The Mann-Whitney U test or Fisher’s exact test was used to compare differences between two independent groups. In addition, all results were calculated with the standard deviation and odds ratio (OR) of the 95% confidence interval (CI). The success rate and shunt revision rate of the endoscopic procedure were measured by the Kaplan-Meier method, and the difference was compared using the log rank test. Univariate logistic regression analysis of the endoscopic procedure was conducted with various factors as independent variables. The ETVSS was calculated as a continuous value for all patients, and the risk of failure was evaluated. All statistical analyses were performed using SPSS ver. 25.0 software (SPSS Inc., Chicago, IL, USA), R (version 3.6.1), and RStudio (version 1.4.1717), and a p-value <0.05 was considered statistically significant.

In this series, there were 36 patients, of which 21 (58%) were male and 15 (42%) were female. The age at shunt insertion ranged from 1 day to 15.4 years, with a mean age of 2.3 years. The mean follow-up duration was 3.0 years (range, 3 days to 18.1 years) after the endoscopic procedure.

Twenty-nine patients, 18 males and 11 females, with previous shunt operations underwent ETV (Table 1). At the time of shunt insertion, the age ranged from 1 day to 15.4 years (mean, 2.4 years). Twenty-one patients were under 1 year old when they underwent shunt operation. The mean age at the time of ETV was 13.1 years (range, 0.7 to 29.6 years). Three patients less than 1 year of age underwent ETV. The etiology of hydrocephalus was congenital aqueductal stenosis in seven cases and aqueductal compression by tumor in nine cases. Nine patients had suspected malabsorption or obstruction due to germinal matrix/intraventricular hemorrhage. Tuberculosis meningitis occurred in one case, and bacterial meningitis occurred in three cases. Of the 29 patients, four patients did not have a MRI performed. Twenty-three patients showed aqueductal flow obstruction, and two patients showed fourth ventricle outlet obstruction. Seventeen out of 25 patients were confirmed to have third ventricular floor bowing (Fig. 1). The mean of the third ventricular floor bowing was 5.3 mm below the reference line. Eight of 17 patients with third ventricular floor bowing failed ETV (success rate, 52.9%). Moreover, three of eight patients who failed ETV had CSF flow obstruction on cine MR before surgery. One in eight patients without third ventricular floor bowing failed ETV (success rate, 87.5%). All patients without third ventricular floor bowing but who were suspected to have CSF flow obstruction on cine MR (Fig. 2) showed 100% success in ETV. Nineteen patients (65.5%) remained shunt revision free. The 5-year shunt revision-free survival rate was 69% (95% CI, 0.54–0.88, Fig. 3A). The other 10 patients needed shunt revision.

Seven patients, three males and four females, with previous shunt operations underwent endoscopic septostomy. At the time of shunt insertion, the age ranged from 0.2 to 12 years (mean, 3.9 years). Four of the patients were under 1 year old when they underwent shunt operation. The mean age at the time of the endoscopic septostomy was 11.9 years (range, 0.5 to 29.5 years). Only one patient less than 1 year of age underwent an endoscopic septostomy. The etiology of hydrocephalus was congenital porencephaly in two cases and postoperative hydrocephalus in three cases (two tumor surgeries and one epilepsy surgery). One patient had suspected malabsorption or obstruction due to germinal matrix/intraventricular hemorrhage, and one patient had bacterial meningitis (Table 2). Four patients remained free of shunt revision or addition. The 5-year shunt revision-free survival rate was 57% (95% CI, 0.3–1.0; Fig. 3B). The other three patients needed shunt revision. All three cases of unilateral hydrocephalus at the initial shunt operation showed symptom improvement after septostomy. Two patients showed evidence of isolated lateral ventricle, one with a trapped trigone area and the other with a trapped fourth ventricle case. These two patients received endoscopic septostomy along with endoscopic fenestration to create passages for entrapment. None of them needed additional shunt revision or addition. No complications were associated with the endoscopic procedures.

Most endoscopic procedure failures occurred within a month. Seven patients needed shunt revision immediately after ETV. For patients who had a problem within a month, the mean time to shunt revision was 16 days (range, 3 to 28 days). Except for one patient, every failure occurred within 3 months. In the cases of endoscopic septostomy, two patients failed within a month, and one patient failed within 2 months. They needed shunt revision after an average of 23 days (range, 12 to 44 days). Similar to the ETV results, failure was revealed within a month or 2.

The univariate results are shown in Tables 3 and 4. The higher the OR was, the higher the probability of shunt revision. Age at shunt insertion, etiology of hydrocephalus, reason for ETV, previous shunt duration and ETVSS were analyzed (Table 5). No significant predictors of procedure success could be found in our study.

Studies reporting ETV results after VP shunt malfunction show a success rate of 42–84% (Table 6). The overall pooled success rate was 68.2% [39]. The reported complication rate was 6.1% (0–15.9%). Studies including adult patients had relatively high ETV success rates [5,29,35,38]. In 2009, a multivariate analysis was conducted on 618 children in the UK, Canada, and Israel to predict the possibility of ETV success [22]. They reported that age, etiology, and previous shunt were the most important factors in predicting ETV success and suggested a way to calculate the success rate through the ETVSS. In our series, all the patients had previous shunts, and the overall ETVSS varied from 40 to 90, with a mean of 72.4, which is not far from the real success rate of ETV. Our success rate was 65.5%, which is similar to the results of other existing studies.

As Marton et al. [25] reported, age at ETV does not have a statistically significant effect on ETV success in our study population. Most ETV failures were revealed within the first month. Seven of 10 patients needed shunt revision immediately. Except for one patient, all failure results were confirmed within three months. The mean time to ETV failure was 16 days. Many studies have also reported that failure occurs within a month [8,24,27,29]. Many authors have reported that a previous etiology of hydrocephalus affects the success of ETV [15,25,29]. However, our data show that regardless of the etiology of hydrocephalus, whether congenital aqueductal stenosis, tumor-related, posthemorrhagic, or postinfectious, none of them affected the success of ETV. This result may be due to selective indications of endoscopic procedures when there are findings that may be suspected to be obstructive hydrocephalus, regardless of the etiology. If there were some indicators, such as third ventricular floor bowing on MRI or CSF flow obstruction on cine MR, the hydrocephalus may be considered an obstructive type. The number of patients in each group was small; thus, nothing was statistically significant. Previous shunt duration was not a factor influencing ETV success.

In the present study, we performed endoscopic septostomy on previously shunted patients with shunt malfunction when secondary obstruction of the foramen of Monro was suspected. Patients had an increased lateral ventricle on the side containing the shunt proximal catheter and a relatively normal contralateral ventricle size. It was obviously caused by shunt malfunction. However, obstruction of foramen of Monro was not confirmed.

Unilateral hydrocephalus is a rare form of hydrocephalus. It mainly develops after intraventricular hemorrhage or meningitis [14,20,33]. In patients with VP shunts, the causes include shunt infection, shunt overdrainage and damage to the ependymal lining during shunt catheter insertion [14,18,28,31]. However, in this study, shunt overdrainage can be excluded because the ipsilateral lateral ventricle containing the shunt catheter inside was larger than the opposite side of the lateral ventricle. It has various forms according to the obstruction site in the ventricular system. The traditional treatment of hydrocephalus with a previous VP shunt malfunction included adding or replacing the shunt [1,14,18,20]. However, multiple shunt insertions lead to higher rates of shunt failure or complication [23,37]. Therefore, neurosurgeons are trying to reduce the number of initial shunts and avoid additional shunts as much as possible by introducing endoscopic septostomy. Uncomplicated asymmetric hydrocephalus was successfully cured with endoscopy. Case series on endoscopic septostomy showed a success rate of 53–87% with asymmetric hydrocephalus [2,16,30]. In the present study, fenestration of the septum pellucidum with a previous VP shunt occlusion showed a success rate similar to that reported in the literature. In some literature, success was defined even if there had been multiple endoscopic procedures on the isolated ventricle. However, in our group, there were only two cases that were tried several times. We categorized the two cases as failures because they needed shunt insertion during the course of treatment. It was difficult to compare simple success rates because the complexity in the literature varies from patient to patient, and the etiology of isolated unilateral hydrocephalus was diverse.

Imaging can be considered at the surgeon’s discretion to assist with surgical planning. A constructive interference in steady state (CISS) study can clearly illustrate even paper-thin membranous obstruction within the aqueduct and fourth ventricle outlet. Since T2-cine MRI gives information on CSF flow, it is very helpful to perform it when obstruction is suspected. With these imaging techniques, we can also visualize the third ventricular floor bowing and the thin membranous structure within the isolated ventricle. When we selected a patient, clinical findings with a high level of possibility of ETV or endoscopic septostomy success results were considered. To perform ETV, we looked at the image and particularly considered the third ventricular floor bowing. Seventeen out of 25 patients had third ventricular floor bowing. Eight of them failed ETV, with a success rate of 52.9%, and one in eight patients without third ventricular floor bowing failed ETV. This result seemed to show ETV to be more effective in patients without third ventricular floor bowing. It did not show a statistically significant relationship. This may be because the turgor of a ventricle decreases after a long duration of shunt, and despite having obstructive hydrocephalus, the third ventricular floor bowing may not have been observed. Although we performed ETV on those who were deemed likely to succeed, sometimes we failed and had to perform shunt revision. Kim et al. [21] explained a possible mechanism of shunt dependency as ‘chronic idling’. After long-term shunting, CSF absorption regresses and needs to be shunted. Even if a hole is made successfully, it can fail due to shunt dependency; therefore, the patient and guardian need a sufficient explanation about the possible problems in advance.

Three of seven patients had unilateral hydrocephalus at the time of the first shunt operation. All of their symptoms improved through septostomy when malfunction occurred. Therefore, performing endoscopic septostomy is reasonable for patients who initially have unilateral hydrocephalus at the first shunt insertion.