A prospective cohort study to evaluate the incidence of febrile neutropenia in patients receiving pegfilgrastim on-body injector versus other options for prophylaxis of febrile neutropenia: breast cancer subgroup analysis

Background Breast cancer chemotherapy often carries a high risk of febrile neutropenia (FN); guidelines recommend prophylaxis with granulocyte colony-stimulating factor (G-CSF), such as pegfilgrastim. Neulasta® Onpro® on-body injector (OBI) is a delivery device administering pegfilgrastim approximately 27 h after application. Methods This prospective study examined patients with breast cancer who received chemotherapy with a high risk of FN, receiving OBI (“OBI”) or other options (other G-CSF or none; “other”). The primary endpoint was FN incidence; secondary endpoints included chemotherapy delivery, adherence (G-CSF in all cycles), compliance (G-CSF day after chemotherapy), and FN incidence in patients receiving curative or palliative treatment. Results A total of 1776 patients with breast cancer were enrolled (OBI, n = 1196; other, n = 580). Across all cycles, FN incidence was lower for OBI (4.4% [95% CI, 3.3–5.6%]) than other (7.4% [5.3–9.6%]). For curative treatment, the FN incidence across all cycles was lower for OBI (4.6% [3.4–5.8%]) than for other (7.1% [5.0–9.3%]). For palliative treatment (OBI, n = 33; other, n = 20), 3 patients (15%) in the other and none in the OBI group had FN. After adjusting for baseline covariates, FN incidence remained lower for OBI (4.6% [3.5–6.1%]) versus other (7.8% [5.7–10.5%]). Adherence was higher for OBI (93.8%) than for other G-CSF (69.8%), as was compliance (90.5 and 53.2%, respectively). Chemotherapy dose delays/reductions were similar for OBI (4.7%/32.3%, respectively) and other (4.7%/30.0%) groups. Conclusion Pegfilgrastim OBI was associated with a lower FN incidence in patients with breast cancer compared to other options for FN prophylaxis. Trial registration www.clinicaltrials.gov, NCT02178475, registered 30 June, 2014


Introduction
Febrile neutropenia (FN) is a potentially life-threatening complication that can occur in patients with breast cancer who are receiving myelosuppressive chemotherapy and is associated with significant morbidity, mortality, healthcare resource utilization, and costs [1,2]. In a recent analysis, the incidence of FN in the USA was as high as 20.6% in patients with breast cancer receiving chemotherapy without prophylaxis for FN [3]. According to real-world data from the USA, the rate of hospitalization for FN in patients with breast cancer was 13.9%, and the mortality rate in hospitalized patients with FN was estimated to be 2.0-2.6%; the mean length of stay ranged from 4.1 to 5.7 days and mean hospital costs ranged from $16,940 to $37,087 USD [1,2]. Among patients with metastatic breast cancer, resource utilization and mortality rate were even higher: FN-associated hospitalization rate was 89%, and the mortality rate in hospitalized patients was 7.3% [4]. Additionally, FN may necessitate chemotherapy dose delays or reductions, which are significantly associated with increased mortality [5,6]. Among patients with breast cancer receiving standard myelosuppressive chemotherapy, up to 60% and up to 96% experienced dose delays and dose reductions, respectively [1,7,8].
The risk of FN varies based on chemotherapy regimen and comorbidities, and National Comprehensive Cancer Network (NCCN) guidelines recommend prophylaxis with granulocyte colony-stimulating factors (G-CSFs) in patients receiving chemotherapy regimens with a high risk (>20%) of FN and intermediate-risk (10-20%) of FN with ≥1 risk factor for FN [9,10]. The NCCN recommends following the FDA-approved dosing schedule, which is G-CSF prophylaxis administered the day following chemotherapy [9,10]. Pegfilgrastim is a long-acting G-CSF that has been shown to reduce the risk of FN by 94% with first-cycle use in patients with breast cancer receiving docetaxel 100 mg/m 2 every 3 weeks [9,11]. The Neulasta ® Onpro ® on-body injector (OBI) is a delivery device that administers pegfilgrastim 6 mg approximately 27 h after application in accordance with NCCN guidelines and FDA recommendations, thus eliminating the need for an office visit the day after chemotherapy administration [12]. In a retrospective study of twenty-eight patients who received chemotherapy and pegfilgrastim OBI between 2016 and 2018, there were no hospitalizations due to FN or chemotherapy dose delays or reductions; however, there was no comparator arm in this study, necessitating additional research [13].
Chemotherapy that carries a high risk of febrile neutropenia is commonly used in breast cancer [14]; however, there is limited evidence of clinical benefits of adherence and compliance to OBI in this patient population. Here, we report analyses of the subgroup of patients with breast cancer with the objective to estimate the incidence of FN in patients with breast cancer who received myelosuppressive chemotherapy and G-CSF prophylaxis with pegfilgrastim OBI or other options; chemotherapy delivery, adherence, and compliance were also evaluated.

Study design and population
The primary analysis of this multicenter, prospective, observational cohort study in patients receiving myelosuppressive chemotherapy and at high risk for developing FN between November 7, 2018, and April 9, 2020, was previously reported [15]. Briefly, in the primary analysis adults were eligible if they had breast, lung, or prostate cancer or NHL, a life expectancy of >6 months, ≥4 planned chemotherapy cycles administered every 3 or 4 weeks, and received a chemotherapy regimen with high FN risk (>20%) or intermediate FN risk (10−20%) and ≥1 risk factor for FN as defined by NCCN guidelines at that time [10]. In this analysis, only patients with breast cancer were included. Patients who received radiation <2 weeks before enrollment or had planned chemotherapy dose reduction for cycle 1, concurrent primary cancers (except non-melanoma skin cancer or adequately treated carcinoma in situ), or significant laboratory abnormalities per the investigator were excluded. Demographic and clinical characteristics, including tumor type, chemotherapy regimen, type of G-CSF prophylaxis received and timing, age, sex, laboratory measurements, comorbidities, and history of other malignancies, were collected. Patients were followed from study enrollment until death, discontinuation of chemotherapy, withdrawal of consent, lose to follow-up, or end of the study.

Endpoints
The primary endpoint was the incidence of FN during the study period. FN was defined as an absolute neutrophil count (ANC) <1000 × 10 6 /L and occurrence of 1 of the following within 24 h of decreased ANC: temperature >38 °C, use of oral antibiotics (i.e., ciprofloxacin, levofloxacin, moxifloxacin, or amoxicillin-clavulanate), or any intravenous antibiotics. Secondary endpoints included the incidence of FN in patients who received treatment with curative or palliative intent, chemotherapy delivery, adherence, and compliance. Adherence was defined as G-CSF support received in all chemotherapy cycles irrespective of G-CSF administration timing. G-CSF support was defined as the administration of a long-acting G-CSF (pegfilgrastim OBI, pegfilgrastim, or biosimilar pegfilgrastim [pegfilgrastim-jmdb, pegfilgrastim cbqv, pegfilgrastim-bmez]) or ≥10 administrations of a short-acting G-CSF (filgrastim, biosimilar filgrastim [filgrastim-sndz or filgrastim-aafi], tbo-filgrastim, or sargramostim) in each cycle. Compliance with pegfilgrastim was defined as pegfilgrastim OBI, pegfilgrastim, or biosimilar pegfilgrastim administered the day after chemotherapy completion in all cycles in which pegfilgrastim was administered.

Statistical analysis
Patients with breast cancer were categorized by the type of G-CSF prophylaxis received in the first cycle of chemotherapy: pegfilgrastim OBI (OBI) or other option (other), which included pegfilgrastim pre-filled syringe (PFS), pegfilgrastim biosimilar PFS, filgrastim, tbofilgrastim, filgrastim-sndz, or no G-CSF prophylaxis selected at the physician's discretion. Patients remained in the originally-assigned group even if a different type of G-CSF prophylaxis was administered in a subsequent cycle. The incidence of FN was calculated as the percentage of patients who experienced FN, and 95% confidence intervals (CIs) were calculated using a normal approximation method. As reported in the primary analysis, the incidence of FN was adjusted using a standardized logbinomial model to account for confounding variables [15]. These included prior surgery within 6 months before study enrollment, antibiotic use 0-7 days before initiation of chemotherapy, and FN risk of the chemotherapy regimen. Adjusted incidences of FN, relative risks, and associated p-values were calculated using a standardized log-binomial model in which weight was assigned to each patient equal to the inverse probability of exposure conditional on that patient's confounder information [16,17]. Associated 95% CIs for adjusted incidences and relative risk were calculated using bootstrap methods. Patient characteristics, chemotherapy delivery, adherence, and compliance were summarized using descriptive analyses. Missing data were not imputed.
f One patient received 4 cycles of bendamustine and rituximab for follicular lymphoma that was incorrectly coded as tumor type "breast" by the site and was included in the analysis. g One patient received carboplatin and paclitaxel, which is not a common chemotherapy regimen for breast cancer. We were unable to obtain clarification from the site regarding the use of this regimen. Among patients in the other group during cycle 1, 73.6 and 7.8% of patients received long-and short-acting G-CSF, respectively, and 18.6% received no G-CSF support.

Incidence of febrile neutropenia
Across all cycles, the incidence of FN was lower in patients who received pegfilgrastim OBI (4.4% [95% CI, 3.3-5.6%]) compared with patients who received other options (7.4%   Fig. 1a). The OBI group had a lower incidence of FN at each cycle. Similar trends were observed in patients who received pegfilgrastim OBI in every cycle; across all cycles, the incidence of FN in patients who received pegfilgrastim OBI in every cycle was 4.0% (95% CI, 2.8-5.2%; Fig. 1b). After adjusting for covariates (prior surgery within 6 months prior to study enrollment, antibiotic use 0-7 days prior to initiation of chemotherapy, and FN risk of chemotherapy regimen), the estimated incidence of FN was 4.6% (95% CI, 3.5-6.1%) for the OBI group compared with 7.8% (95% CI, 5.7-10.5%) for the other group. The risk of developing FN was significantly reduced for the OBI group versus the other group (RR, 0.60 [95% CI, 0.40-0.88]; P = 0.011; Fig. 1c 20), FN was only observed during cycle 1 in 3 patients (15%) in the other group (Fig. 3). Two of these patients received docetaxel with no G-CSF in cycle 1; the other patient received docetaxel plus cyclophosphamide plus biosimilar pegfilgrastim in cycle 1.

Chemotherapy delivery
The percentage of patients who required chemotherapy dose delays or reductions was comparable for patients who received pegfilgrastim OBI or other options. Across all cycles, the percentage of patients with chemotherapy dose delays was 4.7% (95% CI, 3.5-5.9%) for the OBI group and 4.7% (95% CI, 2.9-6.4%) for the other group; at each cycle, the percentage of patients with chemotherapy dose delays was similar between groups. Across all cycles, the percentage of patients

Adherence and compliance
Adherence to G-CSF (1 long-acting G-CSF or 10 shortacting G-CSF per chemotherapy cycle, regardless of timing) was higher in patients who received pegfilgrastim OBI  Fig. 4b).

Discussion
This sub-analysis of a prospective observational study evaluated clinical outcomes in patients with breast cancer who received myelosuppressive chemotherapy and G-CSF support as pegfilgrastim OBI or other options for prophylaxis of FN. The overall incidence of FN was lower in patients who received pegfilgrastim OBI in the first chemotherapy cycle compared with other options, regardless of the use of pegfilgrastim OBI in subsequent cycles. The risk of developing FN was reduced by 40% with pegfilgrastim OBI compared with other options. The incidence of dose delays and reductions was similar among patients who received pegfilgrastim OBI or other options. Adherence with G-CSF prophylaxis and compliance to pegfilgrastim were higher in patients who received pegfilgrastim OBI compared with patients who received other options.
Our findings are consistent with retrospective studies of patients receiving myelosuppressive chemotherapy for non-metastatic solid tumors or NHL. In a study of patients with non-metastatic breast cancer or NHL receiving high or intermediate FN risk chemotherapy regimens, the incidence of FN in the first cycle was 4-9% in patients who received G-CSF prophylaxis [18]. In another retrospective study of patients with non-metastatic solid tumors (including breast) or NHL receiving high or intermediate FN risk chemotherapy regimens, the incidence of FN in the first cycle ranged from 2.2-3.8% in patients who received G-CSF prophylaxis [19]. The incidence of FN reported in a retrospective study of patients with metastatic breast cancer was higher than observed in the present study; the overall incidence of FN was 15.8%, and the rate of G-CSF prophylaxis was 16.7% in this population [4]. However, treatment in the metastatic setting is typically palliative; the American Society of Clinical Oncology guidelines did not find the demonstrable benefit of G-CSF prophylaxis in patients with metastatic disease, which is consistent with the small number of patients who received G-CSF prophylaxis in the palliative intent subgroup of the present study [6]. In this study, the risk of FN in patients receiving "other" treatment with palliative intent appears to be twofold higher for the patients receiving "other" treatment with curative intent; however, the small patient numbers in these groups mean these data should be interpreted with caution. The decreased incidence of FN observed with pegfilgrastim OBI in this study may be associated with increased adherence and compliance in this group relative to patients who received other options [20]. Adherence was 34.4% higher, and compliance was 70.1% higher in patients who received pegfilgrastim OBI compared with patients who received other options.
In prospective and retrospective studies, receipt of pegfilgrastim at least 24 h after myelosuppressive chemotherapy administration resulted in improved adherence and patient outcomes [15,21,22]. Pegfilgrastim OBI allows for next-day administration without a clinic visit, therefore reducing the patient travel burden, reducing noncompliance because of the travel burden, and optimizing outcomes [23,24].

Limitations
The definition of FN used in this study uses an ANC that may be higher than that used to define FN in some clinical settings [25]. However, as the definition was used across both groups, it is not expected to affect the generalizability of the results. Selection bias may exist because of patients being lost to follow-up after enrollment (OBI group, n = 16; 1.3%; other group, n = 10; 1.7%), and therefore it would not be possible to evaluate the risk of FN among these patients. Additionally, these results may be confounded by indication because patients who were perceived to be at higher risk by the investigator were more likely to receive G-CSF prophylaxis. In the analysis of adherence, patients with a better prognosis or lower ECOG may have been prescribed fewer doses of short-acting G-CSF prophylaxis and, thus, may have been inappropriately classified as non-adherent. In addition, reasons for lack of adherence were not captured to the extent required for further analysis. Reasons for lack of compliance were not collected. Patients receiving same-day G-CSF could have been considered noncompliant; however, the percentage of patients receiving same-day G-CSF in the overall study population (~1%; n = 28/2575) was small enough that any difference in the incidence of FN in this subgroup would be unlikely to bias the study results. Factors associated with possible sources of confounding by indication were included in the standardized log-binomial regression, but bias Fig. 4 Adherence to G-CSF and compliance to pegfilgrastim in patients with breast cancer who received pegfilgrastim OBI or other options. a Adherence; percent plus 95% CI. b Compliance; percent plus 95% CI. G-CSF, granulocyte colony-stimulating factor; OBI, onbody injector; Other, other physician choice options can result from unmeasured and unknown risk factors. Finally, the study closed earlier than planned due to the COVID-19 pandemic, and the target sample size was not reached, thus it was not possible to compare patients with curative versus palliative treatment intent.

Conclusions
Patients with breast cancer at high risk of developing FN who received pegfilgrastim OBI had a lower incidence of FN compared with patients who received other options. The decreased incidence of FN in the OBI group may be a result of increased adherence and compliance to G-CSF prophylaxis compared with other options.
Conflict of interest Reshma L. Mahtani reports consulting fees from Biotheranostics, Amgen, Agendia, Eisai, Immunomedics, Genentech, Lilly, Novartis, Merck, Pfizer, Puma, Sanofi, Seagen, AstraZeneca, Daiichi, and participation on an advisory board for AstraZeneca, Amgen, Agendia, Biotheranostics, Eisai, Immunomedics, Lilly, Novartis, Merck, Pfizer, Puma, Sanofi, Seagen, and Daiichi; Rajesh Belani reports employment with Poseida Therapeutics and stock/stock options in Amgen; Jeffrey Crawford reports consulting fees from Glaxo-SmithKline, G1 Therapeutics, Merck, Pfizer, Spectrum, and participation on a data safety monitoring board or advisory board for Beyond Spring, G1 Therapeutics, Merrimack, Mylan and Roche; David Dale reports consulting fees from Amgen, and contracted research support from Amgen; Lucy DeCosta reports employment with Amgen Inc.; Prasad L. Gawade reports employment and stockholding with Amgen Inc.; Chanh Huynh has no disclosures to report; Tatiana Lawrence reports employment and stockholding with Amgen Inc.; Sandra Lewis reports employment and stockholding with Amgen Inc.; William W. MacLaughlin and Mohit Narang have no disclosures to report; Robert Rifkin reports participation in advisory boards for Amgen, BMS (Celgene), Coherus, Fresenius-Kabi, GSK, Janssen, and Pfizer.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.