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

Cancer Sci. 2021 Sep;112(9):3918-3923. doi: 10.1111/cas.15052. Epub 2021 Jul 21.

Abstract

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.

Keywords: cfDNA methylation; cfMeDIP-seq; lung cancer; machine learning; pulmonary nodule.

Publication types

  • Evaluation Study

MeSH terms

  • Adult
  • Aged
  • Aged, 80 and over
  • Case-Control Studies
  • Cell-Free Nucleic Acids / genetics*
  • DNA Methylation*
  • Female
  • High-Throughput Nucleotide Sequencing
  • Humans
  • Immunoprecipitation
  • Lung Neoplasms / genetics
  • Lung Neoplasms / pathology*
  • Machine Learning*
  • Male
  • Middle Aged
  • Multiple Pulmonary Nodules / genetics
  • Multiple Pulmonary Nodules / pathology*
  • Prognosis
  • Sensitivity and Specificity

Substances

  • Cell-Free Nucleic Acids