Acquired isodisomy of chromosome 21 in an acute myeloid leukaemia (AML) patient as an incidental finding during routine chimaerism analysis, and the introduction of a new RUNX1 screening service. I’m going to tell you about a chance finding during routine chimaerism analysis on one of our acute myeloid leukaemia patients that has led to the introduction of a new diagnostic service. Joanne Mason, Registered Clinical Scientist West Midlands Regional Genetics Laboratory, Birmingham Women’s NHS Foundation Trust, 1

Joanne Mason, WMRGL Birmingham Introduction AML is a genetic disease Characterised by enhanced proliferation & differentiation block ~50% cases have cytogenetically visible aberrations The remaining cases have genetic aberrations which are only detectable at the molecular level These genetic lesions help to characterise the subtype of leukaemia, and can be used to guide therapeutic decisions and inform prognosis Molecularly-targeted therapy (e.g. Glivec in CML) AML is a complex heterogeneous disease, caused by the multi-step accumulation of multiple recurring genetic mutations in haematopoetic stem cells leading to enhanced proliferation and a block in differentiation. In approximately half of all cases, lesions are visible cytogenetically, for example as chromosomal translocations. W are only just beginning to discover the underlying molecular genetic aberrations which account for the remaining cases. These genetic lesions help to characterise the subtype of leukaemia, and can be used to guide therapeutic decisions and inform prognosis. Increasingly they are also being used as molecular targets for cancer therapy. Joanne Mason, WMRGL Birmingham

Joanne Mason, WMRGL Birmingham Patient A Diagnosed with AML in May 2006 Karyotype analysis: trisomy 13 (47,XY,+13 [10]) Treated with chemotherapy on the MRC AML15 trial protocol Relapsed November 2007 (47,XY,+13) Salvage chemotherapy, followed by stem cell transplant (SCT) in March 2008 Our patient presented with AML in 2006, when routine cytogenetic analysis showed gain of chromosome 13 in leukaemic blasts, which is a rare but recurrent abnormality. The mechanism by which trisomy 13 is oncogenic is not known. Chemotherapy induced remission which was sustained for 18 months. Sadly the patient relapsed at the end of 2007. Salvage chemotherapy elicited a second remission, prior to the patient receiving a stem cell transplant. Joanne Mason, WMRGL Birmingham

Chimaerism monitoring post-SCT Sex-matched SCT patients are monitored for levels of donor and host DNA post-transplant using polymorphic microsatellite markers. A pre-requisite for chimaerism analysis is to find at least one informative marker that distinguishes donor from host. CAGA 3-15 CAGA Patients who have undergone stem cell transplants are monitored for levels of donor and host chimaerism using polymorphic microsatellites. First it is necessary to find informative markers with different sized alleles in host and donor. This is done by performing a multiplex PCR reaction. Joanne Mason, WMRGL Birmingham

Multiplex microsatellite marker PCR and subsequent fragment analysis This shows the multiplex kit, which we recently overhauled, adding new highly informative markers and using pentanucleotides wherever possible which are less prone to stutter. If anyone is particularly interested in chimaerism analysis then Emma Sach has a poster here at CMGS describing this re-vamped kit and the chimaerism procedure. Joanne Mason, WMRGL Birmingham

Joanne Mason, WMRGL Birmingham Chimaerism analysis DONOR HOST PRE-TRANSPLANT For anyone who’s not familiar with chimaerism analysis, the procedure is pretty straightforward. The PCR is kept in the exponential phase and is therefore semi-quantitative, so that peak heights and areas can be used to estimate the contribution from host and donor in post-transplant samples. In this example we estimated 74% donor and 26% host chimaerism in the post-transplant sample. POST-TRANSPLANT 74% donor 26% host Joanne Mason, WMRGL Birmingham

Microsatellite results Pre-transplant DNA 13 Returning to our patient, we received the first sample for chimaerism analysis about one month after the transplant, and set up the multiplex PCR reaction in order to find informative markers. This is the host-pre-transplant DNA, and I happened to notice something unusual about the pattern of microsatellite markers. Whereas we would normally expect to see a 1:1 ratio CLICK between alleles as a result of amplification of one allele from each of the two homologous chromosomes, some markers appeared to show a skewed ratio. CLICK My first reaction was to check the provenance of the pre-transplant sample, and it transpired that the stored pre-transplant DNA was actually from a blood taken at relapse. If you remember, the patient relapsed with trisomy 13. So all the markers located on chromosome 13 would be expected to show a 2:1 ratio as a result of doubling up of one of the chromosomes 13. However not all the markers showing a skewed ratio were located on chromosome 13. Circle markers The remaining skewed markers suggesting a copy number change were located on chromosome 21. Joanne Mason, WMRGL Birmingham

Chromosome 21 markers Average ratio 4:1 Remission DNA D21S11 Penta D D21S1411 D21S1270 Relapse DNA So I looked at more markers, spread across chromosome 21, and they all showed a skewed ratio, with a mean value of 4:1. I next confirmed that the patient was constitutionally heterozygous by testing remission DNA. So it looks as if the patient has lost a chunk of chromosome 21 at relapse. A deletion spanning all these loci would easily be seen down the microscope..... Joanne Mason, WMRGL Birmingham

Ch 21 markers : copy number change? Cytogenetics 2 normal copies Ch 21 Possible explanations for the discrepancy: 1) Sub-microscopic deletion within chromosome 21 (unlikely as multiple deletions would be required) 2) A cryptic sub-clone with gain or loss of 21 in some cells, not detected by initial cytogenetic analysis (impossible with a microsatellite ratio of 4:1) ...... and yet two apparently normal copies of chromosome 21 were present in every cell. So unlike the markers on chromosome 13, cytogenetic analysis did not offer an obvious explanation for the apparent copy number change. Now there are various potential explanations for the discrepancy: The first two are unfeasible, and the most likely explanation is a recently described genetic phenomenon known variously as acquired isodisomy, acquired uniparental disomy, or copy number neutral loss of heterozygosity. 3) Acquired isodisomy (aka acquired uniparental disomy, or copy number neutral loss of heterozygosity) Joanne Mason, WMRGL Birmingham

Acquired isodisomy (AID) Common mechanism of oncogenesis Prognostic significance in AML? Acquired isodisomy in AML was first reported by our lab and others in 2005. It is a common mechanism of oncogenesis, but is very difficult to detect, hence only being described recently. Importantly it may have prognostic significance in AML, and potentially in other neoplasms. Joanne Mason, WMRGL Birmingham

Mitotic Recombination AID 21 21 21 Acquired isodisomy is thought to arise by somatic recombination between homologous chromosomes at mitosis. Chromosomes identical for most of their length then segregate into one daughter cell. In this way any mutation, indicated by a red star, already present on one chromosome will be duplicated and if that mutation is responsible for driving the leukemia the cell which now has a mutation on both chromosomes will have a significant, survival advantage and undergo clonal expansion. This hypothesis underpins the suggestion that leukemias with acquired isodisomy may have a worse prognosis. Joanne Mason, WMRGL Birmingham

Acquired isodisomy (AID) AID is a mechanism by which homozygosity for a mutation can be achieved without detrimental loss or gain of contiguous chromosome material It is cytogenetically invisible (both chromosomes look the same) and therefore very difficult to detect unless you specifically look for it. DNA microarrays – sub-microscopic & cryptic changes AID is a mechanism by which homozygosity for a mutation can be achieved without detrimental loss or gain of contiguous chromosome material It is cytogenetically invisible (because both chromosomes usually look the same) and it is therefore very difficult to detect unless you specifically look for it. Increasingly DNA microarrays are being used to detect sub-microscopic and cryptic genetic changes in human cancers. As well as being frequent in AML, acquired isodisomy has been reported in other myeloproliferative and myelodysplastic disorders, as well as in lymphoma, adenocarcinoma and cervical cancer to name but a few. Joanne Mason, WMRGL Birmingham

AID21: What genes might be affected? RUNX1 21q22.3 Transcription factor Most frequent target for chromosomal translocation in leukaemia Point mutations in sporadic AML In familial platelet disorder/AML (FPD/AML) So next I wondered what the effect of acquired isodisomy for chromosome 21 might be in our patient. and one strong candidate for mutation was the RUNX1 gene located on the long arm of 21. RUNX1 is involved in many recurrent translocations, with at least 38 other partner genes and it’s therefore an important gene in leukaemia. Its role expanded further in 1999 when two groups reported point mutations, including somatic mutations sporadic AML, and germline mutations in the disease which has become known as familial platelet disorder with predisposition to AML, or FPD, which I’ll return to in a minute. Joanne Mason, WMRGL Birmingham

RUNX1 point mutations in sporadic AML 1.2% of all AML Highly associated with AML FAB M0 trisomy 21 trisomy 13 (80-100%) [Patient AS 47,XY,+13] RUNX1 mutation associated with a poor prognosis in MDS (prognosis in AML not yet known) Discovery of mutations has implications for Risk adapted therapy Molecularly targeted therapy In sporadic AML, RUNX1 point mutations are not that common, but have associations with specific sub-types. They are highly associated with acquired trisomy 21, and it also turns out that trisomy 13 karyotypes have detectable RUNX1 mutations in between 80 & 100% of cases, which makes it even more likely that our patient, with trisomy 13, has a RUNX1 mutation. Presence of a RUNX1 mutation has been associated with a poor prognosis in MDS. Risk-adapted therapy based on cytogenetic findings is standard practice in AML, and increasingly, the data available from molecular genetic studies allows for further stratification based on molecular risk factors, which may in the future include RUNX1. Joanne Mason, WMRGL Birmingham

Joanne Mason, WMRGL Birmingham Familial Platelet Disorder with Predisposition to Acute Myeloid Leukaemia (FPD/AML) Rare autosomal dominant disorder Characterised by inherited thrombocytopenia, platelet function defect and a lifelong risk of myelodysplastic syndrome (MDS) and AML Caused by heterozygous germline mutations in RUNX1 Worldwide, only fifteen pedigrees have been reported to date. In November 2008, request for ?FPD/AML in a West Midlands AML patient. Returning to familial platelet disorder, this is a rare autosomal dominant disorder characterized by inherited thrombocytopenia and a lifelong risk of myelodysplastic syndrome and AML. It is caused by heterozygous germline mutations in RUNX1. Only fifteen pedigrees have been reported to date, although one of those families happens to live in Birmingham, and was originally investigated for RUNX1 mutations by a research lab. Last November, I received a request for RUNX1 mutation analysis from a local haematologist who suspected that she had another FPD family. I explained that testing was not available locally and we would need to send DNA away. This got me thinking about the acquired isodisomy 21 patient from a few months earlier, who I suspected of having a somatic RUNX1 mutation. Joanne Mason, WMRGL Birmingham

Joanne Mason, WMRGL Birmingham RUNX1 Point Mutations RUNX1 mutation screening service AID21 patient AML cases with a strong association with RUNX1 mutations (FAB M0, +13) FPD/AML patient Sequencing of the entire coding region I therefore decided to develop mutation analysis for RUNX1 in the diagnostic lab. The benefits of this would be threefold. Firstly I hoped to find the putative RUNX1 mutation in our acquired isodisomy 21 patient, and therefore tie up the loose ends of this story, but a happy by-product would be development of a mutation analysis service for acquired RUNX1 mutations in other AML patients. Finally we would be able to satisfy requests for germline RUNX1 mutation analysis in patients with suspected FPD. Joanne Mason, WMRGL Birmingham

RUNX1 mutation screening service b d cDNA template PCR under same conditions (‘touchdown PCR’) M13 tag to facilitate high-throughput sequencing The strategy I chose was to sequence the entire coding region using cDNA in four overlapping fragments. Primer sequences courtesy of Dicker et al, Leukemia 2007 Joanne Mason, WMRGL Birmingham

RUNX1 sequencing results.....so far Patient A: p.Asp171Gly (D171G, homozygous) DNA binding domain Previously reported in two AML patients 26% of mutations in RUNX1 are homozygous (wild-type RUNX1 is lost) Wild-type Patient AS I’ll briefly summarise my results. The acquired isodisomy 21 patient did indeed have a mutation. This single nucleotide change, resulting in the conversion of an aspartic acid to a glycine residue, is predicted to disrupt DNA binding, and has been reported before in two other AML patients. As anticipated given that both chromosomes 21 are the same, the mutation is homozygous. Interestingly one of the other AML parients also showed loss of wild-type RUNX1 despite having two apparently normal copies of chromosome 21, so I strongly suspect acquired isodisomy 21 in this patient too. On reviewing the literature, at least 26% of mutations in RUNX1 are homozygous. This suggests that acquired isodisomy for chromosome 21 is common in patients with RUNX1 mutations, but underreported as it cytogenetically invisible. Joanne Mason, WMRGL Birmingham

Joanne Mason, WMRGL Birmingham SNP-based DNA microarrays to investigate cytogenetically cryptic areas of somatically acquired homozygosity (AID) Postulated that such regions contain homozygous mutations in genes known to be mutational targets in leukaemia. In 7 of 13 cases with acquired isodisomy, homozygous mutations were identified at four distinct loci (WT1, FLT3, CEBPA, and RUNX1) The mutation precedes mitotic recombination, which acts as a "second hit" responsible for removal of the remaining wild-type allele. One report of acquired isodisomy for RUNX1 came from Bryan Young’s group, who used DNA microarray technology to look for cytogenetically cryptic areas of somatically acquired homozygosity. They postulated that such regions may contain homozygous mutations in genes known to be mutational targets in leukaemia. In 7 of 13 cases with acquired isodisomy, homozygous mutations were indeed identified in four such genes including RUNX1 In agreement with our model of oncogenesis which I described earlier, they suggest that the mutation precedes mitotic recombination, which acts as a "second hit" responsible for removal of the remaining wild-type allele. Joanne Mason, WMRGL Birmingham

RUNX1 sequencing results.....so far ?FPD/AML patient and three AML patients with trisomy 13 (i.e. highly likely to have RUNX1 mutations) Patient B AML 47,XX +13 p.Val137_Gly138insThr wt B Patient C AML 50,XY +8,+9,+13,+21 p.Met25Lys p.Arg135Lys All de novo, but two other mutations involving arginine 135 have been reported before wt C I am currently in the process of completing the sequence analysis of RUNX1, but so far I haven’t found a mutation in the patient with suspected familial platelet disorder. I identified three AML patients presenting in the past year with trisomy 13. In patient B I have found a novel insertion, and in another AML patient with both trisomy 13 & trisomy 21 there appear to be two different mutations. None of these mutations have been reported before, and all are homozygous, which is particularly interesting for the last patient who has three copies of chromosome 21, which extrapolating from these results all appear to be identical! Joanne Mason, WMRGL Birmingham

Joanne Mason, WMRGL Birmingham Further work Complete the sequence analysis of all four fragments comprising the coding region of RUNX1 Effect of mutations? Inheritance pattern in familial cases Confirm RUNX1 mutations are acquired and not constitutional by sequencing stored remission DNA My plan is to complete the sequence analysis of the entire coding region of RUNX1, and the pathogenicity of any mutations I find needs to be confirmed. If I find a mutation in the FPD patient, I can look at the inheritance pattern to see if it tracks with the disease. In my sporadic AML patients I can confirm that RUNX1 mutations are somatically acquired, rather than constitutional, by demonstrating their absence in DNA taken in remission. Joanne Mason, WMRGL Birmingham

Joanne Mason, WMRGL Birmingham Summary Unexpected microsatellite pattern in pre-transplant DNA taken at relapse Molecular data + cytogenetic data = acquired isodisomy 21 Candidate gene = RUNX1 RUNX1 mutation D171G Sequencing service for other sporadic AML patients, and for suspected FPD/AML referrals. So to summarise, routine chimaerism analysis showed an unexpected pattern in pre-transplant DNA from one of our AML patients. Investigation of this unusual pattern ,and integration with cytogenetic data, showed that the phenomenon of acquired isodisomy was the most likely explanation for the unusual results. This led to a search for candidate genes, which in turn prompted me to introduce a sequencing assay for the RUNX1 gene, and subsequent discovery of a RUNX1 mutation in our acquired isodisomy patient. Now that sequencing of the RUNX1 gene is available within the diagnostic lab, we can process referrals for suspected familial platelet disorder. So RUNX1 mutation analysis is now validated, and ready to use as and when it is incorporated into the risk algorithm for AML, the aim of which is to individualize patient therapy. The challenge for the future is to incorporate all the genetic information into novel risk-adapted therapeutic strategies that will improve the currently disappointing cure rate of only 25-40% in this group of patients. Joanne Mason, WMRGL Birmingham

Joanne Mason, WMRGL Birmingham Acknowledgements Birmingham, WMRGL: Val Davison Mike Griffiths Fiona Macdonald Susanna Akiki Paula White Natalie Morrell Charlene Crosby Birmingham Clinicians: Dr Prem Mahendra Prof Charlie Craddock I will finish by acknowledging my colleagues in the West Midlands Regional Genetics Laboratory, the clinicians and the patients. Thank you for your attention. Thank you for your attention Joanne Mason, WMRGL Birmingham