Inconsistency result

[jeslo328]

Hi

When i plot the simulated data post stat calculation, i.e :

dataPsim_q2wkAa <- res.101a

Data3q2kAa =  dataPsim_q2wkAa  %>%
              group_by(TIME,NDOSESIM) %>%
              summarise(
              #GM  = geo.mean(Value),
              #GCV = geoCV(Value),
              AM  = mean(OIPRED),
              SD  = sd(OIPRED),
              n   = n() ,
              C.P05 = quantile(OIPRED, 0.05,na.rm=T),
              C.P10 = quantile(OIPRED, 0.10,na.rm=T),
              C.P25 = quantile(OIPRED, 0.25,na.rm=T),
              C.P90 = quantile(OIPRED, 0.90,na.rm=T),
              C.P95 = quantile(OIPRED, 0.95,na.rm=T))%>% 
              ungroup() %>%
              arrange(TIME) %>%
              mutate(Schedule ='q2 weeks') %>%
              mutate(xx=TIME/(24*7))

 TIME DOSE           AM                   SD     n       C.P05 C.P10 C.P25 C.P90 C.P95 Schedule    xx
  <dbl> <fct>    <dbl>      <dbl> <int> <dbl> <dbl> <dbl> <dbl> <dbl> <chr>    <dbl>
1  8736      2         22.1                 8.60  1200  10.6  12.5  16.0  33.7  37.5 q2 weeks    52
2  8736     2.5       27.9               11.6   1200  13.0  15.5  19.7  43.3  50.9 q2 weeks    52
3  8736     3         33.4                13.6   1200  15.8  18.5  23.4  51.0  58.1 q2 weeks    52
4  8736     3.2       34.7               14.3   1200  15.9  18.7  24.6  54.0  61.7 q2 weeks    52
5  8736     3.5       38.4               14.7   1200  18.9  21.8  27.6  58.7  65.0 q2 weeks    52
6  8736    4         43.7               18.4   1200  21.0  24.0  30.8  66.4  76.8 q2 weeks    52

For one of the dose i could see post plotting that the mean value is above the pk target, indicating atleast 50% is above te target

However when i did NCA on these at simulated steady state, but i am not getting indicating less than 50%.

dataPsim_q2wk  <- res.101a

dataPsim_q2wkD <- res.101a %>%
                  mutate(OIPRED=((IPRED*90.2) /1000))
                  mutate(TIME = case_when(
                         EVID ==1 ~ (TIME+0.001),
                         EVID !=1 ~ (TIME+0)))  %>%
                  filter(TIME>8399) 


dataP_conce_q2wk <- dataPsim_q2wkD %>%
                    filter(CMT==2)%>%
                    mutate(OTIME=(TIME-8400))

View(dataP_conce_q2wk)
                     
dataP_dose_q2wk <-  dataPsim_q2wkD %>%
                    filter(CMT==1) %>%
                    mutate(OTIME=(TIME-8400.001),
                           AMT2   = (NDOSESIM*WT),
                           AMT_MG = (AMT2*90.2)/1000,
                           AMT_UG = (AMT_MG*90.2)*1000)

View(dataP_dose_q2wk)
o_conc <- PKNCAconc(dataP_conce_q2wk,OIPRED~OTIME|NDOSESIM + ID)
o_dose <- PKNCAdose(dataP_dose_q2wk,AMT_UG~OTIME|NDOSESIM + ID)

data_obj_742_q2wk <- PKNCAdata(o_conc, o_dose,
                          intervals=data.frame(start=0, # Change 1 
                                               end=336,
                                               cmax=T,
                                               cmin=T,
                                               cav=T,
                                               tmax=T,
                                               half.life=F,
                                               aucall=TRUE,
                                               auclast=TRUE,
                                               aucinf.pred=T,
                                               aucinf.obs=T))
results_obj_742_q2wk <- pk.nca(data_obj_742_q2wk)

knitr::kable(as.data.frame(results_obj_742_q2wk))
## a. nca summary result----
summary(results_obj_742_q2wk) 

# sETITNG FOR AUTOMATION

dataxxq2wk_Ind_NCA <- dataxxq2wk %>%
                      filter(PPTESTCD == 'cmin')%>%
                      #mutate(ug_ml = ((PPORRES*90.2)/1000) ) %>%
                      mutate(Frac_over12  = ifelse(PPORRES > 132.5,1,0)) 

prop.table(table(dataxxq2wk_Ind_NCA[c("NDOSESIM","Frac_over12")]),margin = 1) 



#--------------------------------#

Can you advise what is actually going on here ?

[billdenney]

I'm not certain that I understand the question fully, but I have a guess about the answer.

It looks like you're comparing probabilities at individual time points to cmin. If your minimum concentration happens at different time points, then cmin will be less than the individual time point results.

This can happen in almost any steady-state result when you have sampling at the beginning and end of the interval because cmin will be at the beginning half of the time and at the end half of the time. If you have a flat profile (comparing peak to trough ratio relative to your residual error), then it will be even greater.

Does this answer your question?