Same Evidence, Different Verdict

On the morning of 8 May 2026, shares of Atara Biotherapeutics opened more than 37% higher. Nothing about its cell therapy had changed overnight. The trial was the same trial. The dataset was the same dataset the FDA had recently called insufficient. What had changed was that, one week earlier, the director of the agency office that had rejected it had left the building.

That move turned out to be the first clear marker of a pattern that ran through the next six weeks. Between early May and late June, the FDA reversed course on at least five rare-disease therapies it had previously turned away. In three of them, the agency went on to accept the exact evidence it had recently judged inadequate. No company ran a new pivotal trial. No molecule improved. The people doing the reviewing changed, and the answers changed with them.

For anyone building, funding, or staffing a programme in advanced therapeutics, this sequence is worth sitting with. It says something direct about where value actually sits in this market, and about how fast that value can move when none of the underlying science does.

A Cascade, Not a Coincidence

The Atara reversal did not arrive in isolation. It opened a run.

Ebvallo, the allogeneic T-cell therapy from Atara Biotherapeutics and Pierre Fabre for Epstein-Barr virus-positive post-transplant lymphoproliferative disease, had been rejected twice. The sticking point was the single-arm ALLELE study, which the agency had at one point accepted and later called inadequate to support approval. In a Type A meeting held in late April, the FDA agreed that the same study, with an appropriate historical control, could in fact support a resubmission. The companies announced the turn on 7 May, roughly a week after the departure of the official who had overseen the rejection. Atara opened up more than 37% the next morning.

Within the same window, Replimune said it would resubmit RP1, its oncolytic immunotherapy for advanced melanoma, for a third time after reaching agreement with the agency. The programme had been rejected in July 2025 and again in April 2026. The resubmission came roughly two weeks after the FDA commissioner was pushed out. Around the same time, Disc Medicine reached agreement with the FDA to resubmit a new drug application for a recently rejected rare blood disorder programme.

Then, in the space of a single week in June, the agency reversed on two gene therapies in rare disease.

On 17 June, uniQure announced that the FDA had agreed its three-year Phase 1/2 dataset for AMT-130 in Huntington's disease could serve as the primary basis for a biologics license application under accelerated approval. Only weeks earlier the agency had taken a harder line, questioning the external-control data and pointing the company toward a new controlled study, a stance that had the company weighing a parallel route through the United Kingdom. The reversal removed the need for a sham-controlled confirmatory trial, with the FDA agreeing to a confirmatory study using a standard-of-care control instead. uniQure plans to file in the third quarter of 2026.

Five days later, on 22 June, REGENXBIO said the FDA had reversed its February rejection of Navsunli, its gene therapy for Hunter syndrome, and now considered the existing long-term data sufficient to support an accelerated approval resubmission. The agency dropped its earlier demand for additional patients and a new untreated control arm. The stock moved into double digits on the day, rising as much as 17% in morning trading. REGENXBIO also targets a third-quarter resubmission.

Five programmes. Six weeks. One common thread that has nothing to do with the molecules.

The Reversal Ledger

The detail that matters most sits in the last two columns. In the cases of Ebvallo, AMT-130, and Navsunli, the FDA did not respond to new evidence. It re-read evidence it already held. The same single-arm study, the same three-year dataset, the same long-term biomarker package: previously insufficient, now acceptable.

The agency re-read what it already had, and the market priced the result as a regulatory event, in hours, twice.

What Actually Changed at the FDA

To understand the cluster, follow the leadership rather than the science.

For most of the past decade, the FDA's Center for Biologics Evaluation and Research, the division that regulates gene and cell therapies, was run by Peter Marks. Marks was associated with regulatory flexibility and oversaw the review of dozens of cell and gene therapies. He was forced out in March 2025.

His successor, Vinay Prasad, arrived with a reputation for evidentiary stringency and a sceptical view of surrogate endpoints and externally-controlled trials. Under his direction, CBER raised the bar. Multiple developers accused the agency of reversing prior agreements on trial design, and a series of rare-disease programmes that the field had expected to reach the market were turned away. Prasad left the role briefly in mid-2025, returned, and then departed for a second time at the end of April 2026.

Commissioner Marty Makary resigned on 12 May. In the days around his exit, the agency replaced the acting heads of both its drug and biologics centres. Karim Mikhail, a former pharmaceutical executive who spent two decades at Merck and later led Amarin, was named acting director of CBER. By one count he is the fourth acting director and the sixth leader of that center since the start of 2025.

The new leadership has signalled a different posture. In late June, acting CBER and CDER leaders appeared alongside an industry trade body and struck a conciliatory note on rare disease and on rebuilding trust with developers. The reversals are the practical expression of that shift.

This is the uncomfortable part. The science behind AMT-130 and Navsunli did not change between the rejection and the reversal. The agency's reading of it did. A move of 17% in a company's market value, achieved in a single morning, traced to a change in personnel rather than a change in data.

The Real Shift Is About Evidence

It would be easy to read this as politics and leave it there. The more useful read is technical.

The thread connecting these reversals is the agency's willingness to accept externally-controlled trials in ultra-rare disease, where running a traditional randomized or placebo-controlled study is slow, expensive, and in some settings ethically difficult. An external control draws on natural-history data or a matched historical cohort in place of a concurrent placebo arm. Under the prior regime the agency leaned away from these designs and toward conventional controlled trials. The current leadership is leaning back toward them.

For AMT-130, that meant accepting a 75% slowing of disease progression on the composite Unified Huntington's Disease Rating Scale at three years, measured against propensity-matched external controls drawn from a natural-history database, as the primary evidentiary basis for a filing. For Navsunli, it meant accepting long-term biomarker and clinical data in place of a fresh untreated control arm.

This is the substance beneath the headlines. The reversals are not a change of mood. They are a return to a particular evidentiary standard, one that materially changes which programmes are filable, how trials get designed, and how long a path to market takes. A company that can build its evidence package around an external control, and persuade the agency to accept it, operates on a different timeline from one preparing to run a new controlled study.

The Interpretation Layer

Here is the conclusion that should concentrate the mind of anyone allocating capital or talent in this space.

In advanced modalities, a growing share of enterprise value no longer sits in the molecule, or even in the data. It sits in how the agency reads that data in a given quarter. Call it the interpretation layer: the set of judgements, made by a small number of people, about whether a given evidence package clears the bar.

The events of May and June put a number on it. When the interpretation layer moved, Atara moved 37% and REGENXBIO moved into the high teens, in single mornings, with no change to their underlying programmes. That is the interpretation layer being repriced in real time.

For most of the history of drug development, this layer was treated as relatively stable, a fixed feature of the operating environment. It is not stable now. It is one of the most volatile inputs a programme has, and increasingly it is the one that decides whether a therapy reaches patients on a two-year horizon or a five-year one.

Discretion Cuts Both Ways

There is a temptation, in a thaw, to treat the new warmth as permanent. That would be a mistake.

The same discretion that reopened these doors in June is the discretion that closed them earlier in the year. The agency did not change its rules through guidance or rulemaking. It changed its reading, meeting by meeting, as the people in the chairs changed. A regime that can reverse a rejection can reverse a reversal.

The instability is structural rather than incidental. A center on its sixth leader in eighteen months does not give developers a predictable counterparty. The current posture is favourable to rare-disease and gene-therapy sponsors. The next one may not be, and a permanent commissioner has yet to be confirmed. Building or investing as though the window will stay open is a way to misprice the risk.

A favourable regulatory posture is closer to weather than to infrastructure. It can turn, and over the past six months it has turned more than once.

Who Reads the Weather Next

The pattern already has a pipeline. A number of rare-disease developers are reading the uniQure and REGENXBIO reversals as precedent and positioning accordingly. Companies with externally-controlled or natural-history data sets that were previously viewed as marginal now have reason to re-engage. Among those most often named in this context are sponsors with mature long-term data in Huntington's disease, in spinocerebellar ataxia, in Dravet syndrome, and in cardiomyopathy, several of whom held controlled or natural-history packages while the prior regime held firm.

Whether each of them benefits will turn on the specifics of their data and their indication. But the direction of travel is set, at least for now. The agency is more willing to accept the kind of evidence that ultra-rare programmes can realistically generate, and developers are moving to take advantage of it before the weather changes again.

What This Means for How You Build

Strip the story back and a clear operational lesson remains.

You cannot rerun a pivotal trial once it has read out. What you can shape is whether someone in your organisation read the direction of the agency, kept the evidence package ready, and moved the moment the window opened. The companies that turned a reversal into a third-quarter filing did not do so by luck. They did it because they had the regulatory and clinical strategy in place to act inside weeks rather than quarters.

That capability is not something a balance sheet can hold. It lives with a small number of people who read the agency for a living, who understand where the evidentiary bar is moving, and who can build and defend an externally-controlled package under questioning. In a market where the interpretation layer is this volatile, that talent is not a support function. It is one of the most direct determinants of whether a programme reaches patients, and when.

The molecule can be protected. The data can be generated. The capacity can be funded. The judgement to read a moving agency and act on it before the next turn is none of those things, and it is in short supply.

For teams building in gene and cell therapy, that is the seat to fill before the next filing, not after.

Sources and Method

This analysis draws on company disclosures and contemporaneous reporting from May and June 2026, including statements from uniQure, REGENXBIO, Atara Biotherapeutics and Pierre Fabre, Replimune, and Disc Medicine, alongside coverage from BioSpace, BioPharma Dive, STAT, Endpoints, and Reuters, and the FDA's own leadership disclosures. Clinical figures for AMT-130 reflect uniQure's reported three-year Phase 1/2 results, with a data cutoff of 30 June 2025. Market movements reflect intraday trading on the day of each announcement. Where the precise date of an earlier rejection was not disclosed, the timing is described in general terms.

ProGen Search publishes intelligence on the people and infrastructure behind advanced therapeutics, and runs retained search for the operators who build them. If you are standing up the regulatory and technical bench behind a gene or cell therapy programme, it is worth mapping properly.