[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure.
Patients With Early-Stage Breast Cancer Receiving Chemotherapy
Patients With Early-Stage Breast Cancer Receiving Chemotherapy
Table 1.  
Baseline Characteristics of the Study Population by Use of the Recurrence Score (RS) Assaya
Baseline Characteristics of the Study Population by Use of the Recurrence Score (RS) Assaya
Table 2.  
Baseline Characteristics of the Study Population by Receipt of Chemotherapya
Baseline Characteristics of the Study Population by Receipt of Chemotherapya
Table 3.  
Factors Associated With Receipt of Chemotherapy in 44 044 Patients
Factors Associated With Receipt of Chemotherapy in 44 044 Patients
Table 4.  
Factors Associated With Receipt of Chemotherapy Among 11 135 Patients 70 Years and Younger
Factors Associated With Receipt of Chemotherapy Among 11 135 Patients 70 Years and Younger
1.
Theriault  RL, Carlson  RW, Allred  DC,  et al. NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines)—Breast Cancer, Version 3.2013. http://www.nccn.org/professionals/physician_gls/f_guidelines.asp. Accessed April 17, 2015.
2.
Ademuyiwa  FO, Miller  A, O’Connor  T,  et al.  The effects of oncotype DX recurrence scores on chemotherapy utilization in a multi-institutional breast cancer cohort.  Breast Cancer Res Treat. 2011;126(3):797-802.PubMedGoogle ScholarCrossref
3.
Paik  S, Tang  G, Shak  S,  et al.  Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer.  J Clin Oncol. 2006;24(23):3726-3734.PubMedGoogle ScholarCrossref
4.
Arango  BA, Rivera  CL, Glück  S.  Gene expression profiling in breast cancer.  Am J Transl Res. 2013;5(2):132-138.PubMedGoogle Scholar
5.
Kelly  CM, Krishnamurthy  S, Bianchini  G,  et al.  Utility of oncotype DX risk estimates in clinically intermediate risk hormone receptor-positive, HER2-normal, grade II, lymph node-negative breast cancers.  Cancer. 2010;116(22):5161-5167.PubMedGoogle ScholarCrossref
6.
Hornberger  J, Chien  R, Krebs  K, Hochheiser  L.  US insurance program’s experience with a multigene assay for early-stage breast cancer.  Am J Manag Care. 2011;17(5 Spec No):e194-e202.PubMedGoogle Scholar
7.
Tang  G, Shak  S, Paik  S,  et al.  Comparison of the prognostic and predictive utilities of the 21-gene Recurrence Score assay and Adjuvant! for women with node-negative, ER-positive breast cancer: results from NSABP B-14 and NSABP B-20.  Breast Cancer Res Treat. 2011;127(1):133-142.PubMedGoogle ScholarCrossref
8.
Eiermann  W, Rezai  M, Kümmel  S,  et al.  The 21-gene recurrence score assay impacts adjuvant therapy recommendations for ER-positive, node-negative and node-positive early breast cancer resulting in a risk-adapted change in chemotherapy use.  Ann Oncol. 2013;24(3):618-624.PubMedGoogle ScholarCrossref
9.
Hornberger  J, Cosler  LE, Lyman  GH.  Economic analysis of targeting chemotherapy using a 21-gene RT-PCR assay in lymph-node-negative, estrogen-receptor-positive, early-stage breast cancer.  Am J Manag Care. 2005;11(5):313-324.PubMedGoogle Scholar
10.
Reed  SD, Dinan  MA, Schulman  KA, Lyman  GH.  Cost-effectiveness of the 21-gene recurrence score assay in the context of multifactorial decision making to guide chemotherapy for early-stage breast cancer.  Genet Med. 2013;15(3):203-211.PubMedGoogle ScholarCrossref
11.
Rouzier  R, Pronzato  P, Chéreau  E, Carlson  J, Hunt  B, Valentine  WJ.  Multigene assays and molecular markers in breast cancer: systematic review of health economic analyses.  Breast Cancer Res Treat. 2013;139(3):621-637.PubMedGoogle ScholarCrossref
12.
Lo  SS, Mumby  PB, Norton  J,  et al.  Prospective multicenter study of the impact of the 21-gene recurrence score assay on medical oncologist and patient adjuvant breast cancer treatment selection.  J Clin Oncol. 2010;28(10):1671-1676.PubMedGoogle ScholarCrossref
13.
Geffen  DB, Abu-Ghanem  S, Sion-Vardy  N,  et al.  The impact of the 21-gene recurrence score assay on decision making about adjuvant chemotherapy in early-stage estrogen-receptor-positive breast cancer in an oncology practice with a unified treatment policy.  Ann Oncol. 2011;22(11):2381-2386.PubMedGoogle ScholarCrossref
14.
Surveillance, Epidemiology, and End Results Program. Overview of the SEER program. http://seer.cancer.gov/about/overview.html. Accessed April 17, 2015.
15.
Dinan  MA, Mi  X, Reed  SD, Hirsch  BR, Lyman  GH, Curtis  LH.  Initial trends in the use of the 21-gene recurrence score assay for patients with breast cancer in the Medicare population, 2005-2009.  JAMA Oncol. 2015;1(2):158-166.Google ScholarCrossref
16.
Albain  KS, Barlow  WE, Shak  S,  et al; Breast Cancer Intergroup of North America.  Prognostic and predictive value of the 21-gene recurrence score assay in postmenopausal women with node-positive, oestrogen-receptor-positive breast cancer on chemotherapy: a retrospective analysis of a randomised trial.  Lancet Oncol. 2010;11(1):55-65.PubMedGoogle ScholarCrossref
17.
Dinan  MA, Curtis  LH, Hammill  BG,  et al.  Changes in the use and costs of diagnostic imaging among Medicare beneficiaries with cancer, 1999-2006.  JAMA. 2010;303(16):1625-1631.PubMedGoogle ScholarCrossref
18.
Klabunde  CN, Potosky  AL, Legler  JM, Warren  JL.  Development of a comorbidity index using physician claims data.  J Clin Epidemiol. 2000;53(12):1258-1267.PubMedGoogle ScholarCrossref
19.
Wennberg  JE, Fisher  ES, Goodman  DC, Skinner  JS.  Tracking the Care of Patients With Severe Chronic Illness: The Dartmouth Atlas of Health Care 2008. Lebanon, NH: Dartmouth Institute for Health Policy and Clinical Practice; 2008.
20.
Fairfield  KM, Lucas  FL, Earle  CC, Small  L, Trimble  EL, Warren  JL.  Regional variation in cancer-directed surgery and mortality among women with epithelial ovarian cancer in the Medicare population.  Cancer. 2010;116(20):4840-4848.PubMedGoogle ScholarCrossref
21.
Polsky  D, Armstrong  KA, Randall  TC,  et al.  Variation in chemotherapy utilization in ovarian cancer: the relative contribution of geography.  Health Serv Res. 2006;41(6):2201-2218.PubMedGoogle ScholarCrossref
22.
Brooks  GA, Li  L, Sharma  DB,  et al.  Regional variation in spending and survival for older adults with advanced cancer.  J Natl Cancer Inst. 2013;105(9):634-642.PubMedGoogle ScholarCrossref
23.
Davidson  JA, Cromwell  I, Ellard  SL,  et al.  A prospective clinical utility and pharmacoeconomic study of the impact of the 21-gene Recurrence Score® assay in oestrogen receptor positive node negative breast cancer.  Eur J Cancer. 2013;49(11):2469-2475.PubMedGoogle ScholarCrossref
24.
Oratz  R, Paul  D, Cohn  AL, Sedlacek  SM.  Impact of a commercial reference laboratory test recurrence score on decision making in early-stage breast cancer.  J Oncol Pract. 2007;3(4):182-186.PubMedGoogle ScholarCrossref
25.
Asad  J, Jacobson  AF, Estabrook  A,  et al.  Does oncotype DX recurrence score affect the management of patients with early-stage breast cancer?  Am J Surg. 2008;196(4):527-529.PubMedGoogle ScholarCrossref
26.
Henry  LR, Stojadinovic  A, Swain  SM, Prindiville  S, Cordes  R, Soballe  PW.  The influence of a gene expression profile on breast cancer decisions.  J Surg Oncol. 2009;99(6):319-323.PubMedGoogle ScholarCrossref
27.
Early Breast Cancer Trialists’ Collaborative Group (EBCTCG).  Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials.  Lancet. 2005;365(9472):1687-1717.PubMedGoogle ScholarCrossref
28.
Paik  S, Shak  S, Tang  G,  et al.  A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer.  N Engl J Med. 2004;351(27):2817-2826.PubMedGoogle ScholarCrossref
29.
Harris  L, Fritsche  H, Mennel  R,  et al; American Society of Clinical Oncology.  American Society of Clinical Oncology 2007 update of recommendations for the use of tumor markers in breast cancer.  J Clin Oncol. 2007;25(33):5287-5312.PubMedGoogle ScholarCrossref
30.
Giordano  SH, Lin  YL, Kuo  YF, Hortobagyi  GN, Goodwin  JS.  Decline in the use of anthracyclines for breast cancer.  J Clin Oncol. 2012;30(18):2232-2239.PubMedGoogle ScholarCrossref
31.
Warren  JL, Klabunde  CN, Schrag  D, Bach  PB, Riley  GF.  Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population.  Med Care. 2002;40(8)(suppl):IV-3-IV-18.PubMedGoogle Scholar
Original Investigation
November 2015

Association Between Use of the 21-Gene Recurrence Score Assay and Receipt of Chemotherapy Among Medicare Beneficiaries With Early-Stage Breast Cancer, 2005-2009

Author Affiliations
  • 1Duke Clinical Research Institute, Duke University School of Medicine, Durham, North Carolina
  • 2Duke Cancer Institute, Duke University School of Medicine, Durham, North Carolina
  • 3Department of Medicine, Duke University School of Medicine, Durham, North Carolina
  • 4Hutchinson Institute for Cancer Outcomes Research, Fred Hutchinson Cancer Research Center, Seattle
  • 5Division of Medical Oncology, Department of Medicine, University of Washington School of Medicine, Seattle
 

Copyright 2015 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.

JAMA Oncol. 2015;1(8):1098-1109. doi:10.1001/jamaoncol.2015.2722
Abstract

Importance  Guidelines recommend consideration of chemotherapy for most patients with early-stage, estrogen receptor–positive, invasive breast cancer in the absence of additional prognostic information. The 21-gene recurrence score (RS) assay has been shown in limited academic settings to reduce physician recommendations for adjuvant chemotherapy. Associations between the adoption of the assay and receipt of chemotherapy in the general population have not been examined.

Objective  To examine whether adoption of the RS assay in a nationally representative sample of patients with early-stage breast cancer was associated with use of chemotherapy.

Design, Setting, and Participants  Retrospective cohort study of Medicare beneficiaries who received a diagnosis of incident breast cancer between 2005 and 2009 using Surveillance, Epidemiology, and End Results data set with linked Medicare claims.

Main Outcomes and Measures  Receipt of chemotherapy within 12 months after diagnosis.

Results  A total of 44 044 patients had low-risk (24.0%), intermediate-risk (51.3%), or high-risk disease (24.6%, lymph node positive) as defined by National Comprehensive Cancer Network (NCCN) guidelines and met the study criteria. We observed no overall association between receipt of the RS assay and chemotherapy (odds ratio [OR], 1.03 [99% CI, 0.88-1.19]). In multivariable analysis, there was a significant interaction between NCCN risk and use of the RS assay, with assay use associated with lower chemotherapy use in high-risk patients (OR, 0.36 [99% CI, 0.26-0.50]) and greater chemotherapy use in low-risk patients (OR, 3.71 [99% CI, 2.30-5.98]), compared with no receipt of the assay (P = .006 for the overall interaction). Results were similar in prespecified subgroup analyses of patients 70 years and younger, with the exception of a shift toward lower chemotherapy use during 2008 (OR, 0.90 [99% CI, 0.77-.1.05]; P = .09) and 2009 (OR, 0.81 [99% CI, 0.66-0.99]; P = .007). In unadjusted analyses, overall chemotherapy use decreased over time in patients 70 years or younger with high-risk disease and those receiving the assay.

Conclusions and Relevance  The impact of the adoption of the RS assay on receipt of chemotherapy was strongly population dependent and was associated with relatively lower chemotherapy use in groups with high-risk disease and relatively higher chemotherapy use in patients with low-risk disease. Overall use of chemotherapy decreased during the study period in patients who were most likely to receive chemotherapy.

Introduction

National Comprehensive Cancer Network (NCCN) guidelines recommend consideration of adjuvant chemotherapy in estrogen receptor (ER)–positive, node-negative breast cancer for all but the smallest tumors.1 However, clinicopathologically similar tumors can have different metastatic potential,2 raising the need to reconsider the relative benefits and risks of chemotherapy across risk groups. The Oncotype DX 21-gene recurrence score (RS) assay is an independent predictor of prognosis and the benefit of chemotherapy in tamoxifen-treated, ER-positive, node-negative breast cancer.3-7 Several studies have suggested that use of the assay is cost saving or cost-effective, compared with conventional approaches,6,8-10 possibly as a result of more appropriate allocation of chemotherapy to patients most likely to benefit.11

Patients’ treatment decisions can vary substantially depending on the patient’s age, comorbid conditions, finances, and personal preferences and priorities. Studies of patients’ treatment decisions after use of the RS assay have found that many patients do not follow guideline-based or physician-recommended treatment.2,12,13 In addition, even with a low or intermediate genomic risk score, many physicians may still be reluctant to withhold chemotherapy, particularly in younger patients.2

Current evidence suggests that use of the RS assay reduces the likelihood of physicians recommending chemotherapy.2,6,9,12 However, it is unknown whether adoption of the assay has reduced overall chemotherapy use in the general population. Therefore, we examined associations between use of the RS assay and receipt of chemotherapy in a nationally representative cohort of patients who received a diagnosis of incident breast cancer between 2005 and 2009. We hypothesized that adoption of the assay was associated with lower odds of receiving chemotherapy.

Box Section Ref ID

At a Glance

  • We examined whether adoption of the 21-gene recurrence score (RS) assay in a nationally representative sample of Medicare beneficiaries with early-stage breast cancer was associated with use of chemotherapy.

  • A total of 44 044 eligible patients had low- (24.0%), intermediate- (51.3%), or high-risk disease (24.6%, lymph node–positive) as defined by National Comprehensive Cancer Network guidelines.

  • Assay use was associated with lower chemotherapy use in high-risk patients (OR, 0.36 [99% CI, 0.26-0.50]) and greater use in low-risk patients (OR, 3.71 [99% CI, 2.30-5.98]), compared with no assay use.

Methods
Data Source

Data were from a Surveillance, Epidemiology, and End Results (SEER) and Medicare linked data set, a collaborative effort between the National Cancer Institute and the Centers for Medicare & Medicaid Services. The SEER data represent approximately 28% of the US population with cancer.14 We used SEER data from 2005 through 2009 to identify patients’ clinical and demographic characteristics. We used Medicare claims data from 2004 through 2010 to confirm Medicare enrollment and health care resource use. We used claims from the year before diagnosis to identify comorbid conditions. The institutional review board of the Duke University Health System approved this study. Informed consent was waived due to the retrospective nature of the study.

Study Population

We identified all patients who had a diagnosis of breast cancer between 2005 and 2009 in the 12 SEER registries that were continuously active from 2000 onward. To identify patients likely to have complete Medicare claims related to breast cancer management, we required that patients have a primary diagnosis of breast cancer (International Classification of Diseases, Ninth Revision, Clinical Modification code 174.x or 175.x) on an inpatient, outpatient, durable medical equipment, hospice, home health, or carrier-based Medicare claim between 2 months before and 4 months after the SEER-reported diagnosis. We included patients 66 years or older at diagnosis, except those whose diagnosis was made at autopsy.

We excluded patients who had discontinuous Medicare Part A and Part B coverage (ie, fee-for-service Medicare) or were enrolled in Medicare Part C (managed care) from the year before diagnosis through the end of the study period or death. As a result, all patients in the study survived at least 1 year after initial breast cancer diagnosis. We also excluded patients with in situ disease because chemotherapy and use of the RS assay are not recommended for this group. We excluded patients with metastatic disease given the lack of RS assay utility in this group. We limited the analysis to patients with ER-positive tumors, in which the RS assay has been clinically validated and for which its use was initially recommended. Although the initial studies of the RS assay were limited to patients with node-negative disease, we previously observed use of the RS assay in node-positive disease in this cohort,15 and it has since been reported that the RS assay may predict disease-free and overall survival in these patients.16 For these reasons, we elected to include patients with NCCN high-risk disease, including patients with node-positive disease, in our study.

Outcomes

The primary study outcome was receipt of chemotherapy (Current Procedural Terminology [CPT] code Q0083-Q0085, G0355-G0363, J8510-J9999, or 96400-96549) reported on any outpatient or carrier claim between 2 months before and 12 months after diagnosis.17 The primary exposure of interest was use of the RS assay as reported on a carrier claim between 2 months before and 6 months after diagnosis. To identify claims for the RS assay, we rank-ordered each patient’s claims containing CPT code 84999 (ie, unlisted chemistry procedure) from the highest to lowest payment amounts and by the claim date. We then retained the first record, which represented the most costly, earliest procedure. We reviewed these records manually to confirm that those with the highest payment in each year were performed by the same provider (eg, Genomic Health) and reimbursed by the same Medicare carrier (eg, Palmetto GBA), indicating claims for the RS assay. We counted only 1 assay claim per patient. If a patient had multiple claims, we used the claim with the highest line payment on the first submitted claim date. If a patient had only zero-payment claims, we used the first claim and flagged it as not being reimbursed.

Clinical variables, including TNM staging classifications, were available through the SEER Patient Entitlement and Diagnosis Summary. We categorized patients with missing node status as having node-negative disease and patients with microscopic nodal disease (N1mic) as having N1 disease. We used the most recent NCCN guidelines, which are nearly identical to the 2007 guidelines, to categorize patients into 3 risk groups: low, intermediate, and high.1 Low-risk patients were those with ER-positive, node-negative tumors either 0.5 cm or smaller in size or no more than than 1.0 cm with no unfavorable histologic features as assessed by the site-specific extent of disease (ie, Paget disease, invasive disease, inflammatory disease, G3 or anaplastic histopathologic subtype, T4 disease). Intermediate-risk patients were those with ER-positive, node-negative tumors either 0.6 to 1.0 cm in size with unfavorable features or greater than 1.0 cm in size. High-risk patients were those with node-positive disease. Although patients with ER- and progesterone receptor–negative or human epidermal growth factor receptor 2 (HER2)–positive tumors greater than 1.0 cm in size are also considered high risk, our study was limited to ER-positive tumors and HER2 status was unavailable. We used the SEER Patient Entitlement and Diagnosis Summary to obtain demographic variables, including age at diagnosis, sex, race, ethnicity, marital status, and local census tract characteristics. We grouped the SEER registries by census region. We used Medicare inpatient, outpatient, and carrier claims for the year before diagnosis to detect National Cancer Institute comorbid conditions.18

Statistical Analysis

We describe the baseline characteristics of the study population by year of diagnosis using frequencies and percentages for all variables. We used χ2 tests to test for differences between groups. We used a multivariable logistic regression model to examine associations between use of the RS assay and receipt of chemotherapy. To minimize colinearity within the model, clinicopathologic factors included in the multivariable regression analysis were limited to grade, tumor size greater than 2.0 cm, and NCCN risk group. We did not include stage or nodal status because these are effectively captured by tumor size and NCCN risk variables. We conducted a prespecified subgroup analysis of patients aged 65 to 70 years to reduce confounding due to associations between age and use of the RS assay or receipt of chemotherapy. Significance tests and confidence intervals for estimates from all models were based on robust standard errors to account for the clustering of patients by registry. We used SAS, version 9.3 (SAS Institute), for all analyses.

In addition to estimating overall associations, we estimated associations between use of the RS assay and receipt of chemotherapy in prespecified subgroups by adding a subgroup variable and an interaction term between the subgroup variable and the RS assay indicator to the models. We assessed differences between subgroups by testing the significance of the interaction term. We estimated the RS assay associations in each subgroup using model contrasts. Because of the multiple comparisons in this analysis, we report 99% confidence intervals and used α = .01 to establish statistical significance. All tests were 2 sided.

Sensitivity Analyses

To exclude the potential impact of immortal time bias and potential misclassification of chemotherapy in the adjuvant vs recurrent settings, we performed a sensitivity analysis that required all patients to survive 6 months after diagnosis, and we limited ascertainment of both the RS assay and chemotherapy to this same 6-month window. To address issues of selection bias associated with receipt of the RS assay, we also conducted a hospital referral region (HRR)–level analysis to examine associations between adoption of the RS assay and changes in chemotherapy use for the overall and younger than 70 years cohorts and provide a complementary population-based perspective. Hospital referral regions are defined by the 2008 Dartmouth Atlas of Health Care19 and have been used previously to study tertiary medical care variation in regional health care practices, including receipt of cancer-directed care.20-22 We limited analysis to major HRRs, operationally defined as those with at least 100 observations, to compare absolute changes in use of the RS assay and chemotherapy during the before (2005 and 2006) and after (2008 and 2009) periods. We chose not to use a propensity score–matching approach. Although such approaches are able to match for observed factors, they are unable to adjust for unobserved factors such as performance status that would have resulted in persistent bias.

Results

We identified 44 044 Medicare beneficiaries who received a diagnosis of ER-positive, nonmetastatic, invasive breast cancer in the SEER registries between 2005 and 2009. Table 1 shows the baseline characteristics of the study population stratified by use of the RS assay. Use of the assay was higher among patients 70 years or younger, patients with fewer comorbid conditions, and patients with T1c tumors, node-negative disease, and intermediate-risk disease. eTables 1 through 4 in the Supplement show the baseline characteristics of the study population by risk category, receipt of chemotherapy, and use of the RS assay.

Table 2 shows the baseline characteristics of the study population stratified by receipt of chemotherapy. Overall, 14.3% of patients received chemotherapy within 12 months after diagnosis. Among patients who had chemotherapy, more than 80% received it within 4 months after diagnosis. Among patients who underwent chemotherapy and testing with the RS assay, more than 90% received chemotherapy within 3 months after testing. Receipt of chemotherapy was more frequent in patients younger than 70 years, patients with fewer comorbid conditions, and patients with high-risk disease.

In multivariable analysis, we observed no association between use of the RS assay and receipt of chemotherapy (odds ratio [OR], 1.03 [99% CI, 0.88-1.19]) (Table 3). Age and disease risk played dominant roles in predicting chemotherapy use. Compared with patients 70 years or younger, patients in older age groups were successively less likely to receive chemotherapy. Compared with patients with low-risk disease, patients were more likely to undergo chemotherapy if they had high-risk (OR, 20.3 [99% CI, 17.8-23.2]) or intermediate-risk (OR, 3.67 [99% CI, 3.23-4.17]) disease. There was a significant interaction between use of the RS assay and risk group (P = .006). Among high-risk patients, use of the RS assay was associated with a lower likelihood of receiving chemotherapy (OR, 0.36 [99% CI, 0.26-0.50]), whereas patients with low-risk disease who underwent testing were more likely to undergo chemotherapy (OR, 3.71 [99% CI, 2.30-5.98]) (eTable 5 in the Supplement).

Because of the strong dependence of chemotherapy use on patient age, we performed a prespecified analysis limiting the multivariable analyses to patients 70 years or younger (n = 11 135) (Table 4 and eTable 6 in the Supplement). Similar to the primary analysis, we did not observe any association between use of the RS assay and use of chemotherapy (OR, 1.02 [99% CI, 0.81-1.28]). In univariate analysis, use of the assay was associated with decreased use of chemotherapy (OR, 0.65 [99% CI, 0.56-0.75]; P < .001). However, this association disappeared after adjustment for risk, suggesting that the association of the RS assay with chemotherapy in univariate models was largely confounded by risk. We also observed a steady decrease in the likelihood of chemotherapy use over time. Interaction models of risk and use of the RS assay demonstrated trends similar to the main analysis, with use of the RS assay associated with decreased use of chemotherapy in high-risk disease (OR, 0.34 [99% CI, 0.20-0.57]) and increased chemotherapy in low-risk disease (OR, 4.46 [99% CI, 3.15-6.31]; P = .01 for the interaction).

The rate of chemotherapy use remained relatively stable in the overall study population at 14.3%, with use varying by NCCN risk (Figure). However, from 2005 to 2009, receipt of chemotherapy among patients undergoing testing with the RS assay decreased from 20.0% to 14.6%. As expected, chemotherapy use was substantially higher in patients 70 years and younger and varied by risk. In this younger population, chemotherapy use declined from 29.1% in 2005 to 24.0% in 2009 (P = .004). The decline in chemotherapy use was limited to patients with high-risk disease (67.6% in 2005 vs 56.4% in 2009; P = .002) and patients who had RS assay testing (26.4% in 2005 vs 16.1% in 2009; P = .005).

Sensitivity analyses limited to the 6-month window for chemotherapy and survival result in estimates nearly identical to those of the primary analysis. This suggests that neither immortal time bias nor the use of chemotherapy in patients with recurrent disease had any substantial impact on the results of the main analysis. Hospital referral region–level analysis comparing regional adoption of the RS assay and changes in overall chemotherapy use did not reveal significant associations between regional adoption of the RS assay and changes in chemotherapy use in the overall and younger than 70 years populations (eFigure in the Supplement).

Discussion

To our knowledge, ours is the first investigation of the association of use of the 21-gene RS assay and receipt of chemotherapy in a nationally representative sample of patients with early-stage, ER-positive breast cancer in community settings since the introduction of the RS assay in 2004. Contrary to our hypothesis, we did not observe a change in chemotherapy use in the overall study population after the adoption of the RS assay between 2005 and 2009 or a general association between receipt of chemotherapy and use of the assay in multivariable analyses or HRR-level analyses. However, there was a significant interaction between use of the assay and NCCN risk category. Use of the RS assay was associated with decreased chemotherapy use in high-risk patients and increased chemotherapy use in low-risk patients. Because of the substantial impact of age on the use of the RS assay and chemotherapy, we performed a prespecified subgroup analysis of patients aged 66 to 70 years. In these patients, we observed an overall decrease in the receipt of chemotherapy from 29% to 24% that appeared to be limited to patients with high-risk disease and patients who underwent RS assay testing. In this observational analysis, we could not determine to what extent decreased chemotherapy use reflects the influence of RS assay adoption or unrelated changes in practice over the same period. However, our data support the hypothesis that use of the assay may have decreased chemotherapy use in general practice among younger patients with high-risk disease in whom receipt of chemotherapy would have otherwise been likely but that it may have increased chemotherapy use in patients with low-risk disease in whom chemotherapy would have been unlikely.

Most studies have consistently supported the ability of the RS assay to change physician treatment recommendations compared with recommendations made on the basis of clinicopathologic characteristics alone, with reported reductions in chemotherapy use ranging from 10% to 20% with use of the assay.2,8,12,13,23-26 However, not all studies reported declines in chemotherapy use.24 In our study, we observed a differential impact of the RS assay depending on the patient’s pretest likelihood to receive chemotherapy. Evidence consistent with this finding comes from a study by Eiermann et al8 that examined both node-negative and node-positive (ie, high-risk) patients from a total of 15 German medical centers before and after inclusion of RS assay results into their recommendation for chemotherapy. They found that assay results changed recommendations of a tumor board to exclude chemotherapy in 28% of patients and include chemotherapy in 9% of patients for node-positive disease, whereas the recommendations changed less drastically for node-negative disease (18% and 6%, respectively).8 The RS assay risk scores in previous work have been associated with chemotherapy use of 10% (low RS), 36% (intermediate RS), and 72% (high RS), although no clinicopathologic correlates were available in these patients to help disambiguate the relative effect of assay risk score.6

It should also be noted that much of the retrospective evidence regarding use of the RS assay and its impact on the use of chemotherapy did not emerge until after 2009. However, much of the key randomized and prospective evidence regarding use of the RS assay in ER-positive, node-negative disease was available for patients and physicians by the end of the study period. These studies included the major meta-analysis by the Early Breast Cancer Trialists Collaborative Group in 2005 demonstrating a benefit of chemotherapy in all women with early-stage disease, prospective prognostic (2004) and predictive (2006) validation of the RS assay, and guidelines that led to its recommended use by the American Society of Clinical Oncology by 2007.3,27-29

In most published clinical series, roughly half of patients had a low-risk RS, with median or mean ages in the mid-50s. It is unknown how the distribution of the RS from previous clinical series compares with patients in our study population, in which patients were on average at least a decade older. Overall rates of chemotherapy in our study were significantly lower than that reported rates of chemotherapy of roughly 40% in patients with ER-positive, node-negative disease who had the RS assay. In our sample, which included roughly 25% with node-positive disease, we observed 14.3% chemotherapy use among the overall population and 26.5% among patients 70 years or younger, which supports a significant influence of age on chemotherapy use. Although beyond the scope of this study, we previously reported details regarding the adoption of the RS assay in this population.15 In our HRR-based sensitivity analyses, we confirmed Medicare reimbursement of the RS assay in all major HRRs included in the study by the end of the study period.

Our study has some limitations. Only RS assays paid for by Medicare could be detected in the analysis. It is unlikely, but unknown, whether many RS assays among Medicare beneficiaries would be paid by third-party insurers or out of pocket. The claims-based approach used in this study to detect use of the RS assay has been described in detail by our group and others supporting the validity of using administrative claims data to detect use of the assay.15,30 We did not know the results of the assay, only that patients received the test. Any data or discussion of high-, intermediate-, and low-risk disease in this paper refers to NCCN risk categories. Patients in the SEER registry are overall more likely to be non-white and to live in low-poverty and urban areas,31 which may affect the generalizability of our findings. Importantly, the study population was limited to women 65 years or older, and would therefore include just under half of all newly diagnosed cases of breast cancer. Younger women are far more likely to undergo chemotherapy, be heterogeneously insured, have differential access to the RS assay, and therefore the impact of RS assay adoption on chemotherapy in younger women in nongovernment insured settings remains an open question. In our study, HER2 status was unavailable. Patients with ER-positive/progesterone receptor–positive and HER2-positive disease would be much more likely to receive chemotherapy, would not receive the RS assay, and would therefore create selection bias in favor of demonstrating lower chemotherapy rates among RS assay–tested patients. Last, the years of data available represent the initial uptake of the RS assay because the test only became available in 2004, with study limited through 2009 due to a lag in SEER-Medicare data availability. Use of the RS assay during this initial adoption phase may have been enriched for a population of early adopters that may differ systematically from patients in later years who undergo testing following widespread and less selective use of the test.

An additional limitation of the data is that to our knowledge the RS assay risk score is currently not available for study within any nationally representative or population-level cohort, which would be needed to evaluate whether RS assay results are being used to appropriately guide the use of chemotherapy in individual patients. However, the goal of the present study was to examine to what extent adoption of the RS assay at the population or system level has influenced rates of chemotherapy, an effect that has been predicted but never empirically examined at the population level. Payers, policy makers, and institutions seeking to assess the impact of RS assay adoption on rates of chemotherapy use must make these decisions at the system level before knowing the results of the test. To address this question, we believe that the approach used in the present study, which remains agnostic as to the actual RS assay results, is the appropriate approach to evaluate the impact of RS assay adoption on chemotherapy use at the system or population level. Nonetheless, the extent to which the RS results are being used to guide chemotherapy within individual patients in the general population remains an important area of ongoing research.

Conclusions

Patient and clinicopathologic factors significantly influence receipt of chemotherapy and use of the RS assay in routine clinical practice. Previous studies suggested that use of the RS assay, depending on the patient population, may reduce overall chemotherapy rates by up to 20%. Our data suggest that use of the RS assay may have decreased chemotherapy use in general practice among younger patients with high-risk disease in whom receipt of chemotherapy would have otherwise been likely but that it was associated with greater chemotherapy use in patients with low-risk disease. The impact of the RS assay on chemotherapy use is likely population dependent and is influenced by the population’s pretest likelihood of undergoing chemotherapy. The impact and association of RS assay adoption with costs and outcomes at the national level is an important area of future health policy breast cancer research.

Back to top
Article Information

Accepted for Publication: June 19, 2015.

Corresponding Author: Michaela A. Dinan, PhD, Duke Clinical Research Institute, PO Box 17969, Durham, NC 27715 (michaela.dinan@duke.edu).

Published Online: August 27, 2015. doi:10.1001/jamaoncol.2015.2722.

Author Contributions: Dr Dinan had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Dinan, Reed, Lyman.

Acquisition, analysis, or interpretation of data: Dinan, Mi, Lyman, Curtis.

Drafting of the manuscript: Dinan, Lyman.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Dinan, Mi.

Obtained funding: Dinan.

Administrative, technical, or material support: Dinan.

Study supervision: Dinan, Reed, Curtis.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was supported by grant K99HS022189 from the Agency for Healthcare Research and Quality.

Role of the Sponsor: The Agency for Healthcare Research and Quality had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and the decision to submit the manuscript for publication.

Disclaimer: This study used the linked SEER-Medicare database. The interpretation and reporting of these data are the sole responsibility of the authors.

Additional Contributions: Damon M. Seils, MA, Duke University, provided editorial assistance and prepared the manuscript. Mr Seils did not receive compensation for his assistance apart from his employment at the institution where the study was conducted. The authors acknowledge the efforts of the Applied Research Program, National Cancer Institute; the Office of Research, Development and Information, Centers for Medicare & Medicaid Services; Information Management Services, Inc; and the SEER Program tumor registries in the creation of the SEER-Medicare database.

References
1.
Theriault  RL, Carlson  RW, Allred  DC,  et al. NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines)—Breast Cancer, Version 3.2013. http://www.nccn.org/professionals/physician_gls/f_guidelines.asp. Accessed April 17, 2015.
2.
Ademuyiwa  FO, Miller  A, O’Connor  T,  et al.  The effects of oncotype DX recurrence scores on chemotherapy utilization in a multi-institutional breast cancer cohort.  Breast Cancer Res Treat. 2011;126(3):797-802.PubMedGoogle ScholarCrossref
3.
Paik  S, Tang  G, Shak  S,  et al.  Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer.  J Clin Oncol. 2006;24(23):3726-3734.PubMedGoogle ScholarCrossref
4.
Arango  BA, Rivera  CL, Glück  S.  Gene expression profiling in breast cancer.  Am J Transl Res. 2013;5(2):132-138.PubMedGoogle Scholar
5.
Kelly  CM, Krishnamurthy  S, Bianchini  G,  et al.  Utility of oncotype DX risk estimates in clinically intermediate risk hormone receptor-positive, HER2-normal, grade II, lymph node-negative breast cancers.  Cancer. 2010;116(22):5161-5167.PubMedGoogle ScholarCrossref
6.
Hornberger  J, Chien  R, Krebs  K, Hochheiser  L.  US insurance program’s experience with a multigene assay for early-stage breast cancer.  Am J Manag Care. 2011;17(5 Spec No):e194-e202.PubMedGoogle Scholar
7.
Tang  G, Shak  S, Paik  S,  et al.  Comparison of the prognostic and predictive utilities of the 21-gene Recurrence Score assay and Adjuvant! for women with node-negative, ER-positive breast cancer: results from NSABP B-14 and NSABP B-20.  Breast Cancer Res Treat. 2011;127(1):133-142.PubMedGoogle ScholarCrossref
8.
Eiermann  W, Rezai  M, Kümmel  S,  et al.  The 21-gene recurrence score assay impacts adjuvant therapy recommendations for ER-positive, node-negative and node-positive early breast cancer resulting in a risk-adapted change in chemotherapy use.  Ann Oncol. 2013;24(3):618-624.PubMedGoogle ScholarCrossref
9.
Hornberger  J, Cosler  LE, Lyman  GH.  Economic analysis of targeting chemotherapy using a 21-gene RT-PCR assay in lymph-node-negative, estrogen-receptor-positive, early-stage breast cancer.  Am J Manag Care. 2005;11(5):313-324.PubMedGoogle Scholar
10.
Reed  SD, Dinan  MA, Schulman  KA, Lyman  GH.  Cost-effectiveness of the 21-gene recurrence score assay in the context of multifactorial decision making to guide chemotherapy for early-stage breast cancer.  Genet Med. 2013;15(3):203-211.PubMedGoogle ScholarCrossref
11.
Rouzier  R, Pronzato  P, Chéreau  E, Carlson  J, Hunt  B, Valentine  WJ.  Multigene assays and molecular markers in breast cancer: systematic review of health economic analyses.  Breast Cancer Res Treat. 2013;139(3):621-637.PubMedGoogle ScholarCrossref
12.
Lo  SS, Mumby  PB, Norton  J,  et al.  Prospective multicenter study of the impact of the 21-gene recurrence score assay on medical oncologist and patient adjuvant breast cancer treatment selection.  J Clin Oncol. 2010;28(10):1671-1676.PubMedGoogle ScholarCrossref
13.
Geffen  DB, Abu-Ghanem  S, Sion-Vardy  N,  et al.  The impact of the 21-gene recurrence score assay on decision making about adjuvant chemotherapy in early-stage estrogen-receptor-positive breast cancer in an oncology practice with a unified treatment policy.  Ann Oncol. 2011;22(11):2381-2386.PubMedGoogle ScholarCrossref
14.
Surveillance, Epidemiology, and End Results Program. Overview of the SEER program. http://seer.cancer.gov/about/overview.html. Accessed April 17, 2015.
15.
Dinan  MA, Mi  X, Reed  SD, Hirsch  BR, Lyman  GH, Curtis  LH.  Initial trends in the use of the 21-gene recurrence score assay for patients with breast cancer in the Medicare population, 2005-2009.  JAMA Oncol. 2015;1(2):158-166.Google ScholarCrossref
16.
Albain  KS, Barlow  WE, Shak  S,  et al; Breast Cancer Intergroup of North America.  Prognostic and predictive value of the 21-gene recurrence score assay in postmenopausal women with node-positive, oestrogen-receptor-positive breast cancer on chemotherapy: a retrospective analysis of a randomised trial.  Lancet Oncol. 2010;11(1):55-65.PubMedGoogle ScholarCrossref
17.
Dinan  MA, Curtis  LH, Hammill  BG,  et al.  Changes in the use and costs of diagnostic imaging among Medicare beneficiaries with cancer, 1999-2006.  JAMA. 2010;303(16):1625-1631.PubMedGoogle ScholarCrossref
18.
Klabunde  CN, Potosky  AL, Legler  JM, Warren  JL.  Development of a comorbidity index using physician claims data.  J Clin Epidemiol. 2000;53(12):1258-1267.PubMedGoogle ScholarCrossref
19.
Wennberg  JE, Fisher  ES, Goodman  DC, Skinner  JS.  Tracking the Care of Patients With Severe Chronic Illness: The Dartmouth Atlas of Health Care 2008. Lebanon, NH: Dartmouth Institute for Health Policy and Clinical Practice; 2008.
20.
Fairfield  KM, Lucas  FL, Earle  CC, Small  L, Trimble  EL, Warren  JL.  Regional variation in cancer-directed surgery and mortality among women with epithelial ovarian cancer in the Medicare population.  Cancer. 2010;116(20):4840-4848.PubMedGoogle ScholarCrossref
21.
Polsky  D, Armstrong  KA, Randall  TC,  et al.  Variation in chemotherapy utilization in ovarian cancer: the relative contribution of geography.  Health Serv Res. 2006;41(6):2201-2218.PubMedGoogle ScholarCrossref
22.
Brooks  GA, Li  L, Sharma  DB,  et al.  Regional variation in spending and survival for older adults with advanced cancer.  J Natl Cancer Inst. 2013;105(9):634-642.PubMedGoogle ScholarCrossref
23.
Davidson  JA, Cromwell  I, Ellard  SL,  et al.  A prospective clinical utility and pharmacoeconomic study of the impact of the 21-gene Recurrence Score® assay in oestrogen receptor positive node negative breast cancer.  Eur J Cancer. 2013;49(11):2469-2475.PubMedGoogle ScholarCrossref
24.
Oratz  R, Paul  D, Cohn  AL, Sedlacek  SM.  Impact of a commercial reference laboratory test recurrence score on decision making in early-stage breast cancer.  J Oncol Pract. 2007;3(4):182-186.PubMedGoogle ScholarCrossref
25.
Asad  J, Jacobson  AF, Estabrook  A,  et al.  Does oncotype DX recurrence score affect the management of patients with early-stage breast cancer?  Am J Surg. 2008;196(4):527-529.PubMedGoogle ScholarCrossref
26.
Henry  LR, Stojadinovic  A, Swain  SM, Prindiville  S, Cordes  R, Soballe  PW.  The influence of a gene expression profile on breast cancer decisions.  J Surg Oncol. 2009;99(6):319-323.PubMedGoogle ScholarCrossref
27.
Early Breast Cancer Trialists’ Collaborative Group (EBCTCG).  Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials.  Lancet. 2005;365(9472):1687-1717.PubMedGoogle ScholarCrossref
28.
Paik  S, Shak  S, Tang  G,  et al.  A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer.  N Engl J Med. 2004;351(27):2817-2826.PubMedGoogle ScholarCrossref
29.
Harris  L, Fritsche  H, Mennel  R,  et al; American Society of Clinical Oncology.  American Society of Clinical Oncology 2007 update of recommendations for the use of tumor markers in breast cancer.  J Clin Oncol. 2007;25(33):5287-5312.PubMedGoogle ScholarCrossref
30.
Giordano  SH, Lin  YL, Kuo  YF, Hortobagyi  GN, Goodwin  JS.  Decline in the use of anthracyclines for breast cancer.  J Clin Oncol. 2012;30(18):2232-2239.PubMedGoogle ScholarCrossref
31.
Warren  JL, Klabunde  CN, Schrag  D, Bach  PB, Riley  GF.  Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population.  Med Care. 2002;40(8)(suppl):IV-3-IV-18.PubMedGoogle Scholar
×