Skip to main content

METHODOLOGY FOR THE ANALYSIS OF RADIATION CARCINOGENESIS STUDIES AND APPLICATION TO ONGOING EXPERIMENTS

Objective


A computer program has been developed for the analysis of animal carcinogenic data. A preliminary analysis of the data show a distinct dose effect relationship for tumour induction after irradiating the animals by both neutron and gamma sources. The animals are selected for analysis on the basis of instant lethality of the induced tumour. A further refinement in selecting these animals on the bases of the pathological classification is required. The analysis methods developed have also been applied to data from the long lasting monkey experiment performed at The Radiobiological Institute (TNO) Rijswijk. The monkeys were subjected to total body irradiation up to relatively high doses of both X-rays and neutrons, followed by a bone marrow transplantation. The late effects observed in the animals surving more than 3 years include a number of neoplasms. Up to now the prevalence of animals with these induced tumours was quoted to be 9 out of 20 over an observation period of 227 monkey years for X-ray irradiated monkeys (average dose 6.7 Gy) and 7 of the 9 monkeys exposed to fission neutrons (average dose 3.4 Gy) during 87 monkey years. In a control group of 21 animals 2 animals were observed with lethal neoplasms. The pathology findings showed a variety of neoplasms including osteosarcomas, kidney, thyroid and glomus tumours. An apparently shorter latency period was found for fission neutrons. The relative biological effectiveness (RBE) was found to be between 4 and 7 for these relatively hgh doses of fission neutrons.

A more detailed analysis with the Lifestat package serves a dual purpose: a correction for the animals dying intercurrently from causes unrelated to the endpoint, being the induction of a tumour and the analysis of the distribution of tumour induction times will provide the most information on carcinogenicity for these 2 cohorts of monkeys.

Statistical methods were developed for the analysis of animal experiments and applied to the carcinogenesis experiments performed in the Nuclear Center of Fontenay aux Roses (France). New experiments have been designed to investigate an important unsolved problem in radiation protection which is whether protraction of irradiation with high linear energy transfer (LET) particles leads to a reduction of the effect or to an increase (a reverse time factor has been repeatedly observed).

A detailed comparison of the efficiency of neutrons versus gamma-rays whole body irradiation in Sprague Dawley rats in inducing a broad spectrum of neoplasms has been performed. The results show differences in the time course of the various types of neoplasms (sarcomas as compared to carcinomas) and in the relative efficiency of neutrons and gamma-rays. The analysis shows very high relative biological effectiveness (RBE) for neutrons compared to X-rays, again in the range of those found for pulmonary carcinomas. Another aspect of the work was the further development of methods of analysis for incidental tumours. The isotonic regression has been used to estimate the incidence of incidental tumours. The drawback of such a method was the absence of the confidence intervals for the estimated function. In the absence of an analytical answer a modification of the wellknown bootstrap method was introduced. Groups of Sprague Dawley rats were irradiated either with gamma-rays or neutrons. For a selection of tumours (carcinomas of the unrinary of the digestive tract) were are compared with results obtained under the assumption of the nonlethality (isotonic regression). For the isotonic estimate they were obtained with the bootstrap. The purpose of this comparison wasto show that, for the same data set, the 2 assumptions of lethality or nonlethality lead to quite different results.

Assessment of cancer risk following exposure to ionizing radiation at low dose and low dose rate is a problem of major concern.
Observations indicate that linearity for dose response at low dose and dose rate may not be the consequence of single events in single cells.
It appears that there is a need for more information from experimental data in animals exposed to low doses and low dose rate.

In this experiment, 4 groups of rats were irradiated with low doses of fission neutrons from californium source and compared to a control group of 501 animals. Two groups received a brief irradiation of 25 mGy, series S1 (152 rats)at 100 days and series S2 (255rats))at 420 days to assess the role of age in carcinogenesis. The other 2 groups were irradiated over a year, animals were continuously exposed at 2 different distances from a source of californium 252, 6 days a week; series S3 (2490 and S4 (50 rats) recieved respectively 24 nGy and 53 mGY.
The dose rate influenced the life span, the mean survival times for S1 and S2 decreased by 90 days in comparison with controls and were equivalent to survival times fro S4, but the mean survival times for S3 decreased by only 60 days. The dose rate influenced the mean times of tumour occurence, since these times increase with decreasing dose rates particularly for carcinomas.
After neutron exposure of adult rats, for the same dose and dose rate, no significant dependence of age was observed for carcinoma induction, but a decreased risk with increasing age at exposure for sarcoma induction was shown.
A reduced sarcoma excess and a carcinoma increase were observed in exposed groups S1, S2 and S3, but S4 was equivalent to controls.
In conclusion, there is no evidence for a reduction in cancer following protracted exposure to fission neutrons at 25 nGy, in comparison with the same dose delivered as a single short exposure.

Survival and pathology data were gathered rats that received different doses of total body irradiation (as single dose or fractionated) of different radiobiological effectiveness (ie X-irradiation and 0.5 MeV neutrons).

Background tumour incidence in untreated control groups may vary considerably. It was therefore necessary to compare data obtained from experimental groups with those from a corresponding untreated control group from the same period. As the experiment stretched over almost a decade, several control groups had to be included to serve as a reference. Furthermore the spectrum of age related pathology, characteristic for lifetime experiments, is influenced by intercurrent disease. The rats used for the experiment were derived from specific pathogen free stock and kept there after under conventional conditions.

Regular health programme results indicated that a Sendai virus infection occurred during the later part of one of the earliest experiments. It was therefore of paramount importance to check whether this influence could have disturbed the results of this experiment. It appeared that in the control groups, except for interstitial pneumonia no life threatening complications occurred and survival was not decreased. It was therefore decided not to exclude thase experiments from the pathology and survival survey. Later experiments were free of complicating disease.

Part of the histopathological examination of complete necropsies of the dos group receiving 4 Gy X-rays and the corresponding control group has been completed.
Large scale animal experiments at various European institutes have been undertaken in the past decade. The information from different laboratories can be meaningfully combined if the experimental procedures and the pathology are closely coordinated, if similar requirements apply to the design of an experiment with regard to animal numbers, dose groups and other factors and if the methods of statistical analysis are defined at the outset of the experiments. It is accordingly necessary to develop standards of experimental planning and methods of statistical analysis in animal experiments. Analysis of the tumour induction data reveals differences, which may be partly explained by a diversion of the cohorts under study, but are also believed to result from the employed analysis methods.

In particular the different mathematical models (eg the parametric Weibull model versus the nonparametric proportional hazards model) have been under discussion. The analysis approach of both the single dose, fractionated and protracted experiments, either with or without drugs, is also not resolved. The control group appears to play a crucial role in the analysis approach. A unified approach to the analysis of experimental data of animal carcinogenic studies will be the aim of this contract. To this aim a framework will be laid down for the analysis of dose-effect relationships in a concorded way for the existing data in Europe.

The results of the work will be assembled in a laboratory handbook and accompanying software. Essential points will be:
a set of rules and recommendations concerning the choice of the animal strains, the number of animals required in an experiment and their distributions into groups with different doses and time schedules, and the necessity of control groups;
a set of rules defining the conduct of an experiment to ensure comparability and repeatability of the results; special emphasis will be on the necessity of correct randomization;
recommendations on the coordination of the pathology and on the recording of tumours (this can largely draw on work performed by EULEP, but needs to include conventions on data recording for exchange of results);
a set of rules on the extent of data and their format which permit efficient exchange between laboratories for comparison of results.

An extensive section which is central to the project will deal with the models to be used in the analysis, and with the methods and algorithms to be applied. A computer program will be developed under this contract, encompassing the multiinstitutional approaches to the analysis of dose-effect relationships. The department of Clinical Oncology of the Academic Hospital Leiden supports the project through the coordination of the proposed methodological approaches and the associated software development for the comprehensive analysis program. This program is produced in such a way that it can be distributed among the various European institutes that are involved in carcinogenic studies. Extensive development of the computer program has already taken place in 1988 and 1989, as supported by the CEC under a previous contract number BI 6-D-219-NL.

Coordinator

LEIDEN UNIVERSITY
Address
72,Rapenburg 70
2311 EZ Leiden
Netherlands

Participants (3)

Business Unit of TNO Built Environment and Geosciences
Netherlands
Address
Van Mourik Broekmanweg 6
2628 XE Delft
COMMISSARIAT A L'ENERGIE ATOMIQUE
France
Address
Rue De La Federation 31-33
75015 Paris
GSF - FORSCHUNGSZENTRUM FUER UMWELT UND GESUNDHEIT GMBH
Germany
Address
Ingolstaedter Landstrasse 1
85764 Neuherberg