The neutrophil is an important component of the innate immune system. Neutrophils primarily defend the body against microorganisms, and a low number of neutrophils indicates that a person is vulnerable to infection. Additionally, because neutropenic patients lack the leukocytes needed to develop an inflammatory response, common inflammatory manifestations that are observed in patients within the normal range of leukocytes are rarely found. Thus, except in the presence of a fever, an accurate diagnosis is difficult and the most appropriate time for treatment may be missed. Thus, febrile neutropenic patients should be treated differently from other febrile non-neutropenic patients [1].

Many countries, including the US and Europe, have developed and reported guidelines on approaches to and treatments for febrile neutropenic patients. However, the pattern of neutropenic fever has changed over the last 20 years, and the distribution and resistance rate of causative microorganisms are known to differ by region, antibiotic prophylaxis, and the use of catheters [2].

The aim of this study was to investigate the epidemiology of infectious diseases and the patterns of resistance and antibiotic therapy in febrile neutropenic patients, and to develop and suggest empirical treatment guidelines for neutropenic fever that fit the circumstances in Korea through both a foreign literature review and a multidisciplinary study. These guidelines are for adults and refer to data published in Korea. These guidelines are also applicable to other diseases associated with neutropenia, anticancer therapy of malignant tumors, and hematopoietic stem cell transplantation (HSCT) recipients.

In June 2009, the committee for the development of "Guidelines for the Empirical Therapy of Neutropenic Fever Patients based on Literature in Korea" was organized by receiving recommendations from committee members from eight academic societies under the supervision of the National Evidence-based Healthcare Collaborating Agency (NECA): the Korean Society of Infectious Diseases (KSID), the Korean Society for Immunocompromised Host Infections (KSIHI), the Korean Cancer Association (KCA), the Korean Society of Clinical Microbiology (KSCM), the Korean Society of Blood and Marrow Transplantation (KSBMT), the Korean Society of Hematology (KSH), the Korean Society for Chemotherapy (KSC), and the Korean Society of Clinical Oncology (KSCO). The committee consists of five infectious diseases physicians, four hematology-oncology physicians, one laboratory medicine physician, one NECA internist, and one methodologist.

For a systematic literature review, the latest guidelines of Infectious Diseases Society of America (IDSA) [2], National Comprehensive Cancer Network (NCCN) [3], the Infectious Diseases Working Party (AGIHO) of the German Society of Hematology and Oncology (DGHO) [4-13], the First European Conference on Infections in Leukaemia (ECIL-1) [14-18], Asia-Pacific [19], and Japan [20-27] were collected. To search the literature published after the publication of the IDSA guidelines (2002), which are relatively widely used, the PubMed (www.pubmed.gov) search engine was used. The search period was from January 2002 to October 2009. Search entries for neutropenia were "neutrop*nia," "granulocytop*nia," and "leu?op*nia." The search entries for tumor were "cancer," "malignancy," "neoplasm," "leukemia," "lymphoma," "hematolog*" and the combination of "(stem or marrow) AND transplantation." Literature regarding fever and antibiotic therapy were searched by combining "fever or febrile," "anti-infect*," "anti-bacteri*," "anti-microb*," "anti-bio*," "anti-fung*," and "anti-vir*." To find Korean studies published in foreign journals, the Korean literature was also searched through the PubMed engine.

Major reports published in Korea over the last 10 years were searched through the database of Korean Studies Information (http://kiss.kstudy.com) and KoreaMed (http://www.koreamed.org). Search entries were combination of "neutrophil" or "granulocyte," "fever" and "infection" by Korean letters. Reports before 2000 were collected if they were considered to be related to the development of this treatment guideline. Related literature was added by searching references of the collected literature, manually if necessary. The searched Korean literature totaled 39 reports (4 review articles and 35 original articles). In total, 218 references are cited; 27 were from the Korean literature.

To create empirical treatment guidelines for febrile neutropenic patients, the following major categories were selected: definition of neutropenia and fever, initial evaluation and risk of infection, antibiotic prophylaxis, initial antibiotic therapy for febrile neutropenic patients, re-evaluation after 3-5 days and change of antibiotics, use of glycopeptides, catheter-related infections, and antifungal therapy.

The subcommittee of infectious diseases specialists formulated key questions in each area. Key questions were determined by reviewing foreign treatment guidelines and recommendations that could cause problems in Korean circumstances.

Recommended answers to the key questions were based on major guidelines and literature, and the final version of these recommendations was made by a consensus of the guideline development committee.

For strength of recommendations and quality of evidence, the methods used in the latest guidelines of IDSA were accepted (Table 1) [28]

Because neutropenic patients have a high risk of infection, antibiotic prophylaxis can be helpful. However, if antibiotic prophylaxis is applied to all neutropenic patients, including those at a relatively low risk of infection who do not need it, antibiotic-resistant bacteria may emerge and excessive medical costs may be incurred. Thus, it is important to determine which patients will be most helped by antibiotic prophylaxis and the appropriate period for prophylaxis. Patients for whom prophylactic antibacterial, antifungal, or antiviral treatment is recommended are shown in Table 2 [3].

While past studies on antibiotic prophylaxis mainly used sulfamethoxazole/trimethoprim, many studies conducted since the late 1990s have used fluoroquinolones. According to a meta-analysis on the use of prophylactic antibacterial agents [44], the group using prophylactic antibacterial agents showed a lower mortality rate following infection and a lower total mortality rate compared with those not using prophylactic antibiotics or using placebo; the effects in the fluoroquinolones group were particularly evident. Another meta-analysis reported that fever, infection caused by Gram-negative bacteria, microbiologically documented infection, and total infection occurred less in patients using prophylactic fluoroquinolones than in those using sulfamethoxazole/trimethoprim or placebo. However, even prophylaxis using fluoroquinolones did not reduce Gram-positive bacterial infection, fungal infection, or the mortality rate [45]. Additionally, although there was concern about the emergence of antibiotic-resistant bacteria in the group using prophylactic antibacterial agents, resistance was not increased in the fluoroquinolone group. However, some reports have stated that resistant Gram-negative bacteria increased during fluoroquinolone prophylaxis, and this tendency improved after discontinuation of the prophylaxis; close attention to this is necessary [46]. That is, evidence for the use of prophylactic antibacterial agents such as fluoroquinolones exist in intermediate-to-high risk groups, but the long-term effects of antibiotic prophylaxis have not yet been fully determined; the emergence of resistant bacteria should be continuously monitored and antibiotic prophylaxis should be optimized in each hospital.

Ciprofloxacin or ofloxacin, which were widely used for prophylaxis in the past, had good antimicrobial activities against Gram-negative bacteria but relatively poor activities against Gram-positive bacteria. Studies using levofloxacin showed outstanding antimicrobial activity against Gram-positive bacteria as a prophylactic antibacterial agent for solid tumor and lymphoma patients [47] and solid tumor, lymphoma, and acute leukemia patients [48]. Both were large-scale studies that included over 300 subjects in each group. Infection caused by Gram-positive bacteria or bacteremia that was not prevented by fluoroquinolones decreased in the levofloxacin group. Although infection caused by Gram-negative bacteria was also significantly decreased, the total mortality rate and the mortality rate due to infection were not significantly different between the two groups.

In Korea, a study was performed on the effects of prophylactic antibacterial agents in acute leukemia patients undergoing anticancer therapy [49]. The study used ciprofloxacin and roxithromycin for prophylaxis, and the patients using these prophylactic antibacterial agents showed fewer Gram-negative bacterial infections, but more Gram-positive bacterial infections. Additionally, the total infection rate and the mortality rate following infection did not differ between the two groups. As of 2010, the hospital does not use prophylactic roxithromycin.

Because there has been no prospective clinical study comparing the effect of the same drug for different periods to determine the proper administration period of prophylactic antibacterial agents, it is difficult to determine the appropriate end point of antibacterial prophylaxis. Most previous studies that showed effective outcomes of antibacterial prophylaxis used antibacterial agents from the beginning of anticancer therapy or within 48-72 hours after anticancer therapy until recovery of the absolute neutrophil count, and there was no difference in the preventive effects based on the administration period [48,50-53].

When prolonged neutropenia is expected, such as in patients undergoing remission induction therapy or maintenance/consolidation therapy due to a hematologic malignancy or those receiving allogeneic HSCT, antifungal prophylaxis is recommended [2,3]. Azoles have been widely used as prophylactic antifungal agents because of their favorable costs, adverse reaction profiles, and they allow the selection of other therapeutic antifungal agents for breakthrough fungal infections during antifungal prophylaxis.

A meta-analysis on antifungal prophylaxis revealed that the total mortality rate (relative risk [RR], 0.84; 95% confidence interval [CI], 0.74 to 0.95), fungus-related mortality rate (RR, 0.55; 95% CI, 0.41 to 0.74), invasive fungal infection, definite invasive fungal infection, definite invasive Candida infection, and the use of empirical antifungals decreased in the group using prophylactic antifungal agents compared with the groups not using prophylaxis or using a placebo [54]. Although fluconazole as a prophylactic antifungal agent tended to increase invasive aspergillosis, antifungal agents with specificity for filamentous fungi, such as itraconazole, resulted in less invasive aspergillosis infections [54].

Many studies have been performed on fluconazole, which has been shown to be very effective. Prophylaxis using 400 mg fluconazole per day reduced invasive fungal infections and the infection-related mortality rate in allogeneic HSCT recipients [55,56]. However, some reports stated that fluconazole did not significantly prevent invasive fungal infection in acute leukemia or autologous HSCT patients [57-59]. Additionally, studies using less than 400 mg prophylactic fluconazole per day did not find any significant difference in invasive fungal infections or mortality rates [60-62].

Itraconazole has antimicrobial activity against Aspergillus species and its preventive effect is thought to be superior to that of fluconazole; however, a study that directly compared fluconazole and itraconazole showed no significant differences in the all-cause mortality rate, fungus-related mortality rate, definite invasive fungal infection, invasive Candida infection, or superficial fungal infection [63-69]. When the efficacy of itraconazole was compared with that of fluconazole, limiting the studies to those using itraconazole oral solution, not the capsule, invasive fungal and Candida infections were decreased significantly [65-69]. The combination of itraconazole with vincristine or cyclophosphamide should be avoided because of the potential for drug interactions, and the administration of itraconazole should be performed cautiously in patients with a history of heart failure or lower cardiac output because of its cardiotoxicity [70] (intravenous itraconazole is not approved as a prophylactic antifungal agent by the Korea Food and Drug Administration [KFDA] as of 2009).

Posaconazole oral solution has a wide range of antifungal activity against Candida and filamentous fungi [71,72]. An assessment of the effect of prophylactically administered posaconazole in acute myelogenous leukemia and myelodysplastic syndrome patients found that proven or probable invasive fungal infection and invasive aspergillosis were observed less and the survival rate was significantly higher in the posaconazole group than in the fluconazole or itraconazole groups. However, a higher frequency of adverse reactions was found, including increases in bilirubin and liver enzyme levels [73] (posaconazole is approved as a prophylactic antifungal agent for neutropenic fever by the KFDA, but is not on the market and is purchased through the Korea Orphan Drug Center as of 2009).

A study that compared micafungin and fluconazole as prophylactic agents in 882 allogeneic or autologous HSCT recipients reported that the success rate of prophylaxis (80.0% vs. 73.5%; 95% CI, 0.9 to 12) was higher and the frequency of invasive aspergillosis was lower in the micafungin group; however, the mortality rate was not significantly different between the two groups [74]. Moreover, no significant difference was found in the frequency of adverse reactions or the discontinuation rate between the two groups. However, a limitation of this study was that over 70% of the subjects were autologous HSCT or low-risk allogeneic HSCT recipients. A recent study insisted that the incidence of fungal infection was not significantly different between micafungin and fluconazole groups [75].

Although low-dose amphotericin B deoxycholate (0.2 mg/kg/day or 0.5 mg/kg 3 times per week) showed much better preventive effects than fluconazole, it is difficult to use in many cases because of its toxicity [76,77]. Because the amphotericin B lipid formulation has less toxicity than amphotericin B deoxycholate, studies on its use to prevent neutropenic fever have been conducted. Although a preventative effect of 50 mg (low-dose) liposomal amphotericin B was not observed in previous small-scale studies [78-80], recent large-scale studies found that it decreased invasive fungal infection and infection-related mortality rates [81]. Inhalation of amphotericin B deoxycholate has also been attempted to prevent pulmonary fungal infection [82-84].

In Korea, a study investigated the effects of itraconazole oral solution and fluconazole as prophylactic agents and showed that the preventative effects of the two drugs were not significantly different [85]. However, administration compliance was lower due to gastrointestinal adverse reactions in the itraconazole group.

Although it is difficult to find studie s with reliable evidence for the determination of the end point of antifungal prophylaxis, they are generally administered until recovery of absolute neutrophil counts occurs [55,57,59,62,79,81,86-93]. However, allogeneic HSCT recipients may require antifungal prophylaxis even after neutrophil recovery, and NCCN recommends continuing prophylaxis until 75 days after HSCT [3]. Additionally, when graft versus host disease (GVHD) is observed, the period of antifungal prophylaxis can be extended. A recent large-scale study reported that an average of 112-day prophylactic posaconazole therapy effectively prevented invasive fungal infection in patients with GVHD [94].

Sulfamethoxazole/trimethoprim can be used to prevent P. jirovecii in acute leukemia patients and HSCT recipients, and its preventive effect is excellent [95-97]. A meta-analysis on the prevention of P. jirovecii in immunocompromised patients (with the exception of human immunodeficiency virus [HIV] patients) showed that the P. jirovecii-related mortality rate was significantly reduced in the sulfamethoxazole/trimethoprim group (RR, 0.17; 95% CI, 0.03 to 0.94) [96,97]. Because P. jirovecii infection is known to increase in patients using alemtuzumab or fludarabine because of chronic lymphocytic leukemia or lymphoproliferative disorders [98-100], prevention of P. jirovecii can be considered. For this, 160/800 mg or 80/400 mg sulfamethoxazole/trimethoprim is administered, and if there is a concern about adverse events, such as bone marrow suppression, 160/800 mg is administered every other day. When the drug was used every other day, its preventive effect did not differ from that of daily administration (RR of pneumonia, 0.82; 95% CI, 0.61 to 1.09). Significantly more patients who took the drug daily discontinued it due to adverse reactions (RR, 2.14; 95% CI, 1.73 to 2.66) [101]. However, these results should be interpreted carefully because the study was performed not with neutropenic patients, but with HIV infection patients. If sulfamethoxazole/trimethoprim is difficult to administer because of leukopenia, dapsone or aerosolized pentamidine can be used [102]. However, dapsone and aerosolized pentamidine produce a weaker preventive effect against P. jirovecii than sulfamethoxazole/trimethoprim and can lead to additional infections and a higher mortality rate following infection [103].

If antiviral prophylaxis against HSV is not conducted for allogeneic HSCT recipients, approximately 62-80% of HSV IgG-seropositive patients show reactivation of the virus, while only 1-1.5% of HSV IgG-seronegative patients experience viral reactivation [104,105]. For autologous HSCT, a lack of antiviral prophylaxis leads to lesions caused by HSV in approximately 2-6% of cases [106-108]. Thus, antiviral prophylaxis is recommended for HSV-seropositive patients among allogeneic HSCT recipients and autologous HSCT recipients with a high risk of mucositis [3]. According to a study performed in the early 1990s, the antibody-positive rates of HSV type 1 were 100%, 91%, and 82% in populations aged over 30 years, in their 20s, and in their teens, respectively, in Korea. Antiviral prophylaxis is advised in most cases in this country [109].

A meta-analysis on studies using acyclovir to prevent reactivation of HSV revealed that lesions caused by HSV and its isolation rate were significantly decreased in the acyclovir group [110,111]. However, the mortality rate was reduced only when prophylactic antiviral agents were used during engraftment after allogeneic HSCT [111]. Recent studies using valaciclovir, which is more easily administered than acyclovir, reported that the development of HSV lesions was not significantly different between acyclovir and valaciclovir groups [112,113]. Antiviral prophylaxis is generally used until the completion of engraftment or the improvement of mucositis (approximately 30 days in most cases) [114,115].

Although varicella-zoster virus is frequently observed in patients undergoing anticancer therapy, it is not mentioned in these guidelines because it is beyond the scope of empirical therapy for neutropenic fever.

The distribution of etiological agents of neutropenic fever in studies published in Korea over the last 10 years is shown in Table 3. While Gram-positive bacteria account for 60-70% of microbiologically documented infection in Europe and America, Gram-negative bacteria were more frequently observed in studies in Korea until the early 2000s. This is a general characteristic in the Asia-Pacific region, including China, Taiwan, Thailand, and Malaysia [19]. Among Gram-positive bacteria, Streptococcus and coagulase-negative Staphylococcus are the most frequently observed, and Staphylococcus aureus and Enterococcus are next. Among Gram-negative bacteria, Escherichia coli is found most frequently, and Pseudomonas aeruginosa and Klebsiella pneumoniae follow it.

Little data regarding antimicrobial susceptibility to etiological agents has been reported in Korea, and the reported resistance rates vary. The rate of methicillin-resistant S. aureus (MRSA) and those of fluoroquinolone-resistant and third-generation cephalosporin-resistant E. coli were 38-77%, 16-93%, and 0-7.0%, respectively [116-118]. Thus, each hospital needs to choose early empirical antibacterial agents by considering the types of frequently detected bacteria and their susceptibilities. For example, ciprofloxacin combination therapy is difficult to use as an early empirical antibacterial agent in hospitals showing a high fluoroquinolone resistance rate. Additionally, these guidelines do not recommend glycopeptides as early empirical antibacterial agents, but their partial use can be considered in hospitals with high MRSA rates. These guidelines describe general recommendations, and antibacterial agents not mentioned in these guidelines can also be empirically used according to types of detected bacteria and their susceptibilities in each hospital.

Empirical antibiotics as initial therapy should be chosen by considering the susceptibilities of the bacteria detected in each hospital. For hospitals with high rates of resistant bacteria, such as MRSA and multiple-drug-resistant Gram-negative bacteria, appropriate antibiotics should be used based on the circumstances in each hospital. Additionally, these guidelines suggest generally recommended antibiotics, and antibiotics not mentioned in these guidelines can also be used properly if their effects are demonstrated.

Use of antibiotics against Pseudomonas is commonly recommended as an initial empirical antibiotic therapy. Other factors that should be considered in choosing initial empirical antibiotics for febrile neutropenic patients include the infection site (s), history of MRSA infection or colonization, organ dysfunction, history of the use of antibiotics, and bactericidal effects of antibiotics.

When fever persists even after 3-5 days of antibiotic therapy and neither infection sites nor causative organisms are detected, reasons shown in Table 4 can be considered. Re-evaluation of the following is necessary: complete blood cell count, general chemistry, electrolyte test, C-reactive protein (CRP), urinalysis, results of all cultures, a close physical examination, chest X-ray, evaluation of any vascular catheter, additional cultures of blood and specimens from specific infection site (s), imaging studies on sites suspected to have infection (if possible), blood antibiotic levels (particularly aminoglycosides), and ultrasonography or CT for patients with pneumonia, paranasal sinusitis, or enteritis.

The current blood culture system can detect 90-100% of bacteria in blood within 48 hours of blood culture. Thus, it is recommended to repeat blood cultures at 48-hour intervals, as necessary.

Eighteen randomized studies have investigated whether glycopeptides should be added to an initial empirical antibiotic regimen. Of them, only two were double-blind randomized trials [17]. The largest was a multi-center study that included 747 subjects [161], and the smallest had only 46 subjects [162]. When the 747 neutropenic patients in the largest study were randomly divided into addition of vancomycin to ceftazidime + amikacin and no-addition groups, the vancomycin addition group showed faster responses in patients who were found to have bacteremia caused by Gram-positive bacteria; however, the addition group was not significantly different in terms of defervescence and mortality rate compared with the no-addition group. Furthermore, no patient died in the first 3 days among the patients with bacteremia caused by Gram-positive bacteria. However, renal toxicity following the use of antibiotics occurred in 2% of patients in the no-vancomycin group, but was significantly higher (6%) in the vancomycin group (p = 0.02) [161]. Recently, the results of two meta-analyses on the need for the administration of vancomycin as an initial empirical antibiotic regimen were reported [163,164]. One meta-analysis examined a total of 2413 patients by including 14 of 18 randomized studies [163]. It revealed that the addition of a glycopeptide did not significantly reduce the total mortality rate (odds ratio [OR], 0.67; 95% CI, 0.42 to 1.05). In particular, an analysis of 405 patients that included only six studies using the same broad-spectrum antibiotic showed the same finding (OR, 1.05; 95% CI, 0.52 to 2.00). The other meta-analysis investigated a total of 2392 patients by including 13 randomized studies [164]. It also found that the additional administration of a glycopeptide did not significantly decrease the total mortality rate (RR, 0.96; 95% CI, 0.58 to 1.26). For breakthrough infection, the first meta-analysis did not show any significant association with the use of glycopeptides with an OR of 1.18 (95% CI, 0.81 to 1.98) [163], while the second one found that breakthrough infection caused by Gram-positive bacteria was reduced, with a RR of 0.28 (95% CI, 0.11 to 0.37) [164]. However, these findings should be interpreted carefully. All of the studies in the analysis were conducted from 1985 to 1993, before the emergence of vancomycin-resistant enterococci (VRE); thus, VRE breakthrough infection during the administration of vancomycin was not reflected [17]. Moreover, because viridans streptococci bacteremia can deteriorate rapidly, to streptococcal toxic shock syndrome, in neutropenic patients, there is a suggestion that the addition of vancomycin to an initial antibiotic regimen is favorable in hospitals with high penicillin resistance of viridans streptococci [165]. However, most β-lactam antibiotics (e.g., cefepime, imipenem/cilastatin, meropenem, piperacillin/tazobactam), except ceftazidime, have good antibacterial activity against viridans streptococci, so vancomycin is not likely to be of additional help unless ceftazidime monotherapy is used [166].

Although the frequencies of MRSA and VRE are high in Korea, there has been no randomized study on this issue. Only one retrospective study on MRSA bacteremia in not only neutropenic patients, but also others, reported that the addition of vancomycin to an initial antibiotic regimen did not significantly affect the prognosis [167]. According to data from an analysis of 457 febrile neutropenic patients in a university hospital for 10 years [158], S. aureus was identified in 10 (6%) of 172 patients with proven bacteremia, and 77% of them had MRSA. Although data on the rate of viridans streptococci bacteremia in febrile neutropenic patients are insufficient in Korea, approximately 5-7% of the total bacteremia was reported to show viridans streptococci [117,158]. The data on penicillin-resistance of viridans streptococci are also insufficient. One study reported that with the exception of pneumococcus isolated from neutropenic patients, 7 (36%) of 19 streptococci strains were penicillin-resistant [116]. While some researchers have reported that the antimicrobial susceptibility tests of 103 strains of viridans streptococci isolated from various clinical specimens in a university hospital found no penicillin resistance (although they were not from neutropenic patients) [158,168], others have insisted that of 45 strains isolated from blood, 27% were not susceptible to penicillin [169].

However, S. aureus bacteremia is rarely found as a causative organism of bacteremia for the first fever in neutropenic cancer patients after cytotoxic chemotherapy. Its rate was reportedly 1-2% in large-scale clinical studies [132,138,144]. Furthermore, the frequency of resistant bacteria, such as MRSA, is low for the first fever; bacteremia caused by Gram-positive bacteria does not deteriorate rapidly in cases of late initial treatment, unlike bacteremia caused by Gram-negative bacteria, and indiscriminate use of glycopeptides can lead to the emergence of resistant bacteria and nephrotoxicity. Thus, there is insufficient evidence to support the routine inclusion of glycopeptides in the initial antibiotic therapy for febrile neutropenic patients in Korea. Based on these findings, it is recommended not to routinely add vancomycin to the treatment regimen for febrile neutropenic patients in whom the cause has not been clearly determined (A-I).

Two randomized studies investigated whether the addition of a glycopeptide was effective if fever persisted for 3-4 days after beginning initial antibiotic therapy [170,171]. One study randomly administered vancomycin or placebo to 165 of 763 patients whose fever had not improved within 3-4 days after the empirical use of piperacillin/tazobactam. The two groups were not significantly different in terms of defervescence, mortality rate, breakthrough infection, or the frequency of the use of amphotericin B deoxycholate [170]. This result was consistent with that of a recent meta-analysis (RR of treatment failure, 0.61; 95% CI, 0.18 to 2.09) [164]. Based on these findings, it is recommended not to routinely add glycopeptides when fever persists or recurs after 3-5 days (B-I).

Studies on the detailed indications of glycopeptides as an initial empirical antibiotic regimen are not sufficient, but the indications consistently suggested by most specialists, including those who contributed to the IDSA and NCCN guidelines, are as follows:

Positive for Gram-positive bacteria in blood culture (A-II)

The following indications remain controversial among specialists:

Risk of viridans streptococci bacteremia (B-III)

However, even after glycopeptides are administered according to the indications, the discontinuance of glycopeptides is recommended if bacteremia caused by resistant Gram-positive bacteria is not observed in blood cultures (A-I).

As of 2009, 18 randomized studies comparing the efficacy and adverse reactions of teicoplanin and vancomycin had been reported. Of these studies, 13 were performed with neutropenic patients. A meta-analysis on these randomized studies revealed that the mortality rates (RR, 0.95; 95% CI, 0.74 to 1.21) and clinical failures (RR, 0.92; 95% CI, 0.81 to 1.05) were not significantly different between teicoplanin and vancomycin [172]. Additionally, an analysis limited to the studies conducted in neutropenic patients found no statistically significant difference in the mortality rates (RR, 0.95; 95% CI, 0.68 to 1.34) or clinical failures (RR, 0.98; 95% CI, 0.83 to 1.16) between teicoplanin and vancomycin [172]. However, fewer adverse reactions were observed in the teicoplanin group compared with the vancomycin group (RR, 0.61; 95% CI, 0.50 to 0.74). In particular, nephrotoxicity was lower in the teicoplanin group than in the vancomycin group (RR, 0.44; 95% CI, 0.32 to 0.61) [172]. However, teicoplanin is not yet approved by the US FDA as an empirical antibiotic for MRSA infection and neutropenia, and clinical experience and research data on it in severe infections (e.g., endocarditis and encephalomeningitis) are not sufficient compared with those for vancomycin. Moreover, previous randomized studies did not include many patients with MRSA infection itself [172]. As the minimum inhibitory concentration (MIC) of vancomycin against S. aureus has risen, treatment failures have been reported [173], and the vancomycin MIC and levels need to be measured in some cases. Data regarding the association between teicoplanin MIC and treatment failure are insufficient, and levels are difficult to examine; thus, it should be used carefully. Additionally, most studies have stated that the efficacy of teicoplanin was identical to that of vancomycin by taking it once daily after a loading dose [172], but some researchers recommend administering a high dose for infections with complications. Thus, the appropriate dose remains controversial [174].

Duration of antibiotic therapy should be determined by considering the infection site, causative organism, general condition of the patient, treatment response, and neutrophil recovery [3]. If the origin of fever is unclear, antimicrobials should be maintained until the absolute neutrophil count reaches 500/mm³ or higher [3]. When the causative organism or the infection site has been identified, treatment duration should be adjusted to the specific infectious disease by considering the neutrophil recovery [3]. Because there is insufficient evidence regarding treatment duration for these clinical circumstances, this committee suggests general recommendations.

Catheter-related infection is a common complication frequently observed in neutropenic patients [3]. Exit-site infection is defined as a local flare or induration within 2 cm of the exit of a catheter and tunnel infection is defined as a local flare or induration > 2 cm from the exit of a catheter, pus from the exit, or a local flare or induration along the tunnel [175]. If catheter-related infection is suspected, a skin swab for culture from the exit of the catheter should be obtained, and blood cultures from the catheter itself should be conducted [3]. Because the skin swab culture from the exit site has a low specificity for catheter-related infection, but shows a high sensitivity, it can be useful for the exclusion of certain diagnoses [176]. Additionally, a study reported that blood cultures from both central venous catheters and peripheral blood, when performed by the automated blood culture systems used by many hospitals, could also measure the time for bacteria to grow initially. Furthermore, the differential time to positivity (DTP) or the difference between the two times was helpful for the diagnosis of catheter-related infection [3]. That is, when > 120 min of DTP was designated as a cutoff value, its sensitivity and specificity were high for the diagnosis of long-term catheter-related infection in recent studies [177-181]. However, because the studies were not performed with long-term catheters, such as the Hickman catheter, and because the specificity of the examination was lower for patients already treated with antibiotics, the results should be interpreted carefully.

Most catheter-related infections are caused by Gram-positive bacteria, and coagulase-negative Staphylococcus is most frequently isolated [182]. Thus, if catheter-related infection is clinically suspected, a glycopeptide, such as vancomycin, can be used (A-II). Because linezolid was found to increase the mortality rate when it was routinely administered to patients with suspected catheter-related infection in a randomized trial [183], it is not routinely recommended for the treatment of suspected catheter-related infection except in patients with catheter-related infection confirmed to have been caused by Gram-positive bacteria (A-I).

Most catheter-related infections can be improved by antibiotic therapy without removal of the catheter [3]. In particular, the catheter salvage rate of coagulase-negative Staphylococcus reaches 70-80% with intravenous antibiotics alone; thus, the use of antibiotics without catheter removal is generally recommended [182,184] (A-II). However, for catheter-related infection caused by fungi (yeasts or molds) or non-tuberculous mycobacteria (e.g., Mycobacterium chelonae, M. abscessus, or M. fortuitum), the catheter should be removed immediately [3]. Bacillus spp., C. jeikeium, S. aureus, Acinetobacter, P. aeruginosa, S. maltophilia, and VRE can be difficult to treat with antibiotic therapy alone [3]. Thus, catheter-related infections caused by these microorganisms also require initial catheter removal (A-II). In cases with severe inflammation of the mucous membranes, intestinal bacteria flora, such as VRE and Candida, can cause infection through the blood; DTP is helpful in discriminating these cases from those of catheter-related infection [3]. Moreover, when the catheter is not removed because the presence of catheter-related infection is unclear, but the same bacteria are identified in consecutive blood cultures 48-72 hours after beginning appropriate antibiotics, catheter removal should be considered [3,182] (B-II). However, if catheter-related infection is suspected and a patient is clinically unstable, the catheter requires immediate removal [3] (A-II). In cases of bacteremia caused by S. aureus, when catheter-related infection is suspected, catheter removal is generally recommended, because the success rate is low [185] (A-II). However, if the fever of a patient with a catheter and unclear cause of infection improves in 48-72 hours after beginning proper antibiotics and the blood culture result is negative, the process can be closely monitored while maintaining the catheter [184] (B-II). General indications for catheter removal are presented in Table 6.

Empirical antifungal therapy is a standard treatment when broad-spectrum antibacterials are not effective in neutropenic fever patients, based on clinical studies conducted in the 1980s [2,3,11,14,186-190]. Although characteristics of the patients, medications, and epidemiology of fungi may differ compared with those of 20-30 years ago, a lack of antifungal therapy for continuous neutropenic fever can increase invasive fungal infection (IFI) and lead to a higher mortality rate following IFI. Thus, empirical antifungal therapy is recommended in these cases (A-II) [3,14,191].

Currently, 40-50% of neutropenic patients classified as high-risk are known to take empirical antifungal agents [190]. According to a study that analyzed patients after HSCT and anticancer therapy in a single center in Korea from March 2000 to February 2001, 122 of 318 (38.4%) patients used empirical antifungal agents, and 74 (23.8%) among them had IFIs that were caused by Aspergillus and Candida species in most cases (6, 46, and 22 proven, probable, and possible IFIs, respectively) [192]. A foreign study also stated that approximately 15-45% of patients with a continuous neutropenic fever were estimated to have IFI [190]. Other reasons that antifungal therapy is used before a definitive diagnosis of infection are 1) because IFI is difficult to diagnose during the neutropenic period, 2) the delay of antifungal therapy to definitive diagnosis can easily provoke disseminated infection, and 3) in the autopsy of patients who died of neutropenic fever, IFI (Candida or Aspergillus species in most cases) that had not been clinically documented was found, and a continuous fever was the only initial sign of IFI [14,188-190].

If fever persists or recurs even after the administration of antibacterials, empirical antifungal therapy should be conducted [2]. When to begin the empirical antifungal therapy can differ according to the degree of risk. Patients in the low-risk group do not need to start antifungal therapy before diagnosis, while those in the intermediate-risk group are recommended to begin antifungal therapy when fever persists after 6-8 days of beginning broad-spectrum antibacterials due to continuous neutropenic fever and neutropenia. For patients in the high-risk group with over 10 days of neutropenia, empirical antifungal therapy should be started quickly when neutropenic fever persists or recurs after 3-5 days of beginning broad-spectrum antibacterials or when the clinical condition deteriorates [11,193]. When neutropenia is expected to persist for a relatively short period (< 10 days) or estimated to have been resolved for several days from the decision of whether to use empirical antifungal therapy, empirical antifungal therapy is not routinely considered, unless there is a symptom or sign causing suspicion of invasive fungal infection or a history of invasive fungal infection (B-III).

Because many types of antifungal agents have been developed over the last 10 years and antifungal prophylaxis has been widely conducted in the high-risk group, filamentous fungi, as opposed to yeasts, such as Candida, have been found more frequently in IFIs of febrile neutropenic patients. Moreover, the rate of non-albicans species in candidiasis has increased, and fluconazole resistance has become a problem [193-196]. Thus, empirical antifungal agents need to meet the following conditions: 1) appropriate antifungal activity (having susceptibility to prevalent fungi in a region and in a hospital, or at least to Candida and Aspergillus species), 2) acceptable results in randomized controlled studies, 3) recommendations in currently published guidelines, 4) superior tolerance and less adverse reactions, and 5) a reasonable price [197].

To assess the effect of empirical antifungal therapy, the following five composite endpoints are comprehensively considered: 1) resolution of fever during neutropenia, 2) successful treatment of any baseline fungal infection, 3) absence of any breakthrough fungal infection during therapy or within 7 days after completion of therapy, 4) no premature discontinuation of therapy because of drug-related toxicity or lack of efficacy, and 5) survival for 7 days after the completion of therapy. If the therapy does not satisfy any of these endpoints, it is considered to be ineffective [198-200]. When toxicity is observed after initial empirical antifungal therapy, other antifungal agents can be used early in treatment. However, the time to determine whether it is effective remains a controversial issue. For amphotericin B deoxycholate, which is still widely used in Korea, it is recommended to change to other antifungal agents if there is no effect within 3-5 days. When an empirical antifungal agent is changed, a different class of antifungals should be considered first (B-III).

Because amphotericin B deoxycholate has been used as an empirical antifungal for decades and various antifungal agents have since been developed, their efficacy, safety, and adverse reactions have been compared. In Europe and the US, fluconazole, itraconazole, liposomal amphotericin B (not amphotericin B deoxycholate), caspofungin, and voriconazole have been administered empirically (voriconazole is not used as an empirical antifungal agent without the approval of KFDA as of 2009) [14,29,195,201].

Recommended empirical antifungal agents are summarized in Table 7, including doses and characteristics.

Although amphotericin B deoxycholate has been widely used in Korea due to the accumulated experience with its use and its broad antifungal spectrum, the rate of adverse reactions, such as infusion-related toxicity and nephrotoxicity, which is reportedly as high as 50%, is a limitation. ECIL-1 does not recommend amphotericin B deoxycholate in patients 1) with underlying renal impairment, 2) taking co-medications with nephrotoxicity (e.g., taking an immunosuppressant, such as cyclosporin or tacrolimus after allogeneic HSCT, or antibiotics with nephrotoxic potential, such as aminoglycosides), and 3) with a previous history of toxicity. Additionally, current guidelines and reports do not recommend amphotericin B deoxycholate as an empirical agent because of its low efficacy and frequent and severe adverse events [3,14,190,193,195,197,201]. Therefore, although considering the accumulated experiences with its use, broad-spectrum activity, and low price, this committee does not suggest amphotericin B deoxycholate for patients with underlying renal impairment, taking co-medications with nephrotoxicity after allogeneic HSCT, or with a risk of renal failure, such as the elderly (B-I).

Liposomal amphotericin B produces an effect similar to that of amphotericin B deoxycholate (50% vs. 49%), but the frequencies of proven breakthrough fungal infection and adverse reactions (infusion-related toxicity and nephrotoxicity) of the five composite endpoints were significantly lower for liposomal amphotericin B [3,14,190,198,202]. In particular, the frequencies of infusion-related toxicity (17% vs. 44%) and nephrotoxicity (19% vs. 34%) were significantly lower [198]. This committee recommends liposomal amphotericin B for antifungal therapy as A-I.

The overall success rate of caspofungin does not differ from that of liposomal amphotericin B (33.9% vs. 33.7%), but its survival rate for > 7 days after the end of therapy was higher and the mortality rate in underlying fungal infection patients was significantly lower. Moreover, discontinuation of a drug due to drug-related toxicity and adverse reactions was observed less frequently for caspofungin [3,14,190,200]. Thus, this committee recommends caspofungin and liposomal amphotericin B of the same strength (A-I). However, only the intravenous type is available, and it is not effective against Zygomycetes or Cryptococcus species.

Itraconazole in capsule form shows a different absorption rate according to patients, and its absorption rate is improved in the oral and intravenous forms when combined with cyclodextrin. When intravenous itraconazole and amphotericin B deoxycholate were compared, they had similar efficacies and the toxicity of intravenous itraconazole was lower [3,14,190,203,204]. In a recent study, itraconazole provoked fewer adverse reactions and its efficacy was superior to that of amphotericin B deoxycholate, but the time to response and the length of fever were longer [205]. However, its potential for drug interaction should be monitored carefully, and it should be particularly avoided in patients with congestive heart failure.

Although voriconazole shows less breakthrough invasive fungal infection compared with liposomal amphotericin B in neutropenic fever patients (p = 0.02), its overall success rate is lower (26% vs. 31%), "non-inferiority" has not been proven, and it is not approved as an empirical antifungal agent by the FDA [3,14,190,199]. However, it is recommended as a primary therapeutic agent for aspergillosis, with a recommendation strength of A-I, and is used for off-label indications as an empirical antifungal agent due to low rates of breakthrough fungal infection and nephrotoxicity in the US and other countries [193,195,201,206]. Thus, this committee suggests the use of voriconazole as an empirical antifungal agent only in the high-risk group with a high risk of invasive aspergillosis.

Although fluconazole has been reported not to differ from amphotericin B deoxycholate, it has limitations as an empirical antifungal agent because it is not effective against filamentous fungi, such as Aspergillus species, and is used prophylactically [3,14,190,195,207]. A questionnaire survey conducted in HSCT centers of Korea reported that antifungal prophylaxis was administered to 57.6% of anticancer therapy and 90.9% of HSCT patients [29]; thus, fluconazole can be used as an initial empirical antifungal agent for patients not undergoing antifungal prophylaxis and in hospitals with a high rate of Candida infection.

The current trend of prophylaxis with antifungal agents effective against filamentous fungi is thought to largely influence the choice of empirical antifungal agents. When fluconazole or itraconazole is included in antifungal prophylaxis, there is an opinion that it should not be considered an empirical antifungal agent [195]. Future clinical studies to determine which empirical antifungal agents are appropriate according to types of antifungal prophylaxis are expected to be performed.

The Mycoses Study Group of the European Organization for Research and Treatment of Cancer (EORTC/MSG) revised the definition of invasive fungal infection in 2008 to include various methods [208]. However, the definitions were made for communication between clinical researchers, researchers of epidemiological studies, and clinicians, and were not an actual standard practice applied to patients clinically. That is, even when a case does not meet the definition, it does not mean every IFI can be excluded. As clinical characteristics and radiological and laboratory markers have been developed and repeatedly tested, they have been used to identify IFIs and to treat them early. These efforts enhance the survival rate of IFI by treating it early with only a small number of antifungals and reduce unnecessary use, adverse reactions, and medical costs by administering antifungals only to patients in need of them [209]. Additionally, regarding beginning empirical antifungal therapy, there are many efforts to diagnose and treat IFI early and to determine the treatment duration by using one or a combination of repeated fungal cultures, simple X-ray, CT, and non-culture-based diagnostic tests, even when IFI is not clinically suspected [209-211].

In patients with a sustained neutropenic fever, CT is helpful to diagnose IFI, especially invasive aspergillosis. The halo sign or haziness around a nodule or infiltration is a characteristic chest CT finding of infection caused by angio-invasive microorganisms, and it is useful for the suspicion of invasive aspergillosis observed in long-term neutropenic patients; however, it is not found in all patients. Although there is no disease-specific radiological finding characteristic of IFI, empirical antifungal therapy (mold-active therapy) should begin when severe neutropenia persists with symptoms such as fever, cough, and chest pain and pulmonary nodules or infiltration are newly observed on CT images. Additionally, the results of radiological examinations (e.g., CT, MRI) of the abdomen, paranasal sinuses, and head and neck are considered in combination with patients' symptoms and signs and laboratory findings. If there are any abnormalities in the examinations, active efforts, such as bronchoscopy and biopsy, are made to lead to confirmation of the diagnosis and to determine the appropriate use of antifungals through targeted therapy [2,3,12,209].

Galactomannan (GM) is a component of the cell wall of fungi and is used for double-sandwich enzyme-linked immunosorbent assays. It is examined in the plasma, serum, bronchoalveolar lavage fluid, and cerebrospinal fluid, and is known to be helpful for the diagnosis of invasive aspergillosis. Its sensitivity and specificity differ according to the study, and its cutoff value varies according to the type of underlying disease and antifungal prophylaxis [209,212-214]. When antibiotics such as piperacillin/tazobactam or amoxicillin/clavulanic acid are used, a false-positive result may be observed, and antifungals that are currently effective against filamentous fungi can provoke a false-negative result [195,213,215]. Except as an auxiliary measure for diagnosis, GM is used for surveillance in the high-risk group (for early diagnosis before symptoms) and monitoring of responses after antifungal therapy. A recent meta-analysis reported that the sensitivity and specificity of GM analysis were 71% (95% CI, 68 to 74) and 89% (95% CI, 88 to 90), respectively [216], but a study using it clinically found that its sensitivity was < 50% [217]. In particular, because false-negative results can be observed when using antifungals, results should be interpreted carefully.

Moreover, another component of the cell wall of fungi, (1→3)-β-D-glucan, is currently used as an auxiliary measure to diagnose IFI. Unlike GM, it is detected in not only Aspergillus, but also Candida, Fusarium, Trichosporon, P. jirovecii, Acremonium, and Saccharomyces; however, it has not been identified in Zygomycetes [213,215].

Although the polymerase chain reaction (PCR) method has been developed, it has not been standardized and results on clinical validity do not exist. Thus, EORTC/MSG does not recommend it as an auxiliary measure for diagnosis. Despite this, real-time quantitative PCR has been attempted for many genera and species of fungi [208,213,215]. Nucleic acid sequence-based amplification (NASBA) that manufactures a primer with a preserved 18S rRNA base sequence and nucleic acid and performs isothermal amplification has also been developed [218].

Antifungal therapy using described surrogate markers is called "preemptive therapy" and has been recently studied [211], but this committee believes that it is too early to recommend it for routine preemptive therapy in Korea.

When fungal infection is confirmed during the use of empirical antifungal therapy, the treatment duration is determined by any underlying disease(s) and the type and spectrum of fungal infection [2,193,194]. If fungal infection is not confirmed, there is no accurate standard by which to determine the empirical treatment duration, but in previous studies, the duration was an average of 7-14 days (range, 1 to 113) [198-200,203-205]. The treatment duration was limited mainly to experience with amphotericin B deoxycholate. Amphotericin B deoxycholate may be discontinued when neutropenia is improved, the patient's condition is stable, and there are no abnormalities on chest and abdominal CT. If neutropenia persists but the patient's condition is clinically stable, amphotericin B deoxycholate can be discontinued when it has been used for 2 weeks and there are neither suspicious lesions on physical examination nor abnormalities on CT. For stable patients in the high-risk group, continuation of amphotericin B deoxycholate is considered during the period of neutropenia [2]. GM can be helpful to determine the discontinuation of antifungal therapy with no evidence of fungal infection in clinical or radiological examinations despite continuous neutropenia because its negative predictive value is high. On the contrary, when the level is positive, additional examinations, such as CT, are necessary [209,210,213,215]. Moreover, if fever persists even after the recovery of neutropenia, active efforts, such as radiological examinations, biopsy, and cultures, are needed to determine the cause of the fever.