Childs Nerv Syst (2015) 31:1781–1789 DOI 10.1007/s00381-015-2781-8

SPECIAL ANNUAL ISSUE

Management of posterior fossa tumors and hydrocephalus in children: a review Chih-Ta Lin 1,2 & Jay K. Riva-Cambrin 1,3

Received: 26 May 2015 / Accepted: 2 June 2015 # Springer-Verlag Berlin Heidelberg 2015

Abstract Object Most pediatric patients that present with a posterior fossa tumor have concurrent hydrocephalus. There is significant debate over the best management strategy of hydrocephalus in this situation. The objectives of this paper were to review the pathophysiology model of posterior fossa tumor hydrocephalus, describe the individual risks factors of persistent hydrocephalus, and discuss the current management options. Specifically, the debate over preresection cerebrospinal fluid diversion is discussed. Results Only 10–40 % demonstrate persistent hydrocephalus after posterior fossa tumor resection. It appears that young age, moderate to severe hydrocephalus, transependymal edema, the presence of cerebral metastases, and tumor pathology (medulloblastoma and ependymoma) on presentation predict postresection or persistent hydrocephalus. The Canadian Preoperative Prediction Rule for Hydrocephalus (CPPRH), a validated prediction model, can be used to stratify patients at point of first contact into high and low risk for persistent hydrocephalus. Conclusions A protocol is proposed for managing hydrocephalus that utilizes the CPPRH. Low-risk patients can be monitored conservatively with or without an intraoperative

* Jay K. Riva-Cambrin [email protected] 1

Department of Neurosurgery, Clinical Neurosciences Center, University of Utah, Salt Lake City, UT, USA

2

Division of Neurosurgery, University of Vermont Medical Center, Burlington, VT, USA

3

Department of Neurosurgery, Primary Children’s Hospital, University of Utah, 100 N. Mario Capecchi Drive, Salt Lake City, UT 84113, USA

extraventricular drain, while high-risk patients require the use of an intraoperative extraventricular drain, higher postoperative hydrocephalus surveillance, and even consideration for a preoperative endoscopic third ventriculostomy. Keywords Posterior fossa tumor . Hydrocephalus . Cerebrospinal fluid diversion . Upward tentorial herniation . Intratumoral hemorrhage . Ventriculoperitoneal shunt . Endoscopic third ventriculostomy . Extraventricular drain . CPPRH . Pediatric . Oncology

Introduction Posterior fossa brain tumors are the most common solid tumors found in the pediatric population. Approximately 70–90 % of patients with posterior fossa tumors present with hydrocephalus [1–5]. Early studies have suggested that preresection treatment of hydrocephalus improves surgical resection of these tumors, postoperative mortality, and postoperative course [6]; however, numerous studies have also demonstrated that only 10–40 % of patient demonstrate persistent hydrocephalus after tumor resection [1, 2, 6–18]. This presents an interesting problem to pediatric neurosurgeons for which there remains no consensus: What is the best management strategy for hydrocephalus at the time of the initial presentation of the patient [19]? In this review, we will delineate the pathophysiologic models of hydrocephalus specific to pediatric patients with posterior fossa tumors; describe the individual risk factors for persistent hydrocephalus, including the validated predictive model developed to identify at-risk patients; and discuss the current management options, including the debate on preresection cerebrospinal fluid (CSF) diversion. Finally, a protocol of managing hydrocephalus that incorporates an

1782

established predictive model and findings from the contemporary literature will be proposed.

Pathophysiology Posterior fossa tumors often occur near the midline and exert mass effect on the CSF outflow tract. The CSF pathway initiates in the choroid plexus of the lateral ventricle and passes through the bilateral foramen of Monroe, the third ventricle, and the aqueduct of Sylvius before reaching the midline fourth ventricle in the posterior fossa where compression by tumor blocks the egress of CSF along its natural pathway. The result is obstructive hydrocephalus with a triventricular pattern. Alternatively and less commonly, dysfunction of CSF absorption has been proposed as the causative entity for hydrocephalus in children with posterior fossa tumors [2]. This dysfunction may result from disseminated tumor cells themselves, hemorrhage, or more likely chronic inflammation obstructing the CSF reabsorption into the venous system. In this case, a communicating pattern (dilated fourth ventricle and tumor cavity) of hydrocephalus is seen after tumor resection.

Clinical predictors of persistent hydrocephalus Over the years, various risk factors for the development of persistent hydrocephalus in children with posterior fossa tumors have been described, including young age, preoperative hydrocephalus, metastases, midline tumor location, tumor pathology, subtotal tumor resection, use of external ventricular drains (EVDs), and postresection complications. These will be discussed individually below. Young age Younger patients with posterior fossa tumors have been shown to be at higher risk for the development of persistent hydrocephalus. In a retrospective cohort of 42 children with medulloblastoma, Lee et al. found that younger patients were statistically more likely to undergo shunt placement for persistent hydrocephalus (mean age of 5.4 years in patients requiring shunt versus 10 years in those not requiring shunt) [14]. This finding was echoed in a retrospective study of 177 patients performed by Culley et al., who found that persistent hydrocephalus was statistically associated with younger age (4.6 vs. 6.9 years) [2]. They also found that 68 % of patients below the age of 3 years went on to develop persistent hydrocephalus whereas just 23 % of patients above the age of 3 years had persistent hydrocephalus. Two other retrospective studies described an age cutoff of 3 years: Bognar et al. [1] (n=180) reported shunting was necessary in 31 % of patients under

Childs Nerv Syst (2015) 31:1781–1789

3 years and 12 % of patients over the age of 3, while Kumar et al. [13] (n=175) reported that 37 % of children under the age of 3 years and 13 % of those older than 3 with posterior fossa tumors underwent shunt placement. A multicenter study with 454 children by Riva-Cambrin et al. [9] used a cutoff of 2 years of age and found an independent association of young age and persistent hydrocephalus after multivariate analyses. Papo et al. [17] described a similar propensity for younger children to have persistent hydrocephalus, but these authors used a much older age cutoff of 10 years. In their retrospective study of 62 patients, 40.7 % of patients under the age of 10 years required eventual shunt placement after tumor resection versus 10.7 % of patients over the age of 10 years. The authors of this study did not perform a statistical analysis of their results. Preoperative severity of hydrocephalus There are conflicting reports as to whether the radiographic severity of hydrocephalus on initial clinical presentation is associated with persistent or postresection hydrocephalus. Lee et al. [14] measured the degree of preoperative hydrocephalus using the frontal horn ratio (FHR) and ventricular body ratio (VBR). The FHR was defined as the greatest distance between the frontal horns divided by the difference between the greatest distance of the brain parenchyma and the greatest distance between the frontal horns. Similarly, the ventricular body ratio was defined as the greatest distance between the superior thalamus divided by the difference between the brain parenchyma and the greatest distance between the superior thalamus. The preoperative FHR and VBR were both statistically higher in patients requiring a shunt than in those who did not (0.73 vs. 0.48, respectively). Morelli et al. [10] reported a similar correlation using preoperative MRI and the Evans Index to measure preoperative ventricular size. The Evans Index is defined as the maximum width between the frontal horns divided by the maximal width of the inner table. Riva-Cambrin et al. [9] and Tamburrini et al. [18] also confirmed an association between qualitative moderate/severe preoperative ventriculomegaly and persistent hydrocephalus. Despite these findings, the association between preoperative hydrocephalus and persistent hydrocephalus after tumor resection has not been uniformly demonstrated in all studies [1, 2, 11, 15]. When Bognar et al. [1] analyzed their cohort of 180 children using a dichotomized variable of preoperative hydrocephalus or not, they did not find a statistical difference in shunt rates between these two groups. Both Dias et al. [15] and Culley et al. [2] used a qualitative description (none, mild, moderate, or marked hydrocephalus) for preoperative hydrocephalus and found no association with the need for postresection shunting. Similarly, no association was found between the preoperative quantitative ventricular index (VI) and persistent hydrocephalus by either Culley et al. [2] or

Childs Nerv Syst (2015) 31:1781–1789

Gnanalingham et al. [11]. The VI is the ratio between the maximal width between the frontal horns and the largest transverse diameter of the inner table of the skull. Presence of metastasis A correlation between the presence of metastasis on preoperative craniospinal imaging and persistent hydrocephalus has been proposed by Bhatia et al. [3]. This was a small study of 37 patients that underwent preresection endoscopic third ventriculostomy. The authors found a trend toward persistent hydrocephalus in patients presenting with metastasis on craniospinal imaging. Another large study demonstrated a significant independent association with cerebral metastases specifically [9]; however, metastases were not significantly associated with hydrocephalus in the medulloblastoma cohort in another study [14]. Regardless of the possible correlations, many of these studies were not adequately powered to determine an actual relationship between metastases and persistent hydrocephalus. Tumor location A midline location near the fourth ventricle has been correlated with persistent hydrocephalus after resective surgery in several studies [2, 11, 14, 17]. Papo et al. first suggested this correlation in their 1982 study of 62 children, where 29 % of patient presenting with midline tumors developed persistent hydrocephalus after tumor resection requiring shunt placement [17]. The shunt insertion rate for patients with hemispheric tumors was 14.3 %. The authors did not statistically analyze the findings in this study. In a larger retrospective study, Culley at al. [2] found that 40 % of the 103 patients with midline tumors ultimately required a shunt placement but none of the 13 patients with hemispheric tumors developed persistent hydrocephalus. Gnanalingham et al. [11] also demonstrated that 24 % of patients that presented with a midline tumor required shunt placement versus 8 % of patients that presented with a hemispheric tumor. Lee et al. [14] also showed that medulloblastoma patients with significant extension into the fourth ventricle defined by Chang stage T3 or T4 had a statistically higher rate of eventual shunt placement. This correlation between midline tumor location and persistent hydrocephalus has also been challenged by other studies [1, 15, 16]. Riva-Cambrin et al. [9] found that tumor location was not independently associated with postresection hydrocephalus when the multivariate model was adjusted for tumor pathology. Schmid et al. [16] found no significant difference in shunt placement rate of 23 children between midline or lateral tumors. The retrospective study of 58 children by Dias et al. [15] and the larger study of 180 children by Bognar et al. also did not find a correlation. In the Bognar

1783

et al. [1] study, 154 patients initially presented with a midline tumor, 24 (16 %) of which ultimately required a shunt for persistent hydrocephalus. In comparison, only four of the 26 (15 %) patients who presented initially with a hemispheric tumor required shunt placement. Tumor pathology Following an examination of tumor location, most studies naturally also examined tumor diagnosis. Medulloblastomas and ependymomas generally occur in the midline, while astrocytomas generally occur in the cerebellar hemispheres. It is therefore not surprising that most studies found a significant correlation between tumor diagnosis of ependymoma and medulloblastoma and persistent hydrocephalus following tumor resection [2, 10, 13]. Specifically, Culley et al. [2] reported rates of shunt insertion for ependymomas, medulloblastomas, and astrocytomas were 56, 41, and 33 %, respectively. Morelli et al. [10] found a significant risk with medulloblastomas only. Bognar et al. [1] found a difference in shunt insertion rates only between ependymomas (37 %) and astrocytomas (7 %). The study by Kumar et al. [13] demonstrated significantly higher shunt placement rates with ependymomas and medulloblastomas versus astrocytomas (33, 23, and 11 %, respectively). Riva-Cambrin et al. [9] found that medulloblastomas, ependymomas, and dorsally exophytic brainstem gliomas all independently predisposed patients to persistent hydrocephalus, even after adjusting for many other variables delineated above. Running counter to the substance of these findings, no significant association between tumor pathology and persistent hydrocephalus was found in a study by Dias et al., but this discrepancy may be due to the small size of the study (33 patients) [15]. Schneider et al. [7] performed a retrospective study on 130 patients with medulloblastoma that were categorized by their four molecular subtypes (WNT, SHH, group 3, and group 4). The study demonstrated that the shunt insertion rates were significantly different between the WNT subtype and the SHH, group 3, and group 4 subtypes (0 % vs. 29, 29, and 43 %, respectively). While this difference was attributed to the clinical differences between patient subtypes, the results also make it plausible that a molecular mechanism may be responsible for the development of persistent hydrocephalus as well. Extent of tumor resection Gross total resection of the tumor has been correlated with lower incidence of persistent hydrocephalus requiring shunt placement. Kumar et al. [13] found a significant difference in rates of shunt placement between patients that underwent gross total excision and those that had

1784

partial resection (13 vs. 32 %, respectively). This association was also confirmed by Culley et al. [2] and Gnanalingham et al. [11], although all studies utilized only the operative reports to determine extent of resection. Conversely, Dias et al. [15] did not find a relationship between extent of tumor resection and persistent hydrocephalus in their study, and while the larger Bognar et al. [1] study demonstrated a trend toward fewer shunt placements in complete resections compared with partial excisions, the trend was not statistically significant (14 % rate of shunt insertion vs. 18 %, respectively). To confuse the issue further, in their study of 150 children, Stein et al. [20] found a higher rate of persistent hydrocephalus in patients who underwent gross total resection compared with subtotal resection. EVD placement and duration EVD placement prior to or during tumor resection surgery and EVD duration after surgery have been suggested to correlate with developing persistent hydrocephalus [2, 11]. Gnanalingham et al. [11] reported that pre- or intraoperative CSF diversion was independently associated with persistent hydrocephalus on multivariate analysis of 89 patients (odds ratio 23.3 [5–103]); however, their results were confounded by the fact they included preoperative ventricular–peritoneal shunt, endoscopic third ventriculostomy (ETV), and EVD placement in their categorization of preoperative CSF diversion. This result has been countered by other studies. Neither the Kumar et al. study of 175 children nor the Lee et al. study of 42 medulloblastoma patients found any correlation of EVD placement and persistent hydrocephalus [13, 14]. Culley et al. [2] were the first to suggest that a longer EVD placement might be significantly correlated with persistent hydrocephalus. In their study, patients who underwent eventual shunt placement had EVD for a mean of 5.8 days whereas those that did not require shunt placement had the EVD for a mean of 4.3 days. Bognar et al. [1] also found a significant correlation between the need for shunt placement and both EVD placement and duration of EVD. The shunt insertion rate was 41 % for patients with EVD versus 11 % without. Furthermore, 62 % of patients that had an EVD for longer than 8 days underwent eventual shunt placement, but only 21.4 % of those that had an EVD less than 8 days required the shunt. It is important to note, however, that prolonged placement of an EVD might just be a sign of difficulty weaning instead of a physiological predictor of persistent hydrocephalus itself. Postresection surgical complications Culley et al. found a significant relationship between postoperative pseudomeningocele or meningitis and persistent hydrocephalus; however, Bognar et al., Kumar et al., and

Childs Nerv Syst (2015) 31:1781–1789

Gnanalingham et al. did not find this correlation in their respective studies [1, 2, 11, 13].

Predictive models of persistent hydrocephalus development Riva-Cambrin et al. [9] developed a grading model by which a patient’s risk for development of persistent hydrocephalus could be quantified preoperatively upon diagnosis of the posterior fossa tumor (Tables 1 and 2). Labeled the Canadian Preoperative Prediction Rule for Hydrocephalus (CPPRH), the multivariate model identified age less than 2 years, presence of papilledema, moderate/severe hydrocephalus on preoperative imaging, cerebral metastasis, and estimated tumor diagnosis (medulloblastoma, ependymoma, or dorsally exophytic brain stem glioma) as significant risk factors for persistent hydrocephalus. On the basis of these five clinical characteristics, a score of 0 to 10 could be assigned to each patient. Each score could then be correlated to a probability of hydrocephalus development, with a higher score forecasting a higher risk. The CPPRH model was externally validated on a cohort of 111 patients at a separate institution [9]. Foreman et al. [8] subsequently modified the model, replacing papilledema with radiographic evidence of transependymal flow. This modified model, or mCPPRH, was also externally validated on a cohort of 99 consecutive patients. The mCPPRH was used by Schneider et al. to study the incidence of hydrocephalus among patients with different molecular subtypes of medulloblastoma. The results demonstrated that patients with the WNT subtype did not develop persistent hydrocephalus and had lower mean mCPPRH scores in comparison with the SHH, group 3, and group 4 subtypes (1.92, 2.83, 3.38, and 3.36, respectively).

Table 1

The CPPRH in children with posterior fossa neoplasms

Predictor Age <2 years Presence of papilledema Moderate/severe hydrocephalus Cerebral metastases Preoperative estimated tumor diagnosis Medulloblastoma Ependymoma Dorsally exophytic brainstem glioma Total possible Reproduced with permission from [9]

Score 3 1 2 3 1 1 1 10

Childs Nerv Syst (2015) 31:1781–1789 Table 2 The CPPRH’s predicted probabilities of postresection hydrocephalus at 6 months

1785

CPPRH score

Probability of hydrocephalus at 6 months after resection

0 1

0.071 0.118

2

0.191

3 4

0.293 0.422

5 6

0.562 0.693

7

0.799

8

0.875

9

0.925

10

0.956

Reproduced with permission from [9]

Clinical management options The timing of hydrocephalus management in children presenting with both a posterior fossa tumor and preoperative hydrocephalus is the subject of debate. Proponents of preresection permanent CSF diversion (CSF shunts or ETV) argue that preresection shunting reduces the technical difficulty of resection surgery and improves postoperative course and overall mortality. Albright et al. [6] observed that 55 % of patients had a Btaut posterior fossa dura^ without treatment of the hydrocephalus, but this dropped to just 13 % with preresection treatment. Moreover, the study demonstrated an operative mortality of 4 % with permanent CSF diversion compared with 13 % without. Raimondi et al. [5] had similar results. It is important to note, however, that this philosophy will lead to overtreatment for hydrocephalus unless preresection CSF diversion is used judiciously (using the CPPRH). Currently, most surgeons resect the posterior fossa tumor with or without an EVD as they have anecdotally not found that preoperative permanent CSF diversion provides any advantage for the resection itself. The EVD is then weaned carefully over the next few days. This practice minimizes the number of patients treated with a permanent CSF diversion to those that fail to wean from the EVD. Preresection CSF shunting: advantages Multiple early studies support the placement of a permanent shunt system for management of hydrocephalus secondary to a posterior fossa tumor. Abraham et al. [21] observed immediate improvement in symptomatic hydrocephalus after CSF shunt placement, a Bdramatically smooth^ postoperative course, and a mortality of 8 % with preoperative shunt placement. Hekmatpanah et al. [22] also noted that shunt placement allowed for clinical improvement prior to definitive surgery as well as a technically easier resection surgery. Jane et al. [23]

performed postoperative ventriculography through a shunt and noted that preresection shunt placement increased the incidence of fourth ventricular filling and decreased cerebellar swelling. Hemmer et al. [24] also advocated for preresection shunt placement and offered suggestions for eventual shunt removal months after tumor resection. Raimondi et al. performed a large retrospective study of 123 patients with preresection shunt placement. Sixty-two of 78 patients (79 %) demonstrated resolution of papilledema after shunt placement. In their review of 90 operative reports, only four of 67 (6 %) patients that underwent preresection shunt demonstrated a Bbulging^ posterior fossa dura after shunt placement, compared with seven of 23 (30 %) patients that did not undergo preresection shunt [5]. Albright et al. [6] retrospectively compared 47 patients who did not undergo preresection ventriculoperitoneal shunt placement and 27 who did. The difference in the mortality rates prior to hospital discharge between the two groups was statistically significant (12.8 % without shunt placement and 3.7 % with shunt placement). Furthermore, the study observed a significantly greater incidence of clinical deterioration after tumor resection in the nonshunt group (32 % in the nonshunt group, 11 % in the shunt group). Preresection CSF shunting: disadvantages Support of preresection shunt placement has not been universally shared. Goel et al. [25] challenged the observation that the resection was technically easier in their report of 26 patients. The authors argued that shunting caused tumors to move in close proximity to the brainstem, thereby creating a more hazardous resection in 26 % of their patients. Moreover, they observed a shunt infection rate of 19 %, shunt obstruction in 8 %, and clinical deterioration in 31 % of patients. Arguments against preresection shunt placement gained momentum as more studies on the incidence of persistent hydrocephalus were published. The recently reported rates of persistent hydrocephalus after tumor resection range from 16 to 33 % [1, 7–12, 18]. Therefore, routine preresection placement of a permanent shunt system is argued to be unnecessary. This was illustrated by Raimondi et al. [5], who found that a significant number of these shunts were successfully removed over time. Every CSF shunt placed exposes patients to the risk of shunt malfunction, infection, and more surgery. Shunt infection rates in the pediatric population range from 2 to 11 % [5, 26–30] and have been independently correlated with mortality [26]. Intratumoral hemorrhage and upward tentorial herniation following preresection CSF shunt placement have also been described [31–35]. Santhanam et al. [32] reported the case of an 8-year-old girl who experienced intratumoral hemorrhage and neurological deterioration 2 h after a preresection CSF shunt insertion and required emergent tumor resection. The

1786

patient initially survived with severe disability, but ultimately died 18 months after surgery. El-Gaidi et al. [31] attempted to quantify the risk of these phenomena with a large cohort of 214 children treated with a preresection shunt. They found an incidence of 2.8 % for either intratumoral hemorrhage or upward tentorial herniation, and the condition of all of these patients deteriorated within 24 h after shunting. Of concern, five of these six patients (83 %) ultimately died. Finally, metastasis of tumor through the shunt tubing has been described in separate case reports [36–39]; however, tumor metastasis through shunts has not been quantified in large series and, as such, should be considered exceedingly rare [40].

Childs Nerv Syst (2015) 31:1781–1789

hydrocephalus following tumor resection alone. Studies following the Sainte-Rose study [12] have demonstrated persistent hydrocephalus rates of 10–29 % [1, 7–11, 18]. Given the low rates of persistent hydrocephalus, routine ETV may also unnecessarily expose patients to the surgical risks of ETV. A recent meta-analysis demonstrated general ETV complication rates around 9 % [43]. Specific to the preresection posterior fossa tumor population, Sainte-Rose et al. [12] noted a 1.5 % incidence of upward tentorial herniation after preresection ETV, while El-Gaidi et al. [30] reported a 1.1 % risk of intratumoral hemorrhage and death in their series of 87 patients. Others have also hypothesized that the prone patient position required for tumor resection predisposes preresection ETV to failure by a Bsnow globe^ effect [44].

ETV Preresection ETV gained support in the early 2000s. Several study groups argued that preresection ETV reduced the incidence of hydrocephalus after surgery, had a lower complication rate, prevented overdrainage of CSF, and was a faster operation than ventriculoperitoneal shunt placement [3, 4, 12, 41]. Sainte-Rose et al. [12] have championed preresection ETV as it reduced the incidence of persistent hydrocephalus and shunt placement postoperatively. In their retrospective study of 67 patients with hydrocephalus who underwent preresection ETV and 82 patients who did not, the incidences of persistent hydrocephalus were 6.3 and 26.8 %, respectively. On the basis of this data, the group argued that all patients should undergo preresection ETV. However, with the development of the CPPRH predictive model, the group has recently softened their view and suggested preresection ETV may be more appropriate in patients at higher risk for developing persistent hydrocephalus [42]. Bhatia et al. [3] examined a cohort of 37 patients who underwent ETV prior to tumor resection over a follow-up period of 7.5 years. They noted a failure rate of 14 %. Furthermore, they noted an 8 % complication rate (bleeding and infection) but no mortalities. The group argued that the longterm success rate of ETV as well as the low complication rate supported placement of ETV prior to tumor resection. El-Ghandour et al. [40] compared 32 patients that underwent preresection ETV with 21 patients that underwent preresection CSF shunting. They found significantly shorter surgery duration in the ETV group (15 vs. 35 min), lower complication rate (9 vs. 28 %), and a lower rate of procedure failure (6 vs. 38 %). The authors described the ETV procedure as a Bmore physiologic^ method of managing hydrocephalus resulting in the lower complication rate. Given these conclusions, they argued that ETV is a superior procedure in managing hydrocephalus prior to the resection of a posterior fossa tumor. Other groups have challenged the practice of routine preresection ETV, emphasizing the low rate of persistent

Postresection ETV Although preresection ETV is considered controversial, ETV for postresection or persistent hydrocephalus is very effective and, in many centers, is the treatment of choice when EVDs cannot be weaned. Morelli et al. [10] reported an overall ETV success rate of 81 % in the pediatric posterior fossa tumor cohort. More specifically, ETV was used in the postresection setting in four cases, all successful without complications. Tamburrini et al. [18] demonstrated a higher postresection ETV success rate of 90 % in a larger cohort of 30 patients. Kulkarni et al. [45] reported an overall ETV success rate of 67 %, and for those with an etiology of Bother brain tumor (non-tectal),^ in which posterior fossa tumors would be categorized, they also found a success rate of 67 %. Lastly, Drake et al. [46] found an overall ETV success rate for the Canadian hydrocephalus population of 65 % at 1 year and as high as 70– 75 % for those with a tumor etiology. Temporary EVD placement Some authors have advocated pre- or intraoperative management of posterior fossa tumor hydrocephalus with an EVD. These authors argue that an EVD is an effective measure of CSF diversion, allows for continuous pressure monitoring, aids in removing surgical debris and blood products, and has a reduced incidence of upward tentorial herniation or intratumoral hemorrhage [15, 18, 47]. EVDs have the added advantage of being easy to remove after surgery as the majority of patients do not develop persistent hydrocephalus after tumor resection [16, 17]. Richard et al. used EVD placement in the lateral ventricles of children to monitor ventricular fluid pressure before, during, and after posterior fossa tumor surgery [48]. Although the authors did not correlate patient survival with EVD placement, they argued that pressure monitoring was nonetheless helpful in clinical management. In fact, Tamburrini et al. [18]

Childs Nerv Syst (2015) 31:1781–1789

utilized elevated postoperative ICP values from EVDs as criteria for postresection ETV surgery. Rappaport et al. [47] published their experience using EVD in 59 patients. They favored EVD because it allowed for control of intracranial pressure and they believed that it correlated with a lower incidence of persistent hydrocephalus after tumor resection. Dias et al. [15] echoed similar conclusions in their study of 17 patients. Nevertheless, EVDs can still have complications even with careful usage, especially when they are placed hours or days before the posterior fossa tumor resection surgery. Vaquero et al. [49] reported two cases of intratumoral hemorrhage after EVD placement. The first patient was an 8-year-old boy whose condition deteriorated 8 h after an EVD was placed 15 cm above the ear. The second case was an 11-year-old girl who developed a herniation syndrome 2 h after an EVD was placed 20 cm above the ear. Both autopsies demonstrated hemorrhage within the tumor. This report illustrates that, although it is rare, intratumoral hemorrhage and upward herniation are still risks of EVD placement. The most significant arguments against EVD placement are based on the risk of infection. Schmid et al. [16] demonstrated a 4.9 % risk of infection in their retrospective study of 61 patients with posterior fossa tumors (38 adults, 23 children), and Dias et al. [15] demonstrated an alarming 17 % rate of bacterial ventriculitis in a very small series of 17 children. Most pediatric neurosurgeons will try to wean patients from the EVDs as quickly as possible to avoid infection; however, in a series of 58 patients that underwent preresection EVD, Rappaport et al. [47] found a 10 % incidence of infection with a mean duration of catheter placement of only 2.3 (±1.6) days. However, the historical rates of EVD infection may not reflect current risks with the increased diligence of EVD care and evidence suggesting lower infection rates with antibiotic impregnated catheters. In a study of 91 posterior fossa tumor patients, Tamburrini et al. [50] demonstrated infection rates of 2.1 versus 31.8 % between antibiotic impregnated and non-impregnated catheters, respectively.

Recommendations At our institution, we calculate a CPPRH score at the time of first diagnosis of the posterior fossa tumor, firstly, to facilitate a discussion with patients and their families on the risk of hydrocephalus after tumor resection [9]. Our practice is initial management with steroids followed by surgery within 24–48 h of presentation. In terms of hydrocephalus management, the validated CPPRH is then used to stratify each patient as either high or low risk for postresection hydrocephalus. In general, a lowrisk patient would have an intraoperative EVD placed if there was significant preresection hydrocephalus (moderate/severe) as the CPPRH predicts a 24.5 % risk of postresection hydrocephalus. The EVD is weaned postoperatively and removed if

1787

tolerated. If persistent hydrocephalus develops (inability to wean EVD or signs and symptoms of hydrocephalus with recurrent ventriculomegaly), a CSF shunt placement or ETV is then performed. Patients scored as high risk according to the CPPRH have a 75 % risk of postresection hydrocephalus; in these patients, placement of an EVD is mandated intraoperatively before the tumor resection. The high-risk children also need a higher level of postresection surveillance for hydrocephalus, in terms of the pace of EVD weaning and follow-up after discharge. In some institutions, the high-risk children would be considered for a preresection ETV; this is rare in our practice. In the future, we have proposed a randomized trial for this highrisk cohort to compare the use of a preresection ETV versus the conservative approach with EVDs.

Conclusion While 70–90 % of patients with posterior fossa tumors present with hydrocephalus, only 10–40 % demonstrate persistent hydrocephalus after tumor resection. Many risk factors have been examined by a variety of authors with mixed results, but it appears that a young age, moderate to severe hydrocephalus, transependymal edema, the presence of cerebral metastases, and tumor pathology (medulloblastoma and ependymoma) on presentation predict postresection or persistent hydrocephalus. In terms of treatment, most pediatric neurosurgeons currently use EVDs followed by tumor resection with an attempt to wean the EVDs postoperatively; however, the CPPRH, a validated prediction model, can be used to stratify patients at point of first contact into high and low risk for persistent hydrocephalus. Low-risk patients can be monitored conservatively with or without an intraoperative EVD, thus avoiding unnecessary risks of preresection ETVor CSF shunting. Highrisk patients require the use of an intraoperative EVD, higher postoperative hydrocephalus surveillance, and even consideration for a preresection ETV. A preresection ETV has the advantages of avoiding shunting and allowing the tumor resection to proceed under low intracranial pressure but adds the risk of ETV complications and the possibility of unnecessary treatment in 25 % of even the high-risk children.

Conflict of interest None of the authors have any conflicts of interest associated with this publication.

References 1.

Bognár L, Borgulya G, Benke P, Madarassy G (2003) Analysis of CSF shunting procedure requirement in children with posterior fossa tumors. Childs Nerv Syst 19:332–336. doi:10.1007/s00381-003-0745-x

1788 2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

Childs Nerv Syst (2015) 31:1781–1789 Culley DJ, Berger MS, Shaw D, Geyer R (1994) An analysis of factors determining the need for ventriculoperitoneal shunts after posterior fossa tumor surgery in children. Neurosurgery 34:402–407, discussion 407–8 Bhatia R, Tahir M, Chandler CL (2009) The management of hydrocephalus in children with posterior fossa tumours: the role of pre-resectional endoscopic third ventriculostomy. Pediatr Neurosurg 45:186–191. doi:10.1159/000222668 Ruggiero C, Cinalli G, Spennato P et al (2004) Endoscopic third ventriculostomy in the treatment of hydrocephalus in posterior fossa tumors in children. Childs Nerv Syst 20:828–833. doi:10. 1007/s00381-004-0938-y Raimondi AJ, Tomita T (1981) Hydrocephalus and infratentorial tumors. Incidence, clinical picture, and treatment. J Neurosurg 55: 174–182. doi:10.3171/jns.1981.55.2.0174 Albright L, Reigel DH (1977) Management of hydrocephalus secondary to posterior fossa tumors. J Neurosurg 46:52–55. doi:10. 3171/jns.1977.46.1.0052 Schneider C, Ramaswamy V, Kulkarni AV et al. (2014) Clinical implications of medulloblastoma subgroups: incidence of CSF diversion surgery. J Neurosurg Pediatr 1–7. doi: 10.3171/2014.9. PEDS14280 Foreman P, McClugage S, Naftel R et al (2013) Validation and modification of a predictive model of postresection hydrocephalus in pediatric patients with posterior fossa tumors. J Neurosurg Pediatr 12:220–226. doi:10.3171/2013.5.PEDS1371 Riva-Cambrin J, Detsky AS, Lamberti-Pasculli M et al (2009) Predicting postresection hydrocephalus in pediatric patients with posterior fossa tumors. J Neurosurg Pediatr 3:378–385. doi:10. 3171/2009.1.PEDS08298 Morelli D, Pirotte B, Lubansu A et al (2005) Persistent hydrocephalus after early surgical management of posterior fossa tumors in children: is routine preoperative endoscopic third ventriculostomy justified? J Neurosurg 103:247–252. doi:10.3171/ ped.2005.103.3.0247 Gnanalingham KK, Lafuente J, Thompson D et al (2003) The natural history of ventriculomegaly and tonsillar herniation in children with posterior fossa tumours—an MRI study. Pediatr Neurosurg 39:246–253 Sainte-Rose C, Cinalli G, Roux FE et al (2001) Management of hydrocephalus in pediatric patients with posterior fossa tumors: the role of endoscopic third ventriculostomy. J Neurosurg 95:791– 797. doi:10.3171/jns.2001.95.5.0791 Kumar V, Phipps K, Harkness W, Hayward RD (1996) Ventriculoperitoneal shunt requirement in children with posterior fossa tumours: an 11-year audit. Br J Neurosurg 10:467–470 Lee M, Wisoff JH, Abbott R et al (1994) Management of hydrocephalus in children with medulloblastoma: prognostic factors for shunting. Pediatr Neurosurg 20:240–247 Dias MS, Albright AL (1989) Management of hydrocephalus complicating childhood posterior fossa tumors. Pediatr Neurosci 15: 283–289, discussion 290 Schmid UD, Seiler RW (1986) Management of obstructive hydrocephalus secondary to posterior fossa tumors by steroids and subcutaneous ventricular catheter reservoir. J Neurosurg 65:649–653. doi:10.3171/jns.1986.65.5.0649 Papo I, Caruselli G, Luongo A (1982) External ventricular drainage in the management of posterior fossa tumors in children and adolescents. Neurosurgery 10:13–15 Tamburrini G, Pettorini BL, Massimi L et al (2008) Endoscopic third ventriculostomy: the best option in the treatment of persistent hydrocephalus after posterior cranial fossa tumour removal? Childs Nerv Syst 24:1405–1412. doi:10.1007/s00381-008-0699-0 Schijman E, Peter JC, Rekate HL et al (2004) Management of hydrocephalus in posterior fossa tumors: how, what, when? Childs Nerv Syst 20:192–194. doi:10.1007/s00381-003-0900-4

20.

Stein BM, Tenner MS, Fraser RA (1972) Hydrocephalus following removal of cerebellar astrocytomas in children. J Neurosurg 36: 763–768. doi:10.3171/jns.1972.36.6.0763 21. Abraham J, Chandy J (1963) Ventriculo-atrial shunt in the management of posterior-fossa tumours: preliminary report. J Neurosurg 20:252–253. doi:10.3171/jns.1963.20.3.0252 22. Hekmatpanah J, Mullan S (1967) Ventriculo-caval shunt in the management of posterior fossa tumors. J Neurosurg 26:609–613. doi:10.3171/jns.1967.26.6.0609 23. Jane JA, Kaufman B, Nulsen F et al (1973) The role of angiography and ventriculovenous shunting in the treatment of posterior fossa tumors. Acta Neurochir (Wien) 28:13–27 24. Hemmer R, Haensel-Friedrich G, Friedrich H (1976) The value of preoperative shunts in posterior fossa tumors. Mod Probl Paediatr 18:48–50 25. Goel A (1993) Whither preoperative shunts for posterior fossa tumours? Br J Neurosurg 7:395–399 26. Tuli S, Tuli J, Drake J, Spears J (2004) Predictors of death in pediatric patients requiring cerebrospinal fluid shunts. J Neurosurg 100: 442–446. doi:10.3171/ped.2004.100.5.0442 27. Drake JM, Kestle JRW, Milner R et al (1998) Randomized trial of cerebrospinal fluid shunt valve design in pediatric hydrocephalus. Neurosurgery 43:294 28. Cochrane DD, Kestle JRW (2003) The influence of surgical operative experience on the duration of first ventriculoperitoneal shunt function and infection. Pediatr Neurosurg 38:295–301. doi:10. 1159/000070413 29. McLaurin RL (1983) Disadvantages of the preoperative shunt in posterior fossa tumors. Clin Neurosurg 30:286–292 30. Taylor WA, Todd NV, Leighton SE (1992) CSF drainage in patients with posterior fossa tumours. Acta Neurochir (Wien) 117:1–6 31. El-Gaidi MA, El-Nasr AHA, Eissa EM (2015) Infratentorial complications following preresection CSF diversion in children with posterior fossa tumors. J Neurosurg Pediatr 15:4–11. doi:10.3171/ 2014.8.PEDS14146 32. Santhanam R, Balasubramaniam A, Chandramouli BA (2009) Fatal intratumoral hemorrhage in posterior fossa tumors following ventriculoperitoneal shunt. J Clin Neurosci 16:135–137. doi:10. 1016/j.jocn.2008.02.016 33. Waga S, Shimizu T, Shimosaka S, Tochio H (1981) Intratumoral hemorrhage after a ventriculoperitoneal shunting procedure. Neurosurgery 9:249–252 34. Epstein F, Murali R (1978) Pediatric posterior fossa tumors: hazards of the Bpreoperative^ shunt. Neurosurgery 3:348–350 35. Zuccarello M, Dollo C, Carollo C (1985) Spontaneous intratumoral hemorrhage after ventriculoperitoneal shunting. Neurosurgery 16: 245–246 36. Kleinman GM, Hochberg FH, Richardson EP (1981) Systemic metastases from medulloblastoma: report of two cases and review of the literature. Cancer 48:2296–2309 37. Hoffman HJ, Hendrick EB, Humphreys RP (1976) Metastasis via ventriculoperitoneal shunt in patients with medulloblastoma. J Neurosurg 44:562–566. doi:10.3171/jns.1976.44.5.0562 38. Oberbauer RW, Tritthart H, Ascher PW et al (1979) Shunt metastases in posterior fossa tumors. Neuropadiatrie 10:296–300. doi:10. 1055/s-0028-1085332 39. Jamjoom ZA, Jamjoom AB, Sulaiman AH et al (1993) Systemic metastasis of medulloblastoma through ventriculoperitoneal shunt: report of a case and critical analysis of the literature. Surg Neurol 40:403–410 40. Berger MS, Baumeister B, Geyer JR et al (1991) The risks of metastases from shunting in children with primary central nervous system tumors. J Neurosurg 74:872–877. doi:10.3171/jns.1991. 74.6.0872 41. El-Ghandour NMF (2010) Endoscopic third ventriculostomy versus ventriculoperitoneal shunt in the treatment of obstructive

Childs Nerv Syst (2015) 31:1781–1789 hydrocephalus due to posterior fossa tumors in children. Childs Nerv Syst 27:117–126. doi:10.1007/s00381-010-1263-2 42. Di Rocco F, Jucá CE, Zerah M, Sainte-Rose C (2013) Endoscopic third ventriculostomy and posterior fossa tumors. WNEU 79: S18.e15–S18.e19. doi: 10.1016/j.wneu.2012.02.018 43. Bouras T, Sgouros S (2011) Complications of endoscopic third ventriculostomy. J Neurosurg Pediatr 7:643–649. doi:10.3171/ 2011.4.PEDS10503 44. Tamburrini G, Frassanito P, Bianchi F et al. (2014) Closure of endoscopic third ventriculostomy after surgery for posterior cranial fossa tumor: the Bsnow globe effect.^ Br J Neurosurg 1–4. doi: 10. 3109/02688697.2014.987214 45. Kulkarni AV, Drake JM, Mallucci CL et al (2009) Endoscopic third ventriculostomy in the treatment of childhood hydrocephalus. J Pediatr 155:254–259.e1. doi:10.1016/j.jpeds.2009.02.048 46. Drake JM, Canadian Pediatric Neurosurgery Study Group (2007) Endoscopic third ventriculostomy in pediatric patients: the

1789 Canadian experience. Neurosurgery 60:881–886. doi:10.1227/01. NEU.0000255420.78431.E7, discussion 881–6 47. Rappaport ZH, Shalit MN (1989) Perioperative external ventricular drainage in obstructive hydrocephalus secondary to infratentorial brain tumours. Acta Neurochir (Wien) 96:118–121 48. Richard KE, Frowein RA, Vanner GK (1976) Ventricular fluid pressure in posterior fossa tumors of childhood. Mod Probl Paediatr 18:35–39 49. Vaquero J, Cabezudo JM, de Sola RG, Nombela L (1981) Intratumoral hemorrhage in posterior fossa tumors after ventricular drainage. Report of two cases. J Neurosurg 54:406–408. doi:10. 3171/jns.1981.54.3.0406 50. Tamburrini G, Massimi L, Caldarelli M, Di Rocco C (2008) Antibiotic impregnated external ventricular drainage and third ventriculostomy in the management of hydrocephalus associated with posterior cranial fossa tumours. Acta Neurochir (Wien) 150: 1049–1056. doi:10.1007/s00701-008-0022-6