1 Introduction
Fanconi anemia (FA) is the most commonly inherited bone marrow failure syndrome (BMFS) with increased cancer incidence, especially in hematologic malignancies. Patients with homozygous or compound heterozygous FA gene mutations are easily identified, making timely cancer monitoring and prevention strategies practical. The notion that FA heterozygotes might also be predisposed to bone marrow failure and cancers has been hypothesized in the 1970s and investigated extensively thereafter [
1]. Meanwhile, congenital susceptibility to hematological malignancies is gaining much attention. A major change in the 2016 revision to the World Health Organization (WHO) classification of tumors of hematopoietic and lymphoid tissues is the added section on myeloid neoplasms with germline predisposition with FA listed as one of the classic disorders [
2]. However, clarification on whether heterozygotes are also predisposed is lacking.
FA-related genes participate in DNA homologous recombination repair pathway called the FA pathway with some classic tumor-suppressive genes, such as
BRCA2,
BRIP1, and
PALB2 [
3]. Monoallelic mutations in these genes significantly increase the risk of developing various cancers, such as breast, ovarian, and pancreatic cancers [
4]. Whether the risks of hematopoietic degeneration and malignancies are also increased in the context of monoallelic FA genetic deficits is a crucial issue in prevention strategies and family consulting. Some
in vitro clues support the hypothesis that cells carrying heterozygous FA mutations have a direct tumor transformation, and FA heterozygotes have a higher index of diepoxybutane-induced chromosome breakage compared with controls [
1,
5]. However, in terms of population studies, different research groups have yielded conflicting results [
6–
9]. To address this issue, we retrospectively analyzed rare possibly significant variants (PSVs) in the five most obligated FA genes,
BRCA2,
FANCA,
FANCC,
FANCD2, and
FANCG, in aplastic anemia (AA) and patients with hematologic malignancy.
2 Materials and methods
2.1 Patients
Patients diagnosed with AA, myelodysplastic syndrome (MDS), acute myeloid leukemia (AML), and acute lymphocytic leukemia (ALL) from April 2015 to December 2018 were enrolled. Those diagnosed with FA via chromosome breakage tests and/or genetic tests were excluded. Detailed disease histories and workup files were retrieved from electronic medical records. AA was defined as two or three lineages with decreased whole blood cell count, low reticulocyte count, and hematopoietic hypoplasia as indicated by bone marrow morphological analysis via Wright–Giemsa staining and/or pathological analysis through hematoxylin–eosin staining. The diagnosis of AML, MDS, and ALL was in accordance with the WHO classification of tumors of hematopoietic and lymphoid tissues [
2]. Indicative medical history and family history referred to (1) chronic AA with disease course≥3 years; (2) chronic or lineage hypoplasia (≥3 years) before the onset of MDS/AML; (3) with associated or syndromic features, including typically FA-related congenital malformations; (4) any other malignant histories; (5) AA history of next-of-kin within three generations and/or collateral relatives within two generations; (6) chronic or lineage hypoplasia (≥3 years) of next-of-kin within three generations and/or collateral relatives within two generations; and (7) any malignant histories of next-of-kin within three generations and/or collateral relatives within two generations.
All procedures followed were in accordance with the ethical standards of the Ethical Committee of Hebei Yanda Lu Daopei Hospital on human experimentation and with the Helsinki Declaration of 1975, as revised in 2000. Informed consent was obtained from all patients for participating in the study.
2.2 Gene sequencing and variant calling
Amplicon-based high-throughput sequencing was performed as described previously [
10]. Variant calling was performed with Torrent Variant Caller (TVC 5.0-13; Thermo Fisher Scientific). Final confident variants were annotated using Annovar. Variants with minor allele frequency (MAF)≥0.1% in the general population were filtered out according to 1000 Genomes and Exome Aggregation Consortium (ExAC) databases. The significance of the germline missense variants was assessed using
in silico prediction algorithms, including SIFT, Polyphen2, PROVEAN, FATHMM, MutationTaster, and MutationAssessor. Mutations with hypomorphic effect or incomplete penetrance were covered using a loose criterion that defined variants predicted as deleterious/possibly deleterious by≥3/6 scoring tools as rare PSV and further including them in the statistical analysis. For splice site mutations, GeneSplicer, Human Splicing Finder, NetGene2, and FSPLICE were employed, and only when≥2/4 algorithms predicted that it can affect/possibly affect splicing was the variant regarded as possibly significant. The same criteria were adopted when stratifying variants were recorded in the ExAC database. All reported variants in this study were confirmed germline variants by Sanger sequencing with fingernail specimens and/or pedigree analysis.
2.3 Statistical analysis
Fisher exact two-tailed test was adopted for variant frequency comparison between groups. Development of disease was analyzed using cumulative incidence method with NCSS v12.0.2, and P<0.05 was considered as statistically significant.
3 Results
3.1 Demographics
Demographic data, including median age of onset, sex ratio, and case number of each disease subgroup, are shown in Table 1. Patients of the whole cohort were of East Asian ancestry.
3.2 Features of variants
Sixty-eight variants in total met the criteria of rare PSVs in 66 patients (66/788, 8.38%; case AA_41 and MDS_55 harbor a heterozygous
FANCD2 variant and a heterozygous
FANCA variant each), including three stop-gain mutations, three splicing variants, and four
BRCA2 frameshift variants (Table S1, Fig. 1). In line with a previous study, the most frequently mutated gene was
FANCA (
n = 29) followed by
BRCA2 (
n = 20), and the three other genes were rarely mutated [
8]. Two recurrent variants,
FANCD2 c.3973C>A/p.L1325M and
FANCC c.436_438del/p.Y146del, were identified.
FANCD2 c.3973C>A, with uncertain significance as classified using ClinVar database, was carried by three different hosts (AA_026, AML_040, and ALL_181) with the incidence of 0.19% in this cohort and 0.02% in the East Asian assemblage of the ExAC database (odds ratio (OR) = 8.250, 95% confidence interval (CI) 1.378–49.416;
P = 0.025). No other overlaps among carriers were found in the rest of the variants. Eight PSVs were assigned as likely pathogenic or pathogenic by ClinVar and/or Leiden Open Variation Database v. 3.0. 22 PSVs were novel and were absent from population and disease databases and not reported elsewhere (Table S1).
3.3 Indicative medical history
A total of 138 patients (17.51%) in our cohort were positive for indicative medical history and/or family history with higher positive rates in AA and MDS subgroups than ALL subgroup (P<0.001, P = 0.003). This result was in accordance with the characteristics of our cohort with an early onset, especially in AA and MDS subgroups, for the innate predisposition harbored in this cohort. However, the distribution of patients with indicative medical history was even in FA-PSV positive (FA-PSV+) and FA-PSV negative (FA-PSV–) subgroups in all disease categories.
3.4 Risks and cumulative incidence
The annotated exome sequencing data of 4327 East Asians were obtained from the ExAC database, which yielded 8654 alleles of each gene and 270 rare PSVs with the same criteria used in our cohort. Although FANCA alterations are detected in the majority of patients with FA, BRCA2 was the top mutated gene in the general population for 77 mutated alleles with a frequency of 0.89%, followed by FANCD2 (70/8654, 0.81%) and FANCA (68/8654, 0.75%) (Table S2).
The overall frequency of rare PSVs was high in this cohort (270/4327 vs. 68/788, P = 0.016). However, there was no statistical difference in variant frequencies between each disease group and the ExAC data set when calculating separately (Table 2). BRCA2 PSVs showed a higher frequency in ALL (OR= 2.075, 95% CI 1.061–4.056; P = 0.038), and FANCA PSV frequency was significantly higher in AA and AML subgroups compared with that in East Asian assemblage in ExAC (OR= 2.409, 95% CI 1.187–4.888; P = 0.020; OR= 2.833, 95% CI 1.393–5.759; P = 0.008) (Table 2). The patients with FANCA PSVs also tended to show a higher risk for developing MDS, but the risk was not statistically significant (OR= 3.24, 95% CI 0.991–10.595; P = 0.075). The difference in the variant frequencies of FANCC, FANCD2, and FANCG in disease groups and the ExAC East Asian data set was not significant.
Age-related cumulative disease incidence was estimated when each subgroup was divided into FA-PSV+ and FA-PSV− subgroups. FA-PSV status had no effect on disease incidence in all of the four subgroups (Fig. 2A–2D). We further compared the age-related cumulative disease incidence with variant status in specific genes, including FANCA in AA, MDS, and AML and BRCA2 in ALL for the statistically higher mutated rate of the two genes in each disease category (Fig. S1A–S1D). However, the cumulative disease incidence between mutated and nonmutated groups was not different. As such, although variants of particular genes increased the odds of disease onset, the eventual onset of the disease is more complex and related to other innate aberration and environmental exposure.
3.5 Somatic genetic aberrations
We compared acquired genetic abnormalities, including somatic mutations and cytogenic aberrations, between the FA-PSV
+ and FA-PSV
− subgroups. Fifty-six or 86 genes that recurrently mutated in hematologic malignancies were analyzed following the protocol we previously reported [
11]. As a result, 770 somatic mutations were detected in total (Fig. 3). There were 48 and 430 somatic mutations detected in FA-PSV
+ and FA-PSV
− patients (48/25 vs. 430/244,
P = 0.024), respectively, in the MDS/AML subgroup (Fig. 3A), which indicates a higher tumor mutation burden in FA-PSV
+ MDS/AML patients. But genes related to epigenetic regulation and spliceosome were less mutated in FA-PSV
+ patients with statistical significance (8/25 vs. 70/244,
P = 0.024; 1/25 vs. 28/25,
P = 0.013) in this subgroup, while no statistical difference in tumor mutation burden was found between FA-PSV
+ and FA-PSV
− patients in ALL subgroups (Fig. 3B). Within the MDS/AML subgroup, in contrast to a previous report, the rate of trisomy 8 was not significantly increased in FA-PSV
+ patients (2/25 vs. 25/244,
P = 0.47), but the complex and abnormal karyotypes were more prevalent in FA-PSV
+ patients (5/25 vs. 23/244,
P = 0.003; 18/25 vs. 154/244,
P = 0.029). Fusions and cytogenetic abnormalities were distributed evenly within ALL subgroups.
4 Discussion
Previous studies focused on epidemiological investigations involving methodologies to address whether heterozygotes have an increased risk for AA and/or malignancies by starting out from medical history-based pedigree surveys. They compared observed cancer rates within probands’ family members with population estimations, with or without the knowledge of FA mutation states [
6,
7,
12,
13]. Except for two studies that found an increased risk of breast and prostate cancers, most of the studies failed to identify any robust positive results [
6,
7]. Comparing variant frequencies between case groups and the general population is a more direct method, but recent studies revealed limited correlations in specific genes, such as
FANCA and
FANCG [
8,
9]. Conversely, the overall PSV enrichment in our cohort suggests that heterozygous mutations of FA genes abate the capacity of the DNA homologous recombination repair pathway and contribute to hematopoietic failure. In line with a previous study, we identified the increased frequency of
FANCA variants in AA and myeloid malignancy groups [
8]. Furthermore, we present the first evidence that
BRCA2 heterozygotes have a significantly higher risk of developing into ALL. As a prototypical tumor suppression gene,
BRCA2 has been widely studied in hematologic malignancies and multiple solid tumors. However, the relationship between
BRCA2 defect and lymphoid malignancies is unknown. Although rare, some patients with FA and biallelic mutation of
BRCA2 have developed ALL [
14]. Although biallelic mutations of
BRCA2 only constitute approximately 3% of patients with canonical FA,
BRCA2 is the top mutated FA gene in the general population, suggesting the importance of discussing the ALL incidence of
BRCA2 heterozygotes.
A total of 8.38% patients carry rare PSVs of the five FA genes, but the positive rate for indicative medical history and/or family history was 17.51% in the whole cohort and higher in AA and MDS subgroups. In the definition of indicative medical and family history, we excluded many symptoms and syndromes, which are commonly regarded as an indicator of inherited situations, such as pulmonary fibrosis, psoriasis, and any other autoimmune symptoms; thus, the aberrations which might be attributed to the FA pathway defect can be analyzed exclusively. Indicative history may reflect if the patient is genetically suspected. However, this attempt might be hindered by the overlaps of FA and other inherited BMFs and cancer predisposition syndromes, such as neurofibromatosis, Noonan syndrome, and familial platelet disorder. Furthermore, although most patients’ family history is traceable, the genetic information of family members are difficult to obtain. The relatively high ratios of rare PSVs of the five FA genes and indicative medical history and/or family history might reflect both the extensively existing genetic predisposition in patients and the subtle effect of a single mutation in these genes in leukemogenesis. Thus, more comprehensive congenital and external factors should be considered. The study could be more thorough and robust if the mutation status of family members can be established, which implies the importance of comprehensive pedigree investigation and family consulting in both research and clinical facets.
Instead of all or nothing, the impact of different variants imposed on protein might be seen as a continuous variation. Therefore, contributions of the FA pathway defect could be latent and subtle but profound as time goes by. Experiments
in vitro demonstrated the intact protection in FA heterozygote cells that are sensitive to mitomycin C, diepoxybutane, and ionizing radiation [
5,
15]. Without integrated protection from the damage of endogenous or exogenous genotoxin, genomic instability will increase the chance of hematopoietic failure. As such, acquired driver mutations will easily accumulate. We observed the higher incidence of cytogenetic abnormalities and somatic mutations with statistical significance and lower incidence of epigenetic regulation and spliceosome gene mutations in FA-PSV
+ myeloid malignancies. This finding serves as evidence that FA-PSV carriers are prone to accumulating chromosomal structural abnormalities, although without overt FA phenotype, and confer congenital susceptibility to myeloid malignancies.
The current study has several limitations. First, only five FA genes, which are responsible for more than 90% of FA cases, were discussed [
10]. Although 17 genes are barely mutated in typical patients with FA and the general population, they should still be taken into account. Second, the biological effect of variants was evaluated based on minor allelic frequency and
in silico algorithms; thus, variants with
bona fide effects on protein functions might be excluded and vice versa. Besides, all synonymous mutations were excluded without estimating whether they can affect splicing or translation efficiency. The copy number variations (CNV) were not analyzed, but CNV is a non-negligible part of FA variants, especially in
FANCA and
BRCA2 [
16]. Nevertheless, the present results indicated that heterozygous FA mutations participate in AA and leukemogenesis within the carriers without overt FA clinical onset. This finding has clinical significance not only in planning therapeutic strategies and donor selection but also in family genetic consulting and in designing monitoring and prevention protocols for heterozygotes. Along with the growing list of analyzed genes in clinical settings after whole-exome sequencing or whole genome sequencing (WES/WGS) was applied, further strengthened ties between heterogeneous bone marrow failure/tumor disposition gene variants and hematopoietic diseases were noted in our clinical routine. Analysis based on WES/WGS is ongoing to address the limitations of this study.
PDF
(1349KB)
