Normal Haematopoiesis

1. The Haematopoietic Hierarchy

Adult haematopoiesis is a one-way pyramid. At the apex sit ~20,000–50,000 long-term haematopoietic stem cells (LT-HSCs), defined functionally by their capacity to reconstitute the entire blood system after transplantation into an irradiated recipient. They divide rarely and asymmetrically, giving rise to progressively more committed progenitors:

1. LT-HSC (Lin−, CD34+, CD38−, CD90+, CD45RA−) — quiescent, self-renewing.
2. ST-HSC (CD90−) — short-term, ~3–4 month self-renewal.
3. MPP — multipotent progenitor; no self-renewal.
4. CMP / CLP — common myeloid / common lymphoid progenitor (the first major lineage decision).
5. GMP, MEP; Pro-B, Pro-T, NK precursor — downstream restricted progenitors.
6. Mature blood cells — granulocytes, monocytes, erythrocytes, platelets, B cells, T cells, NK cells.

Modern single-cell RNA-seq has softened the rigid “Hodgkin tree” cartoon into a continuum: lineage priming begins very early, and many “CMPs” in fact already lean toward megakaryocyte/erythroid or granulocyte/monocyte fate. Still, the canonical hierarchy remains the indispensable framework for thinking about leukaemia — because each leukaemia type maps to a specific level of arrest.

2. The HSC and Self-Renewal

Self-renewal is the defining HSC property: division yielding at least one daughter that retains stem-cell identity. The choice between self-renewal, differentiation, and apoptosis is governed by:

• Cell-intrinsic transcription factors (HOXA9, MEIS1, BMI1, GATA2, RUNX1).
• Niche signalling (CXCL12 from CAR cells, SCF from endothelial cells, TGF-β from megakaryocytes, Notch ligands).
• Metabolism — LT-HSCs are glycolytic and hypoxic-niche-resident; activation forces oxidative phosphorylation, ROS rise, and exhaustion.

In humans the HSC pool is “polyclonal,” with thousands of clones contributing simultaneously. Lineage tracing (Lee-Six et al., Nature 2018, using somatic-mutation phylogenies) estimated active HSC count at ~50,000–200,000 in adult humans, with each LT-HSC dividing roughly once every 30–40 weeks.

3. Lineage Commitment — Myeloid vs Lymphoid

The first major fork in the haematopoietic tree is between myeloid (CMP) and lymphoid (CLP) lineages:

The choice is governed by competing transcription factors. PU.1 (encoded by SPI1) at high levels drives myeloid fate; at low levels permits lymphoid commitment. PAX5 locks B-cell identity; BCL11B locks T-cell identity. Forced expression of single TFs can reprogram one lineage into another.

4. The Bone-Marrow Niche

HSCs live in specialised microenvironments — niches — that regulate their quiescence, self-renewal, and lineage output. Two anatomical compartments dominate:

The CXCL12/CXCR4 axis is central. CXCL12 produced by stromal cells anchors HSCs in marrow; AMD3100 (plerixafor) blocks CXCR4 and is used clinically to mobilise HSCs into peripheral blood for collection prior to autologous transplant.

In leukaemia the niche becomes complicit. AML blasts remodel marrow stroma into a fibrotic, hypoxic environment that disadvantages residual normal HSCs while supporting LSCs. This explains why patients become pancytopenic before the marrow space is mechanically full of blasts.

5. Cytokines & Growth Factors

The lineages are produced on demand under cytokine direction. Each cytokine binds a receptor that triggers JAK/STAT signalling, ultimately upregulating lineage-specific transcription factors:

These pathways are heavily exploited by leukaemia. JAK2-V617F drives polycythaemia vera; FLT3-ITD drives one third of AML; CSF3R T618I drives chronic neutrophilic leukaemia. Many of these signals converge on STAT5 phosphorylation, which then drives MYC, BCL2, and CCND1 expression — the proliferation/survival programme.

6. Key Transcription Factors

Lineage identity at every level of the hierarchy is enforced by combinations of master TFs. Disruption of any of these is a recurring theme in leukaemia:

Many leukaemia driver fusions act as chimeric transcription factors: RUNX1::RUNX1T1, CBFB::MYH11, PML::RARA, KMT2A::AFF1, ETV6::RUNX1. They corrupt normal lineage-specifying logic, fix cells in a self-renewing precursor state, and await one or more cooperating mutations (in FLT3, NRAS, KIT, IDH1/2) to become fully transformed.

7. Quantitative Output

The numbers are staggering. Daily output in a healthy adult:

• ~2×10¹¹ erythrocytes (~1% turnover/day on a pool of ~2.5×10¹³)
• ~10¹¹ neutrophils with marrow transit ~5–7 days, blood half-life only ~7 hours
• ~10¹¹ platelets from ~10⁸ megakaryocytes
• ~10⁹ lymphocytes generated; most undergo apoptosis

Mathematically, the steady-state size N of a compartment with production rate P and first-order loss rate k is N = P / k, i.e.  $P = k \cdot N$. For neutrophils with $N \approx 5 \times 10^9$ in blood and $t_0.5 \approx 7$ h ( $k = \ln 2 / t_0.5$ ), the required production rate is on the order of $10^10$ cells/day — demanded from a marrow compartment of finite size.

Loss of even modest output (chemotherapy, marrow infiltration, radiation) produces cytopenia within days for neutrophils (short half-life), within weeks for platelets, within ~3 months for red cells (mean lifespan ~120 days). This is exactly the clinical tempo of acute leukaemia and chemotherapy-induced cytopenias.

8. HSC Aging & Clonal Haematopoiesis

HSCs accumulate somatic mutations linearly with age — on the order of ~17 mutations per HSC per decade by whole-genome sequencing of single-HSC-derived colonies (Lee-Six 2018; Mitchell 2022). By age 70, the typical adult’s blood is descended from a few hundred to a few thousand HSCs — oligoclonal rather than polyclonal.

When one of those clones harbours a leukaemia-related driver mutation (most commonly DNMT3A, TET2, ASXL1) and reaches ≥2% variant allele fraction in blood, the condition is Clonal Haematopoiesis of Indeterminate Potential (CHIP) (Genovese 2014; Jaiswal 2014; Steensma 2015). CHIP:

• Affects ~10% of those over 65 and >20% of those over 80.
• Confers ~10–15× relative risk of progression to MDS or AML — though absolute risk is still ~0.5–1% per year.
• Independently predicts cardiovascular disease and all-cause mortality — suggesting clonal myeloid cells exacerbate atherosclerosis through inflammation (the IL-1β / IL-6 axis).

CHIP is the explicit pre-leukaemic phase of myeloid neoplasms and the bridge between normal aging and AML. We will return to CHIP in Part V.

9. Why this Matters for Leukaemia

Every leukaemia subtype is best understood as a corruption of a specific node in this hierarchy:

• CML arises in the HSC or earliest myeloid progenitor — chronic-phase cells differentiate but proliferate without restraint.
• AML arises in HSC or myeloid progenitors with imposed differentiation arrest at the blast level — FAB M0–M7 reflect alternative arrest points (M3 = promyelocyte, M6 = erythroid, M7 = megakaryoblast).
• B-ALL arises at the pro-B / pre-B stage — cells accumulate as immature CD19+ CD10+ blasts.
• T-ALL arises at thymic stages — cortical (CD1a+, CD4+CD8+) or early T-cell precursor (ETP) phenotype.
• CLL is a clonal expansion of post-germinal-centre or naive mature CD5+ B cells — the cell is fully differentiated; the disease is one of inappropriate accumulation.
• MDS is dysplastic differentiation with peripheral cytopenias and a tendency to evolve into AML.

With this framework in hand, the next two parts open the four classical leukaemias. Part III covers the acute leukaemias (AML and ALL); Part IV covers the chronic leukaemias (CML and CLL).