The Challenges with Reproducibility in Organoid Research, and How to Address Them
Team Dynamic Matrices | 27 June 2026
Organoids capture so much real tissue biology that they have become indispensable for development, disease modelling and drug discovery. That same complexity makes them variable. Two labs running the same protocol produce organoids that differ in size, structure and function, and within one lab results drift between batches. We would argue reproducibility is the central unsolved problem in the field, and it is one we deal with every week with customers.
Here are the four sources we see most, and what actually fixes each.
Source 1: Matrix variability
The scaffold is the largest and least appreciated source of variation. Matrigel is an extract from the Engelbreth-Holm-Swarm mouse sarcoma, and its composition changes lot to lot. Hughes et al. (Proteomics, 2010) catalogued more than 1,800 proteins in it; Kozlowski et al. (Communications Biology, 2021) reviewed how this lot variation, together with undefined growth factor content, undermines reproducibility in organoid and stem cell work. Reviews of patient-derived organoid models routinely list Matrigel batch effects alongside specimen source and medium formulation as core problems.
Because the matrix actively shapes development, that variation propagates straight into the organoids. In our own comparisons, lot changes alone shifted organoid formation efficiency by amounts comparable to the effects people publish as biology.
What helps: move to a chemically defined, batch-consistent matrix. This removes one entire uncontrolled variable. It is the single strongest practical argument for synthetic matrices when reproducibility matters.
Source 2: Protocol and handling drift
Organoid protocols are full of steps that are easy to do slightly differently: dissociation intensity, dome volume and geometry, seeding density, timing of medium changes, how long the gel sat before use. Small differences compound over a multi-week culture. Much of this is invisible because it is rarely reported in enough detail to copy.
What helps: written SOPs with numbers (volumes, times, temperatures, cell counts), a standard dome size, and where possible automation of embedding and feeding. A matrix with a simple, temperature-tolerant workflow reduces the number of places technique can intrude. Gels that must be handled on ice or set within a narrow window are a known source of operator variation.
Source 3: Inconsistent readouts
Even isogenic organoids in identical conditions differ in size and morphology. When scoring is manual, that biological spread becomes measurement noise, and different labs scoring "formation efficiency" differently makes cross-study comparison impossible.
What helps: automated imaging and quantification pipelines, pre-registered scoring criteria, and reporting full distributions instead of representative images. Open tools such as OrganoID (Matthews et al., PLOS Computational Biology, 2022) and CellProfiler make this practical for most labs.
Source 4: Mechanical inconsistency
Matrix mechanics influence organoid development, so uncontrolled mechanics is a reproducibility problem in its own right. If stiffness or stress relaxation varies between lots, or simply cannot be specified, organoids experience a different environment each run. Aisenbrey and Murphy (Nature Reviews Materials, 2020) make this point directly in their review of synthetic Matrigel alternatives: the ability to specify and report mechanical properties is a reproducibility feature, not a nicety.
What helps: a matrix with defined, reportable mechanical properties, so every experiment delivers the same cues and every paper can state what they were.
The through-line
Three of the four sources trace back to the matrix. That is not a coincidence. The scaffold is both one of the most influential and one of the least controlled components of an organoid system. Improving reproducibility does not require giving up what makes organoids valuable. It requires removing uncontrolled variables one at a time, and the matrix is the highest-leverage place to start.
Where DyNAtrix® fits
We designed DyNAtrix to attack the matrix-related sources head-on. It is fully synthetic and chemically defined, so it is the same from batch to batch. Stiffness and stress relaxation are set by design rather than inherited from an extract, and we report them on the certificate of analysis. Handling is room temperature and low viscosity, which removes the ice bucket and the pipetting race. If reproducibility is non-negotiable for your work, whether in screening, diagnostics or multi-site studies, that defined foundation is the whole point. Explore DyNAtrix
References
- Hughes CS, Postovit LM, Lajoie GA. Matrigel: a complex protein mixture required for optimal growth of cell culture. Proteomics 10, 1886–1890 (2010).
- Kozlowski MT, Crook CJ, Ku HT. Towards organoid culture without Matrigel. Commun Biol 4, 1387 (2021).
- Aisenbrey EA, Murphy WL. Synthetic alternatives to Matrigel. Nat Rev Mater 5, 539–551 (2020).
- Matthews JM et al. OrganoID: a versatile deep learning platform for tracking and analysis of single-organoid dynamics. PLOS Comput Biol 18, e1010584 (2022).
- Peng Y-H et al. Dynamic matrices with DNA-encoded viscoelasticity support the development of organoids and stem cells. Nat Nanotechnol (2023).
FAQs
Undefined, animal-derived matrices, in our experience and in the literature. Lot-to-lot variation in stiffness, ligand density and growth factor content shifts organoid behaviour independently of whatever biology you are testing.
They help a lot, and you should do them regardless. But protocol discipline cannot compensate for a material that varies underneath it.
At minimum: product and lot, storage modulus at the concentration used, and a stress relaxation half-time if the material is viscoelastic. If your supplier cannot give you those numbers, that itself is informative.
Lot variation is a documented source of noise: Matrigel is an undefined, tumour-derived mixture whose protein and growth factor content change between lots. If you stay with it, use one lot for a whole study, test each new lot side by side before switching and record the lot number. Where reproducibility is critical, a chemically defined matrix removes the variable at its source. See Matrigel alternatives for organoid culture.