Single-Voxel MR Spectroscopy of Gliomas with s-LASER at 7T

Round 1

Reviewer 1 Report

 

In this paper Single-voxel MR spectroscopy of gliomas with s-LASER at 7T, the authors describe a methods for using MR spectroscopy at 7T in a series of 7 patients with low grade gliomas and 7 control patients.

Overall, the topic is interesting and represents a useful reference for moving spectroscopy forward at 7T in tumor patients. The fitting of multiple spectra at high field is a useful demonstration. Limitations are a small patient size, but it remains a useful introductory cohort. There are a few concerns: 

 

1.       The choice of the precuneus as a reference is somewhat problematic, as the spectra of different regions of the brain can differ simply based on the location. It might have been more useful to place the spectrum in a contralateral region of normal brain. This may explain greater than expected variation in ratios. This should be further described in the discussion as a limitation.

2.       The discussion is overly focused on minor differences between the spectroscopy results in this paper and others. While these provide important context, please provide a clearer description of what the advance is in this paper and why it is valuable.

3.       Some of the tables are confusing, particularly Table 2. The combination of CRLB, medians, and means in a single cell makes it very hard to interpret, particularly given the formatting. Please make sure these tables are properly formatted so they can be read, even if it means making the font smaller or separating them into separate tables.

4.       The conclusion makes a comment about the scan time. What was the scan time for each voxel, including the short TE scan and the 110 ms scan?

5.       Figure 1 is missing.

6.       Minor language editing is needed throughout.

7.       Figure 3 is blurry and the lines are quite small. It would be better to show a larger version of the 2HG spectra since that is the goal of that sequence.

 

 

Author Response

Thank you for your comments. They were most valuable. Below is our response to each comment.  

1. The choice of the precuneus as a reference is somewhat problematic, as the spectra of different regions of the brain can differ simply based on the location. It might have been more useful to place the spectrum in a contralateral region of normal brain. This may explain greater than expected variation in ratios. This should be further described in the discussion as a limitation.

We have added it as a limitation. See section 4.6 page 14.

2. The discussion is overly focused on minor differences between the spectroscopy results in this paper and others. While these provide important context, please provide a clearer description of what the advance is in this paper and why it is valuable.

We consider the possibility to investigate the IDH mutation in vivo and to monitor it to be a valuable feature, which could get main interest in the clinical monitoring of the patients. We have added a comment of this aspect in the end of the conclusion.

3. Some of the tables are confusing, particularly Table 2. The combination of CRLB, medians, and means in a single cell makes it very hard to interpret, particularly given the formatting. Please make sure these tables are properly formatted so they can be read, even if it means making the font smaller or separating them into separate tables.

Table has been redesigned for a better overview. See table 1 on page 3. See table 2 on page 7.

4. The conclusion makes a comment about the scan time. What was the scan time for each voxel, including the short TE scan and the 110 ms scan?

This is indicated in section 2.4. H¹-MRS page 4
sLASER scantime was 2min 8s.
110 ms scan had a scan time of 7min 36s.

5. Figure 1 is missing.

Figure 1 is now added. See page 5.

6. Minor language editing is needed throughout.

The manuscript has been carefully checked.

7. Figure 3 is blurry and the lines are quite small. It would be better to show a larger version of the 2HG spectra since that is the goal of that sequence.

Figure 2 now only shows the fit for 2-HG. Figure 2 and 3 are uploaded as separate files in better quality.

Reviewer 2 Report

1. The main question is whether single-voxel 7T MRS can have clinical implications in the diagnosis in intrinsic brain lesions 2. In my opinion, the topic is not original: as the authors recognise in their conclusions, their findings are just consistence with previous studies in 3T MRS 3. This paper adds nothing to the current literature 4. Some methodological drawbacks must be addressed, for instance, patient 6 should be excluded; a longer follow-up period is necessary to assess the nature of the tumore nature;  many MRS markers are useless to be investigated, at least in a new paper: while it is true that 2-HG could be an interesting target of study, the authors only marginally deal with it. 5. The conclusions may be suggested based on the authors findings (but concerns expressed on point 4 are relevant) but they do not add anything new to the current knowledge.

 

 

Author Response

Thank you for your comments.

1. The main question is whether single-voxel 7T MRS can have clinical implications in the diagnosis in intrinsic brain lesions.

We agree, firm conclusion on the clinical impact can not be drawn at present.

2. In my opinion, the topic is not original: as the authors recognise in their conclusions, their findings are just consistence with previous studies in 3T MRS

Still 7T MRS gives higher spectroscopic resolution.

3. This paper adds nothing to the current literature.
4. Some methodological drawbacks must be addressed, for instance, patient 6 should be excluded; a longer follow-up period is necessary to assess the nature of the tumor nature; many MRS markers are useless to be investigated, at least in a new paper: while it is true that 2-HG could be an interesting target of study, the authors only marginally deal with it.

We agree that 2-HG is an interesting part of the study. Patient 6 is still under watchful waiting. We consider the clinical and MRI diagnosis to be sufficient to justify inclusion.

5. The conclusions may be suggested based on the authors findings (but concerns expressed on point 4 are relevant) but they do not add anything new to the current knowledge.

We have made a modification in the in the conclusion with comment on IDH mutation.

Reviewer 3 Report

The article "Single-voxel MR spectroscopy of gliomas with s-LASER at 7T 1" by Martin Prener et al. is interesting and highlights the valuable role of single-voxel MRS with s-LASER to study possible biomarkers to grade glioma.

It is believed that magnetic resonance spectroscopy (MRS) is one of the MRS techniques most likely to benefit from the use of high B0 magnetic field strengths because of the expected improvements in both signal-to-noise ratio (SNR) and spectral resolution. Theory suggests that SNR increases linearly with field strength in biological samples, as does chemical shift dispersion. However, spectral linewidths in brain tissue (measured in Hz) are also known to increase with field strength, so the improvements in resolution may not be as great as predicted by simple theory. However, it has been shown that the overall better spectral quality at higher field strengths, such as 7T, results in lower uncertainty values of metabolite concentrations and also allows estimation of various compounds that are undetectable or require specialized pulse sequences for accurate quantification at lower field strengths. Specifically, previous studies with surface coils have shown that SNR, spectral resolution, and measurement accuracy in the occipital lobe are all higher at 7T than at lower field strengths (3T or 4T). However, there has been only one MRS field strength comparison study using multichannel head coils, which compared measurements made in the parietal white matter at 3T and 7T using an 8-channel head coil. The comparison of measurements made in multiple brain regions using 32-channel head coils at 3T and 7T is of particular importance because 32-channel coils are now increasingly used for cutting-edge research and clinical neuroimaging studies. However, higher field strengths are also associated with technical challenges such as inhomogeneity in the B0 and B1 fields, longer T1 relaxation times and shorter T2 relaxation times, higher radiofrequency power deposition (specific absorption rate (SAR)), and higher chemical shift errors (CSDEs).

 

Some of these problems are inherent physical properties of the biochemical substance under investigation, while others can be addressed by the development of new techniques, both in the design of the pulse sequence and in the scanner hardware. Inhomogeneity of the B1 transmission field causes deviations of the flip angles of the RF pulses from their ideal values, which can lead to a reduction in SNR. In addition, conventional amplitude-modulated RF pulses have a relatively small frequency bandwidth, leading to large CSDE artifacts at high fields such as 7T. Both of these problems can be solved at least in part by the use of frequency-modulated adiabatic RF pulses, which provide uniform B1-independent flip angle rotations as long as B1 levels are high enough to satisfy the adiabatic condition. Adiabatic pulses usually have a high bandwidth, determined by the frequency sweep of the RF pulse. Fully or partially adiabatic pulse sequences have been proposed for MRS in vivo. The partially adiabatic semi-LASER sequence has become popular in recent years because it involves fewer RF pulses, and thus lower SAR, than fully adiabatic sequences such as SADLOVE or LASER. A recent 7T study demonstrated a more than twofold increase in SNR when using sLASER compared with the non-adiabatic STEAM sequence.

Some of the authors of this article previously reported that the performance of MEGA-sLASER and sLASER was evaluated in four patients with IDH1-mutated glioma. Unlike MEGA-sLASER, sLASER was able to detect 2HG concentrations as low as 0.5 mM. In the case of a composite phantom containing 2HG with overlapping metabolites, MEGA-sLASER provided a clean 2HG signal with higher reliability of fit (lower %CRLB). The results demonstrate that sLASER is more robust against field inhomogeneities and experimental or motion-related artifacts, which promotes the adoption of sLASER in clinical implementations.Some of the authors of this article have previously reported that the performance of MEGA-sLASER and sLASER were evaluated in four patients with IDH1-mutated glioma. In contrast to MEGA-sLASER, sLASER was able to detect 2HG concentrations as low as 0.5 mM. In the case of a composite phantom containing 2HG with overlapping metabolites, MEGA-sLASER provided a clean 2HG signal with higher reliability of fit (lower %CRLB). The results demonstrate that sLASER is more robust against field inhomogeneities and experimental or motion-related artifacts, which promotes the adoption of sLASER in clinical implementations.

In contrast, other authors, comparing the absolute uncertainties for GABA estimation for both acquisition methods, found that the CRLBs and absolute uncertainties of GABA estimated for the MEGA-editing method were lower than those for the non-editing acquisition method for all regions analyzed. Compared with the non-editing acquisition, the GABA-editing method provides a much simpler spectral pattern, with only six metabolic signals present, which are relatively well separated. This explains why the spectral fitting accuracy of the MEGA editing results for GABA estimation is higher than that of the sLASER result, despite the fact that the editing method also removes a number of lines from the GABA spectrum. MEGA-sLASER and sLASER showed similar CRLB values for tCr, which is not surprising given that the tCr concentration of MEGA-sLASER was estimated from the MEGA-off spectra using an unedited acquisition. The reasonably good CRLB values of tCr for both approaches confirm that the spectral quality and spectral fitting quality with LCModel were good enough for accurate quantification.

Minor revision

The layout of the paper is sloppy and messy, making it difficult to read in some sections. Figure 1 is missing and a preprint of the work had to be retrieved to verify voxel placement in all three planes in patient sample. For both precuneus and tumor.

Major revision

The authors are invited to discuss whether s-Laser MRS can be invoked as the best standard for the clinic or whether the choice between s-LASER or MEGA s-LASER depends on the biomarkers of interest.

 

Author Response

Thank you for your comments. They were most valuable. Below is our response to each comment.

Minor revision

The layout of the paper is sloppy and messy, making it difficult to read in some sections. Figure 1 is missing and a preprint of the work had to be retrieved to verify voxel placement in all three planes in patient sample. For both precuneus and tumor.

We agree. Our upload happened to be non-optimal. The layout has now been improved.

Major revision

The authors are invited to discuss whether s-Laser MRS can be invoked as the best standard for the clinic or whether the choice between s-LASER or MEGA s-LASER depends on the biomarkers of interest.

Section 4.5 s-LASER or MEGA s-LASER, has been added to the discussion.

Round 2

Reviewer 2 Report

I recognize the Author's attention to the comments made, however, no substantial change has been done on the previous manuscript, and the criticism express previously pertains also this versione of the manuscript

Reviewer 3 Report

The paper has been revised according to the referee's instructions even if at a low extent. It remains a work of medium scientific value, but I believe that its publication is useful to the scientific community.