Treatment patterns and clinical outcomes in chemotherapy-treated patients with HR+/HER2- metastatic breast cancer using real-world data in France and Germany.

Background
Breast cancer (BC) is the most frequent malignancy in women globally, with around 30% of BC patients developing metastatic breast cancer (mBC) [1]. In Europe, mBC is the leading cause of death from all cancers in women, accounting for nearly 4% of all deaths in women and around 2% of all deaths in both sexes [2, 3].
Hormone receptor-positive (HR +, > 10% immunohistochemistry [IHC] expression of oestrogen receptor and/or progesterone receptor, with 1–10% expression considered to represent low positivity [4]) and human epidermal growth factor receptor-2 negative (HER2 −, IHC score of 0, 1 +, or 2 + confirmed as negative by in situ hybridisation) mBC is the most common subtype, accounting for approximately 67–70% of all cases of BC in Europe [5, 6]. The HR status of mBC can be determined by IHC to predict patients who may benefit from endocrine therapy (ET), as recommended in both the European Society for Medical Oncology (ESMO) and American Society of Clinical Oncology/College of American Pathologists (ASCO/CAP) guidelines [2, 4].
ET combined with cyclin dependent kinase 4/6 inhibitors (CDK4/6i) is the current first-line standard of care for patients with HR +/HER2 − mBC [2]; however, high rates of ET resistance can occur in these patients, limiting treatment effectiveness [7]. In ET-resistant patients, single-agent chemotherapy (CT) was, until recently, the only treatment option [2]. However, prognosis for patients with ET-resistant mBC is poor, with data from randomised clinical trials, including the EMBRACE study (patients with mBC who had received 2–5 prior CT regimens), demonstrating a median overall survival (OS) with single-agent CT ranging from approximately 13 to 31 months [8–11]. This highlights the urgent need for alternative treatment options.
OS rates in mBC vary significantly according to tumour characteristics, patient factors and treatment options [12]. For example, treatment with eribulin significantly extended median OS compared with the treatment of physician’s choice for women with advanced/mBC and ≤ 3 prior CT regimens (13.1 vs. 10.6 months; P = 0.041). Moreover, in a subgroup analysis of patients with HER2-negative mBC enrolled in this trial, eribulin prolonged OS by a numerically greater number of months (14 vs 16 months) compared with capecitabine [13]. A real-world study conducted in the United States including 1,545 HR +/HER2 − mBC patients initiating their first CT mostly with capecitabine, demonstrated a median OS of around 23 months, which was observed to decrease with each subsequent line of therapy despite increasing use of eribulin between first and fourth CT regimens [14].
Real-world outcomes and patient characteristics can differ from those of clinical trials, primarily due to differences in patient selection criteria, which may be overly stringent in clinical trials, excluding older patients or those with worse disease characteristics [15]. An evidence gap exists around real-world survival outcomes and treatment patterns in Europe for patients with mBC, with previous real-world research focusing on clinical outcomes for specific treatments [16, 17], rather than multiple treatment types.
The objectives of the present study were to describe current real-world survival outcomes and treatment patterns among CT treated patients with HR +/HER2 − in France and Germany using anonymised healthcare data from Oncology Evidence Network (OEN) hospitals.

Methods

Study design and patient population
This was a descriptive retrospective cohort study analysing data from eligible patients identified between January 1, 2016, and February 28, 2022, (identification period). Eligible patients were adults ≥ 18 years at the earliest date of diagnosis, with HR +/HER2 − mBC (Stage IV [M1 and/or T4] through the International Classification of Diseases [ICD]−10 code C50 Malignant neoplasm of breast) who initiated at least one CT treatment on or after their diagnosis between January 1, 2016, and February 28, 2023 (study period). Data were directly extracted from electronic medical records at three European Oncology Evidence Network (OEN) sites in France and Germany: the Institut de Cancérologie de l'Ouest in France, and Universitätsmedizin Berlin – Charité and University Hospital Würzburg, both in Germany (Figure S1). Where needed, sites curated/enhanced patient records by reviewing clinical notes and medical reports.
Exclusion criteria were minimal to reduce the risk of selection bias and ensure a representative sample of real-world clinical practice. However, to ensure the safety and validity of the findings of this analysis, patients were excluded if they had been treated with HER2-targeted therapies (including trastuzumab, trastuzumab-emtansine, tucatinib, pertuzumab, and lapatinib; patients who received trastuzumab deruxtecan for HER2-low mBC were not excluded from the study) from mBC diagnosis to the end of the study period, had another primary cancer diagnosis in the 5 years prior to mBC diagnosis (except for non-metastatic and non-melanoma skin cancer), or were participating in a clinical trial at the first CT initiation (Table S1). In France, once the cohort meeting eligibility criteria had been identified, the final cohort sample was randomly selected among the eligible patients by the investigators.
The index date changed according to each study objective (described in the outcome section); therefore, a patient could have multiple index dates. The follow-up period was from the study assessment-specific index date until the earliest of death, loss to follow-up, or end of study period for each patient.

Data sources
Anonymised, patient-level data were directly extracted from each OEN Hospital’s oncology centre electronic medical records (EMR) and hospital data warehouses; where needed, the sites enhanced patient records by reviewing clinical notes and medical reports. The data were harmonised into a study-specific Common Data Model, which was validated to restrict the allowed data into clinically reasonable ranges. Patients with no record of death within the study period, or who were lost to follow-up, had their data censored at the last data availability or end of the study period. Each patient’s death status was derived from EMR or national death registry data from each country.

Outcome measures
The primary outcomes were OS (time from index date [that is the date of initiation of first CT] until death from any cause) and time to next treatment or death (TTNTD, time from CT line-specific start date to the earliest of subsequent CT start date or death) by CT line (from first to fourth). Mortality status for patients was ascertained from EMR at each oncology together with national death registry data from each country that was linked to the oncology centre’s EMR.
Other outcomes of interest included mBC treatment patterns, changes in the treatment landscape across the study years, and the discontinuation rate of any line of CT treatment.
Regarding mBC treatment patterns, all systemic treatments (not CT restricted) received as first to fourth line of therapy (LoT) from earliest date of mBC diagnosis (diagnosis index date) to the end of follow-up were recorded. Treatment types, identified using generic treatment names, are described in the supplement (Table S2), and were grouped into CT only, ET + CDK4/6i, ET + other targeted therapy, ET only (monotherapy or multiple ETs), and other regimens. Treatment patterns were described by identifying LoT and CT lines through an ad hoc algorithm developed according to the ESMO treatment guidelines and with input from clinical leads. First, all systemic treatments prescribed for mBC were extracted from each oncology centre’s EMR data and combined in LoTs according to pre-defined algorithm rules (Supplementary material 1). Second, those LoTs including at least one CT were assigned as CT lines and numbered sequentially from first to forth CT by discounting LoTs that did not include any CT. Notably, a CT line may include more than one CT and/or a combination of CT with non-CT.
Changes in treatment landscape for mBC were described as “treatment type” categories in each line of CT, stratified by start year from mBC diagnosis until the end of follow up.
Finally, evaluation of treatment discontinuation focused on patients who received CT only and was defined as the earliest of recorded end date of treatment or start date of a subsequent treatment line.

Subgroup analyses
For patients treated with CT only, baseline patient demographic, clinical, and disease characteristics and survival outcomes were reported in the pre-specified subgroups of interest, specifically for patients who were ≥ 65 years old and those with CDK4/6i exposure before initiation of first CT line.

Statistical analyses
All analyses were performed separately for France and Germany using R software. Patient characteristics and treatment patterns are summarised using descriptive statistics. Time-to-event outcomes (OS and TTNTD) were analysed using the Kaplan–Meier (KM) method, and were calculated from the index date (start date of each treatment of interest) until the date of an event, including death for OS, and the earliest of start of subsequent LoT or death from any cause for TTNTD. Censoring was applied at the earliest date of end of study period or loss to follow-up. The number of patients at risk and the number of censored patients and deaths, the survival probability with 95% confidence interval (CI) at 3, 6, 9, 12, 18, 24, 36, 48, and 60 months after the index date, and the median (25th–75th percentile) survival time are presented for each CT line.

Results

Patients
After applying the inclusion/exclusion criteria and random selection, 339 patients from France and 157 patients from Germany were included in this study (Figure S2). The overall sociodemographic and clinical characteristics were broadly similar in France and Germany, with 99% of patients being female in both countries. However, the median (interquartile range [IQR]) age at first CT initiation differed between 62 (51–71) years in France and 57 (49–67) years in Germany, with approximately 42% of the French cohort aged 65 or over, compared to 31% Germany (Table S3). Most patients were first diagnosed at early BC stage (72% in France and 64% in Germany), and the time from mBC diagnosis until first CT initiation was twice as long in Germany compared to France (6.5 vs. 3.3 months, respectively). The median (IQR) follow-up times were 17 (8–29) and 12 (5–24) months for France and Germany, respectively.

Clinical outcomes

Overall survival
In France, the median OS (IQR) from the start of first CT line was approximately 20 (8–39) months (Fig. 1), decreasing to 12 (6–24) months in patients starting a second, 8 (4–19) months for a third, and 7 (3–15) for a fourth CT line (Table S4). At first CT line, the 1-, 2-, and 3-year OS probabilities were 63%, 42%, and 26%, respectively.
In the German cohort, the median OS (IQR) from the start of the first CT line was approximately 24 (9–NA) months (Fig. 1). The OS dropped to 13 (6–NA) months for second CT line, 10 (5–NA) months for third, and 7 (2–NA) months for fourth CT line. At first CT line, the 1-, 2-, and 3-year OS probabilities were 66%, 49%, and 36%, respectively.
Overall, 42% (n = 144) of patients in France and 31% (n = 48) in Germany were aged ≥ 65 years. These patients had a shorter median OS (IQR) of 13 (5–28) and 17 (7–42) months compared to 24 (10–47) and 25 (10–NA) months in those < 65 years old, for France and Germany, respectively.
Overall, 43% (n = 146) of patients in France and 48% (n = 76) in Germany were exposed to CDK4/6i before starting their first CT treatment. These patients had a median OS (IQR) of 15 (6–27) months in France and 21 (10–24) months in Germany when CDK4/6i exposure was < 12 months, and 19 (11–37) and 25 (6–NA) months in France and Germany, respectively, when exposed for ≥ 12 months. Patients not exposed to CDK4/6i before starting their first CT treatment had a longer median OS (IQR) of 23 (8–47) months in France and 31 (10–NA) months in Germany.

Time to next treatment or death
In France, the median TTNTD (IQR) from the start of first CT line was 5 (3–6) months, declining to 3 (2–5) months by the fourth CT line (Fig. 2A and Table S5). At 1-year, the probability of patients being free from receiving the next systemic treatment or death after the first CT line was 7%.
In Germany, median TTNTD (IQR) was 6 (3–9) months, declining to 3 (1–8) months by the fourth CT line (Fig. 2B and Table S5). At 1 year, the probability of patients being free from receiving the next systemic treatment or death after the first CT line was 17%.

Metastatic BC treatment patterns
Figure 3 shows the treatment patterns by LoT in France and Germany (see also Table S6). Figure 4 shows the distribution of CT agents (monotherapy or combination) by LoT.
In France, 91% (n = 307) of patients received a second LoT, 71% (n = 242) a third, and 53% (n = 178) a fourth. Forty eight percent (n = 162) of patients received their first CT treatment in the first LoT after mBC diagnosis, 29% (n = 98) in their second, 16% (n = 54) in their third, and 6% (n = 22) in their fourth LoT. In France, the most frequent first LoT was CT only (45%, n = 152), followed by ET + CDK4/6i (32%, n = 108), then ET only (16%, n = 54).
In Germany, 90% (n = 142) of patients received a second LoT, 71% (n = 112) a third, and 48% (n = 75) a fourth. Forty three percent (n = 68) of patients received their first CT treatment in the first LoT after mBC diagnosis, 36% (n = 56) in their second, 17% (n = 27) in their third, and 3% (n = 5) in their fourth LoT (Table S6). In Germany, the most common first LoT was ET + CDK4/6i (34%, n = 53), followed by other regimens (29%, n = 45) then CT only (24%, n = 37).

Distribution of treatment types in each LoT, from the first to fourth, by start year
Changes in the treatment landscape across the years by LoT start in France and Germany are shown in Fig. 5A and B, respectively, with the flow of treatments across treatment lines for the entire patient population shown in Figures S3A, S3B, S3C and S3D. In France, ET + CDK4/6i was received as a first LoT by 49% in 2016, 51% in 2018, and 43% in 2019, remaining stable in the following years. Use of CT only as a first LoT increased from 35% in 2018 to 53% in 2021, of which, single-agent taxanes were the most common in both countries.
The use of ET only, ET + other targeted therapy, and other regimens as first LoT decreased over the years. For second LoTs, ET + CDK4/6i use decreased from 40% in 2018 to 20% in 2022, while the use of CT only increased from 30% in 2018 to 45% in 2022. For third LoT, CT only use declined from 63% in 2021 to 48% in 2022. ET + CDK4/6i use decreased until 2021, used by only 6% of patients, but increased to 28% in 2022.
In Germany, ET + CDK4/6i use as a first LoT increased from 7% in 2016 to 45% in 2017, remaining stable in subsequent years. Other regimens used as a first LoT, predominantly CT + targeted therapy generally increased over time (30% in 2016 to 43% in 2020). The use of CT only as the first LoT decreased from 36% in 2016 to 13% in 2020. The use of ET only declined from 2016 to 2020, with no patients receiving ET + other targeted therapies as their first LoT. Second LoT patterns were more heterogeneous in Germany. CT only use increased from 20% in 2016 to 33% in 2022, peaking in 2018 at around 41%. The proportion of patients receiving ET + CDK4/6i increased from 10% in 2016 to 27% in 2018 but fell to 8% in 2019. From 2019, other regimens (mainly CT + targeted therapy) declined from 46% in 2019 to 17% in 2022, whereas the proportion of patients receiving ET only increased from 8% in 2019 to 25% in 2022. For third LoT, CT only was the most common treatment across all years, except for 2019, which was surpassed by other regimens (32% vs 44%, respectively). For fourth-line treatments, over half of patients received CT only treatments.

Treatment discontinuations
The assessment of treatment discontinuations was limited by the low absolute numbers and missing records. Based on available data, paclitaxel was the most common first-line CT monotherapy followed by capecitabine, and disease progression was the primary cause of discontinuation for both these treatments, with 38.7% of those who discontinued paclitaxel, and 71.7% of those discontinuing capecitabine doing so due to disease progression. Disease progression consistently remained the main reason for treatment discontinuation across all CT lines.

Discussion
This real-world study provides insights into the survival outcomes and treatment patterns of patients with HR +/HER2 − mBC initiating their first CT in France and Germany between 1 January 2016 and 28 February 2023. The overall sociodemographic and clinical characteristics of the two patient cohorts were broadly similar, with the exception of median age, and aligned with findings from previous retrospective studies from the same countries. The observed younger age of the German cohort may reflect the overall younger demographic of patients recruited in Germany compared to France, perhaps due to sampling variability, which is noteworthy given the reported older demographic of the overall population in Germany vs France [18].
Following diagnosis, French patients started CT within half the time of German patients (approximately 3 vs. 6 months, respectively) within the current analysis, perhaps reflecting the observed differences in treatment practices between the two countries. A high proportion of patients (81% in France and 75% in Germany) had evidence of bone metastasis and at least one visceral metastasis (76% in France and 89% in Germany). These findings contrast with those of the ESMÉ mBC database study conducted in France, which reported that approximately 65% of patients with HR +/HER2 − mBC had bone metastases at diagnosis [19]. Data from the German real-world PRAEGNANT registry reported bone metastases in 13.0–50.2% of patients, depending on first-line therapy [20].
In both France and Germany, the median OS from first CT initiation was < 2 years (20 months and 24 months, respectively), declining to approximately 8–10 months and 6–7 months for third and fourth CT lines, respectively. These findings align with existing literature showing a poor prognosis for patients with HR +/HER- mBC who initiate CT treatment [14, 19]. The median TTNTD from first CT initiation was approximately 5 months in France and 6 months in Germany, which reduced for subsequent CT lines. Interestingly, patients in this study not exposed to CDK4/6i prior to their first CT LoT had longer median OS. This finding aligns with the results of the IMMU-132–01 trial in adult patients with treatment-relapsed advanced epithelial cancers, which noted a median OS of 11.0 months in those with prior CDK4/6i exposure vs 21.7 months in those with no prior CDK 4/6i therapy [21]. This observation should be interpreted with caution, however, given the relatively small number of patients, and the potential for selection bias within the study population, Further evaluation in a larger patient population is needed to fully understand the implications of this finding.
Treatment patterns throughout the study showed a significant portion of patients in both countries started their first LoT with a CT-based therapy. Analysis of the different treatment types given in the first-line setting revealed vast heterogeneity in the number of regimens administered. Nevertheless, the distribution was generally similar across countries. In France, CT only was most frequently given as a first LoT, followed by ET + CDK4/6i. In Germany, CT + targeted therapy and CT only together accounted for approximately 41% of treatment in this setting, and ET + CDK4/6i was administered to around 34% of patients. Among patients who received CT only as a first-line treatment, single-agent taxanes were the most common in both countries. These results align with other real-world studies in France and Germany in which data from the ESMÉ database showed that, between 2008 and 2014, around 56–57% of patients received CT and 43–65% had ET as first-line treatments [22–24]. An analysis from 2014–2017 from the prospective PRAEGNANT registry found that approximately 43% and 46% of German women with mBC were treated with CT or ET monotherapy, respectively, as their first-line treatment [25].
Real-world treatment patterns observed in this study showed the significant use of CT (40%) over ET (33%) as first LoT. The high use of CT in clinical practice deviates from the 2023 update to the 2021 international ESMO guidelines, which recommend using ET, particularly in combination with CDK4/6i, as a first treatment for this patient population [2]. The guidelines recommend that all ET options should be attempted before switching to CT, unless a visceral crisis, imminent organ failure, or endocrine resistance exists [2]. Several factors may explain this difference. The low use of ET + CDK4/6i is partially explained by CDK4/6i’s lack of availability when clinicians made treatment choices due to the time lag between its European regulatory approval in November 2016 and reimbursement in each country. For example, in France, reimbursement for certain CDK4/6i treatments was granted in 2018, consequently use in clinical practice was mainly limited to early access programmes funded by the French government. Similarly, abemaciclib has been officially reimbursed in France since early 2019, palbociclib since March 2018, and ribociclib since May 2019. However, all these treatments were available before their official reimbursement via funding mechanisms in France. In Germany, new European Medicines Agency-approved drugs are automatically reimbursed at the company's price, providing instant access for patients [26]. Moreover, it is possible that patients who were treated with ET + CDK4/6i did not subsequently receive CT and so were not eligible for this analysis. Additionally, patients’ clinical condition may have influenced the initial treatment choice of CT over ET. Evidence indicates that patients undergoing CT as first LoT are more likely to have visceral metastasis or more than three metastatic sites compared to those treated with ET [22]. In the current study, over 75% of patients in both cohorts presented with visceral metastases, which may have resulted in a higher-than-expected administration of CT in first LoT. This is supported by the fifth European School of Oncology-ESMO guidelines published in 2020; the definition of visceral crisis was better clarified to avoid misunderstandings between the mere existence of visceral metastases and the presence of a visceral crisis, which requires the most rapidly efficacious therapy [27].
Together, the OS, TTNTD, and treatment patterns from this study suggest limited therapeutic benefits from CT alone, consistent with results showing disease progression as the primary cause of CT discontinuation in each line, highlighting the aggressive nature of this metastatic disease.
Recent phase 3 clinical trials in patients with HR +/HER2 − mBC treated with antibody–drug conjugates (ADC; a monoclonal antibody and CT combination) offer some promise to patients [28–30]. The DESTINY-Breast 04 trial showed significantly longer progression-free survival (PFS) and OS in patients treated with trastuzumab deruxtecan, with the risk of progression or death approximately 50% lower and the risk of death 36% lower compared with patients treated with single-agent CT [31]. Similar improvements in PFS with trastuzumab deruxtecan vs CT were noted in the DESTINY-Breast 06 trial in patients with mBC who had received prior ET, with significantly longer PFS seen in patients with HER2-low disease in the trastuzumab deruxtecan group compared with the CT group (13.2 versus 8.1 months, respectively [HR 0.62; 95% CI 0.52–0.75; p < 0.001]), however, it should be noted that this study did not assess OS [32]. Patients who were treated with another ADC, sacituzumab govitecan, demonstrated statistically significant and clinically meaningful benefit vs CT, with a 3.2-month median improvement in OS in the TROPiCS 02 trial [33]. However, final results from the TROPION-Breast01 trial of datopotamab deruxtecan vs investigator’s choice of CT, did not achieve statistical significance in the final OS analysis, despite having previously met the dual primary endpoint of PFS [34]. This study provided valuable insights into treatment utilisation and outcomes in a large cohort of patients with HR +/HER2 − mBC, utilising high-quality patient data. Nevertheless, limitations inherent to retrospective analyses must be acknowledged. Patients were recruited from three study sites, chosen as a convenience sample and not guaranteed to be representative of all centers that treat patients with mBC within the country. Patients may have received their first, second, third, and fourth CT lines at varying time points in their treatment trajectory and may have undergone multiple other treatments between CT lines that could not be accounted for in this study. The study population might be quite heterogeneous in terms of demographics, clinical characteristics and also clinical practice. Such heterogeneity might introduce bias in the estimation of treatment patterns and clinical outcomes. Moreover, ADCs such as trastuzumab deruxtecan only received approval in 2019, resulting in limited observations of their use within this dataset and limiting their relevance to the observed treatment patterns. Additionally, potential inaccuracies due to missing data in EMRs and the unknown impact of COVID-19 on routine clinical care must be considered. The algorithmic stratification of patients by CT line could have introduced bias, as only patients who did not die before initiating the next line of therapy were included in subsequent CT line subgroups. Furthermore, patients who died or were lost to follow-up before initiating their second CT line likely exhibited different clinical characteristics and outcomes compared to those included in this analysis. Finally, the identification of CT lines excluded non-CT treatments, which may have influenced patient outcomes. It should also be noted that given the relatively small sample sizes included annually across all LoTs, caution is warranted when interpreting these findings within the context of the mBC treatment landscape.

Conclusion
The results of the current study indicate that patients with HR +/HER2 − mBC have a median survival of less than 2 years, which decreases with each subsequent CT line; this indicates a poor prognosis for patients who receive CT alone in the real world. The consistent findings in France and Germany highlight the aggressive nature of the disease and underscore the high unmet need for improved treatment options in European countries.

Supplementary Information