Breast cancer is the most frequently diagnosed cancer in US women, and ranks second among cancer- related deaths in women, after lung cancer.1 It is estimated that $8.1 billion (in 2004 $US) in total healthcare costs are spent annually on breast cancer diagnosis and treatment in the United States.2 Chemotherapeutic agents represent a significant portion of the cost of breast cancer treatment, and health plans are managing these costs with care pathways and other utilization management strategies.
Taxanes are among the most frequently used forms of systemic therapy for the treatment of breast cancer.3 These chemotherapeutic agents can be prescribed alone or in combination with other systemic therapies or with local treatment, such as surgery and/or radiation. Taxanes are mitotic inhibitors originally isolated from the bark of the Pacific yew tree, Taxus brevifolia. Three taxanes are currently available in the United States, although this is an area of active drug development.
The first of these agents to be marketed in the United States, paclitaxel injection (Taxol), was approved by the US Food and Drug Administration (FDA) in 1992 as an injectable formulation dissolved in a proprietary version of polyethoxylated castor oil called Cremophor EL and ethanol as a delivery agent.4 The second agent—docetaxel (Taxotere)—received FDA approval in 1996; docetaxel is a semisynthetic taxane dissolved in polysorbate 80 and 13% ethanol and water for injection.5
The toxicity of these 2 solvent-based taxanes includes bone marrow suppression (principally neutropenia), alopecia, and hypersensitivity reactions.6
Irizarry and colleagues reviewed 171 reported cases of anaphylaxis associated with the Cremophor EL–based paclitaxel. A total of 34% were fatal reactions, although the authors suggested that this adverse event (AE) is underreported.7 Although neurotoxicity and myalgias/arthralgias have been cited as potential AEs with paclitaxel and docetaxel, these events are of greater clinical concern with paclitaxel—again, likely because of the Cremophor EL solvent.6 However, docetaxel has been associated with the development of significant fluid retention (eg, edema, ascites, pleural effusions), the incidence and severity of which appear to be reduced by premedication with corticosteroids.6 These toxicities may result in dose reductions, delays in continuation of treatment, or even discontinuation of chemotherapy.
In an attempt to reduce the significant toxicity associated with the solvent-based taxanes, a new formulation of paclitaxel was developed—nab-paclitaxel (Abraxane)—in which the active drug is bound to albumin.8 Nab-paclitaxel received FDA approval in 2005. This albumin-bound paclitaxel has demonstrated superior efficacy over the solvent-based paclitaxel in the treatment of metastatic breast cancer (MBC), as well as a more favorable AE profile.9,10
Although the solvent-based taxanes—paclitaxel and docetaxel—have lower drug acquisition costs than nabpaclitaxel, nab-paclitaxel offers potential clinical advantages, such as better tumor responses with lower rates of severe neutropenia and infusion-related reactions.9,10
Systematic reviews have indicated that taxanes are among the most active regimens in MBC.11 However, differences in the toxicity and efficacy of the taxanes exist. Gradishar and colleagues found that nab-paclitaxel was associated with better response rates compared with the solvent-based generic paclitaxel in MBC (33% vs 19%, respectively; P = .001), with a longer time to tumor progression (23.0 weeks vs 16.9 weeks, respectively; P = .006).10 In addition, the rates of severe neutropenia were significantly lower in the nab-paclitaxel group (9% vs 22%, respectively; P <.001).10
Similarly, in another study, nab-paclitaxel was associated with a significantly longer progression-free survival (12.9 months vs 7.5 months, respectively; P = .0065) compared with docetaxel in women with MBC.12 Grade-3 or -4 neutropenia occurred in 94% of the patients receiving docetaxel and in 38% of patients receiving nab-paclitaxel.12
To evaluate the economic implications of these differences in efficacy and toxicity, Dranitsaris and colleagues performed a pharmacoeconomic analysis of nabpaclitaxel in MBC.13 Nab-paclitaxel was less expensive than docetaxel per quality-adjusted life-year in the costutility model, but more expensive than generic paclitaxel.13 The disadvantages of this type of analysis include a variety of model assumptions that render the real-world implications of the results difficult to interpret by pharmacy and medical directors at health plans.
In 2008, Barron and colleagues evaluated all-cause costs in women treated for breast cancer using claims from 5 US health plans.14 The authors reported mean unadjusted per-patient per-month (PPPM) all-cause costs of $4421 in women receiving treatment for breast cancer compared with PPPM costs of $3352 for agematched controls without breast cancer.14 The top 3 cost categories included hospitalizations, drug therapy, and surgical intervention.14 The components of total cost are of interest to payers, because they manage increasing costs associated with breast cancer treatment. PPPM cost calculations provide plan managers with an estimate of the cost of treating patients with a specific condition. These costs are quite different and generally much higher than typical actuarial estimates of permember per-month dollar amounts.
The benefits and costs with any new treatment option need to be fully evaluated. At a December 2009 ICORE Healthcare Managed Care Advisory Board meeting in Bethesda, MD, payers questioned the value of nab-paclitaxel because of its higher acquisition cost. However, a comparison between nab-paclitaxel and solvent-based taxanes should include the costs associated with toxicity, drug utilization, and ancillary medication use rather than drug acquisition costs alone. These factors contribute to total medical cost, which may be used as the primary outcome measure in a pharmacoeconomic evaluation.
To clarify these issues surrounding the use of the taxanes, we performed a pragmatic, retrospective evaluation of paid medical claims, with the intent of determining the impact of taxane choice on total medical costs in the treatment of MBC.
This study was a cost comparison using paid medical claims from all locations of service from national commercial payer sources. Medical claims include all office visits, hospital charges, procedures (eg, laboratory tests, x-rays), and medications administered in the physician office, infusion center, or hospital. Total medical costs were derived from these data elements. Outpatient pharmacy claims and costs were not available and therefore were not included.
Data were handled in compliance with the Health Insurance Portability and Accountability Act of 1996. Because the data analysis did not involve any patient intervention and used a limited data set with blinded patient identification, a certificate of exemption was obtained from our Institutional Review Board. The sponsor had no role in the study design, analysis, or writing of the paper. Investigators had exclusive access to the data and independently developed the research design and wrote the manuscript.
Study Population and Case Definitions
The data set was derived from a total of 95,530,675 procedure claims from 6,130,088 unique patients (Ingenix Consulting, Eden Prairie, MN). Claims data were analyzed for 24 months, from January 1, 2006, to December 31, 2007. Only allowed charges from patients coded with a diagnosis of breast cancer were included in the final analysis. Patients receiving taxane therapy were identified using stepwise computerized algorithms to define a subset of taxane use in MBC (Figure 1).
Initially, all patients with breast cancer with an International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis code of 174.X were identified. Patients with a single taxane administration were then removed from the analysis. Next, patients who may have developed breast cancer as a secondary site were excluded (1) if any other cancer diagnosis (ICD-9-CM codes 140.XX-173.XX; 175.XX- 195.XX; and 200.XX-208.XX) occurred in the 6 months preceding the initial breast cancer diagnosis, or (2) if they received any chemotherapy other than office- or hospital-administered adriamycin-cyclophosphamide in the 6 months before their first breast cancer diagnosis.
Taxane use within this subgroup was subject to the following criteria to define taxane use in MBC:
- Taxane administration occurring more than 30 days after an established doxorubicin-cyclophosphamidetaxane adjuvant regimen
- Concurrent (±3 days) taxane administration with either doxorubicin or cyclophosphamide in which doxorubicin-cyclophosphamide use did not conform to established adjuvant treatment regimens
- Taxane administration occurring after a secondary (metastatic) cancer diagnosis (ICD-9-CM codes 140.XX-173.XX and 175.XX-199.XX)
- Taxane use not described by these rules was eliminated from further analysis, because this subgroup of patients tended to have incomplete data and therefore an accurate assessment of chemotherapy use could not be done.
Drug use was identified by the following J codes:
- J9170 indicating docetaxel 20 mg
- J9265 indicating generic paclitaxel 30 mg
- J9264 indicating nab-paclitaxel 1 mg.
For analytical purposes, the index date was defined as the date of a patient’s first dose of taxane; from this point onward, total paid charges were captured. Patients were placed in generic paclitaxel, docetaxel, and nab-paclitaxel groups according to their first taxane use. Patients were not required to be continuously enrolled over the study period, because multiple regression analysis was used to adjust for individual patient enrollment period variability (eg, total months with a claim was a control variable).
Analytical Outcomes and Cost Analysis
Mean age was determined for each of the 3 groups based on the patients’ date of birth and date of first taxane use (index date). The number of taxane doses received was determined. In addition, the mean dosing interval for each taxane was calculated; however, intervals longer than 60 days were excluded from the interval calculations. Duration of months in the database was measured from the date of first taxane administration (index date) to the time of last claim, because of disenrollment, end of the data period, or death.
Chemotherapy and biologic agent utilization—specifically bevacizumab (Avastin), trastuzumab (Herceptin), gemcitabine (Gemzar), and vinorelbine (Navelbine)—was determined before, concurrent with, and post-taxane utilization; each was included as a control variable in the multivariate analysis (SAS, SAS Institute, Inc, Cary, NC). The total medical cost statistical model also adjusted for age; number of procedures (ie, office visits, hospitalizations, laboratory tests) after index date; number of unique diagnoses after index date; number of months with claims in the data set after index date; number of taxane doses; and the presence of a taxane claim in the first month. Procedures and diagnoses were included in the multivariate analysis and served as surrogates for severity of illness. Medical costs are reported in actual US dollars, from 2006 and 2007, without discounting.
While adjusting for these variables, total medical cost comparisons were calculated, as well as PPPM total medical costs for managing patients with MBC for the 3 taxane groups. All medical costs were captured from the index date through the end of the data collection period, while controlling for the duration of available data and duration of enrollment for each patient.
Utilization of ancillary medications was defined as a claim for colony-stimulating factors (CSFs) or erythropoiesis- stimulating agents (ESAs) between days 0 and +21, or antiemetics between days 0 and +5 after each taxane administration. Cost comparisons of CSFs or ESAs between days 0 and +21 after taxane administration, or antiemetics between days 0 and +5 after taxane administration, were made using Tobit models.
Tobit models are utilized when the entire study population does not undergo the same interventions. For example, not all patients received CSFs; Tobit models can control for unbalanced utilization and allow for an appropriately adjusted evaluation of cost. For the purposes of this analysis, all the variables included in the multivariate analysis were also included in the Tobit analyses. Throughout the study, all comparisons were considered significantly different at P <.05.
A total of 2245 women with a mean (± standard deviation [SD]) age of 53 years (±10 years) receiving a taxane for MBC were included in the study. The docetaxel group included 1035 patients, the generic paclitaxel group included 997 patients, and the nab-paclitaxel group included 213 patients. Mean (± SD) residence in the database was longer for patients receiving nab-paclitaxel (16.5 ± 6.6 months) compared with docetaxel (12.6 ± 6.3 months) and generic paclitaxel (13.5 ± 6.4 months). Nab-paclitaxel had a significantly higher mean number of doses per patient (9.4; 95% confidence interval [CI], 8.5-10.2), followed by generic paclitaxel (7.0; 95% CI, 6.6-7.3) and docetaxel (5.9; 95% CI, 5.5-6.3), which was adjusted for residence in the database by pairwise comparisons using the Bonferroni method. The mean (±SD) dosing intervals for patients receiving docetaxel, generic paclitaxel, and nab-paclitaxel were 20.5 ± 6.2 days, 16.2 ± 8.8 days, and 15.8 ± 8.9 days, respectively (Table 1).
Other Chemotherapy Utilization
A higher proportion of patients received chemotherapy before the initiation of nab-paclitaxel than before initiating either generic paclitaxel or docetaxel (P <.05 for both); a greater proportion of patients taking generic paclitaxel than patients taking docetaxel (P <.05) received previous chemotherapy (Table 2). The rate of concurrent chemotherapy use was greater in patients taking docetaxel compared with those taking generic paclitaxel or nab-paclitaxel (P <.05 for both); the use for the latter 2 drugs was not significantly different. Posttaxane chemotherapy utilization was not significantly different among the 3 groups (P >.05).
Analyses of the concurrent use of bevacizumabrevealed that a greater proportion of patients in the nabpaclitaxel group received bevacizumab concurrently compared with patients receiving generic paclitaxel and docetaxel, and the rate of patients using generic paclitaxel who received concurrent bevacizumab was greater (P <.05 for all) than patients taking docetaxel (Table 3). A higher proportion of patients receiving nab-paclitaxel also received bevacizumab after the taxane; the rate of bevacizumab use was greater after generic paclitaxel than after docetaxel (P <.05 for both), but the difference was not significant (P >.05) between patients receiving nabpaclitaxel and generic paclitaxel (Table 2). The proportion of patients receiving concurrent or post-taxane trastuzumab was not significantly different (P >.05) among the 3 groups (Table 3).
The rate of patients using nab-paclitaxel who received concurrent and post-taxane gemcitabine was greater than in the generic paclitaxel and the docetaxel groups (P <.05 for both), which were not significantly different (P >.05) from each other in either setting.
The proportion of patients taking nab-paclitaxel who received vinorelbine concurrently was greater than patients taking generic paclitaxel or docetaxel (P <.05 for both), and the proportion of generic paclitaxel patients who received concurrent vinorelbine was more than those taking docetaxel (P <.05). The proportion of patients receiving post-taxane vinorelbine was not significantly different among the 3 groups (P >.05).
Total Medical Cost Analysis
The total medical cost model was robust (r2 = 0.72). Increased total medical cost was attributable to a number of factors, whereas decreased total medical cost was attributable to advancing age and previous chemotherapy (see Appendix at www.AHDBonline.com).
Over the 24-month study period, adjusted median PPPM total medical costs were higher in patients receiving nab-paclitaxel ($3997; 95% CI, $3634-$4396) and docetaxel ($4042; 95% CI, $3844-$4251) (P <.05 for both comparisons) than in patients receiving generic paclitaxel ($3203; 95% CI, $3029-$3388). However, these costs were not significantly different between the nab-paclitaxel and docetaxel groups (Table 4).
When nab-paclitaxel was used as the initial treatment, the adjusted total medical costs were 25% (95% CI, 15%-36%) higher than when generic paclitaxel was used first, holding other factors constant. Similarly, when docetaxel was used as the initial treatment, the adjusted total medical costs were 26% (95% CI, 20%- 33%) higher than when treatment was initiated with generic paclitaxel. There was no significant difference in adjusted total medical costs of care between docetaxel and nab-paclitaxel.
Ancillary Medication Utilization and Cost Analysis
In the analysis of ancillary medication use, rates of CSF use were significantly higher for docetaxel than for generic paclitaxel and nab-paclitaxel (P <.05 for both); the latter 2 agents were not significantly different (P >.05) from each other (Table 5). Rates of ESA use were not significantly different among the 3 groups (P >.05). Antiemetic use was significantly greater for docetaxel than for generic paclitaxel (P <.05) but not significantly different between nab-paclitaxel and generic paclitaxel or nab-paclitaxel and docetaxel (P >.05 for both; Table 5).
Figure 2 outlines the differences in ancillary medication costs associated with generic paclitaxel and nabpaclitaxel compared with docetaxel. Per-patient CSF costs were significantly lower with generic paclitaxel (–$7933) and with nab-paclitaxel (–$12,410) compared with docetaxel (P <.05); patients in the generic paclitaxel group had greater expenditures for CSFs compared with nab-paclitaxel (P <.05). Per-patient ESA costs were significantly lower with nab-paclitaxel (–$552) compared with docetaxel (P <.05). There were no differences in the costs of antiemetics between the groups (Figure 2).
In this retrospective claims analysis, we were able to identify patients with MBC and effectively control for an array of variables, while evaluating total medical costs in this population. This type of analysis reflects the expenditures incurred by a health plan and underscores the importance of a comprehensive analysis that controls for patient and treatment variability; this is preferred to a simple comparison of drug acquisition costs or of mean (or median) total medical costs.
The median adjusted PPPM total medical costs for patients receiving taxanes for MBC were within about $800 of each other (Table 4). Median PPPM total medical costs for docetaxel and nab-paclitaxel were not statistically different, although total medical costs associated with generic paclitaxel treatment were about 25% lower than docetaxel and nab-paclitaxel when analyzed with multiple regression. The cost model explained 72% of the variability in total medical costs.
To evaluate the total medical costs associated with the use of nab-paclitaxel instead of generic paclitaxel, we performed a post-hoc analysis of a hypothetical health plan covering 1 million lives, with approximately 50% women. If the annual incidence of breast cancer is 122.9 per 100,000 women and approximately 33% have MBC, then 203 women ([500,000 X 0.001229] X 0.33 = 203) from the hypothetical cohort would have MBC.15
The median total PPPM medical cost difference between nab-paclitaxel and generic paclitaxel is $794 (Table 4). If this cost were distributed across the entire insured population, each 10% increase in the proportion of patients receiving nab-paclitaxel instead of generic paclitaxel would represent a $0.016 incremental increase in total medical expenditures per member per month. The patients receiving nab-paclitaxel would be expected to realize a reduction in neutropenia and related costs.
In the present study, patients received more doses of nab-paclitaxel than either docetaxel or generic paclitaxel; this may be attributed to a number of factors, including stage of treatment, tolerability, and efficacy. Because nab-paclitaxel is often used in late-stage MBC,8 patients observed receiving this taxane might have required more aggressive treatment. This was confirmed in our study, because nab-paclitaxel was administered more frequently in combination with bevacizumab, gemcitabine, and vinorelbine, which were used more often by patients with advanced MBC.
Other studies have underscored the impact of cost elements in addition to drug acquisition cost in determining total costs.3,14 An analysis by Barron and col-leagues demonstrated that the significant economic burden of breast cancer is driven mainly by hospitalizations, which contribute to the total cost at a rate more than twice that of chemotherapy.14 However, the costs of chemotherapy were significant, and the analysis was performed before the availability of several expensive agents used by patients with MBC.14 More recently, Kruse and colleagues found that chemotherapeutic agents accounted for nearly 60% of costs associated with breast cancer.3
In the present analysis, chemotherapy expenses (taxane plus all other chemotherapy) represented between 44% and 62% of unadjusted mean total medical costs, exclusive of outpatient pharmacy chemotherapy, in the 3 taxane groups. The sum of hospital, office visit, and other charges (eg, laboratory tests, x-rays) represented less than 41% of unadjusted mean total medical costs in the 3 taxane groups. In the analyses by Barron and colleagues and Kruse and colleagues, the mean PPPM total medical costs were $4996 and $4421, respectively,3,14 compared with our median values, which ranged from $3203 to $4042.
In our analysis, patients receiving nab-paclitaxel had significantly lower expenditures for CSFs compared with docetaxel or generic paclitaxel. This may be explained by the improved tolerability and lower rates of neutropenia observed with the albumin-bound formulation. In a study based on a database of patients hospitalized for febrile neutropenia, Kuderer and colleagues concluded that a number of patient-specific factors, such as severity of disease, type of malignancy, and comorbidities contribute to increased mortality, length of stay, and hospitalization costs.16 Such factors should be taken into consideration, and appropriate therapies should be utilized, to address these concerns as a means of improving outcomes and controlling costs.
The most obvious limitation of this analysis is the retrospective nature of the study, which does not control for drug selection biases related to the stage of therapy. We went to considerable effort to identify patients with MBC; however, it is difficult to identify second- and third-line therapies in a relatively short period of followup. Continuous enrollment was not an inclusion criterion for patient selection, but duration of residence in the data set was used as a control variable. Requiring continuous enrollment of our population of patients with MBC would have significantly limited our sample sizes.
Furthermore, in this type of analysis it is difficult to control for patient severity of illness, because the clinical status of the patient is not available. We did, however, control for the number of procedures and the number of unique diagnoses for the patients included in this analysis. Both of these variables were associated with significant increases in total costs.
In addition, this analysis did not specifically identify AE rates, such as infusion-related AEs; however, the costs associated with AEs are included in the total medical cost calculations.
These data represent a snapshot in time rather than a comprehensive view of disease progression from early- to late-stage illness.
Tobit analyses have been criticized in healthcare claims analyses, because of highly variable costs over time. Although outpatient drug therapy costs were not captured, any bias should be equally applied across all 3 taxane groups.
Finally, the clinical efficacy of the taxanes was not assessed in this study.
Analysis of the myriad factors besides drug acquisition cost is necessary to make a determination of the costs of a particular therapy. The efficacy and tolerability of various treatments are reflected in total costs, where reductions in AEs and the use of ancillary medications offset higher drug acquisition costs. In the case of the taxanes, the overall costs of breast cancer management are related to the entire chemotherapeutic regimen, as well as to the costs of hospitalization. These factors contribute to total medical cost, which may be used as the primary outcome measure in a pharmacoeconomic evaluation rather than simply drug acquisition cost. In this analysis of total medical costs, generic paclitaxel was the least expensive, whereas nab-paclitaxel and docetaxel resulted in similar overall medical expenditures.
The authors thank Teri Peterson, MS, for her assistance with the statistical analyses.
This project was sponsored by Abraxis Oncology, manufacturer of Abraxane (nab-paclitaxel). The sponsor had no role in the study design, analysis, or writing of the manuscript. The investigators had exclusive access to the data and independently developed the research design and wrote this report. Dr Force, Dr Pugmire, and Dr Culbertson have received research/grant support from Abraxis and ICORE Healthcare.
Dr Force is also a Consultant to Abraxis and ICORE Healthcare.
- American Cancer Society. Cancer Facts & Figures 2008. Atlanta, GA: American Cancer Society; 2008.
- Brown ML, Riley GF, Schussler N, Etzioni R. Estimating health care costs related to cancer treatment from SEER-Medicare data. Med Care. 2002;40(suppl 8):IV-104-IV-117.
- Kruse GB, Amonkar MM, Smith G, et al. Analysis of costs associated with administration of intravenous single-drug therapies in metastatic breast cancer in a U.S. population. J Manag Care Pharm. 2008;14:844-857.
- Taxol (paclitaxel) injection prescribing information. New York, NY: Bristol-Myers Squibb; July 2007.
- Taxotere (docetaxel) prescribing information. Bridgewater, NJ: sanofi-aventis; 2007.
- Markman M. Managing taxane toxicities. Support Care Cancer. 2003;11:144-147. Epub 2002 Oct 15.
- Irizarry LD, Luu TH, McKoy JM, et al. Cremophor EL-containing paclitaxelinduced anaphylaxis: a call to action. Community Oncol. 2009;6:132-134.
- Abraxane (paclitaxel) prescribing information. Los Angeles, CA: Abraxis BioScience; 2007.
- Desai, N, Trieu V, Yao Z, et al. Increased antitumor activity, intratumor paclitaxel concentrations, and endothelial cell transport of cremophor-free, albumin-bound paclitaxel, ABI-007, compared with cremophor-based paclitaxel. Clin Cancer Res. 2006;12:1317-1324.
- Gradishar WJ, Tjulandin S, Davidson N, et al. Phase III trial of nanoparticle albumin-bound paclitaxel compared with polyethylated castor oil-based paclitaxel in women with breast cancer. J Clin Oncol. 2005;23:7794-7803. Epub 2005 Sep 19.
- Ghersi D, Wilcken N, Simes J, Donoghue E. Taxane containing regimens for metastatic breast cancer. Cochrane Database Syst Rev. 2005;2:CD003366.
- Gradishar WJ, Krasnojon D, Cheporov S, et al. Significantly longer progressionfree survival with nab-paclitaxel compared with docetaxel as first-line therapy for metastatic breast cancer. J Clin Oncol. 2009;27:3611-3619. Epub 2009 May 26.
- Dranitsaris G, Cottrell W, Spirovski B, Hopkins S. Economic analysis of albumin- bound paclitaxel for the treatment of metastatic breast cancer. J Oncol Pharm Pract. 2009;15:67-78. Epub 2008 Nov 26.
- Barron JJ, Quimbo R, Nikam PT, Amonkar MM. Assessing the economic burden of breast cancer in a US managed care population. Breast Cancer Res Treat. 2008;109:367-377. Epub 2007 Aug 3.
- Horner MJ, Ries LAG, Krapcho M, et al (eds). SEER Cancer Statistics Review, 1975-2006. National Cancer Institute. Bethesda, MD. http://seer.cancer.gov/statfacts/ html/breast.html. Accessed June 10, 2010.
- Kuderer NM, Dale DC, Crawford J, et al. Mortality, morbidity, and cost associated with febrile neutropenia in adult cancer patients. Cancer. 2006;106:2258-2266.