Diagnostics & medtech
Organoids & organ-on-chip
The science of self-organizing organoids and perfused organ-on-chip devices: developmental programs that run without a body, the physics of shear and cyclic strain, TEER as a barrier readout, and the maturation, variability and material limits that keep the field honest.
A multicellular organ organizes itself. That fact, not any device, is what makes this field work. Take the stem cells that normally sit at the bottom of an intestinal crypt, isolate them, and hand them the three signals their niche provides in the body — a Wnt agonist such as R-spondin, EGF, and inhibition of BMP — and they rebuild crypt-and-villus architecture without a scaffold or a plan: folded epithelium, polarized enterocytes, even the secretion the tissue exists to perform. Organoids work because development is not orchestrated from outside; it is a default program in the cells, triggered whenever the niche’s signals are present. The laboratory’s contribution is supplying the right signals in the right ratios — an achievement of niche biology, not of engineering.
The chip is engineering’s answer to a different question: a dish is static, and bodies are not. Epithelial barriers in a flat dish differentiate poorly because nothing asks them to cope with flow. Perfuse fluid across the cells and shear stress changes their state — Caco-2 monolayers grown under flow develop taller microvilli and tighter junctions than the same cells grown still. Lung epithelium cycled with vacuum strain, mimicking breathing, models inflammatory responses that static culture misses. The chip’s contribution is physics: physiological shear, cyclic stretch, controlled gradients and, crucially, a geometry small enough that diffusion no longer decides what cells eat. Barrier tightness gets a number from transepithelial electrical resistance, measured continuously, which is what turns the system from a picture into an instrument.
The honest limits are three. First, the maturation gap: induced pluripotent stem cells reset to an embryonic state and their derivatives behave, transcriptionally, more like fetal than adult tissue — a fetal liver models fetal metabolism, not yours. Second, variability: because self-organization is stochastic, no two organoids are identical, and differences between lines and batches are a real confound that the chip’s rigid geometry does not remove. Third, the material itself: the PDMS elastomer that makes soft-lithography cheap is gas-permeable — why it was chosen — but absorbs small hydrophobic molecules, so a drug-screening result in a PDMS chip can understate the dose the cells actually saw. And no current system carries all of an organ: immune cells, vasculature and resident microbiota are added as co-cultures, one axis at a time.
Where this is heading, within diagnostics, is toward tests that cannot be done any other way. A patient’s tumour grown as an organoid is a personalized experiment — the analytical cousin of the companion-diagnostic decision, run on the patient’s own tissue before the prescription. Co-culturing organoids with defined microbial communities extends the microbiome work from association toward controlled cause and effect. Regulators’ guidance now treats microphysiological systems as evidence-bearing models for toxicology, which matters because an animal model whose predictive failure motivated the field in the first place is a poor benchmark to be held to.
See also microbiome diagnostic panels for what microbial co-culture adds, and companion diagnostics platforms for the patient-specific decision these systems increasingly feed.