A new generation of non-invasive tests of liver fibrosis in MASLD

The new generation of non-invasive tests (NITs) for liver fibrosis in metabolic dysfunction-associated steatotic liver disease (MASLD) are advanced diagnostic tools developed using machine learning and multitargeting approaches. These tests aim to improve the accuracy and reliability of diagnosing advanced fibrosis (F3–F4) in MASLD patients, addressing the limitations of conventional NITs.

### 1. **New Generation Markers Developed:**

Three new multitargeted NITs were introduced in the study:

• **FIB-9**: A model based solely on commonly available blood parameters.
• **FIB-11**: A model incorporating additional specialized fibrosis markers.
• **FIB-12**: The most advanced model, combining blood-based markers with liver stiffness measurement (LSM).

These models were developed using **ADORE machine learning software**, which optimized predictions across multiple fibrosis stages simultaneously, surpassing the accuracy of traditional single-target statistical methods.

---

### 2. **Performance of New Generation Markers:**

• **FIB-12**:
• Achieved the highest diagnostic accuracy at **83.3%** in binary classification (AUROC 0.912).
• Outperformed conventional NITs like FIB-4, FibroTest, FibroMeter, ELF, and Elasto-FibroMeter (EFM), all of which had accuracies below 80%.
• Demonstrated superior accuracy over traditional tests in both binary and ordinal segmentation analyses, with over **90% accuracy in 41.5% of patients**.
• Showed robust performance across different patient subgroups, including those with diabetes, older age, or varying disease complexity.
• Was particularly effective for diagnosing fibrosis stages F2–F3, which are critical therapeutic targets for emerging MASLD treatments like resmetirom.

• **FIB-9**:
• A practical and cost-effective option, relying on routine laboratory parameters.
• Achieved accuracy comparable to commercial fibrosis panels (AUROC 0.863).
• Available as a **free online calculator**, making it highly accessible for primary care and mass screening.

• **FIB-11**:
• Showed slightly reduced precision in older or diabetic populations but still demonstrated strong overall performance.
• Can complement FIB-12 in quaternary segmentation strategies to enhance diagnostic reliability.

---

### 3. **Comparison with Conventional Markers:**

Traditional NITs, such as FIB-4, FibroTest, FibroMeter, ELF, and LSM, have been widely used for assessing liver fibrosis. However, these tests often fall short in terms of diagnostic accuracy, particularly in MASLD, due to the metabolic complexity of the disease. For example:

• FIB-4 and FibroTest typically have accuracy rates below 80%.
• The ELF test demonstrated an AUROC of 0.865, which was significantly lower than FIB-12's AUROC of 0.912.

The new generation models (FIB-9, FIB-11, and FIB-12) represent a significant advancement, providing higher accuracy, fewer indeterminate results, and enhanced clinical utility.

---

### 4. **Clinical Implications of New Generation Markers:**

The development of these AI-based NITs has several key clinical implications:

• **Improved Diagnostic Accuracy**:
• The new models, particularly FIB-12, offer better precision for identifying advanced fibrosis, addressing the diagnostic gaps of traditional tests.
• FIB-12 showed strong diagnostic performance for F2–F3 fibrosis, the therapeutic target range for MASLD treatments.

• **Reduced Indeterminate Results**:
• FIB-12 achieved the lowest indeterminate rate (16.4%) and the highest accuracy (up to 92%) in determinate cases when patients were classified into ruled-in, ruled-out, and indeterminate groups.

• **Cost-Effectiveness and Accessibility**:
• FIB-9, as a free online tool, enables healthcare providers to estimate fibrosis stage using basic lab results, making it ideal for mass screening and use in primary care settings.

• **Enhanced Screening and Staging Efficiency**:
• The multitarget design of the new models allows for more reliable staging of fibrosis, which is critical for clinical decision-making and treatment planning.
• The **Fibs+ algorithm**, which integrates FIB-9, FIB-11, and FIB-12, outperformed existing clinical guidelines from the AGA, AASLD, and EASL for fibrosis detection.

• **Broader Applicability Across Patient Populations**:
• FIB-12 maintained high accuracy across different subgroups, including older patients and those with diabetes, ensuring its utility in diverse clinical settings.

---

### 5. **Conclusion:**

The new generation of non-invasive tests—FIB-9, FIB-11, and FIB-12—represents a major advancement in the assessment of liver fibrosis in MASLD. These AI-enhanced models deliver significantly higher accuracy, fewer indeterminate results, and broader clinical utility compared to conventional NITs. Among these, **FIB-12** is the most accurate and reliable tool for advanced fibrosis diagnosis, while **FIB-9** offers a practical, cost-free option for mass screening. These tests are poised to improve clinical workflows, enhance early detection, and facilitate better management of MASLD.