Table of contents

Background and method

Innovations for clinical trials may accelerate bringing new medicines to children, such as through more robustly quantifying treatment effects, combining phases and ages, expanding access, and more often involving patients and their organisations in the design of trials and documenting patient-centric endpoints. The objective of this page is to continually track such innovations using information on clinical trials and to present original insights.

Information from the EU Clinical Trials Register and ClinicalTrials.Gov was used for this analysis . For commenting, the full R code is included below. Paediatric trials were identified based on the protocol-related information that any of the paediatric age groups was to be included in the trial. Interventional trials with medicines were identified in ClinicalTrials.Gov based on type=Drug (which includes agent, pharmaceutical, medications, medicine, medicinal product).

### Options for subsequent chunks
#
# knitr
knitr::opts_chunk$set(dev = "svglite", 
                      fig.width = 6, 
                      fig.height = 4,
                      out.width = "600px", 
                      out.height = "400px",
                      width = "600px",
                      echo = TRUE, 
                      eval = TRUE, 
                      results = "asis",
                      warning = FALSE, 
                      message = FALSE,
                      cache = TRUE)


### Helper function
#
# start date ctgov to posixct
start_date_ctgov <- function(x) {

  # September 30, 2017 -> ok
  # September 2013 -> September 15, 2013
  x <- sub("([a-zA-Z]+) ([0-9]{4})", "\\1 15, \\2", x)
  as.Date(x, format = "%B %d, %Y")

}
#
normalise_string <- function(x) {

  x <- gsub(",", "",  x)
  x <- gsub("-", " ", x)
  x <- tolower(x)
  x <- tools::toTitleCase(x)
  x <- gsub("[ ]+", " ", x)
  x <- trimws(x)
  x

}
#
normalise_number <- function(x) {

  x <- gsub("<=|<|≤|=|>", "",  x)
  x <- gsub("(^| )[.]", "0.", x)
  x <- trimws(x)
  x <- sapply(x, FUN = function(y) {
    z <- try(as.numeric(y), silent = TRUE)
    if (class(z) == "try-error") {
      NA
      message(z)
    } else {
      z
    }
  })
  unname(x)
}



### Initialise

# load libraries
# 
# For plotting:
library(ggplot2)
library(plotly)
library(kableExtra)
#
# This library was specially written
# by Ralf Herold and is available here:
# https://github.com/rfhb/ctrdata
library(ctrdata)

### Get data, manually 

if (FALSE) {

  ## euctr paediatric trials
  dbc <- nodbi::src_mongo(collection = "all_paed_trials_euctr", db = "users")
  # tmp.df <- dbQueryHistory(con = dbc)
  # tmp.qs <- "&age=adolescent&age=children&age=infant-and-toddler&age=newborn&age=preterm-new-born-infants&age=under-18"
  # ctrLoadQueryIntoDb(queryterm = tmp.qs, con = dbc)
  ctrLoadQueryIntoDb(querytoupdate = "last", con = dbc)

  ## euctr adult oncology trials
  dbc <- nodbi::src_mongo(collection = "all_adult_oncology_trials_euctr", db = "users")
  # tmp.df <- dbQueryHistory(con = dbc)
  # tmp.qs <- "https://www.clinicaltrialsregister.eu/ctr-search/search?query=cancer&age=adult&status=completed"
  # ctrLoadQueryIntoDb(queryterm = tmp.qs, con = dbc, only.count = TRUE)
  # ctrLoadQueryIntoDb(queryterm = tmp.qs, con = dbc)
  ctrLoadQueryIntoDb(querytoupdate = "last", con = dbc)

  ## ctgov
  dbc <- nodbi::src_mongo(collection = "all_paed_trials_ctgov", db = "users")
  # tmp.df <- dbQueryHistory(con = dbc)
  # tmp.qs <- "https://clinicaltrials.gov/ct2/results?type=Intr&age=0&intr=Drug&phase=4&phase=0&phase=1&phase=2&strd_e=12%2F31%2F2010"
  # tmp.qs <- "https://clinicaltrials.gov/ct2/results?type=Intr&age=0&intr=Drug&phase=4&phase=0&phase=1&phase=2&strd_s=12%2F31%2F2010"
  # ctrOpenSearchPagesInBrowser(input = tmp.qs)
  # ctrLoadQueryIntoDb(queryterm = tmp.qs, con = dbc, only.count = TRUE)
  # ctrLoadQueryIntoDb(queryterm = tmp.qs, con = dbc)
  ctrLoadQueryIntoDb(querytoupdate = 1L, con = dbc)
  ctrLoadQueryIntoDb(querytoupdate = 2L, con = dbc)

}

100

101

102

103

104

105

106

### Options for subsequent chunks

# knitr

knitr::opts_chunk$set(dev = "svglite",

fig.width = 6,

fig.height = 4,

out.width = "600px",

out.height = "400px",

width = "600px",

echo = TRUE,

eval = TRUE,

results = "asis",

warning = FALSE,

message = FALSE,

cache = TRUE)

### Helper function

# start date ctgov to posixct

start_date_ctgov <- function(x) {

# September 30, 2017 -> ok

# September 2013 -> September 15, 2013

x <- sub("([a-zA-Z]+) ([0-9]{4})", "\\1 15, \\2", x)

as.Date(x, format = "%B %d, %Y")

}

normalise_string <- function(x) {

x <- gsub(",", "", x)

x <- gsub("-", " ", x)

x <- tolower(x)

x <- tools::toTitleCase(x)

x <- gsub("[ ]+", " ", x)

x <- trimws(x)

}

normalise_number <- function(x) {

x <- gsub("<=|<|≤|=|>", "", x)

x <- gsub("(^| )[.]", "0.", x)

x <- trimws(x)

x <- sapply(x, FUN = function(y) {

z <- try(as.numeric(y), silent = TRUE)

if (class(z) == "try-error") {

message(z)

} else {

}

})

unname(x)

}

### Initialise

# load libraries

# For plotting:

library(ggplot2)

library(plotly)

library(kableExtra)

# This library was specially written

# by Ralf Herold and is available here:

# https://github.com/rfhb/ctrdata

library(ctrdata)

### Get data, manually

if (FALSE) {

## euctr paediatric trials

dbc <- nodbi::src_mongo(collection = "all_paed_trials_euctr", db = "users")

# tmp.df <- dbQueryHistory(con = dbc)

# tmp.qs <- "&age=adolescent&age=children&age=infant-and-toddler&age=newborn&age=preterm-new-born-infants&age=under-18"

# ctrLoadQueryIntoDb(queryterm = tmp.qs, con = dbc)

ctrLoadQueryIntoDb(querytoupdate = "last", con = dbc)

## euctr adult oncology trials

dbc <- nodbi::src_mongo(collection = "all_adult_oncology_trials_euctr", db = "users")

# tmp.df <- dbQueryHistory(con = dbc)

# tmp.qs <- "https://www.clinicaltrialsregister.eu/ctr-search/search?query=cancer&age=adult&status=completed"

# ctrLoadQueryIntoDb(queryterm = tmp.qs, con = dbc, only.count = TRUE)

# ctrLoadQueryIntoDb(queryterm = tmp.qs, con = dbc)

ctrLoadQueryIntoDb(querytoupdate = "last", con = dbc)

## ctgov

dbc <- nodbi::src_mongo(collection = "all_paed_trials_ctgov", db = "users")

# tmp.df <- dbQueryHistory(con = dbc)

# tmp.qs <- "https://clinicaltrials.gov/ct2/results?type=Intr&age=0&intr=Drug&phase=4&phase=0&phase=1&phase=2&strd_e=12%2F31%2F2010"

# tmp.qs <- "https://clinicaltrials.gov/ct2/results?type=Intr&age=0&intr=Drug&phase=4&phase=0&phase=1&phase=2&strd_s=12%2F31%2F2010"

# ctrOpenSearchPagesInBrowser(input = tmp.qs)

# ctrLoadQueryIntoDb(queryterm = tmp.qs, con = dbc, only.count = TRUE)

# ctrLoadQueryIntoDb(queryterm = tmp.qs, con = dbc)

ctrLoadQueryIntoDb(querytoupdate = 1L, con = dbc)

ctrLoadQueryIntoDb(querytoupdate = 2L, con = dbc)

}

Topics under preparation

Expansion cohorts of early trials
Estimation of outcomes
Patient-reported outcomes
Clinical trials and disease burden in children
Integrated non-controlled and randomised trials
Adaptive trials

Platform trials

A single clinical trial can be designed and used as platform for investigating different diseases and different investigational medicines at the same time . A basket trial, for example, has different arms corresponding to different diseases, thus permitting a broader range of patients to be included. Both, the effects of treatment with a medicine and the susceptibility of the diseases to the treatment may be documented. An umbrellat trial has different arms corresponding to different medicines to which a patient may be allocated. Other terms in the context of platform trials are complex trials, master protocols and matrix trials. Methods to design, conduct and analyse such trials vary much and are increasingly discussed .

In ClinicalTrials.Gov, basket and umbrella trials with children are identified here by non-randomised allocation to one of at least three arms with largely different labels (not counting control arms), and at least as many conditions as arms. In the plot of the trials, larger symbols are more likely basket or umbrella trials (based on dissimilarity of conditions and arms).

# select database
dbc <- nodbi::src_mongo(collection = "all_paed_trials_ctgov", db = "users")

# get result set
result <- dbGetFieldsIntoDf(
  c("start_date", "condition", "number_of_arms", "arm_group.arm_group_label", 
    "brief_title", "arm_group.arm_group_type", "intervention.arm_group_label",
    "detailed_description.textblock", "official_title", "study_design_info.allocation"), 
  con = dbc)

# get unique trials
result <- result[ 
  result[["_id"]] %in% 
    dbFindIdsUniqueTrials(
      con = dbc), ]

# format data
result$start_year <- as.numeric(format(result$start_date, "%Y"))
result$number_of_arms <- as.numeric(result$number_of_arms)

# count number of conditions
result$number_conditions <- sapply(result$condition, 
                                   FUN = function(x) length(x))

# calculate condition similarity
result$condition_similarity <- sapply(
  result$condition, 
  FUN = function(x) median(RecordLinkage::levenshteinSim(x, rev(x))))

## similarity of arms
#
# see https://prsinfo.clinicaltrials.gov/definitions.html#ArmType
# remove those arms that are labelled as either 
# "Active Comparator", "Placebo Comparator", "Sham Comparator", "No Intervention"
# keep "Experimental" and "Other"
tmp.armskeep <- lapply(result$arm_group.arm_group_type, 
                       FUN = function(x) trimws(unlist(x)) %in% c("Experimental", "Other"))
#
# split arm labels, remove any "group n:" prefixes
# tmp.strings[2188]
tmp.armsgrp <- lapply(result$arm_group.arm_group_label, 
                      FUN = function(x) sub("^.*: (.*)$", "\\1", trimws(unlist(x))))
#
# combine, keep only unique labels, keep only labels with relevant information  
# and with minimum length
tmp.remove <- c("arm", "dose", "mg", "placebo", "control", "rollover", "cohort", "group", 
                "treatment", "regimen", "stratum", "strata", "level", "phase", "subjects", 
                "patients", "study", "scheme", "participants", "part", "with", "open", 
                "label", "intervention", " [0-9A-Z] ", "[A-Z0-1][.] ", " [+]", " and")
tmp.strings <- lapply(seq_along(tmp.armskeep), 
                      FUN = function(x) {
                        tmp <- unique(tmp.armsgrp[[x]][tmp.armskeep[[x]]])
                        for (i in tmp.remove) tmp <- gsub(i, "", tmp, ignore.case = TRUE)
                        tmp <- trimws(tmp)
                        tmp <- tmp[nchar(tmp) > 3]
                        tmp}
)
#
# calculate similarity
result$arms_similarity <- sapply(
  tmp.strings, 
  FUN = function(x) median(RecordLinkage::levenshteinSim(x, rev(x))))
#
result$arms_strings <- sapply(
  tmp.strings, 
  FUN = function(x) paste0(x, collapse = " / "))
# 
# recount number of remaining arms
result$number_of_arms_revised <- stringr::str_count(result$arms_strings, pattern = " / ") + 1

# relevance factor
result$relevance <- result$number_conditions / result$condition_similarity * 
  result$number_of_arms_revised / result$arms_similarity

# the title may increasingly reflect these designs
if (FALSE) {
  with(result, 
       table(grepl("basket study|basket trial|umbrella study|umbrella trial", 
                   official_title, ignore.case = TRUE)))
}

# select trials
result <- subset(
  result, 
  subset = 
    (
      number_of_arms_revised >= 3 & 
        ((number_conditions >= number_of_arms_revised) | arms_similarity < 0.2) &
        # condition_similarity < 0.1) &
        # grepl("tumo[u]?rs|cancers", condition, ignore.case = TRUE)) &
        grepl("Non-", study_design_info.allocation) &
        !grepl("HSCT|transplant", official_title, ignore.case = TRUE)
    ) | (
      grepl("basket study|basket trial|umbrella study|umbrella trial", 
            official_title, ignore.case = TRUE)
    ), 
  relevance > 0
)

# diagnostics
if (FALSE) hist(result$condition_similarity) 
if (FALSE) hist(result$arms_similarity)
if (FALSE) hist(log(result$relevance)) # exp(5)
if (FALSE) plot(result$arms_similarity, result$number_of_arms)
if (FALSE) plot(result$arms_similarity, result$number_conditions)
if (FALSE) plot(result$number_conditions, result$number_of_arms_revised)

# format
result$title <- paste0(gsub("(.{1,40})(\\s|$)", "\\1<br>", result$brief_title), "<br>arms: ",
                       gsub("(.{1,40})(\\s|$)", "\\1<br>", result$arms_strings))

result$title <- gsub("[^[:ascii:]]", "", result$title, perl = TRUE)

vline <- function(x = 0, color = "blue") {
  list(type = "line", yref = "paper", line = list(color = color),
       y0 = 0, y1 = 1, x0 = x, x1 = x)}

p <- plot_ly(data = result, 
             hoverinfo = "text", 
             text = ~title, 
             x = ~jitter(start_year, factor = 2), 
             y = ~jitter(number_of_arms_revised, factor = 2), 
             color = I("black"),
             type = "scatter", 
             mode = "markers", 
             width = 600, 
             height = 400,
             marker = list(opacity = 0.5, 
                           sizemode = "diameter",
                           size = ~relevance / 75)
             ) %>%
  layout(title = "Paediatric basket or umbrella trials",
         xaxis = list(title = "Year of trial start", 
                      gridcolor = "rgb(255, 255, 255)",
                      ticks = "outside",
                      range = c(2000, 2025)), 
         yaxis = list(title = "Number of arms", 
                      gridcolor = "rgb(255, 255, 255)",
                      zeroline = FALSE,
                      ticks = "outside",
                      range = c(1, 18)), 
         font = list(family = "Helvetica", size = 13), 
         autosize = FALSE, 
         shapes = list(vline(as.numeric(Sys.Date()) * 24 * 60 * 60 * 1000)), 
         plot_bgcolor = "rgb(230, 230, 230)"
  ) %>% 
  add_annotations(
    x = 2001, y = 17, 
    text = paste0("paediatricdata.eu ", format(Sys.Date(), format = "%Y-%m-%d"), 
                  "\nSize indicates likelihood of\nbasket or umbrella design"),
    xanchor = "left", yanchor = "top", showarrow = FALSE)

#
# produce code
# htmlwidgets::saveWidget(widget = p, file = "baskettrials.html", selfcontained = TRUE)
# htmltools::browsable(htmltools::tagList(list(p)))
print(htmltools::tagList(p))

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

# select database

dbc <- nodbi::src_mongo(collection = "all_paed_trials_ctgov", db = "users")

# get result set

result <- dbGetFieldsIntoDf(

c("start_date", "condition", "number_of_arms", "arm_group.arm_group_label",

"brief_title", "arm_group.arm_group_type", "intervention.arm_group_label",

"detailed_description.textblock", "official_title", "study_design_info.allocation"),

con = dbc)

# get unique trials

result <- result[

result[["_id"]] %in%

dbFindIdsUniqueTrials(

con = dbc), ]

# format data

result$start_year <- as.numeric(format(result$start_date, "%Y"))

result$number_of_arms <- as.numeric(result$number_of_arms)

# count number of conditions

result$number_conditions <- sapply(result$condition,

FUN = function(x) length(x))

# calculate condition similarity

result$condition_similarity <- sapply(

result$condition,

FUN = function(x) median(RecordLinkage::levenshteinSim(x, rev(x))))

## similarity of arms

# see https://prsinfo.clinicaltrials.gov/definitions.html#ArmType

# remove those arms that are labelled as either

# "Active Comparator", "Placebo Comparator", "Sham Comparator", "No Intervention"

# keep "Experimental" and "Other"

tmp.armskeep <- lapply(result$arm_group.arm_group_type,

FUN = function(x) trimws(unlist(x)) %in% c("Experimental", "Other"))

# split arm labels, remove any "group n:" prefixes

# tmp.strings[2188]

tmp.armsgrp <- lapply(result$arm_group.arm_group_label,

FUN = function(x) sub("^.*: (.*)$", "\\1", trimws(unlist(x))))

# combine, keep only unique labels, keep only labels with relevant information

# and with minimum length

tmp.remove <- c("arm", "dose", "mg", "placebo", "control", "rollover", "cohort", "group",

"treatment", "regimen", "stratum", "strata", "level", "phase", "subjects",

"patients", "study", "scheme", "participants", "part", "with", "open",

"label", "intervention", " [0-9A-Z] ", "[A-Z0-1][.] ", " [+]", " and")

tmp.strings <- lapply(seq_along(tmp.armskeep),

FUN = function(x) {

tmp <- unique(tmp.armsgrp[[x]][tmp.armskeep[[x]]])

for (i in tmp.remove) tmp <- gsub(i, "", tmp, ignore.case = TRUE)

tmp <- trimws(tmp)

tmp <- tmp[nchar(tmp) > 3]

tmp}

)

# calculate similarity

result$arms_similarity <- sapply(

tmp.strings,

FUN = function(x) median(RecordLinkage::levenshteinSim(x, rev(x))))

result$arms_strings <- sapply(

tmp.strings,

FUN = function(x) paste0(x, collapse = " / "))

# recount number of remaining arms

result$number_of_arms_revised <- stringr::str_count(result$arms_strings, pattern = " / ") + 1

# relevance factor

result$relevance <- result$number_conditions / result$condition_similarity *

result$number_of_arms_revised / result$arms_similarity

# the title may increasingly reflect these designs

if (FALSE) {

with(result,

table(grepl("basket study|basket trial|umbrella study|umbrella trial",

official_title, ignore.case = TRUE)))

}

# select trials

result <- subset(

result,

subset =

(

number_of_arms_revised >= 3 &

((number_conditions >= number_of_arms_revised) | arms_similarity < 0.2) &

# condition_similarity < 0.1) &

# grepl("tumo[u]?rs|cancers", condition, ignore.case = TRUE)) &

grepl("Non-", study_design_info.allocation) &

!grepl("HSCT|transplant", official_title, ignore.case = TRUE)

) | (

grepl("basket study|basket trial|umbrella study|umbrella trial",

official_title, ignore.case = TRUE)

relevance > 0

)

# diagnostics

if (FALSE) hist(result$condition_similarity)

if (FALSE) hist(result$arms_similarity)

if (FALSE) hist(log(result$relevance)) # exp(5)

if (FALSE) plot(result$arms_similarity, result$number_of_arms)

if (FALSE) plot(result$arms_similarity, result$number_conditions)

if (FALSE) plot(result$number_conditions, result$number_of_arms_revised)

# format

result$title <- paste0(gsub("(.{1,40})(\\s|$)", "\\1<br>", result$brief_title), "<br>arms: ",

gsub("(.{1,40})(\\s|$)", "\\1<br>", result$arms_strings))

result$title <- gsub("[^[:ascii:]]", "", result$title, perl = TRUE)

vline <- function(x = 0, color = "blue") {

list(type = "line", yref = "paper", line = list(color = color),

y0 = 0, y1 = 1, x0 = x, x1 = x)}

p <- plot_ly(data = result,

hoverinfo = "text",

text = ~title,

x = ~jitter(start_year, factor = 2),

y = ~jitter(number_of_arms_revised, factor = 2),

color = I("black"),

type = "scatter",

mode = "markers",

width = 600,

height = 400,

marker = list(opacity = 0.5,

sizemode = "diameter",

size = ~relevance / 75)

) %>%

layout(title = "Paediatric basket or umbrella trials",

xaxis = list(title = "Year of trial start",

gridcolor = "rgb(255, 255, 255)",

ticks = "outside",

range = c(2000, 2025)),

yaxis = list(title = "Number of arms",

gridcolor = "rgb(255, 255, 255)",

zeroline = FALSE,

ticks = "outside",

range = c(1, 18)),

font = list(family = "Helvetica", size = 13),

autosize = FALSE,

shapes = list(vline(as.numeric(Sys.Date()) * 24 * 60 * 60 * 1000)),

plot_bgcolor = "rgb(230, 230, 230)"

) %>%

add_annotations(

x = 2001, y = 17,

text = paste0("paediatricdata.eu ", format(Sys.Date(), format = "%Y-%m-%d"),

"\nSize indicates likelihood of\nbasket or umbrella design"),

xanchor = "left", yanchor = "top", showarrow = FALSE)

# produce code

# htmlwidgets::saveWidget(widget = p, file = "baskettrials.html", selfcontained = TRUE)

# htmltools::browsable(htmltools::tagList(list(p)))

print(htmltools::tagList(p))

Randomisation in early phase trials

Randomisation throughout all development phases has recently been recommended when small populations are to be studied such as for personalised medicines , including for possibilities to separate prognostic importance of a biomarker from treatment effects and to faster transition phases.

While a seizable portion of early phase paediatric trials that include randomisation also include randomisation to placebo or to no treatment, it is not yet clear how much such approaches are used across paediatric therapeutic areas.

# select database
dbc <- nodbi::src_mongo(collection = "all_paed_trials_ctgov", db = "users")

# get result set
result <- dbGetFieldsIntoDf(
  c("start_date", "brief_title", "study_design_info.allocation", 
    "phase", "arm_group.arm_group_type"), 
  con = dbc)

# get unique trials
result <- result[ result[["_id"]] %in% dbFindIdsUniqueTrials(con = dbc), ]

# format data
result$start_year <- as.numeric(format(result$start_date, "%Y"))

# select trials
result <- subset(
  result,
  grepl("^Rando", study_design_info.allocation, ignore.case = TRUE) &
  grepl("^Phase 1$", phase, ignore.case = TRUE)
)

# identify trials of interest
result$control <- grepl("(Placebo|No Intervention)", 
                        result$arm_group.arm_group_type, 
                        ignore.case = TRUE)

# plot
ggplot(data = result, 
       aes(x = start_year, 
           fill = control)) + 
  geom_histogram(binwidth = 1) + 
  labs(title = "Paediatric phase 1-only interventional trials\nwith randomisation in CT.Gov", 
       x = "Year of first trial authorisation", 
       y = "Number of trials", 
       fill = "Randomisation\nto placebo or\nno intervention?") + 
  xlim(c(2000, 2021)) +
  ylim(c(0, 100)) +
  geom_vline(xintercept = as.numeric(format(Sys.Date(), format = "%Y")), 
             colour = "gray")

# select database

dbc <- nodbi::src_mongo(collection = "all_paed_trials_ctgov", db = "users")

# get result set

result <- dbGetFieldsIntoDf(

c("start_date", "brief_title", "study_design_info.allocation",

"phase", "arm_group.arm_group_type"),

con = dbc)

# get unique trials

result <- result[ result[["_id"]] %in% dbFindIdsUniqueTrials(con = dbc), ]

# format data

result$start_year <- as.numeric(format(result$start_date, "%Y"))

# select trials

result <- subset(

result,

grepl("^Rando", study_design_info.allocation, ignore.case = TRUE) &

grepl("^Phase 1$", phase, ignore.case = TRUE)

)

# identify trials of interest

result$control <- grepl("(Placebo|No Intervention)",

result$arm_group.arm_group_type,

ignore.case = TRUE)

# plot

ggplot(data = result,

aes(x = start_year,

fill = control)) +

geom_histogram(binwidth = 1) +

labs(title = "Paediatric phase 1-only interventional trials\nwith randomisation in CT.Gov",

x = "Year of first trial authorisation",

y = "Number of trials",

fill = "Randomisation\nto placebo or\nno intervention?") +

xlim(c(2000, 2021)) +

ylim(c(0, 100)) +

geom_vline(xintercept = as.numeric(format(Sys.Date(), format = "%Y")),

colour = "gray")

Adolescents in “adult” trials

If adolescent patients are included in trials originally planned only for adults, separate and later trials with these paediatric patients may be avoided. Such an age-inclusive approach would build on scientific and medical considerations for the specific trial, and also on general intentions to improve access for these patients and to shorten timelines of developing a medicines of children. Recommendations for oncology were recently put forward by multiple stakeholders . The approach can also be relevant for later phase trials and perhaps other therapeutic areas. Further analyses may be by type of sponsor or by the authorisation status of the medicine.

To track progress with respect to the recommendations, the analysis is limited to the therapeutic area of oncology and to phase 2 trials (search term "cancer" and phase 2 recorded in the EU Clinical Trials Register), because adult phase 1 trials are not be required to be made public in the registers.

However, phase 2 trials may be therapeutic exploratory trials and therefore for a different cancer(s) than in children; therefore, it is expected that only a portion of such trials is relevant for adolescents with a cancer. The number of phase 2 oncology trials open to adolescents seems to be slightly increasing. More analyses will be done on the subset of trials that are for cancers that occur in the paediatric population.

# select database
dbc <- nodbi::src_mongo(collection = "all_adult_oncology_trials_euctr", db = "users")

# get result set
result <- dbGetFieldsIntoDf(
  c("a1_member_state_concerned", "f116_adolescents_1217_years",
    "n_date_of_competent_authority_decision", 
    "a2_eudract_number", "e72_therapeutic_exploratory_phase_ii", 
    "x6_date_on_which_this_record_was_first_entered_in_the_eudract_database"), 
  con = dbc)

# limit to phase 1
result <- subset(result, 
                 e72_therapeutic_exploratory_phase_ii == TRUE)

# calculate start date
result$startdate <- ifelse(!is.na(tmp <- result$n_date_of_competent_authority_decision), tmp, 
                           result$x6_date_on_which_this_record_was_first_entered_in_the_eudract_database)
class(result$startdate) <- "Date"

# find first member state to authorise trial
tmp <- aggregate(result$startdate, 
                 by = list(result$a2_eudract_number),
                 FUN = function(x) min(x))
result <- merge(x = result, y = tmp, 
                by.x = "a2_eudract_number", 
                by.y = "Group.1", 
                all.x = TRUE)
names(result)[ncol(result)] <- "startdatefirst"
result$startyearfirst <- as.numeric(format(result$startdatefirst, "%Y"))

# find if any member state recorded the inclusion of adolescents
result$adolescents <- ifelse(result$f116_adolescents_1217_years == TRUE, TRUE, FALSE)
tmp <- aggregate(result$adolescents, 
                 by = list(result$a2_eudract_number),
                 FUN = function(x) any(x))
result <- merge(x = result, y = tmp, 
                by.x = "a2_eudract_number", 
                by.y = "Group.1", 
                all.x = TRUE)
names(result)[ncol(result)] <- "adolescentsanyms"

# get unique trials
result <- result[ result[["_id"]] %in% 
                    dbFindIdsUniqueTrials(con = dbc), ]

# plot
ggplot(data = result, 
       aes(x = startyearfirst, fill = adolescentsanyms)) + 
  geom_histogram(binwidth = 1) + 
  labs(title = "Adult phase 2 oncology trials in EUCTR", 
       x = "Year of first trial authorisation", 
       y = "Number of trials", 
       fill = "Adolescents\nincluded?") + 
  xlim(c(2000, 2020)) +
  ylim(c(0, 300)) +
  geom_vline(xintercept = as.numeric(format(Sys.Date(), format = "%Y")), 
             colour = "gray")

# select database

dbc <- nodbi::src_mongo(collection = "all_adult_oncology_trials_euctr", db = "users")

# get result set

result <- dbGetFieldsIntoDf(

c("a1_member_state_concerned", "f116_adolescents_1217_years",

"n_date_of_competent_authority_decision",

"a2_eudract_number", "e72_therapeutic_exploratory_phase_ii",

"x6_date_on_which_this_record_was_first_entered_in_the_eudract_database"),

con = dbc)

# limit to phase 1

result <- subset(result,

e72_therapeutic_exploratory_phase_ii == TRUE)

# calculate start date

result$startdate <- ifelse(!is.na(tmp <- result$n_date_of_competent_authority_decision), tmp,

result$x6_date_on_which_this_record_was_first_entered_in_the_eudract_database)

class(result$startdate) <- "Date"

# find first member state to authorise trial

tmp <- aggregate(result$startdate,

by = list(result$a2_eudract_number),

FUN = function(x) min(x))

result <- merge(x = result, y = tmp,

by.x = "a2_eudract_number",

by.y = "Group.1",

all.x = TRUE)

names(result)[ncol(result)] <- "startdatefirst"

result$startyearfirst <- as.numeric(format(result$startdatefirst, "%Y"))

# find if any member state recorded the inclusion of adolescents

result$adolescents <- ifelse(result$f116_adolescents_1217_years == TRUE, TRUE, FALSE)

tmp <- aggregate(result$adolescents,

by = list(result$a2_eudract_number),

FUN = function(x) any(x))

result <- merge(x = result, y = tmp,

by.x = "a2_eudract_number",

by.y = "Group.1",

all.x = TRUE)

names(result)[ncol(result)] <- "adolescentsanyms"

# get unique trials

result <- result[ result[["_id"]] %in%

dbFindIdsUniqueTrials(con = dbc), ]

# plot

ggplot(data = result,

aes(x = startyearfirst, fill = adolescentsanyms)) +

geom_histogram(binwidth = 1) +

labs(title = "Adult phase 2 oncology trials in EUCTR",

x = "Year of first trial authorisation",

y = "Number of trials",

fill = "Adolescents\nincluded?") +

xlim(c(2000, 2020)) +

ylim(c(0, 300)) +

geom_vline(xintercept = as.numeric(format(Sys.Date(), format = "%Y")),

colour = "gray")

Involvement of paediatric networks

The involvement of paediatric networks is generally recommended to better prepare and conduct trials of medicinal products with children.

Among paediatric trials in the EU Clinical Trials Register with a commercial sponsor (mostly a pharmaceutical company), excluding phase 4 trials, so far a limited proportion (perhaps some 15%) is documented to involve an investigator network. These networks are tabulated below the histogram. Names of networks are spelled in different ways (not normalised or from a dictionary).

# select database
dbc <- nodbi::src_mongo(collection = "all_paed_trials_euctr", db = "users")

# get result set
result <- dbGetFieldsIntoDf(
  c("a1_member_state_concerned", "a2_eudract_number",
    "b1_sponsor.b31_and_b32_status_of_the_sponsor",
    "g4_investigator_networks.g4_investigator_network_to_be_involved_in_the_trial",
    "g4_investigator_networks.g41_name_of_organisation", 
    "e74_therapeutic_use_phase_iv",
    "n_date_of_competent_authority_decision", 
    "x6_date_on_which_this_record_was_first_entered_in_the_eudract_database"), 
  con = dbc)

# calculate start date
result$startdate <- ifelse(!is.na(tmp <- result$n_date_of_competent_authority_decision), tmp, 
                           result$x6_date_on_which_this_record_was_first_entered_in_the_eudract_database)
class(result$startdate) <- "Date"

# find first member state to authorise trial
tmp <- aggregate(result$startdate, 
                 by = list(result$a2_eudract_number),
                 FUN = function(x) min(x))
result <- merge(x = result, y = tmp, 
                by.x = "a2_eudract_number", 
                by.y = "Group.1", 
                all.x = TRUE)
names(result)[ncol(result)] <- "startdatefirst"
result$startyearfirst <- as.numeric(format(result$startdatefirst, "%Y"))

# find if any member state recorded any investigator network involvement
tmp <- ifelse(
  !is.na(result$g4_investigator_networks.g4_investigator_network_to_be_involved_in_the_trial),
  TRUE, FALSE)
tmp <- aggregate(tmp, 
                 by = list(result$a2_eudract_number),
                 FUN = function(x) any(x))
result <- merge(x = result, y = tmp, 
                by.x = "a2_eudract_number", 
                by.y = "Group.1", 
                all.x = TRUE)
names(result)[ncol(result)] <- "networkanyinvolved"

# find if any member state recorded any investigator network names
tmp <- ifelse(
  !is.na(result$g4_investigator_networks.g41_name_of_organisation) &
    result$g4_investigator_networks.g41_name_of_organisation != "" &
    !(grepl(
      paste0("no network|^NA$|^N/A$|not applicable|no aplica|no applica|", 
             "TO BE CONFIRMED|To be completed locally if applicable", 
             "To Be Advised", 
             collapse = ""), 
      result$g4_investigator_networks.g41_name_of_organisation, 
      ignore.case = TRUE)), 
  TRUE, FALSE)
tmp <- aggregate(tmp, 
                 by = list(result$a2_eudract_number),
                 FUN = function(x) any(x))
result <- merge(x = result, y = tmp, 
                by.x = "a2_eudract_number", 
                by.y = "Group.1", 
                all.x = TRUE)
names(result)[ncol(result)] <- "networkanyname"

# get names of involved networks
tmp.names <- tapply(X = result$g4_investigator_networks.g41_name_of_organisation, 
                    INDEX = list(result$a2_eudract_number),
                    FUN = function(x) {
                      tmp <- x
                      tmp <- paste0(tmp, collapse = " / ") # several records
                      tmp <- unlist(strsplit(tmp, " / "))
                      tmp <- trimws(tmp)
                      tmp <- unique(tmp[tmp != ""])
                      tmp <- tmp[!grepl(
                        paste0("no network|^NA$|^N/A$|not applicable|no aplica|no applica|", 
                               "TO BE CONFIRMED|To be completed locally if applicable", 
                               "To Be Advised", 
                               collapse = ""),
                        tmp, ignore.case = TRUE)]
                      tmp
                    })

# add names back to trials
tmp.names.df <- data.frame(
  names(tmp.names),
  sapply(tmp.names, function(x) {
    tmp <- x
    tmp <- sort(x)
    tmp <- paste0(tmp, collapse = " | ")
    if (tmp == "NA" | tmp == "") tmp <- NA
    tmp
    }),
  stringsAsFactors = FALSE)
names(tmp.names.df) <- c("a2_eudract_number", "networknames")
result <- merge(x = result, 
                y = tmp.names.df, 
                all.x = TRUE)

# get unique trials
result <- result[ result[["_id"]] %in% dbFindIdsUniqueTrials(
  con = dbc,
  include3rdcountrytrials = FALSE), ]
# hist(result$startyearfirst)

# select trials
# table(result$b1_sponsor.b31_and_b32_status_of_the_sponsor)
result <- subset(
  result,
  grepl("^Commercial.*|.* Commercial.*", 
        b1_sponsor.b31_and_b32_status_of_the_sponsor, 
        ignore.case = TRUE) &
  e74_therapeutic_use_phase_iv == FALSE
)

# plot
ggplot(data = result, 
       aes(x = startyearfirst, fill = networkanyinvolved)) + 
  geom_histogram(binwidth = 1) + 
  labs(title = "Paediatric phase 1 to 3 trials in EUCTR", 
       subtitle = "that have a commercial sponsor", 
       x = "Year of first trial authorisation", 
       y = "Number of trials", 
       fill = "Investigor\nnetwork\ninvolvement\ndocumented?") + 
  xlim(c(2000, 2020)) +
  ylim(c(0, 250)) +
  geom_vline(xintercept = as.numeric(format(Sys.Date(), format = "%Y")), 
             colour = "gray")

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

# select database

dbc <- nodbi::src_mongo(collection = "all_paed_trials_euctr", db = "users")

# get result set

result <- dbGetFieldsIntoDf(

c("a1_member_state_concerned", "a2_eudract_number",

"b1_sponsor.b31_and_b32_status_of_the_sponsor",

"g4_investigator_networks.g4_investigator_network_to_be_involved_in_the_trial",

"g4_investigator_networks.g41_name_of_organisation",

"e74_therapeutic_use_phase_iv",

"n_date_of_competent_authority_decision",

"x6_date_on_which_this_record_was_first_entered_in_the_eudract_database"),

con = dbc)

# calculate start date

result$startdate <- ifelse(!is.na(tmp <- result$n_date_of_competent_authority_decision), tmp,

result$x6_date_on_which_this_record_was_first_entered_in_the_eudract_database)

class(result$startdate) <- "Date"

# find first member state to authorise trial

tmp <- aggregate(result$startdate,

by = list(result$a2_eudract_number),

FUN = function(x) min(x))

result <- merge(x = result, y = tmp,

by.x = "a2_eudract_number",

by.y = "Group.1",

all.x = TRUE)

names(result)[ncol(result)] <- "startdatefirst"

result$startyearfirst <- as.numeric(format(result$startdatefirst, "%Y"))

# find if any member state recorded any investigator network involvement

tmp <- ifelse(

!is.na(result$g4_investigator_networks.g4_investigator_network_to_be_involved_in_the_trial),

TRUE, FALSE)

tmp <- aggregate(tmp,

by = list(result$a2_eudract_number),

FUN = function(x) any(x))

result <- merge(x = result, y = tmp,

by.x = "a2_eudract_number",

by.y = "Group.1",

all.x = TRUE)

names(result)[ncol(result)] <- "networkanyinvolved"

# find if any member state recorded any investigator network names

tmp <- ifelse(

!is.na(result$g4_investigator_networks.g41_name_of_organisation) &

result$g4_investigator_networks.g41_name_of_organisation != "" &

!(grepl(

paste0("no network|^NA$|^N/A$|not applicable|no aplica|no applica|",

"TO BE CONFIRMED|To be completed locally if applicable",

"To Be Advised",

collapse = ""),

result$g4_investigator_networks.g41_name_of_organisation,

ignore.case = TRUE)),

TRUE, FALSE)

tmp <- aggregate(tmp,

by = list(result$a2_eudract_number),

FUN = function(x) any(x))

result <- merge(x = result, y = tmp,

by.x = "a2_eudract_number",

by.y = "Group.1",

all.x = TRUE)

names(result)[ncol(result)] <- "networkanyname"

# get names of involved networks

tmp.names <- tapply(X = result$g4_investigator_networks.g41_name_of_organisation,

INDEX = list(result$a2_eudract_number),

FUN = function(x) {

tmp <- x

tmp <- paste0(tmp, collapse = " / ") # several records

tmp <- unlist(strsplit(tmp, " / "))

tmp <- trimws(tmp)

tmp <- unique(tmp[tmp != ""])

tmp <- tmp[!grepl(

paste0("no network|^NA$|^N/A$|not applicable|no aplica|no applica|",

"TO BE CONFIRMED|To be completed locally if applicable",

"To Be Advised",

collapse = ""),

tmp, ignore.case = TRUE)]

tmp

})

# add names back to trials

tmp.names.df <- data.frame(

names(tmp.names),

sapply(tmp.names, function(x) {

tmp <- x

tmp <- sort(x)

tmp <- paste0(tmp, collapse = " | ")

if (tmp == "NA" | tmp == "") tmp <- NA

tmp

}),

stringsAsFactors = FALSE)

names(tmp.names.df) <- c("a2_eudract_number", "networknames")

result <- merge(x = result,

y = tmp.names.df,

all.x = TRUE)

# get unique trials

result <- result[ result[["_id"]] %in% dbFindIdsUniqueTrials(

con = dbc,

include3rdcountrytrials = FALSE), ]

# hist(result$startyearfirst)

# select trials

# table(result$b1_sponsor.b31_and_b32_status_of_the_sponsor)

result <- subset(

result,

grepl("^Commercial.*|.* Commercial.*",

b1_sponsor.b31_and_b32_status_of_the_sponsor,

ignore.case = TRUE) &

e74_therapeutic_use_phase_iv == FALSE

)

# plot

ggplot(data = result,

aes(x = startyearfirst, fill = networkanyinvolved)) +

geom_histogram(binwidth = 1) +

labs(title = "Paediatric phase 1 to 3 trials in EUCTR",

subtitle = "that have a commercial sponsor",

x = "Year of first trial authorisation",

y = "Number of trials",

fill = "Investigor\nnetwork\ninvolvement\ndocumented?") +

xlim(c(2000, 2020)) +

ylim(c(0, 250)) +

geom_vline(xintercept = as.numeric(format(Sys.Date(), format = "%Y")),

colour = "gray")

# print most often documented network
result <- subset(result, 
                 subset = !is.na(networknames),
                 select = c("networknames"))[,1]
result <- unlist(strsplit(result, " | ", fixed = TRUE))
# some manual normalisation
result <- sapply(
  result, 
  function(x) switch(
    x,
    "Medicines for Children Research Network MCRN" = "Medicines for Children Research Network",
    "The Medicines for Children Research Network MCRN" = "Medicines for Children Research Network",
    "Cystic Fibrosis - National Knowledge Service UK CF NKS" = "Cystic Fibrosis - National Knowledge Service",
    "Cystic Fibrosis - National Knowledge Service UK CF-NKS" = "Cystic Fibrosis - National Knowledge Service",
    "Cystic Fibrosis - National Knowledge Service UK CFNKS" = "Cystic Fibrosis - National Knowledge Service",
    x
  ))
result <- table(result)
result <- result[rev(order(result))]
result <- as.data.frame(result,
                        stringsAsFactors = FALSE)
names(result) <- c("Name of investigator network documented to be involved in ...", "... number of trials")
# output
#knitr::kable(result[1:15], col.names = c("Network name", "Trials"))
kable_styling(
  kable(
    result[1:40,], 
    col.names = c("Network name", "Trials")),
  bootstrap_options = c("striped", "hover", "condensed", "responsive"),
  full_width = FALSE,
  position = "left")

# print most often documented network

result <- subset(result,

subset = !is.na(networknames),

select = c("networknames"))[,1]

result <- unlist(strsplit(result, " | ", fixed = TRUE))

# some manual normalisation

result <- sapply(

result,

function(x) switch(

"Medicines for Children Research Network MCRN" = "Medicines for Children Research Network",

"The Medicines for Children Research Network MCRN" = "Medicines for Children Research Network",

"Cystic Fibrosis - National Knowledge Service UK CF NKS" = "Cystic Fibrosis - National Knowledge Service",

"Cystic Fibrosis - National Knowledge Service UK CF-NKS" = "Cystic Fibrosis - National Knowledge Service",

"Cystic Fibrosis - National Knowledge Service UK CFNKS" = "Cystic Fibrosis - National Knowledge Service",

))

result <- table(result)

result <- result[rev(order(result))]

result <- as.data.frame(result,

stringsAsFactors = FALSE)

names(result) <- c("Name of investigator network documented to be involved in ...", "... number of trials")

# output

#knitr::kable(result[1:15], col.names = c("Network name", "Trials"))

kable_styling(

kable(

result[1:40,],

col.names = c("Network name", "Trials")),

bootstrap_options = c("striped", "hover", "condensed", "responsive"),

full_width = FALSE,

position = "left")

Network name	Trials
Medicines for Children Research Network	26
PRINTO	10
PRCSG	8
NIHR CRN	5
Innovative Therapies for Children with Cancer ITCC	5
Cystic Fibrosis Foundation	5
Cystic Fibrosis - National Knowledge Service	5
Cystic Fibrosis Foundation Therapeutics	4
CRN: North Thames Central Office	4
Ancillare	4
Sunovion Pharmaceuticals Inc.	3
European Cystic Fibrosis Society - Clinical Trial Network	3
ECFS CTN	3
Cystische Fibrose - Netzwerk Klinische Studien German CF-NKS	3
Cystic Fibrosis Foundation Therapeutics - Therapeutics Development Network	3
Clinical Research Network CRN	3
Vasculair Research Network VRN	2
TREAT-NMD, The University of Newcastle upon Tyne	2
The Experimental Cancer Medicine Centres	2
Teddy Network	2
NIHR CRN National Coordinating Centre	2
NIHR Clinical Research Network	2
NETSTAP e.V.	2
National Cancer Research Network Coordinating Centre NCRN CC	2
Medicines for Children Network MCRN	2
Medicines for Children Network	2
Medical Research Network Ltd.	2
MCRN Medicines for Children Research Network	2
ITCC	2
IQVIA	2
ICON Clinical Research LLC	2
European Cystic Fibrosis Society Clinical Trials Network ECFS-CTN	2
ECFS Clinical Trials Network	2
Cystic Fibrosis Foundation Therapeutics, Inc	2
Cystic Fibrosis Foundation Therapeutics CFFT Therapeutics Development Network TDN	2
CRN North Thames, NIHR Clinical Research Network CRN	2
Comprehensive Clinical Research Network and Medicines for Children Research Network	2
Clinical Research Network North Thames	2
Childrens Oncology Group Operations Center	2
Childrens Oncology Group	2

# DT::datatable(
#   result,
#    filter = list(position = "top", 
#                  clear = FALSE, 
#                  plain = TRUE),
#   options = list(
#     dom = "lprt",
#     search = list(regex = TRUE, caseInsensitive = TRUE),
#     #language = list(search = "Search in any column:"),
#     searchHighlight = TRUE),
#     style = "bootstrap")

# DT::datatable(

# result,

# filter = list(position = "top",

# clear = FALSE,

# plain = TRUE),

# options = list(

# dom = "lprt",

# search = list(regex = TRUE, caseInsensitive = TRUE),

# #language = list(search = "Search in any column:"),

# searchHighlight = TRUE),

# style = "bootstrap")

Innovation coming to paediatric research

Background and method

Topics under preparation

Platform trials

Randomisation in early phase trials

Adolescents in “adult” trials

Involvement of paediatric networks

References

Leave a Reply Cancel reply

Get updates