IVD Analytical Performance Validation: LoD, LoQ, Precision and Stability
Protocol-level guide to IVD analytical validation, covering LoD, LoQ, precision, interference, cross-reactivity, stability, CLSI methods, and FDA/IVDR evidence.
What Is Analytical Performance Validation?
Analytical performance validation is the process of generating evidence that an IVD device correctly, reliably, and consistently detects or measures a particular analyte under defined conditions. It is the laboratory evidence foundation upon which clinical validity and regulatory approval rest. Under both FDA requirements and the EU IVDR (EU 2017/746), analytical validation is one of three pillars of performance evaluation, alongside scientific validity and clinical performance.
Under the IVDR, Article 56 and Annex XIII require manufacturers to establish a performance evaluation procedure that demonstrates conformity with the general safety and performance requirements (GSPRs) in Annex I. MDCG 2022-2 provides guidance on the GSPRs relevant to analytical performance, while MDCG 2025-5 (June 2025) clarifies the regulatory framework for IVD performance studies that generate analytical and clinical performance data.
For FDA submissions, analytical performance data is a required component of 510(k) and PMA applications, with specific expectations articulated in device-specific guidance documents and the eSTAR template.
Core Parameters of Analytical Performance
Parameter Definitions and Regulatory Alignment
| Parameter | Definition | CLSI Standard | IVDR Reference | FDA Expectation |
|---|---|---|---|---|
| Limit of Blank (LoB) | Highest measurement result from a blank sample | EP17-A2 | Annex XIII | Establish baseline noise |
| Limit of Detection (LoD) | Lowest analyte concentration reliably distinguished from blank | EP17-A2 | Annex XIII 1.2.1 | Required for all quantitative and qualitative IVDs |
| Limit of Quantitation (LoQ) | Lowest concentration quantifiable with acceptable precision and bias | EP17-A2 | Annex XIII 1.2.1 | Required for quantitative IVDs |
| Precision (repeatability) | Closeness of results under identical conditions | EP05-A3 | Annex XIII 1.2.2 | Within-run, between-day, between-lot |
| Precision (reproducibility) | Closeness of results across operators, instruments, sites | EP05-A3 | Annex XIII 1.2.2 | Multi-site recommended |
| Trueness (bias) | Closeness of measured value to reference value | EP09-A3 | Annex XIII 1.2.3 | Method comparison or reference material |
| Linearity | Proportional response across measuring interval | EP06-A | Annex XIII 1.2.4 | Required for quantitative assays |
| Analytical specificity | Ability to measure only the target analyte | EP07, MM09 | Annex XIII 1.2.5 | Interference and cross-reactivity |
| Measuring range | Interval of valid results | EP06, EP34 | Annex XIII 1.2.6 | Lower and upper limits defined |
| Stability | Performance over time under defined conditions | EP25-A3 | Annex XIII 1.2.7 | Shelf-life, in-use, transport, open-vial |
Limit of Detection (LoD) and Limit of Blank (LoB)
Conceptual Framework
Per CLSI EP17-A2, three detection capability parameters form a hierarchy:
LoB < LoD ≤ LoQ
LoB: The highest measurement result from a sample containing no analyte. Determined by testing blank samples and calculating the 95th percentile of the blank distribution.
LoD: The lowest analyte concentration reliably distinguished from the blank with stated probability. Calculated as:
LoD = LoB + 1.645 × SD(low concentration sample)
This accounts for both Type I error (false positive from blank) and Type II error (false negative from low concentration sample).
LoQ: The lowest concentration at which the analyte can be quantitatively determined with stated accuracy (commonly ≤20% CV).
LoD Study Design Protocol
| Study Phase | Design Element | Recommendation |
|---|---|---|
| Phase 1: LoB estimation | Blank samples | ≥4 blank matrix or independent negative specimens, ≥2 replicates each, over ≥3 days |
| Phase 2: Preliminary LoD | Low concentration samples | Serial dilutions of quantified positive specimen into negative matrix; 3 replicates per dilution |
| Phase 3: LoD confirmation | Near-LoD samples | Minimum 4 low-level positive specimens near estimated LoD; ≥2 replicates each over ≥3 days; ≥60 blank + ≥60 low-level replicates per lot |
| Phase 4: Verification | Confirm detection rate | 20 replicates on ≥2 lots; confirm ≥95% detection at LoD and <95% at one dilution below |
Sample Selection for LoD Studies
| Sample Type | When to Use | Considerations |
|---|---|---|
| Blank matrix (negative) | LoB estimation | Must match intended specimen type matrix |
| Recombinant or purified analyte | Spiking for low concentrations | Verify commutability |
| Clinical specimens near decision threshold | Most representative | May be difficult to source |
| Contrived specimens | Supplement when clinical samples unavailable | Document preparation method |
Precision Studies
CLSI EP05-A3 Study Design
The gold standard for precision evaluation follows a nested (hierarchical) design:
| Design Element | EP05-A3 Recommendation |
|---|---|
| Number of levels | ≥2 concentrations (near medical decision points) |
| Number of days | ≥20 days |
| Runs per day | 2 runs |
| Replicates per run | 2 replicates |
| Total replicates per level | ≥80 |
| Lots | ≥2 reagent lots (ideally 3) |
| Instruments | ≥1 (multi-instrument for reproducibility) |
Precision Components
| Component | Definition | Calculation |
|---|---|---|
| Repeatability (within-run) | Variation within a single run | SD of replicates within each run |
| Within-laboratory precision | Total variation within one laboratory (between-run + within-run) | Combined SD across runs and days |
| Reproducibility | Variation across laboratories, instruments, operators | Multi-site study; ANOVA decomposition |
Precision Acceptance Criteria Examples
| Assay Type | Level | Typical Repeatability CV | Typical Within-Lab CV |
|---|---|---|---|
| Immunoassay (high concentration) | >100 U/mL | ≤3% | ≤5% |
| Immunoassay (low concentration) | 5–20 U/mL | ≤5% | ≤8% |
| Molecular assay (viral load) | 1,000–10,000 copies/mL | ≤0.5 log | ≤1.0 log |
| Molecular assay (near LoD) | 3× LoD | ≤15% | ≤20% |
| Clinical chemistry | Within reference range | ≤2% | ≤3% |
| Hematology analyzer | Normal WBC | ≤3% | ≤5% |
EP15-A3 Verification Protocol (User-Level)
For laboratories verifying manufacturer claims, EP15-A3 provides a simpler protocol:
5 days, 5 replicates per day = 25 data points per level
2 or more sample materials at different concentrations
Verify that measured precision falls within manufacturer's claimed range
Interference and Cross-Reactivity
Interference Testing (CLSI EP07 / EP37)
Interference testing evaluates whether endogenous or exogenous substances cause systematic bias in test results.
| Interferent Category | Examples | Testing Approach |
|---|---|---|
| Endogenous substances | Bilirubin, hemoglobin, lipids, proteins, antibodies | Test at maximum expected physiological concentrations |
| Exogenous substances | Drugs, metabolites, food additives, supplements | Test at maximum expected therapeutic concentrations |
| Specimen additives | Anticoagulants, preservatives, stabilizers | Test at standard collection concentrations |
| Common disease states | Rheumatoid factor, heterophilic antibodies | Test with confirmed positive samples |
Interference Study Design
| Parameter | Recommendation |
|---|---|
| Interferent concentrations | Test at ≥2 concentrations: near clinical decision point and at maximum physiological level |
| Test concentrations | ≥2 analyte levels: one negative, one positive (near clinical decision point) |
| Replicates | ≥3 per condition |
| Acceptance criterion | Bias within predefined total allowable error (e.g., ≤10% for immunoassays) |
| Screening approach | Test panel of common interferents; investigate any that exceed acceptance criteria |
Cross-Reactivity Testing (CLSI MM09, MM26)
Cross-reactivity evaluates whether the assay detects non-target analytes that are structurally similar to the target.
| Parameter | Recommendation |
|---|---|
| Related organisms/analytes | Test all phylogenetically related species or structurally similar compounds |
| Concentration | Test at high physiological concentrations of cross-reactant |
| Target analyte | Test in presence and absence of target at near-LoD concentrations |
| Acceptance criterion | No false positive (target absent) or ≤20% bias (target present) |
Linearity and Measuring Range
Linearity (CLSI EP06-A)
| Design Element | Recommendation |
|---|---|
| Number of concentration levels | ≥5 (typically 9–11) equally spaced across claimed range |
| Dilution method | Serial dilution from high-concentration pool into blank matrix |
| Replicates | ≥2 per level |
| Direction | Test in both dilution directions |
| Analysis | Polynomial regression; assess deviation from linearity |
| Acceptance | Deviation at each level within clinical allowable error |
Measuring Range Confirmation
The measuring interval (formerly "reportable range") is validated through linearity studies combined with LoQ (lower limit) and saturation or hook effect studies (upper limit):
| Boundary | Determination Method |
|---|---|
| Lower limit | LoQ (not LoD — LoQ ensures acceptable precision at the low end) |
| Upper limit | Linearity plateau or antigen excess (hook effect) testing |
| Verification | Test patient specimens spanning the claimed range |
Trueness and Accuracy
Method Comparison (CLSI EP09-A3)
| Design Element | Recommendation |
|---|---|
| Reference method | FDA-cleared/approved predicate or reference measurement procedure |
| Sample size | ≥40 clinical specimens spanning the measuring range |
| Replicates | Single measurement per method (or duplicate if variability is a concern) |
| Analysis | Deming regression or Passing-Bablok regression; Bland-Altman bias plot |
| Acceptance | Bias within predefined total allowable error at medical decision levels |
Accuracy Hierarchy
| Approach | Preference | When to Use |
|---|---|---|
| Reference measurement procedure | Highest | When available for the analyte |
| FDA-cleared predicate comparison | Standard | Most common for 510(k) submissions |
| Reference material recovery | Supplemental | When clinical samples are limited |
| Proficiency testing samples | Supplemental | For verification of accuracy |
| Spiked recovery | Supplemental | For specific analyte recovery assessment |
Stability Studies
Types of Stability
| Stability Type | Definition | Typical Duration |
|---|---|---|
| Shelf-life (real-time) | Performance from manufacture to expiry | 12–24 months |
| Accelerated stability | Prediction of shelf-life using elevated temperatures | 1–6 months (extrapolated) |
| In-use stability | Performance after first opening or preparation | Hours to weeks |
| On-board stability | Performance while loaded on instrument | Days to weeks |
| Transport stability | Performance after simulated shipping conditions | Per ASTM D4169 |
| Specimen stability | Analyte stability in collected specimen | Hours to days |
Stability Study Design (CLSI EP25-A3)
| Parameter | Recommendation |
|---|---|
| Time points | ≥3 time points for accelerated; ≥5 for real-time |
| Storage conditions | Claimed storage conditions ± stress conditions |
| Test parameters | LoD, precision, accuracy at each time point |
| Acceptance | No statistically significant degradation; performance within claims |
| Lots | ≥3 lots recommended |
Complete Analytical Validation Test Plan
Template: Study Matrix
| Study | CLSI Standard | Samples Required | Sites | Lots | Estimated Duration | Primary Endpoint |
|---|---|---|---|---|---|---|
| LoB / LoD / LoQ | EP17-A2 | 60 blank + 60 low-level per lot | 1 | 2–3 | 4–6 weeks | LoD confirmed at ≥95% detection |
| Precision (repeatability) | EP05-A3 | 2 levels × 80 replicates | 1 | 2–3 | 4–5 weeks | CV within predefined limits |
| Precision (reproducibility) | EP05-A3 | 2 levels × 80 replicates | 3 | 2–3 | 6–8 weeks | Reproducibility CV within limits |
| Linearity | EP06-A | 9–11 levels × 2 replicates | 1 | 1 | 1–2 weeks | Linear across claimed range |
| Trueness (method comparison) | EP09-A3 | ≥40 clinical specimens | 1–3 | 1 | 2–4 weeks | Bias within total allowable error |
| Interference screening | EP07 | ≥20 interferents × 2 analyte levels | 1 | 1 | 2–4 weeks | Bias < acceptance limit |
| Cross-reactivity | MM09 / MM26 | ≥10 related analytes | 1 | 1 | 2–3 weeks | No false positive; bias <20% |
| Stability (real-time) | EP25-A3 | 3 levels × 3 lots × ≥5 time points | 1 | 3 | 12–24 months | Performance within claims |
| Stability (accelerated) | EP25-A3 | 3 levels × 3 lots × ≥3 time points | 1 | 3 | 1–6 months | Performance within claims |
Estimated Total Timeline and Resources
| Parameter | Estimate |
|---|---|
| Total study duration | 6–12 months (excluding long-term stability) |
| Number of studies | 8–12 distinct study protocols |
| Clinical specimens required | 200–500+ (varies by assay complexity) |
| Estimated cost | $150,000–$500,000 depending on assay complexity |
| Regulatory writing | 4–8 weeks for Analytical Performance Report |
FDA vs. EU IVDR Expectations
Comparison of Requirements
| Aspect | FDA (510(k)/PMA) | EU IVDR |
|---|---|---|
| Performance evaluation plan | Implied by design control requirements | Mandatory: Article 56(4), Annex XIII |
| Analytical performance report | Included in 510(k) submission (typically Section 12–14 of eSTAR) | Standalone APR referenced in PER |
| CLSI standards | Expected but not mandatory | Recognized as state of the art; ISO 18113 series |
| Number of lots | ≥2 recommended; 3 for critical parameters | Minimum per risk assessment |
| Multi-site requirement | Recommended for reproducibility | Required for Class C/D; risk-based for Class A/B |
| Software analytical validation | IEC 62304 + FDA software guidance | IEC 62304 + IVDR Article 17(2) |
| Post-market analytical performance | Per post-market study protocol | PMPF (post-market performance follow-up) required per Article 78 |
Common Pitfalls and Strategic Solutions
| Pitfall | Impact | Solution |
|---|---|---|
| Ambiguous intended use claims | Studies misaligned with claims; weak evidence | Define intended use precisely before designing studies |
| Insufficient statistical power | Inconclusive results; study redo | Pre-define sample sizes using power analysis |
| Uncontrolled pre-analytical variables | Results not reproducible | Validate collection, transport, storage conditions |
| Software validation gaps | Algorithm changes invalidate analytical data | Validate software under IEC 62304 before analytical studies |
| Inadequate traceability | Auditors cannot follow evidence chain | Link raw data → APR → PER with document control |
| Using only contrived specimens | Regulators question clinical relevance | Supplement with fresh clinical specimens wherever possible |
| Lot-to-lot variability not assessed | Performance claims not generalizable | Use ≥2 reagent lots (ideally 3) |
| Ignoring lifecycle management | Changes trigger revalidation | Maintain change control linked to PMS/PMPF data |
Checklist: Analytical Validation Readiness
Before Starting Studies
Intended use and performance claims precisely defined
Risk analysis completed (ISO 14971) identifying critical performance parameters
CLSI standards identified for each study type
Acceptance criteria pre-defined for each parameter
Sample sourcing plan documented (clinical, contrived, reference materials)
Test sites qualified
Study protocols reviewed and approved by regulatory affairs
Statistical analysis plan documented
Software frozen (IEC 62304) before study initiation
During Studies
Deviations documented and assessed for impact
Raw data secured with audit trail
Interim analyses performed per protocol
Lot-to-lot comparison tracked
Any out-of-specification results investigated
After Studies
Analytical Performance Report (APR) drafted
All raw data traceable to APR conclusions
APR reviewed against IVDR Annex XIII or FDA eSTAR requirements
Gap analysis: APR completeness vs. regulatory checklist
APR integrated into Performance Evaluation Report (PER) for IVDR
Stability studies ongoing; interim data available
Key Takeaways
Analytical validation is the foundation of regulatory evidence. Without robust analytical data, clinical studies and regulatory submissions lack credibility.
CLSI standards are the international language of analytical validation. Aligning study designs with EP05, EP17, EP07, EP09, EP25 and related standards ensures global regulatory acceptance.
Pre-define acceptance criteria before testing. Regulatory reviewers expect to see acceptance criteria established before study execution, not retrofitted to results.
Use at least 2 reagent lots. Lot-to-lot variability is a common audit finding and regulatory request.
Fresh clinical specimens are preferred. Contrived specimens supplement but do not replace clinical specimen testing for key parameters.
Software must be frozen before analytical studies begin. Changes to algorithms after study initiation can invalidate all analytical data.
The IVDR requires a structured performance evaluation plan. Article 56 and Annex XIII mandate a documented, lifecycle approach to analytical performance that links to PMS and PMPF activities.
Sources
CLSI EP05-A3. "Evaluation of Precision of Quantitative Measurement Procedures." 2014.
CLSI EP17-A2. "Evaluation of Detection Capability for Clinical Laboratory Measurement Procedures." 2012.
CLSI EP07. "Interference Testing in Clinical Chemistry."
CLSI EP09-A3. "Measurement Procedure Comparison and Bias Estimation Using Patient Samples."
CLSI EP25-A3. "Evaluation of Stability of In Vitro Diagnostic Reagents."
CLSI EP06-A. "Evaluation of Linearity of Quantitative Measurement Procedures."
CLSI MM09 / MM26. "Evaluation of Cross-Reactivity."
EU IVDR, Regulation (EU) 2017/746, Articles 56–57, Annex XIII.
MDCG 2022-2. "Guidance on General Safety and Performance Requirements."
MDCG 2025-5. "Questions & Answers Regarding Performance Studies of IVD Medical Devices." June 2025.
MDx CRO. "IVD Analytical Validation under IVDR: Principles and Best Practices." September 2025.
FIND Diagnostics. "Verification and Validation — Analytical Performance Study Design." 2025.
SGS. "Demonstrating Analytical Performance of IVD Devices." Webinar, 2025.
FDA. "eSTAR Electronic Submission Template for 510(k) Submissions."