Understanding tumor metabolism may aid in proper diagnosis and assessment of treatment response¹. The gold standard of positron emission tomography (PET) using ¹⁸F labeled 2-fluoro-2-deoxyglucose (FDG) utilizes the difference in glucose uptake of tumors and normal tissue to enhance the contrast of tumors². However, new techniques have emerged that yield complementary metabolic information. Optical fluorescence lifetime imaging (FLIM), probes the chemical states of metabolites like nicotinamide adenine dinucleotide (NADH) by comparing the relative lifetimes of their fluorescence in cells^3,4. In addition, magnetic resonance spectroscopy (MRS) of hyperpolarized ¹³C nuclei provides atomic information about pyruvate metabolism and lactate production⁵. Combining these techniques in a multi-modal platform would unite three imaging scales (Figure 1) and provide a more complete picture of tumor metabolism. Here, we present our work to develop multi-modal and multi-scale imaging platforms that are both in vitro, in a bioreactor, and in vivo, in a mouse model. Breast cancer has been well classified with the above techniques and was thus chosen as the cancer model.

The bioreactor design (Figure 2) for in vitro FLIM and MRS included precise temperature control, cell culture environmental control (media infusion, oxygen environment, etc.) and an optical imaging window. Temperature was controlled with flowing warm water through a channel that runs around the cell culture and an MR-compatible temperature probe fed into the cell culture volume (Figure 3A-B,4A-B). The optical window of the bioreactor was held in place with glass-filled nylon screws and sealed with two silicon gaskets (Figure 3D). The shape of the bioreactor allows for a ¹³C surface MR coil (Doty Scientific, Inc.) to conveniently fit around the sample volume for improved signal sensitivity.

Several in vitro experiments were performed to test the bioreactor. First, enzyme reactions tested the spectroscopic measurements consisting of approximately 20-U of lactate dehydrogenase (LDH, 5µL by volume of L-LDH from bovine heart, Sigma) and varying amounts, 12, 24 and 36µmol of NADH (Figure 4C-D) in 1mL Tris Buffer. 32µmol of [1-¹³C] pyruvic acid was polarized to ~20% and injected through ports on top of the bioreactor. Dynamic ¹³C spectra were acquired (global RF excitation, 5° FA, 10kHz BW) with a 3s temporal resolution. ¹³C-enriched pyruvic acid was hyperpolarized (HP) using a HyperSense DNP Polariser (Oxford Instruments). To demonstrate the viability of the optical window, an image of GFP-expressing MDA-231 human breast cancer cells⁶ was taken through the window (Figure 3C).

The mouse model used for in vivo experiments was the polyoma virus middle T oncoprotein (PyVT) breast cancer model⁷. In vivo optical imaging was performed through a subcutaneously implanted optical window over a growing mammary tumor (Figure 5A). The window consists of a small custom plastic frame, fabricated using a 3D printer, and a small cover glass. Materials used were MR compatible. All animal experiments complied with institutional IACUC regulations with the [PyVT] mouse anesthetized with 1.5% isoflurane. A 4.7T small animal MR system (Agilent Technologies) and an Inveon Hybrid microPET/CT (Siemens) were used for MR and PET acquisitions, respectively.

For the in vitro enzyme reactions, the time-averaged spectra showed an increase in lactate signal as a function of NADH (Figure 4D), as expected. Once temperature control (37^oC) was incorporated, optimal enzyme kinetics allowed for an enhanced ¹³C lactate signal (Figure 4D). A proof of concept in vivo multi-modal and multi-scale experiment was performed. Optical microscopy data related to extra-cellular matrix (second harmonic generation (SHG)) and metabolism (intrinsic fluorescence from flavin adenine dinucleotide (FAD) and NADH) were acquired in a mammary tumor through an implanted optical window (Figure 5C) and an ¹⁸F-FDG-PET scan with anatomy provided by T₁-weighted gradient echo and T₂-weighted fast spin echo MR scans (Figure 5B) were acquired in the same mouse for a total image session of 3hr. Ongoing work will focus on two aims: adding media flow to the bioreactor to enable its use for cell culture studies and adding metabolic imaging techniques, HP ¹³C MRS and FLIM, to our in vivo studies. The completion of this work will yield two novel platforms capable of providing new insights into how tumor metabolism behaves at different imaging scales within both cell culture and the tumor microenvironment.Feasibility is demonstrated for a system, including workflow, for spatially registered optical microscopy, MRI, and PET of breast tumor metabolism in vivo. The system is coupled with a bioreactor designed to compare cellular metabolism in vitro using both optical microscopy and MR spectroscopy with the macroscopic behavior of tumor cells in the in vivo tumor microenvironment.