2.3 Method Verification, Validation Protocols & CLSI Standards

Method Verification, Validation Protocols & CLSI Standards

Regulatory Mandates: Method Verification vs. Method Validation

Under the Clinical Laboratory Improvement Amendments of 1988 (CLIA '88), codified at 42 CFR § 493.1253, a clinical laboratory must demonstrate and document analytical performance before introducing any new test system or releasing patient results. Federal regulations draw an absolute legal and technical distinction between Method Verification and Method Validation.

1. Method Verification (42 CFR § 493.1253(b)(1))

Method verification applies to unmodified, FDA-cleared or approved test systems of moderate or high complexity. Because the diagnostic manufacturer has already established extensive clinical and analytical performance claims during the FDA premarket review (510(k) clearance or Premarket Approval [PMA]), the laboratory is not required to re-establish these baseline claims from scratch.

Instead, the laboratory must verify that it can reliably replicate the manufacturer's published performance specifications within its own physical environment, using its own personnel, reagents, calibrators, and patient matrices. Four core analytical parameters are legally mandated for verification:

1. Accuracy (Trueness)
2. Precision (Imprecision)
3. Reportable Range (Analytical Measurement Range - AMR)
4. Reference Intervals (Normal Ranges) applicable to the laboratory's patient population

2. Method Validation (42 CFR § 493.1253(b)(2))

Method validation is significantly more rigorous and applies to:

• Laboratory-Developed Tests (LDTs): Assays designed, developed, and validated in-house by the clinical laboratory (e.g., custom mass spectrometry panels or homebrew molecular assays).
• Modified FDA-Cleared/Approved Tests: Any FDA-cleared test altered by the laboratory in a manner that deviates from the package insert. Common modifications include using an unapproved specimen matrix (e.g., cerebrospinal fluid or bronchoalveolar lavage on an assay cleared only for serum/plasma), changing sample incubation times or temperatures, utilizing non-cleared collection tubes, altering reagent ratios, or employing manual dilution protocols not specified by the manufacturer.
• Non-FDA Cleared Tests: Assays labeled as Research Use Only (RUO) or Investigational Use Only (IUO) adapted for clinical diagnostic reporting.

For validated methods, the laboratory director must establish all analytical and clinical performance specifications de novo. In addition to accuracy, precision, reportable range, and reference intervals, full validation mandates establishing:

• Analytical Sensitivity: Limit of Blank (LoB), Limit of Detection (LoD), and Limit of Quantitation (LoQ).
• Analytical Specificity: Cross-reactivity, endogenous interferences (hemolysis, icterus, lipemia), and exogenous interferences (common drugs, preservatives, matrix effects).
• Clinical Sensitivity and Clinical Specificity: True positive and true negative diagnostic rates in confirmed clinical populations.
• Diagnostic Predictive Values: Positive Predictive Value (PPV) and Negative Predictive Value (NPV) across relevant disease prevalence settings.

---

Analytical Verification Parameters and CLSI Protocols

Accuracy (Trueness) & CLSI EP09

Accuracy evaluates the closeness of agreement between the measured laboratory value and the true or accepted reference value. For method verification of an unmodified test, accuracy is evaluated using one or more of the following approaches:

• Method Comparison (Patient Samples): The gold standard per CLSI EP09c. The laboratory tests a minimum of 20 to 40 patient specimens (minimum 40 recommended by CLSI; 20 permitted for simple verification) spanning the entire analytical measurement range in split-sample fashion against an existing validated comparative method or reference laboratory.
• Proficiency Testing / Certified Reference Materials: Testing a minimum of 5 distinct samples with known, peer-reviewed target values (e.g., CAP survey samples, NIST standard reference materials).

Statistical Analysis: Deming vs. Passing-Bablok Regression

Standard Ordinary Least Squares (OLS) linear regression is statistically invalid for method comparisons because OLS assumes that the independent variable (X, comparative method) is measured without error. In clinical laboratory medicine, both methods exhibit measurement error. Therefore, laboratories must use error-in-both-variables models:

• Deming Regression: Parametric regression that accounts for measurement error in both the comparative (X) and candidate (Y) methods. It assumes a constant ratio of analytical variances (λ = s_y² / s_x²) and normally distributed error.
• Passing-Bablok Regression: Non-parametric linear regression based on the median of all pairwise slopes between data points. Passing-Bablok is completely robust against analytical outliers and does not assume a Gaussian distribution of errors.
• Slope and Intercept Interpretation: A proportional bias is indicated if the 95% confidence interval for the slope (β) does not include 1.00. A constant systematic bias is indicated if the 95% confidence interval for the Y-intercept (α) does not include 0.00.

Precision (Imprecision) & CLSI EP05

Precision reflects the closeness of agreement between independent test results obtained under stipulated experimental conditions, quantifying random analytical error. Precision is categorized into two levels:

1. Repeatability (Within-Run Precision, s_r): Closeness of agreement between results obtained with the same method on identical test items in the same laboratory, by the same operator, using the same equipment, within a short interval of time.
2. Intermediate Precision (Within-Laboratory Precision, s_WL): Precision under conditions where test results are obtained across multiple days, different operators, multiple calibrations, and diverse reagent lots.

Experimental Design per CLSI EP05

• Full 20-Day Protocol (Method Validation): The assay evaluates at least two levels of control material (or patient pools) across 20 operating days, running 2 separate runs per day, with 2 replicate measurements per run (20 × 2 × 2 = 80 total data points per concentration level). Analysis of Variance (ANOVA) separates within-run variance from day-to-day and run-to-run variance.
• Modified 5-Day Verification Protocol: For method verification of an unmodified FDA test, laboratories frequently utilize an abbreviated 5-day protocol: running 5 replicates per day across 5 consecutive days (25 total data points per level), verifying that calculated within-laboratory CV remains at or below the manufacturer's published CV claim.

Reportable Range: AMR vs. CRR & CLSI EP06

The laboratory must establish or verify the range of analyte values over which the test system provides clinically valid, linear quantitative results.

1. Analytical Measurement Range (AMR): The range of analyte values that a method can measure directly on the specimen without any dilution, concentration, or other pretreatment.
   • Verification Protocol per CLSI EP06-Ed2: Linear response is verified across the claimed AMR by analyzing a minimum of 4 to 5 distinct concentration levels (e.g., serial dilutions of a high patient pool mixed with a low patient pool). The measured values are plotted against theoretical expected concentrations; the deviation from linearity must not exceed predefined analytical quality goals (TEa).
   • Reporting Rule: Results falling strictly within the AMR can be reported directly without calculation. Results below the AMR are reported as less than the lower limit ("< X"). Results above the AMR must not be reported as absolute numbers unless diluted.
2. Clinically Reportable Range (CRR): The range of analyte values that a method can report, including specimens that are diluted, concentrated, or otherwise pretreated to bring them within the AMR.
   • Dilution Protocol Verification: If the laboratory reports results exceeding the upper AMR limit, it must verify the maximum allowable dilution factor (e.g., 1:10 or 1:100 with normal saline or manufacturer-supplied diluent). The laboratory assays diluted high specimens to prove recovery within ±10% of theoretical values. Results are multiplied by the verified dilution factor before being entered into the EHR. Results exceeding the upper CRR limit are reported as "> Y".

Reference Intervals (Normal Ranges) & CLSI EP28c

A clinical laboratory cannot blindly adopt published reference intervals without verifying that they accurately represent its specific patient population (accounting for local demographics, geographic altitude, ethnicity, and pre-analytical tube types). CLSI EP28c outlines two pathways:

1. Transference and Verification (Standard Approach):
   • When adopting an FDA-cleared manufacturer's reference interval, the laboratory can evaluate transference with the EP28 verification approach using 20 qualified reference individuals meeting defined inclusion criteria (free of overt disease, matched for age and sex).
   • Statistical Verification Criterion: If no more than 2 of the 20 reference subjects (≤ 10%) fall outside the proposed reference interval, the reference interval is statistically verified and may be adopted for clinical reporting.
   • If 3 or 4 individuals fall outside the proposed limits, the laboratory must collect specimens from 20 additional reference subjects; if no more than 2 of the second set of 20 fall outside, the range is verified. If ≥ 5 of the initial 20 fall outside, the proposed interval is rejected.
2. De Novo Establishment (Validation / LDTs):
   • When no established reference interval exists, or when transference fails completely, the laboratory must establish the reference interval de novo.
   • Reference-study design: The nonparametric EP28 approach generally uses at least 120 qualified reference individuals for each partitioned demographic group (e.g., 120 adult males and 120 adult females) to determine the non-parametric central 95th percentile interval with 90% confidence limits.

---

Additional Validation Parameters for LDTs and Modified Assays

Analytical Sensitivity: LoB, LoD, and LoQ (CLSI EP17)

For laboratory-developed tests, establishing analytical sensitivity requires differentiating three critical detection limits:

1. Limit of Blank (LoB): The highest measurement result that is likely to be observed with a stated probability (typically 95%) for a blank sample containing no analyte: LoB = Mean_blank + 1.645 × s_blank
2. Limit of Detection (LoD): The lowest concentration of analyte in a sample that can be reliably distinguished from the blank with 95% certainty, though not necessarily quantified with acceptable precision: LoD = LoB + 1.645 × s_low
3. Limit of Quantitation (LoQ): The lowest concentration of analyte that can be quantitatively determined with predefined analytical performance goals (e.g., total error ≤ 20% or CV ≤ 10%). In clinical medicine, the functional sensitivity of an assay is equivalent to its LoQ.

Analytical Specificity and Interference (CLSI EP07)

The laboratory must evaluate whether non-target substances cause significant positive or negative analytical bias. The laboratory systematically evaluates:

• Common Endogenous Interferents: Hemolysis (free hemoglobin), Icterus (unconjugated and conjugated bilirubin), and Lipemia (turbidity from triglycerides). Tests are performed by spiking varying concentrations of interferent into analyte-containing matrices to establish interference cutoff thresholds (HIL indices).
• Exogenous Interferents: Over-the-counter medications (e.g., acetaminophen, ascorbic acid, biotin), common prescription therapeutics, anticoagulants (EDTA, heparin, citrate), and sample tube gel additives.

---

Summary Comparison: Verification vs. Validation