Search strategy

A systematic search of PubMed, the Cochrane Library, the Web of Science, and Medline was performed for relevant literature dating from February 1992 to 2015 August. We recognized all relevant published and unpublished literature via a thorough search strategy. The key phrases were described in supporting material (S1 File).The search results were only included literature in English and humans for the study category. We reviewed all titles and abstracts, and each potentially relevant article was marked for further examine. The full-text articles of the marked studies were included if satisfied our inclusion criteria. Additionally, the references listed in all obtained studies were also screened for possible inclusion.

GDW did the systematic search, reviewed of all abstracts, marked relevant full-text articles. CT and CYZ independently evaluated the full texts according to the predefined inclusion criteria and screened the references listed in all obtained studies. If a full text was assessed by both authors and satisfied the inclusion criteria, they were included. Any disagreement was determined by a third reviewer CYM.

Inclusion criteria

Studies were considered appropriate for inclusion if they met the following criteria: 1) adoption of either a purely mononostril or binostril endoscopic transsphenoidal approach and 2) series should include 50 or more patients with pituitary adenoma who had undergone surgery in the same center. A cutoff of 50 patients per approach was arbitrarily chosen to exclude those literatures presenting merely their preliminary experience with an approach. Series including variations of the endoscopic-assisted approach or microscopic approaches were excluded.

The items analyzed must include: 1) gross tumor resection (GTR): absence of residual tumor on postoperative MRI; 2) postoperative hormonal remission rate: for patients with somatotrophic tumor, the level of IGF-I norm suitable to sex and age, the level of GH in OGTT <1 ng/ml(2000 year criterion[7]) or < 0.4 ng/mL (2010 year criterion[8]); the PRL levels were normalization in PRL-secreting adenomas; for patients with Cushing disease, post-operative normalized serum ACTH, cortisol levels, and free cortisol level in 24-hour urine collection; 3) improvement in visual function base on comparing preoperative with postoperative visual acuity and visual field examination; 4) the length of hospital stay, and 5) complications: postoperative cerebrospinal fluid [CSF] leaks requiring treated with lumbar drain or second surgery, permanent and temporary diabetes insipidus [DI], anterior pituitary insufficiency, meningitis, epistaxis, or sinusitis. The following data items were also recorded but cannot be used for statistical analysis, due to the number of studies or incidence relevant these data was rare: mortality, the length of operation, visual recovered, hyposmia, cranial nerve injury, intraoperative blood loss, intracavernous carotid artery injury, and recurrence rate.

Statistical analysis

This meta-analysis was done in Stata (version12.0; Stata Corporation, College Station, TX). Considering the articles included in our study were not possible to compare using the traditional meta-analysis method with relative risk or OR., the different parameters of the included studies were analyzed respectively. A random effects-model was conducted for meta-analyses, due to the obvious clinical heterogeneity and the heterogeneity ofvariance was estimated with the DerSimonian and Laird approach [9]. For the dichotomous outcomes, estimated proportions and 95% confidence interval were described. All proportions were compared between the mononostril and the binostril approach using binomial tests and were considered significant difference if the P value was <0.05. Timothy R. DeKlotz [10] and Mario Ammirati [11] used this method when comparing outcomes of endoscopic to microscopic pituitary adenoma surgery. In addition, the Mann-Whitney U test was used for the length of operation and T-test was used for patient characteristics (SPASS 19.0), while the P-value of <0.05 was considered statistically significant.

Search results

From the searches of databases, 2075 studies were initially identified. Part of these studies were excluded as duplicates using reference manager EndNote, and the majority was excluded on the basis of the titles or abstracts, resulting in 107 potentially relevant articles. Subsequently, the full texts of these studies were obtained and reviewed thoroughly, where a further 77 articles did not meet the inclusion criteria. We also attempted to find additional eligible articles from the references from the included articles, but without success. Finally, 30 articles were selected in this meta-analysis. All articles were retrospective clinical researches, among them 13 articles reported outcomes of mononostril approaches, 15 articles described outcomes for binostril approaches, and two articles were directly comparing the mononostrill and the binostril approach. In total, the 30 articles included 4865 patients. A detailed flowchart and the characteristics of the included studies are summarized in S1 Fig and S1 Table, respectively.

Patient characteristics

In the mononostril cohort, there were 2285 patients, mean 45.5 years of age, mean follow-up 30.9 months, 39% male, and proportion of macroadenomas tumors was 75%. In the binostril cohort, there were 2580 patients, mean 48.7 years of age, mean follow-up 35.1 months, 44% male, and proportion of macroadenomas tumors was 83%. We failed to discover any statistically significant difference in the two approaches regarding patient characteristics (S2 Table).

Outcomes analyses

Mononostril or binostril

The primary goal of pituitary surgery is to resect the adenoma as much as possible while minimizing the extent of possible surgical trauma to patients. Compared to traditional microsurgery, endoscopic endonasal pituitary surgery provides a wider angle view, more flexible mobile and close-up shots of the anatomical structures with the ability to obtain significantly higher cure rate, have less complications, as well as shorter operation durations, less blood loss, and shorter hospital stays [12,13]. However, neurosurgeons are dissatisfied with the current situation, resulting in a desire to find a maximally effective surgery approach to resect pituitary adenomas. Some doctors advocate the mononostril approach without crossing the nasal septum to limit the trauma of nose, whereas other doctors prefer the binostril approach to fully exposes the anatomy of the sella turcica and magnify the operative field. The demerits and merits regarding these two wide used surgery approaches have become a hot topic of discussion.

For general pituitary adenomas

This meta-analysis did not demonstrate any significant difference between the mononostril and the binostril approaches within GTR (included GTR of macroadenomas), hormonal remission rate, improvement in visual function, CSF leak, temporary diabetes insipidus, and meningitis. Recognizing the limitations of a meta-analysis in drawing firm conclusions, we assumed it could be the following reasons: 1) advanced endoscopic equipment: as the development of neuroendoscopic technique continues, the high resolution endoscope can provide an excellent panoramic view of the surgical field and enhance the chance of total resection of the tumor [14–16]. Most surgeries were performed with the use of different angled endoscopes for visualization of the parasellar and suprasellar areas, thus allowing surgeons to identify clearly anatomic structures even in a very limited operative field [2,3, 17–25]; 2) Improvements of mononostril approach: although in the mononostril approach, it was unnecessary did no need to cross the nasal septum, many surgeons usuallydetached and pushed the nasal septum to the contralateral nasal cavity to obtain adequate operating space[6,15,17,22,23,25]. Additionally, some surgeons used a speculum to narrow the surgical corridor [2]; 3) other reasons: such as the cooperation with otorhinolaryngologist, the use of intraoperative navigation system or MRI [3,15,23], the characteristic of pituitary adenomas (the majority is soft). All these factors could increase the efficiency of the mononostril approach surgery. These aforementioned parameters demonstrate that the mononostril approach is able to cope with most pituitary adenomas. In other words, the advantage of the binostril approach is unclear within general pituitary adenomas.

In spite of this, this meta-analysis found that the binostril approach had less anterior pituitary insufficiency and temporary diabetes insipidus than mononostril approach. In our opinion, crossing the middle turbinate could enlarge the operational space and improve the view of surgical field to a certain degree; this advantage could result in more freedom of movement with the instruments and makes surgeons more comfortable. Therefore, the binostril approach could make increase identification clarity within the structure while minimizing trauma of the normal pituitary gland.

For invasive pituitary adenomas

Han et al. reported that the rate of GTR was 0% in mononostril group and 28.6% in binostril group (P>0.05) for macroadenomas with Hardy grade 4 and/or Knosp grade 3 and 4 [6]. Linsler et al reported that GTR was 0% in mononostril approach for patients with invasive and giant tumors [4]. Dallapiazza et al reported that rate of GTR was 28% in binostril approach for patients with tumors with Knosp grade 3 and 4 [26]. Oertel et al [2] used the binostril approach for better visualization and to improve the manipulation of the lesion only in few cases: one clivus chordoma, one prepontine arachnoidal cyst and one brainstem cavernous malformation. Rudnik, A et al [20] and D'Haens, Jean et al [21] reported that GH hormonal remission rates were 8.3% (1/12) and 40% (2/5) in the mononostril approach for tumors with Knosp grade 3 and 4, respectively. Whereas John A. Jane [27] and Robert M. Starke [28] reported that GH hormonal remission rates were 33.3% (5/15) and 45% (9/20) in the binostril approach for the same grade tumors, respectively. For tumors with Hardy grade 3 and 4, the GH hormonal remission rates were 33.3% (5/15) in Gondim, J.A et al’s mononostril approach [23] and 52.1% (25/48) in Derya Brcu Hazer et al’s binostril approach [29].

The above literature showed that the binostril group seemed to have a tendency towards better outcomes for invasive macroadenomas pituitary adenomas. Unfortunately, subgroup analysis of the outcome based on Hardy grades and/or Knosp grades was not attained in our analysis because of the lack of suitable data. The Hardy grades and Knosp grades of the tumors was not recorded in most reports. When grades were reported, it was infrequently correlated to postoperative outcomes.

Nasal complications

Since the binostril approach crosses the nasal septum, it results in greater surgical trauma to the nasal cavity. This meta-analysis revealed that the binostril group had a higher rate of epistaxis (1.5% versus 0.4%, p<0.05) than mononostril group. The binostril approach placed the posterior nasal branch of the sphenopalatine artery [27] and some nasal septum nerves at a greater risk than the mononostril approach. Therefore, the risks of epistaxis, olfactory damage, and postoperative discomfort are possibly increased. For olfactory damage, many studies within the binostril group reported some relevant data, for example, Charalampaki et al reported hyposmia was found in 10% of cases and anosmia was found in 2% [30]. Alterations in taste or smell was found in25.7% of cases and 30%was found in Starke et al´s [28] and John A. Jane et al´s [27] report, respectively. Dallapiazza et al reported that alteration in smell was found in 13% of cases [26]. Through above systematic review, we could found that the olfactory damage was more common in the binostril group. However, the number of studies relevant to olfactory damage was rare in the mononostril group; we postulate that olfactory damage was not a common complication of the mononostril group.

We also found the incidence of sinusitis is similar between these two approaches. This was most likely because sinusitis was related to many factors, such as materials of sellar floor reconstruction, antibiotic therapy, clearance of blood clots, and post operation care. We attempted to conduct a more detailed analysis of nasal complications, such as septal perforations, mucosal adhesions, and other sinonasal complaints. Unfortunately, we were unable to perform adequate comparisons with the available data, especially because of the lack data within the mononostril approach.

Other parameters

Additionally, we discovered that the length of hospital stay in binostril group was shorter (3.2 days versus 4.4 days, p = 0.000) than that in the mononostril group. According to our experience, postoperative complications is an influential factor of increasing the length of hospital stay, such as postoperative CSF leaks and diabetes insipidus. The higher rate of anterior pituitary insufficiency and temporary diabetes insipidus we found in the monostril group would increase hospitalization days. Another important factor was the standard of discharge was not same in different medical center. Nevertheless, the exact cause for such result requires further research.

Endoscopic endonasal pituitary surgery is rapidly developing in surgical discipline. With its main targets being more minimally invasive and better results. We believe that this meta-analysis affords new reference evident to the choice of mononostril or binostril endoscopic endonasal transsphenoidal approaches through the inclusion of the vast amount of studies and patients in the analyses than previous reports. Finally, future comparative studies would provide stronger inferences and more credible data, such as meta-analyses with less bias, more rigorous subgroup analysis and multicenter comparisons with attention to surgical expertise, and volume.

Limitations

Almost all included reports are single-armed, uncontrolled studies with an obvious bias for lack of a control group such as no measured publication bias. It is impossible to discern whether authors tend to publish results that appear more favorable from the literature. We were unable to obtain the available data recorded the actual size of the tumors, and we did not know whether the mean size of pituitary adenomas had a statistically significant difference between the two groups. Very few studies reported correlate outcomes with primary/revision surgery and some statistical results of this study were deficient in adequate data. We also had to recognize that this analysis represented only the results of early outcomes and complications, and there were rarely long-term studies following these patients beyond the initial postoperative period published.