SGN-MesoC2 and the mesothelin question: Pfizer shelves a TOP1 ADC in a target class that keeps disappointing

What was tried

Seagen, now part of Pfizer, opened a Phase 1 study of SGN-MesoC2 (PF-08052666), an antibody-drug conjugate that targets mesothelin. The trial was an open-label, single-arm, dose-escalation in advanced solid tumors, with non-small cell lung cancer, ovarian, pancreatic, colorectal, endometrial cancers and mesothelioma all eligible. The primary measures were adverse events, laboratory abnormalities, dose modifications and dose-limiting toxicities, which is the standard shape of a first-in-human safety run. The study enrolled 19 patients and was terminated. The registry reason was direct: "The trial was terminated for strategic reasons. The decision was not based on any safety concerns" (ClinicalTrials.gov).

This is not a molecule that failed on its own data. It is a molecule that was stopped before it could generate much. The question worth asking is why a large sponsor would pull a fresh asset this early, and the answer sits in the target.

The biological hypothesis

SGN-MesoC2 is built from a humanized IgG1 anti-mesothelin antibody, a cleavable tripeptide linker, and a topoisomerase 1 inhibitor payload, which makes it a first-in-class TOP1 ADC against this antigen (AACR 2025 preclinical report). Mesothelin is a GPI-anchored cell-surface glycoprotein. High expression in adults is restricted to the pericardium, pleura, peritoneum and tunica vaginalis, and it has no clear normal function, while it is overexpressed in mesothelioma, serous ovarian cancer, pancreatic cancer and a fraction of lung and gastric tumors (Chu 2023). On paper that profile is what an ADC wants: a surface antigen, broad tumor coverage, limited normal expression.

Open Targets rates the target as modestly associated and heavily pursued. The MSLN association is 0.141 for mesothelioma, 0.138 for the neoplasm basket, 0.123 for ovarian carcinoma and 0.115 for non-small cell lung cancer, and the antibody tractability bucket reads Advanced Clinical (Open Targets, MSLN). The evidence is clinical and literature based rather than genetic, which matches a target chosen for expression rather than for a causal driver role.

What actually happened

The trial stopped at 19 patients with no posted efficacy and no safety trigger (ClinicalTrials.gov). To read a strategic termination you look at the class it joined. Mesothelin has absorbed roughly two decades of antibody, ADC, immunotoxin, bispecific and cell-therapy work, and the therapeutic benefit has stayed modest (Chu 2023). Anetumab ravtansine, a maytansinoid ADC against the same antigen, advanced furthest and did not establish a registration path in mesothelioma or pancreatic cancer (Mauricio 2021). A Genentech program showed why the antigen is treacherous: tumor growth inhibition by a mesothelin ADC tracked with actual antibody uptake on zirconium-89 immunoPET, and several mesothelin-positive xenografts took up little antibody and did not respond despite confirmed expression (Terwisscha van Scheltinga 2016).

Failure mechanism, best guess

The strategic label is honest, and the mechanism behind such decisions is the target, not this antibody. Three properties make mesothelin a hard ADC target. First, the normal serosal expression sets a narrow therapeutic window, since payload delivered to pleura and peritoneum produces on-target, off-tumor toxicity. Second, mesothelin is shed into the circulation, and soluble antigen acts as a sink that intercepts the conjugate before it reaches tumor. Third, expression does not equal delivery. The immunoPET work shows that a mesothelin-positive tumor can still fail to internalize enough antibody to matter (Terwisscha van Scheltinga 2016). A TOP1 payload with a cleavable linker addresses one variable, payload potency and bystander effect, while leaving the antigen-sink and delivery problems untouched. Against that backdrop, and inside a large post-merger portfolio, a first-in-human asset with no differentiated early signal is a rational thing to cut.

How to prevent this next time

Two quantitative tools and one structural one.

Set the prior with the base rate. The cumulative likelihood of approval from Phase 1 in oncology is about 3.4 percent, the lowest of any major therapeutic area (Wong, Siah, Lo 2019). For a fourth or fifth ADC against a target with a long record of modest results, the differentiated prior is lower still, and that number belongs in the go decision before the first patient.

Gate escalation on a Bayesian predictive probability that folds in delivery biomarkers. Mesothelin offers two: soluble mesothelin in serum as a sink marker, and immunoPET or membrane immunohistochemistry as a delivery marker. Enrolling by delivery rather than by expression raises the responder fraction and lowers the number needed to screen. The interim predictive probability of success makes the stop or go explicit:

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

The structural tool is a competitive landscape red-team run before the program starts, which asks what this molecule does that anetumab ravtansine and the auristatin conjugates did not, and refuses to advance without a concrete answer.

The single highest leverage change would have been to enroll by mesothelin delivery, using immunoPET or membrane staining plus soluble-antigen screening, rather than by expression alone.

What this means for similar programs

Mesothelin is not undruggable, but it punishes programs that select on expression and ignore delivery. The targets that reward ADCs combine high tumor expression, low normal expression, efficient internalization and little shed antigen. Mesothelin fails the last two. New entrants should either solve the antigen sink, for example by dosing strategies that saturate soluble mesothelin, or move to formats less sensitive to it, and they should carry a delivery biomarker from first-in-human. Payload innovation alone, even a clean TOP1 conjugate, does not fix a delivery problem.

Open questions

Did SGN-MesoC2 show any objective responses in its 19 patients before the program stopped. How much did soluble mesothelin vary across the enrolled tumor types, and did it predict exposure. Would a delivery-enriched cohort, selected by immunoPET, change the readout for any mesothelin ADC. Is there a tumor context, such as a serous ovarian subset with high membrane mesothelin and low shedding, where a TOP1 payload finally clears the bar.

Sources

1. ClinicalTrials.gov, A Study of SGN-MesoC2 in Advanced Solid Tumors, NCT06466187, 2026, U.S. National Library of Medicine, https://clinicaltrials.gov/study/NCT06466187.

2. Chu Q, Targeting Mesothelin in Solid Tumours, Anti-mesothelin Antibody and Drug Conjugates, 2023, Current Oncology Reports,.

3. Terwisscha van Scheltinga AGT, Ogasawara A, Pacheco G, et al., Preclinical Efficacy of an Antibody-Drug Conjugate Targeting Mesothelin Correlates with Quantitative 89Zr-ImmunoPET, 2016, Molecular Cancer Therapeutics,.

4. Mauricio D, Harold J, Tymon-Rosario JR, et al., Novel Mesothelin Antibody-Drug Conjugates, Current Evidence and Future Role in the Treatment of Ovarian Cancer, 2021, Expert Opinion on Biological Therapy,.

5. Wong CH, Siah KW, Lo AW, Estimation of Clinical Trial Success Rates and Related Parameters, 2019, Biostatistics,.

6. Open Targets Platform, MSLN (ENSG00000102854) Target-Disease Associations, 2026, https://platform.opentargets.org/target/ENSG00000102854.

7. AACR Annual Meeting 2025, Abstract 324, PF-08052666 (SGN-MesoC2, HBM9033), a First-in-Class Topoisomerase 1 Inhibitor-Based ADC Targeting MSLN, 2025, Cancer Research, https://aacrjournals.org/cancerres/article/85/8_Supplement_1/324/754967.