Other laboratory tests should include full blood examination (FBE) with differential white cell count, urea, electrolytes and serum creatinine, liver function. Cultures from other sites should be taken according to clinical indication. A chest X-ray is indicated for patients with respiratory symptoms or signs.
Selection of a first-line antibiotic regimen
In view of emerging evidence regarding efficacy and toxicity differences between empiric treatment regimens, and strong evidence of heterogeneity in clinical practice, the following recommendations were developed to provide clinicians with comprehensive guidance for selecting an appropriate empiric strategy in the setting of neutropenic fever (see Table 2 for summary).
Table 2. Key practice points – empiric therapy
|All patients presenting with fever following chemotherapy should be managed as if they have neutropenic fever and receive empiric antibiotics without waiting for laboratory confirmation of neutrophil count (expert opinion). This management may be modified if neutrophil count and function are adequate.|
|Time to first dose:|
| • Patients with features of systemic compromise (such as hypotension, hypoxia, confusion, major organ dysfunction) should receive antibiotics within 30 min of presentation after the immediate collection of blood cultures and administration of fluid support (grade C recommendation)|
| • Clinically stable patients should receive antibiotics within 1 h of presentation after appropriate cultures have been taken (grade C recommendation)|
| • Administration of antibiotics should not be delayed by the conduct of laboratory or radiological investigations (grade C recommendation)|
|Blood cultures and other tests:|
| • At least two separate blood culture sets should be collected from separate venepuncture sites prior to commencement of antibiotics. At least 20 mL of blood should be drawn from each site and 10 mL inoculated into one aerobic and anaerobic bottle|
| • Collection of blood cultures from vascular catheters in addition to peripheral BC may assist in the diagnosis of catheter-related blood stream infection. Use of differential time to positivity of ≥120 mins is predictive of CRBSI (time from venous catheter to peripheral blood culture become positive)|
| • The administration of empiric therapy should not be delayed in order to perform blood cultures (grade C recommendation)|
| • Repeat blood cultures prior to Day 3 are not recommended for the majority of patients with neutropenic fever unless clinically unstable and/or suspected new infectious foci are present|
|Selection of a first-line antibiotic regimen:|
| • Beta-lactam monotherapy is recommended, unless a life-threatening/immediate hypersensitivity reaction has been documented, for all clinically stable patients (grade A recommendation)|
| • An antipseudomonal beta-lactam antibiotic plus gentamicin once daily adjusted according to levels is recommended for patients with systemic compromise (expert opinion)|
| • Routine addition of vancomycin to the initial empiric regimen within 72 h does not reduce mortality and is associated with increased risk of nephrotoxicity (grade A)|
| • Vancomycin for 48–72 h is recommended for patients in septic shock, and those who are known to be colonized with a resistant Gram-positive bacterium (e.g. MRSA) (expert opinion). There was no consensus whether patients with systemic compromise but not septic shock should receive vancomycin.|
| • Specific patient groups and institution-specific circumstances, such as rural and regional sites, may influence antibiotic selection|
|Modification or Cessation of empiric therapy:|
| • Escalation of antibiotic coverage should not occur prior to day 3–5 in the absence of clinical instability, isolation of a resistant organism or emergence of new infective foci (expert opinion)|
| • It is preferable to stop antibiotic treatment when the neutrophil count recovers to ≥0.5 × 109 cells/L (expert opinion)|
| • If the neutrophil count is <0.5 × 109 cells/L and neutropenia is expected to be prolonged, the decision to discontinue or continue antibiotic therapy should be based upon clinical criteria and individual clinicians' judgement|
The patient's history, allergies, symptoms, signs, recent antibiotic use and culture data, as well as local flora and infection patterns, should guide the initial choice of antibiotic therapy.
Meta-analyses of randomized controlled trials in sepsis have shown that monotherapy with an antipseudomonal beta-lactam (e.g. piperacillin-tazobactam, cefepime, ceftazadime, meropenem) is as efficacious as combination therapy (level I evidence). These studies have shown no clinical advantage of combination therapy over monotherapy, even in subgroups of patients with bacteraemia (including patients with Pseudomonas aeruginosa) or severe neutropenia <0.1 × 109/L (level I evidence).10–12 There is also no evidence that combination regimens prevent the emergence of resistant organisms (level I evidence).10,11 The risk of nephrotoxicity with beta-lactam/aminoglycoside combination therapy usually outweighs any potential benefit (level I evidence).10,11 Therefore, monotherapy regimens are preferred in most circumstances, particularly in elderly patients, and in patients with, or at high risk of, renal impairment.
The choice of piperacillin-tazobactam or cefepime as the institutional first-line monotherapy for neutropenic fever must be supported by local antibiograms data supporting its choice. At times, individual organizations may have clusters of multi-resistant Gram-negative infections that may influence local recommendations,
The following recommendations for initial antibiotic selection apply to specific patient groups (see Table 3 for a summary). Each institution should select its preferred agent/s for each clinical circumstance. Patients with impaired renal function (glomerular filtration rate less than 50 mL/min) will require adjustments to the suggested doses based on calculated creatinine clearance (grade A recommendation). Further caveats to these recommendations are presented in a subsequent section.
Table 3. Initial antibiotic selection (doses for normal renal function).
|Patients without features of systemic compromise |
(beta-lactam monotherapy is recommended unless allergy to the recommended agent/s)
|No penicillin allergy: |
Piperacillin-tazobactam 4.5 g IV six- to eight-hourly (grade A) OR
Cefepime 2 g IV eight-hourly
Other reasonable choice for monotherapy is ceftazidime 2 g IV eight-hourly (grade A)
|Non-life-threatening penicillin allergy (rash): |
Cefepime 2 g IV eight-hourly (grade C)
Other reasonable choices for monotherapy are ceftazidime 2 g IV eight-hourly or meropenem 1 g IV eight-hourly (grade C)
|Life-threatening (immediate) penicillin allergy or beta-lactam allergy: |
Aztreonam 1–2 g IV eight-hourly OR ciprofloxacin 400 mg IV 12-hourly (expert opinion)
+vancomycin 1.5 g IV 12-hourly (if CrCl >90 mL/min) OR
1 g IV 12-hourly (if CrCl 60–90 mL/min)‡)†
|Patients with systemic compromise (The combination of a beta-lactam antibiotic with an aminoglycoside is the regimen of choice)||As for patients without features of systemic compromise (expert opinion): |
+gentamicin 5–7 mg/kg ideal body weight IV once daily, adjusted to level
±vancomycin 1.5 g IV 12-hourly (if CrCl >90 mL/min) OR
1 g IV 12-hourly (if CrCl 60–90 mL/min)‡
|Patients with cellulitis, obviously infected vascular devices, or MRSA carriers with extensive skin breaks/desquamation||As for patients without features of systemic compromise: |
+vancomycin 1.5 g IV 12-hourly (if CrCl >90 mL/min) OR
1 g IV 12-hourly (if CrCl 60–90 mL/min)‡
|Patients with features of abdominal or perineal infection||As for patients without features of systemic compromise: |
+metronidazole 500 mg IV/oral 12-hourly if receiving cefepime, ceftazidime or ciprofloxacin first-line (grade D)
Alternatively, piperacillin-tazobactam or meropenem will provide adequate anaerobic cover, if required (grade B), other than for suspected or proven Clostridium difficile-associated diarrhoea or colitis
Piperacillin-tazobactam is an appropriate agent for the treatment of neutropenic fever (grade A recommendation). It performed better than cefepime and similar to carbapenems in the systematic reviews comparing regimens for monotherapy (level I evidence).7 It may also be associated with less vancomycin use.22 Pharmacokinetic/pharmacodynamic studies support the use of six-hourly piperacillin-tazobactam for Pseudomonas infection (level III-1 evidence).23
Data from the Asia-Pacific SENTRY surveillance program in 2009 reported <3% resistance to piperacillin-tazobactam in Enterobacteriaceae and ∼15% for Pseudomonas aeruginosa (unpublished data from SENTRY program).24 Co-resistance to gentamicin is uncommon supporting its use as a dual agent in severe sepsis.
The use of cefepime in neutropenic fever has been influenced by a published meta-analysis suggesting that it may be associated with an increased mortality in comparison with other monotherapy agents, the cause of which is unclear.7 More recently, a patient level analysis of published and unpublished studies by United States Food and Drug Administration concluded that cefepime was not associated with increased mortality and continues to be an appropriate agent for the empiric treatment of neutropenic fever25 (grade A recommendation).
Pharmacokinetic/pharmacodynamic data in neutropenic fever support the use of eight-hourly dosing.26
Ceftazidime has reduced activity against viridans group streptococci and organisms possessing extended spectrum beta-lactamase (ESBL) in comparison with other monotherapy agents.27–30 Ceftazidime monotherapy should be avoided in patients with extensive mucositis, patients receiving high-dose methotrexate or cytarabine, ciprofloxacin prophylaxis and other patients with risk factors for viridans group streptococcal infection (expert opinion).
Although ticarcillin/clavulanate with gentamicin is an effective and evidence-based empiric regimen, it confers a higher risk of nephrotoxicity than the recommended monotherapy regimens (level I evidence).31 It may still be considered in patients at low risk of nephrotoxicity. Evidence for the use of ticarcillin/clavulanate as monotherapy is restricted to a single study of 83 patients.32 The consensus group concluded that there is insufficient evidence to support its use as monotherapy when alternate agents (with greater numbers of patients studied) are available.
Meropenem is a ‘reserve’ antimicrobial agent in many institutions, making it inappropriate for first-line, empiric treatment in these centres. Carbapenem-resistant bacteria (with few remaining treatment options available or in the pipeline) are an emerging therapeutic and infection control problem in Australia and worldwide.33–36 Their emergence is likely to be amplified by overuse of this drug class. As such, prudent use of carbapenems is strongly advised. There is also a greater incidence of pseudo-membranous colitis associated with this agent compared with other monotherapy agents.7
Use of carbapenems upfront with a de-escalation strategy has been suggested as a way to prevent the emergence of resistance but is not recommended by the consensus group. There are no clinical data in a neutropenic population to support this strategy, and routine carbapenem use is discouraged for the reasons outlined above.
Ciprofloxacin is ‘reserved’ for recalcitrant infections in some institutions, making it inappropriate for first-line, empiric treatment in patients without life-threatening/immediate hypersensitivity to beta-lactam antibiotics (expert opinion). As ciprofloxacin has inadequate Gram-positive coverage, it should be combined with vancomycin if chosen as first-line empiric coverage. Use of ciprofloxacin in combination with an antipseudomonal beta-lactam (in place of an aminoglycoside) is discouraged because of its high potential to drive resistant organism emergence, and the lack of good data demonstrating clinical benefit (expert opinion). Ciprofloxacin is likely to be ineffective in patients with breakthrough neutropenic fever in centres using quinolones as primary prophylaxis due the increased rates of quinolone resistance in Gram-negative bacilli37–41
While gentamicin is the most commonly reported aminoglycoside used in combination therapy for neutropenic fever, amikacin and tobramycin have been well studied in clinical trials.10 While the addition of an aminoglycoside has not been shown to be of clinical advantage compared with beta-lactam monotherapy in systematic reviews, there are particular circumstances where the choice of aminoglycoside may be important. These include severe sepsis where there is a risk of resistance in Gram-negative bacilli and in Pseudomonas infection.
Gentamicin provides adequate coverage in most circumstances, but individual institutions may consider alternatives, such as tobramycin, which has shown marginally less nephrotoxicity and better in vitro activity against Pseudomonas and amikacin, which retains activity against many Gram-negative bacteria which are resistant to other aminoglycosides.24,42 Depending on local patterns of resistance, amikacin may be the preferred agent for serious nosocomial infections caused by Gram-negative bacilli.
Important considerations influencing the choice of first-line antibiotic regimen until susceptibility results are available
Clinically stable patients
Antipseudomonal beta-lactam monotherapy is recommended, unless a life-threatening/immediate hypersensitivity reaction has been documented, for all clinically stable patients without features of systemic compromise (grade A recommendation). For patients with non-life-threatening cases of penicillin allergy, for example non-urticarial rash, cefepime is recommended (grade C recommendation). Other reasonable choices for monotherapy in this situation include ceftazidime or meropenem (grade C recommendation) although the latter is a reserve agent and is not generally recommended for first-line neutropenic fever therapy for reasons discussed below. Ciprofloxacin or aztreonam and vancomycin in combination are recommended for clinically stable patients with life-threatening cases of penicillin or other beta-lactam allergy, unless these patients are known to tolerate an alternative suitable antipseudomonal beta-lactam agent (grade C recommendation). See Table 3 for dose details.
Patients with systemic compromise
Patients with shock and/or features of organ failure are not well represented in published studies and the working group felt that there is insufficient evidence to support the use of monotherapy in this patient group. Therefore, the recommended empiric regimen is an antipseudomonal beta-lactam antibiotic plus gentamicin once daily adjusted according to systemic levels (grade D recommendation).
With regard to empiric vancomycin usage, the consensus group agreed that all patients in septic shock, as well as patients known to be colonized with a resistant Gram-positive bacterium (e.g. methicillin-resistant Staphylococcus aureus (MRSA)) showing any criterion of systemic compromise, should receive vancomycin coverage for 48–72 h (expert opinion). The group did not reach consensus on whether patients with systemic compromise (other than septic shock) not known to be colonized with a resistant Gram-positive organism should also receive vancomycin. In all cases, antibiotic coverage should be regularly reviewed in line with clinical progress and microbiological results. In the absence of positive cultures for resistant organisms, gentamicin should be discontinued after 24–48 h and vancomycin should be discontinued after 48–72 h (expert opinion).
Patients with features of abdominal or perineal infection
Clinicians should follow the antibiotic regimen described for clinically stable patients in Table 3. However, metronidazole 500 mg intravenous (IV) or oral 12 hourly should be added if the patient is receiving cefepime, ceftazidime or ciprofloxacin first-line, as these agents provide inadequate anaerobic coverage (grade D recommendation). Patients receiving piperacillin-tazobactam or meropenem do not require additional adequate anaerobic cover with metronidazole, unless Clostridium difficile-associated diarrhoea or colitis is suspected or proven.
Patients with a pulmonary infiltrate
The management of pulmonary infiltrate is beyond the scope of this clinical guideline; however, bacterial, viral and fungal pathogens should be considered based on the patients' degree of immunosuppression and clinical features. Bacterial pneumonia is the most common cause of pulmonary infiltrate in neutropenic fever. In patients with prolonged neutropenia, invasive fungal infection must be excluded. Investigations may include imaging of the lungs (chest X-ray or high resolution computed tomography scan where possible), sputum and/or broncho-alveolar lavage. Standard empiric therapy will cover most common bacterial respiratory pathogens until microbiology data become available to guide therapy.
Patients with severe penicillin allergy
No published studies exist to guide how patients who cannot receive beta-lactams (because of severe allergy) should best be managed (expert opinion). The recommendations provided represent the consensus opinion of the current working group. A combination of agents should be used that provide antipseudomonal and Gram-positive cover (e.g. ciprofloxacin + vancomycin or aztreonam + vancomycin). Aztreonam does not contain the bicyclic beta-lactam ring structure characteristic of other beta-lactam antibiotics. To date, aztreonam has not demonstrated clinical cross-reactivity in penicillin-allergic patients,43–45 other than cystic fibrosis patients who became sensitized and developed allergic reactions after re-exposure to aztreonam at a later time (although these patients have shown a propensity to develop allergic reactions after numerous exposures to the same or a possibly cross-reacting antibiotic46,47).48 Of note, clinical cross-reactivity between ceftazadime and aztreonam has been demonstrated in vitro and plausibly explained by similarity of their side-chains. Accordingly, extreme caution is advised for patients who are allergic to ceftazidime and subsequently placed on aztreonam therapy.48
Patients at risk of infection with multi-resistant organisms (MRSA, VRE, ESBL)
Patients who have had a past history of infection or colonization, or who are being treated in an institution with high endemic rates of a multi-resistant organism may require modification of the first-line regimen particularly when there is systemic compromise. For MRSA, consider adding vancomycin; for VRE consider adding teicoplanin or linezolid; for ESBLs, consider use of a carbapenem.
Patients in rural/regional setting
Many patients present with suspected neutropenic fever to a rural or regional hospital that do not administer chemotherapy. In many circumstances, access to certain antibiotics and blood products, 24-h medical cover and appropriately skilled health professionals may not be available. Once an initial assessment has been made, local staff should contact the on-call medical oncologist or haematologist (or registrar) from the patient's usual treating centre (facility responsible for administering the patient's chemotherapy) and agree to a management plan. In many circumstances (such as unstable clinical state, the non-availability of certain antibiotics or blood products, lack of 24-h medical cover), the patient will require transfer to the parent unit once initial stabilization has been achieved.
These smaller centres are encouraged to stock empirical antibiotics, such as piperacillin-tazobactam, cefepime or ceftazidime, obtain IV access and commence IV fluid resuscitation, and administer the first dose of antibiotics as per Table 3. Patients suspected of having neutropenic fever should receive rapid assessment. Administration of antibiotics should not be delayed by pending blood results, for example FBE, which can take more than 24 h to turn round in some circumstances. Please refer to Table 2.
Inclusion of a glycopeptide in initial empiric therapy
Routine addition of vancomycin to the initial empiric antibiotic regimen within 72 h does not reduce mortality and is associated with an increased risk of adverse events, mainly nephrotoxicity (level I evidence, grade A recommendation).49,50 There is good evidence to show that empiric (non-targeted) first-line use of vancomycin is unnecessary and potentially harmful.49
- i. Empiric vancomycin is considered unnecessary in most clinically stable patients receiving monotherapy with an antipseudomonal beta-lactam who have no definite sites of Gram-positive infection (grade A recommendation).
- ii. Patients with microbiologically documented Gram-positive infection should receive vancomycin until identification and susceptibility results are available, at which time therapy should be modified appropriately (expert opinion).
- iii. Evidence regarding teicoplanin as an alternative to vancomycin is conflicting. High protein binding is thought to reduce the efficacy of teicoplanin, which has also been shown to be associated with longer duration of fever than vancomycin.51 However, a more recent study did not support the difference in efficacy between the two drugs and promotes the use of teicoplanin as a safer option because of its lower incidence of nephrotoxicity.52 Hospital formulary listings may govern the choice of agent used. In patients with non-life-threatening hypersensitivity to vancomycin, clinicians may consider the cautious use of teicoplanin, although cross-reactivity is possible.
- iv. Systemic compromise. Evidence regarding glycopeptide use in patients presenting with neutropenic fever and systemic compromise is lacking. There is currently no consensus opinion. Vancomycin may be given appropriate consideration.
- v. Patients at risk of resistant Gram-positive infection. Clinicians are advised to review institution-specific bloodstream infection data to ascertain-specific groups (such as high-dose cytarabine or ciprofloxacin prophylaxis patients) who are at higher risk of resistant Gram-positive infections, such as penicillin-resistant viridans streptococci. In such groups, vancomycin use may be considered pending culture results particularly if the antipseudomonal beta-lactam used is one with relatively poor viridans streptococcal coverage, for example ceftazadime.For these patients presenting with shock, addition of vancomycin to empiric therapy is recommended (expert opinion).
- vi. Patients with cellulitis, obviously infected vascular devices or MRSA carriers with extensive skin breaks/desquamation. Vancomycin, according to a validated protocol, should be added to the beta-lactam antibiotic regimen described for clinically stable patients in Table 3, for 48–72 h then reviewed according to culture results (expert opinion)
The recommendations in these consensus guidelines are based on published, peer-reviewed evidence and expert opinion. The recommendations for individual antibiotics may not align directly with the Therapeutic Goods Administration (TGA)-approved indications. Therefore, following these recommendations may lead to off-label use of particular agents.
Modification of empiric therapy during the course of neutropenic fever
As the median time to defervescence in patients successfully treated with frontline antibiotics is 3–5 days, escalation of antibiotic coverage should not occur prior to this period in the absence of clinical instability, isolation of a resistant organism or emergence of new infective foci (expert opinion).
Any subsequent modifications to the initial choice of antibiotic should be guided by repeat clinical assessment (e.g. emergence of focal sites of infection) and microbiological culture results (expert opinion)
When a causative microorganism is identified, consensus opinion supports the commencement of therapy targeted to this pathogen or continuation of the initial antibiotic regimen, provided the patient has become afebrile within 3–5 days of commencing that treatment (expert opinion).53 The duration of therapy for an identified pathogen should be at least that recommended for a similar infection in an immunocompetent host (expert opinion).
Addition of a glycopeptide for persistent fever
Although it is common practice to add vancomycin to empiric coverage in patients with persistent fever at 3–5 days, this practice is discouraged because of a lack of clinical benefit54 (grade B recommendation), as well as nephrotoxicity and resistance concerns. Whether the addition of vancomycin is likely to be beneficial at a later time, such as more than 5 days of persistent fever, has not been adequately studied. Use of vancomycin in this circumstance may be considered, particularly for patients with severe neutropenia, that is, <0.1 × 109 cells/L (expert opinion).
Regardless of fever, vancomycin should be commenced for 48–72 h, pending cultures, in patients who develop haemodynamic instability during antimicrobial therapy for neutropenic fever (expert opinion).
Empiric coverage of viral or fungal infection
In the absence of clinical evidence of viral or fungal infection, empiric frontline coverage for these pathogens is not required (expert opinion). For empiric management of fungal infections, the working group refers the reader to the Australian and New Zealand Consensus Guidelines for the Use of Antifungal Agents in the Haematology/Oncology Setting 2008.55