Historical and methodological developments in clinical trials at the National Cancer Institute.

The first randomized clinical trial at the National Cancer Institute (NCI), planned in 1954, commenced in 1955 for the treatment of patients with acute leukaemia. The programme in clinical trials at NCI had strong influence from the clinician and administrator, C. Gordon Zubrod, who introduced the randomized clinical trial at NCI and organized the co-operative clinical trials programme of the Cancer Chemotherapy National Service Center (CCNSC) beginning about 1955. The biostatistician, Marvin Schneiderman, collaborated on the first randomized trials in acute leukaemia and solid tumours and recruited the biostatisticians and statistical centres in the early phase of the co-operative clinical trials programme of the CCNSC. From the beginning, there was acceptance of the principles of the randomization of patients and the statistical analysis of data. The sequence of clinical trials for a new agent included the non-randomized phase I (dosage finding) and phase II (preliminary efficacy) trials as well as the phase III (comparison of treatments) trials. New concepts for the treatment of patients developed from 1955 to the mid-1960s included the combination of therapies with independent activity to increase response rates and the administration of therapy to patients in a disease-free (remission) state to prolong the disease-free state. Methodological developments related to clinical trials up to the mid-1960s included: a plan for phase II trials (Gehan); a generalization of the Wilcoxon test for the comparison of survival distributions with right-censored data (Gehan); a test of proportional hazards for survival distributions, which later became known as the Mantel-Haenszel test (Mantel), and an exponential regression model with an explanatory variable (Feigl and Zelen).

The statistical analysis of a carcinogen mixture experiment. III. Carcinogens with different target systems, aflatoxin B1, N-butyl-N-(4-hydroxybutyl)nitrosamine, lead acetate, and thiouracil.

This paper describes factorial experiments designed to determine whether two carcinogens that lead to cancers in different organ systems act synergistically to produce cancers in Fischer 344 rats. Four carcinogens, aflatoxin B1 (AFLA), N-butyl-n-(4-hydroxybutyl)nitrosamine (NBBN), lead acetate (LA), and thiouracil (THIO) were studied in pairwise combinations. Each of the six possible pairs were studied by means of a 4 X 4 factorial experiment, each agent being fed at zero and at three non-zero doses. Methods of analysis designed explicitly for this study were derived to study interaction. These methods were supplemented by standard statistical methods appropriate for single agent studies. Neither synergism nor antagonism was demonstrated in these combined exposure studies. Findings for male and female animals were consistent.

Determination of menadione in an animal feed using supercritical fluid extraction and HPLC with electrochemical detector.

Menadione (vitamin K3) is extracted from spiked rat chow using supercritical fluid carbon dioxide at 8000 psi and 60 degrees C. Quantitative extraction requires only 20 min. The extraction does not suffer from the problems associated with conventional solvent extraction of lipophilic materials from animal feeds. Menadione is determined in the extract, which does not require further cleanup, using reversed-phase high-performance liquid chromatography (HPLC) with reductive mode electrochemical detection at a silver electrode at -0.75 V vs. calomel. The minimum detectable quantity by the detector is 125 pg of menadione, and the response is linear over at least 4 orders of magnitude; however, the minimum quantity extractable is about 20 micrograms/g of feed. Repetitive extracts of a spiked feed sample over a five-day period show an average recovery of 90.5% with a relative standard deviation of 2.2% at the 1 mg/g level.

Determination of vitamin K1 in powdered infant formulas, using supercritical fluid extraction and liquid chromatography with electrochemical detection.

Vitamin K1 (phylloquinone) is extracted from commercial soy protein-based and milk-based powdered infant formulas by using supercritical fluid extraction with CO2 at 8000 psi and 60 degrees C. Quantitative extraction requires only 15 min, and does not suffer from the problems associated with conventional solvent extraction of lipophilic materials from media such as formulas. Vitamin K1 is determined in the extracts by using reverse-phase liquid chromatography (LC) with reductive mode electrochemical detection at a silver electrode polarized at -1.1 V vs SCE. LC run time is 9 min. The minimum detectable quantity is 80 pg, and response is linear over at least 5 orders of magnitude. Recovery of vitamin K1 from a milk-based powdered formula was 95.6% with RSD of 7.4%, and from a soy protein-based product, 94.4% recovery with RSD of 6.5%.

The statistical analysis of a carcinogen mixture experiment. II. Carcinogens with different target organs, N-methyl-N'-nitro-N-nitrosoguanidine, N-butyl-N-(4-hydroxybutyl)nitrosamine, dipentylnitrosamine, and nitrilotriacetic acid.

This paper describes factorial experiments designed to determine whether two carcinogens that act on different organ systems act synergistically to produce cancers in Fischer 344 rats. Four carcinogens, N-methyl-N'-nitrosoguanidine (MNNG), N-butanol-N-butylnitrosamine (NBBN), nitilotriacetic acid (NTA), and dipentylnitrosamine (DPN) were studied in pairwise combinations. Each of the six possible pairs was studied by means of a 4 X 4 factorial experiment, each agent being fed at zero and at three non-zero doses. Methods of analysis designed explicitly for this study were derived to study interaction. These methods were supplemented by standard statistical methods appropriate for single agent studies. Antagonism was demonstrated in some chemical mixtures containing NTA. Other chemical mixtures did not interact. Findings for male and female animals were generally, but not always, in agreement.

Statistical analysis of a carcinogen mixture experiment. I. Liver carcinogens.

This paper describes factorial experiments designed to determine whether 2 liver carcinogens act synergistically to produce liver cancers in Fischer 344 rats. Four hepatocarcinogens, cycad flour, lasiocarpine (CAS: 303-34-4), aflatoxin B1 (CAS: 1162-65-8), and dipentylnitrosamine (CAS: 13256-06-9), were studied in pairwise combinations. Each of the 6 possible pairs was studied by means of 4 X 4 factorial experiment, each agent being fed at zero and at 3 non-zero doses. Methods of analysis designed explicitly for this study were derived to study interaction. These methods were supplemented by standard statistical methods appropriate for one-at-a-time studies. Antagonism was not discovered in any chemical mixture. Some chemical mixtures did interact synergistically. Findings for male and female animals were generally, but not always, in agreement.

Formaldehyde risk assessment for occupationally exposed workers.

Formaldehyde has been shown to be carcinogenic in animals and should be considered potentially carcinogenic in humans. The mechanism of action is unknown but formaldehyde is weakly genotoxic and also may act as a late stage carcinogen or promoter. An estimated 1.3 million workers are potentially exposed to formaldehyde through their occupation. Of those exposed workers, about 3.5% were found to be exposed to formaldehyde air concentrations greater than the 3 ppm set by OSHA as a permissible exposure level. Fewer than 12% were exposed to concentrations greater than 1 ppm, but over 88% were exposed to concentrations of 0.5 ppm or more. A quantitative risk assessment, using the multistage low-dose extrapolation model, found the (maximum likelihood) estimate of lifetime risk for excess cancers to be 620 per 100,000 at the OSHA permissible exposure level. The estimated risk is 23 per 100,000 at 1 ppm and 2.8 per 100,000 at 0.5 ppm. Reduction of the OSHA permissible exposure level to 1 ppm would significantly decrease risk with minor economic disruption for most industries involved. However, reduction of risk to levels which have been generally regarded by other regulatory agencies as acceptable, i.e., 10(-5) to 10(-6), would require increased control by all the industries reviewed.

Extrapolation from incomplete data to total or lifetime risks at low doses.

Both epidemiology and laboratory data can contribute to estimates of risks to humans of exposure to low doses of carcinogens. The sum of all these contributions does not permit us to make these estimates with certainty. In chronic disease epidemiology, in looking for possible excessive cancer risks, we sometimes fail to have an adequately long observation time or to observe a population sufficiently aged for cancers to appear in meaningful numbers. In studies of most human exposures, dose data are often lacking, beyond a vague "yes-no" or "lots, not much, hardly any." Thus, without a knowledge of what dose produced an observed result it becomes logically impossible to know what result some other (presumed) dose might yield. Animal data show some promise of being useful in extrapolating to low doses in man. However, several problems exist: (a) man is not a tailless, two-legged mouse, or featherless chicken--that is, we do not know if man is more or less sensitive than the laboratory animal; (b) the mathematical model used for extrapolation leads to large differences in estimates of response; (c) man is genetically heterogeneous and is usually exposed to many more hazards than is the laboratory animal. Thus, existing data, even from well-done studies, are inadequate if we want to make extrapolations in any detail or to apply to specific subgroups in the population. Any risk estimation we do may have to be stated in terms that point out the wide ranges of the estimates.