PCR data analysis if the efficiencies aren't equal - real time PCR troubleshooting (Jun/24/2009 )

Hi everyone!
Now I am trying to perform the real-time quantitative RT-PCR (TaqMan) for the detection changes in gene expression some gene involved in drug resistance of ovarian cancer patients. For comparison the changes of expression in cancer tissue I use pool of the samples from normal ovarian tissue (my calibrator sample). As endogenous control I select 18S RNA. The real-time PCR reactions are performed by Applied Biosystems 7300 Real-Time PCR System. And I have some problem with my experiment. First, CT 18S RNA is very high (close to 9-11). Second, I have positive meaning CT from no template control (NTC) sample and control sample without RT. In this case Ct meanings are low (33-38), but I think CT meanings should be absent. Contamination problem is excluded by repeating analysis with a new set of reagents.
What is the matter? Should I change my endogenous control and dilute sample?
Also I tried to use comparative CT method (ΔΔCt), but validation experiments didn’t show that the efficiencies of the target and endogenous control amplifications are equal. May be, someone can give advice about the methods of data analysis gene expression where standard curve and equal CT for target and endogenous control are unnecessary.
Thanks in advance!

If your Ct for 18S RNA is 11 and your no template control is 33, this suggests a minimum of an over 4 million fold difference in the amount of template present in these two samples. Ct values around 38 are usually due to amplification of noise, amounting to a basically undetectable signal, so your negative controls are probably fine. However, with such a high concentration of template, you probably want to dilute your cDNA 1:1,000 or 1:10,000 (should yield about a 10-14 cycle increase). This way, you'll have more sample for future experiments and your CT values will be in a reasonable range (should be in the linear range of amplification for your primers).
Also, if the efficiencies of the target and endogenous control primer sets are different, you should use the Pfaffl correction to take the efficiencies into account rather than ΔΔCt, which is only an approximation technique and is only applicable when efficiencies are very similar (many users ignore this).

See the paper: Pfaffl, M. (2001) A new mathematical model for relative quantification in real-time RT-PCR. Nucleic Acids Research 29 (9): 00-06

Without going into the Math behind this, if the efficiencies of your assays differ by 3% (assay #1 is 100% and assay #2 is 97%), then the difference between them is 35% after 20 PCR cycles and 57% after 30 cycles (note: a 2-fold difference is a 100% difference). In other words, if you are looking to distinguish a 2-fold difference between your treatments, and your Ct values are around 20-30, you can assume that the error in your experiment due to the differences in assay efficiencies is 35 to 57% or 1.35- to 1.57-fold. Simply put, if you identify a 2-fold difference between sample, up to 1.57-fold of it is due to "background noise" (the difference in efficiency between assays).

Of course the Pfaffl method corrects for this.