*Article* **Intra-Articular Injections Prior to Total Knee Arthroplasty Do Not Increase the Risk of Periprosthetic Joint Infection: A Prospective Cohort Study**

**Jérôme Grondin 1,2, Pierre Menu 1,2,3,4, Benoit Métayer 5, Vincent Crenn 6,7, Marc Dauty 1,2,3,4 and Alban Fouasson-Chailloux 1,2,3,4,\*,†**


**Abstract:** Periprosthetic joint infections (PJI) occur in 0.5 to 2.8% of total knee arthroplasties (TKA) and expose them to an increase of morbidity and mortality. TKA are mainly performed after failure of non-surgical management of knee osteoarthritis, which frequently includes intra-articular injections of corticosteroids or hyaluronic acid. Concerning the potential impact of intra-articular injections on TKA infection, literature provides a low level of evidence because of the retrospective design of the studies and their contradictory results. In this prospective cohort study, we included patients after a total knee arthroplasty, at the time of their admission in a rehabilitation center, and we excluded patients with any prior knee surgery. 304 patients were included. Mean follow-up was 24.9 months, and incidence proportion of PJI was 2.6%. After multivariate logistic regression, male was the only significant risk factor of PJI (OR = 19.6; *p* = 0.006). The incidence of PJI did not differ between patients who received prior intra-articular injections and others, especially regarding injections in the last 6 months before surgery. The use of intra-articular injection remains a valid therapeutic option in the management of knee osteoarthritis, and a TKA could still be discussed.

**Keywords:** knee; total knee arthroplasty; infection; intra-articular injection

#### **1. Introduction**

Periprosthetic Joint Infection (PJI) constitutes one of the most feared complications after total knee arthroplasties (TKA) [1]. PJI increases mortality, with a 71.7% overall survival five years after PJI diagnosis [2] and exposes them to the complications of challenging surgical and medical treatments [3–5]. It also reduces physical function and impairs quality of life [6,7]. Its incidence ranges from 0.5 to 2.8% according to the studies [8–10]. TKA is a frequent surgical procedure, increasing in number every year [11]. There is a great concern about prevention of PJI, and different recommendations have been published [12,13]. Yet, despite these recommendations, the rate of PJI apparently does not decrease over time [2].

TKA improves primary outcomes of knee osteoarthritis (KOA) such as pain and function [14], and is mainly performed after failure of medical treatment. Intra-articular

**Citation:** Grondin, J.; Menu, P.; Métayer, B.; Crenn, V.; Dauty, M.; Fouasson-Chailloux, A. Intra-Articular Injections Prior to Total Knee Arthroplasty Do Not Increase the Risk of Periprosthetic Joint Infection: A Prospective Cohort Study. *Antibiotics* **2021**, *10*, 330. https://doi.org/10.3390/ antibiotics10030330

Academic Editor: Jaime Esteban

Received: 3 March 2021 Accepted: 19 March 2021 Published: 21 March 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

injection remains an usual treatment of non-surgical KOA in the absence of absolute contraindications such as infectious arthritis and drug hypersensitivity [15], but guidelines are contradictory regarding its efficiency and safety [16,17]. During the procedure of intraarticular injections, a contamination of the joint may happen and potentially induce a PJI if an arthroplasty is secondarily performed [1]. In 2017, the Centers for Disease Control and Prevention (CDC) broached the topic, but the issue was considered unresolved, and no recommendation was made [13]. In clinical practice, intra-articular infiltrations of corticosteroids (CS) or hyaluronic acid (HA) are frequently performed [18], and around 30% of the patients who underwent TKA had previously had an intra-articular steroid injection [19]. In this context, many studies have been performed, but have provided a low level of evidence because of their retrospective design and contradictory results [10,19–25] (Supplementary Materials Table S1). Among them, three studies based on large databases have highlighted an increased risk of PJI if prior intra-articular injections had been performed in the few months preceding the surgery [10,19,20], but they were exposed to common limitations with large database studies. A few meta-analyses were performed on PJI after TKA or Total Hip Arthroplasty (THA) [18,26–28], also with contradictory results, and emphasizing the low level of evidence of available studies and the need for prospective trials.

Thus, we aimed to prospectively assess the impact of prior intra-articular injections on the occurrence of periprosthetic joint infection after TKA.

#### **2. Results**

Between January 2016 and May 2019, 304 patients were included, and 279 (91.8%) eventually followed, while 25 patients (8.2%) were lost to follow-up (Figure 1). Mean follow-up was 24.9 months ± 3.8.

#### **Figure 1.** Flow-chart.

Most of the patients were females (72.4%; *n* = 220), and mean age was 71.8 years ± 8.9. Mean body mass index (BMI) was 30.9 kg/m<sup>2</sup> ± 5.3 and 85.5% (*<sup>n</sup>* = 260) of the patients were overweight (BMI > 25) or obese (BMI > 30) at the time of the surgery (35.8% overweight (*n* = 109), 49.7% obese (*n* = 151)) (Table 1). Mean American Society of Anesthesiologists (ASA) score was 2.3 ± 0.6. Two patients were deceased 5 and 7 months after the arthroplasty (1 heart failure due to myocardial ischemia, and 1 cerebral stroke). 68.1% (*n* = 207) of the

patients received infiltration before surgery, 48.8% (*n* = 101) of them with hyaluronic acid alone, 15.5% (*n* = 32) with corticosteroids, and 24.6% (*n* = 51) received both.

**Table 1.** Demographic characteristics.


SD: Standard-deviation; BMI: Body mass index; ASA: American society of anesthesiologists; IA: Intra-articular; CS: Corticosteroids; HA: Hyaluronic acid.

Table 2 summarizes the cases of PJI, mainly males (6 out of 8). Most of the infections (7/8) occurred in the first 6 weeks following arthroplasty and were caused by Staphylococcus aureus (6/8) or Staphylococcus capitis (1/8). The remaining case concerns a patient who initially received a surgery consisting of irrigation and debridement in a context of infectious endocarditis due to a persistent PJI, and a one-stage exchange was secondly performed. One patient died from myocardial ischemia 5 months after diagnosis of PJI. Other surgical and medical strategies performed were all considered successful, and no additional surgery was necessary.



**Table2.**Casesofperiprostheticjointinfections.

Infections;

 Methi-S: Methicillin

 sensitive; Methi-R: Methicillin

 resistant, Staph.:

Staphylococcus;

 Strep.:

Streptococcus.

The overall incidence of infection was 2.6% (8/304). Comparisons of incidence of PJI were completed with Fisher's exact test depending on the "injection" status. Incidence was 2.1% (2/97) in patients without prior injection, and 2.9% (6/207) if any prior intra-articular injection had been performed, OR = 1.42 (CI 95% = 0.28–7.16; *p* = 0.67). It increased to 7.1% (3/42) if injection had been performed within 6 months before surgery, OR = 3.95, but without statistical significance (CI 95% 0.91–17.21; *p* = 0.08).

In univariate regression, the "sex" variable was the only one to be significantly associated with PJI, with an increased risk of infection in males. A trend was found concerning "injection < 6 months" with an OR of 3.46 (*p* = 0.09) (Table 3). Based on these findings, we have investigated potential differences between males and females that could explain the increased risk of PJI in males (Table 4). Therefore, we have highlighted significant differences between the two groups: Smoking, diabetes, alcoholism, and ASA score were significantly higher in males than in females.


**Table 3.** Univariate logistic regression according to patients' characteristics.

CI: Confidence interval; BMI: Body mass index; ASA: American society of anesthesiologists.

**Table 4.** Comparison between males and females.


<sup>a</sup> *t*-test; <sup>b</sup> χ<sup>2</sup> -test. SD: Standard-deviation; BMI: Body mass index; ASA: American society of anesthesiologists.

Multivariate logistic regressions were performed considering differences between males and females. In the total population, only sex was significantly associated with occurrence of infection (OR = 19.6; CI95%: 2.4–164; *p* = 0.006). Knowing existing differences between males and females in our population, we performed multivariate logistic regressions analyzing these 2 groups separately: No factor was significantly associated with PJI occurrence.

#### **3. Discussion**

In this study, the risk of PJI did not significantly increase between patients who had previously received knee infiltration and patients who had not [OR = 1.42 (CI 95% = 0.28–7.16; *p* = 0.67)]. Many studies have been performed concerning the safety of intra-articular infiltrations in the pre-operative period, with a retrospective design and conflicting results [10,19–25]. Four of these studies did not bring out significant associations. However, 3 studies based on large retrospective databases suggested an increased risk of PJI in patients who had received an infiltration in the 3 months preceding surgery [10,19] or even in the preceding 7 months [20]. These findings explain why we compared the oc-

currence of infection between patients who had received an infiltration in the 6 months preceding surgery to the others. There was no significant difference, but a trend toward an increased risk in patients who had received an infiltration in the 6 months preceding surgery (OR = 3.95; CI 95% 0.91–17.21; *p* = 0.08). As discussed below, this trend requires further investigation with larger cohorts in prospective studies. Thus, special attention should be paid to the benefit/risk assessment of a knee infiltration if a surgery is to be scheduled in the next months.

In previous retrospective studies based on large prospective databases, confounding factors such as male sex, BMI, tobacco smoking, prior surgery, and inflammatory arthritis may have been involved in the significant association reported between PJI and prior intra-articular injections [10,19,20]. As recommended in previous systematic reviews [28], we clearly excluded patients with major risk factors of infection: Any prior surgery or septic arthritis of the knee, history of rheumatoid arthritis or hemophilia, and immunosuppressive or immunomodulatory drugs. We also adjusted the results on potential confounding factors previously reported: Male sex, age < 60 years, BMI > 25 kg/m2, diabetes, previous or current tobacco smoking, ASA ≥ 3 [29–32]. In our cohort, male sex was the unique risk factor associated with infection. Smoking, diabetes, alcoholism, and ASA score were significantly higher for males than females. In multivariate logistic regression, excluding male sex, no factor was significantly associated with PJI. Further analysis focusing on male population did not bring out significant results, especially regarding prior intra-articular injection in the 6 months preceding surgery.

Thus, despite conflicting evidence regarding the potential association between PJI and pre-operative joint injection, some pathophysiological hypotheses were suggested: An infectious risk due to the prolonged immunosuppressive effect of glucocorticoids injected [10,24,33], or direct inoculation from the infiltration procedure due to insufficient sterile precautions [10,24]. To investigate these hypotheses, we planned a 24 month-followup. Indeed, the first 2 years are the greatest risk period and represent 60 to 70% of PJI [11,34], and studies with shorter follow-ups have reported lower incidence of infection [8]. Furthermore, early (<3 months) and delayed infections (between 3 and 24 months after surgery) are often exogenous, early infections caused by more virulent organisms than delayed ones; whereas late-onset infections (>24 months) are frequently due to hematogenous infection [1,11], except in cases of very indolent infections due to very low-virulent bacteria [11]. These pathophysiological hypotheses are unlikely to explain late-onset hematogenous infections, which is why we did not follow patients for more than 2 years. In our cohort, every infection occurred within 6 months after surgery, most of them within 6 weeks, consistent with the hypotheses of exogenous pathogenesis.

We selected a telephone follow-up, for it produces higher response rates than postal survey or mail/internet surveys [35–37]. However, the telephone mode brings more positive responses to subjective items than other modes [37], but this bias does not apply in our case, since the interview was closed-ended to detect the occurrence of PJI. A memorization bias may be suggested in principle, but patients would unlikely forget a Periprosthetic Joint Infection with its devastating consequences, revision surgeries, and extended antibiotic therapy.

This study has limitations. Indeed, our cohort was formed with patients admitted in a Physical and Rehabilitation Medicine Hospital, which are usually different from those discharged home directly after surgery: They are usually older, with a higher BMI, and are more frequently females [38]. In our cohort as well, patients were mostly females (72.4%) with a mean BMI of 30.9. Periprosthetic joint infections are usually estimated between 1 and 2% after TKA [30], but may range over 2% [9,10]. In this study, the incidence proportion was 2.6%, which seems consistent with literature knowing that we included more fragile patients. The main limitation was the size of the cohort (around 300 patients), which may have reduced its ability to detect a statistically significant association between intra-articular infiltrations and PJI. However, at the beginning of the study, we calculated that 276 patients were required to detect a doubling of the incidence of infection. Therefore, we included 304 patients and eventually followed 279 of them, but our initial projection may be challenged. The number of patients needed to improve the power and allowing recommendations depends on the incidence of PJI in the population, on the difference in incidence proportion that we aimed to detect, and on the pre-specified power (usually 80%). Further prospective studies should be performed in larger cohorts to clearly establish the safety of intra-articular injections before total knee arthroplasty, in order to improve sensitivity and power. Yet, a single institution is unlikely to sustain such a study. A multicentric study proves to be necessary [28], but exposes us to specific bias of multicentric designs, such as unrecognized heterogeneity across centers [39].

Another limitation is the diagnosis and classification of PJI which are challenging and not consensual [40]. In this study, we used the definition of the International Consensus Meeting on Periprosthetic Joint Infection, and every case fulfilled at least one of the two major criteria of PJI [41,42]. Different classifications exist, mainly based on timing of clinical presentation leading to different surgical strategies [40,43], and therefore decreasing comparability between studies.

Finally, the telephone follow-up might have failed to detect some PJI signs, especially in case of indolent infections due to low-virulent bacteria, which usually provide few clinical manifestations. Indeed, clinical, radiological, and biological parameters may have been more sensitive.

#### **4. Materials and Methods**

#### *4.1. Participants*

Patients were included in the few days following the arthroplasty (2 to 7 days), at the time of their admission in the rehabilitation center. Surgery was performed in the University Hospital of Nantes or in other clinics of Nantes' region, France. Inclusion criteria were: Age > 18 years old, patients hospitalized for rehabilitation after TKA. Exclusion criteria were: Any prior ipsilateral knee surgery, any prior infectious arthritis of the knee, history of rheumatoid arthritis or hemophilia, and immunosuppressive or immunomodulatory drugs.

At the time of the inclusion, we systematically collected the following data: Age, sex, weight, height, BMI, diabetes, tobacco smoking, alcoholism, ASA Score, other significant medical and surgical antecedents, date and place of surgery, prior intra-articular infiltration of the knee: Number, date, and type of medication injected.

#### *4.2. Outcome*

The primary outcome was the incidence of PJI. Every case of infection was reviewed and defined as a PJI if it fulfilled the definition provided by the International Consensus Meeting on Periprosthetic Joint Infection (at least one of the two major criteria: Two positive growths of the same organism using standard culture methods, or sinus tract with evidence of communication to the joint or visualization of the prosthesis) [41,42]. First, we compared incidences of infection between patients who had received prior intra-articular injections and others, and then we focused on patients who had had an intra-articular injection in the 6 months preceding surgery.

#### *4.3. Follow-Up*

Follow-up was performed at 24 months after surgery. A phone call was performed, and occurrences of an infection or an additional surgery were checked based on following questions: "Do you feel any persistent knee pain, erythema and oedema?", "Have you noticed any wound drainage?", "Has a diagnosis of prosthetic infection or any infection of your knee been established? "Have you got any additional surgery?". If any of these occurred, medical, surgical, and bacteriological reports were gathered. If a patient was not able to answer these questions, his general practitioner was called.

#### *4.4. Statistical Analyses*

Statistical analyses were performed using software SPSS 23.0 IBM Corp, Armonk, NY, USA. Comparisons of incidence proportions of PJI were performed with a Fisher's exact test. Logistic regressions were performed with PJI as dichotomous dependent variable, and independent variables were sex, age, BMI, ASA, diabetes, smoking, alcoholism, prior infiltration (Yes/No), infiltration < 6 months (Yes/No). First, we analyzed the association between dependent and independent variables in univariate regression, and then we performed a multivariate logistic regression with forward selection (Wald). We compared demographic characteristics between males and females using *t*-test. *p* < 0.05 was considered significant. To evaluate the number of subjects required, we defined a power of 80%, an alpha risk of 5%, a theoretical incidence of PJI of 2.8% [10], and aimed to detect a doubling of the incidence. We calculated that 276 patients were required.

#### *4.5. Ethics*

This research was conducted in our institution from January 2016 to June 2019. Due to the non-interventional nature of the study, no ethics committee was necessary at the time of the beginning of the study. Yet, necessary processes were performed with the "Direction de la Recherche Clinique" (DRC) of the University Hospital of Nantes, France, and the "Commission Nationale de l'Informatique et des Libertés" (CNIL); the study was registered under the number RC16\_0039. The database was anonymized, and all the patients provided their verbal consent and got an information document.

#### **5. Conclusions**

This study showed no evidence of the causality of prior intra-articular injections in Periprosthetic Joint Infection occurrence, even in the 6 months preceding surgery. In clinical practice, wise use of intra-articular injection remains a valid therapeutic option in the management of knee osteoarthritis, and a total knee replacement could still be discussed.

**Supplementary Materials:** The following are available online at https://www.mdpi.com/2079-638 2/10/3/330/s1, Table S1: Literature review of the impact of intraarticular injection before surgery on the occurrence of infection.

**Author Contributions:** Conceptualization, A.F.-C., M.D.; methodology, J.G., A.F.-C. and M.D.; Formal analysis, J.G., A.F.-C. and M.D.; data curation, J.G., A.F.-C., P.M., B.M., V.C. and M.D.; investigation, J.G., A.F.-C., P.M., B.M., V.C. and M.D.; writing—original draft preparation, J.G., A.F.-C. and M.D.; writing—review and editing, J.G., A.F.-C., P.M., B.M., V.C. and M.D. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and due to the non-interventional nature of the study, no ethics committee was necessary at the time of the beginning of the study. The study was declared to the "Direction de la Recherche Clinique" (DRC) of the University Hospital of Nantes, France, and was registered under the number RC16\_0039. Necessary processes were performed with the "Commission Nationale de l'Informatique et des Libertés" (CNIL).

**Informed Consent Statement:** All the patients gave their oral consent and got an information document. All data were anonymized.

**Data Availability Statement:** The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

**Acknowledgments:** We would like to thank Annie Chailloux for proofreading the manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


## *Perspective* **Controversy about the Role of Rifampin in Biofilm Infections: Is It Justified?**

**Nora Renz 1,2, Andrej Trampuz 1,\* and Werner Zimmerli <sup>3</sup>**


**Abstract:** Rifampin is a potent antibiotic against staphylococcal implant-associated infections. In the absence of implants, current data suggest against the use of rifampin combinations. In the past decades, abundant preclinical and clinical evidence has accumulated supporting its role in biofilm-related infections.In the present article, experimental data from animal models of foreignbody infections and clinical trials are reviewed. The risk for emergence of rifampin resistance and multiple drug interactions are emphasized. A recent randomized controlled trial (RCT) showing no beneficial effect of rifampin in patients with acute staphylococcal periprosthetic joint infection treated with prosthesis retention is critically reviewed and data interpreted. Given the existing strong evidence demonstrating the benefit of rifampin, the conduction of an adequately powered RCT with appropriate definitions and interventions would probably not comply with ethical standards.

**Keywords:** rifampin; biofilm; prosthetic joint infection

**Citation:** Renz, N.; Trampuz, A.; Zimmerli, W. Controversy about the Role of Rifampin in Biofilm Infections: Is It Justified? *Antibiotics* **2021**, *10*, 165. https://doi.org/10.3390/ antibiotics10020165

Academic Editor: Sigrun Eick Received: 17 January 2021 Accepted: 3 February 2021 Published: 5 February 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Introduction**

Rifampin is one of the first-line drugs against tuberculosis. In addition, it has been used against non-mycobacterial microorganisms, mainly staphylococci, for at least 50 years [1]. However, its place in severe staphylococcal infections not involving an implanted device remained unclear for decades because no systematic comparative studies had been performed. In the meantime, few studies have been published on this topic. In five randomized controlled trials and two retrospective cohort studies in patients with *Staphylococcus aureus* bacteremia, no difference of mortality could be shown [2]. A recent multicenter, randomized, double-blind placebo-controlled trial confirmed these data in 758 patients [3]. In the study of Rieg et al. [4], only the subgroup of patients with implants had less late complications related to *S. aureus* bacteremia when treated with combination therapy (4.5% vs. 10.6%, *p* = 0.03). Most of them were treated with a rifampin combination regimen, suggesting a benefit of antibiofilm activity compared to treatment without rifampin. In contrast, the addition of rifampin to standard therapy showed no advantage in patients with native valve infective endocarditis caused by *S. aureus* [5]. Thus, the latest data advocate against the uncritical use of rifampin combination therapy in patients with severe staphylococcal infections in absence of implants.

In contrast, the benefit of rifampin in patients with staphylococcal implant-associated infection is well documented based on abundant in-vitro, animal, and clinical data, as summarized in a recent review [6]. Until recently, only one randomized controlled trial (RCT) existed, in which the added value of rifampin was shown in patients with orthopedic implant-associated staphylococcal infections [7]. In 2020, a second RCT in patients with periprosthetic joint infection (PJI) was published, using different combination therapy regimens, which did not show a better outcome with addition of rifampin to standard treatment [8]. These unexpected data may unsettle clinicians with limited experience in the field of implant-associated infections. Therefore, possible reasons for the failure of demonstrating the benefit of adding rifampin in this trial will be discussed herein in the light of available evidence, including animal data and clinical trials.

#### **2. Short History of Rifampin Use in Patients with Implant-Associated Staphylococcal Infection**

In 1982, the use of rifampin in the treatment of non-tuberculous infections has been initially presented in a large symposium, followed by the publication in a supplemental edition of the Reviews of Infectious Diseases, edited by Merle A. Sande [9]. The special interest in rifampin was based on its unique mode of action, i.e., its inactivation of the bacterial DNA-dependent RNA polymerase. Its main drawback is the single-step mutation of the rifampin-binding enzyme occurring with a frequency of 10−<sup>6</sup> to 10−<sup>7</sup> [10]. This high risk of emergence of resistance explains its occasional failure in infections characterized by a high bacterial load, such as in infective endocarditis or persistent *S. aureus* bacteremia [5,11,12]. Studies of rifampin in non-mycobacterial infection were retarded by the fear that its widespread use could result in resistance to rifampin in *Mycobacterium tuberculosis*.

One of the first observations of the successful use of rifampin combination therapy in implant-associated infections is the report of two patients with *S. epidermidis* infection, one with prosthetic valve endocarditis and the other with ventriculoperitoneal shuntassociated infection [13]. In a case series, Karchmer et al. [14] reported a good outcome with a vancomycin-rifampin, but not betalactam–rifampin combination (87% vs. 43%, *p* = 0.025) for treatment of prosthetic valve endocarditis caused by methicillin-resistant *S. epidermidis*. These data suggest that the combination partner of rifampin matters.

Based on our observation that rifampin could not only prevent, but also cure experimental staphylococcal implant-associated infections [15], we performed additional animal experiments with rifampin combination therapy [16], followed by observational studies and one randomized controlled trial in patients with orthopedic implant-associated infections [7,17–19]. Later, rifampin combination therapy has shown to improve the outcome in patients with other types of implant-associated infections such as staphylococcal prosthetic valve endocarditis [14,20], deep sternal wound infections [21] and vascular graft associated infections [22,23]. However, data from randomized controlled trials are still not available in patients with non-orthopedic implant-associated infections.

#### **3. Evidence for the Efficacy of Rifampin in Animal Studies**

The first observation of the biofilm activity of rifampin has been made >35 years ago in the guinea pig tissue cage model [15]. With four doses of rifampin, implant-associated *S. aureus* infection could be cured in 100% of the tissue cages, if therapy was started up to 12 h after inoculation. If the delay was prolonged to 24 h, the cure rate decreased to 57%. These results unequivocally demonstrate that rifampin is able to eliminate surfaceadhering biofilm staphylococci. However, it also shows that the efficacy of a short-term therapy is limited to a young biofilm. A clear definition of the limit between young and mature (tolerant) biofilm is still lacking. It depends on the microorganism, the antibiotic, and the duration of therapy [24]. Table 1 summarizes several experimental studies with the subcutaneous tissue cage animal model in guinea pigs. In each experimental series, rifampin combinations were significantly more active than other antibiotics [16,25–30]. This animal model does not simulate orthopedic device-related infection. However, it allows following an ongoing infection with the most relevant endpoint, namely complete elimination of the biofilm. Other groups investigated the role of rifampin in animal models of implant-associated osteomyelitis, and corroborated the antibiofilm effect of rifampin, as summarized in a recent review [6].


**Table 1.** Cure rate in the guinea pig tissue cage infection model (copyright© American Society for Microbiology, Antimicrob Agents Chemother 63(2), e01746-18, 2019 [6]).

<sup>a</sup> Fisher's exact test for categorical variables, statistical significance is defined as *p* < 0.05. <sup>b</sup> ABI-0043 is a derivative of Rifalazil, which is a rifamycin derivative.

#### **4. Role of Rifampin in Clinical Studies Involving Orthopedic Implant-Associated Infections**

Based on the animal data showing an impressive antibiofilm activity of rifampin against staphylococci, we started to treat patients with orthopedic device-related infection (ODRI) with rifampin combination in clinical routine. In a first case series, 10 patients with staphylococcal ODRI undergoing debridement and implant retention (DAIR), the success rate was 80% [17]. In this and many subsequent studies, no direct comparison is possible, because either none or all patients were treated with rifampin combinations. In patients treated with DAIR without rifampin combination therapy, the success rates were as low as 31% to 35% [31,32]. However, in these studies, the Infectious Diseases Society of America (IDSA) guidelines regarding the indication for DAIR have not been considered [33].

In the study of Holmberg et al. [34], patients with staphylococcal knee PJI had a better failure-free survival, when treated with a rifampin combination than without rifampin (81% vs. 41%, *p* = 0.01). Similarly, in a study from the Mayo Clinic, patients treated with DAIR according to the IDSA-guidelines including a rifampin-regimen had a better outcome than patients in a historical control group treated without rifampin (93% vs. 63%) [35]. However, in this study, most of the patients received long-term suppressive antimicrobial therapy.

In several studies, all patients undergoing DAIR for staphylococcal PJI were treated with a rifampin-regimen. The failure-free survival ranged between 80% and 100% in patients treated according to the IDSA-guidelines, in whom the rifampin combination could be given for a prolonged time (generally >2 months) [36–43]. In a study, in which 29 patients with acute PJI were treated with ciprofloxacin plus rifampin, the success rate was 83% [39]. Interestingly, in the mentioned Norwegian randomized trial, in which rifampin-combination therapy did not show superiority, another regimen has been used, namely cloxacillin or vancomycin with or without rifampin [8]. Possible reasons for the low success rates and the lack of improvement by the addition of rifampin are presented below. Indeed, diligent choice of antimicrobial agents may be crucial. In the observational study of Puhto et al. [44] in patients with PJI treated with DAIR, treatment success was

significantly higher in patients with ciprofloxacin/rifampin as compared to those with another combination partner or a regimen without rifampin.

Despite the overwhelming evidence for the antibiofilm activity of rifampin, there are a few studies, in which no beneficial effect of rifampin was shown. Bouaziz et al. [45] showed that non-compliance with IDSA guidelines was a risk factor for treatment failure in patients with hip or knee PJI. However, rifampin as single factor was not advantageous because of the strong association between surgical therapy and outcome. Thus, rifampin combination therapy should only be used in patients qualifying for DAIR [33,46]. In an observational study of patients with acute PJI treated with DAIR and linezolid with or without rifampin, patients receiving rifampin did not have an improved outcome. The confounder in this study may be the high prevalence of polymicrobial infection in both groups (41% and 35%, respectively) indicating that many patients may have had wound healing disturbance or even a sinus tract during therapy [47].

Rifampin long-term therapy is complicated by its frequent gastrointestinal side effects, and its strong induction of isoenzymes of cytochrome P450 [6,10]. This is a major clinical challenge, as the effect of rifampin can only be considered in patients in whom it can be given for a sufficient duration. Enzyme induction by rifampin leading to drug-drug interactions requires specific attention prior to and at the end of treatment. However, the interaction of rifampin and other antibiotics in vitro is difficult to interpret, because synergism/antagonism in vitro does not correlate with the effect in vivo [48]. Based on experimental data, the antibiofilm effect seems to be a class effect of all rifamycin derivatives [26,49,50]. First clinical data suggest that rifabutin is a valuable alternative to rifampin with less adverse events and less drug-drug interactions [51].

#### **5. Critical Appraisal of a Randomized Controlled Trial (RCT) Showing no Effect of Rifampin**

The above mentioned RCT compared the outcome of patients with acute staphylococcal PJI treated with prosthesis retention and either monotherapy without rifampin or rifampin combination [8]. In this multicenter study conducted from 2006 to 2012 in eight centers, 48 patients with acute PJI were included in the final analysis. PJI was caused by methicillin-susceptible staphylococci in 38 episodes (among them 36 were *S. aureus*) and 10 by methicillin-resistant staphylococci (of which all were *S. epidermidis*). Twenty-five patients were randomized to receive monotherapy, i.e., cloxacillin (two weeks intravenous, followed by four weeks oral) or vancomycin (six weeks intravenous) and 23 patients received rifampin in addition to the anti-staphylococcal treatment regimen mentioned above.

All patients underwent "soft tissue" revision with retention of the prosthesis. Rerevision with isolation of any pathogen was considered confirmed failure, while clinical signs of infection without revision surgery or isolation of pathogen were categorized as probable failure. Using the Kaplan–Meier method, the infection-free survival rate was similar in the monotherapy group (72%) and rifampin combination group (74%) at two years follow-up (median, 27 months). Success rate in PJI caused by methicillin-susceptible staphylococci was 78% with rifampin combination and 65% with monotherapy. In PJI caused by methicillin-resistant staphylococci, monotherapy was successful in all five patients (100%), whereas rifampin-vancomycin-combination had a success of 60% (three of five). No statistically significant difference was observed in any comparison. The authors conclude that adding rifampin to standard antibiotic treatment in acute staphylococcal PJIs does not improve the outcome.

In view of the above presented role of rifampin as biofilm-active antibiotic, the results of this RCT unsettled clinicians with limited experience in the field. Therefore, some critical points in this study should be highlighted for correct interpretation of the results.

First, the originally registered study protocol at ClinicalTrials.gov (NCT00423982) differs from the published manuscript, suggesting that relevant modifications were performed during the study. In contrast to the initial protocol, in addition to patients with early postoperative PJI those with acute hematogenous PJI were included. In late hematogenous PJI, the duration of infection is less well defined, because it may manifest only delayed after

seeding. This may explain that the success rate of PJI treated with DAIR has shown to be significantly lower in late acute staphylococcal infection as compared to early postoperative infections [52]. Unfortunately, the distribution of the two clinical entities in the analyzed cohort is not provided, making the interpretation of the results of the heterogeneous study population difficult.

Second, the surgical treatment is described in the Methods in detail. Whereas in the trial registration protocol, only a "soft tissue" revision is mentioned, in the manuscript additionally exchange of modular parts, irrigation with 9 L of saline and placement of two gentamicin-containing sponges (10 × 10 cm2) is stated, exceeding the procedure of a soft tissue revision. The adherence to this strict surgical protocol throughout the six-year study in eight study centers is questionable, as inclusion in the study took place most likely only after identification of the causing pathogen. Exchange of mobile parts being a proxy for a thorough debridement was shown to be among most relevant factors for successful outcome in several previous studies in case of retained infected prosthesis [36,53–55]. Noteworthy, no dropouts due to deviating surgical treatment were reported.

Third, the antimicrobial combination partner for rifampin is crucial, as mentioned by the authors in the Discussion. In this study, unusual combinations with oral cloxacillin (low oral bioavailability (37%), poor bone penetration, low maximal dose orally compared to intravenous route [56]) and prolonged intravenous vancomycin (toxic, poorly penetrating into the bone, barely bactericidal, non-therapeutic levels upon initiation of treatment) in case of methicillin-resistance were administered. Substances recommended as antimicrobial combination partner for rifampin are those with a high oral bioavailability and a good bone penetration, such as quinolones, trimethoprim-sulfamethoxazole, doxycycline or clindamycin, none of which was used in the present study. In addition, an unusual rifampin dosage (300 mg three times daily) was used, which is neither approved nor recommended for any indication.

Fourth, the absence of infectious diseases specialists in the author list suggests lack of an interdisciplinary team approach to the management of PJI, which is another important factor determining the treatment success of PJI [57,58]. After discharge, adequate intake or administration of antibiotics, patient compliance and modification in case of intolerance should be ensured. Rifampin is often discontinued due to intolerance or toxicity, as shown by the high number of dropouts (n = 7) due to rifampin discontinuation in this study. The accompaniment by an infectious diseases specialist during the treatment period could probably counteract the high dropout rate and potential selection bias.

Fifth, probably the most relevant drawback of the study is the low number of included patients. The final analysis with 48 patients in eight centers during six years indicates a reluctant recruitment. Since staphylococci are the most frequent pathogens of acute PJI [59,60], the average of one patient per center per year implies that the participating centers are not explicitly centers specialized in septic surgery and that the included patients represent a subgroup of patients bearing the risk of selection bias.

Sixth, due to the low number of included subjects, the study is underpowered, and thus does not allow any conclusion on the effect of rifampin on the outcome of acute staphylococcal PJI. The sample size calculation required at least 62 patients in each group to statistically prove an increase in cure rate of 20% (assuming a high cure rate of 70% in the monotherapy group). The authors aimed to include at least 100 subjects in each group. Only focusing on methicillin-susceptible staphylococci, the success rate with monotherapy was 65% (13 out of 20 patients), whereas the rifampin combination led to treatment success in 78% (14 out of 18 patients). Based on theoretical considerations, by increasing the number sample size sixfold (120 patients in the monotherapy group, 108 patients in the combination group) and assuming the same proportion of success in each group, the results would reach statistical significance. Unfortunately, the study was prematurely stopped without mentioning the reason for discontinuation. Only by increasing the sample size the beneficial effect of rifampin could have probably been shown, if there is one, as suggested by multiple above-mentioned studies.

Finally, there are a few imprecisions regarding the outcome evaluation, the reader should consider while interpreting the study results. It remains unclear to what extend the "probable" failures were true septic failures. Furthermore, it is not indicated, whether non-microbiological criteria (synovial fluid leukocyte count and periprosthetic tissue histopathology) for infection were fulfilled in these cases. In addition, the meticulous analysis of failures to discriminate relapse or infection caused by a new pathogen (superinfection) is missing, however, of utmost importance. The fact that the study was conducted several years ago would have allowed for assessment of long-term follow-up. However, only two-year follow-up was reported. Taking all these aspects into consideration, the discussed study does not allow any deduction on the effect of rifampin on the outcome of acute staphylococcal PJI treated with DAIR.

#### **6. Conclusions**

Taken together, the controversy about the role of rifampin in biofilm infections is not justified. There is abundant data from in-vitro and animal experiments, as well as clinical studies confirming its antibiofilm effect in patients with staphylococcal orthopedic implant-associated infections undergoing DAIR. Thus, one study with multiple weaknesses should not unsettle clinicians. An RCT with appropriate sample size, optimal choice of antimicrobials, standardized surgical interventions and accurate definition of treatment failure would be desirable. However, given the existing strong evidence demonstrating the benefit of rifampin, the conduction of such a clinical study would not comply with ethical standards and would probably not be approved by ethics committees.

**Author Contributions:** N.R., A.T. and N.R. discussed the outline. N.R. and W.Z. performed the literature review and wrote the manuscript. A.T. discussed and critically revised the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Data Availability Statement:** Data is contained within the article.

**Conflicts of Interest:** The authors declare no conflict of interest interpretation of data, in the writing of the manuscript, or in the decision to publish the results.

#### **References**


## *Article* **Tolerance of Prolonged Oral Tedizolid for Prosthetic Joint Infections: Results of a Multicentre Prospective Study**

**Eric Senneville 1,2,3,\*, Aurélien Dinh 4,5, Tristan Ferry 6,7, Eric Beltrand 3,8, Nicolas Blondiaux 3,9 and Olivier Robineau 1,2,3**


**Abstract:** Objectives: Data on clinical and biological tolerance of tedizolid (TZD) prolonged therapy are lacking. Methods: We conducted a prospective multicentre study including patients with prosthetic joint infections (PJIs) who were treated for at least 6 weeks but not more than 12 weeks. Results: Thirty-three adult patients of mean age 73.3 ± 10.5 years, with PJI including hip (*n* = 19), knee (*n* = 13) and shoulder (*n* = 1) were included. All patients were operated, with retention of the infected implants and one/two stage-replacements in 11 (33.3%) and 17/5 (51.5%/15.2%), respectively. Staphylococci and enterococci were the most prevalent bacteria identified. The mean duration of TZD therapy was 8.0 ± 3.27 weeks (6–12). TZD was associated with another antibiotic in 18 patients (54.5%), including rifampicin in 16 cases (48.5). Six patients (18.2%) had to stop TZD therapy prematurely because of intolerance which was potentially attributable to TZD (*n* = 2), early failure of PJI treatment (*n* = 2) or severe anaemia due to bleeding (*n* = 2). Regarding compliance with TZD therapy, no cases of two or more omissions of medication intake were recorded during the whole TZD treatment duration. Conclusions: These results suggest good compliance and a favourable safety profile of TZD, providing evidence of the potential benefit of the use of this agent for the antibiotic treatment of PJIs.

**Keywords:** tedizolid; prosthetic joint infections; prolonged oral treatment; tolerance; compliance

#### **1. Introduction**

Prosthetic joint infection (PJI) is a serious and complex complication following arthroplasty at an incidence rate after hip or knee replacement of 1 to 2% [1]. The aims of the management of patients with PJIs are to restore satisfactory joint function and to eliminate infection. Surgical options include debridement antibiotics and implant retention (DAIR), one- or two-stage replacement, arthroplastic resection and, sometimes, amputation. Given the increasing burden of these infections, especially among the elderly population, developing new therapies such as cell therapy to prevent the progression of osteo-arthritis, and thus, the need for total joint arthroplasty, is an important field of research [2–5]. The antibiotic treatment of patients with PJIs is limited by the tolerance of its prolonged administration and the resistance level of some pathogens [6,7]. Gram-positive cocci, especially coagulase-negative staphylococci (CoNS), are predominant bacteria which are

**Citation:** Senneville, E.; Dinh, A.; Ferry, T.; Beltrand, E.; Blondiaux, N.; Robineau, O. Tolerance of Prolonged Oral Tedizolid for Prosthetic Joint Infections: Results of a Multicentre Prospective Study. *Antibiotics* **2021**, *10*, 4. https://dx.doi.org/ 10.3390antibiotics10010004

Received: 6 December 2020 Accepted: 21 December 2020 Published: 23 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/).

responsible for infections in and around orthopaedic devices [8]. In this context, the use of the oxazolidinone agent linezolid (LZD) has been validated, but potential bone-marrow, neurologic, and metabolic toxicity limit treatment duration to no more than two to three weeks [9–11]. Additionally, the wide use of LZD has resulted in the emergence of CoNS carrying *cfr* genes which are responsible for high levels of LZD resistance [12]. The combination of rifampicin with LZD leads to a reduction in LZD blood concentration, which is associated with a lower rate of adverse hematologic effects but also with lower clinical remission rates [13,14]. Tedizolid (TZD) phosphate is a second generation oxazolidinone which is indicated for the treatment of acute bacterial skin and skin structure infections in adults [15–18]. In the Establish 1 and 2 studies, gastrointestinal disorders and bone marrow toxicity were less frequent in TZD than in LZD patients [19,20]. However, the duration of TZD treatment did not exceed six days. TZD has a high oral bioavailability and can be administered once-daily; furthermore, drug–drug interactions with mono-amine oxidase inhibitors (MAOI), serotonin-reuptake inhibitors (SRI) or rifampicin are unlikely, although the latter has recently been questioned [21,22]. Recent in vitro and animal studies suggested that the addition of rifampicin to TZD was likely to achieve a synergistic effect against methicillin-resistant *Staphylococcus aureus* (MRSA) and *S. epidermidis*, and prevent the emergence of rifampicin-resistant mutants [23,24]. While TZD appears to be an attractive candidate for the treatment of PJIs due to gram-positive cocci, and has shown satisfactory efficacy and tolerability in clinical trials, data about its tolerability and compliance in long-term treatments are lacking. The aim of the present multicentre prospective cohort study was to assess the long-term safety profile and compliance of oral TZD in monotherapy or in combination therapy for the treatment of PJIs.

#### **2. Results**

Thirty-three adult patients (sex ratio female/male 17/16) of mean age 73.3 ± 10.5 years were included from August 2018 to November 2019. A total of 17 patients (51.5%) were enrolled at the Tourcoing Centre, 13 (39.4%) at the Ambroise Paré Centre and 3 (9.1%) at the Lyon Centre. Patient characteristics are presented in Table 1. Despite chronic infection, three patients were treated with a debridement antibiotic and implant retention (DAIR) because of their age and general status which contraindicated the replacement of the implant. TZD use was used to avoid LZD potential toxicities or drug–drug interactions in 16 patients (48.5%), or because of previous, LZD-related adverse events in three patients (9.1%). Among these three patients, one had experienced thrombocytopenia, one anaemia and one gastro-intestinal intolerance. No included patients were receiving MAOI or SRI concomitantly with TZD therapy. Staphylococci were the most prevalent bacterium identified in our patients, accounting for 58% of the total number, and including 21 (42%) methicillin-resistant strains (Table 2). Infection was polymicrobial in 18 cases (54.5%) among which five were associated with gram-negative rods, all of which were susceptible to fluoroquinolones. Geometric mean MIC values for linezolid, as determined by E-test methods, were 1.24 ± 0.83 mg/L, 1.5 ± 0.7 mg/L and 1.64 ± 0.48 mg/L for *Staphylococcus* spp., *Streptococcus* spp. and *Enterococcus* spp., respectively. MIC measurements or TZD blood levels were not routinely performed for tedizolid in this study.

Following postoperative empirical antibiotic therapy (PEAT) of median duration of 7 days (range 4 to 14 days), the mean duration of TZD therapy was 8.0 ± 3.27 weeks (ranging from 6–12 weeks). Among the 27 out of 33 patients (81.8%) who completed the planned therapy, the mean duration of TZD was 8.77 weeks ± 2.79 (range 6–12 weeks). The mean total duration of the antibiotic treatment including PEAT and targeted therapy was 9.15 ± 3.43 weeks (7–12). TZD was associated with another antibiotic in 18 patients (54.5%), e.g., rifampicin in 16 cases (48.5%).


**Table 1.** Demographic data of 33 patients with periprosthetic joint infections.

SD: standard deviation. \*: 4 patients had ≥2 comorbidities.

In total, 20 patients (60.6%) experienced at least one adverse event during TZD therapy. A list of adverse events (AE) potentially attributable to tedizolid is shown in Table 3; the most frequent AE were anaemia (*n* = 4) and pruritus (*n* = 4). Six patients (18.2%) had to stop TZD therapy prematurely because of (i) intolerance which was potentially attributable to TZD (*n* = 2), (ii) early failure of PJI treatment (*n* = 2) or (iii) severe anaemia (*n* = 2). TZD-attributable discontinuation episodes consisted of inflammatory arthritis of the wrist and knee in one patient who also received doxycycline and did not improve after stopping doxycycline but partially recovered after discontinuation of TZD, and vomiting in another patient who received TZD alone (Appendix A Table A1). According to the definitions used to describe the bone marrow toxicity profile of TZD, 8 patients (24.2%) experienced haematological adverse events including anaemia in 4 cases, 2 of which presented acute haemorrhage, leukopenia in 2 cases and thrombocytopenia in 2 cases (Table A1). With the exception of the two patients with acute haemorrhage, none of these adverse events resulted in withdrawing TZD therapy. Although a gastric haemorrhage in one patient and a hematoma at the surgical site in another patient resulted in acute severe anaemia which was most probably unrelated to TZD therapy, the treatment was

discontinued. Haematological adverse events were mild and resolved spontaneously during TZD therapy except in the two patients with severe anaemia who received a blood transfusion. Non haematological adverse events were recorded in 13 (39.4%) patients for whom no premature discontinuation of TZD therapy was required, and were mostly pruritus (*n* = 4), headache (*n* = 2) and insomnia (*n* = 2) (Table A1). There was no safety signal for TZD-associated optic or peripheral neurologic toxicity or metabolic disorder. Overall, the proportion of patients who experienced TZD-attributable adverse event did not differ significantly in patients treated with a combination of antibiotics or with TZD alone [13/18 (72.2%) versus 8/15 (52.3%), respectively; *p* = 0.45], nor did it vary according to the use of rifampicin in combination with TZD or the total duration of TZD therapy (Table 4).

**Table 2.** Microbiology of 33 patients with periprosthetic joint infections.


Legend: MRSA: Methicillin-resistant *Staphylococcus aureus*; MRSE: Methicillin-resistant *Staphylococcus epidermidis*; MR: Methicillin-resistant.

**Table 3.** Episodes of adverse effects reported in 33 patients during tedizolid therapy.


\* Five patients had more than one episode of adverse effects.


**Table 4.** Adverse events according to the duration and the antibiotic regimen in 33 patients treated with tedizolid.

The follow-up of the haematological parameters showed a significant increase of haemoglobin blood levels between baseline and week 6 followed by stabilisation, as well as a significant decrease in platelets, leukocytes and neutrophils counts between baseline and week 6 followed by stabilisation until the end of the treatment (Figure 1A–D).

**Figure 1.** Boxplot of the haematological parameters ((**A**): Haemoglobin, (**B**): Platelets, (**C**): Leukocytes, (**D**): Neutrophils) during Tedizolid therapy. Each box represents median, interquartile range, largest, smallest and outside (point) values.

Regarding compliance to TZD therapy, no cases of two or more omissions of medication intake during TZD treatment were recorded, in accordance with the number of pills present in the returned boxes. Six failures (18.2%), including two early cases, were recorded at one-year following the end of TZD therapy.

#### **3. Discussion**

We report the first prospective cohort study to date providing data on the safety and compliance of prolonged use (i.e., ≥6 weeks) of oral TZD phosphate at a 200-mg, once-daily dose for the treatment of PJIs. As our safety results suggest, oral TZD therapy administered for 6 to 12 weeks according to the current recommendations [25] can be considered for the treatment of PJIs. The overall proportion of patients who experienced an adverse event (60.6%) may appear high, but this may be explained by the design of the study which allowed us to report an exhaustive list of adverse events. Despite the nonoptimal profile regarding the general status of our patients, the tolerance of prolonged oral TZD therapy allowed us to complete the therapy in more than 80% of the cases. Indeed, 19 patients (57.6%) had comorbidities and 27 (81.2%) had an ASA score ≥ 2, which is significantly different from the populations of patients evaluated in other, pivotal clinical trials [19,20]. Our results are close to those reported by Kim et al. on a series of 25 patients with nontuberculous mycobacterial infections treated with a median duration of TZD therapy of 91 days [26]. Eleven of their patients (44%) experienced an adverse event including gastrointestinal intolerance in five patients (20%) and thrombocytopenia in one (4%); no case of anaemia was recorded, while peripheral neuropathy was reported in five patients (20%). The attribution of an adverse event to TZD was, however, difficult, as almost all patients were receiving multidrug therapy. The mean age of our patients was 73.3 years, which is quite high with regard to the risk of developing bone marrow toxicity to LZD, as reported by several authors [9,10]. The correction of LZD-induced bone marrow toxicity after switching to TZD observed in one of our patients has already been reported elsewhere [27,28]. Overall, we only recorded a few significant haematological abnormalities which did not result in discontinuation of TZD therapy, except in patients with acute haemorrhage. There were no differences in safety, especially with regard to haematological laboratory changes, between patients receiving TZD in combination with rifampicin versus patients receiving TZD alone (full data are available upon request), as reported with LZD-rifampicin combination [9,10]. We hypothesise that the increase of haemoglobin values during TZD treatment represents a restoration process after blood spoliation secondary to the surgical intervention, while the decrease of platelets and WBC during TZD treatment might be related to the resolution of the infectious process. The incidence of digestive disorders reported in our patients is close to the results of a meta-analysis by Lan et al., noting, however, that the duration of TZD therapy in the studies included was six days, as recommended for the treatment of acute bacterial skin and skin structure infections [29].

The main limitations of the present pilot study are the small size of the studied population and the assessment of the patients' adherence to TDZ treatment, which was based on the return of the pillboxes and on patient self-reporting. The strengths of the present study are its prospective design and the selection criteria which allowed investigators to include patients in a real-life setting. We strongly believe that the inclusion of patients, regardless of age and risk factors for bone marrow toxicity, enhanced the external validity of our conclusions regarding the tolerance of prolonged oral TZD therapy in patients treated with PJIs.

#### **4. Materials and Methods**

The purpose of the present study was to obtain reliable data on the tolerance, compliance and efficacy of prolonged (i.e., ≥6 weeks) use of TZD alone or in combination therapy for the treatment of PJIs. We present herein data about adherence and tolerance. As post-treatment follow-ups are currently underway, we present data only for the one-year follow-up. We conducted a prospective multicentre cohort study in three French national centres for the management of complex bone and joint infections (also called CRIOAc): Lille-Tourcoing, Paris-Ambroise Paré and Lyon [30].

#### *4.1. Definitions*

Adult patients with PJIs defined according to the MSIS 2018 [31] criteria and for whom TZD treatment was indicated according to the investigator's decision were prospectively included. All patients gave their written informed consent after an explanation of the protocol by the investigating physician. PJIs were characterised according to: acute haematogenous (infection with three-week duration or less of symptoms after an uneventful postoperative period), early postinterventional (infection that manifested within one month after implantation) and chronic (infection with symptoms that persisted for >3 weeks, i.e., beyond the early postinterventional period) according to Zimmerli's definition [32].

Patients demographic data (age, gender, body mass index), comorbidities, microbiology, prior use of LZD and reason for TZD use (e.g., failure and/or toxicity of previous treatment, need to avoid linezolid toxicity or drug–drug interactions), treatment duration, concomitant antibiotics, potential adverse events attributable to TZD were recorded. Laboratory data were recorded at baseline, weekly and at the end of treatment, including haemoglobin, white blood count (WBC), platelet count, alanine aminotransferase (ALT), aspartate aminotransferase (AST) and Protein-C reactive (PCR).

Clinically significant laboratory changes were defined as: (1) anaemia, decrease in haemoglobin ≥2 g/L from baseline after TZD initiation and classified as severe if haemoglobin was <8 g/dL, (2) leukopenia, white blood count (WBC) of <4 G/L after TZD initiation and classified as mild (low limit normal to 3 G/L), moderate (≥2-<3 G/L) and severe (<2 G/L), (3), neutropenia, absolute neutrophil count of <1.5 G/L after TZD initiation, (4) thrombocytopenia, platelet count of <150 G/L after TZD initiation and classified as mild (75–150 G/L), moderate (50-<75 G/L) or severe (<50 G/L); for patients with a baseline platelet count of <150 G/L, thrombocytopenia was defined as a reduction of 25% from the baseline, and (5) elevated AST or ALT 3 times above the upper limit of normal.

Microbiological documentation was based on joint aspiration and/or intraoperative culture samples. During surgical procedures, at least three tissue samples were taken in different areas suspected of infection, using a separate sterile instrument for each sample. The antibiotic susceptibility profile of all pathogens was assessed either by the Vitek 2 cards (BioMérieux, Marcy l'Etoile, France) or by agar diffusion technique using the procedure and interpretation criteria proposed by the Comité de l'Antibiogramme de la Société Française de Microbiologie (CA-SFM EUCAST 2018) (http://www.sfm-microbiologie.org). Methicillin resistance was confirmed by the detection of the *mecA* gene if required.

Adverse events (AEs) were identified from patients' medical records and laboratory data. Associations of adverse events with TZD and related antibiotics were assessed as suggested by the patient's physician and confirmed by the principal investigator according to the chronology of events, as were the need to reduce the daily dosage of the potentially problematic antibiotic, data from any attempt to reintroduce such a mode of treatment, and the type of recorded toxicity (e.g., anaemia, thrombocytopenia and peripheral neuropathy for TZD, tendonitis, and myalgia for levofloxacin and drug–drug interaction for rifampicin). To be attributable to a given antibiotic, a reduction in the daily dosage and/or discontinuation due to intolerance had to be recorded as well as temporal association with event resolution after discontinuation or dose reduction of the agent in question.

#### *4.2. Antibiotic Treatment*

TZD was administered orally at a once daily dose of 200 mg (i.e., one tablet) as a single antibiotic therapy or in combination therapy with another agent with proven activity against the involved pathogen(s) according to the physician's choice. The duration of antibiotic therapy ranged from 6 to 12 weeks. Exclusion criteria were pregnant women or women of childbearing age who were not using contraception, breastfeeding intolerance to TZD, allergy to oxazolidinone, the detection of bacteria which were nonsusceptible to TZD, patients with uncertainty regarding the possibility of achieving a one-year follow-up after the end of treatment or the absence of written consent. Patients were examined during consultations every 3 weeks during treatment and at 6 months and one year after the EOT. During treatment, special attention was paid to potential neurological and optical side effects, as well as to possible drug–drug interactions.

#### *4.3. Statistics*

Data are presented as numbers (percentages) for qualitative variables and as medians (interquartile range: IQR) or means (SD) for quantitative variables. We compared biological variable that might have been affected by the use of oxazolidinone between baseline and day 42 and between day 49 and day 84 using the Student *t*-test, with *p* = 0.05 being set as the threshold of significance.

#### *4.4. Ethics*

Research was conducted in accordance with the Declaration of Helsinki and national and institutional standards. The study was recorded on clinicaltrial.gov under the number NCT03378427, in the EudraCT database under the number 2017-001238-24 and was approved by the French Sud Mediterranean IV *Committee* of *Protection* of the *People* in Biomedical Research on 21 November 2017 under the number 17 10 09. This interventional survey was declared to the National Agency for Medicines and Safety of Health Products under the number 17060A-43. Tedizolid was supplied by *Merck Sharp & Dohme*, Inc.

#### **5. Conclusions**

The results of the present study suggest good compliance and a favourable safety profile of TZD, providing evidence of the potential benefit of the use of this agent for the antibiotic treatment of PJIs.

**Author Contributions:** E.S., E.B. and O.R. conceived the design of the study and wrote the manuscript; A.D. and T.F. reviewed the manuscript, all but N.B. participated in the inclusions and management of the patients; N.B. reviewed the microbiology part of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** No external funding was received except that tedizolid pills were provided by Merck Sharp and Dohme.

**Data Availability Statement:**Data are available upon request to Pr Eric Senneville (esenneville@ch-tourcoing.fr); the French authorities do not authorize the sharing of this type of data without prior consent and infor-mation of the patients on their final use.

**Acknowledgments:** We greatly thank the G. Dron hospital clinical research unit, especially Solange Tréhoux for the technical assistance.

**Conflicts of Interest:** E.S., A.D., O.R. and T.F. declare congress support, speaker invitation, participation to scientific board with MSD; E.B. and N.B. have nothing to declare in relation with the present study. E.S., A.D., O.R. and T.F. declare congress support, speaker invitation, and participation to scientific board with MSD; E.B. and N.B. have nothing to declare in relation with the present study.

*Antibiotics* **2021**, *10*, 4

**Appendix A**

**Table A1.** Details on tolerance

 of TZD therapy in 33 patients treated for

periprosthetic

 joint

infections.


*Antibiotics* **2021**, *10*, 4


#### **Table A1.** *Cont.*

*Staphylococcus*

 *caprae*, \*:

methicillin-resistant.

#### **References**


## *Article* **Long-Term Use of Tedizolid in Osteoarticular Infections: Benefits among Oxazolidinone Drugs**

**Eva Benavent 1,2 , Laura Morata 2,3,4, Francesc Escrihuela-Vidal 1, Esteban Alberto Reynaga <sup>5</sup> , Laura Soldevila 1,2, Laia Albiach 3, Maria Luisa Pedro-Botet 5, Ariadna Padullés 6, Alex Soriano 2,3,4 and Oscar Murillo 1,2,4,\***


**Abstract:** Background: To evaluate the efficacy and safety of long-term use of tedizolid in osteoarticular infections. Methods: Multicentric retrospective study (January 2017–March 2019) of osteoarticular infection cases treated with tedizolid. Failure: clinical worsening despite antibiotic treatment or the need of suppressive treatment. Results: Cases (*n* = 51; 59% women, mean age of 65 years) included osteoarthritis (*n* = 27, 53%), prosthetic joint infection (*n* = 17, 33.3%), and diabetic foot infections (*n* = 9, 18%); where, 59% were orthopedic device-related. Most frequent isolates were Staphylococcus spp. (65%, *n* = 47; S. aureus, 48%). Reasons for choosing tedizolid were potential drug-drug interaction (63%) and cytopenia (55%); median treatment duration was 29 days (interquartile range -IQR- 15–44), 24% received rifampicin (600 mg once daily) concomitantly, and adverse events were scarce (*n* = 3). Hemoglobin and platelet count stayed stable throughout treatment (from 108.6 g/L to 116.3 g/L, *<sup>p</sup>* = 0.079; and 240 <sup>×</sup> <sup>10</sup>9/L to 239 <sup>×</sup> <sup>10</sup>9/L, *<sup>p</sup>* = 0.942, respectively), also in the subgroup of cases with cytopenia. Among device-related infections, 33% were managed with implant retention. Median follow-up was 630 days and overall cure rate 83%; among failures (*n* = 8), 63% were device-related infections. Conclusions: Long-term use of tedizolid was effective, showing a better safety profile with less myelotoxicity and lower drug-drug interaction than linezolid. Confirmation of these advantages could make tedizolid the oxazolidinone of choice for most of osteoarticular infections.

**Keywords:** tedizolid; oxazolidinones; osteoarticular infections; diabetic foot infections; drug-drug interaction

#### **1. Introduction**

Oxazolidinones are a young family of antibiotics that have a wide action against Gram-positive bacteria and include the first designed linezolid and the recent tedizolid. In comparison with the former, tedizolid has shown a higher in vitro activity against some microorganisms (four- to eight-fold lower minimum inhibitory concentration (MIC) values), and its pharmacokinetic/pharmacodynamics parameters allow the once-daily administration, which may improve treatment adherence. Also, tedizolid at approved

**Citation:** Benavent, E.; Morata, L.; Escrihuela-Vidal, F.; Reynaga, E.A.; Soldevila, L.; Albiach, L.; Pedro-Botet, M.L.; Padullés, A.; Soriano, A.; Murillo, O. et al. Long-Term Use of Tedizolid in Osteoarticular Infections: Benefits among Oxazolidinone Drugs. *Antibiotics* **2021**, *10*, 53. https:// doi.org/10.3390/antibiotics10010053

Received: 16 December 2020 Accepted: 6 January 2021 Published: 8 January 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

doses seems to provide a better safety profile and less adverse events than linezolid, especially in relation with myelotoxicity and drug–drug interaction [1–3].

Tedizolid is currently only approved for acute bacterial skin and skin structure infections (ABSSSIs), not including diabetic foot infections [4,5]. However, it seems reasonable that tedizolid can be used in other clinical settings where linezolid has played a relevant role. In this sense, difficult-to-treat osteoarticular infections mainly due to staphylococci constitute a notable scenario where linezolid has provided good efficacy [6–8]. However, adherence to linezolid treatment may involve some difficulties such as (i) the appearance of myelotoxicity at two to four weeks of treatment, since osteoarticular infections usually require longer treatments [8,9]; (ii) the decrease in linezolid serum levels when combined with rifampicin, an anti-staphylococcal agent widely used in device-related infections [10–13]; and (iii) the risk of serotonin syndrome when administered concomitantly with antidepressants broadly used nowadays [14,15].

Despite the potential advantages of tedizolid in osteoarticular infections (higher microbiological activity, advantageous pharmacokinetic/pharmacodynamics parameters, lower myelotoxicity, and drug–drug interactions) [1–3], clinical data on long-term treatment are scarce [16]. Knowledge is limited to a few experimental studies [17–19], case reports [20] and recently some case series in which tedizolid is prescribed for different indication including osteoarticular infections [21,22]. Thus, in the present study, we intend to describe our multicenter experience within the Spanish Network for Research in Infectious Diseases (REIPI) with long-term use of tedizolid in a cohort of patients with osteoarticular and diabetic foot infections and focused on the efficacy and safety in monotherapy or combination.

#### **2. Results**

A total of 51 cases were included in our study. Mean age was 64.8 ± 14.3 and 59% (*n* = 30) were women. Median Charlson Index adjusted by age was 4 (IQR 3–7). Obesity was present in 33% (*n* = 15) of the cases; other frequent comorbidities were diabetes mellitus (*n* = 21, 41%), chronic renal disease (*n* = 16, 31%), malignancies (*n* = 7, 14%), and chronic anemia (*n* = 4, 8%).

There were 53% (*n* = 27) diagnosis of osteoarthritis (*n* = 17 cases of peripheral osteomyelitis, *n* = 6 septic arthritis and *n* = 4 vertebral osteomyelitis; of which 1 case presented simultaneously with septic arthritis of the ankle and osteomyelitis of the tibia), 33% (*n* = 17) cases had a prosthetic joint infection (*n* = 8 post-surgical acute, *n* = 8 chronic and 1 case with intraoperative positive cultures), and there were 18% (*n* = 9) cases of diabetic foot infection, one of them presenting also with vertebral osteomyelitis. Thirty cases (59%) were orthopedic device-related infections. Only two cases (4%) had bacteremia (both due to methicillin-susceptible *Staphylococcus aureus*).

Microorganisms responsible for osteoarticular infections were identified in all but two cases. There were 20 cases (39%) of polymicrobial etiology, 11 of them at the expense of different Gram-positive isolates that were all treated with tedizolid, and the remaining cases were mixed with Gram-positive and Gram-negative microorganisms. All Gram-positive microorganisms involved are presented in Table 1.

Tedizolid was administered at a dosage of 200 mg per day orally for a median of 29 days (IQR 15–44); 63% of the cases (*n* = 32) received tedizolid for more than 21 days, and in 70% of the cases (*n* = 36) time under tedizolid treatment represented more than 50% of the whole antibiotic treatment duration. Causes for prescription of tedizolid are presented in Table 2 (in 14 cases there was more than one reason); most common reason for initiate tedizolid was the potential interaction between baseline treatment and linezolid (65%), followed by the presence of anemia and/or thrombocytopenia (37%) and toxicity caused by a previous antibiotic (16%).


**Table 1.** Gram-positive microorganisms responsible of osteoarticular infections and treated with tedizolid, from the 72 isolates in the whole cohort.

<sup>1</sup> Enterococcus faecium *n* = 3, Enterococcus raffinosus *n* = 1. <sup>2</sup> Streptococcus pyogenes *n* = 1, Streptococcus oralis *n* = 1, Streptococcus agalactiae *n* = 1.

**Table 2.** Reasons for Tedizolid prescription.


<sup>1</sup> All cases were under treatment with Serotonin Reuptake Inhibitors (SRIs). <sup>2</sup> Shortage of Linezolid.

Tedizolid was mainly administered as part of sequential switching therapy (*n* = 47, 92%; 3 cases as salvage therapy after failure), and only in 4 cases (8%) was the initial treatment. Tedizolid was given in monotherapy (*n* = 27, 53%) and in combination (*n* = 24, 47%) almost in similar proportion; among combination therapy, in half of the cases (*n* = 12) tedizolid was combined with rifampicin (600 mg once daily) representing 25% of all staphylococci infections. Less usual tedizolid combinations were used to treat polymicrobial infections with participation of Gram-negative bacteria; among them the most frequent drugs were quinolones (*n* = 7, 14%) and carbapenems (*n* = 4, 5%).

Beside therapy with tedizolid, most of the cases were managed surgically (*n* = 41, 80%); among device-related infections, implants were removed in 57% cases. Among cases with an evaluable outcome (*n* = 48; median of follow-up 630 days, IQR 269–818), the overall cure rate was 83% and 8 cases (17%) failed. There were 3 deaths, all of them non-related neither with the infection nor with tedizolid therapy, in which the outcome could not be evaluated due to short follow-up after treatment. The cure and failure rates among each type of osteoarticular infection are presented in Figure 1. Among failures, 4 of them were prosthetic joint infections (3 of them were put on suppressive antibiotic treatment and the remaining one underwent further surgery to cure the infection), 3 cases of osteoarthritis (one case was a device-related infection managed with implant removal), and a case of diabetic foot infection. Among staphylococci infections, there was no difference in outcome when tedizolid was used in monotherapy vs. combination with rifampicin (failure rate of 21% vs. 0%, respectively, *p* = 0.118).

Failure Cure

**Figure 1.** Failure rates among different types of osteoarticular infection.

Therapy with tedizolid was well tolerated; the only adverse effect observed was gastrointestinal disturbances in three cases (6%; nausea and occasional vomiting), but in any case, treatment was withdrawn.

There was no worsening on the hemoglobin or platelet counts in the blood tests between the beginning and the end of treatment with tedizolid neither in the group of patients with cytopenia nor in those without (Table 3) or those where treatment was prolonged for more than 21 days. In three cases, linezolid was switched to tedizolid because myelotoxicity of the former and patients completed treatment without additional worsening. The use of rifampicin in combination with tedizolid did not produce significant differences in the final levels of hemoglobin or platelets in comparison with tedizolid in monotherapy (Figure 2).


**Table 3.** Analytic values of patients under Tedizolid treatment.

\* In accordance with definition in Section 4.2. No anemia was considered when; Hb > 130 g/L for men and Hb > 120 g/L for women. Mild anemia; Hb 110–129 g/L for men and Hb 110–119 g/L for women. Moderate anemia considered Hb < 109 g/L and severe anemia Hb < 80 g/L for men and women in both cases.

(**a**)

(**b**)

**Figure 2.** Mean comparison of hemoglobin values and platelet count at the beginning and the end of treatment for those cases receiving Tedizolid in monotherapy (**a**) or combination with Rifampicin (**b**).

> Finally, among cases treated with tedizolid because of potential interaction between linezolid and baseline treatment (*n* = 33), they were treated for a median of 34 days (IQR 17–51) and we did not observe adverse events (i.e., serotonin syndrome) or alteration of basal disease (i.e., depressive syndrome) either during therapy with tedizolid or after it was stopped.

#### **3. Discussion**

Tedizolid is recommended for the treatment of ABSSSIs for six days and still, there is limited information in other settings or prolonged treatments. In the present study, we provide data about efficacy, the safety of long-term use (median of 29 days), and benefits of tedizolid in a large cohort of patients with osteoarticular and diabetic foot infections.

Linezolid, the first approved oxazolidinone drug, has provided good outcomes in osteoarticular infections [6–8,12]; however, there are still some concerns for its longterm use regarding adverse events and drug–drug interactions. Comparing with linezolid, tedizolid shows a better microbiological activity and more favorable pharmacokinetic/pharmacodynamics parameters when used at the recommended dose of 200 mg daily. In our experience, tedizolid both in monotherapy or combination provide good efficacy in this field (cure rate 83%), comparable to that of linezolid. These results are difficult to compare with previous work, because to our knowledge, there are no studies focused on tedizolid efficacy in diabetic foot infections and only a recent experience of tedizolid for more than six days in different types of infections, including some osteoarticular infection cases [22]. Thus, larger experience in this setting is needed, but our results seem to be in the line of considering tedizolid a good therapeutic alternative.

Among adverse events observed when using linezolid, anemia and/or thrombocytopenia is common when its use is prolonged beyond two weeks [9,12,23]. It requires monitoring patients, especially those with previous cytopenia or particular risk factors. Tedizolid can also cause dose-related myelotoxicity [16], but at a lower rate than linezolid [24]. Safety of long-term use of tedizolid was evaluated in healthy volunteers for 21 days [16], while most of the information in patients is limited to six days of treatment in ABSSSIs [4,5,21], and there is little information with longer therapies where the appearance of thrombocytopenia and anemia was observed in 7.4 and 1.2%, respectively [22]. In our experience, tedizolid was administered a median of 29 days and was well tolerated without relevant hematologic adverse events appearing or need for withdrawn, even in cases with moderate/severe cytopenia at the start of therapy or those that were switched to tedizolid after developing linezolid myelotoxicity [25].

Among the most relevant drug–drug interactions of linezolid, its use with rifampicin should be emphasized since it is a major anti-staphylococcal agent broadly used in osteoarticular infections always in combination. Previous studies confirmed the interaction between both drugs [3,13,26], as a result, serum linezolid levels decrease [10,11]. Interestingly, this effect between tedizolid and rifampicin was not found in preclinical studies [3,17]; however, well-designed clinical and pharmacokinetic studies are not available. In our experience, despite not determining the comparative serum levels of tedizolid when monotherapy or combination with rifampicin was used, we did not observe differences in the clinical outcome or the impact on hemoglobin or platelet counts in both groups. If the absence of interaction between rifampicin and tedizolid is confirmed, the latter could displace the use of linezolid in those patients who require it in a rifampicin combination.

Finally, treatment with linezolid can also be challenging when given concomitantly with monoamine oxidase inhibitors and other antidepressants due to the risk of serotonin syndrome, a rare but serious complication [14,15]. Tedizolid exhibits a weak reversible monoamine oxidase inhibition in vitro effect, so drug–drug interaction is lower [27]. The potential drug–drug interaction was the main reason (62.8%) for choosing tedizolid in our cohort, and none of the patients interrupted their baseline treatment and no adverse event was observed.

Our study has several limitations inherent to its retrospective nature, as well as the heterogeneity between the different osteoarticular infections, and the limited number of cases. As a result, the inference of our results in particular scenarios should be taken with caution while waiting for wider experience. Also, the different surgical approaches, which are a cornerstone of the treatment of osteoarticular infections, and the use of other antibiotics before tedizolid or in combination (mainly with rifampicin), may have played a role in the overall outcome. Furthermore, to assess the suitability of the rifampicintedizolid combination or the use of tedizolid concomitantly with antidepressants, further specific studies are needed. However, to our knowledge this is the first study assessing long-term

use of tedizolid specifically in osteoarticular infections and carried out by specialists in the field and, therefore, the information provided in terms of efficacy and safety is of interest.

#### **4. Materials and Methods**

#### *4.1. Study Population and Settings*

We conducted a retrospective multicenter study in three Spanish hospitals of the REIPI-GEIO Network (January 2017 to March 2019). We included adult patients attended for osteoarticular and diabetic foot infections caused by Gram-positive bacteria who had received as part of their antibiotic treatment tedizolid at a regimen dose of 200 mg daily for at least 7 days. Polymicrobial infections with the participation of Gram-negative microorganisms were also included, whenever they received tedizolid in combination for their treatment. Those cases where Gram-positive bacteria were involved after a different primary infection (superinfection) were excluded.

We aimed to evaluate the efficacy and safety of cases treated with tedizolid for a long-term period. Additionally, we aimed to evaluate the potential drug–drug interaction between tedizolid and rifampicin or antidepressants.

#### *4.2. Definitions and Data Collection*

Osteoarticular infections were classified into 3 groups: osteoarthritis (including cases with peripheral or vertebral osteomyelitis and septic arthritis), prosthetic joint infections, and diabetic foot infections. All cases met the main diagnostic criteria [28–30] and management was carried out according to the attending medical team, in all cases, antibiotic treatment was tailored by infectious diseases specialists.

Presence of anemia was classified in mild anemia when hemoglobin concentration was between 110–129 g/L for men and 110–119 g/L for women, moderate anemia when hemoglobin was below 109 g/L, and severe when it was below 80 g/L. Thrombocytopenia was considered when platelet count was below 150 × 109/L.

Demographic data and baseline characteristics were collected. The presence of depressive syndrome was specifically registered and the use of drugs with potential major interaction with oxazolidinones such as mono-amino oxidase inhibitors, selective serotonin reuptake inhibitors, opioids, and anticonvulsant drugs. The presence or absence of orthopedic devices was documented and also the need for debridement surgery and removal or exchange of orthopedic devices when necessary. Antibiotic treatment previous and concomitant with tedizolid was also recorded. Microbiologic data were obtained from intraoperative cultures, joint fluid samples, or targeted biopsies.

Written informed consent was considered not necessary for the study, as it was a retrospective analysis of our clinical practice. Data of patients were anonymized for the purposes of this analysis. Confidential information of patients was protected according the Declaration of Helsinki. This manuscript was revised for its publication by Research Ethics Committee of Bellvitge University Hospital (PR459/20).

#### *4.3. Follow Up and Outcome*

To monitor possible hematologic toxicity, we documented laboratory data at the beginning of treatment with tedizolid, during treatment, and at the end of the antibiotic treatment. The patients also underwent clinical follow-up to detect the presence of other adverse events; gastrointestinal or neuropathic toxicity (optical and peripheral).

Cases were considered cured when there was no clinical evidence of infection and no other need for antibiotic or surgical treatment once treatment with tedizolid was concluded. Failure was considered when reappearance of infection signs once treatment was already concluded, in cases of none improvement despite active treatment with tedizolid, the need of suppressive antibiotic therapy to control the infection or death related with the infection.

#### *4.4. Statistical Analysis*

Data were analyzed with Stata 14.2 (Stata Corporation, Texas 77845, USA). Categorical variables were described by counts and percentages, while mean and standard deviation or median and interquartile range (IQR) were used to summarize continuous variables. Comparisons between groups were performed with either the chi-square test or Fisher exact test for categorical variables, and the *t*-test or Mann-Whitney U test was used for continuous variables.

#### **5. Conclusions**

In conclusion, tedizolid was effective and safe providing a valid therapeutic alternative for osteoarticular infections. Its higher microbiological activity and better pharmacokinetic/pharmacodynamics parameters comparing with linezolid, allow it to be used at doses that show a better safety profile with less myelotoxicity and lower drug-drug interaction including rifampicin and antidepressants. If further studies confirm these advantages, tedizolid may become the oxazolidinone of choice in most patients with osteoarticular infections.

**Author Contributions:** E.B., L.M., F.E.-V., E.A.R., L.S., L.A., M.L.P.-B., A.P., A.S., O.M. contributed in the supervision of the clinical cases, data collection, and interpretation; F.E.-V., A.P., and O.M. elaborated the study design; E.B. performed the analysis of the data and wrote the first draft of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** E.B. was supported with a grant of the Instituto de Salud Carlos III—Ministry of Science and Innovation (FI 16/00397). This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of Bellvitge University Hospital (protocol code PR459/20).

**Informed Consent Statement:** Patient consent was waived due to the retrospective nature of the study on the usual clinical practice.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author (omurillo@bellvitgehospital.cat).

**Acknowledgments:** We thank John B. Warren (International House Barcelona) for reviewing the English manuscript. The preliminary results of this study were reported in part at the 30th European Congress of Clinical Microbiology & Infectious Diseases (Paris, France, 2020). We thank CERCA Programme / Generalitat de Catalunya for institutional support.

**Conflicts of Interest:** O.M. has received honoraria for talks on behalf of Merck Sharp and Dohme and Pfizer. A.S. has received honoraria for lectures and advisory meetings from Pfizer, Merck, Angelini, Shionogi, Menarini, and Gilead. L.M. has received honoraria for talks on behalf of Merck Sharp and Dohme, Pfizer and Angelini. All other others declare no potential conflicts of interest relevant to this article.

#### **References**


## *Review* **Dalbavancin for the Treatment of Prosthetic Joint Infections: A Narrative Review**

**Luis Buzón-Martín 1,2,\* , Ines Zollner-Schwetz 3, Selma Tobudic 4, Emilia Cercenado 5,6,7 and Jaime Lora-Tamayo 2,8,9**


**Abstract:** Dalbavancin (DAL) is a lipoglycopeptide with bactericidal activity against a very wide range of Gram-positive microorganisms. It also has unique pharmacokinetic properties, namely a prolonged half-life (around 181 h), which allows a convenient weekly dosing regimen, and good diffusion in bone tissue. These features have led to off-label use of dalbavancin in the setting of bone and joint infection, including prosthetic joint infections (PJI). In this narrative review, we go over the pharmacokinetic and pharmacodynamic characteristics of DAL, along with published in vitro and in vivo experimental models evaluating its activity against biofilm-embedded bacteria. We also examine published experience of osteoarticular infection with special attention to DAL and PJI.

**Keywords:** dalbavancin; prosthetic joint infection; gram-positive

#### **1. Introduction**

Total joint arthroplasties are common worldwide, and the incidence of this surgery is expected to increase steadily in the coming years as the population ages [1]. The most feared complication is infection, which is not associated with high mortality rates, but does carry substantial morbidity, may require many surgeries, and the final results in terms of limb functionality and pain resolution are not always satisfactory. At the same time, prosthetic joint infections (PJI) represent a massive economic burden for healthcare systems that continues to rise, and is expected to be around \$1.62 billion in USA by 2030 [2].

PJIs are complex infections, in which the formation of biofilm, enabling bacteria to evade the host immune system, is crucial. Biofilm-embedded bacteria can also develop phenotypic changes that ultimately lead to antimicrobial tolerance and infection persistence. Not all antimicrobials perform equally in this scenario, and not all antibiotics are ideal for the treatment of PJI. In this context, the arrival of new antimicrobials is very welcome [3].

DAL is a lipoglycopeptide (Xydalba; https://www.ema.europa.eu (accessed on 27 May 2021)) that is almost universally active against Gram-positive bacteria, which are by far the leading cause of PJIs [4]. A number of clinical trials [4–7] have demonstrated its safety and efficacy for the treatment of skin and soft tissue infections, which stand as the

**Citation:** Buzón-Martín, L.; Zollner-Schwetz, I.; Tobudic, S.; Cercenado, E.; Lora-Tamayo, J. Dalbavancin for the Treatment of Prosthetic Joint Infections: A Narrative Review. *Antibiotics* **2021**, *10*, 656. https://doi.org/10.3390/ antibiotics10060656

Academic Editor: Jaime Esteban

Received: 20 April 2021 Accepted: 26 May 2021 Published: 31 May 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

only licensed indication for this antibiotic. However, the wide antimicrobial spectrum of this drug and its unique pharmacokinetic (PK) properties, with a half-life of 181 h [8] and prolonged concentrations in bone tissue [9], along with a good safety profile have led physicians to use it for a number of off-label indications [10,11], which include the treatment of bone and joint infections as well as PJI. In addition, resistance emergence under DAL treatment is, although possible, a very rare phenomenon. In this particular setting, the need for treatment over long periods, coupled with the long half-life of DAL mean that the antibiotic can be used on a convenient weekly basis.

In this narrative review, we assess the role of DAL in the treatment of PJI. We review the drug's PK profile, pharmacodynamic (PD) properties, activity against biofilm-embedded bacteria in in vitro and in vivo experimental models, and finally, we evaluate the available clinical experience in PJI.

#### **2. Search Strategy and Selection Criteria**

The PubMed database was screened for any manuscript published at any time addressing the efficacy of DAL in the setting of biofilm-associated infections, bone and joint infections, and especially PJI. The terms "dalbavancin", "prosthetic joint infection", "biofilm", "foreign-body", "arthroplasty", and "osteomyelitis" were combined. Abstracts and relevant full-length articles were reviewed, and a thorough search was made of the references in these papers in order to select other significant studies. Our review is not exhaustive but focuses on relevant articles regarding the efficacy on DAL on the setting of PJI, and it was restricted to articles written in English and Spanish. We directly contacted the corresponding authors of published cases series of PJI treated with DAL in order to obtain further details.

*Definitions*

MIC50: Minimum inhibitory concentration required to inhibit the growth of 50% of organisms.

MIC90: Minimum inhibitory concentration required to inhibit the growth of 90% of organisms.

MBIC: Minimum biofilm inhibitory concentration. The lowest concentration of an antimicrobial agent required to inhibit the formation of biofilms.

MBBC: Minimum biofilm bactericidal concentration. The lowest concentration of an antimicrobial agent that eradicates 99.9% of biofilm-embedded bacteria.

#### **3. Dalbavancin in Prosthetic Joint Infections**

#### *3.1. DAL Pharmacokinetics*

The pharmacokinetics of DAL are linear and dose-proportional, with the peak concentration (Cmax) and area under the curve (AUC) increasing according to the dose administered, while its half-life (T1/2) of around 7 days remains essentially unchanged [8]. A high protein-bound fraction (>90%) contributes to this prolonged T1/2 [12,13]. It has been proven that serum bactericidal activity remains measurable at 7 days after a dose of 500 mg or higher, which establishes the basis for the weekly based dosing regimen proposed [8,14]. DAL concentrations before the following weekly dose have consistently been shown to range from 33.0 μg/mL to 40.2 μg/mL [12]. For skin and soft tissue infections, the recommended dosage consists of a loading dose of 1000 mg followed by 500 mg seven days later. Cmax and AUC for doses of 500 and 1000 mg of DAL are 133 μg/mL and 312 μg/mL and 11,393 μg·h/mL and 27,103 μg·h/mL, respectively [8].

Solon et al. studied the diffusion of 20 mg/kg of DAL in bone tissue and periarticular structures by administering radioactive [14C]-DAL to rats. Over a 14-day period, the mean bone-to-plasma concentration ratio was 0.63, and the AUC in bone was 1125 μg eq·h/mL [15]. Later, in a phase-1 trial, Dunne et al. showed that DAL concentrations in cortical bone 12 h and 2 weeks after a single infusion of 1000 mg of DAL were 6.3 μg/g and 4.1 μg/g, respectively. In that study, the bone-to-plasma AUC ratio was determined to be 0.13. Of interest, based on population PK modeling, that study proposed a DAL

regimen consisting of two 1500-mg intravenous infusions 1 week apart, which would provide concentrations in bone above the MIC90 for staphylococci for at least 8 weeks [9].

There is little information regarding intracellular concentrations of DAL. In macrophages, it has been observed to be higher than vancomycin and teicoplanin [13]. Still, we are not aware of studies on the activity against intracellular bacteria, which may be important reservoirs of infection in the setting of biofilm-associated infections [16].

In contrast with other glycopeptides (i.e., vancomycin, teicoplanin), one-third of the dose of DAL was observed to be excreted unchanged into urine, suggesting that additional non-renal pathways of elimination, probably feces, are important, as demonstrated previously in rat models [17]. Although dose adjustment does not seem necessary for mild renal impairment, patients with creatinine clearance <30 mL/min would need dose adjustment. In contrast, hemodialysis is not an important route of elimination of DAL, so that dose adjustment is not required as described in the summary of product characteristics (Xydalba; https://www.ema.europa.eu (accessed on 27 May 2021)). DAL is neither a substrate, nor an inhibitor or inducer of liver CYP-450. DAL does not require dosage adjustment in patients with hepatic impairment either [18].

#### *3.2. DAL Pharmacodynamics*

#### 3.2.1. Mechanism of Action and Determination of In Vitro Activity of Dalbavancin

DAL is a semisynthetic drug, structurally derived from the natural glycopeptide A40926 produced by *Nonomuraea* spp. [19], and its structure is closely related to teicoplanin. DAL inhibits the late stages of peptidoglycan synthesis interrupting bacterial cell wall synthesis by binding to the terminal D-alanyl–D-alanine terminus of pentapeptide peptidoglycan precursors [20].

Determination of DAL minimum inhibitory concentration (MIC) must be made by the standard broth microdilution method in cation-adjusted Mueller–Hinton broth supplemented with 0.002% (*v*/*v*) polysorbate-80. In addition, the gradient diffusion method procedure (Etest®) can be used as an alternative that has also demonstrated a high degree of agreement with the standardized broth microdilution method (EUCAST: The European Committee on Antimicrobial Susceptibility Testing. Breakpoint tables for interpretation of MICs and zone diameters. Version 11.0, 2021. http://www.eucast.org (accessed on 27 May 2021) [21]. The disk diffusion method and the agar dilution method are unreliable for the determination of susceptibility to dalbavancin.

The European Committee on Antimicrobial Susceptibility Testing (EUCAST) has defined the breakpoints for interpretation of DAL MICs only against *Staphylococcus* spp., *Streptococcus* groups A, B, C, and G, and *Streptococcus anginosus* group (*S. anginosus*, *S. intermedius*, *S. constellatus*), with those isolates with DAL MICs of ≤0.125 mg/L being susceptible, and those with dalbavancin MIC values of >0.125 mg/L being resistant. In addition, EUCAST has established DAL PK/PD non-species related breakpoints, with the isolates with DAL MICs of ≤0.25 mg/L being susceptible and those with MICs > 0.25 mg/L being resistant.

3.2.2. In Vitro Activity of Dalbavancin against Planktonic Gram-Positive Microorganisms

DAL is bactericidal against most Gram-positive microorganisms commonly involved in the etiology of PJI (essentially *Staphylococcus* spp., *Streptococcus* spp. and *Enterococcus* spp.).

Data from worldwide collections of strains have shown very low DAL MIC values. Of interest, the most recent data on behalf of DAL activity against all these microorganisms, as of January 2021, show that DAL MIC90 values have remained stable, being ≤0.06 μg/mL against different species [22]. In *S. aureus*, resistance to dalbavancin is exceptional, and the MIC90 is 16-fold lower than that of vancomycin (VAN) (0.06 μg/mL vs. 1 μg/mL) [23]. DAL activity has also been observed to be the same irrespective of oxacillin susceptibility [20,24], in contrast to coagulase-negative staphylococci (CoNS), which show a DAL MIC90 of 0.06 and 0.12 μg/mL for strains susceptible and resistant to oxacillin, respec-

tively [25]. Since almost all *S. aureus* strains that are vancomycin-susceptible are also DAL-susceptible, vancomycin susceptibility can be considered a surrogate marker of DAL activity. Consequently, vancomycin-resistant *S. aureus* (VRSA) is also resistant to DAL, and its usefulness against heteroresistant vancomycin-intermediate *S. aureus* (hVISA) is currently a matter for debate [26]. The loss of susceptibility against other anti-Gram-positive antibiotics (i.e., teicoplanin, telavancin, daptomycin, and linezolid) does not correlate with a decrease in DAL activity [27]. In the case of CoNS, Cercenado et al. observed that DAL maintained its activity even against teicoplanin-resistant strains, as long as teicoplanin MIC was ≤8 μg/mL. (P1500: XXVII European Congress of Clinical Microbiology and Infectious Diseases; 22–25 April 2017; Vienna, Austria). In summary, according to published data, DAL is very active against *Staphylococcus* spp. with MIC90 values below the EUCAST susceptibility breakpoint.

Regarding enterococci, DAL activity against vancomycin-susceptible enterococci is comparable to that of staphylococci, although vancomycin-resistant *Enterococcus* spp. pose a challenge for DAL, as this antimicrobial is not active against isolates exhibiting the VanA phenotype. However, DAL is active against strains displaying the VanB phenotype (vancomycin-resistant, with variable susceptibility to teicoplanin), showing MIC90 values around 1 μg/mL, and it is also active against strains of *E. gallinarum* and *E. casseliflavus* that express the VanC phenotype, characterized by intrinsic resistance to vancomycin, but susceptibility to teicoplanin. Overall, it can again be assumed that vancomycin susceptibility is a good surrogate marker of DAL susceptibility in *Enterococcus* spp. and that teicoplanin susceptibility can also be used as a surrogate in vancomycin-resistant strains [23,24].

DAL activity against *Streptococcus* spp. (including penicillin-resistant *viridans* group isolates, penicillin-resistant *S. pneumoniae*, *S. anginosus* group, and ß-haemolytic streptococci) is very high. Resistance to DAL in streptococci is anecdotal, as MIC90 values are below 0.3 μg/mL for *S. viridans* and 0.12 μg/mL for *S. agalactiae* [23,28–30]. Finally, DAL has also been found to be active against other Gram-positive microorganisms eventually found to be the cause of PJI. MIC90 values for *Corynebacterium* spp. range between <0.03 and 0.5 μg/mL, and DAL also shows bactericidal activity against anaerobic Gram-positive cocci, such as *Peptostreptococcus* spp., *Finegoldia magna,* and *Anaerococcus* spp., with MIC90 ranging from 0.12 to 0.5 μg/mL [23,30,31]. Concerning its activity against *Cutibacterium acnes* (formerly *Propionibacterium acnes*), Goldstein et al. [30], in a study including 15 isolates, communicated MICs ranging from 0.03 to 0.5 mg/L, with MIC50 and MIC90 values of 0.25 and 0.5 mg/L, respectively.

As indicated above, it is important to note that EUCAST (www.eucast.org (accessed on 27 May 27 2021)) has not defined a DAL breakpoint for *Corynebacterium* spp. and for anaerobes and defines a non-species related PK/PD breakpoint for DAL of ≤0.25 μg/mL. In this regard, antimicrobial susceptibility testing should be performed in all of the abovedescribed organisms with MIC90 values of 0.5 mg/L.

3.2.3. Activity of DAL against Biofilms of Gram-Positive Microorganisms: In Vitro Experience

A number of studies evaluating the activity of DAL against biofilm formation and eradication are summarized in Table 1 [30–35]. Overall, antimicrobial susceptibility studies on 96-well microtiter plates have shown that very low DAL concentrations are able to inhibit biofilm formation in a very large number of strains of staphylococci (both methicillin-susceptible and methicillin-resistant), streptococci, and enterococci (MBIC90 < 1 μg/mL). These values were lower than those observed for other antimicrobials such as vancomycin (MBIC90 2–4 μg/mL), tedizolid, and daptomycin. Concentrations needed to eradicate biofilm are higher, with MBBC90 ranging from 1 to 16 μg/mL depending on the species, but they were still much lower relative to other comparators (vancomycin MBBC90 > 32–128 μg/mL). The exception were vancomycin-resistant enterococci, which showed very high MBIC90 and MBBC90 for all the anti-Gram-positive antimicrobials tested, including DAL. Regardless of the vancomycin resistance type (VanA or VanB pheno-

#### types), all vancomycin-resistant enterococci had dalbavancin MICs, MBICs, and MBBCs > 16 μg/mL [33].

**Table 1.** Summary of in vitro and in vivo pre-clinical models of dalbavancin activity against biofilm-embedded bacteria.


DAL: Dalbavancin. DAP: Daptomycin. VAN: Vancomycin. LNZ: Linezolid. TDZ: Tedizolid. CLX: Cloxacillin. RIF: Rifampin. MRSA: Methicillin-resistant *Staphylococcus aureus*. MSSA: Methicillin-susceptible *S. aureus*. MRSE: Methicillin-resistant *Staphylococcus epidermidis*. MSSE: Methicillin-susceptible *S. epidermidis*. VSE: Vancomycin-susceptible enterococci. VRE: Vancomycin-resistant enterococci. CV: Crystal violet. MIC: Minimal inhibitory concentration. <sup>1</sup> Pegged-lids confronted to 96-well microtiter plates. MBIC (minimal biofilm inhibitory concentration) is determined by turbidity after confronting the pegs with antibiotics. MBBC (minimal biofilm bactericidal concentration) is determined after incubating the pegged-lid in 96-well microtiter plates with fresh media after having confronted the pegs with antibiotics. <sup>2</sup> MBC was defined as a 50%-reduction in the optic density value as compared with positive controls in the 96-well microtiter plate. <sup>3</sup> PK of 80 mg/kg is comparable to data observed in blister of patients after a single dose of DAL 1000 mg (Cmax 67 μg/mL, and AUC0–7d 6438 μg·h/mL) [41].

Information regarding the anti-biofilm activity of DAL using models with long exposure times was almost non-existent until the end of the last decade, when Di Pilato et al. evaluated the time-kill kinetics of DAL against biofilms of nine clinical strains of *S. aureus* and CoNS, using both a standardized biofilm model and biofilms grown on titanium and cobalt-chrome disks. DAL and vancomycin were used at concentrations of 1, 4, and 16 μg/mL. Against biofilms formed over 7 days on microtiter plates, the response to antibiotics was heterogeneous, although DAL showed faster and greater reduction of biofilm-embedded bacteria in the majority of the strains studied, especially at concentrations of 4 μg/mL and 16 μg/mL. In biofilms formed on Ti and Co-Cr disks, DAL was more active than vancomycin at medium concentrations (4 μg/mL), which may be expected in bone tissue [35].

More recently, Žiemyte et al. proposed a real-time, impedance-based cell analysis in ˙ order to facilitate the determination of antimicrobial susceptibility when bacteria grow in biofilms [37]. In this study, DAL ability to prevent *S. aureus* and *S. epidermidis* biofilm formation was compared with that of other antimicrobials commonly used for treating PJI (linezolid, rifampin, vancomycin, cloxacillin). The MBIC of DAL ranged from 0.5 to 2 μg/mL, and in combination with rifampin showed the highest biofilm inhibitory effect. With respect to the eradication of 6- to 9-h biofilm, DAL stopped or reduced biofilm formation at concentrations of 8–32 μg/mL. The other antimicrobials showed no activity against biofilm formed by *S. aureus*. For biofilms of *S. epidermidis*, low concentrations of DAL were active, although less than the combination of cloxacillin plus rifampin.

3.2.4. Activity of DAL against Biofilm of Gram-Positive Microorganisms: Experimental In Vivo Experience

A few in vivo experimental models [37–39,42] have assessed the efficacy of DAL in biofilm prevention and treatment (Table 1). Darouiche et al. compared DAL, vancomycin, and a placebo for preventing colonization of subcutaneously placed devices in a rabbit animal model inoculated with 10<sup>3</sup> colony-forming units of *S. aureus*. Although not statistically significant, there was a trend toward a lower colonization rate in rabbits that received DAL before the procedure [38]. Nevertheless, the rate of foreign body contamination in rabbits receiving placebo was around 50% (lower as compared with other animal models), thus questioning the validity of the model and its discriminatory power for assessing the efficacy of antimicrobials.

In 2013, Baldoni et al. tested the ability of DAL to eliminate methicillin-resistant *S. aureus* (MRSA) biofilms in an animal model of tissue-cage infection. DAL and rifampin were administered intraperitoneally, the former at different doses (20, 40, and 80 mg/kg, which produced AUC0–7d of 3393, 4298, and 4464 μg·h/mL, respectively). In monotherapy, DAL yielded a very modest killing, but in combination with rifampin, eradicated infection in one third of the cages. Of note, only the higher dosage (80 mg/kg) of DAL was able to prevent the development of rifampin resistance [39].

More recently, Barnea et al. studied the efficacy of DAL for the treatment of sternal osteomyelitis and mediastinitis caused by MRSA using a median sternotomy model in Lewis rats. The efficacy of DAL was proven to be similar to that of vancomycin for the treatment of sternal osteomyelitis and superior to placebo, and also reduced systemic dissemination of staphylococcal infection. DAL concentrations in bone tissue after 10 days of administration were 10.7 μg/g [40].

The models of animal infection suggest a role for DAL in the PJI setting, although some concerns arise after a thorough study of their results. First, in contrast to many of the in vitro studies previously reviewed, the dosages of DAL in some of the in vivo models may have provided lower antibiotic exposure compared to human PK. Second, more data on the combination of DAL with rifampin and comparisons with other rifampin-based combinations would be welcome in order to place DAL in the armamentarium of PJI caused by Gram-positive microorganisms.

#### 3.2.5. Clinical Experience with DAL for Treating Prosthetic Joint Infections

As stated above, the broad antimicrobial spectrum of DAL and its PK properties support its use outside its approved indications. DAL is an attractive alternative in scenarios such as bloodstream infections, endocarditis, and osteomyelitis [10,43–46], even though clinical trials exploring these off-label indications of DAL are scarce.

However, in a randomized clinical trial, Rappo et al. explored the efficacy and safety of DAL for the treatment of osteomyelitis known or suspected to be caused by Gram-positive pathogens [46]. In that single-center study conducted in the Ukraine, DAL was compared with the standard of care (vancomycin was the most frequently used comparator) and the primary endpoint was clinical response at day 42. Failure was defined as the requirement of additional antibiotics, new purulence, the need for new surgery, and/or amputation. A clinical cure at day 42 was 97% in the DAL arm compared to 88% in the standard of care. Reported follow up only extends to 1 year. Even though the patients included did not have orthopedic hardware, the results are encouraging for the use of DAL in the treatment of osteitis persisting after prosthesis removal, in other words, in the setting of a two-step exchange procedure.

Meanwhile, scattered cases have been reported [47]. Furthermore, Buzón-Martín et al. reported their experience of 16 cases of PJI treated with DAL, which is so far the largest single-institution report [48]. Brief details of surgical strategies and antimicrobial treatment were provided. Overall, so as to now, 88% of patients had their infection resolved and there were no major adverse events (Buzón-Martín, unpublished data).

In addition, a number of case series with real-world experience with DAL have been published, also including cases of PJI (Table 2) [45,48–52]. Common limitations found in these case series are the inclusion of small sample sizes, patient heterogeneity, aggregate outcomes of patients with PJI along with other orthopedic-related infections, and lack of details about surgical management. In fact, the goals and difficulties of treatment vary considerably depending on the type of PJI (acute vs. chronic) and whether the prosthesis is retained or removed. The main objectives of the treatment of PJI are to eradicate infection and maintain a pain-free prosthetic joint. In this context, one of three major strategies can be chosen when faced with a given PJI: To attempt eradication and cure with prosthesis retention (debridement, antibiotics, and implant retention—DAIR), attempt eradication and cure with prosthesis removal (followed by prosthesis reimplantation in either a one- or two-stage exchange procedure, or else a joint arthrodesis), or prosthesis retention, abandoning the attempt to eradicate the infection in favor of chronic suppressive antimicrobial therapy [53]. Bearing this in mind, a given antibiotic can perform very differently depending on which surgical strategy has been chosen.


**Table 2.** Clinical series published on the experience with DAL, including cases of bone and joint infection and prosthetic joint infection.

NP: not provided. PJI: prosthetic joint infection.

An additional limitation of these studies is the wide heterogeneity in the use of DAL, even within the same institutions. Loading doses on day 1 ranged from 1000 mg to 1500 mg, and following doses at day 7 ranged from 500 to 1500 mg. The number of doses was also very variable, as some patients were treated with just two doses after prosthesis removal and others received more than 20 doses in the setting of a suppressive strategy [45,48,49,54]. Some authors [48] have even suggested that a biweekly administration strategy might be

useful in this setting. As mentioned before, Dunne et al. [9] settled the rationale basis for a weekly administration of two doses of 1500 mg of DAL, and Rappo et al. proved its efficacy for treating osteomyelitis [46]. Noteworthily, these two 1500 mg doses on day 1 and day 7 of the scheme were only used in 6 out of 12 cases in the Graz series [49], but were not used in the series of Buzón-Martín, Tobudic, and Morata [45,48,50]. So far, the ideal dosing strategy of DAL for PJI remains unanswered, but perhaps two 1500 mg doses on day 1 and 7 after prosthetic removal is the scheme with a more solid investigational and clinical evidence backup [9,46].

In order to overcome the limitation of the studies heterogeneity, we contacted the authors of three of the above-mentioned case series. Dr. Zollner-Schwetz, Dr. Tobudic, and Dr. Buzón-Martín kindly provided more specific data of 36 patients treated at their institutions (Table 3). The majority of patients had already been given other antimicrobials and undergone previous surgeries, and DAL was used as salvage therapy, thus facing greater challenges. The reported etiologies were also heterogeneous, and half the patients were given DAL in combination. DAIR management was anecdotal in these cases and was only performed in two patients. The majority of infections were treated with prosthesis removal (27/36, 75%), a strategy that led to a success rate of 25/27 (92.6%) after a median follow up of 16 months. Within this group, 20/27 (74%) patients were treated with a two-stage revision procedure, two (7.4%) with single-stage revision, and three (11.1%) patients with resection arthroplasty. Of interest, a number of patients were treated with prosthesis retention plus DAL as suppressive antimicrobial therapy (7/36, 19.4%) with successful retention of the prosthesis in the short term in three cases (42.9%). Although large series of suppressive treatment with DAL for other conditions are lacking, there is some evidence to suggest that DAL can be safely administered as compassionate treatment for several months, or even years for non-surgical prosthetic endocarditis (Dr. Buzón-Martín, unpublished data).


**Table 3.** Cases of PJI treated with dalbavancin according to the surgical strategy adopted (data from Buzón et al., Tobudic et al., and Wunsch et al.).

\* Continuous variables are expressed as median and (range). <sup>1</sup> Data available for 20 patients. <sup>2</sup> There were 4 methicillin-susceptible strains (3 managed with prosthesis removal and 1 by suppressive antimicrobial therapy), and 2 methicillin-resistant strains (both managed by prosthesis removal). <sup>3</sup> There were 4 *E. faecium* (all treated with prosthesis removal) and 1 *E. faecalis* (treated with suppressive antimicrobial treatment). <sup>4</sup> One patient died to unrelated causes after three months with no clinical or biochemical signs of failure. Abbreviations: DAL: dalbavancin. CoN: coagulase-negative. GP: Gram-positives. DAIR: debridement, antibiotics, and implant retention.

> Overall, these revisited cases suggest that there is still insufficient experience with the use of DAL in the setting of DAIR, but that good results can be expected in the case of

prosthesis removal. The use of DAL as chronic suppressive therapy could be considered in very carefully selected situations when other alternatives are lacking, although we still need more experience and information regarding the most suitable and sustainable dosage. Finally, as expected, we still need to find out which is the best DAL dosing schedule for treating PJI.

#### *3.3. DAL as a Cost-Saving Strategy*

Cost-saving is an additional issue, which probably justifies the use of DAL in patients with PJI. In the DALBUSE study, Bouza et al. found DAL to be cost-saving [51], and Buzón-Martín et al. observed that the use of DAL allowed an early discharge of most patients, with a presumably relevant impact in terms of healthcare costs. Applying the same cost analysis previously reported by Bouza et al. in the DALBUSE study, an estimated 571 days of hospitalization were avoided and a total of US \$264,769 saved [48].

Several other reports position DAL as a cost-saving alternative [55,56], although, in a recent study, the results of González et al. pointed in the opposite direction [57], finding DAL to be more expensive than the standard of care for the treatment of skin and soft tissue infections. Nevertheless, in the same journal, Bookstaver et al. replied with more specific considerations other than cost and calling for other issues to be taken into account when thinking about antimicrobial stewardship [58]. It is also important to state that cost-saving analyses are quite difficult to extrapolate from the USA to other health systems in Europe, mainly those that are 100% public.

#### **4. Conclusions**

DAL's unique PK properties and high bactericidal activity are attractive characteristics for the treatment of bone and joint infections, including PJI. The possibility of using DAL in an outpatient setting, with the associated cost-saving impact, as well as the obvious improvement in therapeutical adherence compared with oral treatments, increases its value in infections where long treatments are necessary.

With regard to this, although the specific DAL concentrations used in pre-clinical models are not always consistent with human PK, and there is very scarce information on intracellular activity, the results of DAL against biofilm-embedded bacteria are encouraging. In addition, a randomized clinical trial states that DAL is non-inferior to the standard of care in bone infections with no orthopedic hardware. The reported clinical experiences of use of DAL in PJI are scarce and heterogeneous, but its use in the setting of prosthesis removal seems reasonable and effective. We still need more data regarding its use in the setting of prosthesis retention, and also in combination with established antimicrobials such as rifampin.

**Author Contributions:** Conceptualization, L.B.-M. and J.L.-T.; methodology, L.B.-M., I.Z.-S., S.T., E.C. and J.L.-T.; software, L.B.-M. and J.L.-T.; validation, L.B.-M., I.Z.-S., S.T., E.C. and J.L.-T.; formal analysis, L.B.-M.; investigation, L.B.-M., I.Z.-S., S.T., E.C. and J.L.-T.; resources, L.B.-M., I.Z.-S., S.T., E.C. and J.L.-T.; data curation, L.B.-M., I.Z.-S., S.T. and J.L.-T.; writing—original draft preparation, L.B.-M.; writing—review and editing, L.B.-M. and J.L.-T.; visualization, L.B.-M., E.C. and J.L.-T.; supervision, L.B.-M. and J.L.-T.; project administration, L.B.-M. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Ethical review and approval were waived for this study, as this is a review of previously published studies.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** We thank Janet Dawson for reviewing the English Manuscript. We are also indebted to Florian Thalhammer for his collaboration.

**Conflicts of Interest:** J.L.-T. and L.B.-M have received conference grants from Angelini. I.Z.-S. received conference grants from Angelini and served on an advisory board for Angelini.

#### **References**


## *Communication* **Risk Factors of Daptomycin-Induced Eosinophilic Pneumonia in a Population with Osteoarticular Infection**

**Laura Soldevila-Boixader 1,2, Bernat Villanueva 1, Marta Ulldemolins <sup>1</sup> , Eva Benavent 1,2 , Ariadna Padulles 3, Alba Ribera 1,2, Irene Borras 1, Javier Ariza 1,2,4 and Oscar Murillo 1,2,4,\***


**Abstract:** Background: Daptomycin-induced eosinophilic pneumonia (DEP) is a rare but severe adverse effect and the risk factors are unknown. The aim of this study was to determine risk factors for DEP. Methods: A retrospective cohort study was performed at the Bone and Joint Infection Unit of the Hospital Universitari Bellvitge (January 2014–December 2018). To identify risk factors for DEP, cases were divided into two groups: those who developed DEP and those without DEP. Results: Among the whole cohort (*n* = 229) we identified 11 DEP cases (4.8%) and this percentage almost doubled in the subgroup of patients ≥70 years (8.1%). The risk factors for DEP were age ≥70 years (HR 10.19, 95%CI 1.28–80.93), therapy >14 days (7.71, 1.98–30.09) and total cumulative dose of daptomycin ≥10 g (5.30, 1.14–24.66). Conclusions: Clinicians should monitor cumulative daptomycin dosage to minimize DEP risk, and be cautious particularly in older patients when the total dose of daptomycin exceeds 10 g.

**Keywords:** daptomycin; eosinophilic pneumonia; risk factors

#### **1. Introduction**

Daptomycin is a cyclic lipopeptide antibiotic approved for use against complicated skin and soft tissue infection, *Staphylococcus aureus* bacteremia and right-sided infective endocarditis. However, daptomycin has become widely used also in staphylococcal osteoarticular infections because of its remarkable anti-biofilm activity. Indeed, current guidelines advise for its use mainly as an initial induction course of intravenous antimicrobial therapy and often in combination with other antibiotics to avoid the appearance of resistance [1,2]. In this setting, the use of daptomycin for prolonged periods should be balanced between the potential benefits in the outcome and the risk of adverse events [3–5].

Although daptomycin has proven safety, daptomycin-induced eosinophilic pneumonia (DEP) is a rare but severe adverse effect [6,7]. This toxicity is partially related to the usual daptomycin uptake by pulmonary surfactant in the alveoli, which may lead to concentrations high enough to cause injury but also to impair its efficacy; in fact, daptomycin is not recommended to treat pulmonary infections. Despite the fact that the pathophysiology is not totally clear, it seems that DEP is an antigen-mediated process in which alveolar macrophages and T-cells may be activated, which then release interleukin-5 that causes eosinophil production and migration to the lungs. Additionally, alveolar macrophages

**Citation:** Soldevila-Boixader, L.; Villanueva, B.; Ulldemolins, M.; Benavent, E.; Padulles, A.; Ribera, A.; Borras, I.; Ariza, J.; Murillo, O. Risk Factors of Daptomycin-Induced Eosinophilic Pneumonia in a Population with Osteoarticular Infection. *Antibiotics* **2021**, *10*, 446. https://doi.org/10.3390/ antibiotics10040446

Academic Editor: Maria Mezzatesta

Received: 26 February 2021 Accepted: 14 April 2021 Published: 16 April 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

can also excrete cytokines that selectively recruits eosinophils, which may promote further eosinophil accumulation into the lungs [8,9].

Since the introduction of daptomycin, while some cases of DEP have been reported, these have only described the most common clinical manifestations and outcome [10–13]. To date, therefore, we do not know which factors are associated with DEP and thus, in the present study we aimed to determine the risk factors for developing DEP.

#### **2. Results**

In total, 229 cases received at least one dose of daptomycin and among them, 11 (4.8%) had DEP; a comparison of both groups in regard with main clinical and analytical characteristics is presented in Table 1. All DEP cases underwent a chest X-ray while on daptomycin therapy, which showed peripheral lung infiltrates (alveolar or interstitial), and only one patient had a CT scan that showed radiological findings of organizing pneumonia. In contrast, only 26% cases (57/218) of the remaining cohort underwent a chest X-ray, which was considered similar to the baseline one. Of interest, the performance of a chest X-ray significantly increased in accordance with the length of daptomycin therapy, ranging from 21% in cases treated less than 7 days to 42% in those treated more than 14 days (*p* = 0.005). With regard to the age of patients, cases aged ≥70 years underwent a chest X-ray during daptomycin therapy in greater proportion than younger patients (31% vs. 24%, respectively).

**Table 1.** Analysis of risk factors for daptomycin-induced eosinophilic pneumonia (DEP).


Analytical data is presented as median, IQR. The remaining data are presented as *n* (%) unless otherwise noted. <sup>1</sup> TCDD (Total Cumulative Dose of Daptomycin; daily dose X days of treatment; The result was expressed in grams-g-) <sup>2</sup> Cases without DEP in which creatine kinase values were analyzed had a median of 12 days (IQR 6–19.5) of daptomycin therapy.

All DEP cases were treated with daptomycin withdrawal and seven (64%) with corticosteroid therapy. One patient, who had a delay in diagnosis of DEP and therapy, died because of respiratory failure.

In the univariate analysis (Table 1), factors associated with DEP were advanced age, the presence of comorbidities measured by Charlson score, long treatment with daptomycin and high values of TCDD. Concisely, daptomycin therapy for two weeks or longer was associated with high risk of DEP (HR 7.71, 95%CI 1.98–30.09), as well as TCDD values ≥10 g (HR 5.30, 95%CI 1.14–24.66). The presence of blood eosinophilia at the end of daptomycin treatment was significantly higher in DEP cases than in controls (82% and 16%, respectively; *p* < 0.001), as well as leucocyte counts and C-reactive protein values were also higher in DEP cases.

We noted that among older patients aged ≥70 years (*n* = 123), the percentage with DEP (8.1%; 10/123) almost doubled the value of the whole cohort. Also, the percentage of cases with DEP increased significantly among cases aged ≥70 years in comparison with the whole cohort either in cases treated for >14 days or in those with high values of TCDD (Figure 1).

**Figure 1.** Percentage of Daptomycin-induced eosinophilic pneumonia (DEP) cases in the whole cohort and in those aged ≥70 years by (**a**) Length of therapy and by (**b**) The total cumulative dose of daptomycin (TCDD).

Finally, 25 cases had a re-challenge to daptomycin therapy, and two of these presented promptly with DEP (8%) by 4 and 8 days after the re-challenge (3 and 5 months after the first exposure, respectively). Both cases presented blood eosinophilia after the first course of treatment, having received >11 g over >14 days. By contrast, among the remaining patients re-challenged with daptomycin, the eosinophilia was only observed at the prior exposure for four patients (17%).

#### **3. Discussion**

In the present study we reported the main risk factors for developing DEP in a population with osteoarticular infections, providing important new information that may be helpful to clinicians.

Daptomycin has been reported as the leading cause of drug-induced eosinophilic pneumonia [14], and clinicians should maintain a high index of suspicion for DEP because of its potential severity. Although most of cases in our series were resolved by daptomycin withdrawal and corticosteroid therapy, one patient died, which illustrates the inherent risk of failing to identify DEP promptly.

Daptomycin use for the treatment of osteoarticular infection is currently recommended mainly against staphylococcal infections and as initial induction antimicrobial therapy [1,2]. In contrast with the high activity of daptomycin in animal and in vitro studies, its clinical efficacy reported from non-comparative studies appeared to be quite similar to other therapies [5,15,16]. However, prolonged therapy at higher doses than usual seems to be increased in recent years. Our cases were treated with daptomycin for a median of 19 days and resulted in DEP proportions of 4.8% overall and 8.1% among those aged ≥70 years. This data may seem high compared with previous experiences and without placing it in context. Thus, populations with 102 cases of infective endocarditis and 43 cases of complex osteoarticular infections that received high doses of daptomycin (median 8.2 mg/kg/d) for long periods (20–80 days), the authors showed 3% and 4.6% developed DEP, respectively [17,18]. These are consistent with our results given that the populations in both studies were younger (mean age 61.5 years) than in the present study.

Taking all the previous into account, the prolonged therapy with daptomycin against osteoarticular infections or the use of higher doses than usual should be considered on the basis of a list of pros and cons. Probably, the efficacy of daptomycin therapy is related with its anti-biofilm activity and can be benefited through an initial intensive phase of treatment (i.e., 7–14 days). Further therapy should be balanced with inconveniences derived from its use; indeed, monitoring for daptomycin toxicity appears crucial in long therapies and includes not only the risk for DEP but also other adverse events such as rhabdomyolysis. In our experience, performance of chest-X ray was useful to identify DEP and thus, it appears as valid screening to be interpreted together with other clinical signs and analytical parameters.

To our knowledge, no previous studies had been performed to analyze the risk factors of developing DEP. We identified advanced age, high values of Charlson comorbidity index, length of daptomycin therapy and TCDD as the main risk factors for DEP. Of interest, we show that patients older than 70 years, which commonly have more underlying diseases, are at higher risk of DEP; however, further research is needed to evaluate the importance of particular comorbidities in increasing the risk of DEP.

Regarding cumulative dosages of daptomycin and long therapies, our results seem to be consistent with previous works. Hirai et al. [19] reported 40 cases of DEP, 73% of them received a daptomycin dosage >6 mg/kg/d for a median of 14.8 days, whereas the remaining cases were treated with daptomycin at ≤6 mg/kg/d for a median of 23 days. In a systematic review of DEP cases the mean length of daptomycin therapy was 2.8 weeks and main indication for treatment was osteoarticular infection [20]. Overall, it seems that higher risk of DEP is not only dose dependent but also time-dependent. We therefore recommend monitoring the cumulative dose of daptomycin, which is a product of the dosage and length of therapy, rather than considering either variable separately. In our experience, clinicians should be cautious when the TCDD is ≥10 g, and particularly if it increases to ≥15 g, which can be easily attained after 2 weeks of treatment in patients receiving high doses.

Cases with DEP at the end of therapy had higher blood eosinophil counts and more often eosinophilia than controls, a fact that has been mainly reported previously [20,21]. Of interest, we noted a scenario in which severe DEP occurred shortly after a re-challenge with daptomycin, indicating that a drug hypersensitivity mechanism may be play. These cases presented with eosinophilia at the end of their previous course of daptomycin, a finding that was rarely observed in patients given a rechallenge without developing DEP. This clinical situation has been poorly reported to date [22], but it seems that eosinophilia during daptomycin therapy should prompt clinicians to consider avoiding further drug exposure.

The main limitations of the study are those inherent to the retrospective design. Generalizability is affected because patients were recruited from a single center and because the cohort mostly comprised elderly people with heterogeneous clinical presentations of

osteoarticular infections. Also, unfortunately, our sample size of DEP cases was small to allow subgroup analyses or to design other comparative study. These factors must be factored when considering other heterogeneous populations. Irrespective of these shortcomings, however, we believe that our results provide information that can be led to improved management of daptomycin therapy.

#### **4. Materials and Methods**

#### *4.1. Study Design, Setting, and Inclusion/Exclusion Criteria*

This retrospective cohort study was performed at the Bone and Joint Infection Unit of the Hospital Universitari Bellvitge between January 2014 and December 2018. We included all patients with osteoarticular infection (prosthetic joint infection, septic arthritis and osteomyelitis), aged ≥18 years, and treated at least with one dose of daptomycin because of empirical treatment or guided therapy addressed to Gram-positive microorganisms. Polymicrobial osteoarticular infections treated with daptomycin in combination with other antibiotics were also included. We excluded cases attended in our Bone and Joint Infection Unit that received daptomycin due to causes different than osteoarticular infections (i.e., catheter-related sepsis).

To identify risk factors for DEP, cases were divided into two groups: Those who developed DEP and those without DEP.

Written informed consent was considered unnecessary for the study, as it was a retrospective analysis of our clinical practice. Data of patients were anonymized for the purposes of this analysis. Confidential information of patients was protected according National and European normative. This manuscript has been revised for its publication by Research Ethics Committee of Bellvitge University Hospital (PR097/21).

#### *4.2. Definitions and Clinical Data*

All cases fulfilled the main diagnostic criteria for each osteoarticular infection, including those with prosthetic joint infection or osteoarthritis, with or without an orthopedic device.

The modified diagnostic criteria established by Philips et al. were used to define DEP [23], which required exposure to daptomycin with the following features: fever, dyspnea with increased oxygen requirement or requiring mechanical ventilation, new infiltrates on chest X-ray or computed tomography, and clinical improvement following daptomycin withdrawal. In accordance with these criteria, we did not require the previous pre-requisite of a bronchoalveolar lavage with >25% eosinophils.

Demographic, clinical, radiological and analytical data were collected for the included cases. Chronic heart failure, chronic pulmonary disease and chronic kidney disease were defined according to accepted criteria. The total cumulative dose of daptomycin (TCDD) was defined as daily dose of daptomycin × days of treatment; the result was expressed in grams (g).

#### *4.3. Statistical Analysis*

Data were analyzed using Stata software (version 16.0, Stata Corporation, College Station, TX, USA). Categorical variables are described by counts and percentages, while medians and interquartile ranges (IQRs) are used to summarize continuous variables.

Univariate analysis was performed to screen the risk factors for DEP, and logistic regression models were built to estimate unadjusted hazard ratios (HR). In all situations, *p*-values of <0.05 were considered to be statistically significant.

#### **5. Conclusions**

In conclusion, main factors associated with DEP were advanced age, high values of Charlson score, longer treatments and high total cumulative doses of daptomycin. Particularly, clinicians should take care in cases with cumulative doses greater than 10 g, which can be achieved after 2 weeks of daptomycin therapy. In this high risk population and after the beginning of treatment, performing a chest-X ray is useful to identify DEP. Where eosinophilia has previously occurred with daptomycin exposure, further drug challenges should be considered with great care to minimize the risk of DEP.

**Author Contributions:** L.S.-B., B.V., M.U., E.B., A.P., A.R., I.B., J.A. and O.M. contributed in the supervision of the clinical cases, data collection, and interpretation; L.S.-B. and O.M. elaborated the study design; L.S.-B. performed the analysis of the data and wrote the first draft of the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** L.S-B. was supported with a grant of the Ministerio de Ciencia, Innovación y Universidades (FPU (18/02768). This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. This research received no external funding.

**Institutional Review Board Statement:** The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of Bellvitge University Hospital (protocol code PR097/21).

**Informed Consent Statement:** Patient consent was waived due to the retrospective nature of the study on the usual clinical practice.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author (omurillo@bellvitgehospital.cat).

**Acknowledgments:** We thank Dolors Rodriguez-Pardo from Hospital Vall d'Hebron, Isabel Mur from Hospital de la Santa Creu i Sant Pau and Rosa Escudero from Hospital Ramon y Cajal for their collaboration with the manuscript. We thank Michael Maudsley for revising the English manuscript. We thank CERCA Program/Generalitat de Catalunya for institutional support. The preliminary results of this study were reported in part at the 30th European Congress of Clinical Microbiology & Infectious Diseases (Paris, France, 2020).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Javier Cobo and Rosa Escudero-Sanchez \***

Infectious Disease Department, Hospital Ramón y Cajal, IRYCIS, Ctra. Colmenar Viejo, 28034 Madrid, Spain; javier.cobo@salud.madrid.org

**\*** Correspondence: rosa.escudero0@gmail.com

**Abstract:** The treatment of prosthetic joint infections (PJIs) is a complex matter in which surgical, microbiological and pharmacological aspects must be integrated and, above all, placed in the context of each patient to make the best decision. Sometimes it is not possible to offer curative treatment of the infection, and in other cases, the probability that the surgery performed will be successful is considered very low. Therefore, indefinite administration of antibiotics with the intention of "suppressing" the course of the infection becomes useful. For decades, we had little information about suppressive antibiotic treatment (SAT). However, due to the longer life expectancy and increase in orthopaedic surgeries, an increasing number of patients with infected joint prostheses experience complex situations in which SAT should be considered as an alternative. In the last 5 years, several studies attempting to answer the many questions that arise on this issue have been published. The aim of this publication is to review the latest published evidence on SAT.

**Keywords:** suppressive antibiotic treatment; prosthetic joint infection; prolonged antibiotic

**Citation:** Cobo, J.; Escudero-Sanchez, R. Suppressive Antibiotic Treatment in Prosthetic Joint Infections: A Perspective. *Antibiotics* **2021**, *10*, 743. https://doi.org/10.3390/ antibiotics10060743

Academic Editor: Giovanna Batoni

Received: 24 May 2021 Accepted: 15 June 2021 Published: 19 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

#### **1. Therapeutic Options for Prosthetic Joint Infections**

The goal of treating a prosthetic joint infection (PJI) is to eradicate the infection and to maintain or regain implant function. This often involves the replacement of the prostheses, although in some cases (acute infections), the original implant can be salvaged through extensive debridement and prolonged antibiotic therapy, which is referred to as DAIR (debridement, antibiotics and implant retention) [1]. In the remaining situations, the cure can be obtained only by removing the implant, followed by the placement of a new prosthesis, either during the same surgical procedure (one-stage revision) or after a period with antibiotics (two-stage revision) [2]. However, reimplantation is sometimes not possible after removal (resection arthroplasty), and in rare situations, amputation may be necessary. Eventually, due to the patient's conditions or the anticipated sequelae of the intervention, a potentially curative surgical intervention is waived. In this scenario, orthopaedic surgeons turn their gaze to infectious disease (ID) consultants. Can antibiotic treatment help the patient?

#### **2. Concept and Definition of Suppressive Antibiotic Treatment (SAT)**

The term "suppressive antibiotic treatment" (SAT) refers to the administration of antibiotics in the long term or indefinitely over time. In the area of PJI, SAT is considered a "noncurative" strategy, in which antimicrobials are administered with the aim of reducing symptoms and delaying or preventing the progression of PJI that needs a surgical procedure to be cured that, for some reason, will not be performed (at least for a prolonged period of time). SAT can also be used in situations in which adequate surgical treatment is performed and the probability of cure is considered very low.

#### **3. SAT Indications**

SAT appears to be an infrequent therapeutic option in a series (5–14%) that reports the approach of patients with PJI [3–5]. However, in those patients over 80 years of age, the percentage treated by SAT can reach 36.5% [6].

SAT is intended to reduce local symptoms (presence of a sinus tract, inflammation, pain, etc.) and thus delay or elude a surgical intervention that has been rejected or is intended to be avoided. It is possible that SAT may delay or prevent prosthetic loosening by reducing the local peri-implant inflammatory process, although no studies have evaluated this potential effect. Additionally, SAT can be considered a general benefit for the patient's health as a result of the reduction in persistent chronic inflammation [7].

In summary, SAT can be considered for patients with acute PJI for whom conservative treatment (DAIR) has failed, or for patients with chronic-late PJI whose implants are not going to be removed or replaced due to any of the following circumstances:


These situations would therefore be considered PJI with "certain" treatment failure. This would mean that there is evidence of PJI with no curative treatment planned.

There are other situations in which the probability of failure of surgical-medical treatment can be anticipated to be high, although not certain [8,9]. Here, we would cite the following scenarios:


Once the indications are established, certain conditions are required to be able to carry out SAT:


#### **4. Evidence on SAT Efficacy**

#### *4.1. Does SAT Truly Work? What Results Does It Offer?*

Evidence of the efficacy of SAT is scarce. A cohort study in which patients with stable PJI (69% with implants for <90 days) were managed with implant retention and prolonged antibiotic therapy for more than 1 year showed that the failure rate (recurrence of infection or need for surgical revision) was four times higher in patients who discontinued antibiotic treatment [10]. Interestingly, most of the patients with discontinued treatment did not exhibit treatment failure, suggesting that many were actually cured. However, the higher rate of treatment failure in patients who stopped taking antibiotics indicates that, in this series, a proportion of patients not cured by DAIR benefited from continuing antibiotic treatment, via delayed or avoidance of failure, which occurred mostly in the first four months. Further arguments in favour of SAT efficacy are provided by the cases that were "rescued" through SAT after the failure of other strategies [10–12], as well as by the

observation that some SAT failures were temporarily related to the suspension of antibiotic treatment [13].

The interpretation of SAT efficacy is very difficult for three reasons: the absence of controlled studies, the inclusion of patients with acute infections who would be cured by DAIR, and differences in the criteria for evaluating efficacy in published series (Table 1). For example, for some authors, the efficacy criterion was to avoid surgery (even if infection was not controlled) [3], while others required, in addition, control of the symptoms [4,9,11,14]. Success rates varied in the different series from 23% to 84%. However, the series with the highest success rates included patients with early PJI [4,9,14], many of whom would have had the same outcome with much shorter treatments.


**Table 1.** Published Series on SAT in PJI.


**Table 1.** *Cont.*


**Table 1.** *Cont.*

CDI: Clostridioides difficile infection; CoNS: coagulase-negative staphylococci.

We found only one controlled study where patients with PJI at high risk of failure after surgery (DAIR or replacement) managed with SAT were compared with patients in the same conditions who were not managed with SAT. The cases were "matched" using a propensity score. Patients who received SAT had a better outcome at 5 years (68.5% free of infection) than those who did not receive SAT (41.1%) [16]. In a recent multicentre cohort that represents the largest series published to date, we estimated that SAT was effective (control of symptoms and no reintervention) in approximately 75% of the patients after two years and in 50% of patients at 5 years of follow-up [19]. Only patients with persistent infection from whom the implant was not removed were included in this cohort.

#### *4.2. What Factors Are Associated with SAT Failure?*

Few studies have analysed the factors associated with SAT failure. The failure rate seems higher among patients with a sinus tract and in those with infections caused by S. aureus [13,20–22].

In the multicentre study mentioned above, we investigated predictors of failure (defined as the persistence of uncontrolled symptoms of PJI, including sinus tract, or the need for further surgery for debridement or removal of the prosthesis due to infection) [19]. A multivariate analysis showed that the factors associated with failure were the following:


In our opinion, at this moment, there are no firm or clear predictors of failure, which means that SAT should not be excluded if the patient meets the conditions mentioned above.

#### *4.3. Why Could SAT Stop Working? Is the Development of Resistance Frequent?*

In our previously cited cohort study, the coinvestigators were unable to attribute the failure to any specific cause in 52% of the cases. Among the known or attributable causes, the most frequent was the abandonment of treatment or poor adherence (24% of all failures). The development of resistance was not a common cause, as it could only be invoked as a cause of failure in 12% of the cases. This observation has also been made by other authors [18]. In another 11% of patients, the cause of failure was the existence of a previously unsuspected pathogen in cultures that was not covered by the prescribed SAT [19].

#### **5. Practical Aspects of SAT**

#### *5.1. Is a Debridement Mandatory before Starting SAT?*

It seems reasonable to think that the reduction in the inoculum and the debridement of infected tissues favours the success of SAT. In most of the series, patients undergo debridement surgery before starting SAT [21]. The difficulty arises in stable patients who present few symptoms, especially if the surgical risks are high. Thus, in the series of SAT in elderly patients, only 24% were operated on [10].

In our analysis, the failure of the SAT was not associated with the absence of a previous debridement [19]. However, surgical debridement makes it possible to obtain valuable samples for microbiological culture, which is a relevant advantage since culture from sinus tracts is not usually representative of the actual aetiology [23].

#### *5.2. What Are the Most Suitable Antibiotics for SAT? Is a Combination of Antibiotics Necessary?*

From the analysis of the data available in the literature, it is not possible to infer recommendations. The most widely used antibiotic regimens in published series have been the combination of tetracyclines and rifampicin (the last cannot be used alone because of development of resistance) or monotherapy with a beta-lactam or tetracycline antibiotic [3,4,10,14]. In a recent survey of orthopaedists and ID consultants who prescribed SAT, 74% stated that they did not use rifampicin [24].

Since SAT is intended to reduce symptoms and local inflammation, which can be achieved by reducing the bacterial load, antibiotics with activity against stationary growing bacteria are probably not indispensable. In fact, monotherapy with beta-lactams was associated with better outcomes in a large series [10]. It seems reasonable to prioritize tolerability and therapeutic compliance, and for this, it is easier to use monotherapy. In the vast majority of cases, SAT is carried out with orally administered antibiotics. However, there are some recent experiences with intravenous dalbavancin, which have taken advantage of the fact that this drug can be administered once per week or even every two weeks [25], and with the use of beta-lactams such as ceftriaxone or ertapenem subcutaneously [26].

There are no studies on the optimal dosage of antibiotics in SAT. In general, low doses should not be used initially, at least until a reduction in inoculum has been achieved. However, the risks of each antibiotic–bacteria pair must be taken into account. For example, a low dosage of quinolones poses a risk of resistance selection in both staphylococci and Gram-negative bacilli; however, beta-lactam susceptible staphylococci should not develop resistance to a low dose of oral cephalosporins.

#### *5.3. Is Intravenous Treatment Necessary at the Beginning of SAT?*

Similarly, published studies do not provide an answer to this question. In almost all published series, patients receive several weeks of initial intravenous treatment, but in the aforementioned survey, most of the respondents stated that they do so only occasionally [24].

#### *5.4. Can There Be Periods Without Treatment?*

The series in the literature reviewed do not include antibiotic treatment-free periods in their protocols. In fact, in some series, failures are reported coinciding with the interruption of treatment, which, in general, appears in the first 4 months after suspension [9].

#### **6. Safety of SAT**

Information on the safety of prolonged antibiotic treatments can be obtained, not only from studies on SAT in PJI or other osteoarticular infections but also from other areas, such as antibiotic prophylaxis in immunosuppressed patients, the management of specific infections that require very long treatments (multidrug-resistant tuberculosis, actinomycosis, mycobacteriosis, Coxiella endocarditis, etc.) or entities in which infection and bacterial

colonization play a relevant role in the natural history of the disease (cystic fibrosis, acne, suppurative hidradenitis, etc.), for which long-term treatments have been tried.

In SAT series, adverse effects are not uncommon, but they rarely require discontinuation of treatment [19,21,22]. In addition, in many cases, poorly tolerated antibiotics can be substituted for another [10,17]. Data collection on adverse effects has not been systematized in any of the published studies and it was always retrospective. Gastrointestinal disturbances and skin reactions appear to be the most common reported adverse events. It should be borne in mind that in most series, ID consultants with extensive experience in the management of antimicrobials are those who prescribe and monitor treatments. Surprisingly, *C. difficile* infection is an infrequent event despite very long treatments that last many years [19,21].

In a preliminary study including several patients on SAT, colonization by multidrugresistant bacteria was not common. However, the patients who developed infections did so due to bacterial resistance to the antibiotic that they received for SAT [27].

#### **7. Reflections and Conclusions**

The information on SAT is fragmentary, heterogeneous and of low evidence. Despite this, the analysis of the available series suggests that SAT may represent an option with acceptable efficacy for selected cases in which potentially curative surgery cannot be performed or where the probabilities of success of the treatment are low. It is possible to administer antibiotics safely in the long term, provided that the clinician has the appropriate knowledge and experience. More studies are needed to answer the many questions that remain unanswered. To form useful conclusions in future investigations, it would be desirable to establish pragmatic criteria for efficacy, as well as to separate the cases in which SAT is indicated as an alternative to surgical treatment from those where it is indicated due to a high risk of failure of the surgical treatment used.

**Author Contributions:** J.C. and R.E.-S. review of the available information, draft of the manuscript preparation and approve the final version of the manuscript.

**Funding:** This research received no external funding.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data presented in this study are available on request from the corresponding author. The data are not publicly available due to the fact that it is a multicenter study and the complexity of the clinical history in some centers.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

