ALBERT

Description: Abstract 612 Background: Chromosome 5q deletion (del5q) is the most common cytogenetic abnormality in myelodysplastic syndrome (MDS). Although haplodeficiency of several genes may contribute to the disease phenotype, allelic deletion of the ribosomal protein S14 (RPS14) gene is a key effector of the hypoplastic anemia. Disruption of ribosome assembly arising from RPS14 deletion leads to nucleolar stress that triggers p53 activation. In a murine model of the human 5q- syndrome, TP53 inactivation was alone sufficient to rescue the hematologic phenotype, indicating that the molecular pathogenesis of del(5q) MDS is p53-dependent. The tumor suppressor TP53 gene is a key regulator of stem cell homeostasis and senescence. A well described single nucleotide polymorphism (SNP) located at codon 72 in the proline-rich, pro-apoptotic domain of TP53 has been linked to cancer and mutagen susceptibility, and treatment outcome. Substitution of a cytosine (‘C’ allele) for the more common guanine (‘G’ allele) results in translation of a proline rather than arginine residue at position 72, with diminished apoptotic potential. Given the pathogenetic role of p53 in del(5q) MDS, we hypothesized that homozygosity for the ‘C’ allele may be associated with disease predisposition. Methods: Bone marrow and blood samples were investigated from 118 del(5q) MDS patients, 102 non-del(5q) MDS patients, and 98 healthy controls. Genomic DNA was extracted and codon 72 of the TP53 gene was amplified by PCR. Forward and reverse Sanger sequencing was performed to determine genotype. Relationship to disease specific features at diagnosis including cytogenetic risk category, IPSS score, blast percentage, and age were investigated as well as the relationship to response to lenalidomide and AML transformation using SAS software (Version 9.2, SAS Institute Inc., Cary, NC, USA). Results: Genotype distribution significantly differed between del(5q) MDS patients (18% CC, 47% CG, and 35% GG), non-del(5q) MDS patients (9% CC, 58% CG, and 33% GG),and healthy controls (7% CC, 43% CG, and 50% GG) (p=0.01). The frequency of the homozygous CC genotype was 〉2x greater in del(5q) MDS (18%) compared to both non-del(5q) MDS (9%) and healthy controls (7%) (p=0.05). There was no significant frequency difference between non-del(5q) and healthy controls. Del(5q) MDS patients were 〉6 times more likely to carry the CC genotype vs. GG when compared to healthy controls [odds ratio (OR)=6.71, 95% CI: 1.56 to 28.86], whereas non-del(5q) patients were 〉3 times more likely to carry the CC genotype vs. GG when compared to healthy controls (OR=3.87, 95% CI: 0.66 to 22.71). The corresponding ‘C’ allele frequency was significantly greater among del(5q) MDS patients (41.7%) compared to healthy controls (28.6%) (p=0.006), and approached significance in non-del(5q) patients (37.8%) versus controls (p=0.06). There was no association between TP53 R72P genotype and cytogenetic risk group in either del(5q) (p=0.67) or non-del(5q) MDS patients (p= 0.60), IPSS (del5q, p=0.29; non-del5q, p=0.89), or response to lenalidomide (del5q, p=0.57; non-del5q p=0.89). Mean age at diagnosis was significantly (p=0.04) lower in del(5q) MDS (67.4 years; SD=11.1 years) compared to non-del(5q) MDS patients (70.8 years; SD=9.1years), although significant differences in age according to TP53 R72P genotype were not apparent in either MDS cytogenetic group (del5q, p=0.99; non-del5q, p=0.89). Conclusion: Our findings indicate that the TP53 R72P homozygous CC genotype occurs with significantly greater frequency in del(5q) MDS compared to both non-del(5q) MDS patients and healthy controls, suggesting that this polymorphism may play a key role in the pathogenesis of and predisposition to del(5q) MDS. Disclosures: Kurtin: Celgene: Honoraria. Maciejewski:Celgene: Research Funding; Eisai: Research Funding; Alexion: Consultancy. Nevill:Celgene: Honoraria. Karsan:Celgene: Research Funding. List:Celgene: Research Funding.

Description: Abstract 3374 Metaphase spreads have an established value in the routine diagnostic workup of myeloid malignancies and bone marrow failure disorders. In myelodysplastic syndrome (MDS) abnormal karyotypes play a large role in scoring systems and may greatly impact prognosis, or even predict responsiveness to certain therapies. In aplastic anemia (AA) cytogenetic abnormalities detected by metaphase karyotyping may rule out hypoplastic MDS. In myeloproliferative neoplasia (MPN), an abnormal karyotype may distinguish between reactive or malignant proliferation. Due to technical problems, including specimen quality, viability, hypocellularity or a failure of growth, this routine test may fail to yield conclusive results in some patients. With the advent of SNP-A karyotyping, which only requires extracted DNA, non-informative cases can be resolved. As a cytogenetic test, single nucleotide polymorphism array (SNP-A) analysis can provide an opportunity to improve risk assessment and selection of proper treatment modalities. The advantages of SNP-A include excellent resolution, detection of copy neutral loss of heterozygosity (also known as uniparental disomy or UPD), and perhaps most importantly, the ability to test archived DNA samples, rather than actively dividing cells. However, unlike metaphase cytogenetics, this technology cannot detect subsets of abnormal populations or certain classes of genomic rearrangements, such as balanced translocation, inversion or ring chromosomes. In this study, we examined the prognosis and disease characterization for patients with non-informative cytogenetics (N=144) collected over the last 8 years. SNP-A-based karyotyping has been performed for a representative subset of these patients (N=60) to assess whether this technique could provide clinically relevant information. These patients included patients with MDS (N=20), AA (N=20), AML (N=12) and MDS/MPN (N=3). Bone marrow obtained following induction chemotherapy was excluded. We have detected 27 somatic microdeletions and 33 microduplications (25Mb) in 2 cases, including 22q11.23qter and 14q12-q22.1. In 4/60 (7%) a complex karyotype was detected, while 10 had sole lesions (〉10Mb). In presumed AA patients, we have identified 2 patients with monosomy 7, prompting a change of diagnosis to MDS and thereby altering their clinical management. In MDS, when cytogenetic prognostic groupings were applied in previously unscored patients, 10/20 had IPSS scores of 3 or greater. The presence of chromosomal abnormalities detected by SNP-At indictated the presence of advanced risk disease and thereby contributed into poorer survival as predicted by IPSS. Disclosures: No relevant conflicts of interest to declare.

Description: Abstract 4016 Lenalidomide (LEN) is effective in patients with myelodysplastic syndromes (MDS) and the del(5q) cytogenetic abnormality, and has activity in a proportion of patients without this chromosomal defect. To date, no clinically actionable biomarkers, other than the presence of del(5q) detected by metaphase cytogenetics, have been identified to predict response to LEN. The diagnostic yield of metaphase cytogenetics can be enhanced by application of fluorescence in situ hybridization (FISH) for targeted chromosomal lesions including del(5q), as this technique is more sensitive. Similarly, single nucleotide polymorphism array (SNP-A)-based karyotyping, due to its superb resolution, allows for detection of previously cryptic unbalanced chromosomal defects. Both techniques can be preformed on interphase cells and thereby do not require cell division. We hypothesized that application of these technologies may allow for better identification of putative responders to LEN in patients with MDS without del(5q); we stipulated that i) we may detect previously unrecognized cases of del(5q) and that ii) more precise analysis of the karyotypes may allow for recognition of other chromosomal markers of response or refractoriness to LEN. In patients with MDS (N=82), MDS/myeloproliferative neoplasm (MPN) (N=13), acute myeloid leukemia (AML) (N=23), and MPN (N=4), the detection rate of del(5q) increased only marginally with use of additional techniques, from 24% (metaphase cytogenetics (MC) + FISH), to 25% (MC + SNP-A), 25% (FISH + SNP-A) and 26% (all 3 methods). Within this cohort, we then analyzed by FISH and SNP-A karyotype a subset of 42 patients with myeloid malignancies without del(5q) by MC who received LEN. This cohort included 33 MDS (RA, N=5; RARS, N=12; RARS-T, N=1; RCMD, N=1; RAEB1, N=4; RAEB2, N=6; MDS-U, N=4), 6 MDS/MPN and MPN patients (CMML1, N=1; CMML2, N=3, IMF; N=2) and 3 secondary AML. By MC, 32 (76%) showed normal karyotype, 1 (2.4%) no growth and 9 (21%) abnormal karyotype other than del(5q). The overall response rate (ORR) (2006 International Working Group criteria) was 44%, including 3 CR, 3 PR and 8 HI. Previously cryptic del(5q) was detected in an additional 1/18 patient by both SNP-A and FISH (secondary AML with normal metaphase cytogenetics), but this case was refractory to LEN. Del(5q) was also revealed by FISH in 1 patient with unsuccessful MC, but, due to the small size of the clone (8%), SNP-A did not detect this lesion. This patient had a sustained PR with transfusion independence. In 28 patients who received LEN for more than 3 months, the ORR to LEN in patients with normal metaphase cytogenetics was 62%, and 17% for those with chromosomal aberrations (p=.08); the addition of SNP-A did not improve the predictive value of normal cytogenetics. We also analyzed 10 patients without del(5q) by MC who received combination therapy with azacitidine (AZA) and LEN, for whom the ORR was 80% (7 CR, 1 PR). By metaphase cytogenetics, 7/10 patients had a normal karyogram and a response of 86%, compared to 3 patients with chromosomal lesions, 1 of whom responded. Similar to the results in LEN alone, inclusion of defects detected by SNP-A or FISH did not allow for better separation of responders based on normal cytogenetics by MC. Six out of 38 LEN-treated patients had a gain of chromosome 8 material by FISH or SNP-A. 4 out of 6 patients had CR (all of them received combination therapy of AZA and LEN), 1 out of 6 had HI, and 1 with complex karyotype had NR (ORR was 83%), while ORR in patients with other chromosomal abnormalities by FISH or SNP-A was 18%. In conclusion, FISH and SNP-A, when added to routine metaphase cytogenetics, marginally increased the diagnostic yield leading to detection of only 2/42 additional cases with del(5q). In our cohort, the non-del(5q) patients with normal karyotype and those with trisomy 8 or microduplication on chromosome 8 were associated with a favorable response to LEN. Disclosures: Sekeres: Celgene: Consultancy, Honoraria, Research Funding. List:Celgene: Research Funding. Maciejewski:Celgene: Research Funding.

Description: Abstract 607 Using single nucleotide polymorphism arrays (SNP-A) as a karyotyping tool led to the recognition that segmental somatic uniparental disomy (UPD) is a common and important defect in myelodysplastic syndromes (MDS) and related conditions. Homozygous mutations associated with UPD have been found. Additionally, SNP-A-detected submicroscopic deletions have proven helpful in narrowing the search for mutant genes to shared regions of loss of heterozygosity (LOH). Mapping of such microdeletions was instrumental in the identification of TET2 and CBL mutations, frequently found in a homozygous constellation within UPD of the corresponding chromosomal regions. UPD7q is one of the most frequent areas of UPD present in MDS, MDS/MPN and sAML. We hypothesized that this region may be associated with the presence of pathogenic mutations in a homozygous configuration. We utilized two strategies: i) because of the large size of the area commonly affected by UPD7, we applied next generation sequencing (NGS) for the identification of affected genes, and ii) we systematically sequenced genes contained in the overlapping deleted regions. In total, 15 patients with UPD7q were identified. Exome libraries were generated from 2 cases with UPD7q and subjected to sequencing. In one, a missense C to T mutation was found at nt148137376 of EZH2 (100% of sequences), resulting in R690H. The somatic origin of this homozygous mutation was confirmed by sequencing of DNA from a skin biopsy. Simultaneously, we identified 2 patients with overlapping microdeletion involving 7q36.1. The minimally affected area involved 2 genes: CUL1 and EZH2. Sequenced all exons of both genes revealed no mutations in CUL1. However, a mutation in EZH2 exon 19 was identified in proximity to the mutation detected through NGS. The mutation involved position Ile715 producing a frame shift. We detected EZH2 mutations in 10/351 patients (7% of MDS/MPN, 2% of MDS and 0.9% of AML). We found 8 different EZH2 mutations present in 4/15 (27%) of cases with UPD7q, 2/30 (7%) patients with deletion 7q and 4/306 (1%) without LOH. Of note is that both hemizygous mutations were found in patients with microdeletions and none of the patients with large del(7q) or monosomy 7 harbored EZH2 mutation. There were 8 missense and 2 frame shift mutations, located in the SET domain of the EZH2 gene. Diagnoses included 3 cases of refractory cytopenia with multilineage dysplasia, while 5 had myelomonocytoid malignancies, and one each with MDS/MPN unclassifiable and atypical chronic myeloid leukemia (aCML). Overall, EZH2 mutations were detected in 4/53 (8%) cases of CMML, consistent with the high prevalence of somatic UPD7 in this disease. Interestingly, no EZH2 mutant case had a chromosome 7 abnormality by metaphase cytogenetics, but SNP-A karyotyping detected cryptic LOH7q in 6/10 patients with EZH2 mutations. EZH2 is a polycomb associated gene encoding a methyltransferase targeting H3K27, thereby producing a repressive mark. Loss of function or hypomorphic mutations of EZH2 would thus be predicted to result in abrogation of inhibitory chromatin marks and chromatin decompaction, conducive to expression of oncogenes. Immunofluorescence and western blot showed that EZH2 mutations in myeloid malignancies lead to decreased methylation of H3K27 and thereby are functionally relevant. Expression analysis of EZH2 did not show significantly decreased expression in MDS or haploinsuffciency in patients with del7/7q. In a patient with aCML and UPD7q, homozygous R690H EZH2 mutation was detected and trimethylated H3K27 was abrogated while H3K9 methylation appeared normal, consistent with functional impairment of the SET domain. Interestingly, an EZH2 mutation was associated with heterozygous TET2 S733fsX21 and heterozygous ASXL1 W538X mutations, suggesting multiple genes associated with epigenetic regulation may be mutated and act synergistically in malignant evolution. These triple mutant cells ultimately formed lethal tumors when injected to NOD SCID gamma mice. In summary, our investigations demonstrate EZH2 mutations in patients with myeloid malignancies frequently associated with UPD7q, or 7q36.1 microdeletion, but not monosomy 7 and del(7q). These findings suggest that an increasing possibility that mutations in the polycomb gene family represent a new class of molecular lesions conveying a clonal epigenetic instability phenotype and can constitute leukemogenic events. Disclosures: No relevant conflicts of interest to declare.

Description: Germ line mutations in growth factor independent-1 (GFI1) have been described in a small subset of patients (pts) with severe congenital neutropenia. Subsequently, the GFI136N polymorphism, present in 3-5% of controls, has been found overrepresented (11%) in pts with primary (p) acute myeloid leukemia (AML), conveying 1.6 fold risk of development of AML. Mutant GFI136S and N variants lack affinity to HOXA9 (overrepresented in corresponding AML cases), shows increased proliferative potential in vitro and accelerates RAS-driven myeloproliferative neoplasm (MPN) disease in mice. Whole exome next generation (WE NGS) technology facilitates comprehensive screens for the presence of both somatic and germ line genetic alteration. In this study, we used NGS to search for germline variants of the GFI1 gene. We screened 140 pts (mean age 66.8 years, range 44-85) with MDS and related disorders (MDS/MPN and secondary (s) AML) for the presence of GFI1 variants. We found non-synonymous variants in 11 cases (8%), including the previously described pathogenic p.S36N (n=8), p.P107A (n=2), or p.L400F (n=1), while the corresponding frequencies for these alterations were .04 and .001, .002 in the general population. This frequency appears comparable or higher to those previously reported for pAML, but our screen of the TCGA AML cohort, perhaps due to very low coverage for this gene, did not reveal any GFI1 polymorphisms. We next focused on the clinical features of altered GFI1 carriers. A significant proportion of GFI1 cases were younger (age

Description: Abstract 297 Loss of heterozygosity (LOH) involving chromosome-7 is one of the most common unbalanced chromosomal defects found in chronic and acute myeloid malignancies. Recent application of single nucleotide polymorphism arrays (SNP-A) led to realization that recurrent areas of LOH may be not only due to deletion of the whole chromosome-7, its long arm but due to copy-neutral LOH, most often involving 7q. We hypothesized that deletions and uniparental disomy (UPD) involving long arm of chromosome-7 may be associated with pathogenic hemizygous or homozygous mutations, respectively. Such mutations may affect tumor suppressor genes and likely contribute to/drive malignant evolution in myeloid disorders. We identified a large number of patients with lesions of chromosome-7 (monosomy-7, del(7q), and UPD(7q)), including 38 patients with microdeletions 7q. After unsuccessful targeted Sanger sequencing of large numbers of genes on chromosome-7 we set up to apply the next generation sequencing (NGS) of the exome libraries generated from patients with LOH7. Exome chromosome-7 libraries were enriched for the content of coding sequences using the SureSelect capture synthetic biotinylated RNA probes, tiling all the coding regions from chromosome-7. For NGS exome sequencing we selected 6 cases of monosomy-7 (2 sAML, 1 MDS, 1 chronic myelomonocytic leukemia (CMML), 1 juvenile myelomonocytic leukemia (JMML), and 1 aplastic anemia (AA)), 2 cases of del(7q) with MDS, 3 cases of UPD(7q) (2 MDS and 1 MDS/MPN) and also studied paired germline samples when possible. We treated monosomy-7 and del(7q) as one class and analyzed them together. Cases with UPD(7q) were analyzed separately. Averaged sequencing depth was 79. Each monosomy-7 sample had 3788, 3792, 5306, 4013, 5237, 4016 potential alterations (average; 4359/sample), and each del(7q) sample had 5431, 4521 observations (average; 4976/sample). After exclusion of observations outside of affected regions del(7q), we eliminated previously reported SNPs and non-coding lesions and selected 658 observations of which 422 were nonsynonymous. For further analysis, the alteration leading to stop codons were chosen as tier 1 candidates (N=16). After elimination of false positives due to the accumulation of reading errors at specific locations were discarded, 3 candidates were left and verified by Sanger sequencing. All of 3 were confirmed as new SNPs. After elimination of false positives, the alteration present in multiple samples were designated as tier 2 group (N=101). All of these sequence changes were shown to be new SNPs; 2 were not confirmed by traditional sequencing. Mutations for which the non-reference base occurred greater than 50% were designated tier 3 candidates (N=73), of which 25 candidates have been already checked by resequencing. To date, we have identified 2 somatic point mutations confirmed by Sanger sequencing, including NRCAM1Q1040K and LMTK2A1147T and each identified in different monosomy-7 samples. Screening of 30 monosomy-7 or del(7q) samples showed that observed mutations were not recurrent. The similar stepwise analytic approach was applied to 3 cases of UPD(7q) (average; 4312 observations per sample). After exclusion of reported SNPs and synonymous alterations, we selected a total of 147 alterations. The number was further reduced to 5 potential pathogenic changes after elimination of false positive NGS artifacts. By verification with Sanger sequencing, recurrent EZH2 homozygous mutations (both were R690H) were confirmed in 2 cases with UPD7q. Then we sequenced whole the exons in additional 12 cases of UPD(7q) were sequenced yielding 2 EZH2 mutations. In addition, EZH2 mutations were identified in 2 cases of microdeletion 7q36.1 and 4 cases without LOH7q occurring in heterozygous constellation. No mutations of EZH2 were found in monosomy-7 or del(7q) sequenced (N=28). In conclusion, to date, using NGS strategy, we have identified 3 new mutations including NRCAM1Q1040K, LMTK2A1147T, and recurrently occurring mutation of EZH2 (EZH2R690H). Several additional candidate mutations are currently screened. Disclosures: Maciejewski: Eisai: Research Funding; Celgene: Research Funding.

Description: Abstract 307 Whole-exome (WES) sequencing revealed tremendous mutational heterogeneity in leukemia. While WES can be applied for discovery, it also has potential as a diagnostic tool that can overcome the shortcomings of current methods. We theorized that, in addition to mutation discovery, systematic application of WES in MDS may reveal distinct mutational patterns allowing for new molecular classification. We performed WES in 116 paired exomes, including MDS (n=57), MDS/MPN (n=36), and sAML (n=23). We also included comparative analysis with pAML (N=202; TCGA), and other publicly available data for a total of 333 exomes; 10 patients were studied serially. Paired DNA (marrow/CD3+ cells) was subjected to WES, sequence-aligned by BW Aligner, and variants detected via GATK pipeline (Broad Institute). We used defined criteria to minimize false-positives: P

Description: Abstract 1405 Genes involved in congenital genetic cancer susceptibility syndromes are also targets of somatic mutations in various tumors. Examples include WT1, NF1, CBL, TP53 and MLL2 affected both in germ line as well as somatic mutations present in malignant disorders. To apply this approach to investigation of pathogenic mutations in myeloid malignancies, we selected 183 congenital disorders in which germline mutations of disease specific genes are reported to be pathogenic. Their main clinical presentations are skeletal abnormalities (N=54 disorders), skin abnormality (N=24), mental retardation (N=17) and hematological disorders (N=12). In total, we searched for mutations in 204 genes associated with these congenital disorders. We analyzed whole exome of various myeloid malignancies, including 60 cases with myelodysplastic syndromes (MDS), 29 MDS/MPN, 5 with MPN and 122 with acute myeloid leukemia (AML) and found somatic mutations in 62 genes, which also mutated in germ line in various congenital syndromes. Of those, the most frequently mutated genes were TP53 (25 cases) and WT1 (16 cases), associated with germline mutation of Li-Fraumeni syndrome and Wilms tumor, respectively. Some somatic mutations, for example, NF1 (R1276Q) and PTPN11 (D61N), were exactly the same as observed in corresponding congenital disorders (Neurofibromatosis or Noonan syndrome). One of the novel findings is that somatic SET binding protein 1 (SETBP1) mutations (D868N, G870S and I871T) were commonly observed in sAML and CMML, and were identical to germline mutations in Schinzel-Giedion syndrome (see designated abstract). We found recurrent somatic SETBP1 mutations in 15% of each CMML and sAML. Moreover, multiple genes pathogenic in Usher syndrome (congenital hearing and vision loss, complicated by vasoproliferative retinal tumor), were somatically mutated in various myeloid neoplasms. Out of 9 genes which are causative for this syndrome, 15 mutations of 6 genes (MYO7A, USH1C, CDH23, PCDH15, USH2A, and GPR98) were observed in 13 cases, including 2 frameshift and 13 missense mutations. These genes coordinate with each other to form a functional network. CDH23 and PCDH15 are cadherins and act as cell adhesion molecules. MYO7A are actin-based motor molecules with a variety of functions. USH1C serves as an anchor and codes for a scaffolding protein to form a complex with all the other proteins. Through the PDZ binding site, USH1C forms a complex with CDH23, which was the most frequently mutated gene in this family (1 frame shift and 3 misssense mutations). CDH23 mutations were observed in 2 cases with primary AML, sAML and MDS. Specifically, a somatic missense mutation G2771S of CDH23 in a secondary AML case was identical to germline of Usher syndrome. The second most frequently mutated gene, GPR98, is located in 5q14.3 locus; a small hemizygous clone found in del5q of an MDS case. In a serial sample analysis, this mutation increased to become the larger main clone during AML evolution. Moreover, in this case, an additional CDH23 mutation was acquired in the course of leukemic expansion. In such cases with Usher syndrome gene mutations, U2AF1, ZRSR2, EZH2, IDH2 and ETV6 mutations were also observed, suggesting pathogenic cooperation with these well-known tumor suppressor genes and oncogenes. Disclosures: Maciejewski: NIH: Research Funding; Aplastic Anemia&MDS International Foundation: Research Funding. Makishima:Scott Hamilton CARES Initiative: Research Funding.