CDLSO Services
Validation & Verification Support
METHOS SUPPORT BY DISCIPLINE
Validation is not one protocol applied uniformly across a laboratory. The performance characteristics that must be established, the experiments that establish them, and the failure modes that must be ruled out differ substantially by analyte class, platform, and specimen matrix.
The sections below describe how nuCARE’s support is applied within each discipline we work in, and where the analytical risk tends to concentrate in each.
Establishing Scope Before Designing Studies
Before any experiment is designed, the laboratory must determine which regulatory pathway applies under 42 CFR §493.1253. The determination drives everything downstream, and a misclassified scope is among the more consequential findings a laboratory can carry into an inspection.
Verification
Applies to unmodified FDA-cleared or FDA-approved methods. The laboratory demonstrates that it can reproduce the manufacturer's established performance in its own hands, with its own operators and specimens: accuracy, precision, reportable range, and confirmation that the manufacturer's reference intervals are appropriate for the population the laboratory serves.
Validation
Applies to laboratory-developed tests, to modified FDA-cleared methods, and to non-waived methods not subject to FDA clearance. Here the laboratory establishes rather than verifies: accuracy, precision, analytical sensitivity, analytical specificity including interfering substances, reportable range, reference intervals, and any other performance characteristic required for test performance.
Toxicology
Performance Characteristics
• Calibration model and weighting across the intended quantitative range, evaluated on the behavior of the residuals and back-calculated calibrators rather than on the correlation coefficient alone
• Limit of detection, lower limit of quantitation, and cutoff performance, with concordance evaluated at concentrations bracketing the administrative cutoff — commonly ±25% and ±50% of the decision point
• Within-run, between-run, and between-day precision established at low, near-cutoff, and upper-range concentrations
• Bias and accuracy assessed against reference material sourced independently from the calibrator material
• Carryover, evaluated by analyzing blank matrix immediately following the highest calibrator or a supraphysiologic specimen
• Dilution integrity for specimens exceeding the quantitative range
• Stability under the laboratory’s actual handling conditions: benchtop, refrigerated, frozen, freeze–thaw cycling, and processed-extract stability in the autosampler
Ion suppression and enhancement characterized across multiple independent donor lots, since a single blank matrix will not reveal lot-dependent suppression. Co-eluting stable-isotope-labeled internal standards compensate for, but do not eliminate, the need to characterize the effect.
Chromatographic separation demonstrated for compound pairs that share precursor and product ions or interconvert metabolically — morphine and hydromorphone, codeine and hydrocodone, the oxycodone/oxymorphone/noroxycodone set, the benzodiazepine family, and d/l-methamphetamine where enantiomeric reporting is offered.
Where glucuronide conjugates are reported as total analyte, hydrolysis must be demonstrated as efficient and reproducible, with enzyme lot performance verified and hydrolysis-resistant analytes identified rather than assumed complete.
Retention-time windows and qualifier-to-quantifier ion ratio tolerances established as explicit, documented acceptance criteria applied to every reported positive.
Screening assay cross-reactivity profiles characterized against the compounds actually encountered in the laboratory’s population, with expected discordance against the definitive method described in the method documentation rather than treated as error.
Creatinine, pH, specific gravity, and oxidant testing validated as their own methods with their own performance specifications, not as an adjunct to the drug panel.
Oral fluid or blood added to an established urine method requires matrix-specific study. Urine data does not transfer.
Molecular Infectious Disease by PCR
• Limit of detection determined per target and per specimen matrix, using replicate testing across a dilution series bracketing the endpoint and analyzed by probit or comparable regression, rather than assigned from the last dilution that happened to amplify
• Analytical inclusivity across the strains, serovars, and genotypes clinically relevant to each organism target, supported by in silico alignment of primer and probe sequences and by wet testing where isolates are obtainable
• Analytical specificity and exclusivity against phylogenetic near neighbors, commensal flora native to the matrix, and off-panel organisms that produce the same clinical syndrome
• Competitive interference within the multiplex, challenged with high-titer and low-titer target combinations that reflect realistic co-infection rather than single-organism spikes
• Reproducibility across operators, instruments, extraction batches, reagent lots, and days
• Inhibition monitoring through an internal or extraction control, with defined cycle-threshold shift criteria that render a result invalid
• Threshold and cycle-cutoff determination, with documented criteria for indeterminate results and defined repeat or reflex algorithms
• Specimen stability and transport-medium compatibility established for every collection device the laboratory accepts
Urinary specimens are frequently polymicrobial, which makes the interpretive framework as important as the analytical one. Where semi-quantitative organism load is reported, the relationship between cycle threshold and culture density requires its own correlation study and its own reporting thresholds. Contamination from perineal and periurethral flora, and the distinction between colonization and infection, must be addressed in the method’s limitation language rather than left to the ordering clinician.
Nasopharyngeal swab, anterior nasal swab, and lower respiratory specimens behave differently in extraction and carry different organism loads; each accepted source requires its own limit-of-detection work. Seasonal antigenic and genomic drift in viral targets makes periodic re-alignment of primer and probe sequences against current circulating sequences a maintenance obligation, not a one-time validation activity.
Nail presents the most demanding extraction problem in this group. Keratinized specimen requires mechanical pulverization and enzymatic or chemical digestion, and extraction efficiency must be established using spiked nail matrix — spiked buffer will substantially overstate performance. Dermatophytes, non-dermatophyte molds, and Candida species have different lysis requirements, and low-biomass specimens raise the relative significance of environmental contamination.
These are high-consequence results with public health reporting implications, and specificity carries more weight than in most panels. N. gonorrhoeae detection warrants particular attention to cross-reactivity with commensal Neisseria species, which is the reason multiple independent targets are commonly employed. Urine, vaginal, endocervical, and urethral specimens are separate matrices; extragenital rectal and pharyngeal sources are separate again, carry substantially higher background flora, and are the sources most often reported without matrix-specific validation data behind them. Where M. genitalium macrolide resistance markers are reported, the resistance determinant requires its own performance characterization.
Wound specimens are polymicrobial by nature and include anaerobes whose recovery cannot be benchmarked against routine aerobic culture. Comparator-method design deserves early attention here, since discordance with culture is expected rather than exceptional, and the resolution pathway for discordant results should be defined before the study begins rather than after.
• Resistance determinants. Detection of a resistance gene in a polymicrobial specimen cannot be attributed to a specific organism present, and genotype does not establish phenotype. The reporting language and limitation statements that convey this are part of what is validated, not commentary appended afterward.
• Amplicon contamination control. Unidirectional workflow, physical separation of pre- and post-amplification areas, dUTP/UNG carryover prevention where the chemistry supports it, no-template controls, and environmental surface monitoring are treated as validated elements of the method system with defined acceptance criteria.
General Chemistry
• Precision established at two or more concentrations spanning the medical decision points, over multiple days rather than within a single run, and compared explicitly against the manufacturer’s claim
• Trueness and bias by method comparison against the incumbent or a reference method, using patient specimens distributed across the analytical measurement range and evaluated by Deming or Passing–Bablok regression with difference plots — a correlation coefficient does not demonstrate agreement
• Reportable range: linearity across the analytical measurement range, calibration verification at defined intervals, and validated dilution protocols where a clinical reportable range extends beyond it
• Reference intervals verified by transference where the laboratory’s population matches the manufacturer’s, established directly where it does not, with attention to partitioning by age and sex
• Analytical sensitivity — limit of blank, detection, and quantitation — where the low end of the range carries clinical weight
• Interference: hemolysis, icterus, and lipemia thresholds established or verified on the specific platform, alongside analyte-specific and common drug interferents
• Carryover between high- and low-concentration specimens, including sample probe and reagent cross-contamination pathways
• Specimen-type equivalence across every tube type the laboratory accepts — serum, lithium heparin plasma, and gel-barrier tubes are not interchangeable by assumption
Analytes with known inter-method bias require awareness of the traceability chain, since the bias operates directly on clinical decision thresholds — creatinine IDMS traceability and its propagation into reported eGFR is the clearest example, and HbA1c certification is another.
Control rules selected against the observed performance of the method on the specific instrument, rather than a default multirule applied uniformly across the menu. Where a risk-based approach is used, the individualized quality control plan requires its own documented risk assessment.
Estimated and available for quantitative results, and revisited when method performance shifts.
Reagent and calibrator lot changes verified against defined acceptance criteria before patient reporting, with the criteria established during validation rather than negotiated at the moment of the change.
Enrollment for regulated analytes and a defined, documented alternative performance assessment for those not covered by an approved program.
Pharmacogenetics
• Accuracy against well-characterized reference materials, with orthogonal confirmation by an independent method for every interrogated variant. Genotyping accuracy is conventionally held to complete concordance rather than to a statistical acceptance band
• Precision and reproducibility across runs, operators, instruments, reagent lots, and days, including repeat determination on the same specimen
• Analytical sensitivity expressed as the validated input range and quality tolerance for nucleic acid, established at both the low and high ends rather than at a single nominal input
• Specimen-type validation for each accepted source — whole blood, buccal swab, saliva — treated as distinct matrices, with heparinized specimens evaluated specifically for inhibition
• Call rate and no-call rate characterized, with defined repeat and reflex criteria
• Reportable range defined explicitly as the list of variants interrogated, with the corollary that undetected does not mean absent
Gene duplications and multiplications, whole-gene deletion, and CYP2D6–CYP2D7 hybrid and tandem arrangements require copy-number-aware and structure-aware assay design. A SNP-only panel cannot resolve them, and where the assay does not, the limitation belongs in the report.
The UGT1A1 promoter TA repeat requires length resolution rather than allelic discrimination; designing for it as a SNP produces systematically wrong calls.
Unrecognized polymorphism beneath a primer or probe binding site produces confident, incorrect homozygous calls. The risk is assessed during design and monitored afterward.
Star-allele definitions maintained against the current reference nomenclature under version control, with a defined process for adopting updates.
Validated as its own process — activity score assignment, phenotype mapping against current consensus guidance, and an explicit, documented handling of ambiguous or indeterminate diplotypes.
The path from raw genotype call through middleware and LIS mapping to the printed phenotype, drug list, and limitation language, exercised with known genotypes that include the edge cases. This is the step most often omitted.
Adding an allele to the panel, revising a translation table, or adopting an updated guideline is a defined revalidation event with a documented scope.
Proficiency testing where available, and documented alternative performance assessment at least twice annually where it is not.
Cancer Genomics
• Limit of detection expressed by variant class and by variant allele fraction, established across a dilution series of blended reference material rather than at a single dilution point
• Accuracy and concordance against orthogonally characterized specimens and consensus reference materials, with sensitivity, specificity, and positive predictive value reported separately for single-nucleotide variants, small insertions and deletions, large insertions and deletions, copy-number alterations, and fusions — aggregate accuracy across variant classes conceals the classes that perform worst
• Coverage depth and uniformity, with a defined minimum depth per interrogated position and an explicit policy for positions falling below it
• Reproducibility within run, between runs, between instruments, between operators, and across reagent lots
• Specificity and false-positive characterization, including systematic artifact classes: fixation-induced deamination, homopolymer-associated indel artifact, and mapping error in paralogous and pseudogene regions
• Reportable range defined as the specific genomic regions and variant classes the assay interrogates at validated performance, stated as such in the report
Nucleic acid quality metrics, tumor nuclei estimation and macrodissection practice, and the minimum neoplastic cell content at which the claimed limit of detection holds. The detection claim is tied to tumor content, not to the specimen as received.
Enzymatic treatment for deamination artifact and unique molecular identifier-based error correction, validated specifically where low variant allele fraction detection is claimed.
RNA-based fusion calling validated separately from DNA variant detection, with an explicit distinction between known-partner and novel-partner detection capability.
Microsatellite instability and tumor mutational burden validated against their orthogonal reference methods, with panel-size dependence and calibration stated rather than implied.
Validated as a discrete, version-locked exercise: alignment, variant calling, annotation, and filtering evaluated against truth sets, supplemented with in silico data to challenge variant classes not represented in available physical reference material. Every component version is recorded, and version drift is a revalidation trigger.
Somatic tiering and germline classification frameworks applied under knowledge base version control, with a defined policy for possible incidental germline findings arising from a tumor-only assay, established before the first case rather than at the first occurrence.
Run- and sample-level quality metrics trended over time — on-target rate, duplication rate, uniformity, mean and minimum depth — so that gradual degradation is visible before it becomes a reporting problem.
After Implementation
Validation establishes that a method performs. Keeping it performing is a separate discipline, and the two are connected by the decisions made during the study: the acceptance criteria set during validation become the control limits that govern daily operation.
Across every discipline above, nuCARE supports the transition — quality control strategy carried forward from validation data, proficiency testing enrollment or documented alternative performance assessment, calibration verification intervals, competency assessment tied to the methods actually performed, and a defined set of revalidation triggers so the laboratory knows in advance which changes require what response.
As with the validation studies themselves, these remain the laboratory’s activities, performed by the laboratory’s qualified personnel, under the authority of its Laboratory Director. Our role is to provide the scientific expertise, structure, training, and guidance that make the process defensible — and to leave the laboratory able to explain its own methods, in its own words, long after the study is filed.