Guides for diagnostic development
For a lab-developed or modified PCR test, CLIA requires the lab to establish accuracy, precision, analytical sensitivity, analytical specificity including interferences, reportable range, reference intervals and any other characteristic the test needs, before reporting patient results. An unmodified FDA-cleared kit needs only a shorter verification.
Before you start
Get the laboratory director to sign a validation plan that fixes:
The studies
40 or more specimens across the range compared with a reference method, an orthogonal assay or characterized samples. Report positive and negative agreement, or bias and regression.
2–3 levels including one near the cutoff, in replicate across several days, operators, reagent lots and instruments.
Dilution series around the expected limit, about 20 replicates per level. The limit of detection is the lowest level detected in at least 95% of replicates; quantitative tests also set a limit of quantitation.
Related organisms or sequences (homologs, pseudogenes), plus interferents such as hemoglobin, lipids, heparin and extraction carryover.
For quantitative tests, 5 or more levels across the range; for qPCR, a standard curve with efficiency of about 90–110%.
For most qualitative molecular tests the normal result is "not detected"; document the rationale. Also cover specimen and extract stability, extraction recovery, carryover and control performance.
Sample numbers are common starting points, not regulatory minimums; the laboratory director sets the final acceptance criteria.
CAP accreditation
If your laboratory is CAP-accredited, the All Common checklist adds specific validation requirements.
Director approval
A written evaluation of the validation, signed by the laboratory director or qualified designee, before any patient testing.
Accuracy samples
At least 20 samples for an LDT or modified test, across the range; for qualitative tests, positives, negatives and low positives. Fewer needs a documented director justification; QC material does not count.
Precision, sensitivity, specificity
Within-run and between-run precision; the detection limit; analytical specificity; documented interferences with a plan of action; reportable range.
QC and calibration
Written QC and calibration procedures based on the validation data.
Clinical claims
Any sensitivity, specificity or risk claim validated with at least 20 positive and negative samples, unless peer-reviewed literature supports it.
Records and re-verification
Keep a list of LDTs and modified tests; state on reports that the lab developed the test; verify each identical instrument; re-verify within 30 days after a break in testing.
Item numbers are from the August 2017 CAP All Common checklist excerpt; check your current edition. Molecular tests also fall under CAP's Molecular Pathology checklist.
qPCR vs ddPCR
qPCR
ddPCR
After validation
The laboratory director signs a validation report before the first patient result. After that, CLIA expects quality control with every run, proficiency testing (or an alternative assessment) at least twice a year, and revalidation when the workflow changes. To test New York patients, the state's Clinical Laboratory Evaluation Program must approve the test first. Data collected to FDA expectations can later support a 510(k) or De Novo submission.
Source: 42 CFR 493.1253, Establishment and verification of performance specifications
We design validation plans, run the studies and operate the test in our CLIA/CAP laboratory.