Part III
Acute Leukaemias — AML & ALL
Differentiation arrest at the blast stage. The morphological FAB taxonomy is being replaced by a 2022 molecular classification dominated by recurrent fusions and a handful of cooperating mutations. We close with two crystal structures — FLT3 kinase and BCR-ABL bound to imatinib — that explain the targeted-therapy era.
1. The Blast
An acute leukaemia is, by classical definition, a marrow or blood with ≥20% blasts (WHO 2022; ICC 2022). The blast is morphologically:
- Large (15–20 µm)
- Open chromatin (“dispersed,” “immature”)
- One or more prominent nucleoli
- Scant basophilic cytoplasm
- High nuclear-to-cytoplasmic ratio
Two cytochemical markers separate myeloid from lymphoid:
- Auer rods — pink, needle-shaped, MPO-positive azurophilic granules in cytoplasm; pathognomonic for AML.
- Myeloperoxidase (MPO) + — defines myeloid lineage. Sudan Black B is its less specific cousin.
- TdT (terminal deoxynucleotidyl transferase) — nuclear, defines lymphoblast (and a few minimally-differentiated myeloblasts).
- Non-specific esterase — defines monocytic lineage (M4, M5).
Modern practice uses flow cytometry rather than cytochemistry, but the underlying logic survives: lineage is read out from a panel of surface antigens (covered in Part VI).
2. AML — the FAB Classification (Historical)
The 1976 French-American-British scheme (Bennett, Catovsky, Daniel, Flandrin, Galton, Gralnick, Sultan) was the first systematic morphological taxonomy. It still furnishes vocabulary in current use:
| FAB | Name | Distinguishing features |
|---|---|---|
| M0 | AML minimally differentiated | MPO− on light-microscopy; flow needed |
| M1 | AML without maturation | ≥90% blasts (of non-erythroid); rare granules |
| M2 | AML with maturation | ≥10% maturing myeloid; classic t(8;21) RUNX1::RUNX1T1 |
| M3 | Acute promyelocytic (APL) | Hypergranular promyelocytes, faggot cells; t(15;17) PML::RARA |
| M4 | Acute myelomonocytic | ≥20% monocyte-lineage; M4Eo subtype with eosinophils, inv(16) |
| M5 | Acute monocytic / monoblastic | ≥80% monocytic; gum hypertrophy, CNS & skin |
| M6 | Acute erythroid | Pure erythroid (M6b) very rare; redefined in WHO 2022 |
| M7 | Acute megakaryoblastic | CD41/61+; common in Down syndrome |
FAB’s 30%-blast threshold and morphology-only logic are now obsolete — but the M3 (APL) terminology remains in active clinical use because it identifies a uniquely manageable disease.
3. AML — WHO 5th Edition / ICC 2022
Two parallel modern classifications (WHO 5th edition; ICC 2022) split AML into two broad streams:
Genetically defined AML
Diagnosis on the basis of a defining genetic lesion regardless of blast %. These “recurrent genetic abnormalities” include:
- PML::RARA t(15;17) — APL
- RUNX1::RUNX1T1 t(8;21)
- CBFB::MYH11 inv(16)/t(16;16)
- KMT2A rearrangements (11q23)
- NPM1 mutation
- Biallelic CEBPA mutation (b/zip in-frame)
- NUP98 rearrangements
- BCR::ABL1 (rare in AML)
- DEK::NUP214 t(6;9)
- MECOM rearrangements (3q26)
Otherwise classified AML
Requires ≥10% (WHO) or ≥20% (ICC) blasts; further classified by:
- Myelodysplasia-related changes — MDS-related cytogenetics or mutations (SRSF2, SF3B1, U2AF1, ZRSR2, ASXL1, EZH2, BCOR, STAG2)
- Therapy-related (t-AML) — prior alkylator (5–7 yr latency, often complex karyotype, TP53) or topoisomerase-II inhibitor (1–3 yr, KMT2A)
- Down-syndrome-associated
- Germline-predisposed (CEBPA, RUNX1, GATA2, DDX41, SAMD9/SAMD9L)
- NOS — otherwise unspecified
The ELN 2022 risk stratification then collapses these subtypes into three prognostic tiers (favourable / intermediate / adverse) based on cytogenetics and specific mutations. Biallelic CEBPA, NPM1 + low FLT3-ITD ratio, and core-binding-factor fusions are favourable; complex karyotype, monosomal karyotype, TP53, and high FLT3-ITD ratio are adverse.
4. AML Driver Mutations
The TCGA AML 2013 study (NEJM, 200 cases) and subsequent larger studies showed that AML is driven by a small number of recurrent mutations, with a median of only ~13 somatic coding mutations per case — one of the lowest mutational burdens of any cancer.
| Class | Genes | Frequency | Targeted therapy |
|---|---|---|---|
| Signalling / kinase (Class I) | FLT3-ITD, FLT3-TKD, NRAS, KRAS, KIT, PTPN11 | ~50% | Midostaurin (FLT3, KIT); gilteritinib, quizartinib (FLT3) |
| Transcription / differentiation (Class II) | NPM1, CEBPA, RUNX1, GATA2 | ~50% | Menin inhibitors (KMT2A/NPM1) |
| Epigenetic / DNA methylation | DNMT3A, TET2, IDH1, IDH2, WT1 | ~45% | Ivosidenib (IDH1), enasidenib (IDH2); HMA |
| Chromatin modifiers | ASXL1, EZH2, KMT2A, BCOR | ~30% | Menin inhibitors (revumenib, ziftomenib) |
| Splicing factors | SRSF2, SF3B1, U2AF1, ZRSR2 | ~15% AML, much higher in MDS-AML | — |
| Cohesin / spliceosome | STAG2, RAD21, SMC1A | ~10% | — |
| Tumour suppressor | TP53 | ~10% (~30% therapy-related) | Magrolimab (anti-CD47); transplant; poor response to all |
FLT3-ITD deserves special mention — an internal-tandem-duplication of the juxtamembrane domain that constitutively activates the FLT3 receptor tyrosine kinase. It occurs in ~25% of AML, confers poor prognosis (especially at high allelic ratio ≥0.5), and is targeted by midostaurin (RATIFY trial 2017) and gilteritinib (ADMIRAL trial 2019, NEJM).
NPM1 mutations (commonly c.863_864insTCTG) shift the protein from nucleolus to cytoplasm, derange MEIS1/HOXA9 expression, and define a favourable subgroup when isolated; menin-MLL complex inhibitors (revumenib, FDA-approved 2024) target NPM1c-mutant and KMT2A-rearranged AML.
IDH1/IDH2 mutations produce the oncometabolite 2-hydroxyglutarate, which inhibits TET2 and other α-KG-dependent dioxygenases, hyper-methylating DNA and blocking myeloid differentiation. Ivosidenib (IDH1) and enasidenib (IDH2) reverse this and induce remissions, often with a characteristic differentiation syndrome.
5. APL — A Special Case
Acute promyelocytic leukaemia is the success story of haematology. The t(15;17)(q24;q21) PML::RARA fusioncreates a chimeric protein in which the N-terminal of PML (a nuclear-body protein) is fused to retinoic-acid receptor α. The fusion binds DNA as an oligomer, recruits co-repressors at physiological concentrations of retinoid, and locks granulocytic precursors at the promyelocyte stage.
Two non-cytotoxic agents resolve this:
- All-trans retinoic acid (ATRA) — pharmacological retinoid forces co-repressor release and induces terminal granulocytic differentiation.
- Arsenic trioxide (As2O3, ATO) — binds the PML moiety, triggers SUMOylation and proteasomal degradation of PML-RARA.
The Lo-Coco APL0406 trial (NEJM 2013) showed ATRA + ATO without cytotoxic chemotherapy achieved >95% complete remission and >90% 5-year overall survival in low/intermediate-risk APL, formally supplanting anthracycline-based therapy in standard-risk disease.
6. ALL — Lineage and Subtypes
Acute lymphoblastic leukaemia is divided by lineage (B vs T) and immunophenotypic maturation:
| Subtype | Phenotype | Notes |
|---|---|---|
| Pro-B (early-pre-B) | CD19+, CD10−, cyTdT+ | Often KMT2A-rearranged, infant ALL |
| Common B-ALL | CD19+, CD10+ (CALLA), CD22+, cyµ− | Most common paediatric subtype |
| Pre-B | CD19+, CD10+, cyµ+ | t(1;19) TCF3::PBX1 association |
| Mature B (Burkitt) | CD19+, smIg+, CD20+, c-MYC+ (FAB L3) | Treated as Burkitt, not ALL |
| Early T-cell precursor (ETP) | CD7+, CD1a−, CD8−, myeloid markers possible | Adverse risk; mixed myeloid/lymphoid features |
| Cortical T-ALL | CD1a+, CD4+CD8+ double-positive | NOTCH1 mutations; favourable |
| Mature T-ALL | CD1a−, single-positive CD4 or CD8 | Less common |
The historical FAB L1/L2/L3 distinction is obsolete: L1 (small uniform) and L2 (larger pleomorphic) are now both diagnosed simply as B- or T-ALL by flow; L3 with its deep-blue vacuolated cytoplasm is the leukaemic phase of Burkitt lymphoma and is treated separately.
7. ALL Driver Lesions
Recurrent genetic lesions in B-ALL are dominated by chromosomal translocations and copy-number changes; T-ALL is dominated by NOTCH1 mutations and TF rearrangements:
| Lesion | Frequency | Prognosis |
|---|---|---|
| B-ALL | ||
| High hyperdiploidy (>50 chr) | ~25% paediatric | Favourable |
| ETV6::RUNX1 t(12;21) | ~25% paediatric, rare adult | Favourable |
| TCF3::PBX1 t(1;19) | ~5% | Intermediate |
| KMT2A rearrangement t(4;11) etc. | ~5% (~80% of infant ALL) | Adverse |
| BCR::ABL1 t(9;22) (Ph+) | ~3% paed, ~25% adult | Adverse pre-TKI; near-favourable now |
| Hypodiploidy (<44 chr) | ~1% | Adverse; TP53 germline frequent |
| iAMP21 | ~2% | Adverse |
| Ph-like (CRLF2 / JAK / ABL-class) | ~10–15% adolescent/young adult | Adverse |
| DUX4-rearranged | ~5% | Favourable (recently recognised) |
| T-ALL | ||
| NOTCH1 activating | ~55% | Standard risk |
| FBXW7 loss | ~15% | Standard risk |
| CDKN2A deletion | ~70% | Standard risk |
| TAL1, LMO2, TLX1/3 dysregulation | ~50% combined | Subtype-defining |
| ETP-ALL signature | ~10% | Adverse |
IKZF1 deletions, frequent in Ph+ and Ph-like B-ALL, predict inferior outcome. PAX5 alterations are frequent and now serve as subtype delimiters. Ras-pathway mutations (NRAS, KRAS, FLT3, NF1) commonly co-occur and may mark relapse-fated subclones.
8. Ph+ and Ph-like ALL
Ph+ ALL (with t(9;22) BCR::ABL1) was historically the most adverse subtype of adult ALL, with cures <20% by chemotherapy alone. Modern regimens add a TKI (imatinib, then dasatinib or ponatinib) plus corticosteroids, often without classical cytotoxic induction (Foà’s dasatinib + steroid + blinatumomab; Jabbour’s ponatinib + blinatumomab, Lancet Haematol 2024) and reach >75% MRD-negative complete molecular response.
Ph-like ALL (Mullighan, Loh, Roberts, NEJM 2014) is a gene-expression-defined subtype that resembles Ph+ ALL transcriptionally but lacks BCR::ABL1. Drivers include CRLF2 rearrangements (often P2RY8::CRLF2 or IGH::CRLF2), JAK2 fusions and mutations, EPOR fusions, and ABL-class fusions (e.g. NUP214::ABL1, EBF1::PDGFRB) sensitive to TKIs. It is enriched in adolescents and young adults (~25% of AYA B-ALL) and confers poor prognosis with standard chemotherapy.
JAK-pathway Ph-like ALL is being treated with ruxolitinib added to chemotherapy (CHILDREN’S Oncology Group AALL1521); ABL-class with imatinib or dasatinib. Identification requires either a low-density gene-expression panel or RNA-seq up front — testing the limits of standard diagnostic infrastructure.
9. Risk Stratification
AML — ELN 2022
- Favourable: CBF (RUNX1::RUNX1T1, CBFB::MYH11), NPM1 + low-FLT3-ITD ratio, biallelic CEBPA bZIP
- Intermediate: NPM1 + high-FLT3-ITD ratio; FLT3-ITD without NPM1 (low ratio); KMT2A-MLLT3
- Adverse: complex/monosomal karyotype, −5/del5q, −7, 17p abnormalities, TP53 mutation, ASXL1, RUNX1, EZH2, BCOR, STAG2, U2AF1, SF3B1, SRSF2, ZRSR2, KMT2A non-MLLT3 fusions, MECOM, NUP98 fusions
ALL — clinical risk factors
- Age: <1 yr (KMT2A) and ≥35 yr adverse
- Initial WBC: >30 (B-ALL) / >100 (T-ALL) ×10⁹/L adverse
- CNS disease at diagnosis
- Cytogenetics (above)
- Day 14/29 MRD by flow or RT-qPCR — single most powerful predictor; MRD− vs MRD+ separate ~30%-point survival
- Slow morphologic response (M2/M3 marrow on day 14/29)
Risk-adapted therapy is the rule: low-risk AML (favourable cytogenetics, NPM1 mutated, low FLT3-ITD, MRD-negative after induction) avoids transplant in CR1; intermediate and adverse-risk AML proceed to allogeneic stem-cell transplant in CR1 if a donor and fitness allow. In paediatric ALL, MRD at day 29 governs whether the patient de-escalates to standard therapy or escalates to high-risk arms with augmented asparaginase, transplant, or blinatumomab consolidation.
10. Drug-Target Structures
Two structures are at the heart of acute-leukaemia targeted therapy: BCR-ABL kinase bound to imatinib (PDB 1IEP) and the FLT3 kinase domain (PDB 4XUF/4CC8). Imatinib binds the inactive (DFG-out) conformation of ABL; the same trick was later extended to FLT3 by midostaurin and gilteritinib.
BCR-ABL kinase + imatinib (1IEP)
The first targeted-therapy crystal in cancer (Schindler et al., Science 2000). Imatinib (yellow sticks) occupies the ATP pocket of ABL with the kinase locked in DFG-out — explaining substrate competition and selectivity.
FLT3 kinase domain (4XUF)
Catalytic kinase domain of FLT3, the target of midostaurin and gilteritinib in FLT3-ITD AML. The juxtamembrane region — site of the activating ITD insertions — is upstream of this construct.
The BCR-ABL/imatinib structure also explains the most common imatinib-resistance mutation: T315I, the so-called “gatekeeper” residue. The threonine’s side-chain hydroxyl forms a key hydrogen bond with imatinib; isoleucine fills the space and abolishes the interaction. T315I is insensitive to all four first/second-generation TKIs and required design of ponatinib (covered in Part IV).