J Neurosurg Pediatrics 14:160–166, 2014 ©AANS, 2014
Utility of computed tomography or magnetic resonance imaging evaluation of ventricular morphology in suspected cerebrospinal fluid shunt malfunction Clinical article Jonathan N. Sellin, M.D., Jacob Cherian, M.D., James M. Barry, M.D., Sheila L. Ryan, M.P.H., J.D., Thomas G. Luerssen, M.D., and Andrew Jea, M.D. Division of Pediatric Neurosurgery, Texas Children’s Hospital, and Department of Neurosurgery, Baylor College of Medicine, Houston, Texas Object. It is common to evaluate children with suspected CSF shunt malfunctions using CT of the head or, more recently, “quick brain” MRI. However, the reliability of using ventricular behavior, as assessed on cranial imaging during previous presentations with shunt obstructions, is not well defined. The authors conducted a study to determine if CT or MRI of ventricular morphology added useful clinical information in the evaluation of shunt malfunctions. Methods. A retrospective chart review of children operated on at Texas Children’s Hospital from February 20, 2011, to June 18, 2013, for shunt obstruction was conducted. Inclusion criteria involved age 3 years or older in patients who had undergone two or more shunt revisions for intraoperatively confirmed obstructions. Patients with shunt infection but without shunt obstruction and patients with fourth ventricular shunt failure were excluded from the study. Preoperative CT or MRI results were dichotomized into two distinct categories, as determined by a radiologist’s report: either dilation of the ventricular system in comparison with prior scans at points the shunt was deemed functional, or no dilation of the ventricular system in comparison such scans. Determination of the presence of shunt obstruction was assessed by findings documented by the surgeon in the operative report. Each case was then analyzed to see if the patient has a reliable pattern of ventricular dilation, or no dilation, at times of shunt obstruction. Results. Forty-two patients (25 males and 17 females) were included in the study. There were a total of 117 patient encounters analyzed and an average of 2.79 encounters per patient. The mean age at shunt failure presentation was 10.8 years (range 3–23 years). In 4 encounters, patients presented with a CSF leak or pseudomeningocele. Twenty-seven patients (64%) consistently demonstrated dilation of the ventricular system during episodes of shunt obstruction. Four patients (10%) consistently demonstrated no dilation during episodes of shunt obstruction. Eleven patients (26%) demonstrated inconsistent changes in ventricular size at times of shunt obstruction. In those first patient encounters with shunt obstruction presenting with ventricular dilation, 92% (49 of 53) of subsequent encounters demonstrated ventricular dilation with shunt obstruction presentations. Conclusions. Historical CT or MRI data regarding ventricular morphology patterns seen during prior examinations of shunt obstructions may inform a clinician’s judgment of shunt obstruction on subsequent presentations, but they are not conclusive. In the present series, the authors found that changes in the morphology of a given patient’s ventricular system when shunt obstruction occurs were often consistent and predictable, but not always. It remains imperative, however, that cranial images obtained to rule out shunt malfunction be compared with prior studies. (http://thejns.org/doi/abs/10.3171/2014.4.PEDS13451)
Key Words • computed tomography • hydrocephalus • pediatric • neurosurgery • magnetic resonance imaging
U
nfortunately, shunt malfunction is one of the most common clinical problems in pediatric neurosurgery. The diagnosis may be challenging to establish for even the most experienced clinician.1,5 Children with hydrocephalus, and a CSF shunt, presenting for such evaluations often have significant neurological abnormalities and developmental delay. Symptoms such
Abbreviation used in this paper: ETV = endoscopic third ventriculostomy.
160
as headache, vomiting, lethargy, and seizures may accompany shunt malfunction but are nonspecific, making shunt obstruction a routine consideration in this patient population.3,4,6 Moreover, the neurological examination is often limited and unreliable in these patients. Clinicians consequently depend heavily on imaging to evaluate for shunt obstruction.3,7,8 This article contains some figures that are displayed in color online but in black-and-white in the print edition.
J Neurosurg: Pediatrics / Volume 14 / August 2014
CT and MRI patterns for shunt obstruction Computed tomography scans of the brain and a shunt series of radiographs are routinely used to aid in the diagnosis of shunt malfunction.5,7,8,10 More recently, quick brain MRI has emerged as a reasonable alternative to CT for the evaluation of ventricular morphology. The size of the ventricles may be small, normal, or enlarged in the presence of shunt malfunction.3,6,11 In the present retrospective study of our pediatric population, we attempt to evaluate the utility of comparing CT or MRI evaluations of ventricular morphology during prior instances of shunt obstruction to new studies obtained for subsequent presentations for shunt obstruction. Comparing ventricular morphology on presentation to ventricular morphology at first or subsequent shunt obstructions is imperative, and the findings may be predictive in determining the present status of the shunt system.
Methods
A retrospective chart review of patients undergoing shunt revision or externalization at Texas Children’s Hospital was undertaken; records were from patients treated between February 20, 2011, and June 18, 2013. Inclusion criteria were any patient 3 years of age or old with two or more shunt revisions after the age of 3 years that took place during the review period. All data were collected via electronic medical record. This review was approved by the institutional review board. Cases that involved shunt revisions for infection without shunt obstruction were excluded from the study. Patients who underwent shunt revision for infection without obstruction, but who also underwent two or more shunt revisions for shunt obstruction without infection during the study period were included in the study. Shunt revisions or externalizations for abdominal pseudocysts were included, and these cases were considered to represent distal obstructions. Cases that involved posterior fossa shunts were not included in this analysis. Encounters in which individuals were inpatients with externalized shunts or external ventricular drains undergoing a “weaning” trial that was deemed a failure based on clinical grounds, in which cranial imaging was performed, were included. “Weaning” trials also included temporary occlusion of the CSF shunting apparatus after attempted endoscopic third ventriculostomy (ETV). In these instances, inclusion of CT or MRI ventricular morphology data was deemed valuable because these patients had iatrogenic shunt “obstruction.” Demographic data (age, sex, date of birth, and cause of hydrocephalus) were collected for all patients. Operative data (number of shunt revisions, date of surgery, presenting symptoms, shunt type, ventricular location of proximal catheter, and location of obstruction) were collected for all patients. The determination of the presence of shunt malfunction and its location was based on findings documented in the operative report. Results of CT or MRI were dichotomized into two distinct categories, as determined by a radiologist’s report: either 1) dilation of the ventricular system in comparison with prior scans at points when the shunt was deemed functional or 2) no dilation of the ventricular system in comparison with prior scans at points when the shunt was deemed functional. Members of the research team reviewed all images in J Neurosurg: Pediatrics / Volume 14 / August 2014
question. The clinical context of all images reviewed was noted. On the rare occasion when the radiologist’s reading did not explicitly compare new images to baseline images, we made the comparison ourselves by using the appropriate images and the medical record. In essence, the research team curated the review process to ensure that appropriate comparisons to baseline morphology were made. Outcome data and the interval between shunt revisions or time to last follow-up were collected for all patients. Statistical Analysis
The open source R software package (http://www.rproject.org/) was used to calculate 2-sided Fisher’s exact test p values to analyze contingency tables of comparison groups. For analysis of statistical significance, a p value threshold < 0.05 was used.
Results
Forty-two patients were included in our study. There were a total of 117 patient encounters that were analyzed and an average of 2.79 encounters per patient (range 2–7) (Table 1). Demographics
The mean age at shunt failure presentation was 10.8 years (range 3–23 years). Our patient population included 25 males and 17 females.
Clinical and Operative Data
The etiology of hydrocephalus was myelomeningocele in 9 patients, intraventricular hemorrhage of prematurity in 16 patients, tumor in 6 patients, aqueductal stenosis in 2 patients, trauma in 1 patient, postinfectious in 1 patient, and other/unknown in 7 patients (Table 2). In 98 encounters, patients presented with classic signs and symptoms of increased intracranial pressure including headache, nausea and vomiting, lethargy, and seizures (Table 3). In 4 encounters, patients presented with a CSF leak or pseudomeningocele. Of the 117 encounters, 106 involved patients with ventriculoperitoneal shunts, 9 involved patients with ventriculoatrial shunts, and none involved patients with ventriculopleural shunts. Of all of the shunt revisions undertaken, 41 (35%) of the 117 encounters involved a proximal catheter revised via a frontal approach, 60 (51%) involved a proximal catheter revised via a parietal approach, and 23 (17%) involved a proximal catheter revised via an occipital entry site. Fifteen cases involved a CSF shunt system with proximal catheters in more than one location.
Outcome
There were 65 revisions associated with a proximal shunt obstruction, 28 associated with a valvular obstruction, and 36 associated with a distal obstruction. Note that 22 shunt malfunctions were determined to be due to more than one site of obstruction.
Follow-Up
The mean interval between shunt revisions or time 161
J. N. Sellin et al. TABLE 1: Variables collected from patient records
TABLE 3: Clinical presentation
Ventricular System Group (%) Encounters
Dilated
total no. of encounters mean age at shunt failure (yrs) sex male female shunt type ventriculoperitoneal ventriculoatrial ventriculopleural ETV location frontal parietal occipital ETV cause of obstruction* proximal only valve only distal only proximal + valve proximal + distal valve + distal proximal + valve + distal ETV failure trapped horn unknown
No. of Encounters (%)
Not Dilated
p Value
Clinical Presentation
Total
Dilated
Not Dilated
0.2911
47 (51) 45 (49) 82 (89) 9 (10) 0 (0) 1 (1) 34 (37) 45 (49) 20 (22) 1 (1)
25 9.9 16 (64) 9 (36) 24 (96) 0 (0) 0 (0) 1 (4) 7 (28) 15 (60) 3 (12) 1 (4)
0.2678 0.4529 0.2014 1 0.3831 0.4833 0.3722 0.3974 0.3831
total no. of encounters headache vomiting lethargy seizures pseudomeningocele/ CSF leak
117 65 66 49 9 4
92 50 (54) 51 (55) 36 (39) 7 (8) 3 (3)
25 15 (60) 15 (60) 13 (52) 2 (8) 1 (4)
40 (43) 5 (5) 16 (17) 10 (11) 1 (1) 2 (2) 5 (5) 5 (5) 2 (2) 6 (7)
7 (28) 3 (12) 9 (36) 0 (0) 0 (0) 1 (4) 2 (8) 1 (4) 0 (0) 2 (8)
0.177 0.3651 0.056 0.1164 1 0.5172 0.6406 1 1 0.6788
92 11.1
* Distal and proximal refer to the distal and proximal catheter.
to last follow-up was 4.2 months (range 0–18 months). Twenty-seven shunt failures occurred within 1 month of the last shunt presentation. Predictive Value of CT- or MRI-Documented Morphology of Ventricular System for Shunt Obstruction
Overall, CT was the cranial imaging modality in 109
TABLE 2: Etiology of hydrocephalus stratified by presence, absence, or indeterminacy of ventricular dilation* No. of Patients (n = 42) Etiology
Dilators
Nondilators
Indeterminate
Total
myelomeningocele IVH of prematurity tumor aqueductal stenosis trauma postinfectious other/unknown
8 8 6 0 0 1 4
0 2 0 1 0 0 1
1 6 0 1 1 0 2
9 16 6 2 1 1 7
* IVH = intraventricular hemorrhage.
162
p Value 0.6563 0.8208 0.2624 1 1
(93.2%) of 117 encounters, and MRI was the cranial imaging modality in 8 (6.8%) of 117 encounters. Twentyseven patients (64%) demonstrated a predictable pattern of dilation of the ventricular system during instances of shunt obstruction (Table 4). In 4 (10%) of 42 patients, a predictable pattern of nondilation of the ventricular system was seen during instances of shunt obstruction. In 11 patients (26%), we observed inconsistent changes in ventricular size at the time of shunt obstruction and throughout the follow-up period. When ventricular dilation, as opposed to nondilation, was documented at the first presentation, it was then frequently seen in subsequent encounters (Fig. 1). In those encounters with patients in whom ventricular dilation was observed during a first presentation for shunt obstruction, ventricular dilation was demonstrated at shunt failure presentation in 92% (or 49 of 53) of subsequent encounters. This is in contrast to those patient encounters that did not involve ventricular dilation on first presentation: only 41% (or 9 of 22) of subsequent encounters involved patients with ventricular nondilation on shunt failure presentation. Cases without dilation at first presentation for shunt obstruction were more likely to fall in the “radiographically unpredictable” group than those cases with dilation at first presentation for shunt obstruction. This was statistically significant (8 of 12 patients vs 3 of 30 patients, p = 0.000492). Shunt failure encounters in which patients presented with nondilated ventricle were more likely to have distal shunt failures than shunt failure encounters in which patients presented with dilated ventricles, but this was not statistically significant (9 of 25 patient encounters vs 16 of 92 patient encounters, p = 0.056) (Table 1). The analysis in Fig. 2 excludes those encounters where more than one cause of shunt failure was identified. In terms of age, sex, etiology, shunt type, or proximal catheter location, there were no statistically significant differences when examining shunt encounters involving dilation versus nondilation. There was also no statistical TABLE 4: Summary of CT and MRI results and clinical outcomes CT/MRI Findings
Total No. of Patients (%)
no dilation dilation indeterminate
4 (10) 27 (64) 11 (26)
J Neurosurg: Pediatrics / Volume 14 / August 2014
CT and MRI patterns for shunt obstruction
Fig. 1. Schematic representation of imaging results.
significance in the distribution of the etiologies of hydrocephalus and the pattern of shunt failure (dilators, nondilators, or indeterminate). Regarding pseudocysts, in particular, 10 of 117 patient encounters demonstrated shunt failure due to pseudocyst formation. The presence of a pseudocyst was not significantly associated with a nondilated ventricular system (p = 0.1367). Not surprisingly, the presence of a pseudocyst did correlate with abdominal complaints as a presenting symptom (p = 0.02037). Interestingly and in keeping with anecdotal experience, pseudocyst encounters were less likely to present with symptoms of intracranial hypertension (headache, vomiting, lethargy, and seizure), although this association was not statistically significant (p = 0.05578).
Fig. 2. 100% stacked bar graph comparing cause of shunt failure versus radiographic presentation. Highlights show proportions of different causes. Patients who had more than one cause of shunt failure identified intraoperatively were excluded. The denominators for the dilators (n = 61) and nondilators (n = 19) represent the total of patient encounters in which patients had exclusively proximal, distal, or valvular malfunction; encounters in which malfunction resulted from a mixed cause were excluded.
J Neurosurg: Pediatrics / Volume 14 / August 2014
Discussion
Although a patient’s hydrocephalus may be successfully treated with a CSF shunt, there remains the risk of significant long-term complications. The most common of these is failure of the CSF shunt, which happens in over 50% of children within 2 years of surgery and the lifetime risk is at least 80%.2 Missing the diagnosis of shunt malfunction may lead to permanent neurological injury or death. Routine use of imaging modalities in the evaluation of possible CSF shunt malfunction typically includes CT scans of the brain, shunt series radiographs, and more recently quick brain MR images. A CT scan or quick MRI study yields great detail about ventricular morphology, which aids in diagnosis, the modality is readily available, and the study can be done quickly. Radiographic imaging will sometimes, but not always, show an increase in the size of the ventricles compared with a baseline post-shunt image.3 It is important to recognize that an unchanged ventricle size on imaging does not necessarily rule out a CSF shunt obstruction. In the present study, we attempted to evaluate the reliability of historical patterns of ventricular morphology that were documented in the past when patients had presented with shunt obstructions in informing future assessments of shunt function to minimize the chance of misdiagnosis. This evaluation has allowed us to develop a treatment algorithm for patients with suspected shunt obstruction when historical radiographic data are available to provide context (Fig. 3). In our study, 64% of patients presenting with shunt obstruction consistently had ventricular dilation during episodes of shunt obstruction. In 10% of patients, we consistently observed a pattern of ventricular nondilation during episodes of shunt obstruction. However, in 26% of patients there was no predictable pattern of changes in ventricular morphology from one episode of shunt obstruction to the next. Thus, the pattern of changes in ventricular morphology tends to follow a predictable pattern. Approximately 74% of our patients were shown to reliably have dilation or nondilation during shunt obstruction, although this tendency is not absolute. 163
J. N. Sellin et al.
Fig. 3. Flow chart illustrating an imaging approach to evaluate shunt malfunction. ICP = intracranial pressure.
Interestingly, the initial presentation for shunt obstruction seems particularly telling: the vast majority patients who exhibited ventricular dilation on initial presentation also had dilation on subsequent presentations, with 92% (49 of 53) of patient encounters following an initial presentation of ventricular dilation demonstrating ventricular dilation on subsequent presentations. Conversely, of those patient encounters that followed a failure to exhibit ventricular dilation on initial presentation, 59% (13 of 22 patient encounters) patients demonstrated subsequent ventricular dilation. Compared with the first group of patients, who showed dilation on initial presentation, this difference was statistically significant (49 of 52 patient encounters vs 13 of 22 patient encounters, p = 0.0005417), suggesting that patients with dilation on initial presentation are more likely to have dilatation on subsequent encounters. Thus, those in whom dilation was 164
absent at first presentation for shunt obstruction were more likely to fall into a “radiographically unpredictable” group, making subsequent presentations for shunt obstruction more challenging for the clinician. While the historical relationship between ventricular morphology and shunt status seems intuitive, to the best of our knowledge the fact that the relationship is often but not always reliable has not been shown in prior studies. We maintain that CT or MRI in this setting still adds value in the evaluation of possible shunt obstruction. As a first principle, ventricular dilation in the setting of suspected shunt malfunction, regardless of historical patterns, is highly suggestive of shunt obstruction and warrants surgical exploration. A diagnostic dilemma exists, however, when a patient has signs and symptoms of shunt malfunction but no ventricular dilation on CT or MRI. Our findings suggest that if patients present with J Neurosurg: Pediatrics / Volume 14 / August 2014
CT and MRI patterns for shunt obstruction worrisome symptoms but no dilation of the ventricular system on imaging, and if patients have a historical pattern of ventricular dilation on their initial and subsequent shunt obstruction presentations, then the clinician may be more inclined to argue against shunt obstruction and to look for another etiology for the symptoms. This conclusion is born from the fact that the majority of our patients (64%) exhibited such a pattern and that in 92% of patient encounters following an initial presentation of ventricular dilation the patients had ventricular dilation on subsequent presentations. If, however, patients do not show radiographic evidence of consistently having dilated ventricles with occurrences of shunt obstruction, and if their ventricles are not dilated on a new presentation, the value CT or MRI in evaluating shunt obstruction is less clear. Ten percent of our patients consistently showed no evidence of ventricular dilation when they in fact had shunt obstruction, and 36% (15 of 42) of patients showed no ventricular dilation in at least one episode of shunt obstruction. In approximately 21% (25 of 117) of instances of shunt obstructions in our series, there were no changes in ventricular size or morphology on radiographic imaging prior to surgery. Thus, the absence of ventricular dilation in the nondilating or inconsistently dilatory groups is nondiagnostic. Therefore, in the evaluation of new shunt obstruction, the absence of ventricular dilation on CT or MRI in patients with ventricles that do not historically dilate, or inconsistently dilate, should not be used as the only or definitive diagnostic determinant when evaluating shunt function. In a study by Iskandar et al.5 that reported low sensitivity of CT scanning in detecting shunt malfunction, there was no mention of shunt malfunction in the radiologist’s interpretation of CT scans of 24 of 100 children requiring shunt revision. Most of these children had slit-ventricle syndrome or shunt overdrainage. A study by Naftel et al. examined the accuracy of parental diagnoses of shunt failure but also reviewed the test characteristics of a head CT scan in the evaluation of shunt failure.9 This study, which demonstrated a CT sensitivity of 79% and specificity of 86%, as well as the study by Iskandar et al., did not explicitly examine ventricular dilation and did not consider longitudinal imaging patterns observed during multiple instances of shunt malfunction in the same patient. Studies by Mater et al. and Lehnert et al. similarly examined the test characteristics of various imaging modalities in diagnosing shunt failure—defined as an operative shunt revision—but did not examine how consistent the results of such imaging modalities were in the same patient during multiple instances of shunt obstruction.7,8 To the best of our knowledge, the present study is the first to examine the reliability of antecedent cranial imaging patterns during episodes of shunt obstruction as useful clinical information in the evaluation of shunt malfunctions. Limitations
There are several limitations to this study. Most importantly, it is a retrospective chart review and is subject to all of the shortcoming associated with this type of analysis, including documentation omissions, missed relevant cases, and selection bias. The appropriateness of
J Neurosurg: Pediatrics / Volume 14 / August 2014
the evaluation for suspected shunt obstruction and the necessity of obtaining the CT or MRI studies are at the sole discretion of the physician. Differences in imaging technique also made comparing ventricular volume and size between CT and MRI studies difficult, perhaps indicating a change in serial data when in fact there was none. Also, we definitively identified shunt obstruction based on the surgical findings in the operative reports. However, there is no standard definition for shunt obstruction. A shunt obstruction could represent an occluded shunt system or simply a poorly functioning shunt system. We relied on the possibly biased interpretations of intraoperative findings by the operating surgeon to classify shunt status. One could argue that, by only including patients who underwent surgery for shunt obstruction, we have failed to include patients who presented with shunt obstruction but did not receive surgery. It seems a fair presumption, however, that patients with missed shunt obstructions will eventually visit, either because of clinical deterioration or because they re-present, to the emergency center. Our purpose was to characterize ventricular morphology during instances of confirmed shunt obstruction—including only surgical cases, in which obstruction was confirmed intraoperatively. Additionally, our definition of ventricular dilation was broad. We relied on independent radiology reports to determine, in a dichotomous manner, if dilation of the ventricular system was present or absent. However, the degree of ventricular dilation was not measured or quantified. One particular criticism of this study may be our reliance on the impression of radiologists rather than pediatric neurosurgeons. First, one may argue that experienced pediatric neurosurgeons are as or more proficient at assessing ventricular changes than radiologists. Another criticism of our study may be that radiologists were not comparing the index images to previous “well” images and were oblivious to the immediate and historical clinical context that prompted cranial imaging. It is in fact the standard practice of neuroradiologists at our institution to compare new images to antecedent scans, including baseline images (“well” scans) and prior “failure” scans (MRI or CT scans that show evidence of shunt failure) taken at times of confirmed shunt obstruction. The advent of electronic medical records at our institution and the resultant ease of access to clinical information have made this practice possible. Radiologists’ readings routinely include a paragraph in which “comparison studies” are detailed, revealing the radiologists’ attention to the longitudinal view of ventricular changes in a given patient. Our research team ensured this attention to clinical context during our data collection. We hold that independent review by a pediatric neuroradiologist is a strength of our study. These physicians are board certified in radiology and neuroradiology. They practice at a high-volume, free-standing children’s hospital and have extensive experience in reading and interpreting cranial imaging in the setting of shunted hydrocephalus. Unfortunately, neurosurgeons, while indeed very adept at reviewing cranial imaging, are not board certified in radiology or neuroradiology. The use of radiologists eliminates 165
J. N. Sellin et al. evaluation bias. If we were to include a neurosurgery versus radiology assessment, we would then need to address the issue of interobserver variability and reliability, which, to the best of our knowledge, has not been established. To do so would be beyond the scope of our study but is, indeed, a potential topic for a future study. Finally, our study analyzed a subgroup of children with a history of multiple shunt obstructions. We arbitrarily chose to analyze the age of 3 years or older. We surmised that at 3 years of age each patient’s fontanelle would be closed and their sutures fused such that there would be no peculiarities to alter the intracranial CSF dynamics or ventricular compliance in very young children. It would have been particularly interesting to analyze the relationship between total time with a CSF shunt and presentation with nondilation with shunt failure. It is often thought that early and chronic CSF shunting leads to changes in ventricular compliance that resist fulminant ventricular dilation at times of new shunt obstruction. Unfortunately, this information was either inconsistently available or entirely unavailable for our patient population, given the number of outside referrals and only the recent institution of our electronic medical record system. As electronic health record systems evolve, these data may be more readily available for further study.
Conclusions
Routine use of CT, quick MRI, and shunt series radiography is common in the evaluation of CSF shunt malfunction. To the best of our knowledge, no other studies exist that evaluate the efficacy of using radiographic evaluations of ventricular morphology obtained during workup for prior shunt obstructions in the workup of new or suspected shunt malfunction in the symptomatic pediatric patient. Our study indicates that CT or MRI evaluation of ventricular morphology remains valuable, as does clinical examination. In particular, if there is evidence of ventricular dilation at the first presentation of shunt obstruction, then the clinician may suspect each subsequent instance of shunt malfunction to present with CT evidence of ventricular dilation. Furthermore, if a patient has a demonstrated of pattern of ventricular dilation with shunt obstruction, the absence of ventricular dilation during a new presentation argues against shunt obstruction. Suspicion, however, should not lead to conclusions based on radiographic parameters alone. CT or MRI data play an important role in the safe and efficacious evaluation of suspected shunt obstruction, but they do not obviate the need for clinical examination and judgment. Future directions may involve prospective collection of ventricular morphology data, analysis of the relationship between ventricular morphology during obstruction and time shunted, application of clinical decision rules based on radiographic findings, and volumetric calculation to estimate ventricular volumes in children with possible shunt obstruction.
166
Disclosure The authors report no conflict of interest concerning the materials 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: Jea. Acquisition of data: Jea. Analysis and interpretation of data: Jea. Drafting the article: Jea, Sellin. 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: Jea. Statistical analysis: Jea. Study supervision: Luerssen. References 1. Blumstein H, Schardt S: Utility of radiography in suspected ventricular shunt malfunction. J Emerg Med 36:50–54, 2009 2. Drake JM, Kestle JR, Milner R, Cinalli G, Boop F, Piatt J Jr, et al: Randomized trial of cerebrospinal fluid shunt valve design in pediatric hydrocephalus. Neurosurgery 43:294–305, 1998 3. Engel M, Carmel PW, Chutorian AM: Increased intraventricular pressure without ventriculomegaly in children with shunts: “normal volume” hydrocephalus. Neurosurgery 5:549–552, 1979 4. Garton HJ, Kestle JR, Drake JM: Predicting shunt failure on the basis of clinical symptoms and signs in children. J Neurosurg 94:202–210, 2001 5. Iskandar BJ, McLaughlin C, Mapstone TB, Grabb PA, Oakes WJ: Pitfalls in the diagnosis of ventricular shunt dysfunction: radiology reports and ventricular size. Pediatrics 101:1031– 1036, 1998 6. Kim TY, Stewart G, Voth M, Moynihan JA, Brown L: Signs and symptoms of cerebrospinal fluid shunt malfunction in the pediatric emergency department. Pediatr Emerg Care 22: 28–34, 2006 7. Lehnert BE, Rahbar H, Relyea-Chew A, Lewis DH, Richardson ML, Fink JR: Detection of ventricular shunt malfunction in the ED: relative utility of radiography, CT, and nuclear imaging. Emerg Radiol 18:299–305, 2011 8. Mater A, Shroff M, Al-Farsi S, Drake J, Goldman RD: Test characteristics of neuroimaging in the emergency department evaluation of children for cerebrospinal fluid shunt malfunction. CJEM 10:131–135, 2008 9. Naftel RP, Tubergen E, Shannon CN, Gran KA, Vance EH, Oakes WJ, et al: Parental recognition of shunt failure: a prospective single-institution study. Clinical article. J Neurosurg Pediatr 9:363–371, 2012 10. Naradzay JF, Browne BJ, Rolnick MA, Doherty RJ: Cerebral ventricular shunts. J Emerg Med 17:311–322, 1999 11. Sivaganesan A, Krishnamurthy R, Sahni D, Viswanathan C: Neuroimaging of ventriculoperitoneal shunt complications in children. Pediatr Radiol 42:1029–1046, 2012
Manuscript submitted August 29, 2013. Accepted April 28, 2014. Please include this information when citing this paper: published online May 23, 2014; DOI: 10.3171/2014.4.PEDS13451. Address correspondence to: Andrew Jea, M.D., Division of Pediatric Neurosurgery, Texas Children’s Hospital, 6621 Fannin St., CCC 1230.01, 12th Floor, Houston, TX 77030. email: ahjea@ texaschildrens.org.
J Neurosurg: Pediatrics / Volume 14 / August 2014
Utility of computed tomography or magnetic resonance imaging evaluation of ventricular morphology in suspected cerebrospinal fluid shunt malfunction Clinical article Jonathan N. Sellin, M.D., Jacob Cherian, M.D., James M. Barry, M.D., Sheila L. Ryan, M.P.H., J.D., Thomas G. Luerssen, M.D., and Andrew Jea, M.D. Division of Pediatric Neurosurgery, Texas Children’s Hospital, and Department of Neurosurgery, Baylor College of Medicine, Houston, Texas Object. It is common to evaluate children with suspected CSF shunt malfunctions using CT of the head or, more recently, “quick brain” MRI. However, the reliability of using ventricular behavior, as assessed on cranial imaging during previous presentations with shunt obstructions, is not well defined. The authors conducted a study to determine if CT or MRI of ventricular morphology added useful clinical information in the evaluation of shunt malfunctions. Methods. A retrospective chart review of children operated on at Texas Children’s Hospital from February 20, 2011, to June 18, 2013, for shunt obstruction was conducted. Inclusion criteria involved age 3 years or older in patients who had undergone two or more shunt revisions for intraoperatively confirmed obstructions. Patients with shunt infection but without shunt obstruction and patients with fourth ventricular shunt failure were excluded from the study. Preoperative CT or MRI results were dichotomized into two distinct categories, as determined by a radiologist’s report: either dilation of the ventricular system in comparison with prior scans at points the shunt was deemed functional, or no dilation of the ventricular system in comparison such scans. Determination of the presence of shunt obstruction was assessed by findings documented by the surgeon in the operative report. Each case was then analyzed to see if the patient has a reliable pattern of ventricular dilation, or no dilation, at times of shunt obstruction. Results. Forty-two patients (25 males and 17 females) were included in the study. There were a total of 117 patient encounters analyzed and an average of 2.79 encounters per patient. The mean age at shunt failure presentation was 10.8 years (range 3–23 years). In 4 encounters, patients presented with a CSF leak or pseudomeningocele. Twenty-seven patients (64%) consistently demonstrated dilation of the ventricular system during episodes of shunt obstruction. Four patients (10%) consistently demonstrated no dilation during episodes of shunt obstruction. Eleven patients (26%) demonstrated inconsistent changes in ventricular size at times of shunt obstruction. In those first patient encounters with shunt obstruction presenting with ventricular dilation, 92% (49 of 53) of subsequent encounters demonstrated ventricular dilation with shunt obstruction presentations. Conclusions. Historical CT or MRI data regarding ventricular morphology patterns seen during prior examinations of shunt obstructions may inform a clinician’s judgment of shunt obstruction on subsequent presentations, but they are not conclusive. In the present series, the authors found that changes in the morphology of a given patient’s ventricular system when shunt obstruction occurs were often consistent and predictable, but not always. It remains imperative, however, that cranial images obtained to rule out shunt malfunction be compared with prior studies. (http://thejns.org/doi/abs/10.3171/2014.4.PEDS13451)
Key Words • computed tomography • hydrocephalus • pediatric • neurosurgery • magnetic resonance imaging
U
nfortunately, shunt malfunction is one of the most common clinical problems in pediatric neurosurgery. The diagnosis may be challenging to establish for even the most experienced clinician.1,5 Children with hydrocephalus, and a CSF shunt, presenting for such evaluations often have significant neurological abnormalities and developmental delay. Symptoms such
Abbreviation used in this paper: ETV = endoscopic third ventriculostomy.
160
as headache, vomiting, lethargy, and seizures may accompany shunt malfunction but are nonspecific, making shunt obstruction a routine consideration in this patient population.3,4,6 Moreover, the neurological examination is often limited and unreliable in these patients. Clinicians consequently depend heavily on imaging to evaluate for shunt obstruction.3,7,8 This article contains some figures that are displayed in color online but in black-and-white in the print edition.
J Neurosurg: Pediatrics / Volume 14 / August 2014
CT and MRI patterns for shunt obstruction Computed tomography scans of the brain and a shunt series of radiographs are routinely used to aid in the diagnosis of shunt malfunction.5,7,8,10 More recently, quick brain MRI has emerged as a reasonable alternative to CT for the evaluation of ventricular morphology. The size of the ventricles may be small, normal, or enlarged in the presence of shunt malfunction.3,6,11 In the present retrospective study of our pediatric population, we attempt to evaluate the utility of comparing CT or MRI evaluations of ventricular morphology during prior instances of shunt obstruction to new studies obtained for subsequent presentations for shunt obstruction. Comparing ventricular morphology on presentation to ventricular morphology at first or subsequent shunt obstructions is imperative, and the findings may be predictive in determining the present status of the shunt system.
Methods
A retrospective chart review of patients undergoing shunt revision or externalization at Texas Children’s Hospital was undertaken; records were from patients treated between February 20, 2011, and June 18, 2013. Inclusion criteria were any patient 3 years of age or old with two or more shunt revisions after the age of 3 years that took place during the review period. All data were collected via electronic medical record. This review was approved by the institutional review board. Cases that involved shunt revisions for infection without shunt obstruction were excluded from the study. Patients who underwent shunt revision for infection without obstruction, but who also underwent two or more shunt revisions for shunt obstruction without infection during the study period were included in the study. Shunt revisions or externalizations for abdominal pseudocysts were included, and these cases were considered to represent distal obstructions. Cases that involved posterior fossa shunts were not included in this analysis. Encounters in which individuals were inpatients with externalized shunts or external ventricular drains undergoing a “weaning” trial that was deemed a failure based on clinical grounds, in which cranial imaging was performed, were included. “Weaning” trials also included temporary occlusion of the CSF shunting apparatus after attempted endoscopic third ventriculostomy (ETV). In these instances, inclusion of CT or MRI ventricular morphology data was deemed valuable because these patients had iatrogenic shunt “obstruction.” Demographic data (age, sex, date of birth, and cause of hydrocephalus) were collected for all patients. Operative data (number of shunt revisions, date of surgery, presenting symptoms, shunt type, ventricular location of proximal catheter, and location of obstruction) were collected for all patients. The determination of the presence of shunt malfunction and its location was based on findings documented in the operative report. Results of CT or MRI were dichotomized into two distinct categories, as determined by a radiologist’s report: either 1) dilation of the ventricular system in comparison with prior scans at points when the shunt was deemed functional or 2) no dilation of the ventricular system in comparison with prior scans at points when the shunt was deemed functional. Members of the research team reviewed all images in J Neurosurg: Pediatrics / Volume 14 / August 2014
question. The clinical context of all images reviewed was noted. On the rare occasion when the radiologist’s reading did not explicitly compare new images to baseline images, we made the comparison ourselves by using the appropriate images and the medical record. In essence, the research team curated the review process to ensure that appropriate comparisons to baseline morphology were made. Outcome data and the interval between shunt revisions or time to last follow-up were collected for all patients. Statistical Analysis
The open source R software package (http://www.rproject.org/) was used to calculate 2-sided Fisher’s exact test p values to analyze contingency tables of comparison groups. For analysis of statistical significance, a p value threshold < 0.05 was used.
Results
Forty-two patients were included in our study. There were a total of 117 patient encounters that were analyzed and an average of 2.79 encounters per patient (range 2–7) (Table 1). Demographics
The mean age at shunt failure presentation was 10.8 years (range 3–23 years). Our patient population included 25 males and 17 females.
Clinical and Operative Data
The etiology of hydrocephalus was myelomeningocele in 9 patients, intraventricular hemorrhage of prematurity in 16 patients, tumor in 6 patients, aqueductal stenosis in 2 patients, trauma in 1 patient, postinfectious in 1 patient, and other/unknown in 7 patients (Table 2). In 98 encounters, patients presented with classic signs and symptoms of increased intracranial pressure including headache, nausea and vomiting, lethargy, and seizures (Table 3). In 4 encounters, patients presented with a CSF leak or pseudomeningocele. Of the 117 encounters, 106 involved patients with ventriculoperitoneal shunts, 9 involved patients with ventriculoatrial shunts, and none involved patients with ventriculopleural shunts. Of all of the shunt revisions undertaken, 41 (35%) of the 117 encounters involved a proximal catheter revised via a frontal approach, 60 (51%) involved a proximal catheter revised via a parietal approach, and 23 (17%) involved a proximal catheter revised via an occipital entry site. Fifteen cases involved a CSF shunt system with proximal catheters in more than one location.
Outcome
There were 65 revisions associated with a proximal shunt obstruction, 28 associated with a valvular obstruction, and 36 associated with a distal obstruction. Note that 22 shunt malfunctions were determined to be due to more than one site of obstruction.
Follow-Up
The mean interval between shunt revisions or time 161
J. N. Sellin et al. TABLE 1: Variables collected from patient records
TABLE 3: Clinical presentation
Ventricular System Group (%) Encounters
Dilated
total no. of encounters mean age at shunt failure (yrs) sex male female shunt type ventriculoperitoneal ventriculoatrial ventriculopleural ETV location frontal parietal occipital ETV cause of obstruction* proximal only valve only distal only proximal + valve proximal + distal valve + distal proximal + valve + distal ETV failure trapped horn unknown
No. of Encounters (%)
Not Dilated
p Value
Clinical Presentation
Total
Dilated
Not Dilated
0.2911
47 (51) 45 (49) 82 (89) 9 (10) 0 (0) 1 (1) 34 (37) 45 (49) 20 (22) 1 (1)
25 9.9 16 (64) 9 (36) 24 (96) 0 (0) 0 (0) 1 (4) 7 (28) 15 (60) 3 (12) 1 (4)
0.2678 0.4529 0.2014 1 0.3831 0.4833 0.3722 0.3974 0.3831
total no. of encounters headache vomiting lethargy seizures pseudomeningocele/ CSF leak
117 65 66 49 9 4
92 50 (54) 51 (55) 36 (39) 7 (8) 3 (3)
25 15 (60) 15 (60) 13 (52) 2 (8) 1 (4)
40 (43) 5 (5) 16 (17) 10 (11) 1 (1) 2 (2) 5 (5) 5 (5) 2 (2) 6 (7)
7 (28) 3 (12) 9 (36) 0 (0) 0 (0) 1 (4) 2 (8) 1 (4) 0 (0) 2 (8)
0.177 0.3651 0.056 0.1164 1 0.5172 0.6406 1 1 0.6788
92 11.1
* Distal and proximal refer to the distal and proximal catheter.
to last follow-up was 4.2 months (range 0–18 months). Twenty-seven shunt failures occurred within 1 month of the last shunt presentation. Predictive Value of CT- or MRI-Documented Morphology of Ventricular System for Shunt Obstruction
Overall, CT was the cranial imaging modality in 109
TABLE 2: Etiology of hydrocephalus stratified by presence, absence, or indeterminacy of ventricular dilation* No. of Patients (n = 42) Etiology
Dilators
Nondilators
Indeterminate
Total
myelomeningocele IVH of prematurity tumor aqueductal stenosis trauma postinfectious other/unknown
8 8 6 0 0 1 4
0 2 0 1 0 0 1
1 6 0 1 1 0 2
9 16 6 2 1 1 7
* IVH = intraventricular hemorrhage.
162
p Value 0.6563 0.8208 0.2624 1 1
(93.2%) of 117 encounters, and MRI was the cranial imaging modality in 8 (6.8%) of 117 encounters. Twentyseven patients (64%) demonstrated a predictable pattern of dilation of the ventricular system during instances of shunt obstruction (Table 4). In 4 (10%) of 42 patients, a predictable pattern of nondilation of the ventricular system was seen during instances of shunt obstruction. In 11 patients (26%), we observed inconsistent changes in ventricular size at the time of shunt obstruction and throughout the follow-up period. When ventricular dilation, as opposed to nondilation, was documented at the first presentation, it was then frequently seen in subsequent encounters (Fig. 1). In those encounters with patients in whom ventricular dilation was observed during a first presentation for shunt obstruction, ventricular dilation was demonstrated at shunt failure presentation in 92% (or 49 of 53) of subsequent encounters. This is in contrast to those patient encounters that did not involve ventricular dilation on first presentation: only 41% (or 9 of 22) of subsequent encounters involved patients with ventricular nondilation on shunt failure presentation. Cases without dilation at first presentation for shunt obstruction were more likely to fall in the “radiographically unpredictable” group than those cases with dilation at first presentation for shunt obstruction. This was statistically significant (8 of 12 patients vs 3 of 30 patients, p = 0.000492). Shunt failure encounters in which patients presented with nondilated ventricle were more likely to have distal shunt failures than shunt failure encounters in which patients presented with dilated ventricles, but this was not statistically significant (9 of 25 patient encounters vs 16 of 92 patient encounters, p = 0.056) (Table 1). The analysis in Fig. 2 excludes those encounters where more than one cause of shunt failure was identified. In terms of age, sex, etiology, shunt type, or proximal catheter location, there were no statistically significant differences when examining shunt encounters involving dilation versus nondilation. There was also no statistical TABLE 4: Summary of CT and MRI results and clinical outcomes CT/MRI Findings
Total No. of Patients (%)
no dilation dilation indeterminate
4 (10) 27 (64) 11 (26)
J Neurosurg: Pediatrics / Volume 14 / August 2014
CT and MRI patterns for shunt obstruction
Fig. 1. Schematic representation of imaging results.
significance in the distribution of the etiologies of hydrocephalus and the pattern of shunt failure (dilators, nondilators, or indeterminate). Regarding pseudocysts, in particular, 10 of 117 patient encounters demonstrated shunt failure due to pseudocyst formation. The presence of a pseudocyst was not significantly associated with a nondilated ventricular system (p = 0.1367). Not surprisingly, the presence of a pseudocyst did correlate with abdominal complaints as a presenting symptom (p = 0.02037). Interestingly and in keeping with anecdotal experience, pseudocyst encounters were less likely to present with symptoms of intracranial hypertension (headache, vomiting, lethargy, and seizure), although this association was not statistically significant (p = 0.05578).
Fig. 2. 100% stacked bar graph comparing cause of shunt failure versus radiographic presentation. Highlights show proportions of different causes. Patients who had more than one cause of shunt failure identified intraoperatively were excluded. The denominators for the dilators (n = 61) and nondilators (n = 19) represent the total of patient encounters in which patients had exclusively proximal, distal, or valvular malfunction; encounters in which malfunction resulted from a mixed cause were excluded.
J Neurosurg: Pediatrics / Volume 14 / August 2014
Discussion
Although a patient’s hydrocephalus may be successfully treated with a CSF shunt, there remains the risk of significant long-term complications. The most common of these is failure of the CSF shunt, which happens in over 50% of children within 2 years of surgery and the lifetime risk is at least 80%.2 Missing the diagnosis of shunt malfunction may lead to permanent neurological injury or death. Routine use of imaging modalities in the evaluation of possible CSF shunt malfunction typically includes CT scans of the brain, shunt series radiographs, and more recently quick brain MR images. A CT scan or quick MRI study yields great detail about ventricular morphology, which aids in diagnosis, the modality is readily available, and the study can be done quickly. Radiographic imaging will sometimes, but not always, show an increase in the size of the ventricles compared with a baseline post-shunt image.3 It is important to recognize that an unchanged ventricle size on imaging does not necessarily rule out a CSF shunt obstruction. In the present study, we attempted to evaluate the reliability of historical patterns of ventricular morphology that were documented in the past when patients had presented with shunt obstructions in informing future assessments of shunt function to minimize the chance of misdiagnosis. This evaluation has allowed us to develop a treatment algorithm for patients with suspected shunt obstruction when historical radiographic data are available to provide context (Fig. 3). In our study, 64% of patients presenting with shunt obstruction consistently had ventricular dilation during episodes of shunt obstruction. In 10% of patients, we consistently observed a pattern of ventricular nondilation during episodes of shunt obstruction. However, in 26% of patients there was no predictable pattern of changes in ventricular morphology from one episode of shunt obstruction to the next. Thus, the pattern of changes in ventricular morphology tends to follow a predictable pattern. Approximately 74% of our patients were shown to reliably have dilation or nondilation during shunt obstruction, although this tendency is not absolute. 163
J. N. Sellin et al.
Fig. 3. Flow chart illustrating an imaging approach to evaluate shunt malfunction. ICP = intracranial pressure.
Interestingly, the initial presentation for shunt obstruction seems particularly telling: the vast majority patients who exhibited ventricular dilation on initial presentation also had dilation on subsequent presentations, with 92% (49 of 53) of patient encounters following an initial presentation of ventricular dilation demonstrating ventricular dilation on subsequent presentations. Conversely, of those patient encounters that followed a failure to exhibit ventricular dilation on initial presentation, 59% (13 of 22 patient encounters) patients demonstrated subsequent ventricular dilation. Compared with the first group of patients, who showed dilation on initial presentation, this difference was statistically significant (49 of 52 patient encounters vs 13 of 22 patient encounters, p = 0.0005417), suggesting that patients with dilation on initial presentation are more likely to have dilatation on subsequent encounters. Thus, those in whom dilation was 164
absent at first presentation for shunt obstruction were more likely to fall into a “radiographically unpredictable” group, making subsequent presentations for shunt obstruction more challenging for the clinician. While the historical relationship between ventricular morphology and shunt status seems intuitive, to the best of our knowledge the fact that the relationship is often but not always reliable has not been shown in prior studies. We maintain that CT or MRI in this setting still adds value in the evaluation of possible shunt obstruction. As a first principle, ventricular dilation in the setting of suspected shunt malfunction, regardless of historical patterns, is highly suggestive of shunt obstruction and warrants surgical exploration. A diagnostic dilemma exists, however, when a patient has signs and symptoms of shunt malfunction but no ventricular dilation on CT or MRI. Our findings suggest that if patients present with J Neurosurg: Pediatrics / Volume 14 / August 2014
CT and MRI patterns for shunt obstruction worrisome symptoms but no dilation of the ventricular system on imaging, and if patients have a historical pattern of ventricular dilation on their initial and subsequent shunt obstruction presentations, then the clinician may be more inclined to argue against shunt obstruction and to look for another etiology for the symptoms. This conclusion is born from the fact that the majority of our patients (64%) exhibited such a pattern and that in 92% of patient encounters following an initial presentation of ventricular dilation the patients had ventricular dilation on subsequent presentations. If, however, patients do not show radiographic evidence of consistently having dilated ventricles with occurrences of shunt obstruction, and if their ventricles are not dilated on a new presentation, the value CT or MRI in evaluating shunt obstruction is less clear. Ten percent of our patients consistently showed no evidence of ventricular dilation when they in fact had shunt obstruction, and 36% (15 of 42) of patients showed no ventricular dilation in at least one episode of shunt obstruction. In approximately 21% (25 of 117) of instances of shunt obstructions in our series, there were no changes in ventricular size or morphology on radiographic imaging prior to surgery. Thus, the absence of ventricular dilation in the nondilating or inconsistently dilatory groups is nondiagnostic. Therefore, in the evaluation of new shunt obstruction, the absence of ventricular dilation on CT or MRI in patients with ventricles that do not historically dilate, or inconsistently dilate, should not be used as the only or definitive diagnostic determinant when evaluating shunt function. In a study by Iskandar et al.5 that reported low sensitivity of CT scanning in detecting shunt malfunction, there was no mention of shunt malfunction in the radiologist’s interpretation of CT scans of 24 of 100 children requiring shunt revision. Most of these children had slit-ventricle syndrome or shunt overdrainage. A study by Naftel et al. examined the accuracy of parental diagnoses of shunt failure but also reviewed the test characteristics of a head CT scan in the evaluation of shunt failure.9 This study, which demonstrated a CT sensitivity of 79% and specificity of 86%, as well as the study by Iskandar et al., did not explicitly examine ventricular dilation and did not consider longitudinal imaging patterns observed during multiple instances of shunt malfunction in the same patient. Studies by Mater et al. and Lehnert et al. similarly examined the test characteristics of various imaging modalities in diagnosing shunt failure—defined as an operative shunt revision—but did not examine how consistent the results of such imaging modalities were in the same patient during multiple instances of shunt obstruction.7,8 To the best of our knowledge, the present study is the first to examine the reliability of antecedent cranial imaging patterns during episodes of shunt obstruction as useful clinical information in the evaluation of shunt malfunctions. Limitations
There are several limitations to this study. Most importantly, it is a retrospective chart review and is subject to all of the shortcoming associated with this type of analysis, including documentation omissions, missed relevant cases, and selection bias. The appropriateness of
J Neurosurg: Pediatrics / Volume 14 / August 2014
the evaluation for suspected shunt obstruction and the necessity of obtaining the CT or MRI studies are at the sole discretion of the physician. Differences in imaging technique also made comparing ventricular volume and size between CT and MRI studies difficult, perhaps indicating a change in serial data when in fact there was none. Also, we definitively identified shunt obstruction based on the surgical findings in the operative reports. However, there is no standard definition for shunt obstruction. A shunt obstruction could represent an occluded shunt system or simply a poorly functioning shunt system. We relied on the possibly biased interpretations of intraoperative findings by the operating surgeon to classify shunt status. One could argue that, by only including patients who underwent surgery for shunt obstruction, we have failed to include patients who presented with shunt obstruction but did not receive surgery. It seems a fair presumption, however, that patients with missed shunt obstructions will eventually visit, either because of clinical deterioration or because they re-present, to the emergency center. Our purpose was to characterize ventricular morphology during instances of confirmed shunt obstruction—including only surgical cases, in which obstruction was confirmed intraoperatively. Additionally, our definition of ventricular dilation was broad. We relied on independent radiology reports to determine, in a dichotomous manner, if dilation of the ventricular system was present or absent. However, the degree of ventricular dilation was not measured or quantified. One particular criticism of this study may be our reliance on the impression of radiologists rather than pediatric neurosurgeons. First, one may argue that experienced pediatric neurosurgeons are as or more proficient at assessing ventricular changes than radiologists. Another criticism of our study may be that radiologists were not comparing the index images to previous “well” images and were oblivious to the immediate and historical clinical context that prompted cranial imaging. It is in fact the standard practice of neuroradiologists at our institution to compare new images to antecedent scans, including baseline images (“well” scans) and prior “failure” scans (MRI or CT scans that show evidence of shunt failure) taken at times of confirmed shunt obstruction. The advent of electronic medical records at our institution and the resultant ease of access to clinical information have made this practice possible. Radiologists’ readings routinely include a paragraph in which “comparison studies” are detailed, revealing the radiologists’ attention to the longitudinal view of ventricular changes in a given patient. Our research team ensured this attention to clinical context during our data collection. We hold that independent review by a pediatric neuroradiologist is a strength of our study. These physicians are board certified in radiology and neuroradiology. They practice at a high-volume, free-standing children’s hospital and have extensive experience in reading and interpreting cranial imaging in the setting of shunted hydrocephalus. Unfortunately, neurosurgeons, while indeed very adept at reviewing cranial imaging, are not board certified in radiology or neuroradiology. The use of radiologists eliminates 165
J. N. Sellin et al. evaluation bias. If we were to include a neurosurgery versus radiology assessment, we would then need to address the issue of interobserver variability and reliability, which, to the best of our knowledge, has not been established. To do so would be beyond the scope of our study but is, indeed, a potential topic for a future study. Finally, our study analyzed a subgroup of children with a history of multiple shunt obstructions. We arbitrarily chose to analyze the age of 3 years or older. We surmised that at 3 years of age each patient’s fontanelle would be closed and their sutures fused such that there would be no peculiarities to alter the intracranial CSF dynamics or ventricular compliance in very young children. It would have been particularly interesting to analyze the relationship between total time with a CSF shunt and presentation with nondilation with shunt failure. It is often thought that early and chronic CSF shunting leads to changes in ventricular compliance that resist fulminant ventricular dilation at times of new shunt obstruction. Unfortunately, this information was either inconsistently available or entirely unavailable for our patient population, given the number of outside referrals and only the recent institution of our electronic medical record system. As electronic health record systems evolve, these data may be more readily available for further study.
Conclusions
Routine use of CT, quick MRI, and shunt series radiography is common in the evaluation of CSF shunt malfunction. To the best of our knowledge, no other studies exist that evaluate the efficacy of using radiographic evaluations of ventricular morphology obtained during workup for prior shunt obstructions in the workup of new or suspected shunt malfunction in the symptomatic pediatric patient. Our study indicates that CT or MRI evaluation of ventricular morphology remains valuable, as does clinical examination. In particular, if there is evidence of ventricular dilation at the first presentation of shunt obstruction, then the clinician may suspect each subsequent instance of shunt malfunction to present with CT evidence of ventricular dilation. Furthermore, if a patient has a demonstrated of pattern of ventricular dilation with shunt obstruction, the absence of ventricular dilation during a new presentation argues against shunt obstruction. Suspicion, however, should not lead to conclusions based on radiographic parameters alone. CT or MRI data play an important role in the safe and efficacious evaluation of suspected shunt obstruction, but they do not obviate the need for clinical examination and judgment. Future directions may involve prospective collection of ventricular morphology data, analysis of the relationship between ventricular morphology during obstruction and time shunted, application of clinical decision rules based on radiographic findings, and volumetric calculation to estimate ventricular volumes in children with possible shunt obstruction.
166
Disclosure The authors report no conflict of interest concerning the materials 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: Jea. Acquisition of data: Jea. Analysis and interpretation of data: Jea. Drafting the article: Jea, Sellin. 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: Jea. Statistical analysis: Jea. Study supervision: Luerssen. References 1. Blumstein H, Schardt S: Utility of radiography in suspected ventricular shunt malfunction. J Emerg Med 36:50–54, 2009 2. Drake JM, Kestle JR, Milner R, Cinalli G, Boop F, Piatt J Jr, et al: Randomized trial of cerebrospinal fluid shunt valve design in pediatric hydrocephalus. Neurosurgery 43:294–305, 1998 3. Engel M, Carmel PW, Chutorian AM: Increased intraventricular pressure without ventriculomegaly in children with shunts: “normal volume” hydrocephalus. Neurosurgery 5:549–552, 1979 4. Garton HJ, Kestle JR, Drake JM: Predicting shunt failure on the basis of clinical symptoms and signs in children. J Neurosurg 94:202–210, 2001 5. Iskandar BJ, McLaughlin C, Mapstone TB, Grabb PA, Oakes WJ: Pitfalls in the diagnosis of ventricular shunt dysfunction: radiology reports and ventricular size. Pediatrics 101:1031– 1036, 1998 6. Kim TY, Stewart G, Voth M, Moynihan JA, Brown L: Signs and symptoms of cerebrospinal fluid shunt malfunction in the pediatric emergency department. Pediatr Emerg Care 22: 28–34, 2006 7. Lehnert BE, Rahbar H, Relyea-Chew A, Lewis DH, Richardson ML, Fink JR: Detection of ventricular shunt malfunction in the ED: relative utility of radiography, CT, and nuclear imaging. Emerg Radiol 18:299–305, 2011 8. Mater A, Shroff M, Al-Farsi S, Drake J, Goldman RD: Test characteristics of neuroimaging in the emergency department evaluation of children for cerebrospinal fluid shunt malfunction. CJEM 10:131–135, 2008 9. Naftel RP, Tubergen E, Shannon CN, Gran KA, Vance EH, Oakes WJ, et al: Parental recognition of shunt failure: a prospective single-institution study. Clinical article. J Neurosurg Pediatr 9:363–371, 2012 10. Naradzay JF, Browne BJ, Rolnick MA, Doherty RJ: Cerebral ventricular shunts. J Emerg Med 17:311–322, 1999 11. Sivaganesan A, Krishnamurthy R, Sahni D, Viswanathan C: Neuroimaging of ventriculoperitoneal shunt complications in children. Pediatr Radiol 42:1029–1046, 2012
Manuscript submitted August 29, 2013. Accepted April 28, 2014. Please include this information when citing this paper: published online May 23, 2014; DOI: 10.3171/2014.4.PEDS13451. Address correspondence to: Andrew Jea, M.D., Division of Pediatric Neurosurgery, Texas Children’s Hospital, 6621 Fannin St., CCC 1230.01, 12th Floor, Houston, TX 77030. email: ahjea@ texaschildrens.org.
J Neurosurg: Pediatrics / Volume 14 / August 2014
