Interlaboratory variability in fluorescence in situ hybridization analysis

Reliable interpretation of fluorescence in situ hybridization (FISH) data, especially data that have been generated in more than one laboratory, requires knowledge of the sources of variability inherent in FISH analysis. Possible sources of variation may derive from differences in sample preparation, probes used, intrasample heterogeneity, hybridization protocols, counting criteria within and between scorers, fluorescence microscopes, and filters. This study characterized the relative weight of some of these factors in order to determine the degree to which FISH results are comparable between laboratories. We used a hierarchical partitioned chi 2 analysis to measure sources of variation. We found that replicate counts varied no more than expected based on counting statistics (i.e., multinomial variation). However, with replicate hybridizations done in two separate laboratories, the variability increased significantly. Thus, care must be taken when interpreting FISH data that are derived from more than one institution. Previously agreed upon counting criteria as well as standardized FISH hybridization protocols may decrease this variability.