Real-world risk of recurrence and treatment outcomes with adjuvant endocrine therapy in patients with stage II-III HR+/HER2- early breast cancer

Abstract (246 words)

Background

Despite adjuvant endocrine therapy (ET), recurrence is still a concern for patients with HR+/HER2- early breast cancer (EBC). We assessed recurrence risk in real-world patients with stage II/III HR+/HER2- EBC treated with adjuvant ET.

Methods

A retrospective analysis was conducted using the ConcertAI Patient360 database (January 1995 to April 2021) of patients with stage II/III HR+/HER2- EBC ≥18 years who underwent surgery and received adjuvant ET. Risk of recurrence was assessed using invasive disease-free survival (iDFS) with adapted STEEP criteria. An ET subanalysis evaluated iDFS, distant disease-free survival, and overall survival in patients receiving adjuvant non-steroidal aromatase inhibitors (NSAI) vs tamoxifen.

Results

In the full analysis cohort (N = 3133), the risk of an iDFS event was 26.1 % at 5 years, rising to 45.0 % at 10 years. Among patients with stage II disease, the risk of an iDFS event at 5 and 10 years was 22.7 % and 40.5 %, respectively; stage III 5- and 10-year risk was 40.4 % and 62.9 %. Patients with node-negative disease had 5- and 10-year risks of 22.1 % and 36.9 %, respectively; node-positive 5- and 10-year risk was 28.9 % and 49.4 %. ET subanalysis showed improved iDFS with NSAI ± ovarian function suppression vs tamoxifen ± ovarian function suppression (HR, 0.83; 95 % CI, 0.69–0.98; p = 0.031); this trend was observed regardless of menopausal status.

Conclusions

This real-world study highlights the considerable risk of recurrence with adjuvant ET in patients with stage II or III HR+/HER2- EBC (including node-negative disease) and confirms the need for improved treatment options.

Highlights

• •Recurrence remains a concern with HR+/HER2-early breast cancer (EBC).
• •Contemporary data are needed to better understand risk of recurrence.
• •Real-world analysis of 3133 patients (stage II/III) HR+/HER2- EBC was conducted.
• •At 5 and 10 years, risk of invasive disease was 26.1 % and 45.0 %.
• •Results show unmet need remains despite standard-of-care endocrine therapy.

1. Introduction

An estimated 2.3 million new cases of breast cancer (BC) were diagnosed globally in 2022, most in the early (EBC) stage [1,2]. Hormone receptor–positive (HR+)/human epidermal growth factor receptor 2–negative (HER2-) EBC is the most common subtype, for which the standard-of-care treatment includes at least 5 years of adjuvant endocrine therapy (ET) [3,4]. However, despite adjuvant ET, the risk of recurrence remains a concern that persists for decades; ≈50 % of relapses occur after 5 years [5]. A recent meta-analysis that included 14 randomized clinical trials analyzed the risk of recurrence for patients with stage I-IIIC HR+/HER2- EBC within the first 5 years of starting adjuvant ET. Among these patients, the 3-year absolute risk of recurrence was 7 %, which rose to 12 % by 5 years [6].

For longer-term risk of recurrence, a patient-level meta-analysis of 88 studies conducted by Pan et al. included 62,923 women with HR+ BC and examined recurrence rates following a 5-year treatment period with ET [7]. This meta-analysis showed that despite adjuvant ET, recurrences were observed at 5 years and persisted to 20 years. For patients with stage II disease, risk of distant recurrence at 20 years was ≈27–37 %, and for patients with stage III disease, it was ≈46–57 %. Although Pan et al. shed new light on recurrence risk beyond 5 years, the meta-analysis included clinical trials that were enrolled between 1976 and 2011. A contemporary analysis of recurrence risk beyond the 5-year time point is needed to understand long-term risk of recurrence.

For patients with HR+/HER2- EBC, treatment guidelines currently recommend either an aromatase inhibitor (AI) or tamoxifen, based on their menopausal status and clinical risk [3,4]. Real-world studies have shown that adjuvant non-steroidal AI (NSAI) is used more often than tamoxifen overall; however, patterns of use vary by menopausal status, with postmenopausal patients more typically treated with adjuvant AI and premenopausal patients with tamoxifen [8]. Individual randomized clinical trials and meta-analyses comparing AIs and tamoxifen have shown greater efficacy with an AI in HR+/HER2- EBC [[9], [10], [11], [12], [13], [14]]. Although randomized clinical trials have provided valuable data for understanding the efficacy of AIs and tamoxifen, a real-world analysis can provide complementary data, which allows for a broader assessment of patients treated with AI and/or tamoxifen outside of a clinical trial setting and can potentially provide greater generalizability for clinical practice.

There is a need to advance understanding of the risk of recurrence in HR+/HER2- EBC using data that include contemporary standards of care. Conducting real-world studies alongside prospective clinical trials can provide useful insights on diverse patient populations and clinical settings. Therefore, we conducted a real-world study to assess the risk of recurrence over 10 years, from initiation of adjuvant ET, in patients with stage II/III HR+/HER2- EBC [15]. The focus of the analysis was on patients with stage II/III disease due to the higher risk of recurrence in these stages. A subanalysis was also conducted to investigate the outcomes of patients receiving NSAIs vs tamoxifen.

2. Methods

2.1. Overview

This retrospective cohort study was conducted on patients enrolled in the ConcertAI Patient360 BC database (data cutoff dates: January 1, 1995, to April 30, 2021) with HR+/HER2- EBC who received adjuvant ET. The ConcertAI Patient360 database comprises deidentified electronic medical records from 6 million patients receiving treatment at 400+ US academic (30 %) and community (70 %) oncology clinics.

2.2. Study population

Eligible patients in the ConcertAI Patient360 database had a curated diagnosis for BC and formed the initial “BC cohort.” From the initial BC cohort, the key eligibility criteria were non-missing gender data and age ≥18 years at the time of initial diagnosis with stage II or III HR+/HER2- EBC. Patients classified in the database as stage II, IIA, IIB, or IIIA were included (AJCC [American Joint Committee on Cancer], 8th Edition); for those classified as “stage III,” “stage IIIB,” or “stage IIIC” in the database, residual disease status of 0 (ie, complete surgical resection of the tumor) was required. Staging was based on the earliest curated stage from the date of diagnosis (in the absence of curated entry, the worst pathological staging information was prioritized). Patients were required to have undergone primary breast surgery (mastectomy, partial mastectomy, or lumpectomy) and to have received at least one qualifying adjuvant ET regimen (NSAI or tamoxifen, both with or without ovarian function suppression [OFS]). Patients were included if they had an index date (the date of initial ET commencement post surgery) of ≥6 months before the data cutoff, which occurred on April 30, 2021. Before the index date, patients must have had no record of recurrent cancer and no documentation of other concurrent invasive malignancies. Patients meeting all the inclusion criteria were referred to as the "EBC cohort." For the ET subanalysis, patients were split into those who received NSAI ± OFS or tamoxifen ± OFS (patients were required to receive only the respective ET; those who switched ET classes were not included in the ET subanalysis). There were no explicit criteria excluding patients treated with CDK4/6i; however, adjuvant abemaciclib was approved in October 2021, which fell after the database cutoff date used in this analysis. Patients with missing information were excluded from the EBC cohort.

2.3. Statistical analyses

Invasive disease-free survival (iDFS) and distant disease-free survival (DDFS) endpoints were based on the Standardized Definitions for Efficacy End Points (STEEP) 2.0 criteria. The adapted iDFS endpoint incorporated the events of disease recurrence, metastasis, second primary non-breast tumor, or death. These events were identified in the database through observation of a “tumor progression” (“Recurrent tumor” in the disease_status table; after index date) following a “complete response” (disease recurrence event), presence of a metastatic diagnosis date after the index date (metastasis event), curated (source documentation curated by ConcertAI database) date of death (from any cause), or second primary non-breast cancer after the index date (second primary tumor; based on ICD-10 codes identifying second primary cancers, excluding squamous or basal cell skin cancers or new in situ carcinomas of any site). Adapted DDFS events were identified in the database through the presence of a metastatic diagnosis date after index date, second primary non-breast cancer after the index date, or curated date of death (from any cause). Overall survival (OS) events in the database were defined as a curated death (from any cause) date.

Kaplan-Meier methods were used to estimate cumulative probabilities of experiencing an iDFS event at 2, 3, 5, and 10 years from the initiation of ET. The survival package in R was used to generate the models, and the 95 % CIs were generated specifying the log-log transformation. Age (50 years or older) was used as a proxy for menopausal status where the status was unknown/undocumented or documented as perimenopausal [16]. Male patients were analyzed with premenopausal patients. Median follow-up time was calculated based on maximum follow-up date from the index date. Maximum follow-up date was calculated using one of the following: death (if death had occurred), last known data entry in any dataset, or end of study period (final data cut from ConcertAI).

For the ET subanalysis comparing NSAI to tamoxifen, unadjusted Kaplan-Meier analyses were reported stratified by first ET received (excluding patients who had switched ETs). Multivariate Cox regression analyses were performed and were adjusted for age, stage, Charlson Comorbidity Index, and prior adjuvant or neoadjuvant chemotherapy. The final model was selected to avoid multicollinearity of the variables.

3. Results

3.1. Patient demographics and clinical characteristics

A total of 33,714 patients were identified from the database and classified as the BC cohort (Fig. 1). Among this initial BC cohort, 32,057 patients had a non-missing gender record and were ≥18 years old, of whom 25,927 underwent surgical resection. A total of 13,697 patients had HR+/HER2- BC, and 5258 had stage II/III BC. Of these patients, 3750 had treatment with ≥1 qualifying ET therapies. After exclusion of patients with staging entries indicating metastatic disease, there were 3321 patients. Finally, 3133 patients had a qualifying index date, no record of recurrent tumor prior to the index date of recurrence, and no record of another invasive malignancy prior to the index date. Therefore, the final “EBC cohort” used in this analysis consisted of 3133 patients.

Fig. 1.

Patient attrition flow diagram.

BC, breast cancer; EBC, early breast cancer; ER, estrogen receptor; ET, endocrine therapy; HER2, human epidermal growth factor receptor 2; NSAI, non-steroidal aromatase inhibitor; OFS, ovarian function suppression; PR, progesterone receptor.^aStaging was based on the earliest curated stage from the date of diagnosis (in the absence of curated entry, the worst pathological staging information was prioritized). ^bQualifying adjuvant ET regimens: NSAIs and tamoxifen (with or without OFS).

The majority of patients in the EBC cohort (N = 3133) were female (98.8 %). The median age was 59 years (range: 24–87 years). Most patients had stage II disease (80.9 %), and most had positive lymph nodes (per tumor, node, metastasis staging; 57.1 %) (Table 1). Most patients were postmenopausal (70.1 %). Age (≥50 years) was used as a proxy to determine menopausal status when this was unknown/undocumented, which was the case for 33.5 % of patients. A total of 40.5 % received (neo)adjuvant chemotherapy prior to index date, and 48.8 % received radiation therapy as part of their BC treatment plan. Most patients had partial mastectomy (51.8 %), while 30.3 % had mastectomy and 21.61 % had other/unspecified resection. The distribution of included patients by index date is shown in Supplemental Table 1.

Table 1.

Key demographic and clinical characteristics at date of ET initiation.

Characteristic	EBC cohort (N = 3133)	ET subanalysis
		NSAI ± OFS (n = 1854)	TAM ± OFS (n = 653)
Gender, n (%)
Female	3095 (98.8)	1847 (99.6)	630 (96.5)
Male	38 (1.2)	7 (0.4)	23 (3.5)
Race/ethnicity, n (%)
White	2229 (71.1)	1317 (71.0)	438 (67.1)
Black	466 (14.9)	292 (15.7)	94 (14.4)
Asian	98 (3.1)	50 (2.7)	30 (4.6)
American Indian or Alaska Native	NA	9 (0.5)	4 (0.6)
Native Hawaiian or Other Pacific Islander	NA	183 (9.9)	1 (0.2)
Other or unknown	340 (10.9)	183 (9.9)	86 (13.2)
Age
Mean (SD), years	58.37 (12.4)	62.9 (10.2)	48.0 (11.3)
Menopausal status, n (%)a,b
Postmenopausal	2195 (70.1)c	1689 (91.1)	181 (27.7)
Premenopausal	938 (29.9)d	165 (8.9)	472 (72.3)
Stage, n (%)
II	2535 (80.9)	1497 (80.7)	532 (81.5)
III	598 (19.1)	357 (19.3)	121 (18.5)
Nodal status, n (%)
Positive	1790 (57.1)	1060 (57.2)	350 (53.6)
Negative	1089 (34.8)	794 (42.8)	303 (46.4)
Prior chemotherapy, n (%)
Neoadjuvant and/or adjuvant	1270 (40.5)	–	–
Neoadjuvant	164 (5.2)	92 (5.0)	36 (5.5)
Adjuvant	1140 (36.4)	648 (35.0)	253 (38.7)
Radiation therapy, n (%)	1528 (48.8)	902 (48.7)	323 (49.5)

EBC, early breast cancer; ET, endocrine therapy; NSAI, non-steroidal aromatase inhibitor; OFS, ovarian function suppression; TAM, tamoxifen.

^aIf menopausal status was not documented, age (≥50 years) was used as a proxy.

^bIn the EBC cohort, menopausal status was documented as: premenopausal, 21.90 %; perimenopausal, 0.06 %; postmenopausal, 44.21 %; unknown/undocumented, 33.45 %. Twelve additional patients were categorized as other menopausal statuses, including acquired absence of ovaries bilateral, unspecified/other specified perimenopausal and menopausal disorders, postmenopausal atrophic vaginitis, and menopausal and female climacteric states.

^cStatus of 36.9 % of postmenopausal patients were imputed.

^dStatus of 26.8 % of premenopausal patients were imputed.

3.2. Risk of relapse: iDFS in the EBC cohort

The median follow-up time for the analysis was 68.1 months. In the EBC cohort, the median time to an iDFS event was 141.3 months (95 % CI, 131.0–152.0) (Fig. 2). At 5 and 10 years, the risks of an iDFS event were 26.1 % and 45.0 %, respectively. Among patients with stage II disease, median time to an iDFS event was 160.2 months (95 % CI, 145.7–184.5); 5-year and 10-year risk of iDFS event were 22.7 % and 40.5 % (Fig. 3A). For patients with stage III disease, the median time to an iDFS event was 82.6 months (95 % CI, 72.8–91.7). The 5- and 10-year risks of an iDFS event were 40.4 % and 62.9 %. Multivariate Cox regression analysis conducted on the time to iDFS (missing observations dropped [n = 2596]) indicated that patients with stage III disease had a higher risk of iDFS event than those with stage II disease (HR, 1.89 [95 % CI, 1.62 to 2.21, P < 0.001]—using stage II as reference).

Fig. 2.

iDFS in the EBC cohort.

EBC, early breast cancer; iDFS, invasive disease-free survival.

Fig. 3.

iDFS in the EBC cohort by (A) disease stage (B) nodal status and (C) menopausal status.

EBC, early breast cancer; iDFS, invasive disease-free survival.

Patients with node-positive disease had a higher risk of having an iDFS event (5-year: 28.9 %; 10-year: 49.4 %) than node-negative patients (5-year: 22.1 %; 10-year: 36.9 %) (Fig. 3B). This higher risk in node-positive patients was generally consistent in stage II and stage III patients (Supplemental Table 2). Risk of an iDFS event was comparable between patients who were premenopausal at diagnosis (5-year: 24.0 %; 10-year: 41.9 %) or postmenopausal (5-year: 27.0 %; 10-year 46.4 %) (Fig. 3C). Patients who received (neo) adjuvant chemotherapy had a higher risk of an iDFS event (5-year: 32.6 %; 10-year: 51.0 %) than those who did not receive (neo) adjuvant chemotherapy (5-year: 22.2 %; 10-year: 41.4 %).

3.3. Endocrine therapy subanalysis

For the ET subanalysis, 626 patients were excluded from the EBC cohort due to missing information (n = 46) or switching ET class during the follow-up period (n = 580). Of the 2507 patients in the ET subanalysis cohort, 1854 (74 %) received an NSAI with or without OFS and 653 (26 %) received tamoxifen with or without OFS (Fig. 1). The majority of patients in the NSAI and tamoxifen groups were female (99.6 % and 96.5 %, respectively) with stage II disease (80.7 % and 81.5 %, respectively) and had positive lymph nodes (57.2 % and 53.6 %, respectively) (Table 1). The NSAI group had a higher mean age than the tamoxifen group (62.9 vs 48.0 years, respectively) and a higher proportion of postmenopausal patients (91.1 % vs 27.7 %).

The median time to treatment discontinuation was comparable for patients who received NSAIs and tamoxifen (41.3 and 37.1 months, respectively) (Table 2). For all patients, the date of death, loss to follow-up, or study end was used as the date of ET discontinuation. The reason for ET discontinuation was undocumented in 76.8 % of patients in the ET subanalysis (due to missingness in the data).

Table 2.

ET subanalysis cohort: Time to discontinuation of ET treatment.

Subanalysis cohort	Median time to ET discontinuation (range), monthsa
Overall (N = 2507)	40 (0–216)
NSAI group (n = 1854)	41.3 (0–216)
Tamoxifen group (n = 653)	37.1 (0–205)

ET, endocrine therapy; NSAI, non-steroidal aromatase inhibitor.

^aThe minimum time to ET discontinuation was 1 day, but the data are reported in months.

Multivariate Cox regression analyses suggested a greater reduction in the risk of recurrence with NSAIs vs tamoxifen (Table 3). Treatment with an NSAI was associated with a 17 % reduction in the relative risk of having an iDFS event compared with treatment with tamoxifen (HR 0.83; 95 % CI, 0.69–0.98; p = 0.031) and an 18 % reduction in the relative risk of having a DDFS event (HR 0.82; 95 % CI, 0.69–0.98; p = 0.027). OS analysis demonstrated a HR of 0.88 (95 % CI, 0.69–1.13; p = 0.32) for deaths with NSAI vs tamoxifen; however, only 15.3 % and 19.1 % of patients had died in the tamoxifen and NSAI cohorts, respectively). Unadjusted Kaplan-Meier analyses stratified by ET and menopausal status showed that these trends in iDFS, DDFS, and OS were generally consistent in both pre- and postmenopausal patients (Supplemental Fig. 1). However, patient numbers for subgroups in postmenopausal patients (n = 181 treated with tamoxifen) and premenopausal patients (n = 165 treated with NSAI) were relatively low.

Table 3.

Multivariate Cox regression analysis of time to event for NSAI vs TAM (N = 2507).

Time to event	HR (95 % CI)a	P valuea
iDFS	0.83 (0.69–0.98)	0.031
DDFS	0.82 (0.69–0.98)	0.027
OS	0.88 (0.69–1.13)	0.32

DDFS, distant disease-free survival; iDFS, invasive disease-free survival; NSAI, non-steroidal aromatase inhibitors; OS, overall survival; TAM, tamoxifen.

^aPatients treated with TAM were the reference group and those treated with NSAI were the comparators.

4. Discussion

The results of this real-world study of patients with stage II or III HR+/HER2- EBC demonstrated an overall risk of an iDFS event of 26.1 % at 5 years and 45.0 % at 10 years. As expected, patients with stage III disease had a higher risk of an iDFS event than those with stage II, as did node-positive vs node-negative patients. It is noteworthy that patients with stage II (5-year risk 22.7 %) or node-negative (five-year risk: 22.1 %) EBC still faced considerable risk of recurrence. This is particularly noteworthy given that there were nearly four-fold more patients with stage II disease than with stage III in the EBC cohort. Additionally, those with node-negative disease made up about one-third of patients. These findings confirm the persistent unmet need in this broad HR+/HER2- EBC population, including those with stage II or node-negative disease.

Patients with prior (neo) adjuvant chemotherapy represented 40.5 % of the overall population and had a higher risk of an iDFS event at 5 and 10 years than those who did not receive chemotherapy. This was not unexpected, as prior (neo) chemotherapy use may be a proxy for higher-risk disease. On the other hand, no difference in risk of an iDFS event was observed between pre- and post-menopausal patients, which is somewhat unexpected. However, it should be noted that 33.5 % of patients in the EBC cohort had an unknown/undocumented menopausal status. While age was used as a proxy to determine menopausal status, a proportion of patients’ menopausal status may have been misclassified; this is a limitation that may have impacted the outcome of the subanalysis.

Consistent with prior studies, the ET subanalysis demonstrated that more patients (about three-fold greater in this analysis) were treated with NSAI than tamoxifen, with tamoxifen being largely used in premenopausal patients [8]. In multivariate analysis, after adjusting for age, stage, Charlson Comorbidity Index, and prior adjuvant or neoadjuvant chemotherapy, patients treated with NSAIs ± OFS had improved iDFS (17 % reduction in relative risk of event with NSAI), DDFS (18 % reduction in relative risk of event with NSAI), and OS outcomes than those treated with tamoxifen ± OFS. These results were generally consistent, regardless of menopausal status. Collectively, the findings of this study, as well as those from prior meta-analyses of randomized controlled trials, demonstrate a greater efficacy benefit with adjuvant NSAIs over tamoxifen in patients with HR+/HER2- EBC [9,12]. However, it should be noted that there were only 181 postmenopausal patients treated with tamoxifen and 165 premenopausal patients treated with NSAI in our study. Additionally, these results could also have been impacted by use of OFS in combination with ET. Additional analyses need to be conducted to explore these factors.

The meta-analysis conducted by Pan et al. provided insights into the long-term risk of recurrence among patients with HR+ EBC [7]. In their meta-analysis, patients with stage II disease had a risk of any breast cancer event (distant, local, or contralateral) at 10 years that was ≈19–29 %; for patients with stage III disease, this risk at 10 years was ≈34–45 %. However, the relevance of the findings from their meta-analysis to the outcomes in our real-world study is limited. Outcomes reported by Pan et al. were based on clinical trial data and may have differed in several respects from our real-world analysis, which used iDFS based on adapted STEEP criteria. Additionally, real-world studies include a more heterogeneous population than the randomized controlled trials included in the Pan et al. meta-analysis.

There are some limitations to this study. Real-world data were utilized from the ConcertAI Patient360 database, which is US based and representative of commercially insured patients. As clinical practice patterns in real-world settings may vary in older Medicare patients and across different countries, future research should explore real-world data that reflect a more diverse EBC population. Additionally, data on adherence or treatment duration were limited and were not accounted for. This may have impacted recurrence estimates, as poor adherence may have led to poorer outcomes and higher estimated risk of recurrence. Furthermore, this retrospective analysis of an electronic medical record database is limited by missing, incomplete, or duplicate data for some variables. For example, patients were not included if they had multiple HER2 measurements if at least one was HER2 positive. This also applied to staging—if a patient had multiple staging records, any record of stage IV led to exclusion from the analysis. Another example was the use of residual cancer status of 0 after surgery to rule out patients with locally advanced disease; patients who were missing this variable in the records were excluded. These missing, incomplete, or duplicate data may have impacted the composition of the patients included in the EBC cohort, which may in turn have had an impact on the outcomes reported. Finally, there were differences between this real-world study and clinical trials that may lead to differences in their outcomes. The definition of iDFS used in this study was a derivation of the STEEP criteria that was adapted for use with the real-world data. Outcomes data in a real-world setting may have been captured differently than in a clinical trial. Recurrence was also based on provider judgement, which may not have matched the criteria used in clinical trials. Finally, STEEP 2.0 criteria specify alternatives to iDFS (eg, invasive breast cancer–free survival) that may be used as primary endpoints depending on the trial design. Future studies should include these endpoints for analysis.

In conclusion, this real-world evidence study demonstrates that risk of recurrence among patients with stage II or III HR+/HER2- EBC, including in those with node-negative disease, remains unacceptably high despite standard adjuvant systemic therapy, justifying the development and use of additional therapeutic strategies for this population.

CRediT authorship contribution statement

Joyce O'Shaughnessy: Writing – review & editing, Investigation, Conceptualization. Sara M. Tolaney: Writing – review & editing, Investigation, Conceptualization. Denise A. Yardley: Writing – review & editing, Investigation, Conceptualization. Lowell Hart: Writing – review & editing, Investigation, Conceptualization. Pedram Razavi: Writing – review & editing, Investigation, Conceptualization. Peter A. Fasching: Writing – review & editing, Investigation, Conceptualization. Wolfgang Janni: Writing – review & editing, Investigation, Conceptualization. Lee Schwartzberg: Writing – review & editing, Investigation, Conceptualization. Julia Kim: Writing – review & editing, Methodology, Investigation, Formal analysis, Conceptualization. Murat Akdere: Writing – review & editing, Methodology, Investigation, Formal analysis, Conceptualization. Courtney McDermott: Writing – review & editing, Methodology, Investigation, Formal analysis, Conceptualization. Aamir Khakwani: Writing – review & editing, Methodology, Investigation, Formal analysis, Conceptualization. Purnima Pathak: Writing – review & editing, Methodology, Investigation, Formal analysis, Conceptualization. Stephanie L. Graff: Writing – review & editing, Investigation, Conceptualization.

Funding sources

Medical editorial assistance was provided by Nucleus Global and was funded by Novartis Pharmaceuticals Corporation.

Declaration of competing interest

J. O'Shaughnessy reports personal fees from AbbVie, Agendia, Amgen, Aptitude, AstraZeneca, Bristol-Myers Squibb, Celgene, Eisai, G1 Therapeutics, Genentech, Immunome, Ipsen Biopharmaceuticals, Lilly, Merck, Myriad, Novartis, Odonate Therapeutics, Pfizer, Puma, Prime, Roche, Seattle Genetics, Syndax, Carrick Therapeutics, Daiichi Sankyo, Gilead Sciences, Ontada, Pierre Fabre, Samsung, Sanofi. S. Tolaney reports grant funding to institution and personal fees from Eli Lilly, Novartis, AstraZeneca, Merck, Pfizer, Genentech/Roche, Bristol-Myers Squibb, Eisai, Sanofi, Daiichi Sankyo, Gilead, Jazz Pharmaceuticals, Menarini/Stemline; grant funding to institution from Exelixis, Nanostring, OncoPep; personal fees from Seattle Genetics, Arvinas, Cullinan Oncology, eFFECTOR, CytomX, Sumitovant Biopharma, Natera, Tango Therapeutics, SystImmune, Hengrui USA, Blueprint Medicines, Reveal Genomics, Umoja Biopharma, Zentalis, Zymeworks, Circle Pharma, Bayer, Incyte Corp, Aadi Biopharma. D. Yardley reports research funding grants to institution from Ambrx, Amgen, AstraZeneca, BIOMARIN, Biothera Pharmaceuticals, Clovis Pharma, Dana Farber Cancer Institute, Lilly, Roche/Genentech, G1 Therapeutics, Incyte, Innocrin Pharmaceuticals, MacroGenics, MedImmune, Medivation, Merck, Merrimack Pharmaceuticals, Nektar Therapeutics, Novartis, NSABP, Polyphor, Stemline Therapeutics, UT Southwestern; consulting/advisory fees to institution from AstraZeneca, G1 Therapeutics, Gilead Sciences, Immunomedics, Integra Connect, Novartis, Sanofi-Aventis, Stemline Therapeutics. L. Hart reports grants to institution and personal fees from Novartis. P. Razavi reports institutional grants from Grail/Illumina, Novartis, AstraZeneca, Epic Sciences, Invitae/ArcherDx, Tempus, Inivata, BioTheranostics; personal fees from Novartis, AstraZeneca, Epic Sciences, Tempus, Inivata, Foundation Medicine, Pfizer, Daiichi Sankyo, Natera, Paige.AI, SAGA Diagnostics. P. Fasching reports personal fees from Novartis, Pfizer, Daiichi Sankyo, AstraZeneca, Eisai, Merck Sharp & Dohme, Lilly, Pierre Fabre, SeaGen, Roche, Agendia, Sanofi Aventis, Gilead, Mylan, Menarini, Medac, Veracyte, Guardant Health; institutional grants from Biontech, Pfizer, Cepheid. W. Janni reports personal fees from Amgen, AstraZeneca, Daiichi Sankyo, Lilly, MSD, Novartis, Pfizer, Roche, Seagen, Gilead; employment from Universitätsklinikum Ulm; speaker fees to institution from Novartis, GSK, Sanofi, Amgen, Roche, Lilly; chair of AGO Breast Council. L. Schwartzberg reports personal fees from AstraZeneca, Daiichi Sankyo, Genentech, Novartis, Pfizer, Seagen, Merck, Spectrum, Napo, Amgen. J. Kim reports employment from Genesis Research, whom Novartis has paid fees for consulting. M. Akdere, C. McDermott report employment and stock ownership from Novartis. A. Khakwani reports employment and stock ownership from Novartis; prior employment from Imperial Brands PLC. P. Pathak reports employment and stock ownership from Novartis. S. Graff reports personal fees from Novartis, Pfizer, AstraZeneca, Genentech, Lilly, Daiichi Sankyo, AstraZeneca, Gilead Sciences, The Academy for Healthcare Learning, DAVA Oncology, MJH Life Sciences, WebMD/Medscape, IntegrityCE, MedPage Today, MedIQ, Medical Educator Consortium, Research to Practice; consulting fees paid to institution from Seagen; other from HCA Healthcare; research grants to institution from Daiichi Sankyo, Novartis, AstraZeneca.

Glossary

AI

aromatase inhibitor

BC

breast cancer

DDFS

distant disease-free survival

EBC

early breast cancer

ET

endocrine therapy

iDFS

invasive disease-free survival

NSAI

non-steroidal aromatase inhibitors

OFS

ovarian function suppression

OS

overall survival

STEEP

Standardized Definitions for Efficacy End Points

Appendix A. Supplementary data

The following is the Supplementary data to this article: