Comparing malignant monocytosis across the updated WHO and ICC classifications of 2022.

ABSTRACT: The World Health Organization (WHO) classification of hematolymphoid tumors and the International Consensus Classification (ICC) of 2022 introduced major changes to the definition of chronic myelomonocytic leukemia (CMML). To assess its qualitative and quantitative implications for patient care, we started with 3311 established CMML cases (according to WHO 2017 criteria) and included 2130 oligomonocytosis cases fulfilling the new CMML diagnostic criteria. Applying both 2022 classification systems, 356 and 241 of oligomonocytosis cases were newly classified as myelodysplastic (MD)-CMML (WHO and ICC 2022, respectively), most of which were diagnosed as myelodysplastic syndrome (MDS) according to the WHO 2017 classification. Importantly, 1.5 times more oligomonocytosis cases were classified as CMML according to WHO 2022 than based on ICC, because of different diagnostic criteria. Genetic analyses of the newly classified CMML cases showed a distinct mutational profile with strong enrichment of MDS-typical alterations, resulting in a transcriptional subgroup separated from established MD and myeloproliferative CMML. Despite a different cytogenetic, molecular, immunophenotypic, and transcriptional landscape, no differences in overall survival were found between newly classified and established MD-CMML cases. To the best of our knowledge, this study represents the most comprehensive analysis of routine CMML cases to date, both in terms of clinical characterization and transcriptomic analysis, placing newly classified CMML cases on a disease continuum between MDS and previously established CMML.

Developmental trajectories and cooperating genomic events define molecular subtypes of BCR::ABL1-positive ALL.

ABSTRACT: Distinct diagnostic entities within BCR::ABL1-positive acute lymphoblastic leukemia (ALL) are currently defined by the International Consensus Classification of myeloid neoplasms and acute leukemias (ICC): "lymphoid only", with BCR::ABL1 observed exclusively in lymphatic precursors, vs "multilineage", where BCR::ABL1 is also present in other hematopoietic lineages. Here, we analyzed transcriptomes of 327 BCR::ABL1-positive patients with ALL (age, 2-84 years; median, 46 years) and identified 2 main gene expression clusters reproducible across 4 independent patient cohorts. Fluorescence in situ hybridization analysis of fluorescence-activated cell-sorted hematopoietic compartments showed distinct BCR::ABL1 involvement in myeloid cells for these clusters (n = 18/18 vs n = 3/16 patients; P < .001), indicating that a multilineage or lymphoid BCR::ABL1 subtype can be inferred from gene expression. Further subclusters grouped samples according to cooperating genomic events (multilineage: HBS1L deletion or monosomy 7; lymphoid: IKZF1-/- or CDKN2A/PAX5 deletions/hyperdiploidy). A novel HSB1L transcript was highly specific for BCR::ABL1 multilineage cases independent of HBS1L genomic aberrations. Treatment on current German Multicenter Study Group for Adult ALL (GMALL) protocols resulted in comparable disease-free survival (DFS) for multilineage vs lymphoid cluster patients (3-year DFS: 70% vs 61%; P = .530; n = 91). However, the IKZF1-/- enriched lymphoid subcluster was associated with inferior DFS, whereas hyperdiploid cases showed a superior outcome. Thus, gene expression clusters define underlying developmental trajectories and distinct patterns of cooperating events in BCR::ABL1-positive ALL with prognostic relevance.

Enhancer retargeting of CDX2 and UBTF::ATXN7L3 define a subtype of high-risk B-progenitor acute lymphoblastic leukemia.

Transcriptome sequencing has identified multiple subtypes of B-progenitor acute lymphoblastic leukemia (B-ALL) of prognostic significance, but a minority of cases lack a known genetic driver. Here, we used integrated whole-genome (WGS) and -transcriptome sequencing (RNA-seq), enhancer mapping, and chromatin topology analysis to identify previously unrecognized genomic drivers in B-ALL. Newly diagnosed (n = 3221) and relapsed (n = 177) B-ALL cases with tumor RNA-seq were studied. WGS was performed to detect mutations, structural variants, and copy number alterations. Integrated analysis of histone 3 lysine 27 acetylation and chromatin looping was performed using HiChIP. We identified a subset of 17 newly diagnosed and 5 relapsed B-ALL cases with a distinct gene expression profile and 2 universal and unique genomic alterations resulting from aberrant recombination-activating gene activation: a focal deletion downstream of PAN3 at 13q12.2 resulting in CDX2 deregulation by the PAN3 enhancer and a focal deletion of exons 18-21 of UBTF at 17q21.31 resulting in a chimeric fusion, UBTF::ATXN7L3. A subset of cases also had rearrangement and increased expression of the PAX5 gene, which is otherwise uncommon in B-ALL. Patients were more commonly female and young adult with median age 35 (range,12-70 years). The immunophenotype was characterized by CD10 negativity and immunoglobulin M positivity. Among 16 patients with known clinical response, 9 (56.3%) had high-risk features including relapse (n = 4) or minimal residual disease >1% at the end of remission induction (n = 5). CDX2-deregulated, UBTF::ATXN7L3 rearranged (CDX2/UBTF) B-ALL is a high-risk subtype of leukemia in young adults for which novel therapeutic approaches are required.

Classical meets malignant hematology: a case of acquired εγδß-thalassemia in clonal hematopoiesis.

Hemoglobinopathies including thalassemias are among the most frequent genetic disorders worldwide. Primarily, these entities result from germline variants in the globin gene clusters and their cis-acting regulatory elements, and thus the WHO classifies thalassemias as inherited diseases. Non-inherited disorders of globin chain synthesis mimicking the phenotype of thalassemias have also been described and are referred to as acquired thalassemias. These forms mainly affect the alpha-globin genes and are observed at much lower frequencies...

Machine learning integrates genomic signatures for subclassification beyond primary and secondary acute myeloid leukemia.

Although genomic alterations drive the pathogenesis of acute myeloid leukemia (AML), traditional classifications are largely based on morphology, and prototypic genetic founder lesions define only a small proportion of AML patients. The historical subdivision of primary/de novo AML and secondary AML has shown to variably correlate with genetic patterns. The combinatorial complexity and heterogeneity of AML genomic architecture may have thus far precluded genomic-based subclassification to identify distinct molecularly defined subtypes more reflective of shared pathogenesis. We integrated cytogenetic and gene sequencing data from a multicenter cohort of 6788 AML patients that were analyzed using standard and machine learning methods to generate a novel AML molecular subclassification with biologic correlates corresponding to underlying pathogenesis. Standard supervised analyses resulted in modest cross-validation accuracy when attempting to use molecular patterns to predict traditional pathomorphologic AML classifications. We performed unsupervised analysis by applying the Bayesian latent class method that identified 4 unique genomic clusters of distinct prognoses. Invariant genomic features driving each cluster were extracted and resulted in 97% cross-validation accuracy when used for genomic subclassification. Subclasses of AML defined by molecular signatures overlapped current pathomorphologic and clinically defined AML subtypes. We internally and externally validated our results and share an open-access molecular classification scheme for AML patients. Although the heterogeneity inherent in the genomic changes across nearly 7000 AML patients was too vast for traditional prediction methods, machine learning methods allowed for the definition of novel genomic AML subclasses, indicating that traditional pathomorphologic definitions may be less reflective of overlapping pathogenesis.

Novel NPM1 exon 5 mutations and gene fusions leading to aberrant cytoplasmic nucleophosmin in AML.

Nucleophosmin (NPM1) mutations in acute myeloid leukemia (AML) affect exon 12, but also sporadically affect exons 9 and 11, causing changes at the protein C-terminal end (tryptophan loss, nuclear export signal [NES] motif creation) that lead to aberrant cytoplasmic NPM1 (NPM1c+), detectable by immunohistochemistry. Combining immunohistochemistry and molecular analyses in 929 patients with AML, we found non-exon 12 NPM1 mutations in 5 (1.3%) of 387 NPM1c+ cases. Besides mutations in exons 9 (n = 1) and 11 (n = 1), novel exon 5 mutations were discovered (n = 3). Another exon 5 mutation was identified in an additional 141 patients with AML selected for wild-type NPM1 exon 12. Three NPM1 rearrangements (NPM1/RPP30, NPM1/SETBP1, NPM1/CCDC28A) were detected and characterized among 13 979 AML samples screened by cytogenetic/fluorescence in situ hybridization and RNA sequencing. Functional studies demonstrated that in AML cases, new NPM1 proteins harbored an efficient extra NES, either newly created or already present in the fusion partner, ensuring its cytoplasmic accumulation. Our findings support NPM1 cytoplasmic relocation as critical for leukemogenesis and reinforce the role of immunohistochemistry in predicting AML-associated NPM1 genetic lesions. This study highlights the need to develop new assays for molecular diagnosis and monitoring of NPM1-mutated AML.

Molecular characterization and clinical outcome of B-cell precursor acute lymphoblastic leukemia with IG-MYC rearrangement.

Rarely, immunophenotypically immature B-cell precursor acute lymphoblastic leukemia (BCP-ALL) carries an immunoglobulin- MYC rearrangement (IG-MYC-r). This can result in diagnostic confusion with Burkitt lymphoma/leukemia and use of individualized treatment schedules of unproven efficacy. Here we compare the molecular characteristics of these conditions and investigate historic clinical outcome data. We identified 90 cases registered in a national BCP-ALL clinical trial/registry. When present, diagnostic material underwent cytogenetic, exome, methylome and transcriptome analyses. The outcomes analyzed were 3-year event-free survival and overall survival. IG-MYC-r was identified in diverse cytogenetic backgrounds, co-existing with either established BCP-ALL-specific abnormalities (high hyperdiploidy, n=3; KMT2A-rearrangement, n=6; iAMP21, n=1; BCR-ABL1, n=1); BCL2/BCL6-rearrangements (n=15); or, most commonly, as the only defining feature (n=64). Within this final group, precursor-like V(D)J breakpoints predominated (8/9) and KRAS mutations were common (5/11). DNA methylation identified a cluster of V(D)J-rearranged cases, clearly distinct from Burkitt leukemia/lymphoma. Children with IG-MYC-r within that subgroup had a 3-year event-free survival of 47% and overall survival of 60%, representing a high-risk BCP-ALL. To develop effective management strategies this group of patients must be allowed access to contemporary, minimal residual disease-adapted, prospective clinical trial protocols.

Application of precision medicine in clinical routine in haematology-Challenges and opportunities.

Precision medicine is revolutionising patient care in cancer. As more knowledge is gained about the impact of specific genetic lesions on diagnosis, prognosis and treatment response, diagnostic precision and the possibility for optimal individual treatment choice have improved. Identification of hallmark genetic aberrations such as the BCR::ABL1 gene fusion in chronic myeloid leukaemia (CML) led to the rapid development of efficient targeted therapy and molecular follow-up, vastly improving survival for patients with CML during recent decades. The assessment of translocations, copy number changes and point mutations are crucial for the diagnosis and risk stratification of acute myeloid leukaemia and myelodysplastic syndromes. Still, the often heterogeneous and complex genetic landscape of haematological malignancies presents several challenges for the implementation of precision medicine to guide diagnosis, prognosis and treatment choice. This review provides an introduction and overview of the important molecular characteristics and methods currently applied in clinical practice to guide clinical decision making in haematological malignancies of myeloid and lymphoid origin. Further, experimental ways to guide the choice of targeted therapy for refractory patients are reviewed, such as functional precision medicine using drug profiling. An example of the use of pipeline studies where the treatment is chosen according to the molecular characteristics in rare solid malignancies is also provided. Finally, the future opportunities and remaining challenges of precision medicine in the real world are discussed.

Atypical 3q26/MECOM rearrangements genocopy inv(3)/t(3;3) in acute myeloid leukemia.

Acute myeloid leukemia (AML) with inv(3)/t(3;3)(q21q26) is a distinct World Health Organization recognized entity, characterized by its aggressive course and poor prognosis. In this subtype of AML, the translocation of a GATA2 enhancer (3q21) to MECOM (3q26) results in overexpression of the MECOM isoform EVI1 and monoallelic expression of GATA2 from the unaffected allele. The full-length MECOM transcript, MDS1-EVI1, is not expressed as the result of the 3q26 rearrangement. Besides the classical inv(3)/t(3;3), a number of other 3q26/MECOM rearrangements with poor treatment response have been reported in AML. Here, we demonstrate, in a group of 33 AML patients with atypical 3q26 rearrangements, MECOM involvement with EVI1 overexpression but no or low MDS1-EVI1 levels. Moreover, the 3q26 translocations in these AML patients often involve superenhancers of genes active in myeloid development (eg, CD164, PROM1, CDK6, or MYC). In >50% of these cases, allele-specific GATA2 expression was observed, either by copy-number loss or by an unexplained allelic imbalance. Altogether, atypical 3q26 recapitulate the main leukemic mechanism of inv(3)/t(3;3) AML, namely EVI1 overexpression driven by enhancer hijacking, absent MDS1-EVI1 expression and potential GATA2 involvement. Therefore, we conclude that both atypical 3q26/MECOM and inv(3)/t(3;3) can be classified as a single entity of 3q26-rearranged AMLs. Routine analyses determining MECOM rearrangements and EVI1 and MDS1-EVI1 expression are required to recognize 3q-rearranged AML cases.

Molecular landscape and clonal architecture of adult myelodysplastic/myeloproliferative neoplasms.

More than 90% of patients with myelodysplastic/myeloproliferative neoplasms (MDSs/MPNs) harbor somatic mutations in myeloid-related genes, but still, current diagnostic criteria do not include molecular data. We performed genome-wide sequencing techniques to characterize the mutational landscape of a large and clinically well-characterized cohort including 367 adults with MDS/MPN subtypes, including chronic myelomonocytic leukemia (CMML; n = 119), atypical chronic myeloid leukemia (aCML; n = 71), MDS/MPN with ring sideroblasts and thrombocytosis (MDS/MPN-RS-T; n = 71), and MDS/MPN unclassifiable (MDS/MPN-U; n = 106). A total of 30 genes were recurrently mutated in ≥3% of the cohort. Distribution of recurrently mutated genes and clonal architecture differed among MDS/MPN subtypes. Statistical analysis revealed significant correlations between recurrently mutated genes, as well as genotype-phenotype associations. We identified specific gene combinations that were associated with distinct MDS/MPN subtypes and that were mutually exclusive with most of the other MDSs/MPNs (eg, TET2-SRSF2 in CMML, ASXL1-SETBP1 in aCML, and SF3B1-JAK2 in MDS/MPN-RS-T). Patients with MDS/MPN-U were the most heterogeneous and displayed different molecular profiles that mimicked the ones observed in other MDS/MPN subtypes and that had an impact on the outcome of the patients. Specific gene mutations also had an impact on the outcome of the different MDS/MPN subtypes, which may be relevant for clinical decision-making. Overall, the results of this study help to elucidate the heterogeneity found in these neoplasms, which can be of use in the clinical setting of MDS/MPN.

Chromosome banding analysis and genomic microarrays are both useful but not equivalent methods for genomic complexity risk stratification in chronic lymphocytic leukemia patients.

Genome complexity has been associated with poor outcome in patients with chronic lymphocytic leukemia (CLL). Previous cooperative studies established five abnormalities as the cut-off that best predicts an adverse evolution by chromosome banding analysis (CBA) and genomic microarrays (GM). However, data comparing risk stratification by both methods are scarce. Herein, we assessed a cohort of 340 untreated CLL patients highly enriched in cases with complex karyotype (CK) (46.5%) with parallel CBA and GM studies. Abnormalities found by both techniques were compared. Prognostic stratification in three risk groups based on genomic complexity (0-2, 3- 4 and ≥5 abnormalities) was also analyzed. No significant differences in the percentage of patients in each group were detected, but only a moderate agreement was observed between methods when focusing on individual cases (κ=0.507; P<0.001). Discordant classification was obtained in 100 patients (29.4%), including 3% classified in opposite risk groups. Most discrepancies were technique-dependent and no greater correlation in the number of abnormalities was achieved when different filtering strategies were applied for GM. Nonetheless, both methods showed a similar concordance index for prediction of time to first treatment (TTFT) (CBA: 0.67 vs. GM: 0.65) and overall survival (CBA: 0.55 vs. GM: 0.57). High complexity maintained its significance in the multivariate analysis for TTFT including TP53 and IGHV status when defined by CBA (hazard ratio [HR] 3.23; P<0.001) and GM (HR 2.74; P<0.001). Our findings suggest that both methods are useful but not equivalent for risk stratification of CLL patients. Validation studies are needed to establish the prognostic value of genome complexity based on GM data in future prospective studies.

Clonal evolution in chronic lymphocytic leukemia is associated with an unmutated IGHV status and frequently leads to a combination of loss of TP53 and TP53 mutation.

BACKGROUND: Chromosomal abnormalities and gene mutations determine the prognosis of patients with chronic lymphocytic leukemia (CLL). Genetic lesions can be acquired by clonal evolution (CE) likely correlating with clinical progression. METHODS AND RESULTS: Samples of 169 CLL patients were analyzed for cytogenetic clonal evolution (CCE) and CE affecting the genes TP53 and SF3B1. Moreover, the mutational status of IGHV and the clinical outcome was evaluated. CCE was observed in 35% of CLL patients. The most frequently gained cytogenetic aberration was a deletion of TP53. Acquired TP53 deletion was more frequent in patients with SF3B1 mutations compared to those without (19% vs. 7%). CCE showed a tendency to occur more frequently in patients with an aberrant karyotype at first investigation than in patients with a normal karyotype. In 73% of patients with CCE (p = 0.002) and 92% of patients with CE affecting the genes TP53 and SF3B1 (p < 0.001) an unmutated IGHV status was present. CCE and CE affecting the genes TP53 and SF3B1 were significantly associated with each other (p < 0.001). In 7% of patients, CE resulted in the co-occurrence of TP53 deletion and TP53 mutation resulting in a significantly shorter overall survival. CONCLUSIONS: The most frequently gained cytogenetic aberration during CCE was a deletion of TP53, which was associated with SF3B1 mutations. Moreover, CCE was associated with an unmutated IGHV status. Our results indicate the importance of re-evaluation of the TP53 status during the course of the disease to ensure correct treatment guidance.

Consensus criteria for diagnosis, staging, and treatment response assessment of T-cell prolymphocytic leukemia.

T-cell prolymphocytic leukemia (T-PLL) is a rare, mature T-cell neoplasm with a heterogeneous clinical course. With the advent of novel treatment options that will potentially change the management of patients with T-PLL, it has become necessary to produce consensus guidelines for the design and conduct of clinical trials. The T-PLL International Study group (TPLL-ISG) set out to define standardized criteria for diagnosis, treatment indication, and evaluation of response. These criteria will facilitate comparison of results from clinical trials in T-PLL, and will thus support clinical decision making, as well as the approval of new therapeutics by healthcare authorities.

Cytogenetic complexity in chronic lymphocytic leukemia: definitions, associations, and clinical impact.

Recent evidence suggests that complex karyotype (CK) defined by the presence of ≥3 chromosomal aberrations (structural and/or numerical) identified by using chromosome-banding analysis (CBA) may be relevant for treatment decision-making in chronic lymphocytic leukemia (CLL). However, many challenges toward the routine clinical application of CBA remain. In a retrospective study of 5290 patients with available CBA data, we explored both clinicobiological associations and the clinical impact of CK in CLL. We found that patients with ≥5 abnormalities, defined as high-CK, exhibit uniformly dismal clinical outcomes, independently of clinical stage, TP53 aberrations (deletion of chromosome 17p and/or TP53 mutations [TP53abs]), and the expression of somatically hypermutated (M-CLL) or unmutated immunoglobulin heavy variable genes. Thus, they contrasted with CK cases with 3 or 4 aberrations (low-CK and intermediate-CK, respectively) who followed aggressive disease courses only in the presence of TP53abs. At the other end of the spectrum, patients with CK and +12,+19 displayed an exceptionally indolent profile. Building upon CK, TP53abs, and immunoglobulin heavy variable gene somatic hypermutation status, we propose a novel hierarchical model in which patients with high-CK exhibit the worst prognosis, whereas those with mutated CLL lacking CK or TP53abs, as well as CK with +12,+19, show the longest overall survival. Thus, CK should not be axiomatically considered unfavorable in CLL, representing a heterogeneous group with variable clinical behavior. High-CK with ≥5 chromosomal aberrations emerges as prognostically adverse, independent of other biomarkers. Prospective clinical validation is warranted before ultimately incorporating high-CK in risk stratification of CLL.

Genomic arrays identify high-risk chronic lymphocytic leukemia with genomic complexity: a multi-center study.

Complex karyotype (CK) identified by chromosome-banding analysis (CBA) has shown prognostic value in chronic lymphocytic leukemia (CLL). Genomic arrays offer high-resolution genome-wide detection of copy-number alterations (CNAs) and could therefore be well equipped to detect the presence of a CK. Current knowledge on genomic arrays in CLL is based on outcomes of single center studies, in which different cutoffs for CNA calling were used. To further determine the clinical utility of genomic arrays for CNA assessment in CLL diagnostics, we retrospectively analyzed 2293 arrays from 13 diagnostic laboratories according to established standards. CNAs were found outside regions captured by CLL FISH probes in 34% of patients, and several of them including gains of 8q, deletions of 9p and 18p (p<0.01) were linked to poor outcome after correction for multiple testing. Patients (n=972) could be divided in three distinct prognostic subgroups based on the number of CNAs. Only high genomic complexity (high-GC), defined as ≥5 CNAs emerged as an independent adverse prognosticator on multivariable analysis for time to first treatment (Hazard ratio: 2.15, 95% CI: 1.36-3.41; p=0.001) and overall survival (Hazard ratio: 2.54, 95% CI: 1.54-4.17; p<0.001; n=528). Lowering the size cutoff to 1 Mb in 647 patients did not significantly improve risk assessment. Genomic arrays detected more chromosomal abnormalities and performed at least as well in terms of risk stratification compared to simultaneous chromosome banding analysis as determined in 122 patients. Our findings highlight genomic array as an accurate tool for CLL risk stratification.

Clinical application of whole transcriptome sequencing for the classification of patients with acute lymphoblastic leukemia.

BACKGROUND: Considering the clinical and genetic characteristics, acute lymphoblastic leukemia (ALL) is a rather heterogeneous hematological neoplasm for which current standard diagnostics require various analyses encompassing morphology, immunophenotyping, cytogenetics, and molecular analysis of gene fusions and mutations. Hence, it would be desirable to rely on a technique and an analytical workflow that allows the simultaneous analysis and identification of all the genetic alterations in a single approach. Moreover, based on the results with standard methods, a significant amount of patients have no established abnormalities and hence, cannot further be stratified. METHODS: We performed WTS and WGS in 279 acute lymphoblastic leukemia (ALL) patients (B-cell: n = 211; T-cell: n = 68) to assess the accuracy of WTS, to detect relevant genetic markers, and to classify ALL patients. RESULTS: DNA and RNA-based genotyping was used to ensure correct WTS-WGS pairing. Gene expression analysis reliably assigned samples to the B Cell Precursor (BCP)-ALL or the T-ALL group. Subclassification of BCP-ALL samples was done progressively, assessing first the presence of chromosomal rearrangements by the means of fusion detection. Compared to the standard methods, 97% of the recurrent risk-stratifying fusions could be identified by WTS, assigning 76 samples to their respective entities. Additionally, read-through fusions (indicative of CDKN2A and RB1 gene deletions) were recurrently detected in the cohort along with 57 putative novel fusions, with yet untouched diagnostic potentials. Next, copy number variations were inferred from WTS data to identify relevant ploidy groups, classifying an additional of 31 samples. Lastly, gene expression profiling detected a BCR-ABL1-like signature in 27% of the remaining samples. CONCLUSION: As a single assay, WTS allowed a precise genetic classification for the majority of BCP-ALL patients, and is superior to conventional methods in the cases which lack entity defining genetic abnormalities.

Hematologist-Level Classification of Mature B-Cell Neoplasm Using Deep Learning on Multiparameter Flow Cytometry Data.

The wealth of information captured by multiparameter flow cytometry (MFC) can be analyzed by recent methods of computer vision when represented as a single image file. We therefore transformed MFC raw data into a multicolor 2D image by a self-organizing map and classified this representation using a convolutional neural network. By this means, we built an artificial intelligence that is not only able to distinguish diseased from healthy samples, but it can also differentiate seven subtypes of mature B-cell neoplasm. We trained our model with 18,274 cases including chronic lymphocytic leukemia and its precursor monoclonal B-cell lymphocytosis, marginal zone lymphoma, mantle cell lymphoma, prolymphocytic leukemia, follicular lymphoma, hairy cell leukemia, lymphoplasmacytic lymphoma and achieved a weighted F1 score of 0.94 on a separate test set of 2,348 cases. Furthermore, we estimated the trustworthiness of a classification and could classify 70% of all cases with a confidence of 0.95 and higher. Our performance analyses indicate that particularly for rare subtypes further improvement can be expected when even more samples are available for training. © 2020 The Authors. Cytometry Part A published by Wiley Periodicals LLC. on behalf of International Society for Advancement of Cytometry.

Dark-matter matters: Discriminating subtle blood cancers using the darkest DNA.

The confluence of deep sequencing and powerful machine learning is providing an unprecedented peek at the darkest of the dark genomic matter, the non-coding genomic regions lacking any functional annotation. While deep sequencing uncovers rare tumor variants, the heterogeneity of the disease confounds the best of machine learning (ML) algorithms. Here we set out to answer if the dark-matter of the genome encompass signals that can distinguish the fine subtypes of disease that are otherwise genomically indistinguishable. We introduce a novel stochastic regularization, ReVeaL, that empowers ML to discriminate subtle cancer subtypes even from the same 'cell of origin'. Analogous to heritability, implicitly defined on whole genome, we use predictability (F1 score) definable on portions of the genome. In an effort to distinguish cancer subtypes using dark-matter DNA, we applied ReVeaL to a new WGS dataset from 727 patient samples with seven forms of hematological cancers and assessed the predictivity over several genomic regions including genic, non-dark, non-coding, non-genic, and dark. ReVeaL enabled improved discrimination of cancer subtypes for all segments of the genome. The non-genic, non-coding and dark-matter had the highest F1 scores, with dark-matter having the highest level of predictability. Based on ReVeaL's predictability of different genomic regions, dark-matter contains enough signal to significantly discriminate fine subtypes of disease. Hence, the agglomeration of rare variants, even in the hitherto unannotated and ill-understood regions of the genome, may play a substantial role in the disease etiology and deserve much more attention.