Introduction

Chemotherapy-induced (CIN) and febrile neutropenia (FN) are potentially life-threatening complications of myelosuppressive chemotherapy, may often require hospitalization, and may result in disruptions to the planned chemotherapy regimen [1–5]. Known risk factors for CIN/FN enable clinicians to risk-stratify patients and initiate prophylaxis with granulocyte colony-stimulating factors (GCSF) [1, 3, 6–11].

GCSFs are biological growth factors that stimulate the production of white blood cells by promoting the proliferation, differentiation, and activation of neutrophils in the bone marrow [12]. The efficacy of standard and pegylated agents in FN prophylaxis is well established in terms of decreasing the risk of FN, the severity and duration of FN episodes, and chemotherapy disturbances—with no sustained evidence of superiority of either formulation [9, 10, 13–17].

Following the patent expiration for filgrastim (Neupogen®, Amgen) in Europe in 2006, several biosimilar agents have been approved by the European Medicines Agency, including EP-2006 (Zarzio®, Filgrastim Hexal®; Sandoz/Novartis; hereafter, Zarzio®). A biosimilar or “similar biological medicinal product” is a “copy version of an already authorized biochemical medicinal product with demonstrated similarity in physicochemical characteristics, efficacy and safety, based on a comprehensive comparability exercise” [18 (p. 691)]. The clinical development of Zarzio® has been summarized [19] and reviewed in terms of clinical efficacy and safety [20, 21]. Initial clinical experience suggests an effectiveness and safety profile similar to that of the originator product in core and extrapolated indications [22–25].

The MONITOR-GCSF study examined the real-world patterns, outcomes, and associated determinants of Zarzio® prophylaxis in cancer patients receiving myelosuppressive chemotherapy [26, 27]. Being European in scope, the study was framed within the European Organisation for Research and Treatment of Cancer (EORTC) guidelines for the use of GCSF in CIN/FN prophylaxis [9, 10], which specify that the myelotoxicity and associated FN risk of patients’ regimens be assessed at each cycle. Primary prophylaxis is recommended for regimens with an FN risk ≥20 % and for regimens with a risk of 10–20 % if patients present with specific risk factors. GCSF prophylaxis at cycle 1 is not recommended if a regimen’s FN risk is <10 %. We report here on the treatment patterns and associated outcomes of CIN/FN prophylaxis with Zarzio® in 1,447 patients from 140 cancer centers in 12 European countries.

Methods

The methodology of MONITOR-GCSF has been described elsewhere [26, 27]. Methodology elements are summarized below.

Design

MONITOR-GCSF was an international, multi-center, prospective, observational, open-label, pharmaco-epidemiologic study of cancer patients treated with myelosuppressive chemotherapy regimens whose treating physicians prescribed CIN/FN prophylaxis with Zarzio® per their best clinical judgment. Patients were recruited from 140 centers in 12 European countries: Austria (3 centers), Belgium (2), Czech Republic (5), France (34), Germany (27), Hungary (8), Italy (23), Poland (14), Romania (7), Spain (11), Switzerland (2), and the UK (4).

Eligible were male or female adults (age ≥18) with stage 3 or 4 breast, ovarian, bladder, or lung cancer; metastatic prostate cancer; stage 3 or 4 diffuse large B cell lymphoma or multiple myeloma; and receiving primary or secondary prophylaxis with Zarzio®. The safety sample consisted of patients who received at least one dose of Zarzio®. The evaluable sample was limited to patients for whom at least initiation cycle data and follow-up/outcome data were available.

Data model

At enrollment: demographics, anthropometrics, medical and cancer history, prior cancer treatments, histology, history of (repeated) infections, and CIN/FN history.

At all visits: chemotherapy regimen, including changes; surgery, including changes; performance status; current and recent infections; CIN/FN episodes and associated hospitalizations; antibiotic prophylaxis; blood and urine cultures; Zarzio® prophylaxis; selected concomitant medications; adherence; hematology parameters; clinical events; and adverse drug reactions (ADRs).

At study end: performance status, current and recent infections, CIN/FN episodes and associated hospitalizations, blood and urine cultures, hematology parameters, clinical events, and ADRs.

Other indices

Prophylaxis intensity

The study followed the decision schematic in the 2010 EORTC guidelines for primary prophylaxis with GCSF, which were amended per ad hoc expert consensus to include trajectories for secondary prophylaxis. Patients’ were graded accordingly as under-prophylacted, correctly prophylacted, or over-prophylacted.

Patient risk score

This score quantifies the eight risk factors that, according to the EORTC guidelines, may increase FN risk. Each factor was weighted by expert consensus: 3 to age≥65 years and prior FN history; 1.5 to advanced disease and poor performance and/or nutritional status; and 0.5 to no antibiotic prophylaxis, female gender, hemoglobin <12 g/dL, and renal, cardiovascular, or liver disease. The sum of the weighted risk factors constituted the patient risk score (PRS) (range 0–11).

Outcomes

We report clinical outcomes using (a) patients and (b) chemotherapy cycles as the unit of analysis. Patient-level data enable analysis of the association of outcomes with prophylaxis type and chemotoxicity. Incidence refers to a given outcome “ever” experienced during the study. As patients enrolled at different cycles and were in the study for varying amounts of time, we also recorded outcomes at the cycle level. This permits analysis of the association with day of initiation and duration of prophylaxis. Cycle-level analysis evaluates the occurrence of an outcome in a given cycle or, in the case of chemotherapy disturbances, in a cycle subsequent to one in which a CIN/FN event was recorded (lag 1).

The outcomes of interest included the following: rate of CIN of any grade (CIN1/4), grades 3/4 combined (CIN3/4), grade 4 (CIN4), and FN; CIN/FN-related hospitalizations; CIN/FN-related chemotherapy disturbance one cycle after a CIN/FN event (lag 1); and a CIN/FN-related composite outcome that included any occurrence of CIN4, FN, CIN/FN-related hospitalizations, and/or CIN/FN-related chemotherapy disturbances.

Specialized statistical issues

The statistical dependence inherent to the structure of the cycle data being “nested” under patients was taken into account using generalized estimating equations. This procedure adjusts standard errors based on the observed within-cluster correlation.

We applied time-to-event modeling methods to estimate the probability of patients developing CIN4 or FN, being hospitalized, or experiencing chemotherapy disturbances due to CIN/FN in later cycles. For patients receiving primary prophylaxis, we estimated the probability of any one of these outcomes occurring in the cycle in which Zarzio® was initiated and the remaining five chemotherapy cycles. For patients receiving secondary prophylaxis in cycle 2, we estimated the probability of any of these outcomes occurring in cycles 2 through 6. The secondary prophylaxis exercise was limited to patients prophylacted in cycle 2 because too few patients were initiated in cycle 3 or later. In both exercises, chemotherapy disturbances were not estimated for the initiation cycle as such disturbances always occur with a cycle lag = 1.

Human rights

This study was approved by the ethical review committees of participating centers in accordance with national laws and regulations. Patients provided written informed consent.

Results

Patients

In total, 1,496 patients were enrolled yielding an evaluable sample of 1,447 patients (Fig. 1a). The majority of patients (72.3 %) were enrolled in cycle 1 (Fig. 1b). The sample was predominantly female (61.2 %) (Table 1). Median age was 62 years. Most patients (89.1 %) had an ECOG ≤1 score. Safety-relevant history included musculoskeletal pain (15.5 %), headaches (1.9 %), and bleeding (1.2 %).

Fig. 1
figure 1

a Subject enrollment and b follow-up

Full size image
Table 1 Demographics, clinical status, and cancer and CIN/FN history at enrollment
Full size table

Most patients (77.2 %) had a solid tumor, mainly breast (32.2 %) or lung (23.8 %) cancer. The most prevalent hematological malignancy was lymphoma (16.9 %). Proportions of stage 3 versus 4 disease varied across tumor types, with a majority of oncological patients (59.1 %) having stage 4 but equal distribution of either stage in hematological patients. One third of patients (35.8 %) were cancer-treatment-naive; one third (31.8 %) had received at least one line of chemotherapy, while one third (32.6 %) had undergone surgery.

Of the 460 patients with prior chemotherapy, 106 (23.0 %) had experienced ≥1 CIN4 episodes, including 27 (5.9 %) with episodes classified FN. Only 55 (12.0 %) had received GCSF treatment. Thirty-three (7.2 %) required hospitalization, and48 (10.4 %) experienced chemotherapy disturbances.

In the 10–20 % subsample, there were proportionately more patients with the risk factor of age ≥65 (p = 0.043), advanced disease (p < 0.001), and no antibiotic prophylaxis (p = 0.019) than in the rest of the sample. The PRS mean (±SD) was 2.85 ± 1.96, with patients receiving secondary prophylaxis having a higher PRS (p = 0.007). About equal proportions of patients were being treated with regimens with FN risk >20 % (44.3 %) or 10–20 % (45.0 %).

Prophylaxis

Zarzio® was initiated as primary prophylaxis in 72.3 % and as secondary prophylaxis in 27.7 % of patients (Table 2). Relative to the EORTC guidelines, 56.6 % of patients were correctly prophylacted, 17.4 % under-prophylacted, and 26.0 % over-prophylacted (Fig. 2). Under-prophylaxis occurred when secondary prophylaxis was given, but primary prophylaxis was recommended by the guidelines, whereas over-prophylaxis occurred when primary prophylaxis was given when not recommended by the guidelines.

Table 2 Zarzio® prophylaxis patterns
Full size table
Fig. 2
figure 2

Treatment decision relative to EORTC guidelines

Full size image

Proportionately more patients were administered a dose of 30 MIU (p < 0.001). This was not a function of type of prophylaxis (p = n.s.) but of body weight (p < 0.001), chemotoxicity (p = 0.007), and tumor type (p < 0.001).

In half of patients (53.2 %), Zarzio® was initiated in days 1–3 following chemotherapy (M ± SD = 3.08 ± 2.98), yet 13.4 % were exposed on the day of, and the remaining 33.4 % ≥4 days following, chemotherapy. Half (52.9 %) of the hematological patients were initiated on days 4–8. Day of GCSF initiation tended to be earlier for patients with solid tumors (p < 0.001), receiving secondary prophylaxis (p = 0.010), or on chemotherapy regimens with low or moderate FN risk (p < 0.001).

The modal duration of prophylaxis was 5 days (45.7 % of patients); 72.7 % received Zarzio® for 4–8 days. Duration was longer with higher chemotoxicity (p < 0.001) but was independent of prophylaxis or tumor type (both p = n.s.).

Clinical outcomes

Patient level

In total, 504 (34.8 %) patients experienced one or more (ever) CIN1/4 episodes (Table 3). Rates were higher among patients receiving secondary prophylaxis (p < 0.001) but were independent of chemotherapy regimen. The CIN3/4 and CIN4 rates were 22.9 and 13.2 %, respectively, and were independent of prophylaxis type (both p = n.s.) but rose with chemotoxicity (both p < 0.03). The FN incidence was 5.9 %, was independent of prophylaxis type but was associated with chemotoxicity (p < 0.001).

Table 3 Clinical outcomes at the patient and cycle levels
Full size table

Eighty-eight patients (6.1 %) were hospitalized. Hospitalization was independent of prophylaxis type and chemotoxicity (both p = n.s.). The chemotherapy regimen of 138 patients (9.5 %) was disturbed, and this more so among secondary prophylaxis patients (both p < 0.001). Chemotherapy disturbances were independent of chemotoxicity.

Three hundred twenty-three patients (22.3 %) scored positive on the CIN/FN-related composite outcome. This was independent of prophylaxis type (p = n.s.), but associated with chemotoxicity (p = 0.043).

Table 4 summarizes, for patients who received primary prophylaxis, the number who experienced each event; the associated cumulative probabilities derived from time-to-event analyses as patients progressed through chemotherapy cycles; and, as an index of variability, the difference in probabilities between the highest and lowest estimates in a series. Similar data are presented for patients on secondary prophylaxis, but analysis was limited to patients initiated on Zarzio® in cycle 2, as statistically too few patients received secondary prophylaxis at later time points. The frequencies indicate the number of patients who experienced one of the four outcomes while the associated probabilities were derived from the time-to-event modeling calculations.

Table 4 Observed outcomes over the course chemotherapy cycles for patients receiving primary (Zarzio® initiated in cycle 1) and secondary (Zarzio® initiated in cycle 2 only) prophylaxis and associated probabilities
Full size table

Among patients receiving primary prophylaxis, the most infrequently expected event over up to six cycles of chemotherapy was hospitalization in cycle 1 with a modeled occurrence of 2 % of patients. The most frequently observed outcome was a CIN4 episode at 16 % over six cycles. Among patients on secondary prophylaxis, the most infrequently expected event over up to six cycles of chemotherapy was an FN episode in the initiation cycle (cycle 2) at 4 % of patients, while the most frequent event modeled, just as in primary prophylaxis, was CIN4 over six cycles at 16 %.

Cycle level

CIN1/4, CIN3/4, CIN4, and FN episodes were recorded in, respectively, 14.3, 8.0, 3.9, and 1.4 % of cycles (Table 3). These rates were higher for prophylaxis initiated >72 h following chemotherapy (all p < 0.03) and for prophylaxis of ≥6-day duration (all p < 0.02).

CIN/FN-related hospitalizations occurred in 1.5 % of cycles but was independent of initiation or duration of prophylaxis (both p = n.s.). In total, 174 (2.8 %) chemotherapy cycles were disturbed, which was independent of day of initiation (p = n.s.) but higher for prophylaxis lasting ≥6 days (p < 0.001). Chemotherapy delay was the most frequent disturbance (2.3 %) and was associated with prophylaxis initiation (p < 0.001) and duration (p = 0.001).

In total, 507 cycles (6.7 %) were positive on the CIN/FN-related composite outcome. The rate was higher for cycles in which prophylaxis was initiated >72 h (p < 0.001) or lasted ≥6 days (p < 0.001).

Safety

In the safety sample (1,496 patients with 6,392 cycles), 4271events were observed in 777 (53.7 %) patients (Table 5). Bone pain was the most prevalent (24.7 %) and was mostly mild to moderate. Five hundred twenty patients (35.9 %) experienced more than one event. Sixty-one patients died, mainly due to cancer (67.2 %). There were 148 ADRs reported (2.3 % of cycles) in 76 (5.1 %) patients. Most ADRs were mild or moderate (83.8 %) and resolved completely (95.9 %). The most frequent were bone pain (23.0 %), arthralgia (14.2 %), myalgia (7.4 %), diarrhea (6.8 %), back pain (4.7 %), and rash (4.7 %). Four serious ADRs (2.7 % of ADRs in 4 or 0.3 % of patients) were reported, these being bone pain, drug hypersensitivity, vulval abscess, and loss of consciousness. There were no neutropenia-related and no Zarzio®-related deaths.

Table 5 Tolerability and safety
Full size table

Discussion

In this European study of patients with stage 3 or 4 cancer, treated with chemotherapy regimens with varying degrees of FN risk and receiving primary or secondary prophylaxis with Zarzio®, the rates of CIN4 and FN episodes, associated hospitalizations and chemotherapy disturbances, adverse events, and ADRs were statistically within the range of rates reported historically [16, 28, 29]. This documents the effectiveness and safety of Zarzio® in daily practice and extends the efficacy and safety data from its clinical development program [19–21]. These effectiveness and safety findings should address prescribers’ concerns about biosimilars [30]. More importantly, this study revealed the impact of significant variability in prophylaxis patterns, often divergent from the EORTC guidelines for GCSFs. More than a decade of clinical experience with filgrastim has translated into practice patterns that differ from the normative filgrastim regimen in randomized controlled trials; attempt to balance chemotoxicity-associated CIN/FN risk, risk factors, patient safety, and clinician experience; and aim to achieve therapeutic efficiency.

Zarzio® was used mainly for 5 days of primary prophylaxis at a dose of 30 MIU, which tended to be increased to 48 MIU in function of weight, chemotoxicity, and tumor type. Almost six out of ten oncology patients were initiated in the EORTC-recommended time window of 24–72 h post-chemotherapy compared to one third of hematology patients. The latter tended to be started on prophylaxis between days 5 and 8. Patients on a regimen with<10 % risk or on secondary prophylaxis tended to be prophylacted sooner. The modal and median durations of prophylaxis were 5 days, considerably shorter than the 10–11 days in trials, consistent with early findings from claim databases [31], and thus reflective of well-established CIN/FN prophylaxis patterns. Over a quarter of patients received secondary prophylaxis, which is essential in patients with a neutropenic event in a previous cycle. Yet, this relatively high proportion, coupled with the 17.4 % under-prophylaxis data, suggests that, still, too many patients received secondary prophylaxis when primary prophylaxis was indicated. The low rate of antibiotic prophylaxis is likely due to the limited number of hematological malignancy patients, who have been shown to benefit from such prophylaxis [32].

Significant proportions of patients were under- (17.4 %) or over-prophylacted (26.0 %). Under-prophylaxis concerned patients on high FN risk chemotherapy or patients on moderate FN risk regimens but with FN risk factors. This is inconsistent with established evidence, further compounded by the increasingly shorter duration of prophylaxis. The over-prophylaxis involved patients on low-chemotoxicity regimens or patients on moderate-risk regimens but without risk factors. This may signal a real-world trend for clinicians to “over-protect” and reflect a change in practice that heretofore has not been described in real-world studies. If confirmed in future studies, the over-prophylaxis pattern may result in modifications of existing prophylaxis paradigms. The under-prophylaxis patterns are a reminder that, despite risk stratification and evidence-based guidelines, a significant proportion of patients may receive inadequate CIN/FIN prophylaxis.

One in seven patients received their first dose of Zarzio® on the day of chemotherapy. While the evidence is limited, derived mainly from non-controlled studies of pegfilgrastim, and often contradictory in findings, NCCN guidelines now cautiously acknowledge that “same-day administration may be considered in certain circumstances” [15]. The cycle-level analyses showed that same-day initiation was not associated with higher CIN rates.

In both the patient-level and cycle-level analyses, type and duration of prophylaxis, day of initiation, and FN risk were found to impact the CIN1/4, CIN3/4, CIN4, and FN rates; CIN/FN-related hospitalizations and chemotherapy disruptions; and the composite outcome. This is consistent with prior evidence [4, 5, 16, 24].

Seemingly paradoxical, higher rates of CIN, FN, and CIN/FN-related outcomes were found in cycles in which Zarzio® was administered for ≥6 days. However, duration was longer in patients on regimens with FN risk >20 % and presenting with additional risk factors. This may reflect clinician vigilance about chemotoxicity and risk factors and a goal of optimizing outcomes. Noteworthy, rates for chemotherapy disturbances, in general, and for chemotherapy delays specifically were at least twice as high in patients on secondary prophylaxis, thus exposing these patients to the risk of impaired tumor control.

The rates of musculoskeletal, hematological, elevated laboratory values, neurological, and splenomegaly events were consistent with the known safety profile for GCSFs. Likewise, ADR rates were consistent with known rates.

The use of biosimilar filgrastim offers significant cost savings over standard filgrastim (Neupogen®) or pegfilgrastim (Neulasta®). In fact, an economic analysis for the European Union (EU) G5 countries comparing 1 to 14 days of prophylaxis with Zarzio® versus 1 to 14 days with Neupogen® or a singular injection with Neulasta® showed, that under all scenarios, Zarzio® was the most cost-efficient approach to prophylaxis [33]. Further, savings achieved with conversion to biosimilars can be applied, in a budget-neutral way, to purchase additional curative anticancer therapy, and this expands patient access to such treatments as rituximab for diffuse large B cell non-Hodgkin’s lymphoma and trastuzumab for metastatic HER2-positive breast cancer. For the EU G5 countries, it has been estimated that, depending on regimen, such expanded access to rituximab can be achieved by converting between 4 and 14 patients from Neupogen® and between 2 and 24 patients from Neulasta® to Zarzio®. Likewise, expanded access to trastuzumab would require converting between 4 and 14 patients from Neupogen® and between 2 and 24 patients from Neulasta® [34].

Our study has limitations. It was not a population-based study, was limited to centers using Zarzio®, and included only patients prophylacted with this biosimilar. The investigators selected patients, and this may be a source of selection bias. Despite strict inclusion and exclusion criteria, most patients had no or minimal impairment in performance status, and about a quarter had experienced a CIN or FN episode in a prior chemotherapy line or cycle. It is possible that investigators showed preference for less-impaired patients with better prognosis or patients with a prior CIN/FN history and therefore candidates for prophylaxis—some of whom may have been over-prophylacted relative to the EORTC guidelines. The design did not permit calculation of the relative dose intensity (RDI) of chemotherapy regimens, which has been shown to be associated with all-cause mortality [4]. Further analysis is needed to identify determinants of the CIN/FN incidence, hospitalization, and chemotherapy disturbance rates and to compare specific subgroups such as oncological vs. hematological patients; elderly vs. non-elderly patients; under-, correctly, and over-prophylacted patients; and patients with same-day vs. 24–72-h vs. >72-h initiation of prophylaxis.

Conclusion

The clinical and safety outcomes of prophylaxis with biosimilar filgrastim are within the range of historically reported data for originator filgrastim. This underscores not only the clinical effectiveness and safety of biosimilar filgrastim prophylaxis in daily clinical practice, but also the need to improve primary prophylaxis rates in patients with chemotherapy-related and patient-related risk factors for FN.