Abstract
Background.
Breast cancer chemotherapy utilization may differ by race and age, but also varies by genomic risk, tumor characteristics, and patient characteristics. Studies in demographically diverse populations with both clinical and genomic data are necessary to understand potential disparities by race and age.
Methods.
In the Carolina Breast Cancer Study Phase 3 (2008–2013), chemotherapy receipt (yes/no) and regimen type were assessed in association with age and race among hormone receptor (HR) positive and HER2-negative tumors (n=1862). Odds ratios were estimated for the association between demographic factors and chemotherapy receipt.
Results.
Monotonic decreases in frequency of adjuvant chemotherapy receipt were observed over time during the study period, while neoadjuvant chemotherapy was stable. Younger age was associated with chemotherapy receipt (OR [95% CI]: 2.9 [2.4, 3.6]) and with anthracycline-based regimens (OR [95% CI]: 1.7 [1.3, 2.4]). Participants who had Medicaid (OR [95% CI]: 1.8 [1.3, 2.5]), lived in rural settings (OR [95% CI]: 1.4 [1.0, 2.0]), or were Black (OR [95% CI]: 1.5 [1.2, 1.8]) had slightly higher odds of chemotherapy, but these associations were non-significant with adjustment for stage and grade. Associations between younger age and chemotherapy receipt were strongest among women who did not receive genomic testing.
Conclusions.
While race was not strongly associated with chemotherapy receipt, younger age remains a strong predictor of chemotherapy receipt, even with adjustment for clinical factors and among women who receive genomic testing.
Keywords: Breast cancer treatment, adjuvant chemotherapy, neoadjuvant chemotherapy, chemotherapy treatment regimens, HR-positive HER2-negative breast cancer
Introduction
Breast cancer chemotherapy recommendations from the National Comprehensive Cancer Network and the American Society for Clinical Oncology are primarily based on disease characteristics like grade, stage, and hormone receptor status rather than demographics. In clinical practice, however, the relationship between chemotherapy receipt and demographic characteristics such as race and age has been inconsistent in previous literature. While some studies found that Black women were less likely to receive chemotherapy (Griggs 2012, Freedman 2014), others found no association (Lipscomb 2012, Kurian 2013) or increased chemotherapy receipt (Enewold 2018, Zhang 2019). Most of these studies have assessed chemotherapy as a binary variable, and few have incorporated data on specific chemotherapy regimens. Studies that do consider regimen choice commonly compare ‘standard’ vs ‘non-standard’ regimens (Griggs 2007, Kuo 2008, Freedman 2014), with limited information on chemotherapy classes (i.e. taxanes or anthracyclines). Understanding demographic patterns in adjuvant and neoadjuvant chemotherapy in context of clinical characteristics is important to understanding treatment disparities.
Here, we report associations between race and age and chemotherapy receipt within a diverse cohort of women with hormone receptor-positive (HR+), HER2- breast cancer. We limit to women with HR+HER2- tumors for two reasons. First, it allows for a more clinically homogenous cohort. Furthermore, these patients may be eligible to use genomic testing for risk stratification and chemotherapy decision-making. We utilized data from the Carolina Breast Cancer Study (CBCS3), which recruited a race- and age-diverse (50% Black women, 50% women under age 50 at diagnosis) population of women with invasive breast cancer. CBCS includes detailed treatment information together with data on demographics, tumor biology, and genomic testing, allowing us to assess predictors of both any and regimen-specific chemotherapy receipt.
Methods
Study Population
CBCS3 is a prospective, population-based cohort of women with newly diagnosed incident breast cancer in North Carolina, recruited between 2008 and 2013. Cases were identified using rapid case ascertainment in collaboration with the North Carolina Central Cancer Registry. Eligible women were aged 20 to 74 years old. The study was approved by the University of North Carolina Institutional Review Board in accordance with the revised U.S. Common rule. Study participants provided written informed consent prior to study entry. The study oversampled both Black and young (20–49 years old) women to investigate age- and race-related disparities. This analysis was performed under a cells-to-society conceptual framework (Warnecke 2008), in which multiple levels from molecular to community interact to alter breast cancer burden (Emerson 2020). In this work, race is viewed as a social construct, representing exposures and experiences across the life course, including exposure to structural racism.
For the current analysis (Figure 1), we began with the full CBCS3 population (n=2998) and excluded women who did not receive surgery (n=63) or had metastatic disease at diagnosis (n=109). From this subset (n=2826), we further restricted to women who had hormone-receptor (HR) positive, HER2-negative tumors (n=1862). Of these women, 931 (50.0%) did not receive any chemotherapy during their initial course of treatment, 216 (11.6%) received neoadjuvant chemotherapy, and 715 (38.4%) received adjuvant chemotherapy.
Figure 1.
Flowchart for study inclusion
Outcome assessment
The outcomes of interest (chemotherapy receipt and regimen type) were constructed using treatment data abstracted from medical records. These data were collected by certified medical records abstractors at each CBCS3 follow-up, approximately 9–12 months apart. Abstracted data included information on neoadjuvant and adjuvant chemotherapy agents received and timing/dates of receipt. From these data, categorical variables of chemotherapy receipt (yes/no) and regimen were constructed. Chemotherapy regimens included TC (taxane and cyclophosphamide), AC-T (anthracycline and cyclophosphamide, followed by taxane), TAC (taxane, anthracycline, and cyclophosphamide concurrently), TCH (taxane, carboplatin, and HER2-directed therapy), or Other, which includes both uncommon and non-standard regimens. Broader chemotherapy classes were also constructed that represented any taxane (T, including either TC or taxane monotherapy), a combination of an anthracycline and taxane (AT, including AC-T, TAC, and any other non-standard regimens that included a taxane and anthracycline), TCH, and Other. A table detailing numbers of participants who received each regimen is included in supplemental materials (Supplemental Table 2).
Independent variables
Age and race, the primarily independent variables of interest, were self-reported and collected within the baseline-questionnaire. Age was dichotomized at 50 years old (<50 vs 50+) for multivariable analyses, although different categorizations were considered. Race was binarized (Black vs. non-Black) according to self-report. Among women categorized as non-Black (n=1054), the majority of women in this study identified as White (n=991), while a smaller proportion identified as American Indian/Alaska Native (n=9), Asian or Pacific Islander (n=23), or Other (n=31).
Other clinical information including stage, grade, estrogen (ER) and progesterone receptor (PR) status, HER2 status, and receipt of the tumor genomic assay Oncotype DX (ODX) were abstracted from medical records and pathology reports. Tumor and node stage were used to classify women into lower clinical risk (tumor size ≤2cm, N0) and higher clinical risk (tumor size >2cm and/or N1+) categories. Grade was categorized as low (1), intermediate (2), and high (3). ER and PR IHC status were binarized as positive (including borderline) and negative. A tumor that was either ER- or PR-positive was defined as HR-positive. HER2 status was dichotomized as positive (excluding borderline) and negative. ODX receipt was classified as binary (yes/no) and ODX scores among those who received the assay were dichotomized (<26 as low-intermediate risk, ≥26 as high risk as implemented in Sparano 2019).
Other variables related to healthcare access included insurance status (currently covered by insurance yes/no) and insurance type (private, Medicaid, Medicare, other, or uninsured). Women could report multiple insurance sources, so insurance status was categorized such that if a participant reported any Medicare coverage, she was identified as having Medicare. If no Medicare was reported but private insurance was, the patient was coded as having private insurance. If neither Medicare nor any private insurance was reported, but Medicaid insurance was, the participant was coded as having Medicaid. If no insurance was reported, the participant was coded as being uninsured. Urbanicity (urban/rural) of residence at the time of diagnosis was defined using the United States Office of Management and Budget’s (OMB) 2013 rural/urban continuum codes (USDA ERS 2014). This OMB designation separates counties into 9 rural-urban continuum (RUCC) categories based on population. Counties defined as metropolitan based on these categories (RUCC 1–3) were classified as “urban,” while RUCC 4–9 counties were classified “rural.”
Statistical Analysis
To measure the strength of association between chemotherapy use and demographic or clinical variables, univariate logistic regression was used to calculate unadjusted odds ratios (ORs) and 95% confidence intervals (CIs). For age and race only, multivariable logistic regression was used to calculate ORs and 95% CIs. For binary chemotherapy receipt analyses, estimates were stratified by clinical risk defined by tumor size and node status as described in the previous section. Adjustment sets for minimally-adjusted multivariable models included age and race, while fully-adjusted models include age, race, and grade. Sensitivity analyses were performed replacing stage-based clinical risk with ODX risk categories for those women who underwent genomic testing and adjusting for year of diagnosis. To represent trends in chemotherapy use, generalized linear models with binary family and identity link were used to calculate the proportion of individuals in each year who received adjuvant or neoadjuvant chemotherapy among women with HR+HER2- tumors and 95% CIs. All analyses were performed in R version 3.6.3. This is an observational study and was designated exempt by the University of North Carolina at Chapel Hill Institutional Review Board.
Results
Chemotherapy receipt relative to untreated
Chemotherapy receipt (measured by proportion of patients receiving any chemotherapy) decreased monotonically according to diagnosis year from 58.7% in 2008 to 45.2% in 2012 (Figure 2), with the largest decrease between 2008 and 2009 (−8.4%, 95% CI: −16.6, −0.0). Considering adjuvant and neoadjuvant chemotherapy separately, adjuvant chemotherapy declined significantly from 2008 to 2012 (average decrease of 2.7 percentage points per year, 95% CI: −4.4, −1.0), while the proportion receiving neoadjuvant chemotherapy was approximately stable, changing from 12.0% in 2008 to 13.3% in 2012 (Figure 3).
Figure 2. Chemotherapy receipt by diagnosis year in the Carolina Breast Cancer Study Phase 3, 2008–2012.
Chemotherapy receipt in the study population declined from 58.7% in 2008 to 45.2% in 2012 across the study period, corresponding to a decrease of −8.4% (95% CI: −16.6, −0.0).
Figure 3. Chemotherapy receipt by diagnosis year in CBCS3, stratified by adjuvant or neoadjuvant administration.
Receipt of adjuvant chemotherapy in the study population declined from 46.6% in 2008 to 32.0% in 2012. Receipt of neoadjuvant chemotherapy stayed approximately stable, changing non-monotonically from 12.0% in 2008 to 13.3% in 2012.
Women with more advanced tumors had higher odds of chemotherapy receipt (Table 1). Stage (2/3 vs. stage 1, OR [95% CI]: 10.1 [8.0, 12.7]) and grade (grade 3 vs. 1 OR [95% CI]: 11.4 [8.3, 15.8]) were significant predictors of chemotherapy receipt and Oncotype DX score <26 (Low-intermediate vs. High, OR [95% CI]: 0.07 [0.04, 0.11]) was associated with not receiving chemotherapy. Younger age (<50 years old), Black race, Medicaid/uninsured status, and rural residence were associated with higher frequency of chemotherapy in unadjusted models; however, after adjusting for stage and grade, only younger age and Medicare-insured were significantly associated with chemotherapy (Table 1, note: 93.6% of Medicare-insured were over 50 years old).
Table 1.
Associations between participant or tumor characteristics and chemotherapy receipt
| Chemo | Non-Chemo | Unadjusted Prevalence Odds Ratio | Adjusted Prevalence Odds Ratio1 | |
|---|---|---|---|---|
| n (%) or mean (sd) | n (%) or mean (sd) | (95% CI) | (95% CI) | |
| Stage | ||||
| 1 | 185 (19.9) | 723 (77.7) | Ref | Ref |
| 2/3 | 746 (80.1) | 208 (22.3) | 14.0 (11.2, 17.6) | 14.0 (11.0, 18.0) |
| Tumor Size | ||||
| ≤2cm | 336 (36.1) | 771 (82.8) | Ref | Ref |
| >2cm | 595 (63.9) | 160 (17.2) | 8.5 (6.9, 10.6) | 7.4 (5.9, 9.4) |
| Node Status | ||||
| Negative | 384 (41.2) | 832 (89.4) | Ref | Ref |
| Positive | 547 (58.8) | 99 (10.6) | 12.0 (9.4, 15.4) | 13.0 (9.8, 19.5) |
| Grade | ||||
| Low | 140 (15.2) | 376 (40.8) | Ref | Ref |
| Intermediate | 402 (43.8) | 466 (50.5) | 2.2 (1.7, 2.8) | 1.7 (1.3, 2.2) |
| High | 376 (41.0) | 80 (8.7) | 11.4 (8.3, 15.8) | 11.2 (7.8, 16.1) |
| Missing | 13 | 9 | ||
| Oncotype DX2 | ||||
| Low-intermediate | 104 (53.6) | 410 (94.5) | 0.07 (0.04, 0.11) | 0.09 (0.05, 0.16) |
| High | 90 (46.4) | 24 (5.5) | Ref | Ref |
| Age (categorical) | ||||
| 50+ | 373 (40.1) | 617 (66.3) | Ref | Ref |
| <50 | 558 (59.9) | 314 (33.7) | 2.9 (2.4, 3.6) | 2.5 (1.9, 3.1) |
| Race | ||||
| Non-Black | 481 (51.7) | 573 (61.5) | Ref | Ref |
| Black | 450 (48.3) | 358 (38.5) | 1.5 (1.2, 1.8) | 1.1 (0.8, 1.4) |
| Insurance status3 | ||||
| Private | 591 (64.0) | 507 (54.5) | Ref | Ref |
| Medicare | 141 (15.3) | 315 (33.9) | 0.4 (0.3, 0.5) | 0.3 (0.3, 0.5) |
| Medicaid | 126 (13.6) | 60 (6.5) | 1.8 (1.3, 2.5) | 1.0 (0.6, 1.5) |
| Uninsured | 66 (7.1) | 44 (4.7) | 1.3 (0.9, 1.9) | 0.7 (0.4, 1.3) |
| Missing | 7 | 5 | ||
| Urban/Rural | ||||
| Urban | 818 (87.9) | 845 (90.8) | Ref | Ref |
| Rural | 113 (12.1) | 86 (9.2) | 1.4 (1.0, 1.8) | 1.3 (0.9, 1.9) |
Adjusted for stage and/or grade where applicable
Oncotype DX scores were abstracted from clinical record and were categorized as Low: <11, Intermediate: 11–25, High: 26+.
Participants who reported ‘Other ’ insurance status (n=10) were grouped with missing.
We hypothesized that the relationship between demographic characteristics and chemotherapy receipt may depend upon disease severity, and therefore we estimated the relationship between age/race and chemotherapy receipt within categories of tumor size and node status at diagnosis (Table 2, Figure 4). We designated tumors <=2 cm and node-negative as lower-risk and tumors >2cm or node-positive as higher-risk. In both lower and higher risk tumors, women younger than 50 y/o were more likely to receive chemotherapy than those over the age of 50. Black participants with low-risk tumors were slightly more likely to receive chemotherapy than non-Black participants, however these associations were not significant after adjusting for grade.
Table 2.
Associations between first-line chemotherapy receipt (adjuvant or neoadjuvant) and age or race, stratified by Oncotype DX (ODX) receipt
| Untreated n (%) |
Treated n (%) |
Minimal Model OR (95% CI)2 |
Full Model OR (95% CI)3 |
|
|---|---|---|---|---|
| Age and Chemotherapy Receipt | ||||
| Low-risk1 | ||||
| Overall | ||||
| 50+ | 476 (86.2) | 76 (13.8) | Ref | Ref |
| <50 | 230 (71.2) | 93 (28.8) | 2.61 (1.85, 3.68) | 2.41 (1.63, 3.58) |
| Received ODX | ||||
| 50+ | 168 (76.7) | 51 (23.3) | Ref | Ref |
| <50 | 118 (69.4) | 52 (30.6) | 1.51 (0.95, 2.38) | 1.36 (0.81, 2.31) |
| No ODX Assay | ||||
| 50+ | 308 (92.5) | 25 (7.5) | Ref | Ref |
| <50 | 112 (73.2) | 41 (26.8) | 4.61 (2.69, 8.04) | 4.49 (2.41, 8.58) |
| High-risk1 | ||||
| Overall | ||||
| 50+ | 141 (32.2) | 297 (67.8) | Ref | Ref |
| <50 | 84 (15.3) | 465 (84.7) | 2.63 (1.93, 3.58) | 2.54 (1.86, 3.51) |
| Received ODX | ||||
| 50+ | 84 (71.2) | 34 (28.8) | Ref | Ref |
| <50 | 64 (52.9) | 57 (47.1) | 2.21 (1.30, 3.80) | 2.06 (1.18, 3.62) |
| No ODX Assay | ||||
| 50+ | 57 (17.8) | 263 (82.2) | Ref | Ref |
| <50 | 20 (4.7) | 408 (95.3) | 4.41 (2.63, 7.68) | 4.18 (2.47, 7.34) |
| Race and Chemotherapy Receipt | ||||
| Low-risk1 | ||||
| Overall | ||||
| Non-Black | 447 (82.5) | 96 (17.5) | Ref | Ref |
| Black | 259 (78.5) | 73 (21.5) | 1.42 (1.00, 2.01) | 1.34 (0.90, 2.00) |
| Received ODX | ||||
| Non-Black | 195 (75.6) | 63 (24.4) | Ref | Ref |
| Black | 91 (69.5) | 40 (30.5) | 1.43 (0.89, 2.29) | 1.64 (0.95, 2.82) |
| No ODX Assay | ||||
| Non-Black | 252 (88.4) | 33 (11.6) | Ref | Ref |
| Black | 168 (83.6) | 33 (16.4) | 1.60 (0.93, 2.76) | 1.24 (0.66, 2.32) |
| High-risk1 | ||||
| Overall | ||||
| Non-Black | 126 (30.0) | 294 (70.0) | Ref | Ref |
| Black | 99 (28.4) | 250 (71.6) | 1.25 (0.92, 1.70) | 1.01 (0.73, 1.39) |
| Received ODX | ||||
| Non-Black | 94 (61.4) | 59 (38.6) | Ref | Ref |
| Black | 54 (62.8) | 32 (37.2) | 0.91 (0.52, 1.58) | 0.80 (0.44, 1.43) |
| No ODX Assay | ||||
| Non-Black | 32 (8.9) | 326 (91.1) | Ref | Ref |
| Black | 45 (11.5) | 345 (88.5) | 0.76 (0.46, 1.23) | 0.65 (0.39, 1.08) |
Low-risk: (≤2cm, N0); Higher-risk: (>2cm and/or N1+)
Model includes race (for age strata) or age (for race strata)
Model includes race, age, grade
Figure 4. Odds ratios for the association between chemotherapy receipt and age or race.
Age [top, <50 years vs 50+ years (referent)] and race [bottom, Black women vs non-Black women (referent)] were assessed in relation to chemotherapy receipt. Stronger associations are observed between age and chemotherapy receipt, while most associations between race and chemotherapy are null or attenuate with adjustment for clinical factors. (Low-risk: tumor size ≤2cm, N0; high-risk: tumor size >2cm and/or N1+)
To evaluate whether genomic testing influenced demographic patterns, we stratified these analyses by receipt of ODX testing. Among women with clinically low-risk tumors who received the ODX assay, the association between younger age and chemo use was not significant (Low-risk, Oncotype+ fully-adjusted OR<50years [95% CI]: 1.36 [0.81, 2.31] vs. 50+ years). However, among those with clinically higher-risk tumors, young women were more likely to receive chemo (High-risk, Oncotype+ fully-adjusted OR<50years [95% CI]: 2.06 [1.18, 3.62] vs. 50+ years). Black women with clinically low-risk tumors in our study population had a modest increase in odds of receiving chemotherapy relative to non-Black women; however, this was non-significant in models that were adjusted with grade (Table 2, high-risk, fully-adjusted ORBlack [95% CI]: 1.34 [0.90, 2.00] vs. non-Black). Finally, because the proportion of women receiving chemotherapy varied by year of diagnosis, we performed a sensitivity analysis in which we further adjusted age and race estimates for year of diagnosis as a categorical variable. Model estimates did not change substantially when adjusting for calendar year. Furthermore, while the proportion of women receiving chemotherapy did vary by year of diagnosis, distributions of age and race were not significantly different across study years.
Regimen-Specific Chemotherapy Receipt
To assess patterns in use of anthracycline-containing regimens among a clinically homogeneous subset, we considered only ER+/HER2- participants who received adjuvant chemotherapy. There was an increase in anthracycline receipt from 40.2% to 52.4% between 2008 and 2009 and the proportion receiving anthracyclines increased non-monotonically from 40.2% in 2008 to 49.0% in 2012, but these trends were not significant. Among adjuvant chemotherapy regimens, we assessed predictors of taxane (T) or anthracycline-taxane (AT) regimens. We did not compare “Other” regimens to T or AT regimens, as this class was small (n=52) and heterogeneous, making interpretation difficult. Relative to T regimens, AT regimens were most strongly associated with stage (ORStage2/3 [95% CI]: 6.4 [4.2, 10.2] vs. Stage 1), grade (ORHighGrade [95% CI]: 1.8 [1.1, 2.9] vs. low grade), and younger age (OR<50years [95% CI]: 1.7 [1.3, 2.4] vs. 50+) (Table 3). Having Medicaid insurance and rural residence were significantly associated with receiving anthracycline regimens, while having Medicare insurance was significantly with non-receipt of anthracycline regimens. Race was weakly and not significantly associated with AT receipt (OR [95% CI]: 1.3 [0.9, 1.7]).
Table 3.
Associations between clinical or demographic variables and chemotherapy regimen among women receiving adjuvant chemotherapy
| Taxane-based Chemo n (%) |
Anthracycline- and Taxane-based Chemo n (%) |
ORAT (95% CI) (ref = Taxane-based) |
|
|---|---|---|---|
| Stage | |||
| 1 | 132 (82.5) | 28 (17.5) | Ref |
| 2/3 | 211 (42.4) | 287 (57.6) | 6.4 (4.2, 10.2) |
| Tumor Size | |||
| ≤2cm | 181 (62.0) | 111 (38.0) | Ref |
| >2cm | 162 (44.3) | 204 (55.7) | 2.1 (1.5, 2.8) |
| Node Status | |||
| Negative | 212 (77.1) | 63 (22.9) | Ref |
| Positive | 131 (34.2) | 252 (65.8) | 6.5 (4.6, 9.3) |
| Grade | |||
| Low | 68 (63.6) | 39 (36.4) | Ref |
| Intermediate | 147 (50.0) | 147 (50.0) | 1.7 (1.1, 2.8) |
| High | 125 (49.4) | 128 (50.6) | 1.8 (1.1, 2.9) |
| Race | |||
| Non-Black | 201 (54.8) | 166 (45.2) | Ref |
| Black | 142 (48.8) | 149 (51.2) | 1.3 (0.9, 1.7) |
| Age | |||
| 50+ | 175 (51.0) | 118 (37.5) | Ref |
| <50 | 168 (49.0) | 197 (62.5) | 1.7 (1.3, 2.4) |
| Insurance status | |||
| Private | 215 (50.6) | 210 (49.4) | Ref |
| Medicare | 73 (65.8) | 38 (34.2) | 0.5 (0.3, 0.8) |
| Medicaid | 31 (39.2) | 48 (60.8) | 1.6 (1.0, 2.6) |
| Uninsured | 21 (53.8) | 18 (46.2) | 0.9 (0.5, 1.7) |
| Urban/Rural | |||
| Urban | 309 (52.2) | 273 (47.8) | Ref |
| Rural | 39 (54.5) | 42 (45.5) | 1.2 (0.8, 1.9) |
“Other” insurance status (n=5) grouped with missing
To address confounding by clinical disease state, we further stratified by clinical risk (tumor size ≤2cm, N0 vs. tumor size >2cm and/or N+) and adjusted models for age or race, and grade (Supplemental Table 1). When accounting for clinical risk in these multivariable models, we observed that age and race were only associated with anthracycline-inclusive regimens in the higher risk (tumor size >2cm and/or N+) stratum. Similarly, younger age was associated with AT regimens (OR<50 [95% CI]: 2.0 [1.4, 2.9] vs 50+) among the clinically higher risk (tumor size >2cm and/or N+) group, while race was not (ORBlack [95% CI]: 1.3 [0.9, 1.9] vs. non-Black). These associations did not change meaningfully when further adjusting for tumor grade.
Discussion
This analysis used real-world data reflecting a diverse population of women with breast cancer in North Carolina. Using treatment data from medical records, we found evidence of increased odds of chemotherapy use and of anthracycline-inclusive regimen receipt among younger women (<50 years old). We observed that younger women had higher chemotherapy use, even among clinically lower risk patients and among those that receive ODX genomic assays. We also considered other demographic characteristics. Black women, women in rural localities, and women with non-private insurance coverage all had increased unadjusted odds of chemotherapy use; however, only younger age and age-related insurance status were significant after adjustment for clinical stage and grade. These results indicate that age plays a substantial role in clinical decision-making around chemotherapy, even when genomic testing is available. This also demonstrates that many chemotherapy patterns are driven by the clinical patterns and that some subgroups have higher frequency of clinically aggressive subtypes at diagnosis.
Many of our findings with respect to age are consistent with those from other studies assessing predictors of chemotherapy (Morimoto 2010, Griggs 2012, Kurian 2013, Spronk 2017, Neuner 2019) in observing strong univariate associations both with traditional clinical characteristics (stage, grade, Oncotype DX risk score) and demographic characteristics such as younger age, race, and insurance status.
In our study, we observed the strongest associations between chemotherapy and younger age. Consistent with these findings, a 2013 analysis by Kurian et al (Kurian 2013) observed a monotonic association between older age and decreased odds of chemotherapy receipt within a population made up of women with breast cancer in the Kaiser Permanente Northern California system between 2004 and 2007. Using SEER data from 2005 to 2007 and self-reported chemotherapy receipt, Griggs and colleagues similarly observed in 2012 that the proportion of patients receiving adjuvant chemotherapy decreased across each age category (in decades), reporting an odds ratio of 0.91 (95% CI: 0.90, 0.93) per additional year of age (Griggs 2012). These studies utilized data prior to the advent of widespread gene expression assay testing and therefore the current analysis assesses how receipt of therapy may have shifted in the years following genomic testing recommendations in national and international guidelines (in 2007, the American Society of Clinical Oncology and 2008, the National Comprehensive Cancer Network) (Harris 2007, NCCN 2023). We observed a moderate attenuation of the association between younger age and chemotherapy receipt among women who received ODX testing (Table 2).
Previous studies have also observed an association between age and other metrics of chemotherapy receipt (e.g. standard regimen receipt, timely receipt of chemotherapy), particularly among older women. In a 2007 analysis by Griggs et al, authors observed a relationship between older age (≥70 years) and non-standard chemotherapy regimen receipt. The association between increasing age and non-standard chemotherapy regimens was also observed in a 2014 analysis by Freedman et al. Among older women (≥65 years old) in SEER-Medicare, increasing age has been reported to be associated with lower odds of timely receipt of chemotherapy (Wheeler 2012). Younger age is associated with more aggressive disease characteristics, which may cause clinicians to treat it more aggressively than similar breast cancers in older women (Anders 2011). Furthermore, younger women have a longer period of time in which their disease could recur, so these women or their providers may choose or be recommended for a more aggressive treatment regimen. More recent subgroup analysis from the TAILORx trial demonstrated that benefit from chemotherapy varied by age in the intermediate ODX risk group, with younger women having improved DFS in response to both endocrine therapy and chemotherapy (Sparano 2018). Age is an important consideration during chemotherapy decision-making and prescribing patterns reflect this. While we here used age as a chronological variable, we note that there are likely differences in the association between physiological aging and chemotherapy receipt as well. We did not have data to investigate this relationship, but the magnitude of association may differ depending on what metric of aging is addressed.
With regard to the relationship between race and chemotherapy receipt, our results were consistent with prior null associations (Kurian 2013, Lipscomb 2012). Others have observed increased chemotherapy receipt among Black women: Enewold and co-authors reported a moderate but non-significant increase in chemotherapy receipt among non-Hispanic Black women (OR 1.44; 95% CI 0.90, 2.30), but only among women with stage IV disease (Enewold 2018). Comparisons of the current results to those of Enewold et al. are limited by two major factors. First, Enewold did not restrict to HR+HER2- women. Second, Enewold included metastatic patients. Defining chemotherapy patterns in more clinically homogeneous groups (i.e. ER+/HER2-, non-metastatic) is important to designing interventions. Another recent analysis used data from cancer registries to evaluate racial and ethnic differences in chemotherapy. Among women with stage I-III, HR+ HER2- disease, they reported an odds ratio for receipt of any chemotherapy of 1.22 (95% CI: 1.04, 1.42) for non-Hispanic Black women relative to non-Hispanic White women (Zhang 2019). Zhang et al. had similar inclusion criteria (stage I-III, HR+ HER2- disease), but they did observe a statistically significant increase in chemotherapy receipt among Black women. The Zhang study was a much larger cohort drawn from cancer registries, was better powered to detect a small odds ratio. However, we note that the magnitude of effect they observed (OR: 1.22) is similar to that in our work (OR: 1.1). Conversely, Griggs and colleagues observed that Black women had lower odds of treatment with chemotherapy, but this relationship was non-significant (OR 0.83; 95% CI: 0.64, 1.08) (Griggs 2012). Freedman et al observed a similar effect estimate (OR 0.87, 95% CI: 0.84, 0.91) among women in the National Cancer Data Base registry between 1998 and 2005 (Freedman 2014). Beyond differences in inclusion of late-stage patients, associations between race and chemotherapy may vary by geography or by institution. While the study has diversity in race and age, women in CBCS3 do have higher educational attainment and higher proportions of insurance coverage than women on average in NC, so we may have missed or underrepresented patterns of chemotherapy unique to women with lower SES or access to care (Emerson 2020).
In this analysis, we identified not only whether women received chemotherapy, but also what regimen they received. In limiting the analysis to women with HR+/HER2- and adjusting for stage and grade, we could evaluate demographic patterns after controlling for disease aggressiveness. Furthermore, because the CBCS3 sampling schema included 50% younger (<50 years old) women, we were well-powered for evaluating differences in chemotherapy receipt related to age. However, our work also has limitations. First, we did not evaluate substrata of women over age 70 because our study sample was limited to women under age 74 and there were a limited number of women enrolled between the ages of 70–74. Regimens prescribed to older women may differ from those prescribed to women aged 50–70. Another limitation was missing data; not all women in the study received Oncotype DX testing, so we could not compare chemotherapy receipt within genomic risk groups in our study population. Additionally, our models may have some residual confounding. While many factors may have influenced the association between demographics and chemotherapy, we were unable to consider all potential strata due to power, however we did stratify results by stage and found that younger women had higher odds of treatment with chemotherapy in every stratum of stage-based clinical risk, even after adjustment for race and tumor grade. Finally, insurance status is strongly correlated with age, where older women more frequently were covered by Medicare (43.2% of women 50 years or older were covered by Medicare vs 3.3% of women under 50), and younger women were more frequently covered by Medicaid (13.7% of women under 50 vs 6.8% of women 50 years or older). While some associations for age may be mediated by insurance, we were not able to address this due to power and the relatively small cell sizes when stratifying on both age and insurance type.
Our study demonstrates that the strongest predictors of chemotherapy receipt are tumor characteristics, with age also contributing substantially to treatment decision-making. Additional research into shared decision-making and rationale for chemotherapy delivery among young women may indicate to what degree genomic testing, clinician preference, or patient preference is responsible for the increase in chemotherapy receipt in this demographic.
Supplementary Material
Funding
We are grateful to all the participants of the Carolina Breast Cancer Study, whose generous participation has made this research possible, as well as the study staff members. This research was supported by a grant from UNC Lineberger Comprehensive Cancer Center, which is funded by the University Cancer Research Fund of North Carolina, the Susan G Komen Foundation (OGUNC1202 to M.A. Troester), the National Cancer Institute of the National Institutes of Health (P01CA151135 to M.A. Troester), and the National Cancer Institute Specialized Program of Research Excellence (SPORE) in Breast Cancer (NIH/NCI P50CA058223). This research recruited participants &/or obtained data with the assistance of Rapid Case Ascertainment, a collaboration between the North Carolina Central Cancer Registry and UNC Lineberger. RCA is supported by a grant from the National Cancer Institute of the National Institutes of Health (P30CA016086). This work was further supported by funding from the National Institute of Health (U54CA156733 and R01CA253450 to M.A. Troester; R01AG056479 to T. Stürmer), Susan G. Komen for the Cure (TREND21686258 and SAC210102 to M.A. Troester), and a grant from UNC Lineberger Comprehensive Cancer Center, which is funded by the University Cancer Research Fund of North Carolina.
Competing Interests
Dr. Reeder-Hayes has received research funding from Pfizer Global Medical Foundation. Dr. Stürmer receives salary support as Director of Comparative Effectiveness Research (CER), NC TraCS Institute, UNC Clinical and Translational Science Award (UL1TR002489), the Center for Pharmacoepidemiology (current members: GlaxoSmithKline, UCB BioSciences, Takeda, AbbVie, Boehringer Ingelheim), from pharmaceutical companies (Novo Nordisk), and from a generous contribution from Dr. Nancy A. Dreyer to the Department of Epidemiology, University of North Carolina at Chapel Hill. Dr. Stürmer does not accept personal compensation of any kind from any pharmaceutical company. He owns stock in Novartis, Roche, and Novo Nordisk. Other authors report no financial or non-financial interests.
Data Availability
The data analyzed in this study are available from the Carolina Breast Cancer Study (https://unclineberger.org/cbcs/). Restrictions apply to the availability of these data, which were used under data use agreements for this study. Data is not publicly available; however, investigators may submit a letter of intent to gain access upon reasonable request.
Bibliography
- Anders CK, Fan C, Parker JS, Carey LA, Blackwell KL, Klauber-DeMore N, Perou CM. Breast carcinomas arising at a young age: unique biology or a surrogate for aggressive intrinsic subtypes?. Journal of clinical oncology. 2011. Jan 1;29(1):e18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Emerson MA, Reeder-Hayes KE, Tipaldos HJ, Bell ME, Sweeney MR, Carey LA, Earp HS, Olshan AF, Troester MA. Integrating biology and access to care in addressing breast cancer disparities: 25 years’ research experience in the Carolina Breast Cancer Study. Current breast cancer reports. 2020. Sep;12:149–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Enewold L, Penn DC, Stevens JL, & Harlan LC (2018). Black/white differences in treatment and survival among women with stage IIIB-IV breast cancer at diagnosis: a US population-based study. Cancer causes & control : CCC, 29(7), 657–665. 10.1007/s10552-018-1045-9 [DOI] [PubMed] [Google Scholar]
- Freedman RA, Vaz-Luis I, Barry WT, Lii H, Lin NU, Winer EP, Keating NL. Patterns of chemotherapy, toxicity, and short-term outcomes for older women receiving adjuvant trastuzumab-based therapy. Breast cancer research and treatment. 2014. Jun;145:491–501. [DOI] [PubMed] [Google Scholar]
- Griggs et al. , 2007. Social and Racial Differences in Selection of Breast Cancer Adjuvant Chemotherapy Regimens [DOI] [PubMed]
- Griggs et al. , 2012. Factors Associated with Receipt of Breast Cancer Adjuvant Chemotherapy in a Diverse Population-Based Sample [DOI] [PMC free article] [PubMed]
- Harris L, Fritsche H, Mennel R, Norton L, Ravdin P, Taube S, Somerfield MR, Hayes DF, Bast RC Jr; 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. Nov 20;25(33):5287–312. doi: 10.1200/JCO.2007.14.2364. [DOI] [PubMed] [Google Scholar]
- Kuo SH, Lien HC, You SL, Lu YS, Lin CH, Chen TZ, Huang CS. Dose variation and regimen modification of adjuvant chemotherapy in daily practice affect survival of stage I–II and operable stage III Taiwanese breast cancer patients. The Breast. 2008. Dec 1;17(6):646–53. [DOI] [PubMed] [Google Scholar]
- Kurian AW, Lichtensztajn DY, Keegan TH, Leung RW, Shema SJ, Hershman DL, Kushi LH, Habel LA, Kolevska T, Caan BJ, Gomez SL. Patterns and predictors of breast cancer chemotherapy use in Kaiser Permanente Northern California, 2004–2007. Breast cancer research and treatment. 2013. Jan;137:247–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kurian AW, Bondarenko I, Jagsi R, Friese CR, McLeod MC, Hawley ST, Hamilton AS, Ward KC, Hofer TP, Katz SJ. Recent trends in chemotherapy use and oncologists’ treatment recommendations for early-stage breast cancer. JNCI: Journal of the National Cancer Institute. 2018. May 1;110(5):493–500. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lipscomb J, Gillespie TW, Goodman M, Richardson LC, Pollack LA, Ryerson AB, Ward KC. Black–white differences in receipt and completion of adjuvant chemotherapy among breast cancer patients in a rural region of the US. Breast cancer research and treatment. 2012. May;133:285–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Morimoto L, Coalson J, Mowat F, O’Malley C. Factors affecting receipt of chemotherapy in women with breast cancer. International journal of women’s health. 2010. Aug 9:107–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology: Breast Cancer version 4. 2023. [registration required]. https://www.nccn.org/guidelines/guidelines-detail?category=1&id=1419 [DOI] [PubMed]
- Neuner JM, Kong A, Blaes A, Riley D, Chrischilles E, Smallwood A, Lizarraga I, Schroeder M. The association of socioeconomic status with receipt of neoadjuvant chemotherapy. Breast cancer research and treatment. 2019. Jan 15;173:179–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sparano JA, Gray RJ, Makower DF, Pritchard KI, Albain KS, Hayes DF, Geyer CE Jr, Dees EC, Goetz MP, Olson JA Jr, Lively T. Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer. New England Journal of Medicine. 2018. Jul 12;379(2):111–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Sparano JA, Gray RJ, Ravdin PM, et al. Clinical and Genomic Risk to Guide the Use of Adjuvant Therapy for Breast Cancer. N Engl J Med. 2019;380(25):2395–2405. doi: 10.1056/NEJMoa1904819 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Spronk PE, van Bommel AC, Siesling S, Wouters MW, Peeters MV, Smorenburg CH. Variation in use of neoadjuvant chemotherapy in patients with stage III breast cancer: results of the Dutch national breast cancer audit. The Breast. 2017. Dec 1;36:34–8. [DOI] [PubMed] [Google Scholar]
- USDA ERS - Rural-Urban Continuum Codes Documentation [Internet]. Available from: https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/documentation/
- Warnecke RB, Oh A, Breen N, et al. Approaching Health Disparities From a Population Perspective: The National Institutes of Health Centers for Population Health and Health Disparities. Am J Public Health. 2008;98(9):1608–1615. doi: 10.2105/AJPH.2006.102525 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wheeler SB, Carpenter WR, Peppercorn J, Schenck AP, Weinberger M, Biddle AK. Predictors of timing of adjuvant chemotherapy in older women with hormone receptor–negative, stages II–III breast cancer. Breast cancer research and treatment. 2012. Jan;131:207–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Zhang L, King J, Wu XC, Hsieh MC, Chen VW, Yu Q, Fontham E, Loch M, Pollack LA, Ferguson T. Racial/ethnic differences in the utilization of chemotherapy among stage I-III breast cancer patients, stratified by subtype: Findings from ten National Program of Cancer Registries states. Cancer epidemiology. 2019. Feb 1;58:1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data analyzed in this study are available from the Carolina Breast Cancer Study (https://unclineberger.org/cbcs/). Restrictions apply to the availability of these data, which were used under data use agreements for this study. Data is not publicly available; however, investigators may submit a letter of intent to gain access upon reasonable request.