The Tumor Cancer Genome Atlas (TCGA) analysis of gene and protein expression has confirmed major alterations in glucose, amino acid and lipid metabolism in clear cell renal cell carcinoma (ccRCC), the most common type of RCC, which is characterized by histologically prominent storage of glycogen and lipids (1). ccRCC expresses high levels of enzymes necessary to produce fatty acids and other lipids, and two of these, fatty acid synthase (FAS) and stearoyl-CoA desaturase (SCD1), are associated with poor prognosis (2,3). These observations strongly suggest that reprogrammed lipid metabolism in ccRCC might provide not only biomarkers of oncological aggression, but perhaps therapeutic targets as well. Thus, exploring non-invasive MRI methods as predictive metabolic biomarkers in ccRCC is appealing. However, ccRCC is characterized by intratumoral molecular heterogeneity (4), which likely drives the biological behavior of this disease. Moreover, to our knowledge, the heterogeneity of lipid accumulation in ccRCC has not been previously studied. The purpose of this study was to assess intratumoral heterogeneity of lipid accumulation in ccRCC in vivo and to correlate MRI-based fat fraction measures with lipid accumulation at histopathology and the lipidomic profile in the same tumor.Patients and MRI Protocol: This was a prospective, IRB-approved, HIPAA-compliant study. 36 patients with pathologically confirmed ccRCC (after nephrectomy) signed informed consent and underwent 3T dual-transmit MRI with a 16-channel SENSE-XL-Torso coil (n=23, Achieva; n=13, Ingenia, Philips Healthcare, Best, the Netherlands). Axial and coronal T2-weighted single-shot turbo spin-echo images (T2WI) were acquired to localize the tumor followed by axial breath-held three-dimensional (3D) T1-weighted (T1W) fast field-echo (FFE) multi-echo DIXON (mDIXON)(TR/TE = 6.7~8/1.09~1.24 ms, ΔTE = 0.9~1.1ms, 6 echoes, FA = 2~3°, NSA = 1, thickness = 3mm, in-plane resolution = 1.5mm×2mm, acquisition FOV = 402×240×96 mm2, acquisition matrix = 268×120×32, bandwidth = 1413.2 Hz/pixel, acquisition time= 15~19 seconds). Image Analysis: Fat fraction (FF) maps were reconstructed based on a 7-peak spectral modeling (5). Using a DICOM viewer (OsiriX), up to 8 (4 for small lesions) random regions of interest (rROIs) were placed within each tumor on the FF map to assess tumor heterogeneity; half in a superior slice and the other half in an inferior slice through the mass. In addition, targeted ROIs (tROIs) were placed on representative locations of the tumor including an area with high lipid content (when present) on the FF map. Histopathology: After partial (n=23), radical (n=12) or simple (n=1) nephrectomy, tumor specimens were anatomically oriented with the use of fiducial markers placed during surgery and then bivalded through the center to match the location of the MRI acquisitions. Tumor slides were obtained from the selected cutting location and stained with hematoxylin-eosin for assessment of nuclear (Fuhrman) tumor grade. A total of 17 fresh tumor samples matching the location of at least one tROI were obtained from 14 patients (2 samples from one patient, 3 samples from one patient) and stained with Oil Red O. The percentage of cells that stained with Oil Red O was estimated visually to the nearest percentage by a uropathologist who was blinded to the FF results. Additional fresh tissue samples matching tROIs were used for lipidomic analysis with gas chromatography-electron capture negative ionization-mass spectrometry. Statistics: Mean FF and standard deviation (SD) for rROIs were correlated to tumor grade. Spearman correlation and linear regression model were used to assess the relationship between FF, Oil Red O percentage and lipidomic results (P<0.05 considered statistically significant).Fig.1 shows representative images of one tumor and placement of 8 rROIs. We found heterogeneous lipid accumulation both among different tumor grades and within each tumor (Fig.2). Levels of lipid accumulation and heterogeneity (SD) were not correlated with tumor grade or size (P>0.05). FF measures in tROIs correlated statistically with Oil Red O percentage (ρ=0.90, P<0.0001) (Fig.3). Fig. 4 shows representative images in a tumor with low fat content and a tumor with high fat content, tROI in each tumor, and correlative Oil Red O stain in the same location of the tumor. FF measures correlated positively with increased triglycerides (TG) (P=0.0007) and free cholesterol (P=0.006) counts in lipidomic analysis of fresh tissue samples in the same location of the tumor.Dixon-based MRI offers an opportunity to characterize renal tumors in vivo based on lipid accumulation. Furthermore, Dixon-based MRI-directed targeted tissue sampling provides a platform for understanding intratumoral heterogeneity in lipid metabolism that may be leveraged to identify potential therapeutic targets.Quantitative Dixon-based MRI allows for noninvasive assessment of intratumoral heterogeneity of lipid metabolism in ccRCC.