[¹⁸F]Fluorocholine PET/CT Imaging of Liver Cancer: Radiopathologic Correlation with Tissue Phospholipid Profiling

[¹⁸F]fluorocholine PET/CT can detect hepatocellular carcinoma (HCC) based on imaging the initial steps of phosphatidylcholine synthesis. To relate the diagnostic performance of [¹⁸F]fluorocholine positron emission tomography (PET)/x-ray computed tomography (CT) to the phospholipid composition of liver tumors, radiopathologic correspondence was performed in patients with early-stage liver cancer who had undergone [¹⁸F]fluorocholine PET/CT before tumor resection.

Tumor and adjacent liver were profiled by liquid chromatography mass spectrometry, quantifying phosphatidylcholine species by mass-to-charge ratio. For clinical-radiopathologic correlation, HCC profiles were reduced to two orthogonal principal component factors (PCF1 and PCF2) accounting for 80 % of total profile variation.

Tissues from 31 HCC patients and 4 intrahepatic cholangiocarcinoma (ICC) patients were analyzed, revealing significantly higher levels of phosphocholine, CDP-choline, and highly saturated phosphatidylcholine species in HCC tumors relative to adjacent liver and ICC tumors. Significant loading values for PCF1 corresponded to phosphatidylcholines containing poly-unsaturated fatty acids while PCF2 corresponded only to highly saturated phosphatidylcholines. Only PCF2 correlated significantly with HCC tumor-to-liver [¹⁸F]fluorocholine uptake ratio (ρ = 0.59, p < 0.0005). Sensitivity for all tumors based on an abnormal [¹⁸F]fluorocholine uptake ratio was 93 % while sensitivity for HCC based on increased tumor [¹⁸F]fluorocholine uptake was 84 %, with lower levels of highly saturated phosphatidylcholines in tumors showing low [¹⁸F]fluorocholine uptake.

Most HCC tumors contain high levels of saturated phosphatidylcholines, supporting their dependence on de novo fatty acid metabolism for phospholipid membrane synthesis. While [¹⁸F]fluorocholine PET/CT can serve to identify these lipogenic tumors, its imperfect diagnostic sensitivity implies metabolic heterogeneity across HCC and a weaker lipogenic phenotype in some tumors.