Editorial Commentary: Pyogenic Vertebral Osteomyelitis and Antimicrobial Therapy: It's Not Just the Length, but Also the Choice

(See the Major Article by Park et al on pages 1262–9.) The infection of a given vertebra and its local spread to contiguous intervertebral disc, epidural space, paravertebral muscles, and soft tissue is known as vertebral osteomyelitis (VO) [1]. Pyogenic bacteria—namely, Staphylococcus aureus, Streptococcus species, and gram-negative bacilli—account for the most frequent etiology and usually reach the bone tissue by the hematogenous route [1, 2]. In contrast with other types of osteomyelitis, in which surgery is mostly needed and failures may be common, a high cure of VO may be achieved in most cases (>90%) with long-term antibiotic treatment and no need for surgery. The length of therapy had been traditionally established by observational studies and expert opinion, the rationale for this being based on issues such as the impaired penetration of antibiotics into the bone tissue, the possibility of bone sequestration, and the development of bacterial biofilm or other adaptive persistent forms of bacteria in this setting. Recently, in a randomized clinical trial including a large cohort of patients, Bernard et al proved that 6 weeks of antibiotics was noninferior to 12 weeks [3]. As a result, current guidelines recommend 6 weeks of treatment for hematogenous VO [4]. However, limitations for shortening treatments have been highlighted based on the existence of abscesses in surrounding tissues, the presence of multidrug-resistant microorganisms, or the impossibility for using an optimal antibiotic therapy [2, 5]. In this issue of Clinical Infectious Diseases, Park et al have challenged the results of that clinical trial by identifying a group of patients with higher risk of relapse, who presented a worse outcome when treated for <8 weeks [6]. In fact, in this observational retrospective multicenter study including 314 cases of microbiologically confirmed hematogenous pyogenic VO, the identification of a patient's profile with a higher risk of failure is probably the most relevant observation. In addition, this study complements and reasserts Bernard et al's trial by also setting a safe 6-week cutoff in the length of treatment for most patients, who presented no risk factors for relapse. In the present study, Park et al identified the presence of end-stage renal disease (ESRD), undrained paravertebral abscesses, and infection caused by methicillin-resistant S. aureus (MRSA) as risk factors for a worse outcome. However, the authors thoroughly discuss whether the existence of ESRD may itself be independently associated with a poor outcome or may reflect the association with the other factors. Thus, the study of Park et al allows analyzing the impact of the presence of abscesses and MRSA etiology on the final outcome, and underlines the importance of prolonged therapies in these difficult-to-treat cases. Indeed, Park et al's cohort included a high proportion of patients with paraspinal abscesses (49%), the presence of which, when undrained or treated shortly with antibiotics, was associated with recurrence in 48% of cases [6]. In contrast, this clinical scenario was not well drawn in Bernard et al's clinical trial, in which the number of patients with abscesses was very low (19%) [3]. A higher difficulty of treatment in the presence of purulent collections seems logical, especially when large abscesses are involved. However, usually there is no need for surgical treatment in VO, even in the presence of abscesses, this commonly being indicated in cases with spinal cord compression or for sepsis control. The majority of cases may be treated conservatively without initial surgery and can be cured using prolonged therapies plus percutaneous drainage [7]. In any case, the study of Park et al emphasizes the need for long-term therapies in this difficult-to-treat scenario, and illustrates that not only the duration of this therapy but also the type of antimicrobial is crucial for final outcome. In fact, the etiology of VO with abscesses is more frequently staphylococcal and, as we discuss below, the presence of a high percentage of MRSA strains and the use of suboptimal therapy with glycopeptides may be related with the high rate of failure reported even when the authors used therapies for >8 weeks (13%). Indeed, the treatment of drug-resistant microorganisms (herein, MRSA strains) and the specific choice of antimicrobials are probably the most illustrative aspects of this observational study. Actually, Park et al's cohort is conditioned by the high proportion of infections caused by MRSA (43% of staphylococcal cases) and reflects the difficulties of treating this microorganism. Almost 70% of MRSA cases were treated exclusively with intravenous antibiotics, which was mainly vancomycin (98% of cases). Sequential oral treatment was prescribed in a minority of these patients, administration of rifampin and/or fluoroquinolones being anecdotal (10%). The use of glycopeptides or lipopeptides for treatment of osteomyelitis has been widely questioned on the basis of several aspects [8, 9]. In particular, vancomycin is considered not to be an ideal drug for osteomyelitis, where it has shown poor results in both the experimental and clinical setting [9–11]. Vancomycin may not penetrate well into bone, it has slower killing ability than other antistaphyloccocal antibiotics (thus increasing the likelihood of vancomycin heteroresistance), and its activity may be impaired in the setting of biofilm-embedded bacteria or the presence of abscesses [11–14]. All these factors may lead to clinical failure when using vancomycin for treating osteomyelitis, the results of this anti-MRSA therapy being clearly worse than that of β-lactams for methicillin-susceptible S. aureus (MSSA) osteomyelitis. In our opinion, the study of Park et al depicts all of these conditions in the setting of VO, especially when S. aureus is the etiologic agent of these abscesses. MRSA are frequently resistant to other families of antimicrobials, this being highly variable according to local epidemiology and geographical clonal distribution. In the case of susceptible strains, rifampin-based combinations may be used and seem to have an impact in the prognosis. The benefits of rifampin have been proved in experimental and clinical studies of osteomyelitis [15–17], and it is remarkable that no patient treated with rifampin in Park et al's study presented with failure [6]. Besides rifampin, fluoroquinolones maintain good activity against biofilm-embedded bacteria, show a good bone penetration, and have an excellent bioavailability [13]. The fluoroquinolone-rifampin combination may be frequently used against MSSA, as shown by Bernard et al with good results [3], but it is not usually an option for MRSA infections. In the latter case, other rifampin-based combinations (ie, cotrimoxazole, fusidic acid, linezolid) may offer better results than monotherapy with vancomycin or the vancomycin-rifampin combination, which may actually not avoid the emergence of rifampin resistance [9]. Overall, the poor results observed among patients with MRSA infection and concomitant abscesses in Park et al's cohort reflect the limitations of suboptimal therapies (namely, unsuccessful abscess drainage and the use of vancomycin). Under these conditions, prolonged therapies (>8 weeks) showed better results—although they still had a considerable failure rate (approximately 9%–13%)—whereas shortening the treatment significantly increased the risk of failure (approximately 30%–40%). In summary, Park et al have identified a subgroup of patients with VO who are at higher risk of failure, and in whom prolonged antibiotic treatments of at least 8 weeks would be more necessary. The presence of concomitant abscesses and infection by multidrug-resistant microorganisms (herein represented by MRSA) encourages using the more appropriate therapy, which includes selected antimicrobials agents and adequate drainage of collections. In addition, this study complements and reasserts Bernard et al's trial by also setting a 6-week cutoff in the length of treatment as effective for most patients with VO, which presented no risk factors for relapse. Acknowledgments. The authors are indebted to Dr Javier Ariza for his critical review of the manuscript. Financial support. J. L.-T. holds a clinical research contract “Sara Borrell” (CD14/00176) from the Instituto de Salud Carlos III, Spanish Ministry of Economy and Competitiveness. Potential conflict of interest. Both authors: No reported conflicts. Both authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.