Swimerplot RECIST : implémentation

[DanChaltiel]

WIP

[Lpierott]

**#please follow the style of the example of function made by Dan Chaltiel for the waterfall plot
# https://github.com/Oncostat/grstat/blob/main/R/waterfall_plot.R****

library(RColorBrewer)
dim(REQ_bas_pat_EOT_RES_mhyn_th_re_treat_adm)

baseline_v2b=subset(baseline,
                    select=c(SUBJID,BASEDT, BOPTN1, TASGROUP,TASPLSDT, BPCONSDT,BICONSDT))
dim(baseline_v2b)

dim(recist_rep)
dim(eot_v2)

baseline_v2b=baseline_v2b %>% 
  arrange(SUBJID) %>% 
  select(SUBJID ,BPCONSDT ,BICONSDT)  
patient=REQ_bas_pat_EOT_RES_mhyn_th_re_treat_adm %>% 
  arrange(SUBJID) %>% 
  left_join(baseline_v2b, by=c("SUBJID")) %>% 
  mutate(consdt=BPCONSDT) %>% 
  mutate(consdt=ifelse(!is.na(consdt),BICONSDT,NA))

patient$consdt = as.POSIXct(patient$consdt*3600*24 , origin="1970-01-01",format="%Y-%m-%d") %>% 
  set_label("Date")
patient$consdt=as.POSIXct(format(patient$consdt,"%Y-%m-%d"))
table(patient$consdt)


ADM_last_v4b=ADM_last_v4 %>% 
  arrange(SUBJID) %>% 
  select(SUBJID ,CRFNAME ,ADMYN, ADMDT_first,last_ADMDT) 
swim=patient %>% 
  left_join(ADM_last_v4b, by=c("SUBJID")) %>% 
  select(SUBJID, consdt, CRFNAME ,ADMYN, ADMDT_first,last_ADMDT) %>% 
  mutate(T0=consdt) %>% 
  mutate(T0bis=ADMDT_first) %>% 
  mutate(Texam=last_ADMDT) 
table(swim$CRFNAME  , useNA="always")

ADM_vb=adm %>% 
  arrange(as.numeric(SUBJID), ADMCYCLE) %>%
  select(SUBJID ,CRFNAME ,ADMYN, ADMDT) %>% 
  group_by(SUBJID) %>% 
  mutate(subjectid=row_number()) %>% 
  mutate(ADMDT_first=ifelse(subjectid==1,ADMDT,NA)) %>% 
  mutate(ADMDT_first = as.POSIXct(ADMDT_first*3600*24 , origin="1970-01-01",format="%Y-%m-%d") )
swim=patient %>% 
  left_join(ADM_vb, by=c("SUBJID")) %>% 
  select(SUBJID, consdt, CRFNAME ,ADMYN,ADMDT,first_ADMDT) %>% 
  mutate(T0=consdt) %>% 
  mutate(T0bis=first_ADMDT) %>% 
  mutate(Texam=ADMDT) 
table(swim$CRFNAME  , useNA="always")

patient_b=patient %>% 
  arrange(SUBJID) %>% 
  select(SUBJID, consdt ,first_ADMDT) %>% 
  rename(ADMDT_first =first_ADMDT) 

recist_repb=RECIST_v2 %>% 
  left_join(patient_b, by=c("SUBJID")) %>% 
  mutate(CRFNAME="Recist") %>%  
  mutate(T0=consdt) %>% 
  mutate(T0bis=ADMDT_first) %>% 
  mutate(Texam=RCDT) %>% 
  select(SUBJID ,CRFNAME ,RCVISIT, RCDT, RCRESP, Texam, T0, T0bis) %>% 
  distinct() 
table(recist_repb$CRFNAME  , useNA="always")

eot_v2b=eot_v2 %>% 
  left_join(patient_b, by=c("SUBJID")) %>% 
  mutate(Texam=EOTLADDT) %>% 
  mutate(T0=consdt) %>% 
  mutate(T0bis=ADMDT_first) %>% 
  mutate(CRFNAME="end of treatment")  %>% 
  select(SUBJID, CRFNAME, EOTLADDT, Texam, T0, T0bis) 
table(eot_v2b$CRFNAME  , useNA="always")

swim$Texam=as.POSIXct(format(swim$Texam,"%Y-%m-%d"))
recist_repb$Texam=as.POSIXct(format(recist_repb$Texam,"%Y-%m-%d"))
eot_v2b$Texam=as.POSIXct(format(eot_v2b$Texam,"%Y-%m-%d"))

summary(swim)
swim$T0bis=as.POSIXct(format(swim$T0bis,"%Y-%m-%d"))
recist_repb$T0bis=as.POSIXct(format(recist_repb$T0bis,"%Y-%m-%d"))
eot_v2b$T0bis=as.POSIXct(format(eot_v2b$T0bis,"%Y-%m-%d"))

# TO DO: change bind_row to pivot_longer 

swim2=bind_rows(swim, recist_repb,eot_v2b )  %>% 
  mutate(examdl=(Texam-T0)/(3600*24)) %>% 
  mutate(examdlbis=(Texam-T0bis)/(3600*24)) %>% 
  mutate(EXAMDL2=examdl/30) %>% 
  mutate(EXAMDL2bis=examdlbis/30)

dth_death= dth_v3 %>% 
  select(SUBJID, DTHDT ) %>% 
  mutate(CRFNAME="Death")  
dth_death$DTHDT=as.POSIXct(format(dth_death$DTHDT,"%Y-%m-%d"))
summary(swim2)
dth_death2=dth_death %>% 
  left_join(swim2, by = join_by(SUBJID)) %>% 
  mutate(time_to_death=(DTHDT-T0bis)) %>% 
  mutate(time_to_death=time_to_death/30) %>% 
  select(SUBJID, time_to_death ) %>% 
  distinct(SUBJID, time_to_death)

swim3=bind_rows(swim2,dth_death2 ) %>% 
  mutate(CRFNAME=ifelse(!is.na(time_to_death),"Death",CRFNAME )) %>% 
  mutate(EXAMDL2bis=ifelse(!is.na(time_to_death),time_to_death,EXAMDL2bis ))

table( swim3$RCRESP,  swim3$CRFNAME, useNA="always")
table( swim3$time_to_death, useNA="always")


# FU ----------------------------------------------------------------------

names(fu)


fu_fu= fu %>% 
  select(SUBJID, FUDT ) %>% 
  mutate(CRFNAME="Alive at last follow up2")  
table( fu_fu$FUDT, useNA="always")

fu_fu$FUDT=as.POSIXct(format(fu_fu$FUDT,"%Y-%m-%d"))

summary(swim3)
fu_fu2=fu_fu %>% 
  left_join(swim3, by = join_by(SUBJID)) %>% 
  mutate(time_to_fu=(FUDT-T0bis)) %>% 
  mutate(time_to_fu=time_to_fu/30) %>% 
  select(SUBJID, time_to_fu ) %>% 
  distinct(SUBJID, time_to_fu)

swim4=bind_rows(swim3,fu_fu2 ) %>% 
  mutate(CRFNAME=ifelse(!is.na(time_to_fu),"Alive at last follow up2",CRFNAME )) %>% 
  mutate(EXAMDL2bis=ifelse(!is.na(time_to_fu),time_to_fu,EXAMDL2bis ))

table( swim4$RCRESP,  swim4$CRFNAME, useNA="always")
table( swim4$time_to_fu, useNA="always")


add_legend= dth_v3 %>% 
  select(SUBJID) %>%
  mutate(CRFNAME="Alive at last follow up")  %>% 
filter(SUBJID==86 ) %>% 
mutate(EXAMDL2bis=-1 )

add_legend2= dth_v3 %>% 
  select(SUBJID) %>%
  mutate(CRFNAME="Treatment period")  %>% 
  filter(SUBJID==86 ) %>% 
  mutate(EXAMDL2bis=-1 )

add_legend3=bind_rows(add_legend2,add_legend ) 

swim5=bind_rows(swim4,add_legend3 ) 


table(swim5$RCVISIT,  swim5$CRFNAME, swim5$ADMYN,useNA="always")
table( swim5$CRFNAME, useNA="always")
table( swim5$ADMYN, useNA="always")
table( swim5$RCVISIT, useNA="always")
table( swim5$RCRESP,  swim5$CRFNAME, useNA="always")
table( swim5$RCRESP,  swim5$CRFNAME, useNA="always")

swim6=swim5 %>% 
  mutate(visit=ifelse(CRFNAME=="Treatment Administration",1, NA)) %>% 
  mutate(visit=ifelse(CRFNAME=="Recist" & RCRESP=="Complete response" ,2, visit))%>% 
  mutate(visit=ifelse(CRFNAME=="Recist" & RCRESP=="Partial response" ,2, visit))%>% 
  mutate(visit=ifelse(CRFNAME=="Recist" & RCRESP=="Stable disease" ,3, visit))%>% 
  mutate(visit=ifelse(CRFNAME=="Recist" & RCRESP=="Progressive disease" ,4, visit))%>% 
  mutate(visit=ifelse(CRFNAME=="Recist" & RCRESP=="Not evaluable" ,5, visit))%>% 
  mutate(visit=ifelse(CRFNAME=="end of treatment"  ,6, visit)) %>% 
  mutate(visit=ifelse(CRFNAME=="Death"  ,7, visit)) %>% 
  mutate(visit=ifelse(CRFNAME=="Alive at last follow up2"  ,8, visit)) %>% 
  mutate(visit=ifelse(CRFNAME=="Alive at last follow up"  ,9, visit)) %>% 
   mutate(visit=ifelse(CRFNAME=="Treatment period",10,visit )) %>% 
  filter(RCVISIT=="Treatment Period" | RCVISIT=="End of treatment" |RCVISIT== "Follow-up"|is.na(RCVISIT) )  %>% 
  filter(ADMYN=="Yes" | is.na(ADMYN)) %>% 
  distinct() %>% 
  mutate(subjid_num=as.numeric(SUBJID)) %>% 
  arrange(subjid_num, Texam)  

table( swim6$visit, useNA="always")

dim(swim6)
summary(swim6)
suivi=swim6

length(unique(suivi$subjid))

colnames(suivi)=str_to_lower(colnames(suivi))
table( suivi$visit, useNA="always")

# Conversion de la variable visite en facteur 

suivi$visit2 <- factor(suivi$visit, c(1:10), 
                       c("Trt Administration",
                         "CR/PR",
                         "SD",
                         "PD",
                         "Not evaluable", 
                         "End of trt", "Death", "Alive at last follow up2", "Alive at last follow up", "Treatment period"))

table(suivi$rcresp, useNA="always")
table( suivi$visit2, useNA="always")

table(suivi$visit2)

summary(suivi$examdl2bis)
summary(suivi$visit2)
summary(suivi$texam)

# Plot --------------------------------------------------------------------
names(suivi)
suivi_trt = suivi %>% 
  filter(visit2=="Trt Administration")%>%
  mutate(
    subjid_num = fct_reorder(factor(subjid_num), examdl2bis, .fun=max, na.rm=TRUE)
  ) %>% 
  summarise(
    first_trt = min(examdl2bis, na.rm=TRUE),
    last_trt = max(examdl2bis, na.rm=TRUE),
    .by=subjid_num
  )

names(suivi)
suivi_fu = suivi %>% 
  filter(visit2=="Alive at last follow up2" | visit2=="Trt Administration")%>%
  mutate(
    subjid_num = fct_reorder(factor(subjid_num), examdl2bis, .fun=max, na.rm=TRUE)
  ) %>% 
  summarise(
    first_fu = min(examdl2bis, na.rm=TRUE),
    last_fu = max(examdl2bis, na.rm=TRUE),
    .by=subjid_num
  )


dat_swim <- 
  suivi |>
  mutate(max_examdl2bis = max(examdl2bis, na.rm=TRUE),  .by=subjid_num) |>
  mutate(subjid_num = fct_reorder(factor(subjid_num), max_examdl2bis,  na.rm=TRUE))


# plot_final -------------------------------------------------------------------

# Function swimmer plots --------------------------------------------------

#' Create a Swimmer Plot of Overall Response by Patient
#'
#' @param dat_swim Data frame containing visit data
#' @param suivi_trt Data frame with treatment period information
#' @param suivi_fu Data frame with follow-up period information
#' @return A ggplot object representing the swimmer plot
#' @import ggplot2 dplyr
#' @export
swimmer_plot <- function(dat_swim, suivi_trt, suivi_fu) {
  requireNamespace("ggplot2")
  requireNamespace("dplyr")
  
  dat_filtered <- dat_swim %>%
    dplyr::filter(!is.na(visit2)) %>%
    dplyr::filter(!visit2 %in% c("End of trt", "Alive at last follow up2", "Trt Administration")) %>%
    dplyr::select(subjid_num, visit2, examdl2bis) %>%
    dplyr::distinct()
  
  plot <- ggplot2::ggplot(dat_filtered, ggplot2::aes(
    x = examdl2bis, 
    y = factor(subjid_num), 
    color = visit2, 
    shape = visit2, 
    size = visit2
  )) +
    ggplot2::geom_point(position = ggplot2::position_dodge(width = 0.4), size = 2) +
    ggplot2::geom_segment(
      ggplot2::aes(x = first_trt, y = subjid_num, xend = last_trt, yend = subjid_num),
      color = "skyblue",
      alpha = 0.3,
      linewidth = 1.5,
      inherit.aes = FALSE,
      data = suivi_trt
    ) +
    ggplot2::geom_segment(
      ggplot2::aes(x = first_fu, y = subjid_num, xend = last_fu, yend = subjid_num),
      color = "grey",
      inherit.aes = FALSE,
      data = suivi_fu,
      arrow = ggplot2::arrow(length = ggplot2::unit(0.3, "cm"))
    ) +
    ggplot2::scale_shape_manual(values = c(19, 8, 15, 4, 18, 62, 15)) +
    ggplot2::scale_color_manual(values = c(
      "Treatment period" = "skyblue",
      "CR/PR" = "green",
      "PD" = "purple",
      "Not evaluable" = "grey",
      "SD" = "yellow",
      "End of trt" = "pink",
      "Death" = "red",
      "Alive at last follow up" = "grey"
    )) +
    ggplot2::scale_x_continuous(
      name = "Time (in month) since first treatment administration",
      limits = c(0, 28),
      breaks = seq(0, 28, 2)
    ) +
    ggplot2::scale_y_discrete(name = "Patient") +
    ggplot2::geom_vline(xintercept = 0) +
    ggplot2::theme_classic() +
    ggplot2::labs(
      title = "By-patient Swimmer Plot of Overall Response",
      subtitle = paste0("n=", length(unique(dat_swim$subjid_num))),
      shape = NULL,
      color = NULL
    ) +
    ggplot2::theme(
      plot.title = ggplot2::element_text(face = "bold", size = 14, hjust = 0),
      plot.subtitle = ggplot2::element_text(size = 10, hjust = 0),
      axis.title.x = ggplot2::element_text(size = 10, face = "italic"),
      axis.title.y = ggplot2::element_text(size = 10, face = "italic", angle = 90),
      axis.ticks.y = ggplot2::element_blank(),
      axis.text.y = ggplot2::element_blank(),
      legend.text = ggplot2::element_text(size = 10)
    )
  
  return(plot)
}

# Generate the plot using your data frames
plot_final <- swimmer_plot(dat_swim, suivi_trt, suivi_fu)

# Display the plot
print(plot_final)

[Lpierott]

this is the idea of plot