Abstract Biliary tract cancers (BTC) are a rare set of genetically heterogeneous and aggressive malignancies associated with late presentation, poor prognosis and limited effective therapies. Thus, there is an unmet clinical need for the development of novel therapeutic strategies. Identification of uniquely expressed cell-surface tumor-associated antigens (TAA) holds promise in epithelial tumors more generally, as they can serve as ligands for a variety of therapeutics, including CAR-T cell therapy. However, TAA identification in BTCs has been limited by expression in normal liver tissue. Here, we identify the cell-surface protein transmembrane 4 L six family member 1 (TM4SF1) as a potential BTC TAA targetable by CAR-T therapy. We find TM4SF1 expression is upregulated in BTC relative to normal hepatic/biliary tissue at the mRNA level in the TCGA dataset. Correspondingly, we find that TM4SF1 in primary BTC archival tissue is upregulated at the protein level using immunohistochemistry. To validate TM4SF1 as a targetable TAA in BTC, we show that TM4SF1-directed CAR-T cells demonstrate robust, dose-dependent growth inhibition of human-derived BTC cell lines in vitro and significant anti-tumor activity in a heterotopic BTC cell line-derived xenograft model in vivo. Together, our data provide support for TM4SF1 as a promising TAA in BTCs that can be used for the rational development of therapeutic modalities targeting TM4SF1, including CAR-T cells, in a patient population with significant unmet need. Citation Format: Lorraine Nuniz, Franklin Huang. TM4SF1 as a novel tumor-associated antigen in biliary tract cancers targetable by immune effector cell therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 2667.

4569 Background: Enfortumab vedotin plus pembrolizumab (EV±P) is standard-of-care for patients (pts) with locally advanced/metastatic urothelial carcinoma (la/mUC), although not all pts experience durable benefit. Tumor-informed circulating tumor DNA (ctDNA) assays enable a personalized, real-time assessment of disease kinetics, and early ctDNA dynamics (<12 weeks) have been linked to EV±P radiographic response and survival outcomes. However, the prognostic relevance of ctDNA dynamics relative to durable clinical benefit is not well defined. Methods: We conducted a multicenter, retrospective real-world analysis of pts with la/mUC who received ≥1 cycle of EV±P and ≥1 commercial plasma ctDNA test using a clinically validated, personalized, tumor-informed assay (Signatera, Natera, Inc.). Associations between ctDNA features and progression-free survival (PFS) and overall survival (OS) were evaluated using Cox proportional hazards models. Results: A total of 110 pts with 545 plasma samples were analyzed. All pts had ≥1 ctDNA test (median: 4 tests/pt) immediately prior to and/or after EV±P initiation, with median time between serial ctDNA collections of 7.3 (IQR: 5.4-12.0) weeks. In this cohort, 69% (76/110) of pts had ≥1 ctDNA test (355 total) beyond 12 weeks post-EV±P initiation. Among these 76 pts, 31.5% (24/76) were persistently ctDNA negative (≥2 consecutively negative samples with no positives), while 65.8% (50/76) were anytime ctDNA-positive. Time-varying covariate analysis incorporating all serial results after 12-weeks reaffirmed that ctDNA-positivity was associated with worse PFS and OS (PFS: HR: 36.8, p<0.005; OS: HR: 17.7, p=0.01). Serial ctDNA negativity within 4-24 weeks (i.e. ≥2 consecutively negative samples with no positives) was strongly associated with durable OS/PFS beyond 24 weeks (PFS: HR=0.05, 95% CI: 0.01-0.42, p=0.005; OS: HR=0.046, 95% CI: 0.0004-0.33 [Firth], p=0.0002). Furthermore, single-timepoint ctDNA positivity evaluated at 12–24 weeks (HR: 11.8; 95% CI: 2.72–51.60; p=0.001) and >24 weeks (HR: 6.8; 95% CI: 2.5–18.6; p=0.0002) was strongly associated with inferior PFS. Findings were similar for OS (12-24wk: HR: 7.2, 95% CI: 1.7-31.3; p=0.008; >24wk: HR=3.8, CI: 1.2-12.3; p=0.03). Conclusions: Longitudinal ctDNA dynamics are strongly prognostic in pts with la/mUC treated with EV±P, even when assessed beyond 12 weeks from treatment start. Sustained ctDNA negativity further identifies pts with durable clinical benefit. These findings support ctDNA as a potentially complementary biomarker to standard imaging in this clinical context.

Figure S25. PET/CT imaging and biodistribution with [89Zr]DFO-YS5-MMAE in DU145 tumor-bearing mice. (A) Axial, coronal, and MIP PET/CT images of mice (M1–M3) at 24, 48, 96, and 168 hours post-injection show sustained tumor uptake of [89Zr]DFO*-YS5-MMAE. Color scale indicates %ID/g. (B) Organ distribution of [89Zr]DFO*-YS5-MMAE in DU145 bearing mice, at 168 h post-injections.

Overexpression of the proto-oncogene Src is common to a wide variety of cancers. In this work, we found that Src is noncanonically translocated and inverted onto the cell surface in cancer, both in vitro and in vivo. We identified autophagolysosomal exocytosis (ALE) as a secretory mechanism prominent in cancer cell lines. Src represents the prototypical example of a family of membrane-anchored proteins that are transported by this process. Furthermore, this extracellular membrane-associated Src (eSrc) was found in primary tumors, and anti-Src antibody-based therapies mediated tumor cell killing in cell culture systems and in mouse xenograft models. Thus, intracellular N-myristoylated proteins, prototypically Src, can be topologically inverted onto the cell surface in cancer and targeted with antibody therapeutics.

Figure S27. Experimental design and dosing schematic for in vivo combination therapy studies. A schematic timeline illustrating the therapeutic protocol utilized for subcutaneous prostate cancer xenograft models (22Rv1).

Figure S13: A saturation binding assay was performed on DU145 cells, showing Kd values of 3.2 and 3.0 nM for YS5-MMAE and Macropa-PEG4-YS5-MMAE.

Supplementary Table S5. Region of interest analysis data reported as %ID/cc of the [89Zr]PEG-DFB1-TLZ3 on on different subcutaneous tumors from Day 1-4 post injection. (n = 4)

Supplementary Figure S13. Autoradiographic images and H&E staining images of subcutaneous CT26 tumor sections from day 1 to day 4 post injection of 89Zr labeled nanocarriers. *Indicate the presence of necrosis.

Supplementary Figure S4. Coronal μPET/CT fusion and MIP images obtained at 18 h, 48 h, 72 h, and 96 h following administration of 150-170 μCi of 89Zr labeled nanocarriers in nude mice bearing 22rv1 subcutaneous tumors over left flanks.

Figure S9: Experimental design and dosing schematic for in vivo combination therapy studies shown in Figure 2. A schematic timeline illustrating the therapeutic protocol utilized for subcutaneous prostate cancer xenograft models (22Rv1).

Supplementary Figure S2. Pilot PET/CT studies in prostate, pancreatic, and colorectal cancer models to measure the enhanced permeability retention effect using [89Zr]PEG-DFB1-TLZ3.

Figure S17. Biodistribution of [225Ac]Macropa-PEG4-YS5-MMAE in 22Rv1 tumor-bearing mice at day 4 post-injection comparing low (0.1 kBq/µg) and high (6.3 kBq/µg) specific activity formulations (n=4).

Supplementary Table S4. Region of interest analysis data reported as %ID/cc of the [89Zr]PEG-DFB4 on different subcutaneous tumors from Day 1-4 post injection. (n = 4)

Supplementary Figure S23. Organ biodistribution presented in (left) %ID/g, and (right) %ID/organ of 89Zr labeled nanocarriers in nude mice inoculated with 22rv1 cells via intracardiac injection at 72 h postinjection. (n = 4, mean ± SD)

Supplementary Figure S21. Autoradiographic images and H&E staining images of metastatic tumor, and kedney sections from day 1 to day 4 post injection of [89Zr]PEG-DFB1-TLZ3 in nude mice inoculated with 22rv1 cells in the left kidney capsule. (NA = Data not available)

Figure S2: ZIP synergy analysis of the combination of [225Ac]Macropa-PEG4-YS5 and YS5-MMAE in DU145 and LNCaP cells. (A) ZIP synergy score heatmap for DU145 cells treated with matrix of YS5-MMAE and [225Ac]Macropa-PEG4-YS5 to study the synergistic effect. The heatmap indicates regions of synergistic interaction (red regions) and antagonistic interaction (green regions). (B) ZIP synergy score heatmap for C4-2B cells treated with varying dose of [225Ac]Macropa-PEG4-YS5 and YS5-MMAE, and their combinations. The heatmap demonstrates differential synergy and antagonism compared to DU145 cells, highlighting cell-line-specific responses to the combination therapy.

Figure S19. iTLC showing radiolabeling of [134Ce]Macropa-PEG4-YS5-MMAE (A), and purity following the PD10 purification process (B).

Supplementary Table S1. Region of interest analysis data reported as %ID/cc of the [89Zr]PEG-DFB1-TLZ3 on on different subcutaneous tumors on 24 h and 72 h post injection. (n = 4)

Supplementary Table S3. Organ biodistribution data in %ID/Organ tissue for [89Zr]PEG-DFB1-TLZ3 at 72 h in mice model bearing different subcutaneous tumors. (n = 4)

Figure S22. Biodistribution of [134Ce]Macropa-PEG4-YS5 and [134Ce]Macropa-PEG4-YS5-MMAE in 22Rv1 tumor-bearing mice at 168 hours post-injection (n=4). %IA/g values are shown for major organs, including the liver, kidneys, spleen, and tumor.