A multiplex blood-based assay targeting DNA methylation in PBMCs enables early detection of breast cancer

Breast cancer is the most common malignancy worldwide, with 2.3 million new cases and ~685,000 deaths annually. Early diagnosis reduces mortality but current imaging-based screening (mammography, ultrasound) has limitations, particularly in dense breasts and is operator-dependent, leading to false positives and negatives. DNA methylation regulates gene expression and changes early in tumorigenesis, making it a promising biomarker. Prior biomarker efforts focused on tumor tissue and circulating tumor-derived materials (CTCs, ctDNA, cfDNA), which face practical and analytical constraints. Recent studies indicate cancer-related DNA methylation alterations in peripheral blood immune cells across several cancers, suggesting host immune epigenetic reprogramming. The research question here is whether DNA methylation signatures in PBMCs can serve as noninvasive markers for early detection of breast cancer. The study’s purpose is to define BC-specific methylation changes in PBMCs and develop a clinically practical multiplex qMSP assay for early BC detection, evaluating performance against standard markers and imaging.

DNA methylation is a stable epigenetic modification implicated in cancer development, with promoter hypermethylation silencing tumor suppressor genes and widespread changes observed even in precancerous stages. Tissue and tumor-derived analytes (CTCs, ctDNA, cfDNA) have been explored; however, tissues are invasive to obtain, CTCs can be contaminated with normal cells, and ctDNA is low-abundance and fragmented, limiting clinical application. Evidence suggests peripheral blood cell methylation can be a complementary biomarker: cancer-specific methylation alterations have been reported in PBMCs or T cells for hepatocellular carcinoma, prostate, colorectal, and head and neck cancers. Differences between chronic hepatitis and HCC PBMC methylomes indicate immune system epigenetic remodeling during malignancy. For breast cancer specifically, whether PBMC methylation can detect disease progression was unclear, motivating the present work.

Study design and participants: Multicenter retrospective study across ten hospitals in six Chinese provinces (May 2020–July 2022). Of 820 enrolled, 39 were excluded (no pathology data n=5; benign lesions n=24; PBMC isolation failure n=3; insufficient/low-quality DNA n=7), leaving 781 PBMC samples: 366 BC, 290 normal controls (NC), and 125 other tumors (42 colorectal, 41 gastric, 42 lung). Cohorts: Discovery (Cohort I, n=80: 50 BC/30 NC) for methylome profiling and marker discovery; Validation phase (Cohort II, n=200: pyrosequencing set n=100: 50 BC/50 NC; TBS set n=100: 60 BC/40 NC) for marker verification; Multicenter clinical cohort (Cohort III, n=501: 206 BC, 170 NC, 125 other tumors), split into training (130 BC/115 NC) and validation (76 BC/55 NC) sets for assay development and evaluation. Additional 125 non-BC tumors used to assess specificity. PBMC processing and DNA: 1–2 mL peripheral blood collected, PBMCs isolated via Ficoll within 2 hours, stored at −80 °C. DNA extracted (TIANamp), quality-controlled; bisulfite conversion with Zymo EZ DNA Methylation-Gold. Genome-wide methylation profiling: Infinium Human Methylation 850K BeadChip. Data processing via CHAMP in R: probe filtering (detection p>0.01, low bead count, non-CpG, multi-hit, chr Y, SNP-related probes removed); normalization (BMIQ); batch effect analysis (SVA) and correction (ComBat); annotation (EPICanno.ilm10b4.hg19). Differentially methylated CpG positions (DMPs) defined by |Δβ| ≥ 0.06 and p ≤ 0.01 (uncorrected). Pathway enrichment by GO/KEGG (clusterProfiler), significance after multiple testing (p<0.05). Cell-type composition adjustment: PBMC subset proportions estimated by EpiDISH; group differences assessed; markers evaluated after adjustment. Marker selection: Combined approaches: LASSO (10-fold CV) yielded 33 markers; stringent filter |Δβ| ≥ 0.08 and p ≤ 0.0001 yielded eight; six overlapped. Two with primer design issues (cg21723696, cg14928964) excluded. Final eight candidates: four hypomethylated (cg14507403/LMTK2, cg09821790/SLC7A6, cg15694422/MGLL, cg27527887/intergenic) and four hypermethylated (cg11754974/TRDJ3, cg16652347/PLXNA4, cg13828440/KLRD1, cg18637238/KLRK1). Inclusion emphasized TSS1500 sites due to functional relevance. Marker validation: Pyrosequencing (Qiagen PyroMark Q48) and targeted bisulfite sequencing (Genesky; Illumina MiSeq 2×150 bp) in Cohort II. Statistical comparisons via Mann–Whitney U test; relationships to clinical variables assessed (ANOVA with Dunnett’s multiple comparisons). TCGA 450k arrays used to examine breast tumor vs adjacent normal methylation for cg16652347 and cg13828440. Assay development (BC-mqmsPCR): Multiplex qMSP targeting four hypermethylated markers (cg11754974, cg16652347, cg13828440, cg18637238) labeled with FAM; ACTB reference labeled with VIC. ACT computed as ACT = CTreference − CTbiomarker to represent methylation level. Primers/probes designed with MethPrimer; reactions in 10 µL using KAPA PROBE FAST qPCR Master Mix; cycling: 95 °C 3 min; 45 cycles of 95 °C 3 s, 58 °C 30 s. Analytical sensitivity assessed by serial dilutions of bisulfite-converted PBMC DNA (50%, 10%, 1%, 0.1%, 0.01%, 0.001%). Multiplex compared to uniplex qMSP for each marker. Clinical evaluation: In Cohort III training and validation sets, BC-mqmsPCR methylation levels compared between BC and NC. Diagnostic performance quantified by ROC/AUC, sensitivity, specificity; optimal cut-off (Youden’s index) determined in training (cut-off ACT 3.223), applied to validation set. Specificity against other cancers assessed (125 non-BC). Subgroup analyses: early-stage (stage 0/1), minimal tumor size thresholds (≤1.5 cm; also ≤2.5 cm for comparison with serum markers), hormone receptor status (ER/PR/HER2), lymph node status, Ki-67. Comparator tests and follow-up: Serum tumor markers CA153, CEA, CA125 collected where available to compare detection rates in early-stage and small tumors. Follow-up of 170 NC (17 positive, 153 negative at baseline; some lost to follow-up) for up to 2 years with imaging and pathology as indicated to observe incident BC.

• Genome-wide PBMC methylome profiling (850K array) identified 289 DMPs (194 genes) between BC (n=50) and NC (n=30): 112 hypermethylated (38.8%) and 177 hypomethylated (61.2%). TSS regions in BC PBMCs were relatively hypomethylated; DMP-associated genes were enriched in immune surveillance/immune editing pathways. • Eight candidate CpGs were prioritized; validation showed four hypermethylated markers consistently higher in BC by pyrosequencing and TBS: cg11754974 (TRDJ3, TSS1500), cg16652347 (PLXNA4, gene body), cg13828440 (KLRD1, TSS1500), cg18637238 (KLRK1, TSS1500). The four hypomethylated candidates did not validate. Differences persisted after adjusting for PBMC subset composition (higher monocytes, lower NK cells in BC vs NC). • Analytical performance: BC-mqmsPCR multiplex assay generated higher ΔCT signals than uniplex qMSP for each marker and achieved a quantitative detection limit of 0.01% vs 1% (cg11754974) and 0.1% (cg16652347, cg13828440, cg18637238) for single-marker qMSP. • Diagnostic performance (Cohort III): Training set AUC 0.925 (95% CI 0.89–0.96), sensitivity 83.1%, specificity 90.4% (cut-off ACT 3.223); Validation set AUC 0.918 (95% CI 0.86–0.98), sensitivity 80.3%, specificity 89.1%. The assay distinguished BC from other cancers (lung, gastric, colorectal) with AUC 0.913. • Early detection: Stage 0–1 BC AUC 0.940 (95% CI 0.90–0.98) training and 0.927 (95% CI 0.87–0.99) validation; minimal tumors ≤1.5 cm AUC 0.945 (95% CI 0.91–0.98) training and 0.936 (95% CI 0.88–0.99) validation. Sensitivities reached 93.3% (stage 0–1, training), 85.7% (validation); and 93.2% (≤1.5 cm, training), 90.0% (validation). • Clinical utility versus serum markers: In stage 0 BC (n=17), BC-mqmsPCR detected 88.2% (15/17) vs CEA 5.9% (1/17), CA153 0%, CA125 0%. For small nodules ≤1.5 cm, BC-mqmsPCR sensitivity was 91.7% vs CA153 0%, CA125 2.1%, CEA 0%. Overall detection among evaluated subsets: BC-mqmsPCR 82.0% vs CA153 5.3%, CEA 6.0%, CA125 3.5%. Among CA125-, CEA-, and CA153-negative BC patients, BC-mqmsPCR sensitivities were 82.42%, 81.41%, and 81.99%, respectively. • Case examples: Two BC cases (ductal carcinoma in situ) were positive by BC-mqmsPCR but missed by ultrasound, mammogram, and conventional tumor markers. • Follow-up of NC: Of 170 NC (17 baseline positives; 2 lost), two BC cases were confirmed within 1 year among BC-mqmsPCR positives; among 153 negatives (22 lost), no BC was confirmed within 1 year.

This study demonstrates that DNA methylation alterations in PBMCs reflect host immune responses to breast cancer and can be leveraged for noninvasive early detection. The PBMC methylome in BC showed immune pathway enrichment and TSS hypomethylation, consistent with immunologic activation. Four validated hypermethylated CpGs near/within TRDJ3, PLXNA4, KLRD1, and KLRK1—genes implicated in T/NK-cell receptor signaling and immune modulation—were combined into a multiplex qMSP assay (BC-mqmsPCR) with analytical sensitivity down to 0.01%. Clinically, the assay achieved high AUCs (~0.92) and specificities (~89–90%), and showed superior performance in early-stage and minimal tumors compared with traditional serum markers, detecting many cases missed by imaging. Unlike cfDNA/ctDNA methylation assays that often show reduced sensitivity at early stages, PBMC-based markers may benefit from early immune activation, explaining higher early-stage performance. Specificity against non-breast cancers was high, suggesting disease specificity. These findings support PBMC methylation as a practical liquid biopsy reflecting systemic immune epigenetic reprogramming in BC and provide a feasible, rapid, and cost-effective approach for early detection that could complement imaging.

The authors developed BC-mqmsPCR, a simple multiplex methylation-specific qPCR assay targeting four PBMC DNA methylation markers (cg11754974/TRDJ3, cg16652347/PLXNA4, cg13828440/KLRD1, cg18637238/KLRK1). The assay achieved high sensitivity and specificity for BC detection, particularly in early-stage disease and small tumors, outperforming conventional serum tumor markers and identifying cases missed by imaging. It also showed potential to flag cancers earlier than standard clinical methods. Future work should include prospective validation, longer-term follow-up to assess predictive value and patient outcomes, evaluation in populations with inflammatory or autoimmune conditions, and mechanistic studies to understand the roles of these methylation changes in BC development and progression.

• The study excluded patients with inflammatory and autoimmune diseases that may affect PBMC methylation, limiting generalizability to these populations. • Retrospective design; prospective studies are needed for confirmation. • Follow-up of healthy controls was relatively short, and there was no longitudinal assessment of BC patients for recurrence or survival. • Mechanistic understanding of the identified methylation changes remains incomplete. • Potential for false positives exists with PCR-based assays; rigorous laboratory controls are needed to minimize contamination.