FEVER IN THE GRANULOCYTOPENIC PATIENT
Granulocytopenia is associated with conditions as diverse as hematologic malignancy and its treatment, adverse reactions to medications, selected infections, and hereditary disorders. Although fever can be associated with any of these, it is most feared when complicating leukemia and other hematologic malignancies.
Granulocytopenia per se is associated with infections caused by Staphylococcus aureus, selected enteric gram-negative bacilli, (especially Escherichia coli and Klebsiella species), Pseudomonas aeruginosa, Candida species, and Aspergillus species. Hairy-cell leukemia, indwelling Foley or IV catheters, and chemotherapy are associated with risks for other pathogens. In the 1990s, such risks have especially been noted with indwelling IV lines and have resulted in a major increase in the number of gram-positive infections identified. These may be associated with coagulase-negative staphylococci, diphtheroids, and viridans streptococci. Often, these organisms are resistant to commonly employed b-lactam antibiotics. On occasion, fever can be explained by noninfectious processes (underlying malignancy, recent administration of pyrogens, or drug fever), but often it is a consequence of infections that not only are associated with considerable morbidity in the leukemic patient but are also the most important cause of death.
Although at least 40% of patients with fever and neutropenia demonstrate no source of infection, they may clinically improve with antibiotics. The likelihood of response to initial antibiotics reaches 95% for those with neutropenia of less than 7 days’ duration, but it is only 32% in those with fever and neutropenia for longer than 2 weeks. The risk for infection varies directly with both duration of granulocytopenia and rate of decline, and inversely with the absolute granulocyte count. With regard to acute leukemia, infections are noted more commonly during relapse and when granulocyte counts are below 100/mm3.
Fever above 101°F for more than 2 hours that is not associated with the administration of known pyrogens represents infection until proved otherwise. The diagnosis of infection in the granulocytopenic host is made difficult by subtleties of signs and symptoms. Frank pus is rarely encountered because polymorphonuclear leukocytes are necessary for its production. Host responses to infection may be blunted, but pain and fever are usually preserved. Thus, the afebrile patient is unlikely to be infected. Patients with pharyngitis may have pain and erythema without exudate. Skin and anorectal infections demonstrate erythema and local pain but usually lack prominent local heat, swelling, exudate, fluctuation, or regional adenopathy. Urinary tract infections often occur in the absence of irritative voiding symptoms (dysuria, frequency, urgency) and pyuria. Classic hallmarks of pneumonia (cough, sputum production, rales, and clinical consolidation) are frequently not observed. As a general rule, fever attributable to leukemia, drugs, and other noninfectious causes is usually not associated with rigors or hypotension. Alternatively, shaking chills or a “toxic” appearance indicates probable infection, possibly bacteremia.
Evaluation begins with a comprehensive history and physical examination. The history can provide evidence of localized discomfort. Sore throat associated with fever and pharyngeal ulceration caused by antineoplastic drugs may be early signs of bacteremia resulting from P. aeruginosa or infection with herpes simplex. Dysphagia suggests esophagitis caused by Candida species, herpes simplex, or other potential pathogens. Painful defecation can alert the physician to perirectal cellulitis or phlegmon, often involving P. aeruginosa and anaerobes. Fever, vomiting, abdominal pain, and abdominal tenderness may be symptoms of typhlitis (inflammation of the cecum) or pseudomembranous colitis from either underlying leukemia or antibiotics.
Physical examination should be meticulous and repeated often. Special attention should be directed toward painful or erythematous skin and anorectal lesions, pharyngeal erythema, periodontal inflammation, sinus tenderness, and rales. Skin lesions of ecthyma gangrenosum (classically, necrotic centers with surrounding areolae, but having many variants) indicate probable gram-negative bacteria, often P. aeruginosa. Such lesions provide sources for biopsy and culture, often yielding rapid bacteriologic information. Periodontal infection is manifested by local tenderness and fever accompanied by minimal signs and symptoms of inflammation. Careful attention should be paid to indwelling lines. Redness and swelling may be the only indications of infection. Up to 50% of line-related bacteremias may present without localized clinical evidence of infection.
Urine cultures, several sets of blood cultures, and chest roentgenography are indicated in the initial assessment. However, because of granulocytopenia, Gram’s stains (of urine, sputum, other specimens) may fail to demonstrate polymorphonuclear leukocytes. With pneumonia, radiographic evidence is usually present, although findings may be subtle. The presence of lung necrosis generally indicates pneumonia associated with gram-negative enteric bacilli, S. aureus, or Aspergillus or Mucor. Other cultures and x-ray imaging should be obtained as clinically indicated. Routine surveillance cultures of nasal secretions, stool, and other specimens are not generally recommended.
The bacteriology of infections in the granulocytopenic patient includes facultative gram-positive and gram-negative organisms. Anaerobes are uncommonly implicated, except in anorectal infections, where they predominate and require specific therapy. Gram-negative pathogens commonly demonstrated include E. coli, Klebsiella species, Enterobacter, and P. aeruginosa. The frequent use of indwelling venous access devices has been associated with a resurgence of gram-positive infections caused by S. aureus, Staphylococcus epidermidis, streptococci, and occasionally others. Fungal infections are usually noted in patients who are maintained for a long time on broad-spectrum antimicrobials and have prolonged neutropenia. Although Aspergillus and Candida species have been classically implicated, recent studies demonstrate the possibility of infection with diverse organisms that include Trichosporon beigelii, Fusarium species, Geotrichum candidum, and Pseudallescheria boydii. Infection with these more unusual species is often associated with sinusitis, deep-organ involvement, or fungemia. Risk factors for adverse outcome include prolonged neutropenia and organ involvement.
Selected neutropenic patients with fever may be safely managed as outpatients. This concept has been promulgated in part because of the availability of oral antipseudomonal antibiotics, and the availability of an infrastructure that allows for IV therapy outside the hospital. Table 53-1 summarizes the risk stratification of patients with fever and neutropenia. Table 53-2 summarizes the indications for outpatient therapy of febrile neutropenic patients. Forty percent to 60% of patients fall into the category of those who can be safely treated outside the hospital. Patients at low risk for adverse outcome from fever and leukopenia generally do not have uncontrolled cancer and concurrent associated problems requiring hospitalization. In the absence of these features, complications were noted in only 2% of patients, and mortality approached zero. On the basis of similar parameters, patients may be safely discharged to home to complete therapy with either IV or oral antibiotics. All patients receiving outpatient therapy should live within a reasonable distance of the medical center where treatment is being administered. Some centers require this to be no more than 30 miles. Daily follow-up by telephone is indicated, and the patients must have accessibility to a health care provider.
Table 53-1. Risk stratification for febrile, neutropenic patients
Table 53-2. Patients with fever and neutropenia: candidates for outpatient therapy
Treatment of the febrile neutropenic outpatient may be with either oral or parenteral antibiotics. Fluoroquinolones (generally ciprofloxacin or ofloxacin) have been the primary oral antibiotics studied, either alone or in conjunction with either clindamycin or amoxicillin-clavulanate. Response rates of approximately 90% have been noted in many studies after appropriate cultures have been obtained. The optimal antimicrobial therapy is unknown.
A recent retrospective investigation demonstrated that the average patient receives approximately five agents per course, which entails greatly excessive use (based on bacteriologic and clinical indications of infection) and expense and enhances the potential for renal and hepatic toxicity. Selection of an initial antimicrobial therapy should be based on such factors as local pathogens and their antimicrobial susceptibilities, allergy and organ failure, and adjunctive medications. Initial therapy should consist of broad-spectrum bactericidal agents. Controversy continues regarding combination therapy versus monotherapy. Numerous studies now demonstrate the efficacy of broad-spectrum antibiotics (imipenemcilastatin, meropenem, ceftazidime, cefepime) as monotherapy in the uncomplicated patient. However, many centers continue to recommend combinations that include double b-lactams (e.g., ceftazidime plus piperacillin) or b-lactams plus aminoglycosides (e.g., piperacillin or ceftazidime plus gentamicin, tobramycin, or amikacin). If aminoglycosides are employed, a loading dose of more than 2 mg/kg should be used to optimize initial levels. Use of 5-7 mg/kg as a single daily dose should be strongly considered. When P. aeruginosa is strongly considered (e.g., based on epidemiology, presence of ecthyma gangrenosum), two effective antipseudomonal agents are always indicated.
The need for vancomycin in the initial regimen remains controversial. Initial data strongly suggested it was not needed. However, this issue is undergoing reassessment as the role of serious gram-positive infection increases (primarily because of the use of long-standing invasive IV devices in many patients) and methicillin-resistant staphylococci become more frequently encountered. Vancomycin should be employed in the initial antibiotic regimen at institutions where fulminant gram-positive infections have been observed. It may be used when the likelihood of catheter-related infection is high, in patients in whom fever and neutropenia develop while they are on quinolone prophylaxis, in patients known to be colonized with methicillin-resistant S. aureus or highly penicillin-resistant Streptococcus pneumoniae, or in critically ill patients when failure to cover a pathogen could lead to rapid death.
After initial treatment, 3 days are generally needed to define etiology and response to therapy. Further management strategies are guided by these parameters.
Vancomycin should be discontinued from an initial regimen if cultures fail to substantiate its need. If clinical improvement is noted and cultures define a pathogen, therapy should be optimized, but broad-spectrum therapy should be continued. If the patient improves in the absence of a defined organism, therapy should be maintained for at least a week. Stable, compliant patients who can be monitored as outpatients can be switched to oral antibiotics (quinolone, cefixime).
Failure to improve within 3 to 5 days requires meticulous reassessment of the patient and usually an antimicrobial change. Occult sites of infection should be sought. Computed tomography scanning of the abdomen with percutaneous sampling of collections is expeditious and reasonably well tolerated. A recent review of the role of radionuclide imaging in immunosuppressed patients points out the benefits and pitfalls of these modalities. Gallium may not be picked up at infection sites in neutropenic patients, and its accumulation in the bowel may make for difficult interpretations. Most radionuclide modalities take several days for complete evaluations, and this limits their usefulness in critically ill patients.
If no site of infection can be identified after 5 to 7 days, empiric addition of amphotericin B is indicated. The maximum dose is 0.5 to 1.0 mg/kg per day. Up to 50% of patients will respond clinically to this treatment, although the source of infection may not be identified. The results of fungal antigen detection tests are not reliable, and blood cultures are positive in fewer than 40% of patients with fungal infections documented by other means. Fluconazole remains a controversial alternative to amphotericin; it could be used with care in patients who are at hospitals where problems with Aspergillus and Mucor are limited, have not received prior fluconazole prophylaxis, and are unlikely to tolerate amphotericin B. However, a recent pilot study comparing fluconazole with amphotericin B in patients with antibiotic-resistant neutropenic fever demonstrated a poorer response with fluconazole and a significantly higher tendency toward the development of invasive fungal disease. Several forms of liposomal amphotericin B have been approved for use in the United States. All have the advantage of allowing higher dosing with enhanced safety—typically up to 5 mg/kg versus 1 mg/kg with amphotericin B. Their major role appears to be in the management of mold infections (Asper-gillus and Mucor) and of patients in whom significant intolerance to amphotericin has developed. Their high price precludes routine use at this time.
For patients who do not clinically respond to initial antimicrobial therapy, current recommendations do not include careful discontinuation of therapy within the first week. However, some would cautiously curtail antimicrobials if the patient is stable and without clinical infection. With the addition of amphotericin B, broad-spectrum antimicrobials should be modified to delete aminoglycosides to decrease the likelihood of nephrotoxicity. Antiviral therapy with acyclovir (herpes simplex, herpes zoster) or ganciclovir (cytomegalovirus) is indicated for proven or suspected infections but is not recommended empirically.
Antimicrobial chemotherapy is generally continued until granulocyte counts have risen above 500 to 1,000/mm3. Alternatively, treatment may be carefully terminated in selected patients who are clinically stable and without signs of infection.
Colony-stimulating factors, such as granulocyte colony-stimulating factor (G-CSF) and granulocyte-macrophage colony-stimulating factor (GM-CSF), have been employed to shorten the duration of neutropenia following chemotherapy for solid tumors and hematologic malignancies. Several studies have now demonstrated decreased antimicrobial needs, decreased hospital stays, and shortened neutropenic periods when colony-stimulating factors are employed. However, mortality related to infection has not been shown to be reduced. Additionally, most patients respond favorably without them, and they are expensive. Colony-stimulating factors should be considered in selected patients with severe infection, a poor likelihood of rapid marrow recovery, and a poor response to appropriate antimicrobials.
Bacteremia associated with indwelling venous access devices can often be treated without removal of the device. Indications for removal include “tunnel” infections, infections caused by fungi or Corynebacterium jeikeium, failure to respond, or clinical relapse.
Focal hepatic candidiasis usually presents with ongoing fever in the face of resolving neutropenia. Abdominal pain, nausea, and diarrhea occasionally may be noted. Focal hepatic candidiasis often presents as a manifestation of more widespread disease. Striking elevations of the serum alkaline phosphatase have been noted. Hepatic defects are sometimes detected on computed tomography or ultrasonography, but the yield is higher without neutropenia. Liver biopsy confirms the diagnosis, but cultures may be negative if prior antifungal therapy has been rendered. Prolonged intensive parenteral therapy with amphotericin B (total dose >2 g) is the preferred treatment. Several recent studies with fluconazole demonstrate its utility for this syndrome. Dosages have been 100 to 400 mg daily for a median of 30 weeks. Fluconazole has also been used for patients who have failed amphotericin B therapy or have demonstrated intolerable side effects from that agent. (R.B.B.)
Crawford J, et al. Reduction by granulocyte colony-stimulating factor of fever and neutropenia induced by chemotherapy in patients with small-cell lung cancer. N Engl J Med 1991;315:164–170.
This is one of several published studies on the efficacy of colony-stimulating factors in aborting fever and neutropenia in cancer patients who are receiving chemotherapy. This double-blind placebo-controlled investigation demonstrated that G-CSF decreases fever and infection in patients receiving chemotherapy for small-cell carcinoma of the lung. Fever and neutropenia were noted in 77% of controls but in only 40% of those who received G-CSF. Duration of cell counts below 500/mm3 was 6 days (control) versus 1 day (G-CSF). Number of infections, duration of hospitalization, and need for antimicrobials were all significantly decreased in the G-CSF group.
EORTC International Antimicrobial Therapy Cooperative Group and the National Cancer Institute of Canada—Clinical Trials Group. Vancomycin added to empirical combination antibiotic therapy for fever in granulocytopenic cancer patients. J Infect Dis 1991;163:951–958.
Febrile neutropenic patients were randomized to receive ceftazidime plus amikacin with or without vancomycin for empiric therapy. The outcome of patients with gram-positive bacteremias was not improved by the addition of vancomycin, and no patients with gram-positive bacteremia in the absence of vancomycin died within the first 3 days. Patients who received vancomycin had greater nephrotoxicity. The authors do not feel that vancomycin is a required part of an initial regimen but could be safely used later if infection caused by a susceptible pathogen is identified.
Glenn J, et al. Anorectal infections in patients with malignant diseases. Rev Infect Dis 1988;10:42–52.
This is a retrospective review of approximately 60 patients with anorectal infection documented at a tertiary care cancer hospital. The infection itself was associated with death in fewer than 10% of cases. Most cases were associated with a polymicrobial flora that often involved anaerobes. More than 50% of cases were successfully managed with medical therapy alone. Treatment with an aminoglycoside and an antianaerobic agent was usually employed. Indications for surgery included presence of necrotic material, fluctuance, and failure to respond to antimicrobials.
Holleran WM, Wilbur JR, DeGregorio MW. Empiric amphotericin B therapy in patients with acute leukemia. Rev Infect Dis 1985;7:619–624.
This excellent overview of several aspects of fungal infections in acute leukemia addresses risk factors, clinical presentation, and justification for empiric therapy with amphotericin B. Clinical response is often noted. Dosing should maximize at 0.5 mg/kg per day, and treatment is generally discontinued on recovery of granulocytes.
Hughes WT, et al. 1997 Guidelines for the use of antimicrobial agents in neutropenic patients with unexplained fever. Clin Infect Dis 1997;25:551–573.
The Infectious Diseases Society of America published this first in a series of guidelines to aid clinicians in decision making. This document addresses most of the important issues in the management of the febrile neutropenic patient and provides excellent treatment strategies and algorithms that are easy to follow. It also provides alternatives based on the literature. This is an excellent single reference on the subject.
Malik IA, et al. Feasibility of outpatient management of fever in cancer patients with low-risk neutropenia: results of a prospective randomized trial. Am J Med 1995;98: 224–231.
One hundred subjects were inpatients or outpatients. Results demonstrated equal efficacy statistically; however, 21% of the outpatients required hospitalization. Carefully chosen patients with fever and leukopenia can be safely managed outside the hospital. The best antibiotic is not determined. Use of a quinolone should be reserved for those who have not received these agents as prophylaxis.
Powderly WG, et al. Amphotericin B-resistant yeast infection in severely immunocompromised patients. Am J Med 1988;84:826–832.
Recent isolates of Candida species from immunocompromised patients may be more resistant to amphotericin B than similar isolates from patients who are immunocompetent. When resistance is encountered, the outcome is dismal. These findings stress the need for better sensitivity testing of antifungal agents and the availability of suitable alternatives to amphotericin B.
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 (6A):83S–89S.
Cefepime is a fourth-generation cephalosporin similar to ceftazidime but with enhanced stability against selected enteric bacilli and better gram-positive activity. Febrile neutropenic patients were randomized to receive either cefepime as monotherapy or combination antibiotic therapy. Bacterial infections were documented in 40% of patients in both groups. Outcomes as measured by fever resolution, need for alternative therapies, and survival were similar in both groups. This is another study demonstrating that for most febrile neutropenic patients, monotherapy is sufficient. There are now several appropriate agents from which to choose. I would still employ combination therapy for patients at high risk for infection with P. aeruginosa (i.e., those with ecthyma gangrenosum lesions, known positive blood cultures), but in most other instances monotherapy will suffice.
Rolston KV, Rubenstein EB, Freifeld A. Early empiric antibiotic therapy for febrile neutropenia patients at low risk. Infect Dis Clin North Am 1996;10:223–237.
The authors present an excellent summary of the indications and choices for antibiotic therapy in the low-risk patient. Options run from therapy without hospitalization, to early switch from IV to oral antibiotics, to use of home IV therapy. Determinants of low- risk patients are provided.
Rubin RH, Fischman AJ. Radionuclide imaging of infection in the immumocompomised host. Clin Infect Dis 1996;22:414–422.
The authors provide a succinct report on the advantages and pitfalls of radionuclide imaging in immunosuppressed patients. Such processes are of limited value in the diagnosis of infections below the diaphragm, and in the presence of neutropenia. They can occasionally be useful in patients with chest infections and may help target further studies in patients with unexplained fevers.
Talcott JA, et al. The medical course of cancer patients with fever and neutropenia. Arch Intern Med 1988;148:2561–2568.
This interesting retrospective analysis of patients with fever and neutropenia attempts to identify populations that may require less intensive care and evaluation. The authors conclude that in many patients complications are unlikely to develop; the risk factors include concurrent morbidities and uncontrolled cancer.