Evaluation of historical control data in carcinogenicity studies.

Results obtained in long-term carcinogenicity studies with animals should be evaluated, first and foremost, by statistical comparisons of the data obtained from the treated group with that from the concurrent control group. Often the results are compared with data from so-called historical control groups in order to take variations in the incidences of spontaneous tumours into account. Because historical control data change in the course of time and for a variety of reasons, certain requirements must be met before they may be used in the evaluation of the results of long-term studies. The present paper discusses potential sources of variability of tumour incidences in untreated animals, presents databanks for historical control data, mentions the factors that affect tumour incidences in untreated animals and describes biostatistical data evaluation. Finally, details are given of the criteria used by the DFG Commission for the Investigation of Health Hazards of Chemical Compounds in the Work Area to decide whether historical control data may be applied. These include the requirement that the historical control data were obtained with animals of the same species and strain and from the same breeder. The data were obtained in the same laboratory, the study design, experimental methods and assessment criteria were the same, and the studies used for the comparison were carried out within a limited time window. Historical control data that have not been published may be used provided they fulfil the above criteria and have been made available in sufficient detail to be comprehensible.