Epigenome-wide association study of acute lymphoblastic leukemia in children with Down syndrome.

TO THE EDITOR:

Down syndrome (DS) is caused by constitutional trisomy of chromosome 21 and is associated with an up to 30-fold increased risk of acute lymphoblastic leukemia (ALL).[1,2] While DS is associated with alterations in epigenetic markers, including DNA methylation, and gene expression,[3-6] these mechanisms have not been fully explored in relation to DS-ALL etiology.[7] Because the epigenome is sensitive to genetic and environmental influences during fetal development and can be leveraged to characterize blood cell proportions,[8] we sought to evaluate the role of the neonatal methylome in children with DS on subsequent ALL risk.

Our epigenome-wide association study (EWAS) included 126 DS-ALL cases and 198 DS control subjects from the International Study of Down Syndrome Acute Leukemia[7,9] in the Discovery dataset and 24 cases and 24 control subjects from the Michigan-based DS-ALL study[7] in the Replication group. DNA was isolated from neonatal dried bloodspots, bisulfite-converted and assayed using Illumina EPIC methylation arrays. Further details on study subjects, quality control and processing of methylation array data, and statistical analyses are included in the supplemental Methods. The Institutional Review Boards of each participating site approved the study, which was conducted according to the Declaration of Helsinki.

Demographic and birth-related data are summarized in Table 1. Unsupervised hierarchical clustering did not differentiate DS-ALL cases from DS control subjects but did demonstrate variation in blood cell proportions, determined by reference-based deconvolution using the Identifying Optimal Libraries algorithm,[10] and identified a subset of DS newborns with high nucleated red blood cell proportions, as previously shown[6] (supplemental Figure 1).

Table 1.

Demographic and birth characteristics of DS-ALL cases and DS control subjects

	DS-ALL Discovery Study			DS-ALL Replication Study
	DS control subjects(n = 198), n (%)	DS-ALL(n = 126),n (%)	P value	DS control subjects(n = 24),n (%)	DS-ALL(n = 24),n (%)	P value
Sex
Male	91 (46.0)	84 (66.7)	–	14 (58.3)	13 (54.2)	–
Female	107 (54.0)	42 (33.3)	.00037*	10 (41.7)	11 (45.8)	.771*
Race/ethnicity
Asian	10 (5.1)	2 (1.6)	–	1 (4.2)	1 (4.2)	–
Latino	96 (48.5)	86 (68.3)	–	3 (12.5)	2 (8.3)	–
Non-Latino White	54 (27.3)	32 (25.4)	–	15 (62.5)	20 (83.3)	–
Non-Latino Black	10 (5.1)	2 (1.6)	–	5 (20.8)	1 (4.2)	–
Other	28 (14.1)	4 (3.2)	.00037*	0	0	.287*
Missing	0	0		0	0
Age at DS-ALL diagnosis (y)
Median (range)	–	4.0 (0-14.6)	–	–	<4.0, n = 13;≥4.0, n = 11†	–
Blood collection age (d)
Mean (SD)	2.47 (2.03)	2.03 (2.14)	.068	N/A	N/A	–
Median (range)	1.71 (0.17-15.25)	1.46 (0-18.96)	–	N/A	N/A	–
Missing	3 (1.5)	9 (7.1)	–	24 (100.0)	24 (100.0)	–
Gestational age (wk)
Mean (SD)	38.10 (2.33)	38.22 (2.83)	.67‡	N/A	N/A	–
Median (range)	38.29 (26.42-44.71)	38.43 (25.57-44.43)	–	N/A	N/A	–
Preterm (<37)	41 (22.7)	29 (24.0)	.78*	N/A	N/A	–
Missing	17 (8.6)	5 (4.0)	–	24 (100.0)	24 (100.0)	–
Birthweight (kg)
Mean (SD)	3.00 (0.74)	3.08 (0.60)	.31‡	N/A	N/A	–
Median (range)	3.02 (0.81-8.65)	3.12 (0.94-4.58)	–	N/A	N/A	–
Missing	4 (2.0)	1 (0.8)	–	24 (100.0)	24 (100.0)	–

P values calculated using a 2-tailed Fisher’s exact test.

Age-at-diagnosis only available in categories for DS-ALL cases in the Replication Study.

P values were calculated using a 2-tailed t test.

Deconvolution of blood cell proportions in the Discovery study revealed a significant increase in B-cell proportions at birth in DS-ALL cases (mean, 0.0128) compared with DS control subjects (mean, 0.00826; P = 8.58 × 10−4), a difference which was also observed in the Replication study (P = .03) and meta-analysis (effect sizemeta = 0.0056; Pmeta = 1.69 × 10−4; Phet = .15) (supplemental Figure 2 and Table 2). Among cell types, B cells showed the greatest proportional difference between cases and control subjects in both Discovery (+55.57% in DS-ALL) and Replication (+22.23%) studies (supplemental Table 1). An independent deconvolution method, Epigenetic Dissection of Intra-Sample-Heterogeneity (EpiDISH),[11] confirmed the increased B-cell proportions in DS-ALL cases in both studies (Pmeta = 1.67 × 10−4) (supplemental Table 2).

Table 2.

Deconvoluted blood cell proportions in DS-ALL cases vs DS controls

Cell type	Discovery Study			Replication Study			Meta-analysis
	(126 cases, 198 control subjects)			(24 cases, 24 control subjects)
	Effect estimate*	Standard error*	P value*	Effect estimate*	Standard error*	P value*	Effect estimate†	Standard error*	Pmeta†	Phet†	Direction
CD4 T cell	0.0036	0.0055	.51	−0.0147	0.0136	.29	0.0011	0.0051	.83	.21	−+
CD8 T cell	0.0071	0.0030	.016	0.0168	0.0102	.11	0.0079	0.0028	.0055	.36	++
B cell	0.0051	0.0015	8.58 × 10−4	0.0152	0.0069	.03	0.0056	0.0015	1.69 × 10−4	.15	++
NK cells	0.0028	0.0024	.24	0.0048	0.0079	.55	0.0030	0.0023	.19	.81	++
Granulocyte	0.0076	0.0178	.67	−0.0482	0.0376	.21	−0.0026	0.0161	.87	.18	−+
Monocyte	0.0010	0.0040	.81	−0.0003	0.0107	.98	0.0008	0.0038	.83	.91	−+
nRBC	−0.0301	0.0228	.19	0.0163	0.0357	.65	−0.0166	0.0192	.39	.27	+−

NK, natural killer; nRBC, nucleated red blood cells.

P < .05 highlighted in bold.

P values, coefficients, and standard errors calculated using linear regression, testing each blood cell type separately as the dependent variable, with DS-ALL status as the independent variable, and including sex, batch, and ancestry-related principal components from EPISTRUCTURE[12] (n = 10 for Discovery study; n = 3 for Replication study) as covariates. P values were not adjusted for multiple comparisons.

Meta-analysis performed using METAL.[13]

In analyses stratified by self-reported race and ethnicity in the Discovery study, increased neonatal B-cell proportions showed a stronger effect in Latinos (effect size = 0.0058; P = 6.15 × 10−3) than in non-Latino Whites (effect size = 0.0046; P = .098), although this difference was not statistically significant (Phet = .74) (supplemental Table 3).

We performed several sensitivity analyses in the Discovery study to assess potential confounders of the increased B-cell proportions in DS-ALL. First, in subjects with available birth-variable data, we adjusted the regression model for gestational age, birth weight, and bloodspot collection age, and the difference in B-cell proportions between DS-ALL cases (n = 116) and DS control subjects (n = 173) remained significant (effect size = 0.0059; P = 3.38 × 10−4).

Next, in Latino and non-Latino White subjects with single nucleotide polymorphism (SNP) genotype data (117 cases, 130 control subjects), we assessed whether SNPs associated with DS-ALL risk in ARID5B (rs7089424), IKZF1 (rs11978267), CDKN2A (rs3731249), or GATA3 (rs3824662)[7] may confound the association with B-cell proportions, as these loci were previously associated with variation in white blood cell traits.[14] We included the genotypes of these 4 SNPs in the regression model one at a time and also all together, and the significantly increased B-cell proportions in DS-ALL cases remained, with similar effect sizes in Latinos and non-Latino Whites (supplemental Table 4).

Finally, we removed GATA1 mutation-positive control subjects (n = 30 of 184 tested, see supplemental Methods), and the difference in B-cell proportions remained significant (effect size = 0.0043; P = 9.02 × 10−3).

In the Discovery study EWAS of DS-ALL (126 cases, 198 control subjects), the genomic inflation factor was 1.11 after correction with BACON, a Bayesian method to control bias and inflation in EWAS.[15] There were 38 significant differentially methylated probes (DMPs) after false discovery rate (FDR) correction and 10 epigenome-wide–significant DMPs after Bonferroni correction (P < 7.95 × 10−8) (supplemental Figure 3; supplemental Table 5). Pathway enrichment analysis of FDR-significant DMPs revealed significant enrichment of 21 gene ontology pathways (supplemental Table 6). The top DS-ALL–associated CpG (cg27347265; P = 2.90 × 10−12) was located in a putative regulatory region of the B-cell transcription factor gene EBF1 (supplemental Figure 4; supplemental Table 5). For all 10 Bonferroni-significant DMPs, the case-control methylation β-value difference was <0.02, and none were significant in the Replication study at P < .05, although 6 out of 10 had consistent directions of effect.

We identified 31 significant differentially methylated regions (DMRs) associated with DS-ALL in the Discovery study (supplemental Table 7). Although none of the DMRs were statistically significant in the Replication study, 4 of 31 contained significant (P < .05) differentially methylated CpGs with the same direction of methylation changes as the Discovery study (supplemental Table 7).

In summary, an increase in the neonatal proportion of B cells was associated with DS-ALL risk, a finding that persisted after adjustment for potential confounding factors and was consistent between 2 independent case-control datasets. DS is associated with reduced fetal B-cell production[16,17] and reduced numbers of B cells in fetal life[16,18] and childhood.[19,20] We previously observed lower B-cell proportions in newborns with DS than in newborns without DS using reference-based cell-type deconvolution analysis.[6] Results from the current study support that, in the context of DS, children with greater B-cell proportions at birth have an increased risk of developing DS-ALL. A genetic predisposition to overproducing lymphocytes was recently associated with increased ALL risk in the non-DS population.[14] Further studies are required to understand the mechanisms underlying the association between increased B cells and ALL development in children with and without DS, but these may involve effects on the proliferation of preleukemic clones and generation of leukemia-forming mutations, as well as potential impacts on immune function and response to infections.[14,21]

We did not find strong evidence for differences in DNA methylation at birth that might predict subsequent DS-ALL risk, although the Replication dataset was underpowered to reproduce significance for the small differences found between cases and controls in the Discovery study. The significant EBF1 DMP is intriguing given that this gene is frequently deleted in ALL.[22] Investigation of DNA methylation differences in sorted cell populations is required to determine cell-specific epigenetic changes associated with DS-ALL risk.

A strength of our study was the use of newborn DBS, collected before disease onset and, therefore, any case-control differences should not be confounded by the presence of leukemia cells; indeed, in the Discovery study, only 1 DS-ALL case was diagnosed <1 year of age and the B-cell case-control difference was significant both when restricted to cases with age-at-diagnosis ≤4 years (n = 64; effect size = 0.0039; P = .034) or >4 years of age (n = 62; effect size = 0.0061; P value = 1.01 × 10−3).

A study limitation includes the use of a blood cell proportion deconvolution methodology developed in euploid individuals,[10] although the same approach confirmed known differences in blood cell proportions associated with DS.[6] Nonetheless, the increased B-cell proportion in DS-ALL cases requires confirmation using blood cell count measures in newborns. Another limitation was that sequencing data for somatic GATA1 mutations, which cause transient abnormal myelopoiesis,[23] were only available for DS controls in the Discovery study; however, removal of GATA1 mutation-positive control subjects had minimal effect on the B-cell association.

Future studies are needed to understand the role of blood cell trait variation in DS-ALL etiology and examine increased neonatal B cells as a potential risk factor for ALL in the non-DS population.

Supplementary Material

The full-text version of this article contains a data supplement.

Click here for additional data file.

Click here for additional data file.