Genome-wide profiling of 5-hydroxymethylcytosines in circulating cell-free DNA reveals population-specific pathways in the development of multiple myeloma

To the editor,

Multiple myeloma (MM) typically progresses from the precursor conditions of monoclonal gammopathy of undetermined significance (MGUS) and smoldering myeloma (SMM). Compared with European Americans (EA), African Americans (AA) are 2–3 times more likely to be diagnosed with MM. Genetic susceptibility, socioeconomic factors, and obesity do not fully explain the excess risk in AA. Clinical variations of MM between EA and AA suggest a biological cause of racial/ethnic disparities [1]. Although the importance of epigenetics to MM is recognized, previous studies have not investigated genome-wide 5-hydroxymethylcytosines (5hmC), a cytosine modification with a distinct genomic distribution and regulatory role from the more-investigated 5-methylcytosines (5mC) [2]. Reduced global 5hmC levels have been found in MM [3] and MM-specific hydroxymethylome is associated with cell proliferation and prognosis [4]. To improve understanding of the role of 5hmC in disparities in MM, we conducted a genome-wide 5hmC profiling using the 5hmC-Seal and the next-generation sequencing in circulating cell-free DNA (cfDNA) samples from 227 EA and 115 AA patients with newly diagnosed MM, SMM, and MGUS prospectively enrolled at the University of Chicago Medical Center between 2010 and 2017. (Additional file 1: Table S1; Additional file 2).

Overall, the captured 5hmC modifications in cfDNA were more abundant in gene bodies and depleted at the promoter regions (Fig. 1A). Using the Roadmap Epigenomics Project annotations as reference, we found that patient-derived 5hmC profiles were enriched in B-cell and T-cell-derived enhancer marks: H3K4me1 and H3K27ac (Fig. 1B).

Comparing MM and its precursors (MGUS + SMM), we identified 63 differential gene bodies at 5% FDR (false discovery rate) (Fig. 1C; Additional file 3: Table S2) after controlling for sex, age, and race/ethnicity. The KEGG pathway analysis identified several metabolism-related pathways (e.g., citrate cycle) that have been implicated in myeloma cell growth and proliferation as well as the pathogenesis of MM (Fig. 1D) (Additional file 3: Table S3).

Next, we identified 259 differential gene bodies (5% FDR) between EA and AA patients with MM (Additional file 4: Fig. S1; Additional file: 3: Table S4), of which 183 showed higher 5hmC modification levels in AA patients. Of note, LIN28A is one of the most frequently mutated genes reported in MM [5], while KANSL1, LRRC37A3, and ARL17B are in a region (chr17q21) with a segmental duplication that is primarily found in European descents [6]. We identified several metabolism related KEGG pathways (Additional file 3: Table S5). The co-expression network analysis revealed three modules showing different direction of enrichment between AA and EA (Fig. 1E). Furthermore, the protein–protein interaction network analysis identified several relevant hub genes (Fig. 1F–H). For example, FHL2 (Module 1) has been found to regulate hematopoietic stem cell functions [7] and the production of IL6 [8], a cytokine critical to myeloma cell proliferation. Low expression of FHL2 has also been associated with development of IgM myeloma [9].

We then compared MM and its precursors in EA and AA patients separately. We identified 36 and 4 differential gene bodies (5% FDR) in EA and AA patients, respectively (Fig. 2A, B; Additional file 3: Table S6-7). Simulation analyses showed that the number of shared differential genes between EA and AA reached a peak around the 500th rank (Fig. 2C). The majority (94.8%) of the top 500 differential gene bodies were distinct between EA and AA (Fig. 2D), suggesting that although there is mechanistic commonality of myelomagenesis, racial/ethnic heterogeneity exists. The pathway analysis of the top 500 differential gene bodies showed population-specific KEGG pathways (Fig. 2E; Additional file 3: Table S8), including various cancer-related signaling pathways in EA patients, but primarily metabolism-related pathways in AA patients. For example, PIK3CA, which was enriched in several cancer-related signaling pathways in EA patients only, is important for constitutive Akt activity in MM cells and the blockade of PIK3CA induces cell death [10]. In contrast, several ALDH family genes were enriched in metabolism-related pathways in AA patients only. Increased expression of ALDH1 in MM has been identified as a marker of tumor-initiating cells and is associated with chromosomal instability [11]. Specific transcriptional networks related to metabolisms have also been found to contribute to plasma cell growth and proliferation [12].

In conclusion, we identified population-specific 5hmC signatures and pathways that improved our understanding of the epigenetic mechanisms underlying the disparities in MM. These findings could be exploited for novel preventive strategies in high-risk populations in the future.