Gross tumor assessment is not a reliable measure of the efficacy of chemo-preventive agents for breast cancer in preclinical mouse models

Despite success in preclinical animal models, most cancer prevention drugs fail to show efficacy in clinical trials. One potential reason could be that study end points in animal models are not sufficiently robust. Preclinical efficacy studies often use tumor development, as measured by palpation, as an endpoint. A comprehensive study design that includes more rigorous measures by histologic assessment of all the mammary glands is less common. We hypothesized that the efficacy of drug treatment may vary based on the approach to tumor assessment. SV40C3TAg mice that spontaneously develop TNBC-like tumors were used to assess the efficacy of drugs for breast cancer prevention. Histological grading was performed using a 5-point scale. Animals were treated with fluvastatin and/or aspirin for 16 weeks and the experiment was concluded at 22 weeks, and all the mammary glands removed for histologic assessment. Grossly palpable mammary glands (> 3 mm) and /or histologic grade 4-5 was considered tumor bearing and treatment non-responder. 112 mammary glands from 40 mice were assessed. A high concordance (> 90%) was noted between palpably enlarged glands and the finding of grade 4-5 lesions. Conversely, among 74 non-palpably enlarged mammary glands, more than 66% were histologically high-grade and would have been wrongly classified as tumor-free based on gross tumor measurement. In drug efficacy studies, significant differences were noted in the rate of treatment response based on the method of tumor assessment, with a 22% response rate noted using the histological grading system and a 65% response rate using gross palpation. These analyses suggest that there is poor concordance between the gross tumor measurements and histological tumor assessment in a mouse model of breast cancer. Thus, gross tumor measurement alone is insufficient for determining chemopreventive efficacy in preclinical animal models.