Clinical chemistry最新文献

Background: Germline loss-of-function variants in BRCA1 and BRCA2 are established drivers of hereditary breast and ovarian cancer, often acting through aberrant splicing. However, not all spliceogenic changes are pathogenic, and many variants remain classified as uncertain due to insufficient experimental evidence and challenges in applying the ACMG/AMP variant interpretation framework to splicing alterations.

Methods: In this study, we examined the splicing outcomes of 17 variants-10 in BRCA1 [c.135-2A>G; c.135-5T>C; c.5074+1G>C; c.5332+2_5332+4del; c.5333-8C>T; c.5335C>G p.(Gln1779Glu); c.302-24_302-22del; c.302-23A>G; c.547+57T>C; c.4096+34C>G] and 7 in BRCA2 [c.-39-5delT; c.67+3A>G; c.425G>A p.(Ser142Asn); c.425G>T p.(Ser142Ile); c.517-13_517-9del; c.681+5G>C; c.67+84_67+85del]-identified in families with suspected hereditary breast and/or ovarian cancer. Depending on sample availability, we assessed splicing either on carrier-derived mRNA or via splicing-reporter minigene assay.

Results: Eight variants triggered aberrant splicing, while 9 showed no spliceogenic effect. Our findings, combined in some cases with previously published data, allowed us to apply the PVS1_(RNA) criterion at full strength to some variants. For others, residual full-length transcripts or in-frame mis-spliced isoforms precluded full application of PVS1_(RNA).

Conclusions: Following ClinGen ENIGMA BRCA1 and BRCA2 Variant Curation Expert Panel specifications based on ACMG/AMP guidelines, we classified 4 variants as pathogenic or likely pathogenic, 10 as benign or likely benign, and 3 as uncertain significance. This comprehensive analysis of splicing defects refines the clinical classification of BRCA1 and BRCA2 variants and highlights the value of combining experimental and computational evidence to enhance genetic risk assessment in hereditary cancer.

Background: Despite its growing impact, advanced liver fibrosis (ALF) remains substantially underdiagnosed in both primary and specialized care settings. Since liver cirrhosis is typically preceded by a prolonged phase of asymptomatic fibrosis, early detection of ALF, particularly in high-risk individuals, represents a public health priority. From a laboratory medicine perspective, this prolonged subclinical phase offers an interesting opportunity for early detection of ALF using blood-based noninvasive tests (NITs) that can be implemented in primary and nonhepatology care settings before overt disease develops.

Content: By critically appraising the evidence sources in the available literature, this paper provides an overview of blood-based NITs useful for the detection of ALF, with particular emphasis on the aspects and problems related to their implementation in daily laboratory practice. We explore the feasibility of different scenarios using strategies based on routine biochemical parameters and more specialized NITs that incorporate measurements of direct markers of fibrosis activity. Moreover, we highlight discrepancies existing among clinical practice guideline (CPG) recommendations that may hamper their widespread implementation in medical laboratories.

Summary: Advances in understanding and increase in prevalence of ALF require earlier detection and more accurate risk assessment of this condition. Blood-based NITs may provide a widely accessible diagnostic aid, especially in primary care and resource-limited settings. Multidisciplinary collaboration focusing on their integration into clinical pathways to optimize patient evaluation and specialist referral is required. Harmonization of recommendations in international CPGs will certainly contribute to their more effective use.

Background: Circulating cell-free DNA (cfDNA) fragmentomic features, such as fragment size and end motifs, have emerged as promising noninvasive biomarkers for cancer detection. However, the influence of nonmalignant clinical factors on these features remains unclear, potentially confounding liquid biopsy assays.

Methods: We analyzed cfDNA fragmentomic data from 1154 noncancerous individuals undergoing routine health checkups. Three cfDNA features were examined: cfDNA concentration, short fragment ratio (SFR), and cancer-enriched motif (CEM) frequency. Associations with 65 demographic, hematologic, and biochemical variables were assessed using univariate and multivariate analyses. High-resolution correlation mapping of fragment size (110 to 230 bp) and 4-mer end motifs was performed. Patterns were compared with profiles from 283 lung cancer patients, and confounding effects were evaluated using receiver operating characteristic (ROC) analysis.

Results: Multiple nonmalignant variables were significantly correlated with cfDNA features. Age was associated with all three features, while liver function markers (aspartate aminotransferase [AST], alkaline phosphatase [ALP], γ-glutamyl transferase [γ-GTP]) showed strong associations with SFR and CEM frequency. High-resolution analyses revealed that AST-related fragment size profiles closely resembled cancer-associated patterns, whereas age showed partial similarity to cancer-associated end motif alterations. ROC analyses demonstrated that elevated AST or older age reduced the discriminative performance of SFR and CEM, indicating their potential as confounders in lung cancer detection.

Conclusions: Physiological factors such as liver enzyme levels and age can significantly alter cfDNA fragmentomic profiles, generating patterns that resemble lung cancer-associated signals. These results highlight the importance of incorporating strategies to mitigate nonmalignant variability when developing cfDNA-based liquid biopsy assays, to ensure their accuracy, specificity, and clinical applicability.

Background: Artificial intelligence (AI) augmented laboratory tests can improve quality, efficiency, cost-effectiveness, and staff satisfaction. However, the clinical success of these tests can introduce unforeseen challenges for model retraining. This study describes the discovery of an "AI label feedback loop" in a clinically implemented AI-augmented kidney stone composition test.

Methods: An AI-augmented kidney stone composition test (V1) has been previously deployed for clinical kidney stone characterization. After several years of clinical use, a retrained model (V2) was developed using 6 times more data. Model performance of both V1 and V2 were evaluated across 3 datasets: a recent production validation (hold-out) set (mostly AI-influenced labels), the original V1 validation set (pre-AI, entirely human-labeled), and a subset of recent cases with exclusively human-generated or human-corrected labels.

Results: V2 demonstrated a 10% lower concordance rate than V1 when evaluated on the recent production hold-out set, despite a much larger training dataset. Performance between V1 and V2 was similar when applied to the pre-AI validation set. Notably, V2 outperformed V1 on the recent subset of cases with human-only or human-corrected labels, particularly for less-common stone types. These findings revealed an AI label feedback loop, confounding retraining and evaluation.

Conclusions: The integration of AI into clinical practice can potentially influence reported test results, complicating the development and evaluation of future models. To mitigate AI label feedback loops, ongoing human annotation and careful validation set construction are essential. These strategies can ensure reliable performance assessment and support the safe evolution of clinical AI systems.