Allen Eng Juh Yeoh

Clinical Utility of Microarray-Based Gene Expression Profiling in the Diagnosis and Subclassification of Leukemia: Report From the International Microarray Innovations in Leukemia Study Group

PURPOSE The Microarray Innovations in Leukemia study assessed the clinical utility of gene expression profiling as a single test to subtype leukemias into conventional categories of myeloid and lymphoid malignancies. METHODS The investigation was performed in 11 laboratories across three continents and included 3,334 patients. An exploratory retrospective stage I study was designed for biomarker discovery and generated whole-genome expression profiles from 2,143 patients with leukemias and myelodysplastic syndromes. The gene expression profiling-based diagnostic accuracy was further validated in a prospective second study stage of an independent cohort of 1,191 patients. RESULTS On the basis of 2,096 samples, the stage I study achieved 92.2% classification accuracy for all 18 distinct classes investigated (median specificity of 99.7%). In a second cohort of 1,152 prospectively collected patients, a classification scheme reached 95.6% median sensitivity and 99.8% median specificity for 14 standard subtypes of acute leukemia (eight acute lymphoblastic leukemia and six acute myeloid leukemia classes, n = 693). In 29 (57%) of 51 discrepant cases, the microarray results had outperformed routine diagnostic methods. CONCLUSION Gene expression profiling is a robust technology for the diagnosis of hematologic malignancies with high accuracy. It may complement current diagnostic algorithms and could offer a reliable platform for patients who lack access to todays state-of-the-art diagnostic work-up. Our comprehensive gene expression data set will be submitted to the public domain to foster research focusing on the molecular understanding of leukemias.

Childhood Acute Lymphoblastic Leukemia: Progress Through Collaboration

PURPOSE To review the impact of collaborative studies on advances in the biology and treatment of acute lymphoblastic leukemia (ALL) in children and adolescents. METHODS A review of English literature on childhood ALL focusing on collaborative studies was performed. The resulting article was reviewed and revised by the committee chairs of the major ALL study groups. RESULTS With long-term survival rates for ALL approaching 90% and the advent of high-resolution genome-wide analyses, several international study groups or consortia were established to conduct collaborative research to further improve outcome. As a result, treatment strategies have been improved for several subtypes of ALL, such as infant, MLL-rearranged, Philadelphia chromosome-positive, and Philadelphia chromosome-like ALL. Many recurrent genetic abnormalities that respond to tyrosine kinase inhibitors and multiple genetic determinants of drug resistance and toxicities have been identified to help develop targeted therapy. Several genetic polymorphisms have been recognized that show susceptibility to developing ALL and that help explain the racial/ethnic differences in the incidence of ALL. CONCLUSION The information gained from collaborative studies has helped decipher the heterogeneity of ALL to help improve personalized treatment, which will further advance the current high cure rate and the quality of life for children and adolescents with ALL.

Simple rules underlying gene expression profiles of more than six subtypes of acute lymphoblastic leukemia (ALL) patients.

MOTIVATIONS AND RESULTS For classifying gene expression profiles or other types of medical data, simple rules are preferable to non-linear distance or kernel functions. This is because rules may help us understand more about the application in addition to performing an accurate classification. In this paper, we discover novel rules that describe the gene expression profiles of more than six subtypes of acute lymphoblastic leukemia (ALL) patients. We also introduce a new classifier, named PCL, to make effective use of the rules. PCL is accurate and can handle multiple parallel classifications. We evaluate this method by classifying 327 heterogeneous ALL samples. Our test error rate is competitive to that of support vector machines, and it is 71% better than C4.5, 50% better than Naive Bayes, and 43% better than k-nearest neighbour. Experimental results on another independent data sets are also presented to show the strength of our method. AVAILABILITY Under http://sdmc.lit.org.sg/GEDatasets/, click on Supplementary Information.

An international standardization programme towards the application of gene expression profiling in routine leukaemia diagnostics: the Microarray Innovations in LEukemia study prephase

Gene expression profiling has the potential to enhance current methods for the diagnosis of haematological malignancies. Here, we present data on 204 analyses from an international standardization programme that was conducted in 11 laboratories as a prephase to the Microarray Innovations in LEukemia (MILE) study. Each laboratory prepared two cell line samples, together with three replicate leukaemia patient lysates in two distinct stages: (i) a 5‐d course of protocol training, and (ii) independent proficiency testing. Unsupervised, supervised, and r2 correlation analyses demonstrated that microarray analysis can be performed with remarkably high intra‐laboratory reproducibility and with comparable quality and reliability.

NUDT15 polymorphisms alter thiopurine metabolism and hematopoietic toxicity

Widely used as anticancer and immunosuppressive agents, thiopurines have narrow therapeutic indices owing to frequent toxicities, partly explained by TPMT genetic polymorphisms. Recent studies identified germline NUDT15 variation as another critical determinant of thiopurine intolerance, but the underlying molecular mechanisms and the clinical implications of this pharmacogenetic association remain unknown. In 270 children enrolled in clinical trials for acute lymphoblastic leukemia in Guatemala, Singapore and Japan, we identified four NUDT15 coding variants (p.Arg139Cys, p.Arg139His, p.Val18Ile and p.Val18_Val19insGlyVal) that resulted in 74.4–100% loss of nucleotide diphosphatase activity. Loss-of-function NUDT15 diplotypes were consistently associated with thiopurine intolerance across the three cohorts (P = 0.021, 2.1 × 10−5 and 0.0054, respectively; meta-analysis P = 4.45 × 10−8, allelic effect size = −11.5). Mechanistically, NUDT15 inactivated thiopurine metabolites and decreased thiopurine cytotoxicity in vitro, and patients with defective NUDT15 alleles showed excessive levels of thiopurine active metabolites and toxicity. Taken together, these results indicate that a comprehensive pharmacogenetic model integrating NUDT15 variants may inform personalized thiopurine therapy.

Pre- and Postnatal Transplantation of Fetal Mesenchymal Stem Cells in Osteogenesis Imperfecta: A Two-Center Experience

Osteogenesis imperfecta (OI) can be recognized prenatally with ultrasound. Transplantation of mesenchymal stem cells (MSCs) has the potential to ameliorate skeletal damage. We report the clinical course of two patients with OI who received prenatal human fetal MSC (hfMSC) transplantation and postnatal boosting with same‐donor MSCs. We have previously reported on prenatal transplantation for OI type III. This patient was retransplanted with 2.8 × 106 same‐donor MSCs per kilogram at 8 years of age, resulting in low‐level engraftment in bone and improved linear growth, mobility, and fracture incidence. An infant with an identical mutation who did not receive MSC therapy succumbed at 5 months despite postnatal bisphosphonate therapy. A second fetus with OI type IV was also transplanted with 30 × 106 hfMSCs per kilogram at 31 weeks of gestation and did not suffer any new fractures for the remainder of the pregnancy or during infancy. The patient followed her normal growth velocity until 13 months of age, at which time longitudinal length plateaued. A postnatal infusion of 10 × 106 MSCs per kilogram from the same donor was performed at 19 months of age, resulting in resumption of her growth trajectory. Neither patient demonstrated alloreactivity toward the donor hfMSCs or manifested any evidence of toxicities after transplantation. Our findings suggest that prenatal transplantation of allogeneic hfMSCs in OI appears safe and is of likely clinical benefit and that retransplantation with same‐donor cells is feasible. However, the limited experience to date means that it is not possible to be conclusive and that further studies are required.

Nomenclature for alleles of the thiopurine methyltransferase gene

The drug-metabolizing enzyme thiopurine methyltransferase (TPMT) has become one of the best examples of pharmacogenomics to be translated into routine clinical practice. TPMT metabolizes the thiopurines 6-mercaptopurine, 6-thioguanine, and azathioprine, drugs that are widely used for treatment of acute leukemias, inflammatory bowel diseases, and other disorders of immune regulation. Since the discovery of genetic polymorphisms in the TPMT gene, many sequence variants that cause a decreased enzyme activity have been identified and characterized. Increasingly, to optimize dose, pretreatment determination of TPMT status before commencing thiopurine therapy is now routine in many countries. Novel TPMT sequence variants are currently numbered sequentially using PubMed as a source of information; however, this has caused some problems as exemplified by two instances in which authors’ articles appeared on PubMed at the same time, resulting in the same allele numbers given to different polymorphisms. Hence, there is an urgent need to establish an order and consensus to the numbering of known and novel TPMT sequence variants. To address this problem, a TPMT nomenclature committee was formed in 2010, to define the nomenclature and numbering of novel variants for the TPMT gene. A website (http://www.imh.liu.se/tpmtalleles) serves as a platform for this work. Researchers are encouraged to submit novel TPMT alleles to the committee for designation and reservation of unique allele numbers. The committee has decided to renumber two alleles: nucleotide position 106 (G>A) from TPMT*24 to TPMT*30 and position 611 (T>C, rs79901429) from TPMT*28 to TPMT*31. Nomenclature for all other known alleles remains unchanged.

Profiling of somatic mutations in acute myeloid leukemia with FLT3-ITD at diagnosis and relapse

Acute myeloid leukemia (AML) with an FLT3 internal tandem duplication (FLT3-ITD) mutation is an aggressive hematologic malignancy with a grave prognosis. To identify the mutational spectrum associated with relapse, whole-exome sequencing was performed on 13 matched diagnosis, relapse, and remission trios followed by targeted sequencing of 299 genes in 67 FLT3-ITD patients. The FLT3-ITD genome has an average of 13 mutations per sample, similar to other AML subtypes, which is a low mutation rate compared with that in solid tumors. Recurrent mutations occur in genes related to DNA methylation, chromatin, histone methylation, myeloid transcription factors, signaling, adhesion, cohesin complex, and the spliceosome. Their pattern of mutual exclusivity and cooperation among mutated genes suggests that these genes have a strong biological relationship. In addition, we identified mutations in previously unappreciated genes such as MLL3, NSD1, FAT1, FAT4, and IDH3B. Mutations in 9 genes were observed in the relapse-specific phase. DNMT3A mutations are the most stable mutations, and this DNMT3A-transformed clone can be present even in morphologic complete remissions. Of note, all AML matched trio samples shared at least 1 genomic alteration at diagnosis and relapse, suggesting common ancestral clones. Two types of clonal evolution occur at relapse: either the founder clone recurs or a subclone of the founder clone escapes from induction chemotherapy and expands at relapse by acquiring new mutations. Relapse-specific mutations displayed an increase in transversions. Functional assays demonstrated that both MLL3 and FAT1 exert tumor-suppressor activity in the FLT3-ITD subtype. An inhibitor of XPO1 synergized with standard AML induction chemotherapy to inhibit FLT3-ITD growth. This study clearly shows that FLT3-ITD AML requires additional driver genetic alterations in addition to FLT3-ITD alone.

Identification of prognostic protein biomarkers in childhood acute lymphoblastic leukemia (ALL).

Early response to 7 days of prednisolone (PRED) treatment is one of the important prognostic factors in predicting eventual outcome in childhood acute lymphoblastic leukemia (ALL). Using proteomic tools and clinically important leukemia cell lines (REH, 697, Sup-B15, RS4; 11), we have identified potential prognostic protein biomarkers as well as discovered promising regulators of PRED-induced apoptosis. After treatment with PRED, the four cell lines can be separated into resistant (REH) and sensitive (697, Sup-B15, RS4;11). Two dimensional gel electrophoresis (2-DE) and MALDI-TOF/TOF MS identified 77 and 17 significantly differentially expressed protein spots (p<0.05) in PRED-sensitive and PRED-resistant cell lines respectively. Several of these were validated by Western blot including proliferating cell nuclear antigen (PCNA), cofilin 1, voltage-dependent anion-channel protein 1 (VDAC1) and proteasome activator subunit 2 (PA28β). PCNA is a promising protein because of its important roles both in cell cycle regulation and survival control. We subsequently validated PCNA in 43 paired bone marrow samples from children with newly diagnosed ALL (Day 0) and 7 days after PRED treatment (Day 8). ROC curve analysis confirmed that PCNA was highly predictive of PRED response in patients (AUC=0.81, p=0.007) and most interestingly, independent of the molecular subtype, providing a promising universal prognostic marker.

Mutant nucleophosmin deregulates cell death and myeloid differentiation through excessive caspase-6 and -8 inhibition

In up to one-third of patients with acute myeloid leukemia, a C-terminal frame-shift mutation results in abnormal and abundant cytoplasmic accumulation of the usually nucleoli-bound protein nucleophosmin (NPM), and this is thought to function in cancer pathogenesis. Here, we demonstrate a gain-of-function role for cytoplasmic NPM in the inhibition of caspase signaling. The NPM mutant specifically inhibits the activities of the cell-death proteases, caspase-6 and -8, through direct interaction with their cleaved, active forms, but not the immature procaspases. The cytoplasmic NPM mutant not only affords protection from death ligand-induced cell death but also suppresses caspase-6/-8-mediated myeloid differentiation. Our data hence provide a potential explanation for the myeloid-specific involvement of cytoplasmic NPM in the leukemogenesis of a large subset of acute myeloid leukemia.