Liver biopsy in the diagnosis of malignancy

In a National Audit of 1500 liver biopsies, 38% were for suspected malignancy. To measure their contribution to clinical decisions, the initial diagnoses, biopsy diagnoses, final diagnoses, and outcomes were coded by computer and compared. Most patients (92%) were investigated for advanced malignancy. The accuracy of clinical diagnosis was 78% against final diagnosis. Liver biopsy was seen as 'confirming' clinical diagnosis overall. This was achieved in 67% (75% with ultrasound guidance), and specificity was almost 100%. However, hepatocellular cancer was confirmed by biopsy in only 32% and haematological malignancy in 13% of suspected cases. Within 3 months, 44% of patients with histological malignancy had died. Histological tumour type was not used in 36% of final diagnoses. Of patients with a malignancy-negative liver biopsy--showing reactive hepatitis, normality, or cholangitis/cholestasis--25%, 47% and 60%, respectively, had final malignant diagnoses. In 6% of patients, biopsy showed chronic liver disease. Only 12% of deaths were autopsied. Liver biopsy contributes very high specificity to the diagnosis of malignancy, and detects non-malignant disease. Failure to use tumour type may result in sub-optimal therapy. Improving diagnostic practice requires more information on outcomes, including autopsies.