Chapter 7
Phase 4 of WIA-BIO-006 is where a tissue-engineered construct stops being a laboratory result and becomes something a regulator, a clinician, and a quality unit can all accept. Scaffold design, bioprinting, bioreactor conditioning, vascularization, and cell seeding are all upstream of this point. Phase 4 asks whether the finished construct can be shown, batch by batch, to be sterile, viable, mechanically adequate, biocompatible, and functionally tissue-like. It also asks whether the evidence can be organised so that FDA, EMA, KFDA, PMDA, and NMPA reviewers can follow it.
The sections below follow that path. They begin with the release tests that gate every batch. They then cover the assay methods and acceptance thresholds, the regulatory pathways that determine how much evidence a product needs, and the tissue-specific functional readouts. They close with manufacturing integration and a look at where the field is heading.
Every batch passes a sterility gate before any other result matters. The specification lists three methods and one frequency: every batch. Release testing is not a sampling exercise performed once per process validation.
| Test | Method | Criterion or condition |
|---|---|---|
| Sterility | USP <71> | 14-day incubation in media |
| Endotoxin | LAL assay (USP <85>) | <0.5 EU/mL |
| Mycoplasma | PCR or culture (USP <63> / European Pharmacopoeia 2.6.7) | Performed on every batch |
These three tests detect different failures. Bacterial and fungal contamination shows up in the sterility incubation. Endotoxin is a cell-wall residue that can persist after the organisms that shed it are gone, so a construct can pass sterility and still fail the LAL assay. Mycoplasma is the awkward case for cell-based products. It is invisible to routine microscopy and unaffected by the antibiotics commonly used in culture, which is why the specification gives it its own PCR or culture test.
The 14-day incubation also shapes logistics. A product with a short shelf life may reach the patient before the sterility result is final. In that case the release record has to show what was known at the time of shipment, and the certificate of conformance has to reflect that. For cell products the specification extends the panel beyond these three tests. It adds identity (STR profiling for human cells), purity, and potency, each with a documented acceptance criterion and a release-test record.
Terminal sterilization also has to be validated in its own right. The specification points to ISO 11737 for microbiological methods and ISO 11135 for ethylene oxide. Where terminal sterilization would destroy the living component, aseptic-processing validation per the FDA 2004 guidance for sterile drug products applies instead. That choice between terminal and aseptic routes follows directly from the fact that the product contains cells.
The technical core of Phase 4 is a set of assays with explicit protocols and numeric acceptance values. The values matter because they turn a subjective impression ("the cells look healthy") into a pass/fail record that can be audited later.
Viability. Two complementary assays are specified. Live/dead staining uses Calcein AM, which marks live cells green, and ethidium homodimer-1, which marks dead cells red. The protocol is a 15–30 minute incubation, fluorescence imaging, and quantification in ImageJ, with acceptance at >80% viability. Because it counts individual cells, it also shows where dead cells cluster, for example in the core of a thick scaffold. The MTT/MTS assay measures metabolic activity instead. It involves a 1–4 hour incubation with reagent and an absorbance reading between 490 and 570 nm. It reports a population-level number and no spatial information, which is why the two are used together.
Viability (%) = (A_sample / A_control) × 100%
FC = 2^(-ΔΔCt) # qRT-PCR fold change
Mechanical testing. Compression follows ASTM D1621, with a strain rate of 0.5–1 mm/min, a preload of 0.01–0.1 N, and deformation to 10–50%. The outputs are Young's modulus (E), ultimate strength (σ_max), and strain at failure (ε_f). The regulatory cross-walk additionally lists ASTM F2150 and ISO 13314 for compression. Tensile testing follows ASTM D638 on dumbbell-shaped samples, with a gauge length of 10–50 mm and a strain rate of 1–10 mm/min. Reporting the full strain-rate and preload window, and not just the modulus, lets a reader reproduce the test. Hydrated polymer and hydrogel constructs are rate-sensitive, so a modulus without its test conditions is nearly meaningless.
Biocompatibility. The tests are risk-driven. ISO 10993-1:2018 selects the test matrix according to device contact category (surface-contacting versus implanted) and contact duration (limited, prolonged, or permanent).
| Endpoint | Protocol summary | Acceptance |
|---|---|---|
| Cytotoxicity (ISO 10993-5) | Incubate material in medium for 24 h at 37 °C, expose cells to the extract, measure viability by MTT or LDH | >70% viability relative to control |
| Immunogenicity | Cytokine release (IL-1β, IL-6, TNF-α), T-cell proliferation, complement activation (C3a, C5a) | <2-fold increase versus control |
| In vivo implantation (ISO 10993-6) | Subcutaneous implantation for 1–12 weeks; histology with H&E and Masson's trichrome | Score ≤1 on a 0–4 scale |
The histology scale runs from 0 (no reaction) and 1 (minimal reaction) up through 2–4 for increasing severity. Note that the in vitro viability threshold for the extract test (>70%) is deliberately lower than the one for cells on the construct (>80%). The extract test asks whether leachables are harmful. The live/dead test asks whether the finished product supports the cells it carries.
The degradation rates of the common materials show why one test duration does not fit every scaffold:
| Material | Degradation | Biocompatibility |
|---|---|---|
| Gelatin | 1–4 weeks | Excellent |
| Collagen I | 2–8 weeks | Excellent |
| Alginate | 2–8 weeks | Good |
| PLGA | 1–6 months | Good |
| PLA | 6–12 months | Good |
| PCL | 6–24 months | Good |
A gelatin construct may largely degrade within the window of a 1–12 week implantation study. A PCL scaffold will still be mostly intact at its end. Test planning therefore has to consider whether the chosen observation window captures the material's degradation products and the tissue response to them, not only the initial implantation response.
For tissue-engineered products, market access is governed less by price or demand than by regulatory classification. The first decision a developer makes, whether the product is regulated as a device or as a biologic, determines the evidence burden, the timeline, and the jurisdictions that are practical to enter. Phase 4 therefore treats the classification rationale as a first-class part of the record.
The specification describes two overlapping frameworks. The device framework places most tissue-engineered products in FDA Class II, with a 510(k) premarket notification, GMP, and the Quality System Regulation. High-risk products fall into Class III, which requires Premarket Approval and Phase I–III clinical trials. The cellular-product framework follows the HCT/P decision tree under 21 CFR 1271. Section 361 products are minimally manipulated and used homologously, and they carry reduced regulatory burden. Section 351 products require a BLA filing supported by phase I/II/III clinical data. Which framework governs depends on the product's composition and intended claim, and a combination product may touch both.
The regulatory envelope in the standard records three things for each product:
Equivalent envelopes track EMA's four ATMP categories (gene therapy, somatic-cell therapy, tissue-engineered product, and combined ATMP) and submissions to KFDA, PMDA, and NMPA. Capturing the rationale once, in a structured form, avoids re-deriving it for each jurisdiction.
The preclinical sequence is the same regardless of pathway:
Clinical trials then proceed through Phase I (safety, 10–30 patients), Phase II (efficacy, 30–100), and Phase III (confirmatory, 100–1000). Primary endpoints are safety and efficacy. Secondary endpoints are quality of life and cost-effectiveness.
Tissue-specific functional assays show whether a construct does what its target tissue does, which viability and biocompatibility alone cannot establish. The specification defines readouts for three representative tissues and a gene-expression method that works across them.
| Tissue | Readouts |
|---|---|
| Bone | Alkaline phosphatase (ALP); osteocalcin (OCN); calcium deposition (Alizarin Red) |
| Cartilage | Glycosaminoglycans (GAG, Alcian Blue); collagen type II (immunostaining); aggrecan (Western blot) |
| Liver | Albumin secretion (ELISA); urea synthesis (colorimetric); CYP450 activity (luminescent) |
The bone panel moves from an early osteoblast enzyme (ALP) to a later matrix protein (OCN) to mineral deposition. This lets a reviewer see whether differentiation progressed along the expected course. The cartilage panel pairs a matrix-quantity measure (GAG) with a matrix-quality one, since collagen type II is the hyaline-cartilage collagen. The liver panel is explicitly functional: albumin, urea, and CYP450 are things a hepatocyte does, and each one is easy to lose in culture.
For gene expression, the specified qRT-PCR workflow is:
Fold change is then computed as 2^(-ΔΔCt), as in the formulas above. Normalizing to a housekeeping gene is essential, because RNA input varies from sample to sample, especially when it is extracted from dense, scaffold-bound constructs.
Two worked examples in the specification show how upstream design parameters feed these downstream tests. Both come directly from the appendix:
Scaffold porosity (PCL, material density 1.15 g/cm³, apparent density 0.29 g/cm³)
P = (1 - 0.29/1.15) × 100%
P = (1 - 0.252) × 100%
P = 74.8% → ~75%, suitable for cartilage tissue
Cell seeding (scaffold volume 1.0 cm³, target density 5 × 10⁷ cells/cm³)
N_cells = D × V = 5 × 10⁷ × 1.0 = 5 × 10⁷ cells
Suspension: 5 × 10⁶ cells/mL → 10 mL needed
A construct that fails the cartilage panel can be traced back to these numbers. Was the porosity right, and was the seeding density actually achieved? That traceability is one reason the calculations are part of the record and not left in a notebook.
The specification includes an informative cross-walk that maps each concern to its governing standard. Implementers are expected to obtain authoritative copies from the issuing bodies. It is not a substitute for them.
| Concern | Standard |
|---|---|
| Cytotoxicity | ISO 10993-5 |
| Implantation | ISO 10993-6 |
| Sensitization | ISO 10993-10 |
| Finished-product sterility | USP <71> |
| Endotoxin | USP <85> / LAL |
| Mycoplasma | USP <63> / Ph. Eur. 2.6.7 |
| Compression | ASTM F2150 / ISO 13314 |
| Tensile | ASTM D638 |
| Quality management | ISO 13485:2016 |
| FDA HCT/P | 21 CFR 1271 |
| EU ATMP | EC No 1394/2007 |
| Korea, biological products | KFDA Notification 2021-92 |
The table also shows an alignment issue. The in-body compression protocol cites ASTM D1621, while the cross-walk cites ASTM F2150 and ISO 13314. A laboratory should record which standard it actually followed for a given test report, because auditors will compare the report against the cited method.
Manufacturing integration. GMP requires validated processes, SOPs, batch records, in-process testing, and a controlled environment (ISO 5–7 cleanroom). The integration layer adds more detail. Cleanrooms are qualified under ISO 14644, with Class A laminar flow inside a Class B background and environmental monitoring per EU GMP Annex 1. Process performance qualification follows ICH Q7. Batch records, donor-to-construct traceability, an in-process sampling plan, and a certificate of conformance for each shipment complete the picture. Manufacturers map their internal MES or PLM identifiers onto the WIA identifiers scaffoldId, bioinkId, and constructId. Downstream regulators then inherit a provenance graph that runs from donor material through to the finished construct.
Staged adoption. A first implementation typically starts from the reference container with mock data. It then runs the companion conformance suite, replaces mock backends one endpoint at a time, and onboards trusted peers before going to production. Within the 1.x line, existing envelopes and endpoints remain reachable. Hosts may add optional fields but must not remove existing ones.
The roadmap in the specification names several active research tracks, each of which will stress today's release-testing model:
In-situ bioprinting is the sharpest challenge. A 14-day sterility incubation cannot gate a product that is manufactured inside the patient's operating field. Patient-derived and gene-edited cells challenge the batch concept itself, since a "batch" may be a single patient. The standard's proposals endpoint, POST /standards/v1/proposals, tracks such changes through the committee voting process, so the test matrix can change without breaking existing records.
Phase 4 turns the upstream tissue-engineering workflow into a releasable, reviewable product. Sterility, endotoxin, and mycoplasma tests gate every batch. Viability, mechanical, cytotoxicity, immunogenicity, and implantation results are held to numeric thresholds: >80% live cells, >70% extract viability, <2-fold immune response, and a histology score of ≤1. Tissue-specific panels for bone, cartilage, and liver, together with qRT-PCR, show that a construct behaves like its target tissue. Classification under the device and HCT/P routes, and under the ATMP categories in Europe, determines how much clinical evidence is required. GMP integration, with scaffold, bioink, and construct identifiers carried through to the batch record, makes that evidence traceable.
Two lessons recur. Acceptance criteria are only meaningful alongside their test conditions, and test durations must be matched to how fast the material degrades. The natural next step is to ask how an organisation puts this release-and-regulatory structure into everyday practice and keeps it current as the research tracks above mature. The next chapter takes up that question.