Can trained behavior serve as tool for welfare assessment? A pilot study in a cancer model with Wistar rats

This pilot study investigates whether trained behavioral tasks may serve as early, non-invasive indicators of distress in high-severity disease models in laboratory rats. Therefore, six Wistar rats in a diethylnitrosamine-induced hepatocellular carcinoma model, were trained to perform two tasks: lockbox solving and scale climbing. They were assessed over 18 weeks by body weight, hair corticosterone (hCORT), and Grimace Scale. Changes in task performance time and hCORT levels were observed and revealed a three-phase pattern. Initially, poor task performance (mean lockbox solving time 22 s) and elevated hCORT (mean 10.8 pg/mg) indicated a process of adaptation to the experiment. This was followed by improved performance (mean lockbox solving time 2 s) and reduced hCORT (mean 7.9 pg/mg), suggesting acclimatization. Later, performance declined again (mean lockbox solving time 9 s), accompanied by rising hCORT (mean 8.3 pg/mg), likely reflecting tumor progression and repeated anesthesia exposure. Lockbox solving appeared more variable than scale climbing. Acute stressors, such as room changes and prolonged anesthesia, had an impact on lockbox solving and scale climbing. After the initial housing room change lockbox solving time increased from 20 s to 49 s while scale climbing time rose from 12 s to 26 s. Case observations confirmed that task performance deficits coincided with illness or larger tumor burden. These findings support the use of trained behavior as a sensitive tool to detect distress. This approach could improve welfare monitoring in long-term or high-severity animal studies.