Febrile neutropenia (FN) occurrence outside of clinical trials: occurrence and predictive factors in adult patients treated with chemotherapy and an expected moderate FN risk. Rationale and design of a real-world prospective, observational, multinational study

Neutropenia is a major cause of infection-related morbidity and mortality in patients treated with myelosuppressive chemotherapy regimens [1, 2]. Febrile neutropenia (FN) is the most serious manifestation of neutropenia and a key driver of chemotherapy dose delays and/or reductions, which may impact treatment efficacy [2, 3]. The development of FN often leads to increased treatment costs and longer hospital stays, and may also be associated with reduced quality of life (QoL) [1].

FN occurs frequently during chemotherapy. In a retrospective cohort study, FN occurred in 13% to 21% of patients receiving common myelosuppressive chemotherapy regimens for metastatic solid tumours, most frequently during the first cycle (23% to 36%) [4]. Granulocyte colony-stimulating factors (G-CSFs) can be used prophylactically to reduce the risk, severity and duration of FN, and as an adjunct to support the delivery of dose-dense (increased frequency) or dose-intense (increased dose) myelosuppressive regimens [2, 5]. The use of G-CSFs as primary prophylaxis for the prevention of FN has been shown to reduce the relative risks of FN by 46% on average across 15 studies in a systematic review (> 90% in a study of pegfilgrastim), infection-related mortality by 45% and early deaths by 40% [6]. In a study of G-CSF (filgrastim) in patients with early breast cancer treated with epirubicin-cyclophosphamide chemotherapy, the rate of FN was 1% in the G-CSF arms compared with 7% in the control arms (P = .004) [7].

Current evidence-based guidelines from the European Society for Medical Oncology (ESMO), American Society of Clinical Oncology (ASCO), and National Comprehensive Cancer Network (NCCN) recommend prophylaxis with G-CSF for patients treated with chemotherapy with an FN risk ≥20%, and for patients receiving chemotherapy with an FN risk of 10% to 20% if they also present with risk factors including age ≥ 65 years, poor performance status and prior FN [2, 5, 8]. Prophylaxis is not recommended for most patients receiving chemotherapy with an FN risk < 10% [2, 5, 8]; however, in absolute numbers, many patients with an estimated risk of < 20% develop FN with rates of complications similar to those of patients with a high risk of FN development [9]. There is evidence to suggest that patients at lower risk of FN may respond better to G-CSF and that reduced doses of G-CSF might be effective in these patients [8], which would make prophylaxis more cost-effective. In the systematic review of 15 studies, FN risk was reduced with G-CSF, compared with controls across baseline risks from 17 to 78% of FN in control patients, although the trend for greater benefits of G-CSF at lower FN risk levels was no longer significant when the pegfilgrastim or the two studies with baseline risk around 20% were excluded [6]. Supporting data comes mainly from clinical trials and few studies have investigated the occurrence of FN in lower risk patients in real-life settings. Moreover, high-risk sub-populations, such as the elderly or those with comorbidities, are often underrepresented in randomised controlled trials (RCTs), despite convincing evidence that older age and comorbidities increase the risk of developing FN during chemotherapy [10‐12].

This real-world study will include patients who are generally excluded from RCTs and analyses will be performed on data stratified by age (i.e. under and over 65 years of age) to provide a greater understanding of FN risk in elderly patients receiving myelosuppressive chemotherapy regimens with an expected moderate (10% to 20%) FN risk. The risk group of FN between 10 to 20% is very heterogeneous. This study will characterise the real risk of FN with the most common chemotherapy regimens used in clinical practice. It will also identify predictive factors associated with FN in this patient population and will develop a risk model assessment for this group of patients.

Clinicians frequently use data from RCTs to estimate the FN risk associated with a particular chemotherapy regimen [10, 14]; however, reported rates vary considerably for the same chemotherapy regimen. This variation may relate to differences in study populations; furthermore, the patient populations in these trials are often highly selected and may not be representative of the majority of patients treated in the general cancer population [10, 14]. It is also of note that guideline recommendations are generally based on results of RCTs and, as such, may not reflect daily clinical practice.

Evidence suggests that FN rates in the real-world setting may be higher than those reported in RCTs. In a systematic review and meta-analysis of 65 observational (n = 7812 patients) and 110 RCT (n = 42,257 patients) cohorts involving 29 breast cancer chemotherapy regimens, the unadjusted FN rate was 11.7% and 7.9% in the observational and RCT cohorts, respectively. FN rates remained significantly higher in the observational cohorts (P = 0.012) after adjusting for age, chemotherapy intent and regimen [14].

In the PACS 01 trial, which compared 6 cycles of fluorouracil, epirubicin and cyclophosphamide (FEC) with a sequential regimen of 3 cycles of FEC followed by 3 cycles of fluorouracil, epirubicin, and docetaxel (FEC-D) as adjuvant treatment for women with node-positive early breast cancer, the FN rate was 11.2% in the FEC-D arm, which did not meet the 20% risk threshold [15]; however, a meta-analysis of data from 9 studies conducted in routine clinical practice showed that patients treated with adjuvant FEC-D without G-CSF primary prophylaxis had an FN rate of 31%, exceeding the threshold [16].

Real-world data from patients aged < 65 years in Belgium showed the overall incidence of FN with chemotherapy regimens carrying a moderate or high risk of FN for the treatment of breast cancer and Hodgkin lymphoma was higher than expected, based on those reported in the literature from clinical trials [17].

Considering the available real-world evidence for the incidence of FN in patient populations undergoing chemotherapy with a FN risk of 10% to 20%, data are limited but a few observational studies report data from different patient populations.

A retrospective analysis was performed to determine the incidence of FN in 466 Japanese patients with non-Hodgkin lymphoma treated with R-CHOP [18]. Without G-CSF support, the incidence of FN was 9.1% in cycle 1 and 12.3% throughout all cycles. Of the patients with FN, 73.7% developed FN during cycle 1. Risk factors associated with the development of FN included albumin < 35 g/L, relative dose intensity < 85% and lack of G-CSF prophylaxis [18]. These findings suggest that, in this population, patients with these risk factors may benefit from G-CSF prophylaxis from cycle 1. Data from an observational study of 1829 European (97%) and Australian (3%) patients with non-Hodgkin lymphoma receiving CHOP (±rituximab) without G-CSF support showed that 18% of patients experienced at least one FN event [19]. Patients developing FN tended to be older (by about 4 years) and a higher percentage of them had advanced disease and a predicted risk of FN of ≥20% compared with matched controls.

The difference in FN incidence (12.3% vs. 18%) between these two non-Hodgkin lymphoma populations may be due to the ethnicity of the patients and is worthy of further investigation. Nonetheless, it should also be considered that not all patients in the European/Australian study received rituximab.

A retrospective study of 610 Korean women with early-stage breast cancer receiving adjuvant chemotherapy with AC (doxorubicin, 60 mg/m² and cyclophosphamide, 600 mg/m² every 21 days) reported FN in 8.5% of patients [20]. Predictors for FN included the presence of grade 4 neutropenia and a pre-treatment calculated estimated glomerular filtration rate less than 60 ml/min [20].

These real-world studies show differences in FN incidence between different patient populations and highlight that predictors of FN can be identified that may help advise on which patients may benefit most from G-CSF prophylaxis. Differences in FN between patients of different ethnicity is worthy of further investigation.

MONITOR-GCSF was an observational study in patients with solid or haematological malignancies receiving G-CSF support with Sandoz biosimilar filgrastim (EP2006/Zarzio^®) for myelosuppressive chemotherapy [21]. Data from the study were used in a predictive model to identify determinants of febrile neutropenia episodes [21]. Patients were more likely to experience an FN episode if they had Eastern Cooperative Oncology Group status ≥2 anytime during the study or received antibiotic prophylaxis. Notably, patients who were ‘under-prophylacted’ with G-CSF were also more likely to experience FN [22]. Rates of FN were consistently lower in over-prophylacted patients compared to those under- and correctly-prophylacted. Under-prophylacted patients were at higher risk for disturbances to their chemotherapy regimens [23].

Stratified analyses performed in elderly patients showed that the only predictive factor of an elderly patient experiencing an FN episode was receiving antibiotic prophylaxis [24].

Evidence suggests that the availability of G-CSF biosimilars has led to cost savings, compared with use of the reference product [25, 26]; indeed, since 2015, filgrastim has been included on the World Health Organisation (WHO) Model List of Essential Medicines, reflecting its cost-effectiveness [27]. Since cost-savings associated with biosimilars has implications for increasing access to G-CSF supportive care, it is important to characterise the subset of patients at 10% to 20% risk of neutropenia who are most likely to benefit from G-CSF prophylaxis. This is particularly of note since the 10% to 20% subset of patients are a very heterogeneous population.

Our study can build on the predictors identified in observational studies including MONITOR-GCSF and provide information on which patients with a 10% to 20% risk of neutropenia will be most likely to benefit from G-CSF prophylaxis. Since the study will include patients generally not included in clinical trials, the findings will likely differ to published findings from RCTs and may be practice-changing.