Role of Radiological Parameters in Predicting Overall Shunt Outcome After Ventriculoperitoneal Shunt Insertion in Pediatric Patients With Obstructive Hydrocephalus

Devi Prasad Patra, MD, MCh; Shyamal C. Bir, MD, PhD; Tanmoy K. Maiti, MD, MCh; Piyush Kalakoti, MD; Hugo Cuellar, MD, PhD; Bharat Guthikonda, MD; Hai Sun, MD, PhD; Christina Notarianni, MD; Anil Nanda, MD, MPH


Neurosurg Focus. 2016;41(5):e4 

In This Article


Radiological imaging has traditionally been considered the most valuable tool for documenting shunt malfunction. The role of CT and/or MRI in this setting is invaluable. A dilated ventricle, whether the same or increased as compared with its appearance on baseline imaging, is usually suggestive of shunt malfunction; however, the reverse does not hold true in all cases. So a nondilated ventricle does not rule out shunt malfunction and mandates further evaluation and clinical correlation with features of raised intracranial pressure (ICP). Other radiological features that can aid diagnosis include PVL and effacement of sulci. There has been increasing concern regarding the frequent use of cranial CT to monitor shunt function given the risks of high-dose radiation to a developing child. Rapid cranial MRI sequences have been developed to reduce the image acquisition time, thus avoiding motion artifacts and the need for sedation.[6] A few studies have compared the efficacy of these 2 imaging modalities in diagnosing shunt malfunction and have shown that both have nearly equal specificity (89%–93% in rapid cranial MRI vs 76%–93% in CT); however, CT does seem to have better sensitivity than MRI (51%–59% in rapid cranial MRI vs 53%–92% in CT).[1,2,22]

Multiple studies have demonstrated good correlation between ventricular volume and linear indices on CT scans, which include the Evans' index, FHI, FOHR, and the bicaudate index.[7,14,17] These studies collectively considered FOHR as the best method for measuring both absolute and relative ventricular volumes. The normal value of the FOHR has been measured as 0.37 (99% CI 0.36–0.38), which is independent of age.[14] In the present study we also measured OHI and FOIR because a substantial number of patients had preferential dilation of the occipital horn as compared with the frontal horn. Measuring these indices on preoperative scans gives an estimate of the preoperative ventricular volume and in this study was correlated with the overall shunt outcome. However, FOHR correlates best with relative ventricular volumes and so was measured on the postoperative images as well to estimate the degree of ventricular volume reduction.

The morphological pattern of ventricles in the radiological studies at the time of initial evaluation usually does not accurately predict the future circumstances of the patient, especially as related to the need for shunt revision. A few studies in the literature have addressed this specific entity and have come up with specific conclusions. Sellin et al.[19] observed that patients with hydrocephalus initially presenting with ventricular dilation subsequently presented with dilated ventricles in cases of shunt malfunction. Similarly, nondilation at the initial presentation made subsequent presentation with dilated ventricles less likely in cases of shunt obstruction. These authors also observed, though at a low significance level, that distal shunt failures are more often associated with ventricular dilation than proximal obstruction. In the present study ventricular dilation pattern was further categorized into bi- or triventriculomegaly and right-left symmetry. Though not statistically significant, the presence of biventriculomegaly and asymmetrical ventriculomegaly were associated with low shunt revision rates and increased overall shunt revision–free survival.

The pattern of ventricular size reduction has been shown to be different in patients undergoing ETV rather than shunt placement. St. George et al.[20] have demonstrated that ventricular volume decreases steadily up to 3–6 months after ETV and then stabilizes, whereas it decreases even after 6 months post–shunt placement. There has been much controversy over whether this change in ventricular volume has any correlation with clinical outcome. Multiple studies in the literature address this issue; however, most are limited to ETV series. Warf et al. compared the neurocognitive outcome in patients undergoing ETV or VP shunt placement and the reduction in ventricular volume as measured by FOHR and found no correlation between the two.[21] However, in a more recent study these authors suggested that a combination of brain and CSF volumes has a significant effect on overall neurocognitive outcome.[10] In 1998 Buxton et al.[4] described the results of ETV in 27 patients under the age of 1 year. Treatment failure occurred in 21 (78%) of the patients, and the authors found that the postoperative ventricular size does not correlate with treatment success or failure. In a similar study in 2000, Kim et al.[8] found that a reduction in ventricular volume does not independently predict clinical outcome. However, in a series of 29 patients undergoing ETV, Kulkarni et al.[9] showed a significant difference in the degree of ventricular reduction in the patients with clinical success (21 patients, mean FOHR reduction by 16%) compared with that in the patients with failure (8 patients, mean FOHR reduction by 7%; p = 0.03%). There was no significant association between periventricular edema and clinical success or failure after ETV. However, there was a significant positive correlation between clinical success and the presence of flow voids on sagittal T2-weighted MR images. After a systematic review of all of these studies in 2014, Nikas et al.[12] suggested that there is insufficient evidence to consider the degree of ventricular reduction as a measure of treatment effectiveness after both ETV and shunt placement. A similar finding was noted in the present study, with no correlation between preoperative radiological indices and a requirement for shunt revision; however, the presence of PVL, which indirectly suggests an acute hypertensive state,[11] had a lower revision rate. This suggests that ventricular pressure reduction is probably more important than volume reduction to achieve a successful shunt outcome.

The most striking and new finding in the current study was related to the shunt revision interval and shunt revision frequency. The reduction in the FOHR was interestingly high in patients who required early shunt revision. This paradoxical finding is probably related to the fact that CSF overdrainage leads to CSF stasis and a greater chance of mechanical shunt obstruction.[3] Most of the shunts that are routinely used are pressure regulated. A compliant ventricular wall causes rapid decompression of the ventricle, leading to subsequent flow reduction and shunt hardware blockage. However, a less compliant ventricle does not rapidly decrease in size after shunt placement, although a similar reduction in pressure is achieved. This causes less adhesion of the ventricular catheter to the wall and blockage. Note that at this stage this can only be considered a possible hypothesis that needs dynamic studies for further validation. Nevertheless, it possibly carries an important therapeutic implication in that keeping the pressure setting at higher levels when inserting programmable shunts may promote better shunt functioning by avoiding collapse of the ventricles. This further substantiates the fact that one should aim for a ventricular decompression just enough to alleviate the raised ICP rather than an appealing reduction in ventricular size. Another important finding was that greater occipital horn dilation is associated with a lower frequency of shunt revision as dictated by a high OHI and low FOIR in patients with a single shunt revision as compared with those requiring multiple shunt revisions. However, it is worth mentioning again that these indices do not predict the need for shunt revision itself. In a recent multicenter prospective study, Riva-Cambrin et al.[18] found that an age < 6 months is an independent risk factor for shunt failure and an important determinant of shunt survival. Similarly, we found that patients younger than 6 months had significantly fewer days of revision-free survival than their older peers. None of the indices or ventricular characteristics had a significant influence on revision-free survival.

This study has limitations inherent to its retrospective nature. The exclusion of 47 patients because of the unavailability of comparative radiological images may have had some influence on the final outcome. Moreover, the variability in the follow-up also seems to have affected overall revision-free survival. Similarly, the variability in the CT interval had a significant effect on postoperative ventricular indices because of the dynamic nature of CSF drainage through the shunt. These factors can be eliminated by a prospective study with strict clinical and radiological follow-up. In addition, an increase in the sample size may further validate the findings.