Chapter 47. Approach to Infection in Patients Receiving Cytotoxic Chemotherapy for Malignancy

• Risk of infection increases as the circulating absolute neutrophil count (ANC) declines below 1.0 × 109/L. The greatest risk of bacteremic infection occurs when the ANC is <0.1 × 109/L.
• Cytotoxic therapy for remission-induction therapy for acute myeloid leukemia or conditioning therapy for bone marrow transplantation (high-risk patients) is associated with periods when the ANC is <0.1 × 109/L for 14 to 21 days. The time to marrow recovery (ANC >0.5 × 109/L) can vary from 21 to 42 days.
• Intermittent administration of cytotoxic therapy for solid tissue malignancies or lymphoreticular malignancies (low-risk patients) is often associated with a neutrophil nadir at 10 to 14 days from beginning treatment and with periods of neutropenia (ANC <0.5 × 109/L) of less than 5 to 7 days. This pattern of neutrophil recovery influences the natural history of febrile neutropenic episodes.
• Febrile episodes during neutropenia are defined by an oral temperature of 38.3°C (101°F) or more in the absence of other noninfectious causes of fever such as administration of blood products or pyrogenic drugs (e.g., cytotoxic therapy or amphotericin B), the underlying disease, thromboembolic or thrombophlebitic events, or hemorrhagic events.
• A single neutropenic episode may be characterized by one or more febrile episodes, of which one or more may represent infections.
• Body sites most often associated with infection in the neutropenic patient are those associated with integumental surfaces (skin, upper and lower respiratory tract, and upper and lower gastrointestinal tract).
• Antibacterial prophylaxis with oral agents such as cotrimoxazole, norfloxacin, or ciprofloxacin can reduce the frequency of febrile episodes and bacteremic events in patients with protracted neutropenia.
• Patients undergoing remission-induction for acute myeloid leukemia or bone marrow transplantation with a history of herpetic stomatitis or who are IgG seropositive for herpes simplex virus (HSV) are at risk for severe herpetic mucositis. Such patients should be given acyclovir prophylaxis.
• Empiric antimicrobial therapy for suspected infection in the febrile neutropenic patient usually is composed of a broad-spectrum antibacterial regimen of an anti-pseudomonal penicillin or carbapenem administered as a single agent (monotherapy). Aminoglycoside-based combinations do not add to the efficacy of the single agent, but do add toxicity.
• Neutropenic patients responding to empiric antibacterial therapy generally require at least 5 days for the response to be observed in half the cases. Patients remaining febrile at 5 days should be systematically re-evaluated, while consideration of modification of the antimicrobial regimen can be made at day 7 or 8 unless clinical deterioration is evident. Glycopeptides administered as second-line empiric therapy for persistent fever after 3 to 5 days do not influence time to defervescence or febrile episode-related mortality.

Critical care physicians are often called on to provide metabolic, hemodynamic, and respiratory support for patients with various inherited or acquired defects in host defense that render them susceptible to potentially lethal infections. Patients with single host defense system defects, such as those with inherited immune deficiency syndromes such as congenital agammaglobulinemia, are susceptible to particular encapsulated respiratory pathogens such as Streptococcus pneumoniae that require the presence of opsonizing antibody for clearance. In contrast, cancer patients undergoing potentially curative high-intensity myeloablative cytotoxic therapy acquire defects in multiple host defense systems that lead to increased susceptibility to different groups of pathogens normally contained and controlled by the absent or damaged systems. Four broad categories of defects in host defense are clinically relevant: disruption of the integrity of the integumental surfaces, quantitative neutrophilic phagocyte defects, diminished B lymphocyte (humoral) function, and diminished T lymphocyte system function. A working knowledge of the sources of failure in these host defense systems is particularly important for predicting the types of offending pathogens likely to be involved in the kind of life-threatening infections requiring the services of the critical care team. This, in turn, provides a basis for a rational approach to the choice of antimicrobial therapy. This chapter reviews the approach to managing suspected or proven infection in patients with multiple defects in host defense systems, with a particular emphasis on patients undergoing active myelosuppressive cytotoxic therapy, since this represents the largest group of immunocompromised patients who will require critical care services. Infections in patients with the acquired immunodeficiency syndrome (AIDS) are discussed in Chap. 48, and infections in those with organ or bone marrow transplantation are discussed in Chaps. 73 and 90; the problem of lung infiltrates in immunocompromised patients is covered in Chap. 51.

Hematologists and oncologists have long recognized the existence of the direct relationship between dose and response in cancer therapy. Over the last 10 to 15 years, the supportive care strategies for cancer patients undergoing remission-induction or salvage therapy have improved sufficiently to permit the extension of dosing to the very limits of toxicity and beyond. For many malignant diseases this has translated into significantly higher response rates and disease-free survival. Cure is now a goal that can be adopted realistically for many more patients with these diseases.

Despite these encouraging results, newer dose-intensive therapeutic approaches render patients highly susceptible to life-threatening infection, which in itself or in association with tumor-related end-organ damage may be accompanied by multisystem failure. Whereas some investigators have suggested that admission of cancer patients to the ICU is costly and ultimately futile,1–4 others have challenged this assumption.3,5–7 The outcome for cancer patients admitted to an ICU is directly related to the degree of end-organ damage rather than to the status of the underlying malignancy.6,8,9 The rates of mortality are highest (80% to 100%) when cancer patients admitted to intensive care units (ICUs) have serious damage to three or more organ systems,8,9 or those who develop respiratory failure and require mechanical ventilation.6,7,10−13 Further need for critical care services beyond 14 days has been associated with a 100% hospital mortality rate in a number of studies.8,9 Some factors are additive for ICU-related mortality. For example, a German study reported a mortality rate of 55% in patients with persistent severe neutropenia, 72% for neutropenic cancer patients requiring mechanical ventilation, and 90% for neutropenic cancer patients with invasive fungal infection who require mechanical ventilation.7 The critical care consultant must give careful consideration to these factors to estimate the probability that a critically ill febrile neutropenic patient will survive the episode long enough for the antineoplastic treatments to have their desired effect.

Myelosuppression and Neutropenia

The absolute number of circulating segmented neutrophils (ANC) represents the most important single parameter predictive of the risk for life-threatening infection.14 An ANC of 1.5 to 8.0 × 109/L can be considered normal for adults. As the ANC declines below 1.0 × 109/L the risk of infection increases, with greatest risk for bacteremic infection at neutrophil counts below 0.1 × 109/L. Figure 47-1 illustrates the relationship between the neutrophil count and infection for a series of patients undergoing remission-induction therapy for acute leukemia.

The ANC is calculated by multiplying the proportion of white blood cells (WBCs) that are segmented neutrophils on a Romanovsky-stained blood smear by the total number of WBCs in a specified volume of blood measured in an automated blood cell counter. Since neutropenic patients with acute leukemia undergoing cytotoxic therapy frequently have total WBC counts of <0.5 × 109/L, neutrophils may be difficult to detect on a manually reviewed stained smear; accordingly, the range of error for the procedure increases dramatically. Further, automated blood cell counters may give misleading results when abnormal cells such as leukemic blasts of similar size as segmented neutrophils are present in the circulation. This should dissuade the clinician from relying too heavily on a single ANC to judge the risk of infection. Rather, the clinical relevance of the ANC lies in the recognition of the range associated with a specific infection risk.

The pattern of change of the ANC also has a significant independent influence on infection risk. In one study, 29% of the bacteremic episodes occurred as the neutrophil count was falling but before the ANC fell below 0.5 × 109/L.15 Therefore, with a falling neutrophil count, multiple observations over time are necessary to establish a pattern for the neutrophil profile and to estimate the relative infection risk. Survival of an infection during severe neutropenia is also intimately linked to marrow recovery and recovery of the circulating neutrophil count.16,17 The poorest outcomes for the infectious episodes occur among patients in whom the ANC continues to decline or fails to increase.18,19

The duration of severe neutropenia (ANC <0.5 × 109/L) is also related directly to infection risk. For example, bacteremic infections occur 3.5 times and 5.4 times more often when neutropenia lasts 6 to 15 days and >15 days, respectively.20 The duration of neutropenia is related to the degree of hemopoietic stem cell damage caused by the underlying disease process and by myelosuppressive regimens. Following stem cell suppression, the peripheral neutrophil count falls at a rate inversely proportional to the size of the circulating and marginated peripheral neutrophil pools and the size of the marrow storage pool of mature segmented neutrophils. Marrow recovery follows the recruitment of committed stem cells that are the precursors of granulocytic, monocytic, erythroid, and megakaryocytic cell lines from the resting pluripotential stem cell pool. These hemopoietic stem cells must be able to proliferate to increase their numbers and then differentiate to provide the functional effector cell populations.

Patients receiving pulse doses of chemotherapy for solid tissue malignancies or lymphoreticular malignancies sustain only temporary damage to the hemopoietic stem cell pool. The expected circulating neutrophil nadir generally occurs between days 10 and 14. Although the neutrophil nadirs may be <0.5 × 109/L, the duration of neutropenia is rarely longer than 5 to 7 days (median = 3 to 5 days). For example, a patient receiving cyclophosphamide, doxorubicin (hydroxydaunomycin), vincristine (Oncovin), and prednisone (CHOP) beginning on day 1 of a 21-day cycle of treatment for an intermediate- to high-grade non-Hodgkin's lymphoma might develop a febrile episode on day 12 in association with an ANC of 0.1 × 109/L. The patient's neutrophil count would be expected to have reached its nadir, and a rise in circulating neutrophils would be predicted to occur between days 15 and 21. The likelihood that this prediction is correct is increased if a relative monocytosis is observed on the differential WBC count. The recovery of peripheral blood monocytes precedes that of circulating neutrophils in chemotherapy-induced aplasia and often heralds the recovery of the ANC.

In general, the more dose-intensive myelosuppressive regimens are associated with more hemopoietic stem cell damage and longer durations of neutropenia. Standard remission-induction regimens for acute myeloid leukemia (AML) are composed of anthracycline drugs such as daunorubicin administered in intravenous doses of 30 to 60 mg/m2 daily over 3 days, and an antimetabolite, cytarabine, administered as an intravenous bolus or as a continuous infusion at doses of 100 to 200 mg/m2 daily over 5 to 7 days. These regimens predictably produce periods of profound myelosuppression, in which a median of 24 days passes until the circulating neutrophil count rises above 0.5 × 109/L.21 If more than one cycle of therapy is required to achieve a complete remission, then the median time until marrow recovery may be prolonged by as long as 6 weeks. This additional period of myelosuppression is associated with a significant increase in infectious morbidity.21,22

More intensive regimens using high-dose cytarabine (HDA RAC) in doses 15 to 30 times that for standard induction regimens have been successful for salvage therapy of relapsed or resistant leukemia, for initial remission-induction therapy, and for postremission consolidation therapy for acute leukemia.21 The median time until neutrophil recovery (>0.5 × 109/L) following administration of HDARAC (3.0 g/m2 infused over 1 hour every 12 hours for 12 consecutive doses) is 25 to 30 days. Surprisingly, this period of myelosuppression is not substantially longer than for standard induction regimens. However, the period of risk for infection can be significantly longer (e.g., extending the time to neutrophil recovery to 40 to 60 days) for heavily pretreated patients, in whom there may be more prolonged direct effects of cytotoxic therapy on the marrow stem cell pool or the marrow microenvironment.

Other Immunosuppressive Effects of Cytotoxic Therapy

The remission-induction regimens commonly used for acute leukemia have important immunosuppressive effects in addition to the myelosuppressive effects discussed above. Anthracyclines and similar agents (e.g., doxorubicin, daunorubicin, idarubicin, epirubicin, amacrine, mitoxantrone, and rubidazone), antimetabolites (e.g., cytarabine, methotrexate, thioguanine, and mercaptopurine), and alkylating agents (e.g., cyclophosphamide, ifosfamide, melphalan, busulfan, and platinum analogues) have profound suppressive effects on the numbers of circulating T and B lymphoid cells that parallel the acquired functional defects in cell-mediated and humoral immune mechanisms. At the clinical level, the consequences of these effects are reflected by an increased susceptibility to pathogens normally controlled by these mechanisms. The ultimate impact on immune responsiveness appears to depend on the schedule of administration.

T-Lymphocyte Function

Indications from in vitro testing of lymphoid cell responsiveness to mitogen-induced blastogenesis suggest that T-cell function may be moderately depressed in patients with acute leukemia. Among patients undergoing remission-induction therapy for acute leukemia, decreased cell-mediated immune responsiveness can be detected for up to 6 months following chemotherapy-induced remission.23,24 Immune function has been observed in some patients to decline once again as a herald to relapse.23

The clinical consequences of T-cell dysfunction vary with the underlying disease and the cytotoxic regimen. For example, Pneumocystis carinii infection is an uncommon phenomenon among adult patients undergoing remissioninduction for AML, but relatively common among children undergoing consolidation and maintenance-phase chemotherapy for acute lymphoblastic leukemia (ALL).25 An intermediate degree of risk for pneumocystosis appears to be present in those undergoing bone marrow allografting or autografting. The immunosuppressive potential of the conditioning regimens for bone marrow transplantation (BMT) appears to be greater than that associated with AML induction regimens. Accordingly, most centers managing these patients recommend administering primary prophylaxis for P. carinii in BMT recipients or patients with ALL. These infections rarely occur during the primary period of myelosuppressive therapy–induced neutropenia.

B-Lymphocyte Function

Modern cytotoxic therapy for acute leukemia appears to have a more profound effect on humoral immune competence than on T-lymphocyte function. Serum immunoglobulin concentrations and the efficiency of new antigen-induced immunoglobulin synthesis have been observed to decline following institution of remission-induction therapy, reaching a nadir at approximately 5 weeks. It has been difficult to separate the effects of the underlying malignant disease from the effects of the cytotoxic therapy. There does not appear to be a prognostically useful parameter of T- or B-cell function that predicts infection risk in neutropenic patients that is analogous to the predictive value of the ANC for pyogenic bacterial or fungal infection. However, presence of hypogammaglobulinemia may help identify increased risk of infection by encapsulated bacteria.

Integumental Barriers

Integumental barriers are among the most important and most often damaged defense systems for cancer patients. These barriers include the epithelial surfaces of the skin, the upper and lower respiratory tract, the upper and lower gastrointestinal (GI) tract, and the mucosal surfaces lining the genitourinary tract. In critically ill patients, the barrier function of these surfaces also may be compromised by procedures such as percutaneous intravenous catheterization, endotracheal intubation, endoscopic procedures, nasogastric intubation, and indwelling urinary catheterization (Table 47-1).

Integumental damage secondary to cytotoxic therapy has become more prevalent as the dose intensity of the remission-induction regimens has increased.26 The epithelial surfaces of the GI tract appear to be at greatest risk. The antiproliferative effect of therapy prevents cell recruitment into mucosal areas denuded by erosion or by cellular attrition, resulting in the appearance of superficial erosion and ulceration. The absorptive capacity of the GI mucosa also may be impaired significantly among recipients of regimens such as HDARAC, and both anatomic mucosal disruption and absorptive dysfunction appear to temporally parallel that of the neutrophil profile.

A high proportion of patients receiving cytoreductive therapy also experience painful, often debilitating inflammatory lesions within the oral cavity.27 The tissues of the periodontium, gingival surfaces, oral mucosa, and mucosal surfaces of the upper and lower bowel are affected.26 Cytotoxic regimens affect the developing basal epithelial cells of the oral mucosa in a manner that parallels the effect on the marrow system cell pool and the intestinal mucosal surface.28 Mucosal atrophy, cytolysis, and denudation of the mucosal surface result in the painful foci of local ulceration typically observed 4 to 7 days after the administration of the cytotoxic agents and which usually resolve spontaneously between days 14 and 21.27,29

Cytotoxic therapy–induced intestinal mucosal damage has been described in three stages.28 The first stage of initial injury begins during the first week of cytotoxic therapy and is characterized by replacement of the normal crypts and mucus-secreting goblet cells by atypical undifferentiated cells. The second stage represents progressive mucosal injury that occurs during the second and third weeks. This stage is characterized histologically by cellular necrosis, lack of mitotic activity, and focal loss of villous surfaces, and clinically by abdominal pain, diarrhea, electrolyte loss, and invasive infection. The third stage of cellular regeneration occurs after the third week and is characterized by resumption of mitotic activity and cellular proliferation in the crypts with subsequent repopulation of the denuded surfaces by differentiated cells.

The maximum cytotoxic therapy–induced intestinal epithelial damage occurs in the second week between days 10 and 14.30–32 This corresponds to the median time of onset of bacteremic infection on day 14 due to the microorganisms that normally colonize these surfaces.33 To a limited extent, the type of pathogens recovered in bacteremic infections can be predicted from the pool of microorganisms colonizing damaged mucosal surfaces. Oral mucosal ulceration, particularly that involving periodontal tissues, is often associated with viridans group streptococcal bacteremia.34,35 Colonic mucosal damage is more likely to be associated with aerobic gram-negative bacillary infection with Escherichia coli, Klebsiella species, Pseudomonas aeruginosa, or opportunistic yeasts when these pathogens are colonizing the lower GI tract.33

Mucositis not only predisposes patients to invasive infection, but also imposes a significant cost with respect to the resources needed to manage the consequences of mucositis.36,37 Recent cost estimates suggest that an episode of severe mucositis may cost an average of $7985 (U.S. dollars; USD) (year 2002) per patient (95% CI $6164 to $9806).37 Neutropenia with or without infection is estimated to cost an average of $9316 USD (year 2002) per inpatient (95% CI $8522 to $10,110). In times of fiscal restraint, these complications have great significance for institutions charged with the responsibility for managing cancer patients.

Infections and Pathogens

In a review of bloodstream infections occurring in patients with hematologic malignancies over a 14-year period in a tertiary cancer center in Sweden, gram-negative bacilli accounted for 45% and gram-positive organisms accounted for 55%.38 Coagulase-negative staphylococci accounted for 17.5%; Escherichia coli for 16.3%; viridans group streptococci for 12.7% (with S. mitis being the most common in 50 of 178 isolates); Klebsiella and Enterobacter spp. for 12.7%; Staphylococcus aureus for 8.9%; Enterococcus spp. for 7.3%; and Pseudomonas aeruginosa for 5.5%.38 Of note in this experience was the rising incidence of enterococcal bloodstream infections due to penicillin-resistant E. faecium and the high 30-day mortality (24%) compared to other gram-positive 30-day mortality rates (∼15%).38

The infections documented among febrile neutropenic patients have been classified as microbiologically documented with the identification of a pathogen and a focus of infection; as clinically documented with the identification of a clinical focus of infection without isolation of a putative pathogen; and as an unexplained fever wherein neither a clinical focus nor a pathogen are identified.39 Among febrile neutropenic cancer patients not receiving fluoroquinolone chemoprophylaxis managed during the early 1990s, Cornelissen and colleagues reported that microbiologically-documented infections were observed in 33% of patients, with gram-negative infections comprising 18%, gram-positive infections in 9%, and mixed gram-negative and gram-positive infections in 6%.40 Forty-two percent of patients had clinically documented infections and the remaining 24% had unexplained fevers. Among a similar group of patients who had received ciprofloxacin chemoprophylaxis, there were no gram-negative infections. Gram-positive infections were observed in 38% of patients, clinically-documented infections in 47% of patients, and unexplained fevers in only 15%.

A more recent study wherein the investigators identified all potential clinical foci of infections reported the GI tract as the focus of infection in 41% of patients with the oropharynx accounting for 70%, the esophagus 3%, clinical neutropenic enterocolitis 17%, and perirectal soft tissue infection 10%.41 Other foci included the respiratory tract in 10%, urinary tract in 6%, and skin and soft tissue in 10%. Of the skin foci, indwelling central venous catheter sites accounted for 59%, folliculitis for 6%, and cellulitis for 35%.41 Unexplained fevers accounted for only 10% of cases and microbiologically-documented bloodstream infections were seen in 23% of patients, of which gram-negative bacillemia accounted for 37%, coagulase-negative staphylococcemia 19%, streptococcemia 27%, and other gram-positive microorganisms in 16%.41 These observations illustrate that with clinical diligence, clinical foci of infection may be identified in the majority of febrile neutropenic patients receiving cytotoxic therapy.

Fever is the hallmark of infection for most patients during periods of prolonged severe neutropenia. The definition of what constitutes a febrile episode due to suspected infection in a neutropenic patient has varied greatly in different studies.18−20,34,41−52 The International Antimicrobial Therapy Cooperative Group (IATCG) of the European Organization for Research and Treatment of Cancer (EORTC),20,47,48 the Intercontinental Antimicrobial Study Group,50 and others49 have used an oral temperature of >38°C (100.8°F) sustained over a 12-hour period or a single oral temperature of >38.5°C as the criterion for infection-related fever. The recently published German guidelines define an unexplained fever by a single oral temperature of >38.3°C or >38.0°C lasting over an hour or measured twice within 12 hours.53 The National Cancer Institute of Canada Clinical Trials Group also has used a single oral temperature of >39°C together with chills or rigors instead of a single temperature of >38.5°C.18,54 In order to avoid the administration of antimicrobial therapy to patients with noninfectious causes of fever, a qualifier often has been added to these definitions stipulating that other causes of noninfectious fever should be excluded. Such causes include blood products, pyrogenic drugs such as amphotericin B, thrombophlebitis, or hematoma. The Infectious Diseases Society of America (IDSA) has suggested that a single oral temperature of >38.3°C (101°F) in the absence of other obvious environmental causes would be a reasonably safe working definition for an infection-related fever in neutropenic patients.55

The extent to which characteristics of the febrile episode predict a bacteremic event has been somewhat variable in different studies; however, most agree that initial oral temperatures of >39°C (102.2°F), shaking chills, clinical shock, initial ANC of <0.1 × 109/L, and initial platelet count of <10 × 109/L are to some degree predictive of gram-negative bacteremia.15,20,42,56 Viscoli and associates20 demonstrated an 8.4-fold increase in risk for bacteremic infection in neutropenic patients with initial temperatures of >39°C (102.2°F). The duration of fever prior to evaluation, however, does not appear to influence the risk of gram-negative bacteremia.15

The risk of developing a febrile neutropenic episode during each cycle of outpatient cancer chemotherapy for solid tissue malignancies or lymphoreticular malignancies is generally low.57 However, this risk increases with the number of cycles administered and with the dose intensity of the regimen selected.58 In a study of patients with lymphoma from the M.D. Anderson Cancer Center, the cumulative incidence of febrile neutropenic episodes increased from 12% among recipients of cyclophosphamide, vincristine (Oncovin), and prednisone (COP) to 27% among patients receiving a related regimen where doxorubicin (hydroxydaunomycin) was substituted for the cyclophosphamide (HOP) and 46% among recipients of cyclophosphamide, doxorubicin, vincristine (Oncovin), and prednisone (CHOP).59 In recent studies, the incidence of infection among CHOP recipients has been lower, approximately 15% of all cycles.60 Similarly to the findings of the study by Feld and Bodey,59 the addition of cytotoxic agents to a chemotherapy regimen has the effect of increasing the likelihood of myelosuppression. For example, the addition of paclitaxel or docetaxel to carboplatin for the treatment of gynecologic malignancies increased the incidence of severe neutropenia (ANC <0.5 × 109/L) from <1% to 7% and 73%, respectively.61 However, the addition of irinotecan to carboplatin and paclitaxel increased the incidence of severe neutropenia from 7% to 61%, but did not enhance the odds of developing a febrile neutropenic episode (odds ratio 3.56, 95% CI 0.71 to 17.94).61 The median incidence of febrile neutropenic episodes in patients with small-cell lung cancer is reported as 35% for recipients of cyclophosphamide, doxorubicin, and etoposide (CAE) compared with 18% for recipients of a less myelosuppressive regimen containing cyclophosphamide, doxorubicin (Adriamycin), and vincristine (CAV).62 In addition to the factor of neutropenia, the propensity of a given regimen to induce mucosal damage resulting in mucositis correlates directly with the incidence of febrile neutropenic episodes. The severity of mucositis is directly related to the incidence of febrile events, the duration of hospitalization, the costs of medical care, and treatment-related mortality.36

In contrast, febrile neutropenic episodes occur in 70% to 90% of patients receiving intensive cytotoxic therapy for acute leukemia and bone marrow transplantation.54,63 This difference is due to the more prolonged cytotoxic therapy-induced myelosuppression and greater intestinal epithelial damage in these patients.32

Prolonged neutropenic periods may be punctuated by one or more febrile episodes, and a single febrile neutropenic episode may represent more than one infectious process. For example, a febrile neutropenic episode associated with a viridans group streptococcal bacteremia may not be seen to defervesce promptly despite the administration of appropriate antibacterial therapy and despite the documentation of a microbiologic cure on the basis of subsequent sterile blood cultures. This phenomenon of a persistent febrile state may occur in association with the concomitant administration of pyrogenic blood products, the presence of a coexisting infection such as a herpes simplex virus (HSV) mucositis, or a possible fungal superinfection. It is frequently impossible to distinguish clinically the boundaries defining separate sequential infectious processes by the pattern of fever unless a clear pattern of defervescence is seen between one infection and the next. This is particularly frustrating for clinicians managing febrile neutropenic patients without a defined clinical focus of infection or a defined pathogen.

Diagnosis

Most febrile neutropenic episodes are assumed to represent infection. The diagnosis of a febrile state in a neutropenic patient requires a complete but directed clinical history and physical examination designed to identify potentially infected foci for which those patients are at special risk.

Important historical facts may be obtained from the patient, from significant others, and from the patient's medical record. The physician must verify that the patient is neutropenic, the degree of neutropenia, and when the patient is a recipient of cytotoxic therapy, the day of the chemotherapy cycle. The latter is determined relative to the first day of the last cycle of chemotherapy, or in the case of BMT recipients, the day of the BMT.

To avoid omitting the consideration of other noninfectious causes of fever in neutropenic patients, the clinical evaluation should include questions pertaining to the temporal association of the febrile episode to the administration of blood products, to a history of fever associated with the underlying disease, to the administration of chemotherapeutic agents or amphotericin B, to the presence of thrombophlebitis, and to the possible association of the febrile episode with thromboembolic or hemorrhagic events. For example, in a series of neutropenic patients undergoing remission-induction therapy for acute leukemia,17 26 of 72 (36%) febrile episodes were due to noninfectious causes (blood products in 33%, underlying disease or cytotoxic drugs in 15%, and unexplained causes in 47%). However, the majority of these patients had further febrile episodes for which a diagnosis of infection could not be excluded confidently.

The physical signs of inflammation and infection are influenced by the ANC. The incidence and magnitude of localizing findings such as exudate, fluctuance, ulceration, or fissure formation are reduced in a direct relationship to the ANC.64 Other localizing findings, such as erythema and focal tenderness, appear to remain as useful and reliable signs of infection regardless of the ANC.

The body systems most often involved with infection in neutropenic patients are those associated with integumental surfaces (i.e., the upper and lower respiratory tracts, the upper and lower GI tracts, and the skin).42,64,65 Table 47-2 lists the pertinent historical and physical clues to be sought in the clinical evaluation of a febrile neutropenic patient.

Examination of the head and neck area should include eyegrounds, the external auditory canals and tympanic membranes, the anterior nasal mucosa, the vermilion border of the lips, and the mucosal surfaces of the oropharynx. The eyeground examination should look for retinal hemorrhages as evidence of a bleeding diathesis and retinal exudates (often described as “cotton wool” exudates) that would suggest endophthalmitis associated with disseminated candidiasis. Examination of the external auditory canals and tympanic membranes for erythema or vesicular lesions can implicate this as a focus for infection by respiratory pathogens or herpes viruses. The anterior nasal mucosal surfaces should be examined for ulcerated lesions that might suggest the presence of a local filamentous fungal infection such as that due to Aspergillus species. The skin of the external nares should be examined for vesicular or crusted lesions that would suggest HSV infection. Nasal stuffiness and tenderness over the maxillary sinuses suggests that sinusitis is the infectious problem.

The oropharyngeal examination consists of inspection of the dentition, gingival surfaces, mucosal surfaces of the cheeks, hard and soft palate, tongue surfaces, and posterior pharyngeal wall. The presence of decaying teeth and gingival hyperemia implicates those sites as possible sources of bacteremic infection. The presence of shallow, painful mucosal ulcers on an erythematous base suggests herpes mucositis. Progression of this kind of lesion with local tissue necrosis can suggest a polymicrobic infection due to oropharyngeal anaerobic bacteria (e.g., Fusobacterium nucleatum, Bacteroides melaninogenicus, and peptostreptococci), particularly if cultures for HSV are negative or if such lesions develop during prophylactic or therapeutic administration of acyclovir. Oral thrush or pseudomembranous pharyngitis evolves from an overgrowth of opportunistic yeasts such as Candida species. These lesions are characterized by a thick creamy pseudomembrane consisting of masses of fungi existing in both the yeast and the mycelial phases. The distribution may be patchy, confluent, or discrete. The pseudomembrane is frequently closely adherent to the underlying mucosal surface such that attempts at removal reveal an erythematous or hemorrhagic base. The diagnosis is suspected by the clinical appearance and confirmed by the demonstration of the pathogen in culture and by the microscopic appearance of budding yeasts and pseudohyphae on a Gram's stain or potassium hydroxide preparation.

Chest examination should emphasize evaluation of the lower respiratory tract and central venous catheter sites. The typical signs of pulmonary consolidation may be muted or absent in neutropenic patients; however, localized crepitation on auscultation often precedes the appearance of pulmonary infiltrates radiologically, and thus often represents the earliest clue (and often the only clue) to a developing pneumonia in a neutropenic patient. Purulent sputum is similarly reduced in incidence and amount. Therefore the symptoms of the neutropenic patient with a developing pneumonia may manifest only as febrile illness associated with an increased respiratory rate and a few localized crepitations, with or without an associated cough or radiologic changes.66 The clinician must search for additional differential diagnostic clues such as the origin of the suspected pneumonia (community- or hospital-acquired), the tempo of the illness, the association of the illness with other potentially noninfectious factors such as pulmonary edema, exposure to certain chemotherapeutic agents associated with lung injury (bleomycin, busulfan, or cytarabine), radiation therapy, pulmonary thromboemboli, pulmonary hemorrhage, or hyperleukocytosis. The physical assessment of the chest can do little to differentiate infectious or noninfectious causes of pulmonary findings, but it can help identify the lower respiratory tract as the potential infected focus.

The symptoms and signs of an intra-abdominal infection may be obvious or muted, focal or diffuse. The most important finding is focal tenderness.64 For example, tenderness in the right lower quadrant might suggest neutropenic enterocolitis (typhlitis); right upper quadrant tenderness might suggest a biliary tract focus or hepatomegaly; epigastric pain suggests an upper GI focus; and left lower quadrant tenderness suggests colitis or diverticular disease. It is important to examine the perianal tissues for signs of excoriation, local erythema, swelling, tenderness, fissure formation, or hemorrhoidal tissues, since this area is frequently the site of major life-threatening infection in neutropenic patients. Digital examination of the rectum is not recommended in neutropenic patients because of the additional risk of tissue damage, bleeding, and infection. However, a light perianal digital examination can be informative about focal areas of cellulitis without increasing the risk of bacteremic infection.

The examination of the skin should consist of a thorough search for focal areas of pain, swelling, or erythema, especially in association with indwelling vascular access devices. Particular attention should be paid to the venous insertion, tunnel, and exit sites associated with central venous catheters. In contrast, nonspecific local pocket tenderness may be the only clue to infection associated with the totally implantable venous access port-reservoir systems.

Skin rashes are a common phenomenon among neutropenic patients. The differential diagnosis must include both infectious and noninfectious causes. Among the former group are focal ulcerative and necrotic lesions caused by metastatic pyogenic bacterial infection such as that associated with bacteremic P. aeruginosa or Staphylococcus aureus (infections causing ecthyma gangrenosum), or by disseminated angioinvasive filamentous fungi such as that due to Aspergillus species, Pseudallescheria boydii (formerly called Allescheria boydii; the conidial [asexual] state is Scedosporium apiospermum), or Fusarium species (Fig. 47-2). Pustular erythematous lesions diffusely distributed over the skin surface suggest the possibility of disseminated fungal infection such as that caused by Candida tropicalis. Vesicular skin lesions suggest the possibility of infection due to HSV or herpes zoster virus.

The list of possible noninfectious causes of skin rash is long. The three most important considerations are hemorrhagic petechial or ecchymotic rashes associated with profound thrombocytopenia; hypersensitivity rashes associated with specific drugs such as β-lactam antibacterial drugs, allopurinol, or trimethoprim-sulfamethoxazole (TMP-SMX); and specific chemotherapy regimen–related rash syndromes (e.g., the exfoliative palmar/plantar syndrome associated with high-dose cytarabine; Fig. 47-3). These skin rash syndromes may coexist simultaneously. For example, experience has shown that patients undergoing remission-induction therapy with high-dose cytarabine for AML may develop hemorrhagic petechial rashes due to thrombocytopenia, palmar/plantar erythema, and edema due to cytarabine; macular penicillin-related hypersensitivity rash; and hemorrhagic cutaneous infarcts secondary to angioinvasive bacteremic or fungemic infection, all simultaneously. Only careful attention to the pertinent details of history and thorough laboratory investigations can increase the likelihood that the correct diagnosis and therapeutic decisions will be made.

Once the relevant historical details and physical findings are established, the complete evaluation of the febrile neutropenic patient should include a series of laboratory and radiologic investigations designed to complement the clinical examination. Specimens of body fluids such as blood, urine, cerebrospinal fluid, and lower respiratory secretions should be submitted to the clinical microbiology laboratory for culture and antimicrobial susceptibility testing where appropriate. At least two sets of blood cultures should be obtained, at least one of which should be taken from a peripheral venous site. Furthermore, it has been recommended that for patients with multilumen indwelling central venous catheters, each lumen of the catheter should be sampled in addition to blood from the peripheral venous site.55

The basic radiologic investigation is the chest radiograph (posteroanterior and lateral views). When suggested by clinical clues, sinus radiographs are useful for detecting sinus opacification or fluid levels. Panoramic x-ray radiographs can be helpful for evaluating periodontal infection. High-resolution computed tomographic (HRCT) examinations of the lungs has a high yield of abnormalities in febrile neutropenic patients despite normal or nondiagnostic chest radiographs.67,68 In one study, 60% of febrile neutropenic patients with normal chest radiographs had a pulmonary infiltrate demonstrable on the chest HRCT.67 Computed tomography (CT) of the abdomen or hepatic ultrasonography is valuable for assessing the significance of abnormalities in cholestatic enzymes (γ-glutamyltransferase [GGT] and alkaline phosphatase). This is particularly important if the possibility of hepatosplenic candidiasis exists. Abdominal pain and tenderness with diarrhea in a persistently febrile neutropenic patient suggests the possibility of neutropenic enterocolitis. Abdominal CT examination looking for evidence of bowel wall thickening, pneumatosis, wall nodularity, mucosal enhancement, bowel dilation, ascites, and mesenteric stranding is very useful in making the diagnosis.69

Tissue obtained by biopsy of infected sites can be important in the diagnostic evaluation. Biopsied material should be submitted to the clinical microbiology laboratory for culture and to the pathology laboratory for histopathologic or etiologic evaluation. It is relatively common for pathogens to be observed histopathologically in infected tissue specimens without recovery of the organism by culture. Such is the case in the syndrome of hepatosplenic candidiasis, where budding yeasts and pseudohyphae observed in liver biopsy specimens frequently fail to grow in microbiologic cultures. Skin biopsies are often helpful to distinguish drug-related lesions from those caused by specific pathogenic microorganisms. Although ideally desirable for a complete evaluation, tissue biopsies of suspect sites in neutropenic thrombocytopenic patients carry the additional risks of further infection and bleeding. In addition to skin, the body sites most often considered for biopsy are esophagus (to differentiate mucositis secondary to fungi, pyogenic microorganisms, herpes viruses, or chemotherapeutic agents), liver (to evaluate hepatic lesions due to fungi, herpes viruses, pyogenic bacteria, and drug-related or disease-related hepatic damage), and lung (to evaluate the etiology of progressive infiltrates). Institutions caring for these patients should have a protocol for specimen collection and handling developed in consultation with the appropriate laboratories, the operating theaters, and surgical services.

Risk Assessment

Neutropenia-related febrile episodes are heterogeneous with respect to the cause of neutropenia, the duration of neutropenia, the risks of developing fever, the cause of fever, and the cause of infection. Furthermore, patients differ in their response to treatment and in the risks of complications, including those which are serious and life threatening. Accordingly, the practice standard has been to hospitalize all febrile neutropenic patients for assessment, administration of empiric broad-spectrum intravenous antimicrobial therapy,55 and monitoring for and management of complications. Such complications include management of hemodynamic instability and hypotension; respiratory insufficiency requiring oxygen administration; control of pain, nausea, vomiting, and dehydration; investigation and management of confusion, delirium, and altered mental status; hemorrhage requiring blood product transfusion; cardiac dysrhythmia requiring monitoring; changes in metabolic function requiring intervention; and death.

Investigators from the Dana-Farber Cancer Institute70 examined the natural history of febrile neutropenic patients to identify patients at risk for complications due to the neutropenia, the infection, the underlying cancer, and other comorbid conditions. Concurrent comorbidity included hypotension (defined by a systolic blood pressure of <90 mm Hg), altered mental status, respiratory failure (defined by a partial pressure of arterial oxygen of <60 mm Hg), uncontrolled bleeding, severe thrombocytopenia (defined by a platelet count of <40 × 109/L), inadequate outpatient fluid intake or pain control, suspected spinal cord compression, symptomatic hypercalcemia, the need for hospitalization for induction therapy, serious localized infections, acute abdomen, new deep venous thrombosis, syncope, bowel obstruction, and poor performance status (Karnofsky performance status of <50%). Medical complications included hypotension, respiratory failure, new cardiac dysrhythmia or electrocardiographic changes, altered mental status, new focal neurologic abnormalities, hemorrhage requiring >3 units of packed red blood cell transfusion within 24 hours, congestive heart failure, emergency surgery, acute abdomen, pulmonary thromboembolism, acute renal failure, and diabetic ketoacidosis. Based on these comorbidities and complications, febrile neutropenic patients could be classified into three groups at high risk for complications and one low-risk group. Group 1 (39% of the total) was comprised of hospitalized patients usually with hematologic malignancies or hemopoietic stem cell transplant. Complication and morbidity rates were 34% and 23%, respectively. Group 2 (8% of the total) was comprised of outpatients with concurrent comorbidity and had complication and mortality rates of 55% and 14%, respectively. Group 3 (10% of the total) was comprised of outpatients with as yet uncontrolled or progressive cancer and had complication and mortality rates of 31% and 15%, respectively. Group 4 (the low-risk group, 43% of the total) was comprised of outpatients with controlled or responding cancer and no comorbid processes. This group had a complication rate of 2% and no deaths. These observations were prospectively validated in follow-up studies.70,71 These results suggest that high-risk patients with characteristics corresponding to groups 1 to 3 should be admitted and managed as inpatients with careful monitoring for serious complications, whereas low-risk patients corresponding to group 4 can be managed safely and effectively on an outpatient basis.71–79

The Multinational Association for Supportive Care in Cancer developed a scoring system to identify low-risk patients for serious medical complications that would require admission to the hospital.26,80 Identifying factors included absence of symptoms, hypotension, airflow obstruction, hematologic malignancy, invasive fungal infection, or dehydration. Furthermore, status as an outpatient at the onset of the febrile neutropenic episode and age <60 years were also identifying factors. A score of >21 of a possible 26 identified low-risk patients with a positive predictive value of 91%, specificity of 68%, and sensitivity of 71%. This system has been offered as a strategy for identifying patients eligible for studies of more cost-effective, safe, outpatient-based management strategies.

Empiric Antimicrobial Therapy

The empiric initial therapy for suspected infection in febrile neutropenic patients is based on three assumptions:

1. The majority of infections are due to bacteria.81

2. The principal pathogens are aerobic gram-negative bacilli (E. coli, Klebsiella pneumoniae, and P. aeruginosa).42,81

3. Inappropriate therapy for aerobic gram-negative bacillemia is associated with high mortality with a median survival of less than 72 hours.82

Accordingly, the antibiotic regimens that have been recommended for empiric first-line therapy of first fever in neutropenic cancer patients are designed to have excellent activity against these pathogens. With the recognition of various factors favoring infection by gram-positive organisms,83 the addition of agents with an improved spectrum of activity against these pathogens when appropriate also has been recommended.84

Empiric antibacterial therapies may be administered either as combination regimens or as single-agent regimens (monotherapy) (Tables 47-3 and 47-4). Combination regimens include β-lactam agents combined together18,85,86 (double β-lactam regimens), combined with aminoglycosides,42−44,47,49,50,87 or combined with fluoroquinolones.41,88 Single-agent regimens consist of β-lactam agents48−50,52,89−92 with or without β-lactamase inhibitors (tazobactam, clavulanic acid, or sulbactam) or fluoroquinolones.93–95 Monotherapy with aminoglycosides is not recommended.55

β-Lactam antibacterial agents may be categorized as extended-spectrum antipseudomonal penicillins (e.g., carbenicillin, ticarcillin with or without clavulanic acid, piperacillin with or without tazobactam, azlocillin, or mezlocillin), third- or fourth-generation antipseudomonal cephalosporins (e.g., moxalactam, ceftriaxone, ceftazidime, cefoperazone with or without sulbactam, cefpirome, or cefepime), or as carbapenems (e.g., imipenem-cilastatin, meropenem, or ertapenem). The addition of β-lactamase inhibitors to the broad-spectrum antipseudomonal penicillins, ticarcillin and piperacillin, enhances their spectrum of activity against β-lactamase–producing bacteria and their roles in the management of the febrile neutropenic patient.47,51,96−115

In a recent review of physician prescribing behavior for 214 febrile neutropenic patients in Canadian centers, initial empiric therapy with a single agent was administered in 42% of cases in which a third-generation cephalosporin such as ceftazidime was used in 32%, a carbapenem in 2.3%, and a fluoroquinolone in 0.9%.116 Combination therapy was administered in 58% of cases in which an antipseudomonal penicillin plus an aminoglycoside was given in 29% of cases, an antipseudomonal cephalosporin plus an aminoglycoside in 15%, and an antipseudomonal β-lactam plus a glycopeptide in 11%.116Vancomycin was part of the initial empiric antibacterial therapy in 15% of cases. First modification with second-line therapy for persistent fever was administered in 87% of the 214 cases after a median of 5 days. Systemic empiric amphotericin B was administered for persistent fever in 48% of the 214 febrile neutropenic episodes after a median of 9 days. Previous studies have demonstrated that glycopeptides are employed as empiric second-line therapy in 40% to 50% of cases in which extended-spectrum cephalosporins are administered as first-line empiric therapy.48,51,97−100,117−128

The Role of Aminoglycosides in Treatment of Febrile Neutropenic Patients

Aminoglycosides were part of the standard combination empiric antibacterial therapy for the management of febrile neutropenic patients from the early 1970s to the 1990s. The combination of an aminoglycoside with an antipseudomonal β-lactam antibacterial agent was designed to have a broad spectrum of antibacterial activity, achieve bactericidal serum concentrations, exert a synergistic antibacterial effect, and prevent emergence of resistance. Such combinations have been recommended in published guidelines by the Infectious Diseases Society of America (IDSA),55 the National Comprehensive Cancer Network,129 and the Infectious Diseases Working Party of the German Society of Hematology and Oncology,53 but not the Spanish guidelines, however.130 The aminoglycoside antibiotics having proven roles for use with β-lactam antibiotics in neutropenic patients include gentamicin, netilmicin, tobramycin, and amikacin. The choice of aminoglycoside must be based on local institutional bacterial susceptibility patterns, the availability of a mechanism for serum aminoglycoside concentration monitoring, and drug cost.

A recent large randomized controlled trial from the University of Perugia, Italy, compared piperacillin-tazobactam to piperacillin-tazobactam plus amikacin.51 The primary outcome was defervescence of all signs and symptoms of infection without modification of the initial antibacterial regimen. Response was observed in 179 of 364 (49%) monotherapy recipients and in 196 of 369 (53%) combination recipients (p = 0.2). The response rates in single-pathogen gram-positive bacteremias were low (27% and 32 %, respectively) because of the high proportion of coagulase-negative staphylococcal bacteremias (response rates 17% and 18%, respectively; p = 0.8). In contrast, the response rates for streptococcal and enterococcal bacteremias between the two groups were significantly higher (60% and 71%, respectively, p = 0.7). The response rates for single gram-negative bacteremias were also similar (36% and 34%, respectively; p = 0.9). The aminoglycoside failed to enhance the response rates in any circumstance. Studies such as this question the therapeutic value of combination antibacterial therapy with an aminoglycoside.

Two systematic reviews of the literature have examined the safety and efficacy of β-lactam plus aminoglycoside combinations in febrile neutropenic patients.131,132 Furno and associates reviewed 4795 heterogeneously treated febrile neutropenic episodes entered into 29 randomized controlled clinical trials comparing monotherapy (ceftazidime, 9 trials; cefepime, 2 trials; cefoperazone, 1 trial; imipenem-cilastatin, 9 trials; meropenem, 4 trials; ciprofloxacin, 2 trials; and ofloxacin, 2 trials) and aminoglycoside-based combination therapy. The pooled odds ratios for overall treatment failure and for treatment failure in bloodstream infections were 0.88 (95% CI, 0.78 to 0.99) and 0.70 (95% CI 0.54 to 0.92), respectively, demonstrating fewer failures in the monotherapy groups.131 Paul and colleagues examined 7807 febrile neutropenic patients entered into 47 randomized controlled trials comparing β-lactam monotherapy to β-lactam plus aminoglycoside combination therapy. The main outcome was overall mortality. While there was no significant difference in overall mortality (relative risk 0.85; 95% CI 0.72 to 1.02), there were fewer failures among β-lactam monotherapy recipients (relative risk 0.92; 95% CI 0.85 to 0.99).132 Monotherapy recipients also experienced fewer adverse events overall (relative risk 0.85; 95% CI 0.73 to 1.00) and less nephrotoxicity (relative risk 0.42; 95% CI 0.32 to 0.56).132 On the basis of these analyses, β-lactam plus aminoglycoside combinations offer no advantages over broad-spectrum β-lactam monotherapy. Rather they present significant disadvantages with respect to toxicity and costs related to drug monitoring and administration. Based on these observations and considerations, investigators are now recommending that the standard of care for initial therapy of febrile neutropenic cancer patients should be with a broad-spectrum β-lactam agent as monotherapy.132

Fluoroquinolones in the Treatment of Febrile Neutropenic Patients

The fluoroquinolones evaluated in studies of the empiric treatment of febrile neutropenic patients include ciprofloxacin,41,133 perfloxacin,134ofloxacin,73,74levofloxacin,135 and clinifloxacin.94,95 These agents have the advantage of availability in both oral and intravenous formulations, which can facilitate intravenous-to-oral conversion.133,136,137

The studies of empiric fluoroquinolones as first-line therapy for febrile neutropenic patients have largely targeted those patients at lower risk for medical complications.129Ciprofloxacin as well as other agents such as clindamycin, aztreonam, or amoxicillin have been the most completely studied.72,73,75−78,108,135,138−142 The administration of ciprofloxacin 750 mg orally and amoxicillin-clavulanate 625 mg orally both every 8 hours was well tolerated and as effective as intravenous ceftriaxone plus amikacin138 or ceftazidime139 administered on an inpatient basis. Similar results have been reported for trials comparing oral ciprofloxacin-based regimens and intravenous regimens on an outpatient basis.72,73,75−78,108,135,138−142 These strategies appear to be safe and effective for low-risk patients. The National Comprehensive Cancer Network Clinical Practice Guidelines in Oncology have included as an alternative regimen for initial empiric antibacterial therapy of higher-risk febrile neutropenic cancer patients the use of ciprofloxacin plus an antipseudomonal penicillin.129 One large trial compared piperacillin plus ciprofloxacin and piperacillin plus tobramycin in a group of intermediate- to high-risk febrile neutropenic cancer patients.41 Success rates (i.e., defervescence with initial regimen modification) were similar, 27% and 22% for the piperacillin-ciprofloxacin and piperacillin-tobramycin groups, respectively; however, times-to-defervescence were faster among the piperacillin-ciprofloxacin recipients, 5 days versus 6 days (p = 0.005).

Reports of fluoroquinolone resistance among aerobic gram-negative bacilli associated with bloodstream infections in neutropenic cancer patients began to emerge in the early 1990s.143–145 The incidence of fluoroquinolone-resistant Escherichia coli (FREC) bacteremia among patients treated on the EORTC-IATCG clinical trials from 1983 to 1993 increased from zero during the period from 1983 to 1990 to 28% during the period from 1991 to 1993.143 Of note, the incidence of resistance among strains of Pseudomonas aeruginosa and Klebsiella pneumoniae remained largely unchanged at less than 10%.143 This has been more of a problem among cancer patients treated in institutions with high prevalence rates for gram-negative bacillary fluoroquinolone resistance (>10%) despite community-related resistance prevalences of less than 1%.146 Carratala and colleagues reported in 1995 a 37% incidence of fluoroquinolone resistance in 35 of 230 neutropenic cancer patients previously treated with norfloxacin.145

Fluoroquinolone resistance among community-derived gram-negative bacilli, and FREC in particular, has emerged in parallel with the increased prescribing of these products in the community.147–149 There is a significant correlation between the incidence of ciprofloxacin-resistant Escherichia coli bloodstream infection and the increased community and hospital use of fluoroquinolones (r = 0.974, p = 0.005 and r = 0.975, p = 0.005, respectively).147 Among those who had not heretofore received fluoroquinolone therapy, Garau and colleagues reported the prevalence of FREC in the stools of adults and children in the Province of Barcelona, Spain, to be 24% and 26%, respectively.149 These investigators also observed a very high prevalence of FREC in the stools of pigs (45%) and poultry (90%) and argued that the increased prevalence of human carriage of FREC may be linked to the high prevalence in animal-based food products. The increased use of fluoroquinolones in animal feeds and in humans is believed to play a role in the selection for FREC. In an environment with a high (15%) prevalence of fluoroquinolone resistance among gram-negative bacilli, cancer patients receiving ciprofloxacin antibacterial chemoprophylaxis while undergoing high-dose chemotherapy with stem cell rescue became colonized with FREC in one third of cases,150 a phenomenon that increases the likelihood of FREC bloodstream infections in such patients.151

In addition, inappropriate use of fluoroquinolones is common. A recent case-control study of fluoroquinolone use in emergency departments in Philadelphia demonstrated inappropriate prescription of these agents in 81% of cases.152 Lastly, gram-negative bacilli that are co-resistant to fluoroquinolones and other antibacterials such as aminoglycosides, cefotaxime, ampicillin, amoxicillin-clavulanate and trimethoprim-sulfamethoxazole are more frequently being reported.149,153 Thus overuse and inappropriate use of fluoroquinolones in the community is common and is strongly linked to resistance, and ultimately will reduce the likelihood that this class of agents will be useful in a variety of patient populations, including neutropenic cancer patients, who require critical care services.143,145 It is incumbent upon prescription writers of empiric antibacterial therapy to have some understanding of the prevalence of gram-negative bacillary resistance in the institution and the community.

Double Beta-Lactam Combinations

Regimens consisting of an antipseudomonal broad-spectrum β-lactam plus an extended-spectrum third-generation cephalosporin are relatively safe and effective alternatives to regimens of a β-lactam plus an aminoglycoside,18,85,86 but they are costly and do not offer any advantages over those of broad-spectrum β-lactam monotherapies. Cost, hypokalemia, selection of bacterial resistance, and antagonism have been cited as potential disadvantages, although these regimens may have an occasional role in the setting of preexisting renal insufficiency, in which the patient is receiving concomitant nephrotoxic agents such as cyclosporine or cisplatin-containing chemotherapeutic regimens, or where gram-positive organisms such as viridans streptococci are suspected.18,54 The availability of effective β-lactam and fluoroquinolone or β-lactam monotherapy-based regimens has largely rendered double β-lactam regimens obsolete. Of note, these combinations appear in the National Comprehensive Cancer Network guidelines.129

Extended-Spectrum Cephalosporins

The intrinsic activity of many of the third- and fourth-generation cephalosporins such as ceftriaxone, cefoperazone, ceftazidime, or cefepime against aerobic gram-negative bacilli is high. These agents have been used effectively as single agents for empiric treatment of suspected infection.42−45,47−50,87,92,154 Empiric monotherapy has been shown to be effective in both low- and high-risk febrile neutropenic patient populations and in those in whom the expected duration of severe neutropenia (ANC <0.5 × 109/L) is either longer or shorter than 7 days. The experience to date suggests that single-agent empiric regimens will require modification, usually by the addition of other antimicrobials, in one third to one half of patients with neutropenic periods in excess of 1 week.48−50,87 There is a lower likelihood that modifications will be necessary for patients with short-term (<1 week) neutropenia.

A meta-analysis examining the efficacy of ceftazidime monotherapy compared to standard combination therapy for the empiric treatment of febrile neutropenic patients failed to demonstrate a difference with respect to the odds of overall treatment failure and failure in bloodstream infections (odds ratio 1.27; 95% CI 0.79 to 2.03 in 1077 patient episodes; and OR 0.72, 95% CI 0.33 to 1.58 in 248 patient episodes, respectively).92 Another meta-analysis154 compared the efficacy of ceftriaxone with (7 trials) or without (1 trial) an aminoglycoside and ceftazidime with (6 trials) or without (1 trial) an aminoglycoside or azlocillin plus an aminoglycoside (1 trial). The pooled odds ratio for overall treatment failure in the eight trials was 1.04 (95% CI 0.84 to 1.29), demonstrating no differences in the comparison for this outcome. There was also no difference in overall mortality (OR 0.84; 95% CI 0.57 to 1.24). These analyses demonstrate that empiric antibacterial therapy of febrile neutropenic patients with once-daily ceftriaxone is as effective as thrice daily ceftazidime. This has important implications for potential outpatient once-daily intravenous therapy.

Carbapenems for Treatment of Febrile Neutropenic Patients

Both imipenem-cilastatin and meropenem have been studied widely as empiric therapy in febrile neutropenic patients. A meta-analysis examined the efficacy of imipenem-cilastatin compared to a β-lactam plus aminoglycoside combinations (11 trials) and to β-lactam monotherapy (ceftazidime, 4 trials; cefoperazone-sulbactam, 1 trial), or β-lactam plus glycopeptide (ceftazidime, 3 trials) or double β-lactam therapy (cefoperazone-mezlocillin, 1 trial; cefoperazone-piperacillin, 1 trial). There were fewer treatment failures among imipenem-cilastatin recipients compared to β-lactam plus aminoglycoside combinations (OR 0.77; 95% CI 0.61 to 0.98) or to non–aminoglycoside-containing regimens (OR 0.67; 95% CI 0.54 to 0.84).91 These analyses support the superiority of imipenem-cilastatin over control arms largely based on third-generation cephalosporins.

Another meta-analysis comparing carbapenem monotherapy to β-lactam plus aminoglycoside combination therapy demonstrated fewer treatment failures among the carbapenem recipients (OR 0.80; 95% CI 0.66 to 0.96, Peto fixed effects model),131 in contrast to antipseudomonal cephalosporin monotherapy (OR 0.92; 95% CI 0.77 to 1.10, Peto fixed effects model). A similar analysis of the published clinical trials of meropenem monotherapy compared to ceftazidime-based regimens with or without an aminoglycoside demonstrated greater response rates among meropenem recipients (OR 1.28; 95% CI 1.08 to 1.51).155 These systematic observations suggest the inferiority of third-generation cephalosporin monotherapy regimens compared to the carbapenem monotherapy.

Piperacillin-Tazobactam for the Treatment of Febrile Neutropenic Patients

Piperacillin-tazobactam plus amikacin therapy was successful in 210 of 342 (61%) febrile neutropenic patients studied by the European Organisation for the Research and Treatment of Cancer, compared to 196 of 364 (54%) patients receiving ceftazidime plus amikacin (p = 0.05).47 Furthermore, the time-to-defervescence was shorter among piperacillin-tazobactam recipients (p = 0.01), and the frequency with which the initial regimen was modified by the addition of a glycopeptide was lower (24% versus 35%; p = 0.002) in that study. Another large Italian trial examined the role of the aminoglycoside amikacin, when combined with piperacillin-tazobactam.51 The response rates overall between the two groups were similar regardless of the classification of the febrile neutropenic episode as bacteremic, clinically documented, or unexplained fever. Clearly, the aminoglycoside offered no advantage over the piperacillin-tazobactam monotherapy. Piperacillin-tazobactam monotherapy has been studied compared to extended-spectrum cephalosporins with or without aminoglycosides.96–100 Overall unmodified success rates were significantly higher among piperacillin-tazobactam recipients (OR 1.39; 95% CI 1.08 to 1.79).100 Furthermore, the frequency of second-line therapy with glycopeptides was significantly lower among piperacillin-tazobactam recipients (OR 0.77; 95% CI 0.60 to 0.99).100 These observations demonstrate the superiority of initial piperacillin-tazobactam monotherapy over extended-spectrum cephalosporins with or without aminoglycosides. Furthermore, the need to modify the initial antibacterial regimen by the addition of a glycopeptide such as vancomycin or teicoplanin for persistent fever is also significantly less than for the cephalosporin-based comparator groups. This has important implications with respect to cost, drug-related toxicities, and for selection for resistant microorganisms such as vancomycin-resistant Enterococcus spp.

The Role of Glycopeptides as First-Line and Second-Line Therapy in Febrile Neutropenic Patients

There has been a significant increase in the proportion of the microbiologically documented infections observed in neutropenic patients that are due to gram-positive bacteria. Two decades ago approximately 70% of bacteremic isolates were gram-negative bacilli, whereas more recently approximately 60% to 70% are gram-positive cocci.41,83 Previously published clinical trials of ceftazidime-based empiric antibacterial therapy observed more fatal gram-positive superinfections than expected.156 A subsequent small study demonstrated that the addition of vancomycin to ceftazidime therapy reduced the superinfection rate from 24% to zero and the infection-related mortality rate by 91% (OR 0.07; 95% CI 0.01 to 0.63).157 Such observations have led investigators to advocate the inclusion of glycopeptides vancomycin or teicoplanin as part of the first-line empiric antibacterial therapy in febrile neutropenic patients. In contrast, two further studies from the National Cancer Institute suggested that the vancomycin therapy may be safely delayed without increased morbidity or mortality.49,158 In order to examine this question further, the European Organization for Treatment and Research in Cancer and the National Cancer Institute of Canada Clinical Trials Group conducted a large randomized trial comparing empiric therapy with ceftazidime plus amikacin (CA) and ceftazidime plus amikacin plus vancomycin (CAV).45 The overall response rate among the 377 CAV recipients was 288 (76%), whereas the response rate for 370 CA recipients was 232 (63%; p <0.001), the difference largely due to differential response rates for gram-positive bloodstream infections among CAV recipients. Despite the apparent superiority of the CAV arm, persistently febrile CA patients at day +3 received vancomycin, resulting in no differences in overall success rates or mortality.

Treatment failures were due to lack of prompt response and to persistent fever observed at a time very early after the initiation of the allocated regimen rather than objective indicators of failure such as persistence of resistant pathogens or progression at foci of infection. Patients receiving CA were more likely to receive vancomycin with persistence of fever at day +3 than recipients of the CAV regimen.159 Since the median time to defervescence among high-risk febrile neutropenic patients is day 5 (range = day +3 to day +7),41,47,48,50,128 many patients were considered failures unnecessarily because of regimen modification before they would have had a chance to defervesce. The physician compulsion to modify the initial antibacterial regimen for reasons of persistent fever at day +3 is driven in part by previously published protocols49 reinforced by previously published guidelines.160

Three other trials161–163 examining the role of initial glycopeptide therapy in febrile neutropenic cancer patients came to similar conclusions as for the EORTC/NCIC trial;45 that is, the inclusion of glycopeptides as initial therapy in febrile neutropenic patients results in modest improvements in response rates, particularly in the circumstances of gram-positive bacteremia, but has no impact on overall survival of the neutropenic episode. Accordingly, it is recommended in the recent American, German and Spanish guidelines that glycopeptides not be used as part of the initial empiric antibacterial regimen for the treatment of fever in neutropenic cancer patients unless there is evidence of gram-positive infection.53,55,130,164

The above trials demonstrated that among patients with persistent fever due to gram-positive infection at day +3 vancomycin could be added to the regimen at that time with no excess morbidity or mortality.45,49,158,163 Accordingly, second-line glycopeptide-based empiric antibacterial therapy for persistent fever has become quite common. As noted above, almost half of cases enrolled in clinical trials have a glycopeptide administered for these reasons.48,51,97−100,117−128 The efficacy of empiric second-line glycopeptide therapy has been studied in two randomized controlled trials.101,165 Both studies failed to demonstrate a significant treatment effect with regard to defervescence or overall mortality (OR 1.05; 95% CI 0.66 to 1.67 and OR 0.80; 95% CI 0.33 to 1.90, respectively) compared to placebo.166 The results of these two trials together with the small published meta-analysis confirm that in stable, persistently febrile, neutropenic patients, empiric second-line glycopeptide therapy after 48 to 96 hours is unnecessary.166 Furthermore, the initial antibacterial therapy can be safely extended in such patients unmodified for an additional 72 to 96 hours for a total of 4 to 8 days.

The use of glycopeptide antibiotics as part of the initial first-line regimen should be reserved for patients at highest risk for serious gram-positive infection.55,129 Such circumstances include clinical catheter-related infection, infection in patients receiving fluoroquinolone-based antibacterial chemoprophylaxis associated with severe mucositis predisposing patients to viridans group streptococcal bloodstream infections,34,35,167 infection in the setting of colonization by methicillin-resistant Staphylococcus aureus (MRSA), a bloodstream isolate characterized as gram-positive cocci in groups and clusters (suggesting Staphylococcus spp. and the likelihood of a methicillin-resistant coagulase-negative Staphylococcus), and the setting of hypotension or septic shock syndrome without an identified pathogen. The caveat is that the glycopeptide antibiotic should be discontinued in 2 to 3 days if a resistant gram-positive infection has not been identified.

The Role of Hematopoietic Growth Factors in the Management of Febrile Neutropenic Patients

The inclusion of hematopoietic growth factors (HGF), granulocyte and granulocyte-macrophage colony-stimulating factors (G-CSF; GM-CSF) in strategies for the prevention and management of febrile neutropenic patients remains unsettled.168 Berghmans and colleagues published a systematic review of the literature with a meta-analysis to critically examine the evidence for or against use of these products in the treatment of febrile neutropenic episodes.169 Eleven studies encompassing 1218 febrile neutropenic episodes published between 1990 and 1998 were considered in the analysis. Although six trials reported treatment effects for the primary outcome analyzed (mortality), overall there was no effect on mortality (relative risk 0.71; 95% CI 0.44 to 1.15). Furthermore, there were no differences when analysis was conducted by HGF and G-CSF versus GM-CSF. The impact of HGF on length of hospitalization was decreased in four trials, unaffected in four trials, and not reported in three trials. Only three of six trials reporting on the impact of HGF on duration of antibacterial therapy demonstrated a reduction in this outcome. In nine trials in which the impact of HGF on the duration of fever was reported, no treatment effects were observed in seven trials, fever reduction was observed in one trial, and no analysis was provided in one other. The studies included in this analysis were flawed by the failure to control for important variables, including the duration and intensity of neutropenia, type of cytotoxic therapy, and type of malignancy. On the basis of this review, Berghmans and colleagues could not recommend routine use of HGF in the treatment of febrile neutropenic episodes.169

The guidelines of the American Society of Clinical Oncology do not recommend their use as adjunctive therapy for neutropenic patients who are febrile and have suspected infection.170 However, less than half of specialists follow these guidelines, given that usage appears to be influenced more by reimbursement than evidence.171–173

HGF have been evaluated in other non-neutropenic patient populations including patients with community-acquired pneumonia, human immunodeficiency virus–infected patients, neonatal sepsis, diabetic foot ulcer infections, acute hepatic failure or cirrhosis, in orthotopic liver transplantation, and in the critical care unit. While there appears to be some promising results in some subgroups of patients, HGF have no proven clinical benefit in non-neutropenic critically ill patients with regard to morbidity or mortality.174

The Role of Second-Line Therapy for Persistent Fever in Neutropenic Patients

Patients with fever persisting beyond 72 hours on broad-spectrum antibiotic therapy should be re-evaluated carefully.55,129Table 47-5 lists several of the possible explanations for this. A nonbacterial etiology for the febrile episode should be considered. The various noninfectious causes have been discussed already. Factors such as localized tenderness, change in sensorium, hyperventilation, hypotension, progressive renal insufficiency, and acidosis suggest an infectious cause. The re-evaluation should occur between day 3 and day 5 and include a thorough examination to identify a focus. Cultures of blood (one set from each lumen of the central venous catheter and one set from a peripheral vein), urine, and other potentially infected sites should be submitted to the clinical microbiology laboratory. Repeat chest radiography or diagnostic imaging studies such as ultrasonography or high-resolution computed tomographic studies may be performed when abnormalities suggest a specific organ as a potential site for infection. When re-evaluation fails to identify the etiology of the persistent fever, the clinician may elect either to continue the initial empiric antibacterial regimen if the patient's condition shows no clinical change or deterioration, or to modify the empiric regimen appropriate to the findings of the re-evaluation (Table 47-6).

Persistently febrile severely neutropenic patients despite administration of at least 3 to 5 days of broad-spectrum antibacterial therapy are at risk of having an invasive fungal infection as the cause for the ongoing fever.55 Such patients are candidates for empiric antifungal therapy with amphotericin B deoxycholate,175,176 a lipid formulation of amphotericin B,177 an echinocandin, or with an extended-spectrum azole.178,179 The two randomized clinical trials upon which the practice of empiric antifungal therapy was based failed to demonstrate a treatment effect with respect to defervescence compared to an untreated control group (OR 0.13; 95% CI 0.01 to 1.26 for the study by Pizzo and colleagues175 and OR 0.14; 95% CI 0.02 to 1.23 for the EORTC study176).

While it is widely felt that the majority of persistent fevers in the setting of severe neutropenia do not represent invasive fungal infections, approximately 20% may be related to invasive fungal infection with an attendant high mortality rate sufficient to compel physicians to administer empiric antifungal therapy. Such strategies may be considered in patients persistently febrile after 7 days of broad-spectrum antibacterial therapy. Patients should undergo a second work-up for persistent fever to investigate the possibility of invasive fungal infection which includes blood cultures from peripheral and indwelling central venous access sites and sensitive HRCT images of the chest.67,68 In 60% of persistently febrile neutropenic patients with normal or nondiagnostic chest roentgenograms the HRCT scan of the chest demonstrated a pulmonary infiltrate.67 Such procedures can permit the earlier diagnosis and management of invasive mold infections involving the lungs by approximately 1 week.180

Surveillance cultures for detecting potential fungal pathogens have had some limited usefulness for predicting fungal infection. The recovery of filamentous fungi (e.g., Aspergillus species) in a nasopharyngeal surveillance culture in the clinical setting of a persistently febrile, profoundly neutropenic patient receiving broad-spectrum antibiotics and who develops new focal pulmonary infiltrates are to some extent predictive of Aspergillus pneumonia.181 The recovery of Candida albicans from oropharyngeal, rectal, or urine surveillance cultures has had a positive predictive value of 10% to 15% for systematic candidiasis. The recovery of other non-albicans Candida species such as Candida tropicalis, particularly from multiple sites, has had a positive predictive value of >70% for systemic infection. In contrast, the failure to recover Candida species in surveillance cultures has been associated with a negative predictive value of >90% for invasive disease from C. albicans or C. tropicalis.182 This experience suggests that clinicians cannot use surveillance cultures to predict the presence of a candidal infection (except perhaps for C. tropicalis). However, the clinician may be reassured by negative surveillance cultures that are properly obtained and processed, that antifungal therapy may not be indicated.55

Surrogate molecular markers of yeasts and molds are showing promise in the diagnosis of invasive fungal infection. Galactomannan is a component of the cell wall of Aspergillus spp., certain dematiaceous fungi such as Alternaria spp., and some Penicillium spp.,183–185 and has been used to aid in the diagnosis of invasive aspergillosis.180,186 A enzyme-linked immunosorbent serum assay which uses a rat monoclonal antibody EB-A2 that targets the D-galactofuranoside side chains of the Aspergillus spp. galactomannan187 has been developed for the diagnosis of invasive aspergillosis.188 The antibody may cross react with galactomannan-like materials from other molds such as Penicillium spp., Paecilomyces spp., and Alternaria spp.,189 or from foods that are exposed to molds originating from the soil during growth or harvesting, such as rice, pasta, cereals, and vegetables,190 and even milk191 consumed by premature infants.192 Translocation of dietary antigens into the blood of healthy adults is well documented.193 The antibody also cross-reacts with Bifidobacterium spp. lipoteichoic acid, an organism that heavily colonizes the gut of neonates and infants.194 Translocation of Bifidobacterium spp. has occurred in the setting of reduced integrity of the intestinal mucosal barrier.195 The clinical setting of severe mucositis or intestinal graft-versus-host disease with intestinal mucosal damage may be circumstances in which false-positive serum galactomannan tests may be expected. In addition, false-positive serum tests have been reported in association with administration of certain antibiotics such as piperacillin-tazobactam or amoxicillin derived from cross-reacting species of mold such as Penicillium spp.190,196–198 This has implications for clinicians using this test to investigate the possibility of invasive aspergillosis while administering drugs like piperacillin-tazobactam for the treatment of infection.

Duration of Antibacterial Therapy

In general, the IDSA recommendations for duration of the antibacterial regimen encompass the period until neutrophil recovery (ANC >0.5 × 109/L for at least two consecutive days), all signs and symptoms of infections have resolved, and the temperature has remained normal for 48 hours or more.55 For patients with prolonged severe neutropenia who have defervesced and for whom no focus of infection appears to be ongoing, the antibiotic regimen may be discontinued after 2 weeks provided the patient remains under careful observation. Some investigators have advocated substituting a fluoroquinolone-based antibacterial chemoprophylaxis regimen for the systemic antibacterial regimen under these circumstances.199 The response assessment definitions used in clinical trials of antibacterial therapy have often included a stipulation that the patient must remain afebrile for 4 to 5 days in addition to resolution of other signs and symptoms of infection in order for a response to be valid. This seems prudent and consistent with the natural history of these episodes. While the time until response to empiric antibacterial therapy varies with the underlying causes of neutropenia,17,50,200 the median time to response (defervescence) for high-risk patients is 5 to 7 days41,45,47,50,87,88,100 and 2 to 4 days for low-risk patients.138,139 If the antibacterial regimen is to be administered for an additional 4 to 5 days by the above criteria, then high-risk patients and low-risk patients would be expected to have received a 9- to 12-day and 6- to 9-day course of antibacterial therapy, respectively.

Infections occur at a limited number of body sites in febrile neutropenic patients and usually involve the microorganisms colonizing those sites.129 The three systems most commonly involved are the gastrointestinal (GI) tract (oropharynx, gingiva and teeth, esophagus, gut, and perirectal tissues), the respiratory system (sinuses, middle ear, nasopharynx, tracheobronchial tree, and pulmonary parenchyma), and the skin (including biopsy sites and sites of vascular access such as indwelling central venous catheter exit sites, tunnel sites, or insertion sites).

Oropharyngeal Mucosal and Esophageal Infections

The natural history of oral mucositis is influenced by the cytotoxic therapy–induced neutropenia,34 which plays a permissive role in the clinical expression of acute-on-chronic periodontal infections.201 This process usually reaches its maximum intensity at the time of neutropenic nadir, approximately days 10 to 14.27,29–31 At this time, polymicrobial infection becomes superimposed on the chemotherapy-induced mucositis. This in turn extends the morbidity into the third and fourth weeks following the commencement of chemotherapy. Although oropharyngeal bacterial flora (viridans group streptococci, anaerobic gram-negative bacilli, and anaerobic gram-positive cocci) probably contribute to disease in most cases of simple mucositis, in several studies fungi (e.g., C. albicans) have played important pathogenic roles in up to 60% of the oral infections among patients with acute leukemia.202 In addition, reactivated latent HSV infections of the oral cavity have been reported in 50% to 90% of seropositive patients undergoing remission-induction therapy or BMT with a median onset between 7 and 11 days.203,204 Acute exacerbations of preexisting, asymptomatic, chronic periodontitis occurred in 59% of one series of adult patients undergoing remission-induction therapy for acute leukemia.201 These infections typically occurred when the ANC was <0.13 × 109/L. The severity and duration of chemotherapy-associated mucositis correlate to some degree with the extent of preexisting dental plaque and periodontal disease.205

Clinical Approach

Herpetic infections of the oropharynx and esophagus may be anticipated in patients with a history of herpetic stomatitis or in those known to possess IgG antibodies to HSV, indicating infection in the past. Although the typical discrete vesicular lesions on an erythematous base may be observed in neutropenic patients, herpetic infections also may manifest as areas of painful ulceration over a diffusely erythematous base. Such lesions must be distinguished from a typical presentation of oropharyngeal candidiasis or cytotoxic therapy–induced mucositis. Pseudomembranous pharyngitis suggests yeast infection. A thorough examination of the gingival and periodontal tissues for focal areas of pain, erythema, swelling, and bleeding can suggest the periodontium as a potential focus of infection (particularly as a source of bacteremic infection) by viridans streptococci and oropharyngeal gram-negative anaerobic bacilli.206

Laboratory aids include virus culture techniques, direct fungal stains, direct electron microscopic examination for virus particles, cytologic examination of cellular material from the base of the ulcer (e.g., Tzanck preparation for the detection of multinucleated giant cells and intranuclear inclusions), or direct herpes simplex antigen detection techniques. The material from a specific lesion should be submitted to the clinical microbiology laboratory for culture and for direct examination. Routine Gram's stain can be helpful in demonstrating the presence of budding yeasts and pseudohyphae suggestive of Candida species. A potassium hydroxide mount (to digest extraneous unwanted cellular material) also can provide a clue to this diagnosis by demonstrating the presence of these structures.

Management

The morbidity associated with oropharyngeal or esophageal mucositis can be life threatening, particularly when local pain interferes with an adequate nutritional intake. Pain control becomes a high priority. A topical anesthetic such as lidocaine in a 2% water-soluble gel or 5% water-insoluble ointment has been used widely with inconsistent success. Continuous intravenous morphine infusions have been successful for symptom control among BMT recipients with cytotoxic therapy–induced mucositis or acute oral graft-versus-host disease. Herpetic mucositis involving the oropharynx or esophagus should be treated with acyclovir. Intravenous acyclovir (250 mg/m2 q8h) may be administered for severe cases until oral administration (200 mg q4h) can be tolerated for a total course of 7 days. Pseudomembranous candidiasis involving the oropharynx or esophagus may be treated with various approaches. Topical therapy with oral nystatin suspension (2 to 30 million units daily in four divided doses) remains a popular first-line approach. Many physicians prefer to prescribe orally absorbed azole antifungal agents such as ketoconazole (200 to 600 mg daily) or fluconazole (50 to 400 mg daily). It is important to remember that the efficacy of ketoconazole may be compromised by concomitant use of antacid therapy or in the setting of gastric achlorhydria. Invasive candidal esophagitis should be treated with intravenous amphotericin B to a cumulative dose of 500 to 1500 mg (approximately 5 to 15 mg/kg), while oropharyngeal candidiasis has been treated successfully with cumulative doses of about 500 mg (5 mg/kg). Necrotizing polymicrobial anaerobic mucositis responds well to metronidazole (500 mg PO or IV q8h).129

Evidence is accumulating that much of the extra morbidity caused by these infections superimposed on chemotherapy-induced mucositis can be reduced significantly by the prophylactic use of antiviral agents such as acyclovir among HSV-seropositive individuals,207 antiseptics such as chlorhexidine,208 and antifungal agents such as oral azoles.209 Further work is required to determine the optimal dose schedules and routes of administration for these agents.

Enteric Infections

Invasive enteric bacterial infections of the gut due to Salmonella or Shigella species are relatively uncommon in neutropenic patients. Two clinical entities, however, must be considered in febrile neutropenic patients with abdominal pain and diarrhea: toxigenic enterocolitis due to the toxin elaborated from an overgrowth of Clostridium difficile and neutropenic enterocolitis (typhlitis).

Clostridium Difficile–Associated Diarrhea

Clostridium difficile–associated diarrhea (CDAD) is not a rare problem among neutropenic patients receiving broad-spectrum antibiotic therapy, particularly those who are recipients of antibacterial agents that have high biliary excretion rates and are active against intestinal anaerobic bacteria. Approximately 5% to 10% of high-dose chemotherapy recipients have been reported with this complication.210–212 This problem must be considered when such patients develop abdominal pain in association with watery diarrhea with or without blood. The differential diagnosis includes other infectious (cytomegalovirus [CMV] enterocolitis or typhlitis) and noninfectious (cytotoxic therapy–induced mucosal damage or peptic ulcer disease) causes. Protozoa and helminths are unlikely causes of this syndrome unless there is an associated history of travel to or immigration from areas where pathogens such as Giardia lamblia or Strongyloides stercoralis are endemic.

Diarrheal stools from outpatient clinics should be submitted to the clinical microbiology laboratory for bacterial culture (to rule out the remote possibilities of salmonellosis, shigellosis, and Campylobacter or Yersinia infection), examination for ova and parasites (unless this has been done earlier in the patient's course), and culture for C. difficile and the detection of C. difficile toxin. CDAD is much more common among hospitalized patients. If C. difficile is cultured or the toxin detected, this is sufficient to warrant therapy with either metronidazole (500 mg PO q8h for 10 days) or vancomycin (125 to 250 mg PO q6h for 10 days) without a sigmoidoscopic examination for detection of the pseudomembranous changes associated with this infection. In patients with a sufficiently typical clinical presentation, it is also reasonable to begin therapy empirically before laboratory confirmation is available.

Typhlitis and Neutropenic Enterocolitis

Typhlitis, also called neutropenic enterocolitis, necrotizing enterocolitis, or ileocecal syndrome, is a serious, potentially life-threatening infection of the bowel wall seen in up to 32% of patients undergoing remission-induction therapy for acute leukemia.213–216 The pathologic process includes diffuse dilatation and edema of the bowel walls, with varying degrees of mucosal and submucosal hemorrhage and ulceration.217 The cecum appears to be favored for the development of this syndrome, possibly related to its relatively tenuous blood supply. Bacterial invasion through an ischemic gut wall in the setting of neutropenia and cytotoxic therapy–induced mucosal surface damage is the probable pathogenesis.213,214 The syndrome presents a spectrum of severity from mild self-limiting cecal inflammation to fulminant bowel wall necrosis with perforation. The clinical syndrome is typically characterized by a triad of diarrhea, abdominal pain, and fever in the setting of cytotoxic therapy–induced neutropenia.217,218 Abdominal distention, nausea, vomiting, and diffuse watery or bloody diarrhea are also commonly observed. Bacteremia with enteric microorganisms (E. coli or Klebsiella species) and Pseudomonas aeruginosa is associated with typhlitis in up to 28% of cases.213 Clinical examination usually reveals a diffusely tender abdomen; however, localization to the right lower or upper quadrant is not uncommon. Ultrasonographic219 or CT imaging of the abdomen frequently demonstrates thickening and edema of the colonic walls with or without inflammatory changes in the surrounding pericolic tissues (Fig. 47-4). Gas in the intestinal wall (pneumatosis) or an inflammatory phlegmon also may be seen in approximately 20% of cases.69 CDAD is associated with the greatest degree of bowel wall thickening.69 There appears to be a correlation between the thickness of the bowel wall, presumably reflecting the degree of inflammation, and mortality. Cartoni and colleagues, using ultrasonographic examinations, noted mortality rates of 60% among those with mural thickness of more than 10 millimeters.219 Mural thickness also correlated with prolonged duration of symptoms of more than a week compared to patients without thickening (mean duration of symptoms was 7.9 vs. 3.8 days).219

Neutropenic enterocolitis is associated with significant impairment in the functional integrity of the gut mucosal surface. Malabsorption of D-xylose has been shown to precede clinical neutropenic enterocolitis by at least a week.32 This intestinal mucosal damage increases the risk for translocation of bacteria and opportunistic fungi that colonize the intestinal lumen. For example, Candida mannoprotein antigen associated with intestinal colonizing Candida spp. may be detected in patients with neutropenic enterocolitis.220 Translocation of colonizing yeast in the setting of cytotoxic therapy–induced neutropenic enterocolitis is part of the pathogenesis of invasive candidiasis with portal fungemia and subsequent hepatosplenic fungal infection.32,221

Management consists of early recognition, bowel rest with nasogastric decompression, intravenous fluid replacement, blood product transfusion, and broad-spectrum antibacterial agents directed against the aerobic, microaerophilic, and anaerobic enteric microflora (e.g., piperacillin-tazobactam, imipenem-cilastatin, and meropenem plus an agent effective against obligate anaerobic bacteria such as metronidazole). Although surgical consultation early in the evolution of the syndrome is recommended so that all potential management strategies can be planned before an intra-abdominal catastrophe occurs, medical management is recommended for the majority of patients who do not have evidence of a major intra-abdominal catastrophe. Surgical intervention with right hemicolectomy or local resection of necrotic segments of bowel with anastomosis or diverting ileostomy or colostomy should only be considered in the setting of cecal perforation, massive uncontrollable GI bleeding, uncontrollable sepsis, complete bowel obstruction, or pneumatosis cystoides intestinalis.215 With optimal management, the mortality rate has declined from over 50% to less than 20%.217,218 The optimal outcome for these patients appears to require a high index of suspicion for the syndrome, aggressive supportive care, and marrow recovery. It is important to recognize the high risk of recurrence (up to two-thirds of cases) with subsequent cycles of cytotoxic therapy.214

Perirectal Infections

Infections of the perirectal tissues may be life threatening in neutropenic patients. The majority of cancer patients with this complication have received cytotoxic therapy for leukemia or lymphoreticular malignancy within the preceding month and are severely neutropenic (ANC <0.5 × 109/L) at the time of presentation.222 Although the presence of neutropenia may preclude the development of frank suppuration, perirectal infection must be suspected if there is focal tenderness, perirectal induration, or erythema with or without fluctuance or tissue necrosis.64 Although some cases may be associated with a preexisting pathologic process such as an anal fissure or thrombosed hemorrhoidal tissues, most patients present no obvious predisposing factor. Conceivably, small abrasions or tears in the rectal mucosa may be the primary event allowing tissue invasion by the microflora colonizing the rectum, anus, and perineum. The etiology of these infections is polymicrobial.223,224 The most common microorganisms associated with perirectal infections are enteric gram-negative bacilli (e.g., E. coli, Klebsiella spp., or Enterobacter spp.), P. aeruginosa, obligately anaerobic gram-negative bacilli (e.g., Bacteroides spp.), enterococci, and peptostreptococci. Recurrent episodes of infection are frequent (up to 26% of episodes in one series) among patients recovering from one perirectal infection and receiving subsequent cycles of cytotoxic therapy.222

The optimal approach to management is controversial. In general, neutropenic patients should be managed medically unless local care, systemic antimicrobials, and blood product support fail to contain the infection, or if an obvious inflammatory collection must be surgically drained. The likelihood of success of medical management is increased if the antimicrobial regimen contains agents effective in severe intra-abdominal sepsis,225 and should include a single agent such as piperacillin-tazobactam or a carbapenem, imipenem-cilastatin or meropenem, or a combination regimen including an agent effective against anaerobic bacteria (e.g., clindamycin or metronidazole) and either a cephalosporin or a fluoroquinolone. Furthermore, vancomycin may be added if despite these antimicrobials the cellulitis appears to progress. The therapeutic role of antimicrobial agents active against Enterococcus spp. remains difficult to evaluate since the role of these bacterial species in the pathogenesis of these infections is controversial. Cephalosporins are inactive against Enterococcus spp. and the carbapenem-related susceptibility profiles tend to mirror that of ampicillin and therefore are less reliable.226 Enterococcal colonization must be differentiated from infection. Furthermore, the rising incidence of vancomycin-resistant Enterococcus species argues to minimize the use of vancomycin for suspected but unproven enterococcal infection.129 Recent observations with regard to the emergence of vancomycin-resistant Enterococcus species following piperacillin-based antibacterial therapy227 are of concern given the frequency of use of this agent in the treatment of bacterial infections in cancer patients.

Invasive Fungal Infections

Opportunistic Yeast Infections

This group of opportunistic unicellular fungal organisms of the form-class Blastomycetes and form-family Cryptococcaceae includes six genera: Cryptococcus (e.g., C. neoformans, the agent of cryptococcal meningitis), Malassezia (e.g., M. furfur, the agent of pityriasis versicolor), Rodotorula (e.g., R. rubra, an agent causing pulmonary and systemic infections), Candida (e.g., C. albicans, the most common cause of candidiasis), Trichosporon (e.g., T. beigelii, the agent of white piedra and systemic infections in compromised hosts), and Torulopsis (e.g., T. glabrata, now reclassified under the genus Candida). C. albicans, C. tropicalis, C. glabrata, and Trichosporon species are part of the normal microflora of the mouth, colon, and vagina. Consequently, it is not surprising to find these agents involved in the pathogenesis of infections among immunocompromised patients with damaged mucosal and integumental surfaces. This has been recently reviewed.228

Invasive candidiasis encompasses deep infections of various organ sites with Candida species.229,230C. albicans is the most commonly observed; however, C. tropicalis, C. glabrata, C. parapsilosis, C. krusei, and C. guilliermondii are causal agents of infection in neutropenic patients as well. When multiple organ sites are involved, the term disseminated candidiasis may be more appropriate. The most common forms of invasive candidiasis encountered in neutropenic patients are candidemia with or without associated central venous catheter infection, chronic systemic candidiasis (hepatosplenic candidiasis), endophthalmitis, hematogenously spread skin infection, and renal candidiasis. The pathogenesis is the invasion of damaged integumentary surfaces by colonizing yeasts during severe immunosuppression and myelosuppression.

Candidemia without evidence of metastatic infection is frequently associated with indwelling central venous catheters. Such patients should have the catheter removed and should receive antifungal therapy229,231 with intravenous amphotericin B deoxycholate 0.6 to 1.0 mg/kg per day to a total dose of 500 to 1000 mg, or with intravenous or oral fluconazole 400 to 800 mg daily, or intravenous caspofungin 70 mg on day 1 followed by 50 mg daily.232 Patients with evidence of dissemination to other organs should be treated with amphotericin B deoxycholate or a lipid-based formulation of amphotericin B to total doses of 30 to 60 mg/kg (2 to 4 g for a 70-kg patient). The overall response rates for amphotericin B in candidemia have been as high as 70% in teaching hospitals.233 Patients with CNS involvement probably also should receive 5-flucytosine (5-FC). Imidazoles such as ketoconazole or miconazole are not recommended under these circumstances. The role of the newer triazoles fluconazole and itraconazole in the treatment of invasive candidiasis in neutropenic patients appears promising.230

Two randomized trials in non-neutropenic surgical and ICU patients with invasive candidiasis have demonstrated equivalency of fluconazole and amphotericin B.234,235 One retrospective, matched-cohort study236 and one randomized controlled study comparing fluconazole and amphotericin B in a mixture of neutropenic and non-neutropenic cancer patients with invasive candidiasis also have demonstrated similar response rates of 64% to 73%.236,237

The combination of a triazole and a polyene antifungal may be potentially antagonistic given that both classes act on the fungal cell membrane.238,239 Azoles interfere with the 14-demethylation of lanosterol in the synthetic pathway of ergosterol, the predominant sterol in the fungal cell membrane to which polyene antifungal agents such as amphotericin B bind to produce their fungicidal effects. Clinical reports have suggested that sequential treatment with lipophilic azoles such as itraconazole followed by amphotericin B may be deleterious.240 This effect may not be as prominent with the more lipophobic azoles such as fluconazole, with less accumulation of the agent in the lipid-rich environment of the cell membrane. A recent randomized blinded placebo-controlled trial from the National Institutes of Allergy and Infectious Disease Mycoses Study Group compared the safety and efficacy of fluconazole with or without amphotericin B for the treatment of candidemia in severely ill non-neutropenic patients with high Acute Physiology and Chronic Health Evaluation (APACHE) II scores.241 The response rates were somewhat higher in the combination group (60 of 107, 56% vs. 77 of 112, 69%; p = 0.043) and no evidence of antagonism was observed. The odds of treatment failure were increased by high APACHE II scores (OR 1.09; 95% CI 1.03 to 1.14), and decreased if C. parapsilosis was the etiology of the bloodstream infection.241

Most episodes of candidemia are caused by C. albicans;230 however, there is an increasing incidence of non-albicans candidemia.242–245 In a multicenter prospective, observational study of candidemia in 427 consecutive patients from four tertiary care hospitals,242C. albicans accounted for 52% of the bloodstream isolates overall. However, over the course of the 3.5-year study, the proportion of all bloodstream isolates that were non-albicans candidal species increased from 40% to 53%. Risk factors for candidemia due to C. albicans included no prior use of antifungal therapy, solid organ transplant recipient, and having a wound as portal of entry. The single most important risk factor for C. tropicalis fungemia was neutropenia. Prognostic factors for C. glabrata fungemia included prior use of amphotericin B, hemodialysis, and an abdominal portal of entry; for C. krusei fungemia, prior use of fluconazole, and neutropenia; and for C. parapsilosis, prior use of fluconazole. Breakthrough candidemia occurred in 13% of patients receiving an antifungal agent for prophylaxis or empiric therapy. Under these circumstances, the candidemias were due to non-albicans species that were resistant to fluconazole. These observations have important implications for neutropenic patients in whom agents such as fluconazole are being used more widely for prophylaxis246,247 and therapy.236,237,241,248

Chronic systemic candidiasis, also referred to as granulomatous hepatitis, focal hepatic candidiasis, or hepatosplenic candidiasis, has emerged as a significant problem among recipients of high-dose cytotoxic therapy.230,249–253 The typical clinical presentation is a patient who has received high-dose cytarabine, followed by a febrile illness that persists despite broad-spectrum antimicrobial therapy and recovery of the circulating neutrophil count. There may be associated right upper quadrant tenderness, hepatomegaly, splenomegaly, and elevated serum alkaline phosphatase and γ-glutamyltransferase levels. The diagnosis may be established by imaging studies of the abdomen using ultrasonography or CT showing multiple abscesses in the liver and splenic parenchyma (Fig. 47-5). Where possible, a tissue biopsy should be considered for histopathologic and microbiologic confirmation. Biopsy of these lesions demonstrates the presence of necrotizing granulomata, frequently (but not uniformly) containing budding yeasts and pseudohyphal elements consistent with Candida species. Liver biopsy specimens should be submitted in a dry, sterile container to the clinical microbiology laboratory and processed for pyogenic bacteria, mycobacteria, viruses, and fungi. Specimens for histopathologic evaluation may be submitted to the pathology laboratory in fixative and processed for staining with hematoxylin and eosin, with periodic acid-Schiff (PAS), with an acid-fast stain, and with methenamine silver. Multiple sections through the paraffin-embedded tissue fragments may be necessary to detect the pathogens. Open liver biopsy is the most reliable means of definitively establishing the diagnosis. The pathogen is often not grown despite appropriate culture of the tissue specimen. The determination of a correct diagnosis becomes an important factor contributing to the planning of potentially curative postremission therapeutic strategies.

About half the patients respond to systemic amphotericin B (1.0 to 1.5 mg/kg per day) with or without 5-FC (100 mg/kg per day) in cumulative amphotericin B doses as high as 9 g. However, even in those who respond, the difficulty with diagnosis and the requirement for prolonged antifungal therapy frequently result in failure to complete planned postremission therapeutic regimens for the underlying malignancies, leading to shortened disease-free survival.221 The toxicities of prolonged systemic amphotericin B (anemia, renal insufficiency, hypokalemia, fever, chills, and nausea) and 5-FC (myelosuppression) are also limiting. More recently, experience with the newer triazole antifungal agent fluconazole has suggested that this may be a useful approach for patients unable to tolerate or failing amphotericin B–containing regimens.230

Amphotericin B toxicity is often a limiting factor in therapy. Acute infusional toxicities occur in approximately 70% of patients during the first 7 days of therapy,254 and include fever (51%), chills (28%), nausea (18%), and headache (9%). Prophylactic use of diphenhydramine, acetaminophen, and corticosteroids (alone or in combination) has been no more effective than no prophylaxis in preventing these infusional toxicities.254Meperidine (0.2 to 0.5 mg/kg) may ameliorate the rigors. Hydrocortisone has not been shown to be effective, and as a sterol itself, may compete with fungal cell membrane ergosterol for the binding of amphotericin B.255 Metabolic toxicities including potassium and magnesium wastage or elevated serum creatinine occur in more than 80% of patients.256 Approximately one third of amphotericin B deoxycholate recipients can be expected to experience a doubling of the serum creatinine from pretreatment levels.177

The newer lipid-based formulations of amphotericin B have been developed to reduce the amphotericin B–related toxicities and to permit the administration of larger doses.230 These agents include liposomal amphotericin B, amphotericin B colloidal dispersion, and amphotericin B lipid complex (ABLC). At least one randomized controlled trial comparing ABLC (5 mg/kg per day) with conventional amphotericin B (0.6 mg/kg per day) in patients with invasive candidiasis has shown similar results with respect to efficacy and infusional toxicities, but with reduced nephrotoxicity in ABLC recipients.257 Lipid-based formulations of amphotericin B have been studied in clinical trials of empiric antifungal therapy.177,179

The optimal dosing protocol for amphotericin B for empiric therapy in the setting of persistent fever unresponsive to broad-spectrum antibacterial therapy in neutropenic patients has never been established. A daily dose of at least 0.5 mg/kg amphotericin B until marrow recovery has been recommended.258 Medoff suggested a total cumulative amphotericin B dose of 5 to 20 mg/kg given over 3 to 6 weeks for candidemia.259 The recommended doses for invasive candidiasis have ranged from 0.5 to 1.0 mg/kg per day to cumulative doses of 2 to 4 g.260 For a 70-kg patient receiving 0.5 mg/kg per day, this amount would require between 2 and 4 months to administer. It is unknown whether the efficacy of amphotericin B therapy is correlated with total cumulative dose or with duration of therapy.

The IDSA has suggested that patients with prolonged neutropenia who are persistently febrile despite 4 to 7 days of broad-spectrum antibacterial therapy are candidates for empiric antifungal therapy with amphotericin B.55 This recommendation is based on two previously published trials.175,176 The incidence of empiric amphotericin B use is significantly higher compared with the incidence of documented invasive fungal disease in neutropenic patients with acute leukemia (36% to 69% vs. 3% to 11%, respectively).261–264 In BMT recipients, empiric amphotericin B use is also significantly higher than the incidence of invasive fungal infection (38% to 82% vs. 3% to 28%, respectively).265–268 These differences are likely related to the difficulties in diagnosing invasive fungal infection in these patients. At least one randomized controlled clinical trial248 compared fluconazole (6 mg/kg per day up to 400 mg/d) and amphotericin B (0.8 mg/kg per day), and demonstrated similar efficacy but with less drug-related toxicity in the fluconazole recipients compared with amphotericin B recipients (32% vs. 82%, respectively). These results are consistent with those in neutropenic and non-neutropenic patients with invasive candidiasis.234−237,241

The echinocandins, a class of antifungal agents that inhibit the activity of 1, 3-β glucan synthase at a critical step in fungal cell wall synthesis, are fungicidal against species of Candida. Among patients with candidemia, caspofungin recipients have responded only marginally better than conventional amphotericin B deoxycholate recipients, 73% vs. 62%, respectively.269 However, caspofungin was associated with fewer adverse events such as nephrotoxicity. Among the small number of neutropenic patients enrolled into that study, the response rates were 3 of 8 (38%) for amphotericin B deoxycholate recipients and 6 of 8 (75%) among caspofungin recipients (OR 5.0; 95% CI 0.42 to 42.3). Based on these observations and those from the fluconazole trials, the National Comprehensive Cancer Network Clinical Practice guidelines in Oncology recommend the use of either caspofungin 70 mg intravenously as a loading dose followed by 50 mg intravenously daily, or fluconazole 400 to 800 mg daily in adults with invasive candidiasis instead of conventional amphotericin B deoxycholate on the basis of equivalence of efficacy and superior safety profiles.129 Furthermore, caspofungin or a lipid formulation of amphotericin B may be recommended for clinically unstable patients with candidemia due to C. krusei or C. glabrata.

The duration of antifungal therapy for fungemic patients should be at least 14 days following the last positive blood culture and resolution of signs and symptoms of fungal infection in neutropenic and non-neutropenic adult patients.232 The duration for patients with hepatosplenic fungal infections is a period of 3 to 6 months along with resolution of signs and symptoms of infection or calcification of radiologically detectable lesions. The duration for oropharyngeal candidiasis should be 7 to 14 days after clinical improvement and 7 to 21 days for esophageal candidiasis.232

The IDSA guidelines recommend the removal of all indwelling vascular access catheters in candidemic patients.232 The purpose of this recommendation is to reduce the risk for metastatic visceral infections and excess mortality.270 Earlier studies suggested that catheter removal at diagnosis of candidemic episodes reduced the duration of candidemia271 and mortality.270 More recent studies among candidemic patients receiving either caspofungin or amphotericin B deoxycholate with or without central venous catheters left in situ failed to confirm those earlier observations with regard to response rates and to duration of candidemia.269

Opportunistic Filamentous Fungal Infections

Opportunistic filamentous fungal infections are frequently life threatening complications among neutropenic patients undergoing remission-induction treatment for acute leukemia or hemopoietic stem cell transplantation. These infections are most often caused by Aspergillus species, and more often by fumigatus Aspergillus spp. as compared to nonfumigatus Aspergillus spp., including A. flavus, A. niger, and A. terreus, Fusarium spp., the dematiaceous fungi (black molds), or the Zygomycetes (e.g., Absidia spp., Rhizopus spp., Rhizomucor spp., Mucor spp. or Cunninghamella spp.).228 They produce similar clinical syndromes in compromised hosts, including necrotizing nasal mucosal infection, sinusitis, endophthalmitis, cerebral parenchymal infection, pulmonary parenchymal infection, cutaneous infection, typhlitis, hepatosplenic abscesses, osteomyelitis, and intravascular infections.272

Infections by these molds are characterized by blood vessel invasion resulting in thrombosis and blood vessel obstruction, which in turn cause ischemia and infarction of the distal tissues.273–276 This is the mechanism for the clinical manifestations such as pulmonary cavitary disease, hemoptysis, cutaneous infarcts (see Fig. 47-2), stroke-like syndromes due to intracranial infection (Fig. 47-6), and parenchymal hepatic infarction (Fig. 47-7). Infection occurs following the germination of inhaled conidia on the respiratory epithelium with the production of invasive hyphae.275

Characterization of these infections is important in determining prognosis and outcome of antifungal therapy. The National Institute of Allergy and Infectious Diseases Mycoses Study Group (NIAID/MSG) of the National Institutes of Health and the European Organization for the Treatment and Research of Cancer/Invasive Fungal Infections Cooperative Group (EORTC/IFICG) have developed an international consensus of diagnostic criteria for invasive fungal infections (IFI).277 These infections are classified into three levels of probability as proven, probable, or possible, based on the robustness of the evidence supporting the diagnosis. A recent study from Spain conducted among patients undergoing peripheral blood stem cell transplants who developed IFI reported that mortality rates were related to the classification of the IFI as possible (mortality 20%), probable (mortality 57%), or proven (mortality 80%).278 Ascioglu and colleagues caution, however, that these criteria were developed in the context of clinical trials rather than for clinical application.277 The clinical applicability is limited by false-negative observations wherein patients classified as only probable IFI may ultimately have widespread IFI at postmortem examination.279 These observations underscore the need for more sensitive and predictive tests. However, despite the limitations of diagnostic technology, it seems prudent to use these criteria as guidelines for determining the robustness of the clinical diagnosis and the need to proceed to more invasive diagnostic procedures or for epidemiologic research.280

Molds are ubiquitous in the environment183,273 and present in a wide variety of natural and synthetic materials such as soil,274 decaying vegetation,274 fireproofing materials,281 water,282 and air.283Aspergillus species frequently have been detected in the air of hospital rooms, and in particular, in circumstances associated with construction and renovation.284,285 The management of leukemia and BMT patients in specialized self-contained hospital nursing units outfitted with high-efficiency particulate air filtration (HEPA) has reduced the incidence of invasive aspergillosis,286 and overall mortality rates, particularly among hemopoietic stem cell transplant recipients.287 Guidelines are available for planning infection control strategies to contain the excess risk of invasive mold infections among high-risk patients managed in environments where hospital maintenance and renovation projects are underway.288–290 HEPA-filtered nursing units have lower concentrations of airborne fungal conidia than their non–HEPA-filtered counterparts.291

Invasive aspergillosis in leukemia and hemopoietic stem cell transplant patients has been reported to have extremely high mortality rates almost regardless of treatment.292 The risk of this infection increases with the duration of neutropenia to a plateau of 70% to 80% at 5 weeks.293 Marrow recovery represents the single most important factor relating to survival of invasive fungal infection in neutropenic patients.294

The clinical findings relate to the infected organ site. Fever is almost invariably present. Evidence of tissue ischemia and infarction may provide clues to the diagnosis. The most common presentation is that of focal pulmonary infiltrates in a persistently febrile, severely neutropenic patient unresponsive to broad-spectrum antibacterial therapy. Some investigators have suggested that nasal cultures positive for Aspergillus species in this setting can be highly predictive (∼90%) of invasive aspergillosis.181 Positive cultures from other respiratory specimens such as sputum, bronchial brushings, or bronchoalveolar lavage fluid also can be predictive of invasive aspergillosis in high-risk patients;230,295 however, these relationships have not been universally accepted. Based on multivariate analysis of patients with acute leukemia, a number of factors predictive of invasive opportunistic fungal disease have been identified by different investigators.293,296–298 In each of these studies the duration of severe neutropenia was the most important independent variable. Other related variables were the duration of cytotoxic therapy (which gives rise to prolonged neutropenia), the duration of neutropenia associated with antibacterial therapy, and colonization by fungi at surveillance culture sites.

The definitive diagnosis of invasive aspergillosis usually requires microscopic and microbiologic examination of tissue biopsied from sites of infection.273 The demonstration of septate hyphae branching at acute angles in methenamine silver stained or periodic acid-Schiff–stained tissue sections suggests the diagnosis; however, these morphologic characteristics in stained tissue sections are also shared by species of Fusarium and Scedosporium.273 Microbiologic identification in culture is required to confirm the diagnosis. This is important because organisms such as Fusarium spp., S. apiospermum, and S. prolificans are not susceptible to amphotericin B. Where available, immunodiagnostic techniques may increase the positive predictive value of the classic microscopic appearance of these organisms in tissue sections.299

In leukemia patients the reported mortality rates from invasive pulmonary aspergillosis have had a wide range (between 13% and 100%), whereas in BMT patients the mortality rate approaches 100%.292 Marrow recovery represents the single most important determinant of survival.294 Aspergillosis can develop and progress despite the concomitant administration of amphotericin B at daily doses of 0.5 mg/kg.300 Improved responses and survival rates for leukemia patients have been reported with the use of high-dose amphotericin B (1.0 to 1.5 mg/kg per day in contrast to a more standard dose of 0.5 to 0.7 mg/kg per day) without serious or permanent renal damage.300 Anecdotal reports and studies in animal models support the use of 5-FC with amphotericin B.239,292 Doses of 5-FC of 100 to 150 mg/kg per day in four divided doses orally usually result in serum levels of 50 to 100 mg/L. There is an increased risk of excess myelosuppression with serum levels above these values. Since 5-FC is excreted predominantly in the urine and concomitant amphotericin B therapy is almost always associated with a significant decline in renal function, regular monitoring of 5-FC levels and appropriate adjustment of dose to maintain levels in this range are mandatory.

Caspofungin has recently been approved in the United States for the treatment of invasive aspergillosis in patients refractory or intolerant of polyene-based therapy.301 The initial experience in such patients demonstrated response rates of 50% in invasive pulmonary aspergillosis, 23% in disseminated aspergillosis, and 26% in neutropenic patients. Among patients receiving caspofungin for empiric therapy of suspected fungal infection in neutropenic patients and in whom invasive aspergillosis ultimately was determined to be the cause of the persistent fever, the response rate was 42% compared to only 8% among liposomal amphotericin B recipients.302

Amphotericin B lipid complex (ABLC) has been widely used in patients with invasive fungal infections refractory to or intolerant of conventional amphotericin B deoxycholate. In a review of 556 such patients receiving ABLC on an emergency drug release program, the response rate among 130 patients with invasive aspergillosis was 42%.303 A 47% response rate was reported in an historic control study of ABLC in 39 solid organ transplant recipients with invasive aspergillosis.304 In a series of pediatric patients with invasive aspergillosis, a 56% response rate was reported.305

Amphotericin B–related nephrotoxicity impacts upon patient morbidity, mortality, and the cost of management.306–308 The safety profiles of the lipid formulations of amphotericin B in the published literature have been consistent in demonstrating advantages of reduced amphotericin B–related nephrotoxicity.177,307 Even among patients with elevated serum creatinine levels at the outset of ABLC therapy, the advantage persists. The elevated serum creatinine levels have been observed to fall after the first week with continued administration of the drug.309 This reduction in elevated serum creatinine after the first week of therapy may occur even when initially normal serum creatinine levels rise within the first week of ABLC therapy.308 This experience is useful to physicians managing severely ill patients with impaired renal function and for whom polyene-based antifungal therapy is deemed necessary.

The extended-spectrum azoles such as itraconazole, voriconazole, and posiconazole have proven useful in the management of invasive mold infections.310–313 Voriconazole proved to be superior to conventional amphotericin B deoxycholate or other licensed antifungal therapy for the management of invasive aspergillosis in a large multinational trial.312 The treatment effect was observed in different analyses including hemopoietic stem cell transplant recipients, neutropenic patients, patients with proven or probable invasive aspergillosis, and those with pulmonary and extrapulmonary aspergillosis.

There is a high risk of relapse (>50%) in cancer patients surviving an initial episode of invasive aspergillosis during subsequent cytotoxic treatments.314 For this reason many investigators recommend secondary prophylaxis with antifungal agents such as itraconazole or lipid formulations of amphotericin B for patients with treated invasive aspergillosis undergoing subsequent high-dose cytotoxic therapy.315 A combination of high-dose amphotericin B (1.0 mg/kg per day) and 5-FC (100 mg/kg per day) has been used successfully to protect leukemia patients who developed invasive pulmonary aspergillosis during initial remission-induction against reactivation of the infection during subsequent cycles of cytotoxic therapy, without delay of marrow recovery.316–318 This seems to be a rational approach for leukemic patients, but its validity for BMT patients remains to be proven.318 The role of combinations of amphotericin with other agents such as rifampin or the azoles remains unclear and cannot be recommended at the present time.239

Indwelling Vascular Access Device Infections

Indwelling peripheral and central vascular access catheters have long been recognized as a source of sepsis for critically ill cancer patients. This subject has been reviewed recently.231,319,320 Venous and arterial cannulas physically disrupt the integrity of the skin and blood vessels, thus providing an avenue for ingress of bacteria or fungi colonizing the skin surfaces at the site of cannula placement. Determinants of intravenous cannula infection include the type of cannula, the duration of use, the technique of skin preparation for insertion, and the use of venotoxic infusates. The most common microorganisms causing intravenous site infection are gram-positive organisms such as Staphylococcus aureus, the coagulase-negative staphylococci, and Corynebacterium jeikeium; gram-negative bacilli such as the Enterobacteriaceae, P. aeruginosa, Stenotrophomonas maltophilia, and Acinetobacter spp; and fungi such as Candida species. Erythema, swelling, exudate, and focal tenderness at a peripheral intravenous catheter site should always alert the clinician to these etiologic possibilities. Suspect catheters should be removed promptly and carefully using aseptic technique and submitted to the clinical microbiology laboratory in a sterile dry container for microbiologic evaluation.

Venous access is a major problem for patients with complex multisystem problems. The need for repeated blood sampling for monitoring and for the administration of different therapies has led to the widespread use of indwelling tunneled central venous catheters. These devices provide reliable access for venous blood sampling and administration of chemotherapy, blood products, antimicrobial agents, total parenteral nutrition, crystalloids, and vasoactive agents. Infection of these devices may occur at the exit site (the point at which the catheter exits the skin), the tunnel site (that portion of the catheter buried subcutaneously and extending from the exit site to the insertion site), and the insertion site (the site, often associated with an infraclavicular or small lateral neck incision, where the catheter is inserted into a large central vein, such as the subclavian or internal jugular vein).

Erythema, exudate, and focal tenderness at the exit site suggest, but do not prove, the presence of infection. A quantitative increase in bacterial colony counts in culture swabs from the exit site has been associated with an increased probability that central venous catheter infection is present.321,322Staphylococcus epidermidis and C. jeikeium are the most commonly isolated colonizing microorganisms at the exit site and represent the most common etiologic agents of catheter sepsis in cancer patients.323,324 Catheter-related sepsis is suspected if bacteremia or fungemia is present unassociated with any other site of suspected infection. The predominant mechanism of infection appears to be bacterial migration from the exit site along the outside surface of the catheter.320 Suspected catheter exit-site infection, with or without bacteremia, may be treated with antimicrobials without removing the line.325 Unless exit-site surveillance cultures dictate otherwise, the empiric antibacterial therapy should include an agent such as vancomycin that is active against S. epidermidis and C. jeikeium, in addition to S. aureus and streptococci. Infection of the subcutaneous tunnel site is more difficult to control with antimicrobial agents alone and often requires catheter removal.326 Catheter removal is also often recommended in the setting of bacteremia due to more highly pathogenic organisms such as S. aureus, P. aeruginosa, or Serratia marcescens, catheter-related fungemia, or persistent catheter-related bacteremia that has not responded to appropriate antibacterial therapy. The differential time-to-positivity (i.e., the time difference between the time that the blood culture from the peripheral site and central venous catheter site was obtained to the time the cultures became positive) of >2 hours has been used as a method for predicting catheter-related infections.327

Antibacterial Prophylaxis

Antibacterial chemoprophylaxis is used widely for preventing or modifying the etiology of bacterial infection in patients for whom the expected duration of neutropenia is longer than 7 days. Oral nonabsorbable antimicrobial regimens consisting of agents such as aminoglycosides including neomycin, gentamicin, or tobramycin, vancomycin, polymyxin B, colistin, and oral nystatin or amphotericin B have not been consistently effective for reducing the incidence of febrile neutropenic episodes, documented superficial or invasive infection, or overall mortality. In addition, they are unpalatable and costly. On the other hand, oral absorbable antibacterial regimens consisting of trimethoprim-sulfamethoxazole (TMP-SMX) or fluoroquinolones including norfloxacin, ciprofloxacin, enoxacin, ofloxacin, levofloxacin, or perfloxacin have proved useful, although investigators have cautioned that efficacy appears to be intimately linked with compliance,328 personal hygiene, the spectrum of antimicrobial activity of the regimen,329 the cytotoxic potential of the antineoplastic regimen, and the timing of the administration of the regimen relative to the onset of the neutropenia-related risk for bacterial infection.17,330

The experience with oral fluoroquinolones has shown a significant reduction in the morbidity and mortality due to infection by aerobic gram-negative bacilli compared with TMP-SMX.17,331 The tradeoff for this appears to be an increase in the risk of infection due to gram-positive organisms such as coagulase-negative staphylococci, Enterococcus spp., and viridans group streptococci. The presence of a tunneled indwelling central venous catheter adds to the risk for infections due to coagulase-negative staphylococci,332 whereas severe mucositis and periodontal disease appear to predispose to viridans streptococcal infection.34,35,206,333 A syndrome of viridans streptococcal bacteremia has been recognized among high-dose cytarabine or BMT recipients that is often associated with pulmonary infiltrates and hypotension.34,334,335 The pathogenesis is believed to involve severe cytotoxic therapy–induced intestinal mucosal damage in the setting of severe prolonged neutropenia and gastrointestinal luminal colonization by these organisms. Oral fluoroquinolone use tends to select for these microorganisms.336 Of clinical importance is the inconsistent susceptibility of these organisms to penicillin G and the apparent need to modify the empiric antibacterial regimen by the addition of intravenous vancomycin to improve the likelihood of a successful outcome for the febrile neutropenic episode. Bacterial infections among TMP-SMX recipients have been due to coagulase-negative staphylococci, viridans streptococci, and TMP-SMX–resistant aerobic gram-negative bacilli such as P. aeruginosa.330

Four meta-analyses have been published examining the role of fluoroquinolone-based antibacterial chemoprophylaxis in neutropenic cancer patients.63,167,337,338 Within each of these publications are contained multiple meta-analyses. Cruciani and colleagues examined 13 trials comprising 1155 randomized subjects comparing fluoroquinolones to TMP-SMX, nonabsorbable antimicrobials, or placebo.63 These investigators also reported on a meta-analysis examining the efficacy of fluoroquinolones plus additional agents that augmented coverage for gram-positive microorganisms.63,167 Rotstein and associates reported on four meta-analyses; namely, five trials with 394 randomized subjects comparing fluoroquinolones to placebo, 16 trials with 1362 randomized subjects comparing fluoroquinolones to other antibacterial agents, 3 trials with 902 randomized subjects comparing ciprofloxacin to other fluoroquinolones, and 6 trials with 922 subjects comparing fluoroquinolones plus additional gram-positive coverage to control regimens.337 Lastly, Engels and colleagues reported two meta-analyses; first, 9 trials with 731 randomized subjects comparing fluoroquinolones and untreated controls, and second, 9 trials with 677 randomized subjects comparing fluoroquinolones to TMP-SMX.338

These systematic reviews were able to detect prophylactic treatment effects for the fluoroquinolones in a variety of outcomes, including microbiologically documented infection overall, gram-negative infections overall, and gram-negative bacteremia regardless of whether the controls were placebo, no treatment, or TMP-SMX. Infection-related mortality and overall mortality were not affected by fluoroquinolone prophylaxis. Reduction in the incidence of febrile episodes was demonstrable only in the 8 placebo-controlled trials (OR 0.85; 95% CI 0.75 to 0.99); however, this effect was observed only among unblinded trials and not blinded trials.338 Treatment effects were not observed for clinically documented infections or for gram-positive infections.

The combination of fluoroquinolones with agents with additional gram-positive activity such as penicillin, macrolides, or rifampin effectively prevents gram-positive bacteremias as well.54,339–341 A recent updated meta-analysis from the University of Padua encompassing 1202 randomized subjects in 9 trials comparing fluoroquinolones plus augmented gram-positive coverage to fluoroquinolones alone demonstrated a reduction in total bacteremic episodes (relative risk 1.54; 95% CI 1.26 to 1.88), streptococcal infections (RR 2.2; 95% CI 1.44 to 3.37), coagulase-negative staphylococcal infections (RR 1.46; 95% CI 1.04 to 2.04), and incidence of fever (RR 1.08; 95% CI 1.00 to 1.16).167 The incidence of clinically documented infections, unexplained fever, and infection-related mortality were not affected. Gram-positive prophylaxis, however, increased the incidence of prophylaxis-related drug toxicities (RR 0.46; 95% CI 0.28 to 0.76), particularly with the use of rifampin.167

Based upon these observations, antibacterial chemoprophylaxis using oral fluoroquinolones under circumstances in which the prevalence of fluoroquinolone-resistant Escherichia coli (FREC) is low (<3% to 5%) can reliably reduce the risk for invasive gram-negative bacillary infection, and if supplemented by gram-positive agents such as rifampin, penicillin, or macrolides, can reduce the risk for invasive infections due to gram-positive microorganisms including viridans streptococci and coagulase-negative staphylococci. Fluoroquinolone-based chemoprophylaxis does not reliably reduce the incidence of febrile neutropenic episodes, neutropenic episode-related mortality, or physician-initiated systemic antimicrobial prescribing behavior.

Under the appropriate conditions, it is possible that fluoroquinolone-based antibacterial chemoprophylaxis can influence physician prescribing behavior for febrile neutropenic episodes.342–344 A study from Duke University among autologous hemopoietic transplant recipients with ciprofloxacin prophylaxis demonstrated that febrile neutropenic episodes could be safely treated with an empiric glycopeptide-based regimen.343 A study from the University of Manitoba, Canada, demonstrated that patients developing febrile neutropenic episodes while receiving ciprofloxacin prophylaxis during remission-induction therapy for acute myeloid leukemia could be treated safely and effectively with a vancomycin plus ceftazidime–based strategy wherein the ceftazidime was discontinued before the patient defervesced, provided the serial rectal surveillance cultures and 24- to 36-hour blood cultures revealed no evidence of aerobic gram-negative bacilli.344 In both these studies, the oral ciprofloxacin prophylaxis regimen was continued throughout the treatment for the febrile neutropenic episode. A third study from Europe demonstrated that empiric systemic antibacterial therapy for febrile neutropenic episodes could be safely discontinued after 72 to 96 hours if the initial work-up failed to provide evidence for clinically or microbiologically documented infection and if prophylaxis was continued.342 These observations, while provocative, have not been followed up in large randomized controlled studies.

Not all fevers in neutropenic patients represent infection; thus fever is a poor outcome for clinical trials of antibacterial prophylaxis in this patient population. Better discriminators for infection are needed before the kinds of strategies suggested by these trials can enter the domain of standard practice.345 Although the results of antibacterial prophylaxis studies have not yet had a major influence on how physicians use empiric antibacterial therapies, there is reason to believe that they will do so in the future.345 The prophylaxis-related decrease in documented infections has been offset by an increase in unexplained fevers.54 It is possible that these unexplained fevers may be due to increased absorption of pyrogenic endotoxins through cytotoxic therapy–induced damage to the intestinal epithelium.346 This consideration, if true, suggests that at least some unexplained fevers do not require continued antibacterial therapy.

Antifungal Prophylaxis

The major goals of antifungal prophylaxis strategies are to reduce the morbidity and mortality due to superficial and invasive opportunistic fungal infections, and to reduce the use of toxic expensive antifungal therapy. Prophylactic strategies should be applied with a clear understanding of the pathogenesis of the microorganisms involved.

Filamentous Fungi

Filamentous fungi such as Aspergillus species, the dematiaceous fungi, Fusarium spp., Zygomycetes, and Scedosporium spp. are acquired by inhalation of spores called conidia. The conidia germinate on the respiratory epithelium to produce invasive hyphae. There are three possible ways to prevent this. First, patients may be managed in nursing units outfitted with HEPA filtration systems. This reduces the concentration of airborne conidia and the risk of patient exposure. Although this approach is effective for reducing the risk of filamentous fungal infection,286,287,347 it is expensive and has no impact for patients exposed to high concentrations of airborne conidia outside the nursing unit or for those who are already infected before entering the unit. Second, topical agents such as amphotericin B sprayed by aerosol into the nares theoretically might reduce the risk of conidial germination. One randomized study from the Institute Jules Bordet in Brussels348 evaluated the use of intranasal amphotericin B (5 mg/mL in sterile water with a total daily dose of 10 mg in three divided doses) in 90 neutropenic episodes. There was no significant difference in the empiric use of intravenous amphotericin B (35% versus 27%); however, only 1 of 46 recipients of aerosolized amphotericin B developed suspected or proven invasive aspergillosis compared with 7 of 44 controls. Although this is encouraging, intranasal amphotericin cannot be accepted as a satisfactory alternative to air filtration until further studies are done. Inhalation of aerosolized amphotericin B has been studied as a strategy to reduce invasive aspergillosis.349 The incidence of possible, probable, or proven invasive aspergillosis in the aerosolized amphotericin B recipients was 4% compared to 7% in untreated control subjects. Furthermore, there were no differences in overall mortality or in infection-related mortality. Third, systemic antifungal therapy might prevent the progression of hyphal growth once germination occurs. Systemic amphotericin B plus 5-FC has been used successfully to prevent reactivation of previously documented invasive pulmonary aspergillosis among leukemia patients undergoing further postremission cytotoxic therapy.316 Since this combination may be myelotoxic as well as nephrotoxic, it may be prudent to reserve this approach for those in whom opportunistic filamentous fungal infection has been proved by microbiologic or histopathologic methods. The prophylactic role of newer approaches such as lipid formulations of amphotericin B, echinocandins including caspofungin and micafungin, or the extended-spectrum triazole antifungal agents such as itraconazole, voriconazole, or posaconazole, with greater in vitro activity against Aspergillus species is being studied.

Itraconazole, a lipophilic extended-spectrum azole, has been evaluated for antifungal prophylaxis in a number of trials of very heterogeneous patient populations.350–352 A recent meta-analysis of these trials failed to identify a prophylactic benefit of itraconzole against fungal disease due to Aspergillus spp.247 However, studies performed in patients at higher risk for invasive aspergillosis have been more positive.353,354 In hemopoietic stem cell allograft recipients Winston and colleagues evaluated itraconazole administered intravenously 200 mg twice daily for 2 days, then 200 mg intravenously daily for 12 days, then 200 mg orally daily until day 100, compared to fluconazole 400 mg given intravenously daily for 14 days, then 400 mg orally until day 100.353 There was a significant reduction in the overall incidence of proven invasive fungal infection (OR 0.37; 95% CI 0.15 to 0.88), but the protective effect on mold infection was not statistically significant (OR 0.38; 95% CI 0.11 to 1.32). A similar study from the Fred Hutchison Cancer Center was able to demonstrate a significant reduction in the overall incidence of invasive fungal infection (15% to 7%; OR 0.46; 95% CI 0.56 to 0.99) and in invasive mold infections (12% to 5%; OR 0.44; 95% CI 0.17 to 0.98).354 Despite this promising result, treatment-related adverse events necessitating treatment withdrawal resulted in a higher overall mortality among itraconazole recipients. The study drug was discontinued for reasons of intolerance in 36% of itraconazole recipients compared to only 16% of fluconazole recipients.354 A recent meta-analysis from the University of Bonn, Germany, critically examined the efficacy of itraconazole for antifungal prophylaxis against invasive fungal infection in 13 randomized controlled trials encompassing 3597 randomized subjects, and demonstrated reductions in the mean relative risks for invasive fungal infection overall (relative risk reduction [RRR] 40% ± 13%), invasive yeast infection (RRR 53% ± 19%), and fungal infection–related mortality (RRR 35% ± 17%) in neutropenic cancer patients.355 The risk for invasive aspergillosis was reduced only in trials in which the cyclodextrin-based oral or intravenous formulations of itraconazole were used (OR 0.52; 95% CI 0.30 to 0.90; RRR 48% ± 21%).355

Echinocandin antifungal agents may also be useful for prophylaxis in high-risk patient populations. A study comparing a newer echinocandin, micafungin, to fluconazole in a heterogeneous group of autologous and allogeneic transplant recipients demonstrated a significant reduction in the use of empiric systemic antifungal therapy among the micafungin recipients (15% compared to 21%; p = 0.018).356 The incidence of invasive aspergillosis was 0.2% and 1.5% in micafungin and fluconazole recipients, respectively (p = 0.07).

Opportunistic Yeasts

Yeast infections, primarily Candida spp., colonize the patient and cause infection by invading damaged integumental surfaces. HEPA filtering plays no role in the prevention of these infections. The use of topical agents such as nystatin or amphotericin B has had a small impact on colonization profiles, but no significant impact on the incidence of invasive fungal infection or the need to use empiric systemic amphotericin B for suspected invasive fungal infection. Topical chlorhexidine mouth rinses (0.12% chlorhexidine digluconate three times daily) have been effective in reducing the morbidity of oropharyngeal candidiasis in a series of marrow allograft recipients.208 Furthermore, a reduction in candidemia also was noted, suggesting the oropharynx as a possible source for these events. The overall value of this strategy must be evaluated in further trials in different populations at risk.

Systemic antifungal therapy for the prevention of opportunistic yeast infections is better established.246,247,355,357 Published studies using the imidazoles ketoconazole, clotrimazole, and miconazole have demonstrated a reduction in yeast colonization, but have failed to demonstrate a consistent reduction in clinical disease or the need to use empiric systemic amphotericin B. There also has been a selection for more resistant yeasts such as Candida krusei and Candida glabrata in the surveillance cultures of azole recipients.242 The triazole antifungal agents fluconazole and itraconazole have been shown to be effective in reducing the need for empiric antifungal therapy with amphotericin B, superficial fungal infection, proven invasive yeast infection, and fungal infection–related mortality.247 Furthermore, these agents are effective when applied to patient populations with a high risk for invasive fungal infection (10% to 15%),246,247 when doses of 400 mg or more daily are administered,247,355 and when the agents are absorbed.355

It is recommended that antifungal prophylaxis be administered only to defined populations at highest risk for invasive fungal infection and for whom clinical trials have been able to demonstrate a treatment effect. Accordingly, the populations of patients for whom this applies include patients undergoing remission-induction or reinduction therapy for acute leukemia, patients undergoing allogeneic hemopoietic stem cell transplantation, and those undergoing autologous hemopoietic stem cell transplantation without hemopoietic growth factor support.358,359 Furthermore, the duration of prophylaxis should be from day 1 of the cytotoxic regimen until neutrophil recovery for acute leukemia patients. For allogeneic hemopoietic stem cell transplant recipients, treatment from the first day of conditioning until day 75 to 100 is recommended.347,360

Antiviral Prophylaxis

Reactivation of herpes simplex virus (HSV) infection is one of the most common causes of oropharyngeal and esophageal mucositis in patients undergoing remission-induction for leukemia or BMT.361 This complication is painful and can substantially impair adequate nutritional intake and drug administration. Herpes mucositis, stomatitis, or esophagitis can be prevented largely by prophylactic use of acyclovir. Patients at risk for these infections are those with a history of previous herpetic infection. They may be reliably identified by a clear history of the typical vesicular lesions of herpetic stomatitis or by the identification of IgG antibodies against HSV in their sera. Between 60% and 80% of HSV-seropositive patients will reactivate during cytotoxic therapy. It has been recommended that patients undergoing remission-induction therapy, consolidation therapy, or salvage therapy for acute leukemia and those undergoing bone marrow allografting or autografting who are IgG-seropositive for HSV-1 are candidates for acyclovir prophylaxis.207,362,363 Oral and intravenous routes of administration have been studied and found effective. It remains unclear whether oral administration is as effective as intravenous administration in patients receiving regimens highly toxic to intestinal mucosal surfaces, such as high-dose cytarabine or etoposide-containing regimens or BMT conditioning regimens. Acyclovir in doses of 250 mg/m2 administered intravenously every 12 to 8 hours or oral acyclovir administered in doses of 200 to 400 mg four to five times daily has prevented HSV mucositis successfully.207 Prophylaxis for 1 month or less from initiation of treatment has been associated with recurrences in 58% to 70% of patients after the acyclovir was discontinued. Accordingly, it has been recommended that prophylaxis be continued for 6 weeks from the beginning of induction or conditioning,207 or up to 6 weeks following marrow recovery in BMT recipients. Acyclovir doses of 800 mg orally every 12 hours appear to be effective for this. Valacyclovir has been shown to be as effective as acyclovir.