Reproducibility of liver iron concentration measured on a biopsy sample: a validation study in vivo.

ABSTRACT

Determination of liver iron concentration is essential to predict iron related tissue damage and to guide chelation therapy. To assess the reliability of a single biopsy iron concentration determination in representing the whole liver iron concentration, we conducted a prospective study performing two immediately successive liver biopsies from 61 noncirrhotic, iron overloaded thalassemia patients, directing the needle to different direction from the same skin cut. The correlation among sample biopsies was determined by both regression analysis and the Bland-Altman method. The results showed that overall correlation between the two samples was high (Pearson's coefficient of correlation r = 0.970, P < 0.0001; 95% CI 0.951-0.981; R(2) = 0.941). To evaluate if sample dimension had an impact on the analysis we analyzed separately biopsy couples were both sample gross weight were ≥1 mg dry weight (n = 16) from the others [one or both had a specimen gross weight <1 mg dry weight (n = 45)]. In the first case, correlation coefficient r was equal to 0.998 (P < 0.0001; 95% CI 0.995-0.999; R(2) = 0.996) while in the latter was 0.960 (P < 0.0001; 95% CI 0.928-0.977; R(2) = 0.921). In no instance second specimen prediction interval was outside the interval implying different prognostic and therapeutic decision if both liver samples gross weight were ≥1 mg dry weight. The Bland-Altman plot analysis showed the same trend observed using Pearson's correlation coefficient in the analyzed sample categories. Hepatic iron concentration determined in "good quality" biopsy specimen (i.e. sample gross weight ≥1 mg dry weight) is a reliable indicator of whole liver iron concentration.