2022 / 06

Next Paper
DOI: 10.24075/brsmu.2022.056

ORIGINAL RESEARCH

Identification of prognostically significant DNA methylation signatures in patients with various breast cancer types

Kalinkin AI1, Sigin VO1, Nemtsova MV1,2, Strelnikov VV1,3
About authors

1 Research Centre for Medical Genetics, Moscow, Russia

2 Sechenov First Moscow State Medical University (Sechenov University), Moscow, Russia

3 Pirogov Russian National Research Medical University, Moscow, Russia

Correspondence should be addressed: Alexey I. Kalinkin
Moskvorechye, 1, Moscow, 115522; ur.xednay@2akiexela

About paper

Funding: the study was supported by the Ministry of Science and Higher Education of the Russian Federation within the framework of the Federal Scientific and Technical Program for the Development of Genetic Technologies in 2019–2027 (agreement № 075-15-2021-1073).

Author contribution: Kalinkin AI — study design, data acquisition, analysis and interpretation, manuscript writing; Sigin VO — manuscript writing; Nemtsova MV — study concept and design; Strelnikov VV — study concept and design, scientific editing.

Received: 2022-10-18 Accepted: 2022-11-11 Published online: 2022-11-25
|
Fig. 1. cvROC curves (cross-validated receiver operative curve; ROC curve is plotted at every stage of cross-validation, then the resulting curve is constructed) for the best signatures. The vertical axis shows sensitivity (0–1), the horizontal axis shows specificity (0–1), rows show survival outcomes, columns show BC molecular subtypes
Fig. 2. Kaplan–Meier curves for the best signatures. The horizontal axis shows time (years), the vertical axis shows the likelihood of staying alive (0–1). High risk of death, progression and relapse is highlighted in red, low risk of death, progression and relapse is highlighted in turquoise. Rows show survival outcomes, and columns show BC molecular subtypes
Table 1. Clinicopathological characteristics and data on the clinical endpoint status of patients with LumAB, TNBC and HER2-enriched BC molecular subtypes taken from open source (TCGA-BRCA)
Table 2. Total number of signatures and CpG pairs obtained by the LASSO Cox regression method for each survival outcome and BC molecular subtype. For some CpG pairs it was impossible to define genes these belonged to
Table 3. The best signature, number of probes in the signature, values of cvAUC (cross-validated area under curve; the average area under curve obtained at all stages of cross-validation), sensitivity, specificity and accuracy for each survival outcome and BC molecular subtype
Table 4. Multivariate Cox regression results for the best signatures and clinicopathological characteristics. HR — hazard ratio (relative risk), P — responsible for p-val