The innate natural history of brain tumors or their surgical interventions may lead to HC that necessitates temporary CSF diversion procedures, such as external ventricular drainage (EVD) and/or endoscopic third ventriculostomy (ETV), or lead to permanent CSF diversion. Although the incidence and risk factors for development of postoperative HC leading to VP shunt dependency of patients with and without HC prior to a craniotomy for brain tumor have been previously described [19,18], the long-term outcomes of VP shunts in brain tumor patients are largely unknown.
In our study, a total of 85 patients in a consecutive cohort of 4204 adult patients became VP shunt dependent after craniotomies for brain tumors (2% of patients). Of these, 28 patients (33%) had confirmed shunt-failures during the study period of 10 years, yielding cumulative shunt success rates at 1, 5, and 10 years of 77%, 71%, and 67%, respectively (Figure 2, Table 2). In the literature, reports of long-term shunt-failure rates in the adult population range from 11% at 1 year up to 34% at 10 years [24,29,45,36], with the majority of these consisting of various underlying etiologies including congenital diseases, normal pressure hydrocephalus, trauma, tumor, intracranial cysts etc. In a nationwide study on adult HC patients, Donoho et al.  found that 9% of patients required a shunt revision with a median time to shunt revision of only 41 days. However, their study included shunts due to all underlying conditions and 45% had shunt insertions due to obstructive HC. Furthermore, the authors did not state whether an obstructive HC was due to brain tumors and their shunt revision rates reflect the first 6 months only. In another study by Reddy et al.  on VP shunt complications in HC patients with intracranial tumors, 20% and 24% of patients experienced shunt failures requiring shunt revisions within 1 year and 5 years, respectively. Our lower shunt-failure rates might be explained by inclusion of brain tumor patients in adults only. However, the overall median time to shunt-failure was shorter in our study as compared to that of Donoho et al. , but this might be explained by tumor debris and higher protein content in the CSF of patients with brain tumors leading to shunt blockage compared to other non-oncological conditions.
Recently, Hosainey et al.  studied risk factors of early VP shunt failure after brain tumor surgery and found that patients with pre-existing, non-treated HC prior to craniotomy had a significantly shorter shunt-free period before definitive shunting compared to those without pre-craniotomy HC. Interestingly, in the current study, shunted patients with HC prior to craniotomy had significantly longer shunt survival (Figures 3 and 4). This indicates that in patients with distinct pathologies and profoundly deranged CSF dynamics in the early postoperative course after brain tumor surgery, early VP shunting may serve as ‘prophylaxis’ against further CSF disturbances in the future and hence give prolonged shunt longevity due to early ‘normalized’ hydrodynamics by shunting. The median shunt longevity was 457.5 days and 21.5 days for those with and without untreated HC pre-craniotomy, respectively (Figures 3 and 4). In the literature, median shunt survival times range from 19 days in the short-term up to 20.1 years in the long-term [35,29,8,24,12]. However, these studies were not limited to brain tumor patients and include a plethora of underlying conditions. Early changes to CSF dynamics as a result of overloading venous outflow and CSF pathway obstruction caused by disease burden has been described in the literature [40,44]. Further neuronal cell death may also ensue [10,11] if the disease process is left untreated. This requires early VP shunt insertion in order to normalize intracranial processes and avoid brain damage caused by a disrupted hydrocephalic state. Nonetheless, another plausible explanation may be that although some patients with pre- and post-craniotomy HC underwent shunting in the early postoperative course, they might have experienced spontaneous resolution of their hydrocephalic condition in the long-term period. Therefore, a ‘silent’ shunt obstruction may have ensued, making them effectively shunt-independent, wherefore a shunt obstruction would be unnoticed due to lack of signs and symptoms.
Neither patient age at time of shunt placement nor sex were associated with reduced shunt longevity (Table 3). Male gender has been associated with increased risk of shunt dysfunction [29,45], but it is not known if this was related to intracranial tumors. In a study by Reddy et al.  of VP shunt complications for hydrocephalus in patients with intracranial tumors, males had significantly lower 3- or 6-month survival rates compared to females (p < 0.001). This is in contrast to our findings. They also reported a 2% decrease in odds of shunt-failure with increasing age at time of shunt insertion . Comparatively, some studies have associated younger age with higher risk of shunt-failure [12,36], whereas others have not reached this conclusion regarding age in the short-term [1,13,17] nor in the long-term period after shunting .
Tumor location was not significantly associated with reduced shunt longevity despite dichotomizations into supratentorial/infratentorial and intra-axial/extra-axial tumor location (Table 3). Although somewhat surprising, this is in line with previous studies that did not find extra-axial/intra-axial tumor location to be significantly associated with early shunt-failure after craniotomy for brain tumor . In contrast, Khan et al.  studied factors affecting shunt survival in adults and found that extra-axial tumors were more common (13.2%) than intra-axial tumors (9.7%), but in line with our results, they reported that brain tumor location was not a significant risk factor of shunt failure.
With respect to tumor histology, several extra-axial tumors such as choroid plexus tumors, craniopharyngiomas  and schwannomas , and periventricular intra-axial tumors  have been reported to have increased risk of postoperative HC and shunt dependency. Additional stratified risk analysis into those with and without pre-craniotomy HC did not reveal intra-axial/extra-axial tumors as statistically significant risk factors for reduced long-term shunt longevity in our study. Nonetheless, similar reports are scarce in the literature making comparative analysis to our study difficult.
In our study, meningiomas had the highest incidence of shunt failure during follow-up (Table 1). Interestingly, these are extra-axial tumors and not usually located in the ventricles, but they might cause significant CSF dynamics changes after craniotomy if the tumor volume is large, particularly in the posterior fossa region, or if the area of resected/coagulated dura is large. Reddy et al.  reported that patients with benign tumors had higher risk of shunt revision, probably because of a shorter survival rate among patients with malignant brain tumors. In the above-mentioned study by Khan et al. , the effect of brain tumor histology did not reach statistical significance (p = 0.062). In the same vein, Rinaldo et al.  found no difference in the incidence of shunt revision surgery in high grade glioma patients as compared to NPH patients. We believe that the lower number of malignant brain tumor patients with reduced shunt longevity in our study might be due to the short overall survival of these patients, rendering shunt procedures futile when they present at advanced stages in the disease process. In addition to clinical diagnosis of shunt dysfunction, these patients may also suffer from ventriculomegaly as a consequence of radiation-induced brain atrophy, which is diagnosed radiologically. Lastly, patients with high grade gliomas invariably see clinical deterioration due to tumor progression and a shunt dysfunction in this context may be overlooked.
In our study, primary/secondary surgery for brain tumor was not significantly associated with increased risk of reduced long-term shunt longevity (Table 3). Secondary/repeat surgery has been reported as a risk factor for postoperative HC and subsequent VP shunt dependency in patients with pre-craniotomy hydrocephalus  and one would expect repeat surgical intervention for recurrent brain tumor to cause even more CSF disturbance and shunt-failures. However, only seven patients in our study cohort underwent repeat craniotomy for brain tumor, leaving a low statistical power and a high risk of a statistical error type II.
Only 2 patients (7.1%) had shunt infection during the follow-up (Table 1). Although post-shunting meningitis/infection was not significantly associated with reduced shunt longevity (Table 3), infection has been shown to be associated with higher risk of shunt failure in some studies . Our rates of infection lie in the upper range of published reports [1,26,29,24], which can be explained by our inclusion criteria of adult patients with brain tumors only. However, the number of patients with infection were too few for adequate statistical power, possibly giving rise to false negative results in our study. Most of the shunt revisions happened during the first year after shunt insertion (Table 2). Whereas some have reported shunt-failures in the first 6 months [13,26,29,35], others have reported within the first year .
Nine patients in the study cohort (32.1%) underwent multiple shunt revisions (≥ 2 revisions) (Table 1), in keeping with other reports . Korinek et al.  reported that previous shunt revision was an independent risk factor for infection leading to failure and Reddy et al.  reported single shunt revision procedures in 25 patients (13.4%) and multiple shunt revisions in 27 patients (14.4%) after initial shunt placement. Reddy et al.  also found that odds for multiple revisions among those with shunt system replacements were significantly higher (OR 24.39, p < 0.01) than those without any shunt replacement. They also showed that infection, shunt valve replacement and externalization were also significantly associated with multiple revisions. However, the significance was lost when the data was adjusted for the effects of other risk factors such as shunt system replacement and proximal shunt complication. Our study did not find that multiple revisions procedures (≥ 2 revision surgeries) were significantly associated with reduced shunt-longevity in the long-term (Table 3).