Perspectives for the management of febrile neutropenic patients with cancer in the 21st century


  • Nikolaos V. Sipsas M.D.,

    1. Infectious Diseases Unit, Pathophysiology Department, Laikon General Hospital and National and Kapodistrian University of Athens, Athens, Greece
    Search for more papers by this author
  • Gerald P. Bodey M.D.,

    1. Department of Infectious Diseases, Infection Control, and Employee Health, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    Search for more papers by this author
  • Dimitrios P. Kontoyiannis M.D., Sc.D.

    Corresponding author
    1. Department of Infectious Diseases, Infection Control, and Employee Health, The University of Texas M. D. Anderson Cancer Center, Houston, Texas
    • Department of Infectious Diseases, Infection Control, and Employee Health, Unit 402, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Houston, TX 77030
    Search for more papers by this author
    • Fax: (713) 745-6839


Over the past several decades, there has been substantial progress in the management of patients with febrile neutropenia. However, the ever-changing patterns of infection, ecology, and antibiotic-resistance trends do not allow the development of treatment guidelines that could be applied universally. Hence, the institution's predominant pathogens and resistance patterns should guide the empirical choice of antimicrobials. Prompt initiation of antimicrobial therapy remains the gold standard. Monotherapy with the newer broad-spectrum antimicrobials has tended to replace the classic combination therapy. Empirical administration of glycopeptides, such as vancomycin, without documentation of a gram-positive infection is not favored. The development of risk-stratification models has allowed for identification of low-risk patients with additional treatment options, such as early discharge and exclusively outpatient treatment with oral antimicrobials. The initiation of empirical antifungal therapy in persistently febrile neutropenic patients has become common practice, especially recently, since the introduction of new, effective, less toxic antifungal drugs. It is hoped that the development of new nonculture-based diagnostic methods will allow for the early detection of invasive fungal infections and, thus, the replacement of empirical antifungal therapy with pathogen-specific, preemptive therapy. Cancer 2005. © 2005 American Cancer Society.

Over the past several decades, there has been substantial progress in the treatment of patients with neoplastic disease and, thus, decreasing cancer death rates.1 Research has provided a new array of chemotherapeutic agents; and modern treatment modalities, including bone marrow and stem cell transplantation, have been introduced successfully into clinical practice. Unfortunately, the majority of these treatment options continue to face a formidable foe: profound suppression of innate and/or acquired immunity. Neutropenia in particular remains the most prominent chemotherapy-induced immune defect, rendering patients susceptible to infections.2 Hence, despite improvements in long-term survival, infection remains a common complication of cancer therapy and accounts for the majority of chemotherapy-associated deaths, especially when the administration of proper antibiotics is delayed.3 It is noteworthy that infection frequently has a negative impact on the dose intensity of subsequent antineoplastic therapy.4

The Infectious Diseases Society of America (IDSA) first established guidelines to assist infectious diseases specialists, oncologists, internists, pediatricians, and family practitioners in the treatment of febrile neutropenic patients with cancer in 1990 and revised the guidelines in 1997 and in 2002.5 Herein, we present a review of recent developments in the field of treatment of febrile neutropenia and offer clinical perspectives for the 21st century. Data for this review were derived from a Medline search of the English language literature using the terms fever, neutropenia, and cancer as keywords and from references of relevant articles and book chapters.

The Changing Spectrum of Pathogens that Cause Infection in the Neutropenic Patient

There is a myriad of described pathogens in neutropenic hosts.5 This is not surprising, because the main source of these pathogens is the host's endogenous flora.6 However, many other exogenous microorganisms of low virulence that can be acquired from contaminated air or water or from contact with other patients, personnel, or equipment (e.g., vancomycin-resistant enterococci [VRE]), can become invasive and cause infection in neutropenic patients.

Bacterial pathogens

Historically, gram-negative bacilli arising from the alimentary tract have been the prominent pathogens in neutropenic hosts. Between the 1960s to the mid-1970s, Escherichia coli, Klebsiella species, and Pseudomonas aeruginosa accounted for the majority of microbiologically documented infections at most cancer centers.7 Since the introduction of extended-spectrum β-lactams, several institutions in both the United States and Europe have experienced a decrease in gram-negative rod bacteremia and an increase in infections due to gram-positive cocci.8 In a recent study of bloodstream infections (BSIs) in adult patients with malignancies conducted across the United States, the proportion of gram-positive organisms increased from 62% in 1995 to 76% in 2000, whereas the proportion of gram-negative infections decreased from 22% to 15%.9

However, focusing only on BSIs may lead to an underestimation of the continuous threat caused by gram-negative rods in the neutropenic cancer population. In a recent survey conducted at The University of Texas M. D. Anderson Cancer Center, the most common infections were not nosocomial BSIs but, rather, tissue-based infections, such as pneumonia, urinary tract infections, and soft tissue infections, which were caused predominately by gram-negative rods.10 By the same token, the frequency of Pseudomonas septicemia in leukemic patients at The University of Texas M. D. Anderson Cancer Center from both 1972–1981 and 1991–1995 was similar.11

The exact reasons for the shift in the etiology of bacteremia in neutropenic hosts are debatable. Factors implicated in the increase in gram-positive infections include the routine use of central venous catheters (CVCs)8; the widespread use of quinolone prophylaxis12; the administration of high-dose, cytarabine-containing chemotherapeutic regimens in patients with leukemia; and the increased use of proton pump inhibitors.13 Recent evidence suggests that the pendulum may swing back again toward infections caused by gram-negative pathogens.14 One possible reason for this is the declining use of quinolone prophylaxis.

Anaerobic bacteremia occurs in < 5% of patients with febrile neutropenia,15 a percentage that does not appear to have changed much over the past 30 years.9 Patients with cancer who have neutropenic colitis, intraabdominal infections, perirectal abscesses, or periodontal disease are at risk for anaerobic bacteremia, frequently in the context of a polymicrobial infection.

Bacterial pathogens resistant to antibiotics: An emerging problem for the febrile neutropenic patient

Of concern is the emergence of multidrug-resistant, gram-negative rods, such as P. aeruginosa, E. coli, Citrobacter species, Acinetobacter species, and Stenotrophomonas maltophilia, as frequent offenders in neutropenic patients.16 This trend, in all likelihood, is a consequence of antibiotic selection pressure and induction of extended-spectrum chromosomal β-lactamases after the use of β-lactams, including the carbapenems.17, 18

Similarly, the incidence of resistance of the common gram-positive cocci to β-lactams is increasing in neutropenic patients.8 In fact, infection by methicillin-resistant Staphylococcus aureus (MRSA), methicillin-resistant Staphylococcus epidermidis, and penicillin-resistant streptococci has become common in these patients.9 Antibiotic selection pressure with agents that have weak streptococcal activity, such as the quinolones19 and ceftazidime,20 and chemotherapy-induced mucositis in the gastrointestinal tract have accounted for the emergence of antibiotic-tolerant and antibiotic-resistant Streptococcus viridans as significant pathogens. S. viridans can cause fulminant bacteremia and is associated with acute respiratory distress syndrome, renal failure, and rapid death.21 Other relatively uncommon gram-positive organisms that usually are resistant to β-lactams, such as Corynebacterium jeikeum and Bacillus, Leuconostoc, Lactobacillus, and Rhodococcus species, have been recognized increasingly as important pathogens in neutropenic hosts.8

The emergence of antibiotic-resistant, gram-positive pathogens led to the extensive use of vancomycin as part of initial empirical treatment of febrile neutropenia. This has resulted in resistance or decreased susceptibility to vancomycin in some important gram-positive bacteria, most notably VRE, which may account for up to one-third of all enterococcal infections in some cancer institutions.9, 22, 23 VRE has caused nosocomial outbreaks of bacteremia associated with high mortality in leukemic patients, although VRE bacteremia was not the sole factor that effected the poor outcome.24 The emergence of VRE led some institutions preemptively to use newer agents with activity against gram-positive infections, notably linezolid and dalfopristin-quinupristin. However, this strategy may cause the same problems, because increasing resistance to linezolid already has been reported.25 Clearly, in view of the emerging bacterial resistance, antibiotic stewardship involving formulary replacement or restriction is needed.26 Methods of infection control designed to decrease horizontal transfer of multidrug-resistant bacteria in susceptible patients are also of paramount importance.27

Fungal pathogens

Several frequently interrelated risk factors, along with severe, protracted neutropenia, predispose febrile neutropenic patients with cancer to fungal infections. These include prior use of broad-spectrum antibiotics or adrenal corticosteroids,28 advanced age, intensity of chemotherapy, presence of indwelling central catheters,29 tissue damage, and the stage of the underlying disease.

The most common fungal infections in febrile neutropenic patients are candidiasis and aspergillosis. The clinical spectrum of candidiasis is broad, ranging from superficial to disseminated disease. Systemic infections can be classified further as candidemia, acute or chronic disseminated candidiasis, or single-organ candidiasis. Candida species also are common causes of catheter-related infections.29 Candidiasis may be the primary infection in the setting of neutropenia and, more commonly, may be a superinfection after a prior bacterial infection. The most common Candida spp. that cause candidemia are Candida albicans followed by Candida glabrata, Candida tropicalis, and Candida parapsilosis.30 With the extensive use of fluconazole prophylaxis in neutropenic patients, non-albicans Candida spp. and azole-resistant Candida infections have occurred.31, 32 Whether fluconazole use is the sole factor in the changes in the epidemiology of candidiasis is a matter of debate.33

Aspergillosis is the most common invasive mold infection in patients with hematologic malignancies who experience protracted, severe neutropenia.34 Aspergillosis initially affects the lungs or sinuses; in approximately 30% of neutropenic patients, it later disseminates to other organs.34Aspergillus fumigatus is the most common species that causes invasive disease. Recently, the proportion of non-fumigatus Aspergillus species resistant to amphotericin B deoxycholate (AMB), such as Aspergillus terreus, appears to be increasing in these patients.35

Numerous other opportunistic fungi have emerged as important pathogens in persistently and profoundly neutropenic hosts, especially in patients with hematologic malignancies and in bone marrow transplantation recipients.36 These include Fusarium species, Scedosporium species, Acremonium species, Trichosporon beigelii, and Scopulariopsis species among others. Because many of these fungi are resistant to several antifungals, including AMB,37 they tend to present as breakthrough infections. Among these fungi, Fusarium spp. are a common cause of sinopulmonary infection, fungemia, and disseminated skin lesions.38

Initial Assessment of Patients with Febrile Neutropenia

Physical examination

Given the inability of immunocompromised hosts to mount an adequate inflammatory response, the classic signs and symptoms of infection, other than fever, may be minimal or absent.39 In a clinical study, the initial evaluation of febrile neutropenic patients for the presence of infection was erroneous in 33% of patients.40 Hence, physicians should perform a meticulous physical examination and should take notice of every minor and subtle sign and symptom of infection and investigate them further when assessing patients with febrile neutropenia. A focused physical examination should be repeated daily if the symptoms persist and always should include examination of the skin, skin folds, genitalia, anal area, sinuses, oropharynx, and fundi. In addition, intravenous lines should be inspected carefully for signs of inflammation, and any exudates from the catheter site should be stained and cultured. In patients with leukemia, inspection and palpation of the site of a prior bone marrow aspiration along with palpation of the skin and visual inspection for subcutaneous, disseminated, nodular lesions should be performed.

Laboratory evaluation

Similarly, the lack of an adequate inflammatory response renders some laboratory tests unreliable. In a series of febrile neutropenic patients with cancer who had urinary tract infections, only 11% of patients with < 100 neutrophils/mm3 had pyuria.3 In another study among neutropenic patients with pneumonia, 40% of patients had a normal chest radiograph, whereas only 8% of patients had polymorphonuclear cells on sputum Gram stains.41 Initial laboratory evaluations should include a complete blood cell count, measurement of the serum creatinine and urea nitrogen levels, and liver function tests. Whereas neutropenic patients with pneumonia may have a normal chest radiograph, patients with a normal examination may have abnormal chest radiographs. Therefore, initial laboratory evaluation should include a chest radiograph, especially in patients with respiratory symptoms. Concomitant blood cultures obtained from peripheral veins and/or a catheter also should be performed. A differential time to positivity of ≥ 120 minutes between cultures drawn through the catheter and from a peripheral vein is highly specific for catheter-related bacteremia.42 Thus, an efficient clinical microbiology laboratory that provides reliable antibiograms is of paramount importance for the selection of appropriate antibiotics. Culture of stool or urine samples and a lumbar puncture should be performed when there is clinical indication of infection involving these sites. Biopsy and culture of skin lesions also is helpful in many cases.43 High-resolution chest computed tomography may indicate early signs of invasive mold infection, such as the halo sign, even in patients with a normal chest radiograph.44

Risk assessment

Risk assessment of febrile neutropenic patients with cancer gradually is becoming an integral part of their initial evaluation, because it allows for the identification of patients who are suitable for outpatient treatment and for better stratification of patients who may benefit from empirical antifungal therapy.45 It has become evident that patients with fever and neutropenia are not a homogeneous group, because it has been shown that they are not at the same risk for life-threatening complications or death,46, 47 and efforts have been made to formulate a risk-stratification model. Two risk-assessment systems were developed by Talcott et al.48 and the Multinational Association of Supportive Care in Cancer (MASCC).49 According to the latter system, various factors that were associated with a better outcome were assigned an integer weight to develop a risk-index score (Table 1). A risk-index score ≥ 21 identified low-risk patients with a < 5% risk of complications. One drawback of the MASCC study is that it included a relatively low percentage of patients with active acute leukemia, who typically are high-risk patients and, typically, should not be considered for outpatient therapy.

Table 1. The Multinational Association of Supportive Care in Cancer Risk Scoring Index for Identification of Low-Risk Febrile Neutropenic Patients at Presentation
  • a

    A risk index score of ≥ 21 indicates that the patient is likely to be at low risk for complications and morbidity. The scoring system was derived from a study by Klastersky et al., 2000.49

  • b

    Choose one item only.

  • c

    Does not apply to patients age ≤ 16 yrs.

Extent of illnessb 
 No symptoms5
 Mild symptoms5
 Moderate symptoms3
No hypotension5
No chronic obstructive pulmonary disease4
Solid tumor or no fungal infection4
No dehydration3
Outpatient at onset of fever3
Age < 60 yrsc2

Management of Febrile Neutropenia in Patients with Cancer


In the 1960s, bacteremia in neutropenic patients was due primarily to gram-negative bacilli, especially P. aeruginosa.50 It soon became evident that the presence of gram-negative rod bacteremia is associated with high mortality rates; if untreated or treated inappropriately, it is associated with a mortality rate > 50% within 48 hours.51 These facts led to the introduction of the concept of empirical treatment. In pivotal studies from the early 1970s,52, 53 neutropenic patients with fever received treatment immediately with a combination of antibiotics that targeted gram-negative bacilli, including P. aeruginosa. This strategy, which was adopted widely, led to a dramatic reduction in the mortality rate in febrile neutropenic patients with cancer.54

Empirical treatment of febrile neutropenia should be started as soon as possible, even before the results of cultures are available, and antibiotics should be given in maximal therapeutic doses adjusted appropriately to renal/hepatic function (Table 2). Should the cultures yield a specific pathogen, then the regimen can be modified accordingly, but it still should provide broad-spectrum coverage for the possible presence of copathogens and should prevent bacterial superinfection. The infecting organism is confirmed microbiologically in only one-third of neutropenic patients.54 However, the most probable pathogen can be presumed on the basis of the underlying malignancy, the nature of the treatment, the degree and duration of immunosuppression, the type of immune defect, and the predominant pathogens at the hospital where the patient receives care (Table 3).

Table 2. Principles of Initial Antibiotic Management in Neutropenic Cancer Patients with Fever
  1. CNS: central nervous system.

Risk stratification
What are the qualitative and quantitative immune defects?
Individualized therapy
 Any focal signs and symptoms
 Recent or current antibiotic exposure
 Comorbid conditions
 Current or recent hospitalization
 Past history of infection (especially invasive mold infections)
 Site of infection (e.g., CNS)
 Drugs should be given promptly
Table 3. Approach to the Differential Diagnosis of Presumed Infection in the Neutropenic Cancer Patient with Fever
What is the type and spectrum of immunodeficiency?
 Delineates the spectrum of pathogens
Is the clinical presentation suggestive of a particular pathogen/syndrome?
What is the prophylactic regimen?
 Outlines the spectrum of infections breaking through
What is the timetable of infection?
Could a prior history of infections give clues about the current episode?
 If so, then why is recurrence happening?

The issue of which antibiotics should be included in empirical therapy for febrile neutropenia remains debatable despite the continuous publication of numerous studies on the issue. The IDSA has issued guidelines to assist physicians in choosing the optimal regimen.5 Nevertheless, the continuously changing spectrum of pathogens that cause infections in neutropenic patients with cancer, the emergence of previously rare pathogens with unique resistance patterns, and the increasing resistance of common pathogens to antibiotics indicate that there are no treatment guidelines that can be applied universally. Clinicians should be aware of the predominant pathogens and an antibiogram that depicts the in vitro susceptibility patterns of the most prevalent pathogens in their own institution to select an efficient initial empirical therapy.

Initial empirical therapy in moderate-risk to high-risk patients requiring hospitalization

Moderate-risk to high-risk patients should be admitted to the hospital and administered empirical therapy with broad-spectrum antibiotics given intravenously for the entire febrile episode. In early pivotal studies of empirical therapy for febrile neutropenia, administration of a combination of a β-lactam with antipseudomonas activity and an aminoglycoside resulted in an overall response rate of 60–70%.52, 53 The advantages of combination therapy include coverage of a broad spectrum of pathogens, possible synergistic activity against gram-negative pathogens, rapid killing, achievement of bactericidal serum concentrations, a potential decrease in the emergence of resistant strains, and activity against anaerobes. The main drawbacks are suboptimal activity against gram-positive cocci, the ototoxicity and nephrotoxicity of aminoglycosides (which require monitoring of serum levels), and the use of multiple daily doses.

To overcome the disadvantages of combination therapy, several strategies have been proposed, such as once-daily dosing of aminoglycosides55 and substitution of quinolones for aminoglycosides.56 Another option is the early discontinuation of aminoglycoside administration if gram-negative bacteremia is not found within 2–3 days after the onset of fever and if the patient is stable. These strategies have not been validated extensively with rigorous studies in this patient population.

During the 1980s, the advent of new antibiotics with an extended antimicrobial spectrum (third-generation and fourth-generation cephalosporins, carbapenems, fluoroquinolones, and β-lactam/β-lactamase inhibitor combinations) led to the introduction of monotherapy, which is easier to administer and is less toxic than combination therapy. Well designed studies showed that monotherapy with ceftazidime,57 cefepime,58 meropenem,59 imipenem,60 or piperacillin/tazobactam61 was effective equally compared with conventional β-lactam/aminoglycoside combinations. Whether there is a drug that is superior to others when administered as monotherapy is unclear. However, some studies that compared carbapenem with ceftazidime demonstrated the superiority of imipenem or meropenem over ceftazidime monotherapy,62, 63 whereas another study showed comparable efficacy and safety of cefepime and imipenem.64 Limited and contradictory data are available regarding the efficacy of quinolones as monotherapy,65 whereas aminoglycosides clearly are not suitable for monotherapy in neutropenic patients despite in vitro susceptibility.66 Recently, a meta-analysis showed that monotherapy with a suitable agent was associated with a nonsignificant trend toward better survival, a significant advantage in preventing treatment failures, fewer adverse effects, and similar superinfection rates.67 In the current era of cost containment, the large number of treatment options that are similar in safety and efficacy allows physicians to take into consideration economic issues, such as cost minimization, cost-effectiveness, and cost utility, before reaching a decision.68

Addition of vancomycin to the initial empirical regimen

The incorporation of vancomycin into the initial empirical therapy for febrile neutropenia has been controversial. The rationale behind early use of glycopeptides is the coverage of infections due to antibiotic-resistant, gram-positive organisms, including S. viridans, MRSA, enterococci, and Corynebacterium species. Conversely, the extensive use of vancomycin may lead to the emergence of VRE,23, 24 prompting institutions to adopt restrictions on the use of the drug.69

A multinational study showed that not incorporating vancomycin into the initial empirical regimen but, rather, giving the drug only when a resistant, gram-positive organism was isolated did not lead to increased morbidity and mortality rates70 except in patients with bacteremia due to S. viridans. Another study showed that monotherapy with imipenem was as effective in febrile neutropenia as the combination of imipenem and vancomycin.64 Therefore, vancomycin should be included in the initial empirical regimen only at institutions that have a high rate of infections due to MRSA or S. viridans. Otherwise, the initial regimen should contain an agent with good activity against gram-positive organisms (i.e., cefepime, imipenem, meropenem, piperacillin/tazobactam), and vancomycin use should be reserved mainly for documented infections due to gram-positive organisms that are resistant to the initial treatment.71

Options for initial empirical therapy in low-risk febrile neutropenic patients

The reasons for treating febrile neutropenia in the hospital are to monitor patients closely and to treat life-threatening complications promptly, including hemorrhage and infection.72 However, hospitalization may be detrimental to these patients, because they are vulnerable to colonization and subsequent infection by nosocomial, frequently resistant pathogens. The risk-assessment systems developed by Talcott et al.48 and the MASCC49 can predict accurately which patients have a < 5% risk of serious complications. There are treatment options for these low-risk patients along with the conventional hospitalization and administration of intravenous, broad-spectrum antibiotics for the entire febrile episode. These options include early discharge with the use of oral antibiotics after initial stabilization in the hospital with oral73 or intravenous antibiotics65 and outpatient treatment only with oral antibiotics.74–76

Several studies74–76 and a recent meta-analysis77 have shown that oral antibiotics may be substituted safely for intravenous antibiotics in low-risk patients with febrile neutropenia. The best oral regimen studied to date is the combination of a quinolone (e.g., ciprofloxacin) with amoxicillin/clavulanate. Whether the new broad-spectrum fluoroquinolones will obviate the need for combination oral treatment regimens remains to be seen in ongoing studies. Although outpatient treatment with oral antibiotics has many advantages (low cost, avoidance of catheter use/infection, good quality of life, and low risk of superinfection with resistant nosocomial organisms), it still has the potential risk of life-threatening complications, such as septic shock, away from the hospital. In our experience, modern outpatient oral antibiotic therapy is safe in the low-risk patient who lives with a reliable adult within close distance from a hospital and has telephone access. Regardless, the initial clinical and laboratory evaluation should be done in the hospital or in a clinic that can provide rapid laboratory and radiology results, because self-medication at home without a previous evaluation in a health care setting may be dangerous.78 Moreover, patients should be instructed to return immediately to the hospital if they experience high fever, if they are unable to tolerate oral medications because of nausea and emesis, or if their overall condition deteriorates.

Timing of evaluation of the response to the initial empirical regimen

In an analysis of 488 episodes of febrile neutropenia, Elting et al.79 found that the median time to clinical response in hospitalized patients with cancer was 5–7 days. In contrast, in low-risk patients, it has been shown that the median time to defervescence is as short as 2 days.80, 81 Therefore, it would be reasonable not to change the initial regimen for the first 3–5 days, even if the patient remains febrile but otherwise is stable clinically.5 Needless to say, if the patient's condition deteriorates or if a pathogen resistant to initial antibiotics is isolated, then treatment should be modified promptly.

Duration of antibiotic therapy in the case of a clinical response

If a patient has a clinically or microbiologically documented, specific infection (e.g., pneumonia, cellulitis), then antibiotics can be administered until all symptoms and signs are resolved and cultures become sterile. In our experience, administration of therapy for 4 days after resolution of signs and symptoms and a minimum of 7 days is adequate regardless of whether the patient has persistent neutropenia. However, for the majority of febrile neutropenic patients, there is no clinical or microbiologic documentation of an infection, so sterilization of cultures is not a relevant endpoint. Therefore, decisions about the duration of treatment after defervescence must be made on a case-by-case basis, depending on 1) the presence of neutropenia, 2) the patient's risk group, 3) the need for further chemotherapy or invasive procedures, 4) the clinical stability of the patient, and 5) the presence of mucositis. If the patient is stable clinically and is afebrile for > 48 hours, and if his or her neutropenia is resolved, then antibiotic administration can be stopped safely.5

In contrast, it is not clear what the duration of treatment should be if the patient's fever is resolved but he or she remains neutropenic. Some authors state that treatment should be continued until the neutrophil count has risen to 500/mm3.80 However, prolonged antibiotic therapy may be harmful, because it is associated with toxicity, development of resistance, and fungal superinfections. Discontinuation of treatment prior to neutropenia resolution can be considered after 5–7 afebrile days in patients who are at low risk, are stable, do not have mucositis, have no further chemotherapeutic or invasive procedures scheduled, and are under close monitoring.5 Conversely, for unstable patients with profound neutropenia and mucositis, treatment should be continued for at least 2 weeks despite defervescence.

Persistent fever after 3–5 days of empirical antibiotic treatment

A challenging scenario for patients with febrile neutropenia is the persistence of fever after 3–5 days of empirical antibiotic therapy without clinical or microbiologic documentation of infection. A number of factors may cause nonresponsiveness to antibiotics, including occult fungal infections, bacterial infections with cryptic foci (e.g., abscess, endocarditis) or resistant bacterial organisms, atypical infections (toxoplasma, mycobacteria, fastidious pathogens, and viruses), slow response to the initial treatment, noninfectious causes of fever (such as drug-induced fever), underlying malignant disease, graft-versus-host disease, atelectasis, phlebitis, pulmonary embolism, transfusion, marantic endocarditis, and suboptimal serum concentrations of appropriate antibiotics (Table 4).81

Table 4. Causes of Persistent Fever in Neutropenic Patients
  1. VRE: vancomycin-resistant enterococci; AFB: acid-fast bacteria.

Resistant bacterial infection (e.g., VRE)
Bacterial infection associated with tissue necrosis/mucositis (endotoxemia)
Nonculturable cell wall-deficient bacteria
Nonbacterial infection (virus, AFB, toxoplasmosis)
Malignancy-related fever
Superinfection with fungi
Drug or transfusion fever

Careful reevaluation of the patient before modifying the initial empirical regimen is crucial. A meticulous physical examination may reveal new signs and symptoms and may provide clues regarding possible causes of infection.5 Additional laboratory work-up may consist of special cultures, serology for specific pathogens, and imaging studies, including high-resolution chest computed tomography and upper abdomen imaging.44 The utility of repeated blood cultures when the initial cultures are negative has not been proven.82 Should the reevaluation of a persistently febrile neutropenic patient reveal a specific cause, then treatment should be modified accordingly.

In a randomized study by Lazarus et al.,83 it was shown that the removal of a CVC is not helpful in persistently febrile neutropenic patients. Conversely, CVC removal is required when the tunnel or exit site is infected or if there is evidence of septic thrombophlebitis with or without septic emboli. Despite the uncontrolled nature of the existing literature, CVC removal is recommended for patients with clear-cut, CVC-related candidemia and bacteremia in which a Pseudomonas species, fast-growing atypical mycobacteria, S. aureus, Stenotrophomonas species, Bacillus spp., or C. jeikeum is the pathogen.84, 85 CVC-related BSI usually is diagnosed when a patient experiences erythema, purulence, or pain at the catheter site. In addition, concomitant blood cultures obtained from peripheral veins and/or a catheter are helpful.42

If reassessment of the patient yields no new information, then there are three options: Continue the initial regimen, empirically add a glycopeptide if it was not added previously, or empirically add antifungal therapy. Continuation of the initial treatment without modifications is an option reserved for stable patients who have expected, rapid resolution of neutropenia. Empirically changing the initial empirical regimen to one that includes other, broader spectrum antibiotics (i.e., switching from third-generation cephalosporins to carbapenems or cefepime as second-line therapy) is not supported by published evidence.86 Discontinuation of all antibiotics in persistently febrile neutropenic patients can be hazardous.87 The most prudent strategy is to continue intravenous administration of antibiotics until the neutropenia is resolved or for at least 2 weeks and then consider discontinuation under close observation if no infection has been documented and if the patient is stable despite having persistent fever and neutropenia. If the fever persists after the resolution of neutropenia (i.e., > 500 cells/mL), then the patient should be reassessed for cryptic bacterial, fungal, or viral infections.88 If no infection is documented, then treatment can be discontinued 5 days after the resolution of neutropenia despite the presence of persistent fever.5

Empirical addition of a glycopeptide to the initial antibiotic regimen

The second option is to add a glycopeptide if the initial regimen does not include one. However, two recent studies that addressed this issue showed that empirical addition of teicoplanin89 or vancomycin90 to the initial regimen in persistently febrile neutropenic patients without documented catheter-related, skin, or soft tissue infections; lung infiltrates; or septic shock was of no benefit. Therefore, the addition of a glycopeptide should be considered only if there is clinical suspicion or microbiologic documentation of infection due to gram-positive organisms that are resistant to the initial antibiotics. Administration of vancomycin either as part of the initial empirical regimen or as a modification should be discontinued after 3 days of treatment if there is no clinical improvement or isolation of a pathogen that is susceptible to vancomycin to avoid development of vancomycin-resistant, gram-positive organisms.69 The improper use of glycopeptides also may delay early antifungal therapy in patients who are at high risk for “primary” fungal sepsis, especially in patients with leukemia.

Empirical antifungal treatment

Fungal infections account for 2–10% of initial microbiologically confirmed infections in febrile neutropenic patients with cancer.9 The proportion of documented fungal infections increases up to 30% in these patients as neutropenia persists.91 In autopsy studies of neutropenic patients with prolonged fever, 40–69% of patients had evidence of an invasive fungal infection.92 This high incidence, along with the high mortality rate associated with fungal infections (approaching 90% in selected patient groups); the difficulty in making a reliable, timely diagnosis; and the lack of clinical signs and symptoms at the initial stages of the infection until the neutropenia resolves,93 led to the introduction of empirical antifungal therapy in the 1980s.87 The IDSA guidelines recommend introduction of an antifungal drug for patients who remain febrile for ≥ 5 days despite undergoing an appropriate initial regimen and for whom resolution of neutropenia is not imminent. Some authors prefer to institute antifungal therapy earlier (Day 3) or even include antifungals up-front in high-risk patients94; because, in patients who have an initial infection that is fungal, delays in the administration of antifungal therapy may be detrimental. Regardless, physicians should make individualized decisions regarding the perceived risk of fungal sepsis and the timing and type of antifungals to be administered. For example, in a patient with leukemia and a history of invasive pulmonary aspergillosis, the threshold for starting antifungal therapy should be very low, early on in the setting of febrile neutropenia.

To our knowledge, the optimal duration of empirical antifungal treatment has not been established to date. If neutropenia has resolved and the patient is well clinically, then antifungal administration can be discontinued. If neutropenia persists despite defervescence, and if a work-up shows no suspicious lesions, then antifungal administration can be stopped after 2 weeks. In hemodynamically unstable patients who have persistent fever and neutropenia, antifungal treatment should be continued until resolution of both fever and neutropenia.

In recent years, the introduction of effective and less toxic antifungal drugs, such as voriconazole and the echinocandins (caspofungin), has provided suitable alternatives to conventional and lipid formulations of AMB as empirical antifungal therapy in high-risk febrile patients with persistent and profound neutropenia.45 Which of these agents is most effective overall is beyond the scope of this review given the complexities and methodological limitations in the design of modern empirical antifungal therapy trials. In the future, with the introduction of nonculture-based diagnostic methods, such as polymerase chain reaction and antigen detection, and with increasing emphasis on risk stratification,45 empirical antifungal therapy likely will be changed to preemptive antifungal therapy.93


Despite the progress that has been made in the management of febrile neutropenic patients with cancer over the past 3 decades and the subsequent reduction in mortality associated with it, major challenges remain. The objectives for the future include the development of new, effective antimicrobials for the emerging resistant pathogens; the refinement of the existing models of risk stratification to reliably identify low-risk patients; the development of algorithms for safe ambulatory treatment with oral antibiotics in selected patients; and the introduction of new, nonculture-based modalities for early detection of infections (especially fungal infections) and, thus, the replacement of empirical therapy with pathogen-specific, preemptive therapy.