dCas9/CRISPR-based methylation of O-6-methylguanine-DNA methyltransferase enhances chemosensitivity to temozolomide in malignant glioma

BACKGROUND

Malignant glioma carries a poor prognosis despite current therapeutic modalities. Standard of care therapy consists of surgical resection, fractionated radiotherapy concurrently administered with temozolomide (TMZ), a DNA-alkylating chemotherapeutic agent, followed by adjuvant TMZ. O-6-methylguanine-DNA methyltransferase (MGMT), a DNA repair enzyme, removes alkylated lesions from tumor DNA, thereby promoting chemoresistance. MGMT promoter methylation status predicts responsiveness to TMZ; patients harboring unmethylated MGMT (~60% of glioblastoma) have a poorer prognosis with limited treatment benefits from TMZ.

METHODS

Via lentiviral-mediated delivery into LN18 glioma cells, we employed deactivated Cas9-CRISPR technology to target the MGMT promoter and enhancer regions for methylation, as mediated by the catalytic domain of the methylation enzyme DNMT3A. Methylation patterns were examined at a clonal level in regions containing Differentially Methylation Regions (DMR1, DMR2) and the Methylation Specific PCR (MSP) region used for clinical assessment of MGMT methylation status. Correlative studies of genomic and transcriptomic effects of dCas9/CRISPR-based methylation were performed via Illumina 850K methylation array platform and bulk RNA-Seq analysis.

RESULTS

We used the dCas9/DNMT3A catalytic domain to achieve targeted MGMT methylation at specific CpG clusters in the vicinity of promoter, enhancer, DMRs and MSP regions. Consequently, we observed MGMT downregulation and enhanced glioma chemosensitivity in survival assays in vitro, with minimal off-target effects.

CONCLUSION

dCas9/CRISPR is a viable method of epigenetic editing, using the DNMT3A catalytic domain. This study provides initial proof-of-principle for CRISPR technology applications in malignant glioma, laying groundwork for subsequent translational studies, with implications for future epigenetic editing-based clinical applications.

Harnessing clinical annotations to improve deep learning performance in prostate segmentation

Purpose

Developing large-scale datasets with research-quality annotations is challenging due to the high cost of refining clinically generated markup into high precision annotations. We evaluated the direct use of a large dataset with only clinically generated annotations in development of high-performance segmentation models for small research-quality challenge datasets.

Materials and methods

We used a large retrospective dataset from our institution comprised of 1,620 clinically generated segmentations, and two challenge datasets (PROMISE12: 50 patients, ProstateX-2: 99 patients). We trained a 3D U-Net convolutional neural network (CNN) segmentation model using our entire dataset, and used that model as a template to train models on the challenge datasets. We also trained versions of the template model using ablated proportions of our dataset, and evaluated the relative benefit of those templates for the final models. Finally, we trained a version of the template model using an out-of-domain brain cancer dataset, and evaluated the relevant benefit of that template for the final models. We used five-fold cross-validation (CV) for all training and evaluation across our entire dataset.

Results

Our model achieves state-of-the-art performance on our large dataset (mean overall Dice 0.916, average Hausdorff distance 0.135 across CV folds). Using this model as a pre-trained template for refining on two external datasets significantly enhanced performance (30% and 49% enhancement in Dice scores respectively). Mean overall Dice and mean average Hausdorff distance were 0.912 and 0.15 for the ProstateX-2 dataset, and 0.852 and 0.581 for the PROMISE12 dataset. Using even small quantities of data to train the template enhanced performance, with significant improvements using 5% or more of the data.

Conclusion

We trained a state-of-the-art model using unrefined clinical prostate annotations and found that its use as a template model significantly improved performance in other prostate segmentation tasks, even when trained with only 5% of the original dataset.