Neurosurgical Shunt Treatment of Paediatric Hydrocephalus: Epidemiology and Inuencing Factors on Revision Surgeries – a Single-centre Retrospective Analysis of 131 Patients

Paediatric hydrocephalus is a result of cerebrospinal uid circulatory dysfunction and has diverse pathogeneses. This study investigates the epidemiology of paediatric hydrocephalus and inuences of primary aetiology, type of implants, and of complications to develope new therapeutic approaches and strategies. Between 2013-2018 a retrospective analysis of 131 children, suffering from hydrocephalus, was conducted. Medical charts, operative reports and clinical follow-up visits were reviewed. Statistical analysis was performed using t-test/ANOVA and Kruksal-Wallis-test/ Mann-Whitney-U test. Most common pathogeneses of hydrocephalus among our patients were meningomyelocele associated and posthaemorrhagic. The majority received a programmable differential pressure valve (PP valve, 77.8%) or a xed differential pressure valve with gravitational unit (FPG valve, 14.8%). Among 333 shunt associated surgeries 66% were performed because of mechanical shunt dysfunction (61%), infection (12%), or due to other reasons (27%). The median rate of revisions for each patient within one year was 0.15 (IQR 25-75 :0.00-0.68) and was inuenced by aetiology (p=0.045) and used valves (p=0.029). The highest rates were seen in patients with posthaemorrhagic hydrocephalus and in those with FPG valve, the lowest in meningomyelocele associated hydrocephalus and PP valve. The occurrence of mechanical dysfunctions showed a relation to FPG valve (p=0.014). Furthermore, the median time interval between initial shunt surgery and infections was shorter than in mechanical dysfunctions (p=0.033). Conclusion: We could state several inuence factors on revision surgeries in paediatric shunt patients. With the assessment of patients’ risk proles, we can classify paediatric shunt patients and could avoid unnecessary examinations or invasive procedures. Furthermore, we could prevent revisions surgeries, if we choose shunt material in accordance with the patients’ associated shunt complications.


Introduction
The prevalence of hydrocephalus is reported at 88 cases in 100.000 children (age ≤18) [7,14]. The pathogenesis is a cerebrospinal uid (CSF) circulatory dysfunction. The mismatch of production and absorption of CSF leads to neurological symptoms. [6].
The aetiology of this dysfunction is diverse. Many cases occur immediately after birth due to dysfunction of CSF absorption (malresorptive hydrocephalus). Congenital hydrocephalus is often associated with the syndromic incidence of a meningomyelocele. Another cause of malresorptive hydrocephalus is related to perinatal intraventricular hemorrhage. In this pathology, absorption of CSF is dysfunctional due to high protein content or due to destruction of arachnoid granulations. Occlusive hydrocephalus, as in aquaeduct stenosis or tumour blockage, is rare [18].
The most common treatment for hydrocephalus is ventriculoperitoneal (VP) shunting, followed by ventriculoatrial and ventriculopleural shunting. [8]. The valves used mostly in our study were programmable differential pressure valves (PPV) or xed differential pressure valves with gravitational unit (FPgV). The underlying mechanism of PPV is a ball-spring-mechanism. When CSF pressure reaches the opening pressure, the ball moves and opens the valve, allowing CSF to drain. The spring is attached to a rotating spiral, which controls the resistance of the spring towards the ball and can be adjusted by non-invasive methods using a codi ed magnetic eld (opening pressure:30 mmH 2 O-200 mmH 2 O) [21].
The major difference of FPG and PPV is the combination of a ball-in-cone mechanism and a gravitational unit. The ball-cone unit controls the CSF drainage analogous to the ball-spring-mechanism in a horizontal position and, additionally, a second unit controls the CSF drainage in a vertical position to prevent overdrainage when changing posture. The FPgV has xed pressure levels [10]. Besides different mechanisms of the valve, companies set different priorities. Some focus on the adjustability of the CSF drain level or on the development of minor valves for infants [21]. Others focus on valves with high ow velocity to counteract obstruction of the valve or to achieve steadiness to environmental in uences and medical examinations, e.g. magnetic resonance imaging [10].
Although the insertion of a VP-shunt system is one of the most common neurosurgical procedures, dysfunction or infection of the shunt system are frequent. Every second patient undergoes at least one additional surgery within the rst three years after initial implantation because of shunt failure [25]. 3-15 % of paediatric patients suffer from an infection of the shunt system. If this occurs, the whole shunt system has to be removed. After antibiotic treatment a new system has to be implanted. Another frequent cause of shunt failure occurs due to occlusion of the valve or a dislocation of the ventricular/peritoneal catheter. The malfunctioning part will be surgically removed and exchanged.
Overall, an indwelling shunt system implies a high risk for surgeries and longer in-patient stays causing a limitation of quality of life [17]. Presently, there are no established guidelines as to which shunt system should be used in the presence of which pathology of hydrocephalus, nor as to the speci c requirements of shunt systems over the different pathologies.
Therefore, the aim of this study was to investigate the in uence of pathogenesis and shunt materials on incidence and type of complications, in order to establish a better understanding of and better treatment for paediatric hydrocephalus (PH) and, additionally, to reduce complications of the shunt treatment in PH patients.

Materials And Methods
Between 2013 und 2018 a retrospective analysis of 131 children (64 female, 67 male) was conducted. All participants underwent shunt surgery in our institution and were 17 years old or younger at the time of primary shunt surgery. This study was approved by the institutional research ethics board (AZ165/14) and is based on medical charts, operative reports and clinical follow-up documentations. The determined parameters were age at shunt implantation, diagnosis, type of shunt, used catheters and valves, revision surgeries, type of revision, and reason for revision and were assembled anonymous. The database was created with Excel (Microsoft®, USA) and the statistical analysis was conducted with SPSS Statistics 24 (IBM corp, USA). Normal distribution was tested by Kolmogorov-Smirnov test. Differences of two or more cohorts were tested using t-test and ANOVA as parametric tests. As non-parametric tests, Kruskal-Wallis-Test and Mann-Whitney-U-test were used. Chi-square test was used to evaluate the distribution within a cohort or between two patient cohorts. A p-value <0.05 was considered statistically signi cant.

Epidemiology of patients
The most common aetiology in our cohort was the malresorptive hydrocephalus. MMCaH occurred in 35.1%, PPH in 24.4%, and congenital non-meningomyelocele-associated in 9.2% of patients. Hydrocephalus occlusus caused by stenosis was found in 8.4% of patients and by tumour in 7.6% of patients. The remaining patients (15.3%) showed rare underlying pathomechanisms.
Ventriculoperitoneal shunting was performed in 85%. The most used valve was PPV (77.8%), followed by FPgV (14.8%). The remaining 7.4% of patients received an uncommon valve or did not have one.

Revision surgeries
In our study 333 hydrocephalus associated surgeries were performed. 66% (n=220) were revision surgeries, 34% were initial shunt surgeries. The median follow-up time per patient was 41.5 months (IQR 25-75 : 21.0-57.5 months). In the cohort of 112 patients, whose follow-up was completely at our institution and was fully documented, 62 patients (55%) received revision surgery. In the cohort of 112 patients, whose follow-up was complete, 62 patients (55%) received revision surgery. 50% of all patients were revised within the rst year.
To evaluate the characteristics of revision surgery, the cause of revision was analyzed in detail. 61% of surgeries were performed because of mechanical dysfunction. In 63% of cases the valve was changed, making it the most revised part of the shunt system. Ventricular and peritoneal catheters were changed in 56% and 36%, respectively. 12% were performed because of infection of the shunt system, requiring a complete removal. Staphylococcus epidermidis was the most common pathogen (42.3%), followed by enterococcus faecialis (7.7%). In 34.6% no pathogen could be found. Other proven pathogens, each occurring once, were streptococcus pyogenes, escherichia coli, staphylococcus capitis and streptococcus pneumoniae.
Furthermore, the number of revisions within one year of indwelling shunt system (revisions/p/y) and the time interval to rst revision were inspected. In a follow-up of at least one year, the median number of revisions/p/y was 0.15 (IQR  :0-0.68). The lowest rate of revisions/p/y was 0, the highest 3. Time range from initial shunt surgery to rst revision surgery varied from 0 to 76 months. Thereby, an accumulation of low time interval to initial surgery was seen ( gure 2);(p<0.001***). The median interval was 1 month (IQR 25-75 :0-2 months), meaning 38% of revision surgeries were performed one month after initial surgery.
In uences on rate of revisions As a major in uencing factor of the number of revisions/p/y, the aetiologies of hydrocephalus were identi ed (p=0.045*). While the median rate of revisions/p/y in PPH was 0.52 (IQR   In uences on cause of revision Aetiology of hydrocephalus did not only affect revision rate but had signi cant in uence on the reason behind the revision (p=0,040*). Patients with PPH had a signi cantly higher number of revisions caused by mechanical dysfunction than patients with MMCaH ( gure 5). No signi cant difference could be found among infections that occurred with different aetiologies.
In uences on time of revision Age at initial surgery was identi ed as an in uencing factor on the time interval between initial and revision surgery (p<0,001)( gure 8). Patients, who received their shunt system within the rst month of life had the shortest time interval to revision surgery (median 0 months (IQR 25-75 :0-0.75 months)). Surgery after the third month entended the interval to 9 months (IQR 25-75 :4-19 months).

Discussion
The paediatric shunt patient The pathogenesis of hydrocephalus is diverse and could be classi ed in different groups [4,18,22]. In our cohort the MMCaH was most frequent, followed by PPH. Overall, our cohort represents the most common aetiologies of hydrocephalus in accordance with the current literature [11,14]. Furthermore, the most frequent form of treatment in our cohort was VP-shunting [20].
A parameter which is not speci cally described in PH patients is the distribution of age. In this study 48.5% of patients were newborns at the time of initial shunt implantation. Two thirds were in their rst two months of life. In addition, patients with MMCaH or PPH were signi cantly younger than patients with occlusive hydrocephalus. Even though age distribution of hydrocephalus was analyzed in metaanalysis studies. The only statement was, that patients younger than 18 years had a signi cantly higher incidence of hydrocephalus than middle-aged patients [14,26].
Both ndings could be quite helpful to acquire a better understanding of the pathogenesis of PH.

Rate of revisions
The treatment of PH is associated with a high number of complications despite the advances in technology and surgical technique [2]. In our cohort, an overall of 55% of patients were revised, 50% within the rst year after receiving a shunt system. In the current literature, the rate of revisions within one year after surgery ranges from 25% [17] to 50% [11]. Reddy et al. (2014) showed a difference in paediatric and adult patients. In a median follow up of 6.5 years, 78.2% of paediatric patients underwent revision surgery, compared to 32.5% of adult patients [20]. In our study the high number of revisions is in accordance with the current literature and identi es paediatric patients, in contrast to adult patients, as high-risk patients.
We identi ed further factors as high risk for complications within the paediatric group. In our cohort, the younger the patients were at initial surgery, the more frequent reoperations had to be performed. In accordance with Gurbuz et al., who found an association between age at shunt implantation and the occurrence of infections, we concluded, the youngest shunt patients have the highest risk for revision surgeries [12].
In our cohort, 38% of revision surgeries were performed within the rst month after initial surgery. Alatas et al. and Dupepe et al. reported an accumulation of revision surgeries within one year after shunt implantation in patients with MMCaH [1,9]. Wang et al. concluded that patients with PPH had more revisions in the rst three years than in the time thereafter [28]. These results suggest that the rst month after shunt implantation is the most vulnerable period for the shunt system.
Aetiology was associated with the rate of revisions in our cohort. MMCaH occurred with low revision rates and the PPH with high revision rates. Reddy et al. analyzed, among other parameters, the aetiology as an in uencing factor and described a high rate of revisions in MMCaH and a comparatively low rate in PPH [20]. The difference to our cohort was once again the patients' age. Their cohort consisted of paediatric and adult patients, which implicates that aetiology has an impact on paediatric and adult patients, but in different ways. Al-Hakim et al. showed in paediatric patients cohorts a high shunt survival rate in patients with MMCaH compared to other aetiologies, which was in accordance with our paediatric patients [3]. Furthermore, they see regular follow-up examinations as the reason for the low rate of revisions in MMC.
In addition, MMC patients underwent surgical coverage of skin defect, which is accompanied by antibiotic therapy to avoid wound infections. This factor could further explain the low rate of revisions, shunt infections, and could not only help to calculate the risk for necessary surgical treatment for patients on the basis of pathogenesis of hydrocephalus in addition to age and time of indwelling shunt system, but also help to identify approaches for improvement of shunt treatment.
Kestle et al. provided a review about PH and emphasized the lack of current studies comparing different valves [15]. Therefore, in our study we compared the different shunt valves with regard to different mechanisms. We could show that the PPV was associated with a lower rate of revision and the FPgV with a higher rate of revision. Tervonen et al. analyzed the difference of revision rates in programmable and non-programmable valves but could not elaborate any signi cant difference [25]. Alavi et al. showed similar results without any signi cant difference in valves with and without gravitational units [2]. In summary, the differences we uncovered seem to be multifactorial. Because of its small size, the PPV was often used in patients with MMCaH which had the lowest rate of revisions in our cohort. Similarly, the valve implanted mostly in PPH was the FPgV, which had the highest revision rate.
For this reason, we evaluated the in uences of shunt valves as well as age and aetiology on occurrence of shunt infections and mechanical dysfunctions in detail.

Shunt infections
Revision surgeries caused by infection occurred in 12%. Malluci et al. detected infections in 12% within their paediatric cohort, Reddy et al. 21% [17,19]. Similar to our cohort, the bacterial spectrum consisted mostly of staphylococcus epidermidis and staphylococcus aureus [17]. In our patient cohort we could show a signi cant shorter time between initial surgery and the occurrence of shunt infections than the occurrence of a mechanical dysfunction. Overall time until revision was analyzed in several studies with speci c patient cohorts, but not the in uence on cause for revision, especially on the occurrence of infections [9,12]. The infantile immune system of young hydrocephalus patients and the bacterial spectrum indicate that shunt surgery itself is a risk factor and that there is need for improvement to avoid infections of the shunt system.
The chie y evaluated shunt part in this study the valve, does not seem to be the vulnerable part of the shunt for infections. Malucci et al. determined the catheter as the main origin for shunt infections and showed the superiority of the antibiotic catheters [17]. However, in our cohort the shunt valve seems to be more consequential for the occurrence of mechanical dysfunctions.

Mechanical dysfunctions
In our cohort, revision surgeries caused by mechanical dysfunction occurred in 61%. In contrast the current literature describes the majority of revisions by mechanical dysfunctions [13,16,17,19]. Malluci et al. mentioned, that revisions in 81% of patients were not caused by infections, but provided no further analysis [17]. The most revised part of shunt in our cohort was the valve (41%), followed by the proximal (36%) and distal (23%) catheter. Browd et al. described obstruction of proximal and distal catheters as the primary reason for shunt revision [5]. Other studies con rmed these ndings [13,24,27]. Hanak et al. described in ammatory reactions to the shunt catheters, which caused obstruction in a varying time interval after initial surgery. As other possible causes for mechanical dysfunction they described catheter length and the decreasing ability of the peritoneum to absorb CSF, e.g. after abdominal surgeries [13]. These outlined causes are not associated with young age. Therefore, distal catheter obstruction seems to be a problem in adult but not in paediatric patients. As a result, the predominantly revised part in our study cohort, the valve, appears to be the most fragile part of the shunt system for mechanical dysfunctions in paediatric patients and presents a starting point in the combat to avoid shunt revisions.
While the valve did not have any signi cant in uence on the occurrence of infections, we showed an association between the valve and the occurrence of mechanical dysfunctions. The FPgV was associated with signi cantly more mechanical dysfunctions than the PPV. Analogous to the higher revision rate in general, the FPgV was often used in patients with PPH, which was associated with a higher revision rate. If this was the only reason for the higher rate of mechanical dysfunctions, infections should have also been higher in FPgV, which did not occur. Therefore, the underlying reason seems to be the valve itself. An argument could be the technical details of the valves. The CSF must pass through two mechanisms in FPgVs. If the CSF has high protein levels, obstruction is more likely to occur and subsequently to require surgical treatment [10,13]. Additionally the vulnerability of valves could be increased by more sensitive components and the valve could be more vulnerable for minor traumas or a MRI examination [23].
The same mechanism, obstructions caused by high protein levels in CSF, seems to be responsible for the in uence of aetiology on the occurrence of mechanical dysfunctions. The highest rate of mechanical dysfunctions was seen in PPH. In contrast, MMCaH had a different rate of infection occurrence and mechanical obstructions, which leaned more heavily in favor of infections compared to PPH. Alatas et al. showed the same ndings in 31 patients with MCCaH, where 43% got revision surgery due to infection and 14% due to mechanical dysfunction [1]. These results provide a novel approach in the improvement of shunt treatment and a step forward in realizing individually adaptable shunt treatment.
A limitation of this study was its retrospective character. Another aspect was the limited number of patients. Even though a randomized trial could not be conducted, our study cohort could still be seen as a representative cohort of PH patients.

Conclusion
According to the assessment of the individual risk pro les of patients, the shunt material should be chosen individually. To prevent formidable shunt infection in patients with MMCaH, we could perform preventive antibiotic therapy even before surgery. We could help in preventing frequent obstruction of valves in PPH patients if we use a valve, which is associated with a lower rate of mechanical dysfunctions, e.g. a valve with only one unit. A regular follow-up examination plan should be provided to improve the timely recognition of shunt complications. Further prospective, multi-center studies are required to help in preventing revision surgeries.

Declarations
Funding: Authors state no funding involved.
Con ict of interest/Competing interests: Authors declared no con icts of interest.
Availability of data and material: All data was collected in connection with the treatment of the patients, independently of this study. The study data was collected retrospectively and were assembled anonymous.
Code availability: N/A Ethics approval: Approved by the institutional research ethics board (10/2019, AZ 165/14). The research related to human use complies with all the relevant national regulations, institutional policies and was performed in accordance with the tenets of the Helsinki Declaration, and has been approved by the authors' institutional review board or equivalent committee.
Consent to participate: This study does not contain any studies with human participants. The study data was collected retrospectively and were assembled anonymous.