MRD Is Now the Endpoint That Matters: The EMN24 IsKia Trial Just Proved It

Picture a protocol statistician in 2019, sitting across from a regulatory reviewer at a Type B meeting, arguing that measurable residual disease negativity should serve as a primary endpoint in a newly diagnosed multiple myeloma trial. The reviewer’s posture is polite but skeptical. MRD is sensitive, yes — but is it a validated surrogate for survival? Does a 10⁻⁵ threshold translate to a regulatory decision? The meeting ends without consensus. The sponsor files an IND anchored to progression-free survival, like everyone else, and moves on.

That conversation is now obsolete. The EMN24 IsKia trial — a randomized phase 3 study published in Nature Medicine evaluating isatuximab plus carfilzomib, lenalidomide, and dexamethasone in newly diagnosed multiple myeloma — placed MRD negativity at the center of its primary endpoint architecture. And the results didn’t whisper. They declared. The quadruplet regimen produced an MRD negativity rate of 77.5% in the isatuximab arm versus 67.4% in the control arm — a difference that carries both statistical and clinical weight. The hazard ratio for progression-free survival landed at 0.50, meaning the isatuximab-containing arm cut the risk of progression or death in half.

But here is what the abstracts won’t tell you: the EMN24 trial isn’t just a myeloma story. It is a proof-of-concept for a new standard of endpoint ambition in oncology — and it arrives at a moment when the FDA’s position on MRD as a surrogate endpoint remains frustratingly incomplete.

The Surrogate Endpoint Trap

For decades, the field treated MRD negativity as a biomarker, not a primary endpoint. Sponsors ran it as an exploratory outcome, an academic footnote, something to present at ASH while the “real” efficacy data — overall survival, progression-free survival — carried the regulatory argument. The FDA’s 2019 guidance on hematologic malignancies acknowledged MRD’s potential but stopped short of endorsing it as a single-arm approval pathway for most contexts. The agency wanted correlation data. It wanted long-term follow-up. It wanted to see MRD negativity translate, consistently, into survival benefit across multiple datasets before it would commit.

That posture is understandable — but it created a structural problem. Sponsors designing trials around PFS endpoints were engineering for a finish line that takes years to reach. In a disease where therapeutic velocity matters — where a patient diagnosed today will cycle through several lines of therapy before overall survival becomes measurable — PFS-anchored trials impose a temporal penalty on the entire field. The EMN24 trial exposed this tension directly.

The trial enrolled 662 patients, randomized 1:1, across transplant-eligible and transplant-ineligible populations. That is not a small signal. That is a phase 3 dataset with enough statistical power to make MRD negativity at 10⁻⁵ sensitivity speak with authority. The 77.5% versus 67.4% MRD negativity rate difference — an absolute delta of 10.1 percentage points — held up under prespecified analysis. And when you layer on the HR of 0.50 for PFS, the surrogate and the clinical outcome are telling the same story in the same direction.

The counterintuitive read here is worth dwelling on. Most people assume that adding a primary MRD endpoint to a phase 3 trial is a concession — a way to get early readouts when survival data isn’t mature. The EMN24 trial inverts that assumption. MRD wasn’t a consolation endpoint. It was the signal that arrived first and loudest, and PFS confirmed it. The surrogate preceded the outcome, not because the outcome was unavailable, but because the biology said so first.

What Protocol Designers Are Getting Wrong

Open the FDA’s guidance on surrogate endpoints in oncology drug approval and you will find careful, calibrated language about “reasonably likely to predict” clinical benefit. That phrase has done enormous regulatory work over the years — it greenlit accelerated approvals, it structured confirmatory trial obligations, and it created the liminal space where MRD currently lives. The problem is that “reasonably likely to predict” was written for a regulatory context, not a protocol design context. Sponsors have been borrowing regulatory language to justify endpoint choices, when what they actually need is evidentiary language grounded in biology.

The EMN24 IsKia trial provides exactly that. Sanofi, which markets isatuximab as Sarclisa, sponsored the trial through the European Myeloma Network — a collaboration model that deserves its own examination. This was not a single-sponsor-driven pivotal trial optimized for the narrowest possible label. It was a cooperative group effort, which means the endpoint architecture was designed by investigators who had to defend MRD negativity as a primary endpoint to their own statistical committees, their own ethics boards, and ultimately to regulators across multiple jurisdictions. When a cooperative group runs a phase 3 trial with MRD as a primary endpoint and it reads out with an HR of 0.50, that is not a post-hoc rationalization. That is a pre-specified scientific argument that won.

Protocol designers at U.S. sponsors are watching this. They have to be. Because the operational implications are significant: an MRD-primary endpoint requires ultra-deep sequencing infrastructure at the site level, centralized laboratory logistics, and agreement on sensitivity thresholds before a single patient is enrolled. None of this is trivial. I’ve submitted these packages myself — the site qualification burden alone for centralized MRD testing adds weeks to startup timelines and introduces assay standardization risks that PFS endpoints simply do not carry.

But the alternative is worse. Sponsors who default to PFS-only primary endpoints in newly diagnosed myeloma trials in 2026 are building protocols optimized for a regulatory paradigm that is shifting beneath them.

The Endpoint Architecture the FDA Must Codify

The EMN24 trial did something operationally elegant that more sponsors should study: it co-primary-ed MRD negativity with PFS rather than subordinating one to the other. This is not a trivial distinction. A co-primary structure means neither endpoint is a backup. Both carry type I error allocation. Both require pre-specified sample size calculations. Both must read out affirmatively for the trial to fully succeed on its terms. This is rigorous endpoint design — and it forces the FDA to engage with MRD negativity as a regulatory anchor, not an exploratory curiosity.

The FDA’s Project Optimus initiative, which has pushed sponsors to rethink dose optimization in oncology, offers a parallel. When the agency announced Project Optimus in 2021, the industry response was initially defensive — sponsors worried about increased development timelines and additional dose-finding requirements. By 2023, the thinking had shifted. Sponsors began building dose optimization logic into phase 2 designs proactively, because they recognized that a phase 3 readout without dose justification was increasingly vulnerable at advisory committees. MRD is following the same arc. Right now, sponsors treat MRD as optional. Within three to five years, a phase 3 myeloma trial without a pre-specified MRD analysis will look like a dose optimization gap — a hole in the package.

The FDA needs to move faster than that arc. The agency’s guidance on MRD in hematologic malignancies, last meaningfully updated in the early 2020s, does not provide the granular endpoint construction guidance that the EMN24 trial’s design now demonstrates is achievable. Specifically, the agency should address three unresolved questions: what sensitivity threshold is acceptable for registration-grade MRD claims at the 10⁻⁵ versus 10⁻⁶ level; whether MRD negativity duration — not just point-in-time negativity — should factor into primary endpoint definitions; and how MRD endpoints should be handled in the context of treatment discontinuation. The EMN24 trial can’t answer these questions for the FDA. Only the FDA can.

The EMN24 IsKia trial’s 662-patient dataset, its HR of 0.50, and its 77.5% MRD negativity rate will be cited in every myeloma protocol written in the next decade. The question is whether the FDA will write guidance that lets sponsors build on that foundation — or whether the agency will continue to require sponsors to individually re-argue the evidentiary case for MRD, trial by trial, at the cost of years and billions of dollars that patients in newly diagnosed myeloma don’t have the luxury to wait for.

References

1. Nature Medicine — “Isatuximab, carfilzomib, lenalidomide and dexamethasone in newly diagnosed multiple myeloma: a randomized phase 3 trial” (2026)
2. FDA Guidance — “Hematologic Malignancies: Regulatory Considerations for Use of Minimal Residual Disease in Development of Drug and Biological Products for Treatment” (2020)
3. FDA Guidance — “Surrogate Endpoint Resources for Drug and Biologic Development” (updated 2023)
4. FDA — “Project Optimus: Reforming the Dose Optimization and Dose Selection Paradigm in Oncology” (2021)
5. FDA Guidance — “Oncology Therapeutics: Conventional and Novel Endpoints for Accelerated and Traditional Approval” (2019)