A young man with Propionibacterium acnes-induced shunt nephritis

A 23-year-old man presented to our institution in May 2002 at the recommendation of his primary nephrologist in Puerto Rico for a second opinion regarding chronic glomerulonephritis. He had a history of a tectal pilocytic astrocytoma, which was diagnosed in 1992. The tumor was complicated by obstructive hydrocephalus that required ventriculoatrial (VA) shunt placement. He underwent partial tumor resection followed by radiation therapy and had required multiple revisions of his shunt, most recently in 1997. A magnetic resonance imaging of the brain obtained prior to presentation at our institution revealed prominent nodularity in the tectal region suspicious for residual tumor. In addition, there were extensive post-surgical and post-radiation changes in the medial aspect of the left occipital lobe. In 1999, he developed mild renal insufficiency with nephrotic-range proteinuria and microhematuria. A serologic work-up, the details of which were not available to us at the time of his presentation, was reportedly negative. Renal biopsy was performed in Puerto Rico, but only four glomeruli were sampled. Two out of the four glomeruli were reportedly sclerotic, whereas the other two demonstrated mesangial proliferation. The patient received a course of steroids for mesangio-proliferative glomerulonephritis of unclear etiology. After a brief period of initial response, he had several relapses in proteinuria and was kept on chronic prednisone for presumed idiopathic steroid-dependent glomerulonephritis. At the time of presentation at our institution, the patient's medications included prednisone 20 mg once daily, furosemide 20 mg twice daily, metolazone 5 mg once daily, and valsartan 160 mg once daily. The patient denied smoking, alcohol, or recreational drug use. He also denied risk factors for HIV infection. There was no family history of renal disease. On physical examination, he was afebrile with a pulse of 68 bpm and a blood pressure of 130/82 mm Hg. He was cushingoid and had severe acne on the face and anterior chest wall. The shunt was palpable under the skin of the neck, without evidence of erythema or tenderness. The pupils were equal and reactive to light, but the eye movements were notable for right exotropia. In addition, the patient had bilateral nystagmus and his hearing was decreased on the left side. The heart, lung, and abdominal examinations were unremarkable. The skin exam demonstrated multiple striae, predominantly in the abdominal area. Severe pitting edema of both legs was present. Examination of the joints and lymph nodes was unremarkable. Coordination, gait, and detailed peripheral sensory and motor exams were also within normal limits. Initial laboratory results are summarized in Table 1. Most notably, moderate renal insufficiency, hypoalbuminemia, leukocytosis, and anemia were present. Serum creatinine was 1.5 mg per 100 ml (reference 0.5–0.9), a spot urine protein to creatinine ratio was 3.2, and urine microscopy revealed 20 RBCs per high-power field. Hepatic enzymes and thyroid function testing were within normal limits. Additional work-up revealed an elevated erythrocyte sedimentation rate of 65 mm/hr (normal 0–15), low serum complement C3 of 50 mg per 100 ml (normal 83–177), and low-normal C4 of 19 mg per 100 ml (normal 16–47). Other serologic tests, including antibodies to hepatitis B and C, antinuclear antibodies, rheumatoid factor, cryoglobulins, and anti-streptolysin-O were all negative. Renal ultrasound showed normal sized kidneys with no hydronephrosis and normal echotexture. The renal veins and arteries were patent on Doppler examination. Chest X-ray revealed small bilateral pleural effusions. Radiographs of a shunt series verified correct shunt position, with the cephalad tip in the cranial fossa and the distal tip projecting over the expected location of the right atrium of the heart. Upon review of the original kidney biopsy slides from Puerto Rico, it was felt that the sampling of only four glomeruli with predominant sclerosing features was inadequate for rendering a definitive diagnosis and guiding subsequent management. In the light of the patient's ongoing renal insufficiency, proteinuria, hematuria, and hypocomplementemia, a repeat kidney biopsy was performed. The renal biopsy contained 36 glomeruli, 29 of which were globally sclerotic. Several of the obsolescent glomeruli displayed circumferential subcapsular fibrous proliferation with disruption of Bowman's capsule, suggesting old fibrous crescents. The remaining seven glomeruli appeared enlarged with accentuated lobularity (Figure 1a). There was severe diffuse and global mesangial expansion by increased mesangial cell number and matrix associated with membranoproliferative features in the form of circumferential mesangial interposition and narrow duplication of glomerular basement membranes (Figure 1b). Many glomeruli contained infiltrating neutrophils as well as occasional eosinophils. Two glomeruli contained small segmental crescents, one of which was cellular and the other fibrocellular. Approximately 30–40% of the cortex displayed patchy tubular atrophy and interstitial fibrosis accompanied by a moderate focal interstitial infiltrate of lymphocytes, monocytes, plasma cells, and occasional neutrophils. There were focal red blood cell casts. Some proximal tubular cells contained intracytoplasmic protein and lipid resorption droplets. There was mild to moderate arterio- and arteriolosclerosis. No arteritis was identified. The single non-sclerotic glomerulus sampled for immunofluorescence revealed semilinear to granular glomerular capillary wall positivity for IgG (1+), IgM (1+), C3 (2+), C1 (1+), kappa, and lambda (both trace) in a subendothelial distribution with weaker mesangial positivity (Figure 1c). On electron microscopy, glomerular capillary lumina were narrowed or occluded by marked mesangial hypercellularity and increased matrix accompanied by mesangial interposition and duplication of glomerular basement membranes. Some expanded mesangial regions were more sclerotic, with increased matrix and nodularity. Many lobules contained infiltrating monocytes or neutrophils, sometimes associated with endothelial swelling and hypercellularity. There were numerous small mesangial and subendothelial electron-dense deposits with a granular texture, lacking organized substructure (Figure 1d). Foot process effacement involved approximately 90% of the glomerular capillary surface area. Three distinct categories of disease may be associated with a membranoproliferative pattern of glomerular injury on light microscopy. These include immune complex-mediated diseases, thrombotic microangiopathies, and paraprotein deposition diseases.1 In this case, the presence of granular mesangial and subendothelial electron-dense deposits seen on electron microscopy, combined with the presence of polyclonal immunoglobulins and complement components seen on immunofluorescence, point to an immune-complex mediated process. The differential diagnosis includes idiopathic versus secondary forms of membranoproliferative glomerulonephritis. A listing of the differential diagnosis of immune complex-mediated membranoproliferative glomerulonephritis is provided in Table 2. Apart from his renal disease, this patient had no other clinical manifestations of a systemic autoimmune process, and with the exception of hypocomplementemia, rheumatologic serologies were unremarkable. Additional testing for viral, bacterial, and parasitic infections would be needed to exclude infectious causes of secondary membranoproliferative glomerulonephritis. In this patient with a longstanding VA shunt, secondary membranoproliferative glomerulonephritis due to subacute infection of the shunt leading to immune complex-mediated glomerular injury, 'shunt nephritis', was a strong etiologic consideration. Outpatient blood cultures were obtained and grew Propionibacterium acnes. The patient was admitted to the hospital. Cerebrospinal fluid (CSF) cultures were obtained from the VA shunt and they also grew P. acnes. The patient was started on empiric broad-spectrum antibiotics, including intravenous cefotaxime, vancomycin, and ampicillin. In spite of antibiotic therapy, CSF cultures obtained from the shunt remained positive for six consecutive days. He was taken to the operating room for removal of the shunt and an endoscopic frontal third ventriculostomy. After final identification and sensitivity testing of the P. acnes isolate, he was kept on cefotaxime monotherapy for a total of 4 weeks of antibiotic treatment, the latter portion of which he received at home via a peripherally inserted central catheter. The follow-up trends in creatinine, proteinuria, and complement components are summarized in Figure 2. Creatinine level fell from a peak of 1.7 mg per 100 ml prior to shunt removal and antibiotic therapy, down to 1.3 mg per 100 ml within two weeks after the surgery. Valsartan was continued at 160 mg daily and prednisone was initially continued at 20 mg daily. The patient's proteinuria declined, and within 12 months the urine protein to creatinine ratio was down to 0.1. Over this time period, a slow steroid taper was initiated and the prednisone, along with valsartan, were finally discontinued in June 2004, approximately 24 months after shunt removal and antibiotic therapy. In July 2004 the patient developed headache, ataxia, and worsening hearing loss. MRI of the brain demonstrated aqueductal outflow obstruction. Tumor de-bulking procedure was performed and a ventriculoperitoneal shunt was placed. One year later, the patient was noted to develop recurrent low-grade proteinuria (urine protein to creatinine ratio 0.9). Serum creatinine during this time was 1.1 mg per 100 ml; there was no recurrent microhematuria and no evidence of new shunt infection. The proteinuria decreased following re-initiation of low-dose valsartan. Five years after VA shunt removal and antibiotic therapy, the patient's creatinine is 1 mg per 100 ml and spot urine protein to creatinine ratio is 0.3. Membranoproliferative and sclerosing glomerulonephritis with exudative features and focal crescent formation due to P. acnes-induced shunt nephritis. Since the initial description of immune complex glomerulonephritis associated with infection of a VA shunt in 1965,2 over 160 case reports have appeared in the literature (reviewed by Haffner et al.3 and Arze et al.4). Common clinical features in patients with shunt nephritis include renal insufficiency of variable severity, hematuria, proteinuria (usually in the subnephrotic range), hypertension, and anemia. Serum complement levels generally are depressed and cryoglobulins as well as antinuclear antibodies may be detected. Although low-grade fevers may be present, not all patients demonstrate typical signs and symptoms of active infection. As a result, a substantial delay from the onset of the initial renal manifestations to the definitive diagnosis (including demonstration of occult shunt infection) may be present. Moreover, the onset of hematuria, proteinuria, and renal insufficiency may vary from several weeks to several years after the patient's initial shunt surgery or latest shunt revision. Most but not all patients have positive cultures of either the blood, CSF, or both. Renal biopsy is helpful in establishing a causal relationship between the patient's renal manifestations and shunt infection. Typical findings on renal biopsy include a membranoproliferative pattern of glomerular injury seen on light microscopy, with immunofluorescence demonstrating granular subendothelial and mesangial deposits containing polyclonal immunoglobulins (IgM and IgG) and complement (mainly C3). The mechanism of renal injury probably involves persistent antigenemia derived from an infectious agent, which stimulates immune complex formation. It is unclear whether the immune complexes are formed in the circulation and then passively deposited in the glomeruli, or if they form in situ after foreign antigens are first trapped within the glomerular filtration barrier. The presence of immune complexes induces classical pathway complement activation, which in turn mediates direct injury to glomerular cells (via the C5–9 complex) and generates chemotactic peptides (C3a, C5a) that perpetuate local inflammation.1 The most commonly reported organism causing CNS shunt infection with nephritis is Staphylococcus epidermidis,3 accounting for 70% of isolates described in the cases reviewed by Arze et al.4 P. acnes is the next most commonly reported organism. The first reported examples of P. acnes shunt infection were by Beeler et al. in 1976.5 Of the three cases summarized by their report, one had a concomitant immune-complex glomerulonephritis, an entity not previously associated with anaerobic organisms and certainly not expected with P. acnes, which at the time was generally considered to be a non-virulent commensal skin organism. Since then, of the approximately 160 reported cases of shunt nephritis with microbiologic data (including those reported by Arze in 1983,4 Haffner in 1998,3 and the additional 17 cases reported in the last 10 years), at least 10 cases appear to have been mediated by P. acnes. However, the true incidence of P. acnes shunt infection is probably higher than reported, due to the organism's lengthy incubation period and poor growth on certain culture media.6 Propionibacterium species are Gram-positive anaerobic bacilli, typically found among normal skin flora, as well as in the oral cavity, colon, conjunctiva, and external ear canal. In the skin, P. acnes clusters in moist environments, such as the axilla or groin areas, preferentially around hair follicles and sebaceous glands. The organism is slow growing and capable of surviving up to 8 months under anaerobic conditions. In addition to shunt infections, several other serious complications of P. acnes have been reported over the last three decades (Table 3). In general, invasive P. acnes infections tend to occur in the setting of antecedent trauma such as surgery, or in the presence of prosthetic devices. P. acnes has recently been shown to have the ability to form a biofilm (a matrix of extracellular polymeric substances embedding and protecting microorganisms and enabling their communication through biochemical signals).22, 23 Biofilm formation facilitates infections of implanted materials and explains why such infections are difficult to treat without surgical removal of the device, despite the organism's susceptibility to common antibiotics. Recent studies have helped define other virulence factors specific to P. acnes. This search has been greatly facilitated by the recent publication of the complete sequence of the P. acnes genome.24 Detailed analysis of the coding regions has revealed numerous putative virulence genes. These genes may be subdivided into two groups: those coding for proteins involved in tissue invasion and those that modulate the host's inflammatory response. The first group involves enzymes found in a variety of other pathogenic microorganisms25 as well as several genes potentially involved in biofilm formation.22 The second group of genes includes sequences of putative antigenic proteins, which are either present on the microorganism's surface or secreted into the extracellular space. Members of this group are likely responsible for some of the inflammatory conditions associated with P. acnes infection, such as acne vulgaris, nephritis and possibly, sarcoidosis (see Table 3). In contrast to the well-established link between P. acnes and acne vulgaris and shunt nephritis, an association with sarcoidosis remains controversial. It was first suggested by Eishi and co-workers16 who detected P. acnes DNA in mediastinal lymph nodes of patients with pulmonary disease. While some investigators suggest that this finding may be non-specific, as P. acnes has been found in mediastinal lymph nodes of many subjects without sarcoidosis,26 other studies support this intriguing association.17, 18 Interestingly, the membranoproliferative pattern of renal injury has been observed in patients with pulmonary sarcoidosis.27, 28, 29 In addition, P. acnes DNA has been detected by in situ hybridization in a kidney biopsy specimen of a patient with renal sarcoidosis and glomerulonephritis.19 The finding of otherwise unexplained hematuria or proteinuria, with even mild renal insufficiency in a patient with a VA shunt should prompt an aggressive diagnostic approach to exclude subacute shunt infection, even in patients without fever or leukocytosis. P. acnes should always be considered as a potential causative pathogen and anaerobic CSF and blood cultures should be performed in all such cases. The prognosis for stabilization and improvement in renal function and proteinuria tends to be excellent, even in patients with renal insufficiency at the time of diagnosis, provided the infected shunt is removed in a timely manner and an appropriate course of antibiotics is initiated. In the extensive case series reported by Haffner et al.3 (82 cases), the longest reported follow-up period was 22 years, with the patient at that time demonstrating a normal serum creatinine of 0.7 mg per 100 ml and no urinary abnormalities, in spite of renal insufficiency at the time of diagnosis (creatinine 1.1 in a 5-year-old girl) and 50% of glomeruli demonstrating cellular crescents on biopsy. In the earlier case review reported by Arze et al.4 (70 cases), greater than 50% of patients were classified as experiencing 'complete recovery' of renal function after appropriate surgical and antibiotic treatment, although the fact that these were individual case reports brought together for the purposes of a literature review makes it impossible to establish a standard definition of what the authors refer to as 'recovery'. The vast majority of reported cases of shunt nephritis have occurred in the setting of an existing VA rather than ventriculoperitoneal shunt. Therefore, patients such as ours who, after removal of the infected VA shunt, continue to require CSF drainage due to symptomatic obstruction, generally go on to have subsequent ventriculoperitoneal rather than VA shunt replacement. Delayed removal or lack of removal of the infected shunt, even if appropriate antibiotics are administered, may be associated with progressive renal failure leading to ESRD.3 Few, if any, data exist with which to make evidence-based decisions regarding concomitant use of corticosteroid treatment. Although the potential benefit of corticosteroids in terms of reducing glomerular inflammation may seem intuitive, clinicians may understandably be reluctant to initiate immunosuppressive treatment in the setting of documented CNS infection. Depending on the degree of renal insufficiency at the time of diagnosis and the relative severity of the acute (potentially reversible) inflammatory changes noted on biopsy, we feel that once the shunt is removed and appropriate antibiotics are initiated, a course of steroid treatment may be beneficial in helping to reduce glomerular inflammation and thereby stabilize renal function. As with all other forms of glomerular disease, adequate blood pressure control and proteinuria reduction should be attempted with judicious use of angiotensin-converting enzyme inhibitors and/or angiotensin receptor blockers. This case illustrates in a striking manner the potential renal complications associated with chronic, unrecognized VA shunt infection. The diagnosis of shunt nephritis is challenging and, as exemplified by this case, may be overlooked for several years, resulting in significant morbidity in the form of ongoing renal injury with proteinuria and renal failure. The finding of hematuria, proteinuria and even mild renal insufficiency in a patient with a VA shunt should prompt an aggressive microbiologic diagnostic evaluation to exclude subacute infection. Renal biopsy should be considered to document the pathognomonic membranoproliferative lesion and to assess the relative degrees of acute versus chronic injury. Further, this case highlights the critical importance of excluding secondary forms of disease in any patient with a nephritic presentation and biopsy findings demonstrating membranoproliferative features, as identification and treatment of the underlying condition may lead to remission of proteinuria and normalization of renal excretory function, even in patients with a baseline biopsy finding of significant sclerosis alongside ongoing active inflammation, as was the case in this patient. Finally, in this exciting era where the elucidation of entire microbial genome sequences has become possible, future studies correlating bacterial genomes and proteomes with specific mechanisms of disease may hold promise for identifying specific antigens involved in the pathogenesis of immune-mediated conditions, such as shunt nephritis.