3 replies to this topic

[18963422692]

Dear Useers,

   The mouse xenograft PK/PD relationship was established by relating mouse PK concentration to measured xenograft tumor volume data using an optimized cell distribution transduction model . The presented model is a modifified version of the model by Simeoni et al.(Predictive pharmacokinetic-pharmacodynamic modeling of tumor growth kinetics in xenograft models after administration of anticancer agents. Cancer Res. 2004;64(3):1094–101.)and Alison  et al.(A Translational Quantitative Systems Pharmacology Model for CD3 Bispecific Molecules: Application to Quantify T Cell-Mediated Tumor Cell Killing by P-Cadherin LP DART®).

 

    

Why is TSC less than 0 in my model?How can I optimize my model?

.

 Any suggestion?

 

Thank you

Attached Files

•  NOG.phxproj   838.15KB   131 downloads

[bwendt@certara.com]

Hi,

 

TSC is less than 0 simply because kmax is smaller than lambda0.

You can get around this by estimating parameter k2.

I don't have the reference for this specific adaption of the Simeoni model therefore can't check the code in detail.

 

Bernd

[18963422692]

  See the attachment for the literature.

How to set parameter k2？

Attached Thumbnails

• 

Attached Files

•  附件2 A Translational Quantitative Systems Pharmacology Model for CD3 Bispecific2019辉瑞.pdf   2.09MB   861 downloads

[bwendt@certara.com]

I have looked at the model and the data and I got two observations:

 

1. there are perhaps too many parameters to estimate from the data. One thing that you could do is to use the unperturbed data to identify the growth parameters. Once estimated you can freeze those before you start estimating drug potency parameters

2. the potency of you drug candidate is not high. When you look at the curves you can see that the growth is shrinking, but the drug is not potent enough to bend the growth curve and actually shrink the tumor volume. During the study period the tumor volume is not reaching a plateau, so some parameters cannot be estimated

 

In summary, I would suggest you start with a simpler model and look at the unperturbed data first.

 

Bernd