Search

Genomic tools are demonstrating that many human diseases are molecularly heterogeneous and likely to respond differently to molecularly targeted therapeutics. For many widely used treatments, the number of patients needed to treat (NNT) for each patient who benefits is large indicating that many patients are being exposed to the risks of serious adverse effects although they do not benefit from the drug. Consequently, more accurately determining the intended use population for new therapeutics is of increased importance. In this paper, we describe a new paradigm for identifying and internally validating an estimate of the intended use population in randomized phase III clinical trials. The approach preserves the type I error of the trial and approaches determination of the intended use population as a classification problem, not a multiple hypothesis testing problems.

Cancer Immunotherapetics differ fundamentally from most cancer therapies in that they involve manipulation of the immune system to induce an anti-tumor response. This fundamental difference results in differences in the pre-clinical and clinical development of immunotherapeutics. Even the regulatory culture of developing one drug at a time and demonstrating that it, when added to standard therapy, prolongs patient survival, is often not suitable for the development of effective immunotherapy regimens. In this commentary, we explore some of these differences and describe novel clinical trial designs which may be useful in immunotherapeutics regimen development.

This article examines the role of stratification of treatment assignment with regard to biomarker value in clinical trials that accept biomarker-positive and -negative patients but have a primary objective of evaluating treatment effect separately for the marker-positive subset. It also examines the issue of incomplete ascertainment of biomarker value and how this affects inference about treatment effect for the biomarker-positive subset of patients. I find that stratifying the randomization for the biomarker ensures that all patients will have tissue collected but is not necessary for the validity of inference for the biomarker-positive subset if there is complete ascertainment. If there is not complete ascertainment of biomarker values, it is important to establish that ascertainment is independent of treatment assignment. Having a large proportion of cases with biomarker ascertainment is not necessary for establishing internal validity of the treatment evaluation in biomarker-positive patients; independence of ascertainment and treatment is the important factor. Having a large proportion of cases with biomarker ascertainment makes it more likely that biomarker-positive patients with ascertainment are representative of the biomarker-positive patients in the clinical trial (with and without ascertainment), but since the patients in the clinical trial are a convenience sample of the population of patients potentially eligible for the trial, requiring a large proportion of cases with ascertainment does not facilitate generalizability of conclusions.

In a two-treatment trial, a two-sided test is often used to reach a conclusion, Usually we are interested in doing a two-sided test because of no prior preference between the two treatments and we want a three-decision framework. When a standard control is just as good as the new experimental treatment (which has the same toxicity and cost), then we will accept both treatments. Only when the standard control is clearly worse or better than the new experimental treatment, then we choose only one treatment. In this paper, we extend the concept of a two-sided test to the multiple treatment trial where three or more treatments are involved. The procedure turns out to be a subset selection procedure; however, the theoretical framework and performance requirement are different from the existing subset selection procedures. Two procedures (exclusion or inclusion) are developed here for the case of normal data with equal known variance. If the sample size is large, they can be applied with unknown variance and wi...

Dunnett's many-one test to compare several experimental treatment means with a control was extended in our previous paper to a decision procedure for selecting a desirable treatment when there is a strictly monotonic prior preference among the multiple treatments due to other consideration such as toxicity or cost. In this paper, we consider the case where two treatments could have equal prior preference. This formulation includes the problem of comparing a combination with its two components, or the comparison of four arms in a 2 by 2 factorial experiment. Our decision procedures will guarantee a high selection probability for the correct treatment(s) when that selection is appropriate. We establish sample size requirements for our decision procedures which can be applied to clinical trials with normal, binomial, or right censored exponential endpoints.

Test-based variable selection algorithms in regression often are based on sequential comparison of test statistics to cutoff values. A predetermined a level typically is used to determine the cutoffs based on an assumed probability distribution for the test statistic. For example, backward elimination or forward stepwise involve comparisons of test statistics to prespecified t or F cutoffs in Gaussian linear regression, while a likelihood ratio. Wald, or score statistic, is typically used with standard normal or chi square cutoffs in nonlinear settings. Although such algorithms enjoy widespread use, their statistical properties are not well understood, either theoretically or empirically. Two inherent problems with these methods are that (1) as in classical hypothesis testing, the value of α is arbitrary, while (2) unlike hypothesis testing, there is no simple analog of type I error rate corresponding to application of the entire algorithm to a data set. In this article we propose a new method, b...

A loss function approach is used to define the concepts of explained residual variation and explained risk for general regression models. Explained risk measures the ability of the covariates in a correctly specified model to distinguish differing outcomes. Explained residual variation, which is R 2 for a linear model, estimates the explained risk with a penalty for poorly fitting models. Application of the general definitions to linear regression, logistic regression, and survival analysis is given. The importance of distinguishing the concepts of explained residual variation, explained risk, and goodness of fit is discussed.

This study investigated the effects of manipulating the response requirement to the second stimulus (S2) on reaction time (RT) to the first stimulus (S1) in a double-stimulation choice RT task. Forty subjects responded to the 100 msec presentation of a left or right light by pressing the key on the same or opposite side as the light. Treatment conditions included a single-stimulation control (no S2 presented), and two double-stimulation conditions each requiring two responses (R1 and R2) in close succession, in one of these latter conditions, the rule governing R2 was the same as that governing R1 while, in the other, the rule governing R2 changed. Results showed the typical double-stimulation effect; i.e., increased latency of R1 when it was followed by S2 - R2. More importantly, R1 latency was increased further when the rules governing R1 and R2 were different. Results are discussed in terms of divided preparation capacity as well as other theories of the psychological refractory period.

We consider a family of single step allocation methods for the reduction of number of patients given the poorer of two treatments in a sequential comparative clinical trial. The object of the trial is to choose the better treatment with a probability ≧ P∗, where P∗ is assigned, when the difference in success probabilities is greater than or equal to an assigned δ∗. We use the formulation of the trial as a symmetrical selection procedure, Bechhofer, Kiefer, Sobel, 1968, Sobel, Weiss, 1970. If the stopping rule is the difference in successes then either alternating allocation or play the winner allocation appears to be optimal (Robbins, 1956, Sobel, Weiss, 1970.) When the stopping rule takes failures into account alternating allocation is not optimal, and other sirategies become important

A method is presented for the sequential analysis of experiments involving two treatments to which response is dichotomous. Composite hypotheses about the difference in success probabilities are tested, and covariate information is utilized in the analysis. The method is based upon a generalization of Bartlett’s (1946) procedure for using the maximum likelihood estimate of a nuisance parameter in a Sequential Probability Ratio Test (SPRT). Treatment assignment rules studied include pure randomization, randomized blocks, and an adaptive rule which tends to assign the superior treatment to the majority of subjects. It is shown that the use of covariate information can result in important reductions in the expected sample size for specified error probabilities, and that the use of covariate information is essential for the elimination of bias when adaptive assignment rules are employed. Designs of the type presented are easily generated, as the termination criterion is the same as for a Wald SPRT of simple h...

In data monitoring of long-term clinical trials one frequently faces the question of whether the trial should be terminated early either to reject or "accept" the null hypothesis (H0). To help answer this question, this paper suggests that, in the context of a particular proposed analysis at the end of the trial, one might stop early and reject (accept) H0 when the Conditional probability under H0 (the alternative hypothesis, Ha)

In two experiments, subjects pressed a key labeled Red or Green in response to a 100 msec duration stimulus presented to the left or right visual field. In Experiment I, subjects responded to the meaning of Stroop words; the stimulus was the word Red or Green printed in red, green, or white ink. In Experiment II, subjects responded to ink color; the stimulus was either the word Red or Green printed in red or green ink or a red or green color patch. In each experiment, there were 20 strongly right-handed subjects and 20 strongly left-handed subjects. Half the subjects in each handedness group were male and half were female. In both experiments, RT was faster when words were presented to the right visual field than to the left visual field, suggesting that both meaning and ink color of Stroop words were processed more quickly in the left hemisphere. Results of both experiments revealed faster reactions when meaning and ink color of the Stroop words were congruent than when they were not. A comparison with baseline trials indicated that the RT difference between responses to congruent and incongruent Stroop words was due to the incongruent cue interfering with information processing rather than to the congruent cue facilitating processing. Hypothesized interactions between stimulus position, congruence, handedness and sex were not significant.

Summary It was hypothesized that the use of self-reinforcement (SR) would vary as a function of the context in which it was administered. A recognition memory task was administered to 48 college students. They were instructed to administer positive self-reinforcement (SR+) and/or self-punishment (SR-) for their performance under one of four sets of instructions. Total SR administered and rates of SR+ and SR- all varied as predicted.

This study was concerned with the effect of varying the angular orientation of a display on the choice reaction times of two age groups ; a younger group ranging in age from 20 to 30 and an older group aged from 65 to 86. Two stimulus lights mounted on a vertical circular panel were rotated counterclockwise so that the lights formed angles of 0°, 45°, 90°, 135°, and 180° with the horizontal. Changing the spatial relationship of the stimulus lights relative to the fixed position of two response keys increased the complexity of the task and provided a means for testing the hypothesis that age differences in reaction time increase with increasing task difficulty. Results indicated significant differences in reaction time as a function of both display angle and age. With the least compatible display (180°), reaction time was slowed 30%. The role of spatial cues in the process of translating display information into control action was discussed. The predicted interaction of task difficulty and age was not sign...

An experiment is reported which was concerned with the influence of ageing and tins effects of precision of manipulation on the duration of movements. Two different age groups were compared on a task in which two dials had to be sot alternately. The precision required to adjust the dials was systematically varied. A marked increase with ago was noted in the time taken to adjust the dials, but a much smaller increase was observed in the time spent moving the hand from one dial to the other. The time taken to travel between the dials depended on the precision of both the preceding and the following manipulations.

This study investigated the effects of age, stimulus duration, and task difficulty on reaction time and movement time. Subjects made a simple guided movement in response to the onset of a stimulus light. Two levels of task difficulty (simple and two-choice reaction) and three durations of the light (50 msec, response terminated, and movement terminated) were used to define six experimental conditions. Forty-eight subjects were used : 24 college sophomores with a median age of 20. and 24 elderly persons with a median age of 75. Older subjects reacted 11 per cent slower and moved 38 per cent slower than young. Choice reactions wore slower than simple reactions. Old subjects, but not young, moved faster in simple trials than in choice. The effect of stimulus duration was not significant. A predicted interaction of stimulus duration and age was also not supported. Proportionately greater slowing of reaction time for old subjects as difficulty increased was observed but failed to reach an acceptable l...