Disitamab vedotin for HER2-positive breast cancer with liver metastases: phase 3 results from the RC48-C006 phase 2/3 trial

Abstract

Patients with HER2-positive advanced breast cancer with liver metastases (BCLM) have poor prognosis. Study RC48-C006 is a prospective, open-label, multi-center, randomized phase 2/3 trial (ClinicalTrials.gov identifier, NCT03500380). Based on the phase 2 results, the phase 3 stage was initiated to evaluate disitamab vedotin (DV) versus lapatinib plus capecitabine (L + C) in patients with HER2-positive BCLM who were previously treated with trastuzumab and taxanes. The results from the phase 3 stage are reported in this article. A total of 104 patients were randomized at 1:1 to receive DV (53 patients) or L + C (51 patients). Progression-free survival (PFS) assessed by the blinded Independent Review Committee (BIRC) was significantly improved with DV versus with L + C (median: 9.9 vs 4.9 months; stratified hazard ratio [HR]: 0.56 [95.48% CI, 0.35–0.91]; 2-sided P = 0.01), meeting the pre-specified primary endpoint. The investigator-assessed PFS was consistent with the BIRC assessment. Prespecified subgroup analysis of PFS consistently favored DV. Similar incidences of grade ≥3 treatment-related adverse events were reported with DV versus L + C (37.7% vs 36.0%). The present findings support DV as a treatment option for HER2-positive BCLM following trastuzumab and taxane therapy.

Patients with HER2-positive breast cancer with liver metastases (BCLM) have poor prognosis. Here, the authors report phase 3 results from the phase 2/3 RC48-C006 clinical trial comparing disitamab vedotin (HER2 targeted antibody drug conjugate) against lapatinib plus capecitabine in patients with HER2-postivie breast cancer with liver metastases previously treated with trastuzumab and taxanes.

Introduction

Patients with breast cancer with liver metastases (BCLM) represent approximately 45% of all human epidermal growth factor receptor 2 (HER2)-positive advanced breast cancer cases and have a particularly poor prognosis^1,2. The median overall survival (OS) for patients with BCLM is only 2 years, far below the 5-year median OS for the general population with advanced breast cancer³. Although there have been huge advances in HER2-targeted therapies for advanced breast cancer in recent years, patients with BCLM urgently demand more effective treatment methods.

HER2-targeted antibody-drug conjugates (ADCs) are poised to reshape the treatment paradigm for breast cancer. Trastuzumab emtansine (T-DM1) and trastuzumab deruxtecan (T-DXd) have been successively approved in the second-line setting, and many novel ADCs are currently under clinical investigation^4–6. T-DXd is now established as the global standard of care for the second-line treatment of HER2-positive metastatic breast cancer⁷. In China, by contrast, the recommended treatment options have included capecitabine combined with a tyrosine kinase inhibitor (such as pyrotinib) or T-DM1 since 2022 or T-DXd since its approval in China in 2023. However, it was not until 2025 that T-DXd was added to the National Reimbursement Drug List, and its accessibility remains limited due to various socioeconomic factors⁸. Furthermore, T-DXd carries a risk of interstitial lung disease^9,10. T-DM1 is associated with a high incidence of severe thrombocytopenia, particularly in Asian populations^11,12. Meanwhile, the combination of pyrotinib and capecitabine, a commonly used regimen in China, is frequently associated with gastrointestinal and skin toxicities¹³. Patients with BCLM continue to have a poor prognosis, thus representing a high unmet need for more effective alternative therapies.

Disitamab vedotin (DV) is a novel humanized immunoglobulin (Ig) G1 HER2-targeted monoclonal antibody conjugated with monomethyl auristatin E (MMAE) via a cleavable peptide linker, with a drug-to-antibody ratio of 4¹⁴. Different from T-DM1 and T-DXd, DV adopts a uniquely designed humanized anti-HER2 antibody, disitamab, which has a higher affinity to immobilized HER2 than trastuzumab as evidenced by preclinical data¹⁴. With the cleavable linker and membrane-permeable cytotoxin MMAE, DV could specifically kill the target tumor cells and also exerts a bystander effect on neighboring tumor cells regardless of HER2 expression levels¹⁵. DV has been conditionally approved for the treatment of gastric cancer and urothelial cancer in China^16,17.

DV showed promising efficacy with manageable safety in patients with HER2-expressing advanced breast cancer based on the previous pooled analysis of two phase 1/1b studies¹⁸. RC48-C006 was initially designed as a randomized phase 2 clinical trial to compare DV versus lapatinib plus capecitabine (L + C, which was a standard regimen at the time of the trial design) in patients with HER2-positive (defined as immunohistochemistry [IHC] 3+ or fluorescence in situ hybridization [FISH] +) advanced or metastatic breast cancer. The interim analysis of the phase 2 stage showed that DV was much more effective on the liver metastases (BCLM) subgroup: the investigator-assessed median progression-free survival (PFS) was 9.3 months with DV versus 7.1 months with L + C in the whole patient population (n = 200) and was 12.5 versus 5.6 months (hazard ratio [HR]: 0.44, nominal p = 0.002) in the subgroup of patients with liver metastases (n = 81). Additionally, the liver is one of the most common metastatic sites for breast cancer; analysis of matched primary tumor and liver metastasis samples suggested that HER2 expression might be higher in liver metastatic lesions than in primary tumors^19,20. Preclinical studies also demonstrated relatively greater accumulation of both DV and MMAE in the liver, with accelerated MMAE release due to liver-specific protease activity²¹. Based on these clinical and preclinical findings, the protocol was amended to add the stand-alone phase 3 stage only focusing on patients with HER2-positive BCLM after the endorsement of the China National Medical Products Administration (NMPA). In June 2021, DV for the treatment of patients with HER2-positive BCLM was granted the breakthrough therapy designation by China’s regulatory authority.

Here, we show the results from the RC48-C006 phase 3 stage, which evaluated the efficacy and safety of DV versus L + C in patients with HER2-positive BCLM.

Results

Patients and treatment

Between August 2, 2021, and December 30, 2022, 104 patients were randomly assigned (with 53 to the DV group and 51 to the L + C group) and thus included in the intention-to-treat (ITT) analysis set (Fig. 1 and Suppl. Table 1). One patient randomized to the L + C group did not receive the assigned treatment and was excluded from the safety analysis set. The demographic and baseline disease characteristics were generally balanced between groups (Table 1 and Suppl. Table 2). All patients were women, and the median age was 53.0 years (range: 29–68). A total of 41 patients (77.4%) in the DV group and 42 patients (82.4%) in the L + C group had multiple-site metastases. Forty-seven patients (88.7%) in the DV group and 45 (88.2%) in the L + C group had received ≤ 1 prior line of chemotherapy; 32.1% and 27.5% of the patients, respectively, had received previous pertuzumab therapy.

Fig. 1. Trial profile.

^a Some patients had multiple reasons for screening failure. More details are provided in Suppl. Table 1. ^b Patients were randomized at a ratio of 1:1 and stratified by prior chemotherapy lines (≤ 1 versus 2) and lung metastases (presence versus absence). ^c A total of 21 patients crossed over and received disitamab vedotin.

Table 1.

Baseline characteristics of all patients

	Disitamab vedotin (N = 53)	Lapatinib plus capecitabine (N = 51)
Age, years
Median (range)	53.0 (29–68)	52.0 (31–68)
< 55	34 (64.2)	31 (60.8)
≥ 55	19 (35.8)	20 (39.2)
Sexa
Female	53 (100)	51 (100)
HER2 status by IHC/FISHb
IHC 3+	46 (86.8)	41 (80.4)
IHC 2+ and FISH+	7 (13.2)	9 (17.6)
IHC 1+ and FISH+	0	1 (2.0)
ECOG performance statusc
0 (fully active)	24 (45.3)	19 (37.3)
1 (restricted in physically strenuous activity but ambulatory)	29 (54.7)	32 (62.7)
Hormone receptor status
Positive	26 (49.1)	33 (64.7)
Negative	27 (50.9)	18 (35.3)
Metastases sites
Liver	53 (100)	51 (100)
Lymph node	28 (52.8)	26 (51.0)
Bone	24 (45.3)	25 (49.0)
Lung	20 (37.7)	20 (39.2)
Skin	5 (9.4)	2 (3.9)
Pleura	1 (1.9)	4 (7.8)
Others	7 (13.2)	9 (17.6)
Prior lines of chemotherapy in the advanced setting
0	24 (45.3)	19 (37.3)
1	23 (43.4)	26 (51.0)
2	6 (11.3)	6 (11.8)
Prior HER2-targeting treatment
Trastuzumab	53 (100)	51 (100)
Pertuzumab	17 (32.1)	14 (27.5)
Others	0	1 (2.0)
Prior chemotherapy treatmentd	53 (100)	51 (100)
Prior endocrine treatment	19 (35.8)	27 (52.9)
Other prior systemic treatmente	4 (7.5)	9 (17.6)

Data are presented as n (%) or median (range).

IHC immunohistochemistry, FISH fluorescence in situ hybridization, ECOG Eastern Cooperative Oncology Group.

^aSex of the patients was determined based on self-report.

^bHER2 status was evaluated at a central laboratory.

^cThe Eastern Cooperative Oncology Group performance status score 5-point scale is commonly used for selection of appropriate patients in clinical trials, which defines 0 as fully active, able to carry on all pre-disease performance without restriction, and defines 1 as restricted in physically strenuous activity but ambulatory and able to carry out work of a light or sedentary nature.

^dIncludes taxanes (which all patients had received as required), cyclophosphamide, anthracycline, platinum, and others.

^eIncludes anti-cancer herbal medicine and traditional Chinese medicine.

As of the data cutoff date (December 31, 2023), treatment was discontinued in 46 (86.8%) of 53 patients in the DV group, and 45 (88.2%) of 51 in the L + C group. The most common reason for treatment discontinuation in both groups was disease progression (71.7% versus 78.4%). The median treatment duration was 31.1 weeks (range: 6.0–109.4) for DV, 23.9 (range: 2.1–104.4) for lapatinib, and 24.0 (3.0–104.4) for capecitabine. Treatment exposure and compliance were summarized in Suppl. Table 3. In the L + C group, 21 (46.7%) of 45 patients who discontinued randomized phase treatment crossed over and received DV treatment. The subsequent anticancer treatment after stopping the study treatment was provided (Suppl. Table 4).

Primary endpoint

As of the data cutoff date, the median duration of follow-up for PFS was 15.4 months (95% CI, 12.6–18.0) with 38 PFS events observed in the DV group, and was 19.4 months (95% CI, 13.9-NE [not estimated]) with 44 PFS events in the L + C group as assessed by Blinded Independent Review Committee (BIRC). Patients in the DV group had significantly longer median PFS than those in the L + C group (median: 9.9 months [95% CI, 5.8–11.7] vs. 4.9 months [4.2–6.9]; hazard ratio [HR]: 0.56 [95.48% CI, 0.35–0.91]; 2-sided P = 0.01; Fig. 2). The estimated 6-month PFS rate was 62.9% (95% CI, 48.1–74.5) in the DV group versus 37.4% (24.2–50.7) in the L + C group, and the 12-month PFS rate was 35.0% (21.9–48.5) versus 16.6% (7.8–28.3).

Fig. 2. Kaplan-Meier curves for BIRC-assessed progression-free survival in the intention-to-treat population.

PFS was assessed by the blinded Independent Review Committee (BIRC) according to Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1. The hazard ratio and its 95.48% CI was calculated using a stratified Cox proportional hazards model. The statistical significance boundary was a two-sided α level of 0.0452 for this final analysis after adjusting for the interim analysis, thus the corresponding “95.48% CI” for hazard ratio was provided. The 2-sided p-value was derived from the stratified log-rank test. The median PFS follow-up was 15.4 months (95% CI, 12.6–18.0) for the disitamab vedotin group and 19.4 months (95% CI, 13.9–not estimated) for the lapatinib plus capecitabine group. Source data are provided as a Source Data file. PFS progression-free survival, CI confidence interval, HR hazard ratio.

The prespecified subgroup analysis showed that the HR for PFS consistently favored the DV group versus the L + C group across all subgroups, including those defined by hormone receptor status and prior pertuzumab exposure (Fig. 3).

Fig. 3. Subgroup analysis of BIRC-assessed progression-free survival.

The black squares represent the subgroup-specific hazard ratios with error bars corresponding to 95% CI bounds, which were estimated using a stratified Cox model. Source data are provided as a Source Data file. BIRC blinded Independent Review Committee, DV disitamab vedotin, L + C versus lapatinib plus capecitabine, HR hazard ratio, CI confidence interval, ECOG Eastern Cooperative Oncology Group.

Secondary endpoints

The investigator-assessed PFS was consistent with BIRC-assessed results; patients receiving DV had significantly longer median PFS than those receiving L + C (median: 8.3 months [95% CI, 5.7–11.2] vs 5.5 months [4.3–6.9; HR: 0.62 [95% CI, 0.39–0.98]; 2-sided P = 0.04; Suppl. Fig. 1).

As of the data cutoff date, 10 (18.9%) of 53 patients in the DV group versus 15 (29.4%) of 51 patients in the L + C group died. With a median follow-up of 17.2 (95% CI, 15.4–19.8) versus 18.4 (16.9–20.3) months, the median OS was not reached in the DV group versus 25.9 months (95% CI, 20.7-NE) in the L + C group. The stratified HR was 0.56 (95% CI, 0.25–1.29; Suppl. Fig. 2). Considering a high proportion (21 [41.2%] of 51 patients) of patients in the L + C group crossed over and received DV, additional prespecified sensitivity analyses for OS were performed to adjust for the crossover effect (Suppl. Fig. 3 and Suppl. Table 5).

As per BIRC assessment, the objective response rate (ORR) was higher in the DV group (31 [58.5%] of 53 [95% CI, 44.1–71.9]) than the L + C group (28 [54.9%] of 51 [95% CI, 40.3–68.9]; Table 2). The clinical benefit rate (CBR) was 71.7% (38/53, [95% CI, 57.7–83.2]) versus 54.9% (28/51, [95% CI, 40.3–68.9]) and disease control rate (DCR) was 94.3% (50/53, [95% CI, 84.3–98.8]) versus 82.4% (42/51, [95% CI, 69.1–91.6]), in the DV and L + C groups, respectively. Among patients with a complete or partial response, the median duration of response (DoR) was 11.2 months (95% CI, 5.8–14.4) in the DV group and 7.0 months (95% CI, 4.1–9.7) in the L + C group. Median time to progression (TTP) was longer in the DV group, with corresponding values of 9.9 months (95% CI, 5.8–11.7) versus 5.0 months (95% CI, 4.2–7.1).

Table 2.

Summary of objective treatment response as assessed by BIRC and by the investigators

	BIRC assessment		Investigators assessment
	Disitamab vedotin (N = 53)	Lapatinib plus capecitabine (N = 51)	Disitamab vedotin (N = 53)	Lapatinib plus capecitabine (N = 51)
Best overall response, n (%)
Complete response	0	2 (3.9)	0	1 (2.0)
Partial response	31 (58.5)	26 (51.0)	29 (54.7)	22 (43.1)
Stable disease	19 (35.8)	14 (27.5)	20 (37.7)	19 (37.3)
Progressive disease	3 (5.7)	6 (11.8)	4 (7.5)	6 (11.8)
Not evaluable	0	3 (5.9)	0	3 (5.9)
ORR, n (%; 95% CI)	31 (58.5; 44.1–71.9)	28 (54.9; 40.3–68.9)	29 (54.7; 40.4–68.4)	23 (45.1; 31.1–59.7)
CBR, n (%; 95% CI)	38 (71.7; 57.7–83.2)	28 (54.9; 40.3–68.9)	39 (73.6; 59.7–84.7)	24 (47.1; 32.9–61.5)
DCR, n (%; 95% CI)	50 (94.3; 84.3–98.8)	42 (82.4; 69.1–91.6)	49 (92.5; 81.8–97.9)	42 (82.4; 69.1–91.6)
DoR, median months (95% CI)	11.2 (5.8–14.4)	7.0 (4.1–9.7)	9.7 (5.2–14.4)	6.9 (4.1–11.1)
TTP, median months (95% CI)	9.9 (5.8–11.7)	5.0 (4.2–7.1)	9.9 (5.7–11.2)	5.5 (4.3–7.1)

BIRC Blinded Independent Review Committee, ORR objective response rate, CI confidence interval, CBR clinical benefit rate, DCR disease control rate, DoR, duration of response, TTP time to disease progression.

Investigator-assessed treatment responses were consistent with the BIRC assessment (Table 2). The ORR was 54.7% (95% CI, 40.4–68.4) versus 45.1% (95% CI, 31.1–59.7), CBR was 73.6% (95% CI, 59.7–84.7) versus 47.1% (95% CI, 32.9–61.5), DCR was 92.5% (95% CI, 81.8-97.9) versus 82.4% (95% CI, 69.1–91.6), the median DoR was 9.7 months (95% CI, 5.2–14.4) versus 6.9 months (95% CI, 4.1–11.1), and the median TTP was 9.9 months (95% CI, 5.7–11.2) versus 5.5 months (95% CI, 4.3–7.1), in the DV and L + C groups, respectively.

Adverse events

Treatment-related adverse events (TRAEs) of any grade occurred in all patients (100%; grade 3–4: 37.7%) treated with DV and in 49 of 50 patients (98.0%; grade 3–4: 36.0%) treated with L + C. No adverse events led to death (grade 5 events; Table 3). The most common TRAEs of any grade were aspartate aminotransferase increased (77.4% with DV vs. 58.0% with L + C), alanine aminotransferase increased (67.9% vs. 38.0%), neutrophil count decreased (52.8% vs. 42.0%), white blood cell count decreased (49.1% vs. 48.0%), hypoesthesia (47.2% vs. 4.0%), gamma-glutamyltransferase increased (39.6% vs. 18.0%), alopecia (39.6% vs. 2.0%), blood bilirubin increased (17.0% vs. 54.0%), diarrhea(13.2% vs. 46.0%), and palmar-plantar erythrodysesthesia syndrome (0 vs. 42.0%). No patients experienced interstitial lung disease (ILD) or pneumonitis. The most common grade 3–4 TRAEs in either group were neutrophil count decreased (18.9% with DV vs. 8.0% with L + C), gamma-glutamyl transferase increased (7.5% vs. 0.0%), white cell count decreased (5.7% vs 2.0%), hypokalemia (1.9% vs 8.0%), palmar-plantar erythrodysesthesia syndrome (0.0% vs. 10.0%), and diarrhea (0.0% vs. 6.0%) (Table 4). Platelet count decrease of grade ≥ 3 did not occur in the DV group. Serious TRAEs occurred in 6 patients (11.3%) receiving DV and in 4 patients (8.0%) receiving L + C (Suppl. Table 6). TRAEs led to treatment discontinuation in 3 patients (5.7%) in the DV group versus none in the L + C group (Suppl. Table 7).

Table 3.

Summary of adverse events

	Disitamab vedotin (N = 53)	Lapatinib plus capecitabine (N = 50)a
		Lapatinib	Capecitabine	Total
TEAE	53 (100)	/	/	50 (100)
TRAE	53 (100)	49 (98.0)	49 (98.0)	49 (98.0)
Grade ≥ 3 TEAE	22 (41.5)	/	/	20 (40.0)
Grade 3	18 (34.0)	/	/	15 (30.0)
Grade 4	4 (7.5)	/	/	5 (10.0)
Grade 5	0			0
Grade ≥3 TRAE	20 (37.7)	15 (30.0)	16 (32.0)	18 (36.0)
Grade 3	16 (30.2)	11 (22.0)	12 (24.0)	14 (28.0)
Grade 4	4 (7.5)	4 (8.0)	4 (8.0)	4 (8.0)
Grade 5	0	0	0	0
TEAE leading to dose reduction	15 (28.3)	8 (16.0)	21 (42.0)	23 (46.0)
TRAE leading to dose reduction	15 (28.3)	8 (16.0)	21 (42.0)	23 (46.0)
TEAE leading to treatment interruption	37 (69.8)	26 (52.0)	25 (50.0)	28 (56.0)
TRAE leading to treatment interruption	24 (45.3)	20 (40.0)	16 (32.0)	21 (42.0)
TEAE leading to drug discontinuation	3 (5.7)	0	0	0
TRAE leading to drug discontinuation	3 (5.7)	0	0	0
TEAE leading to death	0	/	/	0
TRAE leading to death	0	0	0	0
Serious TEAE	8 (15.1)	/	/	6 (12.0)
Serious TRAE	6 (11.3)	4 (8.0)	3 (6.0)	4 (8.0)

Data are presented as n (%). Adverse events were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 5.0. Grading ranges from 1 through 5 (1, mild; 2, moderate; 3, severe; 4, life-threatening; and 5, death).

^aOne patient in the lapatinib plus capecitabine didn’t receive assigned treatment after randomization and was excluded from the safety analysis set.

TEAE treatment-emergent adverse event, TRAE treatment-related adverse event.

Table 4.

Treatment-related adverse events of any grade and grade 3-4 in the safety population

	Disitamab vedotin (N = 53)		Lapatinib plus capecitabine (N = 50)a
	Grade 3-4b	All grades	Grade 3-4b	All grades
Treatment-related adverse events in 20% or more of treated patients in either groupc
Aspartate aminotransferase increased	1 (1.9)	41 (77.4)	0	29 (58.0)
Alanine aminotransferase increased	0	36 (67.9)	0	19 (38.0)
Neutrophil count decreased	10 (18.9)	28 (52.8)	4 (8.0)	21 (42.0)
White blood cell count decreased	3 (5.7)	26 (49.1)	1 (2.0)	24 (48.0)
Hypoesthesia	2 (3.8)	25 (47.2)	0	2 (4.0)
Gamma-glutamyltransferase increased	4 (7.5)	21 (39.6)	0	9 (18.0)
Alopecia	0	21 (39.6)	0	1 (2.0)
Nausea	1 (1.9)	20 (37.7)	0	13 (26.0)
Hypertriglyceridaemia	2 (3.8)	18 (34.0)	2 (4.0)	15 (30.0)
Weight decreased	0	18 (34.0)	0	12 (24.0)
Blood lactate dehydrogenase increased	0	17 (32.1)	0	5 (10.0)
Hypercholesterolaemia	0	15 (28.3)	0	5 (10.0)
Anemia	0	15 (28.3)	0	8 (16.0)
Hyperglycaemia	2 (3.8)	14 (26.4)	0	5 (10.0)
Asthenia	0	12 (22.6)	0	12 (24.0)
Hyperuricaemia	0	11 (20.8)	0	7 (14.0)
Blood alkaline phosphatase increased	0	10 (18.9)	0	11 (22.0)
Blood bilirubin increased	0	9 (17.0)	1 (2.0)	27 (54.0)
Vomiting	1 (1.9)	8 (15.1)	0	10 (20.0)
Hypokalaemia	1 (1.9)	7 (13.2)	4 (8.0)	10 (20.0)
Diarrhea	0	7 (13.2)	3 (6.0)	23 (46.0)
Rash	1 (1.9)	3 (5.7)	0	10 (20.0)
Palmar-plantar erythrodysesthesia syndrome	0	0	5 (10.0)	21 (42.0)

Data are n (%). Treatment-related adverse events were related to any study drug. Adverse events were classified according to the Medical Dictionary for Regulatory Activities and graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events, version 5.0. Grading ranges from 1 through 5 (1, mild; 2, moderate; 3, severe; 4, life-threatening; and 5, death).

^aOne patient in the lapatinib plus capecitabine group didn’t receive assigned treatment after randomization and was excluded from the safety analysis set.

^bNo grade 5 adverse events occurred.

^cTreatment-related adverse events to any drug occurring in 20% or more of patients in either group are shown in descending order of frequency in the disitamab vedotin group.

All DV-related peripheral neuropathy events were grade 1–3, with the most common (occurring in ≥ 5.0% of patients) being hypoesthesia (47.2%; grade 3: 3.8%), pain in extremities (15.1%; grade 3: 3.8%), neurotoxicity (7.5%; grade 3: 1.9%), and peripheral sensory neuropathy (7.5%; grade 3: 0) (Suppl. Table 8). Peripheral neurological toxicity led to treatment discontinuation in 2 patients (one patient with hypoesthesia and peripheral sensory neuropathy, and one with hypoesthesia).

Discussion

The randomized phase 3 study met its prespecified primary endpoint, with DV demonstrating a statistically significant PFS benefit versus L + C in patients with HER2-positive BCLM who previously received trastuzumab and taxane. Based on these findings, DV was approved for this population by the China National Medical Products Administration on April 30, 2025.

To date, no specific standard treatment has been established for patients with BCLM. Treatment selection for BCLM is typically guided by histologic and molecular characteristics of the primary tumor or biopsied metastatic lesions. HER2-enriched breast tumors exhibit the highest propensity for liver metastasis²². Given the urgent need for effective therapies in this high-risk population, dedicated clinical investigation is warranted. To our knowledge, this represents the first prospective randomized clinical trial specifically designed to evaluate a HER2-targeted therapy exclusively in patients with BCLM.

In the randomized phase 3 stage of this study, DV provided a significantly longer median PFS of 9.9 months versus 4.9 with L + C (HR: 0.56 [95.48% CI, 0.35–0.91]), consistent with the benefit observed with DV (HR for PFS: 0.44) in the phase 2 stage of the study. The median PFS observed with DV in this trial was comparable to that reported for T-DM1 in the EMILIA study, but numerically shorter than that observed with T-DXd in DESTINY-Breast03 or with pyrotinib plus capecitabine in PHOEBE^12,13,23. This difference is probably due to the intrinsically poorer prognosis associated with liver metastases, a well-established negative prognostic factor^2,3. Supporting this, the median PFS in the L + C group was also numerically shorter in the present study (4.9 months) than that reported in EMILIA (6.4 months), PHOEBE (6.8 months), DESTINY-Breast03 (6.8 months), and ACE-Breast-02 (7.1 months) studies^{5,12,13,23,24}. Furthermore, baseline tumor burden in our population was substantial, with the majority of patients having multiple-site metastases. These high-risk features likely contributed to the shortened PFS observed, consistent with prior reports demonstrating inferior outcomes in HER2-positive patients with liver involvement compared to those without^1–3.

Subgroup analyses demonstrated a consistent PFS benefit favoring DV over L + C across all prespecified subgroups, including those defined by hormone receptor status and prior pertuzumab exposure. Although baseline hormone receptor status was imbalanced between arms (HR-positive: 49.1% in the DV group vs 64.7% in the L + C group), the treatment effect was maintained across both HR-positive and HR-negative subgroups, suggesting that the observed benefit is independent of this imbalance.

Within the current treatment landscape, the Chinese Society of Clinical Oncology (CSCO) guidelines recommend T-DXd, pyrotinib plus capecitabine, and T-DM1 for patients with HER2-positive advanced breast cancer previously treated with trastuzumab and a taxane⁸. While these agents are available in China, their distinct toxicity profiles may limit their use in certain patient populations. T-DXd is associated with ILD/pneumonitis; although its incidence appears relatively low in breast cancer populations (10.5% in DESTINY-Breast03), this toxicity remains potentially life-threatening if not detected early. Gastrointestinal adverse events, including nausea (72.8%) and vomiting (44.0%), are also frequently reported with T-DXd⁹. T-DM1 is associated with grade ≥ 3 thrombocytopenia and occurred in 40.4% of patients in the ELAINA trial conducted in a Chinese population¹¹. Pyrotinib plus capecitabine, a widely used regimen in China, is characterized by frequent gastrointestinal toxicities, particularly diarrhea (reported in 95% of patients in the PHOEBE study), as well as capecitabine-related hand-foot syndrome¹³. In this study, no cases of ILD/pneumonitis or grade ≥ 3 thrombocytopenia were observed with DV. Consistently, previous studies of DV in patients with advanced breast cancer rarely reported ILD/pneumonitis¹⁸. DV-related gastrointestinal toxicity in this study was mild. Therefore, DV may offer a valuable therapeutic alternative for patients with preexisting pulmonary conditions who are at increased risk for ILD, as well as for those predisposed to severe gastrointestinal adverse events or thrombocytopenia. DV may fill a critical gap for the clinically heterogeneous patient population in China.

The safety profiles observed in this study were consistent with the known toxic effects of DV, lapatinib, and capecitabine. No new safety signals were observed^16,18. Elevated aminotransferases were the predominant adverse events in patients receiving DV, with an increase of aspartate aminotransferase, alanine aminotransferase, and gamma-glutamyl transferase occurring in more than 30% of patients. This incidence was similar to or higher than that of other HER2-targeted ADCs in patients with breast cancer^5,12,23. Importantly, most of these liver laboratory abnormalities were grade 1–2 in severity, despite all enrolled patients having baseline liver metastases and potentially compromised hepatic function. Hematologic toxicities, primarily decreases in neutrophil and white blood cell counts, were the second most frequent TRAEs and did not lead to treatment discontinuation. Peripheral nervous toxicity is an identified risk of DV and a known class effect of MMAE-containing ADCs. Although DV-related peripheral neuropathy occurred frequently, they were clinically manageable, and the majority of events were grade 1-2. Its incidence and severity observed in this study were consistent with other vedotin ADCs, including tisotumab vedotin²⁵ and enfortumab vedotin²⁶.

This study had a few limitations. First, this study was designed as open-label, considering the different administration protocols of the study drugs, but the use of BIRC assessment for PFS might mitigate the potential bias. Second, the median OS in the DV group was not reached by the cutoff date, thus requiring longer follow-up to fully characterize the long-term treatment benefit of DV versus L + C. Third, the control arm appears to be outdated in the context of the current global treatment landscape. At the time of trial design in early 2021, neither T-DM1 nor T-DXd was approved in China for HER2-positive metastatic breast cancer, and L + C represented the standard second-line therapy. Although T-DM1 was approved in China in 2022, its accessibility remained limited, and it had not yet been widely integrated into routine clinical practice at the time of study conduct. Moreover, Asian populations, including Chinese patients, were shown to exhibit poorer tolerability to T-DM1, particularly with respect to severe thrombocytopenia^11,12. These factors, together with the prescribing preferences and clinical experience of Chinese oncologists at the time, supported the continued relevance of L + C as a comparator. Fourth, the representativeness of the study population is limited. Patients were required to have received prior trastuzumab and a taxane, with or without pertuzumab. The treatment pattern, including the inclusion of first-line patients and the fact that not all patients had received prior pertuzumab, does not fully align with current Western standards for HER2-positive metastatic breast cancer. However, they were consistent with the standard of care in China during the study period. The relatively low rate of prior pertuzumab use (30%) reflects limited access and affordability constraints at the time of enrollment, consistent with another contemporary second-line breast cancer study conducted in China, in which 34.4% of patients in the ARX788 (HER2-targeting ADC) group and 28.2% in the control (L + C) group had received prior pertuzumab²⁷. Importantly, subgroup analysis demonstrated consistent PFS benefits with DV irrespective of prior pertuzumab exposure (HR 0.40 in patients with prior pertuzumab and 0.57 in those without).

Despite these limitations, this study provides clinically meaningful and context-specific evidence for the Chinese population within the current treatment context. It received China regulatory approval in 2025. It addresses the clinical reality of a substantial patient segment for whom, due to comorbidities or healthcare-access considerations, T-DXd or pyrotinib plus capecitabine may not be optimal choices.

In conclusion, this study demonstrates a significant and clinically meaningful PFS benefit with DV compared with L + C in patients with HER2-positive BCLM who have received prior trastuzumab and a taxane. These findings support DV as an additional standard treatment option for this high-risk patient population.

Methods

Study design and participants

This randomized, multi-center, open-label, phase 3 clinical trial was conducted following Good Clinical Practice, the principles of the Declaration of Helsinki, and all relevant local regulations. This trial is registered with ClinicalTrials.gov, number NCT03500380. The study protocol (Suppl. Note 2) was approved by the ethics committees of each participating sites (provided in the Suppl. Table 9). All patients provided signed informed consent before study enrollment. Patients were offered minor compensation (e.g., travel cost coverage). The patients were enrolled between August 2, 2021, and December 30, 2022, across 39 participating sites in China.

Key eligibilities were 18–70 years of age, histologically/cytologically confirmed invasive breast cancer with radiologically confirmed liver metastasis, central lab-confirmed HER2-positive (defined as IHC 3+ or FISH + ) according to the Chinese breast cancer HER2 detection guideline (version 2019)²⁸, prior treatment with trastuzumab and taxanes, the number of prior chemotherapy lines ≤ 2, baseline Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1, and at least one measurable disease as per Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST v1.1). Patients must have adequate organ function, including bone marrow function (hemoglobin ≥ 9 g/dL; absolute neutrophil count ≥ 1.5 × 10⁹/L; platelets ≥ 100 × 10⁹/L), liver function (based on the normal values specified by study site: serum total bilirubin ≤ 1.5 × the upper limit of normal [ULN]; alanine aminotransferase, aspartate aminotransferase and alkaline phosphatase ≤ 5 × ULN), renal function (based on the normal values specified by study site: serum creatinine ≤ 1.5 × ULN, or creatinine clearance ≥ 60 mL/min as calculated by Cockcroft-Gault formula, or 24-hour urine creatinine clearance ≥ 60 mL/min), and cardiac function (New York Heart Association classification <class III; left ventricular ejection fraction ≥ 50%). Specifically, liver function requirements were as follows: serum total bilirubin ≤ 1.5 × upper limit of normal (ULN); alanine aminotransferase, aspartate aminotransferase, and alkaline phosphatase ≤ 5 × ULN. Patients under any of the following situations were excluded: brain metastasis, carcinomatous meningitis, systemic disease that was not under stable control (including but not limited to diabetes, hypertension, pulmonary fibrosis, acute lung disease, interstitial lung disease, and liver cirrhosis), and active infection requiring systemic treatment.

Randomization and masking

Patients were randomized at a ratio of 1:1 to receive either DV or L + C using block randomization (block size of 4) via a centralized interactive web response system. The stratification factors included prior chemotherapy lines (≤ 1 versus 2) and lung metastases (presence versus absence). The patients and investigators were not masked to the assigned treatment because of different administration protocols of the study drugs. Tumor assessment by the Independent Review Committee (IRC) was performed in a blinded manner.

Treatment

Patients randomly assigned to the DV group received DV (2.0 mg/kg) intravenously once every 2 weeks. DV dose is based on calculations using bovine serum albumin (BSA)–based extinction coefficient (EC) in China. Outside of China, DV dose calculation is based on DV-based EC, which is equivalent to 1.07 (BSA-based EC) ÷ 1.41 (DV-based EC) × BSA-based EC dose. Patients assigned to the L + C group received oral lapatinib (1250 mg) once daily and oral capecitabine (2000 mg/m²) once daily on days 1–14 every 3 weeks. Treatment continued until the occurrence of disease progression, death, intolerable toxicity, withdrawal of patient consent, or study termination, whichever occurred first. Patients who progressed on L + C were permitted conditionally, if deemed potentially beneficial by the investigators, to crossover to receive DV treatment.

Assessment

Tumors were assessed using computed tomography (CT) or magnetic resonance imaging (MRI) at baseline, every six weeks until disease progression (defined as radiographic disease progression confirmed by IRC), death, withdrawal of informed consent, initiation of new anti-cancer treatment, or study termination, whichever occurred first, by both the blinded IRC (BIRC) and the investigators according to RECIST v1.1. Radiographic scans of the chest, abdomen, and pelvis were performed at baseline and at all post-baseline tumor assessments. Scans of the bones, head, and other parts were performed at baseline and at post-baseline assessments when clinically indicated. The imaging modality used at baseline was used for individual patients in the subsequent tumor assessments.

Vital signs, physical examination, laboratory examination (including hematology, biochemistry, and urinalysis), and 12-lead electrocardiograms were assessed at the following time points: during screening (within 7 days prior to first dose), within 3 days before each cycle’s dose, at treatment discontinuation, and during safety follow-up. Adverse events were continually monitored from the time of informed consent until 28 days after the last study treatment. Adverse events were coded using the Medical Dictionary for Regulatory Activities (MedDRA version 27.0). The severity of adverse events was assessed using Common Terminology Criteria for Adverse Events (CTCAE) version 5.0. The investigators assessed the attribution of adverse events to any study drugs.

HER2 expression was evaluated by both IHC, using Ventana anti-HER2/Neu (4B5) Rabbit mAb and the ultra-View Universal DAB Detection Kit (Roche), and FISH at an independent central lab. Tumor specimens from primary or metastatic lesions collected before enrollment were acceptable. The IHC staining scores and FISH status were determined according to the Chinese breast cancer HER2 detection guideline, version 2019²⁸.

Endpoints

The primary endpoint was BIRC-assessed PFS (defined as the time from randomization to disease progression according to RECIST v 1.1 or death from any cause, whichever occurred first). Secondary endpoints included investigator-assessed PFS, objective response rate (ORR, the proportion of patients with a confirmed complete or partial response), disease control rate (DCR, the proportion of patients with a confirmed objective response, or stable disease), clinical benefit rate (CBR, the proportion of patients with a confirmed objective response or maintenance of stable status for at least 6 months after randomization), duration of response (DoR, the time from the first complete or partial response to the radiographic disease progression or death due to any cause, whichever occurred first), time to progression (TTP, the time from randomization to occurrence of disease progression), and OS (the time from randomization to death from any cause). Objective response rate, disease control rate, clinical benefit rate, duration of response, and time to progression were assessed by both the BIRC and the investigators as per RECIST v1.1. Adverse events were categorized as treatment-related if the investigator attributed causality to any study drug or if a causal relationship could not be established. Quality of life and other secondary endpoints including pharmacokinetics, pharmacodynamics, and immunogenicity data are not included in this article and will be disclosed in a separate publication.

Sample size estimation

The sample size of the phase 3 stage was based on the findings in the subgroup analysis of the patients with HER2-positive BCLM (median PFS: 12.5 months with DV versus 5.6 months with lapatinib plus capecitabine; hazard ratio: 0.44, nominal P = 0.002) in the phase 2 stage of RC48-C006. Thus, with a more conservative assumption of an HR for PFS of 0.53 (median PFS for the L + C group: 5.6 months) and an annual dropout rate of approximately 5%, the phase 3 stage required to enroll a total of 100 patients for 80 PFS events in both groups to provide an 80% power at a two-sided α level of 0.05.

An interim analysis was planned to be performed when 60% of the total planned BIRC-assessed PFS events were reported in both groups (i.e., a total of 48 PFS events). Sample size re-estimation was planned to be evaluated by an independent data monitoring committee (IDMC) based on the conditional power at the interim analysis, and the sample size would be adjusted to achieve the target conditional power of 80% when the conditional power was 0.35–0.8 according to the promising zone approach in Mehta & Pocock (2011)²⁹. The allowed maximum sample size was set as twice the initial sample size.

At the interim analysis, the IDMC recommended not to proceed with any adjustment to the originally planned sample size. Consequently, the study continued to the final analysis without a sample size modification.

Statistical analysis

The efficacy analyses set included all patients randomly assigned to each treatment group (ie, the intention-to-treat [ITT population), regardless of whether they had received treatment or not. Safety analyses included all patients who received at least one dose of study treatment (ie, the safety population). PFS was analyzed using the Kaplan-Meier method, with median PFS and corresponding 95% CIs provided. A stratified log-rank test was used to test the significance of differences between the DV group and the L + C group; HRs and 95% CIs were calculated using a stratified Cox proportional hazards model. In the PFS analysis, patients who were alive without disease progression at the data cutoff date were censored at the time of the last tumor assessment. Patients without baseline or post-baseline BIRC assessments who remained alive at the data cutoff were censored on the date of randomization. Patients who experienced disease progression or death after two or more consecutive missed tumor assessments were censored at the date of last valid imaging prior to the date of disease progression, death or data cutoff (whichever occurred first). Patients who initiated a new anti-tumor therapy prior to disease progression, death or the data cutoff date were censored at the date of their last imaging assessment prior to initiation of the new anti-tumor therapy.

Secondary time-to-event endpoints (including duration of response, time to progression, and OS) were analyzed using the same methods as for PFS. To reduce the confounding effect due to crossover from the L + C group, OS was adjusted by using data censored at the first dose of crossover treatment and additional analysis methods including rank preserving structural failure time model, two-stage model, and inverse probability of censoring weighting. The 95% CIs for objective response rate, disease control rate, and clinical benefit rate were calculated using the Clopper-Pearson method. Subgroup analyses of PFS were prespecified in the statistical analysis plan prior to the database lock. The HR and corresponding 95% CIs for each subgroup (including age, baseline ECOG performance status, prior lines of chemotherapy, lung metastases, hormone receptor, and prior pertuzumab treatment) were estimated using a stratified Cox model. Safety data were summarized descriptively; safety data after crossover to receive DV treatment were not reported.

An interim analysis and final analysis were planned. This interim analysis was performed according to the protocol by IDMC; the IDMC suggested continuing the study without any adjustment to the originally planned sample size. The statistical significance boundary was a two-sided α level of 0.0452 for the final analyses after adjusting for the interim analysis.

More details are provided in the Statistical Analysis Plan (Suppl. Note 3). All statistical analyses were performed using the SAS software, version 9.4 (SAS Institute).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Suppl. information

Suppl. Information ^{(2.5MB, pdf)}

Reporting Summary ^{(107.8KB, pdf)}

Peer Review file ^{(688.8KB, pdf)}

Author contributions

Jiayu W. and B.X. had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Concept and design: Jiayu W., S.G., Y.H., J.F. and B.X. Acquisition, analysis, or interpretation of data: All authors. Drafting of the manuscript: Jiayu W. and B.X. Critical review of the manuscript for important intellectual content: All authors. Formal statistical analysis: Y.H. and D.G. Administrative, technical, or material support: J Jiayu W., S.G., Y.H. and B.X. Supervision: Jiayu W., Q.O., W.X., Z.N., Q.Z., X.Y., Y.T., J.C., A.W., Jingfen W., H.Y., Z.Y., T.S., Z.T., Xinhong W., Y.W., E.Z., X.K., Xujuan W., D.Z. and B.X.

Peer review

Peer review information

Nature Communications thanks Hui Zhang, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The study protocol is available in the Suppl. file (Suppl. Note 2). All study data supporting the findings of this study are available in the article, Suppl. file, and the Source Data file. Individual participant data are not publicly available due to patient privacy, intellectual property, or confidentiality obligations. Deidentified individual participant data that underlie the results reported in this article can be made available to other qualified medical researchers who provide a methodologically sound proposal, via a request to the corresponding author (xubh@cicams.ac.cn). The data set will be provided in compliance with applicable privacy laws, data protection, and requirements for consent and anonymization, after full publication of the study and related work. A signed data access agreement with the sponsor is required before accessing shared data. No expiration date for data requests has been set once the data are made available. Source data are provided with this paper.

Code availability

No custom code was used for statistical analysis in this study.

Competing interests

Shanshan Gu, Dandan Gao, Zerui Qu, and Chenran Han were employees of RemeGen. Jianmin Fang is a shareholder of RemeGen. The other authors have no conflict of interest to declare.

Supplementary information

The online version contains Suppl. material available at https://doi.org/10.1038/s41467-026-75733-y.