transit absorption

[vjd]

hi @kylebmetrum -

What parts are missing to be able to simulate a transit curve? Not sure about the following

1. is lgamma in the correct representation in the depot compartment?
2. not sure if podo is required here, but it seems to be the dose that is scaled by bio at least in the paper

code <- '
// Table 3 from Savic 2007

$PARAM cl = 17.2, vc = 45.1, ka = 0.38, mtt = 0.37, bio=1, n = 20.1, podo=100
$SET delta=0.1

$CMT depot cen

$MAIN
  double k = cl/vc;
  double ktr = (n+1)/mtt;
  
  $ODE
  dxdt_depot = exp(log(bio*podo)+log(ktr)+n*log(ktr*SOLVERTIME)-ktr*SOLVERTIME-lgamma(n+1))-ka*depot;
  dxdt_cen = ka*depot-k*cen;

$TABLE
double CP = cen/vc;
'




mod <- mcode_cache("transit", code)
mod %>% ev(amt = 100) %>% 
  mrgsim %>% plot

[kylebaron]

Hi @vjd -

I think there is more that needs to go in $MAIN. I tweaked this a bit to look closer to figure 2.

Totally missed you at Rstudio::conf, my friend. Hope you and your family are well!

Kyle

code <- '
// Table 3 from Savic 2007

$PARAM 
cl = 17.2, vc = 45.1, ka = 1.71
mtt = 1, bio=1, n = 5

$SET delta=0.1

$CMT depot cen

$GLOBAL 
double podo = 0;

$MAIN
double k = cl/vc;
double ktr = (n+1)/mtt;
F_depot = 0;
F_cen = 1;

if(self.amt > 0 && self.cmt==1) podo = self.amt;
if(self.amt > 0 && self.cmt==2) podo = 0;

$ODE
double rate_in =  exp(log(bio*podo)+log(ktr)+n*log(ktr*SOLVERTIME)-ktr*SOLVERTIME-lgamma(n+1));
dxdt_depot = rate_in - ka*depot;
dxdt_cen = ka*depot-k*cen;

$TABLE
double CP = cen/vc;
$CAPTURE rate_in
'

mod <- mcode_cache("transit", code)
mod %>% 
  ev(amt = 30) %>% 
  mrgsim(end = 5, delta = 0.1) %>% plot

[vjd]

aha... now I understand this section from the user guide. There are certain sections that can be used and not be used in the main code and self.<param> is one way..

I missed the conference and you guys too. Last year was fun. Will be going to ASCPT. Lots to catch up on, but as of now, buckled in for family duty :)

[kylebaron]

Cool ... catch up with you at ASCPT then. Family duty is a different life, but it's a great one 💯

[safoah]

@kylebaron is the central compartment in this case all of the tissues lumped together or just the highly perfused tissues or is it a presystemic compartment? I am working on a PBPK model and would like to use this absorption model but I have multiple compartments rather than just the depot and central compartment.

[kylebaron]

Hi @safoah -
It's just a lumped compartment without really any physiological interpretation. If you need some sort of transit model for the delay, I'd look at this one instead:
https://github.com/mrgsolve/gallery/blob/master/absorption/transit.md

I guess you could adapt it for splitting doses into multiple compartments etc. But not sure about the model you're using.

Kyle

[2011CAMMARA]

Hi @kylebaron -

I have further questions as to the possibility of modifying your $MAIN logic to allow for multiple dosing events. My issue is that I am inputting post hoc parameter values as well as new dosing events but with the current parameterization I am only able to recognize the first dose event. My model and R code is provided below as well as a sample dataset to use.

reproduce_dosing.csv

code <- '
$PROB

$PARAM
KA = NA_real_,
CL = NA_real_,
VC = NA_real_,
BIO= NA_real_,
NN = NA_real_,
MTT= NA_real_,

$CMT Xa Xc XAUC

$GLOBAL
double podo ; //

$MAIN

// transit CMT variables and settings
F_Xa = 0;
F_Xc = 1;

if(self.amt > 0 && self.cmt==1) podo = self.amt;
if(self.amt > 0 && self.cmt==2) podo = 0;

// defining PK parameters
double KAi=KA;
double CLi=CL;
double VCi=VC;
double NNi=NN;
double MTTi=MTT;
double BIOi=BIO;

// defining transit chain parameters
double KTR=(NNi+1)/MTTi;
double L=log(2.5066)+(NNi+.5)log(NNi)-NN+log(1+1/(12NNi)) ;

// defining rate terms
double K20=CLi/VCi;

$ODE
double depot = Xa;
double CP = (Xc/VC);
double AUC = (XAUC);

dxdt_Xa = exp(log(BIOipodo)+log(KTR)+NNilog(KTRSOLVERTIME)-KTRSOLVERTIME-L)-KAiXa ; // Transit Compartment
dxdt_Xc = KAiXa - K20*Xc ; // Central Compartment
dxdt_XAUC = CP ; // AUC CMT

$CAPTURE CP '
mod <- mcode(code=code, model="model")

doseds <- daeta %>%
filter(EVID>0) %>%
select(ID,TIME,CMT,EVID,AMT,DOSEMG,KA,CL,VC,NN,MTT,BIO)

set.seed(4866519)
out<-mod %>%
data_set(doseds) %>%
carry_out(EVID) %>%
mrgsim_df(delta=0.1,start=0,end=216,atol=1e-10,rtol=1e-12)

[chengshen1992]

Hello @kylebaron ,
Hope all is well.
Working on this model recently, encountering the same issue when multiple execute multiple dosing...
A toy example modified based on yours:

code <- 
// Table 3 from Savic 2007

$PARAM 
cl = 17.2, vc = 45.1, ka = 1.71
mtt = 1, bio=1, n = 5

$SET delta=0.1

$CMT depot cen

$GLOBAL 
double podo = 0;

$MAIN
double k = cl/vc;
double ktr = (n+1)/mtt;
F_depot = 0;
F_cen = 1;

if(self.amt > 0 && self.cmt==1) podo = self.amt;
if(self.amt > 0 && self.cmt==2) podo = 0;

$ODE
double rate_in =  exp(log(bio*podo)+log(ktr)+n*log(ktr*SOLVERTIME)-ktr*SOLVERTIME-lgamma(n+1));
dxdt_depot = rate_in - ka*depot;
dxdt_cen = ka*depot-k*cen;

$TABLE
double CP = cen/vc;
$CAPTURE rate_in
'

mod <- mcode_cache("transit", code)

# Works well when execute a single dose
xx1 <- data.frame(ID=1, EVID=1, AMT=12.6         , CMT=1, TIME=0       )
mod %>% data_set(xx1) %>% 
  mrgsim(end = 24, delta = 0.1) %>% plot

# Only recognize one dose when execute multiple dosing
xx2 <- data.frame(ID=1, EVID=1, AMT=c(12.6, 12.4), CMT=1, TIME=c(0,4.3))
mod %>% data_set(xx2) %>% 
  mrgsim(end = 24, delta = 0.1) %>% plot

Appreciate the insights!
Shen

[kylebaron]

Hi @chengshen1992 -

It's been a while, but I think there might be a way to parameterize this in terms of time since last dose, under the assumption that 100% of all previous doses were absorbed at the time of any dose. I want to say we have that somewhere; I can look.

But I would highly recommend switching to this model instead. It looks more complicated, but it always gives the fewest problems when trying to work with transit models and no assumption of complete absorption of previous doses.

Kyle

[chengshen1992]

Thanks a lot Kyle! The model you suggested solves my issues.