Technical Report
29.12.2002

TR 083 – The Use of T25 Estimates and Alternative Methods in the Regulatory Risk Assessment of Non-threshold Carcinogens in the European Union

TR 083 : The Use of T25 Estimates and Alternative Methods in the Regulatory Risk Assessment of Non-threshold Carcinogens in the European Union | December 2002

The regulation of carcinogens in the European Union, under the 'labelling and use' regulations and in the framework of existing chemicals, relies currently on classification using the strength of evidence. It is self-evident that potency, as well as strength of evidence, should be used to determine concentration limits for carcinogenic substances in preparations and for regulating existing chemicals that have carcinogenic properties.

It has been proposed that a simplified method of estimating carcinogenic potency from the results of experimental animal studies should be used for these purposes. The method, proposed by scientists from Norway and the Netherlands, is known as the T25 method. It relies on a simplified method of deriving the dose of the carcinogen that will produce cancer in 25% of the animals that would not have developed cancer spontaneously. A human HT25 is then calculated based on a number of default assumptions. The ratio between the estimated human daily exposure and the HT25 is then used to calculate the expected incidence (= risk) of cancer.

The T25 has also been used in setting concentration limits that trigger labelling for carcinogens in preparations. The T25 estimate is used to assign the chemical to one of three potency classes. The potency class, taken together with the carcinogen classification category, is then used in a scheme to assign a concentration limit for the chemical in preparations.

In general, provided the method is used with care, it should provide a satisfactory procedure for assigning concentration limits in preparations. workshop was organised by ECETOC in November 2000 to consider these issues. Following the workshop, ECETOC established a Task Force to consider the current proposals and alternatives that had been identified in a preliminary paper drafted by three scientists who attended the workshop.

The Task Force has concluded that the T25 method for risk assessment of non-threshold carcinogens is open to criticism for the following reasons:

  • Estimates based on simple proportional linear extrapolation from the T25 should not be used to predict absolute cancer risk, because of the many unverifiable assumptions used in their calculation;
  • the estimate is based on unproven methodology, which many believe is flawed;
  • the resulting quantitative estimate has a spurious sense of accuracy;
  • there is a strong likelihood of misuse and misinterpretation of derived human cancer risk estimates;
  • the resulting calculated value will be taken as the 'true' risk in communication with the risk managers and with those exposed;
  • the justification of the reliability of the T25 method by comparison of its results with those of the Linearised Multistage model is misleading;
  • species differences and mechanistic data are not taken into account.

These concerns can be summarised by stating that the risk estimates produced by this and similar methods appear precise but almost certainly do not reflect the real risks. This may cause confusion.

In addition to the T25, several possible alternatives methods were considered by the Task Force. Of these, the Task Force recommends a 'margin of exposure' method, based on the weight of evidence from all suitable relevant carcinogenesis bioassays. For each bioassay, the maximum likelihood estimate of the benchmark dose at a 5% risk level is calculated, and a representative value of these estimates converted by allometric scaling to a human benchmark exposure level. The derivation of this and the final decision as to whether the standard is exceeded, should be informed by other data, such as those relating to metabolism, pharmacokinetics, mechanisms of action, the shape of the dose response curve and human experience.

The human benchmark exposure level is divided by the realistic worst-case exposure of workers, consumers and the public exposed through the environment, and the resulting margin of exposure compared with agreed standards, likely to be 104, 105 and 106, respectively, as a basis for judging the acceptability, or otherwise, of the carcinogenic risk.