Zynlonta’s 3-to-1 Death Ratio Is the Accelerated Approval System Grading Its Own Homework

On June 4, 2026, ADC Therapeutics’ stock fell 52% to approximately $1.32 per share. The market was reacting to data from the LOTIS-5 Phase 3 confirmatory trial for Zynlonta — a trial that revealed 13.2% fatal adverse events in the treatment arm versus 4.6% in the control arm across 440 patients. That is 27 patients dead in the Zynlonta-plus-rituximab group against nine in the rituximab, gemcitabine, and oxaliplatin arm. The trial met its primary endpoint. The patients did not.

What this actually reveals goes well beyond one drug and one company. The LOTIS-5 data is a live stress-test of whether the accelerated approval pathway — built on surrogate endpoints and a promise of rigorous post-market accountability — has any real teeth when the confirmatory data comes back wrong.

A Surrogate Promise, a Lethal Gap

The FDA granted Zynlonta accelerated approval on April 23, 2021, for adult patients with relapsed or refractory large B-cell lymphomaB-cell lymphoma after two or more lines of systemic therapy. The basis was a surrogate endpoint — response rate — not overall survival. That is how the pathway works: move fast on promising signal, require a confirmatory trial to verify clinical benefit, pull the drug if the confirmatory data fails. The system’s entire legitimacy rests on that last clause actually executing.

LOTIS-5 launched as that confirmatory vehicle. Zynlonta combined with rituximab was tested against a standard salvage chemotherapy backbone. When the trial met its primary endpoint — progression-free survival, reportedly — the press release language would have been enthusiastic. But buried beneath the efficacy headline was a threefold imbalance in fatal events that no progression-free survival result can paper over. The FDA’s own August 2025 draft guidance, “Approaches to Assessment of Overall Survival in Oncology Clinical Trials,” is unambiguous: sponsors must pre-specify thresholds to rule out harm using appropriate statistical methods, even when overall survival is not the primary endpoint. The question now sitting on the FDA’s desk is whether anyone pre-specified a harm threshold for LOTIS-5 — and if they did, what it was.

ADC Therapeutics has not disclosed the Data Safety Monitoring Board charter language for LOTIS-5. That silence is itself a data point.

When the System Could Have Stopped This

Zynlonta has been on the market for five years under an accelerated approval conditional on LOTIS-5’s results. Five years. The FDA’s December 2024 draft guidance on accelerated approval — “Expedited Program for Serious Conditions: Accelerated Approval of Drugs and Biologics” — explicitly requires that sponsors conduct post-approval confirmatory trials with adequate rigor and that the agency maintain active oversight of enrollment pace and interim safety data. Five years is a long time for the oversight machinery to sit quiet.

The operational timeline here is damning. Zynlonta launched commercially in 2021 while LOTIS-5 enrolled. Physicians prescribed it based on surrogate-endpoint data. Patients — many of whom had exhausted two or more prior lines of therapy — received it in routine clinical practice. All of this occurred while the confirmatory trial that would determine whether the drug actually extended life was still running. When LOTIS-5 closed and the fatal-event imbalance emerged, the commercial ship had already sailed for five years. That is not a hypothetical post-market risk. That is the structure of the accelerated approval pathway operating exactly as designed, and producing exactly the harm critics have warned about since the pathway’s expansion under the Omnibus Consolidated Appropriations Act of 2023.

The comparable case is instructive. Pepaxto (melphalan flufenamide) received accelerated approval in February 2021 for relapsed or refractory multiple myeloma based on surrogate endpoints from the HORIZON trial. Its confirmatory trial later showed increased mortality in the treatment arm. The FDA withdrew it. The sequence from accelerated approval to withdrawal took years — years during which patients received a drug that was shortening their lives. Zynlonta’s trajectory is following the same arc, except the company’s market cap has now been cut in half in a single trading session rather than through any proactive regulatory intervention.

That sequencing — market collapse before regulatory action — is the tell. The signal came from the stock market, not from a safety communication. The FDA has not, as of this writing, issued a clinical hold, a safety communication, or a public statement on LOTIS-5’s fatal-event imbalance. ADC Therapeutics’ shares are trading near penny-stock territory while the agency processes what a 13.2% versus 4.6% death rate means for the drug’s risk-benefit calculation. Investors have already decided. The FDA has not spoken.

The Structural Rot Underneath the Data

The counterintuitive read here is that LOTIS-5 meeting its primary endpoint is actually the most dangerous possible outcome — not because the efficacy signal is fraudulent, but because it creates a regulatory gray zone. A trial that both meets its PFS endpoint and shows a threefold fatal-event imbalance forces the FDA to make a judgment call the accelerated approval framework was never designed to cleanly adjudicate. The agency’s standard operating logic for confirmatory trials is binary: verify benefit or withdraw. LOTIS-5 delivered a third option — verify benefit, demonstrate harm simultaneously — and the existing guidance infrastructure does not have a clean answer for it.

This is not an ADC Therapeutics-specific failure. The antibody-drug conjugate class has been scaling rapidly, with BioNTech’s BNT326/YL202 also generating safety concern headlines in the same period. ADCs carry inherent bystander-toxicity risk from their payload chemistry — in Zynlonta’s case, a pyrrolobenzodiazepine (PBD) dimer linker-payload combination known for DNA cross-linking cytotoxicity. When that payload chemistry moves from a heavily pretreated single-arm study into a randomized combination regimen, the toxicity profile can shift in ways that single-arm Phase 2 data structurally cannot detect. The accelerated approval system, which rewards speed on early single-arm data, is not calibrated to catch that shift before commercial launch.

Sponsors running ADC programs with accelerated approval designations need to read LOTIS-5 as a protocol design warning, not a competitor’s misfortune. The FDA’s August 2025 OS guidance is now functionally a mandatory design input: any randomized confirmatory trial for an ADC must pre-specify a harm-ruling-out analysis with defined thresholds, power calculations for survival inferiority, and a DSMB charter that explicitly triggers a safety review when fatal-event rates diverge beyond a pre-defined margin. If that language is not already in your LOTIS equivalent, it needs to be in the next protocol amendment.

For FDA-watchers, the next critical signal is whether the agency issues a public safety communication or moves toward a clinical hold before ADC Therapeutics files its formal regulatory response to LOTIS-5. The December 2024 draft guidance on accelerated approval gives the FDA explicit authority to demand a withdrawal hearing when confirmatory data raises a serious safety question — not merely an efficacy question. A 13.2% versus 4.6% fatal-event rate is not a borderline efficacy shortfall. It is a safety signal that the agency’s own OS guidance says must be investigated on its own terms, independent of PFS. Watch for an FDA Type A meeting request from ADC Therapeutics in the next 60 days. How fast the agency convenes that meeting — and what it demands as preconditions for keeping Zynlonta on the market — will be the clearest test yet of whether the 2023 accelerated approval reforms produced accountability infrastructure or just accountability language.

References

1. FiercePharma — “ADC Therapeutics’ shares plummet after patient deaths in trial of Zynlonta”
2. Endpoints News — “ADC Therapeutics confirmatory data for Zynlonta muddied by death rates” (LOTIS-5 fatal event rates: 13.2% vs. 4.6%)
3. Oncology Practice Management — “FDA Approves Zynlonta for Relapsed or Refractory Large B-Cell Lymphoma” (April 23, 2021 accelerated approval)
4. Ropes & Gray — “FDA Issues Draft Guidance on Accelerated Approval Requirements” (December 2024)
5. Precision for Medicine — “FDA Draft Guidance on Overall Survival in Oncology Trials” (August 2025)
6. AgencyIQ — “Accelerated Removal: FDA Making Use of New Authority to Withdraw Ineffective Drugs” (Pepaxto/melphalan flufenamide withdrawal)
7. BioPharma Dive — “ADC Therapeutics sinks after deaths in Phase 3 trial” (BNT326/YL202 safety context)