Aggressive chemotherapy with or without total body radiation leads to leukopenia, among other side effects. Infection is the major consequence of neutropenia. Infection incidence and severity have an inverse relation to the granulocyte count and to its rate of decrease. Most infections (and nearly all severe infections) and bacteremias develop when the granulocyte count is less than 0.1 x 10⁹/L (<100/µL). Most severe infections (about 85%) are caused by three gram-positive cocci (-hemolytic streptococcal species, Staphylococcus epidermidis, and Staphylococcus aureus), three gram-negative rods (Escherichia coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa), and two types of fungi (Candida species, especially albicans and tropicalis, and Aspergillus species, especially flavus and fumigatus). Today, gram-positive infections tend to be more common than gram-negative infections, and Pseudomonas aeruginosa infections have become infrequent at many centers. Gram-negative bacillary and streptococcal bacteremias in this setting often lead to rapid death if not treated immediately. Fungal infections are more likely to develop after prolonged periods of neutropenia and antibiotic-induced alterations of alimentary canal and respiratory tract flora [1].

Much effort has been spent attempting to prevent infections during neutropenic episodes. It is convenient to divide these approaches into four major categories. The first is to avoid invasive procedures wherever possible, such as being exquisitely careful about the care of an indwelling venous catheter. The second approach is to prevent new potential pathogens from colonizing the patient through the major routes of acquisition, that is, food, hands, water, and air. No matter how much the acquisition of new organisms is reduced, all patients harbor many of the organisms that can and do cause infection as a result of neutropenia in association with chemotherapy-damaged mucosa, intravenous catheter penetration of the skin, and chemotherapy-damaged respiratory ciliary function. Thus, for the few patients who will have prolonged (>10 days), profound granulocytopenia (a granulocyte count of <0.1 x 10⁹/L), some experts would advise the third approach: administering antimicrobial agents to suppress organisms that are particularly likely to cause infection, that is, the gram-negative rods, gram-positive cocci, and fungi noted earlier. Finally, the fourth approach is to improve host defenses. Thus, we come to the use of growth factors such as G-CSF or GM-CSF, which have dramatically affected our ability to improve host defenses after chemotherapy.

The two currently available growth factors are G-CSF and GM-CSF [2, 3]. Several studies of either G-CSF or GM-CSF have been published [4-6], and it has become clear that the addition of the growth factor reduces the number of days of leukopenia and neutropenia after aggressive chemotherapy in patients with advanced cancer. Most important is the reduction in the number of days of neutropenia with a neutrophil count of less than 0.1 x 10⁹/L. During this period, the patient is at particularly high risk for developing gram-negative rod bacteremia, a serious gram-negative or gram-positive infection, or a fungal infection. Once the granulocyte count begins to increase, the risk for serious infection, but not all infections, decreases sharply.

One of the key issues in cancer therapy today is how to safely give even higher doses of chemotherapy. Many agents have a steep dose-response curve against many tumors [7]. It is similarly recognized that physicians often give less than the known optimal therapy, especially with recurrent cycles of combination or single agents, because of progressive neutropenia and concern for infection. Thus, the advent of therapy with growth factors has allowed the testing of the concept that increased doses of chemotherapy will lead to greater tumor kill and longer patient survival and of the possibility of assuring greater safety in using more standard dosages.

One possible benefit of autologous bone marrow transplantation is that it allows the administration of a greater dose of chemotherapy than could be used otherwise [8]. More recently, peripheral blood progenitor cells have been collected with little difficulty by pheresis techniques; these cells can be stored and reinfused without the need for the more difficult and painful bone marrow extraction process. Growth factors clearly have been proven useful with bone marrow transplantation or peripheral blood progenitor cell transplantation. They are particularly effective if used to prime the peripheral blood progenitor cells and then used to stimulate transplanted or transfused cells in the patient. This approach, used by Peters and colleagues at Duke [9], leads to a substantially shorter period of neutropenia, decreased antibiotic use, shorter hospitalization time, and substantial reduction in cost. For example, until recently, an autologous bone marrow transplantation would have cost about $100 000 or more but is now $50 000 or less because of the dramatically reduced length of hospital stay. Many trials have documented improvements with the use of either G-CSF or GM-CSF. Side effects of G-CSF are largely limited to modest bone pain; those of GM-CSF include fever, rash, and, at doses of greater than 32 µg/kg of body weight, fluid retention [10-12]. The value of these growth factors is that they shorten the duration of severe granulocytopenia; there is little logic in giving them to patients with mild to moderate granulocyte suppression in whom the risk for serious infection is low. The added costs and moderate but annoying side effects outweigh the value of the growth factors.

It is currently less clear whether to begin therapy with growth factors in patients who become neutropenic after chemotherapy, who have not received growth factors, and who then become febrile. Because of the potential for rapid death from gram-negative and streptococcal bacteremia, the standard approach to a new episode of fever during neutropenia is the prompt administration of one or more broad-spectrum antibiotics as empiric therapy until an infection is or is not documented and the offending pathogen is identified with appropriate antimicrobial susceptibilities [13].

Should growth factors be administered concomitantly with empiric antibiotic therapy? In this issue, Maher and colleagues [14] address this question with their report of a double-blind, placebo-controlled trial of G-CSF (filgrastim) given to 216 febrile neutropenic patients with cancer. They observed an accelerated neutrophil recovery (3 days compared with 4 days of a neutrophil count of <0.5 x 10⁹/L) and a shorter duration of febrile neutropenia (5 days compared with 6 days). Overall, the median number of days of hospitalization was not reduced (8 days for both groups), although prolonged hospital stays were reduced by one half. More important, however, is the observation that the duration of profound neutropenia (that is, a neutrophil count of 0.1 x 10⁹/L) was reduced because it has been repeatedly shown that patients with persistent, profound granulocytopenia and concomitant gram-negative bacteremia have a high mortality rate despite prompt administration of antibiotics [15]. For example, an EORTC International Antimicrobial Therapy Cooperative Group trial found that such patients had an improvement rate of only 23% compared with 82% in patients whose granulocyte count recovered to greater than 0.1 x 10⁹/L [16]. It is therefore not surprising that in the study by Maher and colleagues, G-CSF had its greatest efficacy for those with documented infections and a granulocyte count of less than 0.1 x 10⁹/L.

This study did not include patients at greatest risk, that is, those with acute myelocytic leukemia or those who received an allogeneic marrow transplant. Most patients with profound (<0.1 x 10⁹/L), persistent (>7 to 10 days) granulocytopenia and severe gram-negative or streptococcal bacteremia or subsequent infections fall into these categories. Furthermore, despite the reduction in the number of days of neutropenia and febrile neutropenia, the duration of antibiotic therapy, the use of amphotericin B, and the length of hospital stay were not reduced. Thus, although G-CSF had a salutary effect, it must be considered in context, including the possibility of side effects and the cost of $200 to $300 per day.

In vitro and in vivo data also suggest that cytokines may improve neutrophil function during granulocytopenia. Bacterial challenges in animal models showed enhanced antibiotic efficacy. Cytokines may also enhance antimicrobial activity of available neutrophils in addition to stimulating their own further production. Possible mechanisms include increased superoxide production (G-CSF and GM-CSF), membrane depolarization to a stimulus (GM-CSF), chemotactic activity (G-CSF), and phagocytic activity (G-CSF). Granulocyte-macrophage colony-stimulating factors increase the production and function of monocytes and granulocytes, an effect that could be useful in fighting fungal infection [17, 18].

My recommendation, based on preclinical, prophylactic, and therapeutic studies of growth factors and the epidemiology and natural history of infections in neutropenic patients with cancer, is that cytokines (G-CSF and presumably GM-CSF) may benefit patients with febrile neutropenia just beginning to receive empiric antibiotics and in whom the granulocyte count is expected to remain at less than 0.1 x 10⁹/L. These are patients in whom the battle is between "bug and drug," with little or no help from circulating granulocytes. It therefore makes sense that stimulating the production of granulocytes and presumably enhancing the function of the few that are available would be beneficial.

Using cytokines for patients with lesser degrees of neutropenia should be avoided initially and later limited to those who prove over time to have a serious infection (for example, pneumonia, perianal cellulitis, or bacteremia with gram-negative bacilli, streptococci, Staphylococcus aureus, or fungi) and persisting neutropenia with a neutrophil count of less than 0.5 x 10⁹/L. This approach maximizes the advantages of the growth factors while limiting their expense and side effects.

In summary, until more formal guidelines are promulgated, such as those being developed by the American Society of Clinical Oncology, cytokines should only be used in patients expected to have neutrophil levels of 0.1 x 10⁹/L or less for 1 week. Few patients with febrile neutropenia will thus receive both empiric therapy with antibiotics and cytokines, but those who do should have the greatest opportunity for benefit.