Amide Proton Transfer MR Imaging Of Gliomas: Correlation Studies With Histopathology And Genomics

Research BackgroundMalignant glioma is the most common and dismal primary brain tumor, and the median overall survival time remains relatively low despite some improvements in neurosurgery, chemotherapy and radiotherapy. Although the personal gene data-based precise medicine has been a hot area of glioma therapy research, it is still at the initial stage of exploration. Currently, in the context of imaging diagnosis, the clinical dilemmas primarily focus on differential diagnosis among tumor types (for example, high-grade gliomas vs. primary central nervous system lymphomas) before operation, and between tumor recurrence vs. pseudoprogression or radionecrosis during the surveillance in post-operative malignant gliomas. Accuracy diagnosis at this point would help to choose an appropriate regimen, ultimately prolonging survival. Pathology results from lesion tissues remain the mainstream diagnositic and monitoring modality for malignant gliomas. Magnetic Resonance Imaging (MRI) is regarded as the most important imaging modality for the diagnosis and surveillance of brain tumors; however, conventional MRI techniques only provide limited anatomic information rather than molecular biological data. Furthermore, recently, many clinical groups found gadolinium deposits in the deep cerebellar nuclei including the dentate nucleus by autopsy. The Food and Drug Administration of the United States alerts that it is unknown whether these gadolinium deposits are harmful or can lead to adverse health effects. Giving this importance, developing molecular endogenous contrast MRI technique is crucial.Amide proton transfer imaging is a particular Chemical Exchange Saturation Transfer (CEST) MRI technique. It was designed to detect cytoplasm proteins and peptides in vivo, by exploring the water signal changes due to statuation transfer from amide protons. A number of studies showed that APTw imaging can yield unique information about the proteins and peptides in glioma tumor cells, which is consistent with the findings from MRI-guided proteomics and MR spectroscopy in vivo. This research focused on the application of APTw imaging in diagnosis and treatment of malignant gliomas, including differential diagnosis betweeb high-grade gliomas and primary central nervous system lymphomas, image-guided biopsy in pre-and post-operative gliomas, and a preliminary study of radiogenomics that combined APTw image features and messenger RNA expression. The aims were to explore APTw signal as a surrogate biomarker to guide glioma surgery and gene target drugs in the short future, and besides, to investigate the mechanism of APTw imaging on the pathological and genomic level.Part ⅠDifferential diagnosis between primary central nervous system lymphomas and high-grade gliomas using APTw imaging PurposeTo assess the diagnostic performance of APT imaging in the differentiation of primary central nervous system lymphomas (PCNSL) and high-grade gliomas (HGG) and to analyze the correlation between APTW signal intensities and the nuclear-cytoplasm (N/C) ratio.Method1. MRI ProtocolEleven PCNSL patients with 13 lesions and 21 HGG patients were included in the present study. All patients were scanned on a Philips 3 T MRI system. The sequences performed for each patient included:T1-weighted (T1W); T2-weighted (T2W); fluid-attenuated inversion recovery (FLAIR); APT imaging; and gadolinium contrast-enhanced Ti-weighted (Gd-T1W). APTW scanning was performed on the maximum cross-section, and Z-spectra over an offset range of ±6ppm and the conventional semi-solid magnetization-transfer-ratio (MTR) at 15.6 ppm were acquired.2. APTw Image ProcessingThe image analysis was performed using the interactive data language (IDL). After correcting for the Bo inhomogeneity effect, the APTw image was constructed with the so-called magnetization transfer-ratio asymmetry at the offsets of±3.5 ppm with respect to the water signal:MTRasym(3.5 ppm)= Ssat(-3.5 ppm)/So-Ssat(+3.5 ppm)/So, where Ssat and So are the imaging signal intensities with and without selective radiofrequency irradiation, respectively. To account for the contribution of the possible nuclear Overhauser enhancement effect at-3.5 ppm to MTRasy(3.5ppm), the calculated MTRasym(3.5ppm) image is generally called the APTw image.3. APTw Image AnalysisRegions of interest (ROIs) were carefully chosen by two experienced radiologists. For each patient, at least five ROIs were defined in lesion areas encircled by the Gd enhancement. No ROIs included areas of major necrosis or vessels evident on post-contrast imaging. The maximum APTW signal intensity (APTWmax), minimum APTW signal intensity (APTWmin), APTW signal distribution inside the tumor core (APTWmax-min= APTWmax-APTWmin), mean APTW signal intensity [APTWmcan=(APTWmax+APTWmin)/2], total CEST signal intensity (integral of the whole MTRasym spectrum between 0 and 5 ppm, CESTtotai), and MTR value from the lesions, as well as the APTW and MTR values from the peritumoral edema, were used to compare the two groups. CESTtotal and MTR values were recorded in the ROI with APTWmax in each leision.4. Histologic Data AcquisitionDigital images captured using a microscope and a digital camera were analyzed with image analysis software (Image-Pro Plus) by focusing on color-specific features (blue-staining nuclei and pink-staining cytoplasm). The field to capture was selected in areas of solid tumor with the least amount of non-neoplastic tissue, such as blood vessels, necrotic tissue, inflammatory cells, and tissue sectioning artifacts. The N/C ratio was calculated by dividing the nuclear area by the cytoplasmic area.5. Statistical AnalysisThe interrater agreement between the two readers for the assessment was analyzed by using the Intraclass Correlation Coefficient (ICC). The comparisons of APTWmax, APTWmin, APTWmax.min, APTWmean, CESTtotal, and MTR between the PCNSL and HGG groups were performed using an independent samples t-test to analyze the statistical differences. The sensitivity, specificity, and accuracy for the discrimination of PCNSLs and HHGs were calculated for the image parameters that showed significant statistical differences on the previous t-test, and the corresponding optimal cut-off values were determined by a receiver operator characteristics (ROC) analysis. The relationships between the APTWmax or the MTR and N/C ratios were assessed by a simple linear regression analysis. Statistical analyses were performed using the SPSS. The alpha level of all tests was set at P< 0.05.Results1. There was a relatively stronger CEST effect at an offset of-3.5 ppm downfield from water, where the endogenous cellular backbone amide protons of mobile proteins and peptides resonate. This APT effect was significantly larger in both the PCNSLs and HGGs than in the CNAWM (both P< 0.01).2. The PCNSLs usually showed relatively homogeneous hyperintensity (compared to CNAWM) than did the HGGs on the APTW images. APTWmax, APTWmax-min, and CESTtotal were significantly lower (P< 0.05,0.01, and 0.05, respectively), and APTWmin and MTR were significantly higher (both P< 0.01) in PCNSL lesions than in HGG lesions. The APTW values in peritumoral edema were significantly lower for PCNSLs than for HGGs. APTWmax-min had the highest the area under the curve (0.963) in differentiating PCNSLs from HGGs.3. The histologic data calculations showed significantly larger N/C ratios in PCNSLs than in HGGs, which were estimated to be 1.69±0.72 in PCNSLs and 0.55 ±0.21 in HGGs (P< 0.01). There was a strong, significantly negative correlation between APTWmax and N/C ratio (R= 0.576, P< 0.01), and there was a moderate positive correlation between MTR and N/C ratio (R= 0.326, P< 0.084).ConclusionsThe protein-based APTW signal would be a valuable MRI biomarker by which to identify PCNSLs and HGGs presurgically. The results confirmed that APTw intensities logically originated from mobile cytosolic proteins and peptides.Part IIAPTw MRI Signal as a Biomarker for Newly Diagnosed Gliomas Validated by Stereotactic Biopsies and HistopathologyPurposeTo evaluate the accuracy of the APTw signal as an imaging biomarker of malignant gliomas via the neuro-navigation system in patient with newly diagnosed gliomas.Method1. Human Subject RecruitmentThis prospective study enrolled 24 patients with suspected newly diagnosed gliomas (without craniotomy or brain biopsy surgical intervention, radiotherapy or chemotherapy). All the subjects were referred to APT scan prior to biopsy or resection where a neuro-navigation frameless biopsy system was applied. The time interval between APT imaging scanning and surgery was shorter than six days for all patients (except for one case, which was 15 days).2. MRI AcquisitionStudies were performed on a 3T human MRI scanner. A three-dimensional APT imaging sequence developed recently was used in this study. Briefly, this sequence consists of radiofrequency saturation (four block pulses of 200 ms duration and 2μT amplitude); lipid suppression; and three-dimensional gradient-and spin-echo image acquisition. Several standard MR images were acquired for reference, including T2w, FLAIR, T1w and Gd-T1w. Gd-T1w imaging was the last sequence acquired.3. APTw Image ProcessingThe image analysis was performed using the IDL. The acquired APT image series was registered to the saturated image at 3.5 ppm by the analysis of functional neuroimages software (AFNI). After correcting for the Bo inhomogeneity effect, using the determined Bo map from the water saturation shift-referencing method, the APTw image was constructed with the so-called magnetization transfer-ratio asymmetry at the offsets of ±3.5 ppm with respect to the water signal, MTRasym(3.5 ppm).4. Stereotactic BiopsyFor each patient, three-to-six regions of interest (ROIs) that are clinically feasible and ethically appropriate were chosen for tissue when APTw hyperintensity and Gd enhancement were privileged. In case without obvious APTw hyperintensity and Gd enhancement, biopsy targets were put in the T2/FLAIR hyperintense area with the most surgical feasibility. These pre-determined ROIs were labeled on the co-registered clinical MR images in the BrainLab neuro-navigation system. The exact sites of sampling were marked by a screenshot image intraoperatively. Navigated biopsies were thereafter preserved and fixed in 4% paraformaldehyde for further pathological analysis.5. Histopathological AnalysisThe H&E and IHC stained slides were reviewed semi-quantitatively and quantitatively. Tumor grade, tumor cell density, necrosis degree, and pathologic results were evaluated from H&E sections and transferred to ordered categorical variables. Each sample was reviewed by two steps:overall and in a hotspot.The absolute tumor cell density (Cell_count) and proliferation index (Ki-67) were further quantified in the hotspot. Cell_count was calculated semi-automatically by an "analyze particle" function in an open source image-process software ImageJ The unit of Cell_count is/FOV. The proliferation index examination was performed on IHC stained slides. The percentage of Ki-67 positively stained nuclei of tumor cells was counted as the proliferation index.6. APTw Image AnalysisThe clinical routine MR images used in the BrainLab neuro-navigation system were co-registered to the corresponding experimental routine MR images. After this, each sampling site was transferred from the clinical routine MR images to the identical site on the co-registered experimental routine MR images. The sampling site was further pinpointed on the co-registered saturated Ssat image at 3.5 ppm. The ROI, each containing 4-5 voxels was manually drawn at the biopsy site on the co-registered saturated Ssat image at 3.5 ppm. The CNAWM at the same slice was drawn as reference. Finally, relative APTw values were reported as ROI APTw intensities after abstract the CNAWM APTw intensities.7. Statistical AnalysisStatistical calculations were performed using SPSS. Pearson’s or Spearman’s correlation analysis was applied to determine the reciprocal relation between APTw signal intensities and pathologic indexes. The one-way analysis of variance (ANOVA) test, followed by the LSD method or Dunnet’t method, was implemented to evaluate the difference of APTw signal intensities in different grades. The independent t-test or the Mann-Whitney test was used to evaluate the difference of APTw signal intensities for high-and low-grade tumors. To identify significant correlations, the set of potential pathologic predictor variables (tumor grade, cell density, necrosis degree, Ki-67) were entered into a multiple linear regression analysis with APTw signal intensity as a dependent variable. The ROC curve was further employed as an assessment of the diagnostic accuracy of APTw signal intensity in differentiating high-grade specimens from low-grade specimens. Significance levels were tested for P=0.05.Results1. Total 70 biopsied specimens with known tissue sampling sites were obtained from 24 cases with newly diagnosed gliomas.33 specimens were diagnosed as grade-II gliomas,14 specimens as grade-Ⅲ gliomas,15 specimens as grade-IV gliomas, and eight were found to be peritumoral edematous tissues without obvious tumor cells. It was found that multiple grades were found in six patients.21 of these 24 cases underwent resection after the stereotactic biopsy, and the final pathologic diagnosis was consistent with the report depended on the intraoperative biopsy tissues.2. Compared with CNAWM, all of these low-grade lesions showed homogeneous iso-intensity to mild punctate hyperintensity on APTw images, while all of these high-grade gliomas consistently showed APTw-hyperintense foci, no matter whether the lesions were associated with Gd enhancement.3. APTw intensity values showed a strong positive correlation with the grade (R= 0.572; P< 0.001). Notably, there was a very strong positive correlation between APTw intensities and Cellcount (R= 0.757; P< 0.001) and a strong positive correlation between APTw intensities and Ki-67 (R= 0.538; P< 0.001).4. The multiple linear regression analysis showed that the APTw signal was exclusively determined by the Cellcount and Ki-67 index:APTw signal can differentiate high-grade tissue samples (grade-Ⅲ and-Ⅳ) from low-grade tissue samples (grade-II), and the AUC was 0.804 (the cut-off APTw value was 2.74%).ConclusionThe APTw signal is a valuable imaging biomarker for malignant gliomas that can be used to identify the high-grade region of heterogeneous gliomas. APTw image-directed biopsy can potentially increase accuracy of tumor sampling in patients with gliomas.Part IIIApplication of APT MRI to Assessing the Status of Treated High-grade Gliomas: Differentiation between Tumor Recurrence and RadionecrosisPurposeThe capability of neuroimaging to assess glioma response to therapy remains very poor. We aimed to perform the point-by-point correlation analysis between the protein-based APTw MRI signal intensity and the histopathologic index by APTw image-directed stereotactic biopsy, and to evaluate the accuracy of the APTw signal as an imaging biomarker of the status of treated malignant gliomas.Method1. Human Subject RecruitmentThis prospective study enrolled 21 patients with pathologically proved high-grade gliomas. The patients previously underwent surgery, followed by radiotherapy and chemotherapy. Biopsy was administrated due to suspected tumor recurrence. All subjects were referred to an APTw prior to biopsy or resection during which a neuro-navigation frameless biopsy system was applied. The time interval between APT imaging scanning and surgery was shorter than six days for all patients (except for one case which was 25 days).2. MRI acquisition and APTw image processingAll patients underwent a 3D APTw sequence on a 3T MRI scanner, and the details of MRI acquisition and APTw image processing refer to Part II.3. Stereotactic biopsyBiopsy targets were put in the Gd-enhancing areas, with and without APTw hyperintensity. The details of the surgery procedure and equipment refer to Part II.4. Neuropathological AnalysisThe cell density and necrosis observed on hot spots on the H&E slices were recorded as Cell_hot and Nec_hot, while those observed as a whole were recorded as Cell_ove and Nec_ove. The details of pathologic reviewing, reagents refer to Part II. For these treated malignant gliomas, we evaluated tumor status of each specimen, and classified into three levels:quiescent, mixed and active based on the histopathologic features (including nuclear pleomorphism, mitosis, proliferation, neovascularization, patterns of necrosis).5. APTw Image AnalysisThe details of image registration and ROI drawing refer to Part Ⅱ.6. Statistical methodThe differences of each of the pathologic index, as well as APTw signal intensities, among three tumor statuses were evaluated by a one-way analysis of variance (ANOVA) or Kruskal-Wallis rank sum test, followed by the LSD method or Dunnet’t method (for ANONVA) or a Bonferroni correction (for K-W test) as a post-hoc test. Significance levels were set for p= 0.05. Then, Pearson’s, Spearman’s, or Kendall’s rank correlation analysis was applied to determine the reciprocal relations between APTw signal intensities and each pathologic index or between all pathologic indexes. To identify significant correlations, the set of potential pathologic predictor variables (tumor status, Cell_ove, Cell_hot, Nec_ove, Nec_hot, Cell_count, and proliferation) were entered into a multiple linear regression analysis with APTw signal intensity as a dependent variable. A stepwise entry model was employed, and an entry criterion level was set as p< 0.05 to evaluate any cross-correlation between predictor variables.Results1. Total 64 biopsied specimens were obtained from 21 suspected recurrent malignant gliomas.35 tissues were diagnosed as active,13 tissues were quiescent, and 12 tissues were determined as mixed. The remaining four specimens were peritumoral edema without any obvious tumor cell. More than two histopathologic categories tissues (active, quiescent, or mixed) were simultaneously found within the same lesion for nine of 21 patients recruited in this study.19 of these 21 cases underwent resection after the stereotactic biopsy, and the final pathologic diagnosis was consistent with the intraoperative report depended on the APTw image-guided biopsy tissues.2. APTw intensities of quiescent tissues (1.22±0.60%), mixed tissues (1.97± 1.04%), and active tissues (3.14±0.68%) showed statistically significant difference, both inter-groups and between groups.3. We combined quiescent tissues with mixed tissues into radionecrosis group, active tissues were enrolled in tumor recurrence group. The difference of APTw intensities of recurrence and radionecrosis was significant (3.14±0.68% vs.1.61± 0.93%, p< 0.001). ROC analysis was explored to evaluate the diagnostic ability of APTw signal in differentiation tumor recurrence from radionecrosis, the AUC is 0.891 with the sensitivity of 85.7% and specificity of 84.0%.4. APTw intensity showed a very strong positive correlation with tumor status (R= 0.728; p< 0.001). Furthermore, APTw intensity showed a strong positive correlation with Cell_ove, Cell_hot, Cell_count, and Ki-67. On the contrary, there was a weak to moderate negative correlation between APTw intensities and necrosis (Nec_ove and Nec_hot).5. Multiple linear regressions modeling for explaining the origin of APTw intensity identified the tumor status (quiescent, mixed, active) as the only independent, most significant predictor of APTw.ConclusionThe APTw imaging signal is a surrogate biomarker of malignant glioma that has the potential to differentiate recurrent tumor from treatment effects. The APTw hyperintensity is associated with the most malignant areas that harbor active glioma.Part ⅣPreliminary Study of Radiogenomics in Glioblastomas:Relationship between APTw Imaging Features and Messenger RNA Expression PurposeTo explore differentially expressed genes (DEGs) related to the protein-based APT imaging phenotype of GBMs.Method1. Human Subject Recruitment and MRI protocol16 untreated GBM patients were enrolled in this study. All patients were scanned on a 3T MRI scanner, including T2w, FLAIR, T1w, Gd-T1w and APTw imaging. The details of MRI acquisition and APTw image processing refer to Part Ⅰ.2. Tissue Sample AcquisitionThe areas with APTw≥1.57% and Gd enhancement were defined as solid tumor segments, while the areas with APTw≤1.57% and no obvious Gd enhancement were defined as peritumoral zone. Tissues were taken from solid tumor segments and peritumoral zone, respectively, for each patient guided by intraoperative navigation system.3. Image AnalysisTwo experienced neuroradiologists assessed APTw imaging or conventional MR imaging morphologic features indepently.(1) Intratumoral Hemorrhage:The hemorrhage was defined as foci with low signal on T2w or high signal on T1w, which was confirmed by SW images, if necessary.(2) APTw/FLAIR and APTw/Gd-T1w:APTw hyperintense area, FLAIR hyperintense area, and gadolinium T1w enhancing area were calculated. APTw/FLAIR hyperintense area ratio (AFR), as well as APTw hyperintense/gadolinium T1w enhancing area ratio (ATR) were calculated. Based on the AFR cutoff value of 0.8, all patients were divided into two groups:high AFR and low AFR. Based on the ATR cutoff value of 1.0, all patients were divided into two groups:high ATR and lowATR.4. RNA extractionThe total RNA of each of above listed samples was isolated using the Trizol Kit following by the manufacturer’s instructions. Then the total RNA was treated with RNase-free DNase I for 30 min at 37℃ to remove residual DNA. RNA quality was verified using Agilent 2100 Bio-analyzer and were also checked by RNase free agarose gel electrophoresis.5. RNA library construction and sequencingPoly (A) mRNA was isolated using oligo-dT beads (Qiagen). All mRNA was broken into short fragments by adding fragmentation buffer. First-strand cDNA was generated using random hexamer-primed reverse transcription, followed by the synthesis of the second-strand cDNA using RNase H and DNA polymerase I. The cDNA fragments were purified using a QIA quick PCR extraction kit. These purified fragments were then washed with EB buffer for end reparation poly (A) addition and ligated to sequencing adapters. Following agarose gel electrophoresis and extraction of cDNA from gels, the cDNA fragments were purified and enriched by PCR to construct the final cDNA library. The cDNA library was sequenced on the Illumina sequencing platform (Illumina HiSeqTM 2500) using the paired-end technology by Gene Denovo Co. (Guangzhou, China). A Perl program was written to select clean reads by removing low quality sequences (there were more than 50% bases with quality lower than 20 in one sequence), reads with more than 5% N bases (bases unknown) and reads containing adaptor sequences.6. Identifing differentially expressed genesAfter the expression level of each transcript and gene was calculated, differential expression analysis was conducted using edgeR. The false discovery rate (FDR) was used to determine the threshold of the p value in multiple tests, and for the analysis, a threshold of the FDR<0.05 and an absolute value of log2Ratio≥1 were used to judge the significance of the gene expression differences. The DEGs were submitted for GO enrichenment analysis.ResultsBy analyzing the radiogenomic correlation between APTw image features and messenger RNA expression, NFS1 genes were found to be significantly downregulated in GBM with intratumoral hemorrhage compared to GBM without intratumoral hemorrhage. BRCA1 genes were found to be significantly downregulated in the high AFR tumor tissue compared to the low AFR group. SLAMF9 and MIA genes were found to be significantly downregulated in the high ATR tumor tissues compared to the low ATR group. The GO enrichments for DEGs were summaried in Table 4-1.ConclusionOur initial findings demonstrated a novel correlation in molecular imaging and gene characteristics of gliomas, and revealed the potential genetic and biologic significance by the non-invasive protein-based APTw imaging features, which may facilitate the GBM precision medicine that primarily depends on genomics.