: further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate. Very low quality: we are very uncertain about the estimate.

Open in a new tab

¹Downgraded two levels for imprecision; wide CIs and low number of events ² Downgraded one level due to studies at high risk of bias and unclear risk of publication bias

Comparison 1: Embryo culture supernatant versus no intervention

Primary outcomes

1.1 Live birth or ongoing pregnancy rate per woman randomised

This outcome was reported in three out of five studies (Prapas 2012; Tehraninejad 2012; Kamath 2015). We are uncertain whether the endometrial injection of embryo culture supernatant injected before embryo transfer improves live birth or ongoing pregnancy rates compared to the use of placebo/no intervention (OR 1.11, 95% CI 0.73 to 1.70; 3 RCTs; n = 340; I² = 84%; very low‐quality evidence). Our findings suggest that if the chance of live birth/ongoing pregnancy following placebo or no treatment is assumed to be 42%, the chance following the endometrial injection of embryo culture supernatant before embryo transfer is estimated to be between 22 and 81% (Analysis 1.1; Figure 4).

1.1 — Comparison 1: Embryo culture supernatant vs no intervention, Outcome 1: live birth/ongoing pregnancy

Forest plot of comparison: embryo culture supernatant versus standard care, outcome: 1.1 live birth/ongoing pregnancy.

We observed considerable heterogeneity (I² = 84%). Only one study (Tehraninejad 2012) reported an outcome favouring the intervention, while the other two reported a benefit for no intervention. To assess heterogeneity, studies were stratified according to mean age of participants, as pre‐specified in our protocol.

Ιn the subgroup of women of ≤ 37 years of age (Tehraninejad 2012; Kamath 2015), we are uncertain whether embryo culture injection increases live births (OR 1.36, 95% CI 0.73 to 2.54; 2 RCTs; n = 150; I² = 92%; very low‐quality evidence). Concerning the other three criteria, two studies reported a blastocyst transfer and fresh cycles (Prapas 2012; Tehraninejad 2012) and one study had frozen cycles (Kamath 2015).  In the subgroup analysis according to the timing of intervention, one study (Kamath 2015) reported an embryo culture supernatant injection performed two days prior to embryo transfer.  In the < 1 day subgroup analysis (Prapas 2012;Tehraninejad 2012), we are uncertain whether the intervention had an effect on live birth or ongoing pregnancy (OR 1.44, 95% CI 0.90 to 2.30; 2 RCTs; n = 280; I² = 84%; very low‐quality evidence).  In the subgroup analysis according to the type of cycle (fresh or frozen), one study reported frozen cycles (Kamath 2015). We are uncertain whether there was a significant effect of the intervention on the outcome of live birth/ongoing pregnancy (OR 0.30, 95% CI 0.10 to 0.92; 1 RCT; n = 60).

The sensitivity analysis of the studies, using a random‐effects model (OR 1.05, 95% CI 0.31 to 3.53; 3 RCTs; n = 340; I² = 84%; very low‐quality evidence), did not alter the results. There were insufficient data to perform a sensitivity analysis according to the risk of bias assessment, or type of studies (all studies were in full text) or the restriction of the primary outcome to live birth only (all studies reported live birth and ongoing pregnancy as a single outcome).

1.2 Miscarriage rate per woman randomised

This outcome was reported in four studies (Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015). We are uncertain whether the endometrial injection of embryo culture supernatant before embryo transfer improves the miscarriage rate compared to no intervention or placebo (OR 0.89, 95% CI 0.44 to 1.78; 4 RCTs; n = 430; I² = 58%; very low‐quality evidence). Results suggest that if the chance of miscarriage following placebo or no treatment is assumed to be 8.8%, the chance following injection of embryo culture supernatant is estimated to be between 2.9% and 30.3% (Analysis 1.2; Figure 5).

1.2 — Comparison 1: Embryo culture supernatant vs no intervention, Outcome 2: miscarriage

Forest plot of comparison: embryo culture supernatant versus standard care, outcome: 1.2 miscarriage.

In subgroup analyses, in the age subgroups, we are uncertain whether endometrial injection of supernatant improved miscarriage rates (years old subgroup: OR 0.86, 95% CI 0.40 to 1.85; 3 RCTs; n = 240; I² = 72%; very low‐quality evidence). In the timing of injection subgroup analysis, we are uncertain whether the intervention improved miscarriage rates (1 to 24 hours subgroup: OR 0.47, 95% CI 0.20 to 1.14; 2 RCTs; n = 280; I² = 58%; very low‐quality evidence). Finally, in the type of cycle (fresh or frozen) subgroup analysis, we are uncertain whether the intervention improved miscarriage rates in the fresh cycle (OR 0.57, 95% CI 0.25 to 1.27; 3 RCTs; n = 370) and frozen cycle (OR 7.25, 95% CI 0.82 to 64.46; 1 RCT, n = 60).

Sensitivity analysis of the studies using a random‐effects model (OR 1.19, 95% CI 0.31 to 4.52; 4 RCTs; n = 430; I² = 58%; very low‐quality evidence) did not alter the results. There was insufficient data to perform a sensitivity analysis according to the risk of bias assessment, or type of studies (all studies were in full text) or the type of denominator for miscarriage (per pregnancy or per woman), as all studies used 'per woman'.

Secondary outcomes

1.3 Clinical pregnancy rates per woman or couple randomised

This outcome was reported in all five studies (Goto 2009; Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015). We are uncertain whether the endometrial injection of embryo culture supernatant before embryo transfer improves clinical pregnancy rates compared with placebo (OR 1.13, 95% CI 0.80 to 1.61; 5 RCTs; n = 566; I² = 0%; very low‐quality‐evidence). Figure 6

Forest plot of comparison: embryo culture supernatant versus standard care, outcome: 1.3 clinical pregnancy.

Similarly, through subgroup analysis, we are uncertain whether there is a difference in clinical pregnancy between the experimental and the control group in the ≤ 37‐year‐old subgroup (OR 1.26, 95% CI 0.80 to 1.96; 4 RCTs; n = 336; I² = 0%; very low‐quality‐evidence), or by stratification according to the timing of intervention (1 to 24 hours subgroup: OR 1.19, 95% CI 0.78 to 1.81; 3 RCTs; n = 376; I²= 0%; very low‐quality evidence). In subgroup analysis according to type of cycle (fresh or frozen) two studies reported frozen cycles (Goto 2009; Kamath 2015). We are uncertain as to whether there was a significant effect of injection on clinical pregnancy rates (OR 1.05, 95% CI 0.56 to 1.98; 2 RCTs; n = 156; I² = 0%).

The sensitivity analysis of the studies (Goto 2009; Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015) using a random‐effects model (OR 1.13, 95% CI 0.79 to 1.61; 5 RCTs; n = 526; I² = 0%, very low‐quality evidence) did not alter the results. There was insufficient data to perform a sensitivity analysis according to the risk of bias assessment as none of the studies (Goto 2009; Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015) fulfilled the pre‐specified inclusion criteria (low risk of selection and reporting bias, low or unclear risk in all other domains, There were also insufficient data to perform a sensitivity analysis according to the type of studies (all studies were in full text).

1.4 Ectopic pregnancy

This outcome was reported in two studies (Prapas 2012; Kamath 2015). We are uncertain whether there was a difference between the intervention group and the control group with regard to ectopic pregnancy (OR 0.32, 95% CI 0.01 to 8.24; n = 250; 2 RCTs, I² = 41%; very low‐quality evidence). Only one event was recorded in the control group in one of the studies included (Kamath 2015), while the experimental arm of Kamath 2015 and both arms in Prapas 2012 exhibited zero events (Figure 7).

Stratification of patients by age could not generate a meaningful comparison, as this outcome was mentioned only in two studies; one of them had patients younger than 37 years old (Kamath 2015), while the other one had patients older than 37 years old (Prapas 2012).

The sensitivity analysis of the studies using a random‐effects model (OR 0.32, 95% CI 0.01 to 8.24; 2 RCTs; n = 250; I² = 0%, very low‐quality evidence) did not alter the results. There was insufficient data to perform a sensitivity analysis according to the risk of bias assessment or the type of studies (all studies were in full text).

1.5 Multiple pregnancy

This outcome was reported in two studies (Zhu 2010; Kamath 2015). We are uncertain whether the endometrial injection of embryo culture supernatant before embryo transfer improves the incidence of multiple pregnancies compared to no intervention/placebo (OR 0.70, 95% CI 0.26 to 1.83, 2 RCTs, n = 150, I² = 63%, very low‐quality evidence; Figure 8).

Stratification of patients by age was not applicable, as both studies reporting this outcome included patients younger than 37 years old. Through the pre‐specified subgroup analysis according to type of cycle (fresh or frozen), we are uncertain whether endometrial injection of embryo culture supernatant affected multiple pregnancy rates: fresh cycle (OR 1.75, 95% CI 0.39 to 7.81; 1 RCT; n = 90) and frozen cycle (OR 0.31, 95% CI 0.07 to 1.29; 1 RCT, n = 60).

The sensitivity analysis of the studies using a random‐effects model (OR 0.72, 95% CI 0.13 to 4.00; 2RCTs; n = 150; I² = 63%, very low‐quality evidence) did not alter the results. There was insufficient data to perform a sensitivity analysis according to the risk of bias assessment or the type of studies (all studies were in full text).

1.6 Preterm delivery

This outcome was reported in one study (Tehraninejad 2012). We are uncertain whether endometrial injection improved preterm delivery (OR 0.63, 95% CI 0.17 to 2.42, 1 RCT, n = 90, I² = 0%, studies = 1, very low‐quality evidence) Figure 9

There was insufficient data to perform any meaningful sensitivity analysis.

1.7 Foetal abnormalities

This outcome was reported in one study (Kamath 2015).There may have been little or no difference in foetal abnormalities between the intervention group and the control group (OR 3.10, 95% CI 0.12 to 79.23, 1 RCT, n = 60, I² = 0%, low‐quality evidence). Figure 10. Peto OR was 7.39 (0.15, 372.38).

There were insufficient data to perform any meaningful sensitivity analysis.

Other analyses

We performed a sensitivity analysis using a random effects model for the primary outcomes and for the most clinically important secondary outcomes.

(1) Live birth or ongoing pregnancy rate per woman randomised

The sensitivity analysis of the studies (Prapas 2012; Tehraninejad 2012; Kamath 2015) did not alter the results (OR 1.05, 95% CI 0.31 to 3.53; 3 RCTs; n = 340; I² = 84%; very low‐quality evidence).

(2) Miscarriage rate per woman randomised

The sensitivity analysis of the studies (Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015) did not alter the results (OR 1.19, 95% CI 0.31 to 4.52; 4 RCTs; n = 430; I² = 58%; very low‐quality evidence).

Similarly, the effect did not change for the secondary outcomes, either for clinical pregnancy (Goto 2009; Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015) (OR 1.13, 95% CI 0.79 to 1.61; 5 RCTs; n = 526; I² = 0%, very low‐quality evidence) or for ectopic pregnancy (Prapas 2012; Kamath 2015) (OR 0.32, 95% CI 0.01 to 8.24; 2 RCTs; n = 250; I² = 0%, very low‐quality evidence) or for multiple pregnancy (Zhu 2010; Kamath 2015) (OR 0.72, 95% CI 0.13 to 4.00; 2 RCTs; n =150; I² = 63%, very low‐quality evidence).

Each of the remaining outcomes, preterm delivery (Tehraninejad 2012) and foetal abnormalities (Kamath 2015), were reported in one study. We could not perform a meaningful sensitivity analysis.

We also performed a sensitivity analysis using RR to measure magnitude of effect (pooled RR) in the primary outcomes and the most clinically important secondary outcomes. This analysis was not mentioned in the review protocol, but it was conducted towards a more extensive investigation of the heterogeneity detected. The change is stated in the section Differences between protocol and review.

(1) Live birth or ongoing pregnancy rate per woman randomised

The sensitivity analysis of the studies (Prapas 2012; Tehraninejad 2012; Kamath 2015) did not alter the results (RR 1.00, 95% CI 0.52 to 1.93; 3 RCTs; n = 340; I² = 82%; very low‐quality evidence).

(2) Miscarriage rate per woman randomised

The sensitivity analysis of the studies (Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015) did not alter the results (RR 1.17, 95% CI 0.36 to 3.82; 4 RCTs; n = 430; I² = 54%; very low‐quality evidence).

Similarly, the effect did not change for the secondary outcome of clinical pregnancy (Goto 2009; Zhu 2010; Prapas 2012; Tehraninejad 2012; Kamath 2015) (RR 1.06, 95% CI 0.90 to 1.24; 5 RCTs; n = 526; I² = 0%; very low‐quality evidence). This was also true for the secondary outcomes of ectopic pregnancy (Prapas 2012; Kamath 2015) (RR 0.33, 95% CI 0.01 to 7.87; 2 RCTs; n = 250; I² = 0%; very low‐quality evidence) and multiple pregnancy (Zhu 2010; Kamath 2015) (RR 0.73, 95% CI 0.31 to 1.71; 2 RCTs; n = 150; I² = 60%; very low‐quality evidence) .

Each of the remaining outcomes, preterm delivery (Tehraninejad 2012) and foetal abnormalities (Kamath 2015) were reported in one study respectively.Thus, we could not perform a meaningful sensitivity analysis.

There was insufficient evidence to perform further sensitivity analyses in terms of risk of bias assessment, type of study, type of denominator in miscarriage (per woman or per pregnancy) or by restriction of the primary outcome to live birth only.

We conducted the planned analyses using Peto ORs and the results were as follows. For miscarriage, Peto OR was 0.90 (95% CI 0.46 to 1.79; 4 RCTs; n = 430; I² = 58%; very low‐quality evidence). For ectopic pregnancy, Peto OR was 0.14 (95% CI 0.003 to 7.04; 2 RCTs; n = 250, very low‐quality evidence). For multiple pregnancy, Peto OR was 0.70 (95% CI 0.27 to 1.82; 2 RCTs; n =150; I² = 63%; very low‐quality evidence). For foetal abnormalities, Peto OR was 7.39 (95% CI 0.15 to 372.38; 1 RCT; n = 60, moderate quality evidence).