# PharmPK Discussion - Input and interpretation deconvolution results

PharmPK Discussion List Archive Index page
• On 15 Jan 2010 at 16:08:45, Koen Jolling (koen.jolling.aaa.sepspharma.com) sent the message
`The following message was posted to: PharmPKDear,I have some questions related to deconvolution and more specific with respect to the input parameters and the interpretation of the results.I'm using the WinNonlin Phoenix deconvolution toolbox (not the additional IVIVC toolkit)Let me give a concise overview:1) We have in vivo PK data of an immediate release (IR) formulation that will be used as reference. A suitable PK model (1-comp) is developed for these data and the following relevant parameters were derived: K10 = 0.7 and V/F=4000 ml/kg. The dose administered was 10 mg/kg. Deconvolution asks for two parameters alpha1 and A1. For alpha1 K10=0.7 is used. Based on the PK output A1 is calculated as follows: V/F = Dose/A1. This results in 0.0025 mg/mL. Subsequently, this number is dose normalized as is requested for deconvolution becoming A1=0.00025 mg/mL or 250 ng/mL. So as input parameters I used alpha1=0.7 and A1=250 ng/mL. Is that the correct way of doing things.2) In order to calculate the in vivo dissolution/absorption profile deconvolution was performed on in vivo PK data from 4 controlled released (CR) formulations giving as a result the following parameters: input rate (mg/kg/hr), cumulative input (mg/kg) and fraction_input. The question here is what I have to do with these results. My guess here is that multiplying the fraction_input by 100 gives the %absorbed over time. Correct? In my particular case the maximum %absorbed equals approximately 45% which is equal to the absolute bioavailability of the compound. Is this a coincidence? I would think not. But how is F predicted when no iv data were available?3) Based on the results, how could I compare the in vitro dissolution profiles with the deconvoluted absorption profiles? I have both parameters as a function of time, but how to plot them with the in vitro dissolution % on the x-axis and the in vivo %absorbed on the Y-axis? How does a time shifted axis work? Should I fit the in vitro dissolution data with an appropriate equation and use the predicted in vitro dissolution % to corresponding time points of the in vivo %absorbed profiles?4) Since the %absorbed profiles are limited to 45% there is no 1:1 in vivo:invitro relationship. What does this tell me?Thank you,KoenSenior ScientistSEPS Pharma`
Back to the Top

• On 15 Jan 2010 at 13:14:41, Leonid Gibiansky (LGibiansky.-a-.quantpharm.com) sent the message
`The following message was posted to: PharmPKKoen,Several points to consider:1. You do not have IV data, only IR (solution?) but attempt to use a mono-exponential function for the deconvolution. This is correct only if absorption from the IR  is very fast relative to the other processes. If this is a valid assumption, then you can proceed as you described. Otherwise, you need to account for the absorption phase.2. Assuming that mone-exponential assumption is valid, computation of A1 and alpha1 seems to be correct.3. 100*fraction_input can indeed be interpreted as %absorbed over time. Maximum % absorbed should be interpreted as bioavailability relative to the IR, not absolute bioavaialbility. It is equal to absolute bioavailability only if absolute bioavailability of IR is close to 100%.4. To investigate IVIVC, you may relate %absorbed and %dissolved. If they are available at the same time points, you may try regression. If they are available in different time points, you need to proceed as you described: fit the in vitro dissolution data with an appropriate equation and use the predicted in vitro dissolution % to corresponding time points of the in vivo %absorbed profiles. If you would like to introduce a time scaling, scaling parameter become a parameter of the IVIVC model and needs to be estimated.5. Even if the %absorbed profiles are limited to 45%, there could be a 1:1 in-vivo:in-vitro relationship. For example, it may be valid if the tablets leave the body at the point when only 45% of the drug is released.To tell more, it is necessary to perform a more detailed IVIVC analysis.ThanksLeonid--Leonid Gibiansky, Ph.D.President, QuantPharm LLCweb:    www.quantpharm.com`
Back to the Top

• On 15 Jan 2010 at 20:20:18, Koen Jolling (koen.jolling.aaa.sepspharma.com) sent the message
`The following message was posted to: PharmPKDear Leonid,Thank you. As a matter of fact the IR is a commercially available tablet and indeed it turns out that a 2-comp model fits the data much better. A bi-exponential function for deconvolution would be preferred. Is this what you mean by accounting for the absorption phase? Does it matter if a lag time is involved? How would calculation of A, B, alpha and beta go? They are derived in the same way as a 1-comp model but now from V1 and V2 respectively?The absolute bioavailability is known from literature and is based on the comparison of iv administration versus IR solution? We only have IR tablet PK profiles? This is not the ideal scenario, I know. Does this hamper my deconvolution exercise?Thank you,Koen Jolling, Ph.D.SEPS Pharma`
Back to the Top

• On 15 Jan 2010 at 14:56:46, Leonid Gibiansky (LGibiansky.aaa.quantpharm.com) sent the message
`The following message was posted to: PharmPKKoen,By bi-exponential, I meant the equation for the one-compartment model with the first-order absorption:C= -Aa*exp(-ka*t) + A1*exp(-alpha1*t)where Aa part is responsible for quick (hopefully) absorption from the IR.If your drug is better described by the two-compartment model with the first-order absorption, then you should useC=-Aa*exp(-ka*t) + A1*exp(-alpha1*t)+ A2*exp(-alpha2*t)If absorption is fast then you can useC=A1*exp(-alpha1*t)+ A2*exp(-alpha2*t)  (**)(disregarding the part of the profile where concentration increases).I am not sure what software do you use to estimate the model, but if possible, you can use directly equation (**) and estimate parameters A1, A2, alpha1, alpha2 rather than estimate CL/F, V/F, KA, etc.Lag time is important if it is evident in your PK model. If so, you should useC=A1*exp(-alpha1*(t-tlag))+ A2*exp(-alpha2*(t-tlag))   (***)where tlag is also estimated. I do not know whether deconvolution program that you have allows the lag time in the impulse response function. If not, you may want to do some tricks, like shifting observed concentrations (for the ER formulations) by tlag estimated from (***) before performing the deconvolution usingC=A1*exp(-alpha1*t)+ A2*exp(-alpha2*t)(without lag time in the impulse response function).You do not need absolute bioavaialbility to do IVIVC unless you want to use PK parameters from the literature rather than from your IR data. You can do everything relative to the IR formulation.ThanksLeonid--Leonid Gibiansky, Ph.D.President, QuantPharm LLCweb:    www.quantpharm.com`
Back to the Top

• On 15 Jan 2010 at 22:11:32, "Charlie Brindley " (charlie.brindley.at.kinetassist.com) sent the message
`The following message was posted to: PharmPKDear Koen,In addition to Leonid's response, the deconvolution of the MR and IR formswill simulate the in vivo dissolution profile assuming that, inter alia, theabsorption of test compound from the MR form is equivalent to that of the IRform after dissolution.However, a relative bioavailability of only 45% suggests that thisassumption may not be valid (you can simply check the value of relativebioavailability using an area method). Indeed, a MR form of such lowbioavailability (approx. 20% given absolute bioavailability) may not be avery useful formulation. Furthermore, the low bioavailability may indicatedose dumping in the GI tract.Regards,Charlie`
Back to the Top

 Want to post a follow-up message on this topic? If this link does not work with your browser send a follow-up message to PharmPK@boomer.org with "Input and interpretation deconvolution results" as the subject Support PharmPK by using thelink to buy books etc.from Amazon.com

Copyright 1995-2011 David W. A. Bourne (david@boomer.org)