scispace - formally typeset

Journal ArticleDOI

A Bayesian dose-finding design for drug combination clinical trials based on the logistic model.

01 Jul 2014-Pharmaceutical Statistics (John Wiley & Sons, Ltd)-Vol. 13, Iss: 4, pp 247-257

TL;DR: Under the proposed design, the posterior estimates of the model parameters continuously update to make the decisions of dose assignment and early stopping, and the design is competitive and outperforms some existing designs.

AbstractIn early phase dose-finding cancer studies, the objective is to determine the maximum tolerated dose, defined as the highest dose with an acceptable dose-limiting toxicity rate. Finding this dose for drug-combination trials is complicated because of drug–drug interactions, and many trial designs have been proposed to address this issue. These designs rely on complicated statistical models that typically are not familiar to clinicians, and are rarely used in practice. The aim of this paper is to propose a Bayesian dose-finding design for drug combination trials based on standard logistic regression. Under the proposed design, we continuously update the posterior estimates of the model parameters to make the decisions of dose assignment and early stopping. Simulation studies show that the proposed design is competitive and outperforms some existing designs. We also extend our design to handle delayed toxicities. Copyright © 2014 John Wiley & Sons, Ltd.

Topics: Early stopping (50%)

Summary (3 min read)

1 Introduction

  • For oncologists, the objective of phase I dose-finding studies is to determine the maximum tolerated dose (MTD), defined as the highest dose with a relatively acceptable dose-limiting toxicity (DLT) [1, 2].
  • Mandrekar et al. proposed an approach incorporating the toxicity and efficacy of each agent into the identification of an optimal dosing region for the combination by using a continuation ratio model to separate each agent’s toxicity and efficacy curves [8, 9].
  • Most of these existing designs rely on complicated statistical models that typically are not familiar to clinicians, which hinder their acceptance and application in practice.
  • In addition, the performance of these designs seems comparable and there is no consensus which design should be used [13].

Dose-combination model

  • Let there be a two-drug combination used in a phase I dose-finding clinical trial for which the dose-toxicity relationship is monotonic and increases with the dose levels.
  • Before two agents can be combined, each of them typically have been thoroughly investigated individually.
  • Therefore, there is often rich prior information on pj’s and qk’s and their values can be readily elicited from physicians.
  • Using the prior information to define standardized dose has been widely used in dose-finding trial designs, and the most well known example perhaps is the skeleton of the continuous reassessment method (CRM) [21] with a logistic model.
  • Research has shown that this approach improves the estimation stability and trial performance [2, 22].

Likelihood and posterior inference

  • Under the proposed model, the likelihood is simply a product of the Bernoulli density, given by L(β0, β1, β2, β3|data) ∝ (2) We sample this posterior distribution using Gibbs sampler, which sequentially draws each of the parameters from their full conditional distributions .the authors.the authors.
  • Dose finding algorithm and determination of the MTD During the trial conduct, the authors use the dose-finding algorithm proposed by Yin and Yuan [11, 10] to determine dose escalation and deescaltion, and propose a different criterion for MTD selection at the end of the trial.
  • Shown in dark gray is the AUC for a toxicity probability greater than 0.4, which is equal to the probability of overdosing.

BCOPULA and BGUMBEL methods

  • Yin and Yuan proposed two Bayesian methods that use copula regression for combinations.
  • The parameter γ characterizes the drug interactive effect, and α and β the uncertainty of the initial guesses.
  • The combination allocation algorithm is the same as that presented in Section 2.1.
  • The final MTD is the combination with a toxicity probability closest to the target among the combinations already administered in the trial.

LOGODDS

  • By backsolving this equation, an explicit expression for the probability of toxicity is obtained.
  • A normal prior centered in 0 and with variance of 100 was chosen for the interaction parameter.
  • The rest of the dose allocation process, estimation and MTD determination was the same as their proposed design in order to compare their method involving a simple interaction logistic model with other logistic models.

One-dimensional CRM

  • In practice, one-dimensional CRM sometimes is used to conduct dose-combination trials [13].
  • Under this method, the authors first preselect a subset of combinations, for which the toxicity probability order is known, and then apply the standard CRM to find the MTD.
  • The drawback of such an approach is that the authors only investigate a subset of the whole two-dimensional dose space and may miss the true MTD.

3 Simulations

  • The authors simulated 2000 independent replications of phase I trials that evaluate two agents in drug combinations, with five dose levels for Agent 1 and three for Agent 2, giving 15 possible combinations.
  • The authors fixed the toxicity target at 0.3 and used an overall sample size of 60.
  • The design parameters of all the designs (e.g., working model) have been calibrated via simulation before used for the comparison.
  • The authors selected these features in order to employ typical trial set-ups in their simulation study.
  • The authors set the length around the targeted interval, δ, at 0.1.

4 Results

  • For each scenario, the authors present the correct MTD selection rate, or percentages of correct selection (PCS), in Table 2.
  • For these scenarios, all model-based designs gave high PCS, whereas LOGISTIC seemed to perform better than the other methods.
  • The addition of the stopping rule for unacceptable toxicity resulted in PCS that were similar to those presented above (Table 5), except in scenario 4 in which the first dose combination, (1, 1), was the MTD.

4.1 Sensitivity analysis

  • The authors conducted a sensitivity analysis in order to study the performance of their design using different prior distributions and parameters values.
  • According to Table 5, the authors can see that the PCS for all scenarios were very similar under these different prior distributions.

4.2 Time-to-event outcome

  • In practice, a longer follow-up time may be required to assess the toxicity outcome.
  • Before combination assignment, the likelihood is defined as L(β0, β1, β2, β3|data) =.
  • The authors simulated the time-to-toxicity outcomes using an exponential distribution such that the toxicity probabilities at the end of follow-up matched those given in Table 1.
  • Table 6 shows the results of the extended LOGISTIC for all 15 scenarios.
  • The authors observe that the performance of the design decreases only slightly by 2%, and the PCS for all scenarios are still very high.

5 Discussion

  • The authors have proposed a statistical method for clinical trial designs that evaluate various dose combinations for two agents.
  • One benefit of their method compared with the other proposed designs is that it is also efficient when the MTDs are not necessarily located on the same diagonal.
  • When combining several agents, designs developed for single-agent dose-finding trials cannot be applied to combination studies.
  • This approach performs well if the target dose combinations happen to be included in the subset.
  • These files are freely available upon request.

Did you find this useful? Give us your feedback

...read more

Content maybe subject to copyright    Report

HAL Id: hal-01298657
https://hal.sorbonne-universite.fr/hal-01298657
Submitted on 6 Apr 2016
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
A Bayesian dose-nding design for drug combination
clinical trials based on the logistic model
Marie-Karelle Riviere, Ying Yuan, Frédéric Dubois, Sarah Zohar
To cite this version:
Marie-Karelle Riviere, Ying Yuan, Frédéric Dubois, Sarah Zohar. A Bayesian dose-nding design for
drug combination clinical trials based on the logistic model. Pharmaceutical Statistics, Wiley, 2014,
13 (4), pp.247-257. �10.1002/pst.1621�. �hal-01298657�

A Bayesian dose-finding design for drug combination
clinical trials based on the logistic model
Marie-Karelle Riviere
(a,b)
, Ying Yuan
(c)
, Fr
´
ed
´
eric Dubois
(b)
and Sarah Zohar
(a)
(a) INSERM, UMRS 1138, Equipe 22, Centre de Recherche des Cordeliers, Universit
´
e Paris 5, Universit
´
e Paris 6, Paris, France
(b) IRIS (Institut de Recherches Internationales Servier), Suresnes, France
(c) Department of Biostatistics, The University of Texas MD Anderson Cancer Center, Houston, TX 77030, U.S.A.
Abstract
In early phase dose-finding cancer studies, the objective is to determine the maximum tolerated
dose, defined as the highest dose with an acceptable dose-limiting toxicity rate. Finding this dose
for drug-combination trials is complicated due to drug-drug interactions, and many trial designs
have been proposed to address this issue. These designs rely on complicated statistical models that
typically are not familiar to clinicians, and are rarely used in practice. The aim of this paper is
to propose a Bayesian dose-finding design for drug combination trials based on standard logistic
regression. Under the proposed design, we continuously update the posterior estimates of the model
parameters to make the decisions of dose assignment and early stopping. Simulation studies show
that the proposed design is competitive and outperforms some existing designs. We also extend our
design to handle delayed toxicities.
Keywords: Bayesian inference; Dose-finding; Drug combination; Oncology; Phase I trial.
1

1 Introduction
For oncologists, the objective of phase I dose-finding studies is to determine the maximum tolerated
dose (MTD), defined as the highest dose with a relatively acceptable dose-limiting toxicity (DLT)
[1, 2]. DLT is usually defined as a toxicity of grade 3 or higher according to the U.S. National
Cancer Institute toxicity criteria [3]. In practice, patients included in phase I clinical cancer trials
have already been heavily pre-treated and in many cases no alternative therapeutic options are
available to them. For cytotoxic anti-cancer drugs, a dose-toxicity effect is assumed whereby the
higher the dose, the greater the risk of DLT and the greater the efficacy.
Dose finding for drug combination trials is more difficult than that for conventional single-
agent trials due to complicated drug-drug interactions. Moreover, when combining several agents,
the order of the toxicity probabilities is not fully known. Should investigators wish to gradually
increase the acceptable level of toxicity during the trial, the appropriate order in which the doses for
the various drugs in the combination should be increased would be of great interest. For instance,
when combining two cytotoxic agents, it remains difficult to decide how to escalate or de-escalate
the dose combination, even when a partial ordering is known [4, 5].
Recently, many phase I dose-finding designs have been proposed for drug combination trials.
Thall et al. proposed a Bayesian dose-finding method based on a six-parameter model [6]. Wang
and Ivanova developed a 3-parameter model-based method in which the parameters are estimated
using Bayesian inference [7]. Mandrekar et al. proposed an approach incorporating the toxicity and
efficacy of each agent into the identification of an optimal dosing region for the combination by us-
ing a continuation ratio model to separate each agent’s toxicity and efficacy curves [8, 9]. Yin and
Yuan developed a Bayesian adaptive design based on latent 2x2 tables in which the combination’s
toxicity probabilities in the two-dimensional space are estimated using a Gumbel-type model [10].
Yin and Yuan extended their method by changing to a copula-type model to simulate the effect of
two or more drugs in combination [11]. Bailey et al. introduced a second agent as a covariate in
a logistic model [12]. Wages et al. considered an approach based on the continual reassessment
method and taking into account different orderings with partial order between combinations. In this
case, the MTD is estimated for the order associated with the highest model-selection criterion [5].
Most of these existing designs rely on complicated statistical models that typically are not familiar
to clinicians, which hinder their acceptance and application in practice. In addition, the perfor-
mance of these designs seems comparable and there is no consensus which design should be used
[13]. As a result, many of the dose-finding clinical trials conducted to evaluate drug combinations
still use the conventional “3+3” approach, which was developed for single agents and was shown
to be inefficient in terms of dose identification [14, 15, 16, 17, 18].
In this paper, we propose an Bayesian dose-finding design for drug combination trials based
on standard logistic regression. Under the Bayesian paradigm, data monitoring, early stopping
and dose assignment occur continuously throughout the trial by updating the posterior estimates of
the model parameters. Simulation studies show that in general the proposed design provide better
performance than some existing designs. To facilitate the use of the proposed design by clinicians,
R package will be developed for implementing the new design.
This manuscript is laid out as follows: in Section 2, we propose the simple statistical method
for modeling the toxicity probabilities of the drug combination under evaluation. Moreover, we
present the likelihood function and the prior specifications for the unknown parameters. In this
section, we also describe our allocation and dose-finding method, as well as propose stopping rules.
2

We conduct extensive simulation studies to examine the operating characteristics of our design in
Sections 3 and 4, and conclude with a discussion in Section 5.
2 Methods
2.1 Statistical method for combination evaluation (LOGISTIC)
Dose-combination model
Let there be a two-drug combination used in a phase I dose-finding clinical trial for which the
dose-toxicity relationship is monotonic and increases with the dose levels. Let (j, k) denote the
dose level of a combination in which j refers to Agent 1 (j = 1, . . . , J), and k refers to Agent 2
(k = 1, . . . , K). Y
i
is a Bernoulli random variable, denoting the toxicity that is equal to 1 if DLT
occurs in patient i and 0 otherwise (i = 1, . . . , N). We assume that n
j,k
patients are allocated at
combination (j, k) and that a total of t
j,k
DLTs are observed for that combination. We define π
j,k
as
the toxicity probabilities of combination (j, k), θ as the target probability of toxicity, and p
1
, ..., p
J
and q
1
, ..., q
K
as the respective prior toxicity probabilities of Agent 1 and Agent 2 taken alone. For
simplicity purposes, we refer to the selected combination as the MTD in order to maintain the same
designation as in single-agent trials. In this manuscript, we focus on finding one MTD at the end
of the trial. We note that in some cases, it is of interest to find multiple MTDs that can be further
tested in phase II trials, see Yuan and Yin [19] and Ivanova and Wang [20] for related designs.
Let u
j
and v
k
denote the “effective” or standardized doses ascribed to the jth level of Agent 1
and k level of Agent 2, respectively. We model the drug combination-toxicity relationship using a
4-parameter logistic model, as follows:
logit(π
j,k
) = β
0
+ β
1
u
j
+ β
2
v
k
+ β
3
u
j
v
k
, (1)
where β
0
, β
1
, β
2
and β
3
are unknown parameters that represent the toxicity effect of Agent 1 (β
1
),
that of Agent 2 (β
2
) and that of the interaction between the two agents (β
3
). These parameters are
defined such that β
1
> 0 and β
2
> 0, ensuring that the toxicity probability is increasing with the
increasing dose level of each agent alone, k, β
1
+ β
3
v
k
> 0 and j, β
2
+ β
3
u
j
> 0, ensuring that
the toxicity probability is increasing with the increasing dose levels of both agents together, and
intercept −∞ < β
0
< . The standardized dose of two agents are defined as u
j
= log
(
p
j
1p
j
)
and v
k
= log
(
q
k
1q
k
)
, where p
j
and q
k
are the prior estimates of the toxicity probabilities of the jth
dose level of Agent 1 and the kth dose level of Agent 2, respectively, when they are administered
individually as a single agent. Before two agents can be combined, each of them typically have
been thoroughly investigated individually. Therefore, there is often rich prior information on p
j
s
and q
k
s and their values can be readily elicited from physicians. Using the prior information to
define standardized dose has been widely used in dose-finding trial designs, and the most well
known example perhaps is the skeleton of the continuous reassessment method (CRM) [21] with a
logistic model. Research has shown that this approach improves the estimation stability and trial
performance [2, 22]. Our definition of u
j
and v
k
can be viewed as an extension of the skeleton of
the CRM (with a logistic model) to drug-combination trials.
3

Likelihood and posterior inference
Under the proposed model, the likelihood is simply a product of the Bernoulli density, given by
L(β
0
, β
1
, β
2
, β
3
|data)
J
j=1
K
k=1
π
t
j,k
j,k
(1 π
j,k
)
n
j,k
t
j,k
.
We assume the prior distributions of the model parameters are independent. For β
0
and β
3
, we
assign a vague normal prior N(0, 10) centered at 0 to indicate that a priori we do not favor either
positive or negative values for these parameters and let the observed data speak for themselves
through posteriors. For β
1
and β
2
, we assume an informative prior distribution Exp(1) centered
at 1, as those parameters should not be too far from 1. Then, the joint posterior distribution of
parameters β
0
, β
1
, β
2
and β
3
is given by
f(β
0
, β
1
, β
2
, β
3
|data) L(β
0
, β
1
, β
2
, β
3
|data)f(β
0
)f(β
1
)f(β
2
)f(β
3
). (2)
We sample this posterior distribution using Gibbs sampler, which sequentially draws each of the
parameters from their full conditional distributions (see Appendix). These full conditional dis-
tributions do not have closed forms, and we use the Adaptive Rejection Metropolis Sampling
(ARMS) method [23] to sample them. In order to impose the constrain that k, β
1
+ β
3
v
k
> 0
and j, β
2
+ β
3
u
j
> 0, at each iteration of Gibbs sampling, if sampled β
1
, β
2
and β
3
fail to satisfy
the constraint, we re-sample β
1
, β
2
and β
3
. Let (β
()
0
, β
()
1
, β
()
2
, β
()
3
)
=1,...,L
denote the L posterior
samples obtained from Gibbs sampler, the posterior toxicity probabilities can be estimated using
Monte Carlo by:
˜π
j,k
=
1
L
L
=1
exp(β
()
0
+ β
()
1
u
j
+ β
()
2
v
k
+ β
()
3
u
j
v
k
)
1 + exp(β
()
0
+ β
()
1
u
j
+ β
()
2
v
k
+ β
()
3
u
j
v
k
)
.
Dose finding algorithm and determination of the MTD
During the trial conduct, we use the dose-finding algorithm proposed by Yin and Yuan [11, 10] to
determine dose escalation and deescaltion, and propose a different criterion for MTD selection at
the end of the trial. In our design, similar to Yin and Yuan [11], we restrict dose escalation and de-
escalation one level at a time (i.e, we do not allow dose escalate or de-escalate along the diagonal)
based on the practical consideration that physicians are conservative and typically do not allow
two agent to escalate at the same time for patient safety. Nevertheless, we note that Sweeting and
Mander [24] showed that diagonal escalation strategy may be more efficient in reaching the target
toxicity level quicker with fewer patients treated at sub-optimal doses and have a higher percentage
of correct selection at the end of the trial. We take this strategy in our start-up phase described later.
Let c
e
be the probability threshold for dose escalation and c
d
the probability threshold for dose
de-escalation. We require c
e
+c
d
> 1 to avoid that the decisions of dose escalation and deescalation
occur at the same time. Our dose-finding algorithm can be described as follows:
If the current combination is (j, k) and P (π
j,k
< θ|data) > c
e
,
4

Citations
More filters

Journal ArticleDOI
TL;DR: GSK458 plus trametinib is poorly tolerated, due to skin and GI-related toxicities, which may be due to overlapping toxicities precluding sufficient dose exposure.
Abstract: Introduction This Phase Ib trial investigated the safety, tolerability, and recommended phase 2 dose for the pan-PI3K/mTOR inhibitor, GSK2126458 (GSK458), and trametinib combination when administered to patients with advanced solid tumors. Patients and Methods Patients with advanced solid tumors received escalating doses of GSK458 (once or twice daily, and continuous or intermittent) and trametinib following a zone-based 3 + 3 design to determine the maximum tolerated dose (MTD). Assessments included monitoring for adverse events and response, and evaluating pharmacokinetic (PK) measures. Archival tissue and circulating free DNA samples were collected to assess biomarkers of response in the PI3K and RAS pathways. Results 57 patients were enrolled onto the continuous dosing cohort and 12 patients onto an intermittent BID dosing cohort. Two MTDs were established for the continuous daily dosing: 2 mg of GSK458 with 1.0 mg of trametinib or 1.0 mg of GSK458 with 1.5 mg of trametinib; no MTD was determined in the intermittent dosing cohort. The most frequent adverse events were rash (74 %) and diarrhea (61 %). Dose interruptions due to adverse events occurred in 42 % of patients. No significant PK interaction was observed. One patient achieved partial response and 12 patients had stable disease >16 weeks. Mutations in RAS/RAF/PI3K were detected in 70 % of patients, but no pattern emerged between response and mutational status. Conclusion GSK458 plus trametinib is poorly tolerated, due to skin and GI-related toxicities. Responses were minimal, despite enrichment for PI3K/RAS pathway driven tumors, which may be due to overlapping toxicities precluding sufficient dose exposure.

48 citations


Journal ArticleDOI
TL;DR: This book presents a series of intellectual discussions concerning some of the most basic topics in statistics, including interpreting probability and statistical models of induction, including the concepts of inference, p values, and the comparison of evidential theories.
Abstract: This book presents a series of intellectual discussions concerning some of the most basic topics in statistics, including interpreting probability and statistical models of induction, including the concepts of inference, p values, and the comparison of evidential theories. The book is a well-researched and thoroughly professional approach to the various dilemmas faced by statisticians in interpreting data. The author’s knowledge of the literature as it pertains to the various issues discussed in the book is excellent and reminds me of much of the literature I read and studied during my lengthy and varied education in statistics and the practice of statistical science. Furthermore, he provides for teachers of statistics a lengthy list of quotes from many great statisticians on subjects of interest to students of statistics, including Deming, Shewhart, and others, all the way to deFinetti and other Bayesians. The wealth of knowledge present in this 152-page book is remarkable; the book is a worthy of addition to any library of statistical literature. Important questions as to why this book was written, what is its purpose and how does a practicing statistician use the knowledge therein (e.g., “what is the purpose of the nature of statistical evidence?”) is left to the reader to answer for himself or herself. The underlying theme of the book is never stated by the author in a simple, clear-cut manner. Does one use the knowledge of the text to make one a better statistician or a better teacher of statistics, or simply to understand the logic of statistical science in a way that is meaningful to all? As I read the book intensively, I asked myself whether this book is good for teaching others or is simply an intellectual exercise for some. What does the author have in mind when he discusses such common and useful concepts as p values or true value? Does he contribute to the understanding of these concepts, or does he make them more difficult to comprehend? Furthermore, he translates at least one German word, iterationen, incorrectly. The German ending-en indicates that the word is plural and is translated as iterations or runs. This simple error makes one think that there are many places in the text, especially some of the mathematics, which also may be in error. Early in the book, the author attempts to determine some definition of statistical evidence (last paragraph on p. 3). But after reading the paragraph and the extensive discussion of the entire text, one cannot be sure what one has read. When teaching statistics at the college level, I try to keep the ideas as simple as possible without relying on an overwhelming exercise in sophistication as the goal. The author seems to define concepts, redefine concepts, and then define the same concepts yet again, with the end result being imperfect and probably misunderstood. Much of the ideas expressed and discussed in the book appear in a manner that is exasperating to comprehend. As an analogy, reading this book is like entering a room to learn something new but in the end leaving the room with no more knowledge and understanding than when one entered. There is simply too much confusion resulting from lengthy discussions of relatively simple concepts. Because the book is part of a series titled “Lecture Notes in Statistics,” I can only deduce that it is nothing more than a series of lecture notes. There is no flow to the book and little thought given to presenting the details in a style that would induce one to want to read it. When I was young, I would hear that many statisticians were very wise and intelligent persons who should be listened to and understood but presented their knowledge in a direct and understandable way. I fear that this book is full of confusing discussions of very important issues. I would like to see this material rewritten in a manner that we as readers would not have to work so hard to understand the aims of the author. Although the author produces and quotes the works of many great thinkers, he does not bring out their points in a style that is easy for the reader to follow. Often, he places difficult mathematical theorems and proofs in the midst of lengthy paragraphs not needing such mathematical sophistication. When I started reading this book, I expected a thoughtful text introducing the reader to the origins of statistics as a science and methodology. The book did not do this, nor did it present the numerous wonderful applications of statistics to the collection, analysis, and interpretation of data that illuminate our knowledge of the world. Where are the examples in quality control, experimentation, and the like that the author eludes to and how does statistics provide the foundation for us to understand our world? When I finished the book, I felt that the author had failed to communicate to the reader why statistics is so important; instead, he clarifies less than he confuses.

43 citations


Journal ArticleDOI
TL;DR: The waterfall design is safer and has higher probability of identifying the true MTD contour than some existing designs, and is extended to a phase I/II design with the goal of finding the MTD with the highest efficacy.
Abstract: Drug combination therapy has become the mainstream approach to cancer treatment. One fundamental feature that makes combination trials different from single-agent trials is the existence of the maximum tolerated dose (MTD) contour, that is, multiple MTDs. As a result, unlike single-agent phase I trials, which aim to find a single MTD, it is often of interest to find the MTD contour for combination trials. We propose a new dose-finding design, the waterfall design, to find the MTD contour for drug combination trials. Taking the divide-and-conquer strategy, the waterfall design divides the task of finding the MTD contour into a sequence of one-dimensional dose-finding processes, known as subtrials. The subtrials are conducted sequentially in a certain order, such that the results of each subtrial will be used to inform the design of subsequent subtrials. Such information borrowing allows the waterfall design to explore the two-dimensional dose space efficiently using a limited sample size and decreases the chance of overdosing and underdosing patients. To accommodate the consideration that doses on the MTD contour may have very different efficacy or synergistic effects because of drug-drug interaction, we further extend our approach to a phase I/II design with the goal of finding the MTD with the highest efficacy. Simulation studies show that the waterfall design is safer and has higher probability of identifying the true MTD contour than some existing designs. The R package "BOIN" to implement the waterfall design is freely available from CRAN. Copyright © 2016 John Wiley & Sons, Ltd.

31 citations


Cites methods from "A Bayesian dose-finding design for ..."

  • ...[8] proposed a Bayesian dose-finding design based on the logistic model....

    [...]


Journal ArticleDOI
TL;DR: A parametric model is used to describe the relationship between the doses of the two agents and the probability of dose limiting toxicity, and the method is extended to accommodate discrete dose combinations and compared with previous approaches under several scenarios.
Abstract: We present a cancer phase I clinical trial design of a combination of two drugs with the goal of estimating the maximum tolerated dose curve in the two-dimensional Cartesian plane. A parametric model is used to describe the relationship between the doses of the two agents and the probability of dose limiting toxicity. The model is re-parameterized in terms of the probabilities of toxicities at dose combinations corresponding to the minimum and maximum doses available in the trial and the interaction parameter. Trial design proceeds using cohorts of two patients receiving doses according to univariate escalation with overdose control (EWOC), where at each stage of the trial, we seek a dose of one agent using the current posterior distribution of the MTD of this agent given the current dose of the other agent. The maximum tolerated dose curve is estimated as a function of Bayes estimates of the model parameters. Performance of the trial is studied by evaluating its design operating characteristics in terms of safety of the trial and percent of dose recommendation at dose combination neighborhoods around the true MTD curve and under model misspecifications for the true dose-toxicity relationship. The method is further extended to accommodate discrete dose combinations and compared with previous approaches under several scenarios. Copyright © 2016 John Wiley & Sons, Ltd.

26 citations


Journal ArticleDOI
TL;DR: It is found that performance of the model‐based dose‐finding methods varied depending on whether the dose combination matrix is square or not; whether the true MTDCs exist within the same group along the diagonals of the dosecombination matrix; and the number of trueMTDCs.
Abstract: Little is known about the relative performance of competing model-based dose-finding methods for combination phase I trials. In this study, we focused on five model-based dose-finding methods that have been recently developed. We compared the recommendation rates for true maximum-tolerated dose combinations (MTDCs) and over-dose combinations among these methods under 16 scenarios for 3 × 3, 4 × 4, 2 × 4, and 3 × 5 dose combination matrices. We found that performance of the model-based dose-finding methods varied depending on (1) whether the dose combination matrix is square or not; (2) whether the true MTDCs exist within the same group along the diagonals of the dose combination matrix; and (3) the number of true MTDCs. We discuss the details of the operating characteristics and the advantages and disadvantages of the five methods compared.

24 citations


Cites methods from "A Bayesian dose-finding design for ..."

  • ...[24] proposed a Bayesian dose-finding design based on the logistic model, while Mander and Sweeting [25] published a curve-free method that relies on the product of independent beta probabilities....

    [...]


References
More filters

Journal ArticleDOI
TL;DR: A new approach to the design and analysis of Phase 1 clinical trials in cancer and a particularly simple model is looked at that enables the use of models whose only requirements are that locally they reasonably well approximate the true probability of toxic response.
Abstract: This paper looks at a new approach to the design and analysis of Phase 1 clinical trials in cancer. The basic idea and motivation behind the approach stem from an attempt to reconcile the needs of dose-finding experimentation with the ethical demands of established medical practice. It is argued that for these trials the particular shape of the dose toxicity curve is of little interest. Attention focuses rather on identifying a dose with a given targeted toxicity level and on concentrating experimentation at that which all current available evidence indicates to be the best estimate of this level. Such an approach not only makes an explicit attempt to meet ethical requirements but also enables the use of models whose only requirements are that locally (i.e., around the dose corresponding to the targeted toxicity level) they reasonably well approximate the true probability of toxic response. Although a large number of models could be contemplated, we look at a particularly simple one. Extensive simulations show the model to have real promise.

1,277 citations


"A Bayesian dose-finding design for ..." refers methods in this paper

  • ...In practice, one-dimensional CRM sometimes is used to conduct dose-combination trials [13]....

    [...]

  • ...We also compared the performance of this multidimensional designs with a one-dimensional CRM, CRM anti-diag, as described previously....

    [...]

  • ...As the CRM focused on only a subset of doses, it had more patients per dose to find the target doses than the multidimensional designs, given the same total number of patients....

    [...]

  • ...Under this method, we first preselect a subset of combinations, for which the toxicity probability order is known, and then apply the standard CRM to find the MTD....

    [...]

  • ...Using the prior information to define standardized dose has been widely used in dose-finding trial designs, and the most well-known example perhaps is the skeleton of the continuous reassessment method (CRM) [21] with a logistic model....

    [...]


Journal ArticleDOI
TL;DR: In Monte Carlo simulations, two two-stage designs are found to provide reduced bias in maximum likelihood estimation of the MTD in less than ideal dose-response settings and several designs to be nearly as conservative as the standard design in terms of the proportion of patients entered at higher dose levels.
Abstract: The Phase I clinical trial is a study intended to estimate the so-called maximum tolerable dose (MTD) of a new drug. Although there exists more or less a standard type of design for such trials, its development has been largely ad hoc. As usually implemented, the trial design has no intrinsic property that provides a generally satisfactory basis for estimation of the MTD. In this paper, the standard design and several simple alternatives are compared with regard to the conservativeness of the design and with regard to point and interval estimation of an MTD (33rd percentile) with small sample sizes. Using a Markov chain representation, we found several designs to be nearly as conservative as the standard design in terms of the proportion of patients entered at higher dose levels. In Monte Carlo simulations, two two-stage designs are found to provide reduced bias in maximum likelihood estimation of the MTD in less than ideal dose-response settings. Of the three methods considered for determining confidence intervals--the delta method, a method based on Fieller's theorem, and a likelihood ratio method--none was able to provide both usefully narrow intervals and coverage probabilities close to nominal.

730 citations


"A Bayesian dose-finding design for ..." refers methods in this paper

  • ...Therefore, as suggested by other authors [10,16,20,26,27], we implement an algorithm-based start-up phase in order to gather enough information to estimate the j,k ....

    [...]


Journal ArticleDOI
TL;DR: A robust nonlinear full probability model for population pharmacokinetic data is proposed and it is demonstrated that the method enables Bayesian inference for this model, through an analysis of antibiotic administration in new‐born babies.
Abstract: Gibbs sampling is a powerful technique for statistical inference. It involves little more than sampling from full conditional distributions, which can be both complex and computationally expensive to evaluate. Gilks and Wild have shown that in practice full conditionals are often log‐concave, and they proposed a method of adaptive rejection sampling for efficiently sampling from univariate log‐concave distributions. In this paper, to deal with non‐log‐concave full conditional distributions, we generalize adaptive rejection sampling to include a Hastings‐Metropolis algorithm step. One important field of application in which statistical models may lead to non‐log‐concave full conditionals is population pharmacokinetics. Here, the relationship between drug dose and blood or plasma concentration in a group of patients typically is modelled by using nonlinear mixed effects models. Often, the data used for analysis are routinely collected hospital measurements, which tend to be noisy and irregular. Consequently, a robust (t‐distributed) error structure is appropriate to account for outlying observations and/or patients. We propose a robust nonlinear full probability model for population pharmacokinetic data. We demonstrate that our method enables Bayesian inference for this model, through an analysis of antibiotic administration in new‐born babies.

654 citations


"A Bayesian dose-finding design for ..." refers methods in this paper

  • ...These full conditional distributions do not have closed forms, and we use the adaptive rejection Metropolis sampling (ARMS) method [23] to sample them....

    [...]


Journal ArticleDOI
TL;DR: Accelerated titration (i.e., rapid intrapatient drug dose escalation) designs appear to effectively reduce the number of patients who are under-treated, speed the completion of phase I trials, and provide a substantial increase in the information obtained.
Abstract: Background: Many cancer patients in phase I clinical trials are treated at doses of chemotherapeutic agents that are below the biologically active level, thus reducing their chances for therapeutic benefit. Current phase I trials often take a long time to complete and provide little information about interpatient variability or cumulative toxicity. Purpose: Our objective was to develop alternative designs for phase I trials so that fewer patients are treated at subtherapeutic dose levels, trials are of reduced duration, and important information (i.e., cumulative toxicity and maximum tolerated dose) needed to plan phase II trials is obtained. Methods: We fit a stochastic model to data from 20 phase I trials involving the study of nine different drugs. We then simulated new data from the model with the parameters estimated from the actual trials and evaluated the performance of alternative phase I designs on this simulated data. Four designs were evaluated. Design 1 was a conventional design (similar to the commonly used modified Fibonacci method) using cohorts of three to six patients, with 40% dose-step increments and no intrapatient dose escalation. Designs 2 through 4 included only one patient per cohort until one patient experienced dose-limiting toxic effects or two patients experienced grade 2 toxic effects (during their first course of treatment for designs 2 and 3 or during any course of treatment for design 4). Designs 3 and 4 used 100% dose steps during this initial accelerated phase. After the initial accelerated phase, designs 2 through 4 resorted to standard cohorts of three to six patients, with 40% dose-step increments. Designs 2 through 4 used intrapatient dose escalation if the worst toxicity is grade 0-1 in the previous course for that patient. Results: Only three of the actual trials demonstrated cumulative toxic effects of the chemotherapeutic agents in patients. The average number of patients required for a phase I trial was reduced from 39.9 for design 1 to 24.4, 20.7, and 21.2 for designs 2, 3, and 4, respectively. The average number of patients who would be expected to have grade 0-1 toxicity as their worst toxicity over three cycles of treatment is 23.3 for design 1, but only 7.9, 3.9, and 4.8 for designs 2, 3, and 4, respectively. The average number of patients with grade 3 toxicity as their worst toxicity increases from 5.5 for design 1 to 6.2, 6.8, and 6.2 for designs 2, 3, and 4, respectively. The average number of patients with grade 4 toxicity as their worst toxicity increases from 1.9 for design 1 to 3.0, 4.3, and 3.2 for designs 2, 3, and 4, respectively. Conclusion: Accelerated titration (i.e., rapid intrapatient drug dose escalation) designs appear to effectively reduce the number of patients who are undertreated, speed the completion of phase I trials, and provide a substantial increase in the information obtained. [J Natl Cancer Inst 1997;89:1138-47]

560 citations


"A Bayesian dose-finding design for ..." refers methods in this paper

  • ...Our start-up phase shares the spirit of accelerated titration design [28] and can be described as follows: Treat the first cohort of patients at the lowest dose combination ....

    [...]


Journal ArticleDOI
TL;DR: Modifications to the Continual Reassessment Method (CRM) are presented, in which one assigns more than one subject at a time to each dose level, and each dose increase is limited to one level, which makes the CRM acceptable to clinical investigators.
Abstract: The Continual Reassessment Method (CRM) is a Bayesian phase I design whose purpose is to estimate the maximum tolerated dose of a drug that will be used in subsequent phase II and III studies. Its acceptance has been hindered by the greater duration of CRM designs compared to standard methods, as well as by concerns with excessive experimentation at high dosage levels, and with more frequent and severe toxicity. This paper presents the results of a simulation study in which one assigns more than one subject at a time to each dose level, and each dose increase is limited to one level. We show that these modifications address all of the most serious criticisms of the CRM, reducing the duration of the trial by 50-67 per cent, reducing toxicity incidence by 20-35 per cent, and lowering toxicity severity. These are achieved with minimal effects on accuracy. Most important, based on our experience at our institution, such modifications make the CRM acceptable to clinical investigators.

428 citations


"A Bayesian dose-finding design for ..." refers background or methods in this paper

  • ...For instance, the standard algorithm-based method for phase I dose-finding clinical trials in oncology is the so-called “3+3” design, which is referred to as “memory-less” since allocation to the next dose level for an incoming group of 3 patients depends only upon what has happened to the total of 3 to 6 patients previously treated at the current dose level [14, 15, 16, 17, 18]....

    [...]

  • ...As a result, many of the dose-finding clinical trials conducted to evaluate drug combinations still use the conventional “3+3” approach, which was developed for single agents and was shown to be inefficient in terms of dose identification [14, 15, 16, 17, 18]....

    [...]


Frequently Asked Questions (1)
Q1. What are the contributions mentioned in the paper "A bayesian dose-finding design for drug combination clinical trials based on the logistic model" ?

The aim of this paper is to propose a Bayesian dose-finding design for drug combination trials based on standard logistic regression.