Filing an Investigational New Drug application with the FDA is a milestone that people celebrate because it means you've done enough work to legally give your compound to a human being for the first time. It's also a documentation exercise that will make you wish you'd kept better records three years earlier.

CAI-001, our KRAS G12C inhibitor, was our first IND. Filing it taught us things about the regulatory process that no textbook on drug development quite prepares you for, especially as a company where the compound originated from a computational platform rather than a traditional medicinal chemistry campaign.

The Pre-IND Meeting

We requested a Type B pre-IND meeting with FDA about fourteen months before we intended to file. This is the meeting where you lay out your proposed development plan and ask specific questions about what the agency expects to see. Most first-time filers underestimate how much work the pre-IND meeting requires and treat it as an exploratory conversation. It's not.

FDA expects you to arrive with a defined compound, a draft nonclinical study plan, and specific, numbered questions. Vague questions get vague answers. We came with 11 questions, mostly around nonclinical toxicology study design and the adequacy of our rat and dog species selection for a KRAS G12C inhibitor. The FDA responses were detailed and gave us clarity on two points that would have otherwise caused a clinical hold — the duration of our GLP repeat-dose toxicity studies and the need for a cardiovascular safety pharmacology study in an unanesthetized dog model.

Without that pre-IND meeting, we would have run a 4-week GLP tox package. FDA made clear they expected 13-week studies to support the Phase I dose range we were proposing. That difference is eight months of additional study time. Finding out at filing rather than before would have been a serious setback.

The CMC Package

Chemistry, Manufacturing, and Controls is the section of the IND that describes how you make the drug substance, how you make the drug product (the formulation you'll give to patients), and how you know it's what you say it is. For first-in-human studies, the FDA's expectations are relatively flexible compared to what they'll require later, but you still have to demonstrate that your material is consistent and characterized.

For CAI-001, the drug substance synthesis runs eight steps from commercial starting materials. The longest single step involves a palladium-catalyzed asymmetric hydrogenation to set a stereocenter critical for activity — the R-enantiomer has a Ki of 9 nM, the S-enantiomer is essentially inactive at over 2,000 nM. Getting consistent enantiomeric excess above 99% at scale required several months of process chemistry work by our CRO partner that we hadn't fully scoped in the original timeline.

The drug product for Phase I was a simple capsule formulation with a hydroxypropyl methylcellulose capsule shell and a spray-dried dispersion to address solubility limitations — the compound's crystalline form has aqueous solubility of only about 8 µg/mL at pH 7, which would have given unacceptably variable oral absorption. The amorphous solid dispersion improved apparent solubility by roughly 40-fold in the in vitro dissolution testing.

What Surprised Us About the Regulatory Response

FDA did not ask us a single question about the AI platform used to design CAI-001. Not in the pre-IND meeting, not in their review of the IND, not in any subsequent correspondence. The compound was evaluated entirely on its preclinical data package, its physicochemical properties, and the quality of the CMC documentation. The method of discovery is apparently as irrelevant to IND reviewers as whether you found the hit through HTS or fragment screening.

What did generate a clinical hold letter — later resolved — was an ambiguity in our proposed dose escalation schedule. The original protocol used a modified Fibonacci escalation starting from a dose calculated as 1/10th of the NOAEL in the most sensitive species. FDA wanted us to include an explicit definition of dose-limiting toxicity for the stopping rules and to clarify the pharmacokinetic sampling schedule during the first cohort. Those were legitimate criticisms and the protocol was strengthened by addressing them.

Advice for First-Time Filers

Start the IND-enabling package earlier than you think you need to, because the long-pole items — the 13-week GLP tox, the reproductive toxicology package if your indication includes women of childbearing potential — take time that can't be compressed. Lock the salt and polymorphic form of your compound before you start the formal GLP toxicology studies, because changing it afterwards means repeating them.

Invest in the pre-IND meeting. Write specific questions. Treat FDA's written responses as binding commitments and document when you deviate from them and why. The IND review process is much smoother when you can show that your development plan was responsive to the agency's own guidance from the outset.