One can start with immortalized but non-malignant cell lines of epithelial and lymphoid origin. Further, one can also use the primary normal cells from healthy donors.
Normal primary human fibroblasts and keratinocytes are probably the best general choices. These cells have a finite lifespan in culture, are non-transformed, and non-malignant. Another good, but not nearly as good, control would be an immortalized non-malignant cell line that is of the same general type as the tumor cells you are studying. For example, we studied a variety of pancreatic tumor lines and compared tumoricidal activity with NDV to normal human fibroblasts, KCs, and to a pancreas cell line that had been immortalized by inserting a viral gene. This cell line continued to exhibit a number of pancreas specific secretory products and markers, so they were somewhat similar to normal cells even though they were transformed.
The use of "normal" cells growing in a 2D environment as prisoners of a plastic flask in "comparison" with pure populations of cancer cells also growing in a 2D environment and also prisoners of plastic flasks is of no sense with the current scientific knowledge about the tumor microenvironment (see some selected articles for you ...).
And on top of that ... using the "strategy" my colleagues propose here would have led etoposide discarded from clinical development (here also an article for you ...).
In vitro growth assay must be used on cancer cells only to define the GI50 concentration, which in turn must be used to identify the target(s) of the compound.
All the best,
Robert
PS for those of you who trust in in vitro experiments to "develop" anticancer drugs: I attached three articles.