Cerebrospinal Fluid Leak after Transsphenoidal Surgery: A Systematic Review and Meta-analysis

Abstract

Background  Cerebrospinal fluid (CSF) leak is widely recognized as a challenging and commonly occurring postoperative complication of transsphenoidal surgery (TSS).

The primary objective of this study is to benchmark the current prevalence of CSF leak after TSS in the adult population.

Methods  The authors followed the PRISMA guidelines. The PubMed, Embase, and Cochrane Library databases were searched for articles reporting CSF leak after TSS in the adult population. Meta-analysis was performed using the Untransformed Proportion metric in OpenMetaAnalyst. For two between-group comparisons a generalized linear mixed model was applied.

Results  We identified 2,408 articles through the database search, of which 70, published since 2015, were included in this systematic review. These studies yielded 24,979 patients who underwent a total of 25,034 transsphenoidal surgeries. The overall prevalence of postoperative CSF leak was 3.4% (95% confidence interval or CI 2.8–4.0%). The prevalence of CSF leak found in patients undergoing pituitary adenoma resection was 3.2% (95% CI 2.5–4.2%), whereas patients who underwent TSS for another indication had a CSF leak prevalence rate of 7.1% (95% CI 3.0–15.7%) (odds ratio [OR] 2.3, 95% CI 0.9–5.7). Patients with cavernous sinus invasion (OR 3.0, 95% CI 1.1–8.7) and intraoperative CSF leak (OR 5.9, 95% CI 3.8–9.0) have increased risk of postoperative CSF leak. Previous TSS and microscopic surgery are not significantly associated with postoperative CSF leak.

Conclusion  The overall recent prevalence of CSF leak after TSS in adults is 3.4%. Intraoperative CSF leak and cavernous sinus invasion appear to be significant risk factors for postoperative CSF leak.

Introduction

Cerebrospinal fluid (CSF) leak is still widely recognized as a commonly occurring postoperative complication of transsphenoidal surgery (TSS). CSF leak is associated with various complications including meningitis, intracranial infection, and CSF hypotension syndrome. These complications often lead to additional health care costs and substantial morbidity as they may require prolonged hospitalization, reoperation, and external lumbar drainage (ELD). ^{1 2} Grotenhuis reports an additional cost of €10.243 per patient with postoperative CSF leak for transsphenoidal procedures. ² The prevalence of postoperative CSF leak seems increased in patients with an elevated body mass index (BMI) and/or increased intracranial pressure. ³ However, the exact risk of this complication and variables of influence are not clearly defined and reported prevalence rates vary widely (0–40%). ⁴ CSF leak rates among patients undergoing TSS are regarded to be higher than for transcranial neurosurgical procedures due to additional risk factors, such as gravity and a lack of anatomical barriers provided by watertight dural closure and subcutaneous and cutaneous closure. ⁵ However, techniques of closure have been significantly improved by using a vital nasoseptal mucosal flap, the use of sealing materials, and improved neurosurgical techniques. ^{6 7 8 9} TSS has been an evolving field over the last decades, therefore complication rates should be investigated in recent literature and frequently updated as advancements in the surgical technique continue. The objectives of this study are to benchmark the prevalence of CSF leak after TSS in the adult population in the past 5 years, and to define variables affecting this risk.

Methods

The authors followed the PRISMA guidelines for this systematic review and meta-analysis. ¹⁰

Search Strategy and Study Selection

We performed a literature search in the PubMed, Embase, and Cochrane Library databases for articles reporting CSF leak after TSS until April 1, 2020. A combination of free, controlled, and Mesh/Emtree terms for TSS and CSF leak, such as “Transsphenoidal” OR “Endoscopic endonasal” AND “Cerebrospinal fluid leak” OR “Cerebrospinal fluid rhinorrhea,” was used to form a search string ( Appendices A–C [available in the online version], for the search strings per database).

Articles reporting original studies published since 2015 on the adult population reporting CSF leak rates after TSS written in English or Dutch were included. The timeframe 2015 to 2020 was chosen with the aim to provide an up-to-date analysis of CSF leak after TSS and to expand on the existing literature on this topic. ⁴ Extended procedures and use of dural sealants were no restriction for inclusion. Studies including CSF fistula repairs or biopsies were excluded. Furthermore, case reports ( n <30) were excluded, as these were not considered strong evidence due to the risk of publication bias and selected populations.

Two authors (R.S. and E.M.H.S.) independently screened titles and abstracts for eligibility, after which full-texts of all potentially eligible studies were assessed for inclusion. No disagreement regarding the inclusion of an article after full-text assessment was encountered. The reference lists of all included studies and relevant reviews were cross-checked for additional eligible articles.

Data Collection

We extracted the following data from the included studies: study characteristics (authors, publication year, inclusion period, design, country, center name, total number of patients, total number of surgeries); patient characteristics (mean age at surgery, number of females, mean BMI, mean follow-up duration, previous surgery at same site, type of sphenoid sinus, preoperative diabetes mellitus, use of immunosuppressive medication, use of blood thinners, preoperative hydrocephalus, preoperative pneumocephalus, history of skull base radiation, length of stay); surgery characteristics (indication [e.g., pituitary adenoma or craniopharyngioma resection], approach, extended or conventional [based on the article's definition], reconstruction technique, use of sealant, intraoperative placement of a CSF diversion shunt); tumor characteristics (type of tumor, maximal tumor diameter, invasive [Knosp grades 3 and 4] or not, suprasellar extension); outcome parameters (rate of intraoperative CSF leak and rate of postoperative CSF leak, as defined by the article). Studies with a noncomparative design were defined as case series. ¹¹ The study quality of case series was assessed using the National Heart, Lung and Blood Institute of National Institutes of Health (NIH) quality assessment tool for case series studies, ¹² whereas the Newcastle Ottawa Scale ¹³ was used for the quality assessment of cohort studies. Studies with fewer than six points were judged to be of poor quality, studies with six or seven points were deemed of fair quality and studies with more than seven points were classified as being of good quality. Each item was awarded one point if answered with “Yes” or a star.

Statistical Analysis

We performed a meta-analysis of prevalence using the Untransformed Proportion metric in OpenMetaAnalyst for Sierra, version 10.12. A binary random effects analysis using the DerSimonian-Laird method was applied if heterogeneity across studies was significant ( p <0.05). For nonsignificant heterogeneity across studies the binary fixed effects inverse variance model was used. For two between-group comparisons (microscopic versus endoscopic surgery and pituitary adenoma resection vs. other indication) a generalized linear mixed model was applied, using SAS version 9.4 (SAS Institute Inc), as these analyses involved comparisons of groups on study level. Heterogeneity across studies was ascertained through Higgins I. ^{2 14}

The prevalence of CSF leak after TSS with 95% confidence interval (CI) was the primary outcome measure in this study. For between group comparisons of patients with and without certain risk factors for CSF leak the outcome measures were odds ratios (ORs) with 95% CI. We performed three sensitivity analyses: (1) excluding Pines et al, ¹⁵ as this publication accounts for almost half of the total population, (2) a comparison between studies published between 2015 to 2017 and 2018 to 2020 to evaluate a learning curve, (3) high quality studies only.

Results

Included Studies

We identified 2,408 articles through the initial database searches after removing duplicates. Seventy articles met the inclusion criteria for this systematic review. Eight articles were excluded from the meta-analysis due to an overlapping population with another included article (the study with the largest sample size was included). ^{16 17 18 19 20 21 22 23} One article was manually added by hand-searching the reference lists of all included articles. The study selection process and reasons for exclusion are shown in Fig. 1 .

Fig. 1.

Flowchart demonstrating the study selection process.

The included studies yielded 24,979 patients who underwent a total of 25,034 transsphenoidal surgeries as some subjects had more than one surgery. This includes 262 extended procedures and 2,104 conventional procedures. In the remaining 58 articles insufficient information is provided to determine the number of extended and conventional surgeries. An overview of study characteristics is presented in Table 1 . Nineteen studies were judged to be of good quality, 36 studies of fair quality, and 15 studies of poor quality ( Supplementary Material 1 [available in the online version], for an overview of quality assessment).

Table 1. Overview of study characteristics of included studies.

Authors	Year	Study design	Approach	Indication	Population ( N  = 24,979)	Surgeries ( N  = 25,034)	CSF leak (%)	Study quality
Gondim et al 39	2015	RC	Endoscopic	Pituitary tumor	374	374	3.7	Good
Fathalla et al 40	2015	RC	Endoscopic and microscopic	Pituitary tumor	65	65	6.2	Good
Wang et al 22 a	2015	CS	Endoscopic	Pituitary tumor	1,166	1,166	0.6	Fair
Nie et al 41	2015	CS	Endoscopic	Pituitary tumor	52	52	0.0	Fair
Zhan et al 42	2015	RC	Endoscopic	Pituitary tumor	313	318	3.8	Fair
Ishii et al 25	2015	RC	Endoscopic	Pituitary tumor, craniopharyngioma, meningioma, chordoma, ependymoma	48	48	6.3	Poor
Park et al 24	2015	CS	Endoscopic	Pituitary tumor, Rathke cleft cyst, craniopharyngioma, meningioma, chordoma/chondrosarcoma, other	188	197	0.0	Poor
Chabot et al 43	2015	CS	Endoscopic	Pituitary tumor	39	39	10.3	Fair
Pinar et al 44	2015	CS	Endoscopic	Pituitary tumor	32	32	9.4	Fair
Sanders-Taylor et al 45	2015	RC	Endoscopic	Pituitary tumor	264	264	1.9	Poor
Pines et al 15	2015	RC	Unknown	Pituitary tumor	12,938	12,938	1.7	Poor
Xie et al 46	2016	RC	Endoscopic	Pituitary tumor	43	43	14.0	Good
Gao et al 47	2016	RC	Endoscopic and microscopic	Pituitary tumor	105	105	10.5	Fair
Freyschlag et al 48	2016	CS	Endoscopic and microscopic	Pituitary tumor	50	50	0.0	Fair
Park et al 49	2016	RC	Endoscopic	Pituitary tumor, craniopharyngioma, meningioma, chordoma, Rathke cleft cyst, other	106	106	9.4	Fair
Jang et al 50	2016	CS	Endoscopic	Pituitary tumor	331	331	1.8	Good
Zaidi et al 34	2016	RC	Endoscopic and microscopic	Pituitary tumor	135	135	1.5	Fair
Gondim, Albuquerque et al 18 b	2017	CS	Endoscopic	Pituitary apoplexy	39	39	0.0	Poor
Fnais et al 51	2017	RC	Endoscopic	Pituitary tumor, pituitary apoplexy, Rathke cleft cyst, craniopharyngioma, other	145	138	11.6	Fair
Ye et al 52	2017	RC	Endoscopic	Pituitary tumor	1,281	1,281	0.5	Poor
Karki et al 53	2017	RC	Microscopic	Pituitary tumor	123	123	15.4	Good
Wang et al 54	2017	RC	Microscopic	Pituitary tumor	51	51	0.0	Good
Sun et al 55	2017	CS	Endoscopic	Pituitary tumor	42	42	9.5	Poor
Ding et al 26	2017	RC	Endoscopic	Craniopharyngioma	33	33	18.2	Good
Zhou et al 56	2017	RC	Endoscopic	Pituitary tumor	492	492	1.2	Fair
Cebula et al 57	2017	PC	Endoscopic	Pituitary tumor	230	230	0.0	Fair
Levi et al 37	2017	RC	Endoscopic and microscopic	Pituitary tumor	221	221	5.9	Poor
Zoli et al 58	2017	CS	Endoscopic	Pituitary tumor	75	75	1.3	Fair
Fujimoto et al 59	2017	RC	Endoscopic	Pituitary tumor	161	162	4.9	Poor
Yano et al 23 c	2017	CS	Endoscopic	Pituitary tumor	32	34	5.9	Fair
Sasagawa et al 60	2017	RC	Endoscopic and microscopic	Pituitary tumor	78	78	1.3	Fair
Fishpool et al 61	2017	CS	Endoscopic	Pituitary tumor	32	32	0.0	Poor
Ajlan et al 62	2017	RC	Endoscopic	Pituitary tumor	176	176	4.5	Fair
Przybylowski et al 63	2017	RC	Endoscopic	Pituitary tumor	96	96	4.2	Good
Negm et al 19 d	2017	PC	Endoscopic	Pituitary tumor	41	41	2.4	Good
Shin et al 64	2017	CS	Endoscopic	Pituitary tumor	50	50	4.0	Fair
Patel et al 3	2018	RC	Endoscopic	Pituitary tumor, Rathke cleft cyst, craniopharyngioma, other	806	806	4.7	Fair
Eseonu et al 65	2018	CS	Endoscopic	Pituitary tumor	275	275	3.6	Good
Popov et al 66	2018	RC	Endoscopic and microscopic	Pituitary tumor	128	128	3.9	Fair
Hansasuta et al 38	2018	PC	Endoscopic	Pituitary tumor	183	220	3.6	Fair
Han et al 67	2018	CS	Endoscopic	Pituitary tumor	52	52	3.8	Good
Guo et al 68	2018	RC	Unknown	Pituitary tumor	53	53	9.4	Fair
Schuss et al 69	2018	RC	Endoscopic and microscopic	Pituitary tumor	255	255	6.7	Poor
Hajdari et al 70	2018	CS	Endoscopic	Pituitary tumor	170	170	8.2	Fair
Karamouzis et al 71	2018	CS	Endoscopic	Pituitary tumor	90	90	4.4	Fair
Lofrese et al 72	2018	RC	Endoscopic	Pituitary tumor	95	95	5.3	Fair
Cudal et al 73	2018	CS	Unknown	Pituitary tumor, other	47	47	6.4	Poor
Robins et al 74	2018	RC	Endoscopic	Pituitary tumor	142	142	0.7	Poor
Barger et al 35	2018	CS	Endoscopic	Pituitary tumor	43	43	2.3	Good
Wilson et al 75	2018	RC	Endoscopic	Pituitary tumor	135	135	0.0	Good
Rehman et al 76	2018	CS	Endoscopic	Pituitary tumor	63	63	15.9	Fair
Xue et al 77	2019	RC	Endoscopic	Pituitary apoplexy	79	79	12.7	Fair
Chen et al 16 e	2019	CS	Endoscopic	Pituitary tumor	79	79	5.1	Fair
Fallah et al 78	2019	CS	Endoscopic	Pituitary tumor	80	88	4.5	Good
Spina et al 79	2019	RC	Unknown	Pituitary tumor	336	336	0.6	Good
Shen et al 80	2019	CS	Endoscopic	Pituitary tumor	45	45	2.2	Fair
Eichberg et al 17 f	2019	CS	Endoscopic	Pituitary tumor	120	120	1.7	Fair
Chen et al 81	2019	RC	Endoscopic	Pituitary tumor	131	131	8.4	Good
Seltzer et al 21 g	2019	CS	Endoscopic and microscopic	Pituitary tumor	52	52	1.9	Fair
Azab et al 82	2019	RC	Microscopic	Pituitary tumor	205	205	2.9	Good
Memel et al 83	2019	RC	Unknown	Pituitary tumor	115	115	2.6	Fair
Rieley et al 84	2020	RC	Endoscopic	Pituitary tumor, other	427	427	13.1	Poor
Liu et al 36	2020	RC	Endoscopic	Pituitary tumor	189	189	6.3	Fair
Zhang et al 85	2020	CS	Endoscopic	Pituitary tumor	113	113	0.9	Fair
Tardivo et al 86	2020	RC	Endoscopic	Pituitary tumor	81	81	4.9	Good
Castaño-Leon et al 87	2020	RC	Endoscopic and microscopic	Pituitary tumor, other	187	187	5.3	Fair
Pangal et al 20 g	2020	CS	Endoscopic	Pituitary apoplexy	50	50	8.0	Fair
Parikh et al 88	2020	CS	Endoscopic	Pituitary tumor	334	334	3.9	Good
Tafreshi et al 89	2020	CS	Endoscopic	Pituitary tumor, Rathke cleft cyst, arachnoid cyst, xanthogranuloma	47	47	8.5	Poor
Cappello et al 90	2020	CS	Endoscopic	Pituitary tumor, craniopharyngioma, pituitary apoplexy, cyst, other	125	125	3.2	Fair

Abbreviations: CS, case series; CSF, cerebrospinal fluid; N , number; PC, prospective cohort; RC, retrospective cohort.

^a Excluded from primary analysis due to overlapping population with Ye et al. ⁵²

^b Excluded from primary analysis due to overlapping population with Gondim et al. ³⁹

^c Excluded from primary analysis due to overlapping population with Fujimoto et al. ⁵⁹

^d Excluded from primary analysis due to overlapping population with Wilson et al. ⁷⁵

^e Excluded from primary analysis due to overlapping population with Zhang et al. ⁸⁵

^f Excluded from primary analysis due to overlapping population with Chen, Sprau et al. ⁸¹

^g Excluded from primary analysis due to overlapping population with Memel et al. ⁸³

There was insufficient data from the included studies to perform reliable analyses for several risk factors: suprasellar extension, dural invasion, BMI, preventative ELD, reconstruction technique, age at surgery, sex, diabetes mellitus, use of immunosuppressive medication, use of blood thinners, preoperative hydrocephalus, preoperative pneumocephalus, history of skull base radiation, and sealant use.

Outcome and Risk Factor Analysis

The overall prevalence of postoperative CSF leak was 3.4% (95% CI 2.8–4.0%) ( Fig. 2 ). Heterogeneity across studies was substantial ( I ² 81.7).

Fig. 2.

Forest plot prevalence of cerebrospinal fluid leak.

The prevalence of CSF leak found in patients undergoing pituitary adenoma resection was 3.2% (95% CI 2.5–4.2%), whereas patients who underwent TSS for another indication (i.e., craniopharyngioma, meningioma, Rathke cleft cyst) had a CSF leak prevalence rate of 7.1% (95% CI 3.0–15.7%) (OR 2.3, 95% CI 0.9–5.7). Data on further specified diagnosis subgroups is too limited to analyze its influence on CSF leak. In this dataset, there is one study reporting on CSF leak on Rathke cleft cyst separately in which none of the 19 cases had CSF leak. ²⁴ Three small populations of craniopharyngioma's are included in which a total of six out of 49 patients had CSF leak (12.2%). ^{24 25 26} Two studies specify CSF leak in meningioma cases, of which one out of 15 patients had CSF leak (6.7%). ^{24 25} For 2,318 cases the CSF leak rate was not specified as per diagnosis subgroup.

Postoperative CSF leak was observed in 5.5% (95% CI 3.3–9.0%) of microscopically approached cases, as opposed to 4.0% (95% CI 3.0–5.2%) in endoscopic cases (OR 1.4, 95% CI 0.9–2.3).

CSF leak was present in 2.0% (95% CI 0.0–4.9%) of patients with previous TSS as compared with 0.4% (95% CI 0–1.0%) in patients without history of TSS (OR 0.9, 95% CI 0.2–4.5).

The prevalence of CSF leak in patients without intraoperative CSF leak was 0.7%, whereas 4.1% (OR 5.9, 95% CI 3.8–9.0) of patients with intraoperative CSF leak had a postoperative CSF leak. The prevalence of CSF leak in patients without cavernous sinus invasion was 0.5% (95% CI 0.0–1.1%), as opposed to 2.2% (95% CI 0.4–4.1) in patients with cavernous sinus invasion (OR 3.0, 95% CI 1.1–8.7) ( Table 2 ).

Table 2. Risk factors for postoperative CSF leak.

Outcome	OR	Lower bound	Upper bound	p -Value
Pituitary adenoma resection vs. other	2.3	0.9	5.7	0.07
Microscopic vs. endoscopic	1.4	0.9	2.3	0.18
History of TSS vs. no history of TSS	0.9	0.2	4.5	0.87
Intraoperative CSF leak vs. no intraoperative CSF leak	5.9	3.8	9.0	0.00 a
Cavernous sinus invasion vs. no cavernous sinus invasion	3.0	1.1	8.7	0.04 a

Abbreviations: CSF, cerebrospinal fluid; OR, odds ratio; TSS, transsphenoidal surgery.

^aSignificant.

Sensitivity Analyses

When the study from Pines et al ¹⁵ is excluded from the overall analysis, the results are comparable (3.7%, 95% CI 3.1–4.4%) to the primary outcome analysis (3.4%, 95% CI 2.8–4.0%).

The sensitivity analysis only including high quality studies also shows comparable results to the primary outcome analysis with a CSF leak rate of 3.6% (95% CI 2.3–4.8%).

Analysis of studies between 2015 and 2017 shows a CSF leak rate of 2.5% (95%CI 1.9–3.1%). The CSF leak rate is 4.6% (95% CI 3.4–5.8%) in studies published between 2018 and 2020. This does not provide evidence for a learning curve in studies published between 2015 and 2020.

Discussion

This meta-analysis shows that postoperative CSF leak occurs in 3.4% of adults undergoing TSS. Patients with cavernous sinus invasion are significantly more likely to develop postoperative CSF leak compared with those without cavernous sinus invasion (OR 3.0). Another risk factor for postoperative CSF leak is the presence of an intraoperative CSF leak (OR 5.9).

Historically, TSS is thought to pose high risk of CSF leak. The leak rate found in this study is considerably lower compared with a previous meta-analysis including studies published until 2015. ⁴ This previous meta-analysis reports a CSF leak rate between 7.5 and 10.5% for endoscopic endonasal tumor resections (including invasive sinonasal tumors) and 5% for pituitary surgery. ⁴ A similar trend was observed in another recent meta-analysis CSF leak following extended endoscopic endonasal approach for anterior skull base meningioma. ²⁷ In this study CSF leak decreased from 22 to 4% between 2004 and 2020. ²⁷

The reduced CSF leak rate found in the current study most probably results from a combination of three factors. First, improved surgical techniques; approach, sealants, endoscopic visualization and more widely used vascularized nasoseptal mucosal flaps. Second, improved awareness for CSF leak due to initial experiences after more broad indications for (endoscopic) TSS. Third, improved indication for TSS. Endoscopic surgery is no longer chosen for part of the larger tuberculum sellae meningioma and craniopharyngioma (with lateral or suprachiasmatic extensions) cases in most centers. ^{28 29}

No evidence for a learning curve is found within the timeframe of the current study (2015–2020). Analysis of subgroups based on publication year to define a learning curve is limited by the variation in inclusion periods of studies published in the same year, the difference in the number of publications from a certain time period reporting CSF leak and that no differentiation can be made on type of pathology based on year of publication which may influence results. Furthermore, publication bias cannot be excluded as a contributing factor to the difference in CSF leak rate observed between the current and previous meta-analyses. Yet, we do not believe this to be the main factor of influence, considering that publication bias may have also affected studies in the past. Furthermore, there is wide variance in leak rate reported in included studies and studies of small sample size, most vulnerable for publication bias, were excluded.

Moreover, the overall prevalence of postoperative CSF leak after TSS is considerably lower than that reported in meta-analyses for craniotomy (8%) and spinal surgery (14%). ^{30 31} However, this does not apply to all indications for TSS. CSF leak after TSS for other indications than pituitary adenoma resection is comparable to that found for cranial surgery, including infratentorial surgery, known to be more vulnerable to CSF leak. ³⁰ The relatively low overall leak rate in this meta-analysis may be a result of the relatively high number of pituitary adenoma's included, which may represent a patient population with few additional risk factors, ameliorating the risk of postoperative CSF leak.

Furthermore, a broad range of leak prevalences (0.0–18.2%) was reported by the included studies, resulting in substantial heterogeneity in the meta-analysis. The variation between studies could be explained by the fact that we have included TSS for various indications, which may differ in the presence of patient and surgery-related risk factors. This is reflected by the results of our subgroup analyses in which we find a relatively low CSF leak rate of 3.2% for pituitary lesions and a substantially higher prevalence of 7.1% for other indications.

However, CSF leak prevalences vary considerably within different subgroups, for example, including standard extradural pituitary surgery only. This can theoretically be explained by different surgical techniques and closure techniques.

Despite the significant improvement in surgical techniques, cavernous sinus invasion is still a considerable factor in CSF leak due to its need for extensive surgery. ³² This may indicate that tumors infiltrating the cavernous sinus are likely to cross the diaphragm thereby increasing the risk of postoperative CSF leak. As definitions of cavernous sinus invasion may vary, we classified Knosp grade 3 and 4 as invasive for this meta-analysis. This finding also further explains the difference in CSF leak between various surgical indications. As craniopharyngiomas and meningiomas are intradural intra-arachnoid lesions, there will certainly be intraoperative leak and thus higher risk of postoperative CSF leak, compared with extra-arachnoid pathology such as pituitary adenomas.

It was postulated by other authors that reoperation in patients with previous TSS tends to result in incomplete repair of intraoperative CSF leak, which may result in higher rates of postoperative CSF leak. ³³ Although CSF leak was present in 2.0% of patients with previous TSS as opposed to 0.4% in patients who underwent primary TSS, our meta-analysis does not find a significant association between previous TSS and postoperative CSF leak (OR 0.9, 95% CI 0.2–4.5). However, this effect may be influenced by the limited number of studies reporting TSS as a potential risk factor.

To our knowledge this systematic review includes the largest patient population thus far, including over 25,000 cases. Furthermore, it only includes publications from the past 5 years, thereby providing an up-to-date overview of the current situation with state-of-the-art techniques.

One limitation of this study is that the outcome CSF leak is defined differently across studies, this may further explain the variation in reported leak rates across studies. For example, Zaidi et al ³⁴ define CSF leak as “CSF leak requiring intervention,” for other studies CSF leak was taken into consideration only if confirmed by β2-transferrine testing. ^{35 36} Furthermore, the majority of included studies do not clearly describe their definition of CSF leak which may have caused differences in postoperative CSF leak percentages. Although, self-limiting CSF rhinorrhea is very rare, not all patients require intervention by reoperation, which may result in lower reporting of CSF leak in studies incorporating the need for surgical repair in their definition. ^{37 38}

Second, the results of the current meta-analysis are mostly based on retrospective cohort studies and case series, of which a substantial number is of limited sample size. The outcome of this meta-analysis may be subject to publication bias, contributing to the striking difference in postoperative CSF leak rate found for TSS compared with cranial and spinal surgery, as well as previous meta-analyses on TSS.

Third, some of the analyses are based on a limited number of cases. The analysis comparing endoscopic versus microscopic surgery could be performed for a limited number of studies, showing a higher leak rate for microscopic surgery, yet no significant difference. This result should therefore be interpreted with some caution. We find a substantially higher prevalence of CSF leak for TSS for indications other than pituitary adenoma resection. Again, this result is not statistically significant. Yet, the effect may be underestimated by the relatively low number and small sample size of studies reporting on other indications than pituitary adenoma resection.

Fourthly, no meta-analyses could be performed for several potentially important factors due to insufficient data, for example: suprasellar extension, dural invasion, BMI, microadenoma versus macroadenoma, use of preventative ELD, or reconstruction technique. We did not exclude studies based on their skull base reconstruction technique, which means that all types of reconstruction were included. Many recent studies have focused on different sellar reconstruction techniques. In the current review no analyses were possible to compare specific techniques as there was insufficient data from the included studies. Nevertheless, this factor could be a cause of the broad range of leak prevalences. Similarly, factors such as BMI, especially in combination with increased intracranial pressure, and extension of the tumor may have an influence on CSF leak. The effects of these potential influences could not be studied in the current review which limits the generalizability of the overall results.

Lastly, studies with fewer than 30 subjects were excluded from this meta-analysis. Therefore, studies on patients with rare pathology (such as tuberculum sellae meningioma) specifically, may be underrepresented in the current meta-analysis. This may have led to an underestimation of the overall CSF leak incidence after TSS.

The results of this meta-analysis underline that CSF leak after TSS for intradural and invasive lesions, such as craniopharyngiomas or tuberculum sellae meningiomas is a clinically relevant problem. To further improve the advancement of TSS for these indications effective solutions to prevent postoperative CSF leak are warranted. Future research should focus on effective closure techniques including augmented dural repair to prevent intraoperative CSF leak for this type of surgery especially. The outcomes of this meta-analysis could serve as a benchmark for future prospective studies on novel techniques to prevent CSF leak after TSS.

Conclusion

The overall prevalence of CSF leak after TSS in the adult population is 3.4%. Variables of influence are the presence of intraoperative CSF leak and cavernous sinus invasion.

Funding Statement

Funding E.M.H.S. receives a research grant through Polyganics B.V.