Five new pedigrees with inherited RUNX1 mutations causing familial platelet disorder with propensity to myeloid malignancy

Familial platelet disorder with propensity to myeloid malignancy (FPD/AML) is an autosomal dominant disorder characterized by mild to moderate thrombocytopenia, abnormalities of platelet function, and a propensity to develop myelodysplasia (MDS) and/or acute myeloid leukemia (AML). Since the discovery of germline mutations in RUNX1 (previously AML1 or CBFA2) in FPD/AML in 1999,¹  14 pedigrees have been described with confirmed germline RUNX1 mutations.^2-7  Considerable phenotypic and genotypic variability exists. Most mutations are unique to the individual pedigree with only a single incidence of the same mutation in 2 distinct pedigrees.^1,5  The median incidence of myeloid malignancy in FPD/AML is 35% such that these patients with germline RUNX1 mutations are at significant risk of leukemia development

RUNX1 is a key regulator of definitive hematopoiesis and of myeloid differentiation.⁸  The protein belongs to a small family of transcription factors, which share homology for a region designated “runt homology domain” (RHD). The RHD directs binding of RUNX1 to DNA sequences of target genes and also mediates interaction between RUNX1 and the β-subunit, core-binding factor β (CBF-β). The gene is commonly dysregulated in sporadic cases of MDS and AML, mostly by translocations in AML⁹  and by point mutations in MDS.¹⁰  An exception occurs in the most undifferentiated subtype of AML, FAB class M0, in which point mutations occur in approximately 20% of cases.^11-13  A similar frequency (∼ 25%) is observed in certain subtypes of de novo MDS/AML, including refractory anemia with excess blasts (RAEB), AML with multilineage dysplasia, and AML after MDS,¹⁰  with an even higher frequency in therapy-related MDS/AML and radiation-associated MDS/AML (tested in survivors of the atomic bomb in Hiroshima).¹⁴ 

Familial RUNX1 mutations generally cluster to the N-terminal region of the gene within the RHD in exons 3 to 5, often disrupting DNA binding but allowing ongoing dimerization with CBF-β. Familial mutations have been reported less frequently in the C-terminal region, and these typically maintain DNA binding and dimerization with loss of the trans-activation region. The nature of RUNX1 mutations in de novo MDS/AML appears similar to that seen in FPD/AML with a predominance of point mutations in the RHD.^1,2,10,14 

Regardless of the cause of MDS/AML, effective treatment options are limited, and the only curative therapy for MDS is hematopoietic stem cell transplantation (HSCT) with its relatively high treatment-related mortality and morbidity. Inherited RUNX1 mutations can have implications for the selection of donors for HSCT, as one would prefer to exclude a sibling donor with the same leukemogenic mutation. Herein, we describe 5 previously unreported pedigrees with FPD/AML in which some affected patients demonstrate clinical features not previously reported in the disorder and in which several patients underwent HSCT from affected siblings before detection of the inherited RUNX1 mutation, often with poor outcomes. These cases exemplify the risks of sibling donor HSCT in FPD/AML and highlight the need for vigilance for the familial presentation of RUNX1 mutation.

Persons from families with more than 1 first-degree relative with MDS or AML were selected for investigation. Informed consent was obtained from all patients in accordance with the Declaration of Helsinki, and independent ethics approval was obtained from the British Columbia Cancer Agency Research Ethics Board and the Riverside Ethics Committee for Barts and the London NHS Trust (United Kingdom).

Peripheral blood, bone marrow aspirate, and/or saliva samples were obtained from study subjects. Genomic DNA was extracted from peripheral blood or bone marrow aspirate mononuclear cells using phenol and chloroform, following standard procedures. Saliva was collected in Oragene vials (DNA Genotek, Ottawa, ON), and DNA was purified and extracted following the manufacturer's instructions.

Exons 1, 3-6, 7b, and 8 of RUNX1 were separately amplified by polymerase chain reaction (PCR) using intronic primers as previously described.¹  PCR products were purified using gel columns (Montage; Millipore, Bedford, United Kingdom), sequenced using the respective forward and reverse primer, and analyzed on an ABI 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA). Mutations were confirmed either by repeat sequencing or by cloning sequencing using the plasmid vector pCR 2.1 TOPO (Invitrogen, Carlsbad, CA).

Persons from 10 families were investigated; 5 families had a history of at least one person with platelet abnormalities, and all of these demonstrated novel RUNX1 mutations. The remaining 5 pedigrees were heterogeneous, have no detectable mutations in RUNX1 or CEBPA, and continue to be investigated. The 5 FPD/AML pedigrees labeled A, B, C, D, and E, and their corresponding mutations (3 N-terminal, 2 C-terminal) are presented in Figure 1. Clinical characteristics are presented in Table 1. Figure S1 (available on the Blood website; see the Supplemental Materials link at the top of the online article) demonstrates a schematic representation of the positions and consequences of the 5 mutations in relation to the normal RUNX1 protein.

The proband (I-2) was followed for many years with moderate thrombocytopenia and a presumed diagnosis of chronic immune thrombocytopenic purpura (ITP). Her daughter (II-2) developed MDS at age 37 without previous detection of thrombocytopenia. II-2 underwent 2 HSCTs from her HLA-matched sibling (II-1) who had normal blood counts at the time. She relapsed 8 months after the first HSCT but remains well and in complete remission (CR) 6 years after the second HSCT. I-2 subsequently developed MDS at age 63 and died of AML at age 64. The history of MDS in mother and daughter prompted investigations of the other siblings. II-3 was noted to have moderate thrombocytopenia and a history of easy bruising only in association with trauma. His bone marrow aspirate and trephine biopsy demonstrated mild dysplastic features, which were consistent with, but not diagnostic of, MDS. II-1 was also noted to have mild, asymptomatic thrombocytopenia, which was not present when he was selected as the HSCT donor for his sister. The final sibling II-4 had a history of childhood ALL at age 12, treated with intensive chemotherapy. He has been in a CR for nearly 30 years with normal blood counts.

Genomic DNA was extracted from saliva, peripheral blood, and bone marrow from II-1, II-3, and from saliva from II-4. DNA from the diagnostic MDS bone marrow was available for II-2. A 7-bp deletion in exon 8 (1007_1013del) was detected in all available samples from pedigree A causing a frameshift mutation with a new termination at amino acid 563 (G336fsX563).

The proband III-1 presented for investigation of familial thrombocytopenia. She had been treated with splenectomy for a presumptive diagnosis of chronic ITP in her 20s and had maintained a platelet count between 50 and 100 × 10⁹/L with no further treatment. Platelet function studies had revealed an abnormal aggregation response to collagen and epinephrine. She reported a family history of many relatives with thrombocytopenia and mild bleeding tendencies and a history of “leukemia” in her father (II-2) and grandfather (I-1), who were both deceased and for whom further clinical information could not be obtained. Her brother (III-4) had a lifelong history of epistaxis and was also noted to have thrombocytopenia at age 20. He developed AML with normal karyotype at age 28 and was treated with an HSCT from his sister (III-2), who had a normal platelet count. The donor cells failed to engraft after the first HSCT, and he was retransplanted 18 months later from the same donor but subsequently died of graft-versus-host disease. III-1's paternal aunt (II-5) had a long history of excessive bleeding with a release-type platelet aggregation defect in response to adenosine diphosphate, collagen, epinephrine, and arachidonic acid, but with a normal platelet count. She had similarly undergone a sibling HSCT (for AML with monosomy 5) from a brother (II-6). II-6 had mild thrombocytopenia and mild dyserythropoiesis with a normal karyotype on bone marrow analysis, before being selected as the transplant donor. II-5 developed posttransplantation lymphoproliferative disorder and died 3 months after the HSCT. Several of II-5's other siblings (II-1, II-3, II-7) are also followed for moderate thrombocytopenia as are III-1's children (IV-1, IV-2) whose platelet function screening revealed abnormal closure times on PFA-100 testing.

Genomic DNA was extracted from saliva from III-1. DNA from the diagnostic bone marrow aspirates for II-5 and III-4 were also available. A single base pair insertion was detected in all available samples from pedigree B in exon 3 (83insG) causing a frameshift mutation and an early truncated protein at amino acid 109 (A28fsX109).

The proband (II-2) and her sister (II-1) were diagnosed with moderate thrombocytopenia in childhood after their father (I-1) died of AML at age 29. II-2 was later diagnosed with MDS (refractory anemia with multilineage dysplasia, with trisomy 8) at age 45, underwent an unrelated donor HSCT and is currently well. II-2's children, III-2 and III-3, have asymptomatic, mild thrombocytopenia. II-1 has a lifetime history of thrombocytopenia and easy bruising, previously diagnosed as chronic ITP. A bone marrow aspirate and biopsy of II-1 at age 49 revealed MDS (refractory anemia with ringed sideroblasts) with normal karyotype, which has not yet required treatment.

Genomic DNA was extracted from saliva from I-2, II-1, II-2, III-2, and III-3 and from peripheral blood from II-1. A missense mutation was detected in exon 4 (286G→C) in all samples collected from affected members (II-1, II-2, III-2, III-3) from pedigree C (D96H). The unaffected family member (I-2) lacked the mutation.

The proband (II-1) presented at age 21 with mild thrombocytopenia and easy bleeding. Her mother (I-2) had previously been followed for thrombocytopenia, diagnosed at a similar age. Both women were also noted to have platelet function abnormalities with absent aggregation to epinephrine and reduced aggregation to collagen and adenosine diphosphate. I-2 developed MDS (chronic myelomonocytic leukemia with 11q23 deletion) in her late 40s and died 8 years later. I-2's brother (I-3) died of AML at age 64 after a history of easy bruising but a normal platelet count and normal platelet function tests. His daughter (II-3) was noted to have thrombocytopenia at age 12, developed MDS with normal karyotype at age 23, and quickly progressed to AML. She died from complications of pancytopenia during treatment.

Genomic DNA was available from the peripheral blood of II-1. A single nucleotide substitution of G→A was detected at the splice donor site between exon 3 and intron 3, in II-1 predicted to cause altered splicing. Two previously described pedigrees have been reported with splice site mutations at the exon3/intron3 boundary, but these were splice acceptor mutations, resulting in a K90fs predicted protein product.

The proband (I-2) had a history of chronic thrombocytopenia since childhood. She developed MDS with normal karyotype at age 52 and received an HSCT from her brother (I-3) with normal engraftment and no early transplant complications. Her son (II-2) was treated for biphenotypic AML at age 8 and received an HSCT from his sister (II-1) at that time. He had no history of thrombocytopenia or bleeding tendency before his diagnosis of AML. He remains well 26 years after the HSCT. Neither II-1 nor II-2 currently has thrombocytopenia.

Genomic DNA was extracted from peripheral blood and saliva from I-2 and II-2 and from peripheral blood of the donor sibling, II-1. A nonsense mutation was detected in exon 7b (877C→T) in blood and saliva of I-2, resulting in an early termination at amino acid 292 (R292X). II-1 demonstrated a wild-type RUNX1 configuration. Of note, although her son, II-2, had childhood AML, R292X was not detected in his saliva or peripheral blood DNA. Because it is doubtful that the occurrence of AML in this patient represents a sporadic event, it seems probable that the tissues tested for mutation are of donor (II-1) origin.¹⁵ 

Investigation of families with MDS/AML is often limited by the fact that modern pedigrees are small, living-affected family members are few, and DNA is rarely available from deceased patients. Despite these difficulties, previous investigations detected germline mutations in RUNX1 and CEBPA, genes also frequently mutated in sporadic MDS/AML. This study reveals 5 new cases of inherited RUNX1 mutations and provides new clinical insight into the resultant syndrome of FPD/AML. Although familial presentation of MDS and/or AML is rare, the detection of these new FPD/AML families, comprising 50% of the pedigrees investigated, suggests that inherited RUNX1 mutations may occur more frequently than previously appreciated. The increasing availability of molecular testing, along with a heightened awareness of this syndrome, may also lead to the identification of additional families in the near future.

Fourteen families have been previously reported with inherited RUNX1 mutations causing FPD/AML.^1-7  In these cases, there was a strong predominance of N-terminal mutations with just 3 families having germline C-terminal mutations. In our series, 3 N-terminal and 2 C-terminal germline RUNX1 mutations were detected. These new cases suggest that effective screening of the RUNX1 gene should not be restricted to the RHD and that C-terminal mutations may be more frequent than previously recognized.

Although the molecular basis of FPD/AML as caused by germline RUNX1 mutations is constant, the clinical phenotype of affected patients is heterogeneous. The most consistent clinical feature is that of a mild to moderate bleeding tendency because of platelet functional and/or quantitative abnormalities. This feature is absent in many affected patients in our 5 pedigrees. Several affected family members lacked a bleeding history and thus did not have platelet function studies performed. Platelet counts were also frequently normal. Patients within pedigree A were noted to develop mild thrombocytopenia only in their 40s with no preexisting history of bleeding tendency. Similarly, II-5 from pedigree B had a normal platelet count 5 years before her diagnosis and death from AML, but with a mild bleeding tendency and release-type platelet function abnormalities. Our results suggest that neither complete blood counts nor platelet function studies are a sensitive test for inherited RUNX1 mutations. Of note, in pedigree A, a carrier also developed sporadic childhood ALL, suggesting that the phenotype of FPD/AML may not be restricted to myeloid malignancies. This may not be unreasonable given the importance of RUNX1 in ALL (with ∼ 25% of pediatric ALLs involving the TEL-AML1 [ETV6-RUNX1] translocation).¹⁶ 

Overall, fewer than 50% of persons who inherit a RUNX1 mutation develop MDS/AML, and there is a large range in the prevalence of malignancy between families (20%-60%).¹⁷  Our series confirms the median incidence of MDS/AML of 35% among carriers of germline RUNX1 mutations. The age at presentation of MDS/AML (6-75 years) is also variable but not extended by our series.¹⁷  This large age range is an important feature in that it may obscure the diagnosis of FPD/AML if offspring develop MDS/AML before their parents/grandparents. The accumulation of clinical data from additional families will hopefully help to identify the factors that are important in the development of overt malignancy in persons with inherited RUNX1 mutations; however, no information is currently available to predict which patients will develop malignancy or when. The latency period required before onset of overt leukemia suggests that secondary mutations are necessary to trigger MDS/AML and that a heterozygous RUNX1 mutation is insufficient alone.

Investigations of sporadic cases of MDS/AML with RUNX1 mutations have revealed several secondary mechanisms involved in the development of overt malignancy. Half of AML M0 cases demonstrate biallelic RUNX1 mutations, whereas MDS cases frequently harbor additional karyotypic abnormalities.^12,18,23  However, similar recurrent secondary aberrations have not been observed in FPD/AML; in particular, acquired mutations in the wild-type RUNX1 allele are not observed. This is in contrast to familial AML involving germline CEBPA mutations in which somatic C-terminal mutations are acquired, in addition to the germline N-terminal mutations, promoting overt leukemia.^19,20  Secondary cytogenetic abnormalities occur frequently in FPD/AML, but these are not consistent within or between pedigrees. These studies highlight the importance of serial monitoring and collection of material from affected patients to allow for investigation of the secondary events that promote overt malignancy, which is currently lacking.

Previous studies suggest that the occurrence of distinct mutations between families contributes to the variable phenotypes of FPD/AML.¹⁷  Michaud et al demonstrated that some mutant proteins exert dominant-negative effects and that the pedigree with the highest incidence of leukemia had the highest predicted dominant-negative activity.^2,17  Evidence of dominant-negative activity was recently confirmed by Matheny et al, when they examined the effects of different RHD mutations on RUNX1 function.²¹  The family with the highest incidence of leukemia among our pedigrees was pedigree D (75%) in which the mutation occurred early in the RHD and would be predicted to act by a mechanism of haploinsufficiency.

Unfortunately, outcomes for MDS/AML are generally poor, and current treatment options are unsatisfactory, with HSCT being the only potentially curative treatment for MDS. Long-term data on the outcomes of patients treated for FPD/AML-associated leukemias are limited, making a determination of prognosis in these patients difficult.¹⁷  Until recently, only a single family with FPD/AML had been reported in which HSCT was undertaken from a sibling donor who also possessed a germline RUNX1 mutation.³  In that case, the recipient had slow and incomplete engraftment and, within 2 years, developed AML of donor origin. The donor-derived AML was treated with an unrelated donor HSCT, but the patient subsequently died of an Epstein-Barr virus–associated posttransplantation lymphoproliferative disorder. A later report by Kirito et al described a family in which members were noted to have thrombocytopenia once investigated as potential HSCT donors for their MDS-affected sibling.⁷  The presence of thrombocytopenia in several family members prompted screening for germline RUNX1 mutations, confirming an inherited predisposition and preventing the siblings from being selected as HSCT donors.⁷  Our series includes 3 additional families in which siblings were used as HSCT donors, before documentation of an inherited RUNX1 mutation. Patient II-5 from pedigree B received an HSCT from an affected sibling and subsequently died of posttransplantation lymphoproliferative disorder. It is notable that this is the second reported case of a fatal Epstein-Barr virus–related lymphoproliferative disorder in FPD/AML.³  Another person within this family (III-4) had a sibling HSCT with failure of engraftment. In pedigree A, the donor (II-1) was noted to have normal peripheral blood counts and bone marrow morphology before HSCT but developed thrombocytopenia 5 years later. The recipient of the HSCT (II-2) relapsed quickly after a first transplant but was retransplanted with the same sibling donor and is now well 6 years later. Pedigree E also involves sibling HSCT, in this case, with good results. These cases, although few, suggest that results of HSCT from siblings with RUNX1 mutations have a high risk of failure.

The small size of modern families along with the variability in age at presentation of disease in affected patients will result in inherited RUNX1 mutations being easily overlooked. A single case of MDS/AML within a small family may be a subtle presentation of FPD/AML. The current description of 5 new cases of familial RUNX1 mutations illustrates that this syndrome is probably more prevalent than previously appreciated. This highlights the possibility that some presumed cases of sporadic MDS/AML with RUNX1 mutations could represent unrecognized cases of FPD/AML. This occurrence was recently observed in the setting of inherited CEBPA mutations where Corbacioglu et al found germline CEPBA mutations in 2 of 18 AML patients with known CEBPA mutations.²²  Therefore, we recommend that young patients with MDS/AML, irrespective of family history, be investigated for RUNX1 mutations to select those whose families may also be at risk.

The authors thank the families who participated in this study as well as Dr Connie Eaves and Karen Lambie from the Terry Fox Laboratory for their help with processing and shipping samples from Canada.

This work was supported by grants from the Lady Tata Memorial Trust (London, United Kingdom), the Sangara Family Fund through the Department of Leukemia/Bone Marrow Transplantation, Division of Hematology (Vancouver, BC), and from Cancer Research United Kingdom.

Contribution: C.J.O. performed research and wrote the paper; C.L.T., A.K., D.L.F., C.A.S., S.C.J., M.-C.P., G.D.S., B.L., P.R.E.J., A.M., J.A.L.Y., M.J.B., and T.A.L. collected data and edited the paper; J.M.S designed research and obtained ethical approval; and J.F. designed research and cowrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Carolyn J. Owen, Centre for Medical Oncology, Institute of Cancer, Barts and the London School of Medicine, Charterhouse Square, London EC1M 6BQ, United Kingdom; e-mail: cowen@bccancer.bc.ca; or Jude Fitzgibbon, Centre for Medical Oncology, Institute of Cancer, Barts and the London School of Medicine, Charterhouse Square, London EC1M 6BQ, United Kingdom; e-mail: jude.fitzgibbon@cancer.org.uk.