Real-world treatment patterns and outcomes in HER2-low, HR+ metastatic breast cancer patients previously treated with endocrine therapy in the United States.

1
Introduction
Human epidermal growth factor receptor 2 (HER2)-negative, hormone receptor (HR)-positive accounts for 70% of breast cancer cases [1]. The current standard of care for HER2-negative, HR-positive metastatic breast cancer (mBC) is first-line endocrine therapy (ET), typically in combination with a cyclin-dependent kinase 4 and 6 inhibitor (CDK4/6i). Eventual resistance to ET necessitates subsequent treatment with chemotherapy [2,3]. In this population, disease progression following multiple lines of ET, or rapidly following the first line of ET, is associated with limited efficacy of conventional single-agent chemotherapy in later lines of treatment [4,5]. Therefore, despite positive outcomes with first-line (1L) ET + CDK4/6i, the optimal sequence of therapies following disease progression remains unclear [2].
A spectrum of HER2 expression exists within HER2-negative mBC, and cancers with immunohistochemical (IHC) analysis scores of 1+ or 2+ with negative in situ hybridization (ISH) are defined as HER2-low [6,7]. It is estimated that up to 60% of mBC cases historically classified as HER2-negative are in fact HER2-low [8]. This has important therapeutic implications, expanding options for this population beyond ET and chemotherapy towards new HER2-targeted treatments.
Following the phase 3 DESTINY-Breast06 trial (NCT04494425), the HER2-directed antibody-drug conjugate trastuzumab-deruxtecan (T-DXd) was approved in the United States and European Union for the treatment of adult patients with unresectable or metastatic HR-positive, HER2-low or HER2-ultralow (IHC 0 with membrane staining) breast cancer that has progressed on one or more endocrine therapies in the metastatic setting [9,10]. In DESTINY-Breast06,T-DXd was associated with a statistically significant and clinically meaningful longer progression-free survival (PFS) of 13.2 months (95% confidence interval [CI]: 11.4, 15.2) versus 8.1 months (95% CI: 7.0, 9.0) with chemotherapy in patients with mBC previously treated with ≥1 lines of ET and no prior chemotherapy in the metastatic setting [11]. The European Society for Medical Oncology (ESMO) also recommended in 2025 that T-DXd may be considered in HER2-low or HER2-ultralow mBC after ≥2 lines ET or 1 line ET for mBC if progression occurs within 24 months of adjuvant ET or 6 months of 1L ET + CDK4/6i, respectively [12]. T-DXd therefore provides an alternative treatment option to chemotherapy following disease progression on ET.
Given the emergence of new HER2-low and -ultralow-targeted treatments, such as T-DXd, greater contextualization of the effect of existing therapies in this population is needed to understand the potential impact and optimal sequencing of these novel therapies. We conducted a real-world, observational, retrospective database study of a United States (US)-based cohort meeting similar inclusion and exclusion criteria as the DESTINY-Breast06 trial to further define the demographic and clinical characteristics, current treatment patterns, and outcomes among HER2-low, HR-positive mBC patients who initiated a line of chemotherapy after at least one line of ET in the metastatic setting.

2
Methods
2.1
Study design and eligibility criteria
This study used the US-based, electronic health record-derived de-identified Flatiron Health Research database [13]. The Flatiron Health database is a longitudinal database comprising de-identified patient-level structured (e.g. patient birth sex, race/ethnicity, and diagnosis information) and unstructured data (e.g. physician notes, radiology and pathology notes) curated via technology-enabled abstraction. During the study period, de-identified data originated from approximately 280 United States cancer clinics (∼800 sites of care). Most patients in the database originated from community oncology settings; the proportion from community/academic practices varied depending on study cohort. For this analysis, a September 2023 data cutoff was used.
The study included patients diagnosed with HER2-low, HR-positive mBC from January 1, 2011, to August 31, 2023 (study period). Patients initiated chemotherapy (index line of therapy [ILOT]) with an index treatment date occurring ≥90 days prior to the end of the study period. Two subpopulations were defined: “rapid ILOT initiators” initiated ILOT within 6 months of initiating 1L ET + CDK4/6i in the metastatic setting and the “index 3L+ group” initiated ILOT following ≥2 previous lines of ET with or without CDK4/6i in the metastatic setting (Fig. 1). The cohort met additional inclusion and exclusion criteria similar to the DESTINY-Breast06 clinical trial (Supplementary Table 1) [11].
Lines of therapy (LOTs) were defined according to Flatiron Health's oncologist-defined, regimen-based algorithm. Further details on the definition of LOTs and the systemic anti-cancer therapies in each LOT have been described previously [14].

2.2
Study measures
Demographic and clinical characteristics were reported at ILOT initiation. Treatment patterns were described anchored by the ILOT to show how individual regimens were sequenced.
Real-world overall survival (rwOS) was defined as the time from start of ILOT to death; patients without an event were censored at last confirmed activity date. Real-world time to treatment discontinuation or death (rwTTD/D) was defined as the time from start of ILOT to treatment discontinuation or death, whichever occurred first. If a patient had no subsequent LOT and no recorded date of death but had confirmed structured activity ≥120 days after treatment discontinuation date (i.e., indicating drug cessation not due to missing data), the patient was recorded as having a discontinuation event at the treatment discontinuation date. If a patient had no subsequent LOT and no confirmed structured activity ≥120 days after treatment discontinuation date, the patient was censored at treatment discontinuation date. Real-world time to next treatment or death (rwTTNT/D) was defined as the time from start of ILOT to initiation of the next LOT or death, whichever occurred first. If a patient had no subsequent LOT and no recorded death date, the patient was censored at last confirmed activity date.

2.3
Statistical analysis
All analyses were exploratory in nature and were performed in SAS 9.4+ (SAS Institute Inc., Cary, North Carolina, United States). Data were reported for the overall study cohort and stratified by subpopulation.
Descriptive statistics were used to describe patient characteristics and treatment patterns. Continuous data were described by the number of non-missing values, median, mean, standard deviation, as well as lower and upper quartiles. Categorical data were summarized using frequencies and percentages (Supplementary Table 2).
Sankey diagrams were generated to visualize the sequence of treatment regimens received across LOTs. The LOT sequence included ILOT−1, ILOT, ILOT+1, and ILOT+2. Time-to-event analyses were performed using Kaplan-Meier (KM) estimation from the start of ILOT for rwOS, rwTTD/D, and rwTTNT/D. Median follow-up time was calculated by the reverse KM method.

3
Results
3.1
Study population
In total, 35,546 patients aged ≥18 years diagnosed with mBC between January 1, 2011 and August 31, 2023 were included in the database. Following the application of the inclusion and exclusion criteria, 963 patients were selected for the final study cohort (Fig. 2). Within the cohort, 131 (13.6%) “rapid ILOT initiators” initiated 2L chemotherapy within 6 months of initiating 1L ET + CDK4/6i and 832 (86.4%) patients in the “index 3L+ group” initiated 3L+ chemotherapy after ≥2 lines ET.
Patient characteristics at the start of the ILOT are presented in Table 1. Median age was 64.0 years (quartile [Q]1, Q3: 55.0, 72.0); most patients were female (98.3%) and white (71.0%). The most common IHC category was IHC 1+ (70.6%) and most patients were both estrogen receptor (ER) and progesterone receptor (PR) positive (68.8%). At mBC diagnosis, 271 (28.1%) patients presented with de novo stage IV cancer. Per study inclusion criterion, all patients had Eastern Cooperative Oncology Group (ECOG) adequate performance status (ECOG performance score of 0 or 1) within 30 days prior to and 7 days following ILOT initiation, with most patients having a performance score of 1 (60.9%).
Baseline characteristics were broadly similar between the overall cohort, rapid ILOT initiators, and index 3L+ group. Patients presenting with de novo stage IV cancer represented 16.0% of rapid ILOT initiators and 30.0% of the index 3L+ group. Additionally, the proportion of patients positive for both ER and PR was 50.4% and 71.8% for the rapid ILOT initiators and index 3L+ group, respectively.

3.2
Treatment patterns
Most patients (56.2%) received ILOT between 2019 and 2023 (Table 1). Treatment patterns before and after chemotherapy ILOT are presented in Fig. 3 and Supplemental Fig. 1. Prior to ILOT, the most common treatments among the overall cohort were ET + CDK4/6i (52.1%), ET + other targeted therapies (27.5%), and ET monotherapy (18.1%; Supplemental Fig. 1). The most common ILOTs were single agent chemotherapies including capecitabine (45.2%), paclitaxel (9.3%), and nab-paclitaxel (7.0%). Among patients receiving a subsequent LOT after ILOT (N = 702), 57.0% received another chemotherapy-based LOT, while 26.2% returned to treatment with ET ± targeted therapy. A breakdown of the most common treatment regimens by LOT for the overall cohort is provided in Supplementary Table 3.
Stratified by population, treatment patterns were similar from ILOT onwards (Fig. 3). The proportion of patients who received chemotherapy without ET or targeted therapy for ILOT was 73.3% for rapid ILOT initiators and 81.0% for the index 3L+ group. Among rapid ILOT initiator (N = 95) and index 3L+ (N = 607) patients who received a subsequent LOT after ILOT, 55.8% and 57.2%, respectively, received another chemotherapy-based LOT and 27.4% and 26.0%, respectively, returned to treatment with ET ± targeted therapy. The most common treatment regimens by LOT for each subpopulation are reported in Supplementary Table 4.

3.3
Clinical outcomes
Median follow-up time (95% CI) for the overall cohort was 41.8 months (37.4, 48.5) from the ILOT start date. A summary of clinical outcomes overall and stratified by subpopulation are presented in Table 2, Fig. 4, Fig. 5.
Median rwOS (95% CI) from the start of the ILOT was 18.7 months (17.3, 20.2) for the overall population with 1-, 3-, and 5-year survival probabilities (95% CI) of 65.6% (62.4, 68.6), 21.2% (18.2, 24.3), and 7.5% (5.3, 10.2), respectively. Median rwTTD/D and rwTTNT/D (95% CI) were 4.9 months (4.6, 5.2) and 5.6 months (5.3, 6.0) from start of ILOT, respectively (Fig. 4).
Stratified by subpopulation, median rwOS (95% CI) for rapid ILOT initiators was 15.2 months (11.6, 17.1) with 3- and 5-year survival probabilities of 15.6% (9.2, 23.5) and 5.0% (1.5, 11.8), respectively. Median rwOS (95% CI) for the index 3L+ group was 19.5 months (17.9, 21.1), with 3-year and 5-year survival probabilities of 22.1% (18.9, 25.5) and 8.1% (5.6, 11.0), respectively. Median rwTTD/D (95% CI) for the rapid ILOT initiator and index 3L+ group was 4.3 months (3.4, 5.0) and 5.0 months (4.7, 5.5), respectively, and median rwTTNT/D (95% CI) was 4.9 months (4.0, 5.9) and 5.7 months (5.3, 6.1), respectively (Fig. 5).

4
Discussion
HER2-targeted therapies were historically considered ineffective in HER2-negative mBC patients. The recognition that up to 60% of mBC historically classified as HER2-negative is HER2-low expands the population eligible for a new generation of HER2-targeted treatments and warrants a re-evaluation of the current standard treatment sequencing following ET [7,8]. In this real-world study, we characterized patients with HER2-low, HR-positive mBC who initiated a subsequent line of chemotherapy after ≥1 line of ET to better understand the effect of currently available therapies in this population.
The inclusion and exclusion criteria for this study cohort aligned with that of the phase 3 DESTINY-Breast06 trial, and baseline characteristics were similarly distributed between these studies, with around a third of patients in each study presenting with de novo stage IV mBC [11]. More HER2-low patients in the chemotherapy arm of DESTINY-Breast06 had ECOG performance status of 0 compared with the present study (61.6% vs 39.1%, respectively) and, likely due to the selection period of the current study, a higher proportion of these patients had received prior ET + CDK4/6i (89.3% vs 52.1%, respectively) [11]. While median LOTs prior to chemotherapy or study initiation in both this study and the chemotherapy arm of DESTINY-Breast06 was 2, this study included a higher proportion of patients who initiated chemotherapy within 6 months of CDK4/6i (13.6%) vs DESTINY-Breast06 (9.4%).
In the current study, most patients received single agent chemotherapy as their ILOT. Among those who received a subsequent LOT, most received another chemotherapy-based regimen (57.0%) or returned to ET ± targeted therapy (26.2%). Treatment patterns after ILOT were similar between rapid ILOT initiators and the index 3L+ group. The use of ET following chemotherapy was notable and may indicate patients received maintenance ET to prolong disease control and mitigate cumulative dose-related side effects [18]. The preference for a third line of ET suggests physicians may prioritize ET, potentially due to long-lasting responses under previous ET lines and/or concerns about the tolerability and limited benefit of chemotherapy.
Median rwTTD/D and rwTTNT/D, which serve as real-world proxy measures for PFS, were 4.9 and 5.6 months, respectively; median PFS was 8.1 months in the chemotherapy arm of DESTINY-Breast06 [11]. The 1-year survival rate of patients was 65.6% in the current study and 81.7% in the chemotherapy arm of DESTINY-Breast06 [11]. These real-world data show the limitations of standard-of-care chemotherapy following ET as observed in DESTINY-Breast06. The differing distributions of prognostic characteristics between the current study population and that of DESTINY-Breast06, including metastases locations, comorbidities, treatments available during the study period, and study locations, may have contributed to the lower survival rate observed here.
Low median rwOS, rwTTD/D, and rwTTNT/D were reported for rapid ILOT initiators in this study (15.2, 4.3, and 4.9 months, respectively). Second-line ET was not considered for rapid ILOT initiators, suggesting this subpopulation is likely made up of rapid progressors, although reasons for starting a subsequent treatment line were not captured. Notably, 77.9% of rapid ILOT initiators presented with recurrent stage IV mBC. Patients with recurrent disease may lose sensitivity to ET in the metastatic setting more quickly given prior ET treatment, potentially contributing to the rapid initiation of chemotherapy following 1L ET + CDK4/6i observed in this subpopulation [19]. Outcomes are generally worse for patients with recurrent versus de novo mBC which may partially account for the numerically shorter survival observed for rapid ILOT initiators compared with the index 3L+ group [20,21]. Overall, the results of the current study highlight unmet treatment needs among patients initiating chemotherapy shortly after beginning ET + CDK4/6i.
Clinical outcomes for this population could be improved with optimized treatment sequencing. Current standard of care for HER2-low mBC is 1L ET, usually with CDK4/6i, continued until treatments are exhausted or progression is observed with additional factors indicating the need for chemotherapy rather than continued ET [2]. The low rwTTD/D and rwTTNT/D reported here suggest patients did not experience prolonged treatment benefit with chemotherapy after ET. These observations are in line with a real-world study of Italian patients with HER2-negative, HR-positive mBC, which reported a median TTD of 5−6 months for 2L chemotherapy following ET + CDK4/6i [22]. Another real-world study of patients with HER2-negative, HR-positive mBC also reported poor outcomes among those who progressed during ET + CDK4/6i treatment, with a median PFS of 6.0 months following ET + CDK4/6i discontinuation [23]. In the single-agent chemotherapy arms of two randomized clinical trials, patients with HER2-negative, HR-positive mBC receiving paclitaxel or capecitabine had median PFS of 7.1 months and 6.9 months, respectively [4,5]. The current findings address a gap in the literature by defining treatment patterns and outcomes for the HER2-low patient population after prior ET specifically.
The results of our real-world evidence study highlight the challenges of achieving meaningful and durable clinical responses in HER2-low mBC patients using current chemotherapy-based regimens. They also underscore an unmet need for more treatment options among patients who initiate chemotherapy rapidly following 1L ET + CDK4/6i. Incorporating emerging HER2-targeted treatments such as T-DXd earlier into the metastatic setting has the potential to improve patient outcomes. This is supported by clinical trial data demonstrating increased PFS for patients with HER2-low mBC receiving T-DXd compared with chemotherapy [11]. More recent real-world data have further confirmed the effectiveness of T-DXd for HER2-low patients [15,16]. Overall, the results of our study support the incorporation of more effective treatments for patients who are no longer candidates for ET, in line with ESMO 2025 recommendations [12].
Strengths of this real-world study include its large cohort size and extensive follow-up period, longer than that of the DESTINY-Breast06 trial, which enabled a long-term analysis of this population including robust assessment of rwOS. Additionally, due to the data curation of the real-world database used in this study, we were able to specifically analyze HER2-low patients.
Nevertheless, a potential limitation of this study is that, due to cohort selection criteria requiring patients to have received chemotherapy, the observed treatment patterns reflect only a subset of the broader population with HER2-low mBC. Further, the LOT algorithm used was not strictly progression-based, meaning the end of a line could be due to reasons besides a progression or death event (e.g., treatment intolerance). The study period also overlapped with critical developments in mBC treatment, particularly in the use of 1L CDK4/6i. These agents, which received approval between 2015 and 2017 are now recommended as standard-of-care 1L treatment for HER2-low, HR-positive mBC [2,[17], [24], [25]]. Almost half of the patients in this study were not treated with 1L CDK4/6i, reflecting the evolving nature of clinical practice over the study's observation period. The shift to 1L ET + CDK4/6i has changed the disease course of HER2-negative, HR-positive mBC, in most cases delaying the initiation of chemotherapy. Patients who ultimately receive chemotherapy often do so later in the disease course, possibly in the context of more heavily pre-treated and endocrine-resistant disease. This shift in treatment approach may influence observed chemotherapy outcomes, including time to treatment discontinuation. Indeed, real-world studies of patients with HER2-negative, HR-positive mBC who received subsequent chemotherapy following progression on 1L CDK4/6i have observed median rwPFS of approximately 7 months, as compared with the rwTTD/D observed in this study of 4.9 months, and support this study's observations that patients with faster progression on 1L ET + CDK4/6i experience worse subsequent outcomes [26,27].
Additionally, this real-world cohort did not capture HER2-ultralow status, a unique sub-group whose recognition has expanded biological and clinical understanding of the spectrum of HER2 expression. HER-ultralow status is not currently a guideline-defined category [28] and is not routinely captured in clinical or administrative real-world databases. Consequently, tumors classified as HER2 IHC 0 in real-world datasets likely represent a heterogeneous group including both HER2-zero and -ultralow expression. As these patients could not be distinguished, patients with HER2 IHC 0 were not included in this study. Notably, DESTINY-Breast06 demonstrated similar outcomes for HER2-low and -ultralow populations, with PFS in the chemotherapy arms for each population of 8.1 and 8.3 months, respectively, and T-DXd is approved for the treatment of both populations [11]. Patients with HER2-zero (IHC 0) mBC were also not included in this analysis. Although real-world analyses suggest outcomes are similar across the spectrum of HER2 expression in mBC patients treated with 1L ET ± CDK4/6i [[29], [30], [31]], further data are needed on outcomes in this population with subsequent treatment lines. Regardless, the absence of HER2-ultralow and -zero patients and the inclusion of patients who did not receive 1L CDK4/6i limit the applicability of these findings to current treatment practices.
Finally, this study was constrained by the inherent limitations of retrospective, database-derived studies. Reliance on initial documentation quality, which varies among treatment centers, may have resulted in missing or inconsistent data. Additionally, data on potentially relevant disease characteristics may not have been consistently collected. As the US-based Flatiron Health Research database comprises data predominantly from community oncology practices, these findings may not reflect the broader population of mBC patients, particularly those treated in other geographies, due to differences in treatment availability and clinical practices.

5
Conclusion
In this real-world study, patients with HER2-low, HR-positive mBC treated with chemotherapy-based regimens following standard-of-care ET demonstrated an unmet need for more effective therapies. This is particularly true for patients who must quickly initiate subsequent treatment following 1L ET + CDK4/6i, for whom treatment sequencing with chemotherapy following ET is suboptimal. Emerging treatment strategies tailored to the HER2-low population could significantly improve clinical outcomes for these patients.

CRediT authorship contribution statement
Shanu Modi: Writing – review & editing, Conceptualization. Danalyn Byng: Writing – review & editing, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization. Suyuan Zhang: Writing – review & editing, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization. Yan Xiong: Writing – review & editing, Methodology, Investigation, Formal analysis, Conceptualization. Shannon Hunter: Writing – review & editing, Methodology, Investigation, Formal analysis. Kyle Dunton: Writing – review & editing, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization. Tara Harding: Writing – review & editing, Supervision, Project administration, Methodology. William Jacot: Writing – review & editing, Writing – original draft, Visualization, Supervision, Conceptualization.

Data sharing Statement
The data that support the findings of this study were originated by and are the property of Flatiron Health, Inc. Requests for data sharing by license or by permission for the specific purpose of replicating results in this manuscript can be submitted to PublicationsDataAccess@flatiron.com.

Ethical approval
All the results presented in this article are in aggregate form, and no personally identifiable information was used for this study. The data were de-identified and subject to obligations to prevent re-identification and protect patient confidentiality. This study did not constitute human subjects research under the Common Rule and did not require ethical approval or patient informed consent.

Funding sources
This study was sponsored by 10.13039/501100022274Daiichi Sankyo Europe GmbH and 10.13039/100004325AstraZeneca. Support for third-party writing assistance for this article was provided by Kaity McCafferty Layte, BSc, Lillian Dukes, BS, and Brian Vassallo, PhD, from Costello Medical, Inc, and was funded by 10.13039/501100022274Daiichi Sankyo in accordance with Good Publication Practice 2022 guidelines (https://www.ismpp.org/gpp-2022).

Declaration of competing interests
Shanu Modi: Grants or contracts (to institution): AstraZeneca, BioNTech, Daiichi Sankyo, Duality Bio, D3 Bio, Genentech, Nuvation, Pfizer, Seagen; consulting fees: AstraZeneca, Avacta, BioNTech, Boehringer Ingelheim, Daiichi Sankyo, Eli Lilly, Genentech, Gilead, Systimmune; payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events: AstraZeneca, Daiichi Sankyo, Seagen, Pfizer; support for attending meetings and/or travel: AstraZeneca, Daiichi Sankyo, Pfizer. Danalyn Byng: Employee of Daiichi Sankyo Europe GmbH. Suyuan Zhang, Yan Xiong, and Shannon Hunter: Employees of Daiichi Sankyo, Inc. Kyle Dunton: Employee of Daiichi Sankyo UK Ltd. Tara Harding: Employee of AstraZeneca. William Jacot: Grants or contracts: AstraZeneca, Daiichi Sankyo; payment or honoraria for lectures, presentations, speakers bureaus, manuscript writing, or educational events: AstraZeneca, BMS, Daiichi Sankyo, Eisai, Lilly (France), MSD, Novartis, Pfizer, Roche, Seagen; support for attending meetings and/or travel: AstraZeneca, Chugai, Eisai, GSK, Lilly (France), Novartis, Pfizer, Pierre Fabre, Roche, Sanofi Aventis; participation on a Data Safety Monitoring Board or Advisory Board: AstraZeneca, BMS, Daiichi Sankyo, Eisai, Gilead, Lilly (France), MSD, Novartis, Pfizer, Roche, Seagen.