Primary Prophylaxis With Biosimilar Filgrastim for Patients at Intermediate Risk for Febrile Neutropenia: A Cost-Effectiveness Analysis

PURPOSE:

Temporary COVID-19 guideline recommendations have recently been issued to expand the use of colony-stimulating factors in patients with cancer with intermediate to high risk for febrile neutropenia (FN). We evaluated the cost-effectiveness of primary prophylaxis (PP) with biosimilar filgrastim-sndz in patients with intermediate risk of FN compared with secondary prophylaxis (SP) over three different cancer types.

METHODS:

A Markov decision analytic model was constructed from the US payer perspective over a lifetime horizon to evaluate PP versus SP in patients with breast cancer, non–small-cell lung cancer (NSCLC), and non-Hodgkin lymphoma (NHL). Cost-effectiveness was evaluated over a range of willingness-to-pay thresholds for incremental cost per FN avoided, life year gained, and quality-adjusted life year (QALY) gained. Sensitivity analyses evaluated uncertainty.

RESULTS:

Compared with SP, PP provided an additional 0.102-0.144 LYs and 0.065-0.130 QALYs. The incremental cost-effectiveness ranged from $5,660 in US dollars (USD) to $20,806 USD per FN event avoided, $5,123 to $31,077 USD per life year gained, and $7,213 to $35,563 USD per QALY gained. Over 1,000 iterations, there were 73.6%, 99.4%, and 91.8% probabilities that PP was cost-effective at a willingness to pay of $50,000 USD per QALY gained for breast cancer, NSCLC, and NHL, respectively.

CONCLUSION:

PP with a biosimilar filgrastim (specifically filgrastim-sndz) is cost-effective in patients with intermediate risk for FN receiving curative chemotherapy regimens for breast cancer, NSCLC, and NHL. Expanding the use of colony-stimulating factors for patients may be valuable in reducing unnecessary health care visits for patients with cancer at risk of complications because of COVID-19 and should be considered for the indefinite future.

INTRODUCTION

Recent practice guidelines from ASCO and the National Comprehensive Cancer Network recommend hematopoietic colony-stimulating factor (CSF) prophylaxis to patients receiving chemotherapy regimens with a high risk (> 20%) of febrile neutropenia (FN). In patients at intermediate risk of developing FN (10%-20%), the decision to use CSFs as primary prophylaxis (PP) or secondary prophylaxis (SP) is usually made via an individualized risk assessment and patient-physician discussion. In the presence of no additional FN risk factors, practice guidelines recommend the use of SP, whereas PP may be considered if the patient has one or more risk factors.^1,2

However, the pandemic caused by SARS-CoV-2 has resulted in new considerations for cancer care and supportive care. Patients with cancer are a highly susceptible population at risk of transmission of SARS-CoV-2 and the potential consequences of the associated disease, COVID-19. High susceptibility of patients with cancer is primarily due to their frequent contact with the health care system, cancer- or treatment-related immunosuppression, and advanced age and comorbidities.³

Several single- and multicenter studies have described characteristics and outcomes in cancer patients with COVID-19. One cohort study found that of 928 patients with active cancer or history of cancer and COVID-19, 26% developed severe illness, 14% were admitted to the intensive care unit, and 12% required mechanical ventilation. The mortality rate within 30 days of COVID-19 diagnosis was 13%. Of 52 intensive care unit patients with 30-day follow-up data, 31% died.³ Similar findings were published in an earlier Chinese study.⁴ These outcomes emphasize the need to strategically coordinate the care of patients with cancer to minimize risk of infection with COVID-19.

Although cancer providers have been challenged with navigating infection prevention, staffing shortages, and resource limitations during the pandemic, postponing or delaying chemotherapy may not be in the best long-term interest of patients receiving treatment for curative intent.⁵ Sharing of best practices has been important to optimize clinical care while reducing risk of transmission among patients and lessening demand on hospitals. To that end, there has been renewed focus in further reducing the risk of FN for patients with cancer. Improving FN outcomes also aligns with value-based care efforts (eg, the Oncology Care Model [OCM]) that have occurred over the past few years.

Recently, ASCO and the National Comprehensive Cancer Network issued temporary recommendations for the use of CSFs in patients with cancer during the COVID-19 pandemic. Specifically, the threshold for the use of CSF prophylaxis was lowered from only high-risk patients (> 20% risk of FN) to intermediate- (10%-20% risk of FN) or high-risk patients.^6,7 This expansion of CSF prophylaxis is aimed at mitigating the risks associated with COVID-19 for patients with cancer while benefiting facilities by potentially increasing the number of beds available to treat patients of the pandemic.

Considering these new recommendations during the pandemic and the general trend toward value-based care in oncology, we compared the different CSF prophylaxis strategies from a clinical and economic perspective by assessing the cost-effectiveness of PP versus SP using a biosimilar filgrastim (specifically filgrastim-sndz) in patients with breast cancer, non–small-cell lung cancer (NSCLC), and non-Hodgkin lymphoma (NHL) receiving potentially curative chemotherapy.

METHODS

Model Structure

Building on previously published cost-effectiveness analyses in FN, a Markov cycle tree–based model was constructed in Microsoft Excel to evaluate the cost-effectiveness of PP versus SP with a biosimilar filgrastim (specifically filgrastim-sndz) from the US payer perspective.^8,9 The model evaluated CSF prophylaxis strategies for the most common intermediate-risk chemotherapy regimens in patients with breast cancer (adjuvant docetaxel), NSCLC (adjuvant carboplatin and paclitaxel), and NHL (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone [R-CHOP]). Because of the novelty of COVID-19, the rapidity of change, and the underlying complexities of its care patterns, characteristics surrounding the complications of infection with COVID-19 were not incorporated into the model.

The general model structure diagram for breast cancer, NSCLC, and NHL is presented in Appendix Figure A1, online only. The model structure for each analysis was adapted to the specific number of cycles for each cancer type. The first cycle of each regimen was represented as a decision tree with the two arms to pursue either a PP or SP strategy. Based on the risk of FN during the first cycle, patients either developed FN or completed the cycle without FN.

A Markov cycle tree was employed to represent the remainder of chemotherapy. Patients were tracked to determine their history of FN and repeated the Markov cycle until completing up to the maximum number of cycles. In each cycle, patients with a history of FN (v patients without a history of FN) experienced a higher FN risk based on a value cited in previously published economic evaluations. If FN occurred, patients were treated in an inpatient or an outpatient setting and either died from FN or survived to the next cycle of chemotherapy. All deaths during chemotherapy were assumed to be FN-related and to occur at the end of each cycle.

The postchemotherapy phase of the model followed 1-year Markov cycles. Initially, patients were stratified into two groups based on the risks of receiving a suboptimal chemotherapy dose with and without a history of FN. In accordance with previous clinical and cost-effectiveness research, patients with suboptimal chemotherapy delivery were at higher annual risk of cancer-related death. 20 years post-chemotherapy, mortality reverted to standard age- and sex-related rates. The primary outcomes of the analysis were incremental costs per FN event avoided, per life year (LY) gained, and per quality-adjusted life year (QALY) gained. Cost-effectiveness was assessed at the commonly cited willingness-to-pay (WTP) thresholds of $50,000 in US dollars (USD), $100,000 USD, and $150,000 USD, using a biosimilar filgrastim (specifically filgrastim-sndz) SP as the reference comparator.

Model Parameters

All model inputs are presented in Table 1. In each analysis, the age at which patients entered the cohort varied according to their cancer type. Similarly, the baseline FN risks for each cancer type were selected to reflect the real world and focus on intermediate risk. Baseline FN risk for breast cancer was derived from patients having at least one FN risk factor based on the premise that patients received doxorubicin plus cyclophosphamide before docetaxel.² Baseline FN risk for NSCLC was also based on patients having at least one risk factor given that more than 90% of patients with NSCLC receiving carboplatin plus paclitaxel have at least one risk factor.¹⁰ For NHL, no additional FN risk factors were the basis of the baseline risk, as patients receiving R-CHOP have a baseline FN risk of approximately 18% and additional risk factors would place patients above intermediate risk.¹¹

TABLE 1.

Model Parameters

As the risk for FN over the entire course of chemotherapy is greatest in the first cycle, baseline cycle–specific FN risk was calculated as the risk in patients without a history of FN and without CSF prophylaxis. This baseline risk was decreased in patients who received CSF prophylaxis. History of FN increased the likelihood of subsequent FN events. Patients who experienced FN events received treatment in the inpatient or outpatient setting and had an increased likelihood of receiving reduced doses of chemotherapy.

Biosimilar filgrastim costs were based on the average sales price of filgrastim-sndz as of July 2020.¹² Other costs included CSF administration, inpatient FN management, and outpatient FN management and were adjusted to 2020 USD.¹³ Chemotherapy costs were excluded from the analysis as they were assumed to be equivalent between patients receiving PP and SP. Similarly, the difference in chemotherapy costs for patients receiving low versus high relative dose intensity was assumed to be negligible. Postchemotherapy costs were assumed to be unaffected by prophylaxis strategy and were also excluded.

QALYs were calculated by applying utility weights to the estimated LYs. We accounted for quality of life during chemotherapy, during FN hospitalization, and after chemotherapy. The improvement in quality of life experienced 1 year after chemotherapy was assumed to remain constant until death. Mortality during chemotherapy was FN-related. Cancer-related mortality subsequently affected patients until 20 years after chemotherapy, after which standard US death rates applied.¹⁴

Sensitivity Analyses

Alternative parameter values were tested via a one-way sensitivity analysis to evaluate the impact of each parameter on the models’ outcomes. In addition, a probabilistic sensitivity analysis (PSA) accounted for joint uncertainty among all model parameters and assessed the likelihood of cost-effectiveness of PP over a range of WTP thresholds. The PSA simulated 1,000 iterations, with parameter values sampled simultaneously from their individual distributions.

RESULTS

The base-case results for the breast cancer, NSCLC, and NHL analyses are presented in Table 2. Over all three cancer types, biosimilar filgrastim (using filgrastim-sndz) as PP versus SP provided an additional 0.102-0.144 LYs and 0.065-0.130 QALYs at an incremental cost ranging from $650 to $2,463 USD. The incremental cost-effectiveness ratios (ICERs) ranged from $5,660 to $20,806 USD per FN event avoided, $5,123 to $31,077 USD per LY gained, and $7,213 to $35,563 USD per QALY gained, with NSCLC reflecting the lowest ICERs.

TABLE 2.

Base-Case and PSA Results by Cancer Type

According to the one-way sensitivity analysis results on cost per QALY gained, for breast cancer, variation in baseline FN risk, mortality hazard ratio for low relative dose intensity, and the relative risk of FN with filgrastim versus no CSF exerted the greatest influence over the results. For NSCLC and NHL, the most influential parameters were baseline FN risk, mean length of stay for hospitalization (SP), and the cost of an FN event requiring hospitalization.

When the baseline FN risk was adjusted to 10%-20% for the breast cancer model, the results ranged from $86,573 to $18,995 USD per QALY gained, respectively. When similar adjustments were made in the NSCLC model, the results ranged from $53,670 to $1,467 USD per QALY gained. Finally, for the NHL model, the results ranged from $67,238 to $12,884 USD per QALY gained when adjusting the baseline FN risk to 10% and 20%, respectively.

Adjusting the average sales price of filgrastim-sndz by 90%-110% of baseline had less of an impact on the model results than baseline FN risk. For breast cancer, the results varied from $30,056 to $41,070 USD cost per QALY gained, whereas for NSCLC and NHL, the results ranged from $3,250 to $11,177 USD and $14,593 to $23,349 USD, respectively.

The results of the three PSAs are presented as cost-effectiveness acceptability curves ranging from WTP thresholds per QALY gained of $0 to $150,000 USD (Fig 1). For breast cancer, the likelihood of cost-effectiveness at a WTP threshold of $50,000 USD per QALY gained was 73.6%. For NSCLC and NHL, the likelihood of cost-effectiveness at a WTP threshold of $50,000 USD per QALY gained was 99.4% and 91.8%, respectively.

FIG 1.

PSA cost-effectiveness acceptability curves. Cost-effectiveness acceptability curves illustrating the probability of PP with biosimilar filgrastim being cost-effective relative to SP across a range of WTP thresholds for cost per FN avoided, cost per LY gained, and cost per QALY gained. Curves are shown for (A) breast cancer, (B) NSCLC, and (C) NHL. FN, febrile neutropenia; LY, life year; NHL, non-Hodgkin lymphoma; NSCLC, non–small-cell lung cancer; PP, primary prophylaxis; PSA, probabilistic sensitivity analysis; QALY, quality-adjusted life year; SP, secondary prophylaxis; WTP, willingness to pay.

DISCUSSION

Based on our analysis, using a biosimilar filgrastim (specifically filgrastim-sndz) as PP is a cost-effective approach to avoid FN events, which reduces the need for patients to receive hospital or outpatient care. This is especially important for reducing transmission of SARS-CoV-2 among patients with cancer, who are highly susceptible to the complications of COVID-19.⁵ By mitigating the risk of FN in patients receiving chemotherapy with intermediate to high FN risk, the expanded use of CSF further contributes to efficient care management given facility, resource, and staffing labor constraints during the COVID-19 pandemic, while maximizing curative potential.⁵

As SARS-CoV-2 is expected to be endemic,¹⁵ the cost-effectiveness of biosimilar filgrastim in intermediate FN risk regimens raises the possibility that this should be a standing recommendation within practice guidelines. There is historical precedent for re-evaluating risk thresholds informing the use of PP against FN. In 2006, ASCO lowered the definition of high risk for FN, the risk at which PP is recommended, from 40% to 20%.^16,17 Although the recommendation was informed primarily by clinical efficacy data for CSF in patients with an FN risk of approximately 20%, an economic analysis published before the guideline update illustrated that the added costs of more widespread CSF use at the lower threshold were offset by reductions in hospitalization costs.¹⁸

The introduction of biosimilar CSFs has led to the reduction in drug prices for historically expensive therapies. The availability of biosimilar CSFs makes this cost-effectiveness analysis more relevant, as previously published US cost-effectiveness analyses focused only on the originator products and therefore do not reflect the present US market for CSFs.^8,19 The present analysis supports the use of CSFs in patients receiving intermediate-risk chemotherapy in addition to currently available clinical trial data that show benefit of PP for chemotherapy regimens in solid tumors and NHL.

Before COVID-19, utilization of CSF PP was relatively low in patients receiving chemotherapy regimens at intermediate risk of FN. For example, PP with either filgrastim or pegfilgrastim was provided to only about 20.7% of patients and SP to 45.7% of patients receiving R-CHOP-21.²⁰ These real-world prophylaxis rates suggest that a significant portion of patients are at risk for FN and subsequent hospitalization and increased mortality because of this potentially preventable adverse event.

Especially in the current environment where value for money is an urgent focus of providers, governments, and manufacturers, the expanded use of PP also has the potential to contribute to value-based care. In 2016, the Centers for Medicare & Medicaid Services launched the OCM, a system that incentivizes practice advancements and value-based care in oncology. OCM-participating practices must reduce drug costs and meet certain quality measures, including reducing emergency department visits that do not lead to hospital admission.²¹ Expanding the PP use of CSF has the potential to reduce emergency department visits and hospital admissions. As the OCM transitions toward the Oncology Care First Model, it will continue to focus on the concept of value-based care by expanding on the enhanced services provided to beneficiaries. Oncology Care First Model is seen as a likely intermediate step toward an oncology bundled payment structure where reimbursement for all services during an episode will be based on a prospective payment system.²² Under this model, practices will have more incentive to reduce their drug costs and provide the most cost-effective therapies to their patients.

Our study has some limitations. First, in the absence of data for each cancer type, some inputs were based on studies that examined patients with different cancer types. However, we believe the data used in this analysis represent reasonable and conservative estimates of reality. The structure of the analyses likely represents a simplification of the complex interplay between disease, treatments, and costs. The models assumed only one episode of FN per cycle, and no adverse events were included. The analyses also only evaluated short-acting CSFs, whereas long-acting agents are more commonly used for prophylaxis. Furthermore, the inputs for the percentage of patients requiring hospitalization to treat FN were derived from data that predated more recent guidance that emphasizes outpatient management of FN. These are areas of need for future research. Finally, because of the velocity of new information regarding COVID-19 and its novelty, we did not consider how infection introduced through FN management might affect the results. During the time of the COVID-19 pandemic, these results may be viewed as conservative estimates for the cost-effectiveness of using biosimilar filgrastim (specifically filgrastim-sndz) as a PP strategy. Real-world evidence studies should be conducted to evaluate the impact of these recommendations on population-based outcomes. In the future, the use of machine learning may be viable for reducing the uncertainty within complex health economic models, but this is still in its infancy and not yet the preferred approach by health technology assessment organizations.²³

In conclusion, PP with biosimilar filgrastim is cost-effective in patients receiving intermediate-risk, curative chemotherapy regimens for breast cancer, NSCLC, and NHL. In the era of COVID-19 and value-based care, the use of biosimilar filgrastim has valuable potential to reduce complications associated with unnecessary contact with the health care system among patients undergoing potentially curative chemotherapy. This analysis supports the expanded use of PP with biosimilar filgrastim and should be more strongly considered, if not recommended, by patients and providers to improve long-term outcomes.

Appendix

FIG A1.

General model structure. In addition to health care costs, the model was designed to evaluate the total number of FN events avoided, total LYs, and total QALYs over (A) the first cycle of chemotherapy, (B) subsequent cycles of chemotherapy, and (C) postchemotherapy over a lifetime horizon. The RDI threshold (X% in figure) was 90% for NHL and 85% for breast cancer and NSCLC. FN, febrile neutropenia; LY, life year; NHL, non-Hodgkin lymphoma; NSCLC, non–small-cell lung cancer; QALY, quality-adjusted life year; RDI, relative dose intensity.

AUTHOR CONTRIBUTIONS

Conception and design: Edward Li, Dylan J. Mezzio, Kim Campbell, Gary H. Lyman

Collection and assembly of data: Edward Li, Dylan J. Mezzio, David Campbell, Kim Campbell

Data analysis and interpretation: All authors

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Primary Prophylaxis With Biosimilar Filgrastim for Patients at Intermediate Risk for Febrile Neutropenia: A Cost-Effectiveness Analysis

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO’s conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/op/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).