We have demonstrated the feasibility of using the NIR dye labeled 2-deoxyglucose, IRDye800CW 2-DG, for in vivo fluorescence imaging of orthotopic glioma in a mouse model (). Ex vivo fluorescence imaging of tumor sections and microscopic imaging confirmed significantly higher accumulation of the 2-DG dye in intracranial tumors than in normal brain. Good correlations were found between each imaging modality in terms of in vivo evaluation of intracranial tumor burden ( and ).
Small animal imaging has been increasingly adapted for preclinical cancer research. In vivo
imaging promises greater efficiency since each animal serves as its own control and multiple time points can be examined sequentially. In particular, multimodal imaging approaches provide comprehensive information about both tumor anatomy and pathophysiology and even molecular mechanisms 
. In this study, we combined optical imaging (bioluminescence and fluorescence imaging) with MRI to study longitudinal development of intracranial tumors and their uptake of a glucose analogue, 2-DG. There was good agreement between increased BLI signal intensity over time and enlarged tumor volume measured by MRI, as also reported by others previously for intracranial tumors in rodent models 
. Thus, the cheap, fast and high-throughput BLI seems to be just as effective in monitoring the deep-seated orthotopic tumor models as the expensive and time consuming MRI. However, BLI does not provide anatomic details due to the poorer spatial resolution and limited depth of light penetration. Recent progress in BLI tomography should facilitate three-dimensional analysis. In addition to providing detailed anatomic structure, MRI is also useful to obtain pathophysiological information, i.e.
, tumor vascular perfusion and permeability, oxygenation, and apoptosis or necrosis, though not reported here.
Fluorescence imaging can be used to evaluate small reporter molecule pharmacokinetics, avoiding the need for genetic modification of cells. In particular, near infrared fluorescence (NIR) imaging has found a greater potential for clinical application because of its long wavelength (650–900 nm), where light absorbance and scattering are significantly lower, and autofluorescence of normal tissues is also greatly reduced 
. In view of this approach, various target-specific NIR conjugates have been reported for targeting tumor imaging, e.g.
, tumor integrin, αv
, tumor growth factors or their receptors 
, glycoprotein 
, or tumor specific protease 
. Successful applications in various preclinical tumor models have been reported, though most of these studies were performed on surface tumors. In vivo
NIR imaging of deep-seated orthotopic tumor models, in particular, intracranial tumors, remains challenging. By targeting the overexpressed αv
in tumor, Hsu, et al.
recently reported visualization of orthotopic glioma of mouse in vivo
by NIR imaging via Cy5.5-RGD. Peak of tumor uptake was found 2 h post injection and TNR
2.6 was achieved with a craniotomy 
. More recent work with an orthotopic mouse brain tumor McCann, et al.
successfully applied fluorescent molecular tomographic imaging to monitor protease activity in tumor by using protease activatable fluorescence, ProSense680 (peak light emission at 680 nm). By co-registering fluorescence images with MRI, localization of the active protease in tumor was determined 
Here, we applied the commercially available NIR labeled 2-deoxyglucose to imaging intracranial tumors based on the simple mechanism of differential levels of glucose metabolism between tumors and normal tissues. Tumor cells both require more energy for their higher proliferation rate and utilize the inefficient glycolytic pathway to produce energy. Thus, tumor cells need more glucose compared to normal cells. On this basis, 18
FDG has been used as the most common PET radiotracer to visualize clinical tumors and their metastases. However, 18
FDG PET imaging of brain tumor is often compromised by strong background signals of normal brain tissues. Due to the short half life of fluorine-18 (<2 h), PET imaging is normally performed within 1 h after injection of 18
FDG. The lower contrast between tumor and normal brain may be partly attributed to this timing, at which a maximum ratio of uptake is not reached. Indeed, we found no significant difference in light signal in the tumor side of the brain versus the normal side of the brain during the first 4 hours post injection. The peak tumor to normal ratio was actually observed at 24 h, which is consistent with a previous study of subcutaneous U87 tumors using cy5.5 labeled D-glucosamine. However, a control dye, cy5.5-NHS used in that study also produced a high TNR 
. Our study is also in a good agreement with another study that detected essentially no retention of IRDye800CW 2-DG in normal brain after 24 h assessed by ex vivo
. In addition to tumor diagnostic imaging, pyropheophorbide labeled 2-deoxyglucosamine has shown a potential for photodynamic therapy on tumors 
Our results of ex vivo
fluorescence imaging showed that the 2-DG dye distributed well into the whole tumor despite some heterogeneity (). This observation may suggest delivery and distribution of the 2-DG dye to the tumor does not need the BBB disruption, which is the prerequisite condition for the dyes currently used for neurosurgery. Indeed, uncoupling of tumor vascular perfusion and permeability and uptake of FDG has been reported previously 
. However, it is still possible that the dye first leaks out through the disrupted BBB, from which it diffuses into the whole tumor. Further studies will be necessary using earlier stage of tumors, when the Gd contrast leakage is not obviously seen by T1
-weighted contrast enhanced MRI, to prove this hypothesis. Furthermore, the usefulness of the 2-DG dye to stain the infiltrative tumor cells is limited by the tumor model used in this study. The U87 tumor has relatively sharp tumor-brain boundaries. GBM models showing more aggressiveness will be needed to test the ability to stain the finger-shaped infiltrative tumors. Instead of the established GBM cell lines, using surgically resected tumor tissues directly from GBM patients and passing them in animals will generate stable orthotopic GBM xenografts, which show oncogene patterns very similar to primary tumors of patients.
In summary, fluorescence imaging of deoxyglucose uptake in more clinically relevant orthotopic glioma models has not yet been reported. Our results may suggest an optimal time for imaging brain tumors based on the glucose analogues. The near-infrared dye labeled 2-DG may serve as a novel imaging probe to noninvasively monitor intracranial tumor burden in preclinical animal models. From a clinical perspective, development of NIR labeled tumor-specific molecules may have the potential to identify the infiltrating glioblastoma intra-operatively, and to improve the extent of resection of selected tumors.