Introduction

Neuroendoscopy has enjoyed a renewed interest in the last decade. While it is not new as a technique—it has been used as early as the beginning of the 20th century by Walter Dandy and other pioneers—technological advancements have made it safer and easier, making endoscopic treatment of hydrocephalus feasible even in neonates and newborns. Even after the initial enthusiasm and renewed interest has settled and consolidated, endoscopic third ventriculostomy (ETV) for treatment of obstructive hydrocephalus is considered now as first line of management for many patients. For many neurosurgeons who are strong advocates of endoscopy, it is considered unethical not to offer the option of ETV as the treatment of choice in children with obstructive hydrocephalus due to aqueduct stenosis, over the age of 1–2 years (Kulkarni 2005 unpublished data) [5, 9, 11, 16, 23, 33, 56]. Its effectiveness has been demonstrated also in hydrocephalus, resulting from many other causes such as Chiari or Dandy-Walker malformations, myelomeningocele, brain tumors, and NPH, even in situations complicated by a history of hemorrhage, infection, or previous ETV attempts, as well as in cases of shunt malfunction [3, 5, 8–11, 17–19, 28, 29, 34–39, 50–52, 55, 59]. While in older children, the indications for endoscopic treatment have been well defined, much debate has centered in recent years on the value of this treatment in the first few months of life. While different success results have emerged from clinical studies, inevitably, all these studies outline the lack of consensus on the value of third ventriculostomy in infants and neonates. This has led effectively to the organization of the International Infant Hydrocephalus Study (IIHS).

Shunt vs ETV in infants with aqueduct stenosis: debate on the merits of the two treatment types

While there have been more than 30 recent studies that address the question of choice of treatment for hydrocephalus in infants, no consensus opinion has emerged. Standards of inclusion criteria, treatment choices, and outcome measures are inconsistent among studies. Multiple single institution studies with small numbers of patients have produced data of debatable statistical validity. Nonstandardized inclusion criteria have resulted in studies with heterogeneous samples, which may include full-term or premature infants, with simple or complex noncommunicating hydrocephalus, with or without a previous history of shunting, infection, or hemorrhage, and/or patients with myelomeningocele, and/or various forms of communicating hydrocephalus. These heterogeneous samples make valid comparisons virtually impossible, since many current studies suggest that the single most significant factor in the success of an ETV is the etiology of the hydrocephalus [1, 4, 9, 10, 25, 41].

In addition to the problems of inconsistent inclusion standards and small numbers, there is inconsistency in the definition of success and failure among studies making meaningful comparisons impossible. The only criterion for success used in most studies has been some variation on freedom from shunt dependency. Even that single criterion is not applied consistently, with some studies insisting on total shunt freedom after a single ETV attempt, others including shunt freedom after a repeat ETV as well, and in most cases, there are short-term follow-ups that do not define or address the risks of delayed failure. Relying on the single criterion of shunt freedom completely ignores equally important outcomes such as neurodevelopmental and quality of life, hospitalization time, number of surgeries, and economic issues as well. Neurodevelopmental impact is an especially significant factor for infants diagnosed with hydrocephalus, given the known pattern of cognitive and neuropsychological issues associated with shunted hydrocephalic children. This typically includes a pattern of lower IQ scores, hyperverbosity (cocktail party syndrome), deficits in working memory, and problems with attention, planning, organization, and self-monitoring (executive functioning) [13, 22, 23, 44, 53, 60].

Table 1 shows a consolidated comparison of results from various important studies, focusing on success rates of ETV in the first 2 years of life. A wide variation of success has been reported. Critical review of these studies highlights some of their limitations.

Table 1 Synopsis of reported success rate of ETV in published series of the last 15 years
Full size table

Beems and Grotenhuis [4] concluded that etiology rather than age is the significant factor in the success or failure of ETV. However, there is no detailed analysis by patient age, partly due to the fact that the numbers were too small for statistical significance, and they make no attempt to assess neurodevelopment or functional quality of life. Burn et al. [6] broke down their data by age, etiology, and complications. However, their numbers were too small to be statistically meaningful. Nevertheless, they concluded that ETV is not worth attempting in patients with intraventricular hemorrhage (IVH) who are under 1 year old but is still worth considering in patients with other diagnoses. They commented also on the relatively high incidence of subsequent shunt infection after failed ETV (4 of the 13 patients). Buxton et al., in two similar publications, analyzed a general group of infants under 1 year of age [7] and also focused on a group of premature infants [8]. The numbers in each group are small, and they cannot be combined due to the greater rate of complications and problems in the premature group. They concluded that in anatomically suitable patients, ETV is worth trying, as it does not significantly add to the morbidity, it involves little or no mortality, and it can be unpredictably successful in a patient group with a high proportion of shunt complications. They argued in favor of the presumed reduction in morbidity, mortality, and economic burden compared with shunting. However, they acknowledged that the true long-term benefit of such procedures cannot be shown until a randomized, controlled trial comparing shunt against ETV is completed. Cinalli et al. [10] emphasized the need for a definitive study before any conclusions can be reached. They stated that before ETV can be widely accepted as the treatment of choice for these patients, a detailed long-term follow-up study is required. Genitori et al. [15] attempted a larger-scale study, which, unfortunately, did not include an analysis of outcome by age. Hopf et al. [20], in another large study, included only four infants under 1 year old, with multiple assorted complications, providing no statistically significant results for this age group.

The varying standards for success and failure can be highlighted by comparing the following studies:

  1. 1.

    Mori et al. [40] obtained a 25% success rate for 12 children under 1 year of age and concluded that patients under 1 year of age are not candidates for ETV and V-P shunt is the treatment of choice. In those at the age of 1 year or older, ETV was recommended.

  2. 2.

    Buxton et al. [7] documented a success rate of 27% for 27 children under 1 year of age and concluded that ETV is the most appropriate initial treatment choice for infants, stating that although the majority fail, approximately one-third are spared the added morbidity and mortality of having a shunt. With such a low morbidity and zero mortality, the procedure has many benefits over shunting. They suggested that neuroendoscopic third ventriculostomy should be used where possible, rather than a shunt.

  3. 3.

    Kamikawa et al. [30] analyzed a group of 11 infants “mostly” under 1 year old. Their entire group eventually require shunting, yet they found statistically significant advantages to initial treatment with an ETV attempt, despite the nonexistent “success rate,” concluding that even in failed patients under 1 year of age, ETV is safe and clinically effective as the first-line treatment of hydrocephalus in children.

Cinalli [9], in a comprehensive analysis of the options available for treatment of infants with obstructive hydrocephalus, summarized the position by stating that all the available data in the literature seem to show that third ventriculostomy offers at least as good an intellectual outcome as extracranial CSF shunts but also confirm the need for a controlled study.

Historical background leading to the organization of the study

In the last 5 years, as advocates of neuroendoscopy gathered to form the International Study Group on Neuroendoscopy (ISGNE), it became evident that beyond the enthusiasm to employ endoscopy as widely as possible, the lack of robust evidence on its efficacy was obvious. As a result, collaborative studies were undertaken, to try to move beyond the limited single institution studies, on these rare diseases. Gradually, a network of like-minded collaborating clinicians was established and managed to complete three retrospective studies:

  • The first study (20 patients/4 centers) on the merits of repeat ETV was published in 2001 [51].

  • The second study (101 patients/7 centers) on the merits of ETV following infection and/or hemorrhage was published in 2002 [50].

  • Data collection for a third retrospective study on the safety and efficacy of endoscopic biopsies (International Neuroendoscopic Biopsy Study, INEBS) has been completed and is due to be published. Data on 280 patients have been collected from 13 centers worldwide during 2004.

A fourth international study on endoscopic treatment of pineal tumors has been designed already and will commence soon (Principal Investigator Prof. S. Oi).

The IIHS is a significantly more ambitious undertaking than any previous study, moving beyond retrospective data collection to a multicenter prospective randomized, controlled study. Its impact is anticipated to be far reaching:

  • We hope to achieve statistically and clinically meaningful results to help resolve an issue for which the literature has thus far not been able to provide an objective answer. This is the only study to date that focuses on long-term developmental outcomes for hydrocephalic infants. As such, it will provide the means for better prenatal counseling when obstructive hydrocephalus is diagnosed in utero.

  • This study will strengthen international collaboration in this field. The network of participating international investigators will provide the foundation and infrastructure for future international prospective studies.

Concept and rational of the study

The proposed study represents a major departure from most published work on the value of neuroendoscopy in comparison to shunting in the treatment of hydrocephalus. Whereas most studies focus on the “survival” of the created stoma or implanted shunt and surgical complications of each procedure, this study focuses predominantly on the effect of both treatment types on neurodevelopmental outcome of these very young patients. Some neurosurgeons, skeptical on the value of ETV, believe that it may lead to an arrested form of hydrocephalus that compromises the developing brain. This belief is based on the observation that almost always, children who had ETV have ventricles considerably bigger in comparison to children who had shunting [54]. Still, no study has attempted to correlate the size of the ventricles to any developmental measurements, so this belief has not been proven or refuted scientifically. Conversely, other neurosurgeons advocates of ETV are concerned with the long-term complications of shunting on the developing brain, especially the effects of shunt infections. This is based on the observed link between shunt infection and reduction of IQ and the measured memory deficits among shunted children [2, 45–47, 58] Thus, they claim that ETV is a more “physiological” solution and therefore better for the infant’s brain. There has never been a direct controlled comparison of the two treatment types on their impact on intellectual development. For this reason, the primary outcome of the trial has been defined as the neurodevelopmental score at the age of 5 years. In that respect, it may well prove that one treatment type has considerably better neurodevelopmental outcome, but possibly higher complication rate. It will be up to the parents and the treating neurosurgeon in the future to choose one or the other treatment, knowing the full facts in detail.

In the same line of thought, it has been decided that other factors, such as complication rates, hospitalization time, need for repeat surgeries and imaging would be part of secondary measures. It will give us the opportunity to compare directly in controlled circumstances the complication rates between the two types of treatment, whereas, currently, no such information exists. Currently, when we are counseling our patients, we are quoting complication rates from different studies, always choosing subconsciously the most favorable ones, which unfortunately are not directly comparable, because at the very least they are not based on comparable patient populations.

It is clear that the study is not preoccupied with the success rate of ETV or the survival rate of the shunt in the traditional sense, but with the effect of either operative treatment on the developing brain, and as such, it is designed to provide a comprehensive assessment of the relevant relative benefits and risks. Essentially, it is the first randomized, controlled study on the long-term outcome of patients with infantile hydrocephalus due to aqueduct stenosis.

Due to the lack of clear superiority for either surgical technique, it became obvious that randomization in shunt or ETV group, in the clearly defined population of infants under 2 years of age with obstructive hydrocephalus due to pure aqueduct stenosis, is not only ethical but also a duty for all involved in the management of such patients. Nevertheless, families who are presented with both options in a nonbiased way and elect to choose one, possibly on the basis on information that they may have already gathered on their own, may also be included in the study. This does not violate the statistical validity of the study and is designed in the model of a comprehensive cohort design [42].

Neurodevelopmental aspects

Neurodevelopmental assessment of the recruited patients is a key component of the study and, in view of the very young age of the patients, poses certain challenges that needed to be overcome. Hydrocephalus is known to be associated with memory and behavior deficits, and shunt complications are known to influence these functions [13, 22, 23, 44–46, 58, 60]. Nevertheless, all studies that have explored such issues have limited themselves to testing of patients at older ages, well after treatment has been completed, and effectively, they represent incidence studies for that particular problem. This study aims to provide prospective data on preoperative and postoperative testing at fixed ages. By obtaining tests and subsequent scores at predetermined ages, rather than at fixed intervals after surgery, we have streamlined the data that will result and make them comparable, from patient to patient and between the two treatment groups, as well as to normal age matched scores.

For preoperative testing, we have adapted the Denver Developmental Screening Test questionnaire for each month of life, as in these first 24 months of life rapid developmental changes which are normally expected and observed. Postoperative testing for ages of 12, 24, and 36 months is an adaptation of the Denver Developmental Screening Test. For age of 5 and (optional) 7 years, we have employed a combination of the Health Utilities Index (HUI) [21], the Hydrocephalus Outcome Questionnaire (HOQ) [33], and, as an option, the standard Wechsler test. The objective is to have a neurodevelopmental score at each stage, which will be used for statistical comparisons.

Surgical aspects

The study assumes that the participating surgeons fully endorse the rational and the concept of randomization. Hence, the surgeon is expected to explain to the parents the pros and cons of the two surgical methods without compassion and demonstrate the reason behind randomization. While the study does not dictate the use of any particular type of shunt in the shunted group or any particular endoscope or technique in the ETV group, we have defined a minimum of experience from the surgeon of at least ten neuroendoscopic procedures per year and a minimum of five ETV procedures in infants, as a criterion to be accepted to join the study group. We have not set a minimum level of shunt operations, as they are far more common. Hence, all neurosurgeons who have declared an interest to join the study function in environments of established pediatric neurosurgery departments, attracting high enough numbers of patients, sufficient to recruit the rare subgroup of infants with pure aqueduct stenosis. The type of shunt is not considered likely to influence the neurodevelopmental outcome significantly at 5 years of age, and for this reason, we have not controlled it. Moreover, demanding that only one type of shunt is used in the study patients can be viewed as a commercial endorsement of one particular product, which can only lead to credibility and compliance problems, as most neurosurgeons have very fixed views on shunts. The type of endoscope and the technique employed to perforate the floor of the third ventricle are equally not considered likely to influence the neurodevelopmental outcome at 5 years of age. Most endoscopes are of similar caliber, and all instruments used to perforate the floor of the third ventricle cause the same degree of “damage”. Nevertheless, information on such technical issues will be collected and analyzed as part of the secondary analyses.

Patient inclusion criteria extended to previously untreated children presenting under the age of 24 months, of full-term pregnancy (>36 weeks), with triventricular hydrocephalus due to pure aqueduct stenosis diagnosed on MR scan. We strictly excluded children with spina bifida of any variety, prematurity, neonatal IVH, perinatal asphyxia, other congenital malformations such as Dandy-Walker syndrome or severe anatomical dysmorphic features (e.g., heterotopias, corpus callosum agenesis, large cysts) and central nervous system tumor.

Statistical considerations

Sample size

The sample size estimate is based on the ability to detect a statistically significant difference between the treatment groups for the primary outcome: the HUI score at 5 years of age. The following estimations and assumptions have been made:

  1. (a)

    The mean HUI score for the control (CSF shunt) group will be approximately 0.77 and the standard deviation will be 0.20, based on our previous work [32, 33].

  2. (b)

    The magnitude of a clinically meaningful difference in the primary outcome is judged to be an approximately 0.10 increase in the HUI score in the ETV cohort (i.e., a mean utility score of 0.87). This is also based on previous work (Kulkarni 2005 unpublished data).

  3. (c)

    α value=0.05 (two-sided) and β=0.20 (80% power).

Based on the preceding set of baseline assumptions and using the independent-samples t test, it is estimated that at least 64 patients per group are required to reach statistical significance. If we conservatively allow for an approximate 30% loss to follow-up, the sample size estimate can be adjusted to 91 patients per group.

Primary analysis

The primary analysis will be performed on the HUI utility scores obtained at 5 years of age. The primary analysis will be confined to the randomized cohort only and consist of a simple independent-samples t test. An adjusted, multivariable analysis of covariance will also be performed in which the following variables will be included: randomized intervention (ETV or CSF shunt), age at surgery (in weeks), gender, preoperative ventricular size, and baseline neurodevelopmental level.

Secondary analyses

Several secondary analyses will be performed, including survival analysis of time to failure of the initial intervention (to be performed 2 years after closure of enrollment), differences in outcome as measured by the HOQ, and comparison of perioperative complications between ETV and CSF shunt groups. These analyses will use data from both the randomized and nonrandomized cohort, separately and in combination.

All primary and secondary analyses for the randomized group will be done on an intention-to-treat basis, whereby patient outcome will be analyzed according to the group to which they were originally assigned.

Interim analysis

One interim analysis of survival, using Fisher’s exact test, is planned after 50% of the anticipated patient recruitment has completed an initial 9-month period of follow-up in the study. The results will be presented (with treatment allocation blinded) to the monitoring committee. The monitoring committee will then decide whether to continue with the study or stop. If one treatment achieves a statistically significant greater initial success rate at a p value below 0.001, then the monitoring committee will have instructions to recommend stopping the trial. The monitoring committee will not be restricted to an imbalance in the success rate alone. It may also consider other factors, including excessive adverse events.

Administration

For the first time in a neurosurgical prospective randomized trial, we are seeking to administer the entire study using internet technology. Traditionally, trials are administered through paper forms, which can result in a significant time delay between patient recruitment, patient randomization, data collection, and data compilation at the coordinating center. This often leads to collection of inaccurate data, as in most trials, though prospective, the various patient information forms are filled after the event, and not always completely. Use of online internet technology for recruitment, randomization, and data collection ensures contemporaneous data entry and improved accuracy, as the computer technology can force the user to answer certain key questions before proceeding to the next step. In addition, it allows the steering committee to have a real time view of the patients recruited and their eligibility and provide appropriate guidance to the participating centers.

The way forward

The IIHS has been endorsed with great enthusiasm by a great number of neurosurgeons, and as of April 2005, at least 25 centers have committed themselves to participate, and the first few patients have been recruited. It is expected that the study will provide the desired answers and create a close web of collaborating surgeons, who in the future will pursue even more ambitious projects. At the time of evidence-based medicine and ever declining funding for research, a contradiction of trends in its own right, it is refreshing to see surgeons striving to pursue their questions scientifically.