Economic Model of Granulocyte Colony-Stimulating Factors for Patients Receiving Chemotherapy

The most common complication of myelosuppressive chemotherapy is chemotherapy-induced neutropenia, 1 defined as a reduction in the absolute neutrophil count below 1.5 times 109/L. Grade 3/4 neutropenia is a major risk factor for febrile neutropenia (FN), which is defi ned as either an absolute neutrophil count of <0.5 times 109/L associated with fever (single oral temperature of >38.3°C or >38.0°C for >1 hour) or an absolute neutrophil count of <1.0 times 109/L with a predicted decline to <0.5 t 109/L during the next 48 hours associated with fever.

Febrile neutropenia risk depends on a number of factors, including type of chemotherapy regimen, type of cancer, and patient characteristics.^2-4 Febrile neutropenia may result in hospitalization and treatment with broad-spectrum antibiotics, and is associated with signifi cant morbidity, mortality, and cost.^1,2,5,6 In addition, neutropenia and FN can result in suboptimal delivery of chemotherapy, reducing relative dose intensity and adversely affecting survival.^7-9

The granulocyte colony-stimulating factors (G-CSFs) filgrastim and pegfi lgrastim are hematopoietic growth factors, which stimulate the proliferation and differentiation of neutrophils and promote myeloid hematopoietic recovery after chemotherapy. Both agents are indicated to decrease the incidence of infection, as manifested by FN, in patients with nonmyeloid malignancies receiving myelosuppressive chemotherapy associated with a clinically significant incidence of FN.^10,11 Filgrastim is also indicated for reducing the time to neutrophil recovery and the duration of fever after induction or consolidation chemotherapy treatment of adults with acute myeloid leukemia.¹¹ After administration filgrastim is rapidly cleared by the kidneys, necessitating daily subcutaneous injections for up to 14 days per chemotherapy cycle depending on the neutrophil count.¹² Pegfilgrastim, the pegylated form of filgrastim, is cleared via a neutrophil-mediated system and requires only a single dose administered subcutaneously once per chemotherapy cycle.^13-15 Filgrastim and pegfilgrastim are initiated as primary prophylaxis 24 hours after completion of chemotherapy. Current guidelines, including those of the American Society of Clinical Oncology¹⁶ and the National Comprehensive Cancer Network,¹⁷ recommend use of G-CSF primary prophylaxis when the overall FN risk is greater than approximately 20% following myelosuppressive chemotherapy.

Clinical trials have demonstrated that fi lgrastim and pegfilgrastim shorten the duration of chemotherapyinduced neutropenia and reduce the incidence of FN associated with myelotoxic chemotherapy.^12-15,18-23 In addition, G-CSF use is associated with reductions in the incidence of FN-related hospitalizations and intravenous anti-infective use,¹⁵ reductions in the risk of early all-cause mortality,²⁴ and improvements in relative dose intensity.^9,24

In clinical trials the average number of doses of filgrastim administered per chemotherapy cycle was 10 to 11.^13,14,21-23 However, in clinical practice patients often receive shorter courses of fi lgrastim (<7 days), and although this practice may reduce G-CSF costs, it can lead to an increased FN risk and FN-related hospitalizations. 25-29 An additional compliance issue associated with filgrastim use is timing of administration: treatment with filgrastim is often delayed (ie, not administered as prophylaxis), whereas pegfilgrastim is generally initiated as prophylaxis.^28,30,31 In 2 studies designed to evaluate the noninferiority of pegfi lgrastim versus filgrastim in patients receiving myelosuppressive chemotherapy, pegfi lgrastim was as safe and effective as filgrastim.^13,14

Economic modeling has shown that pegfilgrastim is cost-effective or cost saving compared with filgrastim in patients receiving myelosuppressive chemotherapy.^32-34 However, the total direct costs of pegfi lgrastim and fi lgrastim to a managed care organization have not been evaluated over a short time horizon, consistent with a health plan’s budgeting cycle. Additionally, threshold analyses establishing the required levels of G-CSF effectiveness necessary for equivalent total costs among G-CSFs have not been published. To examine the costs from a US managed care perspective of single-dose pegfi lgrastim versus daily fi lgrastim and the required effectiveness of G-CSFs necessary for equivalent total costs among GCSF treatments, we developed a decision-analytic model using a hypothetical cohort of cancer patients receiving myelosuppressive chemotherapy for nonmyeloid malignancies that are associated with an approximate FN risk of at least 20%.

METHODSStudy Population

The hypothetical study population used for the decision- analytic model was a cohort of 100 patients with breast cancer, non-Hodgkin lymphoma, non—small cell lung cancer, gynecologic cancer, head and neck cancer, or sarcoma who received myelosuppressive chemotherapy regimens associated with an approximate FN risk of at least 20%. The cancer types and chemotherapy regimens were chosen because they are typical indications for G-CSFs.

Model Overview and Structure

The decision-analytic model (Figure 1) was constructed in Microsoft Excel 2007 and compared once per-chemotherapy-cycle pegfi lgrastim with 11- and 6-day filgrastim. The model time horizon was based on an average course of chemotherapy (18 weeks, equivalent to 6 cycles of a once-every-3-week regimen); therefore, no long-term costs or clinical consequences (eg, cancer recurrence and mortality) were included in the model. Patients were assumed to be at risk of FN and FN-related hospitalization depending on cancer type, chemotherapy regimen, G-CSF received, and duration of G-CSF treatment. The model was used to estimate total costs by component of care. Budgetary impact was calculated as the difference between the total costs associated with pegfi lgrastim versus those associated with filgrastim (Figure 1). Although meta-analyses of randomized controlled studies have demonstrated statistically significant differences in FN risk reduction favoring pegfi lgrastim over

daily filgrastim,^35,36 we took a conservative approach by assuming equivalent efficacy for pegfi lgrastim and 11-day filgrastim. Using the assumption of equivalent efficacy, the comparison between these treatments can be characterized as a cost-minimization analysis.

Model Inputs and Calculations

Model inputs including G-CSF cost, treatment cost, FN hospitalization rate, and G-CSF effi cacies were obtained from publicly available data, peer-reviewed publications, and databases available by purchase (Table 1).

Cost Estimates. The costs of pegfilgrastim and filgrastim reflect average sales prices effective October 1, 2011, through December 31, 2011, which were obtained from the Centers for Medicare & Medicaid Services 2011 Average Sales Price Drug Pricing Files.37 The average sales price for filgrastim was calculated for each dose, and t he average sales price per day was calculated as a weighted average, with 18.7% of patients expected to receive 300 μg of filgrastim and 81.3% expected to receive 480 μg of filgrastim (based on IntrinsiQ Research Market Data).

The daily administration cost of G-CSF and the complete blood count (CBC) cost were estimated using fee ranges (range of non-Medicare charges) from the MAG Mutual Physician’s Fee & Coding Guide 2011.³⁸ The fee ranges were based on data from a variety of sources (ie, actual physicians’ charges, health insurers’ databases, an analysis of insurers’ explanation of benefi ts to determine their allowable charges) and were considered appropriate for estimating costs to a managed care organization for these services. Because data on the distribution of fees were not provided, the midpoints of the fee ranges were used to estimate base-case parameter values. The daily administration cost of G-CSF was calculated as the midpoint of the fee range for Current Procedural Terminology (CPT) code 96372 (subcutaneous or intramuscular injection).38 The cost of a CBC (per prescribing information10,11) was calculated by summing the midpoints of the fee ranges for CPT codes 36415 (collection of venous blood by venipuncture) and 85025 (CBC, automated and automated differential count).38 Patients who receive pegfi lgrastim are assumed to receive 1 CBC per cycle the day before each chemotherapy dose. Patients who receive filgrastim are assumed to receive 5 CBCs per cycle (once prior to chemotherapy and twice per week during therapy) for the 11-day regimen11 and 3 CBCs per cycle for the 6-day regimen.

The mean total FN-related hospitalization costs for various tumor types have been reported in 3 publications.^2,5,6 Costs range from $13,354 (2000 US dollars) as reported by Kuderer et al2 to $22,839 (2009 US dollars) as reported by Schilling et al.6 The hospitalization cost reported by Caggiano et al5 ($13,400, the median of the 3 published estimates, 1999 US dollars) was used as the base-case cost of FN requiring hospitalization. The hospitalization cost was first adjusted to 2011 US dollars ($21,186) and further adjusted by 50% ($31,779) to account for costs of subsequent FN-related care (40%) and additional physician fees (10%) consistent with published estimates.^32,33,39 Costs of FN events not requiring hospitalization were assumed to be 39% of the FN hospitalization cost ($8263).³²

Chemotherapy regimens were selected based on an approximate FN risk of at least 20% (Table 2). The assumed distribution of cancers and regimens included in the calculations were intended to be typical of the indications for G-CSFs in a hypothetical health plan, but were not meant to be exhaustive. The proportions of patients experiencing FN events with each regimen were derived from published articles,15,43-52 and the percentage of FN events requiring a hospital stay (80%) was based on data from Lyman et al.32 Chemotherapy cost estimates included drug acquisition costs of chemotherapeutic agents and antiemetics; drug acquisition costs of acetaminophen, diphenhydramine, and methylprednisolone for patients receiving rituximab or taxanes; and chemotherapy administration costs.^37-39,53-58

G-CSF Efficacy. Data on FN risk with and without GCSF for breast cancer patients receiving docetaxel/doxorubicin/ cyclophosphamide (TAC) chemotherapy were used to determine the efficacy of pegfi lgrastim, 11-day filgrastim, and 6-day filgrastim for all the tumor types and chemotherapy regimens in the model. Von Minckwitz et al⁴⁰ provide an estimate of the FN risk with pegfilgrastim as well as the only published estimate of the FN risk for 6-day filgrastim among breast cancer patients receiving TAC. The efficacy (relative risk [RR] of FN) of pegfi lgrastim was calculated by dividing the FN risk for breast cancer patients on TAC who received pegfilgrastim40 by the average FN risk in the absence of G-CSF^41,42 (RR = 0.28, Table 1). The RR of FN with 11-day filgrastim versus no filgrastim was assumed to be equivalent to that of pegfi lgrastim, a conservative assumption based on the results of 2 noninferiority trials in patients receiving myelosuppressive chemotherapy that demonstrated that pegfilgrastim is as effective as filgrastim.^13,14 The RR of FN with 6-day filgrastim versus no fi lgrastim (RR = 0.73) was estimated by dividing the FN risk among breast cancer patients receiving 6-day fi lgrastim40 by the FN risk in the absence of G-CSF.^41,42

To calculate the absolute FN risk for patients receiving G-CSF prophylaxis (pegfilgrastim, 11-day fi lgrastim, or 6-day filgrastim), the RR of FN was multiplied by the baseline FN risk (ie, without G-CSF prophylaxis) for each cancer type and chemotherapy regimen in the model (Table 1).

Threshold Analyses

A 2-way threshold analysis was performed on FN risk for patients receiving 6-day filgrastim (estimated as a percentage of FN risk for patients receiving pegfi lgrastim) and the hospitalization cost of FN. Although FN outpatient cost was estimated as 39% of the FN hospitalization cost, it was not expected to vary with fl uctuations in FN hospitalization cost. Accordingly, FN outpatient cost was held constant at the base-case value of $8263.

The FN risk for patients receiving 6-day filgrastim and FN hospitalization cost were varied until parameter combinations were identifi ed for which 6-day filgrastim was equivalent to pegfilgrastim with respect to costs. The FN hospitalization cost was varied from $0 to $100,000, while FN risk for patients receiving 6-day filgrastim was varied from 100% (equivalent to FN risk for patients receiving pegfilgrastim; RR of FN vs no G-CSF = 0.28) to 275% of FN risk for patients receiving pegfi lgrastim (corresponding to an RR of FN with 6-day filgrastim of 0.77). The resulting parameter combinations were used to plot the isocost line for pegfilgrastim versus 6-day filgrastim; the isocost line traced the various parameter combinations for which total costs for 6-day filgrastim and pegfilgrastim were equivalent.

A 1-way threshold analysis was also done to determine the required level of effectiveness of 11-day filgrastim (estimated as a percentage of the FN risk for patients receiving pegfi lgrastim) to have total costs equivalent to those of pegfilgrastim.

RESULTS

In a hypothetical cohort of 100 cancer patients, the total per patient cost of pegfi lgrastim was estimated to be $38,230 compared with $50,067 for 11-day filgrastim and $42,136 for 6-day filgrastim (Table 3). The total budgetary impact was $11,837 less per patient for pegfilgrastim versus 11- day filgrastim and $3906 less for pegfi lgrastim versus 6-day filgrastim. The largest cost savings for pegfilgrastim versus 11-day filgrastim were derived from the drug acquisition cost of pegfilgrastim, daily administration costs, and CBCs. The costs of chemotherapy, FN-related hospitalization, and FN-related outpatient management were equivalent for pegfilgrastim and 11-day filgrastim. The largest cost savings for pegfi lgrastim versus 6-day filgrastim were derived from FN-related hospitalization costs, followed by daily administration costs and the cost of CBCs.

The results of the 2-way threshold analysis are presented in Figure 2. Parameter combinations to the right of the isocost line represent lower total costs for pegfilgrastim versus 6-day filgrastim; those to the left of the isocost line represent lower total costs for 6-day filgrastim versus pegfilgrastim.

When the cost of FN hospitalization was held constant at the base-case value of $31,779, the 2-way threshold analysis indicated that, in order to obtain lower total costs for 6-day filgrastim versus pegfilgrastim, the FN risk for patients receiving 6-day filgrastim would have to be less than 127% of the FN risk for patients receiving pegfi lgrastim (RR of FN with 6-day filgrastim of <0.36).

When the FN risk for patients receiving 6-day filgrastim (as a percentage of the FN risk for patients receiving pegfilgrastim) was held constant at the base-case value of 260% (RR of FN with 6-day filgrastim of 0.73), the FN hospitalization cost would have to be less than $3400 in order to obtain lower total costs of 6-day fi lgrastim versus pegfilgrastim. Results of the 1-way threshold analysis indicate that even if patients receiving 11-day filgrastim had zero FN risk, pegfi lgrastim would still be the less expensive choice.

DISCUSSION

We developed a decision-analytic economic model from a US managed care perspective to examine the costs of single-dose pegfi lgrastim versus multidose fi lgrastim in patients receiving myelosuppressive chemotherapy regimens associated with an FN risk of approximately 20% or greater. The G-CSF, administrative, and laboratory costs are lower for patients who receive pegfi lgrastim primary prophylaxis versus 11-day filgrastim; administrative, laboratory, and FN-related hospitalization costs are lower for patients who receive pegfilgrastim versus 6-day filgrastim. Our results are consistent with previously published cost-effectiveness studies of G-CSF primary prophylaxis for reducing the FN risk. Lyman and colleagues_32,33 found that pegfilgrastim primary prophylaxis was cost-effective compared with 6-day filgrastim in non-Hodgkin lymphoma patients receiving cyclophosphamide, doxorubicin, vincristine, and prednisolone (CHOP) chemotherapy and in early-stage breast cancer patients receiving TAC. Similarly, Eldar-Lissai and colleagues³⁴ estimated that pegfilgrastim resulted in fewer FN events and lower costs compared with 7- to 12-day filgrastim.

Two key points differentiate our study from those published previously. First, we conducted model analyses from the US managed care perspective and included only direct costs, which are frequently used for healthcare reimbursement decisions by US payers. Second, we included several threshold analyses designed to assess the level of effectiveness of daily filgrastim that is required for filgrastim to be equivalent to pegfi lgrastim with respect to costs. Results indicate that even if patients receiving 11-day filgrastim had zero FN risk, pegfilgrastim would still be the less expensive choice. For total costs of 6-day filgrastim to be less than those of pegfilgrastim, either the FN risk for patients receiving 6-day fi lgrastim would have to be less than 127% of the FN risk for patients receiving pegfi lgrastim, or the FN hospitalization cost would have to be less than $3400. These thresholds provide new evidence on the relative level of effectiveness necessary for daily filgrastim to be considered the less expensive choice and serve as reference values that can be used by decision makers when additional data on the relative effectiveness of filgrastim become available.

There are several limitations to our analysis. First, in the absence of data, we assumed that the RR of FN with GCSF versus no G-CSF is the same among different cancer types and chemotherapy regimens. Second, no randomized controlled trials have compared the effi cacy of 6-day filgrastim with either 11-day filgrastim or pegfi lgrastim. Third, as model analyses were performed from the managed care organization perspective, we only considered short-term clinical outcomes and costs and did not include survival, quality of life, or other long-term outcomes. Finally, the model compared G-CSF use for every-3-week chemotherapy regimens; daily filgrastim may be the more appropriate treatment choice for other regimens.

Although alternative treatment options (eg, no G-CSF) may be used in clinical practice, the scope of this modeling study was limited to use of G-CSFs. Providing no GCSF to highrisk patients is outside of currently accepted treatment guidelines (eg, National Comprehensive Cancer Network and American Society of Clinical Oncology^16,17) and accepted clinical practice and would likely result in unacceptable FN hospitalization rates for very high risk myelosuppressive regimens. Furthermore, a metaanalysis has shown increased mortality among patients who received placebo in the fi rst cycle of chemotherapy or secondary prophylaxis after an occurrence of FN.^24,59 Therefore, we chose pegfilgrastim, 11-day filgrastim (based on clinical studies), and 6-day filgrastim (based on clinical practice) as the relevant comparators for patients receiving moderate-risk to high-risk chemotherapy.

Results primarily depend on the overall FN risk, which is determined by the distribution of patients across the cancer types and chemotherapy regimens in a health plan. Randomized controlled trial populations used to derive FN risk are subject to inclusion and exclusion criteria, and may therefore include a relatively healthy cohort of cancer patients. In the community setting, patients may not be as healthy (due to factors such as age and comorbidities) and may have a higher FN risk than presented here. In a setting where FN risk may be higher, pegfi lgrastim would become less costly relative to filgrastim. Accordingly, model results may change based on the distribution of cancers, treatments within the cancers, and demographics of the health plan patient population.

Additional costs not considered relevant from a managed care perspective include FN-related productivity losses from lost work days60 and the opportunity cost of healthcare providers’ potential time savings due to 1-time administration of pegfilgrastim versus daily filgrastim. Of note, Fortner and colleagues61 found that the total number of human resource hours and costs per patient were 2.4 hours and US$57.30 for pegfilgrastim versus 14.8 hours and US$364.⁶⁶ for filgrastim.

Using the best available evidence and cost information, our analysis suggests that in the US managed care setting pegfilgrastim is a less-expensive choice than either 11- or 6-day filgrastim for reducing the risk of chemotherapy- induced FN and FN-related hospitalizations.