Real-world treatment patterns and associated outcomes in patients with HER2-positive unresectable or metastatic breast cancer: the HER2 REAL Study

Abstract

Background

The HER2 REAL (NCT04857619) retrospective study explored treatment practices and survival outcomes in patients with human epidermal growth factor receptor 2 (HER2)–positive unresectable breast cancer/metastatic breast cancer (mBC) in routine clinical care across six countries.

Methods

Adult patients (≥18 years) diagnosed with HER2-positive mBC, having ≥12 months of follow-up data from diagnosis and completing one or more line of therapy (LOT) were enrolled. The primary endpoint was treatment patterns. Secondary endpoints included clinicodemographic characteristics and survival outcomes using the Kaplan–Meier method and presented as median [95% confidence interval (CI)]. Data for the first two LOTs in the metastatic setting are presented.

Results

Of the 763 patients enrolled, 684 patients were analyzed for baseline characteristics. Majority [99.6% (681/684)] were women, 67.3% (379/563) were postmenopausal with median (range) age of 55 (20-91) years. At index date, 99.3% (597/601) of patients had stage IV disease, and 72.1% (465/645) reported visceral metastases. Immunohistochemistry score was available for 64.6% of patients. Median (range) study follow-up was 30.1 (1-167) months. A total of 96.1% (n = 657) and 84.5% (n = 578) of the patients had data on LOT1 and LOT2, available in the medical records. In LOT1, trastuzumab (TRA)+pertuzumab (PTZ)+chemotherapy (CT)-based regimen were prescribed to 58.1% and non–TRA+PTZ+CT-based to 41.6%; TRA+CT-based regimen to 20.2%, which is a subset of the TRA+PTZ+CT group. In LOT2, trastuzumab emtansine (T-DM1)-based regimen was prescribed to 45.5% and non–T-DM1–based, and TRA+CT-based to 53.1% and 11.8% patients, respectively. The median (95% CI) real-world overall survival (rwOS) was 41.4 (38.1-45.8) months for overall cohort. In LOT1, median rwOS was 46.8 (38.7-50.7) months for TRA+PTZ+CT-based regimens. In LOT2, median rwOS was 23.9 (20.5-29.5) months in the T-DM1 cohort. Disease progression or death and toxicity were major reasons for treatment discontinuation.

Conclusions

This retrospective study shows heterogeneity in treatment patterns and survival outcomes in real-world patients with HER2-positive mBC, indicating the need for optimization of treatment in patients with HER2-positive mBC in real-world.

Highlights

• •This study reports real-world outcomes in 684 previously treated patients (one or more LOT) with HER2-positive mBC.
• •In LOT1, 58.1% of patients were prescribed with trastuzumab (TRA)+pertuzumab (PTZ)+chemotherapy (CT)–based regimen.
• •In LOT2, 45.5% of patients were prescribed with trastuzumab emtansine (T-DM1)–based regimen.
• •Median OS was 41.4 months for overall cohort, 46.8 months for TRA+PTZ+CT (LOT1), and 23.9 months for T-DM1 (LOT2) regimens.
• •Disease progression and toxicity were major reasons for treatment discontinuation in real-world practice.

Introduction

Human epidermal growth factor receptor 2 (HER2)–positive breast cancer represents 15% to 20% of all breast cancer cases and typically arises from ductal tissue.¹ This subtype is associated with a more aggressive phenotype, high recurrence rates, and poor survival outcomes.^1,2 Despite advances, 15% to 24% of patients with HER2-positive breast cancer develop metastases after curative treatment, and 3% to 10% of patients present with de novo metastases.^3,4 The brain is a frequent site of metastasis, occurring in 9% to 50% of cases and is associated with poorer overall survival.5, 6, 7 Anti-HER2 therapies such as trastuzumab (TRA)-based regimens have improved the survival outcomes in patients with HER2-positive metastatic breast cancer (mBC).⁸ However, ∼35% of patients display primary resistance, suggesting the need for additional therapeutic options.² The CLEOPATRA trial showed that the addition of pertuzumab (PTZ) to TRA and docetaxel resulted in improved progression-free survival (PFS) and overall survival (OS) in patients who had not previously received anti-HER2 therapy or chemotherapy (CT) for metastatic disease.⁹ Currently, the combination of CT, TRA, and PTZ is the first-line treatment standard of care (SoC) in patients with HER2-positive mBC10, 11, 12 and is approved in several countries.13, 14, 15 Before 2022, trastuzumab emtansine (T-DM1) was the standard therapeutic approach in the second-line setting for HER2-positive mBC in patients who previously received trastuzumab and a taxane, based on the robust PFS and OS data from the EMILIA and TH3RESA trials, which compared T-DM1 with either lapatinib plus capecitabine or treatment of physician’s choice, respectively.16, 17, 18 However, in 2022, trastuzumab deruxtecan (T-DXd), received Food and Drug Administration approval as a second-line treatment based on the results of the DESTINY-Breast03 trial, which showed a significant reduction in the risk of disease progression compared with T-DM1 in the same setting.^19,20 International guidelines recommend T-DXd as a preferred second-line regimen option for HER2-positive recurrent unresectable mBC with T-DM1 listed as an alternative option.^12,21 The recommended treatment options in third-line and beyond include T-DM1, tucatinib with TRA and capecitabine, as well as diverse combinations of CT and HER2-targeted therapies.²¹

Despite these advances, 5-year survival for HER2-positive mBC remains low at 30% to 46%, highlighting the need for accessible treatments with durable responses.^22,23 Although the treatment landscape is evolving with novel therapies, the real-world evidence beyond first-line therapy, especially on long-term outcomes, remains limited and mostly from the United States and Europe.24, 25, 26^,27, 28, 29 Thus, real-world studies can offer insights into the evolving treatment patterns, comparative effectiveness of treatment regimens, associated toxicities, and long-term survival with a broader set of patients, with diverse characteristics representative of the general population. This study evaluated treatment patterns and clinical outcomes in patients with HER2-positive unresectable breast cancer or mBC across Asia Pacific and Brazil.

Patients and methods

Study design

HER2 REAL (NCT04857619) was a retrospective, multicenter, noninterventional study carried out at 43 oncology centers across six countries (Australia, Brazil, Hong Kong, Republic of Korea, Singapore, and Taiwan). The study protocol was approved by the respective independent ethics committee/institutional review boards. A signed written informed consent was obtained from patients during their routine follow-up or next of kin/legal representative (in case of deceased patients), unless a waiver was granted. The study was carried out as per the ethical principles consistent with the Declaration of Helsinki, International Council for Harmonization, Good Clinical Practices Guidelines, Good Pharmacovigilance Practices, and the applicable legislation of the National Institutes of Health Studies. The data presentation in this article adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist.³⁰

Study population and data collection

Adult patients (≥18 years old) diagnosed with HER2-positive unresectable breast cancer or mBC (de novo advanced disease, disease progression while on ongoing treatment, or recurrent disease) since 2017 or the available date for T-DM1 through reimbursement or patient access programs that made it a valid local treatment option (patient enrollment dates—Brazil, Hong Kong, Republic of Korea, and Taiwan: January 2017; Australia: July 2015; and Singapore: January 2014), whichever was earlier, were included per medical chart review. The data extraction date for the study was July 2023. Eligible patients were also required to complete at least one line of therapy (LOT) in the advanced setting and have ≥12 months of follow-up data from the index date (defined as earliest date of unresectable breast cancer/mBC diagnosis). Patients who died within 12 months of diagnosis from the index date were included, irrespective of the follow-up duration. Definitions are provided in Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2026.108326. Patients with HER2-negative unresectable breast cancer or mBC, those with a change in HER2 status from positive to negative at disease progression to metastatic stage, and those who had a synchronous or multiple primary tumors at the time of HER2-positive unresectable breast cancer or mBC diagnosis (except for nonmetastatic nonmelanoma skin cancers or in situ neoplasms), and those participating in a blinded interventional trial at the time of data collection were excluded from the study.

The medical records of eligible and consenting patients were reviewed, and the relevant data on treatment patterns, demographics, clinicopathological characteristics, clinical outcomes, and health care resource utilization were extracted and entered in a centrally designed electronic data capture system. LOT was defined as a treatment regimen, possibly a combination of several drugs, given from either the index diagnosis or disease progression until the treatment fails to control the disease, is not tolerated by the patient, the disease relapses/progresses, or death occurs. Therapies that are paused for a period and then restarted—including at reduced dose—without disease progression, and had occurred were considered as part of the same line rather than a new line. The number of patients receiving LOT was calculated by identifying patients who initiated their earliest available LOT within 30 days before the index date. The number of patients receiving the subsequent LOT was calculated from the stop date of patients receiving the previous LOT (when the treatment was discontinued due to disease relapses/progression or, intolerance to treatment) up to the start of the next LOT. An optional Health Economics Data Questionnaire was used for collecting insurance types and coverage if the data were not available in the medical records. The study did not include any study-specific patient visits or longitudinal follow-up.

Study outcomes

The primary outcomes of the study were the proportion of patients receiving each regimen in LOT1 and LOT2, duration of treatment, and reason(s) for treatment discontinuation (if available) (Supplementary Table S1, available at https://doi.org/10.1016/j.esmoop.2026.108326). Secondary outcomes were demographic characteristics including age, sex, menopausal status, ethnicity, and a family history of breast cancer; and clinicopathological characteristics including Eastern Cooperative Oncology Group Performance Status (ECOG-PS), tumor-node-metastasis staging at index diagnosis, hormone receptor status for estrogen or progesterone, histological subtype, site(s) of metastases, type of HER2 testing [immunohistochemistry (IHC) and/or in situ hybridization (ISH)], and comorbidities. The effectiveness of various regimens was measured by real-world PFS (rwPFS; defined as the time from the index date or date of initiation of LOT to the first documented real-world disease progression or death, whichever occurs first) and real-world OS (rwOS; defined as the length of time from the index date for first LOT or date of initiation of subsequent LOT to death due to any cause) for LOT1 and LOT2. Additionally, adverse events of special interests (AESIs) and the proportion of patients with AESIs that led to treatment discontinuation, hospitalization, and deaths were assessed. The AESIs included cardiac dysfunction [asymptomatic reduced left ventricular ejection fraction (LVEF) of <50%, symptomatic fall in ejection fraction, myocardial infarction, atrial fibrillation, sudden cardiac death, symptomatic heart failure, ventricular arrhythmia, and cardiac dysfunction], interstitial lung disease, and hepatotoxicity. The type of insurance and coverage were also ascertained as a part of the exploratory analyses.

Sample size estimation and statistical methods

Based on the results of sample size estimation conducted to approximate the sample size needed for the primary analysis, a target enrollment of 763 patients was planned for the HER2 REAL study. Analyses were conducted on both the overall population and country levels, using the full analysis set (FAS) of eligible and enrolled patients. Descriptive statistics for clinicodemographic data and treatment patterns were generated with Statistical Analysis Software (SAS version 9.4 or higher). Categorical variables were presented using percentages and corresponding 95% confidence interval (CI); continuous variables as mean standard deviation or median (range). Missing variables were not imputed and were reported. Time-to-event data were evaluated descriptively using Kaplan–Meier survival curves when groups had >20 events. Safety metrics—including AESIs, discontinuations, and deaths—were presented according to available information.

Treatment regimens were classified as SoC or non-SoC according to clinical guidelines. The first- and second-line SoC regimens were TRA+PTZ+CT and T-DM1, respectively. Non-SoC groups included any regimen excluding these standards in their respective lines. A TRA+CT-based subgroup was also analyzed but overlapped with other categories and did not sum to the total patient count.

Results

Patient disposition and baseline characteristics

Overall, 763 patients from six countries were enrolled, of whom 684 (89.6%) were analyzed for baseline characteristic and included in the FAS (Australia 62, Brazil 120, Hong Kong 92, Republic of Korea 157, Singapore 105, and Taiwan 148); 79 patients were excluded [unavailability of ≥12 months of follow-up data (n = 21); did not complete one or more LOT (n = 20); screening failure (n = 20); and other (n = 18)] (Figure 1). The reasons for exclusion are detailed in Figure 1 and Supplementary Table S2, (available at https://doi.org/10.1016/j.esmoop.2026.108326). The median age (range) was 55 (20-91) years; most patients were women [99.6% (681/684)]. Among female patients with available data, 67.3% (379/563) were postmenopausal at the index date. Data on performance scores were available for 28.7% of patients (196/684), of whom 89.8% (176/196) had an ECOG-PS of 0/1. Baseline comorbidities were reported in 44.0% (301/684) of the patients. Data on the family history of breast cancer were available for 84.9% (581/684) of patients; 20.1% (117/581) had a positive family history. Information regarding cost coverage was available for 60.2% (412/684) of patients, of whom 90.8% (374/412) had health care funding (Table 1). Data on insurance coverage is elaborated in Supplementary Table S3, (available at https://doi.org/10.1016/j.esmoop.2026.108326).

Figure 1.

Patient disposition. EDC, electronic data capture; HER, human epidermal growth factor receptor; LOT, line of therapy; SF, screen failure.

Table 1.

Baseline characteristics at the index date

Patient characteristics	Overall (N = 684)	Clinical characteristics	Overall (N = 684)
Countries		Stage at index date (seventh AJCC staging)	601/684 (87.9)
Australia	62 (9.1)	Stage III	4/601 (0.6)
Brazil	120 (17.5)	Stage IV	597/601 (99.3)
Hong Kong	92 (13.5)	Not availablea	83/684 (12.1)
Korea	157 (23)	Tumor grade at index date	293 (42.8)
Singapore	105 (15.4)	Well-differentiated, low-grade (G1)	7/293 (2.4)
Taiwan	148 (21.6)	Moderately differentiated (G2)	131/293 (44.7)
Age (years), median (range)	55 (20–91)	Poorly differentiated, high grade (G3)	155/293 (52.9)
Female	681/684 (99.6)	Not availablea	391/684 (57.2)
Menopausal status	563/684 (82.3)	Histological type at index date	464/684 (67.8)
Premenopausal	147/563 (26.1)	Ductal	407/464 (87.7)
Perimenopausal	37/563 (6.6)	Lobular	15/464 (3.2)
Postmenopausal	379/563 (67.3)	Mixed	7/464 (1.5)
Not availablea	121/684 (17.7)	Other	35/464 (7.5)
Comorbidities	301/684 (44.0)	Not availablea	220/684 (32.2)
Hypertension	146/301 (48.5)	Site of metastasisb on index date	645 (94.3)
Diabetes	83/301 (27.6)	Visceral	465/645 (72.1)
Coronary heart disease	12/301 (4.0)	Visceral only	156/465 (33.5)
Atrial fibrillation	8/301 (2.7)	Nonvisceral	427/645 (66.2)
Other	205/301 (68.1)	Nonvisceral only	126/427 (29.5)
Not availablea	383/684 (56.0)	CNS	149/645 (23.1)
Family history of BC	581/684 (84.9)	CNS only	31/149 (20.8)
Yes	117/581 (20.1)	Not availablea	39/684 (5.7)
No	464/581 (79.9)	Biomarkers at index datec
Not availablea	103/684 (15.1)	IHC for HER2 positivity	442/684 (64.6)
Ethnicity	679/684 (99.3)	0	1/442 (0.2)
Asian	510/679 (75.1)	1+	3/442 (0.7)
Caucasian	103/679 (15.2)	2+	63/442 (14.3)
Other	66/679 (9.7)	3+	375/442 (84.8)
Not availablea	5/684 (0.7)	Not availablea	242/684 (35.4)
Insurance	412/684 (60.2)	ISH for HER2 positivity	113/684 (16.5)
Yesd	374/412 (90.8)	Positive	112/113 (99.1)
Government/national	273/374 (73.0)	Negative	1/113 (0.9)
Employer provided	13/374 (3.5)	Not availablea	571/684 (83.5)
Private	85/374 (22.7)	ISH+ and IHC 2+	54/684 (7.9)
Mixed coverage	3/374 (0.8)	Estrogen receptor	423/684 (61.8)
No (use own funds)	38/412 (9.2)	Positive	218/423 (51.5)
Not availablea	272/684 (39.8)	Negative	205/423 (48.5)
ECOG-PS	196/684 (28.7)	Not availablea	261/684 (38.2)
0/1	176/196 (89.8)	Progesterone receptor	417/684 (61.0)
2	16/196 (8.2)	Positive	145/417 (34.8)
3/4	4/196 (2.0)	Negative	272/417 (65.2)
Not availablea	488/684 (71.3)	Not availablea	267/684 (39.0)
Follow-up duration (months), median (range)	30.1 (1-167)	Time from the initial BC diagnosis	388/684 (56.7)
		<2 years	171/388 (44.1)
		2-5 years	142/388 (36.6)
		>5 years	75/388 (19.3)
		Not availablea	296/684 (43.3)
		Median (range) (months)	26 (0-424)

Data are presented as n (%) unless otherwise specified. Percentages have been calculated for available data.

AJCC, American Joint Committee on Cancer; BC, breast cancer; CNS, central nervous system; ECOG-PS, Eastern Cooperative Oncology Group Performance Status; HER2, human epidermal growth factor receptor 2; IHC, immunohistochemistry; ISH, in situ hybridization.

^aNot available includes not reported, not available, unknown, and missing categories.

^bSite of metastasis for visceral only category contains only patients with metastasis in this location, while visceral includes both these patients, as well as those with visceral + any other location. Corresponding logic applies to the nonvisceral and CNS categories. Nonvisceral metastases include tumors located in the bone, bone marrow, lymph nodes, skin, and soft tissues. CNS metastases include tumors located in the brain, spinal cord, or meninges. All other locations are classified as visceral disease.

^cFor the purpose of matching the biopsy with the index date, a ±30-day time window around the index date was created. If only one biopsy date fell into this window, it was considered a positive match.

^dIncludes private, government-provided/national insurance, employer-provided, and mixed coverage.

Of the patients with available data on metastases (558/684), ∼30.5% (170/558) presented with de novo mBC at the index date. The median [interquartile range (IQR)] duration from the initial BC diagnosis to relapsed mBC (includes patients with de novo mBC as well) was 26.0 (13-48) months. The median (range) duration of study follow-up was 30.1 (1-167) months. At the index date, 99.3% (597/601) of the patients had stage IV disease, and 0.3% (2/601) had stage IIIB and IIIC each. Of the 42.8% (293/684) patients with available tumor grade information at the index date, 52.9% (155/293) had poorly differentiated (grade 3) tumors. Histopathological type at the index date was available for 67.8% (464/684) of the patients, of whom 87.7% (407/464) had invasive ductal carcinoma. Overall, 72.1% (465/645) of patients reported visceral metastases, 66.2% (427/645) reported nonvisceral metastases, and 23.1% (149/645) reported central nervous system (CNS) metastases (Table 1).

Biomarker testing

At the index date, IHC score was available for 64.6% (442/684) of patients. Of these, 84.8% (375/442) had IHC3+, and 14.3% (63/442) were IHC2+. ISH data were available for 16.5% (113/684) of the patients at the index date, of whom 99.1% (112/113) were ISH-positive. In the remaining 35.4% (242/684) patients, IHC or ISH data were not available; HER2 positivity was reported from data available in patient’s medical records. In total, 7.9% (54/684) of patients were IHC2+ and ISH-positive. Of the patients with available data on estrogen receptor (ER) status [61.8% (423/684)], 51.5% (218/423) were ER positive. Progesterone receptor (PR) positivity was reported in 34.8% (145/417) of patients. Data for ER and PR were not reported/missing for ∼38.2% and 39.0% of patients, respectively (Table 1).

Treatment patterns

A median of seven LOTs were analyzed in the overall cohort, of which data from LOT1 to LOT2 are presented in this article. A total of 657 (96.1%) and 578 (84.5%) of patients had data on LOT1 (median duration 8.3 months, IQR 4.4-15.5 months) and LOT2 (median duration 5.1 months, IQR 2.6-10.5 months), available in the medical records (Table 2). The treatment regimens were categorized as TRA+PTZ+CT-based regimen, non–TRA+PTZ+CT-based regimen, and TRA+CT-based regimen. For LOT1, 58.1% (382/657) were prescribed with a TRA+PTZ+CT-based regimen and 41.6% (273/657) with a non–TRA+PTZ+CT-based regimen; 20.2% (133/657) were prescribed with TRA+CT-based regimen, which is a subset of the TRA+PTZ+CT group. Countrywise prescription pattern are summarized in Table 2. TRA+PTZ+CT-based regimens was prescribed to >65% of patients in Australia, Hong Kong, and Republic of Korea in LOT1. For patients receiving LOT2, T-DM1 was prescribed to 45.5% (263/578), whereas non–T-DM1-based and TRA+CT-based regimens were prescribed to 53.1% (307/578) and 11.8% (68/578), respectively. T-DM1-based regimens were prescribed in >50% of patients in Australia, Hong Kong, and Republic of Korea (Table 2). The overall treatment attrition rate was 10.7% (70/657) from LOT1 to LOT2 and 17.8% (103/578) from LOT2 to LOT3. Numerically, highest attrition rates were seen in Australia [LOT1 to LOT2: 26.2% (16/61); LOT2 to LOT3: 29.5% (11/44)] (Supplementary Figure S1, available at https://doi.org/10.1016/j.esmoop.2026.108326). Major reasons for treatment discontinuation were disease progression or death and toxicity.

Table 2.

Treatment pattern by LOT and country

Regimen	Overall (N = 684)	Australia (n = 62)	Brazil (n = 120)	Hong Kong (n = 92)	Republic of Korea (n = 157)	Singapore (n = 105)	Taiwan (n = 148)
LOT1
N1, n/N (%)	657/684 (96.1)	61/62 (98.4)a	102/120 (85.02)b	91/92 (98.9)c	155/157 (98.7)d	104/105 (99.0)e	144/148 (97.3)f
Duration (months), median (IQR)	8.3 (4.4-15.5)	11.0 (6.2-20.0)	7.9 (3.7-18.9)	9.3 (5.8-15.6)	8.3 (4.5-15.3)	7.7 (3.0-13.0)	7.3 (5.0-13.1)
Regimen 1	TRA+PTZ+CT-basedg
n/N1 (%)	382/657 (58.1)	46/61 (75.4)	40/102 (39.2)	62/91 (68.1)	121/155 (78.1)	45/104 (43.3)	68/144 (47.2)
Regimen 2	Non–TRA+PTZ+CT-basedh,i
n/N1 (%)	273/657 (41.6)	14/61 (23.0)	61/102 (59.8)	29/91 (31.9)	34/155 (21.9)	59/104 (56.7)	76/144 (52.8)
Regimen 3	TRA+CT-basedj
n/N1 (%)	133/657 (20.2)	2/61 (8.2)	37/102 (36.3)	10/91 (11.0)	9/155 (5.8)	36/104 (34.6)	39/144 (27.1)
LOT2
N1, n/N (%)	578/684 (84.5)	44/62 (71.0)	79/120 (65.8)	86/92 (93.5)	140/157 (89.2)	98/105 (93.3)	131/148 (91.0)
Duration (months), median (IQR)	5.1 (2.6-10.5)	8.3 (2.3-12.0)	10.3 (4.9-16.3)	5.1 (3.5-8.7)	4.5 (2.1-7.6)	4.3 (2.2-9.5)	4.4 (2.5-9.9)
Regimen 1	T-DM1 basedk
n/N1 (%)	263/578 (45.5)	28/44 (63.6)	15/79 (19.0)	57/86 (66.3)	97/140 (69.3)	29/98 (29.6)	37/131 (28.2)
Regimen 2	Non–T-DM1 basedl,m
n/N1 (%)	307/578 (53.1)	14/44 (31.8)	62/79 (78.5)	29/86 (33.7)	42/140 (30.0)	68/98 (69.4)	92/131 (70.2)
Regimen 3	TRA+CT-basedj
n/N1 (%)	68/578 (11.8)	1/44 (2.3)	14/79 (17.7)	2/86 (2.3)	1/140 (0.7)	20/98 (20.4)	30/131 (22.9)
Regimens received in LOT1 followed by LOT2
Regimen 1	TRA+PTZ+CT (LOT1) followed by T-DM1 (LOT2)
n/N1 (%)	218/684 (31.9)	27/62 (43.5)	12/120 (10.0)	45/92 (48.9)	83/157 (22.1)	24/105 (52.9)	27/148 (138.2)
Regimen 2	TRA+PTZ+CT (LOT1) followed by non–T-DM1 (LOT2)
n/N1 (%)	121/684 (17.7)	7/62 (11.3)	21/120 (17.5)	14/92 (15.2)	25/157 (15.9)	17/105 (16.2)	37/148 (25.0)
Regimen 3	Non–TRA+PTZ+CT (LOT1) followed by T-DM1 (LOT2)
n/N1 (%)	45/684 (6.6)	1/62 (7.7)	3/120 (2.5)	12/92 (13.0)	14/157 (8.9)	5/105 (4.8)	10/148 (6.8)
Regimen 4	Non–TRA+PTZ+CT (LOT1) to non–T-DM1 (LOT2)
n/N1 (%)	183/684 (26.8)	7/62 (11.3)	40/120 (33.3)	15/92 (16.3)	16/157 (10.2)	51/105 (48.6)	54/148 (36.5)

CT, chemotherapy; IQR, interquartile range; LOT, line of therapy; N, total number of patients per country; n, number of patients in the treatment; PTZ, pertuzumab; T-DM1, trastuzumab emtansine; T-DXd, trastuzumab deruxtecan; TRA, trastuzumab.

^aOne patient is excluded due to the gap between imputed LOT end date (year 2020) and the collected LOT end date (year 2016).

^bFourteen eligible patients were excluded because their LOT1 start date was >30 days before the index date; two patients are excluded from the analysis since index date was before 1 January 2017; two patients were excluded from the analysis since the LOT start and end dates were same in the metastatic setting.

^cOne patient was excluded since LOT2 start date was before LOT1 start date.

^dOne patient was excluded since the death date was before LOT start date; one patient was excluded since the LOT start date and end date were same in the metastatic setting.

^eOne patient is excluded from the analysis since death date was before LOT1 start date.

^fThree patients were excluded since the LOT start and end dates were same in the metastatic setting; one patient was excluded due to metastatic setting condition.

^gTRA+PTZ+CT-based regimen included PTZ+TRA+CT, PTZ+TRA+CT+hormone therapy, lapatinib+PTZ+TRA+CT, and lapatinib+PTZ+TRA+CT+hormone therapy.

^hNon–TRA+PTZ+CT-based regimen included CT alone, hormone therapy alone, and T-DM1.

ⁱTwo patients receiving T-DXd-based regimens were excluded from the regimen count.

^jTRA+CT-based regimen included TRA+CT, lapatinib+TRA+CT, TRA+other+CT, and TRA+CT+hormone therapy.

^kT-DM1 based regimens included: T-DM1+hormone therapy+aromatase inhibitor, T-DM1+CT+cyclophosphamide+doxorubicin/liposomal doxorubicin, T-DM1+TRA+tucatinib, T-DM1+hormone therapy+tamoxifen, T-DM1+hormone therapy+aromatase inhibitor+CDK4/6 inhibitor, T-DM1+hormone therapy+aromatase inhibitor+CDK4/6 inhibitor+megestrol acetate, T-DM1+CT+capecitabine, T-DM1+CT+lapatinib+capecitabine, and T-DM1+CT+cisplatin/carboplatin+gemcitabine+vinorelbine.

^lNon–T-DM1 regimen included CT alone, hormone therapy alone, and PTZ+TRA+CT.

^mEight patients receiving T-DXd-based regimens were excluded from the regimen count.

Survival outcomes

With a median (range) follow-up of 30.5 (0.6-98.7) months, the median (95% CI) rwOS was 41.4 (38.1-45.8) months (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmoop.2026.108326). The 2-year OS rate was 74.2%. Kaplan–Meier analysis was carried out at the country level only if the number of events in the LOT was >20. The median OS per country is provided in Supplementary Table S5, (available at https://doi.org/10.1016/j.esmoop.2026.108326). The Kaplan–Meier curves for OS for the overall cohort and by individual country are provided in Supplementary Figure S2, (available at https://doi.org/10.1016/j.esmoop.2026.108326). The 2-year OS rates for patients in LOT1 and LOT2 per country are presented in Supplementary Table S4, (available at https://doi.org/10.1016/j.esmoop.2026.108326). Overall, patients receiving LOT1 and LOT2 had 1-year rwPFS rate of 46.7% and 18.9%, respectively, and 2-year rwPFS rate of 36.5% and 19.6%, respectively. The overall median (95% CI) rwPFS in patients who received LOT1 was 11.1 (9.5-12.4) months and LOT2 was 8.1 (7.1-9.9) months (Supplementary Table S5, available at https://doi.org/10.1016/j.esmoop.2026.108326).

The median (95% CI) rwOS for TRA+PTZ+CT-based regimen in LOT1 was 46.8 (38.7-50.7) months, with a 2-year OS rate of 75.9% (Figure 2; Supplementary Tables S4 and S5, available at https://doi.org/10.1016/j.esmoop.2026.108326). Among 58.1% of the patients in the TRA+PTZ+CT cohort in LOT1, the median (95% CI) rwPFS was 12.3 (10.8-13.1) months, with a 2-year rwPFS rate of 18.9% (Figure 2). The median (95% CI) rwOS in these patients was 23.9 (20.5-29.5) months, with a 2-year OS rate of 48.6% (Figure 2; Supplementary Tables S4 and S5, available at https://doi.org/10.1016/j.esmoop.2026.108326). In LOT2, 45.5% of patients in the T-DM1 cohort had the median (95% CI) rwPFS was 7.2 (6.3-9.6) months, with a 2-year rwPFS rate of 17.0% (Figure 2). Survival outcomes as per treatment patterns are presented Supplementary Table S5, (available at https://doi.org/10.1016/j.esmoop.2026.108326) and Figures 2 and 3.

Figure 2.

Real-world PFS for the top regimen in LOT1 and LOT2. (A) KM curve for median rwPFS: TRA+PTZ+CT-based regimen compared with non–TRA+PTZ+CT-based regimen compared with TRA+CT-based regimen in LOT1. (B) KM curve for median rwPFS: T-DM1–based regimen compared with non–T-DM1-based regimen compared with TRA+CT-based regimen in LOT2. TRA+PTZ+CT-based regimen included PTZ+TRA+CT, PTZ+TRA+CT+hormone therapy, lapatinib+PTZ+TRA+CT, and lapatinib+PTZ+TRA+CT+hormone therapy. TRA+CT-based regimen included TRA+CT, lapatinib+TRA+CT, TRA+other+CT, and TRA+CT+hormone therapy. T-DM1–based regimens included T-DM1+hormone therapy+aromatase inhibitor, T-DM1+CT+cyclophosphamide+doxorubicin/liposomal doxorubicin, t-DM1+trastuzumab+tucatinib, T-DM1+hormone therapy+tamoxifen, T-DM1+hormone therapy+aromatase inhibitor+CDK4/6 inhibitor, T-DM1+hormone therapy+aromatase inhibitor+CDK4/6 inhibitor+megestrol acetate, T-DM1+CT+capecitabine, T-DM1+CT+lapatinib+capecitabine, and T-DM1+CT+cisplatin/carboplatin+gemcitabine+vinorelbine. Non–T-DM1-based regimen included regimens excluding the combination of T-DM1- and T-Dxd–based regimens. For each LOT, an event is said to have occurred when a patient reports disease progression or death within the corresponding LOT. If a patient did not report disease progression or death until the end of the respective LOT, then the patient is censored at therapy end date. If the therapy end date is missing and the LOT under consideration is the last line of therapy for the patient, then the patient is censored at the last contact date. If the therapy end date is missing and the LOT under consideration is not the last line of therapy for the patient, then the patient is censored at day before start date of next LOT. CI, confidence interval; CT, chemotherapy; KM, Kaplan–Meier; LOT, line of therapy; PFS, progression-free survival; PTZ, pertuzumab; T-DM1, trastuzumab emtansine; rw, real-world; TRA, trastuzumab.

Figure 3.

Real-world OS for the top regimen in LOT1 and LOT2. (A) KM curve for median rwOS: TRA+PTZ+CT-based regimen compared with non–TRA+PTZ+CT-based regimen compared with TRA+CT-based regimen in LOT1. (B) KM curve for median rwOS: T-DM1-based regimen compared with non–T-DM1-based regimen compared with TRA+CT-based regimen in LOT2. rwOS is the length of time from the index date or date of initiation of LOT to date of death due to any cause. OS is calculated using the KM technique. The full analysis set included all patients who fulfilled the eligibility criteria. TRA+PTZ+CT-based regimen included PTZ+TRA+CT, PTZ+TRA+CT+hormone therapy, lapatinib+PTZ+TRA+CT, lapatinib+PTZ+TRA+CT+hormone therapy. TRA+CT-based regimen included TRA+CT, lapatinib+TRA+CT, TRA+other+CT, TRA+CT+hormone therapy. T-DM1-based regimens included T-DM1+hormone therapy+aromatase inhibitor, T-DM1+CT+cyclophosphamide+doxorubicin/liposomal doxorubicin, T-DM1+trastuzumab+tucatinib, T-DM1+hormone therapy+tamoxifen, T-DM1+hormone therapy+aromatase inhibitor+CDK4/6 inhibitor, T-DM1+hormone therapy+aromatase inhibitor+CDK4/6 inhibitor+megestrol acetate, T-DM1+CT+capecitabine, T-DM1+CT+lapatinib+capecitabine, and T-DM1+CT+cisplatin/carboplatin+gemcitabine+vinorelbine. Non–T-DM1-based regimen included regimens excluding the combination of T-DM1- and T-Dxd–based regimens. An event was said to have occurred when a patient reported death after diagnosis. If a patient was not reported dead, the patient was censored at the last contact date. Reasons for censoring: For each LOT, an event is said to have occurred when a patient reported disease progression or death within the corresponding LOT. If a patient did not report disease progression or death until the end of the respective LOT, then the patient was censored at therapy end date. If the therapy end date was missing and the LOT under consideration was the last LOT for the patient, then the patient was censored at last contact date. If the end therapy date was missing and the LOT under consideration was not the last LOT for the patient, then the patient was censored at day before start date of next LOT. The KM analysis is carried out only if >20 events are present in the data. CI, confidence interval; CT, chemotherapy; KM, Kaplan–Meier; LOT, line of therapy; OS, overall survival; PTZ, pertuzumab; T-DM1, trastuzumab emtansine; rw, real-world; TRA, trastuzumab.

Safety outcomes

We found AESIs for the overall cohort and for the top regimens of LOT1 and LOT2. Of 684 patients, 34 (5.0%) patients were reported to be hospitalized due to any AESI. For patients receiving LOT1 and LOT2, AESIs were reported by country and those leading to discontinuation are detailed in Supplementary Table S6, (available at https://doi.org/10.1016/j.esmoop.2026.108326). Overall, the most common AESI in LOT1 was cardiac dysfunction (1.6% [11/657]), followed by asymptomatic reduced LVEF of <50% [1.1% (7/657)]; in LOT2, hepatotoxicity was most frequent [0.5% (3/578)] (Supplementary Figure S3, available at https://doi.org/10.1016/j.esmoop.2026.108326).

Discussion

The treatment of HER2-positive mBC is evolving with new HER2-targeted therapies. Decisions on the optimal treatment sequence relies on clinical trial and real-world data. This study analyzed a diverse population from South America, Australia, and Asia, capturing treatment patterns variations influenced by regional health care practices. The findings offer real-world insights into treatment outcomes for patients with HER2-positive mBC. The median age at index diagnosis of our study population was 55 years, which was comparable to that of CLEOPATRA and other studies on patients with HER2-positive breast cancer.31, 32, 33, 34 The majority of patients (99.3%) had stage IV disease at the index date, with about one-fourth experiencing recurrence/relapse within 2 years of initial BC diagnosis. Approximately 68% of patients had visceral only metastases, which was comparable to the results of previous studies.^6,32,33 Additionally, ∼23% of patients exhibited CNS metastasis in our study, which is higher than that reported by the SystHERs US registry study (8.9% of patients at their initial metastatic diagnosis).³⁵ This may have significant clinical implications as the presence of CNS metastasis in HER2-positive mBC is associated with poor prognosis and worse survival outcomes.³⁵

The current SoC for newly diagnosed HER2-positive mBC is TRA+PTZ in combination with a taxane-based CT in the first-line, followed by T-DM1 or the more recently approved T-DXd in the second-line.^16,20 Results from the recent DESTINY-Breast03 trial have demonstrated that among patients previously treated with TRA+CT±PTZ, the risk of disease progression or death was lower for patients treated with T-DXd, with a significantly improved PFS [hazards ratio (HR) 0.30, 95% CI 0.26-0.43, P < 0.001] than for those who received T-DM1, leading to a transition from the previously favored T-DM1 to T-DXd as the standard LOT2 in the metastatic setting.³⁶ Of note, T-DM1 was the recommended second-line treatment regimen during our study period; therefore, the treatment pattern was evaluated based on the utilization of T-DM1 in the second-line. Considering that most patients included in our study may not have had access to T-DXd, the observation of a limited number of patients receiving T-DXd (n = 2) was anticipated. Additionally, tucatinib, which is also currently recommended in third-line for adults with HER2-positive mBC with and without CNS metastases, was not approved in a majority of countries during the study period.^12,37 Our study results showed suboptimal utilization of guideline-recommended treatment strategies in both the first and second lines. Overall, 58.1% patients received first-line TRA+PTZ+CT, with a significant disparity seen across countries (non–TRA+PTZ+CT-based regimen received by >50% of patients from Brazil, Singapore, and Taiwan). This is possibly due to disparity in drug availability, affordability, and reimbursement in study countries.^38,39 Additionally, our study included patients previously treated with HER2-targeted treatment, limiting the use of recommended first and second lines for managing HER2-positive mBC. Moreover, in patients unable to tolerate triplet SoC (TRA+PTZ+CT), non–TRA+PTZ+CT-based combinations are preferred considering the disease burden and patient performance status.

In the second-line, a T-DM1-based regimen was prescribed in almost two-thirds of patients from Australia, the Republic of Korea, and Hong Kong, which aligned with guideline recommendations in both first- and second-line settings. These differences in treatment patterns likely reflect regional variations in guideline adherence, access to medications, and local clinical practices.

Evaluating differences in insurance and reimbursement systems among countries is crucial. In Singapore, the study included only public hospitals, where most patients rely on government subsidies. However, patients with private insurance may preferentially choose to be treated at private oncology clinics rather than at public health institutions to have better access to new treatments. Thus, the Singapore cohort in this study may not reflect the broader national health care situation. Future research should consider these local factors.

The safety profiles of the top regimens used in LOT1 and LOT2 were as expected.²⁰ Cardiac dysfunction and asymptomatically reduced LVEF of <50% were the most commonly reported AESIs, in line with prior studies.²⁰ Due to the retrospective design of this study, only serious AESIs leading to hospitalization were reported; nonserious events could not be captured. Therefore, these results should be interpreted with caution.

The median rwPFS and median OS observed in this study were 12.3 (10.8-13.1) and 46.8 (38.7-50.7) months for TRA+PTZ+CT in LOT1 for the overall cohort. This was lower than that reported by CLEOPATRA [median investigator-assessed PFS 18.7 months; median OS 56.5 (49.3-NC) months] and other trials.^{9,25,33,40,41} The EMILIA trial reported a median PFS of 9.6 months for T-DM1 compared with 6.4 months for lapatinib plus capecitabine (HR 0.65, 95% CI 0.55-0.77, P < 0.001) and a median OS of 30.9 months compared with 25.1 months (HR 0.68, 95% CI 0.55-0.85, P < 0.001). In this study, we report a median rwPFS of 7.2 (6.3-9.6) months for T-DM1 in LOT2, while the median rwOS was 23.9 (20.5–29.5) months for T-DM1 regimen in LOT2.^25,33,40,41 The shorter PFS and OS observed in our real-world study cohort may be due to characteristics of the patients included or suboptimal adherence to guideline-recommended treatment. Notably, patients enrolled in clinical trials tend to have good performance status and fewer comorbidities, and those with CNS involvement are frequently excluded due to strict eligibility criteria. Additionally, the patients included in the T-DM1 cohort in our study included patients who received heterogeneous treatment regimens, combining T-DM1 with endocrine therapy, CDK4/6 inhibitors, tucatinib, and other therapies, limiting the comparative analysis with controlled clinical trials. Results from our study present real-world scenario on patients with HER2-positive mBC. The median rwOS with TRA+PTZ+CT-based regimen was numerically longer than that of TRA+CT; this finding is likely due to a variety of factors that were unaccounted for in this study, including differences in baseline characteristics of patients who received different treatment regimens. Additionally, a higher percentage of patients in our study had CNS metastases, which may have resulted in poor survival outcomes. However, our results highlight a need to further assess the challenges in treatment selection in clinical practice to improve survival outcomes in patients with HER2-positive mBC. Moreover, the study design, requiring inclusion of patients with mBC who had already completed LOT1 with at least 12 months of follow-up (unless deceased), potentially introduced a bias toward enrollment of those with shorter PFS from LOT1. Patients diagnosed with mBC long before their index date and having completed LOT1 in respective participating countries and transitioned to LOT2 at the time of enrollment were excluded as they did not meet the eligibility criteria. Notably, patients maintaining favorable outcomes with the first-line therapy were not included in our study.

In our study, 87.9% of patients moved from first-line to second-line therapy, while 56.3% transitioned from second-line to third-line therapy. Discontinuation was majorly due to disease progression or death and toxicity. The overall attrition rate was 10.7% (70/657) from LOT1 to LOT2 and 17.8% (103/578) from LOT2 to LOT3. The attrition rate in our study was lower than in other reports, possibly because we included only patients with at least one completed LOT and have minimum follow-up.^24,42 In line with other reports, we observed higher attrition rates in later lines, emphasizing the importance of optimal early sequencing of novel therapies to improve management and survival.

Importantly, IHC and/or ISH data were missing approximately one-thirds of patients, which may be due to several factors. Considering the real-world study design, reporting of biomarker testing results may be incomplete in the patient’s medical record, resulting in missing information. Moreover, similar to prospective randomized trials, patients may have been enrolled based on their HER2 status from the primary tumor specimen, and re-biopsy on metastatic sample may not be available or not done.^9,36 This underscores the need for improved and standardized HER2 testing protocols and their expanded access in real-world clinical practices, as HER2 status remains critical for optimal selection of biomarker-guided treatment and inconsistent documentation of HER2 testing results. In our study, most patients (80%) relied on either government-provided/national insurance or self-funding, which may contribute as a barrier to accessing optimal testing facilities in the absence of reimbursement.

This real-world study design used broader inclusion criteria than randomized clinical trials, enabling retrospective data from diverse patients. This approach increased geographic representation and set a benchmark for future HER-positive mBC therapy studies. Data were collected using an electronic data capture system, ensuring consistent and comparable results across regions.

Albeit these strengths, this study had several limitations, most of them inherent to its retrospective nature and the use of chart reviews. Being a retrospective real-word study, the inconsistencies in data collection from the primary health records resulted in missing data for some critical clinical and tumor related parameters such as performance status, hormone receptor status, and metastatic sites, limiting the generalizability of the results. Data on HER2 score was missing for ∼35% of patients, highlighting the need for standardized HER2 testing protocols; some patients with missing LOT end dates were censored at a day before initiation of their next LOT or the last contact date, which may have extended the median rwPFS estimates. Additionally, no standardization for the progression criteria could be applied as this was a chart-based review; thus, variation in imaging frequency and rwPFS interpretation may have led to potential bias. Additionally, Kaplan–Meier analysis could not be carried out in case of <20 events being reported in LOT; thus, we have limited data on OS per country.

The results for the survival outcomes should thus be considered with caution. This was a descriptive study, and no analyses to control for confounding were carried out; therefore, no definitive causal relationships between study endpoints and treatment patterns could be made. Additionally, treatment decisions may have been influenced by contextual factors beyond the scope of the study potentially introducing selection bias, which includes physician judgment, drug accessibility and availability, individual patient characteristics, and institutional protocols. Thus, comparison of results from our study with other clinical trials is limited.

This study aimed to describe treatment patterns and outcomes, with particular emphasis on the importance of guideline-aligned SoC for patients with HER2-positive mBC. Due to the absence of granular chart review, analysis of specific reasons for treatment discontinuation were not possible. Future investigations designed to address this data gap may yield critical insights to support the optimization of therapeutic strategies and enhance patient management.

Conclusion

This retrospective real-world study highlights treatment practices and associated survival outcomes in patients with HER2-positive mBC from Australia, Asia, and Brazil. We observed substantial heterogeneity in first- and second-line regimens across countries, which may be driven by differences in reimbursement pathways, drug availability and affordability, and clinical practice. Uptake of guideline-recommended regimens was poor, underscoring gaps in recommended care in routine practice. Importantly, this analysis is descriptive; it does not establish an optimal treatment sequence. Rather, it provides real-world perspectives on treatment effectiveness and patient outcomes and may serve as a valuable historical benchmark to compare novel treatment paradigms. Future studies are needed to evaluate how improved access to recommended regimens and earlier use of effective targeted agents impact outcomes.