Figure 3. Inter- and intra-observer variability of visual classification.

(A) COX/SDH (top) and quadruple immunofluorescence (bottom: COX-I - green, NDUFB8 - purple, porin - red and laminin - white) were performed in two serial muscle sections from patients P13 and P16. 100 myofibres from each patient were visually classified (COX/SDH) by two independent investigators, and objectively classified (COX-I and porin immunodetection). (B) Bar graphs show the percentage of COX-positive (beige), intermediate(+) (light beige), intermediate(−) (light blue) or deficient (blue) based on visual classification by investigator 1 and 2 (first two bars) and objective classification (last bar). In patient P13, both investigators identified the majority of COX-positive cells correctly (40% versus 42% using immunofluorescence) but differed in other categories. In patient P16, the investigators differed markedly in all categories. Importantly, both investigators demonstrated a high degree of internal inconsistency.