Use your negative control data for the zero concentration data point.

In experimental pharmacology the "ZERO" concentration response can not be there as the dose response curve is always some what ' S' shaped, and lower and higher side of curve is a flat response.

.... I would use the negative control as a zero, as suggested by Stefan Rose.

Thanks very much for the replies folks. How should I use the negative control data? Do you reckon I should select a concentration which is say, an order of magnitude or two below my lowest concentration and use the negative control data there?

For the stats try open source R much better than prism lots of on line guides and free!

http://www.r-project.org/

I agree with Japie (see comment above). In my lab we use GraphPad, Origin, and Sigmaplot. In graphpad it is easy to constrain since you just click the box and set the bottom to zero; but in SigmaPlot and Origin you have to change the formula of the curve that you are using for the fitting purposes. Given your equation, A2 would need to be set to zero to set the bottom of your curve.

Dear David.

I have not experience in this particular topic of pharmacokinetics, however, ¿is it posible that several changes on pharmacodynamics could be produced? ...

The solution proposed by Japie is apprpriate

What is the route of dosing???? How many doses do you have? What are your LLOQ and ULOQ ? How many time points do you have ? At what time point did you take the first sample? Can you give us your concentration data against time? Do you have predose concentration data (assuming an endogenous metabolite interfering with the analysis?) It is very hard to give an answer without the experimental details.

most important thing is whether you are plotting dose or log dose against effect. it is recommended to be logdose to to obtain a more meaningful sigmoid DRC. if it is so, as Dr Chakravarthy says, there is no need to have any zero dose/log dose. it is simply zero response for zero concentration / dose or its log.

Dear all,

Thanks very much for your replies. Those inquiring about pharmacokinetics: to clarify, these responses were measured in isolated vessels ex vivo.

Dr Coulthard: thanks, I did use R for the stats but have not yet forayed into code-driven graphing and curve fitting!

Drs. van Tonder, Henderson and Raina: I should have defined the terms in the formula I provided but you have interpreted very well. This does sound like the solution I was looking for. I will try this today and consider this experiment into crowd-sourced knowledge a success in any case.

Iam sorry not know

Setting A2=0 had exactly the effect I was hoping for. The slope of the curves is much as it was, but they now tend to zero response instead of a value just below that produced by the lowest studied concentration. I'm very grateful for everyone's help.

The question that arises here is: how well do you know the value of "zero"? Most of the time, zero is not actually zero, but is some control or background value that is measured with some degree of error. In this case, constraining the curve to go to control at zero agonist has the effect of moving error in the measurement of control into the fitted parameters. In this situation, the most accurate estimate of the fitted parameters will be obtained by using your control value as a data point and (assuming that you're using log concentration so you can't actually enter a concentration of zero) assigning it a "concentration" many orders of magnitude below the EC50. Of course, if you have a huge number of replicates of control, such that the standard error of control is tiny compared to that of your other data points (i.e. you know the value of "zero" very well), the result will be indistinguishable from simply forcing the curve to zero.