Uterine Septum with or without Hysteroscopic Metroplasty: Impact on Fertility and Obstetrical Outcomes—A Systematic Review and Meta-Analysis of Observational Research

Objective: we performed a systematic review/meta-analysis to evaluate the impact of septate uterus and hysteroscopic metroplasty on pregnancy rate-(PR), live birth rate-(LBR), spontaneous abortion-(SA) and preterm labor (PL) in infertile/recurrent miscarriage-(RM) patients. Data sources: a literature search of relevant papers was conducted using electronic bibliographic databases (Medline, Scopus, Embase, Science direct). Study eligibility criteria: we included in this meta-analysis all types of observational studies that evaluated the clinical impact of the uterine septum and its resection (hysteroscopic metroplasty) on reproductive and obstetrics outcomes. The population included were patients with a diagnosis of infertility or recurrent pregnancy loss. Study appraisal and synthesis methods: outcomes were evaluated according to three subgroups: (i) Women with untreated uterine septum versus women without septum (controls); (ii) Women with treated uterine septum versus women with untreated septum (controls); (iii) Women before and after septum removal. Odds ratios (OR) with 95% confidence intervals (CI) were calculated for the outcome measures. A p-value < 0.05 was considered statistically significant. Subgroup analysis was performed according to the depth of the septum. Sources of heterogeneity were explored by meta-regression analysis according to specific features: assisted reproductive technology/spontaneous conception, study design and quality of papers included Results: data from 38 studies were extracted. (i) septum versus no septum: a lower PR and LBR were associated with septate uterus vs. controls (OR 0.45, 95% CI 0.27–0.76; p < 0.0001; and OR 0.21, 95% CI 0.12–0.39; p < 0.0001); a higher proportion of SA and PL was associated with septate uterus vs. controls (OR 4.29, 95% CI 2.90–6.36; p < 0.0001; OR 2.56, 95% CI 1.52–4.31; p = 0.0004). (ii) treated versus untreated septum: PR and PL were not different in removed vs. unremoved septum(OR 1.10, 95% CI 0.49–2.49; p = 0.82 and OR 0.81, 95% CI 0.35–1.86; p = 0.62); a lower proportion of SA was associated with removed vs. unremoved septum (OR 0.47, 95% CI 0.21–1.04; p = 0.001); (iii) before-after septum removal: the proportion of LBR was higher after the removal of septum (OR 49.58, 95% CI 29.93–82.13; p < 0.0001) and the proportion of SA and PL was lower after the removal of the septum (OR 0.02, 95% CI 0.02–0.04; p < 0.000 and OR 0.05, 95% CI 0.03–0.08; p < 0.0001) Conclusions: the results show the detrimental effect of the uterine septum on PR, LBR, SA and PL. Its treatment reduces the rate of SA.


Introduction
Congenital uterine anomalies (CUAs) are estimated to affect 1% to 4% of the general fertile population [1], although the reported prevalence is significantly higher (8.5-12%) in patients with infertility and recurrent pregnancy loss [2]. The uterine septum is the most common CUA, representing 35% of all of the diagnosed malformations [3,4].
The definition of uterine septum is still a subject of debate. Two main classification systems were developed: the ASRM (AFS) classification initially published in 1988, which was subsequently modified in 2016, and the ESHRE-ESGE classification [5][6][7]. It is accepted that both classification systems have some limitations. In particular, ESHRE-ESGE classification may overestimate the prevalence of septate uterus while that of ASRM may underestimate it, leaving in a gray-zone most of the uteri that could considered as septate. To address this important clinical dilemma, a novel cut-off of >10 mm indentation depth was proposed by the Congenital Uterine Malformation by Experts (CUME) in 2018 [8].
Whether the uterine septum increases the risk of reproductive failure is still uncertain. On the one hand, several observational studies reported an association between the uterine septum and obstetrical complications including recurrent spontaneous abortions (both in the first and second trimester), preterm delivery, intrauterine growth restriction (IUGR) and placental abruption [3,[9][10][11]. Moreover, some authors suggest that hysteroscopic metroplasty of the septum can reduce miscarriage rates and improve the obstetrical outcomes in patients with a history of recurrent miscarriage [1,12,13]. On the other hand, a few studies support a causal effect of the uterine septum on infertility, [13][14][15][16], as demonstrated by an improvement of implantation rates after septum removal [12,[17][18][19].
According to recent studies, the uterine septum may impair the embryo implantation and development through both molecular and mechanical mechanisms. Lower expression of the homeobox protein Hox-A10 (HOXA10) and Vascular Endothelial Growth factor (VEGF) receptor genes, lower number of glandular and ciliated cells in the endometrial lining of the intrauterine septum and uncoordinated uterine contractility (due to increased content of muscle tissue within the septum) are supposed to be the main factors [20,21].
An important effort in summarizing the published data on CUAs and reproductive issues was given by Venetis et al. [13]. Their review found an association among the presence of CUAs and a reduction in clinical pregnancy rates, an increase in the first-or second-trimester spontaneous abortion rates, and preterm delivery rates. However, the study reported by Venetis et al. included all types of uterine anomalies, which limits its clinical significance when limited to the management of the uterine septum [13]. The first randomized controlled trial on hysteroscopic septum resection in women with reproductive disorders was recently published [22]. In this study, Rikken et al. found no improvements in reproductive outcomes from the intervention, thereby questioning any rationale supporting surgical management. Nevertheless, the study by Rikken et al. has been strongly criticized, recognizing several limitations that preclude the interpretation of the data, including the small sample size (n = 68 women) and heterogeneity in terms of patients' characteristics and reproductive disorders. Therefore, the debate on the impact of the uterine septum and its surgical correction on reproductive outcomes is still open.
Over the above-described controversial background, the aim of our systematic review was to summarize evidence from observational studies on the impact of the septate uterus on pregnancy rate, live birth rate, first-or second-trimester spontaneous abortion rates and preterm delivery rates in women suffering from infertility or recurrent miscarriage. We also evaluated the impact of hysteroscopic metroplasty on the same clinical outcomes.

Study Design and Protocol Registration
This meta-analysis was conducted following Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines [23]. The research study, eligibility criteria, study selection and data extraction process were defined a priori. The review protocol was recorded on the international prospective register of systematic reviews PROSPERO (registration ID: CRD42020196157) before starting the literature search.

Eligibility Criteria
We included in this meta-analysis all types of observational studies that evaluated the clinical impact of the uterine septum and its resection (hysteroscopic metroplasty) on reproductive and obstetrics outcomes. We excluded from the analysis narrative or systematic reviews and case reports.
The definition of uterine septum, the diagnostic criteria (according to guidelines) and the diagnostic methods varied among papers; a detailed description is reported in Tables 1-3.
The inclusion criteria applied in this meta-analysis were the following: • Type of study: Case-control studies, cohort studies or case series. • Period of publication: no restriction. • Language: English. • Participants: Women diagnosed with infertility or recurrent pregnancy loss (categories were intended as classified by authors). See Tables 1-3 for details. • Comparators: (i) Women with untreated uterine septum versus women without septum (controls); (ii) Women with treated uterine septum versus women with untreated septum (controls); (iii) Women before and after septum removal. • Outcomes: Pregnancy rate; live birth rate; spontaneous abortion rate; preterm delivery rate. • Outcome definitions: Pregnancy rate (PR-defined as the presence of a gestational sac on transvaginal ultrasound; Live birth rate (LBR-defined as the delivery of one or more living and viable infants). Spontaneous abortion rate (SA-defined as fetal loss prior to the completed 23th week of gestation; Preterm labour (PL-defined as a delivery before 37 weeks of gestation).

Information Sources and Search Strategies
A literature search of relevant papers was conducted using the electronic bibliographic databases (Medline, Scopus, Embase, Science direct, Cochrane library, Clinicaltrials.gov, Cochrane Central Register of Controlled Trials, EU Clinical Trials Register and World Health Organization International Clinical Trials Registry Platform). The strategies for electronic search were the following combined search: Uterine septum OR Septate uterus OR Metroplasty AND Pregnancy rate OR Live birth rate OR Spontaneous miscarriage OR infertility OR preterm delivery.       The resection was performed between the anterior and posterior uterine walls, extending up to the fundus rather than into the fundal myometrium from the lower margin of the septum 3D TV-US AFS (1988) criteria [5]

Study Selection
Titles and/or abstracts of studies retrieved using the electronic search strategy and those from additional sources were screened independently by two review authors (GS, MM) to identify studies that potentially meet the inclusion criteria outlined above. The full text of these potentially eligible studies was retrieved and independently assessed for eligibility by other two review team members (MN, GB). Any disagreement over the eligibility of a study was resolved through discussion with a third external collaborator (AV). A standardized, pre-piloted form was used to extract data from the included studies for assessment of study quality and evidence synthesis. According to this form, we extracted data about studies characteristics (design and time of the study), population (number of enrolled women, general characteristics, patient's selection, definition of septum (ASRM, ESHRE-ESGE and CUME classification) [5][6][7][8], diagnostic technique (3D transvaginal ultrasound and/or hysteroscopy), depth of septum (partial or complete), metroplasty technique performed and specific outcomes including pregnancy rate and live birth rate [LBR from spontaneous conception or after assisted reproductive technology (ART)], spontaneous abortion and preterm labor.
A manual search of a reference list of included studies was also performed in order to avoid missing relevant data. We searched for published (full-text studies and meeting abstracts) and unpublished studies (i.e., for whom only a registered protocol was available) from the aforementioned electronic databases. The results were compared, and any disagreement was resolved by consensus.

Risk of Bias
Two reviewers (G.S. and G.B.) independently judged the methodological quality of the studies included in the meta-analysis. Outcome's selection and measurement was assessed for each distinct outcome. Disagreements between the reviewers over the risk of bias of any study were solved by discussion with a third review author (M.N).
The risk of bias was be assessed for each study using the methodological index for non-randomized studies (MINORS) [59]. The tool consists in 12 items, the first eight are methodological items for non-randomized studies and the last four items are additional criteria to apply in the case of comparative study. Each item is scored as 0 (not reported), 1 (reported but inadequate) or 2 (reported and adequate). The global ideal score is 16 for non-comparative studies and 24 for comparative studies. For comparative studies we considered a paper of high quality with a score of 23-24, medium quality with a score of 21-22 and low quality with a score ≤20. For non-comparative studies we considered a paper of high quality with a score of 15-16, medium quality with a score of 13-14 and low quality with a score ≤ 12.

Statistical Analysis
The data analysis was performed by two authors (F.C. and M.N.) using "metafor" package for R software, version 4.0 (R Foundation for Statistical Computing, Vienna, Austria) [60,61]. A p-value less than 0.05 was considered statistically significant. Odds ratios (OR) with 95% confidence intervals (CI) were calculated for the outcome measures in each study. Meta-analyses were performed using random-effects models. Each outcome was analyzed independently. Heterogeneity was assessed using the I 2 value and potential sources of heterogeneity were explored using meta-regression; an I 2 value over 50% indicated substantial heterogeneity [62]. The risk of publication bias was assessed using funnel plots and the trim-and-fill method [63]. Of note, this method was used to suggest possible missing studies, but not to provide adjusted estimates (those resulting from a meta-analysis including the filled studies) [64].
Subgroup analysis was performed according to the depth of the septum (partial or complete) including the following population: women diagnosed with infertility and women diagnosed with recurrent pregnancy loss. Sources of heterogeneity were explored by meta-regression analysis according to the features of the included studies: assisted re-productive technology/spontaneous conception, study design (prospective/retrospective) and quality of papers included (low/medium/high). Sensitivity analyses including only medium and high-quality studies was also performed.

Spontaneous Abortion during the First Trimester of Pregnancy
SA during the first trimester of pregnancy was investigated in five studies [14,26,28,32,33]. A higher proportion of SA was associated with septate uterus vs. controls (OR 3.15, 95% CI 1.59 to 6.2; p = 0.001) with substantial heterogeneity (I 2 = 78%). The influence of study design was not statistically significant (p = 0.06), while study quality (p = 0.21) did not change the effect size.
The influence of spontaneous conception vs assisted conception (p = 0.06) was not statistically significant, while septum classification (p = 0.43) and study quality (p = 0.41) did not change the effect size. Overall, the funnel plot was symmetrical and the trim-and-fill method by Duval and Tweedie did not suggest any missing studies (Supplementary Figure S1).

Spontaneous Abortion during the First Trimester of Pregnancy
SA during the first trimester of pregnancy was investigated in five studies [14,26,28,32,33]. A higher proportion of SA was associated with septate uterus vs. controls (OR 3.15, 95% CI 1.59 to 6.2; p = 0.001) with substantial heterogeneity (I 2 = 78%). The influence of study design was not statistically significant (p = 0.06), while study quality (p = 0.21) did not change the effect size.

Spontaneous Abortions during the Second Trimester of Pregnancy
SA during the second trimester of pregnancy was investigated in four studies. A higher proportion of SA was associated with septate uterus vs. controls (OR 2.69, 95% CI 1.11 to 6.52; p = 0.03) with moderate heterogeneity (I 2 = 56%) [14,26,28,32]. The influence of study quality was not statistically significant (p = 0.06).

Preterm Labor
Preterm labor was investigated in three studies [14,26,28]. A higher proportion of preterm labor was associated with septate uterus vs. controls (OR 2.56, 95% CI 1.52 to 4.31; p = 0.0004) with low heterogeneity (I 2 = 0%) ( Figure 2D). An analysis of moderators could not be performed due to the small sample size.
Overall, the funnel plot was asymmetrical and the trim-and-fill method by Duval and Tweedie suggested four missing studies (with sample sizes of 50, 185, 255 and 263 pregnancies, and ORs of 3.67, 5.31, 2.15 and 1.23, respectively) in the left area of the plot (Supplementary Figure S2).
Overall, the funnel plot was asymmetrical and the trim-and-fill method by Duval and Tweedie suggested two missing studies (with sample sizes of 255 and 263 pregnancies, and ORs 0.42 and 0.94, respectively) in the right area of the plot (Supplementary Figure S3).

Sub-Analyses
In five studies reporting data from infertile patients, SA was lower after vs. before the removal of the uterine septum (OR 0.04, 95% CI 0.01 to 0.09; p < 0.0001) with moderate heterogeneity (I 2 = 23%).
In seven studies reporting data from recurrent abortion, SA was lower after vs. before the removal of the uterine septum (OR 0.03, 95% CI 0.01 to 0.06; p < 0.0001) with substantial heterogeneity (I 2 = 78%).

Sensitivity Analysis
In six studies with medium/high quality, SA was lower after vs. before the removal of the uterine septum (OR 0.02, 95% CI 0.01 to 0.03; p < 0.0001) with moderate heterogeneity (I 2 = 50%).
Excluding data from three abstracts, SA was lower after vs. before the removal of the uterine septum (OR 0.02, 95% CI 0.02 to 0.03; p < 0.0001) with moderate heterogeneity (I 2 = 54%).
Overall, the funnel plot was slightly asymmetrical and the trim-and-fill method by Duval and Tweedie suggested one missing study (with sample sizes of 29 pregnancies and OR of 0.01) in the left area of the plot (Supplementary Figure S4).

Subgroup-Analyses
In two studies reporting data from infertile patients, preterm labor was not different after vs. before the removal of the uterine septum (OR 11.61, 95% CI 0.05 to 2702.21; p = 0.38) with substantial heterogeneity (I 2 = 84%).
In three studies reporting data from recurrent miscarriages, preterm labor was lower after vs. before the removal of the uterine septum (OR 0.03, 95% CI 0.01 to 0.09; p < 0.0001) with low heterogeneity (I 2 = 0%).

Sensitivity Analysis
In two studies with medium/high quality, preterm labor was lower after vs. before the removal of the uterine septum (OR 0.03, 95% CI 0.02 to 0.08; p < 0.0001) with low heterogeneity (I 2 = 0%).
Excluding data from three abstracts, preterm labor was lower after vs. before the removal of the uterine septum (OR 0.04, 95% CI 0.02 to 0.08; p < 0.0001) with low heterogeneity (I 2 = 0%).

Surgical Complications
Considering the nine papers of the section "Treated uterine septum versus controls (untreated septum)" a total of eleven complications were reported (n = 11 out of 603 procedures; complication rate of 1.8%). The reported complications were five uterine perforations (only one requiring laparoscopic suturing); five abnormal uterine bleeding and only one fluid overload syndrome (Supplementary Table S1).
Considering the nineteen papers of the section "Before and after septum removal" a total of 30 complications were reported (n = 30 out of 1920 procedures; complication rate of 1.5%). The reported complications were 20 uterine perforations, 12 requiring laparoscopic (seven patients) or laparotomy (five patients) management, three abnormal uterine bleeding, one cervical laceration during dilatation, five fluid overload syndrome and one patient developed intrauterine synechiae (Supplementary Table S2).

Discussion
The septate uterus represents a clinical dilemma for the physician who is treating the patient affected with this enigmatic congenital uterine anomaly. Over the last decade, different studies have evaluated the correlation between uterine septum/subseptum (with/without hysteroscopic metroplasty) and reproductive outcomes (PR, LBR, spontaneous abortion, premature delivery). Unfortunately, a comprehensive updated summary of the current available evidence is missing. Two recent publications on this topic questioned the effectiveness of septum surgical correction to improve the patient's reproductive outcomes [39,40]. However, the above-mentioned studies conflict with two recently published position papers that reasonably rejected their findings, continuing to propose metroplasty as the treatment of choice in patients with a septate/subseptate uterus and a history of infertility or miscarriages [40,66]. Hence, based on current scientific uncertainties, we decided to perform a systematic review and meta-analysis in order to summarize the current scientific evidence regarding the role of uterine septum/subseptum and its surgical treatment on human reproduction and pregnancy outcome.

Uterine Septum versus Controls (No Septum)
Regarding the effects of uterine septum on reproductive outcomes (uterine septum versus controls), we found that SA (first and second trimester) and PL were more frequent in patients with uterine septum compared to those without septum; instead, PR and LBR were higher in patients with uterine septum compared to those without septum. However, according to the analysed outcome and our results, some considerations are crucial.
First, based on our analysis, we can affirm that the presence of a uterine septum could have a detrimental effect, especially on LBR, SA. These data were confirmed both on general analysis, subgroup analysis (infertile patients and patients with history of recurrent miscarriage) and sensitivity analysis (excluding low quality papers) with low heterogeneity. This trend was confirmed also in the case of PR and PL in general analysis. At subgroup analysis and sensitivity analysis, these data were not confirmed for PR and PL, even if the low number of included studies and the high heterogeneity precluded generating meaningful conclusions.
To the best of our knowledge, for the first time in a meta-analysis on this topic, detrimental effects on reproductive outcomes were confirmed also in women with uterine sub-septum for SA, LBR and PL. Regarding PR, even if our results were close to significance, the low number of patients with a uterine sub-septum did not allow for the obtaining of meaningful conclusions. Unfortunately, no data could be extracted about the role of the uterine sub-septum length even if we found no statistically significative differences of PR, LBR, spontaneous abortion and preterm delivery comparing patients with uterine septum versus sub-septum.
Our data are in line with two previously published systematic reviews of Venetis et al. and Chan et al. [13,67]. In the former study, the authors, although not distinguishing between partial and complete uterine septum, demonstrated the negative impact of a septate uterus on the ability to maintain pregnancy [13]. In the second, the authors established that the prevalence of uterine anomalies was higher in women with a history of miscarriage compared to the general population [67].
Different to previously published systematic reviews, we also focused our attention on specific subpopulations distinguishing, where possible, the modality of conception (spontaneous versus assisted reproductive technology), the population studied (women with a history of infertility versus recurrent miscarriage), and the trimester of the pregnancy in which the abortion occurred (first versus second trimester abortion).
When we analyzed the SA rate in patient with history of infertility versus recurrent miscarriage, it seems that the detrimental effect of the septum is slightly higher in the population of infertile patients than in those with recurrent abortions (OR 7.56 vs 2.29, respectively). We hypothesize that this finding could be due to the wide range of factors involved in the pathogenesis of recurrent miscarriage, like genetic and chromosomal abnormalities [68], infectious agents [69], hormonal causes, environmental toxicities and oxidative stress [70].
In our analysis we found a clear association between uterine septum and the increased probability of spontaneous abortion both in the first and second trimesters; however, the correlation was higher in the first trimester. This was highlighted in a paper published by Zlopaša et al. [28], who found a percentage of spontaneous abortion of 77.3 % during the first trimester versus 15.9% during the second trimester of pregnancy; in the same way, Saravelos et al. found that the percentage was 72.6 % versus 13.2 % during the first and second trimester of pregnancy, respectively [32].

Treated Uterine Septum versus Untreated Septum
Analyzing the nine papers that compared treated versus untreated uterine septum, we found that hysteroscopic metroplasty could significantly improve LBR on general analysis. This effect was enhanced considering medium/high quality studies. This data was not confirmed in the subgroup analysis (infertile versus recurrent miscarriage). Removal of the septum by hysteroscopy (hysteroscopic metroplasty) would appear to improve the outcome SA, being very close to statistical significance. Indeed, excluding low quality papers, SA was confirmed to be significantly lower in patients who had their uterine septum removed versus unremoved.
In terms of PR and PL, the hysteroscopic correction of septum would seem not to significantly improve the outcomes both in general analysis and in subgroup/sensitivity analysis. These results are in disagreement with previously reported data. However, we ignore if the lack of improvement of PR and preterm delivery rate after surgery could be related to a real inefficiency of the treatment or to the inconsistency of the trials currently available in the literature (high heterogeneity, poor number of high-quality studies). Indeed, Pang et al., one of the few prospective studies, demonstrated that women with a history of recurrent spontaneous abortion who underwent hysteroscopic metroplasty had higher rates of pregnancy and full-term delivery, and lower rates of spontaneous abortion and preterm delivery than women without metroplasty [37]. Conversely, in the recently published paper by Rikken et al. the hysteroscopic metroplasty did not improve reproductive outcomes compared to expectant management in terms of pregnancy rate, pregnancy loss and preterm birth [39]. It is important to highlight that in reference to this paper, a recent letter/position paper that underlined possible biases has been published; the authors strongly criticized the large number of centers from which the data were obtained, the inclusion in the study of patients not infertile at the time of enrollment, the lack of information about other causes of infertility, and the inclusion of multiple operators among other scientific mishaps (probably with heterogeneous techniques of hysteroscopic septum resection).
Recently, the first multicentric randomized trial on this topic was published (TRUST-The Randomised Uterine Septum Trial) [22]. As previously reported, it was methodologically impossible to aggregate this data in our meta-analysis. In this trial, the authors found no significant difference in live birth rate in patients allocated to septum resection versus expectant management (31% versus 15%-relative risk-95% CI 0.47 to 1.65). We certainly recognize the importance of this paper. However, we disagree with the final statement of the manuscript "in light of the lack of any evidence of effectiveness and the potential for harm, we recommend against septum resection as a routine procedure in clinical practice".
First, we must reassure the patients that office based hysteroscopic septum resection without cervical dilatation could be considered a safe procedure that can be performed without anesthesia or with local anesthesia, with a low rate of only minor complications. [71,72].
Secondly, we must underline the different limitations of this trial. As suggested also by Vercellini et al. the number of patients included and the stated effect size were too low and too high, respectively, to identify smaller but still clinically relevant differences between groups [73]. Another limitation is due to the type of patients included; data from patients with a history of subfertility, pregnancy loss or preterm birth were analyzed together, thus limiting the applicability of the results to a different subset of patients. The variability of septum definition and the non-uniformity of the diagnostic technique applied is another important limiting factor which can lead to significant selection bias. Another problem related to this trial's methodology is that if the outcomes were estimated 12 months after randomization (septum diagnosis), it is not clear if the follow-up was the same both in the expectant group and treatment group. Finally, the majority of patients showed a partial septum (without specifying septum length), so generalizing these results to the complete septum is in our opinion incorrect. In conclusion, we recognize the merits of this first randomized trial but, unfortunately, in our opinion, the quality of evidence presented is not adequate to change clinical practice based on decades of scientific literature.
In the 2018, the Cochrane Systematic Review Hysteroscopy was published about treating subfertility associated with suspected major uterine cavity abnormalities. Only two randomised studies were included, one concerning the hysteroscopic removal of submucous fibroids, the second about the hysteroscopic removal of endometrial polyps, and no randomised study with regard to the uterine septum. Therefore, we are unable to compare our results. The authors concluded that more research is needed to measure the effectiveness of the hysteroscopic treatment of suspected major uterine cavity abnormalities in women with unexplained subfertility [74].
A meta-analysis on this topic concluded that septum resection was associated with a lower rate of miscarriage compared with untreated women and no significant effect was seen on live birth, clinical pregnancy rate or preterm delivery [75]. Although interesting, this meta-analysis is incomplete considering the poor number of papers included (only seven) and the fact that all patients were evaluated together without distinction between infertile patients and patients with recurrent miscarriage. Moreover, no meta-regression analysis nor subgroup analysis was performed, thus limiting the clinical applicability of the results.
More recently, a systematic review and meta-analysis comparing reproductive outcomes between women undergoing hysteroscopic resection of the uterine septum and those with expectant management supported that hysteroscopic metroplasty was effective in reducing the risk of miscarriage in patients with complete or partial uterine septum. This study was similar with our results, but they only considered treated uterine septum versus untreated septum without evaluating the uterine septum versus controls (no septum) [76].

Before and after Septum Removal
In the last section of our meta-analysis, we included all observational non-comparatives studies that reported reproductive outcomes before and after septum removal in the same subset of the population. This part of the meta-analysis includes most of the papers, this type of trial being easier to perform than comparative observational studies. Interestingly, in this population we found a significant improvement post-resection in both LBR, SA rate and preterm delivery rate. This data was confirmed both in general analysis and also in subgroup analysis (infertile versus recurrent miscarriage) and sensitivity analysis (excluding low quality papers). The most recent revision of a before/after design studies dates from the year 2000 and is in agreement with our data [77]. Nevertheless, we recommend executing caution when interpreting the results from our last section due to the high heterogeneity of included papers, the presence of publication bias and the intrinsic limits related to before/after non-experimental study design.

Surgical Complications
The total complication rate reported ranged between 1.5% and 1.8%. Of these, the most common complication was uterine perforation (25 events) with or without abdominal surgical correction, followed by eight cases of abnormal bleeding and only six cases of fluid overload syndrome. Concerning uterine perforation, it is interesting to underline that of the 25 cases, 10 cases were reported by one study [12] that used a 26 French operative hysteroscope for metroplasty. This is a fact of considerable importance because the time range of the included studies is large, and so many of them were performed using what now would be considered obsolete medical equipment like the 26 French operative hysteroscope or monopolar energy that could explain the higher complication rate. Currently, recent advances in hysteroscopic technology have created miniaturized safer to use instruments with a diameter of less than 16 French and bipolar energy that are associated with a lower complication rate [71,72].

Biological Rationale
The definitive explanation for the correlation between the septate uterus and impaired reproductive potential is still unclear, although several biological mechanisms have been hypothesized. Accumulating evidence suggested that one of the main pathophysiological noxa could be represented by a different expression pattern of cytokines and inflammatory mediators in the endometrium surrounding the septum which that may lead to altered endometrial receptivity [20,21]. This theory could explain the results of similar impaired reproductive outcomes in patients with either uterine septum or sub-septum [78], supporting that the implantation in patients with uterine septum presented an alteration in the vascularization pattern of the decidua basalis. Histological analyses of the septum found the presence of more muscular tissue than fibroelastic connective tissue [79], therefore an increased content of muscle tissue could generate an increased and uncoordinated uterine contractility [20]. Furthermore, the endometrium located over the septum presented a reduced number of glandular ostia, irregular non-ciliated cells with rare microvilli, and a decreased ratio of the ciliated to non-ciliated cells [78,80]. This may imply a decrease in the response to preovulatory hormonal changes of the endometrium covering the uterine septum which could play a role in the pathogenesis of primary infertility [81,82]. Recent advances also demonstrated a lower expression profile of HOXA genes in the endometrium of patients with a septate uterus; this may be of paramount importance considering that different studies underlined the important role of HOXA genes for the proper development of the female genital tract, including the endometrium and for endometrial receptivity. [21].
Finally, during the second trimester of pregnancy, the space requirements of the developing fetus increase, so the risk of spontaneous abortion and preterm birth could be associated to the reduction of the intrauterine space in patients with a septate uterus [13].

Strengths and Limitations
This meta-analysis critically evaluated all of the current available literature about the uterine septum that has been published over the last 20 years, representing the best available evidence on the implications and the management of septate uterus, a topic that is currently highly controversial. Different from previous systematic reviews and meta-analyses [13,67], we focused our attention on the septate/subseptate uterus in order to provide updated answers to clinically relevant questions on the implications of septate uterus and the potential benefit of hysteroscopic treatment of the uterine septa by using a strictly scientific evidence-based approach. For the first time in the literature, we included data from both comparative (septum versus no septum and treated versus untreated septum) and non-comparative studies (before and after septum removal) providing useful information from different patient populations (recurrent abortion and infertile patients). However, results from before-and-after study designs may have some limitations. We cannot exclude that some other influential events, which could affect the outcome, occurred after the intervention during the follow-up; moreover, a change in outcome measure might be explained by a group with a one-time extreme value naturally changing towards a normal value and, finally, the placebo effect cannot be eliminated.
In the sections "Treated uterine septum versus controls (untreated septum)" and "Before and after septum removal" clinical heterogeneity within and between studies represents a limit of this meta-analysis that is worth highlighting. A random-effects model was used for the calculations to adjust for this heterogeneity. The meta-regression analyses supply the information about the potential moderating effect of some characteristics on the observed effect sizes. In most cases, different moderators such as septum classification, mode of conception, types of studies or quality studies did not change the effets size. Through sub-group analysis we assessed for the first time the role of a partial septum. Despite that fact that the evidence was limited by the poor number of studies included, we confirmed worse reproductive outcomes in this subgroup also.
The quality assessment conducted for the purpose of this meta-analysis suggested that most of the studies were of average quality, with only three out of 25 studies being graded as of high quality. Whether the methodological problems of the individual studies have biased the results cannot be assessed, although it might be considered reassuring that, in most cases, the effect size evaluated based on the quality of the studies did not materially alter the conclusions drawn. Finally, the present meta-analysis is also limited by including predominantly observational studies.

Conclusions and Future Research
The available evidence revealed a correlation between women having a septate uterus and poorer reproductive and obstetrical outcomes (pregnancy rate, live birth rate, miscarriage rate and pre-term delivery rate) compared to women without a septate uterus. Septum resection may be effective in increasing the live birth rate and reducing the risk of spontaneous abortion in women with poor reproductive histories. Non-conclusive evidence can be extrapolated regarding the efficacy of surgery in improving the likelihood of pregnancy in the infertile population. Moreover, the impact of septum resection on the prevention of preterm labor cannot be determined.
Pending further evidence, metroplasty should still be considered as good clinical practice in infertile patients or in patients with a history of repeated spontaneous abortion. [6,83] In case of a newly diagnosed uterine septum in patients seeking pregnancy without a prior history of reproductive failure, any advantages from septum removal cannot be established due to a lack of clinical scientific data. Further research through well designed comparative studies and randomized trials are needed to draw conclusions about the best management of patients diagnosed with a septate uterus.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/jcm11123290/s1, Table S1. Complications of surgical procedures in "Treated uterine septum versus controls (untreated septum)" section. Table S2. Complications of surgical procedures in "Before and after septum removal" section. Figure S1. Funnel plot of studies comparing spontaneous abortions in I-II trimesters between septate uterus vs. controls (no septate uterus). Figure S2. Funnel plot of studies comparing live birth rate before vs. after the removal of septate uterus. Figure S3. Funnel plot of studies comparing spontaneous abortions in I-II trimesters before vs. after the removal of septate uterus. Figure S4. Funnel plot of studies comparing preterm labour before vs. after the removal of septate uterus.

Conflicts of Interest:
The authors declare no conflict of interest.