Choline has been shown to be elevated in cancers as a consequence of increased cell membrane turnover. Thus, intracellular choline measured using MRS may be used as a marker to identify prostate cancer and monitor therapeutic response¹. Choline is also administered extrinsically for imaging prostate cancer: PET studies using ¹¹C-choline and ¹⁸F-labelled choline derivatives² rely on the increased uptake of these agents by cancers to delineate the lesion. In this study, we compared steady-state concentrations of total choline as measured by ¹H-MRS with the uptake of ¹⁸F-Choline on PET in a cohort of 11 prostate cancer patients. The purpose was to interrogate the relationship between intracellular choline and extrinsic choline uptake in prostate cancer.

Patient cohort: Prior to radiotherapy, 11 prostate cancer patients underwent ¹H-MRS and ¹⁸F-Choline PET-CT scans less than 2 weeks apart. Gleason score was: 3+3 (n=1), 3+4 (n=3), 4+3 (n=6), and 4+4 (n=1).

Data acquisition: MRS was performed on a 1.5T Siemens Avanto using an endorectal coil inflated with 60ml of air combined with an external phased-array pelvic coil. T₂-weighted images were used to position the PRESS-localised 3D chemical shift imaging (CSI) grid over the whole prostate. Metabolite spectra (using water/lipid suppression) were acquired with the following parameters: TR/TE = 700/120, vector size 512, 5 averages, 16x16x16 phase encode steps, voxel size 3.75x5.25x4.5mm³ (0.088 ml), 8min acquisition time. Unsuppressed water spectra were acquired from exactly the same location, but with a TE=30ms and 1 average.

PET/CT data were acquired using a Siemens Biograph mCT S128. Patients were fasted for 4 h before a mean ¹⁸F-ethylcholine activity of 282MBq (range 179-326MBq) was injected intravenously. At 60min post-injection a half-body PET/CT scan (skull base to upper thighs) was performed. A delayed pelvic PET/CT acquisition was obtained at 90min post-injection. The co-registered unenhanced CT scans were performed for anatomical localisation and attenuation correction. Semiquantitative image analysis of PET/CT data used Hermes software (Hermes Medical Solutions, Stockholm, Sweden) to draw regions-of-interest and derive standardized uptake values.

Data analysis: Up to 6 tumor voxels within the dominant lesion were selected for MRS analysis for each patient (2 voxels/slice for 3 slices, where available) based on T₂-weighted MRI. Voxel-derived data were processed using jMRUI software³.The peak area ratio between total choline and water was calculated for each voxel, averaged per patient, and multiplied by 2/9 to correct for the number of equivalent spins in the measured peaks. While it is common practice to include only data with low relative values for the estimated Cramer Rao Lower Bounds (CRLB) this can lead to bias by discarding data with low metabolite concentrations⁴. In this study, an estimate of reasonable CRLB was obtained from voxels with CRLB/amplitude <25%, and this absolute value was used as an additional limit for retaining data for voxels with CRLB/amp >25%. No corrections for T₁ and T₂ were performed.

The reported PET parameter was the maximum of the standardized uptake value (SUV_max) measured over the whole volume of the dominant lesion and normalized by the mean of the SUV observed for normal prostate tissue (1ml volume) for both early and delayed acquisitions.

Results of choline measurement using both modalities are quoted as mean ± standard deviation. Data co-registration was achieved by cognitive fusion of the MRI/S and PET data by an experienced radiologist. The relationship between MRS-measured Cho/water ratio and normalized SUV_max of ¹⁸F-Choline PET was investigated using Pearson correlation test.

Figure 1 shows an example of an MR spectrum (right) obtained from the voxel highlighted in the bottom left panel (note the zoomed-in prostate coverage) and corresponding ¹⁸F-Choline uptake (top left). Good quality MRS data were acquired for both water and metabolite spectra. Linewidth for the water acquisitions was less than 10Hz for all spectra. The main metabolites (choline, creatine, citrate) were reliably detected in most tumor spectra (only 5/43 fits were excluded from final analysis). The measured choline/water ratio (x10^-3 units) in tumor varied between 0.05 and 0.12 (mean 0.09±0.02) while the normalized 18F-Choline uptake covered a range from 1.36-4.57 (mean 2.69±1).

PET and MRS results are summarized in figure 2. No significant difference was observed between early or delayed PET measurements. The Pearson correlation test found a weak negative correlation between the measured MRS and PET parameters for early PET acquisition (-0.61, p=0.04).

These results suggest that the uptake of externally administered ¹⁸F-Choline is larger in tumors with lower total intracellular choline concentration, indicating increased avidity of tumors for choline when internal concentrations are low.