Certain lines after many passages (or engineered lines) actually can be resistant to chemotherapy drugs like cisplatin. 

To make things complicated, it is reasonable to speculate that culture conditions also affect some readout substantially: things like serum level, glucose level, the presence or absence of L-glutamine supplement. Previously we even found that cell density affects cell response quite significantly. 

For the purists (I met a die-hard one), cell culture is an unacceptable approach because "in vivo" solid tumors exist as a lump of mixed cells, and bathed in an hypoxic environment. 

***

So, a rational way to approach it, would be first to pick one specific cancer type of interest to you (say breast cancer), choose several related lines, and focus on those first. 

Typical cytotoxicity period is 24 - 48 h, though 72 and 96 h also are used (though certainly less common), depending on cell types and experimental need. 

Promega salesperon told me about this "real-time" viability assay, in which you only add drugs once, but are allowed to read the plate multiple times (1, 2, 3 days). 

Of course, in the negative control, the cells keep growing, and the culture conditions deteriorate over time, until a negative control is just no good. Thus, longer incubation does not always mean better as artifacts creep in. 

I have not tried this assay, but we used Promega's other luminescence assays a lot, and they are facile and cool.

http://worldwide.promega.com/products/cell-health-and-metabolism/cell-viability-assays/realtime-glo-monitor-cell-viability-in-real-time/realtime_glo-mt-cell-viability-assay/

In short, the approach is: 

- decide a cancer type (read literature or consult experts)

- pick cell lines (read literature or consult experts)

- pick a good assay 

- optimize conditions 

- compare (justify and explain your results with literature again)

Good luck to your exp. 

The answer depends on many variables. First variable to consider is the assay endpoint (Apoptosis assay?  Release of LDH? Viability assay such as, e.g. MTT?). For simplicity I assume the use of a viability assay. In this case, a very short incubation times (a few hours) would require very high concentrations of carboplatin to induce measurable effect and the assay will detect only sensitivity of cells to undergo necrosis due to strong chemical insult. Thus, short times/high concentrations would not be recommended, if the purpose is to compare susceptibility of various cell types to carboplatin, because these conditions tend to obscure differences among cells (many cells types, which would under other conditions display differences in sensitivity, would show little difference after this "strong" treatment, because cell specific response mechanisms will be overwhelmed). On the other hand, “more days” of incubation tend to detect differences which are related to the differences in cell proliferation rates rather than biologically meaningful (mechanistic) differences in drug sensitivity. By other words, cells with shorter doubling time will tend to show as more sensitive, and the importance of this effect will increase with increasing incubation time. In this context, “more days of incubation” is relative and reflects differences in doubling times across cells that are to be tested and compared. So to avoid both these effects, incubation times around 2-3 days should be acceptable. However,  this answer is simplified and other variables need to be considered (keeping cells exponentially growing during the assay (which is desirable) depends on the cell load and assay time, and these variables in turn determine whether the assay is within limits of quantitation/linearity at the endpoint.)

Thank you, Roman and Nai-Kei, very useful information! Best regards, Dan

You are welcome Dan! 

There is one additional detail that I would like to share: Carboplatin has very similar activity profile to cisplatin, but carboplatin is much less potent than cisplatin in vitro. In addition, carboplatin and cisplatin have the same mode of action against  cancer cells and they even produce the same drug-DNA bivalent adducts. As a result, it is more reasonable  to use cisplatin instead of carboplatin in many experiments in vitro. Carboplatin has such a low potency in general that it would have been discarded as an inactive compound if its discovery had been entirely based on in vitro screening. On the other hand, I can imagine criticism by some reviewers: e.g. "You should not have used cisplatin in experiments with ovarian cancer cell lines, because ovarian cancer is treated with carboplatin and not with cisplatin!". Nevertheless, carboplatin is clinically used for being less nephrotoxic than cisplatin and not because it displays higher efficacy (or even different efficacy) than cisplatin. In fact, carboplatin and cisplatin display very high degree of cross-resistance (or cross-sensitivity), so validity of this type of criticism is questionable. I have uploaded an image demonstrating strong correlation between IC50 values for carboplatin and cisplatin across 78 cell lines in vitro (Pearson correlation coefficient is 0.79). 

http://s18.postimg.org/imd9mv9a1/cpt_vs_carbo1.png

To conclude, in most cell-based experiments in vitro I would use cisplatin instead of carboplatin. 

Best Regards,

Roman

Dear Roman, Thank you so much to share this with us. We use carboplatin exactly because we are working with ovarian cancer. I went through your publication and see your review article "Interactions of Cisplatin with non-DNA Targets and their Influence on Anticancer Activity and Drug Toxicity: The Complex World of the Platinum Complex", which is extremly interesting for me. I started working in the field of chemoresistance of ovarian cancer a couple of year ago, beginning with establishing high grade serous ovarian cancer cell lines. We generated quite a lots data. It seems to me that interaction of platinum with DNA, which is believed to be the major mechanism of its toxicity, may accounts for only (a small) part of the it. Our data suggest that the response of cells to carboplatin is an accumulating result of the effects of different functional pathways. There is a goog review article from Galluzzi et al about that. It will be very nice if I can have the full text of your article. I cannot download it in pubmed. Best regards, Dan

Dear Dan, thank you for your interesting comment. I will be very interested to see the details when they are available (published). There are facts supporting commonality of the mode of action of these two drugs, and there are also some facts  that may imply diversity in their modes of action. In my view, current evidence points more towards commonality of their modes of action. If we consider the role of non-DNA targets, as you suggest, we may reasonably expect some differences due to different reactivity of cisplatin and carboplatin with various biomolecules and that may imply different targets and different associated mechanisms; however, this is true only for their original molecules and once the drugs are activated (aquated) to form reactive intermediates, the differences in their reactivities are minimal, if any. Also their high cross-resistance and high similarity of their profiles across NCI60 panel suggest similarity of their modes of action (importantly, the similarity of mechanisms of drugs that display similar activity profiles across NCI60 cell panel has been experimentally validated in many examples). On the other hand, it will be very interesting to see the new evidence that challenges this conclusion, as it may bring new insight to the pharmacology of Pt-based anticancer drugs. Best Wishes, Roman