Development of a diagnostic model for biliary atresia based on MMP7 and serological tests using machine learning.

OBJECTIVE: To develop a machine learning diagnostic model based on MMP7 and other serological testing indicators for early and efficient diagnosis of biliary atresia (BA). METHODS: A retrospective analysis was conducted on patient information from those hospitalized for pathological jaundice at Beijing Children's Hospital between January 1, 2019, and December 31, 2023. Patients with serum MMP7, liver stiffness measurements, and other routine serological tests were included in the study. Six machine learning models were constructed, including logistic regression (LR), random forest (RF), decision tree (DET), support vector machine classifier (SVC), neural network (MLP), and extreme gradient boosting (XGBoost), to diagnose BA. The area under the receiver operating characteristic curve was used to evaluate the diagnostic efficacy of the various models. RESULTS: A total of 98 patients were included in the study, comprising 64 BA patients and 34 patients with other cholestatic liver diseases. Among the six machine learning models, the XGBoost algorithm model and RF algorithm model achieved the best predictive performance, with an AUROC of nearly 100% in both the training and validation sets. In the training set, these two algorithm models achieved an accuracy, precision, recall, F1 score, and AUROC of 1. Through model interpretation analysis, serum MMP7 levels, serum GGT levels, and acholic stools were identified as the most important indicators for diagnosing BA. The nomogram constructed based on the XGBoost algorithm model also demonstrated convenient and efficient diagnostic efficacy. CONCLUSION: Machine learning models, especially the XGBoost algorithm and RF algorithm models, constructed based on preoperative serum MMP7 and serological tests can diagnose BA more efficiently and accurately. The most important influencing factors for diagnosis are serum MMP7, serum GGT, and acholic stools.

Serum matrix metalloproteinase-7 for discriminating biliary atresia: a diagnostic accuracy and validation study.

BACKGROUND: Prompt and precise differential diagnosis of biliary atresia (BA) among cholestatic patients is of great importance. Matrix metalloproteinase-7 (MMP-7) holds great promise as a diagnostic marker for BA. This study aimed to investigate the accuracy of age-specific serum MMP-7 for discriminating BA from other cholestatic pediatric patients. METHODS: This was a single center diagnostic accuracy and validation study including both retrospective and prospective cohorts. Serum MMP-7 concentrations were measured using an ELISA kit, the trajectory of which with age was investigated in a healthy infants cohort aged 0 to 365 days without hepatobiliary diseases (n = 284). Clinical BA diagnosis was based on intraoperative cholangiography and subsequent histological examinations. The diagnostic accuracy of age-specific cutoffs of serum MMP-7 were assessed in a retrospective cohort of cholestatic patients (n = 318, with 172 BA) and validated in a prospective cohort (n = 687, including 395 BA). RESULTS: The MMP-7 concentration declines non-linearly with age, showing higher levels in healthy neonates as well as higher cutoff value in neonatal cholestasis. The area under the ROC curve (AUROC) was 0.967 (95% confidence interval [CI]: 0.946-0.988) for the retrospective cohort, and the cutoff of 18 ng/mL yielded 93.0% (95%CI: 88.1-96.3%), 93.8% (95%CI: 88.6-97.1%), 94.7% (95%CI: 90.1-97.5%), and 91.9% (95%CI: 86.4-95.8%) for sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV), respectively. The performance of MMP-7 was successfully validated in the larger prospective cohort, resulting in a diagnostic sensitivity of 95.9% (379/395; 95% CI: 93.5-97.7%), a specificity of 87.3% (255/292; 95% CI: 83.0-90.9%), a PPV of 91.1% (379/416; 95% CI: 87.9-93.7%), and a NPV of 94.1% (255/271; 95% CI: 90.6-96.6%), respectively. Besides, higher cutoff value of 28.1 ng/mL achieved the best sensitivity, specificity, PPV, and NPV for infants aged 0-30 days, which was 86.4% (95% CI: 75.0-94.0%), 95.5% (95% CI: 77.2-99.9%), 98.1% (95% CI: 89.7-100%), and 72.4% (95% CI: 52.8-87.3%), respectively. CONCLUSIONS: The serum MMP-7 is accurate and reliable in differentiating BA from non-BA cholestasis, showing its potential application in the diagnostic algorithm for BA and significant role in the future research regarding pathogenesis of BA.

Relationships of serum MMP-7 and clinical characteristics in choledochal cyst children.

BACKGROUND: Matrix metalloproteinase-7 (MMP-7) is associated with biliary injury. This study aimed to evaluate the relationships of serum MMP-7 with clinical characteristics in choledochal cysts (CDC) children. METHODS: Between June 2020 and July 2022, we conducted a prospective study of CDCs who underwent one-stage definitive operation at our center. Serum MMP-7 was measured using an enzyme-linked immunosorbent assay. We evaluated the relationships between serum MMP-7 and age, laboratory tests, imaging examinations, liver fibrosis, MMP-7 expression, and perforation. RESULTS: A total of 328 CDCs were enrolled in the study, with a median serum MMP-7 of 7.67 ng/mL. Higher serum MMP-7 was correlated with younger age at diagnosis (p < 0.001), larger cyst sizes (p < 0.001), higher liver fibrosis stages (p < 0.001), and higher incidence of perforation (p < 0.01). Liver MMP-7 was mainly expressed in intrahepatic and extrahepatic biliary epithelial cells. The area under the receiver operating characteristic curve (AUROC) was 0.630 (p < 0.001) for serum MMP-7 in predicting perforation. When serum MMP-7 was combined with γ-glutamyl transferase (GGT), the AUROC increased to 0.706 (p < 0.001). CONCLUSIONS: Serum MMP-7 was associated with biliary obstruction in CDCs. Patients with high serum MMP-7 were more likely to have severe liver damage and biliary injury, with higher incidences of liver fibrosis and perforation.

Diagnostic accuracy of serum matrix metalloproteinase-7 as a biomarker of biliary atresia in a large North American cohort.

BACKGROUND AND AIMS: High levels of serum matrix metalloproteinase-7 (MMP-7) have been linked to biliary atresia (BA), with wide variation in concentration cutoffs. We investigated the accuracy of serum MMP-7 as a diagnostic biomarker in a large North American cohort. APPROACH AND RESULTS: MMP-7 was measured in serum samples of 399 infants with cholestasis in the Prospective Database of Infants with Cholestasis study of the Childhood Liver Disease Research Network, 201 infants with BA and 198 with non-BA cholestasis (age median: 64 and 59 days, p = 0.94). MMP-7 was assayed on antibody-bead fluorescence (single-plex) and time resolved fluorescence energy transfer assays. The discriminative performance of MMP-7 was compared with other clinical markers. On the single-plex assay, MMP-7 generated an AUROC of 0.90 (CI: 0.87-0.94). At cutoff 52.8 ng/mL, it produced sensitivity = 94.03%, specificity = 77.78%, positive predictive value = 64.46%, and negative predictive value = 96.82% for BA. AUROC for gamma-glutamyl transferase = 0.81 (CI: 0.77-0.86), stool color = 0.68 (CI: 0.63-0.73), and pathology = 0.84 (CI: 0.76-0.91). Logistic regression models of MMP-7 with other clinical variables individually or combined showed an increase for MMP-7+gamma-glutamyl transferase AUROC to 0.91 (CI: 0.88-0.95). Serum concentrations produced by time resolved fluorescence energy transfer differed from single-plex, with an optimal cutoff of 18.2 ng/mL. Results were consistent within each assay technology and generated similar AUROCs. CONCLUSIONS: Serum MMP-7 has high discriminative properties to differentiate BA from other forms of neonatal cholestasis. MMP-7 cutoff values vary according to assay technology. Using MMP-7 in the evaluation of infants with cholestasis may simplify diagnostic algorithms and shorten the time to hepatoportoenterostomy.

Plasma Matrix Metalloproteinase Concentrations and Risk of Interstitial Lung Disease in a Prospective Rheumatoid Arthritis Cohort.

OBJECTIVE: To evaluate the associations of plasma matrix metalloproteinases (MMPs) with prevalent and incident interstitial lung disease (ILD) in people with rheumatoid arthritis (RA). METHODS: Within a multicenter, prospective cohort of US veterans with RA, we performed a cross-sectional study of prevalent ILD and cohort study of incident ILD. ILD diagnoses were validated by medical record review of provider diagnoses and chest imaging and/or pathology reports. MMP-1, 3, 7, and 9 concentrations were measured in plasma samples, then standardized and categorized into quartiles. The associations of MMPs with prevalent and incident ILD were assessed with logistic (prevalent) and Cox (incident) regression models adjusted for RA-ILD risk factors. RESULTS: Among 2,312 participants (88.9% male; mean age 63.8 years), 96 had prevalent ILD. Incident ILD developed in 130 participants over 17,378 person-years of follow-up (crude incidence rate 7.5/1,000 person-years). Participants with the highest quartile of MMP-7 concentrations had a nearly four-fold increased odds of prevalent ILD (adjusted odds ratio 3.78 [95% confidence interval (95% CI) 1.86-7.65]) and over two-fold increased risk of incident ILD (adjusted hazard ratio 2.33 [95% CI 1.35-4.02]). Higher MMP-9 concentrations were also associated with prevalent and incident ILD, as well as negatively correlated with forced vital capacity among those with prevalent ILD (r = -0.30, P = 0.005). CONCLUSION: MMP-7 and MMP-9 were strongly associated with both prevalent and incident ILD in this large, multicenter RA cohort after adjustment for other RA-ILD risk factors. These population-level findings further support a potential pathogenic role for MMPs in RA-ILD and suggest that their measurement could facilitate RA-ILD risk stratification.