A novel model of acquired hydrocephalus for evaluation of neurosurgical treatments

Four experimental groups were studied in juvenile pigs (Table 1): (1) Untreated Hydrocephalus–animals that developed ventriculomegaly following intracisternal kaolin injections (n = 9); (2) Shunt-treated Hydrocephalus–animals with ventriculomegaly that received ventriculoperitoneal (VP) shunts for diversion of CSF (n = 8); (3) Sham Controls–animals that received intracisternal saline injections (n = 6); and (4) Intact Controls–normal naïve animals, only used for pre-induction MRI (n = 4). All groups were age-matched. Outcomes included pre- and post-induction behavior, neurological status, cognitive testing, neuroimaging with T1- and T2-weighted MRI, ventricular volume quantification from T2-weighted MRI scans, and gross brain morphology including subjective assessments of ventricular catheter patency; e.g., inspection of the catheter lumen for tissue or debris. In addition to the Untreated Hydrocephalus cases described above, 31 kaolin-injected untreated pigs were evaluated with MRI to subjectively assess the success rate of kaolin induction.

All procedures were approved by the Washington University Institutional Animal Care and Use Committee and done in an American Association for Accreditation of Laboratory Animal Care (AAALAC)-accredited facility in compliance with the Guide for the Care and Use of Laboratory Animals and the Animal Welfare Act. Juvenile (∼30-day old) domestic pigs (Sus scrofa domesticus) were obtained from a university approved vendor (Oak Hill Genetics LLC, Ewing, IL) and maintained in standard pens with raised flooring, fed a standard pig chow (Purina Porcine Grower Diet 5084) with ad lib access to water. Pigs were allowed to acclimate to the facility and physical exams were performed by institutional veterinary staff to assure normal health and behavior. Our study corresponds to the USDA Class D category of animal use. All animals undergoing surgery received surgical pre-anesthetics and analgesics, intra-operative anesthetics, and post-operative medications as needed (Additional file 5: Table 1).

Pigs were sedated with a cocktail of telazol, ketamine, and xylazine (1.0 ml/50 lb intramuscular), intubated, and maintained under general anesthesia using 1–4% Isoflurane. The dorsal neck and head were shaved free of hair and the area was surgically prepped using povidone iodine and alcohol after the pig was positioned in a lateral decubitus position. Importantly, the head was temporarily flexed to about 90-degrees to open the atlanto-occipital interval as much as possible, with care taken to maintain a patent airway. Using sterile technique, a well-accepted method was employed to induce hydrocephalus [45, 51‐53]: the cisterna magna was tapped percutaneously with an 18-gauge spinal needle (Becton-Dickenson 405184, McKesson Medical-Surgical. Richmond, VA) inserted in the midline at a 45° angle to the skin of the dorsal neck. The cisterna magna was located by manually sensing the needle tip as it contacted the flexible atlanto-occipital membrane and dura; if the occipital bone was encountered during penetration, the needle was “walked” inferiorly without withdrawing until the atlanto-occipital membrane was encountered. Accurate placement was confirmed when CSF was seen in the hub of the needle when the needle stylet was withdrawn. CSF was allowed to drip from the needle hub to confirm placement within the cisterna magna and checked for blood contamination. An IV extension tubing (B. Braun, ET12SB) was connected to the spinal needle and 1.25 ml of sterile 25% kaolin was injected slowly (about 0.25 ml/min) into the cisterna magna. After a 1–2-min equilibration period to allow dispersion of the kaolin, the IV extension tubing was disconnected from the spinal needle, and continued accurate placement was confirmed by observing a small amount of CSF mixed with kaolin emerging from the needle hub. The needle was withdrawn and the animal allowed to recover from anesthesia using standard veterinary procedures. When the animal could stand, it was placed in a recovery pen for 24–48 h and subsequently returned to its home cage.

Following induction of hydrocephalus, standard VP shunts were placed because they are used almost exclusively to treat pediatric patients with hydrocephalus. We used CSF-Flow Control Shunt Kits (low-pressure valve/reservoir, model 9003D, Medtronic Neurosurgery, Goleta, CA, USA). Shunts were placed at 53–79 days of age, depending on the progression of hydrocephalus (primarily ventriculomegaly severity and neurological status) using similar neurosurgical procedures performed on human patients and feline models [50, 52]. Animals were anesthetized as described previously for the kaolin inductions. A 3–5 cm skin incision was made in the midline dorsal neck over the nuchal crest. The attaching occipital tissue was blunt-dissected off the occipital bone, with minimal use of bipolar cautery to release the muscle attachments. The muscle and skin were retracted to expose the occipital bone ~ 1 cm lateral to the midline on either side. A unilateral burr hole (~ 2.5 cm diameter) was made with a drill, the underlying dura mater was preserved, and any bone bleeding was stopped with bone wax. The location of this burr hole was determined from a pre-shunt MRI; this was usually centered 8.0–10.0 mm lateral to the midline and 10.0–12.0 mm inferior to the nuchal crest. Once the dura was exposed, attention was turned to the lateral thorax and abdomen for placement of the distal catheter. Two small (~ 1 cm) skin incisions were made; one over the left lateral rib cage for passing the distal catheter and inserting the valve with its reservoir and the other over the left abdomen ~ 1 cm caudal to the lowest rib for insertion of the peritoneal end of the distal catheter. The abdominal incision was dissected bluntly until the peritoneal layer was encountered. Two Crile hemostats were used to grasp the peritoneal layer and a small incision ~ 0.25 cm was made to expose the peritoneal cavity. A #1 Penfield instrument was inserted to confirm the peritoneal cavity had been accessed. A tunneler (Medtronic Neurosurgery, Goleta, CA, USA) was passed subcutaneously from the incision over the rib cage to the abdominal wound. A standard shunt catheter, distal slit valves removed, was then passed through the tunneler and inserted into the peritoneal cavity to a depth of about 20 cm. The proximal end of the subcutaneous catheter was attached to the distal port of a shunt valve and secured with 3–0 silk ties. A second portion of shunt tubing was passed subcutaneously from the occipital region to the rib cage incision using the same tunneling technique. The distal end of this catheter was attached to the proximal end of a reservoir with a low-pressure valve and secured with 3–0 silk ties. The valve was then positioned within a subcutaneous pocket over the rib cage and secured to underlying muscle and fascia with absorbable sutures. Once the subcutaneous portion of the shunt system had been completed, attention returned to the occipital area. Within the occipital craniotomy, bipolar cautery was used to coagulate the dura mater, which was then incised in a cruciate fashion with a #11 scalpel blade. The ventricular catheter was inserted through the opening in the dura mater and advanced through the occipital cortex into the lateral ventricle to a depth of about 2.5–3.5 cm from the occipital skull surface, based on pre-shunt measurements calculated from the MR images. Placement of the ventricular catheter in the lateral ventricle was confirmed by the appearance of CSF in the extracranial portions of the catheter. The ventricular catheter was anchored to the skull with Nylon sutures into the pericranium with or without a plastic anchor secured to the skull with self-tapping screws (Titanium cranial fixation system, Medtronic Brain Therapies, Irvine, CA, USA). The abdominal wall and chest incisions were closed in layers in a standard fashion with absorbable sutures. The cranial incision was closed in a layered fashion with absorbable sutures used to approximate the dorsal cervical muscles and overlying fascia in continuity with the galea aponeurotica. The skin was closed with subcuticular absorbable sutures and external interrupted 3–0 ETHILON® nylon non-absorbable sutures. External sutures were removed when the skin had healed, approximately 10-days post-surgery.

During the recovery period (i.e., until a sternal position and/or standing could be achieved), animals were monitored every 15 min and vital signs recorded. Subsequently, monitoring was conducted about every 4–8 h until the animals could nourish themselves and displayed no neurological symptoms or signs of pain and discomfort (i.e., about 1–2 days),. After this period of recovery, daily monitoring was conducted to assure normal recovery from anesthesia following the MRI scans and the surgical procedures (i.e., healing of incisions, tissue swelling, integrity of the distal catheter track). The animal’s neurological status (i.e., general locomotor and sensorimotor behavior, ataxia, balance, alertness, reflexes, and ability to eat and drink) was also monitored. Body temperature and clinical activity during recovery were monitored closely as lethargy and hyperthermia/fever were often observed after kaolin injections. Acute neurological and behavioral signs and symptoms were managed medically with Buprenorphine/Buprenex, Tylenol/acetaminophen suppository, Carprofen/Rimadyl, Dexamethasone, Keppra, and occasional alcohol baths for hyperthermia (Additional file 5: Table 1).

Anatomic MRI imaging was used to guide surgical implantation of ventricular catheters and determine gross morphological changes in the brain, especially volumetric increases in the cerebral ventricles, the location of the kaolin deposits, and the position of ventricular catheters. When possible, neuroimaging was performed pre-induction, post-induction/pre-shunting, and post-shunt on a Siemens Prisma 3.0-Tesla MR scanner with a 60-cm clear bore diameter, a 20-channel head coil, an 80 mT/m gradient field, and a slew rate of 200 mT/ms. The pig was anesthetized as described for hydrocephalus induction and positioned supine in the magnet. Breathing was controlled with a ventilator, respiratory and heart rates were monitored every 15 min, and oxygen saturation was monitored continuously with a pulse oximeter (Nonin® 7500, Nonin Medical, Plymouth, MN) secured to the tail. Depending on brain size, 54–192 slices of T1- and T2-weighted images were collected with a 3D fast spin-echo sequence with an echo train length of 8, FOV 205 × 205 mm (256 × 256 voxels), and a voxel size of 0.8 mm. T1 and T2 MPRAGE scan time varied from 4–11 min (T1:TR 2300 ms, TE 2.36 ms, 2 averages; T2: TR 3200 ms, TE 409 ms, 2 averages). Slice thickness was 900 µm. Data were provided in DICOM format (Digital Imaging and Communications in Medicine, standard for medical neuroimaging data).

Ventricular volume segmentation was performed using the free open-source software programs ITK-SNAP version 3.6.0 (http://​www.​itksnap.​org/​pmwiki/​pmwiki.​php). In ITK-SNAP, semi-autonomous methods were then used to separately label the lateral, third, fourth, and olfactory ventricles in each slice of the sagittal, axial (horizontal), and coronal view planes. Ventricular volume (mm³) for each case was obtained using the Volumes and Statistics tool within the Segmentation menu.

Well-established and validated tests developed by Dilger and Fleming [96‐99] were used to quantify cognitive function prior to and following hydrocephalus induction using novel object recognition (NOR) tasks. NOR can reliably identify cognitive differences in infant pigs and is sensitive to small differences between treatment groups, as exemplified in a recent trial of dietary milk supplementation by Fleming et al. [98]. Using an arena designed from a standard housing pen with opaque walls and door and various objects secured to the floor, pigs were given two 10-min habituation periods over 2 days to explore the empty arena. The following day, they were then given 5 min to become familiar with 2 identical objects (sample trial). After a 24-h delay period, they were then given 5 min to explore 1 familiar object from the sample trial and 1 novel object (test trial). Each trial was recorded with a video camera suspended over the arena. A single, unbiased (i.e., blinded to treatment group), trained observer annotated and scored the continuous recordings to make the following measurements: (1) the number of visits to novel/sample objects, (2) time investigating novel/sample objects, (3) mean length of time per visit to an object, (4) latency to the first object visit, and (5) latency to the last object visit. The primary outcome for cognition was the Recognition Index (RI), which was calculated as the proportion of time each pig spent actively investigating the novel object as a proportion of the total active exploration time of both objects during the test trial. RI significantly > 0.50 indicates novelty preference and thus recognition memory [96, 98].

At designated time points, animals were sedated and anesthetized with Isoflurane and heparin was administered IV to facilitate cardiac perfusion. The anesthetized animals were euthanized using intravenous sodium pentobarbital (120 mg/kg intraperitoneal) or isoflurane (5% inhalation) and following cardiac arrest the chest was opened. The descending thoracic aorta was clamped, the pericardium opened, the right atrium incised to allow intracardiac perfusion, first with Phosphate-Buffered Saline (PBS) or sterile saline delivered via a peristaltic pump (Cole Parmer Masterflex L/S Easy load II, Vernon Hills, IL, USA) through a blunt 13-gauge needle inserted into the left ventricle followed by 4% paraformaldehyde (Sigma-Aldrich, St. Louis, MO, USA) in PBS to complete fixation. Depending on the size of the pig, 5–7 l of each solution were needed for adequate brain fixation. The brain and shunt system were removed using gross dissection and stored in fixative for 48 h and subsequently in 70% ethyl alcohol.

For ventricular volume comparisons between sham controls and kaolin-injected animals, when a Kruskal–Wallis test indicated non-normal distribution of data, an unpaired t-test with Welch’s correction was used for non-parametric analyses. The primary outcome of cognitive function (i.e., learning and memory) was the Recognition Index (RI), which when altered significantly indicates novelty preference and thus Recognition Memory (RM) [96]. The linear mixed effects model was used to evaluate RM; this model is specified with RI as the outcome variable and a random intercept for animal. All data were reported as mean + standard deviation (SD) with statistical significance < 0.05.

Altogether, 27 cases (of the 58 total) contributed ventricular volume and survival data to this report (Table 1), as follows: (A) Untreated, kaolin-induced hydrocephalic pigs (n = 9) survived to euthanization at an age of 55–85 (median 70) days. (B) Shunt-treated hydrocephalic pigs (n = 8) received VP shunts at 53–79 days of age and survived until euthanization at 78–110 days of age (median 81). Amongst the group of shunted pigs, 8 contributed to the untreated ventricular volume analysis (i.e., pre-shunt) whereas only 4 contributed to the post-shunt ventricular volume analysis. Controls included 6 sham/saline-injected pigs that survived 82–93 (median 83) days and 4 intact/non-injected animals that underwent neuroimaging at the approximate age at which experimental animals received kaolin, i.e., 34–41 (median 37) days of age.

In addition to the 17 hydrocephalic cases used for the ventricular volume analysis, 31 kaolin-induced cases were also studied for other purposes (n = 48). Altogether, these results showed that 85.4% (41/48 cases) of kaolin injections produced ventriculomegaly.

In all kaolin-injected animals, at post-mortem examination, kaolin deposits consistently formed a solid cast within the basal cisterns (pontine, cerebellopontine angle, interpeduncular, and prepontine) and occasionally within the cistern of the lamina terminalis (Figures 1, 2, 3; Additional file 1: Figure S1, Additional file 2: Figure S2). Kaolin particles were rarely found within the cistern magna and never within the 4th- or 3rd-ventricles, the cerebral aqueduct, or the lateral ventricles. The basal cistern deposits could be identified on MRI throughout the post-induction and post-treatment survival periods on T2-weighted MRI scans (Figures 1B′ and 3A″) and were accompanied by arachnoiditis and adherence to the dura mater or membrane of Liliquist. In about half of the cases, thin, patchy kaolin deposits were found in the subarachnoid space surrounding the base of the hypothalamus and the pituitary gland. No kaolin deposits entered the parenchyma or any of the cerebral ventricles. The basilar artery and its branches were completely embedded in the kaolin cast.

Bilateral, symmetrical ventriculomegaly occurred in most cases. In 2 cases (<5), subsequent kaolin injections were needed to produce ventriculomegaly. All portions of the cerebral ventricles expanded, but no periventricular edema developed (Figures 1–3; Additional file 1: Figure S1, Additional file 2: Figure S2). It is noteworthy that olfactory ventricles were present normally in the domestic pig brain, each with a narrow channel connected to the frontal horn of each lateral ventricle; both the olfactory ventricle and the connecting channel expanded post-kaolin. The cisterna magna remained open and enlarged, with a thin membranous anterior wall that appeared to separate it from the 4th ventricle. Prominent CSF flow voids, indicative of high pulsatility, were conspicuous in the foramina of Monro, 3rd ventricle, and anterior cerebral aqueduct. The cerebral hemispheres remained gyrencephalic with relatively little distortion of the sulci and gyri. In contrast, the inferior wall of the temporal lobe became extremely thin in response to the considerable expansion of the temporal horn of the lateral ventricle (Figures 1D′ and 3D″, and Additional file 2: Figure S2E).

Volumetric assessments revealed that ventriculomegaly occurred post-induction in all cerebral ventricles (i.e., olfactory, lateral, 3rd, and 4th ventricles; Figs. 4–6). Quantification of ventricular volume confirmed the extent and variability of ventriculomegaly (Fig. 5). Ventricular volume increased significantly (p < 0.001) in all ventricles compared to both intact and sham controls. Regional variations existed within the cerebral ventricles; proportionally, the lateral ventricles expanded the most, followed by the 4th ventricle. All 17 untreated cases (no-shunt or pre-shunt) exhibited lateral and total ventricular volumes that were above 2 standard deviations from the sham control mean. We arbitrarily set this threshold to determine which cases were “hydrocephalic”. Linear regression revealed highly significant correlations between lateral ventricular volume and olfactory, third, and fourth ventricle volumes (Fig. 6).

For about 24-hours following kaolin injections and recovery, pigs were mildly lethargic with most of the cases exhibiting increases in body temperature (103.5–108.0 °C; normal = 100.5–103.5 °C) and some showed mild ataxia (imbalance, wider stance, stretching of hindlimbs). When this acute response occurred, it usually was resolved within 2–3 days with appropriate medical management (Buprenorphine/Buprenex, Tylenol suppository, Carprofen/Rimadyl, Baytril, and Dexamethasone) and with occasional alcohol baths for hyperthermia. Past 2–3 days post-kaolin, these animals did not exhibit any signs of pain or discomfort, and were alert with normal reflexes, consumed food and water, and gained weight throughout the post-kaolin period. Nevertheless, neuroimaging revealed ventriculomegaly in most of these animals (see Additional file 3: Figure S3).

Preliminary results from cognitive testing (Additional file 4: Figure S4 showed a non-significant trend toward a higher Recognition Index and significantly more exploratory time (p = 0.008) in the post-induction animals compared to normal (pre-induction) pigs, suggesting possible learning impairment; i.e., more time was needed to become familiar with the novel object.

Although neuroimaging was not performed on all 8 shunted cases due to logistics, available post-shunt scans revealed optimal (n = 2) and sub-optimal (n = 2) shunt placements. In cases 13 and 17 with optimal shunt placement, ventriculomegaly was reduced (Fig. 7A–C. In the suboptimal shunt cases 10 and 12 the proximal portion of the ventricular catheter was deeply embedded in periventricular tissue (Fig 7D′–F″) and ventriculomegaly increased.. In pilot studies, while attempting different trajectories (including coronal), gross anatomy revealed that the tip of the catheter occasionally penetrated the head of the caudate nucleus, the thalamus, the hippocampus, or the periventricular white matter. Subsequently, when the insertion of the catheter was limited to 3.5 cm from the external surface of the occipital skull to the approximate location of the foramen of Monro, more consistent and optimal placement within the lateral ventricle was achieved (Fig. 7), with the drainage holes open to CSF. In many cases, at least a portion of the catheter contacted the ventricular wall or the choroid plexus (sub-optimal placement, Fig. 7). Contact with the choroid plexus was occasionally associated with hemorrhage throughout this structure, and in these cases, blood was present in the distal valve and catheter. On two occasions, the plastic anchor, which had been sutured to the occipital bone and the catheter as it exited the skull, had become detached. This detachment allowed the ventricular catheter to migrate superficially, in one case dorsally into the parietal and occipital cerebral cortex; in the other case, the catheter had exited the cranial cavity and the proximal tip was located within the cervical musculature. Importantly, because both of these sub-optimal cases were asymptomatic for the entire 30-day post-shunt survival period, it is likely that CSF drainage had been effective for at least a portion of the treatment period. We now use Stealth guidance to optimize the trajectory to the lateral ventricle.

The most prevalent complication was sub-optimal placement of the ventricular catheter, especially in the initial cases. It was difficult to sense tissue penetration as the catheter was advanced (sub-optimal shunt placement, Fig. 7); therefore, it was necessary to rely on the pre-shunt MRI for trajectory planning. As we gained experience, several parameters became very important for optimal placement of the ventricular catheter: (1) The dorsoventral ‘height’ of the lateral ventricle should be at least 6.0 mm for a standard catheter with an outside diameter of 2.3 mm; we were most successful when this dimension was 10–13 mm. (2) Insertion should be no more than 3.5 cm from the surface of the occipital bone; this placed the proximal tip of the catheter near the foramen of Monro and prevented penetration of the caudate nucleus. (3) Insertion should be parallel to the dorsal surface of the skull; the tendency to advance perpendicular to the face of the occipital bone caused the catheter to be angled inferiorly with penetration of the thalamus, caudate nucleus, and/or hippocampus. (4) A plastic anchor, screwed into the occipital bone, should be used to secure the ventricular catheter in place; the Medtronic anchor supplied with standard catheters was ideal for this purpose.

Post-shunt periods less than 28 days were characterized by relatively acute onset of neurological symptoms (n = 4/8, cases 11–14); often demise began 6–24 h before euthanasia was required, and in two cases animals that appeared normal developed profound seizures just 8–10 h later. Shunt patency studies were not performed in vivo, so it is not possible to know unequivocally if these cases represented shunt malfunction. No shunted animals became febrile or developed infections.