An experimental in-vivo canine model for adult shunt infection

The introduction of cerebrospinal fluid (CSF) shunts revolutionized the treatment of hydrocephalus and greatly decreased mortality rate. However, over subsequent decades, difficulties in shunt maintenance and problems resulting from infection or occlusion have persisted [1]. Today, shunt failure rates range from 25% to 40% in the first few months after surgery [2, 3]. After this critical period, the risk remains at 4% – 5%. As a result, some patients may have numerous shunt revisions during their lifetime, representing a major medical risk. Infections have been a leading cause of shunt failure with infection rates ranging from 1% to 30% [4‐8]. Overall, the international reported incidence is around 10% [9]. Gram positive bacteria are most commonly implicated, and approximately 80% are caused by staphylococci, most of these being Staphylococcus epidermidis [9, 10]. These organisms originate on the patient's skin and mucous membranes and investigational and clinical evidence indicates that the inoculation of the shunt system occurs at the time of operation [11]. The severity of a shunt infection may vary from a sub-clinical colonization of the shunt inner lumen surface to fulminating ventriculitis, meningitis, or encephalitis which can be neurologically catastrophic, although the picture with S. epidermidis infections is typically low-grade. A shunt-related infection of the central nervous system (CNS) is of particular concern because of the threat to cerebral functions. Treatment may be prolonged and relapses may occur, complicating control of intracranial pressure and giving rise to life-threatening risks such as cognitive and neurological deficits [3, 6, 12, 13].

Although many studies have investigated the rate, aetiology and epidemiology of shunt infections, preventive measures are difficult to study, and their pathophysiology is largely unknown due to the difficulties inherent in human studies. Animal models of infection in other implantable devices have not hitherto been suitable for these purposes. We have developed a canine model of chronic obstructive hydrocephalus [14] which produces a gradual, well – tolerated increase in ventricular size, resulting in transient and subtle clinical effects. Unlike previous models, there is no focal compression or general inflammation which may confound analysis [15, 16]. Because of the size of the animals this model allows actual shunting procedures to be carried out using shunts designed for human use, and clinically relevant measurements to be made for the purposes of diagnosis, treatment and research. This paper, therefore, describes and characterises an animal model able to address these issues, and reports its use in a pilot study of chronic low-grade shunt infection.

One of the more striking findings from this study was the survival and overall health of each animal, with weight gain and no signs of systemic infection after shunting. The wounds healed well and there was no general spread of bacteria within the ventricular system and no classic signs of inflammation in the brain. The features of overt shunt infection include fever, with in some cases inflammation over the catheter track and possible wound breakdown. Distal obstruction of the peritoneal catheter, with consequent signs of raised intracranial pressure, would be expected. None of these was seen, and benign pathophysiology and lack of bacterial spread may be surprising in view of the large inoculation with S. epidermidis. It could be explained in part by the termination of the experiment at 4 weeks; these low-grade staphylococcal infections often have a longer latency and are more indolent and difficult to diagnose. They cause little or no direct injury, but if untreated they will eventually usually cause distal catheter obstruction due to adhesions or pseudocyst formation [19, 20]. However, there is another possible explanation: S. epidermidis infections are a challenge to diagnose and treat, in part because of the ability of most strains to grow inside shunts as biofilms [21, 22]. Bacteria in biofilms downregulate their synthesis of target sites for antibiotics, thus becoming insusceptible to antibiotic therapy [23]. The biofilm mode may be another explanation for the finding of uniformly positive catheter lumen cultures but no evidence of bacterial spread, invasion, or inflammation in the periventricular brain and ependyma. The negative lumbar CSF cultures as well as the benign central nervous system pathology may indicate the inability of this bacterium to migrate rapidly from the adherent biofilms into the surrounding tissues, though the lumbar CSF would not be likely to show signs of infection in view of the obstructive hydrocephalus. Indeed, the human clinical features of S. epidermidis VP shunt infection, especially at four weeks post surgery, show a similar limited response. While there was no evidence of inflammation using standard histological techniques, other markers of inflammation and infection, such as microglial proliferation and changes in cytokine levels in the brain and/or CSF might have been informative. Also, more pathological abnormalities might become apparent in future studies when the infections are allowed to run for several months before sacrifice. This is suggested by the finding of subgaleal fluid collections in five of the animals on dissection.

Using the model a 100% infection rate (four out of four animals inoculated with S. epidermidis) was achieved and verified by gross examination and culture of hardware and tissue. In addition, the important role of skin flora in seeding of a shunt system during insertion was demonstrated fortuitously by the high rate of S. intermedius isolation, the three dogs that were not inoculated with S. epidermidis all contracting shunt infections due to S. intermedius. The evidence of infection decreased with samples taken downstream of the catheter in the peritoneal cavity, and none was found in the lumbar fluid.

In spite of the inoculation of 10⁸cfu/mL bacteria into the ventricular catheter in four animals, in two cases the animals' own S. intermedius was also isolated, as well as from the three uninoculated controls . S. intermedius comprises 90% of all the staphylococci isolated from canine skin [24]. In two cases where both S. intermedius and S. epidermidis were found in the same shunt system the S. intermedius was found in the more proximal portion, and S. epidermidis in the more distal part. Given the flow of CSF through the shunt system this suggests the possibility of bacterial contamination from the proximal end down to the distal end. Comparison of the S. intermedius isolates by three methods (Table 2) showed that the two isolates from dog 1 (tissue from shunt track and inside of shunt tubing) were identical but different from the isolates from the other dogs. While the differences were only slight, all the other isolates differed from each other in one characteristic or another, suggesting that in each case the bacteria were derived from the dog's own skin flora rather than from cross – contamination from a single source.

S. intermedius has been isolated from a majority of healthy canines [24] although it is rarely isolated from humans. However, transmission to humans has occurred, sometimes resulting in clinical infection [25‐27]. S. intermedius can be confused with Staphylococcus aureus. Both are coagulase positive, and produce a heat – stable nuclease, but S. intermedius does not produce Protein A and most isolates are clumping factor negative. Some variation can occur. Our isolates were clumping factor positive but coagulase negative using human plasma, and they produced a heat – stable nuclease. Their identification was kindly confirmed by Professor François Vandenesch of the University of Lyon, France.

This paper characterizes a canine model of chronic adult obstructive hydrocephalus. The animals survived well with no mortality and low morbidity, providing the opportunity for physiological studies, pathophysiological analyses and diagnostic studies.