CBX-12 turned a 10-patient signal into a randomized Phase 2 and lost it

What was tried

CBX-12 is a peptide-drug conjugate built on Cybrexa's alphalex platform. The vehicle is a pH-low insertion peptide, a sequence that stays unstructured in blood at physiological pH and folds into a transmembrane helix in the acidic extracellular environment characteristic of solid tumours, dragging its payload across the membrane. The payload is exatecan, a topoisomerase I inhibitor listed in ChEMBL as CHEMBL1614650 with a maximum phase of 3 and the development code DX-8951. ChEMBL holds no entry for the conjugate itself.

NCT06315491 was a randomized, open-label, parallel-group Phase 2 trial in platinum-resistant or refractory high-grade serous ovarian, fallopian tube or primary peritoneal cancer. Two arms compared CBX-12 at 125 mg/m2 and at 100 mg/m2 by intravenous infusion every 21 days, with treatment continuing until progression or unacceptable toxicity. The primary endpoint was confirmed objective response rate by RECIST v1.1. Eligibility required progression on platinum or within 26 weeks of the last platinum dose, and allowed up to two additional systemic regimens. Enrolment criteria contained no tumour acidity measure, no topoisomerase I expression threshold and no prior-exposure stratification. The trial ran from 25 September 2024 to 31 October 2025, enrolled 40 participants, and was terminated for limited clinical activity. No results have been posted.

The biological hypothesis

Antibody-drug conjugates in ovarian cancer depend on a surface antigen. Mirvetuximab soravtansine, the one approved agent in platinum-resistant disease, requires high folate receptor alpha expression, and the pivotal data that established it were generated in that selected population (and). That dependency leaves the antigen-low majority untreated.

The alphalex hypothesis was that extracellular acidity is a more general tumour property than any single antigen, and that a peptide responding to pH could therefore deliver a potent payload across tumour types without a companion diagnostic. Exatecan was a defensible payload. It traps the TOP1-DNA cleavage complex efficiently (doi:10.1158/1535-7163.MCT-21-1000) and is the payload in several later-generation conjugates, including a NaPi2b-directed ADC reported in 2025 (doi:10.1158/1535-7163.MCT-25-0254).

Open Targets scores the TOP1 and ovarian carcinoma association at 0.9034, with clinical evidence at 0.996, literature at 0.999 and somatic mutation at 0.946, and lists 26 distinct TOP1 drugs, seven approved, including four antibody-drug conjugates. The target is not in doubt. The question CBX-12 asked was whether pH-directed delivery produces a better therapeutic index than antigen-directed delivery or free camptothecin.

What actually happened

The Phase 1 dataset that justified Phase 2 was presented at ESMO in September 2024. In topoisomerase-naive ovarian cancer, 10 patients produced a 40 percent response rate consisting of one confirmed complete response, one confirmed partial response and two unconfirmed partial responses, and 8 of 12 patients showed clinical benefit. A parallel breast cohort of 7 topoisomerase-naive HR-positive, HER2-negative patients produced a 43 percent rate from two confirmed and one unconfirmed partial response. Dose-limiting toxicity was reversible myelosuppression. Across the Phase 1 population, anaemia occurred in 53.6 percent (24.6 percent grade 3 to 4), leukopenia in 42.0 percent and neutropenia in 40.6 percent. The company reported no interstitial lung disease and no ophthalmic toxicity, both of which limit approved conjugates.

The first Phase 2 patient was dosed on 7 October 2024. Thirteen months later the trial closed at 40 participants with the registry reason limited clinical activity. Cybrexa has released no Phase 2 response rate, duration of response or safety tabulation, and none appear on ClinicalTrials.gov as of 26 August 2026. What is verifiable is the gap between a 40 percent estimate from 10 unrandomized patients and a sponsor decision that 40 randomized patients did not justify continuing.

Failure mechanism, best guess

The arithmetic of the Phase 1 cohort is the most likely explanation. Two confirmed responses in 10 patients is a 20 percent confirmed rate. Counting unconfirmed responses doubles the headline figure without adding evidence, since unconfirmed responses are precisely those that failed to reproduce at the next scan. A 20 percent confirmed rate in a single-arm cohort of 10, in patients selected as topoisomerase-naive and enrolled at investigator discretion, is compatible with a true rate anywhere from the low single digits to the mid thirties. The Phase 2 population was broader, and prior topoisomerase exposure was not an exclusion.

The second candidate is that pH targeting delivered less differentiated exposure than expected. Both doses came from Phase 1, so the Phase 2 arms were not testing whether more drug would help. If tumour acidity in platinum-resistant ovarian disease is heterogeneous, an unselected population dilutes whatever advantage the mechanism has, and nothing in the protocol measured it.

These explanations are not mutually exclusive, and the absence of posted Phase 2 data means neither can be tested from public sources. That absence is itself the finding.

How to prevent this next time

No endpoint-level Phase 2 data exist, so the appropriate tools here are qualitative rather than a Bayesian update on a response rate that has not been released.

Report confirmed and unconfirmed responses separately when a Phase 1 estimate is used to size a Phase 2. The difference between 20 percent and 40 percent in this program is the difference between a signal worth a randomized trial and a signal worth another single-arm cohort.

Carry the enrichment factor forward. The Phase 1 ovarian cohort was topoisomerase-naive and the Phase 2 population was not restricted that way, which changes the population without changing the hypothesis.

Build a mechanism assay into the trial for a delivery platform. A pH-directed conjugate should carry a tumour acidity or intratumoural exposure readout, otherwise a negative result cannot distinguish a bad platform from a bad population.

The single highest leverage change would have been to require confirmed responses only in the Phase 1 estimate and to carry the topoisomerase-naive restriction into Phase 2, so the randomized trial tested the same population that generated the signal.

What this means for similar programs

TOP1 carries a Claidex Mechanism Risk Score of 39 (yellow). The score is dominated by saturation at 14.93, reflecting 27 distinct programs against the target, while the genetic deficit term is 1.45 because the target-disease association is strong. This is the signature of a crowded, well-validated target where differentiation comes from delivery rather than from biology.

For other tumour-microenvironment delivery platforms, the lesson concerns evidence standards rather than the physics. Acidity-based targeting remains untested in a randomized setting, because this trial produced no public efficacy estimate. Programs building on the same idea should expect to be asked for a delivery biomarker, and should assume an antigen-independent claim will be judged against mirvetuximab soravtansine in the selected population rather than against chemotherapy in an unselected one.

Open questions

What confirmed objective response rate did the 40 randomized participants achieve, and will Cybrexa post it?

Did the two dose arms differ, and if not, does that indicate exposure saturation or absence of effect at both levels?

How many Phase 2 participants had prior topoisomerase I exposure, and did the topoisomerase-naive subset behave like the Phase 1 cohort?

Was tumour acidity measured in any participant, and does any assay exist that could have selected for it prospectively?

Sources

1. ClinicalTrials.gov, NCT06315491, A Randomized Phase 2 Study of CBX 12 in Subjects With Platinum Resistant or Refractory Ovarian Cancer. Accessed 26 August 2026. https://clinicaltrials.gov/study/NCT06315491 Cybrexa Therapeutics, "Cybrexa Therapeutics Announces Positive Final Data at ESMO 2024 from Phase 1 Study of Peptide Drug Conjugate CBX-12 in Advanced Solid Tumors," 16 September 2024. https://www.cybrexa.com/press-release/cybrexa-therapeutics-announces-positive-final-data-at-esmo-2024-from-phase-1-study-of-peptide-drug-conjugate-cbx-12-in-advanced-solid-tumors/ Cybrexa Therapeutics, "Cybrexa Therapeutics Announces First Patient Dosed with First-in-Class Peptide-Drug Conjugate CBX-12 in Phase 2 Trial in Ovarian Cancer," 7 October 2024. Open Targets Platform, TOP1 (ENSG00000198900), association with ovarian carcinoma (MONDO_0005140), tractability, safety liabilities and known drugs. Accessed 26 August 2026. ChEMBL, EXATECAN, CHEMBL1614650. Accessed 26 August 2026. Moore KN et al. Mirvetuximab Soravtansine in FRalpha-Positive, Platinum-Resistant Ovarian Cancer. N Engl J Med, 2023.Matulonis UA et al. Efficacy and Safety of Mirvetuximab Soravtansine in Patients With Platinum-Resistant Ovarian Cancer With High Folate Receptor Alpha Expression. J Clin Oncol, 2023.Jo U et al. TOP1-DNA Trapping by Exatecan and Combination Therapy with ATR Inhibitor. Mol Cancer Ther, 2022.Hecht S et al. TUB-040, a Homogeneous and Hydrophilic NaPi2b-Targeting ADC with Stably Linked Exatecan. Mol Cancer Ther, 2025.openFDA FAERS drug event endpoint, queries for CBX-12, EXATECAN, TOPOTECAN and IRINOTECAN. Accessed 26 August 2026.