Superiority of PCNU over AZQ in the treatment of primary brain tumors: results of a prospective randomized trial (81-20) by the Brain Tumor Study Group.

PURPOSE: A two-arm randomized clinical trial was performed to determine the efficacy of PCNU and AZQ in the treatment of de novo or recurrent primary brain tumors. An additional objective was to gather information on the administration and toxicity of these compounds, supplementing that obtained previously in phase I/II studies. METHODS: During 1982 and 1983 the Brain Tumor Study Group randomized 152 adult patients with primary brain tumors to receive PCNU 75-100 mg/m2 intravenously (IV) every 8 weeks or AZQ 15 mg/m2 IV once a week for 4 weeks, every 6-8 weeks. All patients who had not received 'full dose' radiotherapy before randomization received it concurrently with the first course of protocol chemotherapy. The data were analyzed for the total randomized population (RP), and for 130 patients in the valid study group (VSG) formed by excluding 22 patients for whom the histologic diagnosis was not documented by central review. RESULTS: Median survival times were 11.0 months for the PCNU group and 8.4 months for the AZQ group. The difference in survival curves was statistically significant for the RP (p = 0.01) and the VSG (p = 0.02). Life-table analysis of the VSG showed estimated 2-year survivals of 34% for PCNU and 11% for AZQ. The advantage of PCNU remained significant (p = 0.006) after adjustment for histopathologic category, age, initial performance status, and interval from initial reported surgery. Myelosuppression was the principal toxicity in both groups.

Incomplete factorial designs for randomized clinical trials.

Recently there has been increased interest in considering factorial designs for randomized clinical trials when one wishes to study two or more treatments. Such designs may offer impressive gains in efficiency compared with a series of trials studying one treatment at a time. This is especially true when the treatments do not interact with one another. If interactions are of special interest, factorial designs provide one sensible approach for studying them, but larger sample sizes would be required because tests for interactions have lower power than those for main effects. In trials designed to test putative agents for preventing cancer, interactions may be of less interest so that fractions of higher-order factorial designs might be appropriate. Sometimes it may not be reasonable, interesting, feasible, or ethical to study all treatment combinations required in a complete or fractional factorial design, yet one may want to preserve some of the factorial structure to increase efficiency and to aid understanding. For such situations, incomplete factorial designs are proposed. Although not all of the advantages of full factorial designs are preserved, such designs may provide reasonable compromises for certain situations.

The study of markers of biological effect in cancer prevention research trials.

Biological markers may provide a valuable tool for the development of cancer prevention agents, for monitoring patient compliance to a selected intervention, or for further defining the carcinogenic process. This discussion focuses on markers of biological effect and the rationale for their use in cancer prevention trials. Recent studies with biological markers are investigating their incorporation into phase-I, -II, and -III chemoprevention clinical trial designs. Their use in clinical studies is expected to increase the number of agents that may be evaluated and to provide valuable information on the biological effectiveness of agents, doses, and schedules. Markers may also provide information to help in selecting high-risk groups for prevention research, and to indicate the pathways inhibited and the stage of carcinogenesis affected. Such information may prove of crucial importance in strengthening the rationale for long-term trials and other ancillary research. Biomarker research for colon carcinogenesis is discussed, including examples of a number of recent trials that may influence future progress in this area of prevention research. A crucial step in this process is marker validation as an aspect of major prospective observational and intervention studies where cancer incidence is the endpoint. We cannot be fully confident of markers as intermediate endpoints until the evidence from clinical trials is sufficiently strong to support major public health initiatives for prevention.

Aspects of statistical design for the Community Intervention Trial for Smoking Cessation (COMMIT).

We present statistical considerations for the design of the Community Intervention Trial for Smoking Cessation (COMMIT). One outcome measurement, the quit rate in randomly selected cohorts of smokers, is compared with another outcome measurement, the decrease in smoking prevalence, in terms of statistical efficiency and interpretability. The COMMIT study uses both types of outcome measurements. The merits of pair-matching the communities are considered, and sample size calculations take into account heterogeneity among pair-matched communities. In addition to significance tests based on the permutational (randomization) distribution, we also describe approaches for covariate adjustment. The COMMIT design includes 11 pair-matched communities, which should provide good power to detect a 10% or greater difference in quit rates between the intervention and control communities in cohorts of heavy smokers and in cohorts of light or moderate smokers. The power is only moderate to detect intervention effects on the decreases in overall smoking prevalence or in the prevalence of heavy smoking.

Assessing the gain in efficiency due to matching in a community intervention study.

COMMIT (Community Intervention Trial for Smoking Cessation) is a randomized study employing a matched pairs design. Pairs of communities were selected on the basis of their geographical proximity and were chosen to be matched on variables strongly expected to relate to the outcome variable, the smoking quit rate. However, quantitative information was not available to evaluate the efficiency gain from matching. We have used baseline smoking quit rates in the communities as a surrogate for the outcome measure to evaluate the gain in efficiency from the matching. Our method takes account of the possible imperfection of the surrogate as a representative of the true outcome. The method estimates an efficiency gain of at least 50 per cent using the matched design. We also evaluate the further gains in efficiency which would be made by using the baseline quit rate to balance the randomization.

Factorial and reciprocal control designs.

The basic principles of factorial designs are briefly reviewed and estimating equations for a 2(3) factorial design are presented in order to show how easily the basic idea of the 2 x 2 factorial design generalizes to higher dimensions. Fractional factorial designs are explained and advocated for situations in which the number of experimental groups might otherwise become unmanageably large or costly. Actual examples of use in medical settings are presented for both a 2(3) full factorial and a half-fractional 2(4) design. A new class of designs called reciprocal control designs is proposed for certain disease screening studies. An example is provided by a study designed to screen for lung, colorectal, and prostate cancers.

Projecting individualized probabilities of developing breast cancer for white females who are being examined annually.

To assist in medical counseling, we present a method to estimate the chance that a woman with given age and risk factors will develop breast cancer over a specified interval. The risk factors used were age at menarche, age at first live birth, number of previous biopsies, and number of first-degree relatives with breast cancer. A model of relative risks for various combinations of these factors was developed from case-control data from the Breast Cancer Detection Demonstration Project (BCDDP). The model allowed for the fact that relative risks associated with previous breast biopsies were smaller for women aged 50 or more than for younger women. Thus, the proportional hazards models for those under age 50 and for those of age 50 or more. The baseline age-specific hazard rate, which is the rate for a patient without identified risk factors, is computed as the product of the observed age-specific composite hazard rate times the quantity 1 minus the attributable risk. We calculated individualized breast cancer probabilities from information on relative risks and the baseline hazard rate. These calculations take competing risks and the interval of risk into account. Our data were derived from women who participated in the BCDDP and who tended to return for periodic examinations. For this reason, the risk projections given are probably most reliable for counseling women who plan to be examined about once a year.

Some statistical considerations for design of cancer prevention trials.

Carcinogenesis is believed to occur in at least two stages, initiation and promotion, followed by a preneoplastic lesion which develops into cancer. Cancer prevention trials can be classified as primary if the intervention precedes initiation, secondary if it occurs during promotion, and tertiary if it is applied to a preneoplastic lesion. Tertiary prevention trials resemble treatment trials, but primary and secondary prevention trials may be very different in size, duration, and cost. After reviewing some basic questions which must be addressed in designing any cancer prevention trial, some special design considerations appropriate for primary and secondary prevention trials are discussed. These include the use of factorial designs, group or cluster randomization, special sample size calculations needed for large-scale trials of long duration with cancer incidence as the endpoint, and the idea of the case-cohort approach for monitoring and for subsequent exploratory analysis of trial data.

Using permutation tests and bootstrap confidence limits to analyze repeated events data from clinical trials.

In clinical trials comparing treatments for superficial bladder cancer, patients are at risk of repeated recurrences of their disease. Statistical methods of analyzing such data are required. This article presents a nonparametric approach. A statistical test to compare the recurrence or tumor rates in two treatment groups, using the randomization distribution, is described. Confidence intervals for the rate ratio are determined from the bootstrap distribution. The implementation of both requires Monte Carlo methods. Computer simulations support the use of these nonparametric methods when there are more than 60 recurrences in each treatment group. An example illustrating their use is given. The strategy adopted for analysis of these data could be applied to other clinical trials where standard methodology is inappropriate.

The dietary fat--breast cancer hypothesis is alive.

Data from animal experiments and human correlation studies strongly support the dietary fat-breast cancer hypothesis. Moreover, a causal relation between dietary fat and breast malignancy is biologically plausible. Negative findings from recent analytic epidemiologic studies of dietary fat and breast cancer, however, have fueled the notion that the hypothesis is no longer viable. We argue that only limited conclusions should be drawn from epidemiologic studies to date because of the narrow range of dietary fat intake among subjects and the substantial measurement error in dietary assessment. Although many doubts remain about the dietary fat--breast cancer hypothesis, the question is of such importance that intensive efforts at designing better studies of the hypothesis are urgently needed. Such studies might include (1) laboratory investigations in humans that examine possible mechanisms for the effects of fat, (2) large, prospective epidemiologic studies, and (3) randomized, controlled diet trials.

Clinical trials in diet and cancer.

Clinical trials in diet and cancer have special problems. We distinguish between a nutritional supplementation and a dietary intervention trial. Since the latter involves a lifestyle modification the intervention requires more careful planning. To illustrate, the method used to reduce the fat intake of subjects in the Women's Health Trial is described. Any clinical trial should satisfy three basic criteria before initiation: plausibility of hypothesis, feasibility, and justifiable cost. In diet and cancer the plausibility of a hypothesis is often controversial. Our conventional reliance on evidence from case-control and cohort studies for judging the plausibility of dietary hypotheses may be misplaced. Errors in the assessment of individuals' diets and the difficulties of separating the effects of highly correlated dietary variables impose severe limitations on the ability of these studies to elucidate the possible effects of diet on cancer incidence. It is therefore unlikely that a consistent pattern of results will be found. Randomized intervention trials, although expensive, have several advantages over analytic epidemiological studies. For example, whereas dietary assessments are required in epidemiological studies to establish a relationship, in intervention trials they are needed only to explain the relationship. We conclude that there is a serious need for reconsidering the relative importance of evidence from various kinds of epidemiological studies relating to diet and cancer.