David Dockrell MD
Linda L. Lewis MD
Essentials of Diagnosis
The extent and duration of neutropenia are critically important in determining the risk of infection in the patient with febrile neutropenia. For example, patients who undergo bone marrow transplantation are at particular risk because of the extent and duration of neutropenia. Similarly, chemotherapeutic agents such as cytosine arabinoside, which induce prolonged neutropenia, are associated with a greater risk of infection.
Alterations to other aspects of host immunity are also critical. Bone marrow transplantation induces neutropenia as well as defects in T-cell and B-cell function, altering the spectrum of pathogens causing febrile episodes. The administration of glucocorticoids modifies the existing immunodeficiency in patients with cancer or other diseases. The underlying diagnosis may modify the immunosuppression as may occur in patients with Hodgkin's disease who may have underlying T-cell dysfunction, and in patients with HIV-associated neutropenia.
Disruption of the host's normal mechanical barriers further facilitates infection. Indwelling intravenous catheters are a common source of infectious complications. Mucositis induced by chemotherapy allows bacteremia with the normal flora of the mouth and gut. Patients treated with prolonged antibacterial therapy become colonized with fungi, and mucositis allows these organisms to cause invasive disease. This may explain the observation that patients who are neutropenic because of chemotherapy are at greater risk for invasive mycoses than patients with HIV-associated neutropenia. Some diseases may be associated with a lower incidence of infection or an increased susceptibility to particular infecting organisms while the patient is neutropenic. For example, patients with aplastic anemia may have a low risk of bacterial infection but are at particular risk of Aspergillus infections.
Staphylococcal, streptococcal, and enterococcal infections are increasing in frequency. Management of these infections is complicated by increasing methicillin resistance among staphylococci and multidrug resistance among enterococci. Streptococcal infections, including infections with S mitis and viridans group streptococci, are associated with oral mucositis and with ciprofloxacin prophylaxis. Fungal infections are increasing at many centers and include C albicans, non-albicans Candida spp. such as C tropicalis, C kruzei, and Torulopsis glabrata, and filamentous fungi (Aspergillus spp., Mucor, Fusarium, and Pseudoallescheria boydii). Filamentous fungi are increasingly recognized in association with prolonged neutropenia resulting in respiratory tract or disseminated infection.
Gram-negative infections are still encountered, but the spectrum of pathogens has changed. P aeruginosa, once a frequent pathogen in patients with neutropenic fever, has dramatically decreased in incidence for unclear reasons and has been reported to occur in just 1% of cases enrolled in National Institutes of Health treatment protocols in neutropenic populations. It remains a significant pathogen in some subpopulations, such as children with HIV infection. Escherichia coli and K pneumoniae are the most frequently isolated gram-negative bacteria, but the emergence of antibiotic-resistant Enterobacter spp., Serratia marcescens, Stenotrophomonas maltophilia, or Acinetobacter spp. has become an increasing problem. Other pathogens encountered include Corynebacterium jeikeium and Bacillus spp. that cause intravenous catheter-associated infections; rapidly growing mycobacteria that cause exit-site or tunnel infections associated with intravenous catheters; Legionella spp. that cause nosocomial pneumonia; infection with Nocardia spp.; P carinii pneumonia; Clostridium septicum septicemia; C difficile colitis; herpes virus infections, particularly cytomegalovirus in bone marrow transplant recipients; histoplasmosis, blastomycosis, or coccidioidomycosis that cause pneumonia or disseminated infection in endemic areas; cryptococcal infections; and, occasionally, parasitic infections such as Strongyloides stercoralis or Toxoplasma gondii.
An identifiable pathogen is documented in ~ 30%–50% of cases of neutropenic fever, which represents
a decrease in the number of documented infections causing neutropenic fever as compared with the number documented 20–30 years ago.
BOX 25-1 Microbiology of Neutropenia in Adults and Children
The absence of neutrophils and, consequently, signs of inflammation creates a unique challenge for the clinician attempting to determine the cause of neutropenic fever. Although up to 90% of patients become febrile during neutropenia, in most cases, fever is the only sign of infection. This necessitates a thorough history with identification of risk factors for particular types of infection. The history of the underlying illness, its therapy, duration of neutropenia, prophylactic antibiotics administered, previous infections causing neutropenic fever and their treatment, travel and exposure history, thorough knowledge of the most frequent pathogens causing neutropenic fever at the institution and their antimicrobial susceptibilities, and inquiry about specific symptoms and review of systems should be carefully ascertained. A thorough history can help direct the physical examination and subsequent laboratory testing (see Table 25-1). In many cases, no specific clues are identified, and empirical treatment will be initiated without identification of a specific infectious syndrome to explain fever.
Table 25-1. Assessment of patients with neutropenia.
Figure 25-1. The foot of a patient who received high-dose chemotherapy for AML and developed neutropenic fever shows multiple nodules. Biopsy and culture of one of these lesions revealed C albicans.
Techniques such as shell vial for virus or polymerase chain reaction analyses have application in special settings, such as the detection of cytomegalovirus infections after bone marrow transplantation. In addition, an aspirate or biopsy should be obtained from any suspected focus of infection identified by exam and diagnostic imaging. For example, a bronchoalveolar lavage should be considered on any patient with respiratory symptoms or abnormal results from chest radiography. An open-lung biopsy may be necessary in patients who have progressive pulmonary disease without previous diagnosis and who have not responded to empiric therapy.
Specimens should be stained and cultured for bacteria, including anaerobes, and for fungi, mycobacteria, and, in selected patients, viruses. Cryptococcal antigen tests on serum or spinal fluid should be performed in selected patients. Fungal serologies for Histoplasma spp., Coccidiodes spp., or Blastomyces spp. should be performed in patients who reside in or have traveled to endemic areas.
Stool samples should be sent for C difficile toxin assay for patients with unexplained diarrhea, as pseudomembranous colitis secondary to C difficile toxin often complicates the use of antimicrobial agents in patients with neutropenic fever. Other causes of diarrhea in febrile neutropenic patients include neutropenic colitis, enteric pathogens, cytomegalovirus, and parasitic infection. Disseminated strongyloidiasis should be suspected in patients who travel to or reside in endemic areas. If the cause of diarrhea is not established quickly by tests such as C difficile toxin assay, stool culture, and examination for parasites, colonoscopy may be necessary to detect cytomegalovirus or neutropenic colitis. Surveillance cultures of the stool are not useful.
Other potential pathogens that present with patchy infiltrates include Legionella pneumophila, Nocardia asteroides (which is either localized or a diffuse miliary infiltrate), rarely mycobacteria, and in endemic areas Histoplasma capsulatum, Coccidioides immitis, or Blastomyces dermatitidis. Diffuse infiltrates can be seen in infections caused by Legionella spp., Pneumocystis carinii, viruses (especially cytomegalovirus), and parasitic infection.
Computed tomography or magnetic resonance scanning may be useful if head and neck or abdominal/perirectal infection is suspected. The diagnostic utility of nuclear imaging in the evaluation of neutropenic fever has not been established. Abdominal pain and fever associated with an elevation of hepatic enzymes coinciding with the recovery of the neutrophil count is suggestive of hepatosplenic candidiasis (Figure 25-3). Computed tomography or ultrasound of the liver and spleen reveals bull's eye lesions, and diagnosis is confirmed by finding pseudohyphae and granulomata upon computed tomography or ultrasound-guided needle biopsy.
Although only 30–50% of patients with neutropenic fever have documented infections, it is assumed that many of the remainder have occult infection. However, there may be other causes of a febrile episode or of a particular symptom or sign attributed to infection. Numerous tumors can give rise to fever. This has been noted in Hodgkin's lymphoma, non-Hodgkin's lymphoma, and other solid tumors characterized by necrosis. Drugs, including numerous antibiotics used in the management of neutropenic fever, have been reported to cause fever, but this occurs infrequently. Although chemotherapeutic agents like cytosine arabinoside, bleomycin, and 2-CDA are associated with fever, the fever is usually transient and occurs long before the subsequent development of neutropenia. Use of G-CSF or GM-CSF may also cause fever. Drugs that may mask fever, such as corticosteroids, acetaminophen, and nonsteroidal anti-inflammatory drugs, may also be considered when evaluating neutropenic fever. The administration of blood products and, especially, amphotericin B is a common cause of transient increases in temperature associated with the infusion and should be considered in the differential diagnosis. Finally, patients with cancer are at increased risk of deep venous thrombosis and subsequent pulmonary emboli, which are associated with hypoxia and fever.
Figure 25-2. (A) Computed tomography of the chest of a child with neutropenia, persistent fever, and cough revealing a dense, nodular infiltrate in the left perihilar area. (B) Computed tomography of the head of a child with neutropenia, persistent fever, and bloody nasal discharge revealing completely opacified ethmoid and sphenoid sinuses. Surgical exploration revealed pansinusitis with multiple areas of black eschar. Cultured material in both cases grew Aspergillus species.
Figure 25-3. Abdominal computed tomography with contrast in a child with prolonged neutropenia, fever, and abdominal pain. Multiple lucent lesions scattered throughout the liver and spleen are typical of hepatosplenic candidiasis.
The signs and symptoms associated with neutropenic fever may have alternative explanations. Chest radiograph abnormalities and hypoxemia may be caused by noninfectious conditions such as tumor, pulmonary complications of chemotherapeutic agents such as bleomycin or cyclophosphamide, pulmonary emboli, pulmonary edema, radiotherapy, transfusion-related acute lung injury, pulmonary hemorrhage, or transient hypoxemia associated with neutrophil recovery and migration into the lung associated with GM-CSF therapy. Diarrhea may be the result of chemotherapy or use of antibiotics. Skin rashes can be caused by drugs. A rare cause of fever with nodular skin rash, sometimes including vesicles, pustules, or bullae, is Sweet's syndrome (acute febrile neutrophilic dermatosis), which may be associated with acute myelogenous leukemia.
Superinfection is a recognized complication of febrile neutropenia in patients receiving antimicrobial agents. It is usually defined as the development of documented infection during or < 1 week after the discontinuation of antimicrobial therapy. Superinfection occurs in ≤ 20% of cancer patients with neutropenic fever. Fungi account for 25–67% of superinfections. Risk factors for superinfection include a longer duration of profound neutropenia (< 100 neutrophils/mm3), persistence of fever after 3 days of therapy, and presence of a central venous catheter.
The initiation of empirical antibiotic therapy for any febrile neutropenic patient with a single oral temperature elevation > 38.5°C or three or more oral temperatures recorded at > 38.0°C within 24 h has become the standard of care. Even though many patients have no demonstrable infection, it is assumed that most have occult infection. The rapidity with which these patients' clinical status can deteriorate during infectious episodes and the demonstration that empirical therapy decreases mortality have supported the use of empirical therapy for neutropenic fever.
The selection of empirical therapy should be influenced by the epidemiological and pathogenic concerns mentioned. Its design should consider the predominant microorganisms encountered at a given center, their expected susceptibility patterns, and the clinical features and laboratory test results.
BOX 25-2 Empirical Therapy for Neutropenia
The initial empiric regimen should include an antimicrobial agent or combination of drugs with a broad spectrum of coverage against gram-positive and gram-negative bacteria. This regimen should be modified if fever persists or recurs while on treatment. Initial anaerobic coverage is not usually required but should be included in cases with marked oral mucositis or documented bowel or perirectal involvement. Initial empiric antifungal therapy is not necessary but should be added for treatment failure or in cases with signs or symptoms suggestive of fungal infection in association with prolonged neutropenia.
The initial regimen is intended to treat the cause of fever, as well as to prevent subsequent episodes. The selection of an empiric regimen should consider its efficacy as well as its cost, toxicity, and potential to increase antibiotic resistance and superinfection. Because of the emergence of vancomycin-resistant enterococci, many clinicians no longer include vancomycin in initial empiric therapeutic regimens.
Box 25-2 outlines possible empirical regimens. A β-lactam, either alone or in combination with an aminoglycoside or a fluoroquinolone, is often used.
The β-lactam used most often is ceftazidime because of its activity against P aeruginosa and other such gram-negative bacteria as the Enterobacteriaceae. The new fourth-generation cephalosporin, cefepime, may be useful, because of its ability to remain bactericidal against Enterobacteriaceae that have developed resistance to ceftazidime as well as its improved gram-positive coverage compared with that of ceftazidime. The choice of an aminoglycoside (gentamicin, tobramycin, or amikacin) is dictated by local resistance patterns and cost. The administration of two β-lactam antibiotics concurrently should be avoided.
An acceptable alternative to combination therapy is monotherapy with ceftazidime, cefepime, piperacillin-tazobactam, or a carbapenem such as imipenem-cilastin or meropenem. The main concern with ceftazidime, but not cefepime, monotherapy is its poor gram-positive coverage and the emergence of resistance in some gram-negative organisms. For these reasons, centers with high rates of gram-positive infections or infections by Serratia marcescens, Enterobacter spp., or Citrobacter freundii should avoid ceftazidime monotherapy.
Clinical trials have shown imipenem-cilastin or meropenem to be equivalent to ceftazidime-containing regimens. Carbapenems are more active against gram-positive and most gram-negative microorganisms, compared with ceftazidime. Imipenem-cilastin therapy may cause seizures in some patients, whereas meropenem has a low risk of associated seizures. Piperacillin-tazobactam monotherapy is also effective empiric therapy for febrile neutropenia patients. Like cefepime and the carbapenems, piperacillin-tazobactam has a broad spectrum of activity against gram-positive and gram-negative microorganisms. Among these regimens, only the carbapenems and piperacillin-tazobactam have activity against anaerobes. If the clinical suspicion of anaerobic infection is high, metronidazole should be added to therapy with either ceftazidime or cefepime. In β-lactam–allergic patients, a combination of ciprofloxacin or aztreonam, together with either clindamycin or vancomycin is appropriate.
Apart from its use in some penicillin-allergic patients and in centers with high incidences of coagulase-negative staphylococci or methicillin-resistant S aureus, vancomycin is usually not included in the initial therapeutic regimen because of concerns about the emergence of vancomycin resistance. Studies conducted in the early 1980s suggested that empiric therapy with vancomycin-containing regimens was more effective than therapy that did not include vancomycin. More recent studies showed that patients are not adversely affected by a policy of withholding vancomycin use until there is clinical or microbiological evidence that it is required. Vancomycin therapy may be added in patients with an indwelling intravenous catheter who remain febrile after 3 days of empirical treatment with no other obvious source of infection.
Some patients with febrile neutropenia have a low mortality risk (3%) from infection. Their profiles include the development of neutropenic fever while a clinically stable outpatient; the presence of a tumor responsive to chemotherapy; and the absence of any current comorbidity. These patients account for ~ 40% of patients with febrile neutropenia. Strategies being evaluated for the management of these patients include (a) 48 h of intravenous antibiotics followed by oral antibiotics, (b) parenteral antibiotics administered as an outpatient, or (c) oral antibiotics. A recent randomized clinical trial conducted in carefully selected low-risk patients confirmed that oral antibiotics are as effective as intravenous antibiotics in the hospital setting. Whether this conclusion is generalizable to the outpatient population remains to be determined.
Ciprofloxacin has a special place in the management of low-risk adult patients as described above. While it can be used parenterally, its special advantage is that it is an effective oral agent against P aeruginosa and other gram-negative bacteria. Because of its limited gram-positive coverage, however, it should be combined with either amoxicillin-clavulanate or clindamycin. Alternatively, the improved gram-positive spectrum of levofloxacin may make it useful in this setting.
Approximately 66% of patients can be maintained on their empirical regimen throughout the period of neutropenia. The remainder may require modification because of the results of cultures or other tests or because of recurrent or persistent fever.
If microbiological tests identify a specific organism, empiric therapy may be modified if necessary; however, patients should continue to receive broad-spectrum antibiotic coverage throughout the period of neutropenia. The isolation of a gram-positive organism before or during therapy may require the addition of vancomycin or anti-staphylococcal penicillin. If a gram-negative organism is isolated before the initiation of therapy in a patient who has subsequently defervesced, the initial regimen may not need to be modified. However, if a gram-negative organism is isolated from a sample collected after treatment has been initiated, antimicrobial resistance or superinfection should be suspected, and modification of the gram-negative coverage is warranted.
In patients who remain febrile after 5 days of empiric antimicrobial therapy, antifungal therapy should be added. Fungal infections are associated with prolonged periods of neutropenia. Fungi are difficult to recover ante mortem but are a frequent finding at post mortem in patients who died with neutropenic fever. Once fungal infections disseminate, there is a high associated mortality. Amphotericin B (0.5–0.6 mg/kg/d) is the empiric therapy of choice. Amphotericin-lipid complex formulations have lower nephrotoxicity than amphotericin but have not been shown to be more effective and are much more expensive.
Although fluconazole or itraconazole may be equivalent to amphotericin B in managing non-neutropenic patients with candidemia, the role of empiric imidazole therapy in febrile neutropenic patients is not established. Fluconazole resistance is increasing among Candida spp.; C kruzei, C tropicalis, Torulopsis glabrata, and non-albicans species are usually resistant. Aspergillus infections usually require higher daily doses of amphotericin B (1.0–1.5 mg/kg/d). Hepatosplenic candidiasis usually requires prolonged amphotericin B treatment with total doses of 3–5 g. Some patients receiving empirical amphotericin may develop breakthrough candidiasis. This may be owing to infection with resistant strains of Candida or, more commonly, by infection associated with intravascular catheters.
Broad-spectrum antimicrobial therapy should be continued until after recovery from neutropenia. Such therapy results in less relapse of fever than treatment for a 7- to 10-day period. If the period of neutropenia ends shortly after the initiation of antibiotics, therapy should continue for a 7- to 10-day course.
Intravascular catheter infections are usually treated with antibiotics rotated through each port or lumen. As many of the organisms causing these infections are indolent, initial attempts are made to treat without removing the catheter. A catheter should be removed if (a) blood cultures remain positive 48 h after the commencement of antibiotics, (b) Candida or Bacillus infections are present, or (c) exit-site infections, caused by A flavus or mycobacteria, or tunnel infections are present (indicated by tenderness or induration along the subcutaneous track of the catheter).
The hematopoietic growth factors G-CSF and GM-CSF are emerging as important adjuvants to the management of neutropenic fever. By increasing the proliferation and differentiation of bone marrow progenitor cells, they help restore neutrophil and macrophage function. These agents have been demonstrated to reduce the period of neutropenia, shorten the period of hospitalization, and decrease the incidence of bacteremia. However, they have not been consistently demonstrated to decrease overall mortality. These agents should be considered in patients whose expected duration of neutropenia is > 1 week. Side effects of these agents include fever, rash, hypoxia, bone pain, and fluid retention. Various cytokines are being investigated as adjuvants to antimicrobial therapy in the management of neutropenic fever.
The outcome of neutropenic fever can be defined in terms of microbiologic response and clinical outcome and is related to recovery from neutropenia. Among patients with documented bacterial infections, treatment with ceftazidime or cefepime monotherapy, cephalosporin-aminoglycoside combinations, piperacillin-tazobactam, or carbapenem monotherapy have been associated with microbiologic cure rates > 90%. Fungal microbiological response rates are much lower.
Clinical response rates are often defined in terms of resolution of fever after 4 days of treatment, need to alter initial empirical antimicrobial agents, and survival. Overall β-lactam monotherapy has been associated with resolution of fever after 4 days of therapy in ~ 60%, need for subsequent antimicrobial-agent modification in ~ 60%, and overall survival of 90–98%. The mortality rate associated with documented bacterial causes of neutropenic fever has fallen from 90% in the 1950s to <10% in the 1990s.
Fungal and vancomycin-resistant enterococcal infections are associated with a higher mortality—> 50%. The incidence of superinfection is between 20% and 25%, and their mortality is > 50%. The ultimate prognosis for most patients is closely linked to their underlying malignancy. For example, in an analysis of hepatosplenic candidiasis, the prognosis of the underlying leukemia was the most important determinant of outcome.
Prevention & Control
Several strategies have been used to decrease the incidence of neutropenic fever. Trials of antibiotic prophylaxis have had mixed results. Initial trials using trimethoprim-sulfamethoxazole showed some promise, but the emergence of resistance limited its use. In trials, ciprofloxacin therapy reduced the risk of gram-negative infections but not gram-positive infections and did not decrease fever-related morbidity or infection-related mortality. Concerns about the emergence of resistance to this agent have led many authorities to recommend that ciprofloxacin prophylaxis be limited to groups whose length of neutropenia places them at a particularly high risk of infectious complication (eg, bone marrow transplant patients).
Antifungal prophylaxis with fluconazole decreases the incidence of infections by susceptible Candida spp. but is also associated with the emergence of resistant Candida spp. Oral bowel decontamination regimens have been studied, but they are not palatable and are minimally effective. Pneumocystis cariniiprophylaxis with trimethoprim-sulfamethoxazole is effective and recommended for febrile-neutropenia patients receiving high doses of glucocorticoids.
Environmental measures should be taken to lessen the incidence of infection in patients who are neutropenic. Simple isolation with careful hand washing has been demonstrated to be as efficacious as and more practical than reverse isolation. The use of a total protective environment with constant positive airflow and vigorous surface decontamination does not merit the extra expense. Despite decreasing the number of infections, no clear survival benefit has been proven.
Some authorities recommend a cooked diet to avoid the acquisition of organisms in uncooked food. Spices such as freshly ground black pepper are best avoided because of potential contamination by fungal spores. Similarly, potted plants should not be part of the patient's environment because of associated fungal spores and bacteria found in the soil. Environmental sources contribute to fungal colonization. Construction work in or around a hospital will significantly increase the number of Aspergillus spores in the air. In centers where this is a problem, high-efficiency particulate air filters may be helpful. Similarly, water purification systems will decrease the risk of nosocomial Legionella infections.
Freifeld A et al: A double-blind comparison of empirical oral and intravenous antibiotic therapy for low-risk febrile patients with neutropenia during cancer chemotherapy. N Engl J Med 1999;341:305.
Hathorn JW, Lyke K: Empirical treatment of febrile neutropenia: evolution of current therapeutic approaches. Clin Infect Dis 1997;24:(Suppl 2)S256.
Hughes WT, et al: 1997 guidelines for the use of antimicrobial agents in neutropenic patients with unexplained fever. Clin Infect Dis 1997;25:551.
Lee JW, Pizzo PA: Management of the cancer patient with fever and prolonged neutropenia. Hematol Oncol Clin North Am 1993;7:937.
Pizzo PA: Fever in immunocompromised patients. N Engl J Med 1999;341:893.
Ramphal R et al: Clinical experience with single agent and combination regimens in the management of infection in the febrile neutropenic patient. Am J Med 1996;100:(Suppl 6A)83S.
Rolston KVI, Rubenstein EB, Freifeld A: Early empiric antibiotic therapy for febrile neutropenia patients at low risk. Infect Dis Clin North Am 1996;10:223.