Effectiveness and safety of ventriculoperitoneal shunt versus lumboperitoneal shunt for communicating hydrocephalus: A systematic review and meta‐analysis with trial sequential analysis

Abstract Introduction The current standard surgical treatment for cerebrospinal fluid diversion is a ventriculoperitoneal shunt (VPS) implantation. Lumboperitoneal shunts (LPS) are an alternative treatment for communicating hydrocephalus. Prior studies comparing these two included a limited number of participants. Methods We performed a meta‐analysis determined the treatment failure, complications and effectiveness of lumboperitoneal shunt for communicating hydrocephalus. We reviewed studies with clinical and imaging diagnoses of communicating hydrocephalus, all causes and subtypes of communicating hydrocephalus, and studies that analyzed the primary and secondary outcomes listed below. We included randomized controlled trials (RCTs), non‐RCTs and retrospective studies. We performed the meta‐analysis in R, using a random‐effects model and reporting 95% confidence intervals. Results Data from 25 studies, including 3654 patients, were analyzed. The total complication rates were 12.98% (188/1448) for lumboperitoneal shunt and 23.80% (398/1672) for ventriculoperitoneal shunt. The odds ratio for lumboperitoneal shunt versus ventriculoperitoneal shunt complication rates was 0.29 (95% CI 0.19 to 0.45, p < 0.0001), and the I2 was 72%. The shunt obstruction/malfunction rate was 3.99% (48/1204) for lumboperitoneal shunt and 8.31% (115/1384) for ventriculoperitoneal shunt (Odds ratio 0.54, 95% CI 0.37 to 0.79, p = 0.002, I2 = 0%). Based on the Modified Rankin Scale score, there were no differences in effectiveness between lumboperitoneal shunt and ventriculoperitoneal shunt. Nevertheless, lumboperitoneal shunt improved radiological outcomes. Conclusions This analysis demonstrated that lumboperitoneal shunt is a safe and equally effective choice for treating communicating hydrocephalus. More studies are needed to confirm the safety of lumboperitoneal shunt.


| INTRODUC TI ON
Since the first hydrocephalus shunt in 1956, this effective surgical treatment has been developed with biocompatible material and valve system improvements. 1 Ventriculoperitoneal shunt (VPS) implantation is the standard surgical treatment for cerebrospinal fluid (CSF) diversion, especially in North America and Europe. 2 Although VPS provides effective CSF diversion and immediate symptomatic improvement, complications due to the intracranial placement, including brain hemorrhages, brain damage, infections, coma and, rarely, death, may occur. 3 Shunt malfunction and complications causing high revision rate is also a risk. 4 Although LPS has been available for 60 years, this shunt has not gained the same status.
LPS provides an alternative for patients with communicating hydrocephalus. 5 The use of LPS has increased in recent years due to the avoidance of brain damage and extracranial access. LPS is the most popular treatment for idiopathic normal pressure hydrocephalus (iNPH) in Japan. 6,7 The new LPS design, including a programmable valve setting, may provide better safety and lower adverse effects than VPS for patients with hydrocephalus. 8 Only a few studies have compared the complications and efficacy of the LPS and VS. Therefore, we conducted a systematic review and meta-analysis to compare the safety and adverse effects of LPS and VPS in patients with communicating hydrocephalus. We also compared clinical and radiological improvements after treatment with LPS and VPS.

| ME THODS
The meta-analysis followed the reporting guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 9 report. We included randomized controlled trials (RCTs), non-RCTs and retrospective studies. Although we intended to enroll only RCTs, these trials were scarce. Therefore, we included nonrandomized concurrent trials and retrospective studies in the analysis.

| Searchstrategy
We searched the following sources for eligible reports in any language: the Cochrane Library, PubMed, Embase, Clini calTr ials.gov, Cochrane Central Register of Controlled Trials, WanFang database and the China National Knowledge Infrastructure database.
The search using free texts and medical subject headings included "hydrocephalus," "communicating hydrocephalus," "lumboperitoneal shunt," "ventriculoperitoneal shunt," "shunt," "complications," "adverse events," and "efficacy." Two review authors (YJ Ho, WC Chiang) independently searched the databases. We identified other potentially eligible trials, studies or ancillary publications by searching the reference lists of the retrieved trials, reviews and metaanalyses. We also searched gray literature on Open Gray.

| Inclusion/exclusioncriteria
We included studies with a head-to-head comparison between VPS and LPS. We reviewed studies with clinical and imaging diagnoses of communicating hydrocephalus, all causes and subtypes of communicating hydrocephalus, and studies that analyzed the primary and secondary outcomes listed below.

| Selectionofstudiesanddata extractionandmanagement
Two review authors (YJ Ho and WC Chiang) independently extracted data using data collection forms designed to capture information specific to this review. We performed the meta-analysis in R (R Core communicating hydrocephalus, lumboperitoneal shunt, meta-analysis, ventriculoperitoneal shunt variance τ 2 in continuous outcomes and the DerSimonian-Laird estimator 12 in binary outcomes. We used Knapp-Hartung adjustments (Knapp & Hartung) 13 to calculate the confidence interval around the pooled effect. Zero cells were dealt with using a continuity correction by Gart and Zweifel. 14 We performed a sensitivity analysis using a Bayesian approach with the Markov Chain Monte Carlo method. We present the Doi plot with the Luis Furuya-Kanamori index 15 for each endpoint for publication bias.

| Assessmentofriskofbiasinincludedstudies
Two review authors independently assessed the risk of bias of the included studies. We used the Cochrane "RoB 2" assessment tool for randomized trials and ROBINS-1 for nonrandomized studies. The intervention incidence and the control arms were determined from all of the enrolled studies.

| Gradingofthecertaintyofevidence
We evaluated every result in the RCT subgroups using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) 16 methodology. The overall certainty of evidence (CoE) was judged by five downgrading and three upgrading domains. The level of CoE was classified as high, moderate, low or very low.

| Literaturesearch
The study selection process flow diagram is presented as a PRISMA flowchart ( Figure 1). From the initial literature search, we retrieved 531 articles; 507 of these studies were duplicates or irrelevant. A total of 25 studies were identified. Manual searching of the reference lists of these studies did not yield new eligible studies. The characteristics of the included studies are presented in Table 1.

| Includedstudies
Three studies were RCTs, one was a prospective nonrandomized trial, 20 were retrospective studies and one was a nationwide epidemiological survey. Participants were diagnosed with communicating hydrocephalus by clinical symptoms and image studies (CT or MRI). The detailed surgical techniques were slightly different but included no differences that would alter the results among studies, except one study focusing on laparoscopy-assisted VPS and LPS.
The outcomes included clinical symptoms, intraoperative parameters, perioperative parameters, length of hospital stay, radiological improvement, complications, National Institutes of Health Stroke Scale/Score, mRS and the Stein and Langfitt hydrocephalus grade.

| Riskofbiasinincludedstudies
All three RCTs had "some concerns." In the domain of outcome measurement, three RCTs had "some concerns" due to the lack of blinded assessors. Only five nonrandomized studies were at a "low" overall risk of bias; 13 nonrandomized studies were judged as "moderate" risk of bias and four were at "serious" overall risk of bias ( Figure 2).

| Meta-analysisofallstudies
Data from 25 RCTs, non-RCTs, prospective cohorts and retrospective studies, including 3654 patients, were analyzed. LPS was associated with a lower incidence of total complications compared with the incidence in VPS. There was low heterogeneity across the included studies. The total complication rate was 12.98% (188/1448) for LPS and 23.80% (398/1672) for VPS. The odds ratio was 0.29 (95% CI 0.19 to 0.45, p < 0.0001) and the I 2 was 72% (Figures S1-S7).

| Primary outcomes: adverse events
The infection rate for LPS was 1.53% (24/1568), which was significantly lower than the infection rate for VPS of 5.41% (97/1792).

| Secondary outcomes: neurological disability improvement and radiological outcome assessment
No differences in effectiveness between LPS and VPS were detected, based on the mRS score ( Figure S2A). However, radiological outcomes improvement rate was better after LPS than VPS ( Figure S2B).

| Secondary outcomes: neurological disability improvement and radiological outcome assessment
No RCT investigated neurological disability improvements.

| Publicationbias
A review of the Doi plots with the Luis Furuya-Kanamori index for each endpoint could not exclude the potential for publication bias for, total complication rates ( Figure S3), shunt obstruction/malfunction rates ( Figure S4), hemorrhage rates ( Figure S5) and radiological improvement rate ( Figure S6). The results did not change between fixed or random-effects models. Doi plots also validated publication bias in RCT subgroup. Publication bias was strongly suspected in hemorrhages ( Figure S7).

| Trialsequentialanalysis
TSAs were conducted for all RCT subgroups' endpoints. The cu-

| DISCUSS ION
Our meta-analysis demonstrated that adverse effects occur less frequently after LPS implantation than after VPS implantation for the effectiveness of treatment. We compared the safety, neurologic disability and radiological improvement between the two shunts.
Contrary to our initial hypothesis, the two shunts had similar safety and effectiveness. LPS appeared to be better when comparing total complications, including infections, seizures, shunt obstructions and hemorrhage. The use of LPS also resulted in better radiological outcomes. The results from the RCT subgroup analysis, which TSA verified, also favored LPS, which yielded lower total complication rates. Despite still not reaching the required information size, the cumulative z-curve will need to pass through the futility area to reach the area favoring VPS, leaving little chance to overthrow the hypothesis that LPS is superior than VPS as far as complication rates are concerned.
To the best of our knowledge, this meta-analysis is the first report to demonstrate decreased complications in patients treated with LPS. The total complication rate for LPS was 12.98%, and the total complication rate for VPS was 23.80%. Previous studies also revealed similar complication rates of VPS ranging from 13% to 38%, which mostly occurred in the first year after surgery. 17,18 The infection rate for LPS (1.53%) was lower than the infection rate for VPS (5.41%). Infection rates varied between studies, from 1% to 9%. 19,20 Obstruction/malfunction rates were also lower for LPS (3.99%) than   Unfortunately, only five studies included in this meta-analysis mentioned overdrainage but all without a clear definition. Moreover, some of them combined overdrainage with underdrainage into a category. We chose not to pool these outcomes together due to their ambiguity. However, it was indeed a vital topic that should be discussed. More studies are expected to give us more insight into it. Last, our results are rated low and very low CoE when evaluated with GRADE methodology ( Table 2). Again, more high-quality and significant studies are needed.

| CON CLUS ION
Our meta-analysis indicates that LPS is a safe and equally effective treatment for hydrocephalus compared with VPS. LPS had a lower complication rate, including lower infection, seizure, shunt obstruction/malfunction and hemorrhage rate, compared to VPS. Suppose more high-quality studies in the future confirm these beneficial results. In that case, LPS could be a good alternative to VPS or even a first-line treatment option for patients with communicating hydrocephalus who are not a good candidate for VPS.

AUTH O RCO NTR I B UTI O N S
Yi-Jen Ho involved in conceptualization, data curation, formal analysis, methodology, software, visualization, writing the original draft, review, and editing. Wen-Chun Chiang involved in data curation, formal analysis, methodology, and software. Hsin-Yi Huang and Shinn-Zong Lin involved in supervision. Sheng-Tzung Tsai involved in supervision, review, and editing.

ACK N OWLED G M ENTS
This manuscript was edited for English language by Enago, Crimson Interactive Inc. (USA). We thank the following individuals for their expertise and assistance throughout all aspects of our study and for their help in writing the manuscript, Pei-Chun Lai and Chun-Yu Chang.

CO N FLI C TO FI NTE R E S T
The authors declare that they have no conflict of interests and have no funding to support this study.

DATAAVA I L A B I L I T YS TAT E M E N T
The data that supports the findings of this study are available in the supplementary material of this article.