Loss of hematopoietic progenitors heterogeneity is an adverse prognostic factor in lower-risk myelodysplastic neoplasms

Myelodysplastic neoplasms (MDS) are characterized by clonal evolution starting from the compartment of hematopoietic stem and progenitors cells (HSPCs), leading in some cases to leukemic transformation. We hypothesized that deciphering the diversity of the HSPCs compartment may allow for the early detection of an emergent sub-clone that drives disease progression. Deep analysis of HSPCs repartition by multiparametric flow cytometry revealed a strong disorder of the hematopoietic branching system in most patients at diagnosis with different phenotypic signatures closely related to specific MDS features. In two independent cohorts of 131 and 584 MDS, the HSPCs heterogeneity quantified through entropy calculation was decreased in 47% and 46% of cases, reflecting a more advanced state of the disease with deeper cytopenias, higher IPSS-R risk and accumulation of somatic mutations. We demonstrated that patients with lower-risk MDS and low CD34 + CD38+HSPCs entropy had an adverse outcome and that this parameter is as an independent predictive biomarker for progression free survival, leukemia free survival and overall survival. Analysis of HSPCs repartition at diagnosis represents therefore a very powerful tool to identify lower-risk MDS patients with a worse outcome and valuable for clinical decision-making, which could be fully integrated in the MDS diagnostic workflow.


Patients and specimens
BM control samples of cohort #1 (n=9) were from healthy elderly patients (median age of 68 years; min-max: 45-83) obtained from the bone of the femoral head obtained after informed consent, during a surgery for hip replacement.This non-interventional study was approved by the ethical committee of Cochin-Port Royal Hospital (CLEP Decision N°: AAA-2020-08039).In the cohort #2 and #3, non-MDS samples without cytopenia (cohort #2: n=24; median age of 67 years, min-max: 28-91 and cohort #3: n=54; median age of 61 years, min-max: 32-91) were from patients referred to our institutions for BM aspiration in a context of suspicion of different hematological neoplasms (mastocytosis, lymphoma or myeloma) without excess of abnormal cells.Non-MDS samples with cytopenia (cohort #2: n=20, median age of 70 years, min-max: 21-90 and cohort #3; n=126 median age of 70 years, min-max: 21-90) were from patients for who a diagnosis of MDS was excluded based on morphological analysis and/or cytogenetic data and/or to identification of non-MDS related cause.

MFC analysis
MFC analysis for samples of cohort #1 (Supplemental Table 1) using combination #1 of antibodies (Supplemental Table 2) were performed on BM mononuclear cells (BMMNCs) from sternal aspirations or from femoral heads previously isolated on a Ficoll gradient and stored in liquid nitrogen at a concentration of 20 to 30 million cells per mL.The femoral heads were cut in half and collected in a conservation medium (Hanks balanced salt solution with NaHCO3, Eurobio™), supplemented with heparin (7%) and then transported to the laboratory at room temperature.These were scraped with a spatula, grounded in a mortar, and washed with a PBS solution supplemented with DNAse (Sigma Aldrich) at 100ug/mL.BMMNCs were thawed in Iscove's modification of Dulbecco medium supplemented with DNAse (1ug/mL) to avoid the formation of aggregates and limit cell loss.BMMNCs were washed with PBS and sorted on magnetic columns (Miltenyi Biotec™ MicroBead MACS technology) to recover CD34 positive cells.CD34 positive cells were incubated for 20 min in Brilliant Stain Buffer, BD Horizon™ (100µL per tube) with the combination of antibodies shown in supplemental Table 2. Cells were washed with PBS (Dulbecco's PBS™ buffer) and a second labeling was performed with Zoombie acqua Amcyan (BioLegend™) to distinguish dead and live cells.Data were acquired using a LSR Fortessa™ flow cytometer (BD Biosciences).MFC analysis for samples from the cohort #2 and #3 (Supplemental Table 1) with combination #2 and #3 of antibodies (Supplemental Table 2) were performed on fresh BM specimens.BM purity was determined by the method of Holdrinet et al 1 and diluted samples (purity<50%) were excluded.Staining was performed within 24h following aspiration on 2 million BM cells after one wash of whole BM in Dulbecco's phosphate buffered-saline (PBS, Eurobio).
Cells were stained with antibodies for 20mn.Red blood cells were lysed with 1 mL Versalyse TM (Beckman Coulter) and the samples washed once in PBS.Cells were then resuspended in 500µL of PBS and data were acquired using a Navios flow cytometer (Beckman Coulter) with 10 colors, 3 lasers (5+3+2 configuration).Sensitivity of the flow cytometer was daily controlled using Flow-Set Pro beads (Beckman Coulter).Data were analyzed using Kaluza 2.1 version software (Beckman Coulter) using the gating strategy described in Supplemental Figure 1, Supplemental Figure 3 and Supplemental Figure 7.
To ensure a robust calculation of entropy in both CD34+CD38+ and CD34+CD38-compartment of cells, samples with less of 500 cells in the fraction of total CD34+ HSPCs and/or less of 100 cells in one of the two fraction of CD34+CD38-or CD34+CD38+ HSPCs were excluded.The median number of CD34+ HSPCs analyzed was of 6045 (min-max: 1003-192573) and of 12216 (min-max: 845-221903) for samples collected in cohort #2 and cohort #3 respectively.

Genomic testing
We reported all clinically relevant variants with a VAF cut-off at 2%.Libraries were prepared using Ampliseq System, according to the manufacturer's instructions, and sequenced on an ion torrent S5XL (Thermo Fisher Scientific, Inc., Waltham, MA, USA).Average coverage per gene was ≥ 600X.
Reads were aligned against human genome build 19 (hg19) and analyzed for single nucleotid variant (SNV) calling with NextGENe software (SoftGenetics, Chicago, IL) and with an in-house pipeline (Polydiag, Institut Imagine, Université de Paris).IPSS-M was calculated as previously described with missing data for 2 main effects genes (NPM1 and MLL PTD ) and for 4 residuals genes (CEBPA, GNB1, NF1 and PRPF8). 2

SUPPLEMENTAL FIGURES
Supplemental Figure 1 Dussiau C et al

Supplemental Figure 4 :Supplemental Figure 5 :C
analysis of the non-MDS BM sample #72.Examples of abnormal patterns of HSPCs repartition detected in MDS samples compared to non-MDS samples and correlation with MDS characteristics.BM samples are from cohort #2.(A) Upper left panel: increase of MLP (patient #122, IPSS-R high) and of LMPP (patient #218, IPSS-R low).Upper right panel: increase of MPP F2 (patient #204, IPSS-R high) and of MPP F3 (patient #109, IPSS-R very high).Down left panel: increase of CMPMEP (patient #196, IPSS-R low) or GMP (patient #296, IPSS-R intermediate).Down right panel: decrease of CMPMEP F3 (patient #218, IPSS-R low) or increase of CMPMEP F2 (patient #204, IPSS-R high).(B) Examples of abnormal HSPCs sub-populations detected in MDS samples but not in of non-MDS samples without cytopenia such as CD34+ CD38-CD45Ra-CD10+ cells (patients #172, IPSS-R low and #320, IPSS-R intermediate); MPP F4 and CMPMEP F4 (patient #110, IPSS-R very high).(C) Proportion of MDS cases with increased of each HSPCs sub-populations determined according the calculation of z-score >2 compared to the cohort of non-MDS samples without cytopenia.(D) BM Blast cells percentage according the pattern of HSPCs repartition.Lines represent medians.(E) Number of mutations according the pattern of HSPCs repartition.Lines represent medians.Detection of MDS patients with abnormal level of CD38+ or CD38-HSPCs entropy.BM samples are from cohort #2.(A) Repartition of patients (MDS, n=131 and non-MDS with al cytopenia, n=20) according their calculated z-score compared to the cohort of non-MDS samples without cytopenia for CD38+ (left panel) and CD38-(right panel) HSPCs entropy.(B) ROC curve for the CD38-HSPCs entropy in the whole cohort of patients with MDS (black curve, n=131), or MDS low risk (very low/low IPSS-R; orange curve; n=73) or MDS intermediate risk (blue curve; n=17) or MDS high risk (high/very high IPSS-R; red curve; n=33) compared to non-MDS samples with or without cytopenia (n=44).(C) Correlation between level of CD38+HSPCs entropy and blast percentage detected by morphology (left panel) or quantification of CD34+ cells by MFC (right panel).Correlation was determined using non-parametric Spearman correlation test.Supplemental Figure 6: Genomic landscape of MDS patients stratified according their level of CD38+ HSPCs entropy.BM samples are from cohort #2.(A) Mutational landscape of the 102 MDS samples according the level of the CD38+HSPCs entropy (normal or low).Abnormal level of CD38+HSPCs entropy was determined according the calculation of z-score ≤-2 compared to the cohort of non-MDS samples without cytopenia.Patients are shown according their CD38+HSPCs entropy value.Mutational VAF is coded; ice, 2-10%; sky, 10-20%; orchid, 20-30%; blueberry, >30%.If multiple mutations are present in a case, the highest VAF is shown.(B) Prevalence of somatic mutations in mains groups of genes in BM from MDS patients with normal (n=55) or low (n=47) level of CD38+HSPCs entropy.Low level of CD38+HSPCs entropy was determined according the calculation of z-score ≤-2 compared to the cohort of non-MDS samples without cytopenia.Statistical significance was calculated using Fisher's exact tests.EM, Epigenetic Modifiers; SF, Splicing Factors; SIG, Signaling; TF, Transcription Factors; TS, Tumor Suppressors.(C) Prevalence of somatic mutations in oncogenes and leukemia-relevant genes in BM from MDS patients with normal (n=55) or low (n=47) CD38+HSPCs entropy.Genes mutated in ≥2 patients are shown.Low level of CD38+HSPCs entropy was determined according the calculation of z-score ≤-2 compared to the cohort of non-MDS samples without cytopenia.Statistical significance was calculated using Fisher's exact tests.(D) Levels of CD38+HSPCs entropy according the VAF (< or ≥30%) of somatic mutation detected in mutated MDS samples (n=85/102).In cases with multiple mutations, the highest VAF was used.Lines represent means ± s.d.Statistical significance was determined using unpaired two-tailed Mann-Whitney tests.one IPSS-R category, and colors represent the IPSS-M categories.(B) Proportion of restratified patients in each IPSS-R stratum (counting any shift).(C) Proportion of IPSS-M-restratified upstaged patients with either normal or low level of CD38+ HSPCs entropy for each group of IPSS-R.(D) Association between the level of CD38+ HSPCs entropy and restratification of patients with LR-MDS based on their IPSS-R £3.5 (n=64).doublets,white blood cells (WBC) leucocytes were gated according to the SSC/FSC plot and CD45 intensity in a CD45/SSC plot and Mononuclear cells (MNC) were selected on the SSC/FSC plot.Blasts cells were selected on a CD45/SSC plot.Then, CD34+ CD38-and CD34+ CD38+ populations were selected according to their expression of CD38, adjusted to a threshold defined by the lymphocytes, on a CD34/CD38 plot.Identification of HSPCs sub-populations: into the CD34+CD38-stem cell population, HSC/MPP (CD45Ra-CD135+), LMPP (CD45Ra+ CD135+) and EMP (CD45Ra-CD135-) populations were identified on a CD45Ra/CD135 plot.These populations were then separated into 2 different fractions of cells according their CD133 expression: MPP F1 (CD133+), MPP F2 (CD133-), LMPP F1 (CD133+), LMPP F2 (CD133-), EMP F1 (CD133+) and EMP F2 (CD133-).Into the CD34+CD38+ progenitor cell population, CMP (CD45Ra+ CD135-), GMP (CD45Ra+ CD135+) and MEP cells (CD45Ra-CD135-) were identified on a CD45Ra/CD135 plot.Finally, these population were separated into 2 different fractions of cells according their CD133 expression: CMP F1 (CD133+), CMP F2 (CD133-), GMP F1 (CD133+), GMP F2 (CD133-), MEP F1 (CD133+) and MEP F2 (CD133-).In conclusion, this gating strategy allowed the quantification of the CD34+ population into the leucocytes population and of 12 sub-populations.Figures were obtained from analysis of the MDS BM sample #140.Supplemental Figure 9: Analysis of CD38+HSPCs entropy in samples from cohort #3.(A) CD38+HSPCs levels in non-MDS samples with (n=126) or without (n=54) cytopenia and in MDS Mann-Whitney tests.(B) ROC curve for the CD38+HSPCs entropy in the whole cohort of patients with MDS (black curve, n=584), or MDS low risk (IPSS-R≤3.5;orange curve; n=318) or MDS high risk (IPSS-R>3.5;red curve; n= 266) compared to non-MDS samples with or without cytopenia (n=180).(C) Repartition of non-MDS samples without cytopenia and MDS samples according their level of CD38+HSPCs entropy.(D) Correlation between level of CD38+HSPCs entropy and blast percentage detected by morphology.Correlation was determined using non-parametric Spearman correlation test.(E) Overall survival and leukemia free survival of MDS patients stratified by their IPSS-R risk.according to CD38+ HSPCs entropy level.BM samples are from cohort #3.Patients with HR-MDS were defined based on their IPSS-R>3.5(n=371).

Table 5 : Quantification of the different sub-populations in non-MDS samples without cytopenia collected in cohort #2, among CD34+CD38-or CD34+CD38+HSPCs.
Thesevalues were used as references for normal frequency of each HSPCs sub-populations.