Outcomes of Concurrent Endocarditis and Periprosthetic Joint Infection: A Retrospective Case Series of 16 Patients

Introduction Concurrent diagnosis of periprosthetic joint infection (PJI) of total hip arthroplasty (THA) or total knee arthroplasty (TKA) with infectious endocarditis is a devastating clinical scenario infrequently documented in the literature. To date, no studies have fully described the orthopedic and infectious outcomes of patients with these concurrent diagnoses. The purpose of this study was to conduct a case series of patients with these diagnoses and document the orthopedic and infectious outcomes so that surgeons may effectively counsel patients regarding the gravity of the condition and the expected course of treatment. Methods This study is a retrospective case series using patient data from five hospitals within an academic healthcare system in the northeastern United States. Cases of concurrent endocarditis and THA or TKA PJI with a minimum of one-year follow-up were identified from January 2000 to January 2021. Basic statistics such as means, standard deviations, and percentages were used to identify trends within our series. Kaplan-Meier survivorship curves with log-rank tests were performed to determine if there were any differences in two-year mortality and joint survival (defined as needing explant) between patients who had cardiac surgery prior to surgical management for their PJI and those who had surgical management for PJI prior to cardiac surgery. Results A total of 18 joints in 16 patients with endocarditis and concurrent TKA or THA PJI were identified. All PJIs were managed surgically, with 14/18 (77.77%) of joint infections initially being managed by debridement, antibiotics, and implant retention (DAIR) and 4/18 (22.22%) of joint infections initially being managed by explant. Within the first six months of PJI diagnosis, 25% (4/16) of patients died of complications related to their infection, and one additional patient died of bacteremia just over a year after the initial PJI diagnosis. Of the 18 PJIs, 72.23% (13/18) had treatment failure, defined as any outcome equal to or worse than requiring chronic suppressive antibiotics for the infection. Due to low statistical power, we were not able to identify any differences in two-year mortality from PJI diagnosis (p=0.311) or joint survival (in terms of requiring explant) (p=0.420) depending on whether cardiac surgery or DAIR was performed first. Conclusions Concurrent infectious endocarditis and prosthetic joint infection is associated with high morbidity and mortality. Patients with these concurrent infections should be counseled that not only the associated mortality rate is high, but also the surgical treatment of their PJI has a high rate of treatment failure, including an explant following an initial DAIR, an explant with retained spacer, or a requirement of lifelong antibiotic suppression.


Introduction
Periprosthetic joint infection (PJI) of total hip arthroplasty (THA) or total knee arthroplasty (TKA) is a potentially devastating complication that occurs in approximately 1%-2% of patients undergoing the procedures [1]. The infectious source of PJI is commonly acquired at the time of surgery or as a result of hematogenous seeding from a distant source. Hematogenous infection has been noted to occur in anywhere from 20% to 30% of PJI cases [2,3]. Nevertheless, the exact source of hematogenous infection is not always identified, with some studies suggesting identification rates of around 68% [4]. One of the possible sources of hematogenous PJI, especially in the setting of multiple simultaneous PJI, is seeding from infectious endocarditis [5,6]. Concurrent PJI and infectious endocarditis represent a rare but significant clinical scenario, as the one-year mortality for total joint arthroplasty (TJA) PJI may approach 3%-8% and the sixmonth mortality for infectious endocarditis may be as high as 27% [7][8][9]. Furthermore, patients with these concurrent infections may require both surgery to manage their PJI (debridement, antibiotics, and implant retention (DAIR) or explant of the prosthesis) and cardiac surgery for a variety of indications, such as in the setting of acute congestive heart failure (CHF), embolic phenomena, hemodynamic instability, or prosthetic valve endocarditis especially due to Staphylococcus aureus [10,11].
Although prior studies have described that endocarditis may ultimately be found as the source of infection in TJA PJI, most reporting has been in the setting of case reports or as a single variable in retrospective cohort studies with less than 15 cases of endocarditis included [4,5,12]. Furthermore, a thorough characterization of the orthopedic and infectious outcomes of these patients has not been described. Thus, the goal of our study was to characterize the clinical presentation of patients with concurrent PJI and infectious endocarditis and describe the orthopedic and infectious outcomes of these patients in a case series of 16 patients. As patients with these contemporaneous diagnoses are likely to suffer high morbidity and mortality, an understanding of the trends of their clinical courses is needed to sufficiently counsel patients and their families.

Materials And Methods
This is an IRB-approved, retrospective case series using data from the electronic medical records of patients from five hospitals within an academic healthcare system in the northeastern United States. Our institutional database was queried to identify all patients over the age of 18 who had a primary or revision TKA or THA in place (using Current Procedural Terminology (CPT) codes 27447, 27487, 27486, 27130, 27137, 27134, and 27138) and who were then subsequently diagnosed with any of the three endocarditis subtypes native valve endocarditis, prosthetic valve endocarditis, or endocarditis due to a cardiac device (e.g., pacemaker, implantable cardioverter-defibrillator (ICD), and cardiac resynchronization therapy device) (International Classification of Diseases (ICD)-9 and ICD-10 codes I33, I38, I39, T82.6, T82.7, 421, 424, and 996.0) and were also diagnosed with PJI of the hip or knee (ICD-9 and ICD-10 codes 996.6, T84.5, T84.6, T84.7, T84.8, and T84.9). For our study period of January 2000 to January 2021, this query resulted in 330 possible patients. All patients were chart reviewed to confirm that the correct temporal sequence of TJA, endocarditis, and PJI (concurrent with the endocarditis episode) was present.
All patients were also confirmed to have a minimum of one-year clinical follow-up. We also chart reviewed patients to confirm that the three subtypes of endocarditis were diagnosed according to the modified Duke criteria [13] and to confirm that the diagnosis of PJI was made according to the most updated version of the Musculoskeletal Infection Society (MSIS) criteria [14,15]. For patient cases to meet our study's inclusion criteria, we confirmed that the microbiologic organism causing the endocarditis (by blood culture or tissue/device culture) and the PJI (by aspirate or intraoperative tissue culture) were the same organism, which would suggest that the distant infections were truly linked. Patients were excluded from the final cohort in our case series if patients did not meet the inclusion criteria or had insufficient chart information to confirm all of the required diagnoses and temporal constraints. Following the chart review, 18 TJAs in 16 patients were identified.
For each of the patients identified for our final cohort, a variety of demographic, surgical, infectious, and orthopedic variables were obtained. For each patient, data on age, sex, body mass index (BMI), joint (hip or knee), classification of TJA (primary or revision), and all comorbidities listed in the chart, including a history of septic arthritis, diabetes mellitus, hypertension, immunosuppressive medication prescription (systemic chemotherapy, immunosuppressants, or daily steroids), cancer diagnoses, cerebrovascular accident/stroke diagnoses, congestive heart failure, major depressive disorder, rheumatoid arthritis, smoking history, and alcohol use history, were obtained. For each patient, we also obtained data related to the hospital presentation of their concurrent PJI and endocarditis admission; thus, we documented the following: the number of joints infected, erythrocyte sedimentation rate (ESR) at presentation, C-reactive protein (CRP) at presentation, whether symptoms were acute (<4 weeks), if there were documented recent invasive procedures (such as dental procedures), PJI temporal classification related to the most recent TJA (early (<3 months), delayed (3-24 months), or late (>24 months)) [16], whether there was new-onset joint pain after an uneventful recovery from the index TJA, if patients were septic on initial presentation, if there were any signs of local inflammation (warmth, erythema, and edema) at the affected joint, fever of >38°C, and a description of the intraoperative findings by the orthopedic surgeon managing the PJI.
Regarding infectious and orthopedic outcomes, we documented the occurrence and dates of the diagnosis of endocarditis, diagnosis of PJI, and surgery for the management of PJI, including the following: DAIR, prosthesis explant with antibiotic spacer, full two-stage revision with reimplantation of prosthesis, resection arthroplasty (Girdlestone arthroplasty), cardiac surgery, and repeat cardiac surgery (if applicable). We documented the infectious organism of the endocarditis/PJI and whether the organism was a multidrugresistant organism (MDRO), which we defined as organisms with resistance to three or more antibiotic classes during susceptibility testing or the presence of methicillin-resistant Staphylococcus aureus (MRSA) [17]. We also noted the antibiotics prescribed after the hospitalization for the infections in terms of antibiotic type, duration, and route of administration and whether antibiotics were prescribed for the suppression of infection for >6 months duration. We lastly documented the ultimate outcome of the infections for each patient, as well as the dates of death during the study period. Treatment failure for PJI management was defined as a tier 2 to tier 4 outcome using outcome tiers described by Fillingham et al. [18]. The outcome tiers we utilized for our study can be viewed in Table 1.

PJI outcome tiers
Tier 1: Infection control with no continued antibiotic therapy

Statistical analysis
Simple descriptive statistics, such as means, averages, standard deviations, and percentages, were calculated for variables that we obtained in order to identify trends within our cohort. In addition, Kaplan-Meier survivorship curves with log-rank tests were performed to determine if there were any differences in twoyear mortality and joint survival (defined as needing explant) between patients who had cardiac surgery prior to surgical management for their PJI and patients who had surgical management for their PJI prior to cardiac surgery. Only patients who had cardiac surgery were included in this Kaplan-Meier survivorship curve analysis (n=13). The p-value for significance was set as p<0.05 for our study. Statistical analyses were performed using the SPSS software for Windows version 26 (IBM Corporation, Armonk, NY, USA).

Patient demographics
With our inclusion and exclusion criteria, we identified 16 patients with a history of primary or revision THA or TKA who developed concurrent native valve, prosthetic valve, or cardiac device endocarditis and PJI. Table 2 displays the general demographics of our case series. The average age of patients in our cohort at the time of PJI diagnosis was 69.30±9.36 years, with 10 male and six female patients represented. The average BMI in our cohort was 31.65±7.05 kg/m 2 , 56.25% of patients were smokers, 37.5% of patients were diabetic, 62.5% had peripheral artery disease, and 6.25% of patients had a history of septic arthritis. Interestingly, no patients were found to have ever abused injection drugs. Lastly, the average follow-up duration was 43.36 (range: 0.36-143.2) months in our cohort, which was inclusive of patients who died prior to the minimum 12-month follow-up for our study.   Table 3 depicts the characteristics of patient presentations in our cohort. In our cohort, there were 14 (87.5%) patients with PJI affecting a single joint and two (12.5%) patients presenting with multiple concurrent PJI. Of the PJIs, there were 12 knees and six hips. One patient was diagnosed with a knee PJI and concurrent native shoulder septic arthritis, although we excluded the native shoulder septic arthritis from our outcomes analysis. This resulted in a total of 18 total hip and total knee arthroplasties that were infected in 16 patients. In terms of inflammatory markers, the average ESR at PJI diagnosis was 64.00±36.35 mm/hour (reference: 0-13 mm/hour), and the average CRP at PJI diagnosis was 124.31±72.81 mg/L (reference: <8 mg/L).

Characteristics of patient presentations
In addition, 62.5% of patients had symptoms of PJI and endocarditis for less than four weeks prior to presentation to the hospital. We also found that 81.25% of patients experienced new-onset joint pain after an uneventful recovery from their index TJA and that 62.5% of patients were septic on presentation. In terms of temporal PJI classification, 87.5% of patients had a PJI temporally defined as "late," but only 18.75% of patients had a documented invasive procedure as a possible impetus for a hematogenous infection. In addition, we found that all patients in our cohort had signs of local inflammation at the affected joint upon presentation, and 93.75% of patients had a fever documented at presentation greater than 38°C. All patients had surgical management of their PJI, with 9/18 (50%) of joints demonstrating intraoperative purulence.
Regarding the timing of diagnoses, eight (50%) patients had their hip or knee PJI diagnosed prior to endocarditis, while the remaining 50% of patients were diagnosed with endocarditis prior to PJI. In the two patients with multiple simultaneous PJI diagnoses, the diagnoses of PJI were made on the same day of hospital admission.

Infectious and mortality outcomes
In our cohort of 16 patients, we identified seven cardiac device infections, eight native cardiac valve infections, and seven prosthetic/bioprosthetic valve infections, with some patients having multiple classifications of endocarditis simultaneously. In terms of microbiology, 10/18 (55.55%) of joints had methicillin-sensitive Staphylococcus aureus (MSSA) infection, and 3/18 (16.67%) of joints had infections due to MDROs, all of which were due to methicillin-resistant Staphylococcus aureus (MRSA). In addition, we found that 11/18 (61.11%) of joints in our cohort required oral antibiotic suppression therapy >6 months for indications that were determined through shared decision-making between the patient, infectious disease service, and orthopedic service. The rationale for prescribing suppressive antibiotics is documented in Table  4.
We also found that 13/16 (81.25%) of patients required cardiac surgery, but only one patient required repeat cardiac surgery in our study period, which was indicated for worsening heart failure in the setting of prosthetic valve dysfunction. The three patients who did not have cardiac surgery during the study period were determined to not meet the criteria for surgery according to mutual decisions made by the cardiothoracic surgery and infectious disease departments and were only treated with intravenous antibiotics. Lastly, 6/16 (37.5%) of patients died within two years of the endocarditis diagnosis, while 9/16 (56.25%) of patients died during the study period.

Orthopedic outcomes
In our cohort, we analyzed the outcomes of the 18 hip and knee PJIs that were diagnosed, as shown in Table  5. All PJIs were managed surgically, with 14/18 (77.77%) of joint infections initially being managed by DAIR and 4/18 (22.22%) of joint infections initially being managed by explant. Of the 14 joint infections managed with DAIR, 4/14 (28.57%) of joints subsequently required an explant, and 2/4 of these joints went on to reimplantation. Of the joints initially managed with explant, 3/4 (75%) went on to reimplantation. The three joints in our cohort that underwent explant of their prosthesis but did not complete the full two-stage revision were deemed either medically unfit for the reimplantation procedure or the patients did not have the desire to undergo further surgery. Only 5/18 (27.77%) of joints had a successful PJI treatment, while the remaining patients had varying tiers of failure. Notably, 7/18 (38.89%) of joints had treatment failure classified at or above tier 3E, which is representative of severe morbidity and mortality. The ultimate orthopedic outcome for each patient is documented in Table 5 and Figure 1.

Survivorship analyses
We sought to determine if there was a significant difference in mortality within two years of PJI diagnosis and joint survivorship (defined as requiring explant of the prosthesis) between patients who had cardiac surgery performed prior to surgery for PJI management and patients who had surgery for PJI management prior to cardiac surgery. In both of these survivorship analyses, we were able to include a maximum of 13 patients, as three patients did not undergo cardiac surgery and were instead treated with intravenous antibiotics. In our analysis of mortality within two years of PJI, the two patients with synchronous PJIs were counted as one patient. We chose to do this because these two patients had the DAIRs for their synchronous PJIs performed on the same day. In our analysis of joint survivorship (requiring explant), we excluded patients who had explant as the original surgical management for their PJI (thus only including patients with DAIR as the original management).
We found that mortality within two years appeared to be higher in patients who had surgery for PJI management prior to cardiac surgery, although the difference was not significant (log-rank p=0.311), as we were underpowered to detect a difference ( Figure 2). We also found that joint survivorship appeared to be higher in patients who had PJI management surgery prior to cardiac surgery, but the results were not significant (log-rank p=0.420), again due to underpowering ( Figure 3).

Discussion
This study highlights the high morbidity and mortality in patients with concurrent PJI and infectious endocarditis. These concurrent diagnoses, while rare, do occur and should be ruled out in patients with bacteremia and symptoms of a PJI. Rakow et al. studied 106 hematogenous PJIs in a two-year period and found that, while the primary source of infection was identified in only 68% of cases, of the cases with an identifiable source, 19.44% of cases were due to infectious endocarditis [4]. This study shed light on the potential incidence of concurrent PJI and endocarditis cases but did not discuss any of the orthopedic outcomes of these patients. Furthermore, in a study by Tande et al. of 166 patients with at least one arthroplasty in place who were diagnosed with Staphylococcus bacteremia, 36.1% of patients developed at least one PJI, and of these PJI patients, three developed endocarditis [5]. In that study, only one of the three patients had recurrent PJI, but the overall outcome was not discussed. Therefore, a detailed review of our cohort's infectious courses, treatment, and outcomes provides some insight into the seriousness of these concurrent diagnoses.
In our cohort, 87.5% of patients were diagnosed with a PJI classified as "late" (>24 months from index), which is consistent with the previous reporting of the incidence of patients who develop hematogenous PJI associated with another identified distant infection [12]. Within the first six months of PJI diagnosis, 25% (4/16) died of complications related to their infection, and one additional patient died of bacteremia just over a year after the initial PJI diagnosis. This patient who died 13 months after the PJI diagnosis was a male with untreated HIV and was on suppressive antibiotics. Of note, two out of the three patients who did not have cardiac surgery died from overwhelming infection, which is not surprising, as the primary focus of the infection (endocarditis) was not able to be fully addressed surgically. Overall, the mortality rate we found in our study is markedly higher than the reported one-year mortality rate after PJI alone of 3%-8% [7,9].
All 16 patients in our cohort underwent surgical intervention for the 18 PJIs. For the treatment of acute PJI with DAIR, the literature reports success rates of 83%-90% [19][20][21]. However, in our cohort, 2/14 (14.28%) patients undergoing DAIR had treatment success, and only 5/18 (27.77%) of all of the joints had successful PJI management. We found that, in total, 8/18 joints required explant at any point in time, and of these eight joints, three never went on to reimplant, at a rate of 37.5%. This rate of failure to undergo reimplant is higher than the reported attrition rates after the first stage of a two-stage revision, which are normally around 18% [22]. Determining the optimal surgical treatment should continue to be case-specific depending on factors including the chronicity of infection, presence of systemic infection, patients' medical comorbidities, and surgeon's preference. Nevertheless, patients should be counseled on the increased risks of treatment failure with either DAIR or two-stage revision, as 72.22% of all of the joints in our cohort had a failed treatment for PJI.
Regarding survivorship of patients as a function of whether cardiac surgery or PJI management surgery (DAIR of explant) was performed first, we were unable to identify significant differences. While this is due to insufficient power, it is interesting to note that in the three joints that had a cardiac surgery prior to DAIR, two required explant, and all had treatment failure. Previous research has shown that a shorter interval time between infectious symptoms and DAIR of the affected prosthetic joint is associated with improved joint survivorship (due to decreased biofilm formation) [23,24]. While performing cardiac surgery for endocarditis might delay DAIR for PJI, thus increasing the risk for treatment failure, addressing immediate lifethreatening conditions such as hemodynamic instability and acute heart failure likely takes precedent to debriding the joint. Given that treatment failure was also high in patients who had PJI surgery prior to cardiac surgery, we suspect that the high treatment failure in our cohort is representative of the overall gravity of these patients' medical conditions, regardless of whether one surgery took precedent during the hospitalization.
Despite only including cases in which the same organism was identified in both cardiac infection and PJI, for most patients in the cohort, we could not identify an underlying impetus for their infection. Two patients were noted to have had recent foot-related procedures that may have caused a hematogenous infection, and one had a recent pacemaker exchange that became infected. Notably, no patients had any documented recent dental procedures, for which antibiotic prophylaxis is recommended by many orthopedic and cardiac surgeons for both infectious endocarditis and prosthetic joint infection prevention in the setting of preexisting prosthetic implants [25,26]. Additionally, no patients had a documented history of injection drug use despite the high association with infectious endocarditis [27].
The most commonly identified organism in our cohort was Staphylococcus aureus, with MRSA in 16.67% of cases and MSSA in 55.55% of cases, for a total of 72.22% of organisms identified. This is consistent with previous literature describing that Staphylococcus aureus has some of the highest rates of seeding a periprosthetic joint in the setting of bacteremia compared to other organisms, at 18%-21% likelihood of seeding [28,29]. Nevertheless, in the literature, MSSA PJI has high rates of two-year infection-free survival, reported at 93% in one study [30]. In our cohort, 33% (3/9) of patients with MSSA infection died related to the infection, highlighting the high mortality risk with these concurrent infections of PJI and endocarditis.
Although the sample size of this study represents the largest known cohort of this rare concurrent pathology, the limitations of this study include the retrospective design and the small size of our cohort. A larger cohort could potentially allow us to compare and make conclusions regarding outcomes between the initial explant versus DAIR. We lacked sufficient cohort size to identify a statistically significant difference in survivorship between those who had their PJI surgical treatment before versus after the intervention of their cardiac infection. As a chart review-based study, we were limited to only information documented in their medical record; therefore, questions that may have suggested a possible source of infection such as dental procedure or injection drug use may not have been asked or documented.
Currently, there is no consensus diagnostic process for determining whether endocarditis or PJI was the initial infection in the setting of very short intervals between the diagnoses, which limits the interpretability of our study to the outcomes of patients who have "concurrent infections." Furthermore, while we included only patients who had a shared identified organism from both the cardiac infection and PJI to identify a relationship between the infections, we may have created a selection bias against difficult-to-culture organisms, which may explain the predominance of MSSA infections that were identified.

Conclusions
Concurrent infectious endocarditis and prosthetic joint infection is associated with high morbidity and mortality. Of the 18 PJIs, 72.23% resulted in treatment failure, defined as any outcome equal to or worse than requiring chronic suppressive antibiotics. Of the joints that underwent explant, three never went on to reimplant, at a rate of 37.5%. Patients with these two concurrent infections should be counseled that not only the associated mortality rate is high, but also the surgical treatment of their PJI has a high rate of treatment failure, including an explant following an initial DAIR, an explant with retained spacer, or a requirement of lifelong antibiotic suppression.