A Phase II study of SHR-A1921, a TROP2-targeted antibody–drug conjugate, in patients with recurrent or metastatic salivary gland carcinoma

Abstract

Advanced salivary gland carcinoma (SGC) lacking both human epidermal growth factor receptor 2 (HER2) and androgen receptor (AR) expression has limited treatment options. In this single-center, phase II trial, patients with recurrent/metastatic HER2-negative/AR-negative SGC received SHR-A1921, a TROP-2-directed antibody-drug conjugate, every 3 weeks until disease progression or unacceptable toxicity. The primary endpoint was objective response rate (ORR) per RECIST v1.1; secondary endpoints included progression-free survival (PFS), safety, and exploratory analyses of baseline TROP-2 expression by immunohistochemistry. As of August 1, 2025, 17 patients were treated and 15 were evaluable for efficacy, including 10 with non-adenoid cystic carcinoma (non-ACC) and 5 with adenoid cystic carcinoma (ACC). In non-ACC SGC, confirmed ORR was 20.0%, disease control rate was 80.0%, and clinical benefit rate was 60.0%. All patients with ACC achieved stable disease, with median PFS of 15.4 months. Treatment-related adverse events were manageable; oral mucositis was most common. No grade 4–5 treatment-related adverse events or treatment-related serious adverse events occurred; one patient discontinued treatment due to toxicity. Baseline TROP-2 expression was heterogeneous, with numerically higher H-scores in responders but no significant association with PFS. SHR-A1921 showed preliminary activity in non-ACC SGC and disease stabilization in ACC, supporting further evaluation in stratified cohorts. Trial registration: ClinicalTrials.gov, NCT05924256. Registered on June 29, 2023.

Introduction

Salivary gland carcinoma (SGC) comprises a diverse group of rare epithelial malignancies arising in the major and minor salivary glands, accounting for less than 5% of all head and neck cancers^1–3. Despite its histopathologic heterogeneity, encompassing over 20 recognized subtypes, treatment options for advanced or metastatic SGC remain largely limited to surgery, radiation, and empiric chemotherapy^4,5. Patients with unresectable, recurrent, or metastatic salivary gland carcinoma, particularly those lacking human epidermal growth factor receptor 2 (HER2) overexpression or androgen receptor (AR) positivity, have poor prognosis and limited responses to chemotherapy^5–7, while novel therapies remain understudied^6,7. This represents a critical unmet need in precision oncology for rare cancers.

Trophoblast cell surface antigen 2 (TROP-2), a transmembrane glycoprotein implicated in tumor proliferation and metastasis, is overexpressed across SGC histologies, including salivary duct, mucoepidermoid, and adenoid cystic carcinomas, and correlates with poor prognosis^8–10. While TROP-2-directed antibody-drug conjugates (ADCs) like sacituzumab govitecan have demonstrated efficacy in breast, urothelial, and lung cancers^11,12, their potential in SGC remains unexplored, particularly for patients lacking actionable targets (e.g., HER2-negative, AR-negative disease), who face limited treatment options and no approved targeted therapies^5–7. Given the high prevalence of TROP-2 in SGC^8–10 and the clinical success and favorable safety profile of TROP-2 ADCs in other malignancies^11–14, we evaluated SHR-A1921, a novel ADC comprising a humanized anti-TROP-2 IgG1 antibody conjugated to a topoisomerase I inhibitor via a cleavable linker (drug-to-antibody ratio: 3.6–4.8)^14,15, in the first prospective phase II trial for advanced SGC. This study assessed the clinical activity, safety, and potential biomarker correlates of SHR-A1921 as a precision therapy for TROP-2-expressing SGCs.

Results

Patient disposition and treatment status

Seventeen patients with recurrent or metastatic HER2-negative/AR-negative SGC were enrolled and received SHR-A1921, forming the safety population (Fig. 1). At data cutoff (August 1, 2025), seven patients (5/11 [45%] non-ACC; 2/6 [33%] ACC) remained on treatment. Treatment discontinuations were primarily attributed to disease progression (n = 3), consent withdrawal (n = 6), and toxicity (n = 1). Fifteen patients (10 non-ACC and 5 ACC) were evaluable for efficacy. One patient in each group who withdrew consent after one treatment cycle and had no post-baseline tumor assessment was excluded from the efficacy-evaluable population.

Fig. 1. Patient disposition.

Flow diagram illustrating screening, enrollment, and analysis populations. Seventeen patients with salivary gland cancer were screened and initiated treatment with SHR-A1921 (safety population). Of these, 10 patients with non-adenoid cystic carcinoma (non-ACC) and 5 with adenoid cystic carcinoma (ACC) were evaluable for efficacy analyses. As of the data cutoff date (August 1, 2025), seven patients remained on SHR-A1921. Abbreviations: ACC adenoid cystic carcinoma, non-ACC non-adenoid cystic carcinoma.

Baseline characteristics

Baseline characteristics of the safety population (n = 17) are summarized in Table 1. The median age was 53 years (range 33–70 years), with 4 patients (23.5%) aged ≥60 years. Ten patients (58.8%) were male. All patients had ECOG performance status 1. Prior systemic therapy included none in 8 patients (47.1%), one line in 6 patients (35.3%), and ≥2 lines in 3 patients (17.6%). The majority of patients presented with advanced disease: 16 patients (94.1%) had stage IVC disease with distant metastases. The most common metastatic sites were the lungs (n = 16; 94.1%) and lymph nodes (n = 10; 58.8%), followed by the liver (n = 4; 23.5%), pleura (n = 3; 17.6%), and bone (n = 2; 11.8%). Histologic subtypes included adenoid cystic carcinoma (n = 6; 35.3%), with the remaining 11 patients having diverse non-ACC subtypes: clear cell carcinoma of salivary gland (n = 2), myoepithelial carcinoma (n = 2), salivary duct carcinoma (n = 2), and one each of polymorphous adenocarcinoma, sebaceous carcinoma, carcinoma ex pleomorphic adenoma, adenocarcinoma, not otherwise specified, and mucoepidermoid carcinoma. Given the indolent clinical course of certain subtypes, including ACC and clear cell carcinoma of salivary gland, all such patients had documented radiologic progression or worsening symptoms within 6 months before study entry (Supplementary Table 1).

Table 1.

Baseline patient characteristics (n = 17)

Characteristic	n = 17
Age, years
Mean (SD)	51.4 (11.1)
Range	33–70
Median	53
≥60	4 (23.5)
Sex
Female	7 (41.0)
Male	10 (58.8)
ECOG performance status
0	0
1	17(100)
Prior lines of drug therapy
None	8 (47.1)
One	6 (35.3)
≥Two	3 (17.6)
Prior therapy
Systemic Platinum-based chemotherapy	3 (17.6)
Stage at initial diagnosis
ⅣA	1 (5.9)
ⅣC	16 (94.1)
Distant metastases
Yes	16 (94.1)
No	1 (5.9)
Sites of metastases
Lung	16 (94.1)
Lymph node	10 (58.8)
Liver	4 (23.5)
Pleura	3 (17.6)
Bone	2 (11.8)
Histologic subtype
Adenoid cystic carcinomaa	6 (35.3)
Clear cell carcinoma of salivary glanda	2 (11.8)
Myoepithelial carcinoma*	2 (11.8)
Salivary duct carcinomab	2 (11.8)
Adenocarcinoma, not otherwise specifiedc	1 (5.9)
Carcinoma ex pleomorphic adenoma	1 (5.9)
Polymorphous adenocarcinoma	1 (5.9)
Sebaceous carcinoma	1 (5.9)
Mucoepidermoid carcinoma	1 (5.9)

Data reported as No. (%) unless otherwise indicated.

^aAll patients with indolent histologic subtypes had documented radiologic progression or new/worsening symptoms within 6 months before study entry (see Supplementary Table 1 for detailed clinical information).*One case was diagnosed as myoepithelial carcinoma ex pleomorphic adenoma.

^bBoth cases underwent expert pathology review. The salivary duct carcinoma was classified as AR-negative per study definition because only rare/focal AR expression was observed in the recurrent/metastatic specimen, despite AR positivity at initial diagnosis.

^cExpert pathology review revised the diagnosis to adenocarcinoma, not otherwise specified. Abbreviations: ECOG, Eastern Cooperative Oncology Group; NOS, not otherwise specified; SD, standard deviation.

Efficacy outcomes

Non-ACC Population (n = 10): Best overall responses were CR in 1 patient (10.0%), confirmed PR in 1 patient (10.0%), SD in 6 patients (60.0%), and progressive disease (PD) in 2 patients (20.0%). This yielded an ORR of 20.0% (2/10; exact 95% CI, 2.5–55.6%) and DCR of 80.0% (8/10; exact 95% CI, 44.4–97.5%). Clinical benefit was achieved in 60.0% of patients (6/10; exact 95% CI, 26.2–87.8%) (Table 2, Fig. 2). Kaplan–Meier analysis demonstrated early progression events followed by a plateau phase, consistent with durable disease control in a subset of patients (Fig. 2C). One PD event was determined based on unequivocal clinical progression with incomplete imaging data per protocol.

Table 2.

Efficacy outcomes in patients with salivary gland carcinoma

Outcome	ACC (n = 5)	non-ACC (n = 10)
Best response, n (%)
CR	0	1 (10.0)
PR	0	1 (10.0)
SD	5 (100.0)	6 (60.0)
PD	0	2 (20.0)a
ORR, %	0	20.0
DCR, %	100.0	80.0
Clinical benefit rate, n (%)	3 (60.0)	6 (60.0)
PFS, months, median (95% CI)	15.4 (14.4 to 16.4)	NR

Data are reported as No. (%) unless otherwise indicated. Best overall response, clinical benefit, and treatment status as of the data cutoff (August 1, 2025). Outcomes are shown for patients with non-ACC and ACC histology.

^aOne patient met PD criteria based on unequivocal clinical progression; target-lesion assessment was not possible owing to incomplete data.

CR complete response, PR partial response, SD stable disease, PD progressive disease, NR not reach, ORR objective response rate, DCR disease control rate, PFS progression-free survival, CI confidence interval, ACC adenoid cystic carcinoma.

Fig. 2. Efficacy outcomes of SHR-A1921 in patients with non-adenoid cystic carcinoma (non-ACC) salivary gland cancers (n = 10) or adenoid cystic carcinoma (ACC) (n = 5).

A Best change from baseline in target-lesion sum per RECIST v1.1. Each bar represents one patient. One patient was classified as having progressive disease based on clinical deterioration, with insufficient imaging data for formal evaluation. Dashed horizontal lines indicate the thresholds for progressive disease (+20%) and partial response (-30%). B Swimmer plot showing treatment duration and best overall response. Patients who remained on treatment at the data cutoff date of August 1, 2025 are identified by a right-pointing triangle. C Kaplan–Meier curve of progression-free survival (PFS). Abbreviations: non-ACC non–adenoid cystic carcinoma, RECIST Response Evaluation Criteria in Solid Tumors, PFS progression-free survival.

ACC population (n = 5): All patients achieved SD as best overall response (Table 2, Fig. 2). One patient demonstrated near-partial response with 29.6% maximal tumor reduction, while four patients had minimal volumetric changes. The median treatment duration was 15.0 months (range 4.0–23.0 months). Clinical benefit occurred in 60.0% of patients (3/5). With median PFS of 15.4 months (95% CI, 14.4–16.4 months) after one progression event. Overall survival rate remained 100%, with median follow-up of 17.3 months (range 4.3–23.8 months).

Safety profile

Treatment-related adverse events (TRAEs) are summarized in Table 3. The most common any-grade TRAEs were oral mucositis (11/17; 64.7%), anemia (7/17; 41.2%), and alopecia (6/17; 35.3%). Five patients (29.4%) each experienced decreased lymphocyte count, cough, vomiting, weight loss, decreased appetite, fatigue, and elevated γ-glutamyltransferase. Grade 3 TRAEs were primarily oral mucositis (6/17; 23.5%) and decreased lymphocyte count (1/17; 5.9%). No Grade 4 or 5 adverse events occurred, and no serious adverse events were attributed to SHR-A1921. Adverse events led to temporary treatment interruptions in nine patients (52.9%). One patient with ACC discontinued treatment due to inability to tolerate oral mucositis and subsequently withdrew consent; however, in the non-ACC cohort, all toxicities were manageable without requiring treatment discontinuation.

Table 3.

Most common treatment-related adverse events (TRAEs), regardless of attribution, in patients receiving SHR-A1921 (safety population, n = 17)

Adverse event	Any grade, n (%)	Grades 3–4a, n (%)
Oral mucositis	11 (64.7)	4 (23.5)
Anemia	7 (41.2)	0 (0.0)
Alopecia	6 (35.3)	0 (0.0)
Lymphocyte count decreased	5 (29.4)	1 (5.9)
Cough	5 (29.4)	0 (0.0)
Vomiting	5 (29.4)	0 (0.0)
Weight decreased	5 (29.4)	0 (0.0)
γ-glutamyl transferase increased	5 (29.4)	0 (0.0)
Fatigue	5 (29.4)	0 (0.0)
Decreased appetite	5 (29.4)	0 (0.0)
White blood cell count decreased	4 (23.5)	0 (0.0)
Hypoalbuminemia	4 (23.5)	0 (0.0)
Pyrexia (Fever)	2 (11.8)	0 (0.0)
Neutrophil count decreased	2 (11.8)	0 (0.0)
Sinus tachycardia	2 (11.8)	0 (0.0)
Sinus Bradycardia	2 (11.8)	0 (0.0)
Nausea	2 (11.8)	0 (0.0)
ALT increased	2 (11.8)	0 (0.0)
ALP increased	2 (11.8)	0 (0.0)
Rash	2 (11.8)	0 (0.0)
Platelet count decreased	1 (5.9)	0 (0.0)
Infusion-related reaction	1 (5.9)	0 (0.0)
Esophageal ulcer	1 (5.9)	0 (0.0)
Infectious pneumonia	1 (5.9)	0 (0.0)
Constipation	1 (5.9)	0 (0.0)

Data are n (%). For each patient and preferred term, only the highest grade is counted when events occurred multiple times. Events shown are those occurring in ≥2 patients at any grade and/or with any grade ≥2 occurrence. Grading per study protocol Criteria for Adverse Events (CTCAE) version 5.0 in force at trial initiation.

^aNo grade 4–5 adverse events were observed.

ALT Alanine Aminotransferase, ALP Alkaline Phosphatase.

TROP-2 expression analysis and association with efficacy

TROP-2 protein expression was evaluated by immunohistochemistry in baseline tumor specimens from all 10 efficacy-evaluable non-ACC patients (Table 4, Fig. 3A, B). Quantitative analysis revealed marked heterogeneity across histologic subtypes, with H-scores ranging from 0.24 to 65.02 (median 20.6, mean 28.04). Expression patterns varied significantly by histology: clear cell carcinoma demonstrated highest expression (H-scores 42.9–65.0), followed by sebaceous carcinoma (H-score 55.6) and adenocarcinoma, not otherwise specified (H-score 52.7). Minimal expression was observed in mucoepidermoid carcinoma (H-score 0.24) and polymorphous adenocarcinoma (H-score 4.18), while myoepithelial carcinomas showed intermediate levels (H-scores 13.2–16.1).

Table 4.

Quantification of TROP-2 protein expression and clinical response non-ACC salivary gland cancers

Case #	Best response	Positive rate, %	Intensity			H-score	Histologic subtype
			1+	2+	3+
1	SD	32.39	1385	1346	1415	65.02	Clear cell carcinoma of salivary gland
2	PD	27.38	4118	2116	1041	42.90	Clear cell carcinoma of salivary gland
3	PR	14.31	1265	980	580	25.14	Salivary duct carcinoma
4	PD	10.22	184	68	55	16.14	Myoepithelial carcinoma
5	CR	29.50	5231	4309	2714	52.68	Adenocarcinoma, not otherwise specified
6	SD	0.22	25	1	1	0.24	Mucoepidermoid carcinoma
7	SD	4.20	950	213	59	5.32	Carcinoma ex pleomorphic adenoma
8	SD	31.05	1667	1385	846	55.57	Sebaceous carcinoma
9	SD	10.27	1228	282	98	13.23	Myoepithelial carcinoma
10	SD	3.52	109	14	5	4.18	Polymorphous adenocarcinoma

TROP-2 expression was assessed by immunohistochemistry in baseline tumor specimens. Data include positivity rate (%), staining intensity (1 + , 2 + , 3 + ), and H-score for each case, with histologic subtype and best overall response shown.

CR complete response, PR partial response, SD stable disease, PD progressive disease, non-ACC non–adenoid cystic carcinoma.

Fig. 3. Analysis of TROP-2 expression in non-ACC tumors and its association with SHR-A1921 efficacy.

A Representative immunohistochemistry (IHC) images showing the spectrum of TROP-2 staining in non-ACC tumor samples. B Example fields illustrating how H-scores were assigned; scoring was performed in a blinded manner. C Distribution of TROP-2 H-scores and TROP-2 positivity rates stratified by best overall response (CR/PR, SD, PD). D Kaplan–Meier estimates of progression-free survival (PFS) by TROP-2 expression, using the cohort mean H-score (28.04) as the dichotomizing cutoff (low, n = 6; high, n = 4). Numbers at risk are shown below the plot, and the log‑rank p value is indicated. Abbreviations: IHC immunohistochemistry, non-ACC non–adenoid cystic carcinoma, CR complete response, PR partial response, SD stable disease, PD progressive disease, PFS progression-free survival, SGC salivary gland cancer.

Exploratory biomarker analysis suggested numerically higher TROP-2 expression among patients achieving objective responses compared to those with SD/PD (Fig. 3C), though responses occurred across both lower and higher expression ranges. Stratification by the cohort mean H-score (28.04) yielded no statistically significant difference in PFS between TROP-2-high (n = 4) and TROP-2-low (n = 6) subgroups (log-rank p = 0.7077; Fig. 3D). These findings are hypothesis-generating and limited by small sample size and histologic heterogeneity.

Representative objective responses

Figure 4 illustrates radiographic responses in two non-ACC patients. A 70-year-old man with salivary duct carcinoma achieved confirmed PR with marked regression of pulmonary and nodal disease (Fig. 4A). A 33-year-old man with adenocarcinoma, not otherwise specified experienced >50% tumor shrinkage by month 3, subsequently achieving CR with resolution of all measurable disease (Fig. 4B). Both patients maintained ongoing responses on SHR-A1921 at data cutoff. Baseline TROP-2 expression was elevated in both cases relative to the cohort median (H-scores 25.14 and 52.68, respectively; Table 4), though the limited sample size precludes definitive biomarker conclusions.

Fig. 4. Radiographic responses in representative patients treated with SHR-A1921.

Paired baseline and on-treatment imaging from two non-ACC patients: A a 70-year-old man with salivary duct carcinoma who achieved a partial response (PR) and B a 33-year-old man with adenocarcinoma, not otherwise specified (NOS) who achieved a complete response (CR). Arrows indicate the target lesions. Abbreviations: non-ACC non–adenoid cystic carcinoma, CR complete response, PR partial response.

Discussion

This single-arm, phase II study is, to our knowledge, the first prospective evaluation of a TROP-2-ADC (SHR-A1921) in a molecularly defined cohort of HER2-negative/AR-negative SGC, a population with limited treatment options⁵. The trial used a Simon two-stage design and did not cross the Stage-1 continuation boundary for the primary endpoint (details in Methods), and therefore was negative with respect to its prespecified primary endpoint. In the non-ACC subset, the confirmed ORR was 20.0% (2/10; 95% CI, 2.5–55.6), the disease control rate was 80.0% (8/10; 95% CI, 44.4–97.5), and clinical benefit was observed in 60.0% of evaluable patients. Median PFS was not reached in non-ACC and was 15.4 months (95% CI, 14.4–16.4) in ACC at a median follow-up of 17.3 months. Because of small sample size and censoring, these findings are hypothesis-generating and non-comparative; historical chemotherapy outcomes (overall survival ~10–14 months first-line; ~4 months post-progression)^16–18 are provided for clinical context only. The prolonged PFS observed in ACC should be interpreted with caution and may reflect, at least in part, the indolent biology of this subtype rather than drug activity alone. Taken together, despite not meeting the primary endpoint, the limited activity observed in selected non-ACC histologies may warrant cautious further evaluation in biomarker-enriched, histology-informed cohorts with prospectively defined predictive markers and prespecified decision rules. However, these findings remain exploratory and hypothesis-generating rather than definitive evidence of subtype-specific activity.

TROP-2-targeted ADCs have shown antitumor activity in breast, urothelial, and lung cancers, supporting TROP-2 as a druggable target¹¹. Across SGC, TROP-2 expression is common (≥80%) with frequent strong membranous staining^8–10. While TROP-2 overexpression is prognostic in several malignancies¹¹, its predictive role in SGC remains unproven¹⁰. In our cohort, TROP-2 H-score did not associate with PFS, and antigen abundance alone did not predict SHR-A1921 benefit: responses occurred in salivary duct and adenocarcinoma, not otherwise specified, whereas clear cell, sebaceous, and ACC had 0% ORR despite high H-scores. Although both responding cases underwent expert pathology review, these observations warrant cautious interpretation because salivary gland carcinomas are diagnostically heterogeneous and can remain difficult to classify with certainty in challenging cases. Therefore, IHC-based H-scoring remains pragmatic but limited. Consistent with prior work, ADC activity depends on multiple determinants beyond surface expression, including antigen accessibility/membrane localization, internalization and lysosomal processing, stromal/vascular delivery, and payload susceptibility^19–23. To operationalize these factors, immuno-PET with ⁸⁹Zr- or ⁶⁸Ga-labeled anti-TROP-2 antibodies and spatial/pharmacodynamic profiling can noninvasively assess target engagement, biodistribution, and intratumoral heterogeneity, and via window-of-opportunity biopsies link delivery to payload effect^19,24. In addition, given ACC’s distinct biology and potential indolence, future work should treat ACC separately and prioritize deliverability-focused strategies rather than antigen level alone^20,25–29.

While the immunologic effects of SHR-A1921 were not assessed here, preclinical work with topoisomerase-I ADCs suggests potential for immunogenic cell death (ICD), including calreticulin exposure, High Mobility Group Box 1 (HMGB1) release, and Adenosine Triphosphate (ATP) secretion, with downstream dendritic-cell activation and T-cell infiltration^21,22. Accordingly, future SGC studies should prospectively evaluate ICD readouts (calreticulin/HMGB1/ATP), spatial immune contexture (including PD-L1 and tumor-infiltrating lymphocytes), and clonotypic dynamics, to determine whether ICI combinations are biologically justified.

In this study, the safety profile of SHR-A1921 appeared clinically manageable in this small cohort of patients with advanced SGC: oral mucositis was the most frequent TRAE (64.7% any-grade, 23.5% Grade ≥3), with most toxicities being low grade, and no adjudicated interstitial lung disease was reported. Although cross-study comparisons must be interpreted with caution due to differences in study design, patient populations, dosing regimens, ADC constructs, and tumor subtypes, SHR-A1921’s mucosal tolerability showed a higher oral mucositis incidence than in related indications with other TROP2-targeted ADCs²³ (e.g., <15% implied for sacituzumab govitecan in relapsed/refractory head and neck squamous cell carcinoma [HNSCC])³⁰. However, similar elevated rates of stomatitis have been noted as a class effect in other tumor types, including gynecologic cancers (e.g., 21.7–66% any-grade with certain TROP2 ADCs)³¹, indicating on-target mucosal toxicity beyond SGC-specific factors. In addition, many patients with SGC have undergone prior surgery and/or radiotherapy to the head and neck region, which may result in pre-existing mucosal vulnerability and further increase the clinical relevance of oral mucositis in this population, with potential adverse effects on quality of life. Nonetheless, the predominance of low-grade events, low discontinuation rate (5.9%), and absence of severe pulmonary toxicities align with class expectations and indicate manageability with supportive interventions. These findings support the continued evaluation of SHR-A1921 in patients with TROP2-expressing head and neck tumors, with emphasis on supportive measures for mucositis (e.g., prophylactic rinses and dose modifications) in future trials.

This study has several limitations. The single-center, single-arm design and small sample size limit generalizability. The inclusion of indolent histologic subtypes may also complicate interpretation of antitumor activity in the absence of a control group. In addition, the relatively high rate of consent withdrawal may have introduced bias, although most of these patients remained evaluable for efficacy. Interpretation of efficacy and safety is further limited by the fact that only 1 patient received a starting dose of 4.0 mg/kg, precluding any meaningful assessment of dose-related differences. All biomarker analyses were exploratory, and the absence of pharmacodynamic data limits insight into on-target engagement and tumor penetration. Despite these limitations, this study provides a mechanism-anchored rationale, hypothesis-generating activity signals, and a clinically manageable safety profile in this rare disease setting. Future trials should incorporate histology-specific expansion cohorts, integrated biomarker platforms, and immunologic correlates. For example, evaluating SHR-A1921 in salivary duct carcinoma or adenocarcinoma, not otherwise specified using a prospective spatial transcriptomic biomarker panel may help identify the most responsive subsets. In the era of precision oncology, advances in ADC design, patient selection, and biomarker discovery will be critical to improving care for SGC.

In summary, this study provides the first prospective evidence of antitumor activity for a TROP-2-targeted ADC in SGC. The biologically informed patient selection, manageable toxicity, and sustained disease control observed with SHR-A1921 support its further evaluation in this rare and understudied malignancy.

Methods

Study design

This open-label, single-center, phase II umbrella trial was conducted at Fudan University Shanghai Cancer Center. The trial comprised four parallel biomarker-defined treatment arms, with patient assignment based on tumor expression of HER2 and AR³². This report presents results from Arm 3, which evaluated the anti-TROP-2 ADC SHR-A1921 in patients with HER2-negative and AR-negative SGC. The study was conducted in accordance with the Declaration of Helsinki and International Council for Harmonisation Good Clinical Practice guidelines. The protocol was approved by the Institutional Ethics Board of Fudan University Shanghai Cancer Center (No. 2304273-5), and all participants provided written informed consent for trial participation and tissue-based biomarker analysis. Trial registration: ClinicalTrials.gov, NCT05924256. Registered on June 29, 2023.

Patients and interventions

Eligible patients were adults aged 18–75 years with histologically confirmed, unresectable, recurrent, or metastatic SGC, at least one measurable lesion per RECIST v1.1, an Eastern Cooperative Oncology Group (ECOG) performance status of 0-1, and adequate organ function (absolute neutrophil count ≥1.5 × 10⁹/L, creatinine clearance >50 mL/min, left ventricular ejection fraction ≥50%).

For Arm 3, eligible patients required to have HER2-negative (IHC 0 or 1 + ) and AR-negative (defined as ≤10% nuclear staining by IHC) disease. Key exclusion criteria included uncontrolled cardiovascular disease, hypertension, or bleeding disorders; systemic anticancer therapy within 28 days before enrollment; and pregnancy.

SHR-A1921 was administered intravenously on day 1 of each 21-day cycle. Sixteen patients received a starting dose of 3.0 mg/kg, and 1 patient received 4.0 mg/kg at investigator discretion. Dose reduction was allowed for persistent or intolerable adverse events. Treatment continued until disease progression, unacceptable toxicity, withdrawal of consent, or investigator-determined discontinuation.

TROP-2 immunohistochemistry and digital H-score quantification

Formalin-fixed, paraffin-embedded tumor sections were stained with a rabbit monoclonal anti-TROP-2 antibody (ZuoChengBio, catalog no. ZA0412) at 1:7200 dilution, as previously described³³. Staining and digital H-score quantification were performed by Shanghai Zuo Cheng Bio technicians blinded to clinical outcomes, histologic subtype, and diagnosis. Slides were scanned (NanoZoomer S210; .ndpi format) and analyzed using Visiopharm 2025.02. Image analysis employed the following modules: Tissue Detection (for tissue masking), Nuclei Detection-AI (an artificial intelligence-based algorithm for nuclear identification), and Cell Segmentation (for cellular reconstruction). Regions of interest were excluded from necrosis, hemorrhage, and artifacts. TROP-2 intensity was classified into 0/1 + /2 + /3+ categories using three fixed thresholds (T1-T3) calibrated against concurrent positive and negative controls. The H-score was calculated per region as: (1 × % 1+ cells) + (2 × % 2+ cells) + (3 × % 3+ cells) (range: 0–300), with case-level scores reported as cell-count-weighted means across all regions.

End points and assessments

The primary endpoint was objective response rate (ORR) per RECIST v1.1 in the prespecified efficacy population. Secondary endpoints included progression-free survival (PFS), overall survival (OS), and safety. Disease control rate (DCR) was defined as the proportion of patients achieving a best overall response of complete response (CR), partial response (PR), or stable disease (SD). Clinical benefit rate (CBR) was defined as the proportion of patients achieving CR, PR, or durable SD lasting ≥6 months³⁴. An exploratory analysis assessed the association between baseline TROP-2 expression and clinical outcomes.

Tumor response by computed tomography or magnetic resonance imaging was conducted at baseline (within 28 days of random assignment) and then every 6 weeks (plus or minus 7 days) until disease progression. Investigator-assessed complete or partial responses required confirmation by repeat imaging ≥4 weeks after the initial response. Adverse events (AEs) were graded according to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0 at each study visit. Safety evaluations, including laboratory tests and 12-lead electrocardiogram (ECG), were performed on day 1 of each treatment cycle, with safety follow-up continuing for 30 days after the last dose of SHR-A1921.

Statistical analysis

Given the absence of prior TROP-2-targeted therapy trials in SGC, the sample size was based on historical ORRs of ~35% with standard chemotherapy^,35,36,37. We prespecified a target ORR of 45% (10% absolute improvement) as clinically meaningful. A Simon’s two-stage design tested H₀: ORR ≤ 20% vs. H₁: ORR ≥ 45%, with α = 0.049 and 80.7% power. Owing to the biological distinction between ACC and non-ACC SGC subtypes supported by prior evidence, the trial design called for enrolling 10 patients per group in Stage 1. Efficacy outcomes for each cohort are thus summarized descriptively and separately.

The safety population included all patients who received ≥1 dose of SHR-A1921. The efficacy population was initially defined as all HER2-negative/AR-negative SGC patients with at least one post-baseline tumor assessment per RECIST v1.1.

PFS was defined as time from registration to progression or death; OS as time to death from any cause. ORR, DCR, and CBR were reported with exact 95% confidence intervals (CIs). Survival endpoints were estimated using Kaplan–Meier methods. AEs were summarized descriptively by grade and attribution. All statistical analyses were conducted using jamovi (v2.6; https://www.jamovi.org), and figures were generated using GraphPad Prism (v10.0).

Author contributions

These authors contributed equally: Guang-Liang Chen and Yanjing Guo. D.J. contributed substantially to the study conception and design. G-L. C., Y.G., Y.Y., Y.S., S.S., X.L., X.Z., C.S., X.H., R.S., Y.W., Q.J., and D.J. collected the data. G-L. C., Y.G., and D.J. analyzed data. G-L. C. and D.J. drafted the original manuscript. All authors interpreted the data and critically revised the manuscript. All authors approved the final version of the manuscript prior to submission.

Data availability

The datasets generated and/or analyzed during the current study are not publicly available due to patient privacy concerns and institutional/ethical restrictions but are available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41698-026-01463-5.