65© Rapid Science Publishers ISSN 0269-9370

Filgrastim prevents severe neutropenia and reducesinfective morbidity in patients with advanced HIV

infection: results of a randomized, multicenter,controlled trial

Daniel R. Kuritzkes, David Parenti*, Douglas J. Ward†, Anita Rachlis‡,Roberta J. Wong§, Kenneth P. Mallon§, William J. Rich§,

Mark A. Jacobson¶ and the G-CSF 930101 Study Group**

Objective: To assess the effect of filgrastim treatment on the incidence of severeneutropenia in patients with advanced HIV infection, and the effect of initialfilgrastim treatment on prevention of infectious morbidity.

Design: Randomized, controlled, open-label, multicenter study.

Setting: Outpatient centers and physician offices.

Patients: Men and women aged > 13 years, who were HIV antibody-positive, and hada CD4 cell count < 200 × 106/l, absolute neutrophil count (ANC) 0.75–1.0 × 109/l,and platelet count ≥ 50 × 109/l within 7 days of randomization were eligible. Twohundred and fifty-eight patients entered and 201 completed the study.

Intervention: Daily filgrastim (starting at 1 µg/kg daily, adjusted up to 10 µg/kg daily)or intermittent filgrastim (starting at 300 µg daily one to three times per week to amaximum of 600 µg daily 7 days weekly) was administered to maintain an ANCbetween 2 and 10 × 109/l. Patients in the control group received filgrastim if severeneutropenia developed.

Main outcome measures: Incidence of severe neutropenia (ANC < 0.5 × 109/l) ordeath, incidence of bacterial and fungal infections, duration of hospitalization andintravenous antibacterial use, and safety.

Results: The primary endpoint of severe neutropenia or death was less frequent inpatients who received daily (12.8%) or intermittent (8.2%) filgrastim compared withcontrol patients (34.1%; P < 0.002 and P < 0.0001 for comparison with daily andintermittent groups, respectively). Filgrastim-treated patients developed 31% fewerbacterial infections and 54% fewer severe bacterial infections than control patients,required 26% less hospital days including 45% fewer hospital days for bacterialinfections, and needed 28% fewer days of intravenous antibacterials. Filgrastim wasnot associated with an increase in HIV-1 plasma RNA level in a subset of patients inwhom this was measured or any new or unexpected adverse events.

Conclusion: Filgrastim was safe and effective in preventing severe neutropenia inpatients with advanced HIV infection, and may reduce the incidence and durationof bacterial infections, incidence of severe bacterial infections, duration of hospitaldays for infections, and days of intravenous antibacterial agents.

AIDS 1998, 12:65–74

Keywords: AIDS, bacterial infection, hematology, hematopoietic growth factors,hospitalization, neutropenia, outpatient, randomized study

From the University of Colorado Health Sciences Center, Denver, Colorado, the *George Washington University, the †Dupont Circle Physicians Group, Washington, DC, USA, the ‡Sunnybrook Health Sciences Center, University of Toronto,Toronto, Canada, §Amgen Inc., Thousand Oaks, and the ¶University of California San Francisco and San Francisco GeneralHospital, San Francisco, California, USA. **See Appendix.Sponsorship: This study was supported by Amgen Inc., Thousand Oaks, California, USA.Requests for reprints to: Dr Daniel R. Kuritzkes, University of Colorado Health Sciences Center, Division of InfectiousDiseases, 4200 East 9th Avenue, B168, Denver CO 80262, USA. Date of receipt: 28 May 1997; revised: 10 September 1997; accepted: 15 September 1997.

AIDS 1998, Vol 12 No 166

Introduction

Neutropenia is common in patients with advancedHIV infection [1]. The causes of neutropenia in HIVinfection vary, but are primarily due to impairedhematopoiesis resulting from several factors, includingthe use of myelotoxic agents directed against HIV andopportunistic pathogens [2,3], direct infection or infil-tration of bone marrow by HIV, secondary infectionsor neoplasms [4–6], apoptotic death of uninfectedhematopoietic progenitor cells and more mature cells[7], nutritional deficiency [8], and antibodies directedagainst bone-marrow progenitor cells [9].

Several studies have shown that neutropenia is an inde-pendent risk factor for bacterial infection in patientswith HIV disease [10–15]. In addition, neutropeniamay also complicate the use of myelosuppressive ther-apy in HIV-infected patients. Zidovudine, ganciclovir,and trimethoprim–sulfamethoxazole (TMP-SMX) arefirst-line treatments for HIV disease and related oppor-tunistic infections [16,17] and can cause neutropenia.Treatment interruption or discontinuation because ofmedication-induced neutropenia may lead to subopti-mal treatment and disease progression [18,19].

Recombinant methionyl human granulocyte colony-stimulating factor (r-metHuG-CSF, filgrastim) is a bac-terially synthesized recombinant protein that increasesthe production of neutrophils by stimulating prolifera-tion and differentiation of neutrophil progenitors[20–22]. Filgrastim increases absolute neutrophil count(ANC) and enhances the phagocytic and bactericidalfunctions of neutrophils [23]. Several uncontrolled pilotstudies have shown that filgrastim increases ANC inneutropenic patients with HIV infection [24–29].

We therefore performed a randomized, controlled trialto assess the safety and efficacy of filgrastim using dosesmost often prescribed in clinical practice in preventingsevere neutropenia (ANC < 0.5 × 109/l) in HIV-infected patients. In addition, we assessed the rate andduration of bacterial and fungal infections, duration ofhospitalization and intravenous antibacterial use, andsafety.

Material and methods

Study populationThe study protocol was reviewed and approved by theappropriate institutional review boards, and all patientsgave written informed consent before study entry.Individuals aged > 13 years with documented HIV infection (antibody-positive by enzyme-linkedimmunosorbent assay, confirmed by Western blotanalysis or indirect immunofluorescence assay) were

eligible for study if they had an ANC ≥ 0.75 but < 1.0 × 109/l less than 7 days before randomization. Inaddition, patients were required to have a CD4+ lym-phocyte count ≤ 200 × 106/l, platelet count > 50 ×109/l, Karnofsky performance score >50%, and lifeexpectancy ≥ 6 months. Dosages of any antiretroviralagent (e.g., zidovudine, didanosine, zalcitabine, orstavudine), and TMP-SMX, interferon-α, and ampho-tericin B were kept constant for at least 14 days beforerandomization. Patients could be enrolled after ganci-clovir induction therapy for cytomegalovirus retinitisand were required to be on a stable maintenance doseof ganciclovir for ≥ 2 weeks before randomization.Patients were excluded if they had a malignancy (withthe exception of stable Kaposi’s sarcoma not requiringsystemic treatment with myelosuppressive therapy, orlocalized basal or squamous cell carcinoma), knownhypersensitivity to Escherichia coli-derived products,treatment with filgrastim or other hematopoieticgrowth factors (with the exception of erythropoietin)within 14 days of randomization, or if they were pregnant or breastfeeding. Use of investigational anti-retroviral agents was permitted with prior approvalfrom the sponsor.

Study design and treatment regimenThe study was a 24-week, multicenter, open-label,randomized controlled trial. Patients were randomizedto receive daily or intermittent subcutaneous doses offilgrastim, or to an observational control group in aratio of 1 : 1 : 1. Filgrastim (r-metHuG-CSF; Amgen,Inc., Thousand Oaks, California, USA) was providedin 1 ml vials containing 0.3 mg/ml for self-injection.Filgrastim dosage in both treatment groups was titratedto maintain an ANC of 2–10 × 109/l. For patients inthe daily treatment group, filgrastim dosage started at 1 µg/kg daily, and could be increased to a maximum of10 µg/kg daily; for patients in the intermittent treat-ment group, filgrastim dosage started at 300 µg dailyone to three times per week, and could be increased toa maximum of 600 µg daily. Study drug was discontin-ued in the event of a serious adverse event deemed tobe possibly, probably, or definitely related to its admin-istration, or a white blood cell count ≥ 75 × 109/l at the24 h nadir after drug administration. Patients in thecontrol group who developed severe neutropenia werere-randomized to receive filgrastim according to theregimen of one of the two treatment groups for theremainder of the 24-week study period.

Patients continued to receive conventional treatmentfor HIV at the discretion of their physicians.Neutrophil counts were monitored three times weeklyfor the first 2 weeks, and weekly thereafter. Regardlessof treatment group assignment, investigators madeadjustments to myelosuppressive drugs on the basis ofclinically significant changes in hematologic variablesaccording to the patients’ medical status and the stan-

Prevention of severe neutropenia by filgrastim in HIV Kuritzkes et al. 67

dards of clinical practice. Adjustments were not allowedduring the confirmation of severe neutropenia.

Primary study endpointsThe primary endpoint for this study was the occur-rence of confirmed severe neutropenia or death duringthe 24-week study period. A confirmed episode ofsevere neutropenia was defined as two consecutiveANC measurements < 0.5 × 109/l obtained within 7 days, but at least 24 h apart.

Secondary study endpointsSecondary endpoints included incidence of microbialinfections (bacterial and fungal), incidence of severemicrobial infections, days of hospitalization, days ofhospitalization for bacterial infections, use of myelosup-pressive therapy, and days of intravenous antibacterialtherapy.

Microbial infectionsAll presumptive and definitive diagnoses of microbialinfection were included in the analyses. For definitivediagnosis, bacterial infections were confirmed by cul-ture with the exception of pneumonia and sinusitis.Diagnosis of bacterial pneumonia required presence ofa new infiltrate on chest radiograph, and at least one ofthe following: diagnostic sputum Gram stain, or positive culture of a respiratory pathogen from bron-choalveolar lavage fluid or protected brush specimen.Fungal infections were confirmed by culture or biopsy.

The incidence of sinusitis was tabulated, but notincluded in the analysis of bacterial infections becauseof the difficulty in distinguishing bacterial from non-bacterial forms of sinusitis in HIV-infected patients. Inthe absence of confirmatory evidence, a presumptivediagnosis of bacterial or fungal infection could be madeon the basis of a compatible clinical syndrome. Theseverity of bacterial infections was determined by theinvestigators and was graded as mild, moderate, severe,or life-threatening.

HospitalizationsDays of hospitalization for all causes were counted byusing admission and discharge dates. The number ofdays that each hospitalization was prolonged because ofbacterial infections was also recorded.

Myelosuppressive and antibacterial therapyFor the analysis of myelosuppressive medications, full,high and reduced doses of ganciclovir, zidovudine, andTMP-SMX doses were defined. Ganciclovir doseswere categorized as ‘reduced’ if < 75% of the standarddose (5 mg/kg intravenously daily equivalent to 1000 mg orally three times daily), ‘full’ if 75–150%, and‘high’ if ≥ 150%. Similarly, zidovudine was classified as ‘reduced’ for < 500 mg daily doses, ‘full’ for500–600 mg daily doses, and ‘high’ for doses over 600mg daily. For TMP-SMX, doses < 600 mg daily were

considered ‘reduced’, doses 600–1000 mg daily wereconsidered ‘full’, and doses ≥ 1000 mg daily were con-sidered ‘high’ (based on the trimethoprim component).All doses were categorized in a blinded fashion. Thisallowed the calculation of days of full- and high-dosemyelosuppressive medication use.

Statistical analysisAll analyses were performed on an intention-to-treatbasis. The primary efficacy analysis was based on acomparison of daily filgrastim treatment versus control,and intermittent filgrastim versus control. An additionalanalysis of the combined groups was performed due tosimilarities in the treatment effects. The percentage ofpatients who reached the primary endpoint in eachgroup was estimated by the Kaplan–Meier method.The t test was used to compare the difference in per-centages, calculating the variance of the differenceusing Greenwood’s formula [30].

All incidence rates were reported as number of eventsper 1000 patient-days. The number of events thatoccurred from the day of randomization to the day ofstudy termination (or study day 168, whichever camefirst) were counted for each patient. These totals wereconverted to incidence rates by dividing the total num-ber of events in each treatment group by the totalnumber of at-risk days for each treatment group.Generalized linear models were used to calculate thevariances of the point estimates to allow for hypothesistesting. This quasi-likelihood model used a logarithmiclink function, which was adjusted for differential obser-vation time using an offset term, and was based on theassumption that the underlying mean was proportionalto the variance.

SafetySafety was assessed through periodic clinical and labora-tory determinations beginning at date of randomizationand continuing through week 24, or death (if earlier).Deaths that occurred up to 30 days after study termina-tion were recorded. Plasma HIV-1 RNA levels weremeasured by quantitative reverse transcriptase poly-merase chain reaction (National Genetics Institute,Culver City, California, USA) on a subset of patients atbaseline, and at weeks 12 and 24. CD4+ cell countswere obtained from all patients.

Results

Patient populationA total of 258 patients at 30 centers in the UnitedStates and Canada were randomized from July 1993 toAugust 1995. Patients in the three groups were compa-rable at baseline with respect to both demographic andclinical characteristics (Table 1). Patients for whom the

AIDS 1998, Vol 12 No 168

date of HIV infection could be estimated reported amedian duration of 3.8 years. The median ANC at thetime of randomization was 0.9 × 109/l (range,0.136–2.546 × 109/l). The median CD4 cell counts forall patients was 15 × 106/l (range, 0–186 × 106 cells/l; Table 1). The median number of prior opportunisticinfections was similar in the three groups, with theintermittent filgrastim-treated group having the greatestnumber of prior opportunistic infections.

Approximately 80% of patients were receiving antibac-terial agents at study entry. The most frequently usedantibiotics were TMP-SMX (64%), rifabutin (27.1%),and clarithromycin (23.3%). Use of TMP-SMX wassimilar in all three groups. Rifabutin use was greater inboth filgrastim treatment groups compared with thecontrol group, whereas clarithromycin use was morecommon in the intermittent filgrastim and controlgroups than in the daily filgrastim group (Table 1).

Approximately 70% of patients were receiving antifun-gal medications and 80–90% were using antiviralagents. More than 50% of the patients were receivingfluconazole, and approximately 20% were receivingganciclovir at study entry. Use of these medications wasevenly balanced amongst the three groups.

Suspected causes of neutropenia for patients enrollinginto the study were multifactorial. Investigators attrib-uted neutropenia, at least in part, to HIV infection in89.5% of the patients. Additional contributory factorsincluded antiretroviral therapy (53.9%), other myelosuppressive therapy (37.6%), and opportunisticinfection (11.6%).

Patient dispositionAll patients in the filgrastim-treated groups received atleast one dose of filgrastim. Average follow-up forpatients in the daily, intermittent, and control groups

was 147, 158, and 139 days on study, respectively.Fifty-seven patients (22.1%) withdrew from the studyprematurely (before week 24). The most common rea-sons for withdrawal was patient or investigator decision(22 patients), an adverse event (six patients), or death(13 patients). Overall, reasons for early withdrawalwere balanced amongst all three groups.

Filgrastim effect on neutrophil countPatients receiving filgrastim had a prompt and sustainedincrease in ANC (Fig. 1). The median time to reversalof neutropenia was 3 days in the daily treatment groupand 8 days in the intermittent treatment group. Theslight increase in mean ANC observed in the controlgroup was attributable to the 18 patients who receivedfilgrastim after crossing over to filgrastim treatment.

Table 1. Patient characteristics at baseline.

Daily dosing Intermittent dosing Control(n = 83) (n = 89) (n = 86)

Mean (range) age (years) 39.7 (23–65) 39.3 (25–66) 40.3 (20–60)Sex [n (%)]

Female 9 (11) 6 (7) 8 (9)Ethnicity [n (%)]

Non-white 18 (22) 21(24) 24 (29)Median (range)

ANC (× 109/l) 0.9 (0.500–1.846) 0.9 (0.136–2.112) 0.9 (0.473–2.546)CD4+ cell count (× 106/l) 14 (0–180) 12 (0–186) 20 (0–185)Prior opportunistic infections 3 (0–16) 4 (0–20) 3 (0–10)

lndwelling catheters [n (%)]Yes 21 (25.3) 22 (24.7) 25 (29.1)

Medications (%)TMP-SMX 67.5 62.9 61.6Rifabutin 31.3 30.3 19.8Clarithromycin 19.3 25.8 24.4Zidovudine 36.1 38.2 33.7Ganciclovir 22.9 21.3 22.1

ANC, Absolute neutrophil count; TMP-SMX, trimethoprim–sulfamethoxazole.

Fig. 1. Mean absolute neutrophil count (ANC) at weeks1–24. (M), Daily filgrastim dosing (n = 83); (∆), intermittentfilgrastim dosing (n = 89); (j), controls (n = 86).

Prevention of severe neutropenia by filgrastim in HIV Kuritzkes et al. 69

The crossover of these 18 patients was evenly distrib-uted throughout the 24-week study period. Patients inthe daily filgrastim treatment group who were able toreverse their neutropenia required a mean of 1.2 µg/kgdaily to maintain ANC ≥ 2 × 109/l; those receivingintermittent dosing of filgrastim required a mean oftwo 300 µg doses per week.

Primary study endpointFilgrastim significantly reduced the occurrence of theprimary endpoint (confirmed severe neutropenia ordeath; Table 2). Summaries of the individual compo-nents of this composite endpoint indicated that theeffect of filgrastim on the primary endpoint was primarily due to differences observed in the incidenceof confirmed severe neutropenia, because death rates,although slightly lower in filgrastim-treated groups,were comparable across all groups.

Secondary study endpoints

Microbial infectionsEighty-five patients developed a total of 128 new orworsening bacterial infections during the study period.The most common bacterial infections includedMycobacterium avium infection (n = 27), pneumonia (n = 14), catheter-related infections (n = 13), cellulitis

(n = 12), and gastrointestinal infections (n = 11). Themost frequently isolated microorganisms were M. aviumcomplex (MAC), streptococci, Klebsiella pneumoniae,and Pseudomonas aeruginosa. MAC was the most com-mon cause of severe infection in all treatment groups.

Both daily and intermittent dosing with filgrastim low-ered the incidence of bacterial infections comparedwith the control group (Table 3, Fig. 2). Overall, fil-grastim-treated patients developed 31% fewer bacterialinfections than patients in the control group (P = 0.07).The magnitude of this effect was similar in analyses thatadjusted for CD4+ lymphocyte count and number ofprior opportunistic infections and the results did notdepend on baseline concomitant medication use. Theeffect of filgrastim among patients receiving zidovudine[relative risk (RR), 0.63] was similar to those notreceiving zidovudine (RR, 0.73). Patients receivingganciclovir (RR, 0.72) or not receiving ganciclovir(RR, 0.69) had similar effects of filgrastim treatment.

Likewise, filgrastim treatment resulted in statisticallysignificant reductions in the risk of severe bacterialinfections (relative hazard for combined treatmentgroups versus control, 0.46; P = 0.005). Treatmentwith filgrastim did not alter the incidence of fungalinfections in this study (4.14 per 1000 patient-days inthe treated groups versus 4.50 per 1000 patient-days inthe control group).

Although rifabutin use was seen to be greater in the fil-grastim-treated patients at study entry, analyses thatadjusted for rifabutin use did not change the results.

HospitalizationsFilgrastim-treated patients spent fewer days in the hos-pital than control patients. The overall effect of filgras-tim on hospitalization days was not statisticallysignificant; however, its effect on reducing days of hos-

Table 2. Primary endpoint results.

n (%)*

Daily Intermittent Controldosing dosing group

Event (n = 83) (n = 89) (n = 86)

Confirmed severe neutropenia or death 10 (12.8)† 7 (8.2)‡ 26 (34.1)

Confirmed severe neutropenia 1 (1.2) 2 (2.2) 19 (22.1)Death 9 (10.8) 6 (6.7) 10 (11.6)

*Percentages based on Kaplan–Meier estimates. †P < 0.002 versuscontrol (t-test applied to Kaplan–Meier estimates). ‡P < 0.0001 versus control (t-test applied to Kaplan–Meier estimates).

Table 3. Relative risk (RR) of various outcomes for all filgrastim-treated patients versus control patients.

Unadjusted Adjusted*

RR (95% CI) P RR (95% CI) P

Bacterial infectionDaily 0.67 (0.41–1.11) 0.12 0.66 (0.40–1.07) 0.09Intermittent 0.70 (0.44–1.13) 0.14 0.64 (0.41–1.23) 0.06Combined 0.69 (0.46–1.03) 0.07 0.65 (0.30–0.97) 0.03

Severe bacterial infectionDaily 0.60 (0.30–1.20) 0.15 0.59 (0.30–1.16) 0.13Intermittent 0.33 (0.16–0.68) 0.003 0.29 (0.14–0.60) 0.0008Combined 0.46 (0.26–0.79) 0.005 0.43 (0.25–0.74) 0.002

Hospital daysDaily 0.72 (0.42–1.24) 0.24 0.70 (0.41–1.19) 0.19Intermittent 0.76 (0.45–1.28) 0.31 0.70 (0.42–1.16) 0.17Combined 0.74 (0.47–1.16) 0.19 0.70 (0.45–1.07) 0.10

Bacterial infection-related hospital daysDaily 0.51 (0.23–1.11) 0.09 0.50 (0.23–1.09) 0.08Intermittent 0.58 (0.28–1.22) 0.15 0.53 (0.25–1.09) 0.09Combined 0.55 (0.30–1.00) 0.05 0.51 (0.28–0.92) 0.03

*Baseline CD4 cell count and number of prior opportunistic infections. CI, Confidence interval.

AIDS 1998, Vol 12 No 170

pitalization for bacterial infections (Table 3, Fig. 2)appeared stronger. This result suggests that the reduc-tion in hospital days overall was primarily due to thespecific effect of filgrastim on the incidence of bacterialinfections.

Myelosuppressive and intravenous antibiotic therapyNinety-eight of the 258 (38%) patients were takingtwo or more myelosuppressive drugs (including ganci-clovir, zidovudine, and TMP-SMX) at study entry. Six(20%) out of 30 control group patients and none of the68 filgrastim-treated patients taking two or more ofthese agents became severely neutropenic. Seven out of18 control group patients who were receiving ganci-clovir at study entry became severely neutropenic,compared with none of the 39 filgrastim-treatedpatients receiving ganciclovir at study entry (P = 0.0001). Patients in filgrastim-treated groups

received 12% more days of full-dose or high-dosemyelosuppressive medications in filgrastim-treatedgroups compared with the control group (data notshown). This increase was observed with the nearabsence of severe neutropenia. In addition, patients inthe filgrastim-treated groups required 28% fewer daysof intravenous antibacterial use than patients in thecontrol group.

SafetyAll randomized patients were evaluable for safety assess-ments. At least one adverse event was reported by98.3% of filgrastim-treated patients and 94.2% of con-trol group patients. The most frequently reportedadverse events were fever, diarrhea, fatigue, nausea,headache, anemia, abdominal pain, vomiting, andmyalgia. Overall, there were no unexpected clinicallysignificant differences in adverse events in the three

Fig. 2. (a) Incidence of bacterial infections, all patients. (b) Bacterial infection days, all patients. (c) Hospital days, all patients.(d) Hospital days due to bacterial infections, all patients. (M), Daily filgrastim group (n = 83); ( ), intermittent filgrastim group(n = 89); ( ), all filgrastim-treated patients combined (n = 172); (m), control patients (n = 86).

(a) (b)

(c) (d)

Prevention of severe neutropenia by filgrastim in HIV Kuritzkes et al. 71

patient groups. At least one severe adverse event wasreported by 39.5% of filgrastim-treated patients and44.2% of control patients. The most common severeadverse events were thrombocytopenia and anemia.Severe thrombocytopenia occurred in 12 (7.0%) out of172 filgrastim-treated patients and in three (3.5%) outof 86 control patients. The incidence of severe anemiawas similar in all groups.

Filgrastim treatment was not associated with an increasein plasma HIV-1 RNA levels over 24 weeks in thesubset of patients in whom this was tested (Fig. 3). Inaddition, there was no effect of filgrastim on CD4+lymphocyte count.

Discussion

This study shows that filgrastim was safe and effectivein preventing severe neutropenia in patients withadvanced HIV disease and mild neutropenia (ANC0.75–1.0 × 109/l). Administering filgrastim whenpatients are mildly neutropenic rather than waiting forpatients to become severely neutropenic (as in our con-trol group) was associated with reduction in the inci-dence of bacterial infections, including serious bacterialinfections, and reduction in days of hospitalization forbacterial infection in patients. Trends toward reduceduse of intravenous antibacterial medications as well asincreased use of myelosuppressive agents in the filgras-tim-treated group were also seen. These results suggestthat use of filgrastim to maintain a normal neutrophilcount confers important clinical benefits in patientswith advanced HIV disease.

In this study, a median daily filgrastim dose of 1 µg/kgdaily or two 300 µg doses per week were required tomaintain an ANC of 2–10 × 109/l. Similar doserequirements were reported in a non-controlled studyof filgrastim in 200 HIV-infected patients with ANC

< 1.0 × 109/l [31]. In that study, filgrastim reversedneutropenia in 98% of patients with a median dailydose of 1.0 µg/kg during the initial treatment phase.During the maintenance phase, a median of three 300µg doses per week were required to maintain an ANCof 2–5 × 109/l. These doses are substantially lower thanthe median dose required in the oncology setting. As inthe previous report [31], patients who received filgras-tim in the present study were better able to tolerateadministration of myelosuppressive medications,including zidovudine, TMP-SMX, and ganciclovir.

The beneficial effects of filgrastim on neutrophil func-tion are multifactorial. In vitro studies with purifiedneutrophils have demonstrated that filgrastim increasesrespiratory burst and phagocytic activity, and improvesbacterial cell killing [23,32]. Each of these functions isgreater in recently produced neutrophils.Administration of filgrastim to humans increases therate of neutrophil production, resulting in the release oflarge numbers of ‘young’ neutrophils. It is possible thatthe clinical benefits of filgrastim seen in the setting ofadvanced HIV disease are due not only to the increasein neutrophil number, but also to an intrinsic improve-ment in neutrophil function.

Previous studies have shown that the risk of bacterialinfection in HIV-infected individuals increases withdecreasing neutrophil count [11,15]. This risk is great-est among patients with severe neutropenia (< 0.5 ×109/l). Because of concerns for the safety of patients inthe control group, this study included a cross-overdesign that provided filgrastim treatment to subjectswho developed severe neutropenia. The effect of thisdesign was to limit the power of the study to detect sta-tistically significant differences between treatmentgroups with regard to the secondary endpoints. Patientswere limited to the extent of their neutropenia becausetheir physicians acted quickly to cross them over to filgrastim treatment. Nevertheless, a consistently favor-able effect of filgrastim on the incidence of a variety ofsecondary endpoints was observed.

Another consequence of the study design was the lowincidence of bacterial and fungal infections overall. Therate of bacterial infections in the control group was4.25 per 1000 patient-days. This rate was reduced to2.93 per 1000 patient-days in the combined filgrastim-treated groups. It is likely that the rate of bacterialinfections in the control group would have been evenhigher if we had allowed the ANC to decrease to < 0.5 × 109/l for extended periods of time. Evidencefrom previous studies suggests that filgrastim is highlyeffective in correcting neutropenia in this group ofpatients [28,29,31]. Thus, the clinical benefit observedin the present study may be taken as a minimum esti-mate of the efficacy of filgrastim at preventing bacterialinfections in patients with more serious degrees of neu-

Fig. 3. Change in log10 plasma HIV-1 RNA titer from base-line to week 24.

AIDS 1998, Vol 12 No 172

tropenia. Because assignment to treatment group wasnot blinded, the potential for bias existed in the diag-nosis of bacterial infections and grading the severity ofthese infections. This concern may be partially allayedsince 55% of bacterial infections were microbiologicallyconfirmed.

Another study has shown that use of antimicrobialagents such as TMP-SMX significantly reduce the riskof bacterial infection in patients with HIV disease [33].In this study, use of TMP-SMX significantly reducedthe risk of bacterial infection by nearly twofold.However, statistical adjustment for TMP-SMX use didnot change the results.

Use of macrolides or rifamycins for prevention of dis-seminated infection due to MAC might also contributeto a reduction in bacterial infections in addition toreducing the incidence of MAC. At the time this studybegan, MAC prophylaxis was not yet routinely recom-mended for patients with advanced HIV disease.Although rifabutin use increased over the time of thestudy, only 40% of patients received prophylacticrifabutin. Since disseminated MAC infections consti-tuted a significant number of the bacterial infectionsobserved, it is possible that wider use of MAC prophy-laxis might reduce the differences in rates of infectionbetween the filgrastim and control groups.

Patients receiving filgrastim spent fewer days in thehospital overall, and had significantly fewer days of hospitalization for bacterial infection. There was also atrend toward less frequent use of intravenous antibacte-rial agents. Although a formal pharmaco-economicanalysis has not been carried out, treatment that reducesthe incidence of serious bacterial infections and reducesdays of hospitalization is likely to have a substantialeconomic impact on the management of patients withadvanced HIV disease.

Changes in the management of HIV infection broughtabout by the availability of more potent antiretroviralagents and improvements in the management and pre-vention of opportunistic infections may reduce theincidence of neutropenia among HIV-infected individ-uals. For those patients who do develop neutropenia,results of this study suggest that filgrastim was safe andeffective in preventing severe neutropenia and its com-plications in patients with advanced HIV disease. Theresults of the study indicate the clinical benefit of treat-ing mildly neutropenic patients rather than waitinguntil severe neutropenia occurs. Additional studies areneeded to define the optimum strategy for use of filgrastim in HIV-infected patients and to address therole, if any, of filgrastim as adjuvant therapy for specificbacterial or fungal infections.

Acknowledgement

The authors thank M.A. Foote for assisting in thepreparation of the manuscript.

References

1. Aboulafia DM, Mitsuyasu RT: Hematologic abnormalities inAIDS. Hematol Oncol Clin North Am 1991, 5:195–214.

2. Fischi M, Parker CB, Pettinelli C, et al.: A randomized controlledtrial of a reduced daily dose of zidovudine in patients with theacquired immunodeficiency syndrome. N Engl J Med 1990,323:1009–1014.

3. Sawka CA, Shepherd FA, Brandwein J, Burkes RL, Sutton DM,Warner E: Treatment of AIDS-related non-Hodgkin’s lymphomawith a twelve week chemotherapy program. Leuk Lymphoma1992, 8:213–220.

4. Castelia A, Croxson TS, Mildvan D, Witt DH, Zalusky R: Thebone marrow in AIDS. A histologic, hematologic, and microbio-logic study. Am J Clin Pathol 1985, 84:425–432.

5. Moses AV, Williams S, Heneveld ML, et al.: Human immunode-ficiency virus infection of bone marrow endothelium reducesinduction of stromal hematopoietic growth factors. Blood 1996,87:919–925.

6. Nichols L, Florentine B, Lewis W, Sattler F, Rarick MU, BrynesRK: Bone marrow examination for the diagnosis of mycobacter-ial and fungal infections in the acquired immunodeficiency syn-drome. Arch Pathol Lab Med 1991, 115:1125–1132.

7. Zauli G, Vitale M, Re MC, et al.: In vitro exposure to humanimmunodeficiency virus type 1 induces apoptotic cell death offactor-dependent TF-1 hematopoietic cell line. Blood 1994,83:167–175.

8. Rule SA, Hooker M, Costello C, Luck W, Hoffbrand AV: Serumvitamin B12 and transcobalamin levels in early HIV disease. AmJ Hematol 1994, 47:167–171.

9. Donahue RE, Johnson MM, Zon LI, Clark SC, Groopman JE:Suppression of in vitro haematopoiesis following humanimmunodeficiency virus infection. Nature 1987, 326:200–203.

10. Shaunak S, Bartiett JA: Zidovudine-induced neutropenia: are wetoo cautious? Lancet 1989, ii:91–92.

11. Moore RD, Keruly JC, Chaisson RE: Neutropenia and bacterialinfection in acquired immunodeficiency syndrome. Arch InternMed 1995, 155:1965–1970.

12. Caperna J, Barber E, Mathews WC: Estimating neutropenic riskof bacteremia among HIV-infected patients by secondary analy-sis of clinical and laboratory data: a retrospective cohort.American Federation of Clinical Research Regional Meeting.Carmel, California, February 1996 [abstract 125].

13. Eng RHK, Yen K, Tecson-Tumang F, Smith SM, Akgun K: Risk ofgram-negative bacteremia during neutropenia in patients withAIDS. Infect Dis Clin Pract 1994, 3:373–375.

14. Keiser P, Higgs E, Smith J: Neutropenia is associated with bac-teremia in patients Infected with the human immunodeficiencyvirus. Am J Med Sci 1996, 312:118–122.

15. Jacobson MA, Liu RC-C, Davies D, Cohen PT: HIV disease-related neutropenia and the risk of hospitalization for seriousbacterial Infection. Arch Intern Med 1997, 157:1825–1831

16. Carpenter CC, Fischl MA, Hammer SM, et al.: Antiretroviraltherapy for HIV infection in 1996. Recommendations of aninternational panel. International AIDS Society USA. JAMA1996, 276:146–154.

17. Pluda JM, Mitsuyasu H, Yarchoan R: Hematologic effects ofAIDS therapies. Hematol Oncol Clin North Am 1991,5:229–248.

18. Bozzette SA, Finkelstein DM, Spector SA, et al.: A randomizedtrial of three antipneumocystis agents in patients with advancedhuman immunodeficiency virus infection. NIAID AIDS ClinicalTrials Group. N Engl J Med 1995, 332:693–699.

19. Hochster H, Dieterich D, Bozzette S, et al.: Toxicity of com-bined ganciclovir and zidovudine for cytomegalovirus diseasesassociated with AIDS. An AIDS Clinical Trials Group Study.Ann Intern Med 1990, 113:111–117.

Prevention of severe neutropenia by filgrastim in HIV Kuritzkes et al. 73

20. Dührsen U, Villeval JL, Boyd J, Kannourakis G, Morstyn G,Metcalf D: Effects of recombinant human granulocyte colony-stimulating factor on hematopoietic progenitor cells in cancerpatients. Blood 1988, 72:2074–2081.

21. Souza LM, Boone TC, Gabrilove J, et al.: Recombinant humangranulocyte colony-stimulating factor: effects on normal andleukemic myeloid cells. Science 1986, 232:61–65.

22. Morstyn G, Campbell L, Souza LM, et al.: Effect of granulocytecolony stimulating factor on neutropenia induced by cytotoxicchemotherapy. Lancet 1988, i:667–672.

23. Roilides E, Walsh TJ, Pizzo PA, Rubin M: Granulocyte colony-stimulating factor enhances the phagocytic and bactericidalactivity of normal and defective human neutrophils. J Infect Dis1991, 163:579–583.

24. Garavelli PL, Berti P: Efficacy of recombinant granulocytecolony-stimulating factor in the long-term treatment of AIDS-related neutropenia [letter]. AIDS 1993, 7:589–590.

25. Hengge UR, Brockmeyer NH, Goos M: Granulocyte colony-stimulating factor treatment in AIDS patients. Clin Invest 1992,70:922–926.

26. Kimura S, Matsuda J, lkematsu S, et al.: Efficacy of recombinanthuman granulocyte colony-stimulating factor on neutropenia inpatients with AIDS. AIDS 1990, 4:1251–1255.

27. Mueller BU, Jacobsen F, Butler KM, Husson RN, Lewis LL, PizzoPA: Combination treatment with azidothymidine and granulo-cyte colony-stimulating factor in children with human immun-odeficiency virus infection. J Pediatr 1992, 121:797–802.

28. Jacobson MA, Stanley HD, Heard SE: Ganciclovir with recombi-nant methionyl human granulocyte colony-stimulating factorfor treatment of cytomegalovirus disease in AIDS patients [let-ter]. AIDS 1992, 6:515–517.

29. Miles SA, Mitsuyasu RT, Moreno J, et al.: Combined therapywith recombinant granulocyte colony-stimulating factor anderythropoietin decreases hematologic toxicity from zidovudine.Blood 1991, 77:2109–2117.

30. Fleming T, Harrington D: Counting Processes and SurvivalAnalysis. New York: John Wiley; 1991:105.

31. Hermans P, Rosenbaum W, Jou A, et al.: Filgrastim to treat neu-tropenia and support myelosuppressive medication dosing inHIV infection. AIDS 1996, 10:1627–1633.

32. Pitrak DL, Bak PM, De Marais P, Novak RM, Andersen BR:Depressed neutrophil superoxide production in human immun-odeficiency virus infection. J Infect Dis 1993, 167:1406–1410.

33. Hardy DW, Feinber J, Finkelstein D, et al.: A controlled trial oftrimethoprim–sulfmethoxazole or aerosolized pentamidine forsecondary prophylaxis of Pneumocystis carinii pneumonia inpatients with the acquired immunodeficiency syndrome. N EnglJ Med 1992, 327:1842–1848.

Appendix

Members of the G-CSF 930101 Study GroupRonald Mitsuyasu, MD, and Susie McCarthy, RN(UCLA Care Center, Los Angeles, California); HaroldKessler, MD, and Michelle M. Agnoli, RN, BSN(Infectious Disease Outpatient Clinic, RushPresbyterian–St Luke’s Medical Center, Chicago,Illinois); Gary Simon, MD, PhD, and Suzanne Schuck,RN, BSN (George Washington University MedicalCenter, Washington, DC); Kathleen E. Squires, MD,and Donna David, RN (University of Alabama atBirmingham, Birmingham, Alabama); Steve Johnson,MD, and M. Graham Ray, MSN (University ofColorado Health Sciences Center, Denver, Colorado);David T. Scadden, MD (at time of study: NewEngland Deaconess Hospital, Boston, Massachusetts);Jack Fuhrer, MD, and Christine Wallace, MS, RN

(University Hospital, State University of New York atStony Brook, Stony Brook, New York); MichaelLederman, MD, and John T. Carey, MD (UniversityHospital of Cleveland, Cleveland, Ohio); Mary Payne,MD (University of California San Francisco, SanFransisco, California); Rebecca Coleman, PharmD (SanFrancisco General Hospital, San Francisco, California);Diane Antoniskis, MD, and Mark Rarick, MD (KaiserPermanente Immune Deficiency Clinic, Portland,Oregon); Michael Lobell, MD (Maricopa MedicalCenter, Oncology Clinic, Phoenix, Arizona); JohnPost, MD (McDowell Health Care Clinic, ID, HIV,and AIDS Clinic, Phoenix, Arizona); Mary Fanning,MD, and Corina Ouan, MD (Wellesley CentralHospital, Toronto, Canada); Julie Phillips, BSN, andLinda Moran, RN (Sunnybrook Health SciencesCentre, Toronto, Canada); Peter Phillips, MD (StPaul’s Hospital, Vancouver, British Columbia);Raymond Beaulieu, MD, FRCPC, and Emil Toma,MD (Hotel-Dieu de Montreal, Montreal, Canada);Joseph Jemsek, MD, and Jan Caldwell, LPN (NalleClinic, Charlotte, North Carolina); Anthony Lamarca,MD, and Denise Lamarca, MBA (Therafirst MedicalCenter, Ft Lauderdale, Florida); Donna E. Sweet, MD,FACP, and Dairone E. Harrison, RN (University ofKansas Medical School Wichita, Wichita, Kansas);Patricia K. Sharkey, MD, and Michelle Britton(Immuno-Suppression Clinic, San Antonio, Texas);Newton Hyslop Jr, MD, and David Mushatt,MD/MPA (Medical Center of Louisiana at NewOrleans, General Clinic Research Center, NewOrleans, Lousiana); Kent Sepkowitz, MD (The NewYork Hospital, Cornell University, New York, NewYork); Tricia Sarracco, RN (Cornell Clinic TrialsUnit, New York, New York); Ellen S. Yetzer, DO,and Donna S. Morales, RN (Pacific Oaks Research,Beverly Hills, California); Donald Craven, MD, andLisa Hirschhorn, MD, MPH (Boston City Hospital, IDResearch Clinic, Boston, Massachusetts); Perry Pate,MD, and Steven G. Davis, MD (ID Associates, Dallas,Texas); C. Kenneth McAllister, MD, and LTC DavidDooley, MD (Brooke Army Medical Center, Ft SamHouston, Texas); Peter Hawley, MD, and ChristianeJones, RN (Whitman-Walker Clinic, Inc.,Washington, DC); Gary Blick, MD, and Una Hopkins,RN, MSN (Blick Medical Associates, Greenwich,Connecticut); Douglas Dieterich, MD, and KathleenFarrel, RN (NYU Medical Center, New York, NewYork); Samuel Golden, MD, and Roger Anderson,MD (Braddock Medical Center, Pittsburgh,Pennsylvania); P. Samuel Pegram, MD (Bowman GraySchool of Medicine, Winston-Salem, North Carolina);John K. Gartling, PA (Department of InternalMedicine/Section of Infectious Diseases, Winston-Salem, North Carolina); Shannon Schrader, MD, andPatrick McNamara, MD (Houston Clinical ResearchNetwork, Division of Montrose Clinic, Houston,

AIDS 1998, Vol 12 No 174

Texas); Douglas J. Ward, MD (Dupont CirclePhysicians Group, Washington, DC); Ian Mackie, MD,FRCPC (St Joseph’s Health Centre, London, Ontario,Canada); Janet Gilmour, MD, FRCPC (HIV Care

Programme, London, Ontario, Canada); Mallory Witt,MD, and Crystal Lane, RN (Harbor-UCLA MedicalCenter, Torrance, California); Pamela C. Ventra, MD,MA (Novum, Inc., Pittsburgh, Pennsylvania).