In sutsabs/rwToT2: Real World Time on Treatment R Package

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
options(rmarkdown.html_vignette.check_title = FALSE)

Introduction

Real World Time on Treatment (rwToT) has been regarded as a practical surrogate to measure benefits of immunotherapy across several tumors. From business standpoint, treatment duration is a key parameter included in Sponsor’s forecast calculations hence playing a critical role in financial planning and manufacturing supply chain planning. To meet the needs from both internal and external stakeholders, robust analytical techniques are required for real world time on treatment (rwToT) estimation, prediction, understanding of factors associated as well as evaluation of sensitivity and uncertainty in these techniques based on different types of real-world data sources. Sponsor continues to obtain access to multiple oncology real world data sets that will be used to address rwToT needs in an ongoing basis. [@Utsab: we need to add reference at the end of: protocol name and authors: Real World Time on Treatment for anti-PD-1/PD-L1 Based Regimens- Multi Data Source Analysis, author: Sheenu Chandwani, Kaushal Desai,]

rwToT Estimation

• rwToT is defined as length of time (in months) between first and last administration date before discontinuation, for the drug of interest within line of therapy of interest, with differences in determination of discontinuation event, as described below.

  'rwToT duration = Last administration date - First administration + 1'

• All observations will be separated into discontinued vs censored observations. Patient was considered to have discontinued treatment at the earliest of:

  • New Line initiation: If there is record of initiation of next line of therapy (using line of therapy variable)
  • Death: If there is record of patient death during line of therapy of interest [not applied for persistence]
  • Off-treatment: If there is >=120* day gap between last administration date and last known activity date in the dataset If none of the discontinuation criteria are met, patients are considered censored at their last administration date.

• The Kaplan-Meier (KM) methodology will be applied to conduct time to event rwToT analysis and generate the following statistics:

  • KM plot
  • Median (95% CI)
  • Restricted mean (RMST) at specific landmark time(e.g., 6-, 12- months). RMST is defined as the mean rwToT of all subjects in the study population followed up to t, and is simply the area under the rwToT KM curve up to t. When sufficient risk set size is not available at the landmark time due to limited follow-up/data immaturity, parametric functions will be fitted to the KM rwToT. Statistical tests based on the Akaike Information Criterion (AIC) and the Bayesian information criterion (BIC) to assess closeness of fit overall will be used to select the best-fitting parametric distribution. The Pocock 10% rule is used as the decision criteria to determine data maturity. According to this rule the population at risk at time t lower than 10% of the starting population will be considered immature. After the fit is done, the restricted mean calculation can then be performed on this fitted curve at time t.Please refer Protocol(VEAP 7760) for detailed method description
  • On-treatment rate at specific landmark time

Examples

Load Package

library(rwToT)
library(survival)
library(dplyr)

Find the treatment rate using rwToT::get_treatment_rate()

fit_km <- survfit(with(aml, Surv(time,status)) ~ 1)
rwToT::get_treatment_rate(fit_km ,18)

Find the restricted mean using rwToT::get_restricted_mean()

## Get restricted mean at 36 months
rmean_36mon <- rwToT::get_restricted_mean(rwToT_test_file_A %>% 
                                            #converting DD days to months
                                            mutate(DD = (DD/365) * 12), t = 36)
rmean_36mon$mean
knitr::kable(head(rmean_36mon$parametric_km))
knitr::kable(tail(rmean_36mon$parametric_km))

Get the output table with rwToT Statistics using best-fit model rwToT::get_analysis()

```r rwToT_result <- rwToT::get_analysis(rwToT_test_file_B, "Cohort B", database_cutoff_date = "2021-01-31") knitr::kable(rwToT_result$output_table) knitr::kable(head(rwToT_result$parametric_km))

sutsabs/rwToT2 documentation built on Feb. 18, 2022, 2:30 a.m.