A comparison of imaging techniques to monitor tumor growth and cancer progression in living animals

Abstract

Introduction and Purpose. Monitoring solid tumor growth and metastasis in small animals is important for cancer research. Noninvasive techniques make longitudinal studies possible, require fewer animals, and have greater statistical power. Such techniques include FDG positron emission tomography (FDG-PET), magnetic resonance imaging (MRI), and optical imaging, comprising bioluminescence imaging (BLI) and fluorescence imaging (FLI). This study compared the performance and usability of these methods in the context of mouse tumor studies. Methods. B16 tumor-bearing mice (n = 4 for each study) were used to compare practicality, performance for small tumor detection and tumor burden measurement. Using RETAAD mice, which develop spontaneous melanomas, we examined the performance of MRI (n = 6 mice) and FDG-PET (n = 10 mice) for tumor identification. Results. Overall, BLI and FLI were the most practical techniques tested. Both BLI and FDG-PET identified small nonpalpable tumors, whereas MRI and FLI only detected macroscopic, clinically evident tumors. FDG-PET and MRI performed well in the identification of tumors in terms of specificity, sensitivity, and positive predictive value. Conclusion. Each of the four methods has different strengths that must be understood before selecting them for use.

Figure 1

Detection of tumors by in vivo imaging. B16 melanoma cells were subcutaneously injected into shaved C57Bl/6 mice (n = 4 mice for each technique). Mice were repeatedly imaged by: (a) FDG-PET, (b) T2W-MRI, (c) BLI, and (d) FLI. For each technique, a representative mouse is shown, and the smallest detected tumor is reported. Arrows indicate tumors. SUV, standardized uptake value.

Figure 2

Tumor growth monitored in vivo by optical imaging. (a) Signal to background ratio is similar for BLI (B16-luc, left) and FLI (B16-RFP, right) in vitro. Results are shown as the mean and SD of 4 replicate wells from 2 independent experiments. (b) B16-luc and B16-RFP tumors grow equally in vivo. Following subcutaneous injection of cell lines into mice (n = 4 mice for each cell line), tumor volume was calculated from caliper measurements. Four tumors were measured for each cell line. NS, no significant difference between B16-luc and B16-RFP (P > 0.05, test according to [13]). (c) BLI is more sensitive for tumor detection in vivo. B16-luc or B16-RFP cells were injected subcutaneously. Mice (n = 4 mice each for BLI and FLI) were shaved and imaged. Four tumors were measured for each cell line. The dotted line represents the detection threshold calculated based on control tumors not expressing the relevant reporter gene. *, significant difference between B16-luc and B16-RFP (P < 0.05, test according to [13]).

Figure 3

Mouse tissue attenuation is stronger for FLI than BLI. (a) B16-luc and B16-RFP tumors have similar specific signal ex vivo. Excised tumors from a total of 5 mice (n = 3 for BLI and n = 2 for FLI) were subjected to imaging. Data for individual tumors (n = 6 for BLI and n = 4 for FLI) and the median are shown. NS, no significant difference (P = 0.11, Mann-Whitney test). (b) Optical signal is strongly decreased in vivo for FLI on shaved mice. Tumors were scanned in vivo before excision. Data for individual tumors (n = 6 for BLI and n = 4 for FLI) and the median are shown. The tumors are the same as those described in (a). *, significant difference (P = 0.027, Mann-Whitney test). (c) Mouse hair strongly decreases the optical signal for FLI. B16-luc and B16-RFP tumors were imaged in vivo before and after shaving. A representative mouse for each technique (n = 3 and n = 2 mice analyzed for BLI and FLI, resp.) is shown.

Figure 4

Tumor optical signal correlates with tumor volume calculated from caliper measurement. Data shown are BLI detection of luciferase-expressing tumors (n = 22, left) and FLI detection of RFP-expressing tumors (n = 15, right) in shaved mice. Only tumors displaying an optical signal above the background are shown. Tumors were derived from 4 mice each for BLI and for FLI. Tumors were imaged and tumor sizes were measured at various time points. Optical signal and tumor volume were compared using Spearman correlation.

Figure 5

BLI detects both superficial and internal tumors on unshaved mice. (a) Followup of tumor growth in vivo after subcutaneous injection of B16-luc, using in vivo imaging and caliper measurement. A representative mouse of 4 mice is shown. The dotted line represents the detection threshold calculated based on control areas not expressing the relevant reporter gene. (b) Followup of tumor growth using in vivo imaging after intravenous injection of B16-luc. A representative mouse of 4 mice is shown. The dotted line represents the detection threshold calculated based on control areas not expressing the relevant reporter gene.

Figure 6

Identification of spontaneous tumors in vivo by FDG-PET. A RETAAD mouse (representative of 10 mice) was analyzed by FDG-PET, followed by necropsy. Tumors are indicated by red arrows and the numbers show the tumors identified both by FDG-PET and by necropsy. Standardized Uptake Value (SUV) is calculated as defined in Materials and Methods. Bar scale, 5 mm.

Figure 7

Identification of spontaneous tumors by T2W-MRI. A RETAAD mouse (representative of 6 mice) was analyzed by T2W-MRI, followed by necropsy. Tumors are indicated by red arrows and the numbers show the tumors identified both by T2W-MRI and by necropsy. Bar scale, 5 mm.