PH Sensitive MRI Based Resections of Glioblastoma

Standard of care therapy and all FDA approved adjuvant therapy for glioblastoma continue to provide < 12-month progression free survival (PFS) and < 24-month overall survival (OS). Standard of care therapy continues to be defined by the volume of tumor that enhances with gadolinium on standard magnetic resonance imaging (MRI). The investigators have identified a significant tumor burden in non-enhancing (NE) regions beyond the contrast-enhancing (CE) portion of tumor. Furthermore, the investigators have adapted a pH-sensitive technique called amine chemical exchange saturation transfer (CEST) MRI to identify tumor cells in NE regions with high sensitivity and specificity. This study is a randomized trial of CEST based resections versus standard of care in newly diagnosed glioblastoma with primary endpoint of progression free survival and secondary endpoints of overall survival and quality of life metrics. The hypothesis being tested is whether surgical resection of infiltrating tumor cells visualized by CEST MRI contributes to survival in glioblastoma patients.

The investigators have previously addressed the difficulty in identifying tumor in non-enhancing (NE) T2 hyperintense regions surrounding the contrast-enhancing bulk of tumor. The investigators have applied pH sensitive amine chemical exchange saturation transfer echo planar imaging (CEST-EPI) to patients with glioblastoma. CEST imaging can utilize detection of fast-exchanging amine protons to obtain an imaging contrast that is dependent on tissue acidity. Bulk water protons undergo chemical exchange with amine protons. CEST imaging detects the attenuation of bulk water magnetization following a saturation pulse, which can vary by concentration of amine protons. This contrast can be quantitated as the asymmetry in the magnetization transfer ratio (MTRasym) relative to water proton resonance frequency. Tumor cells are known to produce elevated levels of extracellular hydrogen, given 1) a higher metabolic rate secondary to increased cell turnover and 2) an affinity for anaerobic glycolysis - the Warburg Effect. The investigators have shown active tumor burden and active replication occurring in infiltrating glioblastoma cells. and thus leveraged CEST imaging to visualize infiltrating tumor cells. 19 In newly diagnosed and recurrent glioblastoma, higher median MTRasym at 3ppm within CEST+ NE regions p=0.007; p=0.0326] and higher volumes of CEST+ NE tumor [p=0.020; p<0.001] were associated with decreased PFS. Prospective MRI guided biopsies of CEST+ NE regions showed a correlation with presence of cell density and %Ki-67 positivity [p<0.001, p<0.001]. Furthermore, using pathological diagnosis as the gold standard, the investigators saw a threshold of MTRasym > 1.50% as able to identify tumor cells with a sensitivity of 100% and specificity of 71.4%. In unpublished data with additional samples only from newly diagnosed WHO IV glioblastoma, the calculated sensitivity and specificity was 100% when using neuro-pathological review as the gold standard. Additionally, when using single cell RNA sequencing (scRNAseq) detection of tumor cells via copy number variation analysis as the gold standard, from 17 biopsies in CEST+ NE regions, the same threshold of 1.50% identified tumor cells with a sensitivity of 93% and specificity of 100%. Given the strong correlation between tumor cells and CEST+ NE regions, the investigators have preliminary conducted CEST+ NE resections on 5 patients with newly diagnosed WHO IV glioblastoma. In this preliminary data, the investigators found no significant post-operative complications with the larger resections. All patients received standard of care adjuvant therapy (temozolomide + radiotherapy) and none of the patients have experienced progression (median PFS 13 months). With this data, the investigators hypothesize that maximal safe resection of CEST+ NE regions will minimize the post-surgical tumor cell burden and allow for increased survival.

Experimental Approach. The investigators will carry out a prospective, randomized controlled trial of surgical resection of glioblastoma with pH sensitive amine CEST-EPI. Patients will be enrolled by me at the UCLA Ronald Reagan Medical Center. The UCLA department of neurosurgery brain tumor program is a high-volume neuro-oncology center. The investigators will recruit patients with imaging consistent with WHO grade IV glioblastoma. 51 I will directly provide informed consent for participating in the trial along with informed consent for the surgery itself. Prior to randomization, all patients will undergo pre-operative medical clearance and standard pre-operative medication administration (levetiracetam 100mg BID and dexamethasone 4mg TID). All patients will undergo pre-operative standard anatomic MRI 52 within 1 week of surgery (3T Siemens Prisma, Siemens Healthcare, Erlangen, Germany). MRI sequences will include: 1mm 3D inversion-recover gradient recalled echo images prior to and after injection of gadolinium contrast (Gd-DTPA, 0.1 mmol/kg), and dual-echo proton density T2-weighted turbo spin echo images. pH-weighted amine CEST echo planar MRI27 will be obtained (field of view: 256 x 256 mm2, matrix size: 128 x 128, slice thickness 4mm, scan time: approximately 5 minutes).

Imaging data will be downloaded from the PACS server onto lab servers for post-processing and analysis (Figure 7). This will consist of affine motion correction (mcflirt; FSL, FMRIB, Oxford, UK) and registration to anatomic T1 with contrast images. MTRasym will be calculated for all voxels of CEST sequences. All further image analysis will be carried out on AFNI (Software for analysis and visualization of functional magnetic resonance neuroimages). 53 To obtain regions of interest based on MTRasym abnormality, the investigators will first identify CE tumor by a voxel-by-voxel subtraction of image intensity of T1 without contrast from T1 with contrast. 16 Using an intensity threshold, CE will be segmented and for tumors with necrotic or cystic core, these will be filled to obtain CE tumor regions of interest (ROI). This ROI will be expanded volumetrically to obtain NE ROI. The NE ROI will be co-registered to MTRasym maps and filtered by minimum value 1.5 which has been shown to predict presence of tumor cells with high sensitivity and specificity. 19 Lastly, the CE ROI will be subtracted from NE ROI filtered for abnormal MTRasym values to yield a CEST+ NE region of tumor (Figure 7). Prior to surgery, the subjects will be randomized. Patients randomized to the experimental arm will have CEST+ NE segmentations uploaded to intraoperative objects using our intraoperative neuronavigation software (BrainLab Surgical Navigation System, Munich, Germany). Patients randomized to the standard-of-care arm will have standard MRI w/wo contrast sequences uploaded to the intraoperative neuronavigation software. During surgery, I will carry out standard of care resection of the contrast-enhancing portion of tumor (standard-of-care arm) or standard of care resection plus maximal safe surgical resection of CEST+NE regions (experimental arm). Maximal safe surgical resection includes complete avoidance of eloquent (motor, language) tissue. Standard post-operative care will be carried out in all patients. Adjuvant therapy will include standard-of-care synchronous oral temozolomide (6 weeks) and radiation therapy 60 Gy over 30 fractions to the resection cavity. Standard MRI with and without contrast will be used as surveillance imaging and RANO criteria will be used to determine recurrence. Primary endpoint will be progression free survival and secondary endpoints will be overall survival, surgical complication rate, and quality of life differences as measured by pre- and post-operative karnofsky performance scale scores.

Statistical Plan. Using our preliminary data on CEST+ NE resections and median PFS (6 months) for our larger 10-year cohort of patients with newly diagnosed glioblastoma, the investigators calculated a sample size of 60 to provide a power of 95% to detect a meaningful 6 month difference in PFS at a 0.05 significance level. Clinical characteristics associated with survival (age, pre-operative KPS) will be compared using independent sample t-tests to assess for bias despite randomization. Multivariate cox and log rank tests will be used to compare survival outcomes. Independent sample chi squared tests will be used to compare differences in surgical complication rate and post-operative karnofsky performance scale changes.