Introduction

In patients with cancer receiving cytotoxic chemotherapy, chemotherapy-induced neutropenia (CIN) and febrile neutropenia (FN) are frequent complications. CIN is graded according to severity of the reduction of the absolute neutrophil count (ANC) and FN is commonly defined as ANC <0.5 × 109/L with an oral temperature ≥38 °C for more than 1 h [1]. Patients experiencing neutropenic events are more susceptible to subsequent infections [2]. As a consequence of FN, patients often require hospitalisation and antibiotic treatment and frequently have their chemotherapy dose reduced or delayed [3, 4]. Modifications to chemotherapy may decrease its effectiveness, thereby potentially compromising treatment outcomes [4].

Granulocyte colony-stimulating factors (G-CSFs) stimulate the production and maturation of neutrophils during chemotherapy and reduce the incidence and duration of CIN and incidence of FN [5, 6]. Prophylactic G-CSF use from the first cycle of chemotherapy is recommended by the European Organisation for Research and Treatment of Cancer [7] and other international guidelines [1, 8, 9] if the planned chemotherapy regimen is associated with an FN risk of 20 % or more. For chemotherapy regimens with an intermediate FN risk of 10–20 %, guidelines recommend that patient-related and disease-related factors should also be considered when deciding whether or not to give G-CSF support.

Daily G-CSFs are primarily cleared through the kidneys and require dosing until recovery of the neutrophil count. Long-acting G-CSFs are primarily cleared by neutrophils and have significantly reduced renal clearance compared with daily G-CSFs. They therefore require only a single dose per chemotherapy cycle. Pegfilgrastim (Neulasta®; Amgen Inc., CA, USA), consisting of the human recombinant G-CSF filgrastim pegylated at the N terminus with a 20-kDa polyethylene glycol molecule, is administered subcutaneously as a single 6 mg dose [10]. It was approved in both the USA and Europe in 2002. Lipegfilgrastim (Lonquex®; Teva Pharma B. V.), a long-acting filgrastim molecule that is pegylated at a different site from pegfilgrastim, was approved in Europe in 2013 [11]. Other long-acting G-CSFs, such as balugrastim, are in clinical development [12].

The emergence of these recently developed long-acting G-CSFs necessitates a re-evaluation of the evidence. Direct comparative data are limited, and there are no systematic reviews of long-acting G-CSFs that include data from both observational studies and randomised controlled trials (RCTs). Therefore, we conducted a systematic review to capture the available data on the efficacy, safety and effectiveness of long-acting G-CSFs for prophylaxis of CIN and FN in adult patients with cancer.

Methods

Study design

The systematic review was performed according to a pre-specified protocol that was agreed by all authors. We searched the following electronic databases: MEDLINE In-Process & Other Non-Indexed Citations and OVID MEDLINE 1948–present, EMBASE 1980–present and the Cochrane Library. A search of abstract books was also conducted from the annual meetings of the American Society of Clinical Oncology, the American Society of Hematology, the European Hematology Association, the European Society for Medical Oncology, the European Multidisciplinary Cancer Congress, the International Society for Pharmacoeconomics and Outcomes Research and the Multinational Association of Supportive Care in Cancer. Complete search strings are listed in Online Resource 1. The electronic database searches included articles published up to April 2013 and were restricted to English-language studies. Conference abstracts were limited to those published between January 2009 and April 2013. This report follows the PRISMA statement for reporting systematic reviews and meta-analyses [13].

Study selection

Initially, two independent reviewers screened the titles and abstracts of the search results for studies of human adult haematology or oncology patients who were receiving long-acting-G-CSF primary prophylaxis to reduce the risk of CIN during chemotherapy. Studies in which patients received bone marrow transplantation were excluded. Clinical trials and observational studies were included. Editorials, letters, case reports, guidelines, health technology assessment reports, economic evaluations, narrative reviews and research protocols were excluded. Papers were excluded if they did not report neutropenia-related outcomes. Full texts of the remaining articles were then assessed by the reviewers for final inclusion. Additional exclusion criteria were applied at this second stage: studies comparing pegfilgrastim with a daily G-CSF, placebo or no prophylaxis were excluded if fewer than 50 patients received pegfilgrastim; studies with pegfilgrastim alone (which therefore allowed no comparisons) were excluded if fewer than 100 patients received pegfilgrastim. Studies in which pegfilgrastim was used outside of its approved indication were excluded. These additional exclusion criteria were not applied to studies involving new long-acting G-CSFs because we expected to find far fewer papers on these and wanted to ensure that all available data on these other agents were captured. Papers or abstracts reporting results from the same study were indicated as such. If a study included in the form of a congress abstract was published as a peer-reviewed paper after our literature search, we included the paper in place of the congress abstract.

Data extraction

The data collection comprised study and patient characteristics, efficacy (effect of a treatment under controlled, clinical trial conditions), effectiveness (effect of a treatment under uncontrolled, real-world conditions) and safety. Detailed definitions of outcome measures are listed in Online Resource 2. Studies were classified according to their design: ‘RCTs’ where patients were randomised to G-CSFs; clinical trials in which patients were not randomly assigned to neutropenia prophylaxis or no treatment were termed ‘clinical trials’; and studies of routine clinical practice were termed ‘observational studies’. Evidence found in the literature was extracted as presented by the original authors of the study.

Risk of bias assessment

Two independent reviewers assessed risk of bias; disagreements were resolved within the reviewer team by consensus. RCTs were assessed using the Cochrane Collaboration’s assessment tool [14]. Non-randomised studies were assessed using the Methods Guide for Comparative Effectiveness Reviews of the US Agency for Healthcare Research and Quality [15]. Six domains of bias (selection bias, performance bias, detection bias, attrition bias, reporting bias and other bias) were assessed. Based on the reviewers’ judgments, every article was rated as having a ‘low’, ‘high’ or ‘unclear’ risk of bias. Risk of bias was not assessed for conference abstracts.

Results

Eligible trials and study characteristics

Our search identified 731 full publications and 108 congress abstracts (Fig. 1). After removing duplicates, 700 items were left, of which 482 were excluded on the basis of title and abstract screening, leaving 218 articles (Online Resource 3). Three relevant articles were published after completion of the search: Bondarenko et al. (2013) [16], Almenar-Cubells et al. (2013) [17] and Volovat et al. (2013) [18]; these were included to replace congress abstracts identified by the initial search that described the same studies [19–21]. Finally, 33 publications and 11 congress abstracts representing 41 studies were analysed. Key characteristics of the included studies are presented in Table 1.

Fig. 1
figure 1

PRISMA flow diagram

Full size image
Table 1 Overview of study and patient characteristics of included studies
Full size table

Figure 2 illustrates the number of patients exposed to each of the included substances or treatment strategies, the G-CSF interventions used and the study design. The studies included 13 that looked at pegfilgrastim alone, 15 studies in which pegfilgrastim was compared with a daily G-CSF, three studies in which pegfilgrastim was compared with placebo and two studies in which pegfilgrastim primary prophylaxis was compared with no pegfilgrastim primary prophylaxis. We found eight studies that compared other long-acting G-CSFs with daily G-CSFs, pegfilgrastim or placebo. The number of patients who received a long-acting G-CSF was 50,089 (pegfilgrastim = 49,207; lipegfilgrastim = 505; balugrastim = 281; Maxy-G34 = 27; Ro 25-8315 = 28; BCD-017 = 41).

Fig. 2
figure 2

Network diagram. The total number of patients included in randomised controlled trials, clinical trials and observational studies, and in whom a given granulocyte colony-stimulating factor (G-CSF) has been investigated, is stated below the G-CSF’s or comparator’s name. Single-arm studies including pegfilgrastim only are reported in the ‘Pegfilgrastim’ shape. Comparisons between G-CSFs are indicated by arrows specifying the type of study. The arrows point from the investigated G-CSF to the comparator

Full size image

Pegfilgrastim studies included patients with breast, lung, colorectal or gastro-esophageal cancer, Hodgkin’s lymphoma, non-Hodgkin’s lymphoma, acute myeloid leukaemia and various other solid tumours. These studies included patients taking 12 standard chemotherapy regimens and numerous non-standard regimens. All studies of newer long-acting G-CSFs except one (which looked at lipegfilgrastim in non-small cell lung cancer [22]) were conducted in patients with breast cancer, most of whom were receiving docetaxel and doxorubicin.

Risk of bias assessment

Risk of bias was typically higher in non-randomised trials and observational studies than in RCTs (Fig. 3). Most studies excluded patients receiving concomitant antibiotic prophylaxis or who had previously received chemotherapy; therefore, risk of performance bias was low. Risk of reporting bias was difficult to assess across all types of studies because the study protocols were not published.

Fig. 3
figure 3

Risk of bias assessment of included studies. a Randomised controlled trials (RCTs) are defined as those studies in which patients were randomised to a granulocyte colony-stimulating factor (G-CSF). Adequate sequence generation and allocation concealment refers to selection bias, blinding of participants and staff to performance bias, blinding of outcome assessors to detection bias, incomplete outcome data to attrition bias and selective outcome reporting refers to reporting bias. b Non-randomised clinical trials are those in which patients were not randomised to a G-CSF. The risk of bias assessment includes non-randomised clinical trials and observational studies that included more than one G-CSF. c The risk of bias assessment includes non-randomised clinical trials and observational studies that included pegfilgrastim only

Full size image

Efficacy and effectiveness of pegfilgrastim

Table 2 shows efficacy and effectiveness endpoints for studies of pegfilgrastim alone or compared with daily G-CSFs, placebo or no treatment.

Table 2 Efficacy, effectiveness and safety of pegfilgrastim
Full size table

Incidence of FN

Three RCTs reported a significant reduction in FN for pegfilgrastim versus placebo (1 % vs 17 % [23], 2 % vs 6 % [24, 25] and 2 % vs 8 % [26]) in patients with breast or colorectal cancer receiving chemotherapy regimens associated with various FN risk profiles. One RCT designed to demonstrate non-inferiority in duration of severe neutropenia reported a significant reduction in FN incidence for pegfilgrastim versus filgrastim (9 % vs 18 %) in patients with breast cancer [27]. Another RCT with a similar design found a non-significant trend towards lower FN incidence for pegfilgrastim versus filgrastim (13 % vs 20 %) [28].

Ten clinical trials reported FN incidence across numerous tumour and chemotherapy types, including several dose-dense regimens. In eight of these trials, all patients received pegfilgrastim; FN incidence ranged from 1 to 10 % [29–36]. A study in which FN prophylaxis was changed by protocol amendment in subsequent cohorts of patients with primary breast cancer treated with a high FN-risk regimen (docetaxel, doxorubicin and cyclophosphamide) found a significant reduction in the incidence of FN for pegfilgrastim versus daily G-CSF (7 % vs 18 %) [37]. In contrast, another breast cancer trial in which G-CSF schedules were selected at the physician’s discretion reported a higher FN incidence for pegfilgrastim versus filgrastim (11 % vs 4 %) [38].

Observational studies showed FN incidence was higher among patients with haematological malignancies (14–16 %) [39, 40] than in those with solid tumours (4–12 %) [41–43, 17, 44–46]. Five of these observational studies that reported FN incidence compared neutropenia prophylaxis: two studies across various tumour types reported trends towards reduced FN incidence with pegfilgrastim versus daily G-CSF (11 % vs 24 % and 7 % vs 13 %, respectively) [17, 44], one found a significant reduction (5 % vs 7 %) [45] and two did not find a difference for pegfilgrastim versus filgrastim in non-Hodgkin’s lymphoma (NHL) [39] and breast cancer [47]. Significant reductions in FN incidence for pegfilgrastim primary prophylaxis versus no pegfilgrastim primary prophylaxis were also seen in observational studies of patients with breast cancer (4 % vs 30 %) and in patients with various tumour types (odds ratio [95 % confidence interval (CI)] = 0.49 [0.34–0.68]) [48, 41].

Incidence of CIN

An RCT in patients with colorectal cancer treated with chemotherapy with a low FN risk (FOLFOX, FOLFIRI or FOIL) found pegfilgrastim significantly reduced CIN incidence compared with placebo (13 % vs 43 %) [26]. RCTs comparing pegfilgrastim with filgrastim in a non-inferiority setting reported no significant difference in CIN incidence in patients with breast cancer receiving chemotherapy associated with a high FN risk [27, 28].

In clinical trials investigating dose-dense regimens, CIN incidence with pegfilgrastim was low and ranged from 3 to 11 % in patients with breast cancer [29, 31, 32] and 34 % in gastro-esophageal cancer [34]. In studies of standard-dose chemotherapy regimens across various tumour types, CIN incidence ranged from 22 to 30 % [36, 35]. One trial reported that pegfilgrastim significantly reduced the incidence of CIN compared with daily G-CSF (37 % vs 58 %) in patients with breast cancer [37].

Three observational studies reporting CIN incidence compared neutropenia prophylaxis; a difference was not found between pegfilgrastim and filgrastim in patients with breast cancer [47], but in patients with various tumours or NHL CIN incidence was lower in those receiving pegfilgrastim than those receiving daily G-CSF (28 % vs 49 % and 41 % vs 50 %) [17, 49].

Incidence of hospitalisations due to CIN or FN

One RCT reported a significant reduction in FN-related hospitalisations in patients with breast cancer who received pegfilgrastim versus placebo (1 % vs 14 %) [23], while another in patients with colorectal cancer found no significant difference in CIN-related hospitalisations [26].

In a clinical trial including patients with various tumour types receiving pegfilgrastim primary prophylaxis in community-based practices in the USA, the incidence of FN-related hospitalisations was 4 % [35]. A similar study in elderly patients found the incidence of CIN- or FN-related hospitalisations was 5 % [36]. Two clinical trials of patients with breast cancer found no significant difference in incidence and duration of FN-related hospitalisations between pegfilgrastim and daily G-CSFs [37, 38].

Three retrospective observational studies enrolling patients with various tumour types found trends towards reduced incidence of hospitalisations due to FN for pegfilgrastim versus daily G-CSF (9 % vs 20 %, 3 % vs 11 % and 3 % vs 7 %) [17, 44, 50], whereas another found no significant difference between sargramostim and pegfilgrastim [51]. Two other retrospective observational studies [52, 53] reported significant decreases in the risk of CIN-related hospitalisations for pegfilgrastim compared with filgrastim (1 % vs 4 % and 1 % vs 2 %); findings supported by a study of two US databases that found pegfilgrastim reduced the risk of neutropenia-related hospitalisation compared with filgrastim [54].

Incidence of chemotherapy dose reductions and delays

In one RCT in patients with breast cancer receiving pegfilgrastim or placebo, there was no significant difference in the proportion of patients receiving their full chemotherapy dose on schedule [23]; however, cross-over from the placebo to the pegfilgrastim arm was allowed if FN occurred. Another RCT in colorectal cancer reported a significant decrease in dose reductions (3 % vs 11 %) and delays (4 % vs 20 %) due to neutropenia for pegfilgrastim versus placebo [26].

There was a wide range of incidence of dose delays and reductions in the clinical trials (2–77 % and 2–33 %, respectively), but most papers did not specify whether or not the chemotherapy modifications were due to neutropenia [38, 31, 34, 35]. Only one clinical trial compared the incidence of dose delays (due to FN events and non-haematological toxicity) with pegfilgrastim and filgrastim in patients with breast cancer. It found no significant difference between the two arms [38].

Rates of dose delays and reductions in observational studies also varied considerably between trials (5–55 % and 5–42 %, respectively) [39–41, 43, 17, 44]. One study found a significantly lower incidence of delays for pegfilgrastim primary prophylaxis versus no pegfilgrastim primary prophylaxis in patients with breast cancer (5 % vs 12 %), but found no significant difference in dose reductions [41]. In two studies of patients with various tumour types, fewer dose delays (42 % vs 55 %) [17] and dose reductions (32 % vs 38 % and 7 % vs 21 %) [17, 44] due to neutropenia for pegfilgrastim versus daily G-CSF were observed. In a population of Asian patients with NHL, rates of dose reductions and delays were slightly higher in patients who received pegfilgrastim than in those who received filgrastim [39].

Antibiotic use

In one RCT, a non-significant reduction in antibiotic use was reported for pegfilgrastim versus filgrastim (17 % vs 21 %) in patients with breast cancer [28]. Two RCTs reported a significant reduction in the use of antibiotics due to FN for pegfilgrastim versus placebo, one in breast cancer (2 % vs 10 %) [23] and one in colorectal cancer (2 % vs 7 %) [26].

A clinical trial in breast cancer found no significant difference in the use of antibiotics between patients receiving pegfilgrastim and filgrastim (11 % vs 4 %) [38].

An observational study found a significant reduction in the use of antibiotics for pegfilgrastim primary prophylaxis versus no pegfilgrastim primary prophylaxis (28 % vs 46 %) in patients with breast cancer [41]. Two observational studies in patients with various tumour types found lower rates of FN-related antibiotic use in patients who received pegfilgrastim than those receiving daily G-CSF (4 % vs 11 % and 8 % vs 17 %); in the former study, this difference reached significance [17, 44].

Safety of pegfilgrastim

Table 2 shows safety endpoints for studies of pegfilgrastim alone or compared with daily-G-CSFs, placebo or no treatment.

All G-CSF-related adverse events

Two RCTs in patients with breast cancer reported that G-CSF-related adverse events (AEs) were similar for pegfilgrastim and filgrastim [27, 28]. Another RCT found a non-significant increase in G-CSF-related AEs for pegfilgrastim compared with placebo (11 % vs 1 %) in patients with colorectal cancer, primarily due to increased bone pain [26]. Pegfilgrastim-related serious AEs were also infrequent (0.5 %) in patients with various tumours in a clinical trial [35]. Two observational studies in patients with various tumours reported a non-significant decrease in G-CSF-related AEs for pegfilgrastim versus daily G-CSF (6 % vs 10 % and 1 % vs 5 %) [17, 44]. None of the studies reported any fatal AEs that were attributed to G-CSF prophylaxis.

Musculoskeletal pain

In two placebo-controlled RCTs including patients with breast or colorectal cancer, occurrence of any-grade musculoskeletal pain was higher in the pegfilgrastim arms than the placebo arms (31 % vs 27 % and 11 % vs 1 %) [23, 26]. In two further RCTs of patients with breast cancer randomised to pegfilgrastim or filgrastim, overall rates of bone pain were comparable between arms [27, 28], and severe bone pain appeared reduced for pegfilgrastim versus filgrastim (1 % vs 8 %) [28].

In five non-comparative clinical trials, the incidence of any-grade musculoskeletal pain with pegfilgrastim reported ranged from 7 to 26 % [29, 36] and the incidence of severe musculoskeletal pain ranged from 0 to 9 % [29, 31, 32, 35] across patients with breast cancer and various tumour types.

In general, the reported incidence of musculoskeletal pain was lower in observational studies than in clinical trials. The incidence of any-grade musculoskeletal pain with pegfilgrastim in observational studies varied, from 6 % in one study where all patients received pegfilgrastim (with no patients experiencing serious bone or muscle pain) [46] to 50 % in patients receiving either pegfilgrastim or filgrastim [47]. In two other observational studies of patients with various tumour types that compared pegfilgrastim with daily G-CSF, bone pain was less common in the pegfilgrastim arms (2 % vs 6 % and 1 % vs 3 %) [17, 44].

Other long-acting G-CSFs

Table 3 shows the efficacy and safety endpoints for studies involving other long-acting G-CSFs.

Table 3 Efficacy and safety of other long-acting granulocyte colony-stimulating factors
Full size table

Lipegfilgrastim

Lipegfilgrastim is pegylated at a different site from pegfilgrastim (threonine 134) using a carbohydrate linker involving two enzymatic steps. In a placebo-controlled RCT in patients with lung cancer, there was no statistically significant reduction in the first-cycle incidence of FN compared to placebo (2 % vs 6 %) and a significant reduction in the first-cycle incidence of severe neutropenia (32 % vs 59 %) [22]. G-CSF-related AEs were more common in the lipegfilgrastim arm (14 % vs 10 %) [22]. In a non-inferiority RCT comparing lipegfilgrastim with pegfilgrastim in patients with breast cancer, there was no significant difference in FN incidence (1 % vs 3 %) and a non-significant reduction in severe neutropenia incidence (44 % vs 51 %) [16]. Rates of FN-related hospitalisations and antibiotic use were also comparable between the two study arms (1 % vs 2 % and 1 % vs 3 %, respectively) [16]. AEs, including bone pain (14 % vs 10 %), myalgia (9 % vs 6 %) and arthralgia (5 % vs 2 %), were slightly more common with lipegfilgrastim than with pegfilgrastim, but the difference was not significant [16]. In a second RCT in breast cancer, duration of severe neutropenia for lipegfilgrastim and pegfilgrastim was reported to be similar [55].

Balugrastim

Balugrastim is a non-pegylated recombinant fusion protein composed of human serum albumin and G-CSF harvested from yeast. It has been investigated at a dose of 40 mg in two RCTs in patients with breast cancer treated with doxorubicin and docetaxel. In one, incidence (58 % vs 59 %) and duration (1.1 days vs 1 day) of severe neutropenia in cycle 1 were similar for balugrastim and pegfilgrastim [18]. There was no significant difference in FN incidence in cycle 1 between balugrastim and pegfilgrastim (1 % vs 3 %) [18]. The frequency of treatment-related AEs was similar for balugrastim and pegfilgrastim (20 % vs 19 %) [18]. The second RCT found similar durations of severe neutropenia for balugrastim and pegfilgrastim (1.3 days vs 1.2 days) [19].

BCD-017, Maxy-G34 and Ro 25-8315

BCD-017 (empegfilgrastim), Maxy-G34 and Ro 25-8315 are all covalent conjugates of recombinant human G-CSF and polyethylene glycol. Small RCTs compared BCD-017 and Ro 25-8315 with filgrastim in patients with breast cancer but found that neutropenia-related outcomes, including rates of FN, were generally lower in the filgrastim arms [56, 57]. Safety data were reported in the Ro 25-8315 study and suggest G-CSF-related AEs are more common with Ro 25-8315 than with filgrastim [57]. Maxy-G34 was compared with pegfilgrastim in a clinical trial. The incidence of FN and duration of CIN were similar in the two study arms [58]. No safety data were reported.

Discussion

To our knowledge, this is the only systematic review of long-acting G-CSFs that includes newly developed agents and data from both clinical trials and observational studies. We identified 12 RCTs, 12 clinical trials and 17 observational studies, including 58,342 patients in total. Studies in patients with breast cancer were dominant, partly because these were the registration studies for the G-CSFs.

Pegfilgrastim studies included a range of patient populations, cancer types and stages, and chemotherapy regimens. Efficacy and effectiveness results were generally consistent. Although pegfilgrastim did not uniformly show better efficacy or effectiveness in all studies, the vast majority showed better efficacy or effectiveness compared to daily G-CSF, no upfront pegfilgrastim, no G-CSF or placebo in terms of reduction of the incidence of CIN (4/7 studies), FN (11/14 studies), chemotherapy dose delays and reductions (6/8 studies), antibiotic use (6/7 studies) and neutropenia-related hospitalisations (9/13 studies). The observed variation may be partly explained by differences in patient populations and cancer types, or in the way G-CSF was administered. Thirteen (35 %) studies of pegfilgrastim reported safety data and most of these focused on musculoskeletal pain; only three studies reported other G-CSF-related AEs. This suggests that the safety profile of G-CSFs may be generally accepted and studies now investigate only specific AEs known to be associated with their use. The incidence of G-CSF-related AEs was similar between pegfilgrastim and filgrastim. The incidence of bone pain and severe bone pain was lower or no different for pegfilgrastim than filgrastim in most RCTs and observational studies (4/6 studies).

Previously published systematic reviews and meta-analyses of RCTs comparing pegfilgrastim with daily G-CSF or placebo by Cooper et al. and Pinto et al. found that pegfilgrastim more effectively reduced the incidence of FN [59, 60]. The RCT reported by Decaestecker et al. and Pinter et al. [24, 25], showing better efficacy for pegfilgrastim than placebo in reducing the incidence of neutropenia in colorectal cancer patients, reported in this systematic review was not included in these previous systematic reviews. We additionally included non-randomised clinical trials and observational studies that have not been included in former systematic reviews [59, 60]. Nevertheless, the results of our systematic review are generally consistent with these studies. However, while well-designed RCTs have a low risk of bias, inclusion criteria can be restrictive. The observational studies included in our review indicate an advantage for pegfilgrastim over daily G-CSFs or no treatment, suggesting that the efficacy of pegfilgrastim demonstrated in clinical trials has been translated into clinical practice. In fact, we found a greater magnitude of reduction in CIN incidence with pegfilgrastim versus filgrastim in observational studies than RCTs; this could be due to a shorter duration of G-CSF use in current practice (e.g. 5–6 days in clinical practice vs 10–11 days in clinical trials) [7]. Importantly, the safety data from observational studies were consistent with data from RCTs, suggesting that the pegfilgrastim safety profile can be used to guide treatment in a broad patient population. However, care should be taken when interpreting the results of observational studies, owing to the higher risk of bias and confounding factors.

Almost all the studies including other long-acting G-CSFs were RCTs of patients with breast cancer (7/8 studies) receiving doxorubicin and docetaxel (5/8 studies). Lipegfilgrastim has been the most extensively tested (3/8 studies) and appears to be similar to pegfilgrastim regarding the reduction in duration of severe neutropenia in patients with breast cancer. Efficacy of lipegfilgrastim in reducing the incidence of FN was not statistically superior to placebo in a congress abstract describing an RCT in patients with lung cancer [22]. Lipegfilgrastim has now been approved in Europe for reducing the incidence and duration of FN in adults with cancer who are receiving cytotoxic chemotherapy [11]. Further clinical and observational studies in a wider range of tumour types and chemotherapy regimens will confirm whether its efficacy and safety are maintained across a broader patient population in real-world clinical practice. Balugrastim has also been investigated in two phase 3 RCTs of patients with breast cancer and has an efficacy and safety profile comparable to that of pegfilgrastim. Again, further studies will determine whether this translates to other patient populations. Notably, the incidence of FN in the pegfilgrastim arms of the lipegfilgrastim and balugrastim studies (3 % in cycle 1 for both studies [16, 18]) was lower than in the registrational pegfilgrastim studies (9 % and 7 % in cycle 1 [27, 28]), despite a similar study design and patient population. Maxy-G34 also appears to be non-inferior to pegfilgrastim; however, it was tested in only a very small number of patients (n = 35) [58]. BCD-017 and Ro 25-8315 did not appear to be as effective at reducing the incidence of FN as filgrastim [56, 57].

Because very few studies reported long-term outcomes of G-CSF use and two systematic reviews by Kuderer and Lyman et al. [61, 62] looking at survival have previously been published, we did not include overall survival as an endpoint. In 2007, Kuderer et al. [61] published a systematic review of infection-related and early mortality during chemotherapy by type of G-CSF. They reported that there is insufficient data to draw conclusions. An updated analysis in 2013 by Lyman et al. [62] concluded that all-cause mortality is reduced in patients receiving chemotherapy with primary G-CSF support. However, Lyman et al. did not report results by type of G-CSF. We are still awaiting long-term survival data for the newer long-acting G-CSFs. Future research should examine in more detail the effects of long-acting G-CSFs on survival outcomes.

As is true for all systematic reviews, the validity of our findings is limited by the quality of its underlying studies. Another limitation is that some studies did not report how many patients received primary prophylaxis versus secondary prophylaxis. This may have led to an underestimation of effectiveness. Furthermore, the studies were not all consistent in their definitions of FN and CIN and the number of chemotherapy cycles over which they reported data. Finally, combined measures of effect are missing in our analysis.

It is clear that pegfilgrastim is widely used in clinical practice across a broad patient population. Lipegfilgrastim and balugrastim were similar to pegfilgrastim in reducing the duration and incidence of CIN and FN in five studies. Furthermore, the safety profiles of the recently developed long-acting G-CSFs were comparable to pegfilgrastim based on the phase 3 studies identified by this systematic review. These G-CSFs may prove to be valuable therapeutic options; however, there is a need for further studies in broader patient populations to confirm their effectiveness and safety in real-world clinical practice. New biosimilar G-CSFs and next-generation drugs targeting the G-CSF receptor are also in the early stages of development [12] and should be assessed against the current standard of care.