Page 73 - Read Online

Page 4 Moore et al. J Transl Genet Genom 2021;5:200-217 https://dx.doi.org/10.20517/jtgg.2021.08

Disentangling the heritable component of non-Hodgkin lymphoma (NHL) and its subtypes is an active area
of research. An early study of familial aggregation in NHL reported an increased risk of NHL among
[1]
siblings, but not parents or offspring, of an index NHL case . Several subsequent studies found an elevated
[2,3]
risk of NHL associated with a first-degree family history of NHL with the highest risks for siblings .
Studies have also reported higher NHL subtype-specific risks for first-degree relatives of cases affected with
[4,5]
a given NHL subtype , suggesting a degree of subtype specificity. In general, these findings suggest genetic
factors are important in NHL etiology and, in particular, the potential role of recessively acting genetic risk
alleles, but they also underscore the potential for genetic heterogeneity in susceptibility to different NHL
subtypes. Part of the difficulty in characterizing risk and inheritance patterns of NHL subtypes is the limited
study sample size, especially when examining specific subtypes. Genome-wide association studies (GWAS)
have identified multiple susceptibility loci associated with four major subtypes of NHL [6-11] , but a substantial
fraction of the disease heritability remains unexplained. Most GWAS performed assume an additive model
of genetic risk, which has statistical power to detect allelic associations acting through a variety of
mechanisms but may not efficiently detect recessive effects, particularly as minor allele frequency and
[12]
imputation quality decrease . Therefore, recessively acting loci, particularly those with low minor allele
frequency, could be missed by current genome-wide scans and represent potential novel disease-associated
loci.

The widespread use of dense genotyping arrays has led to the identification of sizeable genomic regions
[13]
consisting of consecutive homozygous SNPs in non-consanguineous populations . These runs of
homozygosity (ROH) vary in length, with short ROH persisting from ancient relatedness and long ROH of
several megabases arising from recent parental relatedness . The use of ROH as a measure of the burden of
[14]
homozygosity has been demonstrated to perform better at identifying rare, recessive mutations than a
conventional SNP-by-SNP analysis . Furthermore, studies incorporating whole-exome sequencing have
[15]
uncovered an enrichment of deleterious variants in ROH . In recent years, studies have examined the
[16]
association between ROH and various cancers [17-21] , among other complex common diseases and traits .
[22]
Although no association was observed with the cumulative distribution of ROH, individual ROH were
associated with the risk of childhood acute lymphoblastic leukemia . Hodgkin lymphoma has been
[19]
inconsistently associated with specific ROH and overall homozygosity [20,21] . To our knowledge, no studies
have examined ROH in association with adult NHL.

The goal of the present study was to investigate the association of homozygosity with the risk of four major
NHL subtypes: chronic lymphocytic leukemia/small chronic lymphocytic leukemia (CLL), diffuse large B-
cell lymphoma (DLBCL), follicular lymphoma (FL), and marginal zone lymphoma (MZL). Several measures
of homozygosity were tested against NHL risk using data from eight GWAS.

METHODS
We used data from eight previous GWAS of NHL [6,8,10,11] composed of cases and controls of European
ancestry [Supplementary Table 1 and 2]. The National Cancer Institute (NCI) NHL GWAS included cases
with one of four common NHL subtypes and controls from 22 studies of NHL: 9 prospective cohort studies,
8 population-based case-control studies, and 5 hospital- or clinic-based case-control studies or case series.
These 22 studies comprising the NCI NHL GWAS were genotyped using the Illumina OmniExpress or
Omni2.5 arrays and analyzed as a single study. The other seven GWAS were the University of California at
San Francisco Molecular Epidemiology of Non-Hodgkin Lymphoma study (UCSF2) , the University of
[23]
California at San Francisco Molecular Epidemiology of Non-Hodgkin Lymphoma study (UCSF1) combined
with controls from the Nurses’ Health Study (NHS) , the Scandinavian Lymphoma Etiology Study
[24]
[25]
(SCALE) , the Groupe d’Etude des Lymphomes de l’Adulte (GELA) study combined with controls from