Elicit: Total Analytical Error in Liver Fibrosis Panels (public)
For a liver fibrosis panel, what is total analytical error by population (NAFLD, viral hepatitis, alcohol-related)?
Total analytical error was not explicitly reported for liver fibrosis panels in any population (NAFLD, viral hepatitis, alcohol-related), with studies instead using sensitivity, specificity, and AUROC metrics.
Abstract
No study reported an explicit total analytical error for liver fibrosis panels. Instead, papers quantified diagnostic performance using metrics such as sensitivity, specificity, AUROC, and predictive values. In nonalcoholic fatty liver disease (NAFLD), panels based on FIB-4, NAFLD Fibrosis Score, and APRI showed:
- Sensitivity ranging from 7% to 33% in general NAFLD (with negative predictive values of 98–99% when compared to transient elastography)
- In patients with Type 2 Diabetes Mellitus, FIB-4 reached sensitivities of 81.6–95.9% and AUROC values of 0.79–0.91
- In NAFLD/NASH studies incorporating novel markers (e.g., RAB14, PLIN2), sensitivities of 88–100%, specificities of 89.6–100%, and AUROC up to 0.99 were noted
In viral hepatitis, panels including Hepascore, FIB-4, APRI, and FIBROSpect II demonstrated sensitivities between 71.8% and 82%, specificities around 65–74%, and AUROC values of 0.81–0.83. For alcohol-related liver disease, tests such as Enhanced Liver Fibrosis and FibroTest yielded AUROC figures of 0.90–0.92 with negative predictive values of 97–98%, along with false positive rates of 8–45% and false negatives below 8%.
Methods
We analyzed 40 sources from an initial pool of 996, using 8 screening criteria. Each paper was reviewed for 5 key aspects that mattered most to the research question.
Papers identified with Elicit search
n = 996
Papers screened using: Fibrosis Panel Assessment, Analytical Error Reporting, Target Population, Reference Standard, Study Type and Setting, Adult Population, Publication Type, Outcome Measures
n = 996
Papers screened out
n = 956
Papers included for extraction
n = 40
Data extraction
Study Population and Setting
- Total number of participants
- Population type (NAFLD, viral hepatitis, alcohol-related liver disease)
- Mean/median age
- Gender distribution
- Specific clinical setting (e.g., liver clinic, primary care, research study)
- Inclusion and exclusion criteria
Liver Fibrosis Assessment Method
- Primary method of fibrosis assessment (e.g., liver biopsy, transient elastography, non-invasive scoring system)
- Specific measurement techniques (e.g., FibroScan, specific scoring algorithm)
- Cutoff values used to define fibrosis stages
- Reference standard used for validation
Total Analytical Error Measurements
- Analytical error rates
- Sensitivity
- Specificity
- Positive predictive value
- Negative predictive value
- Area under the receiver operating characteristic (AUROC) curve
Variability and Measurement Precision
- Intra-individual variability of test measurements
- Measurement error sources
- Reproducibility of results
- Factors affecting measurement precision
Study Limitations and Bias Considerations
- Potential sources of bias
- Spectrum effect or spectrum bias discussions
- Limitations in study design
- Generalizability of results
- Conflicts of interest
- Funding sources
Results
Characteristics of Included Studies
| Study | Study Population | Fibrosis Tests Evaluated | Total Analytical Error (TAE) Calculation Method | Sample Size | Full text retrieved |
|---|---|---|---|---|---|
| Hernaez and Thrift, 2017 | General ambulatory (Nonalcoholic Fatty Liver Disease (NAFLD) context) | NAFLD fibrosis score, Aspartate Aminotransferase to Platelet Ratio Index (APRI), Fibrosis-4 (FIB-4), Transient Elastography (TE) | Sensitivity, specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV); no explicit Total Analytical Error (TAE) | 749 | No |
| Ajmera and Loomba, 2023 | NAFLD with Type 2 Diabetes Mellitus (T2DM) | FIB-4, Vibration-Controlled Transient Elastography (VCTE), Agile 3+/4, Magnetic Resonance Elastography (MRE) | Sensitivity, specificity, Area Under the Receiver Operating Characteristic Curve (AUROC); no explicit TAE | 501 | No |
| Angelini et al., 2022 | NAFLD/Nonalcoholic Steatohepatitis (NASH) | RAB14, PLIN2, FIB-4, NAFLD Fibrosis Score (NFS), APRI, Two-Dimensional Shear Wave Elastography (2D-SWE) | Sensitivity, specificity, AUROC; no explicit TAE | 250 | Yes |
| Mózes et al., 2023 | NAFLD | FIB-4, NFS, Liver Stiffness Measurement by VCTE (LSM-VCTE), histology | AUROC; no explicit TAE | 2518 | No |
| Sanyal et al., 2022 | NAFLD | Agile 3+, Agile 4, FIB-4, LSM | AUROC; no explicit TAE | We didn’t find mention of sample size | No |
| Pennisi et al., 2022 | NAFLD | AGILE 3+, FIB-4, LSM | AUROC; no explicit TAE | 614 | No |
| Vali et al., 2023 | NAFLD | 17 biomarkers, FIB-4, FibroScan | AUROC; no explicit TAE | 966 | No |
| Thiele et al., 2018 | Alcohol-related liver disease | Enhanced Liver Fibrosis (ELF), FibroTest, TE, 2D-SWE, indirect markers | AUROC, NPV, Coefficient of Variation (CV); no explicit TAE | 289 | Yes |
| Becker et al., 2009 | Hepatitis C Virus (HCV) | Hepascore, FIB-4, APRI | AUROC, sensitivity, specificity; no explicit TAE | 391 | No |
| Ooi et al., 2016 | Obese NAFLD (bariatric) | NFS, BARD, FIB-4, Forn, APRI | AUROC, sensitivity, NPV; no explicit TAE | 107 | No |
Summary:
- Across the included studies, total analytical error for liver fibrosis panels is not explicitly reported, but related metrics (sensitivity, specificity, AUROC, Negative Predictive Value, Positive Predictive Value, false positive/negative rates, discordance) are widely available.
- FIB-4 and Enhanced Liver Fibrosis are reliable for ruling out advanced fibrosis in Nonalcoholic Fatty Liver Disease and alcohol-related liver disease, with low false negative rates and high Negative Predictive Value.
- In viral hepatitis, performance is generally good but may be compromised in special populations such as those with End-Stage Renal Disease or metabolic syndrome.
- Analytical error is higher for ruling in advanced fibrosis, especially in low-prevalence settings.
- Population-specific factors (obesity, diabetes, ethnicity) and test-specific limitations (cutoff values, measurement variability) must be considered when interpreting results.
- There is a lack of standardized reporting of Total Analytical Error and a need for further validation of novel biomarkers, as indicated by the available studies.