PK/PD SAP Implementation with SDTM/ADaM and gt Tables
David Guy PhD - Biostatistics / Programming
2025-12-08
knitr::opts_chunk$set(
echo = TRUE,
message = FALSE,
warning = FALSE,
fig.width = 10,
fig.height = 6,
collapse = TRUE,
comment = "#>"
)
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(tidyr)
library(gt)
library(zoo)##
## Attaching package: 'zoo'## The following objects are masked from 'package:base':
##
## as.Date, as.Date.numericlibrary(openxlsx)
library(tibble)
library(ggplot2)1. Introduction
Complete PK/PD SAP workflow following CDISC standards for regulatory submissions. High-level workflow summary Use R to implement the SAP logic for PK and PD by creating subject-level ADaM datasets (e.g., ADPPK, ADPPD) that compute PK parameters and PD responses as specified in the SAP.
Use {gt} (and optionally {gtreg}) to create formatted clinical tables that match regulatory expectations and export them to RTF/HTML/PDF for inclusion in ND/BLA/NDA/MAA submissions.
Maintain SDTM mapping specs and ADaM specs as structured metadata (typically Excel), then use {defineR} to convert these specs into SDTM/ADaM metadata spreadsheets and define.xml files with schema validation for submission packages.
If you share your current data structure (variable names and domains), more tailored code templates for your specific SAP (e.g., noncompartmental PK, exposure–response PD models, or population PK) can be provided.
PK/PD stands for pharmacokinetics and pharmacodynamics, two closely linked areas of pharmacology that describe how drugs behave in the body and what effects they produce.
Pharmacokinetics (PK) Pharmacokinetics is the study of what the body does to a drug over time. It focuses on the ADME processes: absorption (how the drug enters the bloodstream), distribution (how it spreads through tissues), metabolism (how it is chemically changed), and excretion (how it is removed).
Pharmacodynamics (PD) Pharmacodynamics is the study of what the drug does to the body. It describes the relationship between drug concentration at the target (such as a receptor) and the resulting biological or clinical effect, including both desired therapeutic effects and side effects.
PK/PD (or PKPD) analysis links drug concentrations over time (PK) to their observed effects (PD) to understand exposure–response relationships. These combined models are widely used in drug development to select doses, design regimens, and predict onset, magnitude, and duration of drug effects in different populations.
2. Input Data
path2File <- "/home/rmlinux/DataBridge Statistical Consulting/CLAUDE/dBase/"
pk_raw <- read.csv(paste0(path2File,"pk_raw.csv"), header=TRUE, stringsAsFactors = FALSE)
glimpse(pk_raw)
#> Rows: 20
#> Columns: 6
#> $ STUDYID <chr> "STUDY1", "STUDY1", "STUDY1", "STUDY1", "STUDY1", "STUDY1", "S…
#> $ USUBJID <chr> "SUBJ001", "SUBJ001", "SUBJ001", "SUBJ001", "SUBJ001", "SUBJ00…
#> $ ARM <chr> "Drug A", "Drug A", "Drug A", "Drug A", "Drug A", "Drug A", "D…
#> $ VISIT <chr> "Visit 1", "Visit 1", "Visit 1", "Visit 1", "Visit 1", "Visit …
#> $ TIME <int> 0, 1, 2, 4, 8, 0, 1, 2, 4, 8, 0, 1, 2, 4, 8, 0, 1, 2, 4, 8
#> $ CONC <dbl> 0.0, 2.5, 4.2, 3.8, 1.9, 0.0, 3.1, 5.0, 4.1, 2.0, 0.0, 1.8, 3.…
pd_raw <- read.csv(paste0(path2File,"pd_raw.csv"), header=TRUE, stringsAsFactors = FALSE)
glimpse(pd_raw)
#> Rows: 16
#> Columns: 7
#> $ STUDYID <chr> "STUDY1", "STUDY1", "STUDY1", "STUDY1", "STUDY1", "STUDY1", "…
#> $ USUBJID <chr> "SUBJ001", "SUBJ001", "SUBJ001", "SUBJ001", "SUBJ002", "SUBJ0…
#> $ ARM <chr> "Drug A", "Drug A", "Drug A", "Drug A", "Drug A", "Drug A", "…
#> $ VISIT <chr> "Visit 1", "Visit 1", "Visit 1", "Visit 1", "Visit 1", "Visit…
#> $ TIME <int> 0, 2, 4, 8, 0, 2, 4, 8, 0, 2, 4, 8, 0, 2, 4, 8
#> $ EFFICACY <int> 10, 15, 18, 16, 12, 17, 20, 19, 11, 13, 14, 13, 9, 12, 13, 12
#> $ BIOMARK <dbl> 1.2, 1.5, 1.7, 1.6, 1.3, 1.6, 1.9, 1.8, 1.1, 1.3, 1.4, 1.3, 1…
pk_raw |>
ggplot(aes(x = TIME, y = CONC, color = USUBJID, linetype = ARM)) +
geom_point(size = 2) + geom_line() +
facet_wrap(~ARM) +
labs(title = "PK Concentration-Time Profiles", y = "Concentration (ng/mL)") +
theme_minimal()
# 3. PK Parameters (SAP)
pk_params <- pk_raw |>
group_by(STUDYID, USUBJID, ARM) |>
arrange(TIME, .by_group = TRUE) |>
summarize(
CMAX = max(CONC, na.rm = TRUE),
TMAX = TIME[which.max(CONC)],
AUC = {
if(length(CONC) > 1) {
dt <- diff(TIME)
cavg <- zoo::rollmean(CONC, 2)
sum(dt * cavg, na.rm = TRUE)
} else 0
},
.groups = "drop"
) |>
mutate(AUC = round(AUC, 2))
pk_params
#> # A tibble: 4 × 6
#> STUDYID USUBJID ARM CMAX TMAX AUC
#> <chr> <chr> <chr> <dbl> <int> <dbl>
#> 1 STUDY1 SUBJ001 Drug A 4.2 2 24
#> 2 STUDY1 SUBJ002 Drug A 5 2 26.9
#> 3 STUDY1 SUBJ003 Drug B 3 2 16.8
#> 4 STUDY1 SUBJ004 Drug B 3.5 2 19.24. PD Summaries
pd_summ <- pd_raw |>
group_by(STUDYID, USUBJID, ARM, VISIT) |>
summarize(
MEAN_EFF = mean(EFFICACY, na.rm = TRUE),
MEAN_BIOM = mean(BIOMARK, na.rm = TRUE),
.groups = "drop"
) |>
mutate(across(c(MEAN_EFF, MEAN_BIOM), ~round(.x, 2)))
pd_summ
#> # A tibble: 4 × 6
#> STUDYID USUBJID ARM VISIT MEAN_EFF MEAN_BIOM
#> <chr> <chr> <chr> <chr> <dbl> <dbl>
#> 1 STUDY1 SUBJ001 Drug A Visit 1 14.8 1.5
#> 2 STUDY1 SUBJ002 Drug A Visit 1 17 1.65
#> 3 STUDY1 SUBJ003 Drug B Visit 1 12.8 1.27
#> 4 STUDY1 SUBJ004 Drug B Visit 1 11.5 1.185. ADaM ADPPK Dataset
adppk <- pk_params %>%
pivot_longer(
cols = c(CMAX, TMAX, AUC),
names_to = "PARAMCD",
values_to = "AVAL"
) %>%
mutate(
PARAM = case_when(
PARAMCD == "CMAX" ~ "Maximum Observed Concentration",
PARAMCD == "TMAX" ~ "Time of Cmax",
PARAMCD == "AUC" ~ "AUC(0-tlast)",
TRUE ~ NA_character_
),
AVAL = round(AVAL, 3)
) %>%
select(STUDYID, USUBJID, ARM, PARAMCD, PARAM, AVAL)
adppk
#> # A tibble: 12 × 6
#> STUDYID USUBJID ARM PARAMCD PARAM AVAL
#> <chr> <chr> <chr> <chr> <chr> <dbl>
#> 1 STUDY1 SUBJ001 Drug A CMAX Maximum Observed Concentration 4.2
#> 2 STUDY1 SUBJ001 Drug A TMAX Time of Cmax 2
#> 3 STUDY1 SUBJ001 Drug A AUC AUC(0-tlast) 24
#> 4 STUDY1 SUBJ002 Drug A CMAX Maximum Observed Concentration 5
#> 5 STUDY1 SUBJ002 Drug A TMAX Time of Cmax 2
#> 6 STUDY1 SUBJ002 Drug A AUC AUC(0-tlast) 26.9
#> 7 STUDY1 SUBJ003 Drug B CMAX Maximum Observed Concentration 3
#> 8 STUDY1 SUBJ003 Drug B TMAX Time of Cmax 2
#> 9 STUDY1 SUBJ003 Drug B AUC AUC(0-tlast) 16.8
#> 10 STUDY1 SUBJ004 Drug B CMAX Maximum Observed Concentration 3.5
#> 11 STUDY1 SUBJ004 Drug B TMAX Time of Cmax 2
#> 12 STUDY1 SUBJ004 Drug B AUC AUC(0-tlast) 19.26. Regulatory PK Table
tbl_pk <- adppk |>
group_by(PARAM, ARM) |>
summarize(
N = n(),
MEAN = mean(AVAL, na.rm = TRUE),
SD = sd(AVAL, na.rm = TRUE),
CV = (SD/MEAN)*100,
MIN = min(AVAL, na.rm = TRUE),
MEDIAN = median(AVAL, na.rm = TRUE),
MAX = max(AVAL, na.rm = TRUE),
.groups = "drop"
) |>
mutate(across(c(MEAN, SD, MIN, MEDIAN, MAX, CV), ~round(.x, 2)))
pk_table_final <- tbl_pk |>
gt(rowname_col = "PARAM") |>
tab_header(
title = "Table 14.1.1.1 PK Parameters by Treatment",
subtitle = "PK Analysis Set (N=4)"
) |>
cols_label(
ARM = "Treatment", N = "n", MEAN = "Mean", SD = "SD",
CV = "CV%", MIN = "Min", MEDIAN = "Median", MAX = "Max"
) |>
fmt_number(c(MEAN, SD, MIN, MEDIAN, MAX), decimals = 2) |>
fmt_percent(CV, decimals = 1) |>
tab_source_note("Linear trapezoidal AUC(0-tlast) per SAP.")
pk_table_final| Table 14.1.1.1 PK Parameters by Treatment | ||||||||
| PK Analysis Set (N=4) | ||||||||
| Treatment | n | Mean | SD | CV% | Min | Median | Max | |
|---|---|---|---|---|---|---|---|---|
| AUC(0-tlast) | Drug A | 2 | 25.45 | 2.05 | 806.0% | 24.00 | 25.45 | 26.90 |
| AUC(0-tlast) | Drug B | 2 | 18.02 | 1.73 | 961.0% | 16.80 | 18.02 | 19.25 |
| Maximum Observed Concentration | Drug A | 2 | 4.60 | 0.57 | 1,230.0% | 4.20 | 4.60 | 5.00 |
| Maximum Observed Concentration | Drug B | 2 | 3.25 | 0.35 | 1,088.0% | 3.00 | 3.25 | 3.50 |
| Time of Cmax | Drug A | 2 | 2.00 | 0.00 | 0.0% | 2.00 | 2.00 | 2.00 |
| Time of Cmax | Drug B | 2 | 2.00 | 0.00 | 0.0% | 2.00 | 2.00 | 2.00 |
| Linear trapezoidal AUC(0-tlast) per SAP. | ||||||||
7. SDTM PC Mapping Spec
sdtm_pc_spec <- tribble(
~SDTM_Var, ~Source_Var, ~Derivation,
"STUDYID", "STUDYID", "Copy as-is",
"DOMAIN", NA, "'PC'",
"USUBJID", "USUBJID", "Copy as-is",
"PCSEQ", NA, "Row number by USUBJID",
"PCTESTCD", NA, "'DRUGCONC'",
"PCORRES", "CONC", "Reported concentration",
"PCSTRESN", "CONC", "Numeric concentration",
"PCSTRESU", NA, "'NG/ML'"
)
sdtm_pc_spec |> gt() |> tab_header("SDTM PC Mapping")| SDTM PC Mapping | ||
| SDTM_Var | Source_Var | Derivation |
|---|---|---|
| STUDYID | STUDYID | Copy as-is |
| DOMAIN | NA | 'PC' |
| USUBJID | USUBJID | Copy as-is |
| PCSEQ | NA | Row number by USUBJID |
| PCTESTCD | NA | 'DRUGCONC' |
| PCORRES | CONC | Reported concentration |
| PCSTRESN | CONC | Numeric concentration |
| PCSTRESU | NA | 'NG/ML' |
8. ADaM Metadata
adam_meta <- tribble(
~Dataset, ~Variable, ~Label, ~Type, ~Origin,
"ADPPK", "STUDYID", "Study ID", "Char", "SDTM.PC",
"ADPPK", "USUBJID", "Subject ID", "Char", "SDTM.PC",
"ADPPK", "ARM", "Treatment Arm", "Char", "SDTM.SV",
"ADPPK", "PARAMCD", "Param Code", "Char", "Derived",
"ADPPK", "AVAL", "Analysis Value", "Num", "Derived"
)
adam_meta
#> # A tibble: 5 × 5
#> Dataset Variable Label Type Origin
#> <chr> <chr> <chr> <chr> <chr>
#> 1 ADPPK STUDYID Study ID Char SDTM.PC
#> 2 ADPPK USUBJID Subject ID Char SDTM.PC
#> 3 ADPPK ARM Treatment Arm Char SDTM.SV
#> 4 ADPPK PARAMCD Param Code Char Derived
#> 5 ADPPK AVAL Analysis Value Num Derived9. Export Files
gtsave(pk_table_final, "pk_table.rtf")
write.xlsx(list(PC=sdtm_pc_spec, ADPPK=adam_meta), paste0(path2File,"metadata.xlsx"))