Skip to main content


Figure 5 | BMC Cancer

Figure 5

From: Optimisation of an immunohistochemistry method for the determination of androgen receptor expression levels in circulating tumour cells

Figure 5

Incurred sample reanalysis in AR staining intensity determined by β-content γ-confidence tolerance intervals. Due to the fact that the set of samples analysed blindly by two different operators in the first (Figure 2) and second (Figure 4) validation experiments were identical, this also allowed a conventional analysis of results by ISR utilising β-content γ-confidence tolerance intervals (±) at β = 95%. The resulting accuracy profiles clearly demonstrated that the training programme impacted almost exclusively on operator 2, correcting the between-operator bias in the process (Figure 2). They also highlight that the degree of ISR achievable in this analysis by operator 1 approached 30%, which is the accepted benchmark for total error for a typical quantitative biomarker assay.

Back to article page