SEARCH

SEARCH BY CITATION

Keywords:

  • lymphoma;
  • meta-analysis;
  • chemotherapy;
  • G-CSF;
  • GM-CSF

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

Summary. The granulopoiesis-stimulating factors, granulocyte colony-stimulating factor (G-CSF) and granulocyte–macrophage colony-stimulating factor (GM-CSF), are used to prevent neutropenia and febrile neutropenia in patients with malignant lymphoma. The question as to whether G-CSF/GM-CSF improves dose-intensity, tumour response and overall survival (OS) has not yet been answered. As the results from single studies are inconclusive, a systematic review was performed to determine the effectiveness of G-CSF/GM-CSF. Randomized controlled trials comparing prophylaxis with G-CSF/GM-CSF versus no prophylaxis in adult patients with malignant lymphoma undergoing conventional chemotherapy were included. Medical databases (Cochrane Library, Medline, Embase) and conference proceedings were searched. All authors were contacted to obtain missing data. We included 11 studies making a total of 1434 patients. Compared with no prophylaxis, G-CSF/GM-CSF significantly reduced the relative risk (RR) for neutropenia {RR 0·64 [95% confidence interval (CI) 0·55–0·75]} febrile neutropenia (RR 0·74 [95% CI 0·62–0·89]) and infection (RR 0·74 [95% CI 0·64–0·85]). G-CSF/GM-CSF did not decrease infection-related mortality (RR 2·07 [95% CI 0·81–5·34]), improve complete remission (CR) (RR 1·06 [95% CI 0·96–1·16]) or OS (HR 0·98 [95% CI 0·81–1·18]). In conclusion, G-CSF/GM-CSF given during conventional chemotherapy in malignant lymphoma patients reduced the RR of neutropenia, febrile neutropenia and infection. However, there is no evidence that G-CSF/GM-CSF improved CR and OS in this clinical setting.

Neutropenia, febrile neutropenia and infections are dose-limiting events that occur during the chemotherapy of malignant diseases. The risk of febrile neutropenia and subsequent infection is directly related to the degree and duration of neutropenia (Bodey et al, 1966). In addition to more frequent days in hospital and a higher mortality, febrile neutropenia and infections often require the dose-reduction of cytostatic drugs or longer intervals between treatment courses. Haematopoietic growth factors, such as granulocyte colony-stimulating factor (G-CSF) and granulocyte–macrophage colony-stimulating factor (GM-CSF) have been shown to stimulate haematopoietic progenitor cells, thereby increasing the number of functional neutrophils (Lopez et al, 1986; Bronchud et al, 1988). Subsequently, these drugs were introduced into clinical trials aimed at the prevention and treatment of neutropenia, neutropenic fever and infections. G-CSF and GM-CSF are different cytokines; however, randomized controlled trials indicate that their effects are very similar in this clinical setting (Beveridge et al, 1998).

Although major endpoints in most studies were related to the prevention of neutropenia, there were hopes that haematopoietic growth factors might also contribute to better tumour control. This rationale was based on the Goldie–Coldman hypothesis suggesting that the application of the intended dose, at the appropriate time, may improve tumour response and overall survival (Goldie & Coldman, 1983; Hryniuk & Bush, 1984). Malignant lymphomas were of particular interest, as they had been shown to be chemosensitive both in experimental models and in retrospective clinical analyses (Skipper, 1990; Armitage, 1993; Lepage et al, 1993).

The rapidly increasing clinical use of G-CSF/GM-CSF intended to prevent neutropenia is associated with substantial costs. Thus, guidelines were developed in order to advise clinicians on the use of haematopoietic growth factors outside clinical trials (Rusthoven et al, 1998; Ozer et al, 2000). These guidelines were based on single clinical trials available at that time and smaller meta-analysis, including data from both solid tumours and haematological malignancies. Both The American Society of Clinical Oncology (ASCO) guidelines and other reviews were either based on limited information or included only a proportion of the existing lymphoma literature (Sweetenham et al, 2000; Lyman et al, 2002).

However, results from prospectively randomized studies using G-CSF/GM-CSF were inconclusive, particularly as far as tumour control and overall survival (OS) were concerned. Therefore, our group has conducted a comprehensive formalized meta-analysis. The complete review has been published in the Cochrane Library (Bohlius et al, 2002). This review includes 11 prospectively randomized studies, with a total of 1434 patients. All patients had either Hodgkin's disease (HD) or non-Hodgkin's lymphoma (NHL) and received standard chemotherapy with or without G-CSF/GM-CSF prophylaxis.

Inclusion criteria.  Only randomized controlled trials with patients aged over 16 years with biopsy-proven HD or NHL were included in this analysis. G-CSF/GM-CSF was given at a minimal dose of 1 µg/kg/d intravenously or subcutaneously to a standard, non-myeloablative chemotherapy prior to the onset of neutropenia. The control group received an identical chemotherapy with the same supportive care, e.g. antibiotic prophylaxis, and a placebo or no treatment. We included both published and unpublished data. Ongoing studies were excluded. Cross-over studies, quasi-randomized studies and studies with less than 10 patients per study arm were disqualified. In addition, trials including HIV-associated lymphoma, multiple myeloma or chronic lymphatic leukaemia cases were excluded. Trials investigating secondary prevention and therapeutic use in established neutropenia and febrile neutropenia were also not subject of the present analysis.

Identification of relevant trials.  Trials were identified by searches of the Cochrane Controlled Trials Register (CCTR), Medline, Embase, Lilacs, CancerLit and smaller databases. Additionally, we searched internet databases of ongoing trials and unpublished literature. We hand-searched the conference proceedings of the American Society of Hematology (ASH) and the ASCO. Citations of all trials identified in the search were checked for additional references. We also contacted experts in the field and pharmaceutical companies for further unpublished data or ongoing trials. All searches were conducted for the period between 1980 and 2000. No language restrictions were used. The full search strategy is published in the Cochrane Library (Bohlius et al, 2002).

Study selection, quality assessment and data extraction.  Study selection, quality assessment and data extraction were carried out independently by two reviewers. Any disagreement between the reviewers was solved by discussion involving a third party. Assessment of study quality included the evaluation of randomization, concealment of allocation, similarity of study groups at baseline, blinding of patients, blinding of clinicians and outcome assessors, reporting of drop-outs and withdrawals and the intention-to-treat analysis.

The primary outcome data were OS and freedom from treatment failure (FFTF). We further extracted information on the following secondary endpoints: rate and duration of neutropenia and febrile neutropenia, infection rate, tumour response, received dose intensity, mortality during chemotherapy, antibiotic treatment, length of hospital stay, side-effects of study medication and quality of life. All original authors were contacted to obtain missing data on study design, characteristics of patients, interventions, primary and selected secondary outcome measures.

Data analysis and statistical methods. Metaview 4·1 (Cochrane Collaboration) and STATA software were used for statistical analyses. A fixed effect model was assumed in all meta-analyses. To estimate the influence of G-CSF/GM-CSF on OS and FFTF, hazard ratios (HR) were calculated. If individual patient data were not available, data were extracted from published results, e.g. survival curves, using methods described by Parmar et al (1998). For binary data, the relative risk (RR) and the risk difference (RD) with corresponding 95% confidence intervals (CI) was determined. The Mantel–Haenszel Method was used for pooling. The estimated overall relative risk differences were used to estimate numbers needed to treat (NNT). Potential causes of heterogeneity were explored by performing sensitivity analyses. Sensitivity and subgroup analysis included GM-CSF/G-CSF, study quality, haematotoxicity of chemotherapy, antibiotic prophylaxis, age of patients, HD or NHL, published or unreported data. For each endpoint including more than four trials, a funnel plot was generated and a linear regression test was performed to examine the presence of bias (Egger et al, 1997). A P-value of less than 0·1 was considered significant for the linear regression test. All data included were based on intention-to-treat or full analysis set. Data based on per protocol analysis were not included in this study.

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

A total of 1858 potentially relevant titles were screened for inclusion. Fifty-five studies were retrieved for more information, of which 19 clinical trials were potentially appropriate for inclusion in the meta-analysis. Of these, six studies were excluded: three trials with 962 patients that were currently ongoing (Doorduijn et al, 2000; Dr J.-Y. Blay, Elypse 2 study, personal communication; Dr D. Cunningham, personal communication, information can be found at http://www.controlled-trials.com/isrctn/trial/PMITCEBO/0/98741793.html), one study in both children and adults, and two studies in which lymphoma and solid tumours were identified, but that did not include 10 or more eligible patients per study arm (Magrath et al, 1996; Yau et al, 1996; Gregory et al, 1998). Two studies were withdrawn: one unpublished study including 100 patients was excluded as we were not able to retrieve any data and another study that was published as an interim-analysis as part of a multicentre trial but did not report any evaluable data (Liberati et al, 1991). In total, 11 randomized controlled trials with a total of 1434 patients were analysed in the present systematic review. Of these trials, three additional follow-up reports and an economic analysis of identical patient data were identified (Aviles et al, 1994; Engelhard et al, 1994; Gerhartz et al, 1994; Souetre & Qing, 1994; Gisselbrecht et al, 1997; Zinzani et al, 1999).

All first authors were contacted to obtain unreported data. In eight trials we succeeded in obtaining additional information on study design, patient characteristics and selected outcome data. Study characteristics are shown in Table I.

Table I.  Summary of size of trial, patients characteristics, interventions, follow up period, concealment of allocation and publication form.
StudyNumber of patients randomizedDiseaseAge (years)Study medicationChemotherapyMedian follow up (years) Concealment of allocationPublication form
  • *

    Unreported data, personal communication.

  • Estimated using published Kaplan–Meier survival curves.

  • NR, not reported. Chemotherapy regimes are defined below: d, days; IV, intravenous; PO, oral; IM, intramuscular.

  • MA/MA Hybrid: mechlorethamine 6 mg/m2 iv, d1, vincristine 1·4 mg/m2, max 2·0 mg/m2 iv, d1, procarbazine 100 mg/m2 po, d1–7, prednisone 40 mg/m2 po, d1–7, doxorubicin 25 mg/m2 iv, d15, bleomycin 10 mg/m2 iv, d15, vinblastine 6 mg/m2 iv, d15, dacarbazine 375 mg/m2 iv, d15, repeated every 28 d, 6 cycles.

  • ESAP: etoposide 40 mg/m2 iv, d1–4, methylprednisolone 350 mg/m2 iv, d1–5, ara-c 2 g/m2 iv, d5, cis-platin 25 mg/m2 iv, d1–4 repeated every 21 d, 3 cycles.

  • m-BECOD: bleomycin 10 mg/m2 iv, d1, epirubicin 70 mg/m2 iv, d1, cyclophosphamide 600 mg/m2 iv, d1, vincristine 1·4 mg/m2 iv, d1, dexamethasone 20 mg/m2 po, d1–5, methotrexate 120 mg/m2 iv, d14, repeated every 21 d, 3 cycles.

  • MVPP-Bleomycin: mitoxantrone 10 mg/m2 iv, d1, vincristine 1·4 mg/m2 iv, d1, prednisone 50 mg/m2 po, d1–14, procarbazine 100 mg/m2 po, d1–14, bleomycin 10 mg/m2 iv, d14, repeated every 21 d, 3 cycles.

  • ACVBP: adriamycin 75 mg/m2, d1, cyclophosphamide 1200 mg/m2, d1, vindesine 2 mg/m2, d1 and 5, bleomycin 10 mg, d1 and 5, prednisolone 60 mg/m2, d1–5, methotrexate 12 mg intrathecal 1–2/week, repeated every 21 d, 3–4 cycles.

  • VIMMM: VP 16 100 mg/m2, d1 and 5, ifosfamide 1000 mg/m2, d1–5, mitoxantrone 10 mg/m2, d1, methyl-GAG 300 mg/m2, d1 and 5, methotrexate 1500 mg/m2, d15, methyl-prednisone 60 mg/m2, d1–5, 2 courses ACVBP alternating with 2 courses VIMMM.

  • CHOP: cyclophosphamide 750 mg/m2 iv, d1, doxorubicin 50 mg/m2 iv, d1, vincristine 1·4 mg/m2 iv, d1, prednisone 100 mg/m2 po, d1–5, repeated every 21 d, 8 cycles.

  • CNOP: cyclophosphamide 750 mg/m2 iv, d1, mitoxantrone 10 mg/m2 iv, d1, vincristine 1·4 mg/m2 iv, d1, prednisone 100 mg/m2 po, d1–5, repeated every 21 d, 8 cycles.

  • ECP: etoposide 50 mg/m2 po, d1–10, cisplatin 60 mg/m2 iv, d1, prednisolone 100 mg po, d1–5, repeated every 21 d, 8 cycles.

  • MOPP: mustine 6 mg/m2 iv, d1 and 8, vincristine 1·4 mg/m2 iv, d1 and 8, procarbazine 100 mg/m2 po, d1–14, prednisolone 25 mg/m2 po, d1–14 repeated every 28 d, 2–8 cycles.

  • MOPP/EVAPP: mustine 6 mg/m2 iv, d1 and 8, vincristine 1·4 mg/m2 iv, d1, procarbazine 100 mg/m2 po, d1–7, prednisolone 25 mg/m2 po, d1–14, etoposide 75 mg/m2 iv, d8–10, adriamycin 25 mg/m2 iv, d8, vinblastine 6 mg/m2 iv, d8 repeated every 28 d, 2–8 cycles.

  • CEOP/IMVP-Dexa: cyclophosphamide 750 mg/m2 iv, d1, epirubicin 70 mg/m2 iv, d1, vincristine 1·4 mg/m2 iv, d1 and 8, prednisolone 100 mg po, d1–5, ifosfamide 2000 mg/m2 iv, d15–17, VP16 100 mg/m2 iv, d15–17, dexamethasone 40 mg/m2 po, d15–19, methotrexate 800 mg/m2 iv, d22, Ca-folinate 15 mg/m2 po, d23–25.

  • COP-BLAM: cyclophosphamide 700 mg/m2 iv, d1, doxorubicin 60 mg/m2 iv, d1, bleomycin 15 mg absolute dose iv, d15, vincristine 1 mg/m2 iv, d1 and 15, procarbazine 100 mg/m2 po, d1–7, prednisolone 50 mg/m2 po, d1–7, repeated every 21 d, 3–6 cycles.

  • ACVB: cyclophosphamide 1200 mg/m2 iv, d1, vindesine 2 mg/m2 iv, d1 and 5, bleomycin 10 mg iv, d1 and 5, prednisone 60 mg/m2 po, d1–5, methotrexate 15 mg intrathecal, d1, adriamycin 75 mg/m2 iv, d1 repeated every 14 d, 4 cycles.

  • NCVB: same as ACVB with mitoxantrone 12 mg/m2 iv, d1 instead of adriamycin, repeated every 14 d, 4 cycles.

  • VAPEC-B: adriamycin 35 mg/m2 iv, d1, 15, 29, 43, 57, 71, cyclophosphamide 350 mg/m2 iv, d1, 29, 57, vincristine 1·4 mg/m2 iv, d8, 22, 36, 50, 64, bleomycin 10 mg/m2 iv, d8, 36, 64, etoposide 100 mg/m2 po, d15–19, d43–47, d71–75, prednisolone 50 mg po, daily for 5 weeks than to be reduced.

  • VNCOP-B: cyclophosphamide 300 mg/m2 iv, d1, 15, 29, 43, mitoxantrone 10 mg/m2 iv, d1, 15, 29, 43, vincristine 2 mg iv, d8, 22, 36, 50, etoposide 150 mg/m2 iv, d8 and 36, bleomycin 10 mg/m2 iv, d22 and 50, prednisone 40 mg im, daily.

Aglietta et al (2000)56HD18–77GM-CSFMA/MA Hybrid7·2*YesFull text
Avilés et al (1994)42NHL34–63G-CSFESAP, m-BECOD, MVPP-Bleomycin7·5*YesFull text
Bastion et al (1993)119NHL55–69G-CSFA: ACVBP B: ACVBP/VIMMMNRUnclearAbstract
Björkholm et al (1999)458NHL60–86G-CSFA: CHOP B: CNOP3·8YesAbstract
D. Cunningham, personal communication39*Relapsed* HD (n = 1) NHL (n = 38)23–71*G-CSFECP6·7*YesUnpublished
Dunlop et al (1998)53HD19–41G-CSFA: MOPP B: MOPP/EVAP Hybrid7·9*YesFull text
Fridrik et al (1997)85NHL19–72G-CSFCEOP/IMVP-Dexa2·7*YesFull text
Gerhartz et al (1993)182NHL15–73GM-CSFCOP-BLAM1·8YesFull text
Gisselbrecht et al (1997)162NHL15–55G-CSFA: ACVB B: NCVB3·3YesFull text
Pettengell et al (1992)80NHL16–71G-CSFVAPEC-B1·25UnclearFull text
Zinzani et al (1997)158NHL60–82G-CSFVNCOP-B1·0–1·5YesFull text

Growth factors were given before the onset of neutropenia at doses of 5 µg/kg/d or 230 µg/m2 for G-CSF (Pettengell et al, 1992) or 400 µg/d for GM-CSF (Gerhartz et al, 1993) subcutaneously during each course of chemotherapy, starting within 48 h after application of cytotoxic drugs. In one study GM-CSF was given between and before the next chemotherapy cycle to investigate whether GM-CSF given before chemotherapy may be myeloprotective (Aglietta et al, 2000). All chemotherapy regimens applied were CHOP- (cyclophosphamide, doxorubicin, vincristine, prednisone) or MOPP (mustine, vincristine, procarbazine, prednisolone)-like and moderately myelosuppressive. Antibiotic prophylaxis was administered in two studies. All studies but one were sponsored by the pharmaceutical industry. All trials were randomized. In nine of 11 trials, adequate measures were taken to conceal allocation to the experimental or control arm. In two studies, the concealment of allocation could not be clarified. At baseline, the distribution of prognostic factors, such as age, sex, and stage of disease, was well balanced in the study groups compared. Five trials were placebo controlled. Ten studies included intention-to-treat or full analysis set calculations in their statistical analysis. In one abstract publication, the specific number of patients in each study arm was not reported but the results were reported in percentages. Therefore, we distributed the total number of patients (n = 119) at random to the four different study arms assuming equal group sizes (Bastion et al, 1993). One study was never published (Dr D. Cunningham, personal communication) but the data included in this review were kindly provided by the principal investigator for inclusion in the present analysis. Heterogeneity between studies was seen only in the analysis of neutropenia and will be discussed.

Neutropenia

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

Neutropenia, defined as an absolute neutrophil count below 0·5 × 109/l, was assessed in six trials with a total of 558 patients. Overall, G-CSF/GM-CSF reduced the risk of neutropenia by 36% (RR 0·64, 95% CI 0·55–0·75; Fig 1). Significant heterogeneity existed between trials (P = 0·0025). A stronger treatment effect was observed in trials with antibiotic prophylaxis (RR = 0·43, 95% CI 0·31–0·60, four trials) compared with trials without antibiotic prophylaxis (RR = 0·78, 95% CI 0·65–0·93, two trials). Older patients had a larger benefit from using G-CSF/GM-CSF (RR = 0·42 [95% CI 0·27–0·66], one trial) than younger patients (RR = 0·71 [95% CI 0·60–0·84] five trials). The estimated treatment effect of published data was larger (RR = 0·55 [95% CI 0·45–0·68] three trials) than for unpublished data (RR = 0·87 [95% CI 0·67–1·13] three trials). The study of Björkholm et al (1999) was not included as the data were based on patients who were compliant to the G-CSF application only. However, inclusion of these data did not influence the overall results (data not shown). Expressed as NNT, four patients would have to be treated with G-CSF to prevent one patient from neutropenia (NNT = 4·35 [95% CI 3·23–6·67]).

image

Figure 1. Meta-analysis of the RR for neutropenia, febrile neutropenia and infection for lymphoma patients undergoing standard chemotherapy treated prophylactically with G-CSF/GM-CSF in comparison to placebo/no treatment. Solid squares indicate risk estimates for the single studies. The size of the squares is proportional to the sample size and the number of events. Horizontal lines denote 95% confidence intervals. The diamond shows the confidence interval for the pooled relative risks. Negative values indicate a RR reduction for neutropenia, febrile neutropenia and infection rates favouring G-CSF/GM-CSF. *unreported data, personal communication. Dunlop et al (1998)° refers to the MOPP/EVAP study arm, Dunlop et al (1998)°° refers to the EVAP study arm, Bastion et al (1993)° refers to the ACVBP study arm, Bastion et al (1993)°° refers to the VIMMM study arm. §1 test for overall effect z = − 5·43, P < 0·0001, test for heterogeneity chi-square = 12·84, d.f. = 5, P = 0·025. §2, test for overall effect z = − 3·19, P = 0·001, test for heterogeneity chi-square = 4·31, d.f. = 4, P = 0·37. §3, test for overall effect z = − 4·28, P = 0·0001, test for heterogeneity chi-square = 12·02, d.f. = 10, P = 0·28. n/N, proportion within total study population.

Download figure to PowerPoint

Febrile neutropenia

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

The RR of febrile neutropenia, defined as an absolute neutrophil count below 1·0 × 109/l and febrile temperatures, was reduced by 26% (RR 0·74 [95% CI 0·62–0·89]; Fig 1). Four studies with 360 patients were included in this analysis. The other studies did not report data on febrile neutropenia or used a different, non-comparable definition of febrile neutropenia. All included studies evaluated G-CSF and data for GM-CSF were not available. Expressed as NNT, six to seven patients would have to be treated with G-CSF to prevent one patient from febrile neutropenia (NNT = 6·25 [95% CI 3·85–16·67]).

Infection

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

Nine studies with 1292 patients reporting microbiologically or clinically documented infection were included in the analysis of infection rates. If both microbiological and clinical documented infections were reported, only the microbiologically documented infections were included. The RR to develop an infection was significantly reduced by 26% (RR 0·74 [95% CI 0·64–0·85]; Fig 1). Expressed as NNT, nine patients wound have to be treated to safe one patient from infectious complications (NNT = 9·1 [95% CI 6·25–16·67]).

Antibiotic treatment

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

Data of four studies including 359 patients were pooled in this analysis. The pooled RR of parenteral antibiotic treatment was reduced by 18% in the G-CSF-/GM-CSF-treated group (RR 0·82 [95% CI 0·57–1·18]) but this difference was not statistically significant. There was no evidence to suggest that G-CSF/GM-CSF reduced the number of patients requiring parenteral antibiotic treatment.

Mortality during chemotherapy

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

No evidence was found to suggest that G-CSF/GM-CSF reduced overall mortality and infection-related mortality during chemotherapy. Overall, 25 of 395 patients treated with G-CSF/GM-CSF and 20 of 386 patients in the control group died during chemotherapy (RR 1·21 [95% CI 0·70–2·10]). Based on the outcome data of 662 patients there was also no significant difference in infection-related mortality (RR 2·07 [95% CI 0·81–5·34]). Twelve of 336 patients in the G-CSF/GM-CSF group and 5 of 326 patients in the control group died due to infection. This effect was largely due to one study in which six patients died during treatment with CEOP/IVMP (cyclophosphamide, epirubicin, vincristine, prednisolone, ifosfamide, VP16, dexamethasone, methotrexate, Ca-folinate) and G-CSF, whereas only one patient in the control group died due to infection (Fridrik et al, 1997).

Complete response

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

In terms of tumour control we found no evidence that G-CSF/GM-CSF improved complete response. The overall RR to achieve complete response for patients treated with G-CSF/GM-CSF was increased by 6%, but this difference was not statistically significant (RR 1·06 [95% CI 0·96–1·16], Fig 2). This analysis was based on nine trials including 1195 patients. Inclusion of 303 patients reported in an interim analysis of an ongoing randomized trial in elderly patients (n = 408) did not influence the overall result (data not shown) (Doorduijn et al, 2000).

image

Figure 2. Meta-analysis of the RR for complete response for lymphoma patients undergoing standard chemotherapy treated prophylactically with G-CSF/GM-CSF in comparison to placebo/no treatment. Solid squares indicate relative risks of single studies. The size of the squares is proportional to the sample size and the number of events. Horizontal lines denote 95% confidence intervals. The diamond shows the confidence interval for the pooled relative risks. Positive values indicate a relatively increased likelihood for complete response favouring G-/GM-CSF. Björkholm et al (1999)° refers to the CHOP study arm, Björkholm et al (1999)°° refers to the CNOP study arm, Dunlop et al (1998)° refers to the MOPP/EVAP study arm, Dunlop et al (1998)°° refers to the EVAP study arm. §, test for overall effect z = 1·12, P = 0·3, test for heterogeneity chi-square = 7·65, d.f. = 10, P = 0·66. n/N, proportion within total study population.

Download figure to PowerPoint

Overall survival

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

When compared with no treatment or placebo, G-CSF/GM-CSF did not have an effect on OS (Fig 3). The pooled HR was 0·97 (95% CI 0·81–1·17). Eight studies including 1048 randomized patients were analysed. Four investigators provided us with individual patient data, in three reports the data were extracted from survival curves, while one author reported HRs. On average, the observation time of the studies that were included was four years (range 1·3–7·9).

image

Figure 3. Meta-analysis of the HR for OS for lymphoma patients undergoing standard chemotherapy treated prophylactically with G-CSF/GM-CSF in comparison to placebo/no treatment. Solid squares indicate hazard ratios of each study. The size of the squares is proportional to the sample size and the number of events (raw data not shown). Horizontal lines denote 95% confidence intervals. The diamond shows the confidence interval for the pooled hazard ratios. Negative values indicate a reduced hazard ratio for overall survival favouring G-CSF/GM-CSF. Dunlop et al (1998)° refers to the MOPP/EVAP study arm, Dunlop et al (1998)°° refers to the EVAP study arm. *based on individual patient data provided by the investigator. §, test for overall effect z = − 0·23, P = 0·8, test for heterogeneity chi-square = 4·10, d.f. = 8, P = 0·85. n/N, proportion within total study population. O-E, observed – expected; var, variance.

Download figure to PowerPoint

Freedom from treatment failure

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

FFTF was defined as freedom from progression, relapse of disease, or death of any cause. Compared to placebo or no treatment, G-CSF/GM-CSF did not have a significant effect on FFTF (HR 1·22 [95% CI 0·83–1·80]). Four studies with 329 patients were included. In all but one study (Gerhartz et al, 1993), the investigators provided individual patient data.

Adverse events

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

Bone pain, thrombo-embolic complications and skin reactions were reported as adverse effects attributable to G-CSF and GM-CSF. Based on seven studies including data of 749 patients, the RR to suffer from bone pain for G-CSF-/GM-CSF-treated patients was increased by 223% (RR 3·23 [95% CI 1·77–5·89]) compared with controls. Data are shown in Table II. There was no evidence that G-CSF/GM-CSF increased the risk of thrombosis or related complications such as transient ischaemic attacks, stroke or myocardial infarction (RR 1·29 [95% CI 0·56–3·01]). Based on two studies with 337 patients, the RR for injection site reactions was increased by 555% (RR 6·55 [95% CI 3·01–14·25]). Due to the small number of patients included, the CI is wide and cautious interpretation of this result is necessary. Chemotherapy-related adverse events such as nausea, vomiting, peripheral polyneuropathy and alopecia were reported to be similar in the compared groups.

Table II.  Summary of adverse events most often reported: bone pain, thrombo-embolic complications, injection site reaction.
OutcomeNumber of trialsNumber of patientsEvent ratesPooled RR [95% CI]
G-CSF/GM-CSFControl group
  • *

    Test for overall effect z = 3·83, P = 0·0001, test for heterogeneity chi-square = 5·03, d.f. = 6, P = 0·54.

  • Test for overall effect z = 0·60, P = 0·5, test for heterogeneity chi-square = 2·12, d.f. = 4, P = 0·71.

  • Test for overall effect z = 4·74, P < 0·0001, test for heterogeneity chi-square = 0·00, d.f. = 1, P = 0·97.

Bone pain774941/38011/3693·23 [1·77–5·89]*
Thrombo-embolic complications542510/216 7/2091·29 [0·56–3·01]
Injection site reaction233743/170 6/1676·55 [3·01–14·25]‡

Continuous data

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

Due to insufficiently reported continuous outcome data such as duration of neutropenia and febrile neutropenia, received dose intensity, duration of antibiotic therapy and length of hospital stay, these data could not be meta-analysed. Two trials reported a shorter duration of severe neutropenia (absolute neutrophil count < 0·5 × 109/l) (Gerhartz et al, 1993; Aglietta et al, 2000), and two studies reported a similar length of febrile neutropenia in the compared groups (Fridrik et al, 1997; Dunlop et al, 1998). The duration of parenteral antibiotic treatment was shorter, hence without statistical significance, in two studies (Souetre & Qing, 1994; Aglietta et al, 2000). None of the studies reported the overall number of days in hospital. Dose-intensity was reported in 9 of 11 studies. In all but one study, the G-CSF-/GM-CSF-treated group received a higher dose intensity than the control group, although the observed differences were only statistically significant in three studies. None of the studies assessed quality of life.

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

The results of this meta-analysis are as follows: (i) G-CSF/GM-CSF reduced the risk of neutropenia, febrile neutropenia and infection in patients undergoing conventional chemotherapy for malignant lymphoma; (ii) there is no evidence that G-CSF/GM-CSF improves tumour control or OS in this setting.

The present study is the first formalized comprehensive meta-analysis evaluating the effects of G-CSF/GM-CSF in patients with malignant lymphoma undergoing conventional chemotherapy. Previous analysis included smaller patient numbers, and heterogeneous populations with solid tumours and haematological malignancies (Ozer et al, 2000; Lyman et al, 2002). In contrast, the present meta-analysis is based on a homogeneous group with far greater patient numbers. An extensive literature search and comprehensive data analysis of existing biomedical databases, ongoing trials, grey literature, conference proceedings and hand-searched journals of potential interest were performed. Overall, 11 prospective, randomized studies including 1434 patients were included. This is more than twice the size of similar reports published by any other group to date (Sweetenham et al, 2000). In addition, we included previously unreported data on outcome, patient characteristics and study design contributed by the authors of the original publications and one previously unpublished study.

The robustness of the results was tested in prespecified sensitivity and subgroup analysis. We found substantial differences for the outcome of neutropenia. The observed heterogeneity between studies (P = 0·025) may be explained by differences in either prophylactic administration of antibiotic treatment during chemotherapy (P = 0·0019), the age of patients (P = 0·022) or publication status (P = 0·023). The interpretation of these findings is complicated as these analyses are interrelated; for example, one study was always included in the group that showed the larger treatment benefit (Zinzani et al, 1997). An additional sensitivity analysis did not reveal any significant difference between G-CSF and GM-CSF, except for bone pain and discontinuation of study medication. Based on these findings and results of randomized trials, comparing G-CSF and GM-CSF directly, we conclude that G-CSF and GM-CSF are clinically comparable. Exclusion of GM-CSF did not significantly influence the results (data not shown).

The most convincing effects of G-CSF/GM-CSF in this study were shown for neutropenia. G-CSF/GM-CSF reduced the risk of lymphoma patients reaching neutrophil counts below 0·5 × 109/l by 36% (RR 0·64 [95% CI 0·55–0·75]). G-CSF reduced the risk of febrile neutropenia by 26% (RR 0·74 [95% CI 0·62–0·89]). In addition, the risk to acquire infection in the G-CSF/GM-CSF group was reduced by 26% (RR 0·74 [95% CI 0·64–0·85]). However, there is no evidence to suggest that G-CSF/GM-CSF decreased overall mortality and infection related mortality after standard-dose chemotherapy.

Data presented in this analysis suggest smaller effects than previously reported, including those published in the ASCO guidelines (Ozer et al, 2000). Differences were most obvious for a comparison of effect-estimates of the RR reduction of infection with a previous smaller study (Sweetenham et al, 2000). Their data of 637 lymphoma patients identified an overall RR reduction of 43% (RR 0·57 [95% CI 0·42–0·77]), whereas our analysis on 1292 cases yielded an overall effect of only 26% relative risk reduction (RR 0·74 [95% CI 0·64–0·85]). A similar observation accounted for the RR reduction of febrile neutropenia. Previous analysis based on a heterogeneous population of patients with solid and haematological malignancies reported in the ASCO guidelines had estimated the RR for febrile neutropenia to be reduced by approximately 50% (Ozer et al, 2000). We observed a RR reduction of only 26% (RR 0·74 [95% CI 0·62–0·89]). The most obvious explanations for these differences relate to a more homogeneous study population in our analysis, a more comprehensive search strategy and a refined study selection.

In addition to reducing infections and related complications, G-CSF/GM-CSF are used in clinical practice to maintain dose intensities. Retrospective analysis indicated that a higher relative dose-intensity may indeed translate into better tumour control (Lepage et al, 1993). A significantly higher received dose intensity in patients receiving G-CSF/GM-CSF was demonstrated in patients with small cell lung cancer, breast cancer and malignant lymphoma (Pettengell et al, 1993; Woll et al, 1995; de Graaf et al, 1996; Fridrik et al, 1997). There was no evidence, however, that the addition of G-CSF/GM-CSF to standard chemotherapy in lymphoma patients improved tumour response (n = 1195, nine trials), FFTF (n = 329, four trials) or OS (n = 1025, eight trials) in the present analysis. Assuming an underlying likelihood of 56% for CR in the pooled control groups and the observed risk difference of 3% (RD 0·03 [95% CI − 0·02–0·09]), approximately 8500 patients would be needed to demonstrate a statistically significant result in a prospective, randomized trial (based on two-sided test, power 80%).

Publication bias due to under-reporting of unexpected or negative data is one of the major obstacles to most meta-analysis. A comprehensive literature search and detection of unreported data can minimize this bias. Figure 4 shows a Funnel plot analysis of the data reported for complete response. The imbalance of positive and negative results indicates that studies with negative findings might be under-represented in our study. We identified two studies on GM-CSF that were never published and were not included in this review as data were either not available or not evaluable (Liberati et al, 1991). Taking this into consideration, the true benefit of G-CSF/GM-CSF to improve complete response may be even less than indicated by this analysis. The most likely explanation for the very similar CR rates in patients receiving or not receiving G-CSF/GM-CSF is that the major endpoint was neutropenia and neutropenia-related infections in the studies included. Studies using higher doses of chemotherapy or time intensification were explicitly excluded from the present analysis.

image

Figure 4. Funnel plot of the complete response meta-analysis. In this graph the standard error (SE) of each trial included in the meta-analysis of complete response is plotted against the effect size. Dots on the right side of the vertical bar indicate studies favouring the use of G-CSF/GM-CSF. This plot shows an imbalance of positive and negative findings, suggesting that small studies with negative findings might be missing.

Download figure to PowerPoint

Adverse events attributable to G-CSF/GM-CSF such as bone pain and skin reactions were more frequently reported in patients treated with G-CSF/GM-CSF compared with the control group and this difference was statistically significant. Thrombosis and related complications occurred more frequently in the G-CSF-/GM-CSF-treated groups than in the control groups, although this difference was not statistically significant (RR 1·29 [95% CI 0·56–3·01]; Odds Ratio 1·31 [95% CI 0·54–3·19]). These findings are consistent with a previous meta-analysis focusing on thrombosis in patients with various malignancies treated with haematopoietic growth factors, which reported a statistically non-significant trend towards increased thrombo-embolic complications for patients treated with GM-CSF (Odds Ratio 1·67 [95% CI 0·92–3·04, n = 838]) (Barbui et al, 1996). In total, a greater number of patients given GM-CSF (24/117) were reported for having discontinued the study due to adverse effects than patients receiving G-CSF. Based on these data, G-CSF seems to be superior in terms of tolerability than GM-CSF. However, it should be taken into consideration that this result is based on an indirect comparison and might be biased due to unreported data in the G-CSF studies. One randomized study which directly compared G-CSF and GM-CSF in 181 neutropenic, afebrile cancer patients undergoing standard chemotherapy does not support this finding (Beveridge et al, 1998).

In addition to the prevention of neutropenia and related effects, growth factors are currently used in high-dose chemotherapy settings to help to generate haematopoietic stem-cells and to support patient recovery after myeloablative treatment (Pettengell et al, 1993). More recently, another role for growth factors is the assistance of haematopoietic recovery in time- or dose-intensified regimens. Currently there are strong arguments for this notion stemming from two large prospectively randomized trials in patients with Hodgkin's disease or non-Hodgkin's lymphoma (Diehl et al, 2000; Pfreundschuh, 2001). Thus, meta-analysis to prove the assumed role of haematopoietic growth-factors in dose escalation are warranted.

In conclusion, we demonstrated that G-CSF/GM-CSF significantly reduce the relative risk for neutropenia, febrile neutropenia and infections, but did not show an effect on tumour control, overall survival and freedom from treatment failure in patients undergoing conventional chemotherapy for malignant lymphoma.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References

We thank the authors of the original studies for their contribution, the editors of the Cochrane Haematological Malignancies Group (CHMG), the members of the CHMG editorial base, the Department of Biometrics at the University of Cologne (Cologne, Germany), Dr Hasenclever (University of Leipzig, Germany) for scientific advice and critical review of study proposal and final draft.

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Neutropenia
  6. Febrile neutropenia
  7. Infection
  8. Antibiotic treatment
  9. Mortality during chemotherapy
  10. Complete response
  11. Overall survival
  12. Freedom from treatment failure
  13. Adverse events
  14. Continuous data
  15. Discussion
  16. Acknowledgments
  17. References
  • Aglietta, M., Montemurro, F., Fagioli, F., Volta, C., Botto, B., Cantonetti, M., Racanelli, V., Teofili, L., Ferrara, R., Amadori, S., Castoldi, G.L., Dammacco, F. & Levis, A. (2000) Short term treatment with Escherichia coli recombinant human granulocyte-macrophage-colony stimulating factor prior to chemotherapy for Hodgkin disease. Cancer, 88, 454460.
  • Armitage, J.O. (1993) Treatment of non-Hodgkin's lymphoma. New England Journal of Medicine, 328, 10231030.
  • Aviles, A., Diaz-Maqueo, J.C., Talavera, A., Nambo, M.J. & Garcia, E.L. (1994) Effect of granulocyte colony-stimulating factor in patients with diffuse large cell lymphoma treated with intensive chemotherapy. Leukemia and Lymphoma, 15, 153157.
  • Barbui, T., Finazzi, G., Grassi, A. & Marchioli, R. (1996) Thrombosis in cancer patients treated with hematopoietic growth factors - a meta-analysis. On behalf of the Subcommittee on Haemostasis and Malignancy of the Scientific and Standardization Committee of the ISTH. Thrombosis and Haemostasis, 75, 368371.
  • Bastion, Y., Bosly, A., Gisselbrecht, C., Reyes, F., Tilly, H. & Herbrecht, R. (1993) A randomized double-blind phase III study of Filgrastim (recombinant human G-CSF) vs placebo during intensive induction chemotherapy in 55–69-year old patients (pts) with poor prognosis aggressive Non-Hodgkin's Lymphoma. Blood, 82, 143a.
  • Beveridge, R.A., Miller, J.A., Kales, A.N., Binder, R.A., Robert, N.J., Harvey, J.H., Windsor, K., Gore, I., Cantrell, J., Thompson, K.A., Taylor, W.R., Barnes, H.M., Schiff, S.A., Shields, J.A., Cambareri, R.J., Butler, T.P., Meister, R.J., Feigert, J.M., Norgard, M.J., Moraes, M.A., Helvie, W.W., Patton, G.A., Mundy, L.J., Henry, D. & Sheridan, M.J. (1998) A comparison of efficacy of sargramostim (yeast-derived RhuGM-CSF) and filgrastim (bacteria-derived RhuG-CSF) in the therapeutic setting of chemotherapy-induced myelosuppression. Cancer Investigations, 16, 366373.
  • Björkholm, M., Osby, E., Hagberg, H., Kvaloy, S., Teerenhovi, J. & Myhre, J. (1999) Randomized trial of r-met-Hu granulocyte colony-stimulating factor (G-CSF) as adjunct to CHOP or CNOP treatment of elderly patients with aggressive Non-Hodgkin's Lymphoma. Blood, 94, 599a.
  • Bodey, G.P., Buckley, M., Sathe, Y.S. & Freireich, E.J. (1966) Quantitative relationships between circulating leukocytes and infection in patients with acute leukemia. Annals of Internal Medicine, 64, 328340.
  • Bohlius, J., Reiser, M., Schwarzer, G. & Engert, A. (2002) Granulopoiesis-stimulating factors in the prevention of adverse effects in the chemotherapeutic treatment of malignant lymphoma. In: The Cochrane Library, Issue 4, 2002.
  • Bronchud, M.H., Potter, M.R., Morgenstern, G., Blasco, M.J., Scarffe, J.H., Thatcher, N., Crowther, D., Souza, L.M., Alton, N.K. & Testa, N.G. (1988) In vitro and in vivo analysis of the effects of recombinant human granulocyte colony-stimulating factor in patients. British Journal of Cancer, 58, 6469.
  • Diehl, V., Franklin, J., Sieber, M., Hasenclever, D., Wolf, J. & Engert, A. (2000) Dose escalated BEACOPP chemotherapy improves failure free survival in advanced Hodgkin's Disease: updated results of the German Hodgkin's Lymphoma Study Group. Blood, 96, 576a.
  • Doorduijn, J.K., Van Der Holt, B., Van Der Hem, K.G., Van Imhoff, G.W., Kramer, M.H.H. & Van Oers, M.H.J. (2000) Randomized trial of granulocyte-colony stimulating factor (G-CSF) added to CHOP in elderly patients with aggressive non-Hodgkin's lymphoma (NHL). Blood, 96, 133a.
  • Dunlop, D.J., Eatock, M.M., Paul, J., Anderson, S., Reed, N.S., Soukop, M., Lucie, N., Fitzsimmons, E.J., Tansey, P. & Steward, W.P. (1998) Randomized multicentre trial of filgrastim as an adjunct to combination chemotherapy for Hodgkin's disease. West of Scotland Lymphoma Group. Clinical Oncology (Royal College of Radiology), 10, 107114.
  • Egger, M., Davey, S.G., Schneider, M. & Minder, C. (1997) Bias in meta-analysis detected by a simple, graphical test. British Medical Journal, 315, 629634.
  • Engelhard, M., Gerhartz, H., Brittinger, G., Engert, A., Fuchs, R., Geiseler, B., Gerhartz, D., Haunauske, A.R., Hartlapp, H.J. & Huhn, D. (1994) Cytokine efficiency in the treatment of high-grade malignant non-Hodgkin's lymphomas: results of a randomized double-blind placebo-controlled study with intensified COP-BLAM +/- rhGM-CSF. Annals of Oncology, 5, 123125.
  • Fridrik, M.A., Greil, R., Hausmaninger, H., Krieger, O., Oppitz, P., Stoger, M., Klocker, J., Neubauer, M., Helm, W., Pont, J., Fazeny, B., Hudec, M., Simonitsch, I. & Radaszkiewicz, T. (1997) Randomized open label phase III trial of CEOP/IMVP-Dexa alternating chemotherapy and filgrastim versus CEOP/IMVP-Dexa alternating chemotherapy for aggressive non-Hodgkin's lymphoma (NHL). A multicenter trial by the Austrian Working Group for Medical Tumor Therapy. Annals of Hematology, 75, 135140.
  • Gerhartz, H.H., Engelhard, M., Meusers, P., Brittinger, G., Wilmanns, W., Schlimok, G., Mueller, P., Huhn, D., Musch, R. & Siegert, W. (1993) Randomized, double-blind, placebo-controlled, phase III study of recombinant human granulocyte-macrophage colony-stimulating factor as adjunct to induction treatment of high-grade malignant non-Hodgkin's lymphomas. Blood, 82, 23292339.
  • Gerhartz, H.H., Engelhard, M., Brittinger, G., Schlimok, G., Thiel, E., Huber, C., Emmerich, B., Schadek, C., Brandmaier, R. & Stern, A.C. (1994) Recombinant human granulocyte-macrophage colony-stimulating factor as adjunct to chemotherapy in aggressive non-Hodgkin's lymphomas. Seminars in Oncology, 21, 2528.
  • Gisselbrecht, C., Haioun, C., Lepage, E., Bastion, Y., Tilly, H., Bosly, A., Dupriez, B., Marit, G., Herbrecht, R., Deconinck, E., Marolleau, J.P., Yver, A., Dabouz-Harrouche, F., Coiffier, B. & Reyes, F. (1997) Placebo-controlled phase III study of lenograstim (glycosylated recombinant human granulocyte colony-stimulating factor) in aggressive non-Hodgkin's lymphoma: factors influencing chemotherapy administration. Groupe d'Etude des Lymphomes de l'Adulte. Leukemia and Lymphoma, 25, 289300.
  • Goldie, J.H. & Coldman, A.J. (1983) Quantitative model for multiple levels of drug resistance in clinical tumors. Cancer Treatment Report, 67, 923931.
  • De Graaf, H., Willemse, P.H., Bong, S.B., Piersma, H., Tjabbes, T., Van Veelen, H., Coenen, J.L. & De Vries, E.G. (1996) Dose intensity of standard adjuvant CMF with granulocyte colony- stimulating factor for premenopausal patients with node-positive breast cancer. Oncology, 53, 289294.
  • Gregory, S., Goh, Y.T., Fuerst, M.P., O'Brien, T. & Giles, F.J. (1998) Fludarabine (F), Cyclophosphamide (C), and GM-CSF (FCGM) is effective in chronic lymphatic leukemia (CLL) and low grade non Hodgkins lymphoma (LGNHL). Blood, 92, 276b.
  • Hryniuk, W. & Bush, H. (1984) The importance of dose intensity in chemotherapy of metastatic breast cancer. Journal of Clinical Oncology, 2, 12811288.
  • Lepage, E., Gisselbrecht, C., Haioun, C., Sebban, C., Tilly, H., Bosly, A., Morel, P., Herbrecht, R., Reyes, F. & Coiffier, B. (1993) Prognostic significance of received relative dose intensity in non-Hodgkin's lymphoma patients: application to LNH-87 protocol. The GELA (Groupe d'Etude des Lymphomes de l'Adulte). Annals of Oncology, 4, 651656.
  • Liberati, A.M., Cinieri, S., Schippa, M., Di Clemente, F., Filippo, S. & Grignani, F. (1991) GM-CSF: clinical trials in non-Hodgkin's lymphoma patients with chemotherapy induced leucopenia. Leukemia, 5, 119122.
  • Lopez, A.F., Williamson, D.J., Gamble, J.R., Begley, C.G., Harlan, J.M., Klebanoff, S.J., Waltersdorph, A., Wong, G., Clark, S.C. & Vadas, M.A. (1986) Recombinant human granulocyte-macrophage colony-stimulating factor stimulates in vitro mature human neutrophil and eosinophil function, surface receptor expression, and survival. Journal of Clinical Investigations, 78, 12201228.
  • Lyman, G.H., Kuderer, N.M. & Djulbegovic, B. (2002) Prophylactic granulocyte colony-stimulating factor in patients receiving dose-intensive cancer. American Journal of Medicine, 112, 406411.
  • Magrath, I., Adde, M., Shad, A., Venzon, D., Seibel, N., Gootenberg, J., Neely, J., Arndt, C., Nieder, M., Jaffe, E., Wittes, R.A. & Horak, I.D. (1996) Adults and children with small non-cleaved-cell lymphoma have a similar excellent outcome when treated with the same chemotherapy regimen. Journal of Clinical Oncology, 14, 925934.
  • Ozer, H., Armitage, J.O., Bennett, C.L., Crawford, J., Demetri, G.D., Pizzo, P.A., Schiffer, C.A., Smith, T.J., Somlo, G., Wade, J.C., Wade, III, J.L., Winn, R.J., Wozniak, A.J. & Somerfield, M.R. (2000) 2000 update of recommendations for the use of hematopoietic colony-stimulating factors: evidence-based, clinical practice guidelines. American Society of Clinical Oncology Growth Factors Expert Panel. Journal of Clinical Oncology, 18, 35583585.
  • Parmar, M.K., Torri, V. & Stewart, L. (1998) Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Statistics in Medicine, 17, 28152834.
  • Pettengell, R., Gurney, H., Radford, J.A., Deakin, D.P., James, R., Wilkinson, P.M., Kane, K., Bentley, J. & Crowther, D. (1992) Granulocyte colony-stimulating factor to prevent dose-limiting neutropenia in non-Hodgkin's lymphoma: a randomized controlled trial. Blood, 80, 14301436.
  • Pettengell, R., Testa, N.G., Swindell, R., Crowther, D. & Dexter, T.M. (1993) Transplantation potential of hematopoietic cells released into the circulation during routine chemotherapy for non-Hodgkin's lymphoma. Blood, 82, 22392248.
  • Pfreundschuh, M. (2001) 2-weekly CHOP (CHOP-14): the new standard for patients with aggressive Non Hodgkin's lymphoma (NHL) > 60 years of age. Blood, 98, #3027.
  • Rusthoven, J., Bramwell, V. & Stephenson, B. (1998) Use of granulocyte colony-stimulating factor (G-CSF) in patients receiving myelosuppressive chemotherapy for the treatment of cancer. Provincial Systemic Treatment Disease Site Group. Cancer Prevention and Control, 2, 179190.
  • Skipper, H.E. (1990) Dose intensity versus total dose of chemotherapy: an experimental basis. In: Important Advances in Oncology (ed. by V. DeVita Jr, S. Hellman & S.A. Rosenberg), pp. 4364. J.B. Lippincott Company, Philadelphia.
  • Souetre, E. & Qing, W. (1994) Economic analysis of lenograstim in the correction of neutropenia following chemotherapy for non-Hodgkin's lymphoma. Pharmacoeconomics, 6, 3643.
  • Sweetenham, J.W., Hackshaw, A. & Knight, A. (2000) Evidence-based review and meta-analysis of the use of hematopoietic growth factors (HGF) in the management of malignant lymphoma (ML). Blood, 96, 139a.
  • Woll, P.J., Hodgetts, J., Lomax, L., Bildet, F., Cour-Chabernaud, V. & Thatcher, N. (1995) Can cytotoxic dose-intensity be increased by using granulocyte colony-stimulating factor? A randomized controlled trial of lenograstim in small-cell lung cancer. Journal of Clinical Oncology, 13, 652659.
  • Yau, J.C., Neidhart, J.A., Triozzi, P., Verma, S., Nemunaitis, J., Quick, D.P., Mayernik, D.G., Oette, D.H., Haynes, F.A. & Holcenberg, J. (1996) Randomized placebo-controlled trial of granulocyte-macrophage colony-stimulating-factor support for dose-intensive cyclophosphamide, etoposide, and cisplatin. American Journal of Hematology, 51, 289295.
  • Zinzani, P.L., Pavone, E., Storti, S., Moretti, L., Fattori, P.P., Guardigni, L., Falini, B., Gobbi, M., Gentilini, P., Lauta, V.M., Bendandi, M., Gherlinzoni, F., Magagnoli, M., Venturi, S., Aitini, E., Tabanelli, M., Leone, G., Liso, V. & Tura, S. (1997) Randomized trial with or without granulocyte colony-stimulating factor as adjunct to induction VNCOP-B treatment of elderly high-grade non- Hodgkin's lymphoma. Blood, 89, 39743979.
  • Zinzani, P.L., Storti, S., Zaccaria, A., Moretti, L., Magagnoli, M., Pavone, E., Gentilini, P., Guardigni, L., Gobbi, M., Fattori, P.P., Falini, B., Lauta, V.M., Bendandi, M., Gherlinzoni, F., De Renzo, A., Zaja, F., Mazza, P., Volpe, E., Bocchia, M., Aitini, E., Tabanelli, M., Leone, G. & Tura, S. (1999) Elderly aggressive-histology non-Hodgkin's lymphoma: first-line VNCOP-B regimen experience on 350 patients. Blood, 94, 3338.