Challenge

An orthopedic device manufacturer was preparing a Premarket Approval (PMA) submission to FDA for a cobalt-chromium/PEEK hybrid permanent implant and needed to revise their existing toxicological risk assessment (TRA) to align with the recently finalized ISO 10993-17:2023 revision. The prior submission package had been built on 2002-era analytical evaluation threshold (AET) assumptions and generic Permitted Daily Exposure (PDE) values, which FDA's Center for Devices and Radiological Health (CDRH) reviewers had flagged as insufficient given the extended (>30-day) patient contact duration and the high-risk permanent implant classification.

In their Additional Information (AI) letter, FDA specifically cited three concerns: (1) the absence of compound-specific Tolerable Intake (TI) values under the new ISO 10993-17:2023 framework; (2) the use of default uncertainty factors without documented justification; and (3) inadequate dose-duration adjustment for the device's permanent contact classification. The sponsor had 10 weeks to respond and could not afford to repeat the ISO 10993-18 extractables and leachables (E&L) testing — doing so would have added 16 weeks and ~$180K to the program and missed the commercial launch window.

The sponsor engaged Chemva to rebuild the TRA using the existing chemistry dataset but re-interpreted through the ISO 10993-17:2023 lens, producing a defensible, compound-specific TI derivation that would withstand FDA review on the first response.

What ISO 10993-17:2023 Changed

The 2023 revision of ISO 10993-17 (Biological evaluation of medical devices — Part 17: Toxicological risk assessment of medical device constituents) represents the most significant update to medical device TRA methodology in over two decades. Four changes materially affect how sponsors build and defend TRAs for medical devices in 2026 and beyond.

Tolerable Intake (TI) Replaces Generic PDE

The 2023 revision formalizes Tolerable Intake (TI) as the primary exposure benchmark for toxicologically significant constituents. Unlike the generic PDE values historically borrowed from pharmaceutical guidance, a TI is compound-specific and must be derived from a documented hierarchy of data sources — published NOAELs, regulatory-source PDEs, authoritative databases such as EPA IRIS or ATSDR, or (as a last resort) analog read-across with justified uncertainty adjustment.

Data-Quality Tiers

Annex D of ISO 10993-17:2023 defines three data-quality tiers for TI derivation: authoritative (regulator-published or peer-reviewed, preferred), recognized (established secondary sources), and provisional (read-across or in-silico (Q)SAR, acceptable only with justification). Submissions that relied on provisional data without tier-appropriate uncertainty factors are now routinely challenged by FDA, Health Canada, and EU Notified Bodies.

Structured Uncertainty Factor Taxonomy

The revision replaces ad-hoc safety factor application with a structured uncertainty factor (UF) taxonomy: UF-A (interspecies), UF-H (intraspecies), UF-S (subchronic-to-chronic), UF-L (LOAEL-to-NOAEL), and UF-D (data-quality modifier). Each factor must be justified against the underlying NOAEL dataset; composite UFs above 3,000 trigger additional scrutiny and typically require explicit regulator-facing justification.

Duration-Adjusted Cumulative Exposure

For long-term and permanent contact devices (greater than 30 days of cumulative exposure), ISO 10993-17:2023 requires exposure adjustment reflecting the cumulative dose delivered over the device's service life — a significant departure from the fixed-interval assumptions in the 2002 standard. For implants expected to remain in place for decades, this changes the shape of the safety argument materially.

Chemva's Solution

Chemva delivered a full ISO 10993-17:2023-aligned TRA, converting the legacy AET/PDE framework into the revision's TI-based methodology in a single integrated work package. The methodology is documented below.

1. Gap Analysis Against ISO 10993-17:2023

Chemva's toxicologists mapped every extractable and leachable identified in the prior ISO 10993-18 study against the 2023 revision's updated data-quality tiers and UF taxonomy. The review identified 42 unique constituents, of which 9 required re-derivation using compound-specific TIs, and 5 required a non-default UF adjustment based on the revised guidance. For the remaining constituents, the prior analysis remained valid but was re-formatted to the 2023 report template.

The gap analysis was captured in a traceability matrix — one row per constituent, columns tracking the prior basis (AET/PDE), the 2023 basis (TI and UF), and the rationale for any change. This matrix became the central exhibit in the FDA response and preempted the "where did each value come from" question that commonly follows a TRA methodology change.

2. Compound-Specific TI Derivation

For the 9 constituents requiring re-derivation, Chemva applied the Annex D hierarchy — prioritizing published NOAELs from peer-reviewed literature, regulatory-source PDEs (ICH Q3C, ICH Q3D, EMA monographs), and authoritative databases (EPA IRIS, ATSDR, EU CoRAP). For two constituents where no authoritative data existed, Chemva built a justified read-across using structurally analogous compounds with documented in-vivo toxicology data.

Each TI was derived using a documented UF composite — for example, a typical UF for a published rat subchronic NOAEL was UF-A (10) × UF-H (10) × UF-S (10) = 1,000. Read-across TIs received an additional UF-D of 3–10 per the 2023 guidance, reflecting the provisional data tier.

3. Margin of Safety (MoS) Recalculation

With compound-specific TIs established, Chemva recalculated the Margin of Safety (MoS) for each constituent using the clinically conservative exposure dose (full E&L-measured leachable concentration × worst-case patient exposure). All 42 constituents demonstrated MoS > 1, with detailed justification provided for two borderline impurities where the MoS fell between 1 and 3 — both were defended using additional read-across data and refined clinical exposure estimates.

Chemva also applied the new duration-adjusted exposure methodology, accounting for the device's permanent contact classification and cumulative release profile over the expected 25-year implant service life, using a kinetic leachable release model bounded by the existing ISO 10993-18 data.

4. FDA-Response Documentation

The final deliverable was a revised TRA report structured to FDA's preferred response format: a side-by-side comparison table (prior methodology vs. 2023-aligned methodology), a response narrative addressing each FDA review question point-by-point, and an executive summary suitable for direct reviewer consumption. Chemva also prepared an internal mock-review package — a simulated FDA reviewer's checklist marked against the revised submission — to stress-test the response before filing.

Impact Delivered

• Chemva's TI-based reassessment was accepted by FDA without a request for additional chemistry or biological testing, closing the Additional Information letter on first response.
• The sponsor avoided a repeat ISO 10993-18 E&L study, saving an estimated 16 weeks and ~$180K in program time and cost.
• PMA clearance was issued within the sponsor's original commercial timeline, preserving an estimated $4–6M in first-year device revenue that would have been lost to delay.
• The TRA framework Chemva established became the sponsor's internal template for three follow-on medical device submissions, compounding the ROI beyond the original engagement.

Key Takeaways

ISO 10993-17:2023 is not a cosmetic revision. Sponsors submitting medical device TRAs on pre-2023 methodology should expect FDA, Notified Body, and Health Canada reviewers to request re-derivation using the TI framework — particularly for implantable, long-term, or high-risk devices where compound-specific TIs and duration-adjusted exposure are material to the safety argument.

The good news: existing ISO 10993-18 extractables datasets remain usable. What changes is the downstream toxicological interpretation, which can be rebuilt on the original chemistry data without repeating the lab work — provided the E&L study was properly scoped (solvent bracketing, AET derivation, quantitative reporting) in the first place. Sponsors preparing submissions in 2026 should engage a toxicology partner experienced with the 2023 revision before finalizing their TRA, not after.