I’m having a problem getting predose concentrations from a steady-state profile to be estimated correctly in NLME. I currently have the dose to be profiled at time = 0 and the inital concentration also set at time = 0. Can NLME handle this as a boundry condition? I’m getting useful estimates when I exclude the predose concentrations, but I would like to use all the available data. Kind regards, Jim
Jim, what is your dose route? If it is IV and you’re setting the CONC at time zero to a value of 0 then I can understand why this might cause some fitting problems. If you have a BLQ value at predose (or elsewhere) you can set a flag for this data and then the Drug Model Engine uses the probability that the censored observation is below the quantification limit in the likelihood function. Perhaps if you post your project, (or a reduced set) we can help you further. Simon
Hi Simon, Thanks for the quick reply. Concentrations being set at Time = 0 are all measureable concentraitons. The aim of therapy is to keep the trough samples above 5g/mL. I would need Sponsor’s approval to post any data but I can easily describe the IgG FDA Guidance which is similar to the EMA IgG Guidance. I’m working with a sponsor’s data for IgG infusions which are not standard between patients. IgG has a half-life of about 28 days and I only have one period to work with Sample include predose, end of infusion, 0.5, 6.0, 12, 24hr after EOF then 3, 7, 14, 21, and 28 days after EOF. Patients can be on either a tau of 28 days or 21 days but few are on the 21 day tau. There are no BLQ values as all samples have measurable concentrations. This data represents the 8th infusion for the study, but I also have predose concentration data for the first 7 infusions but these first 7 predose samples were assayed by clinical labs and not PK labs. I’m concentrating on the 8th infusion profile since that is what is needed from regulatory guidance and from protocol and analysis plans. However since patients changed doses and changed treatement tau, I thought it prudent to get the predose samples on the first 7 infusions to see if it might be needed to clean up my estimates. These patients have primary immunideficiencies and enter the study at steaty-state only on another product. Any help you can provide would be most helpful, else I’ll just have to omit the predose concentations. Kind regards, Jim
Hi Jim, I think I’ve seen an exercise that incorporated a problem similar to this in one of our training sessions but this was for PD data, however similar principles should apply. Could you perhaps blind a few individuals and post the reduced project so we can see if one of us can identify how to improve your fit? Simon.