Longitudinal epigenome-wide associations and pre-chemotherapy prediction of post-chemotherapy inflammation in patients with breast cancer undergoing chemotherapy.

Abstract

e24094 Background: Inflammation is linked with cancer outcomes; systemic cytokine levels are associated with cognitive and functional impairments during chemotherapy. Epigenetic changes, such as DNA methylation, are also associated with cytokine levels, yet the relationship between epigenetic changes and inflammation during breast cancer treatment remains unclear. Methods: We performed epigenome-wide methylation association analyses to identify CpG sites whose DNA methylation levels were associated with circulating inflammatory cytokine concentrations in a nationwide cohort of 241 participants (192 patients with breast cancer and 49 age- and sex-matched healthy controls). DNAm was measured using Illumina 450K and EPIC arrays within 7 days pre-chemotherapy (Timepoint [T] 1) and within 1 month post-chemotherapy (T2). DNA methylation was analyzed on the M-value scale (log2 ratio of methylated to unmethylated probe intensities), allowing more stable variance for linear modeling. Cytokines measured from serum included IL-4, IL-6, IL-8, IL-10, and TNF-α, as well as soluble TNF receptors I and II (sTNFRI/II). Independent CpG-wise linear regression models were fit using limma, adjusting for age, group (chemotherapy vs control), and array type (450K vs EPIC). Analyses examined for cross-sectional associations between DNAm and inflammation at T1 and T2, as well as prediction of T2 inflammation from T1 DNAm. False discovery rate (FDR) < 0.05 defined epigenome-wide significance. Results: Across all analyses (N = 241; mean age = 60 ± 8 years; 78% college degree or above; 71% married/long-term relationship), we tested 449,521 CpG sites and identified five CpG–biomarker associations meeting FDR < 0.05. T1 DNAm at cg08458121 (chr17:1,028,676; CpG island within the gene body/5′UTR region of ABR) was positively associated with T1 IL-10 (log2) (T1→T1; β = 0.224, FDR = 0.032, N = 228). T2 DNAm at cg08458121 (chr17:1,028,676; CpG island within the gene body/5′UTR region of ABR) was positively associated with T2 IL-10 (log2) (T2→T2; β = 0.233, FDR = 0.026, N = 222), demonstrating consistent direction of association across timepoints. T2 IL-4 (log2) was positively associated with T2 DNAm at cg01963906 (chr1:27,677,240; intragenic CpG island in synaptotagmin-like protein 1 [SYTL1]; β = 0.126, FDR = 0.019, N = 222). In longitudinal prediction models (T1→T2), T1 DNAm was inversely associated T2 sTNFRI at cg01479664 (chr16:58,016,655; TEPP; N_Shore; β = −0.188, p = 1.6×10⁻⁷, FDR = 0.041, N = 222) and cg03198678 chr3:194,381,165; LSG1; OpenSea; β = −0.091, p = 2.1×10⁻⁷, FDR = 0.041, N = 222). Conclusions: This epigenome-wide analysis identified DNAm sites associated with circulating inflammatory biomarker levels at pre-chemotherapy and post-chemotherapy in breast cancer.