A Series A biotech I worked with ran a 28-day repeat-dose toxicity study at a CRO that quoted them $180K less than the next-lowest bidder. Clean data. Good science. Clear NOAEL. They were thrilled — until regulatory counsel asked for the GLP compliance statement.
There wasn't one.
The CRO had run the study under "GLP-like" conditions. Their internal quality systems were solid. They had SOPs, trained staff, calibrated equipment. But the study didn't have a designated study director under §58.33, didn't have QAU phase inspections under §58.35, and didn't have a GLP compliance statement in the report. Under 21 CFR Part 58, that study was not GLP-compliant. Under FDA's IND review expectations, it couldn't serve as the pivotal tox package for a Phase 1 application.
They repeated the study. Cost: $620K and 5 months they didn't have. The $180K savings turned into a $440K penalty plus half a year of delay.
This is the most expensive question that doesn't get asked early enough: which studies need to be GLP, and which don't?
What the regulation actually says
21 CFR Part 58 is straightforward about scope. §58.1 says GLP applies to "nonclinical laboratory studies that support or are intended to support applications for research or marketing permits." The key phrase is "intended to support." If a study's results will be in your IND — in Module 4 of the CTD — and if those results are used to establish the safety basis for human dosing, GLP applies.
But "intended to support" leaves room for interpretation. FDA hasn't published a bright-line list of "these studies must be GLP, these don't." Instead, they expect sponsors to use judgment based on how the data will be used. ICH M3(R2) provides the closest thing to a definitive guide: it describes which nonclinical studies are expected before each clinical phase, and by implication, which ones FDA expects to be GLP-compliant.
Let me make this concrete.
Studies that are always GLP
These are the pivotal safety studies. No ambiguity.
Repeat-dose toxicology. Your 28-day or 90-day tox studies in two species. These define the NOAEL, establish the human starting dose calculation, and form the core safety argument for your IND. They must be GLP. Full stop. If your IND's Module 4 contains a repeat-dose toxicity study without a GLP compliance statement, expect a clinical hold.
Safety pharmacology core battery. The three studies required under ICH S7A: cardiovascular (including hERG and telemetry), CNS, and respiratory. These are GLP because they assess whether your compound creates acute safety risks that aren't captured in general toxicity studies. The hERG assay alone has triggered more clinical holds than probably any other single study. Making it non-GLP would be... creative.
Genotoxicity battery. The standard battery under ICH S2(R1): bacterial reverse mutation assay (Ames), in vitro chromosomal aberration or micronucleus, and in vivo micronucleus. All GLP. These studies determine whether your compound damages DNA. The regulatory consequence of a positive result without adequate follow-up is program-stopping. Running them non-GLP is not a cost optimization — it's a submission risk.
Reproductive and developmental toxicity. Segment I, II, and III studies (fertility, embryo-fetal development, pre/postnatal development). Required timing varies by clinical phase and indication, but when required, always GLP.
Carcinogenicity studies. Long-duration, expensive, and always GLP. These are typically required for drugs intended for chronic use and are usually run before NDA, not before Phase 1. But when they're needed, there's no non-GLP option.
The gray zone
This is where sponsors get confused, overspend, or underspend. Neither mistake is harmless.
Dose range-finding studies. You run a short-duration tox study (often 7 days) to select doses for the pivotal study. It's not going into your IND as a safety argument. It's an internal decision-making tool. Does it need GLP?
Strictly: no. §58.1 applies to studies that "support applications." A dose range-finder supports your study design, not your submission. Most CROs will offer a non-GLP option at significantly lower cost. That said, I've seen sponsors run them GLP anyway because the CRO's GLP and non-GLP pricing was close enough that the extra documentation felt like cheap insurance. Not irrational. Just not required.
Exploratory pharmacology. Primary and secondary pharmacodynamic studies, mechanism of action work, selectivity panels. These go in Module 4 under pharmacology, but FDA generally doesn't expect them to be GLP. They're scientific characterization, not safety assessment. The nonclinical overview in Module 2 should describe them and note that they were non-GLP, but this isn't a deficiency.
Exploratory PK and ADME. Single-dose pharmacokinetics, tissue distribution, mass balance, metabolite identification. These studies go in Module 4 under pharmacokinetics. They inform dose selection and help predict human PK, but they're not safety studies. Non-GLP is the norm. Some sponsors do run pivotal PK studies (the ones used to set clinical dose) under GLP, especially for biologics where PK drives the safety assessment. Judgment call.
In vitro studies. Plasma protein binding, CYP inhibition panels, Pgp transport assays. Almost never GLP. The exception: if an in vitro study is the sole basis for a safety claim (rare, but possible with some genotoxicity follow-up studies), GLP applies.
Bridging studies. When you change formulation, route, or species partway through development and need to bridge the old data to the new. Whether these need GLP depends entirely on how you'll use the data. If the bridging study replaces a pivotal study, it's GLP. If it supports a scientific argument, it's judgment.
The cost of getting it wrong
Let me be specific about what GLP overhead actually looks like, because "GLP adds cost" is vague enough to be useless.
A GLP study at a typical CRO costs 20-40% more than the same study run non-GLP. That premium buys you:
- Designated study director with documented oversight
- QAU phase inspections and report review
- Formal protocol with amendment tracking
- Raw data management under archival standards
- GLP compliance statement in the final report
For a 28-day rat tox study, the non-GLP price might be $350-450K. GLP pushes it to $500-620K. The delta is real. For a Series A burning $200K/month, that's 1-2 months of additional runway consumed.
But here's the cost of the alternative mistake: repeating a pivotal study because it was run non-GLP. The repeat costs the full GLP price plus 4-6 months of delay. If you're targeting an IND submission window, that delay can cascade into missed clinical site availability, investor milestone slippage, and competitor timing advantages. I've seen the total cost of a non-GLP mistake reach $1M+ when you account for downstream effects. The $180K savings from the opening story was the worst financial decision that company made.
For studies that genuinely don't need GLP, though, running them GLP anyway is wasted money. A CYP inhibition panel under GLP costs roughly twice the non-GLP version with no regulatory benefit. An exploratory 7-day PK study with full GLP documentation burns budget on QAU inspections and archive management for data that will appear in a supporting role at best. Know where the line is.
The "GLP-like" trap
"GLP-like" is a term the industry uses and FDA does not recognize. There's GLP-compliant (with the compliance statement per §58.3) and there's not GLP-compliant. There's no in-between.
I understand why labs use it. Some facilities run non-GLP studies with strong quality systems — trained staff, calibrated equipment, good documentation. They call these "GLP-like" or "conducted in the spirit of GLP" to signal quality without the full compliance burden. The problem is that this language creates a false sense of regulatory security for sponsors who hear "GLP-like" and mentally round it up to "GLP."
If your CRO proposal says "GLP-like" or "conducted under GLP principles" or "quality-assured per GLP standards," ask directly: will the final report contain a GLP compliance statement per 21 CFR 58.3? If the answer is no, the study is non-GLP regardless of how many quality adjectives are attached to it.
By the way, this isn't a CRO-bashing point. Good CROs are clear about GLP status upfront. The confusion usually happens with smaller facilities that genuinely have strong quality systems but haven't invested in full GLP infrastructure — the QAU, the archive facility, the master schedule. Their science may be excellent. Their regulatory status is still non-GLP.
Bioanalytical studies: the special case
Bioanalytical method validation and sample analysis occupy an awkward regulatory position. The samples come from GLP in vivo studies (typically tox or PK). The analytical work determines drug concentrations in those samples. But the bioanalytical lab may or may not be operating under GLP.
FDA's expectation, articulated in the 2018 Bioanalytical Method Validation guidance, is that bioanalytical work supporting GLP studies should be conducted under GLP. In practice, many sponsors and CROs run the bioanalytical component under GLP while the method development and validation run under a quality system that's not formally GLP. This is acceptable if the method validation data is documented adequately and the study-specific sample analysis is GLP.
There's also GCLP — Good Clinical Laboratory Practice — which applies to bioanalytical labs supporting clinical trials. GCLP isn't a regulation (it's a framework), and it's different from GLP. Confusing the two in your submission is a real mistake, though not a common one.
The practical takeaway: if your nonclinical PK or toxicokinetic study is GLP, the bioanalytical work for that study should also be GLP. If the bioanalytical lab isn't GLP-certified, you need to address this in your submission.
Decision framework
| Study Type | Regulatory Purpose | GLP Required? |
|---|---|---|
| Repeat-dose tox (pivotal) | Safety basis for FIH dose | Yes, always |
| Safety pharmacology core battery | Acute safety risk assessment | Yes, always |
| Genotoxicity battery | DNA damage potential | Yes, always |
| Reproductive/developmental tox | Reproductive safety | Yes, when required by phase |
| Carcinogenicity | Long-term cancer risk | Yes, always |
| Dose range-finding tox | Study design decision | No (but acceptable if run GLP) |
| Exploratory pharmacology | Mechanism of action | No |
| Exploratory PK/ADME | Dose selection, PK prediction | No (exceptions for biologics) |
| In vitro safety panels | CYP, hERG screening | No (except when sole safety basis) |
| Bioanalytical (for GLP studies) | TK/exposure data | Yes, should match parent study |
| Bridging studies | Data continuity | Depends on regulatory use |
This isn't a regulation — it's a practical guide based on how FDA typically evaluates these studies. When in doubt, discuss with your regulatory consultant or raise it in a Pre-IND meeting.
Practical tips for sponsors
Put GLP status in the CRO contract. Not buried in an appendix. In the statement of work, in the study-specific section, with specific reference to 21 CFR Part 58. "Study will be conducted in compliance with GLP per 21 CFR Part 58" is one sentence. Its absence has cost sponsors hundreds of thousands of dollars.
Define what "non-GLP" means for your non-GLP studies. Non-GLP doesn't mean no quality system. Specify what you do expect: trained personnel, calibrated equipment, documented methods, data retention. You're just not requiring the full Part 58 apparatus — the designated study director, QAU, compliance statement, formal archiving. Making this explicit prevents the CRO from interpreting "non-GLP" as "anything goes."
Keep non-GLP data organized. Non-GLP data still goes in your IND. It appears in Module 4 under the appropriate section, clearly labeled as non-GLP. A disorganized non-GLP data package makes FDA reviewers wonder about the quality of your entire program. Clean presentation of non-GLP data signals scientific rigor even without the GLP stamp.
Don't retroactively convert. If a study was run non-GLP, you can't make it GLP after the fact by adding documentation. GLP compliance is prospective — the systems and oversight must be in place during study conduct. If you realize mid-program that a study should have been GLP, the conversation is about whether to continue it as non-GLP (with appropriate justification in your submission) or restart under GLP. Neither option is fun. Both are better than pretending a non-GLP study is GLP.
How non-GLP data fits in the IND
FDA accepts non-GLP data in IND submissions. This isn't a secret, but some sponsors treat any non-GLP study as something to hide or minimize. That's the wrong approach.
In your nonclinical overview (Module 2.4), describe each study's GLP status. For non-GLP studies, briefly explain why GLP wasn't required for that study type. The explanation is usually obvious: "Exploratory PK studies to characterize absorption and distribution were conducted under the sponsor's quality system but were not designated as GLP studies, consistent with standard practice for non-pivotal PK characterization."
In Module 4, organize non-GLP and GLP study reports in the same structure. Don't segregate them into a "lesser" section. The CTD structure doesn't distinguish by GLP status — studies are organized by type (pharmacology, pharmacokinetics, toxicology).
The one thing that will raise a flag: a pivotal safety study labeled non-GLP without explanation. If your 28-day tox study doesn't have a GLP compliance statement, FDA will ask why. "We didn't think it was necessary" is not a good answer when the study supports a First-in-Human dose calculation.
For the full IND requirements by module, see the IND submission checklist.
Related reading:
- GLP Compliance Checklist for Preclinical Studies — the full 21 CFR Part 58 walkthrough
- IND-Enabling Studies: Timeline, Cost, and What You Need — planning the full nonclinical package
- FDA GLP Inspections: What 483 Data Reveals — enforcement patterns at testing facilities
RegFo checks your nonclinical study package for GLP compliance gaps and missing studies before you submit your IND.