FDA says CMC readiness is the bottleneck. The pilot built for it is running out of road.

Last week I wrote that the expedited lane is now the default, and that the compression lands on the drafting. I want to correct the emphasis. I was writing about the clinical documents — the study reports, the summaries, the benefit-risk argument. That was the wrong place to point.

On 23 July FDA published its Strategy Document on Facilitating Chemistry, Manufacturing, and Controls Readiness for Products With Accelerated Clinical Development (91 FR 46443, Docket FDA-2026-N-7232). It sets out what the agency has done and what it plans for FY2026–2027, under the PDUFA VII commitment to advance its capability to facilitate CMC development for products aimed at serious conditions with unmet need. The notice says the actions come from lessons learned on submissions that ran on accelerated timelines.

Read that plainly. The agency looked at what happens when programmes go fast, and the thing it wrote a strategy document about was not the clinical narrative. It was CMC.

The part that makes it interesting

The same week, Endpoints reported that FDA will stop accepting new applications to the CMC Development and Readiness Pilot at the end of April 2027, while INDs already accepted remain in the programme — a date RAPS also reports. The reason Endpoints gave was participation: few sponsors took the programme up across its three-year run, against a ceiling of nine proposals a year and a start date of April 2023.

So within days: here is our strategy for CMC readiness on accelerated programmes, and also, we are winding down the programme we built for CMC readiness on accelerated programmes because not enough of you used it.

I don't think that's incoherence at the agency. I think it's a fairly precise signal about where the problem actually lives, and it isn't at FDA's end.

CDRP was not a demanding thing to join. An active commercial IND, usually before end of Phase 2, an expedited designation, and a CMC development strategy that matched the clinical timeline. What you got was two predesignated CMC-focused Type B meetings, plus follow-up discussions when clarifications were needed — dedicated agency time on manufacturing, early, while you could still act on it. When RAPS covered the industry feedback, the BIO survey reasons for staying out were administrative burden, confidentiality and disclosure worries, and not being sure what CDRP gave you that other expedited programmes didn't. Intellia's Isabella Palazzolo asked for clearer structure, and made the sharper point that frequent focused meetings would beat two broad ones.

Look at the application numbers from the September 2025 workshop: CDER took 12 applications and accepted 5, CBER took 7 and accepted 4. Nineteen applications. In a year. Across the entire US industry, for free agency time on the thing that FDA has now formally identified as the constraint. And the main reason for turning applicants away was failure to meet the minimum eligibility criteria — which suggests that some of the nineteen either fell outside the formal bar or could not present a CMC development strategy aligned closely enough to the accelerated clinical timeline.

What the rejection letters actually say

This is where it connects to something I wrote on 15 July. I said there were 455 rejection letters sitting in the public database and almost nobody drafting submissions was reading them. Here is what you find when you do.

Dilek, Woods, Ballreich, Moore and Alexander at Johns Hopkins analysed the publicly available CRLs for 43 novel therapeutics ultimately approved between 2020 and 2024 (Therapeutic Innovation & Regulatory Science, 2026;60(3):837–846). Manufacturing deficiencies were the most common category by a distance: 65% of letters cited facilities and 51% cited CMC. Labeling came next at 44%. Efficacy and safety were 26% each. PK/PD was 14%.

The remedies FDA asked for run the same way. Facility remediation was required in 65% of cases and additional CMC submissions in 51% — against new clinical trials in 26%.

Jefferies ran a separate count across roughly 350 published letters and found the same direction of travel: more than half cited manufacturing concerns, 41% cited product quality such as impurities or failed stability, and 27% cited insufficient clinical data (BioSpace, 11 May 2026).

Different methods, same signal. Many applications that fail are not failing only on clinical efficacy. They are failing on the part of the dossier that often becomes visible too late: manufacturing readiness, facility readiness, product quality, and the CMC evidence underneath all three.

The cost of getting it wrong is in the same paper. Median time from receiving a CRL to resubmission was 0.60 years; median time from resubmission to approval was another 1.28 years.

I should be straight about where I'm standing. I have not run a CMC programme myself. My work sits on the clinical side of the dossier, and what follows is inference from the public record rather than from inside a manufacturing organisation. If you have run one, you will know things about the next few paragraphs that I don't, and I'd rather hear them than not.

Why I think sponsors stay away

Here's my actual opinion, and I'd like to be argued with on it.

Sponsors didn't skip CDRP because of administrative burden. Nineteen applications a year is not a burden problem — a burden problem looks like two hundred applications and complaints about the forms. This looks like something else. I think most sponsors on an expedited designation genuinely do not know, at end of Phase 2, what their CMC gaps are. You cannot apply for help closing a gap you haven't mapped. The eligibility bar asked for a CMC development strategy aligned to the clinical timeline, and a meaningful number of applicants couldn't clear it.

That is a documentation problem before it is a manufacturing problem. The strategy exists in fragments — in a process development report, in a tech transfer deck, in someone's head at the CDMO — and nobody has written it down as one argument with dates against it. The clinical side gets a designation and the timeline halves. The CMC story doesn't get compressed. It gets discovered late.

Which is why I think the writing question here is not "can we produce Module 3 faster." It's whether anyone can state, in one place and on demand, what the CMC development strategy is and where it is thin. If you can't produce that document, you couldn't have joined CDRP, and you also can't tell whether you're about to spend a year on a CRL response.

What I'd do with this

The Federal Register notice is only two pages, pointing to a longer FDA strategy document, and it is worth reading the underlying document properly rather than taking my summary for it — particularly the FY2026–27 actions, because those are what the agency will be measured against under PDUFA VII.

If you have a programme on an expedited designation, CDRP still accepts applications until end of April 2027. That window is now finite and it was always underused. Going in requires you to write down the CMC strategy against the clinical timeline, which is the exercise, whether or not you get a slot.

I build tooling in this space, so discount accordingly. But the thing I'd want if I were running a programme has nothing to do with a tool. It's the one document nobody owns: what our manufacturing story is, what evidence supports it, and which parts of it we are currently hoping don't get asked about.