https://orcid.org/0000-0001-6305-3570
Figures
Abstract
In pre-clinical models of brain gliomas, Relaxation Along a Fictitious Field in second rotating frame (TRAFF2), continues wave T1rho (T1ρcw), adiabatic T1rho (T1ρadiab), and adiabatic T2rho (T2ρadiab) relaxation time mappings have demonstrated potential to non-invasively characterize brain gliomas. Our aim was to evaluate the feasibility and potential of 4 different spin lock methods at 3T to characterize primary brain glioma. 22 patients (26–72 years) with suspected primary glioma. T1ρcw was performed using pulse peak amplitude of 500Hz and pulse train durations of 40 and 80 ms while the corresponding values for T1ρadiab, T2ρadiab, TRAFF2 were 500/500/500Hz and 48 and 96, 64 and 112, 45 and 90 ms, respectively. The parametric maps were calculated using a monoexponential model. Molecular profiles were evaluated from tissue specimens obtained during the resection. The lesion regions-of-interest were segmented from high intensity FLAIR using automatic segmentation with manual refinement. Statistical descriptors from the voxel intensity values inside each lesion and radiomic features (Pyrad MRC package) were calculated. From extracted radiomics, mRMRe R package version 2.1.0 was used to select 3 features in each modality for statistical comparisons. Of the 22 patients, 10 were found to have IDH-mutant gliomas and of those 5 patients had 1p/19q codeletion group comparisons. Following correction for effects of age and gender, at least one statistical descriptor was able to differentiate between IDH and 1p/19q codeletion status for all the parametric maps. In the radiomic analysis, corner-edge detector features with Harris-Stephens filtered signal showed significant group differences in IDH and 1p/19q codeletion groups. Spin lock imaging at 3T of human glioma was feasible and various qualitative parameters derived from the parametric maps were found to have potential to differentiate IDH and 1p19q codeletion status. Future larger prospective clinical trials are warranted to evaluate these methods further.
Citation: Jambor I, Steiner A, Pesola M, Gardberg M, Frantzén J, Jokinen P, et al. (2024) Relaxation Along a Fictitious Field, continuous wave T1rho, adiabatic T1rho and adiabatic T2rho imaging of human gliomas at 3T: A feasibility study. PLoS ONE 19(4): e0296958. https://doi.org/10.1371/journal.pone.0296958
Editor: Syed M. Faisal, University of Michigan Medical School, UNITED STATES
Received: September 26, 2023; Accepted: December 21, 2023; Published: April 1, 2024
Copyright: © 2024 Jambor et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data cannot be shared publicly because of IRB regulations. Data are available from the Ethics Committee, Hospital District of Southwest Finland for researchers who meet the criteria for access to confidential data. Contact: Secretary of the Ethics Committee P.O. Box 52, FI-20521 Turku, Finland U-Hospital A3 Email: eettinen.toimikunta@varha.fi.
Funding: The authors wish to acknowledge CSC – IT Center for Science, Finland, for computational resources used in creation of the parameter maps in this study. This study was financially supported by grants from Instrumentarium Research Foundation, Sigrid Jusélius Foundation, Turku University Hospital, TYKS-SAPA research fund, Finnish Cancer Society, Finnish Cultural Foundation, and Orion Research Foundation. HM was supported by the Cultural Foundation of Finland, Orion Pharma Research Fellowship, and Academy of Finland (#26080983). TL was supported by Academy of Finland #340761. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Gliomas are the most common primary brain neoplasms. They range from low grade to anaplastic gliomas and glioblastoma. In the most recent 2021 WHO classification of central nervous system tumors, astrocytic gliomas and oligodendrogliomas are primarily differentiated based on molecular markers [1]. IDH mutation and 1p/19q codeletion are the two most important markers predicting longer survival and distinguishing oligodendrogliomas from other diffuse astrocytomas, respectively. The current standard criterion for tumor grading is based on the combined histological and molecular diagnosis of tumor specimen obtained during surgery. However, this may have limitations due to sampling errors especially when tumor-grade heterogeneity is present within the tumor. These sampling errors could have consequences on the management of the disease and survival. Therefore, novel more accurate non-invasive imaging tools which could predict cancer aggressiveness might complement the integrated molecular and histopathologic grade [2].
Relaxation along a fictitious field (RAFF) is an MRI technique applying amplitude and frequency-modulated irradiation in a sub-adiabatic regime [3]. The use of radiofrequency pulse is based on sine and cosine amplitude and frequency modulations of equal amplitudes, which give rise to a stationary fictitious magnetic field in a doubly rotating frame. The RAFF relaxation time constant was found to differ from laboratory frame relaxation times (T1 and T2) and rotating frame relaxation times, such as continues wave T1rho (T1ρcw), adiabatic T1rho (T1ρadiab), and adiabatic T2rho (T2ρadiab) [4]. Rotating frame relaxations have shown to be quantitative MRI markers to follow up disease progression, including brain ischemia [5]. Moreover, RAFF relaxation values have shown excellent correlation with cell density in a rat glioma model, which makes it a potential biomarker to follow up cancer therapy outcome [6].
Our hypothesis is that improved preoperative characterization delineation of brain tumors using novel MRI techniques, including spin lock imaging could provide tools for better treatment selection. Thus, the purpose of this study was to evaluate the feasibility of spin locking methods performed using Relaxation Along a Fictitious Field in second rotating frame (TRAFF2), continuous wave T1rho (T1ρcw), adiabatic T1rho (T1ρadiab), and adiabatic T2rho (T2ρadiab) and their potential to non-invasively characterize brain glioma.
Material and methods
Study design and study population
The study was approved by the local ethics committee (Turku University Hospital, Turku, Finland) and the ClinicalTrials.gov Identifier is NCT02186262. Each patient with suspected primary glioma gave written inform consent.
MRI protocol and MRI reporting
The MRI examinations were performed using 3T Philips system (Ingenuity PET/MR, Philips, Cleveland, OH). Two channel volume whole body RF coil was used for the excitation while 32 channel manufacture’s head coil was used as a receiver coil. All four spin locking imaging methods, TRAFF2, T1ρcw, T1ρadiab, and T2rho,Table were measured using 3D T1-FFE sequence. The spin lock weightings were repeated once in every 3 s with 20 acquisitions after each weighting. Other parameters were: repetition time of individual acquisitions/echo time (TR/TE) 4.9/2.4 ms, acquisition voxel size 1.25x1.84x2.50 mm3, reconstruction voxel size 1.25x1.25x2.50 mm3, 24 slices, centric k-space coding, acquisition time 2 minutes 15 seconds per each frame. T1ρcw was performed using a block pulse with radiofrequency (RF) peak amplitude 500 Hz (corresponding to 11.74 mT, B1) and pulse durations of 40 and 80 ms. T1ρadiab was performed using hyperbolic secant pulses with RF peak amplitude 500 Hz (corresponding to 11.74 mT, B1) and pulse durations of 48 and 96 ms. The pulse train duration was 48 ms and 96 ms. TRAFF2 was performed with RF peak amplitude of 500 Hz (corresponding to 11.74 mT, B1) and pulse train duration of RAFF2 68 ms and 135 ms. In addition, FLAIR-weighted images were acquired using a 3D TSE sequence with the following parameters: TR/TE 8000/337 ms, inversion time 2400 ms, acquisition voxel size with whole brain coverage 1.0x1.0x2.0 mm3, reconstruction voxel size with whole brain coverage 1.0x1.0x1.0 mm3, FOV 256x256 mm2, SENSE factor 2.0 in right-left and 2.0 and anterior-posterior direction, acquisition time 5 minutes 28 seconds. All MRI examinations were performed within specific absorption rate limit.
Histology and immunohistochemistry
All histopathological material was reviewed by one of the co-authors (MG), board certified pathologist since 2008 with neuropathologist subspecialty since 2011, with 15 years of experience in neuropathology. The histological and molecular pathology was performed at Tyks Laboratories, Turku, Finland, which is accredited by the Finnish Accreditation services, according to the ISO standard 15189:2013. Histology of brain lesions is within the accreditation field, and thus quality control is well documented. Tumor tissue samples were formalin-fixed and paraffin embedded (FFPE). FFPE sections were H&E stained, and immunohistochemistry was performed on a Ventana Benchmark XT Autostainer (Ventana, Strasbourg, France) using an anti-IDH1 R132H antibody at 1:50 dilution (clone H09, Dianova, Hamburg, Germany). Tumors negative for an immunohistochemically detectable IDH1 mutation were genetically tested for IDH1/IDH2 mutations. All tumors were tested for 1p19q codeletion by fluorescent in situ hybridization using Vysis 1p36/1q25 and 19q13/19p13 FISH probe kit (Abbot Laboratories, Abbot Park, IL). Two diffuse astrocytic gliomas without IDH1/2 mutations were further tested for EGFR amplification using silver in situ hybridization. With all stainings and molecular information available, an integrated diagnosis according to the current WHO classification.
For Ki-67 staining, a rabbit monoclonal antibody (clone 30–9, Ventana) was used. The proliferation index was evaluated in the most highly proliferating hotspot area by counting positively stained nuclei/all nuclei in a high-power microscopic field (40x objective).
Data analysis
The relaxation time values TRAFF2, T1ρcw, T1ρadiab, and T2ρadiab were calculated using two parameter monoexponential model. Rigid co-registration was performed using FSL [7] 5.0.4 so all modalities were aligned to the FLAIR images. The lesion ROIs (Region of Interest) were segmented from FLAIR images using automatic segmentation procedure described by Artzi [8]. Tumor region was defined manually from segmented ROIs by applying 1–2 voxel erosion after the automatic procedure.
Voxel level parametric maps of the fitted relaxation values were analyzed by calculating statistical descriptors from the voxel intensity values inside each lesion ROI: 10th, 25th, 50th, 75th, and 90th percentile; kurtosity; skewness; median. For radiomics, we applied radiomic feature extraction using Pyrad package [9] and MRC package [10]. From extracted radiomics, mRMRe R package version 2.1.0 [11] was used to select 3 features in each modality for statistical comparisons.
Statistical analysis
Student’s unpaired t-test was applied for comparison between groups of IDH-mutant vs IDH-wildtype, and astrocytic gliomas without 1p/19q codeletion vs oligodendrogliomas with 1p/19q codeletion, and Pearson’s r for correlation between relaxation values and Ki-67 index. All statistical tests were done in RStudio environment (v 1.1.383, 2017 RStudio, Inc.). FDR-corrected p-values smaller than 0.05 were considered statistically significant, unless otherwise noted.
Results
In total, 38 patients were enrolled between 02/2014 and 9/2015, of those 37 underwent the brain MRI examination. Due to slow enrolment, also patients with suspected recurrent gliomas were enrolled (n = 8) but these patients were not included in the final analyses due to large heterogeneity of primary treatment between these cases. In the remaining 29 patients, 2 were found to have metastatic disease following surgical resection, 2 were found to have non-glial lesions (one meningioma, one with epidermoid), and 3 withdrawn from the trial precluding the use of histopathological material for the analyses in this trial. Thus, 22 patients were included in the final IDH group and Ki-67 proliferation index analysis, and 9 patients in 1p/19q codeletion group comparisons (Fig 1).
Visual inspection of TRAFF2, T1ρcw, T1ρadiab, and T2ρadiab demonstrated good tissue contrast between glioma and normal brain tissue as presented in Fig 2.
Two representative cases: 1: Diffuse astrocytoma IDH-mutant, grade 2 (A), (upper row) and 2: oligodendroglioma, 1p/19q codeleted, grade 2 (O) (bottom row) demonstrate automatic FLAIR lesion segmentation and imaging contrasts for TRAFF2, T1ρcw, T1ρadiab, and T2ρadiab parametric maps. Same scaling was applied within the same sequence to allow comparison between the gliomas. Diffuse astrocytoma, IDH-mutant shows elevated values in all parametric maps compared to oligodendroglioma, IDH-mutant and 1p/19q codeleted. This trend was observed across the cohort, however, without statistical significance.
In the group comparisons, at least one statistical descriptor per each spin lock imaging method was able to differentiate between IDH and 1p/19q codeletion groups after correcting for age and gender (Tables 1 and 2, S1 and S2 Tables in S1 File), however, no statistically significant difference was found with ROI mean before correction for effects of age and gender. In the ROI based analysis, only FLAIR derived variables were found to have significant correlation with Ki-67 (Table 3, S3 Table in S1 File).
Model Variable~Group+Age+Gender-1 was used to correct for effects of age and gender.
Model Variable~Group+Age+Gender-1 model was used to correct for effects of age and gender.
Age and Gender were evaluated with Pearson correlation, while other MRI variables were evaluated with model Ki67~VariableMRI VARIABLE+Age+Gender-1. Only Age and FLAIR were found to have statistically significant correlation after correction for multiple comparisons over evaluations on basic statistical descriptors.
In the radiomic analyses, corner-edge detector features with Harris-Stephens filtered signal were selected in feature selection step for most of the modalities, and also showed significant group differences in IDH and 1p/19q codeletion evaluations (Tables 4 and 5). Various features prominently from pyradiomics package were selected and demonstrated significant correlation with Ki-67 correlations (Table 6).
P-values are fdr-corrected 110 over evaluated radiomics. For MRI variables GroupIDH~VariableRADIOMIC+Age+Gender-1 model was used to correct for effects of age and gender.
P-values are fdr-corrected over 110 evaluated radiomics. For MRI variables GroupIDH~VariableRADIOMIC+Age+Gender-1 model was used to correct for effects of age and gender.
The radiomic feature values were evaluated with model Ki67~VariableRADIOMIC FEATURE +Age+Gender-1.
Discussion
In prior preclinical studies spin lock imaging demonstrated promising results for non-invasive characterization of brain gliomas. Gliomas tend to have high signal on T2 based sequences. The hyperintensity on T2 based sequences (including fluid-attenuated inversion-recovery, FLAIR) is due to prolongation of transverse relaxation times (T2 relaxation time) mainly caused by the increase in tissue water content and ultrastructural changes. The areas of haemosiderin appear as foci of signal dropout. However, anatomical (T2-/T1-/FLAIR weighted imaging) have limited potential for predication of IDH mutation and 1p/19q codeletion [12]. Accurate detection of IDH mutation and 1p/19q codeletion is important for risk stratification since these markers have predicative power for longer survival and distinguishing oligodendrogliomas from diffuse astrocytomas, respectively [2]. Thus, development of novel imaging methods for non-invasive detection IDH mutation and 1p/19q codeletion could offer new possibilities for patients’ risk management and tailored treatment.
We used whole tumor FLAIR lesion segmentation, histogram and radiomic analysis to differentiate gliomas according to their IDH gene and 1p/19q chromosomal status. Following correction for effects of age and gender, at least one statistical descriptor per each spin lock imaging method and various radiomic features were able to significantly differentiate between IDH and 1p/19q codeletion groups following correction for multiple comparison. Although we demonstrated statistical significance, the clinical implications of these findings remain to the explored.
Our study has a number of limitations. Only 22 patients were included and further studies are need to explore our preliminary findings in a larger study cohort. The current study did not aim to explore relaxation mechanisms of T1ρadiab, and TRAFF2. Different relaxation methods, including dipolar interactions, diffusion, exchange relaxation pathways, can all contribute to various degree to TRAFF2, T1ρcw, T1ρadiab, and T2ρadiab [3, 6, 13–15]. Although this study demonstrated clinical significance for several variables, the clinical relevance of the findings and impact on patient outcome is left to be explored in larger clinical trials. Three patients withdrawn from the trial precluding the use of histopathological material for the analyses which may have introduced a bias and affect the completeness of the study. Finally, this study used only a limited number of radiomics due to the small sample size and future studies using large number of radiomics and various machine learning methods are warranted [16–18]. Please note, the interpretability of these radionics features and their direct biological or clinical relevance is beyond the scope of this study.
Conclusion
Our clinical trial demonstrated feasibility of quantitative TRAFF2, T1ρcw, T1ρadiab, and T2ρadiab imaging of human gliomas using a clinical 3T MRI scanner. Our initial results indicate the potential of these methods to improve non-invasive glioma characterization.
Supporting information
S1 Checklist. Human participants research checklist.
https://doi.org/10.1371/journal.pone.0296958.s001
(DOCX)
S2 Checklist. Reporting checklist for diagnostic test accuracy study.
https://doi.org/10.1371/journal.pone.0296958.s002
(DOCX)
Acknowledgments
The authors wish to acknowledge CSC–IT Center for Science, Finland, for computational resources used in creation of the parameter maps in this study.
Data cannot be shared publicly because of IRB regulations. Data are available from the Ethics Committee, Hospital District of Southwest Finland for researchers who meet the criteria for access to confidential data. Contact:
Secretary of the Ethics Committee
P.O. Box 52, FI-20521 Turku, Finland
U-Hospital A3
Email: eettinen.toimikunta@varha.fi
References
- 1. Louis DN, Perry A, Wesseling P, et al.: The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol 2021; 23:1231–1251.
- 2. Law M, Yang S, Wang H, et al.: Glioma grading: sensitivity, specificity, and predictive values of perfusion MR imaging and proton MR spectroscopic imaging compared with conventional MR imaging. AJNR AmJNeuroradiol: 1989–1998. pmid:14625221
- 3. Liimatainen T, Mangia S, Ling W, et al.: Relaxation dispersion in MRI induced by fictitious magnetic fields. J Magn Reson 2011; 209:269–276. pmid:21334231
- 4. Jokivarsi KT, Liimatainen T, Kauppinen RA, Gröhn OHJ, Närväinen J: Relaxation Along a Fictitious Field (RAFF) and Z-spectroscopy using Alternating-Phase Irradiation (ZAPI) in Permanent Focal Cerebral Ischemia in Rat. PLoS One 2013; 8:1–10. pmid:23874898
- 5. Grohn OH, Lukkarinen JA, Silvennoinen MJ, Pitkanen A, van Zijl PC, Kauppinen RA: Quantitative magnetic resonance imaging assessment of cerebral ischemia in rat using on-resonance T(1) in the rotating frame. Magn Reson: 268–276. pmid:10440951
- 6. Liimatainen T, Sierra A, Hanson T, et al.: Glioma cell density in a rat gene therapy model gauged by water relaxation rate along a fictitious magnetic field. Magn Reson Med 2012; 67:269–277.
- 7. Hernández M, Guerrero GD, Cecilia JM, et al.: Accelerating Fibre Orientation Estimation from Diffusion Weighted Magnetic Resonance Imaging Using GPUs. PLoS One 2013; 8:61892. pmid:23658616
- 8. Artzi M, Aizenstein O, Jonas-Kimchi T, Myers V, Hallevi H, Ben Bashat D: FLAIR lesion segmentation: Application in patients with brain tumors and acute ischemic stroke. Eur J Radiol 2013; 82:1512–1518. pmid:23796882
- 9. Van Griethuysen JJM, Fedorov A, Parmar C, et al.: Computational radiomics system to decode the radiographic phenotype. Cancer Res 2017; 77:e104–e107. pmid:29092951
- 10. Merisaari H, Taimen P, Shiradkar R, et al.: Repeatability of radiomics and machine learning for DWI: Short-term repeatability study of 112 patients with prostate cancer. Magn Reson Med 2019:mrm.28058. pmid:31703155
- 11. De Jay N, Papillon-Cavanagh S, Olsen C, El-Hachem N, Bontempi G, Haibe-Kains B: MRMRe: An R package for parallelized mRMR ensemble feature selection. Bioinformatics 2013; 29:2365–2368. pmid:23825369
- 12. van der Voort SR, Incekara F, Wijnenga MMJ, et al.: Predicting the 1p/19q codeletion status of presumed low-grade glioma with an externally validated machine learning algorithm. Clin Cancer Res 2019; 25:7455–7462. pmid:31548344
- 13. Liimatainen T, Sierra A, Hakkarainen H, et al.: MRI Contrasts Generated Using Fictitious Fields in High-Rank Rotating Frames Correlate With Myelin Content in Normal Rat Brain ex vivo. Proc 21st Sci Meet Int Soc Magn Reson Med 2013; 21:0869.
- 14. Liimatainen T, Mangia S, Ling W, et al.: Relaxation dispersion in MRI induced by fictitious magnetic fields. JMagn Reson: 269–276. pmid:21334231
- 15. Jambor I, Pesola M, Taimen P, et al.: Rotating frame relaxation imaging of prostate cancer: Repeatability, cancer detection, and Gleason score prediction. Magn Reson Med 2016; 75:337–344. pmid:25733132
- 16. Le VH, Kha QH, Minh TNT, Nguyen VH, Le VL, Le NQK. Development and Validation of CT-Based Radiomics Signature for Overall Survival Prediction in Multi-organ Cancer. J Digit Imaging. 2023; 36: 911–922. pmid:36717518
- 17. Nguyen HS, Ho DKN, Nguyen NN, Tran HM, Tam KW, Le NQK. Predicting EGFR Mutation Status in Non-Small Cell Lung Cancer Using Artificial Intelligence: A Systematic Review and Meta-Analysis. Acad Radiol 2023; 27:S1076–6332(23)00179-4. pmid:37120403
- 18. Taha B, Boley D, Sun J. State of Radiomics in Glioblastoma. Neurosurgery. 2021; 15; 89:177–184. pmid:33913492