Tisotumab vedotin vs ADCE-T02: why only one of them causes nosebleeds
Tisotumab vedotin and ADCE-T02 both target tissue factor, yet their toxicity profiles could hardly be more different. Which design decisions produced that gap, and how much of the difference can we actually attribute with confidence?
Tisotumab vedotin and ADCE-T02 both target tissue factor (TF/CD142), which makes their divergent toxicity profiles an unusually controlled experiment in ADC design. We traced each of tisotumab vedotin's signature toxicities to a specific molecular decision, then assessed how completely ADCE-T02's redesign addresses each mechanism. Three design changes, four predicted consequences, and one genuinely unbounded new risk.
Who this is for: ADC discovery, translational and clinical development teams designing, in-licensing or benchmarking a next-generation conjugate.
- epistaxis, TV in patients vs ADCE-T02 in primates
- 30% → 0%
- epitope, payload and linker, redesigned at once
- 3 changes
- our confidence in the best prediction vs the ILD unknown
- 90% vs 25%
- trials, registries and literature, in one 8-hour run
- 18 sources
epistaxis, TV in patients vs ADCE-T02 in primates
epitope, payload and linker, redesigned at once
our confidence in the best prediction vs the ILD unknown
trials, registries and literature, in one 8-hour run
Key takeaways
- Tisotumab vedotin and ADCE-T02 both target tissue factor, which makes their toxicity gap a controlled experiment.
- Epistaxis is an epitope decision: HuMax-TF occupies the Factor X exosite and blocks coagulation wherever it binds.
- Poulsen et al. (2026) confirmed experimentally that Ab-T02 does not inhibit Factor X activation.
- Alopecia and neuropathy are MMAE payload-class liabilities; exatecan does not bind tubulin, so both mechanisms are unavailable.
- A hydrophilic linker supports a drug-to-antibody ratio near 4.5 without the usual pharmacokinetic penalty.
- Exatecan's membrane permeability kills antigen-negative neighbours, which heterogeneous tumours such as pancreatic cancer require.
- The redesign trades rather than eliminates: topoisomerase I payloads carry an interstitial lung disease risk tisotumab vedotin never had.
- ADCE-T02 has no clinical data. Every tolerability comparison here sets primate observations against a Phase 3 record.
Tissue factor is not a tumour-specific antigen
Drug development rarely produces a controlled comparison. Two ADCs against the same receptor, for the same patients, designed in the same era, differing in almost every downstream molecular decision, is about as close as the field gets.
Tissue factor is a physiologically essential receptor. It initiates the extrinsic clotting cascade, and it is constitutively expressed on corneal and conjunctival epithelium and on skin keratinocytes. Any TF-binding ADC will reach all of those tissues. The only question is what the payload does when it arrives.
Tisotumab vedotin was approved on real benefit: an overall survival hazard ratio near 0.70 against investigator's choice chemotherapy in recurrent cervical cancer. It also carries epistaxis in 30 to 69% of patients across trials, conjunctivitis and keratitis in 26 to 43%, and alopecia in 38 to 44%. Managing it requires a mandatory eye care plan and an ophthalmology pathway, which is a real access barrier in the low- and middle-income countries where most cervical cancer mortality occurs.
ADCE-T02 is a next-generation TF-targeting ADC in Phase 1 (Tiffany-01, NCT06597721). Its published preclinical package reports none of those toxicities in non-human primates. The interesting question is not whether it is better. It is which of its three design changes is doing the work, and how much of the improvement we can predict rather than hope for.
What teams in this space search for
- Why does tisotumab vedotin cause epistaxis?
- Is exatecan safer than MMAE in an ADC?
- Which ADC design change actually reduces toxicity?
How we solved it with Hydra
“Compare ADCE-T02 against the TF-ADC tisotumab vedotin. Both target TF, but ADCE-T02 appears not to suffer the same side effects, at least preclinically. Work out the most critical differences in ADC design between the two, and what they teach us about ADC development as a whole. Use the databases and published literature to reanalyse both antibodies where needed. Act as a senior researcher contextualising the learnings for the team: strengths, limitations, and intuitions in both the data and the biology.”
What Hydra ran
Retrieved the ADCE-T02 preclinical paper (Poulsen et al. 2026, PMID 41996633), the Tiffany-01 registry record (NCT06597721), the full innovaTV 201/204/205/301 trial series, and the TIVDAK FDA label (BLA 761208)
Built an annotated knowledge base across 18 primary sources with explicit gaps flagged, so unsupported claims could not propagate downstream
Mapped every major ADC design parameter side by side: antibody backbone and epitope, linker chemistry, payload class, drug-to-antibody ratio, and internalisation
Assembled the tissue factor expression landscape across 17 normal and tumour tissue types to identify which normal tissues each design would expose
Attributed each tisotumab vedotin adverse event to its causal molecular feature, then scored how completely the ADCE-T02 redesign interrupts that causal chain
Graded five hypotheses on a 0 to 100 confidence scale and wrote falsifiable confirm-if and refute-if criteria for each, with the patient numbers needed to decide
What it found
Epistaxis is an epitope decision. HuMax-TF binds the Factor X exosite on TF, so every dose also delivers an anticoagulant to the nasal submucosa, where TF's job is to seal constant small breaches. Poulsen et al. confirmed experimentally that Ab-T02 does not inhibit Factor X activation, which breaks the causal chain outright.
Alopecia and peripheral neuropathy are payload-class liabilities, not target liabilities. MMAE disrupts tubulin; hair follicle matrix cells and axonal transport both depend on it. Exatecan is a topoisomerase I inhibitor and does not bind tubulin, so those mechanisms are structurally unavailable.
The linker change is real but harder to attribute. A hydrophilic self-immolative linker reduces non-specific tissue partitioning and supports a drug-to-antibody ratio near 4.5 without the pharmacokinetic penalty that hydrophobic linkers incur above 4.
The payload switch also changes which tumours are reachable. MMAE has limited membrane permeability, so it works best in TF-homogeneous tumours such as cervical cancer. Exatecan diffuses into TF-negative neighbours, which is what heterogeneous tumours like pancreatic cancer require.
It is a trade, not an elimination. Removing the auristatin toxicity cluster introduces the topoisomerase I class risk of interstitial lung disease, which trastuzumab deruxtecan established at roughly 15% all-grade. Primate studies did not predict that for trastuzumab deruxtecan either.
What we learned
The epitope is a clinical decision, not just a binding decision. For any target with a physiological function, ask what the antibody does to that function before Phase 1, not after.
Payload class is the highest-leverage variable for tolerability. No amount of antibody or linker engineering removes auristatin neuropathy from an MMAE conjugate.
Bystander killing is an architectural requirement for heterogeneously expressed targets, not a bonus feature. A low-permeability payload leaves the antigen-negative fraction alive to repopulate the tumour.
A first-generation drug's toxicity profile is a mechanistic inventory for the second generation. Tisotumab vedotin's label is a readable list of what each component does when it reaches normal tissue.
The honest conclusion is conditional. Three predictions are near-certain, one is moderate, and the ILD question cannot be answered from preclinical data at all. Publishing that spread is the point, not a weakness in it.
ADCE-T02 vs Tisotumab Vedotin: ADC Design Lessons from a Shared Target
The full 22-page analysis: five design pillars compared parameter by parameter, five hypotheses with confidence scores, falsifiable Phase 1 confirmation criteria, and the complete reference set.
Run this analysis on your question
Hydra plans, executes, and validates, so you reach a defensible answer in hours, not weeks.
What you get
- Each tisotumab vedotin toxicity traced to the specific molecular feature that causes it
- Five hypotheses graded 25 to 90 on confidence, with the reasoning behind each score
- Falsifiable confirm-if and refute-if criteria for Phase 1, with minimum patient numbers
- Five generalizable ADC design principles that transfer to any target expressed on normal tissue
- The one unbounded risk named explicitly rather than buried in a limitations paragraph
Glossary
| Term | What it means |
|---|---|
| Tissue factor (TF, CD142) | A 47-kDa transmembrane receptor that initiates the extrinsic clotting cascade and is overexpressed in several solid tumours |
| ADC | Antibody-drug conjugate: a monoclonal antibody joined by a linker to a cytotoxic payload |
| Epitope | The precise surface a given antibody binds, which decides what it does to the target's normal function |
| Factor X exosite | The region of TF where Factor X docks before FVIIa activates it; occupying it blocks coagulation |
| MMAE | Monomethyl auristatin E: a tubulin-disrupting payload that arrests dividing cells at G2/M |
| Exatecan | A camptothecin analogue that inhibits topoisomerase I, active across the cell cycle and highly membrane-permeable |
| DAR | Drug-to-antibody ratio: how many payload molecules each antibody carries |
| Bystander killing | Payload diffusing from a killed antigen-positive cell into antigen-negative neighbours |
| Hydrophilic linker | Linker chemistry that limits aggregation and non-specific tissue uptake, permitting a higher DAR |
| ILD | Interstitial lung disease: the characteristic class risk of topoisomerase I payloads |
| NHP | Non-human primate: the preclinical species used for ADC tolerability, reliable for some endpoints and not for ILD |
| innovaTV 204 | The pivotal single-arm Phase 2 trial (n=101) behind tisotumab vedotin's toxicity numbers |
| Tiffany-01 | NCT06597721, the ongoing Phase 1 dose-escalation trial of ADCE-T02 |
Sources & methods
- 01Poulsen TT, et al. ADCE-T02, a next generation ADC targeting tissue factor shows superior preclinical efficacy and tolerability compared to tisotumab vedotin. Mol Cancer Ther, 2026. doi:10.1158/1535-7163.MCT-25-0632 (PMID 41996633) Link
- 02Coleman RL, Lorusso D, Gennigens C, et al. Efficacy and safety of tisotumab vedotin in previously treated recurrent or metastatic cervical cancer (innovaTV 204). Lancet Oncol, 22(5):609-619, 2021. doi:10.1016/S1470-2045(21)00056-5 Link
- 03Vergote I, Gonzalez-Martin A, Lorusso D, et al. Tisotumab vedotin as second- or third-line therapy in recurrent cervical cancer. N Engl J Med, 391(1):44-55, 2024. doi:10.1056/NEJMoa2309945 Link
- 04de Bono JS, Concin N, Hong DS, et al. Tisotumab vedotin in patients with advanced or metastatic solid tumours (innovaTV 201). Lancet Oncol, 20(3):383-393, 2019. doi:10.1016/S1470-2045(18)30859-3 Link
- 05Tsumura R, et al. Anti-tissue factor antibody-drug conjugates conjugated with MMAE or DXd in pancreatic cancer models: influence of TF expression heterogeneity. Cancer Sci, 115(11):3675-3687, 2024. doi:10.1111/cas.16322 Link
- 06Ogitani Y, Aida T, Hagihara K, et al. DS-8201a, a novel HER2-targeting ADC with a novel DNA topoisomerase I inhibitor. Clin Cancer Res, 22(20):5097-5108, 2016. doi:10.1158/1078-0432.CCR-15-2822 Link
- 07Modi S, Jacot W, Yamashita T, et al. Trastuzumab deruxtecan in previously treated HER2-low breast cancer. N Engl J Med, 387(1):9-20, 2022. doi:10.1056/NEJMoa2203690 Link
- 08Marshall RF, et al. Ocular toxicities associated with antibody-drug conjugates. Curr Opin Ophthalmol, 35(5):405-415, 2024. doi:10.1097/ICU.0000000000001048 Link
- 09Huang R, Shih HA, Murakami MA, et al. Antibody-drug conjugates: hydrophilicity and its impact on DAR, aggregation, and pharmacokinetics. Mol Pharm, 17(7):2518-2527, 2020. doi:10.1021/acs.molpharmaceut.0c00258 Link
- 10Ahmadi SE, Rahimian E, Rahimi S, et al. Tissue factor: a potential double-edge molecule in cancer biology. Biomark Res, 11(1):60, 2023. doi:10.1186/s40364-023-00498-5 Link
- 11U.S. Food and Drug Administration. TIVDAK (tisotumab vedotin-tftv) prescribing information, BLA 761208, 2024. Link
Figures reflect analyses PharosBio ran on public datasets and public benchmarks; the methods and results shown are real and repointable to your own target.
Frequently asked questions
Why does tisotumab vedotin cause nosebleeds?
Its antibody, HuMax-TF, binds tissue factor at an epitope overlapping the Factor X binding exosite. That blocks formation of the TF-FVIIa-FX complex, so every dose also acts as an anticoagulant at mucosal surfaces. The nasal submucosa is TF-dense and constantly breaching, which is why epistaxis is the most common adverse event.
Is exatecan safer than MMAE in an antibody-drug conjugate?
It is a different toxicity profile rather than a safer one. Exatecan does not bind tubulin, so alopecia and auristatin-class neuropathy are structurally unavailable. In exchange it carries the topoisomerase I class risk of interstitial lung disease, which trastuzumab deruxtecan established at roughly 15% all-grade.
Does ADCE-T02 have clinical data?
No. As of this analysis it has published preclinical data and an ongoing Phase 1 trial (Tiffany-01, NCT06597721) in dose escalation. Every tolerability comparison here sets primate preclinical observations against tisotumab vedotin's Phase 3 clinical record, an asymmetry that inherently favours the newer molecule.
Which ADC design change reduces toxicity the most?
Payload class. No amount of antibody or linker engineering removes auristatin-class neuropathy from an MMAE conjugate; only changing the payload does. The epitope choice is a close second when the target has a physiological function the antibody can block, as tissue factor does with coagulation.
What is the biggest unknown for ADCE-T02?
Interstitial lung disease. Exatecan shares the camptothecin scaffold with DXd, and primate studies are known to be poor predictors of this endpoint: trastuzumab deruxtecan's preclinical package did not flag it either. The plausible range runs from near-zero to trastuzumab-deruxtecan-comparable, and preclinical data cannot narrow it.