Back to homepage


Addendum for Pitfalls, Common Errors, and Frequently Asked Questions

[For more details, see Chou T.C. Am J, Cancer Res. I:925-959; Free access; Cancer Res. 70:440-446, 2010, and Synergy 1:3-21, 2014]

  1. Synergism/Antagonism quantification is a mass-action law issue (determined by the CI values), not a statistical issue (not determined by the p values).
  2. For any drug combination synergy determination, the dose-effect curves of each drug alone are "always required" [e.g., (Dm)1, m1; (Dm)2, m2]; For constant ratio combinations,(Dm)1,2, and m1,2 values are required since these parameters are used to calculate/simulate the CI values, where CI<1, =1, and >1 indicate synergism, additive effect, and antagonism, respectively.
  3. Two-fold serial dilutions are usually performed for in vitro experiment for each drug alone and their mixture to create 5-6 concentrations, with their Dm values (IC50 values) located in about the middle of the concentration ranges. The diluted mixture (e.g., in [(IC50)1/(IC50)2] ratio is always at a constant ratio when it is diluted. The non-constant ratio combination can also be used for CI calculation. But it is less efficient in analysis or in computation due to changing ratios, which does not allow a simulation.
  4. You only need 15-18 dose-effect data points (5 to 6 points for each drug and their mixture plus an uninhibited control) for two-drug combination in vitro to determine synergism or antagonism. The in vitro experiments ususally take 1-2 weeks to complete and computer simulation usually takes about one second after the data entries.
  5. Never enter un-reliable dose-effect data into the computer [e.g., 0% inhibition, fa = 0, which means <<0.1% inhibition, fa <<0.001; or 100% inhibition; fa = 1, which means >>99.9999% inhibition], unless the assay is extremely accurate. These will cause the computer to crash since log 0 is -∞. Delete the unreliable or erroneous data points (usually at low and high extreme of doses or concentrations). There is no specific number of multiple doses (or concentrations) that is required [e.g., D1: 5 concentrations, D2: 6 concentrations; Mixture: 4 concentrations]. They are flexible as long as they are accurately determined. Inaccurate data points of assays (e.g., poor r values) cannot provide accurate synergism or antagonism determination.
  6. For the combination of activators, e.g., fractional survival or fractional cure, you need “Vmax” equivalent or “cure” equivalent. For example, if immunosuppressant for organ transplant to have 70 day survival is a long term survival (fa≈1), then 35 day survival is fa=0.5. Alternatively, in the median-effect equation, fa= (C-T)/C, (fa/fu) = (C-T)/T for inhibitor ; and fa= (T-C)/T, (fa/fu) = (T-C)/C for activator, where C is control and T is tested with a drug or an effector. The CI equation was derived for the inhibitors or the reference ligands. For activatiors further indeepth studies may be needed for the development of the generalized methodology.
  7. For review article, also see Chou, T-C Am. J. Cancer Res. 1:925-954, 2011. ( and Chou TC. Synergy 1:3-21,2014 (Inaugural article, )