Clinical chemistry and laboratory medicine最新文献

Laboratory errors are an important component of medical errors and are predominantly associated with the extra-analytical phases of the total testing process, particularly the pre-pre-analytical and post-post-analytical phases, which are largely dependent on clinical activities but also require laboratory support. The post-post-analytical phase is the stage in which clinicians interpret patients' laboratory data using population-based reference data, such as population-based reference intervals (popRIs) and population-based decision limits (popDLs), typically provided by the laboratory. To minimize errors in this phase, it is essential that more accurate tools - such as personalized reference intervals (prRIs) and personalized decision limits (prDLs) - are made available to clinicians. However, population-based references are still widely used, whereas their personalized counterparts have not yet been implemented in routine practice. The discrepancy between inadequate population-based references and more appropriate personalized references can introduce systematic yet latent errors in the interpretation of patients' laboratory data, potentially compromising patient safety even when clinicians are highly competent in data interpretation. In this opinion paper, we (1) summarize the limitations of popRIs and popDLs, (2) develop the concept of latent errors, and (3) discuss how personalized RIs and personalized DLs can be used to reduce latent errors and enable more accurate interpretation of patient laboratory data.

Objectives: An isotope dilution liquid chromatography-tandem mass spectrometry (ID-LC-MS/MS)-based candidate reference measurement procedure (cRMP) was developed and validated to measure serum and plasma concentrations of the total and free form of valproic acid.

Methods: Quantitative nuclear magnetic resonance spectroscopic methodology was used to determine the absolute content (g/g) of the reference material, ensuring traceability to SI units. Separation of valproic acid from potential unknown interferences was achieved with reversed-phase chromatography. A protein precipitation protocol was established for sample preparation for total valproic acid, while the free form was separated by ultrafiltration. Assay validations and measurement uncertainties were aligned with guidelines from the Clinical and Laboratory Standards Institute, the International Conference on Harmonization, and the Guide to the Expression of Uncertainty in Measurement.

Results: The cRMPs were highly selective and specific with no evidence of matrix effects, allowing quantifying total and free valproic acid in a range of 2.40-145 μg/mL and 1.60-42.0 μg/mL, respectively. Intermediate precision was <4.0 % and repeatability CV ranged from 0.9 to 3.5% for all concentrations of free and total valproic acid. The relative mean bias ranged from -0.4 to 4.1 % for native serum and from -0.3 to 3.5 % for Li-heparin plasma levels for total valproic acid. Free valproic acid showed mean biases between -2.9 and 4.0 % for native serum and ultrafiltrates. Measurement uncertainties for single measurements and target value assignment ranged from 1.7 to 3.4 % and 0.9-1.3 %, respectively, for total valproic acid. Free valproic acid ranged from 2.0 to 4.1 % and from 0.8 to 1.5 % for single measurements and target value assignment, respectively.

Conclusions: We present novel ID-LC-MS/MS-based cRMPs for total and free valproic acid in human serum and plasma which provides a traceable and reliable platform for the standardization of routine assays and evaluation of clinically relevant samples.

Objectives: Precise and accurate determination of urinary cortisol is important for diagnosis and monitoring of cortisol excess and deficiency. To achieve this, a selective and specific, isotope-dilution two-dimensional heart-cut liquid chromatography-tandem mass spectrometry (LC-MS/MS)-based method was developed.

Methods: A novel, two-dimensional heart-cut LC-MS/MS method was developed using a Waters Acquity BEH C18 column and a Raptor Biphenyl column, with water (A1) and with 0.2 mM ammonium fluoride in water (A2) and methanol (B1+B2) as mobile phases. This approach was complemented by a supported liquid extraction (SLE) sample preparation protocol. Matrix effects were assessed through post-column infusion experiments and comparisons of standard line slopes across matrices. A multi-day validation study evaluated accuracy and precision of the candidate reference measurement procedure (RMP) and measurement uncertainty (MU) was estimated in compliance with current guidelines. Reproducibility was assessed by performing a method comparison study between two independent laboratories.

Results: In the first LC column dimension, cortisol and prednisolone were separated from cortisone, prednisone, and 20-beta-dihydrocortisone. Baseline separation of cortisol from prednisolone was achieved in the second dimension. No interfering signals were detected at the retention time for quantifier and internal standard transitions in any urine samples analysed. No significant matrix effect, ion suppression, or enhancement was observed. Linearity was successfully demonstrated (r≥0.999); residuals were randomly distributed in a quadratic model via linear regression with an 1/x weighting. The lower limit of the measuring interval was determined at 2.00 ng/mL (5.52 nmol/L), displaying a relative deviation of -0.1 %, a coefficient of variation (CV) of 2.9 %, and recoveries of 96-103 %. Intermediate precision ranged from 1.9-3.2 % and repeatability CV from 1.6-2.7 %. Relative mean bias ranged from -3.5-1.2 %, indicating method accuracy was unaffected by matrix variance. Measurement uncertainties (k=1) for cortisol were ≤3.4 % across different concentrations and sample types. The expanded uncertainty at a 95 % confidence level (k=2) was ≤6.8 %. An increased number of measurements (n=6) for target value assignment reduced the uncertainty to 0.8-1.3 % (k=1), yielding an expanded uncertainty (k=2) of 1.7-2.5 %.

Conclusions: The validation results confirm the robustness, accuracy, and reliability of this RMP for determination of cortisol in human urine within a working range of 2.00-220 ng/mL (5.52-552 nmol/L).

The definition of analytical performance specifications (APS) by the Milan model 1b is based on indirect approaches investigating the impact of analytical performance of the laboratory test on clinical classification and thereby on the probability of patient outcomes. As direct diagnostic outcome studies (Milan model 1a) for defining APS are now considered very difficult and costly to be performed in practice, expert groups have gathered to reach consensus on how to use available information and apply Milan model 1b to the definition of APS. They have highlighted three major aspects: a) the definition of the clinically acceptable misclassification rate(s); b) the influence of the clinical pathway and patient population and setting (disease prevalence) when diagnostic thresholds are defined, e.g., in guidelines; and c) the intended use of the test. The basic question calling for an answer is how to move forward and provide specific APS for certain measurands that are key in clinical decision making. Here, cardiac troponin testing is used as a practical example for the application of model 1b-derived APS. Proposals are made for moving to practice with the application of this model to APS definition.

The integration of artificial intelligence (AI) and machine learning (ML) into laboratory medicine shows promise for advancing diagnostic, prognostic, and decision-support tools; however, routine clinical implementation remains limited and heterogeneous. Laboratory data presents unique methodological and semantic complexities - method dependency, analyte-specific variation, and contextual sensitivity-not adequately addressed by general-purpose AI reporting guidelines. To bridge this gap, the EFLM Committee on Digitalisation and Artificial Intelligence (C-AI) proposes an expanded checklist to support assessment of requirements and recommendations for the development of AI/ML models based on laboratory data. Building upon the widely adopted ChAMAI checklist (Checklist for assessment of medical AI), our proposal introduces six additional items, each grounded in the CRoss Industry Standard Process for Data Mining (CRISP-DM) framework and tailored to the specificities of laboratory workflows. These extensions address: (1) explicit documentation of laboratory data characteristics; (2) consideration of biological and analytical variability; (3) the role of metadata and peridata in contextualizing results; (4) analyte harmonization and standardization practices; (5) rigorous external validation with attention to dataset similarity; and (6) the implementation of FAIR data principles for transparency and reproducibility. Together, these recommendations aim to foster robust, interpretable, and generalizable AI systems that are fit for deployment in clinical laboratory settings. By incorporating these laboratory-aware considerations into model development pipelines, researchers and practitioners can enhance both the scientific rigor and practical applicability of AI tools. We advocate for the adoption of this extended checklist by developers, reviewers, and regulators to promote trustworthy and reproducible AI in laboratory medicine.

Introduction: The increasing use of therapeutic monoclonal antibodies (t-mAbs) has improved cancer and autoimmune disorder treatment. These therapeutics can interfere with serum protein electrophoresis (SPEP) and immunofixation (IF), potentially leading to the appearance of monoclonal bands that may be misinterpreted as monoclonal gammopathies. Identifying the migration patterns and detection thresholds of t-mAbs is crucial to avoid misinterpretation in clinical laboratories.

Content: A systematic review following PRISMA guidelines was conducted using Pubmed and ScienceDirect databases, with algorithm-based searches and double-blind article selection. Data on the matrix used, separation methods and type of interference were collected into an extraction table.

Summary: A total of 30 articles were included and 30 t-mAbs were described. 11 t-mAbs migrated at the end of the gamma region, 12 in the mid-gamma region, 5 in the early gamma region, one in the beta-2 globulin region and one in the alpha-2 globulin region. Most t-mAbs were detectable by SPEP and IF at concentrations above 100 mg/L.

Outlook: Caution is needed when a new peak appears on SPEP, as it may be mistaken for a monoclonal spike leading to misdiagnosis. Therefore, understanding the migration profiles of these t-mAbs is essential. Different methods are available to remove t-mAbs interference and could be used in daily practice.

Objectives: Accurate quantification of aldosterone is critical for screening and diagnosing primary aldosteronism (PA). Current competitive chemiluminescence immunoassays (cCLIA) overestimate plasma aldosterone concentration (PAC) compared to liquid chromatography-tandem mass spectrometry (LC-MS/MS). However, LC-MS/MS is technically demanding and time-consuming, limiting its widespread clinical utility. Therefore, a novel two-step sandwich chemiluminescence immunoassay (sCLIA) for accurate quantification of PAC was systematically evaluated.

Methods: Precision, trueness, linear range, and maximum dilution factor of the new immunoassay were comprehensively validated. In a multicenter study involving 2,696 samples from seven Chinese centers, PAC measurements were performed in parallel using sCLIA, cCLIA, and LC-MS/MS. The study specifically focused on evaluating the assay's performance at low aldosterone concentrations and in patients with chronic kidney disease (CKD), investigating potential interference from renal impairment by comparing the consistency between immunoassays and LC-MS/MS results across different CKD stages.

Results: The sCLIA exhibited excellent analytical performance for PAC measurement, with intra-assay imprecision <4.64 % and bias <5.71 % against certificated reference materials. The assay exhibited a wide reportable range (30-100,000 ng/L) with a limit of quantification at 30 ng/L and dilution capability ≥50-fold. Compared to cCLIA, sCLIA showed superior agreement with LC-MS/MS, particularly at low PAC concentrations (<110 ng/L) and in subjects with reduced renal function (eGFR<60 mL/min/1.73 m²).

Conclusions: This novel sCLIA method exhibited excellent analytical performance, combining the practical advantages of immunoassays with LC-MS/MS accuracy, thereby offering an ideal solution for large-scale primary aldosteronism screening in clinical practice.

Objectives: Glial fibrillary acidic protein (GFAP) is a well-established biomarker of astrocytic activation associated with neurodegenerative diseases, neuroinflammatory disorders, and traumatic brain injury. With increasing interest in blood-based biomarkers, the need for analytically validated assays and reliable reference intervals is critical for routine clinical implementation. This study aimed to analytically validate the MSD S-Plex^® GFAP immunoassay for plasma and to establish age-stratified reference intervals in an apparently healthy population.

Methods: This study was conducted in two phases. First, key analytical validation parameters - including repeatability, intermediate precision, measurement range, interferences, and sample stability - were evaluated following Clinical and Laboratory Standards Institute (CLSI) and published protocol guidelines. Second, reference intervals were derived from 579 apparently healthy individuals aged 17-91 years using a right-sided non-parametric percentile method. Age-specific upper reference limits were calculated for three predefined age groups, and a continuous age-dependent centile model was applied.

Results: MSD S-Plex^® GFAP assay demonstrated strong analytical performance, with coefficients of variation for repeatability and intermediate precision below 12 %. After accounting for the 1:2 dilution ratio, the validated measurement range was 0.425-1760 ng/L, with all calibration residuals remaining within ±15 %. GFAP concentrations were unaffected by hemolysis (p=0.85) and remained stable for up to 7 days at 4 °C and under frozen storage conditions. Age-stratified upper reference limits for plasma GFAP were established as 38 pg/mL (18-<50 years), 73 pg/mL (≥50-<70 years), and 156 pg/mL (≥70 years). Additionally, sex-related differences were observed after age 50, with females showing higher absolute GFAP levels than males. A strong positive correlation between age and plasma GFAP levels was observed (Spearman's r=0.832, p<0.0001).

Conclusions: This study demonstrates the robust analytical performance of the MSD S-Plex^® GFAP assay and establishes age-related reference values for plasma GFAP. These findings support its suitability for routine clinical use and enhance its applicability in the diagnosis and monitoring of central nervous system (CNS) pathologies, such as neurodegenerative diseases, neuroinflammatory disorders, and acute brain injuries, within biomarker-supported clinical algorithms.

Objectives: Measurement uncertainty (MU) plays an important role in the interpretation of laboratory results, but data on serum proteins analyzed by immunoturbidimetry according to ISO/TS 20914 are limited.

Methods: MU of 11 serum proteins, including CRP, RF, ASO, IgG, IgA, IgM, C3, C4, ceruloplasmin, transferrin, and β2-microglobulin, were estimated using 1-year internal quality control (IQC) data obtained from Roche Cobas analyzers. MU was calculated using uncertainty and calibrator uncertainty according to ISO/TS 20914, assuming negligible deviation from external quality assessment data. Analytical performance specification (APS) models were selected according to the EFLM APS selection criteria, and maximum allowable uncertainty (MAU) values were determined based on sources such as EFLM models and literature.

Results: IgA and RF were the only two analytes that met the required and minimum MAU values, respectively, at both IQC levels. MU values for CRP, ceruloplasmin, transferrin, and β2-microglobulin exceeded targets at both levels. MU for C3, C4, IgG, and IgM exceeded the minimum MAU at IQC1 but remained acceptable at IQC2. MU values for ASO were calculated as 10.01 and 7.22 % but could not be evaluated due to a lack of reference data. Assay precision should be improved for CRP, IgG, IgM, ceruloplasmin, transferrin, and β2-microglobulin. Use of updated calibration materials for CRP may help reduce MU.

Conclusions: Maintaining acceptable precision over a long period remains a challenge for serum proteins analyzed by immunoturbidimetry, highlighting the need for methodological improvements and stricter quality monitoring. In this context, MU assessment is crucial.

Laboratory data can be meaningful only when compared with reliable reference data; therefore, the estimation of reliable reference data is just as important as the accurate measurement of measurands in patient samples. Since analyte concentrations in the human body are influenced by both random variations (such as biological fluctuations) and systematic variations (such as physiological rhythms and age-related changes), the conventional model for estimating reference data - based solely on the statistical distribution of single-sample measurements from reference individuals - may not provide sufficiently reliable information for interpreting patient results. Therefore, a paradigm shift from relying solely on single-sample measurement distributions to incorporating metabolic changes observed in the human body when estimating reference intervals may enhance the clinical value of laboratory data for an effective clinical decision making and patient care. This opinion paper aims to summarize how to facilitate this transition and to identify the most suitable model for estimating reference intervals that reflect underlying metabolic dynamics.