Dear Sachin Khiste, 

The simplest one to know is: if you need to estimate parameters, e.g. clearance rate, the best is compartment analysis.  Non compartmental analysis gives you generic information, e.g. area under the curve; it sets no boundaries on your system, it is not modeling based, rather, analysis based. It is generally applied when you do not need internal details of your system, dynamics.  

I have selected some references that might help to understand better. 

Based on your comments: "Im working on some of the pharmacokinetic parameters of our new drug. " it seems like compartment analysis. 

What may happen is that you can apply non compartmental analysis to estimate the parameters for a compartmental analysis (https://www.mathworks.com/videos/integrating-pkpd-and-mechanistic-modeling-with-simbiology-and-matlab-81930.html)

"I have inject single dose my compound in mice IP and collected serum from 0 to 36 Hr. " from this part, you can apply both.

"I calculated the serum concentration using HPLC." I have no idea what is HPLC, but must be a measuring technique. 

"Now I have to do the PK analysis using software Kinetica. " try also simBiology, they have very nice tools. 

"So just want to know is my study is compartmental or non-compartmental." In case of doubt, apply both, they are complementary; non compartmental is easier. 

Some nice links and references:

• Non-compartmental Analysis (https://it.mathworks.com/help/simbio/ug/non-compartmental-analysis.html)
• Non-compartmental analysis (https://www.ncbi.nlm.nih.gov/pubmed/23007438)
• How does Matlab Simbiology calculate pharmacokinetic (PK) non-compartmental (NCA) area under the curve (AUC)? (https://it.mathworks.com/matlabcentral/answers/214569-how-does-matlab-simbiology-calculate-pharmacokinetic-pk-non-compartmental-nca-area-under-the-cur)

PS.  I have uploaded the following video, I hope that helps somehow: https://www.youtube.com/watch?v=cbVULmI18do&feature=youtu.be

Since you are working with a new drug and the route of administration is extravascular, you may want to fit your data with 1 or 2 compartmental model and see how it explains your drug kinetic profile.

Dear Showande Johnson Segun,

Well said. He may start from simple to complex, according to the level of details he needs. First NCA, then one compartment and then two compartments. 

However, the shift from two compartments and more is justifiable just if he can prove that the drug has different absorption rates, like kidneys and similar organs and other tissues.

The rule is always the same, from simple to complex. 

Hi, First NCA without any assumption, to calculate AUC, Cmax, This should Support your objective. You may try to fit a compartmental model later on, depending on the type and Quality of data.

I agree, first Cmax and AUC can be calculated without any adjustments. Then for the rest of parameters: A pharmacokinetic adjustment to find the best model to define the experimental data helps to estimate better results for PK parameters. If we don't check if our drug describes or not a compartmental behaiviour we can do more errors in the estimated parameters. If we check what is the best model to define the PK behaviour we are more proxim to real values. As example if the plasma concentration versus time curve had 2 phases (with 2 different slopes in the log adjustment) and at first we assume (without check) that it is non compartmental = only pone solpe , the slope give us the half-live, only 1 half life, when realy the experimental points defined 2 solpes with 2 half lives (alpha and beta). All estimated parameter wuill be different in value than parameters estimated using the compartmental model. If we use the best adjustment we find the best estimated parameters. It is a recomendation to check before waht is the best model because the estimated values for PK parameters could be very differents and we want the values more adjusted to real points

Dear Dr Marcus Schultze-Mosgau and Laura Garcia Capdevila, 

Thanks for the further details. 

Dear Sachin Khiste, 

I believe now you have all you need, good luck and if possible, post here if you succeeded. 

Best regards,

Jorge Pires,

Dear Sachin,

I agree on the approach suggested by Marcus and Jorge performing NCA first.

Keep also in mind that NCA is usually the method to assess toxicokinetic data and it is often useful to be able to compare parameters over studies based on the same evaluation method.

I would like to mention an additional aspect (besides starting wth the easiest model first) when deciding on the number of compartiments which can be included in a model: Every compartment (elimnination rate) need to be based on a sufficient number of time points (should be at least 3) - and the fraction of the partial AUC in relation to the total AUC should be meaningful for each compartment. 

Unfortunately, I have seen "mathematical model optimizations" which have not been justified by the number of available data. Thus, I made this comment as you mentioned the last sampling time but neither the number of blood draws taken nor the value of the terminal elimination half-life of your compound.

Best regards,

Clemens

Dear Clemens Guenther, 

I agree with you regarding the number of data.

One possible way to "solve" the problem, if he cannot generate new data, is creating "fake" data, I have used this technique recently, and it may be useful when we cannot generate more data. However, these methods must be applied just when it is not possible to generate new experimental data, since the information content is always smaller.

Regading "mathematical model optimizations", indeed, the optimization method cannot do miracles, we need to pay attention to the quality of the data, information richness. 

Best regards for all,

Jroge Pires,

Thank you so much guys