The CMC chasm. Why most biotech IND submissions fail at chemistry, not biology.

The biology is usually right. The science deck looks clean. The data tells a real story. And the IND still gets a clinical hold or a refuse-to-file letter. Almost every time, the reason is the same. The Chemistry, Manufacturing, and Controls section was treated as paperwork instead of a program.

CMC is roughly 40 percent of the IND by page count and roughly 80 percent of the reasons FDA pushes back. If you are running a biotech program and you have not built the CMC story as carefully as you built the science story, you are about to find that out in writing from CBER or CDER.

What CMC actually covers

The CMC section under 21 CFR 312.23(a)(7) is a three-part claim. Chemistry: what the product is, characterized in measurable terms. Manufacturing: how it is made, by whom, under what controls. Controls: how you know each batch is the same as the last one and safe to dose.

For a small molecule, CMC means impurity profile, polymorph identification, stability under defined storage. For a biologic, it means host cell protein levels, aggregation state, glycosylation patterns, potency assays that actually correlate with activity. For an exosome or cell therapy, it means donor screening, surface marker identity, cargo content, sterility, and a release panel that the agency will accept as a meaningful predictor of behavior in a patient.

None of that is paperwork. All of it is bench work performed under documented procedures, generating data that FDA can read and audit.

Where teams fail

There are four failure patterns I see repeatedly in pre-IND meetings and warning letter reviews.

Analytical method validation done late. The team has a sensitive HPLC method or a flow cytometry panel. They have data. But the methods were not validated to ICH Q2(R1) standards before the data was generated. FDA reads the IND, recognizes the assay, asks for the validation report, and there is none. The submission stalls until the team goes back and validates retroactively.

Stability data with too few timepoints. The product is stored at one temperature for three months. FDA expects accelerated and long-term stability across the intended storage range with timepoints sufficient to support an expiration date. The IND proposes a six-month shelf life with two stability timepoints. The agency disagrees.

Process consistency demonstrated on one or two batches. Process consistency is a claim about what your manufacturing yields, not what it can yield. FDA wants three consecutive engineering or process-performance batches that meet release criteria. One batch with good results is not data. It is anecdote.

Reference standard logic missing. Every potency assay needs a reference standard. The IND proposes a potency claim against a reference. The reference’s preparation, characterization, and stability are not documented. The whole potency story collapses.

What the agency reviewers actually do

CBER and CDER reviewers read the CMC section against a mental model built from hundreds of prior submissions. They are not looking for elegance. They are looking for completeness, traceability, and consistency.

When the reviewer sees a potency assay, they ask: did you validate it, what is your acceptance range, what is your reference standard, and what is your statistical justification for the range. If any of those four questions does not have a clear answer in the IND, the reviewer writes a clinical hold question.

The clock starts when the IND is filed. The 30-day review window is real. If your CMC section opens questions the reviewer cannot resolve from the document alone, you will get a clinical hold and the timeline shifts by months.

What teams should do differently

Build the CMC story in parallel with the science story, not after. The mistake almost every early-stage biotech makes is treating CMC as something the regulatory consultant handles in the last six weeks before filing. By then, the data has been generated under methods that were not validated, on batches that were not run consistently, against references that were not characterized. The fix is retrospective and costs months.

Run the pre-IND meeting on CMC, not biology. Most pre-IND meetings I review focus 70 percent of the time on the clinical protocol and 30 percent on CMC. The ratio should be reversed for any program that is not a follow-on of a well-characterized predicate. CBER and CDER will tell you what they want to see in CMC. Listen, then build to that spec.

Hire a CMC lead who has shipped a successful IND. Not a regulatory consultant who advises. A CMC lead who has done it. The pattern recognition is the asset. Pay for it.

Treat the analytical methods as products. Validation is not overhead. It is the foundation of every claim you make about the molecule. Plan validation studies the same way you plan toxicology studies. Resource them the same way.

The CMC chasm is real. Most programs fall into it. The ones that do not, did not start their CMC program in the last quarter before filing.

Andrew J. Hillman writes The Hillman Letter for operators, investors, and patient families navigating biotech development. More at https://andrew-hillman.com.