OMIP: Analysis of Human Myelopoiesis and Myeloid Neoplasms

Purpose and Appropriate Sample Types

Cells undergoing myelopoiesis in both bone marrow and the mature myeloid compartments in the blood represent a broad range of phenotypes in different stages of maturation and activation. Neoplasms that arise from this lineage are often heterogeneous and frequently display unique characteristics and outcomes depending on the originating subset or degree of differentiation(1, 2). This is particularly true for myeloid neoplasms (MN) which rely on multiple, separate flow cytometry panels for differential diagnosis(3, 4). While standard practice, this approach has inherent disadvantages which include the inability to observe co-expression patterns between markers contained in separate panels and increased time and labor associated with multiple flow cytometry assays. Use of a single, high-order panel that is able to differentiate multiple hematopoietic lineages, or distinguish different MNs would aid this effort and provide the opportunity to further characterize disease subsets. This panel was designed to detect cell surface markers associated with hematopoiesis with a special emphasis on the progression of myelopoiesis from hematopoietic progenitors. Because the study of hematopoietic disease requires a broad combination of early hematopoietic and mature myeloid markers(5), this panel is also appropriate for the general analysis of myelopoiesis, granulopoiesis, erythropoiesis, and megakaryocytopoiesis, in any human cell source that contains mature myeloid cells, myeloid progenitors, or hematopoietic progenitors (Table 1).

Table 1:

Summary Table for Application of this OMIP

Purpose	Subset analysis of hematopoiesis and hematopoietic disease
Species	Human
Cell Types	Any source containing human myeloid cells
Cross-References	No similar OMIPs

Background

Hematopoiesis is the process by which all types of blood cells arise from a common multipotent hematopoietic stem cell (HSC) identified by a lack of lineage-specific markers and by KIT (CD117) and CD34 co-expression(6, 7). While more immature subsets of HSC can be identified using CD133 and CD38(8, 9), lineage⁻CD117⁺CD34⁺ cells encompass the most mature group of progenitor phenotypes that retain multipotent potential, and further differentiation of these cells proceeds down one of five hematopoietic lineages: erythropoiesis, lymphopoiesis, myelopoiesis, granulopoiesis, or megakaryocytopoiesis(10) (Figure 1A). Once committed, progenitors of each lineage express cell surface markers that can be used to identify their increasingly restricted step-wise progression into mature phenotypes. Of these lineages, myelopoiesis is arguably the most complex regarding the diversity of fully-differentiated phenotypes it can give rise to, and the numerous branch-points and subsets therein(11). Thus, flow cytometry analysis is a mainstay in the study of hematopoiesis, and in particular, myelopoiesis.

Figure. 1:

Analysis of bone marrow mononuclear cells using this OMIP. (A) An example gating schema for healthy bone marrow mononuclear cells. Single cells were gated using forward scatter height versus width parameters followed by side scatter height versus width parameters. Dead cells along with CD3+ and CD19+ cells were excluded from analysis prior to subsequent gating schemes. Gates were placed according to fluorescenceminus‐one controls or higher. Red arrows = gated cells. Roman numeral markers correspond to the schematic in part B. (B) A schematic overview of the major hematopoietic lineages analyzed in this panel. Roman numerals indicate discrete phenotypes. (C) Histograms depicting CD15, CD32, CD36, and CD163 staining on healthy bone marrow mononuclear cells (black), and two patients with CMML (red = patient 1; blue = patient 2). Cells in part C were first gated down to FSC versus SSC characteristics as in part A.

Cancers that arise from the myeloid compartment are like-wise complex, retaining many of the surface markers reflective of their originating cell types, as well as acquiring aberrant expression of other myeloid markers or lineage types(1, 2). Immunophenotyping MNs is important for diagnosis, and in particular is critical to stratifying myeloid and lymphoid leukemia’s, which are managed distinctly. MNs, including the myelodysplastic syndromes (MDS), are a heterogeneous group of diseases with increased risk of transformation to acute myeloid leukemia (AML), and poor survival rates(2).

Assessment of MNs relies largely upon morphological assessment of bone marrow aspirate, and a series of lower-parameter flow cytometry panels run in parallel or succession (3, 4). Such panels were first outlined as a standardized set in 2008 by the International Workshop on Standardization of Flow Cytometry in MDS. An example screening would include ten four-color panels, each of which include CD45, seven of which would include the progenitor markers CD34 and/or CD117, five of which would include a combination of myeloid markers such as CD33, HLA-DR, CD13, CD36, CD64, and/or CD11b. The remaining parameters were occupied by lineage markers CD2, CD5, CD7, CD10, CD19, CD14, CD15, CD16, CD56, and/or CD71 such that each tube focuses on specific disease subsets (3). More recently, the EuroFlow Consortium has proposed a series of panels that would be run in a progressive manner beginning with one of several screening panels that is selected for based on clinical or laboratory indications. In regard to MNs, the results of this initial screening panel would orient each case toward one of three more comprehensive panel sets focusing on B cell precursors, acute lymphoblastic leukaemia, or AML/MDS disease subsets using markers associated with B cells, T cells, or myeloid cells respectively. Considering the AML/MDS panel set along with the initial screening panel, the EuroFlow Consortium recommendations encompass 36 markers for MNs cases (4).

The panel presented here, includes a framework that allows a broad overview of myelopoiesis from HSC through mature monocyte/macrophage populations, myeloid-derived suppressor cells (MDSC), and a broad spectrum of MDS/MPN disease subtypes(2, 12, 13-15) in a single 18-color panel (Table 2). The panel was constructed by first titrating the staining dilution of each antibody and calculating the optimal staining index (Supplemental Figures 1 and 2), and then in a step-wise manner building the panel based on category sets of parameters (Table 2), adding each category in turn (Supplemental Figure 3A-D). The panel includes as a utility anti-CD3 and anti-CD19 conjugated to brilliant violet® 510 (BV510), which can be excluded from analysis in the same detector as the viability marker Live/Dead™ Aqua (ThermoFisher Scientific), and anti-CD138 in order to gate out mature lymphocyte populations and plasma cells before analyzing hematopoietic populations. Similarly, erythropoietic and megakaryocytopoietic lineages can be separated from the myeloid lineages using CD71 and CD41a respectively, and committed progenitors therein can be identified using CD117, or a combination of CD45, CD34, and CD117. Within CD41a⁻CD71⁻ cells, uncommitted HSC can be identified by gating on CD45⁺ or CD45^low cells, followed by CD34⁺CD117⁺ cells, and then finally on CD33⁻HLA-DR⁻ cells. Conversely, multipotent progenitors (MPP) and common myeloid progenitors (CMP) can be distinguished from committed monoblasts within the corresponding CD33⁺HLA-DR⁺ gate using CD64 expression. Common lymphoid progenitors (CLP) can be identified as CD34⁺CD117⁻(11), although more mature lymphocyte populations are removed from analysis using CD3 and CD19 as part of a common “dump gate” along with the viability marker. More mature monocytic and granulocytic populations can be analyzed within the CD34⁻ gate. Here, mature monocytes can be identified by CD14 expression, whereas promonocytes can be identified as CD33⁺CD64⁺HLA-DR⁺ and CD11c⁺ after gating on CD117⁻ cells. The hierarchy of granulocyte maturation can be analyzed by first gating on CD33⁻CD64⁻ cells within the CD117⁻ gate, and then by increasing CD16 expression on CD11b^low cells. Finally, fully mature neutrophils can be distinguished from band cells using SSC and CD45 expression (Figure 1B, Supplemental Table 3). Importantly, the samples stained here were processed by density gradient centrifugation to obtain bone marrow mononuclear cells (BMMNC), as many patient bone marrow biopsies are done. While standard practice, this methodology excludes polymorphonuclear cells, such as mature neutrophils. Alternative sample processing, such as RBC lysis can retain these populations if desired (Supplemental Figure 4).

Table 2:

Reagents

Specificity	Clone	Fluorochrome	Purpose	Category
CD138	MI15	APC-Cy7	Plasma Cell	Utility
CD19	HIB19	BV 510	Dump	Utility
CD3	SK7	BV 510	Dump	Utility
CD45	HI30	BUV 805	Leukocyte	Utility
		Live/Dead Aqua	Viability	Utility
CD117	104D2	BB 515	Progenitor	Lineage
CD11b	ICRF44	BV 650	Myeloid Lineage	Lineage
CD11c	3.9	PE-CF594	Dendritic Cell	Lineage
CD14	M5E2	BV 786	Monocyte	Lineage
CD16	3G8	BUV 496	FC Receptor	Lineage
CD33	WM53	BV 421	Myeloid Lineage	Lineage
CD34	581	PE-Cy7	Progenitor	Lineage
CD36	CB38	PerCP-Cy5.5	M2	Mac. Subset
CD64	10.1	BUV 737	Activation; M1	Mac. Subset
HLA-DR	G46-6	BUV 395	M1	Mac. Subset
CD15	W6D3	AF 700	Myeloid Maturation	Disease Diff.
CD163	GHI/61	BV 605	Macrophage	Disease Diff.
CD32	FUN-2	APC	FC Receptor	Disease Diff.
CD41a	HIP8	PE	Megakaryocyte	Disease Diff.
CD71	M-A712	BV 711	Erythroid Marker	Disease Diff.

BUV, Brilliant Ultra Violet™; PE, R-phycoerythrin; BV, Brilliant Violet™; Cy, cyanine; PerCP-Cy5.5, Peridinin-chlorophyll Cy-5.5; APC, allophycocyanin; Mac., macrophage; M₁, type-I macrophage; M₂, type-II macrophage; FC, fragment crystallizable region; Diff, differentiation

In addition to classifying normal hematopoietic lineages, CD15, CD41a, and CD71 are key markers that can differentiate between several types of MNs that in combination with CD32 and CD163 can, in theory, distinguish between the majority of MDS subtypes, and further sub-fractionate disease subsets including CMML (Figure 1C, and Supplemental Figures 6-8) (16-22). While the basic design of this panel focuses on myeloid lineages, the inclusion of CD38 and CD45RA in place of disease differentiating markers, such as CD32 or CD163, could aid in the analysis of long-term HSC and HSC subtypes(11).

With high-order cytometry, algorithm-guided analyses offer a more comprehensive and objective analytic option compared to conventional gate-based approaches which require foreknowledge of cellular phenotypes. Here, we used spanning-tree progression analysis of density-normalized events (SPADE) to assign cells to one of 200 distinct nodes based on the expression of each parameter (with the exception of the viability/dump, forward scatter, and side scatter parameters) and to then cluster these nodes using a previously described hierarchical algorithm(23). This resulted in 37 distinct node clusters, or phenotypes (Supplemental Figures 9 and 10, Supplemental Table 4). Thirty of these phenotypes correspond to populations identified by gate-based analysis, and in many cases identified further sub-populations, as annotated by a numeric suffix (Supplemental Figure 10 Supplemental Figure 4). SPADE also identified seven phenotypes that do not correspond to populations identified by the gate based analysis. While not demonstrated here, this panel is suitable for other algorithm-guided analysis in addition to SPADE including t-SNE/viSNE, Wanderlust, FlowSOM, or PhenoGraph(24-26).

In summation, this optimized panel is suitable for the study of hematopoiesis with an emphasis on myelopoiesis. Because this panel is able to distinguish phenotypes across a broad range of the myeloid compartment, it can characterize numerous MPNs and in theory differentiate a large portion of MDS subtypes, including CMML. It also has the potential to aid the study of MPNs by further stratifying disease subsets or defining novel biomarker combinations therein, particularly when combined with algorithm-guided analyses such as SPADE.