A year ago, when researchers at Purdue University argued that environmental standardisation in laboratory experiments involving mice was likely to lead to more, rather than less, variation between different laboratories, they met with some resistance since it was not clear what factors should be varied to improve reproducibility. Following an analysis of data from behavioural tests commonly used in drug discovery studies, they have now shown that introducing only two controlled environmental variables to preclinical studies using mice can greatly reduce false positives and the number of animals needed for testing. The tests, which compared behaviours between two inbred strains of mice, were repeated in four different model laboratories that varied in details such as background noise, the age of the mice, environmental enrichment, familiarity with handler, lighting levels and cage size. In each laboratory, one group of mice (standardised) were treated identically whilst the other group (heterogenised) were tested under four different sets of conditions produced by varying two environmental factors in a controlled manner. Mice of the same strain would have been expected to show the same behaviours in each laboratory but, in 33 out of 36 behavioural characteristics such as fear and curiosity, the standardised group showed as much as five times more variation between laboratories compared with the heterogenised group.
When conditions are highly standardised, the variation in data produced within a particular laboratory will be very low, but variations between laboratories will be large and unpredictable. The researchers believe that tests in mice using a heterogeneous test design more closely resemble human clinical trials and should reduce both the number of animals needed for preclinical studies and the number of false positives. A reduction in false positives could have important implications for reducing the number of compounds that fail in expensive downstream clinical trials.
The study is published in Nature Methods.