Most clinical trial managers I talk to don't read tox study reports. Why would they? The studies are signed off by preclinical, the IND is filed by regulatory, and CTM picks up on day one of dosing.
Then the FDA letter arrives. Trial paused. Someone forwards it to operations and asks: how long until we restart?
The honest answer is six months. Sometimes more. And in nine cases out of ten the gap that triggered the hold was sitting in the preclinical package the whole time.
Below are five gaps I've seen pause Phase 1 trials. Not theoretical. Actual lines in actual FDA clinical hold letters. If you run clinical operations and you've never asked your regulatory lead about these, ask before site activation.
1. hERG assay run, but not under GLP
The cardiac safety story for a small molecule sits in Safety Pharmacology. ICH S7B asks for an in vitro IKr assay (almost always hERG) plus an in vivo QT study in a relevant species. Both under GLP.
The gap is almost never "we didn't do hERG." It's "we did hERG at the CRO that does our discovery work, and they're not GLP-certified for that assay."
FDA reads the report, sees the IC50, and notices the GLP statement is missing or qualified. That's a hold. The fix is repeating the study at a GLP lab. Three months minimum, $40-80K depending on the CRO.
What CTM should check: ask regulatory whether the hERG study is GLP-compliant, full stop. Not "we have hERG data." GLP-compliant.
2. Repeat-dose toxicity shorter than the planned trial
ICH M3(R2) Table 1 maps clinical trial duration to required preclinical duration. The rule isn't subtle: a 14-day clinical study needs a 2-week repeat-dose tox study in two species. A 6-month clinical study needs a 6-month rodent study and a 9-month non-rodent.
I've watched a Series B team file an IND for a 28-day Phase 1b extension supported by 14-day rodent tox. They knew the rule. They thought the FDA would accept it because the molecule was clean.
It was not accepted. The hold letter cited M3(R2) Section 1.2 specifically.
What CTM should check: ask what the longest planned dosing duration is across all Phase 1 cohorts (including extension and MAD), and whether the Repeat-Dose Toxicity studies cover that duration in both species.
3. Genotoxicity battery incomplete
ICH S2(R1) defines the standard battery: an Ames test, an in vitro mammalian assay (chromosomal damage or mouse lymphoma), and an in vivo test (usually micronucleus). All three. Not two.
The most common shortcut I see is Ames plus in vitro micronucleus, with the in vivo test scheduled for "post-IND." That doesn't work for a small molecule heading into healthy volunteers. FDA wants the full battery before first dose in humans.
The Genotoxicity section in CTD Module 4 expects all three. Skipping the in vivo arm is the most expensive shortcut in preclinical: typically a 4-month delay because the bioanalytical and pathology read takes time.
What CTM should check: confirm three genotox studies, not two. Ames, in vitro mammalian, and in vivo.
4. MRSD math doesn't show its work
The maximum recommended starting dose comes from the NOAEL of the most sensitive species, scaled to humans by body surface area, divided by a safety factor. FDA's 2005 guidance lays out the math.
Where this goes wrong: the protocol says "starting dose 0.1 mg/kg," but the IND doesn't show how that number falls out of the NOAEL. Or the safety factor used is 10 when the molecule has a steep dose-response in non-rodents and FDA expects 100.
This isn't a missing study. It's a missing paragraph. The data is in the Toxicology Written Summary, but the bridge from animal NOAEL to human starting dose isn't documented well enough for the reviewer to follow.
What CTM should check: read the MRSD justification in Module 2.4. If you can't follow it as a non-expert, the reviewer probably can't either.
5. One species when FDA wants two
ICH M3(R2) requires repeat-dose toxicology in two species — one rodent, one non-rodent. The non-rodent is usually dog or non-human primate. There are exceptions for biologics and ATMPs, but for a small molecule heading into Phase 1, two species is the rule.
I've seen teams skip the non-rodent because the dog facility quoted six months and the program needed to file. They submitted with rodent only and a justification arguing the non-rodent wasn't needed.
The FDA reviewer disagreed. Hold.
The frustrating part: the justification might have worked at a pre-IND meeting. Filing it cold as part of the IND doesn't give you a conversation, just a letter.
What CTM should check: confirm both rodent and non-rodent tox are in the package, or that there's documentation from a Type B pre-IND meeting where FDA agreed to the single-species approach.
What to actually do with this list
These aren't edge cases. They're the most common preclinical gaps FDA cites in clinical hold letters. The data behind them is sitting in your IND package right now. Someone just has to read it against ICH M3(R2), S7B, and S2(R1) and check whether each requirement is met.
That's the job we built Regfo to automate. The rules engine reads the toxicology, pharmacology, and genotoxicity reports, extracts the actual data points, and tells you which ICH requirement is met, which is partially met, and which is missing. With the page number where the gap lives.
Run it on your own IND package before site activation. Or, more useful, run it before the IND is filed. The cost of fixing a gap pre-submission is a phone call. The cost of fixing it post-hold is six months and your enrollment window.
Try it on your study set. Paste the studies, get the gaps in 30 seconds.