Targeted mutation screening of 292 candidate genes in 38 children with inborn haematological cytopenias efficiently identifies novel disease‐causing mutations

Summary Establishing a precise diagnosis is essential in inborn haematological cytopenias to enable appropriate treatment decisions and avoid secondary organ damage. However, both diversity and phenotypic overlap of distinct disease entities may make the identification of underlying genetic aetiologies by classical Sanger sequencing challenging. Instead of exome sequencing, we established a systematic next generation sequencing‐based panel targeting 292 candidate genes and screened 38 consecutive patients for disease‐associated mutations. Efficient identification of the underlying genetic cause in 17 patients (44·7%), including 13 novel mutations, demonstrates that this approach is time‐ and cost‐efficient, enabling optimal management and genetic counselling.


Introduction
Haematological cytopenias, i.e., deficiencies of one or several blood cell components, are the shared hallmark of an extremely heterogeneous group of either environmentally or genetically triggered diseases. The identification of the underlying genetic aetiology of these diseases is particularly important in children, in whom inborn disorders predominate. These disorders comprise red blood cell membrano-and enzymopathies, bone marrow failure syndromes, familial haemophagocytic lymphohistiocytosis, mitochondriopathies as well as primary immunodeficiencies. Except for haemoglobinopathies and hereditary spherocytosis, all of these are orphan or even ultra-orphan diseases.
Our study was approved by the responsible institutional review board and performed with the written informed consent from patients, parents and/or legal guardians. We included 38 children and young adults (17 males and 21 females) with a median age of 7Á5 years (range 0Á1-21Á4 years; Fig 1A), who were referred consecutively to our hospital and who fulfilled established cytopenia criteria. The patients' DNA was analysed with a custom-designed targeted enrichment panel (HaloPlex TM ; Agilent Technologies, Santa Clara, CA, USA) that comprised the respective 292 candidate genes (Table S1) similar to a recently established targeted panel for genetic investigation of primary immunodeficiencies (Erman et al, 2017). Regions of interest included all exons, exon-intron boundaries and promoter regions.
Sequencing was performed on a HiSeq3000 platform (Illumina, San Diego, CA, USA) as described (Salzer et al, 2016;Erman et al, 2017). 98Á5% of enriched exonic bases were considered callable with a minimum read depth of 2. The same calculation for a minimum read depth of 10 and 40, revealed 96Á4% and 88Á2% coverage, respectively (not shown). Sequencing data were analysed with a previously established next generation sequencing (NGS)-data analysis   pipeline that is based on current Genome Analysis Toolkit (GATK) best practice recommendations (Salzer et al, 2016;Erman et al, 2017). The potential relevance and recurrence of variants was assessed with prediction tools (e.g. CADD), ExAC and our internal dataset of more than 400 sequenced individuals, as described (Salzer et al, 2016). Quality metrics were generated using CollectHsMetrics (http://broadinstitute. github.io/picard/command-line-overview.html, last accessed 13 April 2017) for analysis of target-capture sequencing data. Additional details and genetic results are outlined in the Supplementary Results. All identified variants which were deemed potentially disease-causing were validated with capillary sequencing as described (Salzer et al, 2016;Erman et al, 2017) or multiplex ligation-dependent probe amplification. Collectively, we identified a causative genetic defect in 17/ 38 patients (44Á7%) including novel mutations in 13 of them ( Fig 1A). In 8/17 (47%) patients the results of molecular testing were considered as 'confirmatory' and 'diagnostic' in 9/17 (53%), respectively. (Table I). Eleven of the patients presented with anaemia (clinical and haematological details are provided in Table S2) five of them with a normocytic (spherocytic), chronic, haemolytic anaemia (CHA; Patients 1-5; Table I) and six with a macrocytic anaemia (Patients 6-11, Table I). Within the subgroup of five patients with hereditary spherocytosis (HS), we identified heterozygous mutations in the Spectrin Beta, Erythrocytic gene (SPTB) in two, and mutations in the Ankyrin 1 (ANK1) membrane protein encoding gene in the remaining three patients (Table I). Contrary to the originally suspected congenital dyserythropoietic anaemia (CDA), Patient 6 had a pyruvate kinase deficient anaemia due to a homozygous PKLR mutation (Table I) (Unal & Gumruk, 2015). Two patients with mild macrocytic anaemia had mutations in the gene encoding the piezo-type mechanosensitive ion channel component 1 (PIEZO1): Patient 7 bore a previously described pathogenic p.R2456H variant (Zarychanski et al, 2012), and Patient 8 a novel p.Q1361R mutation located adjacent to the already known p.R1358P mutation (Table I) (Albuisson et al, 2013). This finding, together with the subsequent perception of dessicytes in the respective blood smears (Fig 1B,C) was particularly relevant, given that splenectomy had been considered prior to identifying the disease as hereditary xerocytosis (HX); this form of treatment is contraindicated in HX because of an associated inherent increased risk of thromboembolic complications (Andolfo et al, 2016). In Patients 9 and 10, novel compound heterozygous mutations in codanin 1 (CDAN1) confirmed the respective diagnosis of CDA type 1 (CDA1), which originally had solely been based on the morphological abnormalities of erythropoietic components in the peripheral blood and bone marrow (Fig 1D-F; Table I). Furthermore, we also identified a novel mutation in the ribosomal protein S29 gene (RPS29, p.A47T) in patient 11 with Diamond-Blackfan anaemia (DBA) ( Table I).
Disease-relevant mutations were discovered in two of four patients with thrombocytopenia. The severe form of macro-thrombocytopenia in an infant (Patient 13; Table II; Fig 1G,  H; Table S3), was the consequence of a heterozygous mutation in the gene encoding the myosin heavy chain 9 (MYH9, p.S96L), while a novel heterozygous p.G165R mutation in the RUNX1 gene, which encodes the haematopoietic runt related transcription factor 1, was responsible for a mild form of chronic normocytic thrombocytopenia in an eightyear-old girl (Patient 12; Table II; Table S3) (Sood et al, 2017).
Of the four leucopenic patients, the only mutation detected was in a 17-year-old female with moderate neutropenia and B cell deficiency, who suffered from recurrent fever bouts and oral aphthous lesions (Patient 14: Table II;  Table S4). Her phenotype can probably be explained by a heterozygous variant (p.P41A) in the gene encoding the haematopoietic transcription factor GATA binding protein 2 (GATA2), which had been documented previously in a patient with myelodysplastic syndrome (MDS) (Holme et al, 2012), despite a frequency of heterozygotes of 0Á21% (134 of 123 458 total alleles) in gnomAD (http://gnomad.broadinsti tute.org/, accessed March 2018) amongst non-Finnish Europeans (Table II).
Similarly, we only discovered a clear genetic disease cause in one of seven patients with bicytopenia. He suffered from steroid-responsive DBA with anaemia and leucopenia and carried a novel heterozygous splice-site donor frame shift mutation in the ribosomal protein RPL5 gene (Patient 15; Table II; Table S5).
We also discovered disease-predisposing mutations in two of seven patients with pancytopenia (clinical details of the two patients are provided in Table S6). We identified a unique deletion encompassing exons 6-31 of the FANCA gene in an eight-year-old girl with Fanconi anaemia (Patient 16; Table II; Figure S1) and a novel mutation in the nonhomologous end joining factor 1 gene (NHEJ1, also termed Cernunnos) (Patient 17; Table II). This finding not only disclosed that he suffered from a special form of immunodeficiency with a severe B cell deficiency, but also helped to explain his physical problems, growth retardation, microcephaly and clinodactyly, which are also salient features of this syndrome (Buck et al, 2006). He developed a mild pancytopenic form of MDS with monosomy 7, which was eventually replaced by a 20q deletion.
As an alternative approach to whole exome or genome sequencing (Yang et al, 2013), targeted sequencing panels have hitherto been used for more narrowly defined groups of haematological disorders, such as inherited anaemias (Agarwal et al, 2016;Roy et al, 2016), inherited bone marrow failure syndromes (Ghemlas et al, 2015), DBA (Gerrard et al, 2013) and thrombocytopathies (Lentaigne et al, 2016). Given both the phenotypic overlap and diverse genotypic spectrum of many of these conditions, we deliberately chose to use an extended gene panel that covered a broad spectrum of inborn haematological diseases. This approach enabled us to include patients with diverse haematological conditions and thereby to also discover uncommon and unexpected diseases, such as HX (Patients 7 and 8), CDA1 (Patients 9 and 10), RUNX1-related thrombocytopenia (Patient 12), GATA2related neutropenia (Patient 14), or NHEJ-related combined immunodeficiency with transformation to MDS (Patient 17). Our overall diagnostic success rate of 44Á7% is therefore higher than similar targeted sequencing panel studies performed previously. Given that 13 of our 17 genetically identified cases carried novel mutations and that these also included several cases with ultra-rare haematological diseases, of which only few cases are known so far, suggests that many such cases may remain unrecognized and undiagnosed at tertiary referral centres. Considering the inherent difficulties often experienced with the conventional diagnostic evaluation of childhood cytopenias, their prevalent genetic origin, the plethora of potentially involved genes and the heterogeneous types and distribution of mutations, together with the reassuring results presented herein strongly argue for the implementation of such targeted sequencing screening programmes in the initial diagnostic work-up of such diseases.

Acknowledgements
We gratefully acknowledge the support of all study participants as well as nurses and treating physicians at St Anna Children's Hospital Outpatient Clinics. This study was supported through intramural funding of the Ludwig Boltzmann Institute for Rare and Undiagnosed Diseases (LBI-RUD) Vienna, by the Austrian Science Fund (FWF): Project I2250-B28, and by a charitable donation of the Kapsch group (http://www.kapsch.net/kapsc hgroup) to OAH.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Table S1. Genes included in the Hematology panel design divided by disease group. Table S2. Pertinent data of the 11 patients with anemia. Table S3. Pertinent data of the two patients with thrombocytopenia. Table S4. Pertinent data of the patient with leukopenia. Table S5. Pertinent data of the patient with bicytopenia. Table S6. Pertinent data of the two patients with pancytopenia. Figure S1. MLPA results with the FANCA kits P031-B2 (left) and P032-B2 (right). Results from normal controls (peak ratio 1 corresponding to 2 copies) and the twin girls harboring a biallelic deletion of exons 6-31 of the FANCA gene (peak ratio 0) are shown.