The Not-So-Good Prognosis of Streptococcal Periprosthetic Joint Infection Managed by Implant Retention: The Results of a Large Multicenter Study Free

Abstract

Periprosthetic joint infection (PJI) is a dreaded complication of joint replacement [1, 2]. Removal of the infected foreign body is the rule for any given device-associated infection. However, debridement, antibiotics, and implant retention (DAIR) may be attempted in some acute cases of PJI [2–4]. When strict selection of patients is followed, the success rate may reach >85% [4–7].

Streptococci are responsible for PJI in 4–12% of cases [8, 9] especially in hematogenous infections [10, 11]. Some studies have suggested that streptococcal PJI may have a more favorable outcome compared with other etiologies [12–14], but this has been contested by others [15]. In fact, the success rate of streptococcal PJI (mostly Streptococcus agalactiae) treated with DAIR varies from 22% to 100%, presumably depending on the selection criteria used [6, 13, 15–18] (Supplementary Table 1). Thus, the real success rate for patients managed by DAIR remains uncertain.

The optimal antimicrobial treatment for streptococcal PJI is also unknown. Current guidelines recommend the use of β-lactams [2, 4], but these antibiotics may have a very high minimal biofilm eradication concentration [19, 20]. The role of alternative compounds with a better antibiofilm profile [21] has not been consistently explored in clinical studies.

Our aim was to analyze the clinical presentations and outcomes of a large cohort of patients with streptococcal PJI managed by DAIR, focusing on the impact of antimicrobial therapy.

METHODS

Setting and Patients

This was a multicenter retrospective study performed in 52 hospitals from 15 nations between 2003 and 2012. Patients were included if they had suffered a PJI that was caused by streptococci and initially managed by DAIR. Eighty-one cases included here have previously been published [6, 15, 22].

PJI was defined according to Infectious Diseases Society of America (IDSA) guidelines as the presence of a sinus tract communicating with the prosthesis, acute inflammation on histologic examination, purulence surrounding the prosthesis, and/or ≥2 evaluable samples yielding the same organism [4]. Polymicrobial cases were also included if streptococci were isolated from the beginning, but we excluded cases of streptococcal superinfection. Microorganisms were identified following standard criteria [23], after samples had been inoculated in liquid and solid media and incubated for ≥7 days. Enterococci, obligate anaerobes (i.e., Peptostreptococcus spp.) or nutritionally variant streptococci (i.e., Abiotrophia spp.) were not included.

PJI was classified as early postoperative, if the symptoms began within the first 3 months after the prosthesis was placed, and late post-surgical, if they started thereafter. The episode was considered acute hematogenous, if it occurred after an uneventful postoperative course and after microbiologically confirmed or clinically suspected streptococcal bacteremia. A contiguous spread was considered, if the PJI occurred in a limb with either infectious cellulitis, or a soft tissue abscess. New radiographical signs of infection were taken as a surrogate marker of chronicity (i.e., periprosthetic radiolucency, bone sclerosis, or osteolytic lesions). Chronic renal failure was defined as a baseline creatinine >150 µmol/L; immunosuppressant therapy was recorded if the patient received, was receiving glucocorticoid, or other immunosuppressant drug therapy.

Data were recorded in a Microsoft-Access database. All cases were critically reviewed by one author (J. L.-T.), and any doubts or inconsistencies were double-checked by the investigator at each hospital.

Clinical and Surgical Management

DAIR has been described elsewhere [2, 3, 24]. Briefly, it comprises thorough surgical debridement of all purulent collections and necrotic tissues surrounding the prosthesis. Mobile parts of the device (i.e., the polyethylene liner) are exchanged if feasible. DAIR is recommended in patients who meet the criteria proposed by the IDSA guidelines [4]. Patients with early postoperative (<1 month) or acute hematogenous PJI with ≤3 weeks of symptoms qualify for DAIR if they have a soundly fixed prosthesis, good periprosthetic soft tissues condition, and antibiotics are available with a reasonable activity against biofilm-embedded bacteria. In the present study, these criteria were not strictly met by many patients, and the decision to undergo DAIR was taken by individual medical group on a case by case basis.

Outcome and Follow-up

Patients were followed until death, treatment failure, removal or replacement of the prosthesis, or until loss to follow-up. Overall Failure was the primary endpoint and was considered in cases of: (12) death related to the infection;64(ii) need for salvage therapy to control the infection, including supplementary surgical debridements >30 days after the first debridement, prosthesis removal (due to any cause during the first year after debridement, or due to streptococcal persistence or relapse, or superinfection by other microorganisms), or the need for supplementary courses of antibiotics beyond the initially scheduled treatment (including chronic suppressive antimicrobial therapy); and/or (iii) persistent signs of infection at the last visit or follow-up appointment.

Given the retrospective nature of this study, and to avoid a survivor bias when analyzing the impact of antimicrobial therapy, several failure dynamics were studied:

• Early Failure was considered to have occurred in patients who met the failure criteria within the first 30 days after surgical debridement.

• Late Failure was considered to have occurred in patients who met the failure criteria beyond the first 30 days after debridement but who were still under antimicrobial therapy. In this group, only antimicrobials received during the first 30 days were analyzed.

• Failure after Therapy was considered to have occurred in patients who met the failure criteria once they had finished the scheduled therapy. In this analysis, the antibiotics received throughout treatment were included.

Statistical Analysis

Categorical parameters were compared with the χ² test or Fisher exact test, and continuous variables were compared with the Mann–Whitney U test or Kruskal–Wallis test. Parameters associated with Overall Failure, Late Failure, and Failure after Therapy were identified by Kaplan–Meier curves (long-rank test), univariate, and multivariate Cox regression. For the analysis of Early Failure, logistic regression were performed. All analyses were 2-tailed, and a P value < .05 was considered statistically significant.

RESULTS

Description of the Series

Overall, 922 cases of PJI were recorded, of which 92 (10.0%) were excluded for various reasons, leaving a cohort of 830 cases. We initially managed 462 (55.7%) by DAIR, and these cases were used as the focus of this analysis (Supplementary Figure 1).

The median age was 72 years (interquartile range [IQR], 65–78 years), and 50% were men. The most frequent type of PJI was hematogenous (52%), which occurred more frequently in men, in patients with malignancy and in those with knee prostheses. Patients with hematogenous PJI more frequently presented with bacteremia and elevated temperature, along with higher leukocyte counts and C-reactive protein (CRP) levels (Table 1).

The most frequent species was S. agalactiae (159 cases [34.4%]) (Table 2). There were 63 (14%) polymicrobial infections that were typically postoperative (83%), presented less frequently with fever (51% vs. 68%, P = .007) and more frequently with a sinus tract (34% vs. 10%, P < .001), and had lower CRP levels (80 mg/L [IQR 41–150] vs. 202 mg/L [IQR 110–291], P < .001).

Baseline features, clinical presentation, and management were similar among the streptococcal species (Supplementary Table 2). Exceptions to this were the higher rate of patients with rheumatoid arthritis among episodes caused by S. pyogenes, and the higher rate of chronic lung disease and malignancy in PJI due to S. pneumoniae. Pneumococcal PJI was also more frequently hematogenous, occurred more frequently with knee prostheses, and presented with a higher leukocyte count. Penicillin minimum inhibitory concentration (MIC) was >0.125 mg/L in 24/425 cases (6%).

DAIR Management

Patients underwent debridement after a median of 5 days (IQR 2–13) from the onset of symptoms. Removable components were exchanged in 53% of cases, this being highly variable across participating centers (Supplementary Figure 2). The median number of different antimicrobial classes prescribed per patient was 2 (range 1–6). Patients were usually treated with β-lactams, which were given intravenously for a mean time of 21 days ± 20 days. Rifampin-based combinations were significantly used (i.e., during >21 days) in 37% of patients, but this fraction was also highly variable across the participating hospitals (in those recruiting >10 patients, it ranged from 18% to 88%) (Supplementary Figure 2). Alternative antimicrobials such as fluoroquinolones, clindamycin, or linezolid were used less often (Supplementary Table 3). In patients not failing while on treatment, antimicrobial therapy was continued for a median of 91 days (IQR, 58–171 days).

Outcome

The primary endpoint was evaluable in 444 patients (96.1%). Overall Failure occurred in 187 patients (42.1%; 95% confidence interval [CI], 37.5%–46.7%) after a median of 62 days from debridement (IQR, 25–160 days); by contrast, 257 patients (57.1%) did not fail and were followed up for a median of 802 days (IQR, 507–1339 days) (Figure 1A). Success rates were highly variable among the participating centers (Supplementary Figure 2), with it ranging from 44% to 91% among hospitals recruiting >10 patients.

Figure 1.

Kaplan-Meier curves of patients with streptococcal periprosthetic joint infection according to the criteria for indicating debridement and implant retention. A, Kaplan-Meier curve of all evaluable patients (n = 444, 187 failures). Causes of failure were due to the streptococcal infection in 147 cases (79%), the other reasons being prosthesis removal due to orthopedic causes (15 patients [8%]), and superinfection by other microorganisms (25 cases [13%]). Death related to periprosthetic joint infection was observed in 11 cases (2%). B, Black continuous line: patients meeting IDSA criteria for DAIR (see text): 81 failures in 221 episodes of infection; grey dotted line: patients not meeting IDSA criteria for DAIR: 106 failures in 223 episodes of infection; long-rank test, P = .017. Reasons for not fulfilling the IDSA criteria were (more than 1 motive per patient is possible): in 67 patients (30%) symptoms duration was longer than 21 days; 90 patients (40%) had a post-surgical infection with symptoms beginning beyond the first month after the placement of the prosthesis; 61 patients (27%) presented with a sinus tract; and in 80 cases (36%) there were radiographic signs of prosthesis loosening and/or chronic infection. C, Post-surgical cases (i.e., non-hematogenous cases) (n = 189, 82 failures): black continuous line: cases with symptoms beginning within the first 30 days after the placement of the prosthesis (n = 78, 25 failures); grey continuous line: cases with symptoms beginning within 31 and 90 days after the placement of the prosthesis (n = 41, 13 failures); black dotted line: cases with symptoms beginning beyond 90 days after the placement of the prosthesis (n = 70, 44 failures). Long-rank test, P < .001. Abbreviations: DAIR, debridement, antibiotics, and implant retention; IDSA, Infectious Diseases Society of America.

Open in new tabDownload slide

Independent predictors of a poor outcome were rheumatoid arthritis (hazard ratio [HR], 2.36), late post-surgical infection (HR, 2.20), and bacteremia (HR, 1.69). The exchange of removable components was independently associated with a favorable outcome (HR, 0.60) (Table 3). No one streptococcal species was associated with a higher likelihood of Overall Failure, although a nonsignificant better prognosis was observed for S. pneumoniae (24% failure). A high penicillin MIC (>0.125 mg/L) was also not associated with failure. Also, polymicrobial cases were not associated with a higher likelihood of failure, even when S. aureus was involved (data not shown).

Late post-surgical infection was indeed a predictor of bad prognosis, when defined as onset of symptoms beginning >3 months after the prosthesis placement (Figure 1C). Cases with symptoms beginning within the first and third month had a similar prognosis to that of cases with symptoms beginning within the first month after prosthesis placement. No relevant differences were observed in these 2 groups of patients (data not shown).

The failure rate was higher in patients not fulfilling the IDSA criteria for DAIR, namely, 106/223 (48%) versus 81/221 (37%) (long-rank test, P = .017) (Figure 1B). Again, indication of DAIR according to the IDSA criteria was highly variable among participating centers (Supplementary Figure 2), ranging from 33% to 83% in those recruiting >10 patients. Independent predictors of failure among patients meeting the IDSA criteria were rheumatoid arthritis (HR, 2.46 [95% CI, 1.34–4.53]), bacteremia (HR, 1.92 [95% CI, 1.22–3.02]), and male sex (HR, 1.85 [95% CI, 1.18–2.91]). Interestingly, the exchange of removable components during debridement was especially beneficial in patients not meeting the IDSA criteria (37% failures vs. 62%, P < .001), in comparison with patients fulfilling them (failures 33% vs. 39%, P = .286).

Failure Dynamics and Antimicrobial Therapy

Among the 187 patients who failed, 55 (29%) developed Early Failure, 71 (38%) developed Late Failure, and 61 developed Failure after Therapy (33%). Variables independently associated with Early Failure were age, rheumatoid arthritis, late post-surgical infection, bacteremia, and infection by S. pyogenes (Table 4).