Antitumor efficacy testing in rodents

Abstract

The preclinical research and human clinical trials necessary for developing anticancer therapeutics are costly. One contributor to these costs is preclinical rodent efficacy studies, which, in addition to the costs associated with conducting them, often guide the selection of agents for clinical development. If inappropriate or inaccurate recommendations are made on the basis of these preclinical studies, then additional costs are incurred. In this commentary, I discuss the issues associated with preclinical rodent efficacy studies. These include the identification of proper preclinical efficacy models, the selection of appropriate experimental endpoints, and the correct statistical evaluation of the resulting data. I also describe important experimental design considerations, such as selecting the drug vehicle, optimizing the therapeutic treatment plan, properly powering the experiment by defining appropriate numbers of replicates in each treatment arm, and proper randomization. Improved preclinical selection criteria can aid in reducing unnecessary human studies, thus reducing the overall costs of anticancer drug development.

Figure 1

Activity of tamoxifen in human tumor xenografts in mice. A) MDA-MB-435 estrogen receptor–negative melanoma xenografts. B) MDA-MB-361 estrogen receptor–positive breast cancer xenografts. Cells of both lines were implanted orthotopically into the mammary fat pad of athymic nu/nu NCr mice (Animal Production Program, NCI-Frederick), and treatment was initiated when the tumors reached 150–175 mg in size. The MDA-MB-361 tumor-bearing mice were treated weekly with estradiol cypionate (20 μg per mouse) to support tumor growth. Exogenous estradiol is not required for progressive growth of MDA-MB-435 xenografts. For both studies, the vehicle control was 100% sesame oil given by oral gavage once daily for 20 days (n = 20 mice). Tamoxifen was administered by oral gavage once daily for 20 days at a dose of 45, 22.5, or 11.25 mg/kg (n = 10 mice per dose). Individual tumor weights were calculated as weight in mg = (length × width²)/2. Data are plotted as median tumor weight ± the 95% confidence interval of the median.

Figure 2

Activity of temozolomide and topotecan in human tumor xenografts. A) A375 melanoma xenografts. B) Colo 829 melanoma xenografts. Cells of both lines were implanted subcutaneously in female athymic nude (nu/nu NCr) mice (Animal Production Program, NCI-Frederick). Treatment was initiated when the tumors reached 150 mg. Temozolomide was administered by oral gavage as a single dose of 400 mg/kg or as three 200-mg/kg doses given 4 days apart (temozolomide 200 mg/kg × 3) (n = 10 mice per dose group). Topotecan was administered intraperitoneally at 1 mg/kg 5 days per week for 2 weeks (n = 10). The vehicle control group (n = 20) was treated with three doses of saline given 4 days apart. Individual tumor weights were calculated as weight in mg = (length × width²)/2. Data are plotted as median tumor weights ± 95% confidence interval.

Figure 3

Kaplan–Meier analysis of survival of athymic nude mice bearing intraperitoneal OVCAR-5 human ovarian cancer xenografts. Mice were nu/nu NCr (Animal Production Program, NCI-Frederick). The therapeutic agent was administered at a dose of 1 mg per mouse given intraperitoneally once every other day, for a total of seven doses (n = 10 mice per group) using two different vehicles (lipid vehicle and phosphate-buffered saline [PBS]). The lipid vehicle (n = 20 mice) and PBS (n = 20 mice) were used in separate vehicle control groups following the same dosing schedule. Mice were treated with vehicle alone or with one of the therapeutic agents solubilized in the vehicles. Dotted lines indicate 95% confidence intervals.

Figure 4

Tumor weight plots for MDA-MB-361 human breast tumors implanted subcutaneously in athymic nude mice. The data are from Table 1. The main graph presents the median and average tumor weights for a group of six mice (nu/nu Ncr; Animal Production Program, NCI-Frederick), each implanted with 1 × 10⁷ cells in 0.1 mL. The inset presents the individual growth curve for each of the six mice. Individual tumor weights were calculated as weight in mg = (length × width²)/2. The error bars indicate the 95% confidence intervals of the averages or the medians, as appropriate.

Figure 5

Tumor weight plots for MDA-MB-361 human breast tumors implanted in the mammary fat pads of athymic nude (nu/nu Ncr; Animal Production Program, NCI-Frederick) mice. The data are from Table 2. The main graph presents the median and average tumor weights for a group of 13 mice, each implanted with 1 × 10⁷ cells in 0.1 mL. The inset presents the individual growth curves for each of the 13 mice. Individual tumor weights were calculated as weight in mg = (length × width²)/2. The error bars are the 95% confidence interval of the average or the median, as appropriate.