Early Adoption of Biosimilar Growth Factors in Supportive Cancer Care

Abstract

This retrospective observational analysis of health claims data compares the safety, effectiveness, and cost of biosimilar filgrastim and filgrastim.

In 2010, the US Food and Drug Administration established an approval pathway for biosimilar agents, biological agents with an active ingredient highly similar to the reference biological agent, to facilitate drug competition and lower costs. It is unclear whether biosimilar drugs could yield large cost savings in the United States given the previously documented market and regulatory barriers.^1,2,3 Moreover, few studies have investigated the real-world safety and effectiveness profile of biosimilar agents compared with that of brand-name (reference) biological agents.

The colony-stimulating growth factors (CSF) filgrastim-sndz and tbo-filgrastim reduce risk of chemotherapy-induced neutropenia and were the first biosimilar agents approved in the United States. We examined the incidence of febrile neutropenia (FN), CSF-related adverse events (AEs), and drug cost among commercially insured US patients with cancer treated with chemotherapy.

Methods

This retrospective observational study analyzed administrative health claims data from a large commercially insured population. Only deidentified data were used, and the study was exempt from review by an institutional review board. The study included commercially insured adults with cancer treated with chemotherapy who received reference filgrastim or biosimilar filgrastim from September 1, 2012, through April 30, 2017. Patients with atypical risk for FN (ie, those with previous bone marrow or stem cell transplant or prior FN) were excluded from the study. The population was divided into either biosimilar or reference filgrastim groups according to the earliest filgrastim use (index date).

Patients were observed for 21 days postindex to assess filgrastim treatment cost (total and per day of use) and the incidence of FN and AEs (eg, spleen rupture, acute respiratory syndrome, serious allergic reactions). Two validated claim-based algorithms (neutropenia and/or fever requiring hospitalizations) were used to capture the incidence of FN.⁴ Study outcomes were compared using multivariable regression model adjusted for baseline risk factors (Table 1).

Table 1. Incidence Rate of Febrile Neutropenia and Growth Factor–Related Adverse Events in Biosimilar Filgrastim and Filgrastim Treatment Groups.

Measures	No. (%)		Adjusted Odds Ratio (95% CI)a	P Valueb
	Biosimilar Filgrastim (n = 1531)	Filgrastim (n = 9671)
Hospitalization for febrile neutropenia
Broad definition (neutropenia or fever requiring hospitalization)c	173 (11.3)	912 (9.4)	1.15 (0.96-1.37)	.12
Strict definition (neutropenia and fever requiring hospitalization)c	69 (4.5)	373 (3.9)	1.12 (0.86-1.46)	.39
Growth factor–related adverse eventsd	106 (6.9)	621 (6.4)	1.04 (0.84-1.29)	.72

^aDerived from logistic models adjusted for age, sex, race, census region, rural/urban indicator, modified Deyo-Charlson Comorbidity Index; myelosuppressive risk status of chemotherapy agents during the 7 days before the index date; metastases; history of heart disease; history of liver disease; history of infection; previous radiation therapy; count of all-cause emergency department visits; and all-cause medical cost. The baseline covariates were captured within 6 months preindex.

^bP values were 2-sided; P < .05 was considered statistically significant.

^cEmergency department visits for febrile neutropenia followed by hospitalization were counted as hospitalization.

^dAdverse events included spleen rupture, acute respiratory syndrome, serious allergic reactions, capillary leak syndrome, thrombocytopenia, leukocytosis, cutaneous vasculitis, or bone and muscle ache.

Results

Of 11 202 patients included in the study, 1531 (13.7%) used biosimilar CSF. The median patient age was 61 years, 6927 patients (61.8%) were female, and 5782 patients (51.6%) had at least 1 non–cancer-related comorbid condition. No meaningful differences in baseline risk factors were observed between the filgrastim and biosimilar filgrastim groups. Biosimilar filgrastim accounted for 7% of filgrastim users in 2014 (171 of 2580), which increased to 16% in 2015 (384 of 2420) and to 36% in 2016 (778 of 2180).

Incidence rates of strict (neutropenia and fever) and broad (neutropenia or fever) definitions of FN were similar between the 2 groups. The incidence of broad FN in patients receiving biosimilar agents was 11.3%, whereas the incidence in those receiving brand-name agents was 9.4% (P = .12). Incidence rates in patients with strict FN were 4.5% and 3.9% in patients receiving biosimilar agents or brand-name agents, respectively (P = .39; Table 1). No difference in the rate of AEs was observed (biosimilar agents, 6.9%; brand-name agents, 6.4%; P = .72). The mean filgrastim drug costs per patient were similar between the biosimilar and reference filgrastim groups (biosimilar agents, $2522; brand-name agents, $2516; P = .27; Table 2); the cost per day of use was 2.3% less for biosimilar filgrastim (biosimilar agents, $731; brand-name agents, $748; P = .02). Compared with reference filgrastim, the mean total costs per patient and per day of use were 6.1% and 10.8% less for filgrastim-sndz (P = .07 and P < .001), respectively, and were marginally higher for tbo-filgrastim (2.7% and 2.3%, respectively; P = .64 and P = .50, respectively).

Table 2. Treatment Cost for Biosimilar Filgrastim and Filgrastim Treatment Groups.

Measures	Biosimilar Filgrastim (n = 1531)	tbo-Filgrastim (n = 963)a	Filgrastim-sndz (n = 568)a	Filgrastim (n = 9671)
Filgrastim Treatment Cost per Day of Use, US Dollars
Mean (SD)	731 (1076)	765 (1256)	667 (663)	748 (1054)
Median (IQR)	462 (336-878)	470 (328-917)	449 (337-809)	504 (359-793)
Cost ratio (95% Cl)b	0.95 (0.90-0.99)	0.98 (0.93-1.04)	0.88 (0.82-0.95)	1 [Reference]
P valuec	.02	.50	<.001
Filgrastim Treatment Cost per Patient, US Dollars
Mean (SD)	2522 (4437)	2585 (4832)	2363 (3667)	2516 (3487)
Median (IQR)	1340 (739-2807)	1319 (705-2906)	1354 (749-2581)	1501 (814-2861)
Cost ratio (95% Cl)	0.97 (0.92-1.02)	0.98 (0.92-1.05)	0.92 (0.84-1.01)	1 [Reference]
P value	.27	.64	.07

Abbreviation: IQR, interquartile range.

^aFilgrastim-sndz was approved in March 2015 and was the first biosimilar agent approved in the United States. Tbo-filgrastim received US Food and Drug Administration (FDA) approval as a biological agent in August 2012 because a biosimilar-agent approval pathway had not been established at the time of FDA submission. In our analysis, tbo-filgrastim users and filgrastim-sndz users were both considered as part of the biosimilar filgrastim cohort.

^bDerived from multivariable regression models with gamma distribution and log link adjusted for age, sex, race, census region, rural/urban indicator, modified Deyo-Charlson Comorbidity Index; myelosuppressive risk status of chemotherapy agents during the 7 days before the index date; metastases; history of heart disease; history of liver disease; history of infection; previous radiation therapy; count of all-cause emergency department visits; and all-cause medical costs. The baseline covariates were captured within 6 months preindex.

^cP values were 2-sided; P < .05 was considered statistically significant.

Discussion

The findings suggest biosimilar filgrastim is similar to reference filgrastim in terms of safety and effectiveness in the real-world setting. The results are reassuring regarding the intended clinical outcomes associated with filgrastim. The savings associated with biosimilar filgrastim have been modest because early adoption of biosimilar agents has been slow, and current pricing for biosimilar drugs is closely tracking that of the brand-name product. However, the market for biosimilar agents seems to be increasing over time, indicating broader acceptance of these drugs. As more biosimilar agents enter the market, continuing research is needed to assess the drug uptake, clinical outcomes, and realized cost savings. Our study, which was conducted using claims data, has the advantage of rich cost information, but it is limited by the lack of clinical detail for most patients.