Reproducibility of the measurement of colonic proliferation using bromodeoxyuridine across laboratories

Although measures of colonic cell proliferation are being used as potential intermediate markers in chemoprevention studies, measurement standardization is still ongoing. This study was designed to assess the reproducibility of the labeling index quantification, as measured by bromodeoxyuridine, across four laboratories experienced in its use. Each institution submitted 10 slides, with one circled area of each slide to be scored. Each site followed its standard procedures for scoring colonic crypts; no attempts to standardize these procedures were made. There was high concordance among the laboratories on whether scorable crypts were present on a particular slide, but only two pairs of laboratories demonstrated agreement statistically greater than that predicted by chance. The overall difference among the sites on the number of scorable crypts was marginally significant (P = 0.083), and there was a highly significant overall difference in the magnitude of the labeling index (P < 0.0001). Sites 1 and 2 tended to have similar results, as did sites 3 and 4, most likely due to common training. Even with these discrepancies, high correlation (r > 0.75) was observed among the reported labeling index values for each pair of laboratories. Without standardized training, these laboratories may differ in the crypts considered appropriate for counting and in whether cells are counted as labeled or unlabeled. These results suggest that standardized training in scoring across all sites performing labeling index determinations is required to assure reproducibility across sites or studies. These results may also help explain discrepancies in the average values of the labeling index reported in the literature.