Invasive Candida Infection in Patients With Bacterial Infective Endocarditis

Abstract Over 21 months, 12 patients with invasive Candida infections detected during the course of treatment of bacterial endocarditis, including 11 with candidemia, were identified. Invasive Candida infections can occur as a complication of bacterial endocarditis and may occur more frequently in patients with injection drug use and broad-spectrum antibiotic exposure.

Infective endocarditis (IE) is a complex disease entity that carries significant morbidity and mortality.The vast majority of cases are caused by bacterial microorganisms.Approximately 1%-6% of IE is caused by fungal pathogens, and Candida species account for half of these cases with an associated mortality of >30% [1,2].Injection drug use, presence of a prosthetic valve or cardiac device, and chronic indwelling venous access are known risk factors for fungal IE [1].While prior endocarditis has been associated with increasing the long-term risk for subsequent episodes of Candida IE, the median duration between infections was 426 days in that study [3].In 1964 Henderson and Nickerson reported the only 2 cases of Candida albicans superinfection in patients with bacterial IE, although in 1 patient the Candida and bacterial organism were isolated simultaneously [4].Since then there has been little literature on the topic of Candida infections occurring in patients undergoing treatment for bacterial IE.Given the potential devastating effects of missed or delayed diagnosis of candidemia, it is crucial for providers to be aware of this phenomenon and to identify at-risk patients.To further understanding of this disease process, we report our experience with the largest case series of invasive Candida infections diagnosed in patients with bacterial IE.

METHODS
Institutional review board exemption was obtained from the University of Kentucky.Cases of candidemia in the first 12 weeks after diagnosis of bacterial IE were identified using the institution's multidisciplinary endocarditis team (MDET) database.All patients discussed by MDET who developed positive blood or tissue cultures with a Candida species after originally being diagnosed with bacterial endocarditis were included.Detailed retrospective chart review was then performed by all of the study investigators.Twenty-seven demographic and clinical variables were recorded, including patients' Pitt bacteremia score, Charlson Comorbidity Index (CCI) score, modified Duke criteria, and Duke-International Society of Cardiovascular Infectious Diseases (ISCVD) criteria (Table 1) [5][6][7].For continuous variables, descriptive statistics were utilized.

Definitions
Previous Candida infection was defined as any documented Candida infection within the electronic medical record of the University of Kentucky.Superficial infections, such as cutaneous Candida infections or vulvovaginal candidiasis, were excluded.Initial organisms were defined as the pathogens found in the first positive blood culture during the index admission.Duration of bacteremia was determined from the first positive to the first documented negative blood culture, including cultures from outside hospitals.History of injection drug use (IDU) was defined as any documented history of IDU within 30 days of admission.Timing of candidemia was determined by the number of days between the first positive bacterial blood culture and the first positive blood culture for Candida.Candidemia complications were defined as foci of infections that occurred after the onset of candidemia.Antibiotic treatment prior to candidemia was defined as any antibiotic given for the diagnosis of IE and prior to the onset of candidemia.Duration of therapy prior to candidemia was defined as the days of antibiotic therapy for IE prior to the first culture positive for Candida.Duration of candidemia was defined as the time between the first positive culture for Candida and the first documented negative blood culture.BRIEF REPORT • OFID • 3 total of 234 patients with definite endocarditis.The median age was 39.5 years (range, 23-59 years) and 7 patients were male (Table 1).Seven patients had a history of previous IE.In all 7, the previous IE was secondary to a bacterial organism and was managed medically.No patients had a prior documented history of Candida infection at any site or a previous valve replacement.

Between
The median Pitt bacteremia score was 2 (Range: 0-5) and the median CCI score was 0 (Range: 0-6).Eleven patients had a history of IDU within 30 days of the index admission and 7 patients had positive hepatitis C RNA polymerase chain reaction testing on admission.Two patients had central venous catheters at the time their Candida infection was diagnosed, and 1 patient had a peripherally inserted central catheter.One patient required initiation of dialysis during their hospitalization.No other patients had a history of requiring dialysis.All 12 patients met modified Duke and Duke-ISCVD criteria for definite IE.The median duration of bacteremia was 5 days (Range: 1-17 days).Eight patients had isolated bacteremia with Staphylococcus aureus.Five patients grew methicillin-susceptible S aureus and 3 patients grew methicillin-resistant S aureus.Three patients had polymicrobial bacteremia, and in all 3 instances S aureus was one of the pathogens.A single patient's cultures were positive for Enterococcus faecalis.Patients had a wide range of antibiotic exposures during their admission.Seven received broadspectrum gram-positive therapy with vancomycin.Nine patients received broad-spectrum gram-negative therapy with cefepime, ceftaroline, ceftriaxone, ertapenem, or piperacillintazobactam.Two patients were treated with either nafcillin or cefazolin alone.
In 11 patients, candidemia was identified at a median of 4 days (Range:1-70 days; Figure 1).One patient developed Candida osteomyelitis at the site of a cranioplasty with bone cultures growing Candida parapsilosis but did not have a corresponding positive blood culture.The median duration of candidemia was 2 days (Range: 1-8 days).Candida glabrata was isolated in 4 patients and C albicans and C parapsilosis were isolated in 4 patients each, including 1 patient who grew both C albicans and C parapsilosis.One patient's blood cultures were positive for Candida tropicalis.The duration of antifungal therapy in patients who survived to discharge ranged from 14 days to indefinite therapy, which was recommended in 5 cases.One patient received only 6 days of antifungal treatment but passed away during the index hospitalization while on treatment.Eleven patients received micafungin at some point in their course.Three patients received liposomal amphotericin, 1 in conjunction with flucytosine.Nine patients were ultimately transitioned to oral therapy, 7 with fluconazole, 1 with voriconazole, and 1 with isavuconazole.Two patients underwent surgical valve replacement and 1 patient underwent open pulmonary embolectomy for a saddle pulmonary embolism from a large tricuspid valve vegetation.In 2 of these cases, surgical cultures (1 from a valve specimen and 1 from the pulmonary embolus) also grew the concordant Candida species.
Eleven patients survived to hospital discharge and 10 survived to 90 days postdischarge.The median time to last documented follow-up after the diagnosis of Candida infection for the 11 patients who survived to discharge was 532 days (Range: 65-830 days) and there were no documented relapses with Candida.Two additional patients died at 162 days and 555 days.One passed from a new episode of endocarditis with a bacterial organism and the other died from an opioid overdose.

DISCUSSION
We report a case series of 12 patients with definite bacterial IE who developed invasive Candida infection manifested principally by candidemia.While Candida is a well-described cause of IE, most previous studies have described its role as the primary etiologic pathogen rather than an organism isolated during the course of treatment of bacterial IE.Although this is a retrospective case series, it highlights a concerning phenomenon with significant treatment implications that is affecting patients with bacterial IE.Notably, this outcome occurred in 5.1% of all patients with definite IE treated at our institution during the study period.
There were a number of similarities between the 12 patients in this series.Eleven reported IDU and 7 had prior bacterial IE.No patients had previously documented infections with Candida.In all but 1 case, S aureus was either the primary or, in the case of polymicrobial bacteremia, 1 of the primary pathogens.All 12 cases involved native valves, including 10 that affected the tricuspid valve.Additionally, 10 patients received broad-spectrum gram-positive or gram-negative antibiotic therapy in the days prior to detection of their Candida infection.Although there was a wide range with respect to the timing of candidemia, 8 patients had positive blood cultures for Candida within 7 days of starting antibiotic therapy.
Broad-spectrum antibiotic exposure and IDU are both known risk factors for invasive candidiasis [1,8].Several of the patients were also admitted to the intensive care unit and had indwelling central venous access, which increases the risk of candidemia [2].However, 5 patients developed candidemia within 48 hours of admission.Recent literature has suggested that S aureus is able to invade host tissue and disseminate via adherence to hyphal elements of C albicans, suggesting that Candida could facilitate the invasion of S aureus in cocolonized patients [9].It is possible that patients were colonized with both organisms and detection of Candida was delayed due to decreased sensitivity of blood cultures for detection of invasive candidiasis [10].
It remains unclear whether these episodes of candidemia led to superinfection of the valvular vegetations.Only 3 patients underwent surgical procedures with pathologic evaluation of valve or embolism specimens.Two of the 3 patients had positive operative cultures for Candida and these patients were candidemic for 6 and 8 days.In contrast, the patient with negative valve cultures and pathology only had 1 day of candidemia.Ultimately, 7 patients were believed to have Candida IE by the MDET, prompting recommendations for longer durations of treatment.
In our clinical practice, we do not routinely perform valve surgery during the index hospitalization for patients with tricuspid valve IE given the relatively low in-hospital mortality for this condition.Our approach advocates for treating the patients' substance use disorder and having them return as an outpatient for evaluation for valve surgery when hemodynamic indications arise.Seven patients had isolated native tricuspid valve endocarditis and were managed accordingly.Of the 5 patients with left-sided endocarditis, 2 underwent valve surgery.In the other 3, 2 did not have surgical indications and 1 was deemed not a surgical candidate due to their neurologic condition (unresponsive wakefulness syndrome).
Another feature of this study was the relatively successful treatment with step-down oral therapy.Nine patients were transitioned to an oral triazole, including 7 who received fluconazole.These results are consistent with other previously published findings supporting the use of oral fluconazole in treatment of Candida IE [11].Eleven patients survived to hospital discharge (8.3% mortality), which is substantially lower than the commonly reported 25% mortality associated with candidemia or 30% mortality seen with Candida IE [2,12].One additional patient passed away within 90 days postdischarge for an overall mortality of 16.7%.

CONCLUSIONS
While additional data and prospective studies would add further validity to these observations, some tentative inferences can be drawn from this series.Patients undergoing treatment for bacterial IE, particularly individuals with IDU and S aureus bacteremia, may be at risk for candidemia, which can progress to superinfection of valvular vegetations.Clinical changes in patients, particularly those who inject drugs, undergoing bacterial IE treatment should prompt collection of blood cultures to evaluate for candidemia.A subset of patients may present asymptomatically, but the optimal surveillance strategy for detecting this complication remains unclear.Providers could consider repeating blood cultures 4-7 days after bacteremia clearance to assess for previously undetected candidemia.Finally, patients with candidemia, including Candida IE, can be successfully transitioned to oral azoles to complete therapy.Further research is needed to better understand the pathogenesis and optimal management of this phenomenon.

Notes
Patient consent.The design of this study has been approved by the University of Kentucky Institutional Review Board and per their approval, patient consent was not required.
Potential conflicts of interest.The authors: No reported conflicts of interest.

Figure 1 .
Figure 1.Scatter plot outlining the time to detection of the invasive Candida infection (blue), time to surgical intervention (yellow), and time to in-hospital death (red) for all 12 patients.