Pyogenic spine infections: variables affecting the treatment and prognosis

Abstract

Nitin Tandon, MD, Kelly Frasier, MD, Dennis G. Vollmer, MD, San Antonio, TX, USAIntroduction: The incidence of primary pyogenic infections of the spine has been steadily increasing over the past decade. However, the characterization and management of these infections remains challenging and is often a subject of debate. In this analysis we seek to determine if characterizing these infections based on their anatomic locus assists or obscures the selection of optimal management strategy; evaluate the relationship of the vertebral level of infection to the likelihood of neurological deficits at presentation; ascertain the safety of the implantation of spinal instrumentation and/or devitalized bone (allograft or autograft) in the presence of infection; evaluate the impact of surgical decompression on neurological outcome and create an algorithm, which would provide an objective means of identifying patients who should be managed with surgical debridement up front rather than a trial of medical management.Methods: One hundred and one sequential admissions for noniatrogenic bacterial infections of the spine in 82 patients, managed between May 1996 and December 2000, were reviewed. Data collected include level of spinal involvement, the locus of the infection (intervertebral disc, vertebral body or epidural space), number of segments involved and their imaging characteristics on magnetic resonance imaging (MRI) and computed tomography, the presence of neurological deficits, kyphotic deformity, relapse of infection (in 18 admissions) and outcomes after surgical management. Infections that occurred in a postoperative setting and infections caused by either mycobacteria or fungi were eliminated from this review. Patient data were analyzed for correlations between spinal level of involvement and the presence of neurological deficit, relapses after instrumentation or decompressive surgery and their neurological condition at 6 months after surgery compared with before surgery (graded using the Frankel score). All patients who presented with new deficits underwent surgical decompression unless this was contraindicated by their medical comorbidities or if they refused operative intervention. Surgery for patients without deficits was reserved to those who failed adequate medical therapy and those with significant or progressive kyphosis on plain films.Results: Thirty admissions were managed surgically, 17 of which involved placement of instrumentation and allograft/autograft after debridement of the nidus of the infection. Eleven cases were managed with decompression alone, and one patient underwent grafting and fusion without instrumentation. No patient who underwent placement of bone grafts and/or hardware experienced a relapse of infection nor needed any longer duration of antibiotic therapy. Of the 27 admissions where patients presented with neurological deficits, the average level of the infection was at T3, whereas a similar average for all admissions was at T10. Patients who underwent decompression experienced an average of 1.5 grade improvement in their Frankel score (p<.001), whereas patients with deficits who were managed medically had no significant improvement. The correlation between the locus of infection (eg, epidural space) and the presence of deficits was poor. Surgery directed at draining epidural collections was unrewarding unless there was a clear loculation of purulent material. This could be detected on the preoperative MRI scan as a nonenhancing component in the epidural space with fluid-like intensity signals on the T2-weighted sequences. Relapse rates correlated with shorter duration of antibiotic therapy and multisegmental osseous involvement.Discussion: Instrumentation and bone grafting in the presence of infection may be safely used if indicated for spinal stabilization. Patients presenting with deficits may experience substantial improvement with surgical decompression. Imaging and intraoperative findings suggest that infections of the spine rapidly traverse anatomic boundaries. The broader term “pyogenic spine infection” with or without neurological deficits is preferable to diagnoses that focus on the locus of the infection. This terminology along with an analysis of number and extent of bony segments involved facilitates the decision to manage a patient medically or surgically. An algorithm that takes these multiple factors into account is under development and will assist in a more objective analysis of these patients.