Antibióticos profilácticos o FEC‐G(M) para la prevención de infecciones y la mejora de la supervivencia en pacientes con cáncer que reciben quimioterapia mielotóxica
Resumen
Antecedentes
La neutropenia febril (NF) y otras complicaciones infecciosas son algunas de las más graves toxicidades relacionadas con la quimioterapia para el cáncer, con una tasa de mortalidad de un 2% a un 21%. Los dos principales tipos de regímenes profilácticos son los factores estimulantes de colonias de granulocitos (FEC‐G) o de granulocitos‐macrófagos (FEC‐GM) y los antibióticos, con frecuencia quinolonas o cotrimoxazol. Ciertas guías actuales importantes recomiendan el uso de los factores estimulantes de colonias cuando el riesgo de neutropenia febril está por encima del 20%, pero no mencionan el uso de antibióticos. Sin embargo, se ha mostrado que ambos regímenes reducen la incidencia de infecciones. Ya que ninguna revisión sistemática ha comparado los dos regímenes, se realizó una revisión sistemática.
Objetivos
Comparar la eficacia y seguridad del FEC‐G(M) en comparación con los antibióticos en pacientes con cáncer que reciben quimioterapia mielotóxica.
Métodos de búsqueda
Se realizaron búsquedas en The Cochrane Library, MEDLINE, EMBASE, bases de datos de ensayos en curso y en resúmenes de congresos de la Sociedad Americana de Oncología Clínica y la Sociedad Americana de Hematología (1980 a diciembre de 2015). Se planeó incluir tanto publicaciones de texto completo como de resúmenes. Dos autores de la revisión realizaron el cribaje (screening) de los resultados de la búsqueda de forma independiente.
Criterios de selección
Se incluyeron ensayos controlados aleatorizados (ECA) que comparaban la profilaxis con FEC‐G(M) versus antibióticos para la prevención de la infección en pacientes con cáncer de todas las edades que recibían quimioterapia. Todos los brazos del estudio tenían que recibir regímenes de quimioterapia idénticos y otro tratamiento de apoyo. Se incluyeron los textos completos, así como los resúmenes y los datos no publicados si se disponía de información suficiente sobre el diseño de los estudios, las características de los participantes, las intervenciones y los resultados. Se excluyeron los ensayos cruzados (crossover), los ensayos cuasialeatorizados y los ensayos retrospectivos post‐hoc.
Obtención y análisis de los datos
Dos autores de la revisión analizaron de forma independiente los resultados de las estrategias de búsqueda, extrajeron los datos, evaluaron el riesgo de sesgo y analizaron los datos de acuerdo con los métodos estándar de Cochrane. La interpretación final se realizó junto con un médico experimentado.
Resultados principales
En esta revisión actualizada, no se incluyeron nuevos ensayos controlados aleatorizados. Se incluyeron dos ensayos en la revisión, uno con 40 pacientes con cáncer de mama que recibían quimioterapia de alta dosis y FEC‐G en comparación con los antibióticos, y otro que evaluaba a 155 pacientes con cáncer de pulmón de células pequeñas que recibían FEC‐GM o antibióticos.
Se juzgo el riesgo general de sesgo tan alto en el ensayo del FEC‐G, ya que ni los pacientes ni los médicos fueron cegados y no todos los pacientes incluidos fueron analizados al azar (7 de 40 pacientes). Se consideró que el riesgo general de sesgo en el FEC‐GM era moderado, debido al riesgo de sesgo de rendimiento (no se cegó ni a los pacientes ni al personal), pero el riesgo de sesgo de selección y desgaste era bajo.
En el ensayo que comparaba el FEC‐G con los antibióticos, no se informó de la mortalidad por todas las causas. No hubo evidencia de una diferencia en la mortalidad relacionada con la infección, con cero eventos en cada brazo. No se notificaron infecciones documentadas microbiológica o clínicamente, infecciones graves, calidad de vida y acontecimientos adversos. No hubo evidencia de una diferencia en la frecuencia de la neutropenia febril (riesgo relativo (RR) 1,22; intervalo de confianza (IC) del 95%: 0,53 a 2,84). La calidad de la evidencia de los dos resultados comunicados, la mortalidad relacionada con la infección y la frecuencia de la neutropenia febril, fue muy baja, debido al escaso número de pacientes evaluados (alta imprecisión) y al alto riesgo de sesgo.
No hubo evidencia de una diferencia en términos de tiempo medio de supervivencia en el ensayo que comparaba el FEC‐GM con los antibióticos. Los tiempos de supervivencia a dos años fueron del 6% (0 al 12%) en ambos brazos (alta imprecisión, baja calidad de la evidencia). Hubo cuatro muertes tóxicas en el brazo del FEC‐GM y tres en el brazo de los antibióticos (3,8%), sin evidencia de diferencia (RR 1,32; IC del 95%: 0,30 a 5,69; P = 0,71; baja calidad de la evidencia). Hubo un 28% de infecciones de grado III o IV en el brazo de FEC‐GM y un 18% en el brazo de antibióticos, sin ninguna evidencia de diferencia (RR 1,55; IC del 95%: 0,86 a 2,80; P = 0,15, baja calidad de la evidencia). Hubo 5 episodios de 360 ciclos de infecciones de grado IV en el brazo del FEC‐GM y 3 episodios de 334 ciclos en el brazo del cotrimoxazol (0,8%), sin evidencia de diferencia (RR 1,55; IC del 95%: 0,37 a 6,42; P = 0,55; baja calidad de la evidencia). No hubo una diferencia significativa entre los dos brazos en cuanto a toxicidades no hematológicas como la diarrea, la estomatitis, las infecciones, los eventos adversos neurológicos, respiratorios o cardíacos. La trombopenia de grado III y IV se produjo con una frecuencia significativamente mayor en el brazo del FEC‐GM (60,8%) en comparación con el brazo de los antibióticos (28,9%); (RR 2,10; IC del 95%: 1,41 a 3,12; P = 0,0002; baja calidad de la evidencia). En este ensayo no se informó de la mortalidad relacionada con la infección, la incidencia de la neutropenia febril ni la calidad de vida.
Conclusiones de los autores
Como sólo se encontraron dos pequeños ensayos con 195 pacientes en total, no es posible llegar a ninguna conclusión para la práctica clínica. Se necesitan más ensayos para evaluar los efectos beneficiosos y perjudiciales del FEC‐G(M) en comparación con los antibióticos para la prevención de infecciones en pacientes con cáncer que reciben quimioterapia.
PICO
Resumen en términos sencillos
Antibióticos profilácticos o FEC‐G(M) para la prevención de infecciones en pacientes con cáncer sometidos a quimioterapia
Pregunta de la revisión
Se revisó la literatura existente que examina la eficacia y la seguridad de los factores estimulantes de colonias de granulocitos (macrófagos) (FEC‐G(M)) en comparación con los antibióticos para prevenir infecciones en pacientes con cáncer que reciben quimioterapia.
Antecedentes
El tratamiento del cáncer con quimioterapia (fármacos contra el cáncer) altera el sistema inmunológico y reduce el recuento de glóbulos blancos. Esto aumenta el riesgo de infección en las personas. Tanto los factores estimulantes de colonias de granulocitos (FEC‐G(M)) como los antibióticos pueden reducir el riesgo de infección asociada a los tratamientos de cáncer. La revisión comparó la eficacia de los antibióticos con la de los FEC‐G(M) para la prevención de la infección.
Características de los estudios
Se realizaron búsquedas en varias bases de datos médicas y se identificaron dos ensayos controlados aleatorizados (ECA) que cumplieron con los criterios de inclusión; no se identificaron nuevos ensayos para esta actualización de la revisión. Un ensayo incluyó 40 pacientes con cáncer de mama que recibieron altas dosis de quimioterapia. Dieciocho pacientes recibieron FEC‐G y 22 antibióticos (ciprofloxacino y anfotericina) para prevenir la infección. Otro ensayo evaluó el FEC‐GM frente a los antibióticos en pacientes con cáncer de pulmón de células pequeñas, con 78 pacientes en el brazo del FEC‐GM y 77 pacientes en el brazo de los antibióticos.
Resultados clave
El estudio que analizó el FEC‐G frente a los antibióticos no informó de todas las causas de mortalidad, infecciones documentadas microbiológica o clínicamente, infecciones graves, calidad de vida o eventos adversos. No se encontró evidencia de una diferencia entre las dos opciones profilácticas para los resultados de la mortalidad relacionada con la infección (ningún paciente murió a causa de la infección), o la neutropenia febril.
El ensayo que evaluó el FEC‐GM versus los antibióticos no encontró ninguna evidencia de una diferencia en la mortalidad por todas las causas, la mortalidad del ensayo, las infecciones o las infecciones graves. La única diferencia entre los dos brazos se encontró para el evento adverso de trombocitopenia, que favorece a los pacientes que reciben antibióticos. En este ensayo no se informó sobre la calidad de vida.
Se necesita más investigación para determinar la mejor prevención contra la infección en los pacientes de cáncer.
Calidad de la evidencia
La calidad de la evidencia de mortalidad relacionada con la infección y la frecuencia de la neutropenia febril en el ensayo del FEC‐G fue muy baja, debido al pequeño número de pacientes que se evaluaron y al diseño del estudio (alto riesgo de sesgo). El ensayo que analizó el FEC‐GM frente a los antibióticos informó sobre la supervivencia general, las muertes tóxicas, las infecciones, las infecciones graves y los eventos adversos. Debido al muy pequeño número de pacientes incluidos, se consideró que la calidad general de todos estos resultados era baja.
La evidencia está actualizada hasta diciembre de 2015.
Authors' conclusions
Summary of findings
| G‐CSF compared with antibiotics for the prevention of infections and improvement of survival in cancer patients receiving myelotoxic chemotherapy | ||||||
| Patient or population: cancer patients receiving myelotoxic chemotherapy Intervention: G‐CSF Comparison: antibiotics | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Antibiotics | G‐CSF | |||||
| All cause mortality | see comment | not reported | ||||
| Infection‐related mortality | see comment | 40 (1 RCT) | ⊕⊝⊝⊝1,2 | no patient died of infectious causes during the 18‐week duration of the trial | ||
| Quality of life | see comment | not reported | ||||
| Incidence of febrile neutropenia | 318 per 1000 | 388 per 1000 (169 to 904) | RR 1.22 (0.53 to 2.84) | 40 (1 RCT) | ⊕⊝⊝⊝1,2 | |
| Incidence of severe infections | see comment | not reported | ||||
| Adverse events | see comment | not reported | ||||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 due to the low number of very low number of events, the result is highly imprecise (downgraded by 2 points) 2 high risk of performance bias (neither patients nor physicians blinded) and detection bias (no intention to treat analysis) (downgraded by 1 point) | ||||||
| GM‐CSF compared with antibiotics for the prevention of infections and improvement of survival in cancer patients receiving myelotoxic chemotherapy | ||||||
| Patient or population: cancer patients receiving myelotoxic chemotherapy Intervention: GM‐CSF Comparison: antibiotics | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Antibiotics | GM‐CSF | |||||
| All cause mortality | see comment | 115 (1 RCT) | ⊕⊕⊝⊝1 | Two‐year survival times were 6% (0 to 12%) in both arms | ||
| Infection‐related mortality | see comment | not reported | ||||
| Quality of life | see comment | not reported | ||||
| Incidence of febrile neutropenia | see comment | not reported | ||||
| Incidence of severe infections (Grade III or IV) | 182 per 1000 | 282 per 1000 (156 to 509) | RR 1.55 (0.86 to 2.80) | 115 (1 RCT) | ⊕⊕⊝⊝1 | not reported |
| Adverse events Toxic deaths | 39 per 1000 | 51 per 1000 (12 to 222) | RR 1.32 (0.30 to 5.69) | 115 (1 RCT) | ⊕⊕⊝⊝1 | |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 due to the low number of very low number of events, the result is highly imprecise (downgraded by 2 points) | ||||||
Background
Description of the condition
Cancer patients receiving myelosuppressive therapy or haematopoetic stem cell transplantation are at increased risk of febrile neutropenia and infectious complications. The risk of febrile neutropenia and subsequent infection is directly related to the duration and severity of neutropenia (Bodey 1966; Bodey 1986). Infectious complications constitute major dose‐limiting side effects in patients undergoing myelosuppressive therapy. Special risk circumstances, such as patient age greater than 65 years or poor performance status, impact the associated morbidity and mortality (Kuderer 2006; Pizzo 1999). The mortality rate associated with febrile neutropenia in cancer patients is between 2% and 21% (Smith 2015). In addition, infectious complications are a common cause of dose reductions during chemotherapy treatment.
Febrile neutropenia (FN) can be prevented by a prophylactic regimen. Prophylaxis started at the beginning of the first chemotherapy cycle or in parallel with documented or anticipated neutropenia is called primary prophylaxis, whereas prophylaxis given to patients who had already experienced episodes of FN in an earlier chemotherapy cycle, is referred to as secondary prophylaxis. Effective prophylaxis, using either colony‐stimulating factors (CSF) or antibiotics (or both), would decrease clinically relevant negative outcomes such as all cause mortality, infection‐related mortality, and infectious complications. Given the high costs of the consequences of FN, and also of the colony‐stimulating factors themselves, economic arguments are introduced into discussions on the best prophylactic strategy (Kuderer 2006; Leibovici 2006).
In clinical trials addressing the prevention of FN, granulocyte‐macrophage colony‐stimulating factors (GM‐CSFs) have been reported to be effective in reducing the duration and severity of chemotherapy‐induced febrile neutropenia (Johnston 2000; Holmes 2002). Prophylaxis, using antibiotics, has also been shown to be beneficial with reduced fever, incidence of infections and hospitalisations (Bucaneve 2005; Cullen 2005).
Description of the intervention
Colony‐stimulating factors (CSF)
The current American Society of Clinical Oncology (ASCO) guidelines justify the administration of CSFs in clinical settings where the expected risk of suffering FN is approximately 20% (Smith 2015). In addition to the myelotoxicity of the planned chemotherapy regimen, patient‐specific risk factors should be taken into account. Secondary prophylaxis with CSFs is recommended for patients who have developed a neutropenic complication in a previous chemotherapy cycle, and in whom a reduced dose might compromise disease‐free or overall survival, or treatment outcome. The guidelines from the Infectious Diseases Working Party (AGIHO) of the German Society for Haematology and Medical Oncology (DGHO) give similar recommendations (Vehreschild 2014).
Thus far, randomised controlled trials (Crawford 1991; Trillet‐Lenoir 1993), and subsequent meta‐analyses, have shown that primary prophylaxis with CSFs is effective in reducing FNin patients with both solid and haematological malignancies (Bohlius 2008; Hackshaw 2004; Lyman 2002; Sung 2004; Sung 2007; Wittman 2006). Furthermore, GM‐CSFs may decrease hospitalisation and the use of intravenous therapeutic antibiotics (Crawford 1991; Trillet‐Lenoir 1993). In a meta‐analysis on the use of G‐CSFs in cancer patients hospitalised with established FN, the authors observed a possible benefit of adding GM‐CSFs to antibiotic treatment on infection‐related mortality and length of hospitalisation(Clark 2005). A meta‐analysis by Kuderer 2006 showed that under certain standard dose chemotherapy regimens, early and infection‐related mortality were also reduced with primary G‐CSF prophylaxis. However, none of the meta‐analyses with less restrictive inclusion criteria were able to demonstrate that prophylactic administration of GM‐CSFs improved overall survival when compared to placebo or no treatment. None of these analyses addressed the question of GM‐CSFs versus antibiotics, which is a question closer to clinical reality. One group did a subgroup analysis of studies in which the published report mandated antibiotic prophylaxis compared to those that did not, and found no difference between the groups (Sung 2007). This may be due to the high number of trials where no information about antibiotic prophylaxis use is available. In addition, this meta‐analysis included studies that analysed cycles of chemotherapy as opposed to patients. The distorting effect of such an analysis is difficult to estimate.
Of the many meta‐analyses looking at GM‐CSF versus placebo or no treatment, only one meta‐analysis, restricted to patients with lymphoma, was published in The Cochrane Library (Bohlius 2008). This analysis found a reduction in the rate of infections (odds ratio (OR) 0.74; 95% CI 0.64 to 0.85) but no effect on infection‐related mortality (OR 1.37 favouring control; 95% CI 0.66 to 2.82).
GM‐CSF is usually well tolerated, with only a moderate number of adverse events, mostly bone pain and headaches, however, there are some hints of increased risk of acute myeloid leukaemia or myelodysplastic syndromes (Lyman 2010).
Antibiotics
During the last decade, prophylaxis with antibiotics was studied in a number of randomised clinical trials. The evidence provided was not considered to be entirely convincing, because none of the studies were sufficiently large to provide conclusive evidence on the real efficacy of prophylaxis (Bucaneve 2005; Cullen 2005; Karp 1987; Lew 1995). Subsequent meta‐analyses suggested that prophylaxis using antibiotics reduced the incidence of gram‐negative bacterial infection, total infection, fever episodes, and hospitalisation (Cruciani 2003; Engels 1998). Moreover, a meta‐analysis of data on antibiotic prophylaxis (or more specifically, fluoroquinolones) compared to placebo or no intervention demonstrated that not only infections were reduced, but all cause mortality, and infection‐related mortality were too (Gafter‐Gvili 2005; Gafter‐Gvili 2012; Leibovici 2006). One important question which is still unanswered is whether prophylaxis should be considered for all patients with cancer and neutropenia. In another meta‐analysis on antibiotic prophylaxis, the majority of patients were suffering from haematological malignancies and received high‐dose chemotherapy and bone marrow transplantation, with only a few studies focusing on solid tumours (Cullen 2005; Gafter‐Gvili 2012). Another factor possibly compromising the results of the main meta‐analysis was that studies were included that randomised chemotherapy cycles and not patients, or reported cycle‐based outcomes, as opposed to a true incidence (where the number of patients and not cycles are analysed). No information on GM‐CSFs compared to antibiotics was available from these analyses.
How the intervention might work
Colony‐stimulating factors
Granulocyte colony‐stimulating factors (G‐CSF) predominantly augment the proliferation, maturation, and release of neutrophils, resulting in a dose‐dependent increase in circulating neutrophils (Bronchud 1988; Morstyn 1988). It is a growth factor for the myeloid lineage that stimulates the growth of granulocytes and eosinophil colonies; granulocyte (macrophage) colony‐stimulating factors (GM‐CSF) also stimulate the growth of macrophages (Griffin 1990). Both colony‐stimulating factors have shown comparable results in decreasing the incidence and duration of neutropenia and fever after chemotherapy. However, there is a lack of formal comparisons between the two drugs. Probably due to the macrophage activation caused by GM‐CSF, but not G‐CSF, tolerability of GM‐CSF has been reported to be inferior. Injection site reactions in particular, seem more frequent with GM‐CSF (Alvarado 1999; Beveridge 1997; Beveridge 1998; Fischmeister 1999; Hovgaard 1992). Given the undesired additional effects of GM‐CSF and concerns of tumour stimulation by GM‐CSF, the drug has become more or less disregarded by recent clinical studies and guidelines (Smith 2015). Granulocyte (macrophage) colony‐stimulating factors is no longer commercially available in several European countries for infection prophylaxis. It is licensed for mobilisation of stem cells, and after autologous or allogeneic stem cell transplantation (Smith 2015).
Antibiotics
Antibiotic prophylaxis, most often using flouroquinolones, reduces infections by targeting potential pathogens, and in contrast to G‐CSFs it does not provoke the dose‐limiting effect of haematological toxicity. A major concern of a routine prophylactic use of antibiotics in patients with cancer and neutropenia is that it increases bacterial resistance to these agents. This, in turn, may compromise the treatment success of both current and future serious infections by expanding (multi)resistance. In addition, hypersensitivity reactions, gastrointestinal toxicities, and the promotion of fungal overgrowth after antibiotics put the patient at risk of potentially serious adverse events. These factors may limit their efficacy in reducing infection‐related morbidity or mortality (Carratala 1995; Gafter‐Gvili 2007; Somolinos 1992).
Why it is important to do this review
The best prophylactic treatment of febrile neutropenia and infections in cancer patients receiving antineoplastic therapy remains controversial, and in general, international guidelines concentrate on either antibiotics or G‐CSFs. The evidence outlined above suggests that prophylaxis with an antibiotic might be as effective as with G‐CSFs for reducing both infections and mortality.
The aim of this systematic review is to provide a comprehensive overview on the benefits and harms of G‐CSF compared to antibiotics for infection prophylaxis in cancer patients. By systematically identifying all randomised trials conducted to date and by conducting a critical review of their reliability and validity, we will mitigate the statistical limitations of individual studies.
Objectives
To compare the efficacy and safety of G‐CSF or GM‐CSF compared to antibiotics in cancer patients receiving myelotoxic chemotherapy.
Methods
Criteria for considering studies for this review
Types of studies
We included only randomised controlled trials (RCTs). We excluded cross‐over trials and quasi‐randomised trials. We included full‐text, abstracts, and unpublished data if sufficient information on study design, participant characteristics, interventions and outcomes was available.
Types of participants
We planned to include paediatric and adult, male and female patients with a confirmed diagnosis of any type of cancer who were undergoing myelotoxic chemotherapy. Both solid and haematological malignancies were eligible.
Types of interventions
We included trials comparing G‐CSF or GM‐CSF to antibiotics in the primary prophylaxis of infection‐related complications. Trials that examined pegylated G‐CSF (pegfilgrastim) were eligible, provided pegfilgrastim was given once, 24 hours after the completion of chemotherapy.
Comparison 1
-
G‐CSF versus antibiotics
Comparison 2
-
GM‐CSF versus antibiotics
Trials looking at secondary prophylaxis, defined as prophylaxis in a patient who suffered from FN in an earlier course of chemotherapy, were also eligible, but a subgroup analysis was planned. However, we did not identify any trial evaluating secondary prophylaxis.
We included studies in which the intended chemotherapy regimen and supportive care did not differ between study arms. Therefore, we excluded studies that compared dose‐intensified, dose‐accelerated, or dose‐dense regimens with standard chemotherapy, as this resulted in different chemotherapy protocols in the arm that received antibiotic prophylaxis and the arm that received CSF prophylaxis. Trials with more than two arms were included, provided at least two arms with the relevant comparison had the same chemotherapy protocol.
We excluded trials using G‐CSF, GM‐CSF, or antibiotics to treat febrile neutropenia, fever, or infections.
Types of outcome measures
Primary outcomes
-
Overall survival
-
All cause mortality (including infection‐related, treatment‐related, or on‐trial mortality)
-
Infection‐related mortality
Studies focusing solely on the efficacy of prophylaxis will most likely have short‐term follow‐up only, mainly providing information on early mortality. Determining the cause of death in severely ill patients can be associated with measurement bias. Therefore, we extracted all cause mortality, comprising infection‐related and treatment‐related mortality.
Secondary outcomes
-
Microbiologically or clinically documented infections, or both
-
We accepted any definition of clinically documented or microbiologically documented infections given by authors. If available, we extracted data on all, not only severe, clinically or microbiologically documented infections. Microbiologically documented infections were required to have some kind of cultural confirmation of the infection. Infections reported without information on microbiological confirmation were considered to be clinically documented infections.
-
-
Severe infections
-
Frequency of febrile neutropenia (FN; any definition of fever and neutropenia accepted)
-
Quality of life (QoL; if measured with a validated QoL instrument)
-
Adverse events
Search methods for identification of studies
For this updated review, we revised the search strategy used for the first review. We used search strategies based on those described in Chapter 6 of the Cochrane Handbook for Systematic Reviews of Interventions (Lefebvre 2011). We did not use any language constraints.
Electronic searches
We searched the following electronic databases:
-
Cochrane Central Register of Controlled Trials (CENTRAL; The Cochrane Library, December 2015; see Appendix 1)
-
MEDLINE (1980 to December 2015; for search strategy see Appendix 2)
-
EMBASE (1980 to January 2008; for search strategy see Appendix 3)
Since we revised our searches, we re‐ran them for CENTRAL and MEDLINE for the entire period, i.e. 1980 to 2015.
Searching other resources
We searched conference proceedings of the following annual meetings, which were not included in CENTRAL for abstracts:
-
American Society of Hematology (ASH) from 2000 to 2015
-
American Society of Clinical Oncology (ASCO) from 2000 to 2015
-
European Hematology Association (EHA) from 2000 to 2015
We electronically searched the database of ongoing trials:
We handsearched the following references:
-
References of all identified trials, relevant review articles and current treatment guidelines
Data collection and analysis
Selection of studies
Two review authors (NS, OB) independently screened the results of the search strategies for eligibility by reading the abstracts. In the case of disagreement, we obtained the full‐text publication. If no consensus could be reached, we consulted a third review author, in accordance with Chapter 7 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
We documented the study selection process in a flow chart as recommended in the PRISMA statement (Moher 2009), showing the total numbers of retrieved references and the numbers of included and excluded studies.
Data extraction and management
Two review authors independently extracted the data according to the guidelines proposed by Higgins 2011b. If required, we contacted authors of individual studies for additional information. We used a standardised data extraction form containing the following items:
-
General information: author; title; source; publication date; country; language; duplicate publications.
-
Quality assessment ('Risk of bias' assessment): sequence generation; allocation concealment; blinding (participants, personnel, outcome assessors); incomplete outcome data; selective outcome reporting; other potential sources of bias.
-
Study characteristics: trial design; aims; setting and dates; source of participants; inclusion and exclusion criteria; comparability of groups; subgroup analysis; statistical methods; power calculations; treatment cross‐overs; compliance with assigned treatment; length of follow‐up; time point of randomisation.
-
Participant characteristics: age; diagnosis; stage of disease; prior treatments; number of participants recruited, allocated, and evaluated; participants lost to follow‐up; noticeable differences in risk factors for developing FN.
-
Interventions: duration; type; dose and timing of GM‐CSF, G‐CSF, antibiotics, and other infection prophylaxes (e.g. antimycotics); concomitant treatment (setting, duration, type of chemotherapy); and supportive care (e.g. type of empirical antibiotic therapy).
-
Outcomes: all cause mortality; infection‐related mortality; microbiologically or clinically documented infections, or both; severe infections; QoL; frequency of FN; adverse events.
Assessment of risk of bias in included studies
Two review authors (NS and OB) independently assessed the risk of bias for each study using the following criteria outlined in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions(Higgins 2011a).
-
Sequence generation
-
Allocation concealment
-
Blinding (participants, personnel, outcome assessors)
-
Incomplete outcome data
-
Selective outcome reporting
-
Other potential sources of bias
We made a judgement for every criterion, using one of three categories.
-
'Low risk': if the criterion was adequately fulfilled in the study, i.e. the study was at a low risk of bias for the given criterion.
-
'High risk': if the criterion was not fulfilled in the study, i.e. the study was at high risk of bias for the given criterion.
-
'Unclear': if the study report did not provide sufficient information to allow for a judgement of 'Yes' or 'No', or if the risk of bias was unknown for one of the criteria listed above.
Measures of treatment effect
We used intention‐to‐treat data. For binary outcomes, we calculated risk ratios (RRs) with 95% confidence intervals (CIs) for each comparison. We did not identify or extract time‐to‐event or continuous outcomes.
Unit of analysis issues
We evaluated the number of patients with events rather than number of episodes, as the second one could be biased (e.g. a patient with one episode of febrile neutropenia is at increased risk to have a second episode of febrile neutropenia).
Dealing with missing data
As suggested in Chapter 16 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b), there were many potential sources of missing data that had to be taken into account: at the study level, outcome level, and summary data level. It is important to distinguish between 'missing at random' and 'not missing at random'. As we only identified one trial without missing data, we did not contact the original investigators.
Assessment of heterogeneity
As we only found two trials, which we did not meta‐analyse, we did not assess heterogeneity of treatment effects between trials.
Assessment of reporting biases
In meta‐analyses with at least 10 trials included for one outcome, we would have explored potential publication bias by generating a funnel plot and statistically testing this by using a linear regression test (Sterne 2011). We would have considered a P value of less than 0.1 to be significant for this test. However, as we analysed two trials only, we did not generate a funnel plot.
Data synthesis
As we only identified one trial for each comparison, we could not pool data. However, to analyse data for individual studies we entered data into Review Manager (RevMan) 5.3.
Moreoever, we created 'Summary of findings' tables for each comparison on absolute risks in each group with the help of the GRADE approach, and will use it to summarise the evidence of all cause mortality, infection‐related mortality, quality of life, incidence of febrile neutropenia, incidence of severe infections and adverse events.
Subgroup analysis and investigation of heterogeneity
We had considered performing subgroup analyses using the following characteristics:
-
Different types of underlying malignant disease;
-
Different baseline risk for febrile neutropenia or infection;
-
Study setting (in‐patients or out‐patients);
-
Different type of treatment (e.g. haematologic stem cell transplantation versus standard chemotherapy);
-
Different types of G‐CSFs used;
-
Age (<18 versus ≥ 18 years); and
-
According to whether regimens included antimycotic prophylaxis.
However, as we had insufficient data to meta‐analyse, we could not perform these analyses.
Sensitivity analysis
We had considered performing sensitivity analyses using the following quality criteria:
-
Quality components with regard to low and high risk of bias;
-
Fixed‐effect modelling versus random‐effects modelling;
-
Duration of study; and
-
full‐text publication versus abstract publication only.
Again, as we identified only two trials, which were too heterogenous to pool, we could not perform these analyses.
Results
Description of studies
Results of the search
The literature search was designed to find all relevant articles where G‐CSFs, GM‐CSFs, or antibiotics were used as prophylactic agents. For this update, we set up a new search covering all time periods, i.e. after removing duplicates, we screened titles and abstracts of 11,785 references and excluded 11,696 at the initial stage. We assessed the full text of the remaining 89 references and excluded 87 references with reasons (see Excluded studies). As we identified no new trial fitting the inclusion criteria for this review update, we included the two already known trials in this review. See Figure 1 for study flow diagram.
Study flow diagram.
Included studies
Two studies fulfilled the inclusion criteria of this review. One study involved adults with breast cancer receiving high‐dose chemotherapy, and compared prophylaxis for at least six cycles (Schroder 1999). The other trial evaluated patients with small‐cell lung cancer receiving accelerated chemotherapy (Sculier 2001). For more details see Characteristics of included studies.
Design
Schroder 1999 was an open‐label randomised (1:1) study. Sculier 2001 was a three‐arm trial, two arms of which could be analysed in this review. The third arm evaluated standard chemotherapy without any infectious prophylaxis.
Sample sizes
Schroder 1999 included 40 patients, 18 in the G‐CSF prophylaxis arm and 22 in the antibiotics arm. Sculier 2001 included 243 patients, 233 of whom were eligible. However, 78 of these patients received an intervention not applicable for this review, therefore 155 patients were analysed in this review.
Locations
Location is not reported by Schroder 1999, the Sculier 2001 trial took place in several European countries.
Participants
Schroder 1999 randomised chemotherapy‐naive patients with breast cancer who received three, three‐week courses of intravenous cyclophosphamide (1500 mg/m²), epirubicin (80 mg/m²), and 5‐fluouracil (1500 or 1000 mg/m²) given on day one; followed by three cycles of intravenous cyclophosphamide (1500 mg/m²), 5‐fluouracil (600 mg/m²) on day one and intravenous methotrexate (1500 mg/m²) on day two. Sculier 2001 included patients with small‐cell lung cancer receiving six courses of EVI (epirubicin 90 mg/m², vindesine 3 mg/m² and ifosfamide 5 g/m²) every 14 days.
Interventions
In the G‐CSF arm in the Schroder 1999 trial, patients received 263 µg subcutaneous of G‐CSF (lenograstim) on days 3 through to day 12 of each cycle. Patients in the antibiotics arm received two oral prophylactic agents, a combination of ciprofloxacin (250 mg twice daily) and amphotericin B (500 mg four times per day) on days 3 through to day 17 of each cycle, without blinding of the study participants. Patients in the Sculier 2001 study received either GM‐CSF as a daily subcutaneous dose of 5 µg/kg, from day 3 through to day 13 or until the neutrophil count reached ≥ 4000 mm³ after nadir, or cotrimoxazole (160 mg trimethoprim plus 800 mg sulfamethoxazole). This was administered orally every 12 hours from day three until the end of the courses of chemotherapy.
Outcomes
Schroder 1999 evaluated infection‐related mortality, episodes of hospitalisation for febrile neutropenia, duration of hospitalisation for febrile neutropenia, grade IV leucopenia, and analysed costs of prophylaxis. Sculier 2001 assessed overall survival, tumour response, absolute and relative dose intensity, incidence of infections and severe infection, and adverse events. None of the trials evaluated quality of life.
Conflict of interest
Funding not reported.
Excluded studies
We excluded 87 trials with reasons (one trial included two comparisons (Tjan‐Heijnen 2003):
-
14 trials compared antibiotics to placebo or no further treatment (Attal 1991; Carlson 1997; Cullen 2005; Dickgreber 2009; Karp 1986; Lamy 1993; Lee 2002; Petersen 1988; Pignon 1990; Rafecas 1989; Schuette 2011; Talbot 1993; Tjan‐Heijnen 2003; Yamada 1993).
-
46 trials compared antibiotics and G‐CSF or GM‐CSF to antibiotics only (Aarts 2013; Alonzo 2002; Altman 1996; Ardizzoni 1994; Bishop 2000; Bradstock 2001; Burton 2006; Clarke 1999; Dibenedetto 1995; Ernst 2008; Faber 2006; Garcia 2000; Garcia‐Saenz 2002; Geissler 1997; Gonzalez‐Vicent 2004; Greenberg 1996; Gulati 1992; Heath 2003; Hecht 2010; Heil 1997; Joshi 2003; Ladenstein 2010; Lee 1998; Lehrnbecher 2007; Little 2002; Maiche 1993; McQuaker 1997; Michel 2000; Miles 1994; Nemunaitis 1995; Nolan 2007; Ojeda 1999; Ottmann 1995; Pettengell 1992; Piccirillo 1999; Przepiorka 2001; Pui 1997; Schmitz 2004; Spitzer 1994; Stahel 1994; Timmer‐Bonte 2005; Trigg 2000; Welte 1996; Witz 1998; Yau 1996; Zinzani 1997a).
-
7 trials compared antibiotics and G‐CSF to G‐CSF only (Eleutherakis‐Papaiakovou 2010; Feng 2014; Kim 2005; Lalami 2004; Lee 1998; Suh 2008; Tjan‐Heijnen 2003).
-
21 trials compared G‐CSF to placebo or no further treatment (Bennett 2001; Björkholm 1999; Brugger 2009; Chevallier 1995; Crawford 1997; Doorduijn 2005; Dunlop 1996; Fridrik 1997; Godwin 1998; Hartmann 1997; Holowiecki 2002; Kosaka 2015; Larson 1998; Michon 1998; Osby 2003; Patte 2002; Romieu 2007; Seymour 1995; Trillet‐Lenoir 1993; Veyret 2006; Vogel 2005).
Risk of bias in included studies
See Figure 2 and Figure 3 for risk of bias summary.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Allocation
Both trials were described as randomised, but the randomisation procedure was not reported. Therefore, we judged the risk of selection bias as unclear.
Blinding
There was no blinding of the participants or personnel due to the use of either an oral antibiotic or subcutaneous injections of GM‐CSF; no information was given about whether or not the assessors were blinded. Therefore we judged potential risk of performance bias as high and of detection bias as unclear.
Incomplete outcome data
As 23 courses from seven patients from the antibiotics group, who switched to rhG‐CSF, were not included in the analysis by Schroder 1999, we judged the risk of attrition bias as high in this trial. All patients in the Sculier 2001 trial were evaluated as randomised, reasons for ten patients not being eligible after randomisation were given. Therefore, we judged risk of attrition bias for this trial as low.
Selective reporting
As we did not identify study protocols; it is unclear if all the planned outcomes are reported. We judged the risk of reporting bias as unclear.
Other potential sources of bias
As no other potential source of bias was reported, we judged this bias as "unclear".
Effects of interventions
See: Summary of findings for the main comparison ; Summary of findings 2
Comparison 1: G‐CSF versus antibiotics
Overall survival
Not reported by Schroder 1999.
All cause mortality (including infection‐related, treatment‐related, or on‐trial mortality)
Not reported by Schroder 1999.
Infection‐related mortality
Infection‐related mortality was the same in both groups of the Schroder 1999 trial: no patient died of infectious causes during the 18‐week duration of the trial.
Microbiologically or clinically documented infections
Not reported.
Incidence of severe infections
Not reported
Quality of life (QoL)
Not reported.
Incidence of febrile neutropenia (FN)
Schroder 1999 reported febrile neutropenia in 7/18 patients receiving G‐CSF and in 7/22 patients receiving ciprofloxacin and amphotericin B (relative risk (RR) 1.22; 95% confidence interval (CI) 0.53 to 2.84).
Adverse events
Not reported.
Comparison 2: GM‐CSF versus antibiotics
Overall survival
There was no evidence of a difference in median survival time, with 264 (95% CI 220 to 308) days for patients in the GM‐CSF arm and 264 (95% CI 223 to 305 days) in the antibiotics arm (Sculier 2001). Two‐year survival times were 6% (0 to 12%) in both arms.
All cause mortality (including infection‐related, treatment‐related, or on‐trial mortality)
There were four toxic deaths in the GM‐CSF arm (5.1%) and three in the antibiotics arm (3.8%), without evidence for a difference (RR 1.32; 95% CI 0.30 to 5.69; P = 0.71).
Infection‐related mortality
This outcome was not reported in Sculier 2001.
Microbiologically or clinically documented infections
There were 22 grade III or IV infections (28%) in the GM‐CSF arm in the Sculier 2001 trial and 14 infections (18%) in the antibiotics arm, without any evidence of a difference (RR 1.55; 95% CI 0.86 to 2.80; P = 0.15).
Incidence of severe infections
There were 5 episodes out of 360 cycles (1.3%) of grade IV infections in the GM‐CSF arm and 3 episodes out of 334 cycles in the cotrimoxazole arm (0.8%), without evidence of a difference (RR 1.55; 95% CI 0.37 to 6.42; P = 0.55).
Quality of life (QoL)
Not reported.
Incidence of febrile neutropenia (FN)
Not reported.
Adverse events
There was no significant difference between the two arms for non‐haematological toxicities like diarrhoea, stomatitis, infections, neurologic, respiratory or cardiac adverse events. Grade III and IV thrombopenia occurred significantly more frequently in the GM‐CSF arm (60.8%) compared to the antibiotics arm (28.9%); with a RR 2.10; 95% CI 1.41 to 3.12; P = 0.0002.
Discussion
Summary of main results
The striking finding of this review is that there is only one very small study comparing granulocyte colony‐stimulating factors (G‐CSF) to antibiotics for infection prophylaxis in cancer patients receiving myelosuppressive chemotherapy, and one trial with 155 patients evaluating granulocyte macrophage colony‐stimulating factors (GM‐CSF) versus antibiotics. The trial evaluating G‐CSF did not report all cause mortality, incidence of documented or severe infections, quality of life, or adverse events. We did not find evidence of a difference in infection‐related mortality (none of the 40 included patients died because of infection), or in incidence of febrile neutropenia.
The trial that evaluated GM‐CSF reported overall survival, toxic deaths, infections and severe infections and non‐haematological adverse events, without any evidence of a difference between the GM‐CSF arm and the antibiotics arm. Patients in the antibiotics arm had fewer thrombopenic adverse events. Quality of life was not reported.
Overall completeness and applicability of evidence
As only two small trials were identified, it is not possible to come to a final conclusion regarding the best prophylactic regimen in cancer patients at risk of neutropenia. Therefore, this clinically important question remains unanswered. Moreover, the trial assessing G‐CSF evaluated only a few of the outcomes of interest (incidence of febrile neutropenia and infection‐related mortality), but all cause mortality, incidence of documented or severe infections, quality of life, and adverse events were not assessed.
The trial evaluating GM‐CSF versus antibiotics reported more of the outcomes of interest (overall survival, toxic deaths, infections and severe infections and adverse events), however, due to the small sample size, there was no evidence of a difference, except for the adverse event, thrombocytopenia.
The 41 trials that were excluded because they evaluated the influence of the combination of GM‐CSF and antibiotics compared to GM‐CSF or antibiotics only, underline the huge imbalance between the number of direct comparisons of the two drugs we evaluated in this review, and the number of trials that were conducted in this field.
Quality of the evidence
The risk of bias in Schroder 1999 was high, as this trial was not blinded and not all patients of the included 40 patients were analysed as randomised (seven of 22 patients from the antibiotics arm crossed‐over to G‐CSF and were excluded from analysis). The risk of bias for Sculier 2001 could be considered to be moderate, as risk of performance bias was high, but risk of selection and attrition bias was low.
As only two trials could be included in this review, one evaluating G‐CSF, the other evaluating GM‐CSF, no meta‐analysis was possible.The trial evaluating G‐CSF reported infection‐related mortality and incidence of febrile neutropenia. We judged the quality of evidence for both outcomes to be very low, due to the small number of events, which lead to high imprecision (downgraded by two levels), and the high risk of bias (downgraded by one level).
The trial that analysed GM‐CSF versus antibiotics reported overall survival, toxic deaths, infections, severe infections and adverse events. Because of the very small number of patients included, we downgraded overall quality of the evidence for all outcomes by two levels (high imprecision). As risk of bias was moderate in this trial, we did not downgrade the quality of evidence for this reason. Therefore, overall quality for all the outcomes mentioned above was considered to be low.
Potential biases in the review process
To prevent bias within the review, we considered only RCTs and performed all relevant processes in duplicate. We developed a sensitive search strategy, and searched all relevant data from international cancer congresses by hand to minimise potential publication bias. We are not aware of any obvious deficiencies in our review process. The small number of trials included in this review could lead to publication bias as a funnel plot could not be generated.
Agreements and disagreements with other studies or reviews
One comprehensive meta‐analysis of GM‐CSF versus control includes 148 trials with more than 16,000 patients (Sung 2007). However, in this publication it is not reported how many patients received additional antibiotics, and how many patients received either G‐CSF or GM‐CSF. Similarly, the most comprehensive antibiotics versus control meta‐analysis includes 49 trials with more than 6000 patients (for the outcome all cause mortality; Gafter‐Gvili 2005). The low number of trials directly comparing antibiotics to G‐CSFs is surprising, considering the higher cost of GM‐CSFs compared to standard antibiotics. However, a high number of trials comparing GM‐CSFs to control received funding from pharmaceutical companies that produce GM‐CSFs. As there are only two small trials directly comparing G‐CSF or GM‐CSF versus antibiotics, no final conclusion on the best prophylactic regimen is possible. Clearly, more trials with larger numbers of patients are required to answer this question, in particular, with regard to early all cause and infection‐related mortality. In addition, GM‐CSF is no longer commercially available for infection prophylaxis in several European countries; it is licensed instead for mobilisation of stem cells or after autologous or allogeneic stem cell transplantation (Smith 2015).
Study flow diagram.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
| G‐CSF compared with antibiotics for the prevention of infections and improvement of survival in cancer patients receiving myelotoxic chemotherapy | ||||||
| Patient or population: cancer patients receiving myelotoxic chemotherapy Intervention: G‐CSF Comparison: antibiotics | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Antibiotics | G‐CSF | |||||
| All cause mortality | see comment | not reported | ||||
| Infection‐related mortality | see comment | 40 (1 RCT) | ⊕⊝⊝⊝1,2 | no patient died of infectious causes during the 18‐week duration of the trial | ||
| Quality of life | see comment | not reported | ||||
| Incidence of febrile neutropenia | 318 per 1000 | 388 per 1000 (169 to 904) | RR 1.22 (0.53 to 2.84) | 40 (1 RCT) | ⊕⊝⊝⊝1,2 | |
| Incidence of severe infections | see comment | not reported | ||||
| Adverse events | see comment | not reported | ||||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 due to the low number of very low number of events, the result is highly imprecise (downgraded by 2 points) 2 high risk of performance bias (neither patients nor physicians blinded) and detection bias (no intention to treat analysis) (downgraded by 1 point) | ||||||
| GM‐CSF compared with antibiotics for the prevention of infections and improvement of survival in cancer patients receiving myelotoxic chemotherapy | ||||||
| Patient or population: cancer patients receiving myelotoxic chemotherapy Intervention: GM‐CSF Comparison: antibiotics | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Antibiotics | GM‐CSF | |||||
| All cause mortality | see comment | 115 (1 RCT) | ⊕⊕⊝⊝1 | Two‐year survival times were 6% (0 to 12%) in both arms | ||
| Infection‐related mortality | see comment | not reported | ||||
| Quality of life | see comment | not reported | ||||
| Incidence of febrile neutropenia | see comment | not reported | ||||
| Incidence of severe infections (Grade III or IV) | 182 per 1000 | 282 per 1000 (156 to 509) | RR 1.55 (0.86 to 2.80) | 115 (1 RCT) | ⊕⊕⊝⊝1 | not reported |
| Adverse events Toxic deaths | 39 per 1000 | 51 per 1000 (12 to 222) | RR 1.32 (0.30 to 5.69) | 115 (1 RCT) | ⊕⊕⊝⊝1 | |
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 due to the low number of very low number of events, the result is highly imprecise (downgraded by 2 points) | ||||||
