Similar blood-borne DNA methylation alterations in cancer and inflammatory diseases determined by subpopulation shifts in peripheral leukocytes

Abstract

Background: Although many DNA methylation (DNAm) alterations observed in peripheral whole blood/leukocytes and serum have been considered as potential diagnostic markers for cancer, their origin and their specificity for cancer (e.g., vs inflammatory diseases) remain unclear.

Methods: From publicly available datasets, we identified changes in the methylation of blood-borne DNA for multiple cancers and inflammatory diseases. We compared the identified changes with DNAm difference between myeloid and lymphoid cells extracted from two datasets.

Results: At least 94.7% of the differentially methylated DNA loci (DM loci) observed in peripheral whole blood/leukocytes and serum of cancer patients overlapped with DM loci that distinguish between myeloid and lymphoid cells and >99.9% of the overlapped DM loci had consistent alteration states (hyper- or hypomethylation) in cancer samples compared to normal controls with those in myeloid cells compared to lymphoid cells (binomial test, P-value <2.2 × 10(-16)). Similar results were observed for DM loci in peripheral whole blood/leukocytes in patients with rheumatoid arthritis or inflammatory bowel diseases. The direct comparison between DM loci observed in the peripheral whole blood/leukocytes of patients with inflammatory diseases and DM loci observed in the peripheral whole blood of patients with cancer showed that DM loci detected from cancer and inflammatory diseases also had significantly consistent alteration states (binomial test, P-value <2.2 × 10(-16)).

Conclusions: DNAm changes observed in the peripheral whole blood/leukocytes and serum of cancer patients and in the peripheral whole blood/leukocytes of inflammatory disease patients are predominantly determined by the increase of myeloid cells and the decrease of lymphoid cells under the disease conditions, in the sense that their alteration states in disease samples compared to normal controls mainly reflect the DNAm difference between myeloid and lymphoid cells. These analyses highlight the importance of comparing cancer and inflammatory disease directly for the identification of cancer-specific diagnostic biomarkers.

Figure 1

Overlap of the DM loci of cancer and inflammatory diseases with M-L DM loci. For each dataset, light and dark grey bars represent percentages of DM loci detected in this dataset that overlapped/did not overlap with the M-L DM loci list, respectively. DM loci numbers are depicted in brackets after the corresponding percentages. DM=differentially methylated; M-L DM loci=DM loci in myeloid cells compared to lymphoid cells; HNSCC_PB and OVC_PB=DM loci detected from peripheral whole blood of head and neck squamous cell carcinoma and ovarian cancer samples, respectively; HNSCC_SR=DM loci from the serum samples of head and neck squamous cell carcinoma; PAC and SCLC=DM loci from peripheral leukocytes of pancreatic and small-cell lung cancer samples, respectively; RA_PL and IBD=DM loci detected from peripheral leukocytes of rheumatoid arthritis samples and peripheral whole blood of inflammatory bowel disease samples, respectively.