shunt survival after failed endoscopic treatment of hydrocephalus

8
J Neurosurg: Pediatrics / Volume 10 / December 2012 J Neurosurg Pediatrics 10:463–470, 2012 463 G IVEN the difficulties intrinsic to safely maintain- ing shunt function in rural East Africa, we have pursued endoscopic methods for treating infant hydrocephalus with the goal of preventing shunt depen- dence when possible. 14,15 We found that, for infants less than 1 year of age, combining CPC with ETV signifi- cantly increased the likelihood of treatment success over ETV alone, and have subsequently demonstrated the ef- ficacy of the combined ETV-CPC procedure in treating infant hydrocephalus of a variety of etiologies. 14,16,18,21,22 Nearly all endoscopic treatment failures were evident be- Shunt survival after failed endoscopic treatment of hydrocephalus Clinical article BENJAMIN C. W ARF , M.D., 1,2 SALMAN BHAI, B.S., 3 ABHAYA V. KULKARNI, M.D., PH.D., 4 AND JOHN MUGAMBA, M.D. 5 1 Department of Neurosurgery, Children’s Hospital Boston; 2 Department of Global Health and Social Medicine, 3 Harvard Medical School, Boston, Massachusetts; 4 Department of Neurosurgery, Hospital for Sick Children and University of Toronto, Ontario, Canada; 5 CURE Children’s Hospital of Uganda, Mbale, Uganda Object. It is not known whether previous endoscopic third ventriculostomy (ETV) affects the risk of shunt failure. Different epochs of hydrocephalus treatment at the CURE Children’s Hospital of Uganda (CCHU)—initially placing CSF shunts in all patients, then attempting ETV in all patients, and finally attempting ETV combined with choroid plexus cauterization (CPC) in all patients—provided the opportunity to assess whether prior endoscopic surgery affected shunt survival. Methods. With appropriate institutional approvals, the authors reviewed the CCHU clinical database to iden- tify 2329 patients treated for hydrocephalus from December 2000 to May 2007. Initial ventriculoperitoneal (VP) shunt placement was performed in 900 patients under one of three circumstances: 1) primary nonselective VP shunt placement with no endoscopy (255 patients); 2) VP shunt placement at the time of abandoned ETV attempt (with or without CPC) (370 patients); 3) VP shunt placement subsequent to a completed but failed ETV (with or without CPC) (275 patients). We analyzed time to shunt failure using the Kaplan-Meier method to construct survival curves, Cox proportional hazards regression modeling, and risk-adjusted analyses to account for possible confounding differences among these groups. Results. Shunt failure occurred in 299 patients, and the mean duration of follow-up for the remaining 601 was 28.7 months (median 18.8, interquartile range 4.1–46.3). There was no significant difference in operative mortality (p = 0.07 by log-rank and p = 0.14 by Cox regression adjusted for age and hydrocephalus etiology) or shunt infection (p = 0.94, log-rank) among the 3 groups. There was no difference in shunt survival between patients treated with pri- mary shunt placement and those who underwent shunt placement at the time of an abandoned ETV attempt (adjusted hazard ratio [HR] 1.14, 95% CI 0.86–1.51, p = 0.35). Those who underwent shunt placement after a completed but failed ETV (with or without CPC) had a lower risk of shunt failure (p = 0.008, log-rank), with a hazard ratio (adjusted for age at shunting and etiology) of 0.72 (95% CI 0.53–0.98), p = 0.03, compared with those who underwent primary shunt placement without endoscopy; but this was observed only in patients with postinfectious hydrocephalus (PIH) (adjusted HR 0.55, 95% CI 0.36–0.85, p = 0.007), and no effect was apparent for hydrocephalus of noninfectious etiologies (adjusted HR 0.98, 95% CI 0.64–1.50, p = 0.92). Improved shunt survival after failed ETV in the PIH group may be an artifact of selection arising from the inherent heterogeneity of ventricular damage within that group, or a consequence of the timing of shunt placement. The anticipated benefit of CPC in preventing future ventricular catheter obstruction was not observed. Conclusions. A paradigm for infant hydrocephalus involving intention to treat by ETV with or without CPC had no adverse effect on mortality or on subsequent shunt survival or infection risk. This study failed to demonstrate a positive effect of prior ETV or CPC on shunt survival. (http://thejns.org/doi/abs/10.3171/2012.9.PEDS1236) KEY WORDS shunt outcome endoscopic third ventriculostomy choroid plexus cauterization hydrocephalus Africa developing countries Abbreviations used in this paper: CCHU = CURE Children’s Hospital of Uganda; CPC = choroid plexus cauterization; ETV = endoscopic third ventriculostomy; HR = hazard ratio; PIH = postin- fectious hydrocephalus; VP = ventriculoperitoneal. See the corresponding editorial in this issue, pp 461–462.

Upload: john

Post on 09-Mar-2017

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Shunt survival after failed endoscopic treatment of hydrocephalus

J Neurosurg: Pediatrics / Volume 10 / December 2012

J Neurosurg Pediatrics 10:463–470, 2012

463

Given the difficulties intrinsic to safely maintain-ing shunt function in rural East Africa, we have pursued endoscopic methods for treating infant

hydrocephalus with the goal of preventing shunt depen-dence when possible.14,15 We found that, for infants less than 1 year of age, combining CPC with ETV signifi-cantly increased the likelihood of treatment success over ETV alone, and have subsequently demonstrated the ef-ficacy of the combined ETV-CPC procedure in treating infant hydrocephalus of a variety of etiologies.14,16,18,21,22 Nearly all endoscopic treatment failures were evident be-

Shunt survival after failed endoscopic treatment of hydrocephalus

Clinical article

Benjamin C. Warf, m.D.,1,2 Salman Bhai, B.S.,3 aBhaya V. KulKarni, m.D., Ph.D.,4 anD john mugamBa, m.D.5

1Department of Neurosurgery, Children’s Hospital Boston; 2Department of Global Health and Social Medicine, 3Harvard Medical School, Boston, Massachusetts; 4Department of Neurosurgery, Hospital for Sick Children and University of Toronto, Ontario, Canada; 5CURE Children’s Hospital of Uganda, Mbale, Uganda

Object. It is not known whether previous endoscopic third ventriculostomy (ETV) affects the risk of shunt failure. Different epochs of hydrocephalus treatment at the CURE Children’s Hospital of Uganda (CCHU)—initially placing CSF shunts in all patients, then attempting ETV in all patients, and finally attempting ETV combined with choroid plexus cauterization (CPC) in all patients—provided the opportunity to assess whether prior endoscopic surgery affected shunt survival.

Methods. With appropriate institutional approvals, the authors reviewed the CCHU clinical database to iden-tify 2329 patients treated for hydrocephalus from December 2000 to May 2007. Initial ventriculoperitoneal (VP) shunt placement was performed in 900 patients under one of three circumstances: 1) primary nonselective VP shunt placement with no endoscopy (255 patients); 2) VP shunt placement at the time of abandoned ETV attempt (with or without CPC) (370 patients); 3) VP shunt placement subsequent to a completed but failed ETV (with or without CPC) (275 patients). We analyzed time to shunt failure using the Kaplan-Meier method to construct survival curves, Cox proportional hazards regression modeling, and risk-adjusted analyses to account for possible confounding differences among these groups.

Results. Shunt failure occurred in 299 patients, and the mean duration of follow-up for the remaining 601 was 28.7 months (median 18.8, interquartile range 4.1–46.3). There was no significant difference in operative mortality (p = 0.07 by log-rank and p = 0.14 by Cox regression adjusted for age and hydrocephalus etiology) or shunt infection (p = 0.94, log-rank) among the 3 groups. There was no difference in shunt survival between patients treated with pri-mary shunt placement and those who underwent shunt placement at the time of an abandoned ETV attempt (adjusted hazard ratio [HR] 1.14, 95% CI 0.86–1.51, p = 0.35).

Those who underwent shunt placement after a completed but failed ETV (with or without CPC) had a lower risk of shunt failure (p = 0.008, log-rank), with a hazard ratio (adjusted for age at shunting and etiology) of 0.72 (95% CI 0.53–0.98), p = 0.03, compared with those who underwent primary shunt placement without endoscopy; but this was observed only in patients with postinfectious hydrocephalus (PIH) (adjusted HR 0.55, 95% CI 0.36–0.85, p = 0.007), and no effect was apparent for hydrocephalus of noninfectious etiologies (adjusted HR 0.98, 95% CI 0.64–1.50, p = 0.92). Improved shunt survival after failed ETV in the PIH group may be an artifact of selection arising from the inherent heterogeneity of ventricular damage within that group, or a consequence of the timing of shunt placement. The anticipated benefit of CPC in preventing future ventricular catheter obstruction was not observed.

Conclusions. A paradigm for infant hydrocephalus involving intention to treat by ETV with or without CPC had no adverse effect on mortality or on subsequent shunt survival or infection risk. This study failed to demonstrate a positive effect of prior ETV or CPC on shunt survival.(http://thejns.org/doi/abs/10.3171/2012.9.PEDS1236)

Key WorDS      •      shunt outcome      •      endoscopic third ventriculostomy      •      choroid plexus cauterization      •      hydrocephalus      •      Africa      •      developing countries

Abbreviations used in this paper: CCHU = CURE Children’s Hospital of Uganda; CPC = choroid plexus cauterization; ETV = endoscopic third ventriculostomy; HR = hazard ratio; PIH = postin-fectious hydrocephalus; VP = ventriculoperitoneal.

See the corresponding editorial in this issue, pp 461–462.

Page 2: Shunt survival after failed endoscopic treatment of hydrocephalus

B. C. Warf et al.

464 J Neurosurg: Pediatrics / Volume 10 / December 2012

fore 6 months, and these patients were ultimately treated with CSF shunting.14–16,18–22

It is not known whether a previous, but failed, ETV, with or without CPC, affects the future risk of shunt fail-ure. It has been suggested by others that an ETV per-formed at the time of or prior to shunt placement might decrease the subsequent risk of shunt malfunction.2,12 Since our treatment paradigm at CCHU had evolved to-ward first attempting endoscopic treatment in all patients, we were interested to know whether, compared with pa-tients treated with primary shunt placement, those with failed or abandoned endoscopic treatment might have an altered incidence of shunt malfunction or infection over time. It seemed plausible that prior endoscopic treatment might decrease the risk of future shunt malfunction be-cause, for instance, 1) a persistently patent ETV might decrease shunt dependence over time, or 2) decreased choroid plexus in the ventricle following CPC might re-duce the risk of future ventricular catheter obstruction. Alternatively, however, prior endoscopic treatment might increase the risk of subsequent shunt failure for various reasons, including the following: 1) blood or suspended debris in the CSF might increase the risk of subsequent ventricular catheter or valve obstruction; 2) there could be an increased risk of shunt infection when placement follows a concomitant or recent endoscopic procedure; 3) CPC could increase the risk of shunt infection, given the potential role of choroid plexus in the inflammatory response.24

Different epochs of treatment at the CURE Chil-dren’s Hospital of Uganda (initially treating all hydro-cephalus patients with shunt placement, then attempting ETV in all patients, and finally attempting ETV-CPC in all patients), without a change in the patient population or the surgeons involved, provided the opportunity to look for an effect. The retrospective study presented here was carried out to assess whether prior endoscopic surgery af-fected shunt survival.

MethodsAfter obtaining appropriate institutional approval

from both Children’s Hospital Boston and CCHU, we reviewed the CCHU database containing prospectively collected clinical data on patients requiring treatment for hydrocephalus at CCHU in Mbale, Uganda, between De-cember 2000 and May 2007. This period covers 2 distinct epochs of treatment philosophy at CCHU: primary place-ment of a CSF shunt without the use of endoscopy, and primarily attempting endoscopic treatment in all patients in an effort to avoid shunt placement. The present study includes an extended follow-up of some endoscopically treated patients who have been reported on in previous publications (cited below). Before June 2001, all patients with hydrocephalus were nonselectively treated by place-ment of a VP shunt. We previously reported our initial 1-year outcome for CSF shunting at CCHU, demonstrat-ing that shunt failure and infection risks at 1 year using the inexpensive Chhabra shunt were equivalent to those using the Codman-Hakim Micro Precision Valve shunt system and that these outcomes were consistent with

those reported from other series in North America.13 Af-ter June 2001, our treatment protocol changed, and nearly all patients underwent endoscopy for attempted ETV,15 and later on ETV-CPC,14 without regard for age or etiol-ogy of hydrocephalus. In some of these patients, the ETV attempt had to be abandoned for shunt placement because of difficult anatomy, poor visibility, or a severely scarred prepontine cistern discovered upon creation of the ETV. Patients with nontransparent CSF were not treated with immediate shunt placement but, rather, underwent place-ment of a ventricular access device for serial tapping until the CSF cleared. These patients subsequently underwent repeat endoscopy for attempted ETV or, in the case of the primary shunt treatment group, VP shunt placement. The results for the initial intention to treat by endoscopy for all patients during the same time period as the pres-ent study—including aborted ETV attempts, failed ETV (with or without CPC), and successful ETV (with or with-out CPC)—have been previously reported.20

Thus, all patients in this study received a VP shunt un-der one of the following circumstances: 1) those primarily and nonselectively treated with CSF shunting with no en-doscopy; 2) those treated with CSF shunting after failure of successfully completed endoscopy (ETV with or without CPC); 3) those treated with CSF shunting when the ETV attempt was abandoned (with or without CPC).

Patients were predominantly from Uganda, although some were from other regions in East Africa. Of 2329 pa-tients treated for hydrocephalus during this study period, we identified 914 in whom a VP shunt had been placed. Fourteen patients (1.5%) were eliminated from further study because they had previously undergone shunt placement elsewhere or because of incomplete informa-tion. Thus, 900 patients formed the basis of this analysis. When possible, patients lost to follow-up were contacted by telephone or home visit. The database record for every patient was reviewed independently in duplicate, includ-ing by the principle author, who was also the primary surgeon in the majority of procedures performed during this period (B.C.W.). For each patient the following infor-mation was obtained: date and nature of initial and sub-sequent operations, age at treatment, etiology of hydro-cephalus (postinfectious, myelomeningocele, or another noninfectious cause) as defined previously,15 date of shunt failure or death, and most recent follow-up. The time to shunt failure was defined as the time between initial shunt placement and any subsequent shunt operation or death related to shunt failure.

Shunt failure was defined as a subsequent shunt-re-lated operation for any reason, including infection, shunt obstruction, subdural collection, abdominal complication such as CSF pseudocyst, or death known to be related to shunt failure or infection. We used standard treatment success criteria to determine the adequacy of shunt func-tion: a shift in head circumference growth to a normal or less-than-normal rate, as plotted on a standard growth chart; decompression of the anterior fontanel; relief from symptoms of elevated intracranial pressure (such as irrita-bility and vomiting); resolution of abnormal eye findings (for example, sunsetting or cranial nerve VI palsy); and a decrease or arrest in ventriculomegaly as determined on

Page 3: Shunt survival after failed endoscopic treatment of hydrocephalus

J Neurosurg: Pediatrics / Volume 10 / December 2012

Shunt survival after ETV

465

ultrasonography or CT scanning by using the Evans ratio or frontal-occipital horn ratio, respectively. These are the same criteria for success that we have previously reported for endoscopic treatment. Shunt infection was determined by CSF profile, Gram stain, and culture result as well as clinical picture. Open or infected incisions and exposed shunt hardware from skin erosion were included as shunt infections. The criteria for diagnoses and clinical indica-tions for treatment were consistent among all groups and over the period of the study.

Statistical AnalysisWe analyzed time to shunt failure using the Kaplan-

Meier method to construct survival curves and Cox pro-portional hazards regression modeling. Our primary ob-jective was to determine if significant differences existed in time to shunt failure depending on what, if any, previ-ous endoscopic treatment the child had received before VP shunt insertion.

Patients were divided into the following groups for analysis: 1) primary nonselective VP shunt placement without endoscopy (255 patients); 2) immediate VP shunt placement at the time of aborted ETV attempt without CPC (340 patients); 3) immediate VP shunt placement af-ter aborted ETV attempt with CPC (30 patients); 4) VP shunt placement following ETV failure (141 patients); 5) VP shunt placement following CPC or ETV-CPC failure (134 patients). Preliminary analysis suggested that there was little difference between Groups 2 and 3 (p = 0.69, log-rank test) and Groups 4 and 5 (p = 0.97, log-rank test) Therefore, our analyses were limited to comparing dif-ferences among 3 treatment groups: primary nonselec-tive VP shunt placement, immediate VP shunt placement after aborted endoscopy, and VP shunt placement after completed, but failed, endoscopy.

We performed risk-adjusted analyses to account for possible confounding differences among these groups. Age at VP shunt insertion and etiology of hydrocephalus (myelomeningocele, postinfectious, nonpostinfectious, other/unknown), both known to have some prognostic effect on hydrocephalus treatment,20 were included in a multivariate Cox regression model. We also considered adjusting for the patency of the cerebral aqueduct (open, closed, or unknown) and presence of prepontine cisternal scarring (yes, no, or unknown), which were determined by direct inspection at the time of endoscopy, as we have demonstrated these to be independent factors that affect the likelihood of ETV success.19 We decided, however, not to include these latter 2 variables in the final model because they were impossibly confounded with the use of endoscopy (they were almost never assessable in patients who did not undergo endoscopy), and it would therefore not be possible to provide any independent interpretation of these variables. The proportional hazards assumption for each independent variable was confirmed by visual examination of survival curves (for categorical variables) and assessment of partial residual plots as a function of time (for all variables). All variance inflation factors were less than 3, suggesting no concerning collinearity among the variables in the multivariate model.6 The analyses were repeated using time to death (from any cause) and

time to shunt infection as the dependent variable. All analyses were done with SPSS Advanced Statistics 17.0 (SPSS Inc.).

ResultsPatient Characteristics

The characteristics of the 900 patients analyzed are shown in Table 1. Among the 3 treatment groups, there was no significant difference in age at shunt insertion (p = 0.8, ANOVA), but the distribution of etiologies was sta-tistically different (p < 0.001, chi-square). Postinfectious hydrocephalus accounted for more than 78% of the group with abandoned ETV and immediate shunt placement.

Patient Follow-UpShunt failure was known to occur at some point in

299 patients. The mean follow-up time for the 601 pa-tients who were not known to have had shunt failure was 28.7 months (median 18.8 months, interquartile range 4.1–46.3 months). The duration of follow-up was less than 6 months for 179 patients; but 31 of these were known to have died of causes that were thought to be unrelated to shunt malfunction or infection. Thus 148 patients (16.4%) were lost to follow-up before 6 months after shunt place-ment. Follow-up data for more than 1 month were avail-able for 536 (89.2%) of those with no known shunt mal-function.

MortalityThere were 13 known deaths from any cause within

30 days after VP shunt placement, for an overall operative mortality of 1.4%. Ten of these deaths did not appear to be related to shunt failure, and 3 occurred in the context of shunt failure. Of 99 confirmed deaths, 6 were directly attributed to shunt failure, including 2 shunt infections, one on postoperative Day 1, and the other on postopera-tive Day 545.

Patients with abandoned or failed endoscopic proce-dures did not have a higher operative mortality for shunt placement than those who did not undergo endoscopy, and there was actually a trend toward a lower mortality rate in these cases (Fig. 1) (p = 0.07 by log-rank and p = 0.14 by Cox regression adjusted for age and etiology).

Shunt FailureFor the entire sample, the shunt survival rate at 6

months, 1 year, 2 years, and 5 years was 74.3%, 69.5%, 64.6%, and 56.5%, respectively (Kaplan-Meier method). Of 299 shunt failures, 206 (68.9%) occurred within 180 days of placement. Etiology did not have a clear impact on shunt survival, with no significant difference in fail-ure between infants with PIH and myelomeningocele and those with nonpostinfectious hydrocephalus (p = 0.91, log-rank).

Compared with those who were treated with primary shunt placement without endoscopy, survival analysis demonstrated a lower risk of shunt failure for those who underwent shunt placement after a completed but failed

Page 4: Shunt survival after failed endoscopic treatment of hydrocephalus

B. C. Warf et al.

466 J Neurosurg: Pediatrics / Volume 10 / December 2012

endoscopy (p = 0.008, log-rank) with a hazard ratio (ad-justed for age at shunting and etiology) of 0.72 (95% CI 0.53–0.98), p = 0.03 (Fig. 2). There was no significant dif-ference between patients shunted nonselectively with no endoscopy and those shunted at the time of an abandoned endoscopy attempt (adjusted HR 1.14, 95% CI 0.86–1.51, p = 0.35).

Given that PIH accounted for a disproportionate number of patients who underwent shunt placement at the time of the abandoning of an ETV attempt (78.4%), there was concern that this could have created selection bias. The ventricles of PIH patients treated with shunt place-ment after ETV failure would have been less involved by ependymal scarring and destruction than their coun-terparts in whom the ETV attempt had been abandoned due to distorted anatomy or poor visibility. Because PIH composed such a distinct and dominant group, we sub-

sequently performed a stratified subgroup analysis of all PIH cases (578 cases) (Fig. 3A) and all other etiologies with PIH excluded (322 cases) (Fig. 3B). This demonstrat-ed a robust effect of prior ETV in the PIH group (adjusted HR 0.55, 95 % CI 0.36–0.85, p = 0.007) but no significant effect for the noninfectious etiologies (adjusted HR 0.98, 95% CI 0.64–1.50, p = 0.92). For both the PIH group and all others, there was no statistically significant difference in shunt survival between those who underwent primary shunt placement and those who underwent shunt place-ment at the time of an aborted endoscopy (p = 0.38 and p = 0.89, respectively). Thus, the PIH group entirely ac-counted for the positive effect of previous ETV on shunt survival in the overall group, and there appeared to be no such effect among the other etiologies.

Shunt InfectionAmong the 900 new shunt placements, there were

119 total known shunt infections. The infection rate at 30

TABLE 1: Patient characteristics*

Characteristic Total Primary VPS Immediate VPS at Abandoned ETV† VPS After Failed ETV†

no. of patients 900 255 370 275hydrocephalus etiology PIH 578 (64.2) 125 (49.0) 290 (78.4) 163 (59.3) myelomeningocele 110 (12.2) 41 (16.1) 25 (6.8) 44 (16.0) non-PIH 156 (17.3) 65 (25.5) 31 (8.4) 60 (21.8) other 56 (6.2) 24 (9.4) 24 (6.5) 8 (2.9)mean age at VPS (mos) 9.2 ± 18.6 9.8 ± 20.1 8.9 ± 18.6 9.0 ± 17.3

* Values represent numbers of patients (%) unless otherwise indicated. Mean values are given with SD. Abbreviation: VPS = VP shunt.† With or without CPC.

Fig. 1. Time to death from any cause by treatment group. Kaplan-Meier survival curve for time to death from any cause after VP shunt placement. Group A = immediate VP shunt placement at abandoned endoscopy, Group B = VP shunt placement after failed endoscopy, and Group C = primary VP shunt placement without endoscopy.

Fig. 2. Time to shunt failure by treatment group. Kaplan-Meier sur-vival curve for time to shunt failure after VP shunt placement. Group A = immediate VP shunt placement at abandoned endoscopy, Group B = VP shunt placement after failed endoscopy, and Group C = primary VP shunt placement without endoscopy.

Page 5: Shunt survival after failed endoscopic treatment of hydrocephalus

J Neurosurg: Pediatrics / Volume 10 / December 2012

Shunt survival after ETV

467

days, 90 days, 6 months, 1 year, and 2 years was, respec-tively, 5.7%, 9.5%, 11.7%, 13.3%, and 15.2% (Kaplan-Meier method). It is notable that, although the majority of known infections occurred in the first 6 months, late shunt infec-tions continued to occur over the 7-year time span at a rate of about 1% per year (Fig. 4). Of 119 shunt infections, 10 presented primarily with wound infection or breakdown or skin erosion over the shunt hardware, a complication typi-cally associated with severe malnutrition.

Survival analysis demonstrated no difference in shunt infection among any of the 3 treatment groups (pri-mary shunt placement, shunt placement after abandoned endoscopy, and shunt placement after failed endoscopy) (p = 0.94, log-rank). As with shunt failures, it is notable

that the infection rate was not higher among infants with PIH or myelomeningocele, for whom CSF contamination might be more likely, when compared with those with nonpostinfectious hydrocephalus of other causes (p = 0.17, log-rank).

DiscussionThe purpose of this study was to investigate whether,

compared with patients treated with primary shunt place-ment, those with failed or abandoned endoscopic treat-ment might have an altered incidence of shunt malfunc-tion or infection over time. It was our hypothesis that prior completed ETV or CPC might decrease the risk of shunt failure, and it was our hope that an intention to treat by endoscopy in all patients did not increase the risk of shunt infection or failure. These data demonstrate no ef-fect of an abandoned ETV attempt on shunt survival or infection risk relative to primary shunt placement without endoscopy. Interestingly, patients with PIH who under-went shunt placement following failure of a completed ETV had significantly better shunt survival than those in whom a shunt was placed as a primary treatment or at the time of abandoned ETV, while no such effect was observed among those with noninfectious etiologies. Al-though we had thought that CPC might decrease the risk of future shunt malfunction, this study failed to demon-strate such an effect, and CPC did not increase the risk of future shunt infection. That all the significant effect of prior ETV came purely from the PIH group limits the conclusions that can be drawn.

Patients with abandoned or failed endoscopic proce-dures did not have a higher operative mortality for shunt placement than those without endoscopy, and there was actually a trend toward a lower mortality in these cases. In geographically defined cohort studies with very success-ful follow-up, we have previously reported that childhood mortality for infants with both myelomeningocele and PIH (the 2 most common individual etiologies of hydro-cephalus at CCHU) is significantly greater than for their

Fig. 3. Upper: Time to shunt failure by treatment group for all PIH patients. Kaplan-Meier survival curve for time to shunt failure after VP shunt for PIH patients only (n = 578). Lower: Time to shunt failure by treatment group for all other etiologies. Kaplan-Meier survival curve for time to shunt failure after VP shunt with all PIH patients excluded (n = 322). Group A = immediate VP shunt at abandoned endoscopy, Group B = VP shunt after failed endoscopy, and Group C = primary VP shunt without endoscopy.

Fig. 4. Time to shunt infection. Kaplan-Meier survival curve for time to shunt infection after VP shunt for the entire cohort.

Page 6: Shunt survival after failed endoscopic treatment of hydrocephalus

B. C. Warf et al.

468 J Neurosurg: Pediatrics / Volume 10 / December 2012

unaffected peers.17,23 The under-5-year mortality among infants treated for myelomeningocele and those treated for PIH was 37% and 30%, respectively, compared with the overall reported childhood mortality of 16% in Ugan-da. We demonstrated that most deaths were due to com-mon treatable conditions unrelated to hydrocephalus or its treatment 17,23 and that survival in myelomeningocele patients was equivalent to that in unaffected Ugandan children in the context of a community-based rehabilita-tion program.23 Thus, the mortality rate observed in the present study was completely expected.

Gallo et al.2 suggested that previous ETV might de-crease future shunt malfunction, and Shim et al.12 sug-gested that simultaneous ETV and shunt placement had the same effect. But, these reports involved small num-bers of patients and lacked statistical significance. Our findings in a much larger population fail to indicate a statistically significant protective effect of a successfully completed ETV (with or without CPC) against future shunt malfunction in patients with noninfectious causes of hydrocephalus. Surprisingly, a robust effect was ob-served in the PIH group. There are a few possible expla-nations for and interpretations of our findings:

Could Previous ETV With or Without CPC Have a Physiologically Protective Effect Against the Failure of a Subsequently Implanted VP Shunt?

Among patients undergoing endoscopy for hydro-cephalus treatment during the study period, 60.4% were younger than 6 months, and 85.2% were younger than 12 months of age.20 Young age is a negative factor for ETV success,7,8,20 and this may be related to the evolving ef-ficacy of CSF absorption in infants.9 This had formed the basis for our original hypothesis that CPC might improve the outcome for ETV in infants, in that an even tempo-rary reduction in the rate of CSF production might bal-ance any residual “communicating” hydrocephalus (from insufficient absorption) that may remain after ETV had successfully bypassed the “obstructive” component.14 It is possible that a persistently patent ETV might become more effective at treating hydrocephalus over time, ren-dering a subsequent shunt malfunction clinically irrel-evant. However, the data presented here failed to sup-port this in patients with hydrocephalus of noninfectious causes. Although it is possible that an initially failed ETV may allow some children to become shunt-independent over time, it seems equally possible that a patent ETV at the time of shunt placement may close as a result of ven-tricular decompression from the shunt. Finally, if ETV itself decreased the risk of future shunt malfunction, the effect should be observed in the noninfectious etiologies. That this would only be observed for PIH is difficult to explain.

Likewise, we had considered the possibility that for some patients, CPC alone could have provided some pro-tection against future shunt failure. It seemed possible that CPC might become sufficient to treat a “communicating” hydrocephalus after the child is older and CSF absorptive mechanisms have become sufficient to achieve the neces-sary balance. But there was no significant difference in shunt survival between those with failed ETV alone and

those with failed ETV-CPC. Alternatively, CPC, by reduc-ing the choroid plexus to a line of scar tissue, should prevent ventricular catheter obstruction by choroid plexus, which is often regarded as the most common cause of shunt mal-function.3,4,10,11 However, we found no difference in shunt survival in regard to whether CPC had been performed, which fails to support this attractive hypothesis.

Why Did a Prior ETV Appear to Protect Against Shunt Malfunction in the PIH Group?

Patients with completed ETVs are distinct from those in whom completion of the procedure was not possible, and the latter group might have had an intrinsically higher risk of shunt failure. This initially seemed less likely, since the treatment groups represented different epochs of treatment and, within each epoch, treatment was offered nonselec-tively. Initially, all patients were treated with CSF shunting, and subsequently all patients were treated with attempted ETV or ETV-CPC in an unselected manner. But patients with PIH are a heterogeneous group with varying degrees of prior ventricular destruction, those with the worst in-volvement accounting for the majority of those for whom ETV had been abandoned. This raises a strong possibility of selection bias within the PIH group. However, it is im-portant to observe that there was no significant difference in outcome between patients who underwent shunt place-ment without endoscopy and patients treated with shunt placement at the time of an aborted ETV attempt. The latter group (78.4% of whom had PIH) primarily included those with severe anatomical distortion and significant cis-ternal scarring. If the improved shunt outcome following failed ETV were only from the aforementioned effect of selection, then one would anticipate a higher shunt failure risk in the remaining group (that is, those in whom ETV was abandoned), than that observed in patients nonselec-tively treated with shunt placement without endoscopy, which would have included all patients. That this was not the case suggests a possibly real protective effect of the pri-or ETV. However, as noted above, it is counterintuitive that such an effect would be confined to those with PIH. It is possible that endoscopic fenestration of membranes in PIH patients at the time of attempted ETV and shunt placement could have improved the ultimate shunt outcome in a man-ner sufficient to account for the equivalence of outcome in the group of patients treated with primary shunt placement and the group that underwent shunt placement at the time of an abandoned ETV.

Was There a Protective “Second-Surgery Effect” in Which There was Some Advantage to the Second, Delayed Procedure of Shunt Placement?

This could have been the case for those patients with PIH whose CSF at the time of the initial operation might have been contaminated or might have contained debris that had cleared by the time of the second operation. It would not be an anticipated benefit for those with nonin-fectious etiologies.

Were There Other Unaccounted Residual Confounders?Given that the patient groups were treated in different

Page 7: Shunt survival after failed endoscopic treatment of hydrocephalus

J Neurosurg: Pediatrics / Volume 10 / December 2012

Shunt survival after ETV

469

epochs, it is possible that the more recently treated endos-copy group received better overall medical care because the medical team had acquired more experience over the study period. Given the relative technical ease of shunt insertion, it seems unlikely that better surgical skill was contributory. As well, there was no obvious shunt surviv-al advantage to those who were underwent shunt place-ment at the time of aborted endoscopy, who theoretically should also have benefited if there was an improvement in overall care for the more recent patients.

Also, this effect would have been the same regardless of hydrocephalus etiology. Another possible confounder is shunt location and insertion technique. Shunts placed at aborted, or after failed, endoscopy were inserted fron-tally, while primary shunts were placed occipitally. But, this would have been true for patients with infectious and noninfectious etiologies of hydrocephalus, alike. In addi-tion, shunt location is unlikely to have had an effect, since the frontally placed shunts that were placed after aborted endoscopy did not show a survival advantage compared with the shunts placed as a primary treatment. Also, the advantage of one shunt location over the other remains unconfirmed.1 The same is true for any advantage from endoscopic assistance of ventricular catheter placement, which was used in both the abandoned endoscopy and completed endoscopy groups. In addition, endoscopically assisted ventricular catheter placement has been previ-ously found to be of no benefit in avoiding shunt mal-function.5

ConclusionsWe report, in a large population of patients treated

with CSF shunting for hydrocephalus, that the paradigm of primarily attempting ETV with or without CPC had no adverse effect on mortality or on subsequent shunt sur-vival or infection risk. An apparent benefit of prior ETV on shunt survival was observed only in patients with PIH, but this seems most likely an artifact of selection or tim-ing of the shunt placement arising from the inherent het-erogeneity of ventricular damage within that group. The anticipated benefit of CPC in preventing future ventricu-lar catheter obstruction was not observed.

Disclosure

The authors report no conflict of interest concerning the mate-rials or methods used in this study or the findings specified in this paper.

Author contributions to the study and manuscript preparation include the following. Conception and design: Warf. Acquisition of data: Warf, Bhai, Mugamba. Analysis and interpretation of data: Warf, Bhai, Kulkarni. Drafting the article: Warf. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Warf. Statistical analysis: Warf, Kulkarni. Admin-istrative/technical/material support: Warf, Bhai, Mugamba. Study supervision: Warf.

References

1. Dickerman RD, McConathy WJ, Morgan J, Stevens QE, Jolley JT, Schneider S, et al: Failure rate of frontal versus parietal

approaches for proximal catheter placement in ventriculoperi-toneal shunts: revisited. J Clin Neurosci 12:781–783, 2005

2. Gallo P, Szathmari A, De Biasi S, Mottolese C: Endoscopic third ventriculostomy in obstructive infantile hydrocephalus: remarks about the so-called ‘unsuccessful cases.’ Pediatr Neurosurg 46:435–441, 2010

3. Ginsberg HJ, Drake JM, Peterson TM, Cobbold RS: Recana-lization of obstructed cerebrospinal fluid ventricular cath-eters using ultrasonic cavitation. Neurosurgery 59(4 Suppl 2):ONS403–ONS412, 2006

4. Gnanalingham KK, Lafuente J, Brew S, Thompson D, Hark-ness W, Hayward R: Percutaneous coagulation of choroid plexus to unblock the ventricular catheter using the Seldinger technique: preliminary report. Surg  Neurol 64:440–443, 2005

5. Kestle JR, Drake JM, Cochrane DD, Milner R, Walker ML, Abbott R III, et al: Lack of benefit of endoscopic ventriculo-peritoneal shunt insertion: a multicenter randomized trial. J Neurosurg 98:284–290, 2003

6. Kleinbaum DG, Kupper LL, Muller KE: Applied Regression Analysis and Other Multivariable Methods, ed 2. Belmont, CA: Duxbury Press, 1988, pp 209–214

7. Kulkarni AV, Drake JM, Kestle JR, Mallucci CL, Sgouros S, Constantini S: Predicting who will benefit from endoscopic third ventriculostomy compared with shunt insertion in child-hood hydrocephalus using the ETV Success Score. Clinical article. J Neurosurg Pediatr 6:310–315, 2010 (Erratum in J Neurosurg Pediatr 7:221, 2011)

8. Kulkarni AV, Warf BC, Drake JM, Mallucci CL, Sgouros S, Constantini S: Surgery for hydrocephalus in sub-Saharan Af-rica versus developed nations: a risk-adjusted comparison of outcome. Childs Nerv Syst 26:1711–1717, 2010

9. Oi S, Di Rocco C: Proposal of “evolution theory in cerebro-spinal fluid dynamics” and minor pathway hydrocephalus in developing immature brain. Childs Nerv Syst 22:662–669, 2006

10. Pang D, Grabb PA: Accurate placement of coronal ventricular catheter using stereotactic coordinate-guided free-hand pas-sage. Technical note. J Neurosurg 80:750–755, 1994

11. Sekhar LN, Moossy J, Guthkelch AN: Malfunctioning ven-triculoperitoneal shunts. Clinical and pathological features. J Neurosurg 56:411–416, 1982

12. Shim KW, Kim DS, Choi JU: Simultaneous endoscopic third ventriculostomy and ventriculoperitoneal shunt for infantile hydrocephalus. Childs Nerv Syst 24:443–451, 2008

13. Warf BC: Comparison of 1-year outcomes for the Chhabra and Codman Hakim Micro Precision shunt systems in Uganda: a prospective study in 195 children. J Neurosurg (4 Suppl) 102: 358–362, 2005

14. Warf BC: Comparison of endoscopic third ventriculostomy alone and combined with choroid plexus cauterization in in-fants younger than 1 year of age: a prospective study in 550 African children. J Neurosurg (6 Suppl) 103:475–481, 2005

15. Warf BC: Hydrocephalus in Uganda: predominance of infec-tious origin and primary management with endoscopic third ventriculostomy. J Neurosurg (1 Suppl) 102:1–15, 2005

16. Warf BC, Campbell JW: Combined endoscopic third ventricu-lostomy and choroid plexus cauterization (ETV/CPC) as pri-mary treatment of hydrocephalus for infants with myelome-ningocele: long-term results of a prospective intention to treat study in 115 African infants. Clinical article. J Neurosurg Pediatr 2:310–316, 2008

17. Warf BC, Dagi AR, Nsbuga B, Schiff SJ: Five-year survival and outcome of treatment for post-infectious hydrocephalus in Ugandan infants. Clinical article. J Neurosurg Pediatr 8: 502–508, 2011

18. Warf BC, Dewan M, Mugamba J: Management of Dandy-Walker Complex-associated infant hydrocephalus by com-bined endoscopic third ventriculostomy and choroid plexus

Page 8: Shunt survival after failed endoscopic treatment of hydrocephalus

B. C. Warf et al.

470 J Neurosurg: Pediatrics / Volume 10 / December 2012

cauterization (ETV/CPC). Clinical article. J Neurosurg Pe-diatr 8:377–383, 2011

19. Warf BC, Kulkarni AV: Intraoperative assessment of cerebral aqueduct patency and cisternal scarring: impact on success of endoscopic third ventriculostomy in 403 African children. Clinical article. J Neurosurg Pediatr 5:204–209, 2010

20. Warf BC, Mugamba J, Kulkarni AV: Endoscopic third ven-triculostomy in the treatment of childhood hydrocephalus in Uganda: report of a scoring system that predicts success. Clinical article. J Neurosurg Pediatr 5:143–148, 2010

21. Warf BC, Stagno V, Mugamba J: Encephalocele in Uganda: Ethnic variation in lesion location, endoscopic management of hydrocephalus, and survival in 110 consecutive children. Clinical article. J Neurosurg Pediatr 7:88–93, 2011

22. Warf BC, Tracy S, Mugamba J: Long-term outcome for com-bined endoscopic third ventriculostomy alone or in combina-tion with choroid plexus cauterization compared with choroid plexus cauterization for congenital aqueductal stenosis in Af-

rican infants. Clinical article. J Neurosurg Pediatr 10:108–111, 2012

23. Warf BC, Wright EJ, Kulkarni AV: Factors affecting survival of infants with myelomeningocele in southeastern Uganda. Clinical article. J Neurosurg Pediatr 7:127–133, 2011

24. Wewer C, Seibt A, Wolburg H, Greune L, Schmidt MA, Berg-er J, et al: Transcellular migration of neutrophil granulocytes through the blood-cerebrospinal fluid barrier after infection with Streptococcus suis. J Neuroinflammation 8:51, 2011

Manuscript submitted January 17, 2012.Accepted September 10, 2012.Please include this information when citing this paper: pub-

lished online October 5, 2012; DOI: 10.3171/2012.9.PEDS1236.Address correspondence to: Benjamin C. Warf, M.D., Children’s

Hospital Boston, Department of Neurosurgery, 300 Longwood Avenue, Boston, Massachusetts 02115. email: [email protected].