The study protocol was registered at http://www.crd.york.ac.uk/PROSPERO/ (registration number: CRD42024499584). It was exempt from institutional review board approval because it did not involve human intervention. The review was conducted in accordance with the PRISMA guidelines, as illustrated in Figure 1 and supplementary materials. ²¹

Eligibility criteria, information sources, search strategy literature search was conducted on PubMed, EMBASE and the Cochrane database of systematic reviews. Publications up to February 29, 2024 were considered, and reference lists of relevant reviews and articles were hand-searched. Inclusion criteria were: full-text articles in English, involving human subjects and reporting pregnancy outcomes (i.e., implantation rate, clinical or ongoing pregnancy rate and live birth rate). Articles on factors affecting implantation were included regardless of whether the population studied were RIF patients. This approach was adopted because the underlying pathophysiology of implantation failure is likely similar, regardless of the number of failures. For articles on treatments of RIF, the population was restricted to women with RIF. Patients were considered RIF if they had three or more implantation failures or were defined as such in the respective article.

The search specifically targeted publications that employed a set of keywords and MESH terms that covered various aspects of assisted reproductive technologies (ART) and embryo implantation. The keywords included, but were not limited to: Assisted Reproductive Technologies, ART, embryo implantation, implantation failure, recurrent implantation failure, RIF, recurrent pregnancy loss, RPL, early embryo miscarriage, early abortion, body mass index, BMI, habits, age, embryo quality, embryo nidation, endometrial omics and embryo implantation, endometrial decidualization, aneuploidy and embryo loss, fresh embryo transfer, frozen embryo transfer, coagulation, uterine factors, endometriosis, chronic endometritis, microbiome, thyroid, implantation window, endometrial receptivity, genetic factors, progesterone, endometrial scratching, hydrosalpinx, intrauterine adhesions. The complete search strategy and the specific selection criteria and PICO tables for each topic are available as supplementary material. ²¹

Titles and abstracts underwent a two-stage screening process to identify relevant studies. Full-text articles that met the inclusion criteria were retrieved. Two reviewers independently extracted data from eligible studies using standardized data collection forms. Two authors for each topic examined the articles included in the systematic review to identify articles to be included in the meta-analyses, namely those with comparable population and outcomes were included in meta-analyses.

The GRADE pro Guideline Development Tool ²² was used to assess the quality of evidence and to summarize the findings, in accordance with the GRADE methodology. ²³

Indirectness. Whenever possible, studies were pre-selected to ensure there were no relevant issues of indirectness, with only those directly relevant to the research question being included in the analysis. In presence of one or more sources of heterogeneity among populations considered for the study and/or the definition of the factor considered, indirectness issues have been rated as serious or very serious, respectively.

Risk of bias. The risk of bias was assessed prior to the meta-analysis to guide the selection of studies, whenever possible. For randomized controlled trials (RCTs), the Cochrane risk of bias tool version 2 ²⁴ was used. Non-randomized studies of interventions were evaluated with the ROBINS-I tool, ²⁵ while observational studies on exposures were assessed using ROBINS-E. ²⁶ Where possible, studies at high risk of bias were excluded from the meta-analysis. In cases with varying levels of risk of bias, subgroup analyses were conducted. Supplementary table 1 reports the risk of bias for the subgroup with the lowest risk of bias. The overall risk of bias was considered very serious in the presence of studies at high risk of bias, and serious in the presence of studies with moderate risk or with some concerns.

Inconsistency. Inconsistency was assessed based on the I ² statistic and the number of studies included. Specifically, inconsistency was rated as serious when I ² was greater than 50% or when fewer than three studies were available. It was rated as very serious when both conditions were met.

Imprecision. Imprecision was downgraded to serious if either the total number of events across studies was less than 500 or if the confidence interval (CI) of the Odds Ratio (OR) included 1. It was downgraded to very serious if both issues were present.

Other considerations. The potential presence of publication bias was assessed using a funnel plot. The strength of the association was evaluated based on the pooled OR values. An OR between 0.25 and 0.5 or between 2 and 4 was considered a strong association, while more extreme OR values (less than 0.25 or greater than 4) were identified as a very strong association.

To synthesize the findings of these studies, we used either common (also referred to as “fixed”) or random effects models to pool the odds ratios with their corresponding 95% confidence intervals. Although random effect models allow to take into account additional sources of heterogeneity, the precision in the estimation of the between-study variance may decrease if the number of studies is not sufficiently high, thereby negatively affecting the accuracy of the odds ratios pooled. For this reason, according to Murad et al., ²⁷ we used the common effects model when there were fewer than five studies. When the studies included were more than five, we used the random effects model. To gauge the variability among the studies selected, we computed and reported the I ² and τ2 values for each meta-analysis. These metrics enabled us to quantify and assess the extent of inconsistency among the study outcomes. Additionally, Cochran’s Q test was performed to evaluate the presence of heterogeneity and to ensure the robustness of our analytical approach. Analyses were carried out in the R version 4.3.1 ²⁸ using the “meta” package. ²⁹

In the meta-analysis, clinical pregnancy rate (CPR) and live birth rate (LBR) served as the measured outcomes to evaluate the embryo transfer success.

We conducted searches on 23 topics that resulted in 4,380 articles retrieved ( Figure 1 and Table 1). Figure 1 depicts the overall PRISMA flow diagram (PRISMA for individual systematic reviews of each topic are available as supplementary material ²¹ ). After eliminating duplicates and excluding those with titles and/or abstracts not pertinent to the objective of this review, 520 studies remained. Following full-text review, 302 studies were further excluded. Ultimately, 218 articles met the inclusion criteria ( Figure 1). Eighty articles covering 18 distinct topics were deemed suitable for inclusion. Following the evaluation of the risk of bias, 75 studies were included in the meta-analyses ( Figure 2).

The evaluation of the risk of bias are available as supplementary material. ²¹ Although every effort was made to summarize key evidence for all factors and treatments, strict adherence to an evidence-based approach was sometimes challenging due to inconsistencies in defining RIF and the sporadic availability of direct and high-quality evidence for numerous factors, diagnostic tools, and interventions.

Table 1 shows an overview of the results. The topics included in the systematic review were selected due to established association or relationship as well as for their presumptive role in implantation failure. When the articles identified lacked comparable data in terms of population or outcomes meta-analyses were not performed. The results of the meta-analyses are reported in this section, whereas factors and interventions for which the meta-analysis were not performed are discussed in the Discussion.

Meta-analyses of factors contributing to successful implantation

Body mass index

In women undergoing ART, normal BMI ((18.5–25 kg/m ²) is significantly associated with improved LBR as compared with women with overweight (>25 kg/m ²) with significantly higher odds of live birth for normal weight ( Figure 3). ^{30– 33} Although the effect is statistically significant (OR 1.11, 95% CI 1.02 to 1.22), the confidence interval is close to 1, which results in a moderate certainty of the evidence (Suppl. Table 1, line no. 1).

Age

In women of 35 years of age or younger undergoing ART, there is a significantly increased odds of live birth compared to women of older age (OR 1.26, 95% CI 1.15 to 1.37) ( Figure 4). ^{34, 35} This result is based on only two studies, therefore the certainty of the evidence has been downgraded to moderate (Suppl. Table 1, line no. 2).

Embryo euploidy

In women with RIF, transfer of euploid embryos, identified by Preimplantation Genetic Testing for Aneuploidies (PGT-A), significantly improved chance for successful embryo implantation. According to the common effects model, PGT-A has a significant positive effect on the odds of clinical pregnancy (CP) (OR 3.21, 95% CI 1.88 to 5.47) ( Figure 5). ^{36– 38} Despite the strong association, the certainty of the evidence has been downgraded to moderate due to the non-randomized study design, some concerns regarding the risk of bias, and the small sample size (Suppl. Table 1, line no. 3).

Hydrosalpinx

In women undergoing ART, the presence of untreated hydrosalpinx decreased the odds of CP versus women without hydrosalpinx (OR 0.22, 95% CI 0.13 to 0.37) ( Figure 6). ^{39– 41} Despite non-negligible issues regarding risk of bias, mainly related to the study design, the very strong association suggests that the certainty of evidence is moderate (Suppl. Table 1, line no. 4).

Coagulation abnormalities

The fixed effect model revealed that the abnormal coagulation (MTHFR 677C>T mutant genotype and activated protein C resistance) resulted in a decrease in the odds of clinical pregnancy (OR 0.49, 95% CI 0.31 to 0.78) ( Figure 7). ^{42, 43} Despite the strong association, the certainty of the evidence was downgraded to moderate due to the high heterogeneity between the two studies (Suppl. Table 1, line no. 5).

Intrauterine abnormalities

A significant percentage of RIF patients may have undetected intrauterine anomalies. ^{44– 47} Patients who underwent hysteroscopy during previous IVF cycle were compared to patients who did not. Hysteroscopy before IVF had a significant positive effect on clinical pregnancy in RIF patients (OR 1.47, 95% CI 1.20 to 1.80) ( Figure 8). ^{44, 45, 48} However, some concerns regarding risk of bias, combined with the high heterogeneity, result in low certainty of evidence for this finding. Thus, the decision on whether to perform uterine cavity evaluation with hysteroscopy before IVF should be carefully evaluated based on level of suspicion of abnormalities, such as intrauterine adhesions (IUA), polyps, or submucosal myomas (Suppl. Table 1, line no. 6).

Endometriosis

Women with endometriosis, with and without RIF, showed no significant difference in the odds of live birth versus those without endometriosis (OR 1.22, 95% CI 0.99 to 1.50) ( Figure 9). ^{49, 50} This may also be due to the small number of studies, which additionally have some concerns regarding the risk of bias. Therefore, the evidence for an effect of endometriosis has been rated as very low (Suppl. Table 1, line no. 7).

Chronic endometritis

Chronic endometritis (CE) causes a noticeable decline in the likelihood of live birth. Specifically, based on our analysis, live birth odds for women with CE (RIF and non-RIF) are roughly one-fifth less compared to those without the condition (OR 0.21, 95% CI 0.10 to 0.44) ( Figure 10). ^{51– 53} While meta-analyses of the general IVF population have demonstrated this effect, data specifically addressing patients with RIF are lacking. Furthermore, the comparability of available studies is impaired by the variability of the diagnostic criteria, since there is no agreement yet in both the definition or the specific number of plasma cells required for diagnosis. ⁵⁴ For these reasons, along with the small sample size, the quality of the evidence has been downgraded to moderate, despite the very strong association detected (Suppl. Table 1, line no. 8).

Vaginal microbiome

Women with a dysbiotic microbiome, e.g., bacterial vaginosis, have lower odds of achieving a clinical pregnancy than women with an eubiotic microbiome (OR 0.45, 95% CI 0.22 to 0.92) ( Figure 11). ^{55– 57} The high heterogeneity, combined with the small sample size, led to the certainty of the evidence being downgraded to low, despite the strong association observed (Suppl. Table 1, line no. 9).

Hypothyroidism

Hypothyroidism, both clinical (overt) and subclinical, may impact endometrial embryo implantation in IVF cycles. Our analysis showed that it significantly reduced the CPR in infertile patients (OR 0.74, 95% CI 0.60 to 0.91) ( Figure 12). ^{58, 59} When distinguishing cases of subclinical hypothyroidism with and without thyroid antibodies, no differences in implantation outcomes were found. The analysis included only two studies ^{58, 59} with a low risk of bias, due to thorough adjustment for confounders, systematic classification of interventions, adherence to protocols, and comprehensive reporting. Despite the findings, additional well-structured studies are needed to confirm these results, specifically in the setting of embryo transfer, with and without prior implantation failure. Some additional concerns about risk of bias result in the certainty of the evidence being downgraded to low (Suppl. Table 1, line no. 10).

Endometrial receptivity

The Endometrial Receptivity Analysis (ERA) test, as well as other tests developed based on gene sequence codification through IVF cycles, rely on the idea that certain key factors are consistently expressed together over time in a systematic manner. However, studies have shown that this procedure, aimed at identifying an endometrial window of implantation (WOI), is not effective in improving the implantation rates in general IVF population. ^{60– 62} The most convincing perspective is that the WOI is defined only by successful implantation. ⁶³ Indeed, the number of implantation failures is not relevant to evaluate the WOI because the physiology and pathology of this reproductive step remain the same, regardless of the numbers of attempts. The ERA test conducted in women undergoing IVF/ET or frozen embryo transfer (FET) cycles did not result in a significant improvement in the CPR (OR 0.85, 95% CI 0.69 to 1.04) ( Figure 13). ^{60, 64, 65} Due to the width of the CI, and some concerns about the risk of bias, the certainty of the test’s effectiveness was assessed as low (Suppl. Table 1, line no. 11). High-quality studies are needed to evaluate the efficacy of endometrial receptivity arrays in RIF patients specifically.

Meta-analyses of treatments of recurrent implantation failure

Assisted hatching

In our analysis, women with RIF undergoing assisted hatching (AH) had no significantly higher odds of clinical pregnancy (OR 0.95, 95% CI 0.58 to 1.58) ( Figure 14). ^{66– 71} The pooled OR is slightly below 1; however, the wide CI and the small sample size, lead to assessment of the effect of AH on clinical pregnancy as very low (Suppl. Table 1, line no. 12). Our results are in line with recent evidence suggesting that AH does not increase CPR or LBR in patients with RIF, ^{20, 68, 72, 73} highlighting the need for further robust studies in patients with RIF.

Anticoagulant treatment

The meta-analysis of limited number of studies investigating the use of heparin in IVF treatments for women with RIF, with or without acquired or inherited thrombophilia, reported significant findings regarding LBR. More specifically, heparin was found to increase the odds of live birth (OR 3.05, 95% CI 1.37 to 6.82) ( Figure 15). ^{74, 75} Such strong association was observed in a sample of fewer than 200 subjects from two heterogeneous studies, one of which is at high risk of bias. For this reason, the certainty of the evidence for the result must be considered very low (Suppl. Table 1, line no. 13). Further studies are needed to confirm this conclusion in clearly delineated patient populations, including RIF, before clinical recommendations for use.

Platelet-rich plasma

Platelet-rich plasma (PRP) intrauterine infusion therapy has been shown to significantly enhance implantation rates in RIF patients, especially following transfers of euploid embryos. Our analysis indicates that PRP treatment can improve the odds of clinical pregnancy in RIF patients (OR 3.48, 95% CI 2.47 to 4.89) ( Figure 16). ^{76– 83} This result was obtained from a subset of four randomized studies with low risk of bias. The subgroup analysis of studies with moderate risk and/or non-randomized studies is in agreement with these findings. Considering the strong association observed, there is a high degree of certainty in the evidence in favor of the positive effect of PRP treatment in RIF (Suppl. Table 1, line no. 14). However, well designed RCTs, particularly those including euploid embryo transfers, are needed to validate this treatment for clinical use. Furthermore, the risk of PRP treatment needs to be thoroughly validated.

Peripheral blood mononuclear cells

Intrauterine infusion of peripheral blood mononuclear cells (PBMCs) in RIF patients resulted in higher odds of clinical pregnancy compared to control groups (OR 2.28, 95% CI 1.51 to 3.43) ( Figure 17). ^{84– 90} This result has been observed in a small subgroup of randomized studies with low risk of bias, but also with limited sample size, and in presence of heterogeneity (Suppl. Table 1, line no. 15). Thus, more properly sized studies are needed before any robust conclusions can be drawn regarding PBMCs. Further risk profiles are not fully validated.

Granulocyte colony-stimulating factor

In RIF patients, administering granulocyte colony-stimulating factor (G-CSF) before embryo transfer has reportedly resulted in an increase in clinical pregnancy rates (OR 2.09, 95% CI 1.48 to 2.97) ( Figure 18). ^{91– 98} However, our finding is impaired by the heterogeneity between studies in terms of doses, routes, and timing of G-CSF administration. Despite a strong positive association observed across a set of eight randomized studies, the certainty of the evidence has been downgraded to moderate (Suppl. Table 1, line no. 16). Therefore, this intervention cannot be recommended.

Endometrial scratching

Based on our analysis, endometrial scratching significantly increased the odds of clinical pregnancy in women with RIF (OR 2.19, 95% CI 1.67 to 2.87) ( Figure 19). ^{99– 107} The result are based on a considerable number of non-randomized studies with a moderate risk of bias and on subgroup analysis of RIF patients in one RCT with a low risk of bias. ¹⁰⁶ Therefore, the certainty of the evidence regarding these results on endometrial scratching is to be considered high (Suppl. Table 1, line no. 17). However, its effectiveness remains debated. Further well-conducted RCTs in this specific population are needed to establish definitive conclusions.

Antibiotics for chronic endometritis

The pooled odds ratio obtained by comparing the CPR of women with CE treated with antibiotics to those without CE was higher than 1 (OR 1.19, 95% CI 0.90 to 1.58) and was not reaching statistically significant ( Figure 20). ^{108– 111} This result, combined with the previous findings on the negative effect of CE, supports the effectiveness of antibiotics in its treatment. However, due to some concerns regarding the risk of bias, the certainty of the evidence for a significant difference between the two groups has been considered low (Suppl. Table 1, line no. 18). Further well-structured RCTs are needed to confirm this finding.

Below are additional factors resulted from the systematic review for which we did not find studies suitable for meta-analysis.

Lifestyle factors such as alcohol consumption, smoking, and stress affect fertility and should be addressed during consultations. Yet, their exact impact on implantation success, particularly in cases of repeated failures, is not fully understood due to limited research. Our systematic review did not find studies on the effects of lifestyle factors that were suitable for a meta-analysis, likely because these factors are being eliminated by patients at the beginning of their conception journey. Smoking is known to significantly reduce fertility and increase health risks for mothers and infants. It is linked to decreased number of retrieved oocytes, higher rates of IVF cycle cancellations, and lower implantation rates. ^{170– 173} Alcohol is a recognized risk factor for stillbirth and other adverse pregnancy outcomes. Couples trying to conceive are generally advised to avoid alcohol to improve their chances of a successful healthy pregnancy. ^{170– 177} High stress levels, indicated by elevated cortisol, have been associated with a significantly increased risk of miscarriage. ¹⁷⁸ However, the impact of stress on IVF outcomes is less clear, with some studies showing no significant effect on the success of the first IVF cycle, while others suggest that stress from previous IVF failures may increase future risks. ^{179, 180} We did not find specific data for patients with RIF in relation to the above-mentioned lifestyle factors.

The use of FET cycles is on the rise, with varying endometrial preparation methods like programmed (medicated, hormone replacement therapy [HRT], controlled), natural and modified natural cycles, and stimulated cycles with letrozole or gonadotropins, each having different implications for pregnancy outcomes. Programmed FET cycles have been associated with higher risks of hypertensive disorders, pre-eclampsia, and postpartum hemorrhage, highlighting the potential need of the corpus luteum to mitigate these risks ¹⁸¹ ; a large multicenter RCT is currently being conducted to elucidate this issue. ¹⁸² Our literature search identified only two RCTs on success rates of FET vs fresh transfers in women with RIF, both of which had a critical risk of bias and questionable randomization procedures. ^{183, 184} In both studies, fresh day six blastocyst transfer was included in the fresh transfer group. In Pontré et al. cleavage-stage biopsy for PGT-A was performed. Furthermore, the study included the same women more than once. ¹⁸³ In Magdi et al., the randomization procedure was not transparent ending with an unequal number of women in each group. ¹⁸⁴ Due to these limitations, we deemed these studies unsuitable for meta-analysis. A recent large multicenter retrospective cohort study analyzing over 8,000 first autologous blastocyst transfers without PGT-A found comparable live birth rates, clinical pregnancy rates, and miscarriage rates between fresh and frozen transfers after adjusting for multiple confounders. ¹⁸⁵ More research is needed before firm conclusions can be drawn. Deciding between fresh and frozen transfers should account for specific patient factors and conditions. Fresh cycles might impair endometrial morphology and receptivity due to elevated progesterone levels from the day of ovulation trigger, potentially leading to dyssynchronous endometrial and embryo development and lower implantation rates. ^{186– 188} Sequential cleavage and blastocyst transfers have also been proposed for RIF patients with a good number of embryos. ^{189– 191} Further high-quality studies are needed to confirm this advantage.

Although numerous studies suggest a potential link between endometrial thickness (EMT) and implantation success, conclusive evidence is lacking. ^{192, 193} Therefore, EMT cannot yet be reliably established as an evidence-based criterion for determining endometrial receptivity before embryo transfer. In natural cycles, specific endometrial characteristics are essential for successful implantation. ¹⁹⁴ The goal of endometrial preparation in ART is to replicate these conditions. RCTs have shown that an endometrial thickness of ≥9 mm on the day of hCG administration correlates with higher pregnancy rates, with an optimal range between 8 and 14 mm. ^{138, 195} Women with an endometrial thickness between 10 and 12 mm have shown the highest pregnancy rates. However, these studies are not without limitations. some include too many uncontrolled variables in their inclusion and exclusion criteria. Other studies, despite improved controls, provide only suggestive findings. ^{164, 196– 198} The association between EMT and RIF is challenging to analyze, largely due to the evolving and inconsistent definitions of RIF. ^{73, 114} Historically, RIF has been defined more by expert consensus than by a solid pathophysiological framework.

Ideally, the role of EMT in successful embryo implantation should be analyzed in a sample of patients undergoing euploid embryo transfers, excluding other confounding factors. Despite numerous studies suggesting that EMT has a role in implantation success, it remains uncertain to what degree EMT is a clinically relevant prognostic parameter.

Progesterone is essential for transforming the estrogen-primed endometrium into secretory endometrium, crucial for embryo implantation and pregnancy maintenance. ¹⁶⁰ Its absence or deficiency can impair implantation and increase miscarriage risks. The correlation between serum and endometrial progesterone levels is complex, not linear and is affected by factors like BMI and smoking. ¹⁷² These differences might be influenced by factors such as microcirculation, the balance between free and protein-bound progesterone, and variations in the responsiveness of epithelial cells to progesterone. The role of progesterone and its serum levels in predicting pregnancy success even varies across different cycle types. ^{3, 159, 160, 165, 167, 199– 205}

Hyperinsulinemia, insulin resistance and obesity are associated with diminished uterine receptivity, potentially influenced by altered endometrial gene expression. ^{206, 207} Obese women present a different endometrial gene expression profile during the WOI, which is more pronounced if associated with infertility or PCOS. ²⁰⁸ Treatment strategies like metformin and lifestyle modifications may not consistently improve LBR. ²⁰⁹

Conditions such as antiphospholipid syndrome and subsequent anticoagulation are considered in diagnostic assessments, but many factors influencing successful implantation remain unproven. ^{16, 210}

Progesterone influences endometrial features like natural killer cells, which are significant for embryo implantation success. However, the clinical ability and utility of testing for endometrial natural killer cells is currently limited. ²¹¹

Current evidence suggests that additional (adjuvant) treatments could prevent RIF more cost-effectively than multiple IVF cycles. However, not all proposed treatments have robust evidence supporting their efficacy. ²¹² In our systematic review and meta-analysis of treatments for RIF, we focused on those with the most compelling evidence in the literature ( Table 3).

Although there is some evidence that AH has a positive effect on implantation ( Figure 14), routine use in RIF patients is not recommended due to potential risks and lack of consensus. ^{66, 213– 215} A previous systematic review found that AH did not improve CPR, ²⁰ whereas other studies not included in the systematic review found that it increased pregnancy rate. ^{66, 70, 216, 217} ESHRE does not recommend AH for RIF patients, nor as an add-on in general. ^{73, 210}

Heparin appears to be effective in improving live birth in RIF women with coagulation disorders like antiphospholipid syndrome and thrombophilia ( Figure 15). However, although no significant heterogeneity presence was detected, it is worth mentioning that the two studies included evaluated different coagulopathies, as well as slightly different treatments, and were assessed as having a very serious bias. ^{74, 75} In the setting of RIF, the use of aspirin is not supported by sufficient studies. ^{218– 220}

Intrauterine infusion of PRP has demonstrated potential in improving EMT before FET and increasing CPR in RIF patients ( Figure 16). It is known for promoting cell migration and proliferation and possibly increased expression of receptivity markers. ²²¹ Although based on the four RCTs analyzed, the GRADE approach led to strong recommendations, more large-scale well-defined controlled studies are needed to confirm the effectiveness and long-term safety of PRP. ^{222– 224}

Intrauterine infusion of PBMCs enhanced EMT and receptivity, resulting in significant increases in implantation and LBR in RIF patients when used in frozen-thawed cycles ( Figure 17). ^{86, 89} Further studies are needed to validate these findings as only RCTs with very modest sample sizes were included.

As shown in Figure 18, G-CSF may improve CPR in both fresh and FET cycles. However, results varied depending on the route, dosage and time of administration, and the RCTs included reported contradictory results. ^{91, 98, 225, 226}

Despite ongoing criticism and lack of robust evidence, endometrial scratching is a common practice in some clinical settings. While its utility in all-comers is not recommended, ²²⁷ the data for RIF is not as straightforward. ²²⁸ It is believed that this procedure induces decidualization and prepares the endometrium for implantation by increasing cytokine production, potentially enhancing synchronization between the endometrium and the embryo. Our analysis found that endometrial scratching may improve IVF outcomes in women with RIF. Nevertheless, this result should be taken with caution since several biases arose from the evaluation of the 11 articles included. Therefore we cannot recommend this procedure for clinical management of RIF patients at this point. ^{114, 229– 232}

Antibiotics are effective in treating CE, a common issue among RIF patients. As shown in Figure 20, patients with resolved CE after antibiotic treatment exhibit increased pregnancy and LBR compared to those with persistent CE. ^{111, 142, 233– 235} However, significant biases were identified in the analyzed studies, and the GRADE analysis supports only weak recommendations with low-quality evidence.

Hysteroscopy and 3D Ultrasonography are crucial for evaluating factors critical to effective embryo implantation. ²³⁶ Hysteroscopy is particularly useful to diagnose and treat uterine conditions, allowing for the removal of lesions and restoration of the uterine cavity. Based on low quality evidence, corrective procedures for uterine abnormalities like removal of polyps, fibroids, and scar tissue have shown to improve LBR and reduce miscarriage rates, particularly in patients with primary infertility and recurrent miscarriage ( Figure 8), which is in agreement with The ESHRE Good Practice Recommendations on RIF. ⁷³ However, diagnostic hysteroscopy in RIF patients with normal uterine findings has shown no benefits. ^{48, 237– 242}

Below are additional treatments resulted from the systematic review for which we did not find studies suitable for meta-analysis.

Immunotherapy and Immunomodulators are being investigated with the aim of enhancing maternal-fetal immune tolerance. Treatments such as intravenous immunoglobulin, tacrolimus, and cyclosporine have shown potential benefits but lack sufficient evidence and risk assessment for routine use. ^{243– 250}

Intralipids have shown potential in modulating natural killer cell activity and increasing LBR in RIF patients, while Lymphocyte Immunization Therapy has been explored for its ability to modulate the maternal-fetal immune balance. However, both lack strong support for routine use, ^{251– 258} and risk/benefit ratio should be strongly considered.

Endometrial ultramorphology evaluation can identify specific endometrial features, such as pinopodes, which are morphological markers of a receptive endometrium and can help determine the optimal phase for embryo transfer. ^{259, 260} Unfortunately, their evaluation is quite cumbersome.

Among patients with RIF, treatment with corticosteroids (prednisone) did not improve LBR compared with placebo. Data suggested that the use of prednisone may increase the risk of preterm delivery and biochemical pregnancy loss. ²⁶¹ Our results challenge the value of prednisone use in clinical practice for the treatment of RIF.

We focused on the above-mentioned treatments because they have been proposed for RIF patients. This review discusses the complex challenge of managing RIF, contributing to the understanding of potential and limitations in clinical applications of diagnostic procedures and treatments. The results highlight the need for continued research in order to improve pregnancy outcomes.

The biggest hurdle for research efforts is the lack of established diagnostic criteria on what exactly qualifies for RIF, and hence range of definitions used in various studies. Despite advancements in clinical procedures and embryology labs, RIF remains a significant issue. The various influencing factors, including immunological, genetic, anatomical, hematologic, microbial, and endocrine aspects are yet to be fully understood.

While providing an overview of infertility with a focus on embryo implantation and RIF, this review does not comprehensively cover all aspects of infertility management. Additionally, many studies discussed herein lack high-quality data. Furthermore, for the meta-analysis, we considered CPR or LBR. These outcomes might underestimate the actual implantation rates. Further the risk profiles of many of the suggested treatments for RIF are not sufficiently validated. Finally, the search was restricted to English language publications, which may exclude pertinent studies written in other languages.