Predicting tumor radiation response from patient-derived organoid irradiation assays
Radiotherapy is delivered to roughly half of all cancer patients, yet no validated functional test predicts which tumors will respond before treatment begins. Patient-derived organoids preserve the 3D architecture, hypoxic gradients, and genomic background of the original tumor, making them a physiologically realistic platform to measure radiosensitivity directly, ex vivo, before a single fraction is given to the patient.
Radiotherapy outcome varies enormously between patients with histologically identical tumors — some achieve complete pathological response, others show radioresistant regrowth within months. Currently, radiation oncologists prescribe a standardized dose (e.g., 60 Gy in 30 fractions for head and neck cancer) largely irrespective of the tumor's intrinsic radiosensitivity, because no functional pretreatment assay has been validated for routine clinical use.
Organoid-based radiosensitivity testing aims to close this gap: by irradiating a patient's own tumor organoids across a clinically relevant dose range before treatment starts, oncologists could in principle escalate dose for radioresistant tumors or spare radiosensitive patients from unnecessary toxicity.
A 2019 study (Vlachogiannis et al.) and several follow-ups demonstrated that organoid drug and radiation response can predict patient outcome with roughly 80–90% sensitivity and specificity across colorectal and rectal cancer cohorts.
Organoid plates are irradiated using the same physical sources used clinically: a ⁶⁰Co teletherapy unit, a cesium-137 irradiator, or more commonly today a clinical or small-animal linear accelerator (linac) delivering megavoltage X-rays. Dosimetry is calibrated with ionization chambers so that the dose delivered to the organoid plate (measured in Gray, Gy — joules absorbed per kilogram of tissue) precisely matches the intended experimental dose, typically a range spanning 0, 2, 4, 6, 8, and 10 Gy delivered as a single acute fraction to mimic the biological effect of a full fractionated course.
Because organoid number and size vary between wells, each irradiation condition is run in triplicate or greater across sister plates split from the same expanded organoid line, alongside a sham-irradiated (0 Gy) control plate that undergoes identical handling. This paired design allows radiation-specific effects to be isolated from handling or passage-related variability.
The biological effect of radiotherapy begins in femtoseconds, as high-energy photons deposit energy along stochastic tracks through the organoid, ejecting electrons that ionize water and shatter the DNA double helix. The double-strand break (DSB) is the molecular lesion that ultimately determines whether a cell survives, repairs, or dies.
When megavoltage photons traverse tissue, they eject fast secondary electrons via Compton scattering. These electrons deposit energy in two ways: direct ionization of the DNA backbone itself (roughly 30% of damage, dominant for high-LET radiation), and indirect damage via radiolysis of the abundant intracellular water into highly reactive hydroxyl radicals (•OH), which diffuse a few nanometers and attack the sugar-phosphate backbone and bases (roughly 70% of damage for the low-LET X-rays used in conventional radiotherapy).
A single Gray of radiation induces roughly 1,000 single-strand breaks, 500–1,000 base damages, but critically also ~35–40 double-strand breaks per cell — the lesion most strongly linked to cell death, because unlike single-strand breaks, DSBs lack an intact complementary template strand for straightforward repair.
A DSB is lethal not because DNA is broken per se, but because incorrect or failed repair can produce dicentric chromosomes, chromosome fragments, and other mitotic catastrophes that kill the cell at its next division attempt.
Because indirect damage depends on oxygen to "fix" (chemically stabilize) the radical damage into a permanent lesion, tumor regions with low oxygen tension (hypoxia) — common in the necrotic cores of solid tumors and in 3D organoid interiors — are radioresistant. The oxygen enhancement ratio (OER) is typically 2.5–3, meaning hypoxic cells require 2.5–3 times more radiation dose to achieve equivalent killing versus well-oxygenated cells.
Organoids, being avascular 3D structures with a diffusion-limited oxygen gradient, spontaneously recapitulate this hypoxic-core biology, making them uniquely suited (compared to 2D monolayers) to model clinically relevant radioresistance driven by tumor microenvironment.
The relationship between dose and cell survival is classically described by the linear-quadratic (LQ) model:
SF(D) = exp(−αD − βD²)
where α represents damage from single radiation tracks causing lethal lesions directly (dominant at low dose), and β represents damage from the additive/interacting effect of two independent sub-lethal tracks (dominant at high dose, giving the characteristic curve "shoulder"). The α/β ratio, empirically ~10 Gy for early-responding tumors and ~3 Gy for late-responding normal tissue, underlies fractionation strategy in the clinic.
Within seconds of a double-strand break, the cell mounts a visible, quantifiable molecular alarm: phosphorylation of histone H2AX spreading for megabases around the break site, recruiting the DNA damage response machinery and marking the lesion for repair. Counting these foci by immunofluorescence microscopy provides a direct, single-cell readout of radiation-induced damage and repair kinetics.
The MRN complex (MRE11-RAD50-NBS1) senses the free DNA ends at a double-strand break within seconds and recruits and activates the kinase ATM (Ataxia-Telangiectasia Mutated). Activated ATM phosphorylates serine 139 of the histone variant H2AX — creating "γH2AX" — across a chromatin domain spanning roughly 1–2 megabase pairs flanking the break, amplifying a single molecular lesion into a microscopically visible focus containing thousands of γH2AX molecules.
MDC1 binds γH2AX and recruits further ATM, creating a positive feedback loop that expands and stabilizes the focus, while simultaneously recruiting downstream repair factors (53BP1, BRCA1) that commit the break to one of two repair pathways.
Non-homologous end joining (NHEJ) directly ligates the two broken DNA ends together, is active throughout the cell cycle, is fast (minutes to hours), but is error-prone — small insertions or deletions are common.
Homologous recombination (HR) uses the intact sister chromatid as a template for error-free repair, but is restricted to S/G2 phase (when a sister chromatid is available) and is slower. Cells with defective HR — for example, BRCA1/2-mutant tumors — are markedly more radiosensitive because they must rely on the more error-prone and less efficient NHEJ or alternative pathways, accumulating lethal misrepair.
Counting residual γH2AX foci at 24 hours post-irradiation — after the initial wave of fast repair is complete — is one of the strongest single predictors of intrinsic radiosensitivity: cells or organoids that still show >5 foci per nucleus at 24h are consistently more radiosensitive in clonogenic assays.
In 3D organoid culture, γH2AX immunofluorescence requires optical clearing and confocal z-stack imaging to resolve foci through tens to hundreds of microns of tissue depth. Automated image analysis identifies individual nuclei (DAPI channel) and counts discrete γH2AX puncta per nucleus over a time course (typically 30 min, 2h, 8h, and 24h post-irradiation), generating a repair kinetics curve whose slope and plateau reflect the efficiency of the DNA damage response.
The gold-standard functional endpoint in radiobiology is not molecular damage but reproductive survival: does an irradiated cell retain the capacity to proliferate indefinitely and form a colony? Adapting the classical 2D clonogenic assay to 3D organoids preserves this rigorous definition of survival while capturing tumor-realistic architecture.
The classical clonogenic assay (Puck & Marcus, 1956) plates single cells at low density and counts colonies of ≥50 cells after 1–2 weeks — the operational definition of a surviving, reproductively intact cell. The organoid adaptation dissociates irradiated organoids to single cells or small fragments, re-embeds them in fresh matrix at low density, and counts newly formed organoid-forming units (OFUs) after 10–14 days, each representing a stem/progenitor cell that retained full clonogenic and self-renewal capacity after irradiation.
Surviving fraction (SF) at each dose is calculated as:
SF = (OFUs formed at dose D / plating efficiency) ÷ (OFUs formed at 0 Gy / plating efficiency)
Among all points on the survival curve, the surviving fraction at 2 Gy (SF2) — matching the standard clinical fraction size — has the strongest historical correlation with clinical radiocurability across multiple tumor types (West et al., cervical cancer cohorts; malignant melanoma and glioma cohorts). Tumors with SF2 below ~0.3–0.4 tend to be locally controlled by standard radiotherapy; tumors with SF2 above ~0.5–0.6 tend to recur locally despite full-dose treatment, and are candidates for dose escalation, radiosensitizing agents, or altered fractionation.
Because organoid SF2 assays take 2–3 weeks, they are currently better suited to retrospective biomarker validation and adjuvant/neoadjuvant treatment planning than to same-day treatment decisions — but they remain far faster than waiting for clinical outcome, which can take months to years.
Surviving fractions measured across the dose series (0, 2, 4, 6, 8, 10 Gy) are fit by nonlinear regression to SF(D) = exp(−αD − βD²), yielding patient-specific α and β coefficients. A high α/β organoid line shows a steep initial slope with little shoulder (radiosensitive, favors hypofractionation); a low α/β line shows a pronounced shoulder (more resistant to single large doses, may benefit from conventional fractionation).
A functional organoid assay only has clinical value once it is validated against real patient outcomes. Combining the organoid-derived surviving fraction with genomic and transcriptomic radiosensitivity signatures, and testing the combined score against actual radiotherapy response, is the final and most important step before any predictive biomarker can influence treatment decisions.
Independent of organoid functional assays, a 10-gene expression-based Radiosensitivity Index (RSI) was derived and validated by Eschrich, Torres-Roca and colleagues from clonogenic SF2 data across dozens of cancer cell lines, capturing pathways including RAS, PI3K/AKT, and cell-cycle regulation. RSI produces a continuous score used clinically to compute the Genomic-Adjusted Radiation Dose (GARD) — an individualized estimate of the biological effect of a given physical radiation dose — which has now been retrospectively validated as prognostic across 11 different tumor types and multiple independent cohorts.
Organoid SF2/α-β parameters and RSI/GARD scores capture partially overlapping but non-identical biology: RSI reflects baseline transcriptional state, while organoid clonogenic survival captures the fully integrated cellular response including DNA repair capacity, hypoxia, and stem-cell fraction. Combined multivariable models integrating both data types consistently outperform either signature alone in predicting pathological complete response to neoadjuvant chemoradiotherapy, particularly in rectal and esophageal cancer cohorts.
In rectal cancer neoadjuvant chemoradiotherapy cohorts, organoid drug-and-radiation response classified patients into responder/non-responder groups with reported sensitivity around 84% and specificity around 93% for predicting actual clinical/pathological response.
The next validation frontier is prospective use: growing a patient's organoids and running the full irradiation-to-RSI pipeline within the 2–4 week window between diagnosis and treatment planning, then using the resulting radiosensitivity score to guide actual dose or fractionation decisions in a randomized trial. Multiple academic centers have begun such prospective organoid-guided radiotherapy trials, with the eventual goal of incorporating a functional organoid radiosensitivity assay into routine tumor board decision-making alongside imaging and pathology.