Deep-learning-based reconstruction of undersampled MRI to reduce scan times: a multicentre, retrospective, cohort study.

BACKGROUND: The extended acquisition times required for MRI limit its availability in resource-constrained settings. Consequently, accelerating MRI by undersampling k-space data, which is necessary to reconstruct an image, has been a long-standing but important challenge. We aimed to develop a deep convolutional neural network (dCNN) optimisation method for MRI reconstruction and to reduce scan times and evaluate its effect on image quality and accuracy of oncological imaging biomarkers. METHODS: In this multicentre, retrospective, cohort study, MRI data from patients with glioblastoma treated at Heidelberg University Hospital (775 patients and 775 examinations) and from the phase 2 CORE trial (260 patients, 1083 examinations, and 58 institutions) and the phase 3 CENTRIC trial (505 patients, 3147 examinations, and 139 institutions) were used to develop, train, and test dCNN for reconstructing MRI from highly undersampled single-coil k-space data with various acceleration rates (R=2, 4, 6, 8, 10, and 15). Independent testing was performed with MRIs from the phase 2/3 EORTC-26101 trial (528 patients with glioblastoma, 1974 examinations, and 32 institutions). The similarity between undersampled dCNN-reconstructed and original MRIs was quantified with various image quality metrics, including structural similarity index measure (SSIM) and the accuracy of undersampled dCNN-reconstructed MRI on downstream radiological assessment of imaging biomarkers in oncology (automated artificial intelligence-based quantification of tumour burden and treatment response) was performed in the EORTC-26101 test dataset. The public NYU Langone Health fastMRI brain test dataset (558 patients and 558 examinations) was used to validate the generalisability and robustness of the dCNN for reconstructing MRIs from available multi-coil (parallel imaging) k-space data. FINDINGS: In the EORTC-26101 test dataset, the median SSIM of undersampled dCNN-reconstructed MRI ranged from 0·88 to 0·99 across different acceleration rates, with 0·92 (95% CI 0·92-0·93) for 10-times acceleration (R=10). The 10-times undersampled dCNN-reconstructed MRI yielded excellent agreement with original MRI when assessing volumes of contrast-enhancing tumour (median DICE for spatial agreement of 0·89 [95% CI 0·88 to 0·89]; median volume difference of 0·01 cm3 [95% CI 0·00 to 0·03] equalling 0·21%; p=0·0036 for equivalence) or non-enhancing tumour or oedema (median DICE of 0·94 [95% CI 0·94 to 0·95]; median volume difference of -0·79 cm3 [95% CI -0·87 to -0·72] equalling -1·77%; p=0·023 for equivalence) in the EORTC-26101 test dataset. Automated volumetric tumour response assessment in the EORTC-26101 test dataset yielded an identical median time to progression of 4·27 months (95% CI 4·14 to 4·57) when using 10-times-undersampled dCNN-reconstructed or original MRI (log-rank p=0·80) and agreement in the time to progression in 374 (95·2%) of 393 patients with data. The dCNN generalised well to the fastMRI brain dataset, with significant improvements in the median SSIM when using multi-coil compared with single-coil k-space data (p<0·0001). INTERPRETATION: Deep-learning-based reconstruction of undersampled MRI allows for a substantial reduction of scan times, with a 10-times acceleration demonstrating excellent image quality while preserving the accuracy of derived imaging biomarkers for the assessment of oncological treatment response. Our developments are available as open source software and hold considerable promise for increasing the accessibility to MRI, pending further prospective validation. FUNDING: Deutsche Forschungsgemeinschaft (German Research Foundation) and an Else Kröner Clinician Scientist Endowed Professorship by the Else Kröner Fresenius Foundation.

No benefit from TMZ treatment in GB with truly unmethylated MGMT promoter: Reanalysis of the CE.6 and the pooled Nordic/NOA-08 trials in elderly GB patients.

BACKGROUND: The treatment of elderly/ frail patients with glioblastoma is a balance between avoiding undue toxicity, while not withholding effective treatment. It remains debated, whether these patients should receive combined chemo-radiotherapy with temozolomide (RT/TMZ➜TMZ) regardless of the O6-methylguanine DNA methyltransferase gene promoter (MGMTp) methylation status. MGMT is a well-known resistance factor blunting the treatment effect of TMZ, by repairing the most genotoxic lesion. Epigenetic silencing of the MGMTp sensitizes glioblastoma to TMZ. For risk adapted treatment, it is of utmost importance to accurately identify patients, who will not benefit from TMZ treatment. METHODS: Here, we present a reanalysis of the clinical trials CE.6 and the pooled NOA-08 and Nordic trials in elderly glioblastoma patients that compared RT to RT/TMZ➜TMZ, or RT to TMZ, respectively. For 687 patients with available MGMTp methylation data, we applied a cutoff discerning truly unmethylated glioblastoma, established in a pooled analysis of four clinical trials for glioblastoma, with RT/TMZ➜TMZ treatment, using the same quantitative methylation specific MGMTp PCR assay. RESULTS: When applying this restricted cutoff to the elderly patient population, we confirmed that glioblastoma with truly unmethylated MGMTp derived no benefit from TMZ treatment. In the Nordic/NOA-08 trials RT was better than TMZ, suggesting little or no benefit from TMZ. CONCLUSION: For evidence-based treatment of glioblastoma patients validated MGMTp methylation assays should be used that accurately identify truly unmethylated patients. Respective stratified management of patients will reduce toxicity without compromising outcome and allow testing of more promising treatment options.

Zotiraciclib (TG02) for newly diagnosed glioblastoma in the elderly or for recurrent glioblastoma: The EORTC 1608 STEAM trial.

BACKGROUND: Zotiraciclib (TG02) is an oral multi-cyclin dependent kinase (CDK) inhibitor thought to inhibit tumor growth via CDK-9-dependent depletion of survival proteins such as c-MYC and MCL-1 which are frequently overexpressed in glioblastoma. METHODS: EORTC 1608 (NCT03224104) (STEAM) had a three parallel group (A,B,C) phase Ib, open-label, non-randomized, multicenter design in IDH wild-type newly diagnosed glioblastoma or anaplastic astrocytoma. Groups A and B explored the maximum tolerated dose (MTD) of TG02 in elderly patients, in combination with hypofractionated radiotherapy alone (group A) or temozolomide alone (group B), according to O6-methylguanine DNA methyltransferase promoter methylation status determined centrally. Group C explored single agent activity of TG02 at first relapse after temozolomide chemoradiotherapy with a primary endpoint of progression-free survival at 6 months (PFS-6). Tumor expression of CDK-9, c-MYC and MCL-1 was determined by immunohistochemistry. RESULTS: The MTD was 150 mg twice weekly in combination with radiotherapy alone (group A) or temozolomide alone (group B). Two dose-limiting toxicities were observed at 150 mg: one in group A (grade 3 seizure), one in group B (multiple grade 1 events). Main toxicities included neutropenia, gastrointestinal disorders and hepatotoxicity. PFS-6 in group C was 6.7%. CDK-9, c-MYC and MCL-1 were confirmed to be expressed and their expression was moderately cross-correlated. High protein levels of MCL-1 were associated with inferior survival. CONCLUSIONS: TG02 exhibits overlapping toxicity with alkylating agents and low single agent clinical activity in recurrent glioblastoma. The role of CDK-9 and its down-stream effectors as prognostic factors and therapeutic targets in glioblastoma warrants further study.

3D volume growth rate evaluation in the EORTC-BTG-1320 clinical trial for recurrent WHO grade 2 and 3 meningiomas.

BACKGROUND: We previously reported that tumor 3D volume growth rate (3DVGR) classification could help in the assessment of drug activity in patients with meningioma using 3 main classes and a total of 5 subclasses: class 1: decrease; 2: stabilization or severe slowdown; 3: progression. The EORTC-BTG-1320 clinical trial was a randomized phase II trial evaluating the efficacy of trabectedin for recurrent WHO 2 or 3 meningioma. Our objective was to evaluate the discriminative value of 3DVGR classification in the EORTC-BTG-1320. METHODS: All patients with at least 1 available MRI before trial inclusion were included. 3D volume was evaluated on consecutive MRI until progression. 2D imaging response was centrally assessed by MRI modified Macdonald criteria. Clinical benefit was defined as neurological or functional status improvement or steroid decrease or discontinuation. RESULTS: Sixteen patients with a median age of 58.5 years were included. Best 3DVGR classes were: 1, 2A, 3A, and 3B in 2 (16.7%), 4 (33.3%), 2 (16.7%), and 4 (33.3%) patients, respectively. All patients with progression-free survival longer than 6 months had best 3DVGR class 1 or 2. 3DVGR classes 1 and 2 (combined) had a median overall survival of 34.7 months versus 7.2 months for class 3 (P = .061). All class 1 patients (2/2), 75% of class 2 patients (3/4), and only 10% of class 3 patients (1/10) had clinical benefit. CONCLUSIONS: Tumor 3DVGR classification may be helpful to identify early signals of treatment activity in meningioma clinical trials.

Lomustine with or without reirradiation for first progression of glioblastoma, LEGATO, EORTC-2227-BTG: study protocol for a randomized phase III study.

BACKGROUND: Chemotherapy with lomustine is widely considered as standard treatment option for progressive glioblastoma. The value of adding radiotherapy to second-line chemotherapy is not known. METHODS: EORTC-2227-BTG (LEGATO, NCT05904119) is an investigator-initiated, pragmatic (PRECIS-2 score: 34 out of 45), randomized, multicenter phase III trial in patients with first progression of glioblastoma. A total of 411 patients will be randomized in a 1:1 ratio to lomustine (110 mg/m2 every 6 weeks) or lomustine (110 mg/m2 every 6weeks) plus radiotherapy (35 Gy in 10 fractions). Main eligibility criteria include histologic confirmation of glioblastoma, isocitrate dehydrogenase gene (IDH) wild-type per WHO 2021 classification, first progression at least 6 months after the end of prior radiotherapy, radiologically measurable disease according to RANO criteria with a maximum tumor diameter of 5 cm, and WHO performance status of 0-2. The primary efficacy endpoint is overall survival (OS) and secondary endpoints include progression-free survival, response rate, neurocognitive function, health-related quality of life, and health economic parameters. LEGATO is funded by the European Union's Horizon Europe Research program, was activated in March 2024 and will enroll patients in 43 sites in 11 countries across Europe with study completion projected in 2028. DISCUSSION: EORTC-2227-BTG (LEGATO) is a publicly funded pragmatic phase III trial designed to clarify the efficacy of adding reirradiation to chemotherapy with lomustine for the treatment of patients with first progression of glioblastoma. TRIAL REGISTRATION: ClinicalTrials.gov NCT05904119. Registered before start of inclusion, 23 May 2023.

Epigenetic landscape reorganization and reactivation of embryonic development genes are associated with malignancy in IDH-mutant astrocytoma.

Accurate grading of IDH-mutant gliomas defines patient prognosis and guides the treatment path. Histological grading is however difficult and, apart from CDKN2A/B homozygous deletions in IDH-mutant astrocytomas, there are no other objective molecular markers used for grading. Experimental Design: RNA-sequencing was conducted on primary IDH-mutant astrocytomas (n=138) included in the prospective CATNON trial, which was performed to assess the prognostic effect of adjuvant and concurrent temozolomide. We integrated the RNA sequencing data with matched DNA-methylation and NGS data. We also used multi-omics data from IDH-mutant astrocytomas included in the TCGA dataset and validated results on matched primary and recurrent samples from the GLASS-NL study. We used the DNA-methylation profiles to generate a Continuous Grading Coefficient (CGC) that is based on classification scores derived from a CNS-tumor classifier. We found that the CGC was an independent predictor of survival outperforming current WHO-CNS5 and methylation-based classification. Our RNA-sequencing analysis revealed four distinct transcription clusters that were associated with i) an upregulation of cell cycling genes; ii) a downregulation of glial differentiation genes; iii) an upregulation of embryonic development genes (e.g. HOX, PAX and TBX) and iv) an upregulation of extracellular matrix genes. The upregulation of embryonic development genes was associated with a specific increase of CpG island methylation near these genes.

Marizomib for patients with newly diagnosed glioblastoma: a randomized phase 3 trial.

BACKGROUND: Standard treatment for patients with newly diagnosed glioblastoma includes surgery, radiotherapy (RT) and temozolomide (TMZ) chemotherapy (TMZ/RT→TMZ). The proteasome has long been considered a promising therapeutic target because of its role as a central biological hub in tumor cells. Marizomib is a novel pan-proteasome inhibitor that crosses the blood brain barrier. METHODS: EORTC 1709/CCTG CE.8 was a multicenter, randomized, controlled, open label phase 3 superiority trial. Key eligibility criteria included newly diagnosed glioblastoma, age > 18 years and Karnofsky performance status > 70. Patients were randomized in a 1:1 ratio. The primary objective was to compare overall survival (OS) in patients receiving marizomib in addition to TMZ/RT→TMZ with patients receiving only standard treatment in the whole population, and in the subgroup of patients with MGMT promoter-unmethylated tumors. RESULTS: The trial was opened at 82 institutions in Europe, Canada and the US. A total of 749 patients (99.9% of planned 750) were randomized. OS was not different between the standard and the marizomib arm (median 17 vs 16.5 months; HR=1.04; p=0.64). PFS was not statistically different either (median 6.0 vs. 6.3 months; HR=0.97; p=0.67). In patients with MGMT promoter-unmethylated tumors, OS was also not different between standard therapy and marizomib (median 14.5 vs 15.1 months, HR=1.13; p=0.27). More CTCAE grade 3/4 treatment-emergent adverse events were observed in the marizomib arm than in the standard arm. CONCLUSIONS: Adding marizomib to standard temozolomide-based radiochemotherapy resulted in more toxicity, but did not improve OS or PFS in patients with newly diagnosed glioblastoma.