Chronic Pleural Effusion in Ventriculoperitoneal Shunt due to Diaphragmatic CSF Fistula: Report of a Case Treated by Endoscopic Choroid Plexus Coagulation and Literature Review

Implantation of ventriculoperitoneal (VP) shunts is the standard therapy of hydrocephalus in case an endoscopic procedure cannot avoid a VP shunt. Infection and mechanical shunt failures are the most frequent complications of cerebrospinal fluid (CSF) shunt therapy [1]. Thoracic complications like pleural effusion or intrathoracic migration of the distal end of the VP shunt have been rarely reported [2, 3]. The diaphragm is normally a watertight physical barrier between the peritoneal and pleural cavities. Foramen of Morgagni (sternocostal triangle) and Foramen of Bochdalek (lumbocostal triangle) are embryonic defects in the fusion of components, which can cause a congenital diaphragmatic hernia [4], pneumothorax as complication of a laparoscopy [5], acute pleural effusion as complication of peritoneal dialysis [6], or a transdiaphragmatic migration of VP shunt [7‒9]. In case of an acquired pleural effusion due to CSF fistula without an intrathoracic shunt catheter migration, it is believed that diaphragmatic defects or diaphragmatic perforation due to inflammatory processes play a role [9, 10]. Thoracentesis with chest tube placement under ultrasound guidance is an easy and safe way to acutely treat an effusion-related life-threatening respiratory decompensation [11], however, might not solve the problem of diaphragmatic fistula. We report on an unusual acquired diaphragmatic CSF fistula in coincidence of an acquired CSF hypersecretion in the presence of a VP shunt, happening in a child with severe syndromal neurological impairment with an additional postinfectious hydrocephalus after devastating influenza A sepsis and meningoencephalitis 3 years earlier.

The patient was a 4-year-old girl with a severe global developmental disorder based on a neurogenetic disease with complex unbalanced chromosomal aberration involving chromosomes 3, 9, 14, and X. The complex clinical picture includes laryngo-/tracheomalacia, hiatal hernia, gastroesophageal reflux, and neurogenic gulp disorder. At the age of 10 months, the girl had a septic shock with cardiopulmonary arrest in the context of severe pneumonia with influenza A and meningoencephalitis. The girl went into severe acute respiratory distress syndrome, requiring extracorporeal oxygenation and underwent several resuscitations. Over some months, she developed global brain atrophy plus a pressure active hydrocephalus requiring placement of a VP shunt 3 months after the influence infection. Due to a neurogenic swallowing disorder recurrent aspiration occurred, demanding a fundoplication and percutaneous endoscopic jejunostomy 1 month later. Afterward, she was stable over 3 years in a severely reduced neurological status.

At the age of 4 years, she was hospitalized because of mycoplasma pneumonia with 3 weeks persistent clinical signs. C-reactive protein was increased (19.8 mg/dL) and calculated antibiotic therapy with Piperacillin/Tazobactam and Clarithromycin was started. Microbacterial and virological findings were negative in all body fluids. Routine echocardiography showed the known 10 mm secundum atrial septum defect (ASD) with the new finding of a left-right shunting and tricuspid regurgitation grade II. A large right-sided pleural effusion was seen. Because of progressive respiratory deterioration the pleural effusion was drained by a chest tube. The clear serous fluid showed laboratory criteria of a transudate (pleural total protein 0.9 g/L, serum total protein 5.9 g/dL, ratio pleural/serum protein 0.15, pleural lactate dehydrogenase 167 U/L) and was interpreted as a complication of the pneumonia. The continuing high drainage rates up to 1,400 mL in 24 h, however, could not be explained by a pneumonia effusion and a differential diagnostic work-up trying to differentiate transudate from exsudate was not successful. Radiologically, the VP shunt was confirmed in a regular intraperitoneal position, however was partially traveling through the subphrenic suprahepatic region, with the shunt tip however in the mid abdomen. Beta trace protein, specific to CSF, could be detected in the pleural drain fluid at a concentration of 2.7 mg/L, CSF concentrations are >6 mg/L. Due to the low concentration of beta trace protein clearly below normal CSF concentration, CSF as the source of pleural secretion was considered very unlikely. To rule out any participation of the shunt system, a replacement of the shunt valve (proGAV^® 5 cmH₂O, gravitational unit: 25 cmH₂O) and optimized placement of abdominal catheter away from the right upper abdominal quadrant resulted in no improvement of high pleural drainage fluid output. Finally, the rare occurrence of an acquired abdomino-thoracic CSF fistula was considered despite the low β-trace protein concentration. As described in literature, an abdomino-thoracic CSF fistula was visualized by single photon emission computerized tomography (SPECT), in our setting combined with low-dose computer tomography for 3D visualization. The examination demonstrated the transition of tracer injected into the shunt system from the intraperitoneal to the right pleural space in three-dimensional imaging and thus confirmed an abdomino-thoracic fistula anterior to the esophagus (shown in Fig. 1).

The externalization of the VP shunt (with removal of valve and abdominal catheter) resulted in complete cessation of the right pleural fluid discharge and the chest tube was removed without any recurrence of pleural effusion. The option of a ventriculoatrial (VA) shunt to permanently solve the dilemma could not be chosen because the hemodynamically relevant ASD and the barely balanced cardiopulmonary system would not tolerate an additional intravascular volume load of up to 1,000 mL of CSF/24 h.

The hydrocephalus and CSF hypersecretion problem were solved with a bilateral endoscopic total choroid plexus destruction with laser from the Foramen of Monroi to the tip of the temporal horn using a bi-occipital burr hole approach. A rigid Little Lotta Pediatric Endoscope (Storz, Tuttlingen, Germany) and a Thulium Laser with 15 W power applied via 0.5 mm laser fiber (Lisa Laser, Katlenburg, Germany) were used for the procedure and the choroid plexus was coagulated and shrunk as radical as possible through both lateral ventricles. The external ventricular drain (EVD) was kept which showed that the CSF production rate came down to 0 within a few days after surgery and was consequently removed thereafter. The CT scan after choroid plexus coagulation (CPC) showed a marked opening of the previously compressed external CSF spaces and slight decrease of ventricular width (see Fig. 2). 1 Year after CPC the ASD was successfully closed using an Amplatzer^® Septal Occluder.

In the follow-up of 2.5 years, there was a regular course of head circumference and no neurological symptoms suspicious of active hydrocephalus. The child is shunt free and neurologically stable on her low level of cognitive and motor function at present. Therefore, no further imaging was performed until now.

As shown in big prospective trials the complication rate 1 year after VP shunt insertion is 38%, after 2 years 48%, and after 3 years 54%. Shunt obstruction was most frequently seen in the first 2 years with 75% of all complication, followed by infection with 16% and overdrainage with 7% [12]. Pleural effusion as in our patient is only very rarely reported [2, 3]. There are pleural effusions with and without migration of VP shunt in pleural cavity [9]. In the review of Ulus et al. [13], a ratio of patients with and without migration was described as 61% (n = 22) to 39% (n = 14). We have reviewed the literature and collected 18 more cases with this rare condition of pleural effusion in connection with a non-migrated VP shunt (see Table 1). Only 4 cases occurred in adults. Pleural effusions without migration are mostly a problem of childhood (14 of 19 cases, 74%, Table 1). Averaging all 19 cases, a pleural effusion occurred with a mean delay of 30 ± 73 months to insertion of VP shunt. 13 cases happened in the 1st year after insertion (63%) and 4 cases in 2nd year (21%).

In case of a correct position of the distal part of the VP shunt in the abdominal cavity, the true cause of pleural effusion can easily be overlooked and is not self-evident as this case shows. Bronchopulmonary infections present a difficulty in interpreting the pleural effusion and delay diagnosis and thus therapy as in our case. In our literature search, tachypnea, dyspnea, or tachydyspnea are the most common presenting symptoms (14 of 19 cases, 74%, Table 1). One case report describes symptoms of pneumonia but did not report a delay in therapy [13].

Congenital or acquired diaphragm defects, negative intrathoracic pressure, positive intra-abdominal pressure, and peritoneal reduced absorptive capacity of CSF are described as enabling factors for pleural effusion [9]. 5 of 19 cases have used SPECT detecting abdomino-thoracic fistula [13‒17]. This case report is the first using SPECT combined with low-dose CT. The exact location of the tracer transfer into the pleural space was nevertheless difficult to visualize but is interpreted to be in the area of the fundoplication ventral of the esophagus and not in the area of a suspected foramen Morgagni. In 2 further cases, fundoplication was as well reported in medical history [17, 18]. Therefore previous manipulation near to diaphragm could be a further risk factor for developing a VP shunt associated pleural effusion. Based on a gap in the diaphragm, which acts as a valve, large amounts of fluid may be suctioned to the pleural space that cannot return to the abdominal cavity. The one-way flow is probably due to a pressure gradient resulting from negative intrathoracic and positive abdominal pressure during the inspiration [17]. Due to progressive respiratory deterioration, 18 patients (95%) received thoracentesis.

Treatment in all 19 cases is shown in Table 1. One patient did not require any further treatment because the patient was clinically stable after removal of the chest tube and no renewed pleural effusion could be detected [19]. 11 of 19 patients (58%) received a change to a VA shunt [10, 13‒16, 20‒24] as definite solution. In 6/11 cases, the switch to VA shunt was performed directly, in 1/11 after externalization of the VP shunt via EVD. 2/11 cases had an EVD, then abdominal surgery with either new placement of distal catheter or closure of the diaphragmatic hole that was unsuccessful, and a secondary switch to a VA shunt was required. In 2 cases, the first unsuccessful surgery was new placement of distal catheter followed by switch to VA shunt. 3/19 patients (16%), however, were successfully treated with new placement of the distal VP shunt catheter. One patient (5%) was successfully treated with a ventriculostomy [25], 1 patient (5%) with the easy maneuver of increasing the valve pressure [26] to decrease CSF drainage rate, and 2 patients (11%), including the presented one, with CPC. The other case, who was treated with CPC, had an unsuccessful attempt of pleurodesis as first surgical intervention to treat the fistula [17].

This case, reported by Chuen-im et al. [17], did also present with an unusually high CFS production rate of approximately 800 mL/day corresponding of our case. Choroid plexus epithelium generates approximately 80% of CSF [27] and is among the most efficient secretory epithelia in the human body with a maximum production rate of 0.4 mL/min per gram of tissue rivalling the secretion rate in proximal tubule of kidney [28]. The total CFS volume of approximately 150 mL in adults is replaced 3–4 times per day corresponding with a maximum CFS production rate of 500–600 mL per day [28]. However, there is a large variability (0.11–0.73 mL/min) in CSF production rate in healthy adults [29]. The methods of measurement of CSF production rate are based on calculation by intracranial pressure measurement [29] or two-dimensional cine phase-contrast MR imaging [30] and are not very precise, especially in healthy children almost no systematic investigations exist. In a cohort of 100 children with hydrocephalus, the hourly CSF output measured through an external ventricular drainage fluctuates between 0.001 and 0.44 mL/min [31]. Chronic large amounts of CSF associated with hydrocephalus are compatible with choroid plexus hyperplasia or nonobstructive tumors of the choroid plexus, such as a choroid plexus papilloma (CPP) [32], which could not be seen in our patient. Idiopathic intracranial hypertension [33], infectious hydrocephalus [34], and intraventricular hemorrhage-associated hydrocephalus [35] are also named as reason for an increased CSF production rate. Our patient had history of an infectious hydrocephalus years ago, but it is unlikely to consider this event in the past to be the cause of hypersecretion, after the patient had no evidence of high-volume CSF shunting rates, like ascites, in the 2 years prior to the occurrence of the fistula. As described for patient with choroid plexus hyperplasia or CPP, showing a significant decrease of CSF production after endoscopic CPC [28], our patient did benefit of a total bilateral CPC. Because of the high CSF production rate, we wanted to be as radical in lateral ventricle CPC as possible and treat all choroid plexus in the lateral ventricles including all plexus in the temporal horns. Since we did lack a flexible endoscope, we chose the unusual approach with a bilateral occipital burr hole that allowed to reach with a rigid scope bilaterally the entire choroid plexus from the tip of the temporal horn to the foramen of Monroi. After CPC, the production rate of the external CSF drainage was very low and ceased after 2 days. After removal of the EVD, there was no further need of a shunt corresponding with some case reports in patients with CPP [28, 36‒38]. Chuen-im et al. [17] assumed that the increased CSF production rate presented presumably throughout the whole life, without having any valid findings to support this claim. In interpreting our patient, it is be difficult to separate a congenital and a postinfectious component for hydrocephalus and we have no explanation why the girl developed hypersecretion. We cannot even rule out, that the unusually high CSF output rates through the chest tube were secondary to the loss of the usual intra-abdominal counter pressure, that ranges between 4 and 5 mm Hg in supine [39] position that usually exists in VP shunts and decreases the pressure differential that allows shunt flow. In view of the still insufficient understanding of the genetic components of developmental hydrocephalus [28], genetic involvement appears to be possible in our patient with a pronounced chromosomal disorder.

Abdomino-thoracic CSF fistulas in patients with VP shunts are a rare occurrence but have been described in all together 19 case reports, 15 of which were in children. The existence of such a fistula needs to be ruled out in children with VP shunts if a pleural effusion is detected and no other obvious cause can be identified. A β-trace protein concentration in pleural fluid below the usual CSF concentration of this protein does not exclude CSF as origin of the pleural effusion. Single photon emission tomography after injection of the traces inside the shunt system combined with computer tomography can reveal the transition of CSF from intraperitoneal to the pleural space and with a better 3-dimensional interpretation than in SPECT alone might identify the presumable location of the transition point in the diaphragm. In addition to diaphragmatic defects, negative intrathoracic pressure, positive intra-abdominal pressure and peritoneally reduced CSF uptake capacity, increased CSF production and manipulation near to diaphragm are enabling factors for a pleural effusion. Radical CPC in both lateral ventricles can be an alternative therapy option, also in hyper-secretive choroid plexus, if a switch to a VA shunt is either not possible or not wanted.

Written informed consent was obtained from the parents of the patient for publication of the details of their medical case and any accompanying images. This study protocol was reviewed and approved by Ethics Committee at the Medical Faculty of Eberhard Karls University and at the University Hospital Tübingen, approval number 142/2022A.

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The authors received no specific funding for this work.

S. Schmid and M. Schumann initiated and designed the article. M. Hofbeck, A. Bevot, M. Reimold, M. Kumpf, J. Michel, and F. Neunhoeffer provided important aspects for analysis and interpretation and critically revised the final version.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.