The logic behind targeting splice variants is elegant. Because alternative splicing generates protein isoforms that exist only in specific tissues or disease states, drugs aimed at a single isoform can, in principle, spare the other isoforms produced by the same gene. That selectivity reduces off-target toxicity while concentrating destructive power on tumor cells. The review’s authors, Masuko and Masaru Katoh of M & M Precision Medicine in Tokyo, note that dysregulated splicing regulators such as ESRP1, ESRP2, SF3B1, SRSF2 and U2AF1 can globally rewire which isoforms a tumor produces, while splice-site mutations in oncogenes and tumor suppressors can locally generate neo-antigens that exist nowhere in healthy tissue. Both mechanisms make splice variants attractive targets for monoclonal antibodies, antibody-drug conjugates, bispecific antibodies and engineered T cells.

CLDN18.2 is the most advanced example. The CLDN18 gene produces two isoforms through alternative promoters: CLDN18.1, restricted to lung tissue, and CLDN18.2, the stomach isoform. The two proteins are nearly identical except for a short N-terminal region, the first transmembrane domain and the first extracellular loop, all encoded by isoform-specific first exons. In healthy gastric epithelium, CLDN18.2 sits buried within tight junctions, inaccessible to circulating antibodies. But in gastric and gastroesophageal junction (G/GEJ) cancers, disrupted cell polarity redistributes the protein to non-junctional cell surfaces, exposing it to drug attack. An early immunohistochemical study detected moderate-to-strong CLDN18.2 staining in 56 percent of gastric cancers, 60 percent of pancreatic ductal adenocarcinomas and 50 percent of esophageal adenocarcinomas, establishing the antigen as a pan-gastrointestinal target.

Zolbetuximab, a mouse-human chimeric IgG1 antibody against CLDN18.2, became the first therapy in this class to reach regulatory approval. As monotherapy it produced an objective response rate of just 9 percent in a phase 2 trial, but when combined with standard chemotherapy in the phase 3 SPOTLIGHT and GLOW studies, it extended median overall survival to 18.2 and 14.4 months respectively, versus 15.5 and 12.2 months with chemotherapy alone. The US Food and Drug Administration has now approved zolbetuximab plus fluoropyrimidine- and platinum-based chemotherapy as first-line treatment for CLDN18.2-positive G/GEJ cancer patients. Yet the modest single-agent activity of the first-generation antibody left substantial room for improvement, and that room is now being filled.

Second-generation antibodies have been engineered for higher binding affinity and enhanced effector functions. Omectatug, a humanized IgG1, achieved a single-agent response rate of 22 percent in early-phase trials—more than double that of zolbetuximab. When combined with capecitabine and oxaliplatin chemotherapy plus an immune checkpoint inhibitor, omectatug produced a combination response rate of 76 percent. FG-M108, engineered with an afucosylated Fc region to boost antibody-dependent cellular cytotoxicity, reached 78 percent in combination with the same chemotherapy backbone. Osemitamab, which uses a diagnostic antibody targeting the CLDN18.2-specific first extracellular loop rather than a region shared with CLDN18.1, produced a 68 percent response rate with a median overall survival of 21.7 months in first-line chemoimmunotherapy. All three have advanced into phase 3 randomized trials.

Antibody-drug conjugates represent an even more aggressive escalation. These molecules couple antibodies to cytotoxic payloads—typically monomethyl auristatin E or topoisomerase I inhibitors—via chemical linkers, delivering lethal doses directly to antigen-positive cells while sparing healthy tissue. Eight CLDN18.2-targeting ADCs have now reported single-agent response rates ranging from 23 to 47 percent in G/GEJ cancer patients, with ciletatug vedotin leading at 47 percent. Arcotatug tavatecan, designed with a silenced Fc region specifically to attenuate complement-mediated toxicity and mitigate gastrointestinal side effects, achieved 35 percent as monotherapy. Six of these ADCs are now in phase 3 trials comparing monotherapy against standard second-line chemotherapy, and the review suggests that ADC monotherapy may eventually challenge the zolbetuximab-chemotherapy combination in the first-line setting on both cost-effectiveness and quality-of-life grounds.

Bispecific antibodies and cell therapies are pushing the envelope further. Bispecific T-cell engagers such as emaretamig and givastomig bridge CLDN18.2 on tumor cells to either CD3 or 4-1BB on T cells, redirecting cytotoxic activity toward the cancer. Givastomig, engineered with a silenced Fc region, achieved a 16 percent single-agent response rate without triggering systemic cytokine release, whereas emaretamig and IBI389, which lack such engineering, produced response rates of 14 and 22 percent but caused cytokine release syndrome in 43 and 60 percent of patients respectively. On the cellular front, conventional CAR-T cells targeting CLDN18.2 have produced response rates of 30 to 55 percent, but next-generation designs are dramatically outperforming them: IMC002, built with a heavy-chain-only antibody variable domain, achieved 67 percent with an 11 percent complete response rate, while KD-496, which dually targets CLDN18.2 and NKG2D ligands, reached 75 percent in a small phase 1 cohort. A phase 3 trial of IMC002 is now underway.

FGFR2b, the second splice variant highlighted in the review, presents a different but equally compelling story. The FGFR2 gene produces two mutually exclusive isoforms—FGFR2b in epithelial cells and FGFR2c in mesenchymal cells—through alternative splicing of exons 8 and 9, a process controlled by the epithelial-specific RNA-binding proteins ESRP1 and ESRP2. FGFR2b is overexpressed or amplified in roughly 17 percent of G/GEJ cancers, and bemarituzumab, a humanized afucosylated IgG1 antibody against it, demonstrated a 16 percent single-agent response rate in early trials. In the phase 2 FIGHT study, bemarituzumab plus chemotherapy produced a 48 percent response rate versus 33 percent with chemotherapy alone. However, the phase 3 FORTITUDE-101 trial revealed a sobering complication: while interim analysis showed significant survival benefits, longer follow-up attenuated the clinical benefit, and ophthalmological toxicities—dry eye, keratitis and corneal epithelial defects—occurred in 32 percent of treated patients, reflecting FGFR2b’s physiological role in corneal epithelial maintenance.

Next-generation FGFR2b-targeted biologics are being explicitly engineered to solve this problem. Preclinical studies show that ADCs such as ALK201 and BG-C137, and bispecific antibodies pairing FGFR2b with EGFR or HER2, can spare corneal tissue by weakening FGFR2b signaling inhibition or by using epitopes that preserve intact FGF7/10-FGFR2b signaling in healthy cells. ALK201 has already achieved a 24 percent single-agent response rate with manageable safety in early-phase trials, including only grade 1 ocular adverse effects in 11 percent of patients. The review argues that bispecific ADCs combining FGFR2b with CLDN18.2, EGFR, HER2 or MET represent the logical next step, potentially allowing dual targeting of heterogeneous tumors while mitigating single-target toxicities.

Resistance remains the field’s central challenge. Tumors can evade CLDN18.2- or FGFR2b-targeted therapies by switching isoforms—transdifferentiating to produce other claudins or splicing FGFR2 toward the mesenchymal FGFR2c variant—through therapy-induced epithelial-mesenchymal transition. The immunosuppressive tumor microenvironment, populated by exhausted CD8-positive T cells, angiogenic endothelial cells and fibrotic cancer-associated fibroblasts, further blunts antibody and cell therapy efficacy. The review argues that companion diagnostics capable of detecting both the target antigen and its alternative isoforms in primary and metastatic lesions will be essential for monitoring epitope loss and predicting resistance. Combination strategies pairing splice variant-targeted biologics with immune checkpoint inhibitors, anti-angiogenic agents and cancer-associated fibroblast inhibitors are already in clinical trials, on the premise that remodeling the tumor microenvironment can convert it from an immunologically hostile to a therapeutically permissive state. As the field matures, the authors contend, comparative studies of ADC monotherapy versus antibody-chemotherapy combinations, alongside the clinical development of bispecific ADCs and allogeneic CAR-T cells, will determine which of these splice variant-targeted platforms delivers the greatest benefit to patients with gastric, gastroesophageal and other solid cancers.

Subject of Research: CLDN18 and FGFR2 splicing variants are emerging tumor-associated antigens for cancer therapy

Article Title: CLDN18 and FGFR2 splicing variants are emerging tumor-associated antigens for cancer therapy

Article References: Katoh, M., & Katoh, M. (2026). CLDN18 and FGFR2 splicing variants are emerging tumor-associated antigens for cancer therapy. Frontiers in Molecular Medicine, 6, Article 1927389. https://doi.org/10.3389/fmmed.2026.1927389

Image Credits: AI Generated

DOI: 10.3389/fmmed.2026.1927389

Keywords: cancer therapy, tumor-associated antigens, alternative splicing, CLDN18.2, FGFR2b, antibody-drug conjugates, CAR-T cells, gastric cancer, CLDN18, FGFR2, splicing, variants