Crm {UBCRM}

Dose-escalation with the Continual Reassessment Method

Description

The function gives the next level to include patients following a model-based design. Needs an updated input dataframe with the CreData() structure.

Usage

Crm(Dk, prior, target = 1/3, nptmax = 24, nmaxmtd = 6, nmaxdose = nptmax, sd = 1.34,
approach = "bayes", model = "power", method = "fpost", nextlevel = "ntarget",
upskipping = F, downskipping = F, lastdose = NA)

Arguments

Dk

Study dataframe with CreData() structure.

prior

Numeric vector of prior DLTs probabilities.

target

Target used for the MTD determination.

nptmax

Maximum number of patients to include in the study.

nmaxmtd

Maximum number of patients to be treated at the designated MTD. Assign a high value (=nptmax) to avoid such a stopping rule.

nmaxdose

Maximum number of patients to be treated at the same dose. Assign a high value (=nptmax) to avoid such a stopping rule.

sd

Standard deviation used in case of a normal distribution assumption for the parameter.

approach

Character indicating the estimation method: "bayes" (default value) for CRM or "mle" for CRML.

model

Character indicating the dose-DLT relationship model: "power", "tangent" or "logistic". More informations in the details section.

method

Estimation method for the posterior probabilities. "fpost" (default) estimates the mean of the posterior distribution of the parameter alpha (hat_alpha=E[alpha]) and uses it in psy(hat_alpha,...). "ppostp" and "pposts" directly estimate the mean of the posterior DLT probability. "ppostp" uses prior as singletons whereas "pposts" calculates appropriates singletons (see ail, ait or aip functions).

nextlevel

Character option used for determining the next dose level. "ntarget" (default) if the next level is chosen as the closest level to the desired target (may be higher than target). "utarget" if the next level is the closest level with the restriction to be lower than the target value.

upskipping

Boolean option used for determining the next dose level. If TRUE no level skip in escalation will be allowed. If FALSE (default) the level skips will be permitted.

downskipping

Boolean option used for determining the next dose level. If TRUE no level skip in desescalation will be allowed. If FALSE (default) the level skips will be permitted.

lastdose

Integer representing the last experimented dose level.

Details

Details of the 3 dose-DLT relationship proposed models: "power" for the power model psy(s,a)=s^exp(a), "tangent" for the hyperbolic tangent model psy(s,a)= ((tanh(s)+1)/2)**a, "logistic" for the logistic model psy(s,a) = exp(3+a*s)/(1+exp(3+a*s)). Note: power and tangent models are equivalent after an appropriate transformation on the parameter.

Value

nextdose

An integer representing the next recommended dose to experiment.

mtd

If reached, an integer representing the MTD.

prob

Posterior DLTs probabilities.

Author(s)

Benjamin Esterni, Baboukar Mane. Unite de Biostatistique et de Methodologie, Institut Paoli-Calmettes, Marseille, France.

References

O'Quigley J., Pepe M., Fisher L. (1990). Continual Reassessment Method: a practical design for Phase I clinical trials in cancer. Biometrics 46, 33-48. <https://doi.org/10.2307/2531628>

O'Quigley J., Shen LZ. (1996). Continual Reassessment Method: a likelihood approach. Biometrics 52, 673-684. <https://doi.org/10.2307/2532905>

Paoletti X., Kramar A. (2009). A comparison of model choices for the Continual Reassessment Method in phase I cancer trials. Statistics in Medecine 28, 3012-3028. <https://doi.org/10.1002/sim.3682>

Chamorey Emmanuel. (2009). Methodologie des essais de phase precoce en cancerologie: evolution des schemas et apport de la pharmacologie. These.

Garret-Mayer Elizabeth. (2006). The Continual Reassessment Method for dose-finding studies: a tutorial. Clinical Trials: 57-71. <https://doi.org/10.1191/1740774506cn134oa>

See Also

simCrm, ssimCrm

Examples

data<- CreData(5)
data<- updata(data,1,3,0)
data<- updata(data,2,3,1)
data<- updata(data,2,3,1)
data
Crm(data,prior=c(0.1,0.15,0.25,0.35,0.45),target=0.3,nextlevel="ntarget",nptmax=24,nmaxmtd=6)
data<- updata(data,3,3,2)
data
Crm(data,prior=c(0.1,0.15,0.25,0.35,0.45),target=0.3,nextlevel="ntarget",nptmax=24,nmaxmtd=6)