If you want to analyze a dose-response relationship, then an experimental design with as many different concentrations as possible is preferable. There is no need to measure the response repeatedly for the same concentration. The data would then be analyzed by a regression (of a generalized additive model, if the functional relationship between concentration and response can't be expressed in a simple formula). There are many helpful and sensible models for the analysis of done-response curves, and it would allow you to investigate specific diofferences in the curves, like donse-sensitivity, maximal response, parameters like LD50 (dose needed for a half-maximal response) or Dmax (dose at which the response is maximal, if the curve has a peak or maximum), asymptotic response etc.

Thinking about these things is statistics - the question if one should go for one-way or two-way ANOVA after having the data is not.

As the horse has already bolted, the best advice (apart from thinking about the experimental design next time) depends on if the "before" and "after" values are measured in the same experimental units (->"repeated measures"). If this is the case, you can calculate the "after-before" differences for each exp. unit and analyze these differences with a one-way ANOVA. Otherwise, things are a bit complicated (either you have a BACI design [->two-way] or you a confounding of time and conc [-> not analyzeable in a meaningful way]).

Note that ANOVA requires that the assumption of a conditionaly normal distribution of the response makes sense. If not, you may consider a transformation of the response or a different error model (generalied linear model).

Thank you for the answer.