Plasma cell-free DNA methylome-based liquid biopsy for accurate gastric cancer detection.

Early detection plays a critical role in mitigating mortality rates linked to gastric cancer. However, current clinical screening methods exhibit suboptimal efficacy. Methylation alterations identified from cell-free DNA (cfDNA) present a promising biomarker for early cancer detection. Our study focused on identifying gastric cancer-specific markers from cfDNA methylation to facilitate early detection. We enrolled 150 gastric cancer patients and 100 healthy controls in this study, and undertook genome-wide methylation profiling of cfDNA using cell-free methylated DNA immunoprecipitation and high-throughput sequencing. We identified 21 differentially methylated regions (DMRs) between the gastric tumor and nontumor groups using multiple algorithms. Subsequently, using the 21 DMRs, we developed a gastric cancer detection model by random forest algorithm in the discovery set, and validated the model in an independent set. The model was able to accurately discriminate gastric cancer with a sensitivity and specificity of 93.90% and 95.15% in the discovery set, respectively, and 88.38% and 94.23% in the validation set, respectively. These results underscore the efficacy and accuracy of cfDNA-derived methylation markers in distinguishing early stage gastric cancer. This study highlighted the significance of cfDNA methylation alterations in early gastric cancer detection.

Probabilistic modeling methods for cell-free DNA methylation based cancer classification.

BACKGROUND: cfMeDIP-seq is a low-cost method for determining the DNA methylation status of cell-free DNA and it has been successfully combined with statistical methods for accurate cancer diagnostics. We investigate the diagnostic classification aspect by applying statistical tests and dimension reduction techniques for feature selection and probabilistic modeling for the cancer type classification, and we also study the effect of sequencing depth. METHODS: We experiment with a variety of statistical methods that use different feature selection and feature extraction methods as well as probabilistic classifiers for diagnostic decision making. We test the (moderated) t-tests and the Fisher's exact test for feature selection, principal component analysis (PCA) as well as iterative supervised PCA (ISPCA) for feature generation, and GLMnet and logistic regression methods with sparsity promoting priors for classification. Probabilistic programming language Stan is used to implement Bayesian inference for the probabilistic models. RESULTS AND CONCLUSIONS: We compare overlaps of differentially methylated genomic regions as chosen by different feature selection methods, and evaluate probabilistic classifiers by evaluating the area under the receiver operating characteristic scores on discovery and validation cohorts. While we observe that many methods perform equally well as, and occasionally considerably better than, GLMnet that was originally proposed for cfMeDIP-seq based cancer classification, we also observed that performance of different methods vary across sequencing depths, cancer types and study cohorts. Overall, methods that seem robust and promising include Fisher's exact test and ISPCA for feature selection as well as a simple logistic regression model with the number of hyper and hypo-methylated regions as features.

Prediction model for malignant pulmonary nodules based on cfMeDIP-seq and machine learning.

Cell-free methylated DNA immunoprecipitation and high-throughput sequencing (cfMeDIP-seq) is a new bisulfite-free technique, which can detect the whole-genome methylation of blood cell-free DNA (cfDNA). Using this technique, we identified differentially methylated regions (DMR) of cfDNA between lung tumors and normal controls. Based on the top 300 DMR, we built a random forest prediction model, which was able to distinguish malignant lung tumors from normal controls with high sensitivity and specificity of 91.0% and 93.3% (AUROC curve of 0.963). In summary, we reported a non-invasive prediction model that had good ability to distinguish malignant pulmonary nodules.

Detection of ovarian cancer using plasma cell-free DNA methylomes.

BACKGROUND: Ovarian cancer (OC) is a highly lethal gynecologic cancer, and it is hard to diagnose at an early stage. Clinically, there are no ovarian cancer-specific markers for early detection. Here, we demonstrate the use of cell-free DNA (cfDNA) methylomes to detect ovarian cancer, especially the early-stage OC. EXPERIMENTAL DESIGN: Plasma from 74 epithelial ovarian cancer patients, 86 healthy volunteers, and 20 patients with benign pelvic masses was collected. The cfDNA methylomes of these samples were generated by cell-free methylated DNA immunoprecipitation and high-throughput sequencing (cfMeDIP-seq). The differentially methylated regions (DMRs) were identified by the contrasts between tumor and non-tumor groups, and the discrimination performance was evaluated with the iterative training and testing method. RESULTS: The DMRs identified for cfDNA methylomes can well discriminate tumor groups and non-tumor groups (ROC values from 0.86 to 0.98). The late-stage top 300 DMRs are more late-stage-specific and failed to detect early-stage OC. However, the early-stage markers have the potential to discriminate all-stage OCs from non-tumor samples. CONCLUSIONS: This study demonstrates that cfDNA methylomes generated with cfMeDIP-seq could be used to identify OC-specific biomarkers for OC, especially early OC detection. To detect early-stage OC, the biomarkers should be directly identified from early OC plasma samples rather than mix-stage ones. Further exploration of DMRs from a k larger early-stage OC cohort is warranted.

Longitudinal monitoring of cell-free DNA methylation in ALK-positive non-small cell lung cancer patients.

BACKGROUND: DNA methylation (5-mC) signals in cell-free DNA (cfDNA) of cancer patients represent promising biomarkers for minimally invasive tumor detection. The high abundance of cancer-associated 5-mC alterations permits parallel and highly sensitive assessment of multiple 5-mC biomarkers. Here, we performed genome-wide 5-mC profiling in the plasma of metastatic ALK-rearranged non-small cell lung cancer (NSCLC) patients receiving tyrosine kinase inhibitor therapy. We established a strategy to identify ALK-specific 5-mC changes from cfDNA and demonstrated the suitability of the identified markers for cancer detection, prognosis, and therapy monitoring. METHODS: Longitudinal plasma samples (n = 79) of 21 ALK-positive NSCLC patients and 13 healthy donors were collected alongside 15 ALK-positive tumor tissue and 10 healthy lung tissue specimens. All plasma and tissue samples were analyzed by cell-free DNA methylation immunoprecipitation sequencing to generate genome-wide 5-mC profiles. Information on genomic alterations (i.e., somatic mutations/fusions and copy number alterations) determined in matched plasma samples was available from previous studies. RESULTS: We devised a strategy that identified tumor-specific 5-mC biomarkers by reducing 5-mC background signals derived from hematopoietic cells. This was followed by differential methylation analysis (cases vs. controls) and biomarker validation using 5-mC profiles of ALK-positive tumor tissues. The resulting 245 differentially methylated regions were enriched for lung adenocarcinoma-specific 5-mC patterns in TCGA data and indicated transcriptional repression of several genes described to be silenced in NSCLC (e.g., PCDH10, TBX2, CDO1, and HOXA9). Additionally, 5-mC-based tumor DNA (5-mC score) was highly correlated with other genomic alterations in cell-free DNA (Spearman, ρ > 0.6), while samples with high 5-mC scores showed significantly shorter overall survival (log-rank p = 0.025). Longitudinal 5-mC scores reflected radiologic disease assessments and were significantly elevated at disease progression compared to the therapy start (p = 0.0023). In 7 out of 8 instances, rising 5-mC scores preceded imaging-based evaluation of disease progression. CONCLUSION: We demonstrated a strategy to identify 5-mC biomarkers from the plasma of cancer patients and integrated them into a quantitative measure of cancer-associated 5-mC alterations. Using longitudinal plasma samples of ALK-positive NSCLC patients, we highlighted the suitability of cfDNA methylation for prognosis and therapy monitoring.

Finding new cancer epigenetic and genetic biomarkers from cell-free DNA by combining SALP-seq and machine learning.

The effective non-invasive diagnosis and prognosis are critical for cancer treatment. The plasma cell-free DNA (cfDNA) provides a good material for cancer liquid biopsy and its worth in this field is increasingly explored. Here we describe a new pipeline for effectively finding new cfDNA-based biomarkers for cancers by combining SALP-seq and machine learning. Using the pipeline, 30 cfDNA samples from 26 esophageal cancer (ESCA) patients and 4 healthy people were analyzed as an example. As a result, 103 epigenetic markers (including 54 genome-wide and 49 promoter markers) and 37 genetic markers were identified for this cancer. These markers provide new biomarkers for ESCA diagnosis, prognosis and therapy. Importantly, these markers, especially epigenetic markers, not only shed important new insights on the regulatory mechanisms of this cancer, but also could be used to classify the cfDNA samples. We therefore developed a new pipeline for effectively finding new cfDNA-based biomarkers for cancers by combining SALP-seq and machine learning. In this study, we also discovered new clinical worth of cfDNA distinct from other reported characters.