Device and Delivery System Stability: Integrating Extractables/Leachables with Time–Temperature Data
Regulatory Frame & Why This Matters
For combination products and advanced delivery systems—prefilled syringes, autoinjectors, on-body pumps, inhalers, IV sets—the question is no longer “do we have stability data?” but “do our extractables and leachables (E&L) controls and stability testing form a single, mechanistically consistent argument for quality and patient safety across the labeled lifecycle.” Classical drug-product stability programs are anchored in ICH Q1A(R2) principles (long-term/intermediate/accelerated conditions, significant change) and, where applicable, photostability under Q1B. That framework proves chemical and physical stability in time–temperature space. Delivery systems add another axis: the material and processing chemistry of the container–closure–device, where extractables (compounds released from materials under exaggerated conditions) define the universe of concern, and leachables (those actually migrating into the product under normal conditions) define real exposure. Regulators in the US/UK/EU will accept shelf-life and in-use claims only when these two lines of evidence converge: (1) compositionally plausible leachables are identified and qualified toxicologically, (2) sensitive, stability-stage methods actually measure them (or their worst-case surrogates) in the product across aging, and (3) device function and integrity (e.g., container-closure integrity, dose delivery mechanics) remain stable so that migration profiles and clinical performance do not shift late in life.
This integration matters operationally and scientifically. From an operational perspective, E&L and stability workstreams often live in different organizations (device development vs analytical development vs toxicology). If they are not synchronized, dossiers tend to show a perfect E&L study that is not reflected in stability methods, or pristine stability trends that measured everything except the compound toxicology flagged as a risk. Scientifically, migration is governed by polymer chemistry, additives (e.g., antioxidants, plasticizers, curing agents), lubricants (e.g., silicone oil in prefilled syringes), and process residues, all modulated by the product’s solvent system, pH, ionic strength, surfactants, and storage temperature. Without a unifying plan, teams can over-rely on exaggerated extractables profiles that are not thermodynamically relevant or, conversely, on long-term drug-product testing that lacks the sensitivity or specificity to see the low-ppm/ppb leachables that actually define patient exposure. The defensible posture is therefore to treat E&L as the source model and stability as the exposure measurement, with toxicology providing the acceptance rails that both must meet. When these pieces are aligned, reviewers see a coherent causal chain from material to molecule to patient, which is the standard for modern combination products.
Study Design & Acceptance Logic
Design begins with a simple mapping exercise that too many programs skip: list every wetted or vapor-contacting component in the delivery system (barrels, stoppers, plungers, O-rings, adhesives, inks, cannulas, bags, tubing, reservoirs, coatings, lubricants), assign material families and additives, and identify their interaction compartments with the drug product or diluent (e.g., long-term product contact in a prefilled syringe barrel; short, high-surface-area contact in an IV set during infusion; storage in an on-body pump cartridge). For each compartment, define three linked studies. (1) Controlled extractables using exaggerated, yet chemically meaningful conditions (solvent polarity ladder, high-temperature soaks, time), geared to reveal a comprehensive marker list and response factors. (2) Leachables-in-product stability—analytical methods at least as sensitive and selective as the extractables suite, run on real lots across long-term/intermediate/accelerated conditions, ideally using orthogonal LC/GC/MS approaches to track the specific marker set likely to migrate. (3) Function/integrity tracking—container-closure integrity (deterministic CCIT), dose delivery metrics, and mechanical/aging characteristics (e.g., break-loose/glide forces, pump flow curves) at the same timepoints to confirm that device aging does not open new migration pathways or change delivered dose.
Acceptance logic must be numeric and predeclared. For toxicological qualification, construct permitted daily exposure (PDE) or analytical evaluation thresholds (AET) per component of concern, considering worst-case dose and patient population. Translate these into batch-level acceptance criteria for the measured leachables in stability pulls (e.g., “Compound X ≤ A μg/mL at any timepoint; cumulative exposure ≤ B μg over the labeled use”). For compounds with structure alerts or genotoxic potential, adopt tighter thresholds and, when appropriate, conduct targeted spiking/recovery to prove method robustness around decision levels. For functionality, define device acceptance windows that reflect real clinical performance: dose accuracy and precision, priming success, occlusion detection, needle shield engagement, and any human-factor-critical behaviors. Then link these to leachables where plausible (e.g., plasticizer migration that could alter viscosity or surfactant efficiency, thereby affecting dose delivery). Finally, planning must account for in-use states (reconstitution or dilution, secondary containers like IV bags/tubing). Create a short in-use matrix—time and temperature brackets with the same leachables panel—so label statements (“use within X hours at Y °C”) rest on data for both product quality and leachables exposure, not on extrapolation.
Conditions, Chambers & Execution (ICH Zone-Aware)
Delivery systems piggyback on climatic zones but add unique stresses. Establish long-term storage at the labeled condition (e.g., 25/60 or 2–8 °C for liquids; 30/75 for certain markets), include intermediate when triggered per ICH Q1A(R2), and keep accelerated for mechanism reconnaissance, not expiry replacement. Overlay device-specific factors: (i) orientation (plunger-down vs plunger-up), which can alter lubricant pooling and effective contact surface; (ii) headspace oxygen control for oxidation-sensitive products; (iii) thermal gradients and freeze–thaw cycles for pumps and reservoirs; (iv) agitation/transport profiles for on-body or wearable systems that experience motion and vibration; and (v) light exposure for clear polymers, where photolysis of additives can generate secondary leachables. For inhalation devices, add humidity cycling and actuation stress; for IV sets, include clinically relevant flow rates and dwell times.
Execution rigor determines credibility. Use device-representative lots (materials, molding/cure conditions, silicone oil levels, sterilization modality and dose). Align stability pulls with CCIT and mechanical tests on the same aged units where feasible; if destructive testing prevents this, ensure statistically matched cohorts with clear traceability. For prefilled syringes, track silicone oil droplets and subvisible particles alongside leachables; a rise in droplets may confound or mask migration, and both can influence immunogenicity risk. For tubing and bags, ensure contact times and temperatures reflect realistic infusion scenarios; include priming/flush steps if clinically routine. Document actual ages (pull times) precisely, and preserve chain of custody, since migration is time–temperature-history dependent. When excursions occur (e.g., temporary high-temperature exposure), characterize their impact through targeted leachables checks and function tests; report how affected data were handled (included, excluded with rationale, or bracketed by sensitivity analysis). Zone awareness remains essential for market alignment, but the decisive question is whether the device–product system exposed to real stresses maintains both chemical/physical quality and safe leachables profiles throughout shelf life and in-use.
Analytics & Stability-Indicating Methods
Analytical strategy must connect the extractables library to stability monitoring. Begin with comprehensive profiling for extractables using orthogonal techniques—GC–MS for volatiles/semi-volatiles, LC–MS for non-volatiles and oligomers, and ICP–MS for elemental species. For each detected family (antioxidants such as Irgafos/Irgaflex derivatives; plasticizers like DEHP/DEHT; oligomeric cyclics from polyolefins or polyesters; silicone oil fragments; photoinitiators; residual monomers), curate marker compounds with reference standards where available. Develop targeted, validated LC–MS/MS and GC–MS methods for those markers in the actual drug-product matrix with adequate sensitivity to meet the AET. Establish specificity via accurate mass, qualifier ions/transitions, and retention time windowing; prove robustness by matrix-matched calibrations and isotope-dilution when practicable.
Stability-indicating here means two things. First, the methods must be capable of tracking change over time in the product (i.e., detect migration kinetics at relevant ppm/ppb levels across aging and in-use). Second, they must be able to discriminate leachables from product-related degradants and excipient breakdown products so trending is interpretable. Build an interference map early—forced degradation of the product and stress of excipients—so that candidate leachables are not misassigned. For silicone-lubricated systems, couple chemical assays with particle analytics (light obscuration, micro-flow imaging) to quantify droplets and morphology; tie these to chemical markers (e.g., cyclic siloxanes) to understand origin. Where trace metals are plausible leachables (e.g., needle cannula corrosion, catalysts), include ICP-MS with low blank burden and validated digestion/solubilization protocols. Finally, make data integrity visible: vendor-native raw files, version-locked processing methods, reintegration audit trails, and serialized evaluation objects so reviewers can reproduce targeted-quant results and trend overlays. The goal is not maximal assay count but a tight suite whose selectivity, sensitivity, and robustness map cleanly to the toxicological thresholds and to real-world exposure conditions.
Risk, Trending, OOT/OOS & Defensibility
Risk management should be designed into trending, not appended. Create a Leachables Risk Ladder that ranks markers by: (1) toxicological concern (genotoxic alerts, sensitizers), (2) likelihood of migration (partition coefficient, solubility, volatility, matrix affinity), and (3) analytical detectability. Assign monitoring intensity accordingly: high-risk markers receive lower reporting limits, tighter action thresholds, and more frequent checks at late anchors and in-use windows. For each marker, predefine decision rails: Reporting Threshold (RT), Identification Threshold (IT), Qualification Threshold (QT/PDE), and an internal action threshold below QT to trigger investigation before nearing patient-risk boundaries. Build trend cards that show concentration vs age with the PDE band overlaid, together with confidence intervals where applicable. These cards must coexist with classical quality attributes (assay, impurities, particulates) and device metrics so an executive can see, on one page, whether any migration trend threatens the claim or the label.
Define OOT/OOS logic in the same quantitative grammar as your thresholds. An OOT event is a confirmed upward inflection exceeding a predeclared slope or variance boundary yet still below QT; it should launch mechanism checks (batch-specific material lot? sterilization dose shift? silicone application drift? storage orientation?). OOS relative to QT/PDE demands immediate risk assessment: confirmatory re-measurement, exposure calculation at the maximum clinical dose, and an evaluation of device function/integrity (e.g., CCIT failure that increased ingress). Investigation outcomes must be numerical (“measured 0.9× AET with repeatability ≤ 10%; exposure at max dose = 0.6 × PDE”) and tie to control actions (tighten supplier specifications, adjust cure/flush, change lubricant deposition, add label safeguards). Defensibility rests on transparent math: timepoint concentration → per-dose exposure → daily exposure vs PDE → margin. Pair this with demonstrated method fitness (recoveries, matrix effects) so numbers are trusted. Where leachables are undetected, report quantified LOQs and exposure upper bounds; “ND” without context is weak evidence. This disciplined framing converts migration uncertainty into controlled, reviewer-friendly risk management.
Packaging/CCIT & Label Impact (When Applicable)
Container-closure integrity (CCI) and functional performance are not side notes; they determine whether migration pathways expand and whether dose delivery remains within claims. Use deterministic CCIT (vacuum decay, helium leak, HVLD) at initial and aged states, bracketed by extremes of orientation and storage condition. Present pass/fail with leak-rate distributions and tie any outliers to material or assembly variance. For prefilled syringes and cartridges, characterize silicone oil (deposition process, total load, droplet trends in product) because it intersects both E&L (chemical markers) and particles (SVP morphology), and can influence immunogenicity risk via protein adsorption/aggregation. For bags and sets, assess welds, ports, and seals—common ingress points that can also harbor unreacted monomers/oligomers.
Translate evidence to label language. For in-use holds (“stable for 24 h at 2–8 °C and 6 h at room temperature after dilution in 0.9% NaCl”), show that both quality attributes and leachables remain within acceptance for those conditions—ideally in the same table—so the sentence reads like a conclusion, not a convention. Where device mechanics matter (e.g., autoinjector priming, maximum allowed dwell before use), base instructions on aged-state tests that include leachables trending; do not assume functionality is invariant as materials age. For light-sensitive polymers, justify “store in the carton” when photolysis products were observed in extractables, even if not quantifiable as leachables under protected storage. Finally, align CCIT outcomes with microbiological integrity where sterility is relevant; a chemically safe but leaky system is not acceptable, and reviewers expect both lines of defense. A well-written label clause is simply the shortest path from your numbers to patient practice.
Operational Playbook & Templates
Make integration repeatable with a documented playbook. (1) Material & Process Ledger: a controlled bill of materials that lists polymers/elastomers/metals, additives, sterilization modality/dose, curing/aging conditions, and supplier change controls, each linked to extractables histories. (2) E&L–Stability Bridging Matrix: a table mapping each extractable family to the targeted leachables method(s), LOQ/AET, matrix, timepoints (including in-use), and toxicology owner; highlight “no method” gaps and resolve before pivotal builds. (3) Device Integrity & Function Plan: CCIT method and sampling, mechanical test battery, dose delivery accuracy/precision, and the schedule tied to stability pulls. (4) Toxicology Workbook: calculation templates for PDE/AET by clinical scenario, uncertainty factors, cumulative exposure logic, and decision trees for qualification (read-across vs specific tox studies). (5) Authoring Templates: one-page “Migration Summary” per marker family (trend figure with PDE band, table of max concentration and exposure vs PDE, method ID/LOQ, and action statement), and a “Function & Integrity Summary” (CCI pass rates, mechanical metrics, any drift, linkage to migration). These blocks slot directly into protocols, reports, and responses to regulator queries.
Execute with disciplined data governance. Pin data freezes and archive vendor-native raw files, processing methods, and evaluation objects so that trends and exposure calculations can be reproduced byte-for-byte. Establish cross-functional reviews at each major anchor (e.g., M6, M12, M24) where analytical, device, toxicology, and regulatory leads sign off on the integrated picture. Pre-approve deviation categories and laboratory invalidation rules for targeted leachables assays (e.g., matrix suppression beyond acceptance, qualifier transition failure) to avoid ad hoc retesting. For supply changes or material substitutions, run delta extractables studies with focused stability checks before implementation; treat device/material changes like CMC changes that can ripple into E&L and stability simultaneously. When the playbook is internalized, the organization produces consistent, defendable E&L-stability dossiers without last-minute reconciliation.
Common Pitfalls, Reviewer Pushbacks & Model Answers
Pitfall 1: Orphaned extractables libraries. Teams generate exhaustive extractables profiles but never translate them into validated, matrix-qualified targeted methods for stability. Model answer: “Here is the bridging matrix; targeted LC–MS/MS/GC–MS methods for markers A–F meet LOQs below AET; trends across M0–M36 show max exposure ≤ 0.3 × PDE.” Pitfall 2: AET mis-calculation. Using nominal dose instead of worst-case clinical exposure or failing to account for multiple device contacts leads to inappropriate thresholds. Model answer: “AETs derived from maximum labeled daily dose and multi-component contact; cumulative exposure across two syringes per day evaluated.” Pitfall 3: Ignoring in-use. Stability looks fine in vials but leachables appear during dilution/infusion. Model answer: “In-use matrix (PVC and non-PVC bags; standard sets) included; markers B and D measured ≤ 0.2 × PDE over 24 h at room temperature.” Pitfall 4: Device aging unlinked to chemistry. Function drifts (e.g., increased glide force) but chemical migration is not reassessed. Model answer: “Aged CCIT/mechanics run in lockstep with leachables; no increase in leak rate or marker concentrations at M36.” Pitfall 5: “ND” without context. Reporting “not detected” without LOQ and exposure bounds invites challenge. Model answer: “LOQ = 0.5 ng/mL; at maximum daily dose, exposure ≤ 0.05 × PDE.”
Expect reviewer questions in three clusters. “How were markers selected and tied to stability?” Answer with the bridging matrix and method IDs. “Are thresholds patient-relevant?” Show PDE/AET math for worst-case dose and population (pediatrics, chronic use), including uncertainty factors. “What about silicone oil and particles?” Provide joint chemical-particle evidence at aged states and any label mitigations (“do not shake”). Where genotoxic alerts exist, cite the most conservative threshold and confirm targeted detection at or below it. Always end with a decision sentence: “Max marker C at 36 months = 0.12 μg/mL (0.24 μg/dose; 0.08 × PDE); function/CCI unchanged; shelf life 24 months maintained; in-use 24 h at 2–8 °C/6 h RT supported.” Precision, not prose, closes reviews.
Lifecycle, Post-Approval Changes & Multi-Region Alignment
E&L–stability integration must persist through change. For material substitutions (new elastomer formulation, different syringe barrel polymer, alternate adhesives/inks), run targeted delta-extractables, update the marker panel, and execute a focused stability check on high-risk markers at late anchors and in-use. For process changes (sterilization dose/method, silicone deposition), confirm both chemical migration and device mechanics are unchanged or improved; if migration increases but remains below PDE, document margin and rationale. For presentation changes (vial → PFS, PFS → autoinjector), treat as new contact geometry and restart the mapping; do not assume read-across unless materials and contact modes are demonstrably equivalent. Across US/UK/EU, maintain one statistical and toxicological grammar—same PDE math, same AET derivation, same reporting format—so regional wrappers vary but the science does not. Divergent thresholds or marker lists by region signal process, not science, and attract queries.
Post-approval surveillance should include metrics that forecast risk: (i) max concentration as a fraction of PDE for each high-risk marker over time (aim to see stable or declining trends as suppliers mature); (ii) CCIT pass-rate stability; (iii) mechanical metric stability (glide force distribution, pump flow profiles); (iv) complaint signals that might reflect device–chemistry interactions (odor, discoloration, particulate spikes); and (v) change-control cycle time with evidence packs. When metrics drift, respond with engineering: supplier specification tightening, sterilization optimization, lubricant process control, or packaging geometry changes—paired with data that show the quantitative improvement in exposure or function. The target state is a portfolio where every device-enabled product has a living, testable link from materials to markers to migration to patient exposure and label, refreshed as the product evolves. That is how E&L ceases to be a separate report and becomes the chemical foundation of a stable, approvable delivery system.