Recently we got a review of a manuscript saying that "data did not follow dose-response patterns". I know about linear models of response, as well as log-logit and probitos-like distributions. However, I was wondering should all biological responses fit in those models?
Hi Abel, of cause there are several models that can conveniently describe the response of a biological system to a certain dose or concentrations, some of which you already listed. I would, however, not necessarily regard the fit of any single model as a prerequisite for the existence of a relationship between both dose and response. More important is that the relationship is somewhat continuous or even monotone. If your data was just randomly going up and down without any pattern or trend, one might doubt that there is such relationship (although there might be one that is just masked by another effect). Hope that makes sense. Best, Markus
Yes Markus, it does make sense. But you touched a very interesting point "the monotone". Consider a situation where you have responses significantly and consistent differences, but not monotonically variated (which is quite common in biological data). In this scenario, if experimental design/methods is appropriate, you might have mainly two options to understand those results. First option is to consider that there wasn't real effect and differences are randomly distributed. Second option is to consider that we don't know the proper dose-response curve and/or it might be not monotone.
We trend to choose the first option, but for me with this we imply that our stats aren't good enough and we should change experimental methods or implement additional stats.
Your response conduces to another 2 related question:
1- Is there real data showing that at a properly designed experiments, the existence of dose response is conditioned or likely to be monotone (or any distribution in particular)? (2) or this is more something that we expect without much guessing much why?
Hi Abel, yes, I am totally with you. A missing (obvious) dose- or concentration-response does not mean that there is no effect. Furthermore, if your data is not monotone, it does not mean that there is n no relationship, only that it is not as simple. However, if you observe a polyphasic response, it would be necessary in my opinion to know why this is the case if you would want to describe this using a mathematical model, right? Best, Markus
In Ames test a toxic effect is some times observed at higer doses inducing the highest mutagenic response
Let's take a simple biphasic response (very common...). Can anyone give an example for a good mechanistic explanation for its shape/parameterization?
Many compounds have a dual response at different doses. For example ascorbic acid at low doses is antioxidant and at higher doses is pro-oxidant.
yes, it is common, but I was more after a good description of a mechanism of such biphasic response.
Elena that is really a very interesting question. I've been looking for recent scientific articles in this subject and actually it's hard to find a mechanistic information about that. However, we can find in some mathematical model & description books is that biological systems are very dynamic and likely not monophasic. It seems to me is that Myriam's answer is the first step of (at least) three to understand those processes. Firstly, chemicals change properties (most of time not linearly) according to their concentrations and surrounding features. Secondly, and maybe more important, biological systems are able to detect those non-linear/non-monophasic changes and respond in order to maintain homeostasis, generating also non-monophasic responses. Finally, these biological responses are expressed in different biological hierarchy levels, which have distinct reaction and interaction times, which are interdependent and related. Lets follow the example of Myriam ascorbic acid in one oxidative stress enzyme. This acid could induce or reduce expression superoxide dismutase (SOD) genes in a biphasic concentration dependent way. These genes responses are typicaly quick and exponential. In other hand, transcription of genes in SOD enzyme can be longer depending of gene (exponential) expression but also subtract concentration (oxidant species related to ascorbic acid biphasic influence) as well as "building blocks" availability. Finally, the activity of this enzyme (which is usually measured) would be also dependent on the presence of others antioxidants (linear or log-linear dependent) as well as daily physiological metabolism (sinusoid dependent). My impression is that at the end, if biological systems still can maintain homeostasis, physiological responses are commonly to be nom-linear and monophasic. Does it make sense for you guys?
Dear All,
Thanks for all the responses on this topic. As we continued the research we could clarify some features of the non-monotonic toxic effects of a pesticide considered safe.
Thus, it might be of interest to some of you.
All the Best
Abel
http://pubs.acs.org/doi/abs/10.1021/acs.est.6b03056
Dear All,
I just found this blog http://ourstolenfuture.com/newscience/lowdose/nonmonotonic.htm a bit old but potentially interesting to some of you.
Cheers
thanks, great, especially because endocrine effects should follow the biphasic curve very often.
The question still stands, especially with regard to the variety (or similarity?) of the mechanisms involved:)...
- Badges
- Science topic
- Similar topics
- Microbiology
- Assays