Find synonyms Find exact match
INTRODUCTION — Cancer patients receiving cytotoxic antineoplastic therapy sufficient to adversely affect myelopoiesis and the integrity of the gastrointestinal mucosa are at risk for invasive infection due to colonizing bacteria or fungi that translocate across intestinal mucosal surfaces. Patients with profound prolonged neutropenia are at particularly high risk for serious infections; profound prolonged neutropenia is most likely to occur in the preengraftment phase of hematopoietic cell transplantation (HCT, particularly allogeneic) and in patients undergoing induction chemotherapy for acute leukemia.
Because neutropenic patients are unable to mount robust inflammatory responses, serious infection can occur with minimal symptoms and signs. In such patients, fever is often the only sign of infection. Infections in neutropenic patients can progress rapidly, leading to hypotension and/or other life-threatening complications. It is critical to recognize neutropenic fever early and to initiate empiric systemic antibacterial therapy promptly in order to avoid progression to a sepsis syndrome and possibly death [1,2].
The use of empiric antibacterial and antifungal therapy for high-risk neutropenic adults presenting with fever will be reviewed here. The management of neutropenic fever in cancer patients at low risk for complications is discussed separately. An overview of neutropenic fever syndromes, the risk assessment of patients with neutropenic fever, and the diagnostic approach to patients presenting with neutropenic fever are presented elsewhere. The use of antimicrobial prophylaxis and colony stimulating factors to prevent infections in HCT recipients and patients with chemotherapy-induced neutropenia is also discussed elsewhere. (See "Treatment and prevention of neutropenic fever syndromes in adult cancer patients at low risk for complications" and "Overview of neutropenic fever syndromes" and "Risk assessment of adults with chemotherapy-induced neutropenia" and "Diagnostic approach to the adult presenting with neutropenic fever" and "Prophylaxis of infection during chemotherapy-induced neutropenia in high-risk adults" and "Prophylaxis of invasive fungal infections in adults with hematologic malignancies" and "Prophylaxis of invasive fungal infections in adult hematopoietic cell transplant recipients" and "Prophylaxis of infections in hematopoietic cell transplant recipients" and "Use of granulocyte colony stimulating factors in adult patients with chemotherapy-induced neutropenia and conditions other than acute leukemia, myelodysplastic syndrome, and hematopoietic cell transplantation".)
Neutropenic fever in children is also presented separately. (See "Fever in children with chemotherapy-induced neutropenia".)
GUIDELINES — Guidelines have been developed for the management of fever in neutropenic patients with cancer, including hematopoietic cell transplant (HCT) recipients [1-5]. The recommendations below are generally in keeping with the 2010 Infectious Diseases Society of America (IDSA) guidelines .
Fever — Fever in neutropenic patients is defined as a single oral temperature of >38.3°C (101°F) or a temperature of >38.0°C (100.4°F) sustained for >1 hour . The definition of fever and appropriate methods for measuring body temperature are discussed in greater detail separately. (See "Overview of neutropenic fever syndromes", section on 'Fever' and "Overview of neutropenic fever syndromes", section on 'Temperature measurement'.)
Neutropenia — The definition of neutropenia varies from institution to institution but neutropenia is usually defined as an absolute neutrophil count (ANC) <1500 cells/microL, and severe neutropenia is usually defined as an ANC <500 cells/microL, or an ANC that is expected to decrease to <500 cells/microL over the next 48 hours [1,6]. The risk of clinically important infection rises as the neutrophil count falls below 500 cells/microL and is higher in those with a prolonged duration of neutropenia (>7 days). For the purposes of this discussion, we will define neutropenia as an ANC <500 cells/microL.
The ANC can be calculated by multiplying the total white blood cell (WBC) count by the percentage of polymorphonuclear cells (PMNs) and bands (calculator 1).
(See "Overview of neutropenic fever syndromes", section on 'Neutropenia' and "Overview of neutropenia in children and adolescents", section on 'Definitions' and "Approach to the adult with unexplained neutropenia", section on 'Terminology and definition of neutropenia'.)
RISK OF SERIOUS COMPLICATIONS — The initial clinical evaluation focuses on assessing the risk of serious complications. This risk assessment dictates the approach to therapy, including the need for inpatient admission, IV antibiotics, and prolonged hospitalization (table 1).
●Low-risk patients are defined as those who are expected to be neutropenic (absolute neutrophil count [ANC] <500 cells/microL) for ≤7 days and those who have no active comorbidities or evidence of significant hepatic or renal dysfunction. This group of patients has been well studied in randomized trials and has been shown to be at low risk for serious complications . Most patients receiving chemotherapy for solid tumors are considered to be low-risk for complications requiring hospitalization or prolonging hospitalization.
●We define high-risk patients as those who are expected to be neutropenic (ANC <500 cells/microL) for >7 days. Patients with neutropenic fever who have ongoing comorbidities or evidence of significant hepatic or renal dysfunction are also considered to be high-risk, regardless of the duration of neutropenia. Other criteria that confer a high-risk status can be found in the Table (table 1).
Some experts have defined high-risk patients as those expected to have profound neutropenia (ANC ≤100 cells/microL) for >7 days based on experience that such patients are the most likely to have life-threatening complications [1,2]. However, formal studies to clearly differentiate between patients with an ANC <500 cells/microL and ≤100 cells/microL are lacking. For the purposes of this discourse, we will combine these groups. Profound prolonged neutropenia (ie, ANC ≤100 cells/microL expected to last >7 days) is most likely to occur in the preengraftment phase of myeloablative hematopoietic cell transplantation (particularly allogeneic) and in patients undergoing induction chemotherapy for acute leukemia.
An alternative to using the clinical criteria described above is to use the Multinational Association for Supportive Care in Cancer (MASCC) risk index, which is a validated tool for measuring the risk for neutropenic fever-related medical complications (calculator 2) .
The risk assessment of patients with neutropenic fever is discussed in greater detail separately. (See "Risk assessment of adults with chemotherapy-induced neutropenia".)
INITIAL ASSESSMENT — In patients presenting with neutropenic fever, a reliable method for obtaining body temperature must be used and a mechanism for estimating the absolute neutrophil count (ANC) is mandatory. The temperature should be taken using oral or tympanic thermometry. (See "Overview of neutropenic fever syndromes", section on 'Temperature measurement'.).
Risk of neutropenia — Patients and their families should be instructed by their hematologist or oncologist to inform healthcare providers in the triage setting about recent chemotherapy, and providers in the triage setting should ask cancer patients who do not offer this information about recent chemotherapy. The ANC can be estimated based upon the timing of the febrile episode following the first dose of the current cytotoxic chemotherapy or from a laboratory-based measurement of the ANC from a complete blood count (calculator 1). Knowledge that the ANC reaches its nadir of <500 cells/microL at a median of 12 to 14 days from day one of chemotherapy can guide the clinician to the correct index of suspicion regarding the likelihood of neutropenia.
Over 70 percent of cancer recipients who develop systemic therapy-related complications present to emergency triage facilities within four to six weeks of systemic anti-cancer treatment . The use of a sensitive but non-specific historical indicator, receipt of systemic anti-cancer therapy within the preceding six weeks, has been advocated for use in emergency triage departments to identify patients who are likely to be neutropenic .
Risk of complications — As discussed above, it is crucial to assess the risk of serious complications in patients with neutropenic fever, since this assessment will dictate the approach to therapy, including the need for inpatient admission, IV antibiotics, and prolonged hospitalization. (See 'Risk of serious complications' above.)
Identification of sepsis — Although the signs and symptoms of infection may be significantly muted in neutropenic patients [10,11], an early part of the clinical assessment at triage should include an examination for evidence of a systemic inflammatory response syndrome (SIRS), and a determination of the criteria for sepsis, severe sepsis, or septic shock . (See "Sepsis and the systemic inflammatory response syndrome: Definitions, epidemiology, and prognosis", section on 'Definitions'.)
Patients presenting with evidence of new organ dysfunction (altered mental status, hypotension, or hypoxia) should be managed emergently for severe sepsis. Patients with septic shock should be managed in a critical care hospital environment. (See "Evaluation and management of severe sepsis and septic shock in adults".)
Diagnostic evaluation — The diagnostic evaluation of patients presenting with neutropenic fever is summarized in the following Table (table 2), and is discussed in detail separately. (See "Diagnostic approach to the adult presenting with neutropenic fever".)
General principles — Fever in a neutropenic patient should be considered a medical emergency. Broad-spectrum antibacterials should be given as soon as possible (within 60 minutes of triage) and at full doses, adjusted for renal and/or hepatic function. In addition, the diagnostic evaluation should be obtained quickly. (See "Diagnostic approach to the adult presenting with neutropenic fever".)
The aim of empiric therapy is to cover the most likely and most virulent pathogens that may rapidly cause serious or life-threatening infection in neutropenic patients (table 3) . The following general principles apply:
●Antibiotics are usually administered empirically, but should always include appropriate coverage for suspected or known infections. Even when the pathogen is known, the antibiotic regimen should provide broad-spectrum empiric coverage for the possibility of other pathogens, unlike the treatment strategy adopted in many immunocompetent hosts.
●In high-risk patients, antibiotics should generally be administered intravenously in a hospital setting.
●Initial antibiotic selection should be guided by the patient's history, allergies, symptoms, signs, recent antibiotic use and culture data, and awareness of the susceptibility patterns of institutional nosocomial pathogens .
●Ideally, antibiotics should be bactericidal.
●Clinical response and culture and susceptibility results should be monitored closely, and therapy should be adjusted in a timely fashion in response to this information .
Febrile neutropenic patients should be monitored frequently with respect to vital signs (blood pressure, heart rate, respiratory rate, and temperature), performance status (the clinical burden of the neutropenic fever syndrome), and the ability to achieve adequate oral intake in the presence of oral or gastrointestinal mucositis. Temporarily holding administration of systemic chemotherapy should be considered during the management of the sepsis syndrome until the patient stabilizes. Attention to fluid and electrolyte management is important given the dehydrating effects of fever, vomiting, and/or diarrhea. Urine output of >0.5 mL/kg per hour should be maintained.
Afebrile neutropenic patients with new signs or symptoms that are consistent with infection should be evaluated and managed as if they are febrile .
Timing of antibiotics — In all febrile neutropenic patients, empiric broad-spectrum antibacterial therapy should be initiated immediately after blood cultures have been obtained and before any other investigations have been completed [3,9]. Antimicrobial therapy should be administered within 60 minutes of presentation [9,14,15]. This is discussed in greater detail separately. (See "Overview of neutropenic fever syndromes", section on 'Timing of antibiotics'.)
Spectrum of coverage — Although gram-positive bacteria are the most frequent pathogens identified during neutropenic fever episodes, it is important to cover broadly for gram-negative pathogens because of their virulence and association with sepsis [15,16]. Furthermore, gram-negative organisms continue to cause the majority of infections in sites outside of the bloodstream (eg, respiratory tract, biliary tract, gastrointestinal tract, urinary tract, and skin) , and a rising number of infections are polymicrobial [15,16]. Clinicians need to be aware of the current microbiology surveillance data from their own institution, which can vary widely from center to center and over time [12,15].
Although anaerobic bacteria are present in abundance in the gastrointestinal tract, it is usually not necessary to include specific anaerobic antibiotic coverage in the initial empiric regimen. Anaerobic bacteremia occurred in only 3.4 percent of episodes in a large series of cancer patients from France , and anaerobic bacteria are often part of polymicrobial infections [19,20]. Anaerobic coverage should be included in the regimens of patients with infections that are known or expected to be caused by anaerobes, as discussed below. (See 'Initial regimen' below.)
As the duration of profound neutropenia increases, the risk of invasive fungal infections (eg, candidiasis, invasive aspergillosis) becomes substantial in patients with neutropenic fever. (See "Overview of neutropenic fever syndromes", section on 'Fungal pathogens' and 'Addition of an antifungal agent' below.)
The epidemiology of infections in febrile neutropenic patients is discussed in greater detail separately. (See "Overview of neutropenic fever syndromes", section on 'Epidemiology'.)
Initial regimen — The choice of antibiotics is driven by multiple factors, including the degree of immunocompromise, prior antibiotic and infection history, and local patterns of antibiotic resistance, and whether an agent is bactericidal or not [1,15]. Some drugs, such as beta-lactams, exhibit time-dependent killing. When using beta-lactams, correct dosing intervals should be used to ensure that drug concentrations are greater than the minimal inhibitory concentration for the pathogen. Other antibiotics, such as aminoglycosides and fluoroquinolones, exhibit concentration-dependent killing and are important in the treatment of gram-negative sepsis.
The Infectious Diseases Society of America (IDSA) recommends the following approach for the initial therapy of high-risk neutropenic patients with fever (algorithm 1) ; we agree with this approach:
●Initiation of monotherapy with an anti-pseudomonal beta-lactam agent, such as cefepime, meropenem, imipenem-cilastatin, or piperacillin-tazobactam. Ceftazidime monotherapy has also been shown to be effective, and continues to be used at some cancer centers. However, many experts avoid ceftazidime monotherapy because of rising resistance rates among gram-negative bacteria and its limited activity against gram-positive bacteria, such as streptococci, compared to newer alternatives. The dosing of these agents for patients with normal renal function are:
•Cefepime – 2 g IV every eight hours
•Meropenem – 1 g IV every eight hours
•Imipenem-cilastatin – 500 mg IV every six hours
•Piperacillin-tazobactam – 4.5 g IV every six hours
•Ceftazidime – 2 g IV every eight hours
●Other antibiotics (eg, aminoglycosides, fluoroquinolones, and/or vancomycin) may be added to the initial regimen in patients with complicated presentations (eg, hypotension and/or mental status changes), focal findings (eg, pneumonia or cellulitis) or if antimicrobial resistance is suspected or proven.
●Vancomycin (or other agents that target gram-positive cocci) is not recommended as a standard part of the initial regimen but should be added in certain patients, such as those with suspected catheter-related infection, skin or soft tissue infection, pneumonia, or hemodynamic instability. (See 'Addition of gram-positive coverage' below.)
●Modifications to the initial regimen should be considered for patients at risk for infection with antibiotic-resistant organisms, patients who are clinically unstable, and patients with positive blood cultures that are suggestive of a resistant infection. Risk factors for infections caused by resistant bacteria include previous infection or colonization by the organism and/or treatment in a hospital with high rates of resistance. (See 'Modifications to the regimen' below and 'Antibiotic resistance' below.)
In addition, we suggest that anaerobic coverage be included if there is evidence of necrotizing mucositis, sinusitis, periodontal cellulitis, perirectal cellulitis, intraabdominal infection (including neutropenic enterocolitis [typhlitis]), pelvic infection, or anaerobic bacteremia. (See "Neutropenic enterocolitis (typhlitis)" and "Enterotoxicity of chemotherapeutic agents".)
Comparisons of regimens
Monotherapy — Monotherapy with a beta-lactam agent with activity against Pseudomonas aeruginosa (eg, cefepime, meropenem, imipenem-cilastatin, piperacillin-tazobactam, or ceftazidime) is frequently employed; clinical trials with ceftazidime, imipenem-cilastatin, or meropenem have demonstrated equivalent outcomes compared to two-drug regimens [21-23]. In addition, fewer adverse events have generally been seen with monotherapy regimens compared with combination regimens . The majority of the regimens evaluated provided coverage targeted at gram-negative bacilli, especially P. aeruginosa.
Single drugs have also been compared with each other in various clinical trials of empiric therapy for patients with neutropenic fever, as illustrated below:
●In a multicenter randomized trial, treatment with meropenem was compared with ceftazidime in 411 patients and the outcomes were similar for clinically defined and microbiologically defined infections .
●Another trial assessed cefepime versus imipenem in 251 patients . In an intent-to-treat analysis, the response to treatment in the two groups was comparable (75 versus 68 percent, respectively) as was the side effect profile.
●A multicenter open-label randomized trial compared the efficacy and safety of piperacillin-tazobactam and cefepime in 528 patients . On multivariate analysis, treatment with piperacillin-tazobactam was associated with greater treatment success (odds ratio, 1.65, 95% CI 1.04-2.64).
●A meta-analysis of clinical trials showed increased mortality among patients who received cefepime compared with other beta-lactam antibiotics, particularly in those with neutropenic fever; both groups included patients who received either monotherapy or combination therapy of a beta-lactam plus a non-beta-lactam antibiotic . However, in a subsequent meta-analysis performed by the United States Food and Drug Administration (FDA), which included 50 more trials than the original meta-analysis as well as patient-specific information, there was no difference in mortality between patients who received cefepime and patients who received other antibiotics [29,30]. Accordingly, the IDSA guidelines continue to recommend cefepime as a suitable option .
One concern about monotherapy is the possibility that increasing rates of antibiotic resistance in a number of pathogens may reduce the efficacy of this strategy. Single agents, especially ceftazidime, may actually promote the outgrowth of resistant organisms in this group of patients who require frequent antibiotic administration . It is therefore important to maintain vigilance for the emergence of antibiotic resistance locally that may necessitate a change in antibiotic practices. (See 'Antibiotic resistance' below.)
Combination therapy — Numerous combination antibiotic regimens have been studied as initial empiric therapy in neutropenic fever, but none has been shown to be clearly superior to others or to monotherapy [32,33]. One approach is to use an extended-spectrum beta-lactam (eg, piperacillin, ceftazidime) in combination with an aminoglycoside. Other examples of combination regimens include double beta-lactams or a beta-lactam and a fluoroquinolone. A meta-analysis of eight randomized controlled trials that compared ciprofloxacin/beta-lactam combinations to aminoglycoside/beta-lactam regimens for the empiric therapy of neutropenic fever demonstrated similar overall efficacy for clinical cure and all-cause mortality . Double beta-lactams are usually avoided because of overlapping toxicities.
Penicillin-allergic patients — Many patients with a history of allergy to penicillin tolerate cephalosporins  (see "Penicillin-allergic patients: Use of cephalosporins, carbapenems, and monobactams"). However, those with a history of an immediate-type hypersensitivity reaction (eg, hives and/or bronchospasm) should not receive beta-lactams or carbapenems. Alternative empiric regimens for such patients include ciprofloxacin plus clindamycin or aztreonam plus vancomycin . In general, fluoroquinolones should not be used in patients who have received them recently (including those receiving a fluoroquinolone for neutropenic prophylaxis). The decision of which alternative regimen to use in penicillin-allergic patients should be made based upon the susceptibility patterns of bacteria (especially gram-negative bacilli) at the individual institution as well as the patient’s previous microbiologic data.
Addition of gram-positive coverage — Routine addition of gram-positive antibiotic coverage to the initial empiric antibiotic regimen has not been associated with significant clinical benefit [34-36]. A meta-analysis of seven randomized trials found that addition of gram-positive antibiotic coverage to standard empiric therapy did not reduce all-cause mortality in patients with cancer and neutropenic fever .
Even in febrile neutropenic patients with skin and soft tissue infections who had a higher incidence of proven gram-positive bacteremia compared with patients with other infections (31 versus 17 percent), the addition of empiric vancomycin did not improve outcomes and was associated with increased toxicity . The risk of promoting resistance among enterococci and Staphylococcus aureus is an important reason to avoid empiric vancomycin use .
Vancomycin (or other agents that target gram-positive cocci) is not recommended as a standard part of the initial regimen, but gram-positive coverage should be added in patients with any of the following findings :
●Hemodynamic instability or other signs of severe sepsis
●Positive blood cultures for gram-positive bacteria while awaiting speciation and susceptibility results
●Suspected central venous catheter (CVC)-related infection (eg, chills or rigors during infusion through a CVC and/or cellulitis around the catheter entry site)
●Skin or soft tissue infection
●Severe mucositis in patients who were receiving prophylaxis with a fluoroquinolone lacking activity against streptococci and in whom ceftazidime is being used as empiric therapy. Addition of gram-positive coverage is recommended in this situation because of the increased risk of Streptococcus viridans infections, which can result in sepsis and the acute respiratory distress syndrome [1,38-41].
Empiric gram-positive coverage is particularly important for patients who are colonized with methicillin-resistant S. aureus (MRSA), vancomycin-resistant enterococcus, or penicillin- or ceftriaxone-resistant streptococci who become hemodynamically unstable or develop bacteremia with gram-positive cocci.
Vancomycin is used most commonly when an agent with specific gram-positive activity is indicated. Linezolid is an alternative for patients intolerant of vancomycin. However, a concern with linezolid is that it may cause myelosuppression, typically after two or more weeks of therapy. Whether myelosuppressive effects can be seen earlier in patients receiving chemotherapy remains unknown [42,43].
A multicenter randomized trial of 611 febrile neutropenic patients compared the safety and efficacy of linezolid and vancomycin . Patients with proven or suspected infection due to a gram-positive pathogen were randomly assigned to receive linezolid (600 mg IV every 12 hours) or vancomycin (1 g IV every 12 hours) for 10 to 28 days. The following findings were noted:
●In patients with documented gram-positive infections, those who received linezolid defervesced more quickly than those who received vancomycin (5.9 versus 9.1 days); there was no difference in time to defervescence among patients without documented infections.
Daptomycin is another alternative to vancomycin, but it has been less well studied and should not be used for pulmonary infections because it is inactivated by surfactant and therefore does not achieve sufficiently high concentrations in the respiratory tract. (See "Treatment of hospital-acquired, ventilator-associated, and healthcare-associated pneumonia in adults", section on 'Other agents'.)
Modifications to the regimen — Modifications to the antimicrobial regimen during the course of neutropenic fever can be made based upon the following principles :
●The initial treatment regimen should be modified based upon clinical and microbiologic data.
●Unexplained persistent fever in a patient who is otherwise stable rarely necessitates an empiric adjustment to the initial antibacterial regimen. However, if an infection is identified, the regimen should be adjusted accordingly.
●Documented infections (based on clinical findings and/or microbiologic data) should be treated with antibiotics that are appropriate for the site and susceptibility patterns of organisms that are isolated.
●If vancomycin or other gram-positive coverage was started initially, it may be stopped after two to three days if there is no evidence of a gram-positive infection. Vancomycin overuse has been associated with the development of resistance (eg, vancomycin-resistant Enterococcus spp).
●Patients who are or become hemodynamically unstable after initial doses of a standard antimicrobial regimen for neutropenic fever should have their regimen broadened to include coverage for resistant gram-negative, gram-positive, and anaerobic bacteria, as well as fungi. (See 'Antibiotic resistance' below and 'Addition of an antifungal agent' below.)
●Empiric antifungal coverage should be considered in high-risk patients who have persistent fever after four to seven days of a broad-spectrum antibacterial regimen and no identified source of fever. (See 'Addition of an antifungal agent' below.)
●Oral ulcerations may be due to herpes simplex virus or Candida spp. Thus, addition of acyclovir and/or fluconazole may be warranted if these findings are present. (See "Treatment of oropharyngeal and esophageal candidiasis".)
●In patients with diarrhea, if there are abdominal signs, empiric therapy of Clostridium difficile can be instituted while assays are pending. (See "Clostridium difficile in adults: Treatment".)
Persistent fever — The median time to defervescence following the initiation of empiric antibiotics in patients with hematologic malignancies, including hematopoietic cell transplant (HCT) recipients, is five days, in contrast to only two days for patients with solid tumors . As stated above, modification of the initial antibacterial regimen is not needed for persistent fever alone. However, patients who remain febrile after the initiation of empiric antibiotics should be reevaluated for possible infectious sources. (See "Diagnostic approach to the adult presenting with neutropenic fever".)
Management algorithms have been developed for the reassessment of neutropenic patients with persistent fever after two to four days (algorithm 2), and after four or more days (algorithm 3) . Key factors in the management of patients with persistent fever include whether the patient is clinically stable, whether there is an identified site of infection, and when the patient is expected to recover from neutropenia. Consideration should be given to the addition of empiric antifungal therapy, as described below. (See 'Addition of an antifungal agent' below.)
Antibiotic resistance — The increasing frequency of antibiotic-resistant organisms is a major concern [45,46]. The patient’s risk for the following resistant organisms should be considered when choosing an empiric regimen:
●Among gram-positive organisms, pathogens with acquired resistance include coagulase-negative staphylococci, MRSA, vancomycin-resistant enterococci (VRE), and penicillin- and ceftriaxone-resistant Streptococcus pneumoniae and other streptococci.
●Multidrug-resistant gram-negative bacilli, such as P. aeruginosa, Escherichia coli, and Citrobacter, Acinetobacter and Stenotrophomonas spp . The use of fluoroquinolones for prophylaxis has contributed to the emergence of antibiotic resistance.
●The presence of extended-spectrum beta-lactamases (ESBL), plasmid-mediated AmpC-type beta-lactamases, and carbapenemase-producing bacteria (eg, Klebsiella pneumoniae carbapenemase [KPC]) can severely limit treatment options [15,47].
Risk factors for infections with resistant bacteria include previous infection or colonization by the organism, recent use of antibacterial agents (eg, for prophylaxis), and treatment in a hospital with high rates of resistance . It is important to be aware of institutional resistance patterns since a variety of nosocomial outbreaks involving cancer patients have been reported. Some centers have reported an increased incidence of resistant pathogens with the routine use of prophylactic antibiotics [48-50]. Antibiotic history, recent culture results, exposure to prophylactic antibiotics, and the susceptibility patterns for organisms in the institution should be used to help guide selection of initial antibiotic therapy.
Strategies to reduce drug resistance include limiting prophylaxis, using targeted therapy when feasible, discontinuing unneeded empiric therapies (eg, vancomycin) when cultures remain negative, and initiation of antibiotic stewardship programs . (See 'Addition of gram-positive coverage' above.)
The following antibiotics can be used when resistant infections are suspected :
●MRSA – Vancomycin, linezolid, or daptomycin; daptomycin should be avoided in patients with pneumonia because it does not achieve sufficiently high concentrations in the respiratory tract. (See "Treatment of hospital-acquired, ventilator-associated, and healthcare-associated pneumonia in adults", section on 'Other agents' and "Treatment of invasive methicillin-resistant Staphylococcus aureus infections in adults" and "Treatment of skin and soft tissue infections due to methicillin-resistant Staphylococcus aureus in adults".)
●ESBL-producing gram-negative bacilli – A carbapenem (eg, imipenem, meropenem). (See "Extended-spectrum beta-lactamases", section on 'Treatment options' and "Gram-negative bacillary bacteremia in adults", section on 'Extended-spectrum beta-lactamases'.)
●Carbapenemase-producing bacteria, including Klebsiella pneumoniae carbapenemase (KPC) – Colistin or tigecycline. (See "Carbapenemases", section on 'Treatment' and "Gram-negative bacillary bacteremia in adults", section on 'Carbapenem resistance'.)
Addition of an antifungal agent
Indications — An empiric antifungal agent should be added after four to seven days in high-risk neutropenic patients who are expected to have a total duration of neutropenia >7 days who have persistent or recurrent fever and in whom reassessment does not yield a cause (algorithm 3) . The rationale for this approach is that undiagnosed fungal infection was found in early studies in many patients who died during prolonged neutropenia [51,52]. The incidence of fungal infection (especially those caused by Candida or Aspergillus spp) rises after patients have experienced more than seven days of persistent neutropenic fever [51-53].
In patients who are clinically unstable or have a suspected fungal infection, antifungal therapy should be considered even earlier than what is recommended for empiric therapy. (See "Treatment of candidemia and invasive candidiasis in adults" and "Treatment and prevention of invasive aspergillosis".)
An ongoing question is whether all high-risk neutropenic patients with persistent fever need to receive empiric antifungal therapy, since fungal infection is not documented in most patients. The following observations provide limited support for this practice:
●In an autopsy study of patients who died after prolonged neutropenic fever between 1966 and 1975, 69 percent of patients had evidence of systemic fungal infection . It should be noted that over one-half of the patients in this early series had Candida infections, which may have been effectively prevented with antifungal prophylaxis strategies. For example, in one trial, fungal infections were documented in only 1 percent of persistent fevers in patients receiving fluconazole prophylaxis .
●Resolution of fever occurs in approximately 40 to 50 percent of patients given antifungal therapy [52,53,55,56]. However, this does not prove that the patient had an occult fungal infection. Since slow responses to empiric antibacterial therapy are frequent in high-risk patients, the defervescence noted in temporal association with antifungal therapy may be explained by antibacterial therapy.
Choice of drug — The choice of agent for empiric antifungal therapy depends upon which fungi are most likely to be causing infection, as well as the toxicity profiles and cost (algorithm 3) . In patients who have not been receiving antifungal prophylaxis, Candida spp are the most likely cause of invasive fungal infection. In patients receiving fluconazole prophylaxis, fluconazole-resistant Candida spp (eg, C. glabrata and C. krusei) and invasive mold infections, particularly Aspergillus spp, are the most likely causes.
The 2010 IDSA guidelines for empiric antifungal therapy recommend amphotericin B deoxycholate, a lipid formulation of amphotericin B, caspofungin, voriconazole, or itraconazole as suitable options for empiric antifungal therapy in neutropenic patients . The evidence to support the use of each of these agents is presented below. (See 'Studies' below.)
We favor the following approach:
●For persistently febrile patients who have not been receiving antifungal prophylaxis and who have no obvious site of infection, such as pulmonary nodules, we favor caspofungin (or another echinocandin) since Candida spp is a likely cause in such patients and the echinocandins provide excellent coverage for Candida spp and are well tolerated.
●For persistently febrile patients with pulmonary nodules or nodular pulmonary infiltrates, invasive mold infection should be strongly suspected and treated. Prompt assessment frequently requires bronchoscopy with bronchoalveolar lavage with cultures, stains, and Aspergillus galactomannan antigen testing to distinguish bacterial from mold pathogens, while simultaneously initiating antibacterial and anti-mold therapy until the specific etiology is established. (See "Diagnosis of invasive aspergillosis", section on 'Approach to diagnosis'.)
●Voriconazole or a lipid formulation of amphotericin B are preferred in patients with pulmonary findings suggestive of an invasive mold infection due to higher failure rates with caspofungin in preventing and treating invasive aspergillosis, which is the most common cause of mold infections . Current data are insufficient to conclusively determine whether voriconazole or a lipid formulation of amphotericin B is optimal; the choice of the initial antifungal agent may vary based on an institution's experience (ie, epidemiology and susceptibility patterns) and patient risks for specific mold infections (eg, Aspergillus versus the agents of mucormycosis). Most experts prefer voriconazole if aspergillosis is thought to be most likely, but if mucormycosis is suspected, an amphotericin B formulation should be given since voriconazole has no activity against the agents of mucormycosis. It is important to note that at most centers, aspergillosis accounts for the majority of invasive fungal infections in neutropenic patients. (See "Prophylaxis of invasive fungal infections in adults with hematologic malignancies", section on 'Epidemiology' and "Prophylaxis of invasive fungal infections in adult hematopoietic cell transplant recipients", section on 'Epidemiology' and "Epidemiology and clinical manifestations of invasive aspergillosis" and "Treatment and prevention of invasive aspergillosis" and "Mucormycosis (zygomycosis)".)
●For persistently febrile patients who have been receiving anti-mold prophylaxis, a different class of antifungal agent with activity against molds should be used for empiric therapy. For example, if voriconazole or posaconazole has been used for prophylaxis, an amphotericin B formulation should be used. We favor a lipid formulation of amphotericin B rather than amphotericin B deoxycholate in order to minimize toxicity.
It is also important to remember that caspofungin and other echinocandins are not active against Cryptococcus spp, Trichosporon spp, and filamentous molds other than Aspergillus spp, such as Fusarium spp. In addition, some yeasts can demonstrate relative resistance to these drugs (Candida parapsilosis, C. rugosa, C. guilliermondii, and non-candidal yeasts). Failure of caspofungin to prevent aspergillosis has also been reported, even though it has in vitro activity against Aspergillus spp . Moreover, the clinical efficacy of the echinocandins for endemic fungi (Histoplasma, Blastomyces, Coccidioides spp) has not been demonstrated. (See "Cryptococcus neoformans infection outside the central nervous system" and "Treatment of Cryptococcus neoformans meningoencephalitis and disseminated infection in HIV seronegative patients" and "Infections due to Trichosporon species and Blastoschizomyces capitatus" and "Diagnosis and treatment of disseminated histoplasmosis in non-HIV-infected patients" and "Treatment and prevention of Fusarium infection".)
Dosing — The dosing of the various antifungal agents recommended above is as follows:
•Caspofungin – Loading dose of 70 mg IV on day one, then 50 mg IV once daily
•Voriconazole – Loading dose of 6 mg/kg IV every 12 hours on day one, followed by 4 mg/kg IV every 12 hours
•Amphotericin B lipid complex – 5 mg/kg IV once daily
•Liposomal amphotericin B – 3 to 5 mg/kg IV once daily
Studies — The following observations have been made about each of the agents used for empiric antifungal therapy.
Amphotericin B formulations
Amphotericin B deoxycholate has historically been the most common agent given. In a randomized trial of patients with persistent neutropenic fever, liposomal amphotericin B was compared with amphotericin B deoxycholate using a composite of five end points: survival, resolution of fever during neutropenia, resolution of preexisting fungal infection, prevention of breakthrough fungal infection, and absence of premature discontinuation of the drug because of toxicity or lack of efficacy . Both drugs performed similarly in terms of survival (93 versus 90 percent) and resolution of fever (58 percent each), but the liposomal preparation was associated with significant reductions in breakthrough fungal infections; infusion-related fever, chills, or rigors; and nephrotoxicity. The lipid formulations of amphotericin B have replaced amphotericin B deoxycholate in most centers due to toxicity considerations, but they are more costly. Amphotericin B deoxycholate should be avoided in patients with antecedent renal disease and in those receiving other nephrotoxic drugs.
Echinocandins — A randomized trial compared the echinocandin, caspofungin, to liposomal amphotericin B in 1095 patients with persistent neutropenic fever despite four days of empiric antibiotic therapy . The overall success rates (34 and 34 percent), and the rates of breakthrough fungal infections and resolution of fever were similar in both arms.
In the small subset of patients who had a fungal infection at baseline (only 27 per group), a successful outcome was significantly more likely with caspofungin than liposomal amphotericin (52 versus 26 percent). Caspofungin was also associated with a significantly higher rate of survival seven days after the completion of therapy (93 versus 89 percent) and was significantly less likely to be associated with nephrotoxicity (2.6 versus 11.5 percent), infusion-related events (35 versus 52 percent), or cessation of therapy for drug-related adverse events (5 versus 8 percent).
Micafungin and anidulafungin are other echinocandins, but they have not been adequately studied or approved by the FDA for patients with neutropenic fever. Micafungin appeared to be effective in small studies in patients with neutropenia [61,62]. These agents can be used as alternatives to caspofungin when caspofungin is not available, since the spectrum and antifungal activity of all three agents is similar. (See "Treatment of candidemia and invasive candidiasis in adults", section on 'Echinocandins' and "Pharmacology of echinocandins".)
Voriconazole — An international open-label, randomized trial compared liposomal amphotericin B to voriconazole in 837 neutropenic patients with persistent fever . Mortality was similar in both groups. There was a trend toward a better response with liposomal amphotericin B compared with voriconazole in four of five composite endpoints, including overall response (31 versus 26 percent), but voriconazole was associated with significantly fewer documented breakthrough fungal infections (2 versus 5 percent). Voriconazole was associated with fewer infusion-related adverse effects and less nephrotoxicity, but with more cases of transient visual changes and hallucinations.
The results of this study are difficult to interpret and may have been affected by the open-label design . More patients had voriconazole stopped for perceived lack of efficacy (ongoing fevers), although fevers persisted equally in the liposomal amphotericin B group. The FDA reviewed the results of the trial along with additional information from the sponsor and recommended not to approve a licensed indication for voriconazole for empiric antifungal treatment, since voriconazole did not fulfill the criterion for noninferiority compared with liposomal amphotericin B [58,65]. Whether the failure to meet the statistical definition of noninferiority reflects a true difference in efficacy between voriconazole and liposomal amphotericin B or a problem in the study design awaits further clinical trials.
Nevertheless, the IDSA guidelines recommend voriconazole as an option for empiric antifungal therapy due to its proven efficacy against Candida and Aspergillus spp infections, the chief fungal pathogens in patients with neutropenic fever, and the lower risk of serious toxicity compared with amphotericin B formulations .
●Posaconazole – Posaconazole is a broad-spectrum triazole that has been approved by the FDA for the prophylaxis of fungal infections in neutropenic patients and for the treatment of mucocutaneous candidiasis. It has in vitro activity against yeasts and molds (such as Aspergillus spp and the Mucorales), but it has not been studied for the empiric treatment of invasive fungal infections in neutropenic patients or for the treatment of established mold infections. Furthermore, its availability only in an oral formulation and its variable absorption limit its use for treatment of suspected infection in most experts’ opinion. (See "Prophylaxis of invasive fungal infections in adults with hematologic malignancies" and "Prophylaxis of invasive fungal infections in adult hematopoietic cell transplant recipients" and "Treatment and prevention of invasive aspergillosis", section on 'Posaconazole'.)
●Itraconazole – A randomized trial compared IV followed by oral itraconazole to amphotericin B deoxycholate for up to 28 days as empiric therapy in 384 febrile neutropenic patients . Itraconazole was at least as effective as amphotericin B (response rate 47 versus 38 percent), and significantly fewer patients had drug-related adverse effects (5 versus 54 percent). The median duration of neutropenia in this study was 10 and 8 days for the itraconazole and amphotericin groups, respectively. Five breakthrough fungal infections were observed in each group.
Despite its efficacy, we do not recommend itraconazole for patients with neutropenic fever because it has several important drawbacks. It should not be used in patients with an estimated creatinine clearance below 30 mL/min, and appropriate dosing in hepatic failure has not been well characterized. Itraconazole has negative inotropic properties and can induce or exacerbate congestive heart failure. The IV formulation of itraconazole is no longer available in the United States, and oral formulations have variable bioavailability. It also has a number of important drug interactions, most notably with cyclosporine, quinidine, and HMG-CoA reductase inhibitors (“statins”). (See "Pharmacology of azoles", section on 'Drug interactions'.)
●Fluconazole – Fluconazole is generally not recommended for empiric antifungal therapy because of its lack of activity against molds, a major concern in patients with hematologic malignancies and those undergoing HCT.
Preemptive antifungal therapy — An alternative to empiric antifungal therapy is preemptive therapy; this involves serial screening of high-risk patients for markers of fungal colonization and/or infection in an attempt to prevent invasive infection. Such markers include laboratory tests (eg, Aspergillus galactomannan antigen, beta-D-glucan assay) and imaging tests (high-resolution chest computed tomography). If one of these markers suggests a fungal infection, antifungal therapy is started. This approach is best suited for patients receiving prophylaxis with an anti-yeast agent, such as fluconazole, where the concern is mainly mold pathogens and one is considering broadening the coverage to include an anti-mold agent.
Some studies support the use of preemptive antifungal therapy. However, further study is necessary to establish the efficacy and safety of this approach . This is discussed in greater detail separately. (See "Treatment and prevention of invasive aspergillosis", section on 'Preemptive therapy'.)
Duration — If an infectious source of fever is identified, antibiotics should be continued for at least the standard duration indicated for the specific infection (eg, 14 days for Escherichia coli bacteremia); antibiotics should also continue at least until the absolute neutrophil count (ANC) is ≥500 cells/microL, or longer if clinically indicated .
When no source is identified and cultures are negative, the timing of discontinuation of antibiotics is usually dependent on resolution of fever and clear evidence of bone marrow recovery. If the patient has been afebrile for at least two days and the ANC is >500 cells/microL and is showing a consistent increasing trend, antibiotics may be stopped . An alternative approach in patients who remain neutropenic involves continuing antibiotics until an appropriate treatment course has been completed and all signs and symptoms of infection have resolved, and then switching to oral fluoroquinolone prophylaxis until marrow recovery has occurred; however, this approach has not been well studied .
In patients with an identifiable cause of fever (eg, a urinary tract infection caused by E. coli) who have had steady clinical improvement but who require continued antibiotic therapy following the resolution of neutropenia to complete the course of therapy, a switch from IV antibiotic therapy to appropriate oral agents can be considered.
CATHETER REMOVAL — Central venous catheter (CVC)-related infections are common in patients with neutropenic fever. If blood cultures drawn from the CVC become positive at least 120 minutes before peripheral blood cultures drawn at the same time, then the CVC is likely to be the source of the bacteremia . In addition to antibiotics, CVC removal is recommended for patients with catheter-related bloodstream infections in which any of the following organisms is implicated :
●Candida species (see "Treatment of candidemia and invasive candidiasis in adults", section on 'Catheter removal')
●Rapidly growing nontuberculous mycobacteria
This recommendation is based upon observational studies showing improved clearance of infection, and, in some cases, a mortality benefit, among patients with S. aureus, P. aeruginosa, or Candida spp bloodstream infections in whom the CVC was removed [67-72]. In a study of cancer patients with bacteremia caused by rapidly growing mycobacteria, CVC removal was associated with a significantly reduced rate of relapse of bacteremia . There are no studies that have demonstrated the benefit of CVC removal in patients with CVC-related bloodstream infections caused by fungi other than Candida spp, but CVC removal is generally recommended in such patients based upon the severity of these infections and the biologic plausibility of the potential benefit of this intervention.
Antibiotics should be administered for a minimum of 14 days following catheter removal and clearance of blood cultures. (See "Treatment of intravascular catheter-related infections", section on 'Removal'.)
Catheter removal is also recommended for tunnel infection, port pocket infection, septic thrombosis, endocarditis, sepsis with hemodynamic instability, and bloodstream infection that persists despite ≥72 hours of therapy with appropriate antibiotics, even when pathogens other than those described above are isolated .
A prolonged duration of treatment of four to six weeks is recommended for patients with complicated CVC-associated infections, such as those with deep tissue infection, endocarditis, septic thrombosis, or persistent bacteremia or fungemia occurring >72 hours following catheter removal in a patient receiving appropriate antimicrobial therapy . The evidence to support the need for an extended duration of therapy is strongest for S. aureus, but this approach is also suggested for complicated infections caused by other pathogens.
For CVC-associated bacteremia caused by coagulase-negative staphylococci, the CVC may be retained; in this setting, patients are treated with systemic antibiotics with or without antibiotic lock therapy . (See "Treatment of intravascular catheter-related infections", section on 'Antibiotic lock therapy' and "Tunneled, cuffed hemodialysis catheter-related bacteremia", section on 'Antibiotic lock'.)
Although the Infectious Diseases Society of America (IDSA) guidelines recommend CVC removal in patients with candidemia, some authors have suggested that catheter removal may not be necessary in neutropenic patients with candidemia, in whom the source is often the gastrointestinal tract, rather than the CVC. This is discussed in detail separately. (See "Treatment of candidemia and invasive candidiasis in adults", section on 'Catheter removal'.)
COLONY STIMULATING FACTORS — Colony stimulating factors (CSFs; also known as myeloid growth factors or hematopoietic growth factors), such as granulocyte and granulocyte-macrophage colony stimulating factors (G-CSF and GM-CSF), are generally not recommended for the management of patients with established fever and neutropenia, with some exceptions. These issues are discussed in detail separately. (See "Use of granulocyte colony stimulating factors in adult patients with chemotherapy-induced neutropenia and conditions other than acute leukemia, myelodysplastic syndrome, and hematopoietic cell transplantation", section on 'Neutropenic fever'.)
MYELOID RECONSTITUTION SYNDROME — Clinicians should be aware of the myeloid reconstitution syndrome, a circumstance in which there may be onset or progression of an inflammatory focus defined clinically or radiologically that manifests at the time of neutrophil recovery . Because such processes may appear in the context of a persistent neutropenic fever syndrome, the likelihood of superinfection must be considered with respect to the antimicrobial spectrum of the patient’s current empiric antibacterial therapy and the microbiologic differential diagnosis applicable to those circumstances. (See "Overview of neutropenic fever syndromes", section on 'Neutropenic fever syndromes'.)
OUTCOMES — The efficacy of the treatment of patients with neutropenic fever syndromes has improved greatly as demonstrated by a progressive decline in mortality rates since the prompt initiation of empiric coverage was implemented in the 1970s [17,75,76]. Studies from the 1960s, before the routine use of empiric therapy, documented mortality rates of 90 percent in neutropenic patients with bacteremia caused by gram-negative bacilli [77,78]. Sepsis due to P. aeruginosa or Escherichia coli resulted in death within 48 hours after the first blood culture had been drawn in approximately one-half of patients [79,80].
In contrast, in a study of 41,779 adults with cancer who were hospitalized with neutropenic fever in the United States between 1995 and 2000, the in-hospital mortality was 9.5 percent . The mortality rate depended upon the underlying malignancy: 8 percent in patients with solid tumors, 8.9 percent in patients with lymphoma, and 14.3 percent in patients with leukemia. The number of major comorbid conditions also significantly affected the mortality rate. Patients without any comorbidities had a 2.6 percent risk of dying, compared with 10.3 percent for patients with one comorbidity, 21.4 percent for patients with two comorbidities, 38.6 percent for patients with three comorbidities, and 50.6 percent for patients with four comorbidities.
Outcomes are poor in neutropenic patients who are critically ill. In a cohort study conducted in 428 neutropenic patients with severe sepsis or septic shock in France between 1998 and 2008, the hospital mortality was 50 percent .
Factors that influence the risk of treatment failure are discussed separately. (See "Risk assessment of adults with chemotherapy-induced neutropenia", section on 'Risk of treatment failure'.)
PROPHYLAXIS — The prophylaxis of infection in patients at risk for neutropenic fever is discussed elsewhere. (See "Prophylaxis of infection during chemotherapy-induced neutropenia in high-risk adults" and "Prophylaxis of invasive fungal infections in adults with hematologic malignancies" and "Prophylaxis of invasive fungal infections in adult hematopoietic cell transplant recipients" and "Prophylaxis of infections in hematopoietic cell transplant recipients" and "Use of granulocyte colony stimulating factors in adult patients with chemotherapy-induced neutropenia and conditions other than acute leukemia, myelodysplastic syndrome, and hematopoietic cell transplantation".)
SUMMARY AND RECOMMENDATIONS
●Cancer patients receiving cytotoxic antineoplastic therapy sufficient to adversely affect myelopoiesis and the integrity of the gastrointestinal mucosa are at risk for invasive infection due to colonizing bacteria or fungi that translocate across intestinal mucosal surfaces. Because neutropenic patients are unable to mount robust inflammatory responses, serious infection can occur with minimal symptoms and signs. In such patients, fever is often the only sign of infection. It is critical to recognize neutropenic fever early and to initiate empiric systemic antibacterial therapy promptly in order to avoid progression to a sepsis syndrome and possibly death. (See 'Introduction' above.)
●It is crucial to assess the risk of serious complications in patients with neutropenic fever, since this assessment will dictate the approach to therapy, including the need for inpatient admission, IV antibiotics, and prolonged hospitalization. High-risk neutropenic patients are those with an absolute neutrophil count (ANC) <500 cells/microL expected to last >7 days or evidence of ongoing comorbid conditions (table 1). Profound prolonged neutropenia is most likely to occur in the preengraftment phase of hematopoietic cell transplantation (particularly allogeneic) and in patients undergoing induction chemotherapy for acute leukemia. (See 'Risk of serious complications' above.)
Initial empiric therapy
●Fever in high-risk neutropenic patients should be considered a medical emergency. Empiric broad-spectrum antibacterial therapy should be initiated immediately after blood cultures have been obtained and before any other investigations have been completed (algorithm 1). Antibiotics should be administered within 60 minutes of presentation in patients presenting with neutropenic fever. (See 'General principles' above and 'Timing of antibiotics' above.)
●Initial antibiotic selection should be guided by the patient's history, allergies, symptoms, signs, recent antibiotic use and culture data, and awareness of institutional nosocomial infection and susceptibility patterns. Ideally, antibiotics should be bactericidal. (See 'General principles' above.)
●For high-risk patients with neutropenic fever, we recommend empiric monotherapy with an anti-pseudomonal beta-lactam agent, such as cefepime, meropenem, imipenem, or piperacillin-tazobactam (algorithm 1) (Grade 1B). Because of high rates of resistance, ceftazidime is no longer recommended at most centers, but it is reasonable to use it at centers with low rates of resistance. (See 'Initial regimen' above.)
●We recommend against the use of vancomycin (or another agent that targets gram-positive cocci) as a standard part of the initial regimen (Grade 1A). In contrast, gram-positive coverage should be added in patients with suspected central venous catheter-related infection, skin or soft tissue infection, pneumonia, or hemodynamic instability. For patients with complicated presentations (eg, hypotension; central venous catheter, skin, or soft tissue infections; pneumonia), coverage should be broadened to cover the likely pathogens (eg, resistant gram-negative, gram-positive, and anaerobic bacteria, as well as fungi). (See 'Initial regimen' above and 'Antibiotic resistance' above.)
Empiric therapy for persistent fever
●Persistent fever alone in an otherwise stable patient is not sufficient justification for modification of the initial antibiotic regimen. However, modifications to the initial regimen should be considered for patients at risk for infection with antibiotic-resistant organisms, for patients who are clinically unstable, and for patients who have positive blood cultures that are suggestive of a resistant infection. (See 'Modifications to the regimen' above and 'Persistent fever' above.)
●Patients who remain febrile after the initiation of empiric antibiotics should be reevaluated for possible infectious sources. Management algorithms have been developed for the reassessment of neutropenic patients with persistent fever (algorithm 2 and algorithm 3). (See 'Modifications to the regimen' above.)
●We recommend the empiric addition of an antifungal agent for neutropenic patients who are persistently febrile after four to seven days despite broad-spectrum antibacterials and who have no documented source of infection (algorithm 3) (Grade 1C). (See 'Indications' above.)
●The choice of antifungal agent for empiric therapy depends upon which fungi are most likely to be causing infection, as well as toxicity profiles and cost:
•For persistently febrile patients who have not been receiving antifungal prophylaxis and who have no obvious site of infection, such as pulmonary nodules, we favor caspofungin (or another echinocandin) since Candida spp is the most likely cause in such patients.
•For persistently febrile patients with pulmonary nodules or nodular pulmonary infiltrates, invasive mold infection should be strongly suspected. In such patients, we favor voriconazole or a lipid formulation of amphotericin B.
•For persistently febrile patients who have been receiving anti-mold prophylaxis, a different class of antifungal agent with activity against molds should be used for empiric therapy. (See 'Choice of drug' above.)
●In addition to antimicrobial therapy, we recommend central venous catheter removal for patients with catheter-related bloodstream infections caused by S. aureus, P. aeruginosa, Candida spp, or rapidly growing nontuberculous mycobacteria (Grade 1B). We also recommend central venous catheter removal for patients with catheter-related bloodstream infections with fungi other than Candida spp (Grade 1C). The central venous catheter should also be removed in patients with complicated infections (eg, tunnel infection, port pocket infection). (See 'Catheter removal' above and "Treatment of intravascular catheter-related infections", section on 'Removal' and "Treatment of candidemia and invasive candidiasis in adults", section on 'Catheter removal'.)
ACKNOWLEDGMENT — The editorial staff at UpToDate Inc. would like to acknowledge Dr. Gregory K Robbins, who contributed to earlier versions of this topic review.
All topics are updated as new information becomes available. Our peer review process typically takes one to six weeks depending on the issue.