How effective is a 'clinical fibrosis assessment' using non-invasive fibrosis tests in patients with nafld?

Introduction

Application of at least two non-invasive tests (NITs) for fibrosis is recommended by international guidelines to stage liver fibrosis in primary/secondary care. Generally, NITs differentiate advanced fibrosis (F3–4) from milder stages (F0–2) with reasonable accuracy, providing a high negative predictive value. Many clinicians do not employ the structured approach advocated in guidelines and it remains unclear how reliable clinicians are in interpreting NIT results. Our aim was to assess the concordance between NIT-based ‘clinician fibrosis assessments’ by hepatologists and histology in patients with NAFLD, and compare this with algorithmic approaches.

Methods

Six consultant hepatologists, blinded to histology, independently staged fibrosis (F0–2 vs F3–4) for 230 patients with NAFLD using FIB-4 alone, FIB-4 plus ELF, and FIB-4 plus Liver stiffness measurement (LSM; Fibroscan). The same patients were used for each assessment with patients anonymised and presented in a different random order each time. Clinicians were also provided: age, sex, BMI, ALT, AST, ALP, albumin and platelets. Concordance between histology and clinical fibrosis assessments or algorithmic approaches using combinations of FIB-4, ELF and LSM were assessed.

Results

230 patients were included (median age 54 [22–78] years; 55% female; median FIB-4 1.12 [IQR 0.78–1.91]; ELF 9.3 [IQR 8.6–10.2]; LSM 9.4 [IQR 6.3–14.3]; 41% F0–1, 22% F2, 21% F3 and 16% F4). Overall, raw test performance was good for staging histologic F3–4 (AUROCs: 0.84 (CI 0.79–0.89) for FIB-4, 0.86 (0.80–0.91) for ELF and 0.86 (0.81–0.91) for LSM). Concordance between the 6 clinicians for F0–2 vs. F3–4 fibrosis was good (κ=0.64 [0.60–0.67] for FIB4 alone; κ=0.70 [0.67–0.74] for FIB-4 plus ELF; κ=0.69 [0.66–0.72] for FIB-4 plus LSM). Table 1 shows the concordance between histology and clinical fibrosis assessments and algorithmic NIT fibrosis assessments for F0–2 vs. F3–4. Overall, there was variability in concordance between individual clinician fibrosis assessments and histology and this was reflected in variability in the false positive and negative rates for F3–4. Concordance with histology was better when clinicians used FIB-4 plus LSM or ELF rather than FIB-4 alone. Purely algorithmic approaches, particularly sequential use of FIB-4 then LSM, tended to perform better than the clinical fibrosis assessments, with only one clinician having marginally higher concordance than the best performing algorithmic approach.

Conclusions

Overall, adhering to the recommended algorithmic approaches using NITs to stage fibrosis (F0–2 vs. F3–4) performed more consistently than less-structured ‘clinical’ NIT-based assessments conducted by consultant hepatologists. Adoption of algorithmic approaches will reduce variability in patient care.