By some estimates, as many as two-thirds of studies about new drug targets cannot be replicated owing to factors including statistical flaws, incomplete methodological descriptions, publication bias and outright fraud. Researchers have now added another irreproducibility culprit: genetic instabilities in cancer cell lines that cause seemingly standardized cells to behave differently from one another.

Todd Golub, CSO at the Broad Institute, and his colleagues analysed whole-exome sequences from 106 cell lines that were housed at both the Broad Institute and the Sanger Institute to explore their clonality and genetic stability. A median of 19% of the mutations they detected were only present in one set of paired cells.

In a deeper dive into whether these genetic inconsistencies translate into differences in drug responses, they administered 321 anticancer compounds to 27 strains of a common breast cancer cell line called MCF7. At least 75% of the compounds that blocked growth in some strains were completely inactive in others, they report in Nature.

“Established cancer cell lines, generally thought to be clonal, are in fact highly genetically heterogeneous,” they conclude. “Cancer cell lines remain a powerful tool for cancer research, but their genomic evolution leads to a high degree of variation across cell line strains, which must be considered in experimental design and data interpretation.”

Last year, Golub's team also showed that genetic inconsistencies can muddy the reliability of patient-derived xenograft mouse models of cancer, in which cancer cells are harvested from patients and injected into mice.