Proton MRS of cervical cancer at 7 T

The differentiation grade of cervical cancer is histologically assessed by examining biopsies or surgical specimens. MRS is a highly sensitive technique that images tissue metabolism and can be used to increase the specificity of tissue characterization in a non‐invasive manner. We aim to explore the feasibility of using in vivo 1H‐MRS at 7 T in women with cervical cancer to study tissue fatty acid composition. 10 women with histologically proven Stage IB1‐IIB cervical cancer were scanned with a whole‐body 7 T MR system with a multi‐transmit system and an internal receive only monopole antenna. A STEAM sequence was used to obtain 1H‐MRS data. Fatty acid resonances were fitted with Lorentzian curves and the 2.1 ppm/1.3 ppm ratios were calculated. 1H‐MRS data showed fatty acid signals resonating at 2.1 ppm, 1.9 ppm, 1.5 ppm, 1.3 ppm and 0.9 ppm. Mean 2.1/1.3 ppm ratios were 0.019 ± 0.01, 0.021 ± 0.006, 0.12 ± 0.089 and 0.39 ± 0.27 for normal, Grade I, Grade II and Grade III groups respectively. Poorly differentiated tumor tissue (Grade III) showed elevated fatty acid ratios when compared with the well differentiated tumor (Grade I) or normal tissue. 1H‐MRS in cervical cancer at 7 T is feasible and individual fatty acid signals were detected. In addition, poorly differentiated tumors show more fatty acid unsaturation. The 2.1 ppm/1.3 ppm ratio has potential for tumor characterization in a non‐invasive manner for uterine cervical cancer.

fatty acids arise from novo-synthesis. 9,10 In the uterine cervix, alterations in fatty acid metabolism (the α-carboxyl (2.24 ppm) and α-olefin (2.02 ppm), observed as one peak resonating at 2.1 ppm, methylene at 1.3 ppm and methyl at 0.9 ppm) have been used to discriminate pre-invasive and invasive tumors. 11,[14][15][16][17] In addition, the methylene fatty acid resonance at 1.3 ppm has been used to differentiate between cancer and healthy control groups. 18 An important limitation of these studies was the data acquisition at 1.5 T, where fatty acid resonance frequencies overlap.
Ultra-high magnetic field strengths such as 7 T offer an increased signal to noise ratio (SNR) and spectral resolution, which can be exploited for individual detection of metabolites that overlap at lower fields. This allows exploration of fatty acid profiles and the possibility to detect otherwise overlapping fatty acids or metabolites at ultra-high field. However, the strong non-uniformities found at these field strengths, due to wavelength shortening, make the use of a body coil (based on loop coils) inefficient. Instead, external antennas with more penetrating B 1 power can be used.
The use of an additional receive only (Rx) internal antenna can further boost SNR in the region of interest (i.e. cervix). [19][20][21][22][23][24] In the particular case of uterine cervix imaging, internal Rx coils can be introduced transvaginally. However, given their limited diameter, transvaginal coils restrict the imaging to the forward view of the coil. In the case of larger tumors, transvaginal coils may have a suboptimal performance. Endorectal monopole antennas have an improved far-field performance and can cover larger regions. [25][26][27] We propose to exploit the SNR available at 7 T in combination with an Rx internal antenna to explore the feasibility of tumor characterization in uterine cervix by the fatty acid profile measured with 1 H MRS.

| MATERIALS AND METHODS/EXPERIMENT
This prospective study was approved by the institutional review board (clinicaltrials.gov: NCT02083848). Ten women with histologically proven Stage IB1-IIB2 cervical cancer were included in this study after giving informed consent. Patients were excluded when definitive therapy had already been initiated. They all underwent a 1.5 T MRI (standard) clinical examination in addition to the 7 T MR protocol. Afterwards, all patients underwent radical therapy and a final histology report was available. In addition, five healthy female volunteers were scanned after giving signed informed consent. The coil setup used for the MR examination is schematically shown in Figure 1, and consisted of seven (or eight in the volunteers' case) external fractionated dipole antennas (MRCoils, Zaltbommel, The Netherlands) 28   High resolution T 2 weighted (T2w) images (multi-slice SE, TSE factor 16, T E /T R = 70/7000 ms, 250 × 4000 × 59 mm FOV, 0.8 × 0.81 mm resolution, 3 mm slice thickness, 15 slices) were acquired in the transverse, sagittal and oblique (to the cervix) planes for anatomy localization. This T2w MRI was used for the planning of the MRS measurement. Single-voxel (SV) and CSI measurements were obtained with a stimulated echo acquisition mode (STEAM) sequence (repetition/echo times T R /T E SV,CSI = 1400/36-75, 10 ms, varying voxel size from 20 to 50 mm 3 or 30 × 30 mm 2 FOV with 5 × 5 × 5 mm 3 voxels, for the SV or the CSI case respectively, 192 signal averages and 16 phase cycles for the SV case, 2048 acquired points, 4000 acquisition bandwidth). The STEAM sequence was chosen in this study as it only requires low B 1 amplitudes and can reach shorter T E times. In the SV MRS acquisitions, the voxels were located such as to include at least 80% of the tumor. A VAPOR (variable power and optimized relaxation delays) 33 scheme with a fixed 150 Hz bandwidth was used for water suppression in all cases. The CSI grid was positioned in the tumor in all cases.
The were also fitted with the same approach. Fitted peak integrals were used to calculate the fatty acid ratios.

| RESULTS
Median age for this patient population was 38.6 years old (25-66 years Median age for the volunteer group was 27.5 years old (24-35 years). None of the volunteers had any previous history of uterine cervix disease. Therefore, all these cases were considered healthy.
High resolution T2w images were obtained successfully in all patients and volunteers for anatomy localization and for 1 H MRS planning.
The resulting spectra obtained with the internal antenna at 7 T had sufficient SNR (99 on average) to detect metabolite (including choline and creatine) and fatty acid signals. The SNR of the MRS acquired in volunteers was 54 on average, which corresponds to 55% of the SNR in patients.
The fatty acids resonating at 2.1 ppm and at 1.3 ppm were present in all data sets. Therefore, the 2.1 ppm over 1.3 ppm ratio was calculated for all groups. In addition, the measured baselines were strongly influenced by first order phase artifacts, due to sub-optimal B 0 shimming and therefore residual water signals. However, the fitting algorithm was able to fit the baselines. Average 2.1 ppm/1.3 ppm fatty acid ratios found were 0.019 ± 0.010, 0.021 ± 0.006, 0.12 ± 0.089 and 0.39 ± 0.270 (a.u.) for the normal, Grade I, Grade II and Grade III groups respectively. These ratios are increased in poorly differentiated tumors, but this is not statistically significant. Figure 2 shows the T2w images with the STEAM localization planning (yellow box) for two patients with histologically proven AC, International Federation of Gynecology and Obstetrics (FIGO) Stage IB2, Grade III and tumor size 60 cm (Figure 2A and 2B) and Stage IB1 with Grade II and a microscopic tumor that was not observed with MRI ( Figure 2C and 2D) respectively. The 60 mm tumor was also visible in the MRI at this spatial resolution. Both fitted spectra show the total fitted spectrum in red, the individual fits in gray and the residual in green. Figure 2C and 2D also shows low choline levels. In addition, the fatty acids resonating at 0.9 ppm, 1.1 ppm (as a shoulder-peak of the 1.3 ppm peak), 1.3 ppm, 1.5 ppm and 2.1 ppm are present in both cases. Only the

| DISCUSSION
In vivo 1 H MRS of uterine cervical carcinomas at 7 T to measure fatty acid profiles has been shown to be feasible for the first time. We have found that the use of an internal antenna in combination with an external array of radiating antennas provides sufficient SNR for MRI and STEAM MRS acquisitions at 7 T. The inclusion of internal antennas to enhance SNR in MRI has been already explored at lower magnetic field strengths to image the uterine cervix, with positive results, 15,19,20 and has been validated in this pilot study at the field of 7 T. The positive effect of this internal monopole antenna has been recently investigated at our institution for the same patient population. 27 In particular, the quality of the T2w MRI obtained for all patients was found to be superior to the T2w MRI routinely acquired at 1.5 T. 27 The MRS results found in this study agree with previous published MRS studies at lower magnetic field strengths, where mainly fatty acid signals were observed in uterine cervical ACs. 11,[14][15][16][17][18]36 The fatty acid oxidation pathway in cancer cells has been recently given more attention and proven to be as important in cancer metabolism as the Warburg effect. Tumor cells prefer the fatty acid oxidation pathway as a source of energy, which can come from either external or newly formed (i.e. from novo-synthesis) fatty acids, which are oxidized and stored as lipid droplets in the tissue. 12,13 Therefore, it seems reasonable to characterize the lipid composition in cancer and its relationship to disease, if any, with 1 H MRS. The the healthy contra part. 37 Finley et al. observed in serums that a higher aggressiveness in prostate cancer tissue presented elevated peri-prostatic adipose tissue. 38 Fatty acid content differences in non-alcoholic liver disease, which is the precursor of liver cancer, have also been found between the different stages of the disease. 39,40 All these points stress the relevance of fatty acid profiling in diseased tissue.

FIGURE 3
Zoomed-in T2w images are shown for three different patients with SCC and Grades III, II and III respectively. Volume localization for the STEAM acquisition planning is shown in yellow. Two spectra (D, F) had detectable (black) and fitted (red and gray) levels of choline as well as fatty acids. Metabolites observed and fitted are labeled as 1, choline, 2, creatine, 3, 2.1 ppm, 4, 1.9 ppm, 5, 1.5 ppm, 6, 1.3 ppm, 7, 1.1 ppm, and 8, 0.9 ppm. Overlapped in red is the total fit, in gray the individual fits and in green the residual The increased spectral resolution available at 7 T enabled depiction of fatty acid peaks that overlap at lower fields. Choline and creatine were detected and fitted in tumor grades II and III.
A limitation of using the methylene 1.3 ppm peak for ratio calculation is that it overlaps with the methyl resonance of lactate, which also appears at 1.3 ppm and is typically present in tumors. [41][42][43] Therefore, the fitting of this fatty acid resonance can be somewhat overestimated.
Moreover, the presence of microscopic intra-and extra-cellular lipids could cause a resonance shift due to their differences in bulk magnetic susceptibility, which could explain the 1.1 ppm shoulder on the 1.3 ppm resonance. Some of these effects could be overcome using editing techniques such as simple editing or double quantum coherence schemes. 44,45 Although the STEAM sequence has been used extensively in vivo at different magnetic field strengths in many different in vivo investigations, it still suffers from low SNR due to the necessity of using strong dephasing gradients between three 90°RF pulses, which lead to signal loss. This was somewhat compensated in our study by the inclusion of the internal Rx antenna in patients and the additional receivers in the volunteers. In addition, the STEAM RF pulses are sensitive to B 1 non-uniformities. Therefore, in extensive 2D or 3D CSI volumes the effect of the RF pulses over the whole volume could vary, particularly at higher field strengths, which may lead to incorrect metabolite concentrations and phase artifacts. The calculation of metabolite ratios corrects for these differences. In addition, the poor profiles of the STEAM pulses can introduce spurious chemical shift artifacts, mainly from water in the surrounding tissues, even when water suppression schemes are included. Optimization of the MRS sequence for future studies would be favorable.

| CONCLUSION
We were able to obtain lipid profiles in cervical carcinomas with 1 H MRS at 7 T. For the first time, it was observed that the 2.1 ppm/1.3 ppm fatty acid ratio might be associated with tumor grade in cervical cancer, as seen by the increased unsaturation in poorly differentiated tumors. Therefore, this ratio may have the potential of characterizing tumor grade in a non-invasive manner to aid clinical diagnostics.

FUNDING INFORMATION
STW Technology Foundation (grant no 10822).