Introduction ALT is one of the most commonly used tests to detect liver disease for further investigation, but its accuracy is uncertain. Currently there is no systematic review of diagnostic accuracy of ALT in detecting liver fibrosis using liver biopsy as reference standard.

Methods Standardised methods for conducting systematic reviews were used. MEDLINE, EMBASE, and reference lists from articles were searched. Studies were included if they evaluated paired samples of liver biopsy and serum, with extractable data for ALT, >30 participants, and presented data as sensitivity, specificity, predictive values, likelihood ratios, or ROC curves or included sufficient data to calculate these parameters. ALT above upper limit of normal threshold were defined using local values and ranged from 30 to 60 u/l.

• Download figure
• Open in new tab
• Download powerpoint

Abstract PTH-085 Figure 1

Results Initial search located 9563 abstracts, 344 papers were assessed, of which 9 met inclusion criteria. The majority of included studies were published in last 10 years, were from secondary specialist care, and conducted in USA (3), Europe (3), and rest of world (3). Median study size was 206 (IQR 124–341). Median proportion of moderate/severe fibrosis on biopsy was 22% (range 6–34), any fibrosis 81% (range 58–100). Pooled sensitivity for ANY fibrosis was 87% (95% CI: 85–88) specificity 34% (95% CI: 30–38), positive likelihood ratio 1.3 (95% CI: 1.2–1.4). Moderate/severe fibrosis sensitivity was 89% (95% CI: 87–91), specificity 35% (95% CI: 32–37), positive likelihood ratio 1.36 (95% CI: 1.3–1.42). AUROC values were reported in 2 studies - moderate/severe fibrosis AUROC=0.815, and severe fibrosis AUROC=0.59.

Conclusion ALT accuracy in predicting fibrosis has only been studied in specialist clinic populations where it has low specificity for both any or moderate/severe fibrosis, but high sensitivity for both. Predictive values in clinic populations with prevalences found in these studies show that ALT has high PPV for any fibrosis (because of high prevalence) but low NPV. In contrast it has high NPV for ruling out moderate/severe fibrosis but poor PPV. Results from this review suggest that ALT should be used in conjunction with other tests and clinical features and not alone to identify liver fibrosis in clinic populations. Spectrum effects are likely to be important especially in those with no fibrosis so affecting the specificity. Further studies are needed in primary care populations of the accuracy of ALT versus robust non invasive markers, but these findings would suggest that a normal ALT can rule out moderate/severe fibrosis with reasonable certainty due to its high sensitivity and likely low prevalence in primary care though this may not be the case for mild fibrosis.

Disclosure of Interest None Declared.