Combining imaging flow cytometry and machine learning for high-throughput schistocyte quantification: A SVM classifier development and external validation cohort.

BACKGROUND: Schistocyte counts are a cornerstone of the diagnosis of thrombotic microangiopathy syndrome (TMA). Their manual quantification is complex and alternative automated methods suffer from pitfalls that limit their use. We report a method combining imaging flow cytometry (IFC) and artificial intelligence for the direct label-free and operator-independent quantification of schistocytes in whole blood. METHODS: We used 135,045 IFC images from blood acquisition among 14 patients to extract 188 features with IDEAS® software and 128 features from a convolutional neural network (CNN) with Keras framework in order to train a support vector machine (SVM) blood elements' classifier used for schistocytes quantification. FINDING: Keras features showed better accuracy (94.03%, CI: 93.75-94.31%) than ideas features (91.54%, CI: 91.21-91.87%) in recognising whole-blood elements, and together they showed the best accuracy (95.64%, CI: 95.39-95.88%). We obtained an excellent correlation (0.93, CI: 0.90-0.96) between three haematologists and our method on a cohort of 102 patient samples. All patients with schistocytosis (>1% schistocytes) were detected with excellent specificity (91.3%, CI: 82.0-96.7%) and sensitivity (100%, CI: 89.4-100.0%). We confirmed these results with a similar specificity (91.1%, CI: 78.8-97.5%) and sensitivity (100%, CI: 88.1-100.0%) on a validation cohort (n=74) analysed in an independent healthcare centre. Simultaneous analysis of 16 samples in both study centres showed a very good correlation between the 2 imaging flow cytometers (Y=1.001x). INTERPRETATION: We demonstrate that IFC can represent a reliable tool for operator-independent schistocyte quantification with no pre-analytical processing which is of most importance in emergency situations such as TMA. FUNDING: None.

Accurate classification of plasma cell dyscrasias is achieved by combining artificial intelligence and flow cytometry.

Monoclonal gammopathy of unknown significance (MGUS), smouldering multiple myeloma (SMM), and multiple myeloma (MM) are very common neoplasms. However, it is often difficult to distinguish between these entities. In the present study, we aimed to classify the most powerful markers that could improve diagnosis by multiparametric flow cytometry (MFC). The present study included 348 patients based on two independent cohorts. We first assessed how representative the data were in the discovery cohort (123 MM, 97 MGUS) and then analysed their respective plasma cell (PC) phenotype in order to obtain a set of correlations with a hypersphere visualisation. Cluster of differentiation (CD)27 and CD38 were differentially expressed in MGUS and MM (P < 0·001). We found by a gradient boosting machine method that the percentage of abnormal PCs and the ratio PC/CD117 positive precursors were the most influential parameters at diagnosis to distinguish MGUS and MM. Finally, we designed a decisional algorithm allowing a predictive classification ≥95% when PC dyscrasias were suspected, without any misclassification between MGUS and SMM. We validated this algorithm in an independent cohort of PC dyscrasias (n = 87 MM, n = 41 MGUS). This artificial intelligence model is freely available online as a diagnostic tool application website for all MFC centers worldwide (https://aihematology.shinyapps.io/PCdyscrasiasToolDg/).

Transcriptomic and genomic heterogeneity in blastic plasmacytoid dendritic cell neoplasms: from ontogeny to oncogenesis.

Oncogenesis and ontogeny of blastic plasmacytoid dendritic cell neoplasm (BPDCN) remain uncertain, between canonical plasmacytoid dendritic cells (pDCs) and AXL+ SIGLEC6+ DCs (AS-DCs). We compared 12 BPDCN to 164 acute leukemia by Affymetrix HG-U133 Plus 2.0 arrays: BPDCN were closer to B-cell acute lymphoblastic leukemia (ALL), with enrichment in pDC, B-cell signatures, vesicular transport, deubiquitination pathways, and AS-DC signatures, but only in some cases. Importantly, 1 T-cell ALL clustered with BPDCN, with compatible morphology, immunophenotype (cCD3+ sCD3- CD123+ cTCL1+ CD304+), and genetics. Many oncogenetic pathways are deregulated in BPDCN compared with normal pDC, such as cell-cycle kinases, and importantly, the transcription factor SOX4, involved in B ontogeny, pDC ontogeny, and cancer cell invasion. High-throughput sequencing (HaloPlex) showed myeloid mutations (TET2, 62%; ASXL1, 46%; ZRSR2, 31%) associated with lymphoid mutations (IKZF1), whereas single-nucleotide polymorphism (SNP) array (Affymetrix SNP array 6.0) revealed frequent losses (mean: 9 per patient) involving key hematological oncogenes (RB1, IKZF1/2/3, ETV6, NR3C1, CDKN2A/B, TP53) and immune response genes (IFNGR, TGFB, CLEC4C, IFNA cluster). Various markers suggest an AS-DC origin, but not in all patients, and some of these abnormalities are related to the leukemogenesis process, such as the 9p deletion, leading to decreased expression of genes encoding type I interferons. In addition, the AS-DC profile is only found in a subgroup of patients. Overall, the cellular ontogenic origin of BPDCN remains to be characterized, and these results highlight the heterogeneity of BPDCN, with a risk of a diagnostic trap.

Multicentric MFI30 study: Standardization of flow cytometry analysis of CD30 expression in non-Hodgkin lymphoma.

CD30 transmembrane receptor, a member of the tumor necrosis factor receptor family, is expressed in different lymphomas. Brentuximab vedotin (BV), a CD30 monoclonal antibody (Ab)-drug conjugate, is effective in CD30-positive lymphomas. However, the response to BV is not always correlated to CD30 expression detected by immunohistochemistry (IHC). The objectives of this study were to standardize and evaluate CD30 intensity by flow cytometry (FCM) in non-Hodgkin's lymphomas. Twelve centers analyzed 161 cases on standardized cytometers using normalized median fluorescence intensity (nMFI30) of three different Abs, of which one clone can recognize the same epitope as BV. FCM distinguished four groups of cases: negative group (n = 110) which showed no expression with the three clones; high positive group (n = 13) which gave nMFI30 > 5% with all tested clones; dim positive group (n = 17) which showed nMFI30 > 1% with all tested clones and <5% for at least one; discordant group (n = 21) with positive and negative expression of the different clones. In consistency with the literature, CD30 was positive in all anaplastic large cell lymphomas, in some diffuse large B-cell lymphomas (DLBCL), and in other rare lymphomas. FCM results were concordant with those of IHC in 77% of cases. Discrepancies could be explained by clones-related differences, microenvironment, or intracytoplasmic staining. Interestingly, FCM was more sensitive than IHC in 11% of cases, especially in DLBCL. Multicenter standardized FCM of specific CD30 could improve case detection and extend the treatment of BV to various CD30-positive lymphomas.

How should we diagnose and treat blastic plasmacytoid dendritic cell neoplasm patients?

Blastic plasmacytoid dendritic cell neoplasm (BPDCN) is a rare and aggressive leukemia for which we developed a nationwide network to collect data from new cases diagnosed in France. In a retrospective, observational study of 86 patients (2000-2013), we described clinical and biological data focusing on morphologies and immunophenotype. We found expression of markers associated with plasmacytoid dendritic cell origin (HLA-DRhigh, CD303+, CD304+, and cTCL1+) plus CD4 and CD56 and frequent expression of isolated markers from the myeloid, B-, and T-lymphoid lineages, whereas specific markers (myeloperoxidase, CD14, cCD3, CD19, and cCD22) were not expressed. Fifty-one percent of cytogenetic abnormalities impact chromosomes 13, 12, 9, and 15. Myelemia was associated with an adverse prognosis. We categorized chemotherapeutic regimens into 5 groups: acute myeloid leukemia (AML)-like, acute lymphoid leukemia (ALL)-like, lymphoma (cyclophosphamide, doxorubicin, vincristine, and prednisone [CHOP])-like, high-dose methotrexate with asparaginase (Aspa-MTX) chemotherapies, and not otherwise specified (NOS) treatments. Thirty patients received allogeneic hematopoietic cell transplantation (allo-HCT), and 4 patients received autologous hematopoietic cell transplantation. There was no difference in survival between patients receiving AML-like, ALL-like, or Aspa-MTX regimens; survival was longer in patients who received AML-like, ALL-like, or Aspa-MTX regimens than in those who received CHOP-like regimens or NOS. Eleven patients are in persistent complete remission after allo-HCT with a median survival of 49 months vs 8 for other patients. Our series confirms a high response rate with a lower toxicity profile with the Aspa-MTX regimen, offering the best chance of access to hematopoietic cell transplantation and a possible cure.

Cryopreservation and thawing of hematopoietic stem cell CD34-induced apoptosis through caspase pathway activation: Key role of granulocytes.

INTRODUCTION: Cell damage inescapably occurs during both the freezing and the thawing graft processes for autologous hematopoietic stem cell (HSC) transplantation. To estimate HSC injury, a quality control is performed including: (i) CD34+ quantification; (ii) percentage of CD34+ viability and (iii) evaluation of HSC functional ability to form colony forming unit-granulocyte macrophage (CFU-GM). Apoptosis involves complex pathways such as caspase enzymes. Here, we assess the extent of apoptosis that is caspase-dependent before and after cryoconservation of CD34+, using a Fluorescent Labeled Inhibitor of CAspases (FLICA). METHODS: Caspase pathway activation status was evaluated in 46 patients (multiple myeloma [n = 24], lymphoma [n = 22]), by flow cytometry, using a 7-aminoactinomycin-D (7AAD)/FLICA staining test, in CD34+, CD3+, CD14+ and CD56+ cells. Viable 7AAD-/FLICA+ cells were then correlated with various parameters. RESULTS: We showed a significant caspase pathway activation, with 23% CD34+/7AAD-/FLICA+ cells after thawing, compared with the 2% described in fresh CD34+ cells (P < 0.0001). Moreover, caspase pathway was significantly activated in thawing CD3+, CD56+ and CD14+ cells. We also report a significant correlation between the rate of CD34+/7AAD-/FLICA+ cells and post-thawing granulocytes count (P = 0.042) and their potential to be differentiated into CFU-GM (P = 0.004). DISCUSSION: Our results show substantial cell death, induced by the increase of caspase pathway activation, secondary to the thawing process, and across all study cell types. This observation may affect the immune response quality during recipient aplasia, without detecting a clinical impact. Moreover, caspase pathway activation through CD3+ and CD56+ subpopulations could modify the therapeutic result of donor lymphocytes infusion (DLI).

Multicenter validation of the flow measurement of classical monocyte fraction for chronic myelomonocytic leukemia diagnosis.

Peripheral blood monocytes include three subsets defined by CD14 and CD16 surface markers. An increase in the CD14++CD16- classical monocyte fraction ≥ 94% of the total monocytes was proposed to rapidly and efficiently distinguish chronic myelomonocytic leukemia from reactive monocytosis. The robustness of this assay required a multicenter validation. The flow cytometry assay designed to quantify peripheral blood monocyte subsets was implemented by multiple diagnosis laboratories in France. A nationwide survey was performed to evaluate its performance. All the 48 French laboratories answered the questionnaire, revealing that 63% use this assay routinely. Central blind reanalysis of 329 cytometry files collected from five laboratories demonstrated an excellent correlation in classical monocyte fraction measurement (r = 0.93; p < 0.0001). The cutoff value of 94% classical monocytes being the critical readout for diagnosis, we then compared 115 patients with classical monocytes ≥ 94% and 214 patients with a fraction < 94% between initial analysis and reanalysis. An agreement was obtained in 311 files. Finally, an overt diagnosis, available for 86 files, confirmed a good sensitivity (93.6%) and specificity (89.7%). This survey demonstrates the robustness of the flow assay with limited variability of classical monocyte percentage between centers, validates the 94% cutoff value, and confirms its sensitivity and specificity.

The Importance of an In-depth Study of Immunoglobulin Gene Rearrangements When Ascertaining the Clonal Relationship between Concomitant Chronic Lymphocytic Leukemia and Multiple Myeloma.

Chronic lymphocytic leukemia (CLL) and multiple myeloma (MM) are hematological disorders that occur at different stages of B-cell development. It has been shown that CLL B-cells can differentiate into plasma cells in vitro and in vivo. CLL is the most frequent adult leukemia in the western world. It is a heterogeneous disease, characterized by clonal proliferation and the accumulation of mature CD5+ B lymphocytes (1). MM is a clonal plasma cell malignancy that accounts for more than 10% of all hematologic cancers (2). Although secondary cancers [particularly solid tumors (3-5)] can occur with CLL and MM, the concomitant occurrence of these two disorders in the same patient is rare [for a review of the few reported cases, see Ref. (6)]. The clonal relationship between these diseases has not always been clarified but is important in terms of understanding the pathogenesis and optimizing treatment. The clonal relationship between CLL and MM can be evaluated by (i) analyzing immunoglobulin (Ig) heavy chain and light chain (Ig kappa light chain and Ig lambda light chain) gene rearrangement, (ii) identifying and comparing somatic mutations, and (iii) studying chromosomic aberrations. Nevertheless, Ig rearrangements must always be interpreted in the light of specific phenomena such as allelic exclusion, B-cell receptor (BCR) revision (VH and DH gene replacement), BCR editing, and somatic mutations-events that were not considered in previous studies. These issues can be addressed by sequencing the rearranged Ig genes from sorted populations and interpreting the generated data. In the present study, we evaluated the putative clonal relationship between the two diseases by combining DNA copy number analysis with an assessment of Ig gene rearrangements [clonality assessment, V(D)J sequencing, and somatic hypermutation analysis] in highly enriched CD19+ CD5+ (CLL) and CD38+ CD138+ (MM) cell populations. Array comparative genomic hybridization data suggested a possible phylogenic progression from CLL to MM. Moreover, V(D)J sequencing indicated that both CLL and MM cells used the same VH and JH genes but different DH genes. However, in-depth analysis and interpretation of Ig gene rearrangements ultimately suggested that the two diseases had distinct clonal origins.

Hyperferritinemia at diagnosis predicts relapse and overall survival in younger AML patients with intermediate-risk cytogenetics.

The prognostic value of ferritin level at diagnosis in AML patients is unknown. We studied 162 younger AML patients with intermediate-risk cytogenetics who received intensive chemotherapy. The median ferritin level at diagnosis was 633 µg/L and 128 (79%) patients had a ferritin level above the upper normal limit. Hyperferritinemia was significantly associated with a higher cumulative incidence of relapse as well as poorer disease-free and overall survival. In multivariate analysis, hyperferritinemia remained an independent poor prognosis factor. The level of ferritin at diagnosis has a major impact on relapse suggesting a link between inflammation, oxidative stress and chemoresistance in AML.

5-Azacytidine treatment for relapsed or refractory acute myeloid leukemia after intensive chemotherapy.

Despite progress in the understanding of leukemia pathophysiology, the treatment of acute myeloid leukemia (AML) remains challenging. In patients with refractory or relapsed (R/R) AML, the prognosis is still poor and this group is targeted for new drug development. We reviewed the outcome of 47 patients, with R/R AML after at least one course of intensive chemotherapy, treated with 5-azacytidine in three different French institutions. The overall response rate was 38% including complete remission in 21%, partial remission in 11%, and hematological improvement in 6% of cases. Median time to relapse was 6 (range, 1-39) months. Median overall survival was 9 months (not reached by responders vs. 4.5 months for nonresponders patients, P = 0.0001). Univariate analysis identified the absence of peripheral blood blasts and <20% bone marrow blasts as prognostic factors for both overall response and survival, but not age, ECOG/PS, type of AML, cytogenetic, status of the disease, number of previous lines of therapy, previous hematological stem cell transplantation, or white blood cells count. Bone marrow blasts percentage <20% was the only independent prognostic factor identified by multivariate analysis for overall response (P = 0.0013) and survival (P = 0.0324). Six patients in remission could proceed to an allogenic hematological stem cell transplantation. The drug-related grade 3/4 adverse events were hematopoietic toxicities (38%) and infection (32%). In conclusion, this study suggests that a salvage therapy with 5-azacytidine is an interesting option for patients with R/R AML after intensive chemotherapy. Prospective randomized studies are needed to demonstrate a superiority of this approach over others strategies.

Intracytoplasmic detection of TCL1--but not ILT7-by flow cytometry is useful for blastic plasmacytoid dendritic cell leukemia diagnosis.

Diagnosis of blastic plasmacytoid dendritic cell neoplasm (BPDCN) or plasmacytoid dendritic cell leukemia (pDCL) is mainly based on immunophenotypical characterization of leukemic cells in blood or bone marrow samples. We tested by flow cytometry intracellular expression of the proto-oncogene T-cell leukemia 1 (TCL1), as well as membrane and intracellular expression of immunoglobulin-like transcript 7 (ILT7) in 21 pDCL samples and 61 non-pDC acute leukemia samples [i.e., 14 B-acute lymphoblastic leukemia (B-ALL), 9 T-ALL and 38 acute myeloid leukemia (AML)]. TCL1 is highly expressed in all pDCL samples while at a statistically lower level in all B-ALL and 34% of AML. Statistical analysis shows that intensity of TCL1 expression is a good marker for differential diagnosis of pDCL versus other acute leukemia (area under the receiver-operating characteristic curve, [AUC]: 0.96). By contrast, ILT7 positivity is limited to few pDCL samples and cannot be useful for diagnosis purpose. In conclusion, high intracellular intensity of TCL1 expression is currently the best marker for pDC lineage assignment by flow cytometry, which is particularly useful to distinguish pDCL from CD4(+) CD56(+/-) undifferentiated or monoblastic acute leukemia. Thus, intracellular TCL1 detection should be included in acute leukemia diagnosis panels used in hematology laboratories. © 2012 International Society for Advancement of Cytometry.

Concordance of assays designed for the quantification of JAK2V617F: a multicenter study.

BACKGROUND: Many different techniques have been designed for the quantification of JAK2V617F allelic burden, sometimes producing discrepant results. DESIGN AND METHODS: JAK2V617F quantification techniques were compared among 16 centers using 11 assays based on quantitative polymerase chain reaction (with mutation-specific primers or probes, or fluorescent resonance energy transfer/melting curve analysis), allele-specific polymerase chain reaction, conventional sequencing or pyrosequencing. RESULTS: A first series of blinded samples (granulocyte DNA, n=29) was analyzed. Seven assays (12 centers) reported values inside the mean +/- 2SD; the mean coefficient of variation was 31%. Sequencing techniques lacked sensitivity, and strong discrepancies were observed with four techniques, which could be attributed to inadequate standards or to different modes of expression of results. Indeed, quantification of JAK2V617F in relation to another control gene produced higher than expected values, suggesting the possibility of more than two JAK2 copies/cell. After calibration of assays with common 1% to 100% JAK2V617F standards (dilutions of UKE-1 cells in normal leukocytes), 14 centers tested ten new samples. JAK2V617F allelic burdens greater or equal than 1% were then reliably quantified by five techniques -- one allele specific-polymerase chain reaction and four TaqMan allele-specific quantitative polymerase chain reaction assays, including one previously giving results outside the mean +/- 2SD -- with a lower mean coefficient of variation (21%). Of these, only the two TaqMan allele-specific quantitative polymerase chain reaction assays with primer-based specificity could detect 0.2% JAK2V617F. CONCLUSIONS: Techniques expressing the allelic burden as JAK2V617F/total JAK2 and using a common set of standards produced similar quantification results but with variable sensitivity. Calibration to a reference standard improved reproducibility.