FDA Evaluation of Reagent Stability for In Vitro Diagnostics: Shelf-Life Expectations in Guidance and 510(k) Precedent

How FDA evaluates reagent stability for in vitro diagnostics: shelf-life expectations in guidance and 510(k) precedent

For an in vitro diagnostic (IVD), "stability" is not one claim but a family of claims: how long an unopened kit lasts (shelf-life), how long a reagent lasts once opened or loaded on an analyzer (in-use / on-board), how long a stored calibration holds, and whether performance survives shipping. FDA does not publish a single numeric shelf-life expectation. Instead, it expects each claim to be established experimentally, anchored to a recognized consensus standard, and supported by real-time data. This article summarizes the standard FDA leans on and how stability is actually presented and cleared in 510(k) and De Novo precedent.

The anchor standard: CLSI EP25

The central reference is CLSI EP25, "Evaluation of Stability of In Vitro Medical Laboratory Test Reagents." FDA recognizes the second edition as a consensus standard (recognition number 7-318, recognized in full, entered December 18, 2023) ¹, and it recognized the earlier approved guideline, EP25-A, under recognition number 7-235 ¹⁶⁷. In practice, submitters design and describe stability programs "in accordance with CLSI EP25," across chemistry, immunoassay, molecular, and allergy platforms ^{54649165169170}. Older submissions also cited neighboring references such as CEN 13640 (Stability Testing of In Vitro Diagnostic Reagents) and FDA's own calibrator- and quality-control-specific guidances for calibrator and control materials ⁴.

The practical consequence: FDA reviewers expect an EP25-style protocol with defined study conditions, timepoints, and acceptance criteria, and the review record typically states that "all protocols and acceptance criteria were reviewed and found acceptable" ¹⁰.

Real-time versus accelerated: the governing principle

The recurring principle across cleared submissions is that real-time data defines the shelf-life claim, while accelerated (elevated-temperature) studies establish or support an initial claim that real-time testing then confirms. Reviewers routinely accept a claim on accelerated data with the real-time study still ongoing at the time of clearance.

• The Pantex AM/PM Salivary Cortisol EIA set its 9-month expiration at 2 to 8 °C on the real-time study, supported by room-temperature data and a stress-chart (Arrhenius-type) prediction; the expiration was assigned from the shortest-lived reagent in the kit ⁷.
• For the Atellica CH Creatinine_2 assay, the 12-month shelf-life claim was established by an accelerated study, "a real time stability study to support a claim of 12 months shelf life is ongoing" ¹⁰.
• The Eon Cholesterol reagent's accelerated heat-stress data supported 21 months, but the initial claim was deliberately capped at 18 months pending real-time confirmation ³¹.
• Hycor's NOVEOS specific-IgE submissions repeatedly ran both an ongoing real-time and an accelerated study per CLSI EP25; accelerated data supported unopened shelf-life claims of 12 to 48 months for individual components, while available real-time data supported a shorter interim claim (for example 6 months at 2 to 8 °C) until the real-time study matured ¹²¹³¹⁴.

Where accelerated modeling is used explicitly, submitters apply an Arrhenius relationship. The Nova StatStrip control solutions were aged at 25, 40, and 50 °C, and shelf-life was predicted with the Arrhenius equation (roughly a doubling of stability per 10 °C decrease), so four months at 50 °C predicted 24 months at room temperature ⁴⁷. A caution FDA has flagged: accelerated predictions can over-assign shelf-life if the test matrix is wrong. In the OraQuick Ebola De Novo, whole blood was used as the testing matrix because it was "the limiting factor to correct shelf life assignment," and testing in clinical matrix was described as curtailing the risk that shelf-life is over-assigned ⁴³.

Study design elements FDA expects

Cleared submissions converge on a consistent design vocabulary that mirrors EP25:

• Multiple lots, typically three. Accelerated and real-time studies are run on three lots of the relevant components ^35114547132.
• Multiple timepoints, with margin beyond the claim. For open-vial studies, the Lyphochek Diabetes Control defined the study period as "at least 20% longer than the claimed open vial stability," tested at a minimum of three timepoints (T0, Tfinal, and Tfinal+20%), with a failure criterion of Tfinal drifting more than ±10% from T0 ⁵⁰.
• Testing across the measuring range with native samples. The Access Thyroglobulin real-time study tested native patient serum samples within the assay's measuring range at each timepoint ⁴⁶.
• Predefined acceptance/drift limits. Examples include control-recovery change limits of <10% or <15% from the Day 0 value ⁶⁵⁵⁰, applied consistently to each claim ¹⁰.

Unopened shelf-life claims

Unopened shelf-life durations in precedent vary widely by product chemistry and storage condition, and FDA expects the storage condition to be stated with the claim:

• Longer-dated materials: a bicarbonate calibrator carried a 13-month and 36-month closed/shelf-life depending on level ⁶; NOVEOS components ranged 12 to 48 months ¹⁴; a FilmArray GI Control panel supported a 36-month shelf-life at −25 to −15 °C from a real-time study run out to many months ¹³²; an Xprecia Prime PT/INR strip supported 24 months at 5 to 30 °C, tested at 5 °C and at 30 °C/75% RH across timepoints from 4.5 to 27 months ¹³⁴.
• Shorter-dated materials: rapid tests and some kits support 9 to 12 months, e.g., an iHealth Flu A&B/COVID-19 rapid test at up to 9 months (2 to 30 °C) ¹³⁷ and an Avantik viral transport medium at 12 months (2 to 25 °C) ¹³⁸.

In-use claims: open-vial and on-board stability

FDA treats "opened" performance as distinct claims that must each be supported. Precedent separates open-vial (bench) stability from on-board (analyzer-loaded) stability, often with different durations:

• A C-reactive protein kit for the SPAPLUS supported an unopened 5-month, open-vial 3-month, and on-board 30-day claim, each from real-time studies on multiple lots ⁷⁴.
• The Atellica CH Creatinine_2 supported 17 days on-board at 15 to 25 °C from an open-bottle study, alongside a separate open-vial/on-board calibrator claim of 48 hours ¹⁰.
• Elecsys CYFRA 21-1 reagent was shown stable for 12 weeks at 2 to 8 °C after first opening ⁷⁹; NOVEOS allergen capture reagents carried 28-day on-board and 15-day open-vial claims ⁵¹⁷.

Calibration stability

Calibration stability (how long a stored calibration curve remains valid) is a specific, separately supported claim, frequently framed as "calibration stability" or "lot calibration stability" and verified under EP25:

• Stored calibration curves have been cleared at 28 days (Free T3 calibrators, verified by assaying controls in replicate across day 0 to day 28) ¹⁰⁹, 30 days (ARK Lamotrigine calibration curve) ¹⁰³, and 19 days (ARK Methotrexate) ¹⁰⁶.
• On-board and lot calibration stability are handled as distinct sub-claims: the Elecsys CMV IgM system met an 8-day on-board calibration claim ⁷⁵, and the Elecsys CMV IgG met a 28-day lot calibration claim ⁸⁰.
• The Beckman DxC 700 submission ran an explicit "In-Use and Calibration Stability Verification (EP25-A)" to verify 30- and 90-day on-board reagent claims and a 1-day calibration stability claim ¹⁶⁹.

Shipping and transport stability

FDA expects evidence that a product survives realistic distribution stress, usually via a simulated-shipping or temperature-stress cycle applied before (or as part of) the stability study:

• The OSOM iFOB kit underwent a shipping stress test (8 hours at 50, 55, and 60 °C; 2 days at 45 °C) plus low-temperature and freeze/thaw exposure ⁵².
• Molecular kits pair closed-bottle (shelf-life) studies with open-bottle in-use and transport studies that include freeze-thaw, as in the ipsogen JAK2 De Novo ¹³⁰ and the Eonis SCID-SMA De Novo, where components were exposed to a defined worst-case shipping stress cycle before the in-use/on-board study ¹³⁵.
• Immunoassay reagents are often "transport-simulated" then tested for on-board stability, e.g., ImmunoCAP Tryptase calibrators held at 32 °C for a week, then on-board at 32 °C for 28 days ¹⁰⁴; the Access Thyroglobulin real-time study incorporated simulated summer and winter shipping conditions before storage ⁴⁶.
• Frozen products validate cold-chain shipping directly, as in the FilmArray GI Control panel's simulated 5-day dry-ice shipping study ¹³².

Practical takeaways for a submission

The precedent points to a consistent set of expectations a reviewer will look for:

1. Conform the protocol to CLSI EP25 (current recognized edition 7-318) and cite it ¹⁵⁴⁶.
2. Support the labeled shelf-life with real-time data; use accelerated data to establish an initial or interim claim, and disclose that the real-time study is ongoing if it is ^7103112.
3. Test multiple lots (typically three), multiple timepoints with margin beyond the claim, and native samples spanning the measuring range, against predefined recovery/drift acceptance criteria ^3504665.
4. Support each in-use claim separately: open-vial, on-board, calibration/curve, and reconstituted stability are distinct and each need data ^7479103109.
5. Demonstrate shipping/transport robustness through a stress or simulated-shipping study, including freeze-thaw where relevant ^52130135.

A note on scope: this overview synthesizes the FDA-recognized consensus standard for IVD reagent stability and a cross-section of cleared 510(k)/De Novo precedent spanning chemistry, immunoassay, molecular, allergy, and rapid-test products. Exact durations, acceptance limits, and study depth are product- and platform-specific, so the cleared examples above are illustrative of FDA's expectations rather than fixed thresholds.