Dispersion-Limited Nanocarrier Delivery, Measured in Patient Organoids

A nanoparticle drug carrier is an engineered particle roughly a hundred nanometres across that is supposed to ferry a chemotherapy payload through the bloodstream and into a tumor. They routinely kill cancer cells in laboratory dishes and then fail in patients: the file's synthesis attributes more than 90% of oncology preclinical failures to the gap between flat, static cell cultures and living tissue. A major, under-measured reason is physical rather than molecular. Because of their enormous surface area, nanoparticles spontaneously clump together, and a clump that grows larger than the pores of the dense protein mesh around a tumor simply cannot get in. This was shown directly in a synthetic porous matrix, where electrostatic repulsion effectively narrowed 149 nm pores by as much as 47 nm and stretched the time a particle stayed trapped in a cavity from 2.60 s to 9.49 s as salt concentration dropped from 0.1 M to 0.001 M Electrostatic Barriers to Nanoparticle Accessibility of a Porous Matrix. No standard assay sees any of this: flat monolayers have no matrix at all, and every existing patient-derived organoid drug study in the file tests only freely dissolved small molecules.

The proposal is to make that invisible physics visible and test whether it is predictive. Patient-derived organoids — miniature tumors grown from a patient's own cancer cells inside a gel — recreate the diffusion barriers, cell-matrix signalling, and stromal fibroblast crosstalk of real tissue. I will build a platform that watches fluorescent nanocarriers move, clump, and stall inside living patient-derived pancreatic tumor organoids co-cultured with the patient's own fibroblasts, extracts quantitative transport and agglomeration features from three-dimensional single-particle movies, and then asks a blunt question: do these organoid-measured numbers predict how the same carriers accumulate in and affect tumors grown in mice from the same patients, better than a conventional dish screen does?

Core Claim

The claim is twofold. First, the colloidal dispersion state of a nanocarrier inside complete culture medium and inside the organoid's extracellular matrix — not its molecular potency — is the dominant, measurable determinant of how deep it penetrates and which tumor cells it kills. Second, transport features measured in patient organoids predict in vivo carrier accumulation and efficacy with a materially higher rank correlation than either 2D uptake measurements or organoid viability readouts that ignore particle physics. The claim is falsifiable: if organoid transport features do not out-predict the standard 2D screen, the dispersion-barrier hypothesis is wrong for this carrier class and the field should spend its effort elsewhere. Currently, no reported result found on Lacuna for nanoparticle penetration depth or agglomerate size distributions in any patient-derived organoid system, and no reported result found on Lacuna for any correlation between organoid transport metrics and xenograft drug accumulation.

Research Program

The biological model follows the patient-specific pancreatic cancer co-culture of Patient-specific modeling of stroma-mediated chemoresistance of pancreatic cancer using a three-dimensional organoid-fibroblast co-culture system, which pairs primary pancreatic ductal adenocarcinoma organoids with patient-matched cancer-associated fibroblasts in a Matrigel/collagen-I hydrogel and showed that fibroblasts drive a chemoresistant mesenchymal shift via secreted hepatocyte growth factor. Their reported constraints shape the study design honestly: expansion takes a median of 40 days and succeeds for only 50% of lines, so I will enroll roughly 30 patients to bank 12–15 usable organoid–fibroblast pairs, and I will complement them with organotypic slice cultures, which preserve native architecture and resident stroma without any expansion delay, as demonstrated for prostate cancer in A Feasibility Study of Co-Established Patient-Derived Subcutaneous Xenograft and Organotypic Slice Cultures in Hormone-Naive Primary Prostate Cancer Preclinical Modeling. As an acellular comparator I will use decellularized tumor extracellular matrix scaffolds, whose Tris-Trypsin-Triton protocol drops residual DNA to under 1 ng/mg while doubling porosity from about 23% to about 46% Development of an Acellular Tumor Extracellular Matrix as a Three-Dimensional Scaffold for Tumor Engineering.

The particle panel holds the drug chemistry fixed and varies only the dispersion state. One fluorescently labeled carrier (a liposomal or polymeric doxorubicin formulation) will be prepared in surface variants spanning a controlled range: a stably coated, well-dispersed form and forms designed to flocculate under physiological ionic strength and serum. This borrows the logic of surface functionalization from the coating literature — where grafting a cardanol diluent onto graphene oxide shifted its water contact angle from 63° to 114° and prevented agglomeration in epoxy Synergistically enhancing the performance of cardanol-rich epoxy anticorrosive coatings using cardanol-based reactive diluent and its functionalized graphene oxide — and the electrolyte-control logic of the DLVO study above Electrostatic Barriers to Nanoparticle Accessibility of a Porous Matrix. Agglomerate size distributions will first be characterized in plain buffer versus complete culture medium, testing the documented concern that dispersions stable in simple solvents flocculate in ionic, protein-rich biological media.

Measurement combines three instrumented readouts on the same live cultures. Light-sheet microscopy with the DeepSPT analysis pipeline — which segments motion regimes along single trajectories with a reported 96% median per-trace accuracy, classifies endosome populations at roughly 72% accuracy, and was validated on 900,000 synthetic trajectories Deep learning assisted single particle tracking for automated correlation between diffusion and function — yields per-particle trajectories inside the organoid and its surrounding matrix, from which I extract confinement fraction, stalling location, penetration depth, and in situ agglomerate size. Diffusion tensors in the peritumoral matrix come from the LiFT-FRAP platform, which images full 3D diffusion anisotropy non-invasively at 8 volumes per second for diffusivities up to 51 µm²/s A noninvasive fluorescence imaging-based platform measures 3D anisotropic extracellular diffusion. Because the particle-tracking grand challenge found that most tracking algorithms fail below a signal-to-noise ratio of 4 Objective comparison of particle tracking methods, I will report imaging SNR per experiment and only trust motion classifications above that floor. The therapeutic readout reuses the spatio-temporal video pipeline of Spatio-Temporal Analysis of Patient-Derived Organoid Videos Using Deep Learning for the Prediction of Drug Efficacy, which predicted terminal ATP viability with a Pearson correlation of 0.9214 and a mean absolute percentage error of 0.1755 — halving the prior best of 0.4375 — and forecast 100-hour outcomes from the first 7.5 hours of bright-field video; I extend its feature set with the particle-transport measurements. Segmentation label noise is a known hazard when stromal cells make organoids hard to count MultiOrg: A Multi-rater Organoid-detection Dataset, so organoid detection will carry the multi-rater uncertainty of that dataset forward rather than assume a single ground truth.

The in vivo anchor implants the same organoid lines as patient-derived xenografts in mice, following the avatar design of Patient-derived organoids and orthotopic xenografts of primary and recurrent gliomas represent relevant patient avatars for precision oncology, which paired organoids with matched xenografts and showed drug responses tracking clinical outcomes — but, like every organoid study in the file, only for freely dissolved drugs. Mice receive the identical formulations at matched doses; endpoints are fluorescence-quantified tumor accumulation and growth suppression. Across all line–formulation pairs, each candidate predictor (2D uptake and IC50, morphology-only organoid viability, organoid transport features, and the combined model) is scored by Spearman rank correlation against the in vivo endpoints.

Baselines

The naive reference point is the conventional 2D monolayer screen: nanoparticle formulations applied to flat cultures of the same patient-derived cells, read out by uptake fluorescence and a lysis-based viability assay at 72 hours, exactly as small molecules were screened in Fangchinoline Induced G1/S Arrest by Modulating Expression of p27, PCNA, and Cyclin D in Human Prostate Carcinoma Cancer PC3 Cells and Tumor Xenograft. It is the strongest competitor in ubiquity and cost, and the literature already flags its distortion: cells that died rapidly under doxorubicin in 2D survived the same concentrations as 3D spheroids Uniform Tumor Spheroids on Surface-Optimized Microfluidic Biochips for Reproducible Drug Screening and Personalized Medicine, and nanostructured lipid carriers co-delivering doxorubicin and paclitaxel showed up to 9 times greater effectiveness than free drug mixtures in 2D lung cancer models while the review reporting this sits inside a >90% clinical attrition record Nanoparticle drug delivery systems for synergistic delivery of tumor therapy. It will be run with the same particle batches, the same patient lines, and the same exposure schedule, so the only difference is the culture geometry. Beating it means a statistically significant gain in rank correlation with in vivo accumulation and efficacy; my explicit prediction is that 2D uptake will show near-zero or negative correlation.

The strongest existing methodological baseline is the Fillioux spatio-temporal video viability model itself Spatio-Temporal Analysis of Patient-Derived Organoid Videos Using Deep Learning for the Prediction of Drug Efficacy, which holds the best reported numbers on organoid response prediction (Pearson 0.9214 with terminal ATP, MAPE 0.1755 against a prior best of 0.4375). It is the strongest competitor because it already solves non-destructive, early forecasting in organoids — but it predicts an in-well endpoint and has never been applied to nanocarriers or to in vivo outcomes. I will run it unmodified on the nanoparticle-treated wells, morphology features only, on the same microscope, with the same compute and the same tuning budget as my combined model. Beating it means the addition of particle-transport features improves in vivo prediction where morphology alone does not; the decisive signature is the fixed-payload dispersion experiment, where morphology-only predictions should stay flat across surface variants while transport features track the in vivo outcome.

The key ablation isolates dispersion as the causal ingredient: the same carrier chemistry, payload, and fluorescence label presented in a stably dispersed form and a deliberately flocculated form, dosed into the same wells. If organoid penetration and in vivo accumulation move together between these two arms, dispersion — not chemistry — is established as the operative variable. A second ablation runs the same particles in the acellular decellularized matrix versus the live fibroblast co-culture, separating passive matrix filtration from cellular uptake, efflux, and stromal remodeling; the decellularized scaffold is expected to underpredict because decellularization lowers elastic modulus by removing cellular tension Development of an Acellular Tumor Extracellular Matrix as a Three-Dimensional Scaffold for Tumor Engineering.

Why not just…?

Why not just run the screen in 2D plates? Because the specific failure mode this proposal targets — agglomerates too large for matrix pores — is structurally absent from a monolayer: there is no matrix, no gradient, and every cell is bathed uniformly. The biochip study cited above showed 2D cultures artificially inflating doxorubicin sensitivity, and the nanocarrier review documents large in vitro advantages that do not survive translation. Still, this is not an argument from authority but a measured claim: the 2D arm is included as the primary baseline, and if its correlation with in vivo accumulation matches the organoid arm, the proposal's premise fails and the experiment says so.

Why not just use standard cell-line spheroids in Matrigel? Cheaper and faster, but the file gives three reasons for doubt: Matrigel forces uniform spherical morphologies and hides the state-dependent drug vulnerabilities that appear only in collagen branching cultures, where chemotherapy spared mesenchymal star-like organoids that radiation killed and vice versa Heterogeneity-driven phenotypic plasticity and treatment response in branched-organoid models of pancreatic ductal adenocarcinoma; established cell lines drift genetically and lose patient mutations while patient avatars retain them Patient-derived organoids and orthotopic xenografts of primary and recurrent gliomas represent relevant patient avatars for precision oncology; and Matrigel is a mouse-sarcoma extract with batch variability. This is a real threat, though: if the question is purely physical filtration, cell-line spheroids might suffice. The design settles it empirically — a head-to-head subset runs A549 spheroids alongside matched organoids, and if the cell-line transport metrics predict in vivo outcomes as well, I will report that screening can drop the patient-derived requirement and keep patient organoids only for validation.

Why not just measure diffusion in acellular gels? The DLVO study already gives clean physics for that: salt-tunable escape times and electrostatically narrowed pores in silica films Electrostatic Barriers to Nanoparticle Accessibility of a Porous Matrix, and LiFT-FRAP already gives 3D anisotropic diffusion tensors in cornea and tendon A noninvasive fluorescence imaging-based platform measures 3D anisotropic extracellular diffusion. What acellular systems cannot reproduce is the part that matters for translation: cellular endocytosis, P-glycoprotein efflux upregulation, stromal HGF signalling that reprograms cells toward chemoresistance Patient-specific modeling of stroma-mediated chemoresistance of pancreatic cancer using a three-dimensional organoid-fibroblast co-culture system, and active matrix remodeling. The acellular arm is retained as an ablation rather than dismissed: if it predicts in vivo delivery as well as the live co-culture, the cheaper platform wins and that too would be a useful, publishable result.

Why not just go straight to mouse xenografts? Xenografts are the validation endpoint here, not a replacement for the screen, because they are low-throughput, expensive, hard to image in real time, and confounded by murine stroma that does not match human pharmacology — the same fangchinoline study needed very high systemic doses to reach 57.8% tumor suppression in PC3 xenografts, precisely the dosing discrepancy this proposal wants to explain before animals are spent Fangchinoline Induced G1/S Arrest by Modulating Expression of p27, PCNA, and Cyclin D in Human Prostate Carcinoma Cancer PC3 Cells and Tumor Xenograft. The point of the organoid layer is to make the xenograft cohort smaller and better targeted, and the point of the matched-patient design is that the in vivo comparison stays within human tissue.

Proposed Experiments and Decisive Checks

The program runs five linked experiments. First, media-induced agglomeration: measure size distributions of each formulation in buffer, serum-free medium, and complete medium, expecting physiological ionic strength and serum protein to push a fraction of particles above the matrix mesh size. Second, dispersion versus penetration: dose the stabilized and flocculated forms into organoid co-cultures and slice cultures, quantify penetration depth, stalling fraction, and in situ agglomerate growth over 48 hours by light-sheet tracking, with LiFT-FRAP mapping the anisotropic diffusion tensor of the surrounding matrix. Third, phenotype-resolved killing: apply the extended video-viability model to ask which morphological classes are spared, testing the expectation that carriers which stall peripherally kill organized epithelial organoids while sparing mesenchymal ones, mirroring the differential killing seen with free drugs in the branched-organoid work. Fourth, the matched in vivo cohort: dose the same formulations in xenografts of the same lines and measure tumor accumulation and growth suppression. Fifth, the prediction tournament: score every predictor against the in vivo endpoints across all line–formulation pairs.

Two checks are decisive. If the fixed-chemistry dispersion pair produces concordant changes in organoid penetration and in vivo accumulation, dispersion is established as a causal, screenable variable — and the platform becomes the first assay that measures it in human tumor architecture rather than in silica or salt. If, instead, the organoid transport features fail to beat the 2D baseline at predicting in vivo outcomes, the stromal-filtration hypothesis is not the binding constraint for this carrier class, and the result would redirect nanocarrier optimization toward whatever predictor does win. Either outcome closes a gap that, on the evidence assembled here, no study has yet touched: the file contains no work that measures nanoparticle colloidal behavior inside patient-derived tumor models and validates it against in vivo delivery.

References

  1. Electrostatic Barriers to Nanoparticle Accessibility of a Porous Matrix
  2. Patient-specific modeling of stroma-mediated chemoresistance of pancreatic cancer using a three-dimensional organoid-fibroblast co-culture system
  3. A Feasibility Study of Co-Established Patient-Derived Subcutaneous Xenograft and Organotypic Slice Cultures in Hormone-Naive Primary Prostate Cancer Preclinical Modeling: A Single-Institution Experience
  4. Development of an Acellular Tumor Extracellular Matrix as a Three-Dimensional Scaffold for Tumor Engineering
  5. Synergistically enhancing the performance of cardanol-rich epoxy anticorrosive coatings using cardanol-based reactive diluent and its functionalized graphene oxide
  6. Deep learning assisted single particle tracking for automated correlation between diffusion and function
  7. A noninvasive fluorescence imaging-based platform measures 3D anisotropic extracellular diffusion
  8. Objective comparison of particle tracking methods
  9. Spatio-Temporal Analysis of Patient-Derived Organoid Videos Using Deep Learning for the Prediction of Drug Efficacy
  10. MultiOrg: A Multi-rater Organoid-detection Dataset
  11. Patient-derived organoids and orthotopic xenografts of primary and recurrent gliomas represent relevant patient avatars for precision oncology
  12. Fangchinoline Induced G1/S Arrest by Modulating Expression of p27, PCNA, and Cyclin D in Human Prostate Carcinoma Cancer PC3 Cells and Tumor Xenograft
  13. Uniform Tumor Spheroids on Surface-Optimized Microfluidic Biochips for Reproducible Drug Screening and Personalized Medicine
  14. Nanoparticle drug delivery systems for synergistic delivery of tumor therapy
  15. Heterogeneity-driven phenotypic plasticity and treatment response in branched-organoid models of pancreatic ductal adenocarcinoma