Introduction

Pancreatic ductal adenocarcinoma (PDAC) has poor clinical survival, with urgent need for biomarkers for tumour differentiation. Multimodal imaging, including measures of glucose metabolism and cellularity, has the potential to non-invasively assess tumour metabolic subtype and intra- and inter-tumoural heterogeneity to guide and monitor treatment.

Methods

Murine PDAC tumours were imaged with diffusion weighted MRI (DWI) and hyperpolarized ¹³C-pyruvate-lactate spectroscopic imaging (CSI), and with ¹⁸F-FDG PET.

Tumour models: Murine epithelial (N=3) and mesenchymal (N=4) PDAC cell lines were orthotopically transplanted in mice (CrTac:NCr-Foxn1^nu). Separately, CKP_lox mice (Ptft1a^wt/cre; Kras^wt/G12D; p53^fl/fl) developed endogenous PDAC tumours (N=23 mice, with 29 histologically correlated regions classified epithelial by histology, with mesenchymal and anaplastic excluded from analysis due to low numbers in this cohort).

Imaging Systems: Small animal 7 T preclinical scanner (Agilent/GE magnet, Bruker AVANCE III HD electronics) using a dual-tuned ¹H/¹³C volume coil (31 mm ID; RAPID Biomedical). Small animal PET/CT scanner (Siemens Inveon).

MR Imaging: For all tumours, T2-weighted anatomical RARE, DWI, and ¹³C CSI were acquired. DWI used an EPI readout with 12 b-values up to 1528 s/mm² and 0.25x0.25x1.0 mm³ voxels and fit for the apparent diffusion coefficient (ADC) in MatLab. [1-¹³C]pyruvate was hyperpolarized (Oxford instruments HyperSense DNP), injected (80 mM), and imaged with multi-frame CSI with 2x2x3 mm³ voxels and 5 s / frame. Ratios of the area under the curves (AUCr) of lactate to pyruvate spectral peak time-courses were calculated.

PET Imaging: For orthotopic tumours, 15 min static [¹⁸F]-FDG PET images were acquired 1 hour post-injection. Mean standardized uptake values (SUV) were calculated within ROIs derived from coregistered anatomical MRI.

Analysis: Mean ADC, AUCr, and SUV were calculated within regions of interest (defined manually based on histopathological evaluation of H&E staining). Necrotic areas were excluded. Statistical group comparisons and correlations were calculated in GraphPad PRISM.

Histology: Endogenous tumours were extracted, fixed in formalin, and H&E stained and classified as epithelial, mesenchymal, or anaplastic by experienced pathologist (KS). Epithelial tumours were characterized by growth pattern¹ (classical ductal, papilliform, small ductal, or solid squamous-like) and by tumour cellularity (low, medium, or high)².

Results

Orthotopic tumour imaging (Fig. 1) showed no significant differences in ADC between epithelial and mesenchymal groups (Fig. 2). Mean SUV and ¹³C AUCr were significantly different and positively correlated (Fig. 2).

Endogenous epithelial tumours were grouped as classical ductal with low tumour cellularity, papilliform with medium tumour cellularity, or small ductal or solid with medium or high tumour cellularity. Tumour histology examples are shown in Figure 3.

ADC and AUCr values grouped by tumour cellularity and growth pattern and compared with ANOVA or t‑tests are shown in Figure 4. Considering all tumours, ADC was separated well by tumour cellularity, and between classical ductal and papilliform, small ductal, or solid tumours, but not between the latter three. AUCr was separated between high and med or high and low tumour cellularity, between solid and classical ductal, and marginally between solid and papilliform, but not between other pairs.

Endogenous tumour ADC and AUCr values are also correlated (Fig. 5). Plotting separately the small ductal and solid tumour data (those with substantial variation within the growth pattern group in Fig. 4) shows correlation within the solid tumour group as well.

Discussion

These results indicate that mesenchymal tumors rely more on glucose metabolism than do epithelial tumours. The positive correlation of SUVmean and AUCr in orthotopically implanted tumours suggests that these methods provide similar information about tumour glucose metabolism. The variation of AUCr (and SUVmean) indicate that ¹³C metabolic imaging has potential to distinguish between tumours of similar cellular density.

Combined DWI and hyperpolarized ¹³C CSI differentiates between endogenous PDAC tumours of different tumour cellularity and growth pattern classification. Classical ductal tumours have distinctly low AUCr and higher ADC, while solid and small ductal tumours have low ADC and may have high AUCr. Particularly within solid tumours, ADC and AUCr are inversely correlated, suggesting combined DWI and ¹³C metabolic imaging may help correct metabolic results for variations in tumour cellularity.

Not all tumour types observed could be separated with ADC and AUCr; papilliform and small ductal tumours had similar presentation, and may require additional or alternative biomarkers to identify.

PET imaging was omitted for endogenous tumours in part because the orthotopic results were highly correlated, indicating minimal additional benefit to acquiring both modalities.

Conclusions

Mesenchymal tumors strongly rely on glucose metabolism in comparison with epithelial tumors. The combination of diffusion-weighted and ¹³C-pyruvate-lactate spectroscopic imaging can characterize endogenous mouse pancreatic tumours better than either method alone. Classical ductal tumours are well separated from others by their high ADC and low lactate to pyruvate AUC ratio. Solid tumours have an inverse correlation between ADC and AUC ratio, suggesting the combination of these imaging methods can be used to correct metabolic results for cell density.