Organoids and stem cells / Published data reanalysis
Read the function analysis
Relative spheroid response is not absolute function
These are source cell-line spheroids, not patient grafts. Donor, lot and circuit crosswalks are unavailable; the pairs cannot be counted as independent people.
# Functional response in a perfused pancreatic spheroid model AI-assisted reanalysis, 2026-10-19. This follows the checklist's organoid and engineered-tissue goal with actual published functional measurements. It replaces no earlier result and does not establish rejuvenation or organ replacement. ## What was acquired and checked We retrieved the primary XML and the unmodified **29,553-byte Source Data workbook** for [Quintard et al., 2024](https://doi.org/10.1038/s41467-024-45710-4). The workbook SHA-256 is pinned in the [analysis plan](analysis_plan.json). Read-only extraction from `Figure 7` retains 48 paired low/high insulin observations and their exact source-cell coordinates. Every recomputed high/low ratio matches the workbook's Figure 7d stimulation index exactly. The source workbook remained byte-identical after analysis. The source construct uses EndoC-βH5 cells with fibroblasts and endothelial cells. It is a pancreatic islet spheroid model, not a patient-islet graft or the blood-vessel organoid experiment elsewhere in the paper. Each Figure 7 point is described as a spheroid observation. Donor, manufacturing-lot and chip/circuit mappings are unavailable in this workbook. The methods mention duplicate assay measurements; duplicates do not add biological units. ## Results from measured source observations Stimulation index (SI) is high-glucose insulin concentration divided by the paired low-glucose concentration. Both concentrations use pg/mL. | Source condition | Reported observations | Mean SI | Median SI | Sample SD | Mean low insulin | Mean high insulin | | --- | ---: | ---: | ---: | ---: | ---: | ---: | | Conventional well | 22 | 1.442 | 1.254 | 0.359 | 27,523 | 38,346 | | Static, no surrounding vascular bed | 5 | 1.311 | 1.366 | 0.185 | 29,332 | 38,399 | | Static, with vascular bed | 5 | 1.470 | 1.572 | 0.347 | 41,553 | 59,364 | | Flow, no surrounding vascular bed | 5 | 1.398 | 1.243 | 0.356 | 21,354 | 29,278 | | Flow, with vascular bed | 11 | 2.631 | 2.725 | 0.903 | 16,169 | 43,668 | The combined flow/vascular-bed condition has the highest mean and median SI. Its mean SI exceeds the well condition by **1.189**, a **1.825-fold ratio of group means**. It exceeds the static/no-bed condition by **1.320**. The descriptive four-chip-condition contrast is +1.075 SI units, calculated as combined − flow-only − bed-only + neither. This is not a donor-independent or causal interaction estimate. Higher SI does **not** mean that absolute stimulated insulin is highest: the static/vascular-bed mean high-glucose concentration is larger than the combined condition's. The baseline concentrations also differ substantially. The combined condition's mean of individual SIs is 2.631, whereas the ratio of its group mean concentrations is 2.701. Those are different quantities; the former reproduces Figure 7d. Do not substitute one for the other. Across **81 single-observation omissions**, the combined-versus-well mean difference ranges +1.022 to +1.304 SI units. Every corresponding chip-control contrast also remains positive. These are influence checks on the reported assay observations, not leave-one-donor validation or confidence intervals. All rows, condition summaries and contrasts are retained in [derived outputs](derived/analysis_receipt.json). ## Source statistics and inference limits The Figure 7 caption describes unpaired two-sided t tests. The workbook's SI analysis includes ANOVA with Dunnett comparisons, and reports a Brown–Forsythe variance-test P value of 0.004. We preserve that discrepancy and the source outputs in [a separate context record](derived/source_statistics_context.json). We do not resolve it by silently selecting one test or by treating source P values as new confirmatory results. No donor/batch-adjusted inference is possible without the missing mapping. The numerical direction survives individual-row omissions, but unobserved manufacturing or cell-source differences could still explain some of it. This is exploratory source-data reanalysis selected after reading the paper, not an untouched validation cohort or a calibrated treatment prediction. ## What this advances for regeneration The data make **coupled transport and niche architecture** a concrete functional comparator. Comparing flow and surrounding endothelium separately and together can discriminate an interaction that a viability marker or vessel image alone would miss. SI, absolute secretion, viability and lineage identity must remain separate outcomes. The next evidence intake should obtain traceable donor/lot, spheroid and circuit identifiers plus paired perfusion, viability, identity and functional measurements. An independently replicated construct comparison can then test whether the functional difference persists at the correct biological unit. Do not infer those joint outcomes from the marginal SI result. For the dental programme, this supports evaluating perfused niches as one component of pulp/attachment constructs. It supplies no inserted PDL fibers, enamel/dentin recovery, tooth positioning, sensory integration or durability after support withdrawal. Those remain required in the [restoration contract](../../periodontal-attachment-restoration-2026-10-11/restoration_contract.json). Genome correction would add its own product/identity/function requirements; no genetic editing was evaluated in this source experiment. ## Reproduce and credit With the source workbook in the recorded local data directory and `openpyxl` available, run from the repository root: ```text python studies/checklist-followup-2026-10-19/organoid/analyze_islet_function.py python -m unittest discover -s tests -p test_islet_source_function.py -v ``` Eight tests check ratio reconstruction, units, incomplete pairs, measured zero versus missing values, observation counts and influence calculations. The [receipt](derived/analysis_receipt.json) pins source bytes, script, plan and outputs in canonical UTF-8/LF form. Raw workbook editing was not performed. Credit: **Quintard C. et al. (2024), A microfluidic platform integrating functional vascularized organoids-on-chip, Nature Communications 15:1452**, [DOI 10.1038/s41467-024-45710-4](https://doi.org/10.1038/s41467-024-45710-4). The article is [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). Source-derived numerical tables retain that attribution/license. Changes: extracted source-cell-linked pairs, recomputed SI, generated descriptive summaries and single-row influence analyses. No source images, diagrams, full text or complete workbook are redistributed in GitHub. Generic study code is separate from this source-data license.