Objective: To determine actual first-line treatment patterns for metastatic triple-negative breast cancer (mTNBC) in the United States by analyzing records of patients treated in community practices.
Methods: We derived our data from the Georgia Cancer Specialist Database (GCSD) (2003-2008) and the International Oncology Network’s (ION) Treatment and Outcomes Database (2003-2008). In both data sets, patients with stage IV TNBC were selected and followed for up to 1 year after the initial mTNBC diagnosis. Chemotherapy was deemed first-line if initiated within 120 days of mTNBC diagnosis.
Results: The study included 30 and 29 mTNBC patients from GCSD and ION, respectively. In the GCSD sample, 16 patients (53%) were treated with monotherapy, namely capecitabine and the taxanes (44% each of the monotherapy recipients). Fourteen patients received combination therapy, with the taxanes and cyclophosphamide/doxorubicin the most frequently used agents (79% and 36%, respectively). The ION data analysis showed that 38% of patients received monotherapy, most commonly with the taxanes (64% of monotherapy recipients); 62% received combination therapy, with cyclophosphamide/doxorubicin and bevacizumab/paclitaxel being the most frequent concurrent agents (33% for each). We determined ION mTNBC patients had higher treatment costs, due at least in part to the greater use of bevacizumab.
Conclusions: First-line treatment patterns for mTNBC were diverse, reflecting a lack of consensus and an unmet need for standardized therapy. These results will need to be confirmed with data derived from other community settings.
Triple-negative breast cancer (TNBC) is characterized by aggressive tumors whose cells lack estrogen receptors (ERs) and progesterone receptors (PRs), and do not overexpress HER2.1 Approximately 10%-17% of all breast cancers in the United States are triple negative.2 Compared with women diagnosed with other subtypes of breast cancers (non-TNBC) in the United States, patients diagnosed with TNBC are more likely to be younger (<35 y) and African-American.3,4 Patients diagnosed with TNBC also have a higher rate of mortality (42.2% vs 28.0%; P < .0001) and distant recurrence within the first 5 years of diagnosis when compared with non-TNBC patients.1,5 TNBC is further associated with higher tumor grade, increased risk for visceral or cerebral metastasis, and poor survival after recurrence.1,3,5-7
Early-stage TNBC is treated with a combination of surgery, radiation, and neoadjuvant/adjuvant chemotherapy, which can often lead to a good prognosis. However, compared with other breast cancers, TNBC patients experience a higher proportion of distant recurrence (33.9% vs 20.4%; P < .0001).1 Furthermore, prognosis is poor in advanced disease, with a median overall survival of only 13.3 months following diagnosis of metastatic TNBC (mTNBC).5,6
Despite the reported benefits from treatment with chemotherapy, prognosis remains poor for TNBC patients. Furthermore, the shorter survival times may also be affected by the decreased treatment options currently available compared with options for other forms of metastatic breast cancer (mBC).2,4 Standardized therapies such as trastuzumab and lapatinib are available for HER2-expressing breast cancer in the metastatic and adjuvant settings.8,9 Cancers that are ER-positive and/or PR-positive can also be treated with endocrine therapies.8 However, lack of hormone-receptor expression and HER2 overexpression render TNBC resistant to treatment with these agents.10 Consequently, standard cytotoxic chemotherapy remains the preferred treatment for TNBC.8,11
Sequential single-agent chemotherapy is often preferred to combination chemotherapy in the treatment of patients with mBC because of its similar antitumor efficacy and decreased toxicity.8 However, for rapidly progressive visceral disease, improved response rates and time to progression can justify use of a combination strategy, despite greater toxicity.9,12 In HER2-negative mBC, conventional chemotherapy and biologic therapy (Table 1) with the vascular endothelial growth factor (VEGF) ligand inhibitor bevacizumab are potential first-line treatment choices that are often used for mTNBC. When combined with taxane therapy, bevacizumab significantly improved the objective response rate (36.9% vs 21.2%; P < .001) and prolonged median progression-free survival (PFS) (11.8 mo vs 5.9 mo; hazard ratio [HR] = 0.60; P < .001) in first-line treatment of mBC patients, compared with taxane alone.13 Furthermore, taxane/bevacizumab combination therapy displayed greater efficacy than single-agent taxanes in a broad range of breast cancer patient subgroups.14 Bevacizumab in combination with capecitabine, nab-paclitaxel, docetaxel, or anthracycline compared with the chemotherapy agent alone resulted in an increase in median PFS for each combination in a first-line setting in the RIBBON-1 trial.15 There were, however, no significant differences in overall survival. The HR for triple-negative status was 0.78 (95% CI, 0.53-1.15); the HR for non–triple-negative status was 0.61 (95% CI, 0.48-0.77); and the HR for unknown status was 0.52 (95% CI, 0.15-1.78). In the second-line setting, bevacizumab in combination with chemotherapy again resulted in an increase in median PFS from 5.1 to 7.2 months, but there was no significant difference in overall survival.16 In an analysis of the 159 patients in this trial with triple-negative cancer, the median PFS was 6.0 months with bevacizumab/chemotherapy compared with 2.7 months with chemotherapy alone.17 However as the subsequent studies failed to confirm the same degree of benefit observed in the initial trial, the indication of bevacizumab for mBC has been withdrawn by the FDA.18
The debate continues over the merits of sequential monotherapy versus combination therapy, especially given the emergence of innovative, less-toxic formulations. In a nonrandomized study following 631 mBC patients pretreated with at least 2 chemotherapeutic regimens, the antimetabolite capecitabine achieved an overall response rate of 34.7% (N = 345 evaluable patients), with a median time to progression of 6.6 months and median overall survival of 10.0 months.19 A more recent study conducted in mBC patients who received prior adjuvant anthracycline-based chemotherapy (including doxorubicin itself) showed that pegylated liposomal doxorubicin plus cyclophosphamide achieved an overall clinical benefit rate (complete response plus partial response plus stable disease for >6 months) of 71%. Unlike standard doxorubicin, the liposomal formulation appeared to confer no significant cardiotoxicity.20 In a randomized phase II study, gemcitabine in combination with vinorelbine versus sequential monotherapy showed no significant benefit in median overall survival (10.6 mo vs 8.9 mo; P = .436) in mBC.21 Similarly, previous work demonstrated that combination chemotherapy achieved higher response rates and longer times to progression, but with greater toxicity and no prolongation of survival, compared with single-agent therapy.22,23
Several combination therapies have proven to be effective in the first-line treatment of mTNBC. For example, the nontaxane microtubule inhibitor ixabepilone combined with capecitabine yields a higher mTNBC response rate than capecitabine alone. In phase III trials, the combination achieved a 31% overall response rate compared with 15% for capecitabine monotherapy. It also resulted in increased median PFS (4.2 mo vs 1.7 mo, respectively; HR = 0.63).24 These trials took place in a taxane-experienced mBC population. In a partially taxane-naïve population, another metabolite inhibitor, gemcitabine, was added to paclitaxel therapy, resulting in a 35.7% response rate in 14 mTNBC patients with a 6-month median time to progression.25 Nanoparticle albumin-bound paclitaxel administered in combination with bevacizumab and gemcitabine resulted in an overall clinical benefit rate (overall response rate plus stable disease) of 84.6% (11/13 patients with mTNBC) with an 18-month overall survival rate of 82.5% and minimal toxicity.26
In a retrospective review of phase II and III studies in first-line therapy of mBC, carboplatin/docetaxel and carboplatin/paclitaxel combination therapies attained response rates of 53%-62%.27 When administered as first-line treatment in phase II trials, carboplatin with docetaxel resulted in an overall response rate of approximately 60% and a median time to progression of 6.5-9.6 months.28,29 On the other hand, a randomized phase III trial in mBC showed no significant differences in the overall response rates between carboplatin/paclitaxel and single-agent paclitaxel (38% vs 49%; P = .20).30
Evidence suggests that platinum-based combination therapy with cisplatin or carboplatin may enhance response rates in TNBC to a greater extent than in other breast cancer types, though results are inconsistent for metastatic disease.31,32 A small, retrospective study found that mTNBC patients taking cisplatin plus gemcitabine had improved median PFS compared with the patients with other types of mBC (5.3 mo vs 1.7 mo; P = .058).33
Several novel treatments are currently under investigation for the treatment of mTNBC. These include inhibitors of PARP1, a cellular enzyme that plays a central role in repairing single-strand DNA breaks.34 Other agents under investigation for TNBC have biologic targets that include VEGF, mammalian target of rapamycin (mTOR), and epidermal growth factor receptor (EGFR).35
Despite the lack of approved, standardized TNBC-specific therapy, the above discussion identifies some of the chemotherapeutic options available for mTNBC. An evaluation of actual first-line treatment patterns for mTNBC would provide insights into how patients with this clinically aggressive breast cancer type are treated in current practice.
In this study, we sought to identify current patterns of first-line mTNBC chemotherapy by analyzing data from community practices in the United States. We further compared the costs of anticancer drugs administered as first-line therapy between mTNBC and other mBC (metastatic non-TNBC) patients during the first year following diagnosis of stage IV TNBC.
We analyzed de-identified records for patients with breast cancer stored in the Georgia Cancer Specialist Database (GCSD) from 2003-2008 and the International Oncology Network’s Treatment and Outcomes Database (ION) from 2003-2008.
Georgia Cancer Specialists is one of the largest private oncology practices in the United States. It records 160,000 cancer patient visits annually at its offices in north and central Georgia, southeast Tennessee, and southwest North Carolina. The GCSD includes patient demographics (eg, age, sex, race, smoking status); disease diagnosis; staging at diagnosis; chemotherapy protocols including protocol name, line of therapy, and start and end dates of the protocol; longitudinal records of infusion drugs (chemotherapy and nonchemotherapy); outpatient services; outpatient prescriptions; laboratory results; insurance information; and survival status.
ION’s Treatment and Outcomes Database houses longitudinal information on more than 105,000 unique oncology patients from 2003 to 2008. It represents more than 100 private practice oncologists across the United States (Alabama, California, Georgia, Nebraska, North Carolina, Ohio, Texas, Virginia, and Wisconsin). The available data include demographics, diagnoses, staging, death, visit information, treatment plans, medication administered (eg, date of infusion or prescription, drug dosage), and laboratory results. Since these were de-identified patient databases, the study was exempt from institutional board review.
In both data sets, patients with stage IV mTNBC or metastatic non-TNBC were selected and followed for up to 1 year after initial diagnosis. We identified patients as having breast cancer based on the presence of a 174. International Classification of Diseases, Ninth Revision diagnostic code in their records. The selected patients had complete records on their ER, PR, and HER2 status. Those with negative values for all ER, PR, and HER2 statuses were categorized as having mTNBC, while others with at least 1 positive ER, PR, or HER2 were classified as having metastatic non-TNBC. We excluded patients with other primary tumor types prior to the initial stage IV breast cancer date.
Patients’ drug use history was captured for the year following diagnosis of stage IV metastatic disease. In the ION database, all drugs (chemotherapeutic and nonchemotherapeutic) were identified by drug names from the “Agent” data table (drug administration). Drugs were identified by national drug code, J-code, or drug names in the GCSD.
We flagged a particular course of therapy as first line if the first drug was initiated within 120 days following the initial mBC diagnosis. We considered concurrent drugs as part of the first-line combination regimen if they were commenced within 30 days of the first drug. These agents could be chemotherapeutic, targeted (eg, trastuzumab and bevacizumab), or hormonal (tamoxifen, anastrozole, letrozole, exemestane, leuprolide, fulvestrant, and goserelin) drugs, in oral or infusion forms.
First-line therapy was considered terminated when a patient switched to a new chemotherapeutic agent after receiving at least 2 courses of a first-line chemotherapy regimen, or if there was a gap of more than 30 days in administering any of the first-line component drugs.
The average total drug costs for first-line therapy were calculated within the first year after the mBC diagnosis. The total cost of each drug was equal to the year’s total doses multiplied by the drug’s price per dose.
Drug prices were obtained from 2 sources. Prices of infusion drugs and some oral drugs (eg, capecitabine) were based on the average sales price (ASP) marked up by 6%, as obtained from the Centers for Medicare & Medicaid Services website.36 Payment allowance limits subject to the ASP methodology are based on third-quarter 2009 ASP data.
For drugs with no ASP, we applied the wholesale acquisition cost, which was obtained from First DataBank. When an oral drug had multiple national drug codes, a standardized price per unit was created by averaging the prices for the multiple codes.
The study included 59 mTNBC and 155 metastatic non-TNBC patients treated with chemotherapy or hormonal drugs. A total of 30 mTNBC patients were from the GCSD and 29 were from the ION database. Table 2 shows the comparative use of mTNBC chemotherapeutic drugs in the 2 databases, as single agents or in combination.
In GCSD, 16 patients (53%) received monotherapy, with capecitabine (7 patients, or 44%) and 1 of the taxanes being the most frequently used therapy. For 14 patients (47%) receiving combination therapy, cyclophosphamide plus doxorubicin (36%) was the most frequently used regimen. Three patients (21%) received gemcitabine plus paclitaxel. Other combinations used in multiple patients were bevacizumab/paclitaxel and capecitabine/docetaxel.
Different treatment patterns occurred in the ION database sample. Eleven patients (38%) received monotherapy, with taxanes administered to 7 (64%) of those patients. Eighteen (62%) patients received combination regimens, with 6 (33%) patients receiving cyclophosphamide plus doxorubicin and another 6 (33%) patients receiving bevacizumab plus paclitaxel. Four (22%) patients received paclitaxel plus either carboplatin or gemcitabine, while 2 (11%) received combinations containing cyclophosphamide plus 5-fluorouracil.
Single agents more often appeared as first-line therapy in the GCSD than in the ION database (53% vs 38%, respectively). Single-agent capecitabine was used more often as first-line therapy in GCSD patients than in patients documented in ION, whereas taxane use was similar in both databases. The taxanes and capecitabine were the most common single-agent therapies in the GCSD, while taxanes and vinorelbine were the leading single agents in the ION database.
Combination therapy was more frequent in the ION database than in the GCSD (62% vs 47%, respectively). The combination of a taxane with bevacizumab or carboplatin was administered to patients in both databases, but this use occurred more often in the ION database than in the GCSD. Cyclophosphamide plus 5-fluorouracil was used in combination with doxorubicin in GCSD patients, and in combination with epirubicin or methotrexate in the ION population.
Treatment costs were substantially greater in the ION cohort than in the GCSD cohort (Table 3). Compared with the ION mTNBC population, GCSD patients with mTNBC had lower average chemotherapeutic drug costs, but this result was not statistically significant ($17,275 vs $9969, respectively; P = .2265). The more frequent use of bevacizumab in the ION mTNBC cohort (Table 2) likely contributed to the group’s higher cost. The average drug cost for patients treated with bevacizumab was $64,863 (range, $6710 to $137,642).
ION patients with metastatic non-TNBC also had higher average first-line mBC therapy costs than the analogous GCSD patients ($17,779 vs $10,514, respectively; P = .2297). There was a trend in both cohorts toward higher first-line treatment costs in patients with metastatic non-TNBC compared with those with mTNBC (Table 3). However, this difference was not statistically significant (ION P = .3237; GCSD P = .5200).
This study compared the first-line treatment patterns of mTNBC versus metastatic non-TNBC patients from the GCSD and ION databases. In the GCSD sample, patients were frequently treated with capecitabine and taxane monotherapy, and with cyclophosphamide/doxorubicin in combination with taxanes. Analysis of the ION data set showed that taxanes, cyclophosphamide/doxorubicin, and bevacizumab/paclitaxel were the most common single-agent and combination therapies. In addition, ION TNBC patients had higher treatment costs compared with GCSD TNBC patients due partly to the greater use of bevacizumab.
Taken together, this analysis of the GCSD and ION databases suggests that there is no general consensus in the current treatment for stage IV TNBC. Physicians limit their first-line therapeutic choices for mTNBC to the mainstay cytotoxic chemotherapies. Taxanes and capecitabine were the most frequently used monotherapies, while the most common combination therapies contained paclitaxel with bevacizumab, carboplatin, or gemcitabine. The other leading combination was cyclophosphamide and doxorubicin.
The lack of uniform treatment practice is consistent with the lack of specific guidelines for TNBC. The National Comprehensive Cancer Network (NCCN) breast cancer management guidelines currently do not provide recommendations focused on TNBC treatment. Instead, the NCCN leaves physicians to choose from its list of standard chemotherapies.8
Several potential factors may have contributed to the observed divergence in TNBC treatment patterns between the 2 data sets. GCSD represented community practice data in a local region, in which treatment patterns tend to be homogeneous. The ION data were derived from different communities located across the United States, and may reflect more varied patterns of TNBC treatment.
There are several limitations to the current study. The sample size of the data sets was small, which limits the comparison of treatment patterns and cost. Furthermore, the results from the ION and GCSD data sets represent a proxy for the mBC population in the United States, but differences in the 2 databases may have given rise to the observed divergence in treatment pattern. The data in GCSD represent a localized perspective and yield a more complete and in-depth record of medical practice in both hospital and clinical settings. In contrast, the ION data set encompasses populations from more diversified regions, with the data primarily collected from private practices and a lack of hospital representation. In both cases, the data were compiled from electronic medical records that physicians manually entered. Key-in error, in addition to possible missing data, may have also affected our results. The finding that total anticancer drug cost within the first years of disease diagnosis for TNBC was comparable to that for non-TNBC may be due in part to the small sample size of the study and potential confounding factors for which we did not adjust. Consequently, the small sample size may have caused fluctuations in the average drug costs, which in turn may have contributed to the significant differences between bevacizumab and 5-fluorouracil. The costs associated with treatments were not directly obtained from the data sets, which may have limited the results of the study. Prices were derived from ASP or wholesale acquisition cost information. These imputed costs may not completely reflect actual costs. Finally, the definition of first-line chemotherapy was arbitrary. Hence, a change in the criteria for days of chemotherapy and drug initiation, or in the allowable gap for first-line component drugs, might influence the results. Our findings, therefore, need to be confirmed in studies drawing on data from a large, nationally representative population. Nonetheless, this study underscores a current unmet need for standardized, first-line drug therapy administered to mTNBC patients in the United States.
The different mTNBC treatment patterns observed in the 2 databases highlight a lack of standard and/or approved therapy for this heterogeneous and more clinically aggressive breast cancer subtype. We also found a difference in the range and distribution of treatments and costs between the 2 community health databases in this study. Additional studies using a larger sample size and national, representative data are necessary in order to confirm our results.