Minimally invasive therapy for gynaecological symptoms related to a niche in the caesarean scar: a systematic review

Authors


Abstract

Background

Various therapies are currently used to treat symptoms related to the niche (an anechoic area) in the caesarean scar, in particular to treat abnormal uterine bleeding (AUB).

Objective

To systematically review the available literature reporting on the effect of various therapies on niche-related symptoms.

Search strategy

A systematic search of MEDLINE, Embase, Cochrane, trial registers and congress abstracts from AAGL and ESGE was performed.

Selection criteria

Articles reporting on the effectiveness of therapies other than hysterectomy in women with niche-related symptoms were included. Studies were included if they reported one of the following outcomes: effect on AUB, pain relief, sexual function, quality of life (QOL), and surgical, anatomic, fertility, or pregnancy outcome.

Data collection and analysis

Two authors independently selected the articles to be included. The Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines were followed. A standardised checklist was used to score the methodological quality of the included studies.

Main results

Twelve studies were included, reporting on hysteroscopic niche resection (eight studies, 384 patients), laparoscopic repair (one study, 13 patients), (laparoscopic assisted) vaginal repair (two studies, 47 patients), and oral contraceptives (OCs) (one study, 11 patients). Reported AUB improved in the vast majority of the patients after these interventions, ranging from 87 to 100%. The rate of complications was low. Pregnancies were reported after therapy; however, sample sizes and follow-up were insufficient to study fertility or pregnancy outcome. The methodological quality of the selected papers was considered to be moderate to poor, and was therefore insufficient to make solid conclusions.

Author's conclusions

More evidence is needed before (surgical) niche interventions are implemented in daily practice.

Introduction

Worldwide, the rates of caesarean section are rising.[1] Caesarean section scars may be associated with complications in later pregnancies, such as uterine rupture, abnormally adherent placenta, and scar dehiscence or rupture.[2-5] The presence of an anechoic area at the site of a caesarean section scar has been observed in women after a caesarean section. In random populations, it was present in 24–69% of women evaluated with transvaginal sonography (TVS), and in 56–78% of women evaluated with transvaginal sonohysterography (SHG).[6-12] Such an anechoic area is also known as a niche (Figure 1).[6, 8, 9, 13, 14] A generally accepted definition of a niche is still the subject of debate, however.[15]

Figure 1.

Schematic diagram of a niche at the site of the caesarean section scar. Figure modified from Gubbini et al. (2011).[29]

A common symptom reported to be associated with the presence of a niche is postmenstrual spotting. The first publications on caesarean section scar defects in relation to bleeding symptoms date from 1975.[16] Since then many articles have reported a high prevalence of niches in women with abnormal uterine bleeding (AUB), including prolonged menstruation or postmenstrual spotting (PS).[9, 14, 17-21] Two studies reported a significantly higher prevalence (34 and 29%) of postmenstrual spotting in women with niches compared with women who did not have niches during SHG (OR 2.8 and 5.5) in a random population of women with a history of caesarean section.[6, 12] One other study reported a higher prevalence of postmenstrual spotting in women after a caesarean section than after vaginal delivery, but could not identify a significant relationship with the existence of a niche. The same study did, however, report a higher frequency of AUB in women with diverticula (anechoic round structures) and deformation of the cervical canal at the scar site, identified during TVS.[10] Postmenstrual spotting after a caesarean section in women with a niche is not a rare event. Approximately 60% of women have a niche after a caesarean section, and of these approximately 30% experience postmenstrual spotting.[6, 12] Other reported symptoms in women with a niche were dysmenorrhea (53.1%), chronic pelvic pain (36.9%), and dyspareunia (18.3%).[18]

It has been assumed that AUB may be caused by the retention of menstrual blood in the niche, which is intermittently expelled after menstruation has almost completely ceased, causing postmenstrual spotting and pain.[4, 17, 22] The presence of fibrotic tissue below the niche may impair the drainage of menstrual flow.[14] Additional, newly formed fragile vessels in the niche may also contribute to the accumulation of blood produced in situ.[23]

Niche-related menstrual bleeding disorders do not always respond to hormonal therapies.[24] Therefore, a hysterectomy is often performed to treat niche-related gynaecological symptoms.[17, 25, 26] In women who wish to preserve their fertility, however, less invasive, more conservative treatments have been developed. These therapies aim to: facilitate the drainage of menstrual blood and to reduce the in situ production of blood by coagulating niche vessels (hysteroscopic resection);[22, 24, 27-29] reconstruct the uterine defect in the caesarean section scar [abdominal, (robotic) laparoscopic, or vaginal repair];[30-37] reduce menstrual periods (continuous hormonal therapies); or reduce menstrual blood (cyclic hormonal therapies or other medical therapies).[38] Before surgical interventions such as niche resections or niche repair are implemented on a large scale, their effectiveness should be proven and compared with medical therapies or expectant management.

The objective of the current review is to provide an overview of the available literature on the effect of niche therapies, except for hysterectomy, on AUB and other gynaecological symptoms, including surgical, anatomical, and fertility outcomes.

Methods

We performed this systematic review according to the Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines, as all of the studies included were observational.[39]

Literature search

The MEDLINE, Embase, and Cochrane computerised bibliographic databases were searched. Also the following trial registers, www.clinicaltrial.gov and www.trialregister.nl were searched for continuing or not yet published studies. To avoid missing recent and not yet indexed studies, the 2012 annual abstract compilations by ESGE and AAGL were checked. The search was performed within the following limits: reports in English, and published between 1 January 1950 and 15 August 2013. The search strategy was developed with the assistance of a research librarian specialised in medical sciences, and included the following terms appearing in the title, abstract, or keywords: treatment (medical subject heading, MeSH), surgical procedures, operative (MeSH), caesarean section (MeSH), uterus, cicatrix, scar, isthmocele, niche, pouch, wound dehiscence, diverticula, and scar defect (Appendix S1). The reference lists of retrieved articles were cross-checked for relevant studies.

Eligibility criteria

All randomised controlled trials, cohort studies, case–control studies, and case series of at least five patients were considered for inclusion if they: (1) reported on conservative therapies (hysteroscopic niche resection, laparoscopic repair, abdominal repair, vaginal repair, endometrial ablation, levonorgestrel intrauterine system, or medical treatment) for the treatment of niche-related symptoms in non-pregnant patients; and (2) reported on one of the following outcomes – AUB, pain relief, sexual function, quality of life (QOL), surgical outcome, and anatomic reconstruction, fertility, or pregnancy outcomes. All other studies (including case reports and video reports) were excluded.

For reports that were published twice we included the study with the largest number of included patients. An eligibility assessment was performed independently in an (unblinded) standardised manner by two reviewers trained in performing systematic reviews (L.V. and J.H.). Disagreements between reviewers were resolved by having three reviewers (L.V., J.H., and A.V.) review the original article until a consensus was reached.

Data presentation

Data extraction was performed according to a standardised format including items about the general study level characteristics (such as design, population, and setting), diagnosis, the intervention studied, and the outcomes. To get an impression of the quality of the included studies, all studies were systematically evaluated on possible bias (selection bias, verification bias) and a structured checklist was used with the following items: clear definition of primary outcome; (validated) instruments used for measurement outcome; clear definition of population; clear definition of the inclusion and exclusion criteria; clear definition of niche and of niche measurements; evaluation pre- and post-treatment similar regarding diagnosis of niche and bleeding outcomes; clear methods of follow–up; and reporting on non-treated patients and complications (Tables 2 and 3). Every included study was evaluated by two reviewers independently (L.V. and J.H.). Data in the tables are subdivided according to the type of intervention.

Statistical analysis

To study the effect on postmenstrual bleeding, a meta-analysis was performed by use of comprehensive meta-analysis 2, (2005; Biostat, Englewood, NJ, USA). Results were presented in a forest plot, comparing outcomes after therapy with baseline.

Results

Literature identification

The electronic search in MEDLINE, EMBASE, and Cochrane generated 1861 records. Two additional records were identified through cross-checking and 244 additional records were identified in the trial registers. Two continuing studies were identified, with no data yet reported: one prospective cohort and one randomised controlled trial (NTR 3269 and NTR 1922). After the removal of duplicates, 1629 records were screened, and 28 articles were thought to meet the inclusion criteria and were selected for full article assessment. An additional 16 articles were excluded because of the following reasons: study design (six case reports, five video reports); two double publications; two that included other subjects; and one review without original data. Ultimately, 12 studies from eight countries were included in this review (Figure 2). The studies, 11 peer-reviewed articles,[22, 24, 27-29, 35, 37, 38, 41-43] and one abstract,[45] were published between 1996 and 2013, and reported on a total of 455 women (Table 1).

Table 1. Study characteristics
Author yearDesignPeriod n Intervention (n) Follow-up (months)Control (n) IndicationPrimary outcomeSecondary outcomes
Hysteroscopic niche resection
Chang[27]Prosp. cohort2005–200857hys. resect. (22)NRNonePAUBNRNA
Gubbini[29]Case series2005–200841hys. resect. (41)NRNonePAUBInfertilityNA
Gubbini[24]Case series2001–200526 hys. resect. (26)NRNonePAUBPAUBInfertility
Marra[45]Case series2001–200978hys. resect. (78)NRNonePAUB, HMBPAUBInfertility
Feng[43]Retro. cohort2006–200962hys. resect. (57)12NonePAUB, PMAUBNA
Wang[28]Retro. cohort2003–200857hys. resect. (57)45NonePAUB, PMAUB, DMProg fact.
Fabres[22]Retro. cohort1993–200124hys. resect. (24)14–24NonePAUBAUBInfertility
Florio[42]Case–control2007–200939hys. resect. (19)3OC/20PAUBAUB, DMSatisfact.
Laparoscopic niche repair
Marotta[41]Case seriesNA13lap. repair (13)3–48NoneRM <3 mmUterine anat. RMPregnancy outcome
Vaginal niche repair
Klemm[31]Case series1999–20045vag. repair (2) lap. vag. (3)30 (3–46)NoneAUB, painUterine anat.Pregnancy outcome
Luo[37]Retro. cohort2010–201142vag. repair (42)10–23NoneAUB, PAUB, PMUterine anat.NA
Medical treatment
Tahara[38]Case series1997–200311OC (11)12NonePAUBPAUBNA
Figure 2.

prisma flow diagram of the systematic literature search; DA, dissertations and abstracts; SCI, Science Citation Index.

Large randomised controlled trials are lacking. Three studies were reported as prospective cohort studies; however, based on the reported methodology we qualified them as case series.[24, 28, 45]

Methodological quality of included articles

None of the articles met all of our quality criteria. A clear definition and standardised measurements of outcomes were lacking in most articles. Most studies did not report how they evaluated bleeding symptoms or niche measurements. None of the studies reported on the use of validated instruments (Table 2).

Table 2. Quality assessment
ReferencesClear definition of primary outcomeClear instrument (validated) to measure primary outcome Clear description of populationDefinition of niche/ minimal size for diagnosisClear description of inclusion criteriaClear definition of exclusion criteriaEvaluation pre- & post-treatment similar (bleeding)Evaluation pre- & post-treatment similar (niche) Follow-up period definedComplete follow-upOutcome of non-treated patients Complications reported
  1. NA, not applicable; NR, not reported.

Hysteroscopic niche resection
Chang[27]NoNRYesYes/NRYesYesNRNRNRNRNRYes
Gubbini[29]NoNRNoYes/NRYesNoNAYesNRNRNAYes
Gubbini[24]NRNRNoYes/NRNoNoNRYesNoNRNANo
Marra[45]NRNRNoNo/NRNoNoNRYesNoNRNAYes
Feng[43]NRNRYesNo/NRNoYesNRNRNRNRNRYes
Wang[28]Yes: days of menstruationNRNoYes/NRNoYesNRNoNRNRNAYes
Fabres[22]NoNRYesYes/NRNoYesNRNRNRNRNRNo
Florio[45]NoNRNoYes/NRNoYesNoNoYesNRNAYes
Laparoscopic niche repair
Marotta[41]NRNRNoNo/YesYesNoNRYesYesYesNANo
Vaginal repair
Klemm[31]NoNoNoYes/NRNoNoNRYesNoNoNAYes
Luo[37]NoNoNoYes/NRNo NoNRYesYesYesNAYes
Medical treatment
Tahara[38]NoNRYesYes/NRNoNoNRYesNRNoNAYes

None of the included articles described efforts to address potential sources of bias or confounding. For example, the method of patient selection and enrolment were not reported. Data on excluded patients, those lost to follow-up, or missing data for outcome measurements are lacking. Methods of post-treatment outcome measurements were not reported or were different from pre-treatment evaluation in most studies, indicating a high risk of verification bias (Table 3). Publication bias is suspected because there were only articles with positive effects and small sample sizes, and large randomised trials and prospective cohort trials are lacking.

Table 3. Risk on selection and verification bias of included studies: results of our assessment
References Risk on selection bias Risk on verification bias
  1. NA, not applicable; NR, not reported; PAUB, postmenstrual abnormal uterine bleeding.

  2. High risk of selection bias: two or more items missing.

  3. Moderate risk of selection bias: one item missing.

  4. Risk of selection bias was assessed on following items: clearly defined population; clearly defined inclusion or exclusion criteria; clear definition of niche or bleeding outcome.

  5. Risk on verification bias was defined as high if any of the following items was present: pre- and post-treatment evaluation differed; evaluation between groups differed.

  6. Risk on verification bias was defined as moderate if any of the following items was present: incomplete follow-up; follow-up not defined; no reporting on non-treated patients.

Hysteroscopic resection
Chang[27]LowClear definition of population, and inclusion and exclusion criteriaNAEvaluation post-treatment not reported, no information on follow-up
Gubbini[29]HighNo clear population, no clear exclusion criteria, only patients with pregnancy after therapy includedLow–moderateNo clear definition of outcomes; however, biopsy taken but possibly not relevant
Gubbini[24]HighNo clear population described, no clear in/exlusion criteria, no information on excluded patientsLow–moderateNo clear definition of outcomes; however, biopsy taken but possibly not relevant
Marra[45]HighNo clear description of population, no clear in/exclusion criteria, no clear outcomeLow–moderate No clear definition of outcomes; however, biopsy taken but possibly not relevant
Feng[43]ModerateClear description of population but no clear definition of nicheNAEvaluation post-treatment not reported, no information on follow-up
Wang[28]HighNo clear inclusion criteria and population descriptionHighDifferent methods of evaluation pre- and post-treatment
Fabres[22]ModerateNo clear inclusion criteriaNANo information on post-treatment evaluation
Florio[45]HighPAUB not defined, no clear population described, no information on criteria for treatment optionsHighDifferent evaluation for treatment groups; different evaluation pre- and post-treatment
Laparoscopic repair
Marotta[41]HighPopulation not described, no clear definition of nicheLow 
Vaginal repair
Klemm[31]HighNo clear population described, no clear in/exclusion criteriaModerateIncomplete follow-up
Luo[37]HighNo clear population described, no clear in/exclusion criteria, PAUB not definedLow 
Medical therapy
Tahara[38]ModeratePAUB not defined, clear populationNAEvaluation post-treatment not reported

The duration of follow-up was reported in eight studies,[22, 28, 35, 37, 38, 41-43] with a mean of 20 months, ranging between 3 and 48 months. The mean sample size was 38 patients (Table 1).

Reported interventions

Reported interventions were hysteroscopic niche resection (Figure 3),[22, 24, 27-29, 42, 43, 45] laparoscopic niche repair,[41] laparoscopic assisted vaginal repair,[35] vaginal repair (Figure 4),[37] and oral contraceptives.[38] Details about the technical aspects are given in Table 4.

Table 4. Intervention characteristics
ReferencesInterventionDuration of surgery (min.) Success of therapyAdmission Hospital (days) Anesthesia appliedType of hysteroscopeDistension mediumSafety check during surgeryCompli- cations
  1. coag., coagulation of the niche bottom; dist., distal part of the niche; M–blu, methylene blu; NR, not reported; prox., proximal part of the niche; US, ultrasound.

  2. a

    Used CO2 laser to cut niche out.

  3. b

    Postoperative infection (= 1).

  4. c

    Side effects: one patient, breast tenderness; one patient, nausea.

Hysteroscopic niche resection
Chang[27]Distal, coag. loopNR100%NRNR9 mm mon.Sterile waterUSNone
Gubbini[29]Dist. & prox., coag. rollerball14.1 (± 3.5)100%NRGeneral9 mm mon. StorzGlycerine/manitolM-bluNone
Gubbini[24]Dist. & prox., coag. rollerballNR100%NRNR9 mm mon. StorzSorbitol/manitolM-bluNone
Marra[45]Dist. & prox., coag. rollerballNR100%NRNR9 mm mon StorzSorbitol/manitolM-bluNone
Feng[43]Distal, coag. rollerballNR92% (57/62)NRGeneral6.5 and 8 mm bipol.Olympus5% manitolNoneNone
Wang[28]Distal, coag. rollerball30.2 (±6.6)100%OutpatientNR26F Storz resectoscopeDistilled waterNoneNone
Fabres[22]Distal, coag.NR100%NRNR9 mm mon. StorzSorbitol/manitolNoneNone
Florio[45]Dist. & prox., coag. rollerballNR (only for subgroups)100%NRGeneralNRSorbitol/manitolNoneNone
Laparoscopic repair
Marotta[41]Cut out, two-layer closureNR100%<1 GeneralNonea Hyster.None
(Laparoscopic assisted) vaginal repair
Klemm[31]Cut out, one-layer closure117 (27–92)100%NRNRNANATVUNone
Luo[37]Cut out, two-layer closure55.3 (± 23.5)100%4 (±2.1)NRNANANone2.4%b
Medical therapy
Tahara[38] oral contraceptive (ethinyl estradiol/norgestrel)NANANANA2c  
Figure 3.

Schematic diagrams of two approaches of hysteroscopic niche resection: (A) resection distal part; (B) resection distal and proximal part. Modified figure from Gubbini et al. (2011).[29]

Figure 4.

Schematic overview of a niche repair: this can be performed by a vaginal or abdominal approach, by laparotomy, laparoscopy, or robot assisted. Modified from Gubbini et al. (2011).[29]

The use of OCs in the treatment of niche-related AUB was studied in two studies.[38, 42] Women used OCs for 21 days followed by 7 days of OC withdrawal in both studies. The OCs contained 0.5 mg norgestrel and 0.05 mg ethinyl-estradiol,[38] or 0.075 mg gestodene and 0.030 mg ethynylestradiol.[42]

There were no reports on abdominal interventions, levonorgestrel intrauterine system (IUS) or endometrial ablation that met our selection criteria.

Diagnosis and indications

All studies included women with a niche identified via TVS, hysteroscopy, or magnetic resonance imaging (MRI). Three studies did not provide a definition of a niche. Excepting one study, none of the studies provided an objective reproducible method of niche measurement for inclusion. Only Marotta used an objective cut-off level for inclusion.[41] All studies, except two, included women with AUB, in particular postmenstrual AUB. Two studies also included women without symptoms, because of the presence of a large niche and a desire to maintain fertility (Appendix S2; Tables 1, 2 and 4).[35, 41]

Abnormal uterine bleeding

All of the studies included reported on women with AUB (= 455); nine reported specifically on postmenstrual spotting (Appendix S2). None of the studies used validated questionnaires to assess bleeding disorders. Four studies did not report any quantitative outcome with respect to AUB.[22, 37, 41, 45] AUB was reported to be improved in 87% (hysteroscopic resection),[22, 24, 26-29, 42, 43, 45] 100% (laparoscopic repair),[41] 93% (vaginal repair),[37] and 91% (OCs) of women.[38] Three studies, including 121 women, reported on bleeding parameters both at baseline and after surgery: hysteroscopic niche resection reduced the number of days of bleeding by 2–4 days,[27, 43] and vaginal repair reduced the number of days of bleeding by 4–7 days.[37] These three studies reported sufficient data for evaluation in a meta-analysis. The weighted mean difference was –4.48 days (95% CI −6.59 to −2.37 days; Figure 5). A retrospective study comparing the reduction in menstrual bleeding after hysteroscopic niche resection with OCs was in favour of the hysteroscopic niche resection: the reduction of menstrual bleeding was 2.5 days more after hysteroscopic niche resection.[42]

Figure 5.

Meta-analysis.

Pain relief

Data on pain were reported in three studies, including 93 women.[29, 37, 42] At baseline, 38–63% of these patients reported pain. After hysteroscopic resection pain symptoms were reported to be resolved in 97% of the patients and in 100% of the patients after laparoscopic repair (Appendix S2). Data on methods of pain assessment are lacking.

Patient satisfaction, urinary symptoms, sexual function, and quality of life

The reported satisfaction rates regarding relief of bleeding symptoms were: 87% (range 59–100%) after hysteroscopic resection (310 patients);[22, 24, 27-29, 43, 45] four out of four patients after laparoscopic repair;[41] 45 out of 47 patients after vaginal repair;[37] and ten out of 11 patients after treatment with OCs (Appendix S2).[38] No validated questionnaires or other validated tools were used to asses urinary symptoms, sexual function, or QOL. Additional therapies applied because of the failure of the initial methods were reported in two studies.[28, 37] Ten women received additional OCs and one underwent a hysterectomy after hysteroscopic niche resection.[28] One woman underwent a second vaginal repair after the failure of the first.[37]

Surgical outcome

A systematically performed comparison of niche characteristics at baseline and after surgery was reported only by Marotta (Appendix S3; Table 2).[41]

The success rates reported after surgery were 92–100% after hysteroscopic resection,[22, 24, 27-29, 41, 43, 45] and 100% after laparoscopic and vaginal repair.[37, 41] Four studies included a diagnostic hysteroscopy 8–12 weeks after hysteroscopic surgery for the evaluation of uterine and niche anatomy; however, detailed information on the criteria used and the outcomes is lacking.[24, 28, 29, 42] Direct endometrial biopsy confirmed the presence of cuboid endometrial cells covering the niche after 8–12 weeks in 60 out of the 60 women evaluated.[29, 42]

The mean thickness of the residual myometrium after laparoscopic repair increased by 8.2 mm, as measured on MRI, after a mean follow-up period of 3 months (Appendix S3).[41]

Adverse outcomes and complications

No complications or adverse effects were reported after hysteroscopic resection or laparoscopic repair. Reported side effects in two of the 11 women treated with OCs included breast tenderness and nausea. After vaginal repair, an infection was reported in one of the 42 treated patients (Table 4).

Fertility and pregnancy outcome

Seven studies reported on fertility and pregnancy outcomes. A total of 78 women were reported to have fertility problems, of whom 72 later conceived (92%): 67 after hysteroscopic resection, four after laparoscopic repair, and one after vaginal repair (Appendix S4).[21, 23, 27, 28, 34, 40, 44] Four of the 72 pregnant women had a spontaneous miscarriage, and 58 women delivered at term by a planned caesarean section. The pregnancy outcomes for an additional ten pregnant women were not reported. No additional obstetric complications were reported. The duration of subfertility and details on previous diagnostic work-up in relation to the therapies received was reported in one study.[29] This study included 41 subfertile women, all with fertility problems of unknown aetiology and with a duration of more than 2 years. All were reported to have conceived within 2 years without additional therapy. After vaginal repair or medical treatment, no information on pregnancies was reported.

Discussion

Main findings

The effect of therapy on bleeding symptoms or other gynaecological symptoms could be evaluated in 384 women after a minimally invasive surgical intervention, and in 31 women after OC therapy; however, the methodological quality of the majority of papers included should be regarded as moderate to poor. The included studies reported a reduction of AUB and pain and a high rate of satisfaction among most women after hysteroscopic niche resection or vaginal or laparoscopic niche repair. A significant difference was reported using a meta-analysis on AUB after treatment compared with baseline. The thickness of the residual myometrium after laparoscopic niche repair increased substantially. Quality of life, sexual function, and urinary symptoms were not studied. Seven studies reported on pregnancy outcomes without an adverse outcome in 52 women after hysteroscopic niche resection, in three women after laparoscopic niche repair, and in one woman after vaginal niche repair.

Strengths and limitations

We included only articles published in English and we excluded case reports, which involves a risk of reporting bias. As a consequence we may have missed relevant publications or rare complications. In general, a funnel plot could be helpful to assess the level of publications bias; however, only three observational studies reported sufficient data to be included in a funnel plot, and as a consequence this will not provide reliable information.[46, 47] We used a structured checklist in an attempt to weight the scientific quality of included studies; however, this review was hampered by the fact that several of the individual studies did not report all relevant information needed to do this in a proper way. Clear definitions of primary outcome, abnormal uterine bleeding, abnormal postmenstrual bleeding, pain, satisfaction, or a successful procedure are lacking. Validated instruments for the measurements of these outcomes were not used.

Most studies do not report the method of patient enrolment, the number of excluded patients, or the number of patients lost to follow-up; therefore, besides selection bias, verification bias cannot be excluded. In addition, the sample size in most of the included studies is insufficient to draw solid conclusions, in particular with respect to fertility or pregnancy outcome or to enable the eventual correction for confounding. Long-term outcome after various therapies is not reported in most studies, so information on the risk of recurrence cannot be addressed. Although many studies reported a high success rate for the procedure they examined, none underscored their statements or conclusions with data. It is surprising that none of the studies assessed the effect of the interventions on patient QOL. Only seven studies reported on pregnancy outcome after hysteroscopic niche resection or laparoscopic repair. These women were all scheduled for an elective caesarean section, so the success rate for trial of labour after various niche therapies cannot be addressed.

Interpretations of the evidence

As far as we are aware only one review has previously been published on this topic, which reported on the effect of hysteroscopic niche resections on fertility outcome. That review concluded that hysteroscopic niche resection was effective in improving symptoms related to menstruation, and that it restored fertility in a high percentage of women treated;[48] however, they did not perform a systematic literature search and did not perform a critical appraisal.

Given the methodological shortcomings of the included studies in this review we have to conclude that the current evidence is not sufficient to draw solid conclusions on the effectiveness of the interventions under study. And as a consequence, the outcomes of our meta-analyses to assess improvement of bleedings symptoms should be critically assessed. In addition, the number of pregnant patients is insufficient to analyse pregnancy outcomes or to estimate the risk of pregnancy-related complications after hysteroscopic niche resection or laparoscopic niche repair. Although the thickness of the residual myometrium (RM) increased after laparoscopic niche repair and the majority of the patients had an RM of over 8.3 mm after therapy, it is unknown what this means for the risk of uterine rupture. We do not know if an increased thickness of the RM is related to an improved functionality of the lower uterine segment. In addition, we do not know the effect of hysteroscopic niche resection on the risk of uterine rupture, malplacentation or cervical incompetence. In theory one could assume the hysteroscopic resection reduces the thickness of the residual myometrium and its functionality; however, information on this topic is currently lacking. Some studies reported on a laparoscopic (assisted) niche repair in asymptomatic patients.[31, 41] Given the lack of information of the effect of surgical interventions on pregnancy outcome, we believe that these interventions should only be considered in symptomatic patients. This was also stated by Demers and colleagues[49] The effect of OC on niche-related bleeding symptoms was assessed in only two studies, with a total of only 31 patients.[38, 42] Given the high rate of successful levonorgestrel IUS insertions for abnormal uterine bleeding,[50] it is surprising that its effect on niche-related AUB has not yet been evaluated.

Because of the dearth of comparative studies of sufficient methodological quality, evidence on the most optimal therapy is lacking. In view of the lack of evidence, the least invasive therapy, i.e. OC, should be considered the first choice for the treatment of gynaecological symptoms. This is of course less suitable for women who desire to conceive in the short term.

Given the lack of sufficient evidence on the outcome of hysteroscopic, laparoscopic, and vaginal repair on fertility and pregnancy outcome, we propose that these interventions should only be applied in a research setting with long-term follow-up.

Conclusion

The included studies report a reduction of AUB, a substantial reduction of pain or prolonged bleeding, and a high rate of satisfaction in most patients after hysteroscopic niche resection, or vaginal or laparoscopic niche repair, with a low complication rate; however, concern about the methodological quality of the included studies exists. Because of the small sample sizes and incomplete long-term follow-up no conclusions can be drawn on the risk of complications during pregnancy or delivery after these therapies. Comparative studies of different therapies with sufficient methodological quality are lacking. Future studies that include a sufficient sample size, with long-term follow-up and using validated instruments, are needed. Surgical interventions should be compared with expectant management or medical therapies, including levonorgestrel IUS. Until this information is acquired surgical niche therapies should only be applied in a research setting, preferably in randomised trials.

Disclosure of interests

None of the authors have relevant financial, personal, political, or religious interest linked to the subject of this article.

Contribution to authorship

The review was conceived by L.V., A.V., and J.H., and they selected the articles and collected the data. Analysis of the data was performed by L.V. and J.H. The first draft was written by L.V. and J.H., and A.V., M.B., H.B., and S.V. supervised the article to the final draft.

Details of ethics approval

Ethics approval was not needed because this is a literature study.

Funding

This study is funded by ZonMw, The Netherlands Organization for Health Research and Development (project no. 80-82305-97-12030).

Acknowledgements

We thank Mrs Boerstoel, assistant librarian, for her help with the literature search.

Ancillary