Cost impact of oral capecitabine compared to intravenous taxane-based chemotherapy in first-line metastatic breast cancer

Abstract

Objective: Few studies have examined the costs associated with differing first-line chemotherapy regimens in patients with metastatic breast cancer (MBC). This study compares the relative cost impact of women starting first-line chemotherapy with capecitabine versus taxanes.

Methods: Women receiving first-line chemotherapy for MBC from 1998 to 2002 were identified from a hybrid North Carolina Medicaid-claims-tumour registry linked database and Medicare records, and were followed through to 2005 with claims data. Statistical t- and chi-square tests were used to compare baseline characteristics between patients who received first-line chemotherapy with capecitabine versus taxanes. Projected mean costs for 12 months continuous eligibility were estimated using an ordinary least squares linear regression. Overall cost impact of capecitabine after start of therapy was then examined using a multivariate log-linear regression model.

Results: While patients starting taxanes had significantly lower total costs in the pre-index year than patients starting capecitabine (mean: $20,042 vs. $35,538, p<0.001), in the post-index year, the patients on taxanes experienced significantly higher healthcare utilisation and associated costs compared to patients on capecitabine (mean: $43,353 vs. $35,842, p=0.0089). The differences were primarily attributable to lower expenses in chemotherapy related claims and fewer visit days to outpatient settings for patients on capecitabine. After adjustment with propensity scores and other confounders, the capecitabine group was associated with 32% lower healthcare costs compared to the taxane group (p=0.0001).

Conclusions: In this population-based study, women who received capecitabine as first-line treatment for MBC had significantly lower costs compared to women starting taxane therapy.

Key words::

• breast cancer
• capecitabine
• cost analyses
• Medicare and Medicaid
• oral chemotherapy

Introduction

Breast cancer is one of the most common forms of cancer among American women. In 2007, the American Cancer Society estimated there were 40,460 female deaths caused by breast cancer in the US, resulting in 15% of total female cancer deaths and ranking second after lung cancer¹. It is estimated that 40–45% of all patients with breast cancer will eventually develop metastatic disease², leading to an increased cost burden on health insurance programmes.

Usually, chemotherapy is applied to treat metastatic breast cancer by using cytotoxic drugs. Capecitabine, an orally-administered prodrug of fluorouracil, was approved by FDA in 1998 for the treatment of patients with metastatic breast cancer (MBC) resistant to paclitaxel and anthracycline-containing therapy. In 2001, combination therapy of capecitabine and docetaxel was approved by the FDA to treat MBC, based on favourable survival data compared to single-agent therapy³. The flexible schedules, ease of administration and tolerable toxicity profile, as well as desirable efficacy in MBC, contribute to the utility of capecitabine in the management of breast cancer^4–7. Recently, more and more clinical studies have focused on the role of capecitabine in first-line treatment for MBC patients, demonstrating promising therapeutic efficacy^8–13. O’Shaughnessy et al ⁸ conducted a randomised phase II trial of capecitabine as first-line therapy for MBC compared with intravenous cyclophosphamide, methotrexate and a 5-fluorouracil (CMF) regimen. The overall objective response rate (ORR) was 30% for capecitabine versus 16% for the CMF regimen. A subsequent randomised trial by Stockler et al ⁹ confirmed the comparative efficacy of capecitabine, delivered by continuous or intermittent schedule, versus CMF and demonstrated significant improvement (p<0.01) in select quality of life (QoL) measures, although global QoL was similar between combined capecitabine and CMF treatment arms. Bajetta et al ¹¹ investigated the efficacy of capecitabine first-line monotherapy in elderly patients. Both the low-dose and high-dose cohorts achieved ORRs of 35% and 37%, respectively. Furthermore, the low-dose cohort appeared to show a greater median survival (16 vs. 10 months). Overall, these studies suggest that first-line capecitabine treatment is active and results in equivalent efficacy or survival time in MBC treatment.

As well as the therapeutic efficacy, pharmacoeconomics are another concern when evaluating the first-line use of capecitabine. Because the FDA has approved capecitabine for second-line use as monotherapy and combination therapy, currently almost all of the cost-impact and cost-effectiveness studies concentrate on this approved field^14–19. Verma has examined the cost-effectiveness of combination therapy for the patients with anthracycline-pretreated MBC^15,16 which showed that costs for treatment of side-effect events were less for patients having capecitabine plus docetaxel therapy than for those having docetaxel as a single agent. Although the total cost per patient of combination therapy was 8.9% higher than that of docetaxel monotherapy, creating a net increase of $1998, potential cost-saving due to lower docetaxel dose and hospital utilisation offset the costs of combination. Similarly, Hornberger et al ¹⁷ reported the total cost of capecitabine and docetaxel combination therapy ($20,701) and docetaxel alone ($19,839). The costs associated with combination therapy were offset by the reduced cost of hospitalisations ($5,167 vs. $5,861). The mean cost per quality-adjusted year of life gained was $5,520. Another pharmacoeconomic study estimated the relative cost effectiveness of four chemotherapies (infusion 5-FU, gemcitabine, vinorelbine and capecitabine)¹⁹. Capecitabine appeared to be the most cost effective with a cost-effectiveness ratio of $1,436 and a marginal cost-effectiveness ratio of $687 per quality-adjusted life month.

As described above, more and more clinical trials have shown supportive and promising results for first-line use of capecitabine. Few studies have examined the costs associated with differing first-line chemotherapy regimens in patients with breast cancer. Thus, there is a need to evaluate the pharmacoeconomics of first-line use of capecitabine to further enhance our understanding. The current study may be the first to examine the relative cost impact on women receiving first-line chemotherapy with capecitabine versus paclitaxel or docetaxel.

Methods

The study was conducted using patient data from a linked database of the North Carolina (NC) Central Cancer Registry and Medicaid claims²⁰, with the addition of Medicare records. The Central Cancer Registry (CCR) database collects detailed data on all cancer cases diagnosed among North Carolina residents, as required by law. Medicaid claims files were collected for all available patients in the study sample through the NC department of Health and Human Services. Medicare information was procured through the Centers for Medicaid and Medicare Services.

Cases consisted of women (aged 18 or above) receiving first-line chemotherapy for metastatic or recurrent breast cancer from 1998 to 2002 and followed through to 2005 with claims data. Cases were also included if they had used either capecitabine or taxanes (but not both) and no prior evidence of first-line chemotherapy 90 days before start of therapy. However, they may have had other concurrent chemotherapy alongside capecitabine or taxane. Concurrent cases were included in the sample because the cost differences between capecitabine and taxane administration could be compared under both mono and concurrent therapy scenarios using multivariate regression modelling, and the interaction between concurrent therapy and capecitabine group membership could be further examined for statistical significance.

Measures used in the study included patient sociodemographic (age, race), eligibility (Medicare only, Medicaid only or dual eligibility), and health characteristics (clinical status and comorbidity). Clinical status measures comprised positive or negative oestrogen receptor (ER) or progesterone receptor (PR), metastasis at diagnosis or recurrent disease, cancer stage at diagnosis and tumour size. Medical comorbidity was evaluated by the D’Hoore Charlson comorbidity score²¹.

Healthcare costs were calculated up to 1 year before and after chemotherapy start date and were further broken down into inpatient, outpatient, durable medical equipment (DME), home health, and Medicaid prescription costs. Medicare prescription costs were not included in the analysis as Medicare Part D had not yet taken effect during the time period of this study. Claims related to hospice and skilled nursing facility services were not included in the total. Inpatient-related costs were calculated over all available claims in Medicare and Medicaid that had date of service occurring during patient stay at hospitals, including costs of hospitalisations and treatment during stay.

Medicare and Medicaid outpatient claims were calculated over periods not in inpatient stay and consisted of all claims submitted by outpatient institutions, clinics and non-institutional providers. Total healthcare costs were assumed to capture costs resulting from chemotherapy treatment in addition to other healthcare, including treatment costs, hospitalisations, and supportive care like antiemetics and premedications for infusions.

Using logistic regression, a propensity model was first used to predict membership in the capecitabine group. Predictors included age, race, tumour size, Medicare/Medicaid insurance status, stage, Charlson comorbidity score, ER/PR negative status, healthcare utilisation variables (outpatient, inpatient, medication charges and number of visits), number of months with claims during pre-year, prior adjuvant hormone therapy for metastatic disease, time from initial diagnosis to recurrence, and time from estimated date of recurrence to start of capecitabine/taxane chemotherapy.

Statistical t- and chi-square tests were used to compare baseline characteristics between patients who received capecitabine versus taxanes (paclitaxel or docetaxel). Because patients may have been eligible in Medicare or Medicaid for a fraction of a year either pre- or post-start of treatment, it became necessary to adjust cost data for partial eligibility and/or drop-out. After adjusting for inflation to 2005 dollars, predicted mean costs were projected to 12 months of continuous eligibility by using an ordinary least squares (OLS) linear regression with cost as a dependent variable, capecitabine/taxane indicator as an independent variable and number of months eligible in Medicare/Medicaid as a covariate.

The total cost impact of capecitabine was further examined by using a multivariate regression least squares model with log costs in the year after start as the outcome, as described in reference 22. Capecitabine/taxane indicator, propensity scores, sociodemographics, number of months with claims in order to adjust for partial eligibility and other potential confounders were included as predictors. Regression diagnostics including investigation of the standardised residuals were used to investigate possible violation of modelling assumptions. Furthermore, empirical standard errors adjusting for possible model misspecification were used to assess variability of parameter estimates. All analyses were conducted using SAS version 9.1.

Results

Inclusion criteria for the study population are shown in Figure 1. The initial sample size consisted of 18,859 female patients diagnosed with breast cancer in North Carolina from 1998 to 2002 and followed until 2005. Among the patients diagnosed as having MBC or with an indication of a recurrence (n=2,236), 733 patients were found initiating capecitabine or taxane, but had no chemotherapy 90 days before initiation of treatment and did not have both first-line capecitabine or taxane present. Of these patients, 114 started with capecitabine (n=114) and 619 started with taxanes (n=619).

Table 1 illustrates the patient characteristics of the final sample. There was no significant difference between treatment groups in age, race, tumour size, ER/PR negative status, cancer stage at diagnosis, Charlson comorbidity score and metastasis at diagnosis. However, insurance status was found to vary between groups (p=0.0004), with the percentage of Medicare-only patients significantly lower in the capecitabine group (46%) than in the taxane group (63%). Number of months eligible in Medicaid or Medicare during the year after start was also significant, with the capecitabine showing a mean of 8.33 months compared to taxane with 10.10 months (p<0.0001).

Table 2 presents the unadjusted combined costs adjusted for inflation up to 2005 before and after the year of starting the treatment for each group in addition to a breakdown of total health costs into its constituent components. Patients starting taxanes (mean: $20,042) had significantly lower healthcare costs (p<0.001) in the pre-index year compared to those starting capecitabine (mean: $35,538). However, after start of treatment, patients on taxanes experienced significantly higher healthcare costs (mean: $43,353) compared to capecitabine (mean: $35,842, p=0.0089). The increase was primarily attributable to outpatient costs (capecitabine mean: $21,206 vs. taxane mean: $33,200, p<0.0001).

Table 3 shows the top outpatient/durable medical equipment/and home health HCPCS (Healthcare Common Procedure Coding System) codes as determined from the Medicaid/Medicare dataset together with their mean associated costs. Overall unadjusted costs for taxane-specific chemotherapy administration (docetaxel and paclitaxel) combine to a mean of $5,275 per patient compared to capecitabine specific mean of $3,122 (p<0.0001). Additional chemotherapy agents associated with the taxane group have substantially higher mean costs compared to agents in the capecitabine group.

Comparison of pre-year cost, utilisation, and demographic characteristics between capecitabine and taxane groups, after adjusting for propensity score, suggest that the propensity score was able to correct for discrepancies in Medicare, Medicaid, and dual insurance status between groups (p>0.10). However, propensity-adjusted imbalances are not fully eliminated with regard to pre-year prescription fills (capecitabine 19.8 vs. taxane 12.2, p=0.0171) and pre-year outpatient costs (capecitabine $18,981 vs. taxane $10,025 p<0.0001).Thus in order to address possible imbalance in cost and resource utilisation between the groups during the pre-year, total pre-year direct costs were added into the multivariate regression.

Table 4 describes the relationship between total health costs and first-line use of capecitabine. The regression model explains about 22% of the total variance in log total costs. Standardised residual plots conformed to normality assumptions and suggested no gross functional misspecification of the model. After adjusting for using the propensity score plus other confounders, capecitabine was associated with 32.0% lower healthcare costs compared to taxane (p=0.0001). Other significant predictors included monotherapy, which was associated with a 36% decrease in healthcare costs (p<0.001), insurance status (p=0.01), Charlson comorbidity score (p<0.01), total costs in year before starting treatment (p<0.001), and months eligible in Medicare/Medicaid during year after treatment (p <0.001).

Discussion

The economic burden of cancer is significant and was likely to be in the range of $300–400 billion in 2001 on a worldwide basis, among which $100–140 billion were projected as direct costs²³. Breast cancer represents an important part of the total financial resources allocated to treat cancer. In the US, the annual costs for a breast cancer patient were projected to be at least $12,828 higher than those for women without breast cancer in a managed-care population²⁴.

This is a population-based, descriptive study of the cost impact of capecitabine as first-line chemotherapy. The study found that women who received capecitabine as first-line treatment for MBC had significantly lower costs compared with women starting taxane therapy. The differences were primarily attributable to lower expenses in chemotherapy-related claims and fewer visit days to outpatient settings for patients on capecitabine. Our results are consistent with the cost-effectiveness study of capecitabine plus docetaxel combination therapy in patients with anthracycline-pretreated advanced breast carcinoma¹⁵. In this study, cost savings were due to reduced docetaxel dose and hospital utilisation resulting from the adverse events compared with that of single-agent docetaxel. On the other hand, capecitabine was less likely to be used by MBC patients than taxane. This may represent a general pattern of underuse, or issues more specifically related to oral chemotherapy agents. Capecitabine usage may have been driven by clinical factors like prior adjuvant therapy, prior hormonal therapy, extent of metastatic disease, performance status, and availability of intravenous access, in addition to the issues raised by oral agents in general. Healthcare system access and elderly patients with chronic conditions may also affect the use pattern of capecitabine.

Health insurance status was seen to predict total healthcare costs. Compared to patients with Medicare only, patients with Medicaid alone and dual status showed higher healthcare costs. Potentially this may represent the differences of healthcare system access or residential setting, general pattern of chemotherapy and overall treatment burden for elderly patients with various chronic conditions. However, the significant difference may be in part affected by missing prescription cost information in our Medicare records during the time frame of the study, as Medicaid recipients have prescription costs available during this time period, but Medicare does not have full prescriptions costs. Drug prescription benefit claims under Medicare part D have only become available since 2006.

Given that capecitabine administration is delivered orally and is liable to be acquired via the pharmacy counter, whereas taxane administration is as a rule intravenously administered by a healthcare provider, there is the concern that the difference in costs between taxane and capecitabine may be also affected by missing prescription cost information for the Medicare-only patients. This would imply that capecitabine acquisition and administration costs shown in this paper may be greater than stated; a projection from this study’s data suggests that capecitabine costs are understated by at most 20%. However, this concern is tempered by knowledge that the interaction between insurance type and capecitabine group membership is not significant (p=0.2268), providing no evidence that the capecitabine effect varies within insurance strata. Secondly, capecitabine costs are partially captured by Medicare as capecitabine-related claims submitted by non-institutional providers and DME suppliers.

It is worth noting that in the year after the start of first-line therapy, the discrepancy between outpatient costs for taxane compared to capecitabine seems impressive (median $21,715 vs. $7,953). Cost increases may not only be due to taxane-related services but, when additional chemotherapy agents are concurrently delivered, these agents are more expensive than when administered in the capecitabine group, as suggested by Table 3. These increased expenses seem to be attributable to increases in frequency of visits or increased scheduling and may indicate a relative benefit of capecitabine use resulting in physician visit savings. There is no evidence that the presence of multiple therapy modifies the cost difference between capecitabine and taxane group, since the interaction between multiple therapy and capecitabine group is not significant (p=0.3164).

Traditionally, in terms of chemotherapy, most patients tend to receive intravenous treatments rather than oral administration. However, intravenous therapy places a burden on patients’ daily lives. Patients spend a lot of time travelling to, waiting for and receiving therapy, which increases the costs of the treatment indirectly. With the development of oral anticancer drugs, oral chemotherapy is becoming an attractive and promising strategy with its convenience and ease of administration. Chau et al ²⁵ investigated the pharmacoeconomic impact of oral capecitabine as home-based therapy and demonstrated savings in healthcare utilisation and in costs compared with intravenous 5-FU. Therefore capecitabine, as an oral chemotherapy option for breast cancer, has a favourable financial impact on patients and payers.

Limitations

Cases in the study were treated in one state (North Carolina), affecting generalisability of results to other regions. Some patients may not have been fully eligible in Medicaid/Medicare 1 year before or 1 year after starting capecitabine or taxanes, leading to biased costs before or after the index year. However, as a way to correct for this potential bias, the number of months with claims during the pre-year was used in the propensity model, and furthermore the number of months with claims the during post-year was used as a potential confounder in the multivariate regression model. Moreover, the database does not capture additional utilisation made through out-of-pocket, or via commercial insurance, which may lead to additional bias in our cost estimates and comparisons. Although the database’s sample may be relatively small, given the availability of SEER-Medicare data, which would provide larger sample sizes, the authors believe that this limitation is offset by the dataset’s link to Medicaid claims information, from which the capecitabine oral prescription stream. Administration can be captured through the Medicaid pharmacy component which was not readily available in Medicare during the time periods studied. Lastly, due to the altered reimbursement encoded in the Medicare Modernization Act of 2003, the conclusion of the current study may be less relevant to today’s environment. Additionally, the availability of generic paclitaxel may affect the calculations from this historical dataset.

Conclusions

In this population-based study, women (with NC Medicaid or Medicare) who received capecitabine as first-line treatment for metastatic breast cancer had significantly lower costs compared to women starting taxane therapy.

Figure 1. Study population.

Table 1. Description of study population (n=733) (taxane n=619, capecitabine n=114).

Study population	Capecitabine		Taxanes		t-test/χ^2p-value
	n	%	n	%
Age: mean, years (SD)		60.5 (15.4)		61.6 (12.5)	0.4915
Race					0.9666
White	77	67.54	420	67.85
African-American	36	31.58	192	31.02
Other	1	0.88	7	1.13
Tumour size 20 mm plus	59	51.75	345	55.74	0.4323
ER/PR negative*	31	27.19	155	25.04	0.6274
Insurance					0.0004
Medicaid only	30	26.32	117	18.90
Medicare only	52	45.61	388	62.68
Dual	30	28.07	114	18.42
Stage^†					0.5779
Distant	21	18.42	117	18.90
Local	29	25.44	185	29.89
Regional	57	50.00	293	47.33
Undetermined	7	6.14	24	3.88
Charlson comorbidity score: mean (SD)		2.40 (2.38)		2.24 (2.32)	0.4878
Metastatic at diagnosis	21	18.42	117	18.90	0.9040
First-line treatment					–
Capecitabine monotherapy	85	74.56	–	–
Capecitabine combination therapy	29	25.44	–	–
Taxane^‡ monotherapy	–	–	261	42.16
AT^§ combination therapy	–	–	39	6.30
ACT^¶ combination therapy	–	–	69	11.15
Other/undetermined chemotherapy	–	–	250	40.39
Months eligible, after start: mean (SD)		8.33 (3.57)		10.10 (3.00)	<0.0001

*ER/PR: estrogen receptor/progesterone receptor; ^†Stage: Surveillance Epidemiology and End Results (SEER) summary cancer stage at diagnosis (2000); ^‡Taxanes: paclitaxel or docetaxel; ^§AT: anthracycline and taxanes observed in first line; ^¶ACT: anthracycline, cyclophosphamide and taxanes observed in first line.

Table 2. Annual costs of women with MBC before and after starting first-line capecitabine and taxane chemotherapy.

Annual costs (US$)*	Capecitabine		Taxanes		p-value^†
	Adjusted mean (SE)	Median	Adjusted mean (SE)	Median
Prior to chemotherapy
Total costs	35,538 (2,068)	19,416	20,042 (1,017)	12,992	<0.001
After chemotherapy
Total costs	35,842 (2,847)	21,833	43,353 (1,271)	31,558	0.0089
Inpatient costs	7,070 (1,333)	4,785	6,953 (595)	2,269	0.9306
Outpatient costs	21,206 (2,182)	7,953	33,200 (974)	21,715	<0.001
DME costs	2,374 (225)	1,096	591 (100)	29	<0.001
Prescription costs(Medicaid)	3,633 (380)	491	1,527 (170)	0	<0.001
Home health costs	1,559 (310)	0	1,082 (138)	0	0.1270

Total costs=inpatient + outpatient costs (includes MD/other outpatient) + DME + Medicaid prescription costs + home health costs.

*Costs were adjusted for inflation of the year of 2005 in US dollars. Furthermore, projected means for 12-month continuous eligibility were estimated by using an OLS linear regression with costs as dependent variables, and capecitabine/taxane and months eligible in Medicaid/Medicare as independent variables. Unmodified median costs are shown.

^†p-values compare capecitabine/taxane means from OLS model.

DME, durable medical equipment.

Table 3. Top six outpatient/DME/home health services by treatment group during year.

HCPCS codes	Capecitabine			HCPCS codes	Taxane
	Description	Mean costs ($)	Mean no. claims		Description	Mean costs ($)	Mean no. claims
*	Capecitabine	3,122	5.2	J9265	Paclitaxel 30 mg	2,660	4.3
J3487	Zometa 1 mg	891	1.5	J9170	Docetaxel 20 mg	2,615	3.0
J9355	Trastuzumab 10 mg	360	0.3	J9355	Trastuzumab 10 mg	2,613	3.6
Q0136	EPO alpha	312	0.8	Q0136	EPO alpha	1,107	3.3
J0880	Darbepoetin 5 μg	245	0.4	J0880	Darbepoetin 5 μg	593	1.0
J9001	Doxorubicin HCl 10 mg	181	0.07	77413	Radiation Tx	474	4.4

*Capecitabine services based on claims with prescription costs and HCPCS administration codes J5820, J5821.

Costs adjusted for inflation to US$2005.

HCPCS, Healthcare Common Procedure Coding System.

Table 4. Relationship between total health costs at year 2 and the use of capecitabine as first-line therapy.

Dependent variable →	Natural logarithm total costs
Independent variables ↓	Coefficient estimates (SD)
Intercept	9.04 (0.17)***
First-line use of capecitabine	–0.39(0.10)***
Age (>65 years)	0.073(0.09)
White race	–0.035 (0.08)
Tumour size 20 mm plus	–0.053 (0.07)
Medicaid/Medicare dual	0.28 (0.10)**
Medicaid only	0.25 (0.12)*
Medicare only	REF
Charlson comorbidity score	0.048 (0.016)**
Monochemotherapy	–0.45 (0.071)***
Total costs in the year before starting treatment	0.0000071 (0.0000017)***
Metastasis at diagnosis	0.11 (0.089)
Propensity score probability	0.13 (0.035)
Number of months eligible	0.086 (0.011)***
Adjusted R^2	0.22

*p<0.05, **p<0.01, ***p<0.001.

Reference groups: first-line use of taxanes, Medicare only, combination therapy.

Costs were adjusted for inflation in US$2005.

Acknowledgement

Declaration of interest: This study was funded by Roche as a supplemental study to an R01CA12137 on adjuvant treatment for breast cancer in North Carolina (R. Anderson, PI).

Drs Anderson, Balkrishnan, and Kimmick are paid consultants for Roche Pharmaceuticals. Dr Wei was an employee of Roche Pharmaceuticals at the time of study conduct. Dr Wu and Mr Camacho have no conflicts of interests to declare.