Background:
Pancreatic ductal adenocarcinoma (PDAC) is the third leading cause of cancer death in the United States, with five-year survival below 13%¹. One-third of PDAC deaths are attributable to cachexia, a wasting syndrome in which patients lose muscle and adipose mass and become too weak for therapy². No FDA-approved therapy exists for PDAC cachexia, and its severity does not correlate with tumor size, implicating the peritumoral microenvironment as a possible initiating driver³,⁴. During PDAC progression, peritumoral white adipose tissue (WAT) can undergo tumor-induced browning, with uncoupled thermogenesis and creatine cycling depleting systemic energy reserves⁵⁻⁷. No non-invasive tool currently detects this metabolic conversion before irreversible deterioration occurs. Creatine chemical exchange saturation transfer (CrCEST) MRI is a probe-free, radiation-free molecular imaging method that maps tissue creatine on standard scanners and is sensitive to thermogenic creatine cycling in brown adipose tissue (BAT)⁸.
Hypothesis/Objective:
Tumor-induced adipose browning produces a detectable CrCEST metabolic signature at the tumor-adipose interface. The objective was to characterize tumor and adipose CrCEST phenotypes across subcutaneous PDAC models implanted in BAT versus WAT environments, with genetic validation by creatine transporter (SLC6A8) and UCP1 knockout.
Methods:
C57BL/6 wild-type, SLC6A8 knockout (creatine-transporter deficient), and UCP1 knockout (thermogenesis deficient) mice received subcutaneous syngeneic KPC-derived (KPC-K8484; 1×10⁵ cells) PDAC allografts into either the interscapular BAT depot or the flank WAT environment. Animals were imaged once tumor volume exceeded 50 mm³ by anatomical MRI; those below threshold were excluded. All imaging was performed on a 9.4T Agilent scanner. Anatomical volumes used 2D T2-weighted fast spin-echo (RARE, ETL 8). Keyhole-accelerated CEST FSE employed a high-resolution reference at -100 ppm (128×128, 6 averages) followed by 61 keyhole offsets (ETL 64, 128×64, zero-filled reconstruction). Normalized cross-correlation motion rejection excluded respiratory-corrupted frames. Z-spectra were fit with a five-peak Lorentzian model generating fat-fraction, creatine (Cr_Amp at ~2 ppm), amide proton transfer (APT, ~3.5 ppm), water, and magnetization transfer maps. Voxels were retained only if R²≥0.8. Tumor, tumor boundary, adjacent adipose, and muscle ROIs were manually segmented from anatomical images, and ROI-level Lorentzian amplitudes reported. UCP1, Ki67, and Oil Red O staining validated adipose phenotype and tumor proliferative activity.
Results:
NOE amplitude reliably distinguished adipose (BAT 32.3 ± 1.1%; WAT 27.4–29.3%) from tumor (7.1–7.6%) and muscle (<12%), providing a consistent anchor for adipose-tumor interface segmentation. In WT BAT-adjacent tumors, CrCEST was higher in tumor (1.73 ± 0.10%) than in adjacent BAT (0.90 ± 0.28%), with tumor APT at 1.68 ± 0.02%. In the WT flank WAT model, peritumoral WAT exhibited elevated CrCEST (3.49%) and APT (4.50%), a signature consistent with tumor-induced WAT browning. In the SLC6A8 KO animal, peritumoral WAT CrCEST fell to 1.66% (52% reduction vs. WT flank WAT), consistent with transporter-mediated creatine import in browning WAT. In the UCP1 KO animal, BAT CrCEST fell to 0.49% and tumor APT fell to 0.76% (>2-fold lower than WT BAT-adjacent tumors), consistent with thermogenesis-dependent creatine cycling and attenuated tumor proliferative signal in the absence of BAT-driven metabolic support. Histology (Ki67, Oil Red O, UCP1) confirmed proliferative activity and adipose phenotype.
Conclusion:
Multi-parametric CEST MRI resolves distinct creatine and APT signatures across the PDAC tumor-adipose interface, with modulation by adipose type and by genetic perturbation of creatine transport and thermogenesis. Because CrCEST is probe-free and clinically translatable on standard scanners, these findings support its further development as a non-invasive biomarker of tumor-adipose metabolic crosstalk and early cachexia risk in PDAC.
Image/Figure:
Click to view full size
Click to view full size
Image/Figure Caption:
Figure 1. Multi-parametric CEST MRI of adipose-tumor metabolic crosstalk across PDAC models. (A-D) Leftmost column: anatomical T2-weighted reference with tumor and adipose depots marked by arrowheads. Remaining columns: overlaid parametric maps of lipid (NOE), creatine, APT, and water. (A) WT mouse with KPC-K8484 tumor implanted in the interscapular BAT depot. (B) WT mouse with flank tumor adjacent to WAT. (C) Creatine transporter knockout (SLC6A8-/y) with interscapular tumor. (D) UCP1 knockout with interscapular tumor, showing reduced creatine and APT signal across BAT consistent with loss of thermogenic creatine cycling. (E-H) Representative five-peak Lorentzian Z-spectrum fits for Tumor (top row) and Adipose Boundary (bottom row) ROIs from (E) WT flank-implanted, (F) WT BAT-implanted, (G) creatine transporter KO (SLC6A8-/y), and (H) UCP1 KO animals. Red dashed line marks APT (~3.5 ppm); green dashed line marks the creatine offset (~2 ppm). Inset values report ROI-level APT and Cr Lorentzian amplitudes.
Author
Victoria Laney
University of Illinois Chicago