# Modeling Post-Infusion Data in Monolix

Good afternoon everyone,
I have pharmacokinetic data from a population only post-infusion plasma concentrations. S, plasma drug levels were obtained following the end of the infusion. So, the peak plasma concentrations are at time zero and decreases from there.

My question is that, how do I implement this popPK of post-infusion data in Monolix?

I plan to use “X” as the regressor with the initial conditions at time=0. Do you have any suggestions?

Here is what I get when I try the initial parameter estimates: Here is the model that I am using:

[LONGITUDINAL]
input = {Cl, V1, Q, V2, Cc_init}
Cc_init = {use=regressor}

EQUATION:
V = V1
k = Cl/V1
k12 = Q/V1
k21 = Q/V2

Cc = pkmodel(V, k, k12, k21)
Cc_out = Cc+Cc_init

OUTPUT:
output = Cc_out

Thanks…

1 Like

Dear Andy,

I think that your model is well implemented but there are incoherent informations
1/ Using this model, the output can not go to values less than Cc_init due to its structure. Thus, on your check initial fixed effects, the red curves can not decrease a lot and not toward 0.
2/ The parameters you set (at least in terms of V and Cl) are big. Thus, if I make a unitary bolus on it for example, I’ll have a concentration that can not go over 2.5e-5
3/ If you make a bolus or an infusion, the calculated concentration (Cc) will not be zero at the end of the infection, thus setting Cc_init to this value might not do what you want

Two suggestions

• Let Monolix estimate the Cc_init
• Modify Cc_init to take the initial concentration (post infusion) into account. For that, I think you need to calculate the concentration (explicit formulas of the solution are given here (http://mlxtran.lixoft.com/libraries/)) and take the dosing into account (key words to use administration information in Mlxtran can be found here)

Best

Jonathan

1 Like

Hi,

It seems that you have no AMT column in your data set, such that the pkmodel macro has no input and stays at 0 all the time. Consequently, CC_out stays at Cc_init all the time too.

Here is a “trick” that might be useful in your case:

• format your data set like this: • set the Cc_init column as AMT column-type when loading the data
• your model is then just:
[LONGITUDINAL]
input = {Cl, V1, Q, V2}

EQUATION:
V = V1
k = Cl/V1
k12 = Q/V1
k21 = Q/V2
Cc = pkmodel(V, k, k12, k21)

OUTPUT:
output = Cc

Best,
Géraldine

1 Like

(I could not put 2 images per post, so here is the result)

After the steps in the previous message, I get with my dummy data: Best,
Géraldine

1 Like

I forgot to add: one should hack the bioavailability to convert the amount into a concentration. Like that:

[LONGITUDINAL]
input = {Cl, V1, Q, V2}

EQUATION:
V = V1
k = Cl/V1
k12 = Q/V1
k21 = Q/V2
Cc = pkmodel(V, k, k12, k21,p=V)

OUTPUT:
output = Cc

With this trick, the volume V does not need to be fixed to 1 to recover the right initial concentration, and can be estimated.

1 Like

Hi Geraldine,
That was simply amazing instructions…here are the updated PopPK fit results from Monolix…simply amazing. Thank you as well Jonathan.

Geraldine, I used the updated model you sent and it worked very well.

This was really helpful to post this question…I have been trying this for some time now.

Thanks again…this is great.

Andy

1 Like