Extractables and leachables studies answer a specific question: which chemical constituents can leave a device under defined laboratory or clinical conditions, and at what approximate quantity? They do not by themselves prove safety. They feed a toxicological risk assessment, which applies exposure estimates and uncertainty factors. ISO 10993-18 frames this as a stepwise exercise: identify materials of construction, characterize their constituents, estimate release under exaggerated extraction, and measure release under simulated clinical use. FDA's 2024 draft guidance on chemical analysis for biocompatibility warns that the recommendations are nonbinding and adapt to device contact type. A well-formed study therefore begins not with equipment, but with a documented rationale for why each extraction and analytical condition represents the worst case for the tissue contact at issue [1][2][3].

Where Extractables End and Leachables Begin

The distinction is operational, not merely semantic. An extractables study uses laboratory conditions chosen to obtain worst-case estimates of analytes to which tissue may be exposed. A leachables study measures chemical substances released under the device's clinical conditions of use. ISO 10993-18 separates the estimation of release potential under laboratory extraction from the measurement of clinical release, and the same document covers degradation products as part of the chemical characterization picture [2][2].

FDA's draft guidance describes a leachables or simulated-use study as a supplemental option that can refine the exposure estimate for a more accurate toxicological risk assessment. Such studies sit outside the scope of that draft guidance, and FDA suggests discussing the planned approach before study initiation through the Q-submission process. A manufacturer deciding between exhaustive extraction and simulated use is therefore choosing between conservative screening data and a more representative but harder-to-justify exposure refinement [1][1].

Extraction Design as a Source of Bias

Small preparation choices can change the extractables profile. FDA recommends test article preparation that mimics clinical preparation where relevant, because physical transfer of chemicals onto surfaces may otherwise be missed. Pre-rinsing test articles can remove residuals and should generally not be performed unless the instructions for use include pre-rinsing. Cutting devices may generate particles or expose inner components that would not be relevant to the biocompatibility evaluation [1][1].

FDA's September 2026 town hall on extractions added detail. Test article selection should account for manufacturing steps that produce the greatest quantity of residuals, such as cleaning detergent, or that alter materials to release more extractables, such as polymer heating during molding. For models offered in different sizes, the configuration with the largest surface area of each material becomes the test article to produce a worst-case profile. Coupons can be challenging to justify, though a homogeneous material delivered by volume may support alternatives [3][3].

Documenting an Extractables Table

For each solvent, identified and quantified or semi-quantified extractables above the analytical evaluation threshold should be tabulated with retention time, IUPAC name, CAS registry number when available, a structural descriptor or molecular image when a CAS number is absent, major ions observed, and the type of data used to establish identity. Tying each entry to this minimum set makes the later toxicological review auditable and lets a reviewer distinguish confident identifications from tentative structural assignments [1].

What Analytical Tooling Can and Cannot Standardize

Analysis choices follow the same burden of justification. For non-targeted analysis, a study published in 2025 compiled 106 reference standards of polymer additives and found no significant correlation between six physicochemical properties and the relative response factors measured by GC-MS and LC-MS. The authors selected standards for commercial availability, chemical relevance, toxicological coverage, occupational safety, frequency of use, and analytical compatibility within detectable and separable ranges [8][8].

CDRH's liquid-liquid extraction tool addresses a narrower decision: how pH, extract volume, organic solvent volume, and extraction iterations affect recovery of extractables from aqueous extracts. The model combines a modified Nernst distribution equation with validated structure-property relationships for distribution coefficient and pKa. Testing across nine analytes produced total root mean square error below 20%, but the supporting framework requires each laboratory to establish its own empirical recoveries before internal use [6][6].

Exhaustive Extraction and the Chronic Implant Caution

One peer-reviewed study collected explanted cardiac leads implanted for 3 to 130 months and analyzed exhaustive extracts by liquid chromatography with mass spectrometry. No new species formed over implant times exceeding a decade, and the quantity of each class of mobile species remained constant or decreased. Polyurethane oligomers stayed at constant levels, while the antioxidant Irganox 1076 decreased at a rate consistent with diffusional release rather than reactive consumption [7][7].

The authors interpret this as evidence that the assumption that all mobile species from an exhaustive extraction of a new device are released upon implant is extremely conservative. They also suggest additional end-of-life biocompatibility testing for biostable chronic implants may be unnecessary. Those conclusions address polymer components of the studied leads only and are based partly on internal, non-published reports for exposure estimates [7][7].

A Suggested Evaluation Workflow for a Hypothetical Device

Consider a hypothetical blood-contacting catheter with a polymeric shaft and a metal hub. A defensible study would first document the materials of construction, then select a test article size with the largest surface area for each material. Manufacturers should withhold pre-rinsing unless the labeling specifies it, and avoid cutting unless the clinical preparation requires sectioning. Extraction would follow a worst-case rationale tied to contact duration and tissue type [1][3].

For analysis, the team could tabulate extractables above the analytical evaluation threshold with retention time, chemical name, CAS registry number when available, structural descriptor, major ions, and the evidence used for identification. A blank solvent-only control separates analyte contributions from background. If a liquid-liquid step follows the aqueous extraction, the laboratory would record pH, volumes, and iterations, then establish its own empirical recoveries before interpreting any absence of response as absence of a species [1][6].

Replicates, Degradation, and the Limits of a Worst Case

Replication selection affects study timeline, cost, and validity. FDA's town hall distinguished analytical replicates, which assess variability in the analysis by injecting one extract three times, from extraction replicates, which assess variability in preparing the extract itself. A single extraction replicate may not represent the extractables profile for fixed tissues, certain in-situ polymerizing devices, or devices with historically variable extractables. For well-established materials, pooling multiple devices into one sufficient extract can sometimes replace triplicate extraction [5][5].

Aggressive conditions can also destroy the worst-case logic. FDA cautions that a large non-volatile residue alone does not indicate degradation. Destructive swelling may matter when it damages a device, for example silicone shredding from a metal lead during extraction, thereby exposing non-contacting portions. When a first solvent is incompatible, the suggestion is to evaluate at least two additional alternative solvents of similar polarity and maintain post-extraction photographs to document what actually happened [3][3].