Laparoscopically Assisted Anorectal Pull-Through versus Posterior Sagittal Anorectoplasty for High and Intermediate Anorectal Malformations: A Systematic Review and Meta-Analysis

Objective Anorectal malformations (ARMs) are one of the commonest anomalies in neonates. Both laparoscopically assisted anorectal pull-through (LAARP) and posterior sagittal anorectoplasty (PSARP) can be used for the treatment of ARMs. The aim of this systematic review and meta-analysis is to compare these two approaches in terms of intraoperative and postoperative outcomes. Methods MEDLINE, Embase, Web of Science and the Cochrane Library were searched from 2000 to August 2016. Both randomized and non-randomized studies, assessing LAARP and PSARP in pediatric patients with high/intermediate ARMs, were included. The primary outcome measures were operative time, length of hospital stay and total postoperative complications. The second outcome measures were rectal prolapse, anal stenosis, wound infection/dehiscence, anorectal manometry, Kelly's clinical score, and Krickenbeck classification. The quality of the randomized and non-randomized studies was assessed using the Cochrane Collaboration's Risk of Bias tool and Newcastle-Ottawa scale (NOS) respectively. The quality of evidence was assessed by GRADEpro. Results From 332 retrieved articles, 1, 1, and 8 of randomized control, prospective and retrospective studies, respectively, met the inclusion criteria. The randomized clinical trial was judged to be of low risk of bias, and the nine cohort studies were of moderate to high quality. 191 and 169 pediatric participants had undergone LAARP and PSARP, respectively. Shorter hospital stays, less wound infection/dehiscence, higher anal canal resting pressure, and a lower incidence of grade 2 or 3 constipation were obtained after LAARP compared with PSARP group values. Besides, the LAARP group had marginally less total postoperative complications. However, the result of operative time was inconclusive; meanwhile, there was no significant difference in rectal prolapse, anal stenosis, anorectal manometry, Kelly's clinical score and Krickenbeck classification. Conclusion For pediatric patients with high/intermediate anorectal malformations, LAARP is a better option compared with PSARP. However, the quality of evidence was very low to moderate.

Introduction Eligibility criteria Study selection and data collection process Two reviewers independently reviewed and collected data from the included studies; a third reviewer was required for the final decision in case of discrepancies.

Data items
The following data were sought: authors and year of study, the country, study type, single or multi center, ARMs type, number of participants, male/female sex ratio, age at surgery, weight at surgery, with/without colostomy done, associated anomalies, follow-up, operative time, length of hospital stay, total postoperative complications, rectal prolapse, anal stenosis, wound infection/dehiscence, anorectal manometry, Kelly's clinical score and Krickenbeck classification.

Quality assessment
The quality of the included randomized study was assessed by the Cochrane Collaboration's Risk of Bias tool [16]. The meta-analysis would only include the studies at low risk or unclear risk of the overall bias. Non-randomized studies were assessed using the Newcastle-Ottawa scale (NOS) [17]. The meta-analysis would include the studies deemed moderate or high methodological quality which was at least five stars. The studies published in the professional or high quality journal of pediatric surgery were considered first. And the studies published in a general journal or low quality journal would be included after comprehensive discussion. The overall quality of the evidence of main outcomes was assessed by GRADEpro (Version 3.6). There were four levels: high, moderate, low, or very low; and the results were presented in the Summary of Findings table [18].

Data synthesis and analysis
Statistical analysis was performed using Review Manager (RevMan, version 5.3, the Nordic Cochrane Centre, the Cochrane Collaboration, Copenhagen). Weighted mean differences (WMDs) with 95% confidence intervals (CIs) were presented for continuous data. Pooled risk ratios (RRs) were calculated for dichotomous data. The Cochrane's Q-statistic and I 2 index were used to assess statistical heterogeneity in the meta-analysis. For heterogeneous data, a random-effects model was used; otherwise, a fixed-effects model was employed. P<0.05 was considered statistically significant.

Study selection
A total of 332 potentially eligible studies were identified and reviewed. According to inclusion criteria, 226 studies remained after removing the duplicates. 216 of records, of which titles or abstracts were screened, were excluded. 50 studies remained were evaluated in detail. And 40 of these studies were excluded, 6 of which were reviews and meta-analysis, 15 of which with no comparative studies, 12 of which were indicating irrelevant topics or other anorectoplasty technology, 6 of which were of the same center and with the overlapped patients, and 1 of which with no sufficient data. As a result, 10 of the records were included in this meta-analysis, including one randomized controlled trial (RCT), eight retrospective studies, and one prospective study were included (Fig 1). All 10 articles came from professional journals of pediatric surgery: Journal of Pediatric Surgery, South African Journal of Surgery, and Pediatric Surgery International. Although the authors were contacted, we were unable to obtain additional information to include in the analysis.

Study characteristics
The 10 studies assessed 360 pediatric participants, including 191 and 169 that had undergone LAARP and PSARP, respectively. Table 1 summarizes the features and basic information of each included study, as well as patient characteristics, including publication year, country, study type, single vs multi center, ARM type, group, male/female sex ratio, age at surgery, weight at surgery, with/without colostomy done, associated anomalies, and follow-up. Mean age at surgery ranged from 2.7 to 22.6 months; mean follow-up duration ranged from 17.4 to 261 months. Table 2 presents primary and secondary outcome results from each included study: operative time, length of hospital stay, total postoperative complications, rectal prolapse,  anal stenosis, and wound infection/dehiscence. Tables 3, 4 and 5 show secondary outcome resulting from each study, including anorectal manometry, Kelly's clinical score [15], and Krickenbeck classification [3]. Table 6 shows overall analysis between LAARP and PSARP.

Risk of bias of included studies
Only randomized clinical trial was judged to be of low risk of bias (Fig 2). All nine cohort studies were judged to be of moderate to high quality (Table 7).

Safety
There were no reports of any adverse events following LAARP or PSARP in the studies reviewed.

Primary outcome measures
Operative time. Two studies [21,28] reported longer operative time of LAARP compared with that of the PSARP group (P = 0.13, P<0.001, respectively); while two studies [24,27] were in the contrary (P>0.05, P<0.01, respectively); and one study [22] reported the two groups had the same operative time (P = 0.92) ( Table 2). Therefore, the result of operative time was inconclusive. As only two [24,27] from the five studies [21,22,24,27,28] reporting operative time were suitable for the meta-analysis, and the heterogeneity was substantial (Q statistic = 11.60, P = 0.0007; I 2 = 91%), we only made a qualitative systematic review for it.

Heterogeneity
Two variables in the analysis were detected with obvious heterogeneity, i.e. length of hospital stay and sphincter squeeze.

Quality assessment of evidence
The quality of the evidence as assessed with GRADEpro was rated as very low to moderate. And it was summarized in the Summary of Findings table (Table 8).

Summary of evidence
Posterior sagittal anorectoplasty (PSARP) is a reconstruction which allows pediatric surgeons to operate under direct visualization; it is used as a standard technique since 1982, described by deVries and Peña [4]. Because of extensive perineal dissection in PSARP, favorable outcome is problematic [29]. With the development of small-size instruments and laparoscopic techniques in pediatric surgery, laparoscopically assisted anorectal pull-through (LAARP), gradually accepted by pediatric surgeons since 2000, was first introduced by Georgeson et al. [6]. Feasibility and safety of the LAARP approach has been demonstrated for high/intermediate type ARMs [27,30]. To some extent, LAARP has many advantages, including minimal surgical trauma, excellent visualization of rectal fistula and gynecologic anatomy, potentially fewer wound complications, and accurate placement of the bowel into levator ani and the sphincteric complex [11,30,31]. Furthermore, Wong et al. [25] considered that it was more accurate and required fewer other variables to compare LAARP with PSARP using high/intermediate anorectal malformations in studies. Therefore, to evaluate the efficacy and safety of LAARP in treating high/intermediate anorectal malformations from the above ten included studies, this systematic review/meta-analysis was performed through primary and secondary outcomes.  Voluntary bowel movements were defined as feeling an urge to defecate, the capacity to verbalize this feeling, and the ability to hold the bowel movement To compare LAARP with PSARP for the treatment of high/intermediate anorectal malformations, 10 studies were included in this meta-analysis. Interestingly, LAARP was associated with shorter hospital stay, less wound infection/dehiscence, higher ACRP, and better functional results (grade 2 or 3 constipation). In addition, the LAARP group had marginally less total postoperative complications compared with PSARP treated pediatric patients (RR 0.66, 95%CI 0.44-0.99; P = 0.05). However, no significant differences were found between the LAARP and PSARP groups in rectal prolapse, anal stenosis, anorectal manometry (RAIR, HPZL), Kelly's clinical score (fecal incontinence, fecal staining, sphincter squeeze, average score, and good ranking) and Krickenbeck classification (voluntary bowel movements, soiling grade 1, soiling grade 2 or 3, and grade 1 constipation). In addition, the result of operative time was inconclusive, and the uncertainty of the operative time of different centers may be caused by the different operating skills of surgeons and the complexity of patients' condition. These results indicated that LAARP was relatively more effective and safer in comparison with PSARP.
Not only LAARP but also PSARP is able to treat ARMs with success, however they both will cause specific postoperative complications [23]. Postoperative complications in the included studies were rectal prolapse, anal stenosis, wound infection/dehiscence, rectal retraction, and incontinence, among others. The increased wound infection/dehiscence incidence after PSARP might be due to the extent of the dissection performed as well as incision size [32]. However, occurrence rates of rectal prolapse and anal stenosis increased compared with the PSARP group, although no statistical differences were obtained. In the LAARP group, rectal prolapse may be due to the fact that the rectum was inadequately fixed [28]. To prevent the morbidity of rectal prolapse, the rectum should be secured to presacral fascia during LAARP, while dissection of rectum and pelvis should be limited [33]. Tong et al. [24] suggested that in the development of muscle channel, it might prevent stenosis to start anal dilation two weeks postoperatively using radially dilating trocars. Although most complications could be treated effectively or even cured, they surely affected recovery, defecation functions, and long-term outcomes.
In addition to postoperative complications, postoperative anorectal manometry is also useful for outcome comparison. Postoperative anorectal manometry is often used to evaluate functional results after surgical reconstruction of ARMs, and a good defecation status correlates well with the presence of normal anal canal resting pressure and an adequate anorectal pressure difference [34]. Meanwhile, RAIR reflects normal relaxation of the internal anal sphincter in response to rectal distension [35]. Anorectal manometry values may vary with age [36]. In this review, with the limited manometric data obtained, higher incidence of RAIR, increased ACRP and longer HPZL were observed in patients after LAARP compared with those that underwent PSARP, although only ACRP showed a statistically significant difference. Yazaki et al. [28] suggested that the quality of the patient's nerves and muscles in the pelvis was a true determinant of outcomes, since clear visualization during operation could reduce damage to the muscle complex and nerves around the puborectal muscles [20].
It is largely accepted that the postoperative defecation status is of great importance. Three studies [19,20,24] used the Kelly's clinical scoring system [15] to compare midterm outcomes between the LAARP and PSARP groups, and found higher general scores and more good rankings in the former group, although no statistical difference was obtained. Five studies [21-23, 25, 27] evaluated functional results according to the Krickenbeck classification [3], and found more voluntary bowel movements and improved status of soiling and constipation in each grade. Indeed, this simple system was already validated in previous studies assessing patients that underwent PSARP [37]. In addition, to assess bowel function in patients who received LAARP or PSARP, a structured fecal continence evaluation (FCE) questionnaire developed by Yazaki et al. was used [28]. With time, functional results may improve, and LAARP seems to provide better outcomes [27]. Consistency of classification and standard scoring systems is suggested to help develop the standardized protocols to evaluate postoperative conditions and improve postoperative outcome as well as the quality of life [38].
For the long-term prognosis of ARMs, it is mainly to evaluate the defecation function. In the study by Ming et al. [27], there was no statistically significant difference between the two groups in good voluntary bowel movements and soilings. However, De Vos et al. [23] found that the continence of both groups was poor in long-term evaluation and many patients needed a bowel management programme. Therefore, long-term prognosis is uncertain.

Limitations
There are several limitations in this systematic review and meta-analysis. Firstly, only one randomized clinical trial was included; some included studies were retrospective in nature. Therefore, results were likely to be confused, with the lack of control. In addition, surgery approach was often determined by physician's experience and patient's condition. Besides, all studies were single center trials, with small sample sizes, and results might be biased. And some data like length of hospital stay, anorectal manometry (RAIR, ACRP, HPZL) and KCS were analyzed in only three papers with a limited number of patients. The length of hospital stay reflects the condition of the patient during the treatment period, and it is significant; anorectal manometry and KCS can evaluate the effect of surgical treatment, and reflect the quality of children's life. However, these results should be interpreted with caution given the low number of participants considered.
Furthermore, some data showed overt heterogeneity which included length of hospital stay and sphincter squeeze. Three articles [20,24,27] reported the length of hospital stay, of which Yang et al. [20] and Tong et al. [24] had the similar hospital stay, and Ming et al. [27] had obvious short-term hospitalization. The different postoperative complications and hospital discharge standards might be the reasons for the emergence of heterogeneity. Three articles [19,20,24] reported the sphincter squeeze. The reasons of the existence of its heterogeneity were that the assessment was performed by the examining finger of the surgeon and that the children were too young to comprehend. The realities of clinical practice inevitably result in certain degree of heterogeneity which could cause significant statistical heterogeneity, leading to inaccurate conclusions in a medical meta-analysis [39]. Moreover, among the included studies, patient age, follow-up time, and disease degree varied; such differences may affect the final results. Finally, unpublished works not included or omission of other data might lead to biased findings.

Conclusions
In conclusion, LAARP is a safer, more feasible and effective surgical procedure compared with PSARP in treating high/intermediate anorectal malformations in pediatric patients. LAARP has shorter hospital stay, reduced wound infection/dehiscence, and higher ACRP compared with PSARP. In addition, LAARP has marginally significant advantage of less total postoperative complications. Furthermore, the result of operative time is inconclusive; meanwhile, LAARP and PSARP have similar statuses of rectal prolapse, anal stenosis, anorectal manometry, Kelly's clinical score, and Krickenbeck classification. However, follow-up may not have been long enough, only one RCT was included, and the quality of evidence was very low to moderate. Long term follow-up, large, multi-center studies, and high quality randomized controlled trials are needed in the future to confirm the current findings.
Supporting Information S1