Creating ADRS with Lugano 2014 Criteria

Introduction

This article describes creating an ADRS ADaM dataset for lymphoma studies based on the Lugano 2014 response criteria.

Lymphoma response assessment under Lugano 2014 is based on a combination of imaging-based evaluations:

PET-CT based assessment, providing metabolic response evaluation using the 5-point Deauville scale.
CT-based assessment, providing anatomic response evaluation of nodal and extranodal disease.

Depending on the study and the disease subtype, response evaluation may use PET-CT as the primary modality (for FDG-avid lymphomas) or CT alone (for non–FDG-avid lymphomas). Some studies collect both PET-CT and CT response components, and a combined overall response is derived.

Please check Lugano 2014 Classification for more details.

Note: In many Lugano 2014 studies, the overall timepoint response may be collected directly from the investigator or independent review committee. In such cases, the collected overall response should generally be used according to the study protocol and statistical analysis plan, and the derivation shown below may not be needed.

The derivation below is provided only as an example of how an integrated timepoint response could be derived when the combined overall response is not collected directly. It is not intended as general Lugano 2014 implementation guidance. Study-specific rules may vary and should be aligned with the protocol, SAP, CRF design, and data review conventions.

For extended guidance on common steps in ADRS creation and additional response endpoints, refer to the examples in Creating ADRS (Including Non-standard Endpoints).

Lugano 2014 Response Categories for Lymphoma

The Lugano 2014 response criteria define lymphoma response using PET-CT based metabolic assessment and CT-based anatomic assessment.

The following tables summarize the response categories used in this vignette for PET-CT and CT assessments. These summaries are intended to support the example derivations and should be aligned with the study protocol and statistical analysis plan.

PET-CT Based Response Categories

PET-CT Response	Description
Table 1: PET-CT Based Response Categories
Lugano 2014 response categories used in this vignette
CMR	Complete metabolic response
PMR	Partial metabolic response
NMR or SMD	No metabolic response or stable metabolic disease
PMD	Progressive metabolic disease
NE	Not evaluable
ND	Not done or not determined
NED	No evidence of FDG-avid disease, generally BICR or IRC only
PSP	Pseudoprogression

CT-Based Response Categories

CT Response	Description
Table 2: CT-Based Response Categories
Anatomic response categories used in this vignette
CAR	Complete anatomic response
PAR	Partial anatomic response
SAD	Stable anatomic disease
PAD	Progressive anatomic disease
NE	Not evaluable
ND	Not done or not determined
NED	No evidence of disease

In this example data, NMR is used for no metabolic response. Some implementations may use SMD for stable metabolic disease. For the purpose of the combined overall response derivation in this vignette, both NMR and SMD map to SD.

Values such as NED, PSP, NE, and ND require study-specific handling. For example, NED by PET-CT is generally expected only from a blinded independent central review (BICR) or independent review committee (IRC) and may indicate that no FDG-avid disease was identified at baseline. In that case, the integrated timepoint response often defaults to the CT response if one is available.

Programming Workflow

Read in Data
Pre-processing of Input Records
Derive PET-CT and CT Response Parameters
Derive Combined Overall Timepoint Response (OVRLRESC)
Other Endpoints

Required Packages

The examples of this vignette require the following packages.

library(admiral)
library(admiralonco)
library(pharmaversesdtm)
library(pharmaverseadam)
library(dplyr)
library(tibble)

Read in Data

To begin, all data frames needed for the creation of ADRS should be read into the environment. This will be a company-specific process. For this vignette, the main input datasets are ADSL and RS.

For demonstration purposes, the SDTM and ADaM datasets based on CDISC Pilot test data from pharmaversesdtm and pharmaverseadam are used.

In this vignette, the RS SDTM dataset is expected to contain lymphoma response assessments based on Lugano 2014 criteria. The example RS dataset contains separate records for:

PET-CT based response assessments, identified by RSSCAT = "INCLUDING PET-CT SCAN".
CT-based response assessments, identified by RSSCAT = "NOT INCLUDING PET SCAN".

# Lymphoma SDTM data
rs <- pharmaversesdtm::rs_onco_lymphoma

# Convert blanks to NA
rs <- convert_blanks_to_na(rs)

# ADaM data
adsl <- pharmaverseadam::adsl

Pre-processing of Input Records

At this step, it may be useful to join ADSL to your RS domain. Only the ADSL variables used for derivations are selected at this step.

adsl_vars <- exprs(TRTSDT)
adrs <- derive_vars_merged(
  rs,
  dataset_add = adsl,
  new_vars = adsl_vars,
  by_vars = get_admiral_option("subject_keys")
)

Partial Date Imputation and Deriving `ADT`, `ADTF`, `AVISIT`, `AVISITN` etc.

If your data collection allows for partial dates, you could apply a company-specific imputation rule at this stage when deriving ADT. For this example, here we impute missing day to last possible date.

adrs <- adrs %>%
  derive_vars_dtm(
    dtc = RSDTC,
    new_vars_prefix = "A",
    highest_imputation = "D",
    date_imputation = "last"
  ) %>%
  derive_vars_dtm_to_dt(exprs(ADTM)) %>%
  derive_vars_dy(
    reference_date = TRTSDT,
    source_vars = exprs(ADT)
  ) %>%
  mutate(
    AVISIT = VISIT,
    AVISITN = VISITNUM
  )

Derive `PARAMCD`, `PARAM`, `PARAMN`

In this RS dataset, both PET-CT and CT response records use RSTESTCD = "OVRLRESP" and are distinguished by RSSCAT and RSMETHOD. For this vignette, RSTESTCD and RSSCAT are used to derive PARAMCD, PARAM, and PARAMN.

# Prepare param_lookup for SDTM RSTESTCD and RSSCAT to add metadata
param_lookup <- tibble::tribble(
  ~RSTESTCD,  ~RSSCAT,                   ~PARAMCD, ~PARAM,            ~PARAMN,
  "OVRLRESP", "INCLUDING PET-CT SCAN",   "PETRSP", "PET-CT Response",       1,
  "OVRLRESP", "NOT INCLUDING PET SCAN",  "CTRSP",  "CT Response",           2
)

adrs <- adrs %>%
  derive_vars_merged_lookup(
    dataset_add = param_lookup,
    by_vars = exprs(RSTESTCD, RSSCAT)
  ) %>%
  mutate(
    PARCAT1 = RSCAT,
    AVALC = case_when(
      RSSTAT == "NOT DONE" ~ "ND",
      TRUE ~ RSSTRESC
    )
  )

Derive Combined Overall Timepoint Response(`OVRLRESC`) Parameter

For this vignette, the combined overall timepoint response parameter, OVRLRESC, is derived from the PET-CT and CT response records collected at each visit.

This example represents a scenario where the combined overall response is not collected directly on the CRF. Instead, it is derived using the available PET-CT and CT response records.

General Derivation Assumptions Used in This Vignette

The following table summarizes the assumptions used in this vignette to derive the combined overall timepoint response from PET-CT and CT response records under Lugano 2014. These assumptions are intended for demonstration purposes. Please refer to your study protocol, statistical analysis plan, and other study documentation before using in production analyses.

Table: Combined Overall Timepoint Response Based on Lugano 2014 Response Categories

PET-CT Response	CT Response	Combined Overall Response
Table 3: Combined Overall Timepoint Response
PET-CT, CT, and Combined Overall Response Mapping
CMR	Any	CR
PMR	Any	PR
NMR or SMD	Any	SD
PMD	Any	PD
PSP	Any	PSP
NED	Any	Use current CT response
NE / ND, with prior evaluable PET-CT	CAR / PAR / SAD / NE / ND / NED	Carry forward prior PET-CT response
NE / ND, with prior evaluable PET-CT	PAD	PD
NE / ND, no prior evaluable PET-CT	CAR / PAR / SAD / PAD / NED	Use current CT response
NE / ND, no prior evaluable PET-CT	NE / ND / Missing	NE or ND
Missing	Any	Use current CT response
Missing	Missing	ND
This table is example-only and should be aligned with the study protocol and statistical analysis plan.
For evaluable PET-CT responses CMR, PMR, NMR or SMD, and PMD, the PET-CT response determines the integrated response in this example.
When PET-CT is NE or ND and the current CT response is not progressive, prior evaluable PET-CT response may be carried forward.
When PET-CT is reported as NED, the integrated response generally defaults to the CT response if available. If both PET-CT and CT are NED, the integrated response is NED.
Pseudoprogression handling is study-specific and is not implemented further in this example.

Combined Overall Timepoint Response(`OVRLRESC`) Records referenced from above table

# Pre-processing for Overall values
map_pet_to_overall <- function(x) {
  recode_values(
    x,
    "CMR" ~ "CR",
    "PMR" ~ "PR",
    c("NMR", "SMD") ~ "SD",
    "PMD" ~ "PD",
    "NED" ~ "NED"
  )
}

map_ct_to_overall <- function(x) {
  recode_values(
    x,
    "CAR" ~ "CR",
    "PAR" ~ "PR",
    "SAD" ~ "SD",
    "PAD" ~ "PD",
    "NED" ~ "NED",
    "NE" ~ "NE",
    "ND" ~ "ND"
  )
}

Derive prior evaluable PET-CT response for carry-forward logic

adrs <- adrs %>%
  restrict_derivation(
    filter = PARAMCD == "PETRSP",
    derivation = derive_vars_joined,
    args = params(
      dataset_add = adrs,
      filter_add = PARAMCD == "PETRSP" &
        AVALC %in% c("CMR", "PMR", "NMR", "SMD", "PMD"),
      by_vars = get_admiral_option("subject_keys"),
      order = exprs(ADT, AVISITN),
      mode = "last",
      join_type = "before",
      filter_join = ADT.join < ADT,
      new_vars = exprs(
        AVALC_P = AVALC,
        ADT_P = ADT
      )
    )
  )

Derive Combined Overall Timepoint Response

Please note that the by variables used below depend on the data collection. In the example it is assumed that PET-CT and CT response records are collected at the same day. If this is not the case, the variables ADT, ADY, ADTM, and ADTF shouldn’t be used. In addition, you may use RSSPID to identify the records that should be combined.

adrs <- derive_param_computed(
  dataset = adrs,
  by_vars = exprs(
    !!!get_admiral_option("subject_keys"), !!!adsl_vars, DOMAIN, ADT, ADY, ADTM,
    ADTF, VISIT, VISITNUM, AVISIT, AVISITN
  ),
  parameters = c("PETRSP", "CTRSP"),
  set_values_to = exprs(
    AVALC = case_when(
      # PET-CT evaluable: metabolic response determines overall response
      AVALC.PETRSP %in% c("CMR", "PMR", "NMR", "SMD", "PMD") ~
        map_pet_to_overall(AVALC.PETRSP),

      # PET-CT NED: default to CT if CT is available
      AVALC.PETRSP == "NED" &
        AVALC.CTRSP %in% c("CAR", "PAR", "SAD", "PAD", "NE", "ND", "NED") ~
        map_ct_to_overall(AVALC.CTRSP),

      # PET-CT NED and CT missing: keep NED
      AVALC.PETRSP == "NED" & is.na(AVALC.CTRSP) ~
        "NED",

      # PET-CT is NE or ND and CT indicates progression
      AVALC.PETRSP %in% c("NE", "ND") & AVALC.CTRSP == "PAD" ~
        "PD",

      # PET-CT is NE or ND and prior evaluable PET-CT exists
      AVALC.PETRSP %in% c("NE", "ND") &
        !is.na(AVALC_P.PETRSP) &
        AVALC.CTRSP %in% c("CAR", "PAR", "SAD", "NE", "ND", "NED") ~
        map_pet_to_overall(AVALC_P.PETRSP),

      # PET-CT is NE or ND and no prior evaluable PET-CT exists
      AVALC.PETRSP %in% c("NE", "ND") &
        is.na(AVALC_P.PETRSP) &
        AVALC.CTRSP %in% c("CAR", "PAR", "SAD", "PAD", "NED") ~
        map_ct_to_overall(AVALC.CTRSP),

      # PET-CT is NE and CT is also NE or ND or missing
      AVALC.PETRSP == "NE" &
        (AVALC.CTRSP %in% c("NE", "ND") | is.na(AVALC.CTRSP)) ~
        "NE",

      # PET-CT is ND and CT is also NE or ND or missing
      AVALC.PETRSP == "ND" &
        (AVALC.CTRSP %in% c("NE", "ND") | is.na(AVALC.CTRSP)) ~
        "ND",

      # PET-CT missing; use CT response if available
      is.na(AVALC.PETRSP) &
        AVALC.CTRSP %in% c("CAR", "PAR", "SAD", "PAD", "NED", "NE", "ND") ~
        map_ct_to_overall(AVALC.CTRSP),

      # No valid response available
      TRUE ~ "ND"
    ),
    PARAMCD = "OVRLRESC",
    PARAM = "Overall Response - Derived",
    PARAMN = 3,
    PARCAT1 = "LUGANO 2014"
  ),
  keep_nas = TRUE
)

Derive `AVAL` (Numeric tumor response from `AVALC` values)

The AVAL values are not considered in the further parameter derivations below, and so changing AVAL here would not change the result of those derivations.

adrs <- adrs %>%
  mutate(
    AVAL = recode_values(
      AVALC,
      c("CR", "CMR", "CAR") ~ 1,
      c("PR", "PMR", "PAR") ~ 2,
      c("SD", "NMR", "SMD", "SAD") ~ 3,
      c("PD", "PMD", "PAD") ~ 4,
      "NE" ~ 5,
      "NED" ~ 6,
      "ND" ~ 7
    )
  )

Other Endpoints

The OVRLRESC parameter can be used as input for the derivation of standard endpoints, such as Best Overall Response (BOR), Confirmed Best Overall Response (CBOR), and other oncology response endpoints. Please see Creating ADRS (Including Non-standard Endpoints) for guidance on how to derive them.