Development of a prediction model for radiosensitivity using the expression values of genes and long non-coding RNAs

Developing Tumor Radiosensitivity Signatures Using LncRNAs

Long non-coding RNAs (lncRNAs) are involved in diverse biological processes, including DNA damage repair, and are of interest as potential biomarkers of radiosensitivity. We investigated whether lncRNA radiosensitivity signatures could be derived for use in cancer patients treated with radiotherapy. Signature development involved radiosensitivity measurements for cell lines and primary tumor samples, and patient outcome after radiotherapy. A 10-lncRNA signature trained on radiosensitivity measurements in bladder cell lines showed a trend towards independent validation. In multivariable analyses, patients with tumors classified as radioresistant by the lncRNA signature had poorer local relapse-free survival (P = 0.065) in 151 patients with muscle-invasive bladder cancer who underwent radiotherapy. An mRNA-based radiosensitivity index signature performed similarly to the lncRNA bladder signature for local relapse-free survival (P = 0.055). Pathway analysis showed the lncRNA signature associated with molecular processes involved in radiation responses. Knockdown of one of the lncRNAs in the signature showed a modest increase in radiosensitivity in one cell line. An alternative approach involved training on primary cervical tumor radiosensitivity or local control after radiotherapy. Both approaches failed to generate a cervix lncRNA radiosensitivity signature, which was attributed to the age of samples in our cohorts. Our work highlights challenges in validating lncRNA signatures as biomarkers in archival tissue from radiotherapy cohorts, but supports continued investigation of lncRNAs for a role in radiosensitivity.

Developing a Prognostic Gene Panel of Epithelial Ovarian Cancer Patients by a Machine Learning Model

Epithelial ovarian cancer patients usually relapse after primary management. We utilized the support vector machine algorithm to develop a model for the chemo-response using the Cancer Cell Line Encyclopedia (CCLE) and validated the model in The Cancer Genome Atlas (TCGA) and the GSE9891 dataset. Finally, we evaluated the feasibility of the model using ovarian cancer patients from our institute. The 10-gene predictive model demonstrated that the high response group had a longer recurrence-free survival (RFS) (log-rank test, p = 0.015 for TCGA, p = 0.013 for GSE9891 and p = 0.039 for NTUH) and overall survival (OS) (log-rank test, p = 0.002 for TCGA and p = 0.016 for NTUH). In a multivariate Cox hazard regression model, the predictive model (HR: 0.644, 95% CI: 0.436⁻0.952, p = 0.027) and residual tumor size < 1 cm (HR: 0.312, 95% CI: 0.170⁻0.573, p < 0.001) were significant factors for recurrence. The predictive model (HR: 0.511, 95% CI: 0.334⁻0.783, p = 0.002) and residual tumor size < 1 cm (HR: 0.252, 95% CI: 0.128⁻0.496, p < 0.001) were still significant factors for death. In conclusion, the patients of high response group stratified by the model had good response and favourable prognosis, whereas for the patients of medium to low response groups, introduction of other drugs or clinical trials might be beneficial.

Long Non-Coding RNAs in Gliomas: From Molecular Pathology to Diagnostic Biomarkers and Therapeutic Targets

Gliomas are the most common malignancies of the central nervous system. Because of tumor localization and the biological behavior of tumor cells, gliomas are characterized by very poor prognosis. Despite significant efforts that have gone into glioma research in recent years, the therapeutic efficacy of available treatment options is still limited, and only a few clinically usable diagnostic biomarkers are available. More and more studies suggest non-coding RNAs to be promising diagnostic biomarkers and therapeutic targets in many cancers, including gliomas. One of the largest groups of these molecules is long non-coding RNAs (lncRNAs). LncRNAs show promising potential because of their unique tissue expression patterns and regulatory functions in cancer cells. Understanding the role of lncRNAs in gliomas may lead to discovery of the novel molecular mechanisms behind glioma biological features. It may also enable development of new solutions to overcome the greatest obstacles in therapy of glioma patients. In this review, we summarize the current knowledge about lncRNAs and their involvement in the molecular pathology of gliomas. A conclusion follows that these RNAs show great potential to serve as powerful diagnostic, prognostic, and predictive biomarkers as well as therapeutic targets.

PR-LncRNA signature regulates glioma cell activity through expression of SOX factors

Long non-coding RNAs (LncRNAs) have emerged as a relevant class of genome regulators involved in a broad range of biological processes and with important roles in tumor initiation and malignant progression. We have previously identified a p53-regulated tumor suppressor signature of LncRNAs (PR-LncRNAs) in colorectal cancer. Our aim was to identify the expression and function of this signature in gliomas. We found that the expression of the four PR-LncRNAs tested was high in human low-grade glioma samples and diminished with increasing grade of disease, being the lowest in glioblastoma samples. Functional assays demonstrated that PR-LncRNA silencing increased glioma cell proliferation and oncosphere formation. Mechanistically, we found an inverse correlation between PR-LncRNA expression and SOX1, SOX2 and SOX9 stem cell factors in human glioma biopsies and in glioma cells in vitro. Moreover, knock-down of SOX activity abolished the effect of PR-LncRNA silencing in glioma cell activity. In conclusion, our results demonstrate that the expression and function of PR-LncRNAs are significantly altered in gliomagenesis and that their activity is mediated by SOX factors. These results may provide important insights into the mechanisms responsible for glioblastoma pathogenesis.

Machine Learning and Radiogenomics: Lessons Learned and Future Directions

Due to the rapid increase in the availability of patient data, there is significant interest in precision medicine that could facilitate the development of a personalized treatment plan for each patient on an individual basis. Radiation oncology is particularly suited for predictive machine learning (ML) models due to the enormous amount of diagnostic data used as input and therapeutic data generated as output. An emerging field in precision radiation oncology that can take advantage of ML approaches is radiogenomics, which is the study of the impact of genomic variations on the sensitivity of normal and tumor tissue to radiation. Currently, patients undergoing radiotherapy are treated using uniform dose constraints specific to the tumor and surrounding normal tissues. This is suboptimal in many ways. First, the dose that can be delivered to the target volume may be insufficient for control but is constrained by the surrounding normal tissue, as dose escalation can lead to significant morbidity and rare. Second, two patients with nearly identical dose distributions can have substantially different acute and late toxicities, resulting in lengthy treatment breaks and suboptimal control, or chronic morbidities leading to poor quality of life. Despite significant advances in radiogenomics, the magnitude of the genetic contribution to radiation response far exceeds our current understanding of individual risk variants. In the field of genomics, ML methods are being used to extract harder-to-detect knowledge, but these methods have yet to fully penetrate radiogenomics. Hence, the goal of this publication is to provide an overview of ML as it applies to radiogenomics. We begin with a brief history of radiogenomics and its relationship to precision medicine. We then introduce ML and compare it to statistical hypothesis testing to reflect on shared lessons and to avoid common pitfalls. Current ML approaches to genome-wide association studies are examined. The application of ML specifically to radiogenomics is next presented. We end with important lessons for the proper integration of ML into radiogenomics.

A multiobjective Bayesian networks approach for joint prediction of tumor local control and radiation pneumonitis in nonsmall-cell lung cancer (NSCLC) for response-adapted radiotherapy

PURPOSE

Individualization of therapeutic outcomes in NSCLC radiotherapy is likely to be compromised by the lack of proper balance of biophysical factors affecting both tumor local control (LC) and side effects such as radiation pneumonitis (RP), which are likely to be intertwined. Here, we compare the performance of separate and joint outcomes predictions for response-adapted personalized treatment planning.

METHODS

A total of 118 NSCLC patients treated on prospective protocols with 32 cases of local progression and 20 cases of RP grade 2 or higher (RP2) were studied. Sixty-eight patients with 297 features before and during radiotherapy were used for discovery and 50 patients were reserved for independent testing. A multiobjective Bayesian network (MO-BN) approach was developed to identify important features for joint LC/RP2 prediction using extended Markov blankets as inputs to develop a BN predictive structure. Cross-validation (CV) was used to guide the MO-BN structure learning. Area under the free-response receiver operating characteristic (AU-FROC) curve was used to evaluate joint prediction performance.

RESULTS

Important features including single nucleotide polymorphisms (SNPs), micro RNAs, pretreatment cytokines, pretreatment PET radiomics together with lung and tumor gEUDs were selected and their biophysical inter-relationships with radiation outcomes (LC and RP2) were identified in a pretreatment MO-BN. The joint LC/RP2 prediction yielded an AU-FROC of 0.80 (95% CI: 0.70-0.86) upon internal CV. This improved to 0.85 (0.75-0.91) with additional two SNPs, changes in one cytokine and two radiomics PET image features through the course of radiotherapy in a during-treatment MO-BN. This MO-BN model outperformed combined single-objective Bayesian networks (SO-BNs) during-treatment [0.78 (0.67-0.84)]. AU-FROC values in the evaluation of the MO-BN and individual SO-BNs on the testing dataset were 0.77 and 0.68 for pretreatment, and 0.79 and 0.71 for during-treatment, respectively.

CONCLUSIONS

MO-BNs can reveal possible biophysical cross-talks between competing radiotherapy clinical endpoints. The prediction is improved by providing additional during-treatment information. The developed MO-BNs can be an important component of decision support systems for personalized response-adapted radiotherapy.

Differential miRNA expression profiling reveals miR-205-3p to be a potential radiosensitizer for low- dose ionizing radiation in DLD-1 cells

Enhanced radiosensitivity at low doses of ionizing radiation (IR) (0.2 to 0.6 Gy) has been reported in several cell lines. This phenomenon, known as low doses hyper-radiosensitivity (LDHRS), appears as an opportunity to decrease toxicity of radiotherapy and to enhance the effects of chemotherapy. However, the effect of low single doses IR on cell death is subtle and the mechanism underlying LDHRS has not been clearly explained, limiting the utility of LDHRS for clinical applications. To understand the mechanisms responsible for cell death induced by low-dose IR, LDHRS was evaluated in DLD-1 human colorectal cancer cells and the expression of 80 microRNAs (miRNAs) was assessed by qPCR array. Our results show that DLD-1 cells display an early DNA damage response and apoptotic cell death when exposed to 0.6 Gy. miRNA expression profiling identified 3 over-expressed (miR-205-3p, miR-1 and miR-133b) and 2 down-regulated miRNAs (miR-122-5p, and miR-134-5p) upon exposure to 0.6 Gy. This miRNA profile differed from the one in cells exposed to high-dose IR (12 Gy), supporting a distinct low-dose radiation-induced cell death mechanism. Expression of a mimetic miR-205-3p, the most overexpressed miRNA in cells exposed to 0.6 Gy, induced apoptotic cell death and, more importantly, increased LDHRS in DLD-1 cells. Thus, we propose miR-205-3p as a potential radiosensitizer to low-dose IR.