What is in vitro / early toxicology and when do you need it?
In vitro and early toxicology is the first real safety look at a candidate, run mostly in cells and biochemical systems rather than in a full animal study. It sits late in discovery and through the preclinical stage, after you have a lead or a short list of leads, but before you commit the budget and the calendar to a definitive GLP toxicology program. The job is blunt: find the liabilities that will kill the molecule, and find them while killing it is still cheap.
You reach for it at two moments. The first is candidate selection, when you have two or three molecules that all look good on potency and DMPK and you need a safety read to break the tie. A clean structure that throws an Ames positive or a strong hERG signal is not the candidate you want to nominate. The second is right before IND-enabling, where you run dose-range-finding (DRF) and a few exploratory screens to set the doses and design for the pivotal GLP studies that follow. Get the DRF wrong and your expensive two-species tox study reads at the wrong doses.
Almost all of this work is non-GLP, and that is by design. The point is decision-grade science that informs your go/no-go, not a regulatory filing. The definitive, GLP-compliant battery (repeat-dose tox in two species, the safety pharmacology core, the full genotox set) belongs to the IND-enabling stage that comes next. Treating early tox as a cheap filter rather than regulatory paperwork is what keeps the program affordable. You retire weak candidates here, before you have sunk six figures into a GLP study on a molecule that was never going to survive.
What does an in vitro / early toxicology CRO actually do?
These CROs run the assay panels that surface the safety signals regulators and your own development team care about most: mutagenicity, chromosomal damage, cardiac ion-channel effects, and liver and general cell toxicity. Some are pure in vitro shops; others pair the in vitro panels with small in vivo dose-range-finding studies so you can hand off a coherent package into IND-enabling. The strongest partners also tell you, plainly, which version of each study is exploratory and which you should later repeat under GLP.
Most early-tox sourcing breaks into a handful of recognizable workstreams. You rarely buy all of them at once. You buy the two or three that answer the question in front of you, whether that is breaking a tie between candidates or de-risking a known structural alert before you spend on the pivotal program.
- Genetic toxicology screens: the bacterial reverse mutation test (Ames), often a mini-Ames or Ames II for early triage, plus an in vitro micronucleus or chromosomal aberration assay to flag clastogenic and aneugenic liability before first-in-human.
- Cardiac safety: a hERG patch-clamp assay (manual or automated, such as a QPatch or IonWorks panel) for QT and arrhythmia risk, sometimes broadened to a multi-ion-channel panel or a cardiomyocyte (iPSC-CM) assay for a fuller cardiac read.
- Hepatotoxicity and general cytotoxicity: liver liability screens in HepG2 or primary human hepatocytes, mitochondrial toxicity and reactive-metabolite (glutathione trapping) assays, and broad cytotoxicity panels to catch general cell stress.
- Off-target and secondary pharmacology: receptor, enzyme, and ion-channel safety panels (the kind of broad pharmacological profiling that flags unexpected interactions a target-focused program would miss).
- Dose-range-finding (DRF) and MTD studies: short, usually non-GLP in vivo studies that establish tolerability and set the doses for the later GLP repeat-dose program, plus exploratory toxicokinetics to tie dose to exposure.
- Photosafety, skin sensitization, and irritation screens where the modality or route of administration calls for them, often using validated in vitro alternatives (3T3 NRU phototoxicity, KeratinoSens or h-CLAT for sensitization).
How do you choose an in vitro / early toxicology CRO?
Start with whether the assay actually fits your molecule, not the size of the menu. A shop with a flawless small-molecule genotox panel may have nothing useful for an oligonucleotide, an antibody, or a cell therapy, where the relevant questions shift to immunogenicity, off-target hybridization, or tumorigenicity. Match the supplier to your modality and the specific liability you are chasing before you look at a price.
The other thing that separates a useful partner from a frustrating one is honesty about the GLP line. You want a CRO that says clearly which of its studies are exploratory and which you should later run under GLP, and that builds an early assay (a fit-for-purpose hERG read, say) in a way that maps cleanly to the validated version you will need for the filing. A cheap non-GLP study you cannot connect to your regulatory package is a study you will end up paying for twice.
Run two or three candidates against the same written scope and weigh these points:
- Quality and GxP status: confirm whether each study is non-GLP exploratory or GLP-compliant (21 CFR Part 58, OECD GLP for ex-US work), and ask which validated guideline each assay follows, for example OECD 471 for Ames or ICH S2(R1) for the genotox battery.
- Capacity and lead time: a great lab booked solid for months is slower than a good lab with an open slot. Ask about the current queue, in-life and reporting turnaround, and what historically causes slippage.
- Modality and indication fit: confirm the panels and any animal models are relevant to your molecule (small molecule, ADC, oligonucleotide, biologic, cell or gene therapy) and your therapeutic area, not a generic small-molecule default.
- Region and regulatory track record: check that the data will be accepted where you plan to file (FDA, EMA, PMDA, NMPA) and ask how many comparable programs the team has carried into an IND.
- Data quality and reporting: ask for a sample report, the assay acceptance criteria and historical control data, and how they handle a borderline or equivocal result rather than just reporting a flag.
- IP and confidentiality: settle who owns the data and any structural information disclosed, and confirm the CDA covers a target or chemotype you may not want named.