Burrowing Behavior as Robust Parameter for Early Humane Endpoint Determination in Murine Models for Pancreatic Cancer

Abstract

Due to late-stage diagnoses and limited treatment options, pancreatic cancer is predicted to be the second leading cause of cancer deaths by 2030. Many different murine models were developed over the past decades to test new therapies for this tumor entity. The laws and regulations demand the continuous refinement of animal experiments in order to improve animal welfare sustainably. A key aspect here is the definition of early humane endpoint criteria to avoid severe and lasting suffering of the animals through timely euthanasia. The present study analyzed the welfare of mice in different pancreatic cancer models, various mouse strains, and under different therapeutic interventions in preclinical testing. Their welfare was monitored before any intervention, during tumor progression, and on the days before the individual humane endpoint for each mouse by assessing body weight change, distress score, perianal temperature, burrowing behavior, nesting activity, and mouse grimace scale. The data was retrospectively analyzed via receiver operating characteristic curve analysis to quantify the predictability of each parameter for humane endpoint determination. Burrowing behavior proved to be a robust predictor of the humane endpoint two days in advance under various conditions, including diverse pancreatic cancer models and different therapeutic approaches.

Keywords: animal welfare; early humane endpoint; pancreatic cancer; refinement.

Figure 1

Assessment of the various welfare parameters on healthy mice (pre), during tumor progression (tp), and on the days (4–1) directly before or at the humane endpoint (hep), respectively, in the orthotopic, subcutaneous, and intravenous pancreatic cancer model. The body weight change (A), distress score (B), and perianal temperature (C) were monitored on the four days before and on the day of hep (4, 3, 2, 1, and hep). The number of burrowed pellets after 2 and 17 h (D,E), nesting activity (F), and MGS score (G) were assessed for three days prior to the hep. The statistical analysis was carried out with repeated measures two-way ANOVA (A,B) or a mixed model (G), followed by Dunnett’s test for comparisons with tumor progression values (tp) within the models (# p ≤ 0.05) and Tukey’s test for inter-model differences (* p ≤ 0.05). The grey area represents the baseline values of healthy mice before the tumor cell injection. Orthotopic (n = 3–9); subcutaneous (n = 3–7); intravenous (n = 4–7).

Figure 2

Discriminative power of the welfare parameters for early humane endpoint prediction in different pancreatic cancer models. The predictive power was calculated for each parameter on the days before (4–1) and on the day of the humane endpoint (hep) via receiver operating characteristic curve (ROC) analysis, in comparison to the respective values during tumor progression. The AUC values were calculated, respectively, for the orthotopic (A), intravenous (B), and subcutaneous (C) animal models. The AUC values were displayed via heat maps, indicating the parameters of high predictive power in green (AUC: 0.80–1.00) and low predictive power (AUC: 0.50–0.70) in red. BW: body weight change; DS: distress score; Temp: perianal temperature; B2h: burrowing activity after 2 h; B17h: burrowing activity after 17 h; Nest: nesting activity after 17 h; MGS: mouse grimace scale; orthotopic (n = 3–9); intravenous (n = 4–7); subcutaneous (n = 3–7).

Figure 3

Monitoring of the various welfare parameters on healthy mice (pre), during tumor progression (tp), and on the days (5–1) directly before or at the humane endpoint (hep), respectively, for the different therapy groups in an orthotopic pancreatic cancer model. The percentage of body weight change (A), the distress score (B), and perianal temperature (C) were monitored on the six days before and on the day of the hep (6, 5, 4, 3, 2, 1, and hep). The number of burrowed pellets after two hours (D) was analyzed for four days prior to hep and on the day of hep. The number of burrowed pellets after 17 h (E), nesting score (F), and MGS score (G) were analyzed for three days before and on the day of the hep. The statistical analysis was carried out with repeated measures two-way ANOVA (A,B), followed by Dunnett’s test for comparisons with tumor progression values within the models (# p ≤ 0.05) and Tukey’s test for inter-model differences. The grey area represents the baseline values for each parameter of all the mice before the tumor cell injection. Vehicle (n = 3–5); Erlotinib (n = 4–7); LXH254 (n = 3–6); Erlotinib + LXH254 (n = 4–7).

Figure 4

Discriminative power of the welfare parameters for early humane endpoint (ehep) determination for the different therapy groups in an orthotopic pancreatic cancer model. The diagnostic power for ehep prediction was calculated for each parameter on the days before (6–1) and on the day of the humane endpoint (hep) via receiver operating characteristic curve (ROC) analysis, in comparison to the respective values during tumor progression. The AUC values were calculated for the vehicle- (A), Erlotinib- (B), LXH254- (C), and dual Erlotinib/LXH254-treated (D) animals. The AUC values were displayed via heat map, indicating parameters of high predictive power in green (AUC: 0.80–1.00) and low predictive power (AUC: 0.50–0.70) in red. BW: body weight development; DS: distress score; Temp: perianal temperature; B2h: burrowed activity after 2 h; B17h: burrowed activity after 17 h; Nest: nesting activity after 17 h; MGS: mouse grimace scale; vehicle (n = 3–5); Erlotinib (n = 4–7); LXH254 (n = 3–6); Erlotinib + LXH254 (n = 4–7).

Figure 5

Analysis of the various welfare parameters on healthy C57BL/6NTac mice (pre), during tumor progression (tp), and on the days (5–1) directly before or at the humane endpoint (hep) for the different therapy groups in a metastasizing tumor model. The percentage of body weight change (A), distress score (B), perianal temperature (C), number of burrowed pellets after two and 17 h (D,E), and nesting score (F) were assessed for five days prior to the hep and on the hep (5, 4, 3, 2, 1, and hep). The statistical analysis was carried out with repeated measures two-way ANOVA (A,B), followed by Dunnett’s test for comparisons with tumor progression values within the models (# p ≤ 0.05) and Tukey’s test for inter-model differences. The grey area represents the baseline values for each parameter of all the mice before the tumor cell injection. Vehicle (n = 4–6); BI + Tram (n = 2); BI + Tram + BKM (n = 3).

Figure 6

Diagnostic ability of the welfare parameters for early humane endpoint prediction for the therapy groups in a metastasizing pancreatic cancer model. The discriminatory power was calculated on the days before (5–1) or on the day of the humane endpoint (hep) via receiver operating characteristic curve (ROC) analysis, in comparison to the respective tumor progression values. The AUC values were calculated for the different treatment groups, namely vehicle (A), dual treatment with trametinib and BI-3406 (B), and triple treatment with trametinib, BI-3406, and BKM120 (C). The AUC values were displayed via heat maps, indicating the parameters of high predictive power in green (AUC: 0.80–1.00) and low predictive power (AUC: 0.50–0.70) in red. BW: body weight development; DS: distress score; Temp: perianal temperature; B2h: burrowed amount after two hours; B17h: burrowed amount after 17 h; Nest: nesting activity after 17 h; MGS: mouse grimace scale; vehicle (n = 4–6); Tram + BI (n = 2); Tram + BI + BKM (n = 3).

Figure 7

Monitoring of animal welfare using the different parameters on healthy mice (pre), during tumor progression (tp), and on the days (5–1) directly before or at the humane endpoint (hep) for different therapy groups in a metastasizing pancreatic cancer model. The percentage of body weight change (A), distress score (B), perianal temperature (C), number of burrowed pellets after 2 (D) and 17 h (E), and nesting activity (F) were monitored for the four to five days before and on the day of the humane endpoint (5, 4, 3, 2, 1, and hep). The statistical analysis was carried out with repeated measures two-way ANOVA (A,B) or a mixed-effects model (D), followed by Dunnett’s test for comparisons with baseline values within the values with the models and Tukey’s test for inter-model differences. The grey area represents the baseline values for each parameter in the healthy mice before the tumor cell injection. Vehicle (n = 3); BI + Tram (n = 1); BI + Tram + BKM (n = 1).

Figure 8

Predictive power of the welfare parameters for early humane endpoint determination for a therapy group in a metastasizing pancreatic cancer model. The discriminatory power was calculated for each parameter on the days before (5–1) or on the day of the humane endpoint (hep) via receiver operating characteristic curve (ROC) analysis, in comparison to the respective tumor progression values. The AUC values were calculated for vehicle-treated animals. The AUC values were displayed via heat maps, indicating the parameters of high predictive power in green (AUC: 0.80–1.00) and low predictive power (AUC: 0.50–0.70) in red. BW: body weight development; DS: distress score; Temp: perianal temperature; B2h: burrowed amount after two hours; B17h: burrowing activity after 17 h; Nest: nesting activity after 17 h; MGS: mouse grimace scale; vehicle (n = 3).