Synopsis

The new paradigm in oncology to tailor patient-specific therapies triggered the invention of powerful imaging techniques to non-invasively quantify tumor biology in depth. PET/MR is an emerging new hybrid imaging modality that allows the simultaneous acquisition of high resolution anatomical, functional and metabolic information.

We established a multimodal imaging workflow on a whole-body PET/MR system to quantify tumor cellularity, glucose uptake and aerobic glycolysis. We applied this workflow on a pre-clinical breast cancer model in a longitudinal study to analyze the effect of necrosis and tumor growth on metabolic parameters and measure the correlation of glucose uptake and LDH activity.

Inroduction

PET/MR (positron emission tomography/magnetic resonance) is an emerging hybrid imaging technology that is potentially superior to PET/CT (computed tomography). The system allows the acquisition of high resolution anatomical images with excellent soft tissue contrast and the quantification of cellular and metabolic data with functional imaging and MR spectroscopic imaging (MRSI).^1,2 Here, we established a multimodal imaging workflow on a clinical PET/MR system to simultaneously quantify tumor cell density with diffusion weighted imaging (DWI), glucose uptake with FDG-PET, and LDH activity with MRSI of hyperpolarized [1-¹³C]pyruvate in a pre-clinical breast cancer model.

Methods

MAT-B-III tumors were induced in eleven Fischer344 female rats by subcutaneous implantation of 1x10⁶ cells/200µL. Three animals were used to validate static and dynamic FDG uptake measured on the PET/MR (mMR Biograph, Siemens) with uptake measurements on a pre-clinical PET/CT (Inveon, Siemens). For this purpose, animals were measured on the PET/CT and ten hours later at the PET/MR.

Multimodal imaging was performed eight, ten and thirteen days after cell injection. 11±2MBq [¹⁸F]FDG were injected (tail-vein) followed by 90min PET. Meanwhile, 2D and 3D T₁- and T₂-weighted imaging, 2D-dynamic chemical shift imaging (CSI, 20 time-steps, 5.8s/time-step, resolution_nominal: 10x10x13mm³) of hyperpolarized pyruvate (90±3mM, pH7.6±0.2) and 2D-diffusion weighted imaging was performed.

PET/CT images were reconstructed in 12x10s, 6x30s, and 11x300s frames using an ordered-subsets expectation-maximization (OSEM) 3D-algorithm (2 iterations, 16 subsets). PET/MR data was reconstructed in 12x10s, 6x30s and 17x300s bins using a fully iterative OSEM 3D-algorithm (3 iterations, 21 subsets). Spatially resolved ¹³C-data were zero-filled by a factor of four in the two spatial dimensions and a factor of two in spectral dimension before Fourier transformation. Pyruvate and lactate time curves of whole tumor regions of interest (roi) were fitted to a two-site exchange model, yielding k_PL as an expression of lactate dehydrogenase (LDH) activity.³ Tumor cellularity was quantified by calculating mean apparent diffusion coefficients (ADCs) of whole tumor rois^,5. Data are represented as mean±std. Statistical analysis was performed with GraphPad prism (GraphPad Software). Correlations of quantitative parameters were calculated with a two-tailed t-test yielding Pearson product-moment coefficients (significance level: *p<0.05).

Results

Before performing a longitudinal study with the PET/MR we validated FDG uptake with measurements at a small-animal PET/CT. SUV_mean/max show no significant difference between the two systems (Figure 1). However, we observed that image derived input functions measured in the vena cava were underestimated by a factor of ~1.8, which leads to an overestimation of K_i-values obtained by Patlak analysis (not shown).

A scheme of the multimodal imaging workflow established on a whole-body PET/MR is shown in Figure 2. We acquired dynamic PET images and simultaneously measured 2D and 3D T₁- and T₂-weighted images to localize tumors for an accurate positioning of 2D axial ¹³C-CSI and 2D DWI images that were measured 60-90min after FDG application.

A Qualitative (Figure 3 A) and quantitative (Figure 3 B) comparison of static late-frame FDG images and ¹³C-MRSI data show a good positive correlation of augmented glucose uptake and elevated LDH activity (n=24). The Pearson coefficients for the correlation of SUV_mean and k_pl and the correlation of SUV_max and k_pl were 0.62 (p=0.0012) and 0.58 (p=0.0031), respectively.

We further analyzed the effect of progressive necrosis ad tumor growth on longitudinal metabolic PET and ¹³C data. Pearson correlation coefficients are displayed in Table 1. We measured increasing ADC values with growing MAT-B-III tumors, while FDG uptake decreased. A similar but not significant trend was observed with k_pl values.

Discussion

We developed a multimodal imaging workflow on a clinical system to simultaneously quantify FDG uptake, LDH activity and cell density in rodents. FDG uptake can be measured accurately at the clinical system, while activity recovery in smaller structures (e.g. vena cava) are underestimated, which we attribute to limited spatial resolution (data not shown). Notably, we show a good quantitative and qualitative correlation of augmented glucose consumption and increased LDH activity. This indicates that a major portion of consumed glucose is metabolized by oxidative glycolysis in the investigated tumor model. This metabolic abnormality is known as the Warburg effect⁶ and is one of the hallmarks of tumors⁷. Inverse correlations of DWI and FDG-PET further show that both modalities can be used to address cell density of tumors.⁸ The correlation of ADC with k_pl showed similar but not significant trends, which is probably due to a low spatial resolution of ¹³C-images.

Conclusion

We established a workflow that allows multiparametric tumor characterization at a hybrid PET/MR. Used methods can readily be transferred to bigger tumor models with potential for clinical applications paving the way for tailored and patient-specific therapy approaches.

Acknowledgements

We thank Cambridge Isotope Laboratories Inc. for providing [1-¹³C]pyruvate and appreciate support from EU Grant No. 294582 (MUMI), BMBF (FKZ 13EZ1114), DFG (SFB 824) and the DFG Major Instrumentation Initiative PET/MR.

We acknowledge Birgit Blechert and Michael Michalik for culturing cells, Sylvia Schachoff for her help at the PET/MR and Sybille Reder and Markus Mittelhäuser for performing the PET/CT measurements.

References

1 Rosenkrantz, A. B. et al. Current Status of Hybrid PET/MRI in Oncologic Imaging. American journal of roentgenology 206, 162-172 (2015).

2 Sauter, A. W., Wehrl, H. F., Kolb, A., Judenhofer, M. S. & Pichler, B. J. Combined PET/MRI: one step further in multimodality imaging. Trends in molecular medicine 16, 508-515 (2010).

3 Day, S. E. et al. Detecting tumor response to treatment using hyperpolarized 13C magnetic resonance imaging and spectroscopy. Nature medicine 13, 1382-1387 (2007).

4 Padhani, A. R. et al. Diffusion-Weighted Magnetic Resonance Imaging as a Cancer Biomarker: Consensus and Recommendations. Neoplasia. 11, 102-125 (2009).

5 Koh, D.-M. & Collins, D. J. Diffusion-Weighted MRI in the Body: Applications and Challenges in Oncology. American journal of roentgenology 188, 1622-1635 (2007).

6 Warburg, O. On the Origin of Cancer Cells. Science 123, 309-314 (1956).

7 Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: the next generation. Cell 144, 646-674, (2011).

8 Baba, S. et al. Diagnostic and prognostic value of pretreatment SUV in 18F-FDG/PET in breast cancer: comparison with apparent diffusion coefficient from diffusion-weighted MR imaging. Journal of nuclear medicine 55, 736-742 (2014).