Select the MTD from Completed Trials

Choose the maximum tolerated dose for each of a set of completed trials from the final patient and DLT counts.

Usage

boin_select_mtd(
  n_pts,
  n_tox,
  target,
  cutoff_eli = 0.95,
  extrasafe = FALSE,
  offset = 0.05,
  bound_mtd = FALSE,
  p_tox = NULL,
  mtd_max_estimate = NULL,
  min_mtd_sample = 1
)

Arguments

n_pts: Integer matrix with one row per trial and one column per dose, or a vector for a single trial, giving the number of patients treated.
n_tox: Integer matrix or vector of the same shape as n_pts, giving the number of DLTs observed.
target: Numeric scalar. Target DLT probability.
cutoff_eli: Numeric scalar. Posterior probability cutoff for dose elimination. Defaults to 0.95.
extrasafe: Logical scalar. Apply the stricter safety rule at the lowest dose, in which case no dose is selected when the lowest dose is judged too toxic. Defaults to FALSE.
offset: Numeric scalar between 0 and 0.5. Amount by which cutoff_eli is relaxed for the safety rule. Defaults to 0.05.
bound_mtd: Logical scalar. Require the isotonic estimate at the selected dose to be at or below the de-escalation boundary. Defaults to FALSE.
p_tox: Numeric scalar. Lowest DLT probability deemed overly toxic, used to derive the de-escalation boundary when bound_mtd is TRUE. Defaults to 1.4 * target.
mtd_max_estimate: Numeric scalar or NULL. Largest isotonic estimate a dose may have and still be selected. Supplying it bounds the selection whatever bound_mtd says, and replaces the de-escalation boundary that bound_mtd would otherwise use. Defaults to NULL.
min_mtd_sample: Integer scalar. Smallest number of patients a dose must have received to be eligible. Defaults to 1, which admits every dose that treated a patient.

Value

A data frame with one row per trial and columns

trial: Trial index.
mtd: Selected dose level, or NA when no dose is selected.
reason: Why the dose was or was not selected.

Details

Dose elimination is re-derived from the final data of each trial rather than carried over from the dose-finding stage, which is what the reference implementation does and which matters when a trial ends at its maximum sample size. The isotonic fit is computed over the admissible doses only, so that eliminated doses do not influence the estimates of the doses that remain. The dose whose estimate is closest to target is selected, with ties broken by a small increasing perturbation across the admissible doses.

bound_mtd caps the estimate at the de-escalation boundary, which always lies above target. A cap at or below the target rate therefore cannot be expressed that way, and needs mtd_max_estimate. The two caps are independent of the decision table, so bounding the selection does not change how the trial was run.

Values of reason are "selected", "lowest_dose_eliminated", "no_admissible_dose" and "no_dose_below_bound".

References

Liu S. and Yuan, Y. (2015). Bayesian Optimal Interval Designs for Phase I Clinical Trials. Journal of the Royal Statistical Society: Series C, 64, 507-523.

Examples

n_pts <- matrix(c(3, 6, 9, 3,
                  3, 6, 9, 3), nrow = 2, byrow = TRUE)
n_tox <- matrix(c(0, 1, 3, 2,
                  0, 1, 2, 1), nrow = 2, byrow = TRUE)

boin_select_mtd(n_pts, n_tox, target = 0.30)
#>   trial mtd   reason
#> 1     1   3 selected
#> 2     2   4 selected

boin_select_mtd(n_pts, n_tox, target = 0.30, bound_mtd = TRUE)
#>   trial mtd   reason
#> 1     1   3 selected
#> 2     2   4 selected

# Require the isotonic estimate at the selected dose to be at most 0.30,
# which the de-escalation boundary cannot express
boin_select_mtd(n_pts, n_tox, target = 0.30, mtd_max_estimate = 0.30)
#>   trial mtd   reason
#> 1     1   2 selected
#> 2     2   3 selected