Creating a Metabolic ADLB ADaM • admiralmetabolic

Introduction

This article describes creating a laboratory ADaM for metabolic clinical trials.

We advise you first consult the admiral Creating a BDS Finding ADaM vignette. The programming workflow around creating the general set-up of an ADLB using admiral functions is the same. In this vignette, we focus on common ADLB derivations in metabolic studies and avoid repeating information and maintaining the same content in two places.

Note: All examples assume CDISC SDTM and/or ADaM format as input unless otherwise specified.

Required Packages

The examples of this vignette require the following packages.

library(admiral)
library(admiralmetabolic)
library(pharmaversesdtm)
library(dplyr)
library(stringr)

Read in Data

To start, all data frames needed for the creation of the ADaM dataset should be loaded into the global environment. Reading data will usually be a company specific process, however, for the purpose of this vignette, we will use example data from pharmaversesdtm and admiralmetabolic. We will utilize LB and ADSL data.

lb_metabolic <- pharmaversesdtm::lb_metabolic
admiralmetabolic_adsl <- admiralmetabolic::admiralmetabolic_adsl

lb <- convert_blanks_to_na(lb_metabolic)
adsl <- convert_blanks_to_na(admiralmetabolic_adsl)

Define Lookup Tables

Define parameter lookup table used to derive PARAMCD, PARAM, and PARAMN variables.

# Assign PARAMCD, PARAM, and PARAMN
param_lookup <- tibble::tribble(
  ~LBTESTCD, ~PARAMCD, ~PARAM, ~PARAMN,
  "ALB", "ALB", "Albumin (g/L)", 1,
  "ALP", "ALKPH", "Alkaline Phosphatase (U/L)", 2,
  "AST", "AST", "Aspartate Aminotransferase (U/L)", 3,
  "CHOL", "CHOLES", "Cholesterol (mmol/L)", 4,
  "GGT", "GGT", "Gamma Glutamyl Transferase (U/L)", 5,
  "GLUC", "GLUC", "Glucose (mmol/L)", 6,
  "HBA1CHGB", "HBA1CHGB", "Hemoglobin A1C/Hemoglobin (mmol/mol)", 7,
  "INSULIN", "INSULIN", "Insulin (mIU/L)", 8,
  "TRIG", "TRIG", "Triglycerides (mg/dL)", 9
)

Derive Core ADLB Variables

The basic parameters and timing variables can be derived similarly to other BDS finding ADaMs. For the derivation of Glucose and Insulin, only fasted results (where LBFAST = "Y") are considered. For all other laboratory tests, the fasting status does not affect their inclusion in the dataset at this stage.

# Define required ADSL variables for derivations
adsl_vars <- exprs(TRTSDT, TRTEDT, TRT01A, TRT01P)

adlb <- lb %>%
  # Remove non-fasted GLUC and INSULIN results
  filter(!(LBTESTCD %in% c("GLUC", "INSULIN") & LBFAST != "Y")) %>%
  # Join ADSL with LB (need TRTSDT for ADY derivation)
  derive_vars_merged(
    dataset_add = adsl,
    new_vars = adsl_vars,
    by_vars = get_admiral_option("subject_keys")
  ) %>%
  # Calculate ADT, ADY
  derive_vars_dt(
    new_vars_prefix = "A",
    dtc = LBDTC
  ) %>%
  derive_vars_dy(reference_date = TRTSDT, source_vars = exprs(ADT))

adlb <- adlb %>%
  # Add PARAMCD PARAM and PARAMN - from parameter lookup table
  derive_vars_merged_lookup(
    dataset_add = param_lookup,
    new_vars = exprs(PARAMCD, PARAM, PARAMN),
    by_vars = exprs(LBTESTCD)
  ) %>%
  ## Calculate PARCAT1 PARCAT2 AVAL AVALC ANRLO ANRHI
  slice_derivation(
    derivation = mutate,
    args = params(
      PARCAT1 = LBCAT,
    ),
    # Handle specific parameters requiring conventional units (CV)
    derivation_slice(
      filter = LBTESTCD %in% c("TRIG", "INSULIN"),
      args = params(
        PARCAT2 = "CV",
        AVAL = as.numeric(LBORRES),
        AVALC = NA_character_,
        ANRLO = as.numeric(LBORNRLO),
        ANRHI = as.numeric(LBORNRHI)
      )
    ),
    # Handle other parameters using standard units (SI)
    derivation_slice(
      filter = TRUE,
      args = params(
        PARCAT2 = "SI",
        AVAL = LBSTRESN,
        # Only populate AVALC if character value is non-redundant with AVAL
        AVALC = if_else(
          is.na(AVAL) | as.character(AVAL) != LBSTRESC,
          LBSTRESC,
          NA_character_
        ),
        ANRLO = LBSTNRLO,
        ANRHI = LBSTNRHI
      )
    )
  )
#> All `LBTESTCD` are mapped.

Derive Visit Information

Derive Analysis Visit (AVISIT) and Analysis Visit Number (AVISITN).

adlb <- adlb %>%
  mutate(
    AVISIT = case_when(
      str_detect(VISIT, "SCREEN") ~ "Baseline",
      !is.na(VISIT) ~ str_to_title(VISIT),
      TRUE ~ NA_character_
    ),
    AVISITN = case_when(
      AVISIT == "Baseline" ~ 0,
      str_detect(VISIT, "WEEK") ~ as.integer(str_extract(VISIT, "\\d+")),
      TRUE ~ NA_integer_
    )
  )

For deriving visits based on time-windows, see admiral Visit and Period Variables.

Derive Metabolic Parameters

In addition to the core ADLB variables, several metabolic parameters and scores will be derived based on laboratory data and vital signs. These derivations include:

Fatty Liver Index (FLI): Derived from triglycerides (mg/dL), GGT (U/L), BMI (kg/m2), and waist circumference (cm). \[ FLI = \left[\frac{\exp(\lambda)}{1 + \exp(\lambda)}\right] \times 100 \] where \[ \lambda = 0.953 \times \ln (\text{triglycerides}) + 0.139 \times BMI + 0.718 \times \ln (GGT) + 0.053 \times \text{waist circumference} – 15.745. \] FLI is interpreted as a score between 0 and 100, where higher values indicate a worse condition.
Homeostasis Model Assessment – Insulin Resistance (HOMA-IR): Calculated from fasting plasma glucose (mmol/L) and fasting plasma insulin (mIU/L). \[ \text{HOMA-IR} = \frac{FPI \times FPG}{22.5}. \] An alternative formula is used when fasting plasma glucose is assessed in mg/dL: \[ \text{HOMA-IR} = \frac{FPI \times FPG}{405}. \]

To derive these parameters and scores, the following variables will be needed:

From ADLB: TRIG (Triglycerides), GGT, GLUC (fasted), INSULIN (fasted).
From ADVS: BMI, WSTCIR (Waist Circumference).

Add variables required for derivation

In this section, we will merge relevant variables needed to derive metabolic parameter from the ADVS (Analysis Dataset for Vital Signs) dataset into the ADLB dataset. This is necessary to include variables such as BMI and Waist Circumference, which are required for the derivation of FLI score parameter in subsequent section.

# Load ADVS dataset (assuming it has been created by ad_advs.R)
admiralmetabolic_advs <- admiralmetabolic::admiralmetabolic_advs
advs <- convert_blanks_to_na(admiralmetabolic_advs)

# Merge BMI and WSTCIR from ADVS to ADLB based on subject and date
adlb <- adlb %>%
  derive_vars_transposed(
    advs,
    by_vars = exprs(!!!get_admiral_option("subject_keys"), ADT),
    key_var = PARAMCD,
    value_var = AVAL,
    filter = PARAMCD %in% c("BMI", "WSTCIR")
  )

Please note that in this example we merge on ADT, but it is also possible to merge on AVISIT or other windowing algorithms specified in the statistical analysis plan.

Derive HOMA-IR index

This part describes how to calculate the Homeostasis Model Assessment – Insulin Resistance (HOMA-IR) index from fasting plasma glucose and fasting plasma insulin. This derivation does not require merging data from other datasets.

# Derive HOMA-IR using derive_param_computed
adlb <- adlb %>%
  derive_param_computed(
    by_vars = exprs(!!!get_admiral_option("subject_keys"), AVISIT, AVISITN, ADT, ADY, !!!adsl_vars),
    parameters = c("INSULIN", "GLUC"),
    set_values_to = exprs(
      AVAL = AVAL.INSULIN * AVAL.GLUC / 22.5,
      PARAMCD = "HOMAIR",
      PARAM = "Homeostasis Model Assessment - Insulin Resistance",
      PARAMN = 10
    )
  )

Derive FLI score

To obtain the FLI score parameter, we’ll utilize triglycerides, GGT, BMI, and waist circumference. Note that this derivation requires merging data from the ADVS dataset (BMI and Waist Circumference), as demonstrated in the previous section. FLI is provided here as an example; users can derive other metabolic scores like HSI, NAFLD, and others using the same logic and required input variables.

# Derive FLI using derive_param_computed
adlb <- adlb %>%
  derive_param_computed(
    by_vars = exprs(!!!get_admiral_option("subject_keys"), AVISIT, AVISITN, ADT, ADY, BMI, WSTCIR, !!!adsl_vars),
    parameters = c("TRIG", "GGT"),
    set_values_to = exprs(
      AVAL = {
        lambda <- 0.953 * log(AVAL.TRIG) + 0.139 * BMI + 0.718 * log(AVAL.GGT) + 0.053 * WSTCIR - 15.745
        (exp(lambda) / (1 + exp(lambda))) * 100
      },
      PARAMCD = "FLI",
      PARAM = "Fatty Liver Index",
      PARAMN = 11
    )
  )

adlb <- adlb %>%
  arrange(!!!get_admiral_option("subject_keys"), ADT, PARAMN) # Arrange for consistency

Remaining ADLB Set-up

The admiral Creating a BDS Finding ADaM vignette describes further steps, including how to calculate baseline and change from baseline variables, add analysis flags (e.g., ANL01FL), handle reference ranges, categorizations, and other common ADLB requirements.

Example Scripts

ADaM	Sample Code
ADLB	ad_adlb.R