Propionibacterium acnes Vertebral Osteomyelitis

Seek and Ye Shall Find?

Abstract

The pathogenic potential of Propionibacterium acnes in implant-associated spine infections is increasingly recognized. However, the clinical findings and outcomes associated with this pathogen in nonimplant-associated spine infections are unknown. We retrospectively reviewed nine male patients who met our strict case definition for P. acnes vertebral osteomyelitis to determine whether the laboratory, radiographic, and histopathology found in these infections differs from that typically associated with vertebral osteomyelitis. We also determined clinical outcomes as they related to medical and surgical treatments rendered. Radiographic results, laboratory results, and histopathology results of patients with P. acnes vertebral osteomyelitis were often not those classically described for this disease. Two of the nine patients‘ imaging examinations did not suggest infection. Five of six patients had normal erythrocyte sedimentation rates at diagnosis. Three of six patients’ histopathology did not show evidence of infection. Two patients whose treatment included spinal surgery with implant placement developed relapse of their infection. The remaining seven patients had rapid improvement with antimicrobial therapy. Patients with P. acnes vertebral osteomyelitis may present with atypical findings. Clinicians should be aware of the pathogenic potential of this organism, particularly following invasive spine procedures.

Level of Evidence: Level IV, therapeutic study. See the Guidelines for Authors for a complete description of levels of evidence.

The preceding two decades have witnessed remarkable changes in the diagnostic and therapeutic approach to pyogenic vertebral osteomyelitis. Dramatically rising numbers of spinal fusion procedures and epidural injections for pain relief and anesthesia have increased the possibility of postoperative disc space infections.^6,12 Image-guided percutaneous needle aspiration for biopsy and culture is commonly used to establish a microbiologic etiology facilitating directed therapy, thereby minimizing the need for surgery.⁷ Typically patients with vertebral osteomyelitis present with marked elevations in erythrocyte sedimentation rates, characteristic findings on magnetic resonance imaging (MRI), and, if obtained, histopathology suggestive of acute inflammation.^3,4 In this context Propionibacterium acnes has emerged as an increasingly recognized etiology of musculoskeletal infections and of vertebral osteomyelitis specifically.^8,13,23

P. acnes is a nonspore-forming anaerobic gram-positive bacillus. Long considered a nonpathogenic skin commensal that rarely caused disease but frequently contaminated blood cultures, P. acnes has been increasingly recognized as a cause of musculoskeletal infections, postprocedural neurosurgical infections, and endocarditis.^{8,9,13,17,23,27,31} Accumulating evidence implicates P. acnes as a common cause of delayed onset spinal implant infections.^10,18,28,29 Whether P. acnes plays a pathogenic role or is an innocent bystander contaminant presents a difficult clinical challenge, particularly in the absence of foreign bodies, which are considered to play an important role in the pathogenesis of these infections.²⁵ A greater understanding of the clinical findings that typify P. acnes spine infections in the absence of implants will serve to enhance clinicians' awareness of this microbe's pathogenic potential.

We aimed to assess whether radiographic and laboratory characteristics of P. acnes vertebral osteomyelitis differed from those traditionally described for vertebral osteomyelitis. We also sought to determine the potential effects of medical and surgical management strategies on outcomes in this group of patients.

MATERIALS AND METHODS

We retrospectively evaluated the clinical and radiographic findings, treatment, and outcome of nine patients with Propionibacterium acnes vertebral osteomyelitis (in the absence of spinal implants). We first reviewed all potential cases of spine infections (disc space infections, vertebral osteomyelitis, epidural abscess) from 1994 to 2002 by searching the Mayo Clinic Medical and Surgical Indexes,²¹ a database of interventional radiology procedures, and positive spine site cultures from our microbiology lab. This search yielded 333 patients. We included only patients age 18 years or older with back pain, clinical suspicion of spine infection, and at least two separate spinal site cultures positive for only P. acnes for this report. Though not necessary for inclusion in the study, all patients either responded to appropriate therapy or had infection relapse with the same organism, furthering the likelihood that P. acnes was acting as a true pathogen. We excluded patients with spinal implants at the time of diagnosis and patients with mixed infections that included P. acnes. This left nine patients, all of whom were men with back pain, as the presenting complaint (Table 1). Six patients had previously undergone an invasive procedure of the spinal column (four patients had laminectomies; two patients had discectomies) a median of 169 days (range, 79-489 days) before diagnosis. The median duration of symptoms before diagnosis was 135 days (range, 44-474 days). The minimum followup was 2 months (mean, 40 months; range, 2-100 months); only one patient did not have followup of at least 1 year. The study was approved by our institutional review board prior to data collection.

TABLE 1: Demographic and Clinical Characteristics of Nine Patients with P. acnes Vertebral Osteomyelitis

Medical and surgical diagnostic approaches and therapies were not standardized and were performed at the discretion of the treating physicians. Various radiologists interpreted imaging tests as part of routine clinical practice, and diagnostic impressions were recorded per routine clinical practice (Table 2). Five patients had blood cultures performed, but none were positive (Table 3). Seven patients had at least one spinal aspiration with culture performed. Six of these patients went on to have confirmatory cultures obtained at an open surgical procedure. P. acnes was isolated in anaerobic culture media in all nine patients, exclusively anaerobically in eight. Among the patients with available information, cultures turned positive after 3 to 7 days. Susceptibility testing was obtained for six patients; all isolates tested were penicillin-susceptible.

TABLE 2: Radiographic Diagnostic Impression of Nine Patients with P. acnes Vertebral Osteomyelitis

TABLE 3: Diagnostic Features of Nine Patients with P. acnes Vertebral Osteomyelitis

Eight patients underwent a spinal surgical procedure for diagnosis and/or treatment of their infection. Six patients underwent a surgical procedure to confirm the diagnosis of infection and/or establish a microbiologic etiology. Of these, three patients had transpedicular biopsies, two patients underwent anterior débridement with spinal fusion using autologous bone grafting material, and one patient had posterior débridement with biopsies and cultures. The seventh patient was unexpectedly found to have a P. acnes infection of a spinal pseudarthrosis at a prior spinal fusion site during a revision procedure. The eighth patient that underwent a surgical procedure had combined anterior/posterior débridement and spinal fusion to confirm the diagnosis, eradicate local infection, and achieve spinal stability. This patient received only perioperative prophylactic antimicrobials. Antimicrobials were not prescribed initially because cultures did not turn positive until after the patient had been discharged from the hospital. Subsequent to the patient's discharge from the hospital intraoperative cultures yielded P. acnes from multiple specimens, confirming the preoperative imaging findings suggesting infection. The patient was doing very well when culture results were belatedly noted, and therefore antimicrobials were not prescribed.

The eight patients who received parenteral antimicrobial therapy were treated for a median 42 days (range, 28-50 days). Six of these patients were treated with β-lactam agents (three with cefazolin, three with ceftriaxone) and two with vancomycin.

From the records of these nine patients we noted clinical data (laboratory values, radiographic findings, histopathology results), and the clinical outcomes at the time of last followup. One of us (TJK) used a coded data collection instrument to record data to help standardize data collection.

RESULTS

Radiographic results (Table 2), laboratory results (Table 3), and histopathology results (Table 3) of patients with P. acnes vertebral osteomyelitis were often not those classically described for this disease. Of the six patients who had erythrocyte sedimentation rates (ESR) available at diagnosis, five had values within range of normal; the sixth patient had a markedly elevated ESR (135 mm/hr). Six patients had histopathologic examinations performed on spinal aspirates or surgical specimens. The histopathology was strongly suggestive of infection in three patients, but was not suggestive in the other three patients (Table 3).

Two of nine patients were subsequently diagnosed with infection relapse. The first patient initially was diagnosed with osteomyelitis and was treated with surgical débridement and concomitant spinal fusion with implant placement. This was followed by 4 weeks of parenteral antimicrobial therapy for culture-positive P. acnes infection. Over 2½ years later the patient complained of persistent back pain and a revision surgery was offered for the diagnosis of pseudarthrosis. At the time of revision surgery the periimplant area looked infected. Implants were removed, and the patient was treated with 6 weeks of empiric antimicrobial therapy. Intraoperative cultures were negative. Four months later the patient underwent a staged second spinal fusion procedure with implant placement. Multiple cultures obtained during the second fusion procedure were once again positive for P. acnes. The patient was retreated with ceftriaxone for 6 weeks and was subsequently placed on long-term oral antimicrobial suppression with cephalexin. With over 4 years of followup, there is no evidence of recurrent infection, though the patient takes prescription pain medication for continuing back pain. The second patient that developed treatment failure was initially diagnosed with vertebral osteomyelitis and underwent débridement by an anterior spinal approach in combination with cefazolin treatment for 6 weeks. Subsequently the patient underwent spinal fusion with implant placement. The patient complained of persistent low back pain 4½ years later, and an MRI suggested disc herniation. The patient's ESR was 6 mm/hr and C-reactive protein (CRP) level was 0.125 mg/dL. The patient underwent a revision surgery. Multiple cultures were positive for P. acnes. This patient received only perioperative parenteral antimicrobial therapy. Despite receiving essentially only surgical débridement for treatment, this patient has over 2½ years of followup with no recurrent infection or pain.

The remaining seven patients in the series had rapid improvement in their symptoms following institution of therapy and no evidence for treatment failure over the followup periods. The one patient who had combined anterior/posterior débridement and spinal fusion to confirm the diagnosis received only perioperative prophylactic antimicrobials, but with over 4 years of followup there was no suggestion of recurrent spine infection.

DISCUSSION

The laboratory, radiographic, and histopathologic findings of P. acnes nonimplant associated vertebral osteomyelitis are not well-characterized. Medical and surgical management strategies for such infections are poorly characterized and may impact clinical outcomes. This report demonstrates these infections often have atypical findings, and that outcomes are generally quite good in the absence of spinal implants being placed as part of the treatment. The study is limited by its retrospective nature and limited sample size.

This case series describes P. acnes spine infections in the absence of implants and emphasizes the diagnostic challenges and therapeutic quandaries posed by this emerging but indolent pathogen. Though historically not considered pathogenic, P. acnes has been more recently considered a primary pathogen only in the setting of foreign material. This series suggests that P. acnes may not be an uncommon cause of native spine disc space infections, particularly following invasive procedures of the spine. A consistent clinical finding in these patients is back pain. Our findings also demonstrate that P. acnes native spine infections often present with atypical laboratory and radiographic findings. Furthermore it is well-known that P. acnes recovery in culture is optimal when medium for anaerobic growth is used, and cultures often do not turn positive for a number of days. Not all physicians and facilities routinely perform anaerobic cultures or have extended durations of culture. It is possible that we may have missed cases of P. acnes spine infection during the study period if appropriate cultures were not taken. Accordingly, we suggest that P. acnes may be an underrecognized and underreported etiology of native spine disc space infections.

Upon comprehensive review of the published English literature using PubMed, MEDLINE, and a review of the bibliographies of pertinent articles, we identified 22 reported cases of native spine vertebral osteomyelitis caused by P. acnes in addition to our series of nine patients.^{1,8,11,13,15,16,19,20,22,24,26,27,32,33} Patients with P. acnes native spine infections seem to demonstrate some notable similarities.

Of the 22 cases we identified on literature review, 20 had information regarding gender. Fourteen of 20 patients were male, suggesting an apparent gender disparity in patients with P. acnes vertebral osteomyelitis. We found infection exclusively in men in our series of nine patients. The reason for this apparent difference is unclear. It could be related to differential skin colonization and subsequent procedure-related wound contamination in men versus women. McLorinan²⁵ recently described quantified P. acnes growth from the skin overlying the lumbar spine in four men and six women. The mean colony forming units/cm² was 1 × 10⁵ for the men and 3 × 10³ for the women.

Of those patients identified on literature review, 10 of 17 patients had some predisposing invasive procedure of the spine that presumably contributed to the infection, consistent with the findings in our series where six of nine patients had previously undergone a predisposing procedure. While some have advocated hematogenous seeding of previously manipulated spinal tissue as the likely source of infection, there is substantial evidence to suggest that direct contamination at the time of the procedure is the likely route of seeding. In a recently published study of 79 patients,²⁵ 21.5% of patients undergoing spine surgery had positive cultures from surgical tissue cultures, and 65% of those cultures contained P. acnes.

Erythrocyte sedimentation rates and CRP, while notoriously nonspecific, are considered relatively sensitive tests in vertebral osteomyelitis.⁵ In our series, only one of the six patients tested had an elevated ESR rate. Seven of 17 patients with diagnostic ESR information available upon literature review had rates less than 35 mm/h.^{1,8,11,13,16,20,22} Furthermore, we and others^1,8,11,13,32 have found atypical radiographic findings can accompany P. acnes spine infections, particularly the lack of substantial paravertebral and epidural inflammation about the bony lesions. Of the patients in our series with histopathology available, three of six did not show evidence of acute infection. This may have been due to either sampling error or, potentially, that low virulence P. acnes infections do not always evoke typical inflammatory responses. Gram stains of appropriate specimens are typically negative even in clinically documented disease when P. acnes is the etiologic pathogen.¹⁴ Culture results often take many days to turn positive. Patients may be discharged from the hospital or even lost to followup before their spinal cultures turn positive.¹¹ Finally P. acnes is a well-known common skin contaminant of blood and other cultures.² As such, discerning the clinical relevance of a single culture growing P. acnes from a spine aspiration procedure is very challenging.

To consider all spinal aspiration cultures positive for P. acnes pathogenic would almost certainly result in over-treatment (if there was no infection present) or mistreatment (if another microbe was the true pathogen but not isolated on culture). Conversely, if one were to consider all such isolates contaminants, it would almost certainly result in undertreatment (if physicians decided no infection was present) or further potentially unnecessary invasive procedures (eg, surgical biopsy to confirm the microbiologic findings). P. acnes infections of the native spine thus present a tremendous diagnostic challenge for even the wariest physician. At our institution we usually obtain a confirmatory culture via second spinal aspiration or surgical biopsy before proceeding with prolonged parenteral antimicrobial therapy. However, in the correct clinical context of a suspected disc space infection following an invasive procedure with heavy growth of P. acnes in pure culture, it is not unreasonable to initiate antimicrobial therapy in the absence of a second confirmatory culture. Using an antimicrobial with adequate staphylococcus coverage, such as cefazolin, is a reasonable choice in this situation.

In previously published reports, most patients with P. acnes vertebral osteomyelitis were treated successfully with antimicrobial therapy alone, though in a few cases surgical débridement was used as well.^13,15,19 Though eight of nine patients in our series had surgical procedures performed, six patients underwent surgery, at least in part, to achieve or confirm a microbiologic etiology. In the seventh patient the infection was an unexpected finding at a planned stabilization procedure. If the diagnosis of P. acnes vertebral osteomyelitis can be confirmed without extensive surgical intervention via either positive aspirate cultures or transpedicular biopsy, our results in concert with others suggest that patients respond well to antimicrobial therapy alone. It seems likely that antimicrobial therapy alone is appropriate therapy for P. acnes vertebral osteomyelitis in the absence of implants or significant abscess, similar to the approach used when other bacteria cause vertebral osteomyelitis. We typically treat confirmed cases of P. acnes vertebral osteomyelitis with 4 to 6 weeks of parenteral β-lactam antimicrobials at our institution. Ceftriaxone provides a convenient dosing schedule that facilitates outpatient antibiotic administration if the organism is penicillin susceptible. In our series and in a report by Do et al,¹³ surgical débridement alone without prolonged antimicrobial therapy seemed to cure the two patients in which it was attempted.

Each of the patients in our series who suffered relapsed infection had spinal implants placed as part of the surgical treatment of their infection. The placement of foreign material into the infected bed of a patient with P. acnes spine infection may increase the likelihood of treatment failure based upon this small series. Presumably the presence of this foreign material contributed heavily to treatment failure. The delay in diagnosis with a relapsed infection was striking at 2½ and 4 years, highlighting the low virulence but persistent nature of P. acnes musculoskeletal infections in the presence of foreign material.

P. acnes may be a more common cause of native spine vertebral osteomyelitis than has been previously recognized. The indolent yet persistent nature of this anaerobic organism contributes to unique diagnostic challenges. Clinicians should be aware of P. acnes pathogenic potential and propensity to present with atypical findings. Obtaining appropriate anaerobic cultures that are followed for an adequate length of time will enhance the diagnostic yield.