CDK6 - the kinase the KMT2A-r rationale actually rests on, and the pan-essentiality flag and the selectivity are both true

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

• The window is a lineage window, and the lineage is the toxicity. CDK6 dependency is deepest in haematopoietic cells; neutropenia is the dose-limiting toxicity of this class; the one DLT in the COG paediatric palbociclib trial was haematologic. Going CDK6-selective concentrates the pressure where the marrow already is.
• Cdk6-null HSCs fail competitive repopulation and Cdk6-null mice are more sensitive to 5-fluorouracil [E5]. That is an essential-function toxicity in exactly the compartment under stress during induction, and it is invisible in steady state - so a tolerability signal in an unstressed patient does not bound it.
• Part of CDK6's leukaemic function is kinase-independent [E6]. An ATP-competitive inhibitor removes the catalytic arm and leaves the scaffold, so the DepMap knockout number is an upper bound on what any current drug can deliver, not a prediction of it.
• CDK6 induces p16INK4a [E6]. Inhibiting CDK6 in a cell where CDKN2A is intact - which is about 91% of infant KMT2A-r ALL - may relieve the very brake the therapy depends on, and the direction of that feedback has never been measured in this disease.
• Cytostatic, not cytotoxic. A G1 arrest leaves cells alive in a disease whose hazard is front-loaded, and an arrested clone is a clone available to acquire resistance.
• RB1 loss is the canonical resistance route for this class, and live DepMap gives RB1 a POSITIVE mean gene effect (+0.1801 across 1,258 lines, +0.1127 in leukaemia lines) - losing RB1 is fitness-neutral to favourable, so nothing stops a treated clone from taking that exit.
• The target is a fusion output. Any escape that lowers dependence on the fusion programme - including the documented lineage switch - takes the dependency with it.
• the inhibitor targets a pool that may not be the relevant one - Only nuclear CDK6 complexes are catalytically active and they are a minority of cellular CDK6; the transcriptional function at promoters is partly kinase-independent. An ATP-site CDK4/6 inhibitor and a CDK6 degrader are therefore not interchangeable in KMT2A-rearranged disease, and nobody has measured the CDK6 partition or chromatin occupancy in a KMT2A-rearranged cell to say which matters.
• paralogue escape - CDK4, not CDK6, is absent from the FLT3/PIM1 promoters in AML, so a pan-CDK4/6 inhibitor perturbs two proteins with different chromatin behaviour and a resistance route through CDK4 up-regulation would look like on-target failure.

1. Does CDK6's kinase-independent transcriptional function operate in KMT2A-rearranged cells at all? Every published demonstration is in BCR-ABL+ or normal HSC systems. A kinase-dead CDK6 knock-in rescue in a KMT2A::AFF1 line would settle whether a CDK4/6 inhibitor can phenocopy the knockout this graph's rank-13 claim is built on.
2. ANSWERED IN PART, and the answer is the opposite of the worry: menin-inhibitor treatment ATTENUATES CDK6 levels in MLL1-rearranged AML, and yet the menin-inhibitor plus abemaciclib combination is SYNERGISTIC rather than redundant [E8]. What remains open is whether this holds in B-lineage KMT2A::AFF1, where the differentiation mechanism that both agents are said to act through has no myeloid counterpart.
3. Is the leukaemia-lineage CDK6 dependency present in B-lineage lines specifically, or is the 57-line mean carried by myeloid lines? That split was not computed this cycle and the index arm of the project's scan is myeloid-enriched.
4. What is the CDK6 gene effect in an infant-derived KMT2A::AFF1 line? None exists in DepMap; ALL-PO (CVCL_1069, age 4 months) is in the collection with no CRISPR value, which is already ledger priority 16.
5. Is the menin-inhibitor/CDK4-6 synergy CDK6-specific or generic? In NUP98-rearranged leukaemia, palbociclib and gilteritinib - two unrelated kinase inhibitors - converged on the same pathways when combined with a menin inhibitor [E8 counter-evidence], which suggests the synergy may be a proliferation-plus-differentiation effect available to many partners.
6. What is the nuclear/cytoplasmic partition of CDK6 in KMT2A-rearranged blasts, and is CDK6 on chromatin at HOXA9, MEIS1 or FLT3 there? If the relevant pool is the chromatin-bound transcriptional one, a degrader and an ATP-site inhibitor would separate cleanly in this disease and nobody has tested it.

1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In MLL-rearranged infant ALL and human MLL-rearranged cell lines, CDK6 mRNA is higher than in wild-type infant ALL (P<0.001), falls on siRNA knockdown of MLL-AF4 or MLL-ENL, and knockdown of CDK6 - but not CDK4 - inhibits growth of the KMT2A::AFF1-positive line SEM, with MLL-rearranged lines undergoing G1 arrest on the CDK4/6 inhibitor PD0332991.

   MLL fusion-driven activation of CDK6 potentiates proliferation in MLL-rearranged infant ALL · van der Linden MH, Willekes M, van Roon E, Seslija L, Schneider P, Pieters R, Stam RW · Cell Cycle 13:834-844 2014
   PMID 24736461DOI
   ▸What this evidence cannot show

   Abstract read this cycle; full text not retrieved. Cell lines and patient expression, no in vivo or clinical readout. SEM is a t(4;11) line from a five-year-old (CVCL_0095), not infant-derived. Nothing in the abstract demonstrates fusion occupancy at the CDK6 locus, so fusion-driven is an inference from knockdown and not from ChIP. The PD0332991 arm uses a dual CDK4/6 agent and cannot attribute the arrest to CDK6.

   ▸Evidence that argues against this (1)

   • The one trial ever run with this drug class in MLL-rearranged acute leukaemia, NCT02310243, completed in July 2019 and has posted no results, twelve years after this pre-clinical case was first made. University of Ulm; Pfizer; National Center for Tumor Diseases et al. 2015
2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In murine and human MLL-AF9-driven AML, cells are exceptionally reliant on CDK6 but not CDK4; the growth inhibition from CDK6 depletion is mediated through enhanced myeloid differentiation; the requirement extends to alternative MLL fusions and to a mouse MLL-AF9 model; CDK6 is transcriptionally activated by mutant MLL; and a small-molecule CDK6 inhibitor closely phenocopies depletion.

   Requirement for CDK6 in MLL-rearranged acute myeloid leukemia · Placke T, Faber K, Nonami A, Putwain SL, Salih HR, Heidel FH, et al. · Blood 124:13-23 2014
   PMID 24764564DOI
   ▸What this evidence cannot show

   Abstract read this cycle. Murine and human MYELOID MLL-AF9 disease, not infant B-lineage KMT2A::AFF1 - the mechanism claimed is myeloid differentiation, which has no direct counterpart in a pro-B blast. Rule 73 applies: this is the system, and the system is not the index disease.

   ▸Evidence that argues against this (1)

   • If CDK6 inhibition acts through differentiation, it targets the same final common path claimed for menin inhibition, DOT1L inhibition and LSD1 inhibition in this disease, so combination with a menin inhibitor is as likely to be redundant as synergistic. Maiques-Diaz A et al. 2018
3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

   In this project's unbiased genome-wide scan of 17,728 protein-coding genes for selective requirement in 11 Cellosaurus-annotated KMT2A-rearranged leukaemia lines versus 43 other leukaemia lines, CDK6 ranks 13th (difference -0.5266; KMT2A-r mean -1.7918, other leukaemia mean -1.2652) while CDK4 ranks 16,052nd (difference +0.0716; KMT2A-r mean -0.3742, other leukaemia mean -0.4458). MEN1 ranks 1 and KMT2A 11.

   Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets); analyses/depmap-kmt2a-selective-scan-full/results.json · Broad Institute DepMap; Cellosaurus (SIB); Open Targets Platform; analysis run by this project · Databases and derived analysis 2026
   ▸What this evidence cannot show

   This project's own unreviewed computation, re-read from the persisted results file on 2026-09-20, not re-run. Eleven KMT2A-rearranged lines, none infant-derived, and the comparison is not conditioned on lineage within the index arm - that arm is roughly 7 myeloid to 4 lymphoid, so a difference mixes fusion status with lineage. Gene effect is knockout, not drug.

   ▸Evidence that argues against this (1)

   • The same scan ranks HOXA9 3,649th with a difference of -0.0479 - the gene most often described as the fusion's output is not a selective dependency in it, which is a reason to treat rank in this scan as a statement about these eleven lines rather than about the disease. analysis run by this project 2026
4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

   Fetched live on 2026-09-20, DepMap Chronos gene effect for CDK6 across 1,258 cell lines has mean -0.6321 and median -0.5267, with 51.9% of lines at or below -0.5 and 23.1% at or below -1.0; for CDK4 the mean is -0.7371 and median -0.5645, with 58.4% at or below -0.5. Restricted to the 57 leukaemia-annotated lines, CDK6's mean is -1.3767 and CDK4's is -0.4464. CDK4 is therefore the broader dependency across all tissues and CDK6 the deeper one in haematopoietic lines.

   DepMap CRISPR (Chronos) gene effect for CDK6 (ENSG00000105810) and CDK4 (ENSG00000135446), retrieved via the Open Targets Platform GraphQL API, depMapEssentiality.screens · Broad Institute DepMap; Open Targets Platform; query run by this project 2026-09-20 · Database query 2026
   ▸What this evidence cannot show

   A database fact opposable on scope, per the standing rule. The leukaemia split is by the screen's own diseaseFromSource string matched on leukemia/leukaemia/lymphoblastic/myeloid, which is a coarse text match and not a curated lineage annotation, and it is not conditioned on fusion status. The DepMap release version is not stated by the API. Chronos gene effect is a CRISPR knockout fitness score in immortalised lines over roughly 21 days: it cannot see a self-renewal or differentiation phenotype, and it is not a measurement of what a drug does. No infant-derived line is in the panel.

   ▸Evidence that argues against this (1)

   • The same two genes are reported to have redundant functions in cell-cycle progression, so single-gene knockout scores understate whichever kinase its paralogue can cover - which would depress CDK4's leukaemia score if CDK6 is present to compensate, and vice versa in solid tissue. Scheicher R et al. 2015
5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In mice, CDK6 is required for stem-cell activation rather than for steady-state haematopoiesis: Cdk6-null HSCs do not efficiently repopulate upon competitive transplantation, Cdk6-deficient mice are significantly more susceptible to 5-fluorouracil, and the mechanism is transcriptional suppression of Egr1 by a CDK6-containing complex, with the same repopulation defect extending to BCR-ABL(p210)+ leukaemic stem cells.

   CDK6 as a key regulator of hematopoietic and leukemic stem cell activation · Scheicher R, Hoelbl-Kovacic A, Bellutti F, Tigan AS, Prchal-Murphy M, Heller G, et al. · Blood 125:90-101 2015
   PMID 25342715DOI
   ▸What this evidence cannot show

   Abstract read this cycle. Murine germline knockout in a BCR-ABL model, not KMT2A-rearranged and not human. A germline knockout is a lifetime absence and is not the same exposure as weeks of partial reversible pharmacological occupancy - the standing database-scope rule applies to mouse genetics too. What it does establish is that the normal haematopoietic compartment has a CDK6 requirement revealed by stress, which is the condition during induction chemotherapy.

   ▸Evidence that argues against this (1)

   • The same abstract states haematopoiesis appears normal under steady-state conditions in Cdk6-null mice, so the toxicity argument is conditional on stress and regeneration rather than being a baseline marrow requirement. Scheicher R et al. 2015
6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In a murine p185 BCR-ABL+ B-acute lymphoid leukaemia model, CDK6 is part of a transcription complex that induces expression of the tumour suppressor p16INK4a and of VEGF-A, and this function is independent of CDK6's kinase activity; high CDK6 therefore suppresses proliferation by upregulating p16INK4a unless p16INK4a is absent, in which case CDK6 exerts its full tumour-promoting function.

   A kinase-independent function of CDK6 links the cell cycle to tumor angiogenesis · Kollmann K, Heller G, Schneckenleithner C, Warsch W, Scheicher R, Ott RG, et al. · Cancer Cell 24:167-181 2013
   PMID 23948297DOI
   ▸What this evidence cannot show

   Abstract read this cycle. Murine BCR-ABL+ B-ALL, not KMT2A-rearranged, and the angiogenesis half of the claim is not obviously transferable to a marrow-resident leukaemia. The load-bearing part for this entry is narrower and better supported: a documented non-catalytic CDK6 function in a lymphoid leukaemia, which is enough to break the knockout-to-inhibitor inference without the VEGF biology having to hold here.

   ▸Evidence that argues against this (1)

   • Against reading kinase-independence as the whole story: two independent groups report that small-molecule CDK6 or CDK4/6 inhibition phenocopies CDK6 depletion in MLL-rearranged cells, which argues the catalytic arm does carry real weight in this specific context. Placke T et al. 2014
7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   CCND3 is indispensable for the maintenance of B-cell acute lymphoblastic leukaemia independent of mutational background, and its anti-apoptotic effect is independent of the kinase activity of the CCND3-CDK4/6 complex, leading the authors to conclude that downregulating CCND3 might be superior to inhibiting CDK4/6 kinase activity.

   CCND3 is indispensable for the maintenance of B-cell acute lymphoblastic leukemia · Ketzer F, Abdelrasoul H, Vogel M, Marienfeld R, Muschen M, Jumaa H, Wirth T, Ushmorov A · Oncogenesis 11:1 2022
   PMID 35013097DOI
   ▸What this evidence cannot show

   Abstract read this cycle. Cited here only for the kinase-independence point, which converges with E6 from the cyclin side. B-ALL generally, not selected for KMT2A rearrangement. One paper, not independently replicated in anything read here.

   ▸Evidence that argues against this (1)

   • Live DepMap shows CCND3 is a much deeper dependency in leukaemia lines (mean -1.0076 across 57) than across all lines (-0.3459), so the cyclin and the kinase track each other by lineage - which is also what a genuinely kinase-dependent complex would look like. Broad Institute DepMap; Open Targets Platform 2026
8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In human AML cell lines and patient-derived AML cells harbouring MLL1 rearrangement or mutant NPM1, menin-inhibitor treatment ATTENUATES CDK6 levels, and co-treatment with the menin inhibitor SNDX-50469 and the CDK4/6 inhibitor abemaciclib induces SYNERGISTIC lethality; the same study reports synergy with venetoclax and shows the combinations are preserved against MLL1-r AML cells expressing mutant FLT3 and CRISPR-edited to introduce mutant TP53.

   Effective Menin inhibitor-based combinations against AML with MLL rearrangement or NPM1 mutation (NPM1c) · Fiskus W, Boettcher S, Daver N, Mill CP, Sasaki K, Birdwell CE, Davis JA, Takahashi K, Kadia TM, DiNardo CD, et al. · Blood Cancer Journal 12:5 2022
   PMID 35017466DOI
   ▸What this evidence cannot show

   Abstract read this cycle; full text not retrieved. The system is AML - human AML cell lines and patient-derived AML cells with MLL1 rearrangement or mutant NPM1 - NOT infant B-lineage KMT2A::AFF1, and the in vivo arm of the paper is the venetoclax combination, not the abemaciclib one, so the CDK4/6 synergy is reported in vitro and ex vivo only. No synergy statistic, concentration or line identity was read. Abemaciclib is a dual CDK4/6 agent, so the synergy cannot be attributed to CDK6 alone.

   ▸Evidence that argues against this (1)

   • The parallel result in NUP98-rearranged leukaemia found that palbociclib and gilteritinib, despite inhibiting two unrelated kinases, affected SIMILAR pathways when combined with a menin inhibitor - cell cycle, proliferation and differentiation - which is an argument that the synergy is a generic proliferation-plus-differentiation effect rather than a CDK6-specific one. Miao H et al. 2024
9. E_q65_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

   In human T cells CDK6 partitions across at least three complexes with different compartments - the abundant 450 kDa Hsp90/CDC37 complex is predominantly cytoplasmic while the 170 kDa cyclin D/CDK6 complex is in both - and only the NUCLEAR CDK6 complexes have kinase activity, so catalytically active CDK6 is a minority of total CDK6.

   Active cdk6 complexes are predominantly nuclear and represent only a minority of the cdk6 in T cells. · Mahony D, Parry DA, Lees E · Oncogene 1998;16:603-11 1998
   PMID 9482106DOI
   ▸What this evidence cannot show

   1998 biochemistry in normal T cells, not leukaemia and not a KMT2A-rearranged cell. Fractionation leaks between compartments, and the study predates the kinase-independent transcriptional functions later attributed to CDK6, which a kinase-activity readout scores as zero. Abstract only.

   ▸Evidence that argues against this (1)

   • CDK6 has documented kinase-INDEPENDENT transcriptional functions, so 'only nuclear CDK6 is active' is true of catalytic activity and may be false of function; a cytoplasmic pool scored inactive by kinase assay is not thereby inert. Uras IZ et al. 2016
10. E_q65_loc2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In FLT3-ITD acute myeloid leukaemia cells CDK6, but not its functional homolog CDK4, is found at the promoters of the FLT3 and PIM1 genes where it regulates their transcription, and palbociclib kills these cells through that transcriptional activity.

    Palbociclib treatment of FLT3-ITD+ AML cells uncovers a kinase-dependent transcriptional regulation of FLT3 and PIM1 by CDK6. · Uras IZ, et al. · Blood 2016;127:2890-902 2016
    PMID 27099147DOI
    ▸What this evidence cannot show

    FLT3-ITD AML, not KMT2A-rearranged, and cell-line rather than patient material. ChIP is a genomic address, not a subnuclear compartment. The CDK4-versus-CDK6 discrimination is the load-bearing detail and is a paralogue distinction of exactly the class this graph's registry warns about. Abstract only.

    ▸Evidence that argues against this (1)

    • CDK6 is a recurrent KMT2A-fusion target gene and the CDK4/6-inhibitor rationale in this disease rests on the cell-cycle kinase function; if the relevant pool in KMT2A-rearranged cells were the chromatin-bound transcriptional one, an ATP-site inhibitor and a degrader would not be interchangeable - and nobody has measured the partition here. Mahony D et al. 1998
11. E_q65_loc3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    Elevated NUCLEAR CDK6 specifically, scored by compartment on immunohistochemistry, is associated with unfavourable prognosis in lung adenocarcinoma patients - evidence that the nuclear/cytoplasmic partition of CDK6 is discriminating in tissue and not only in fractionation.

    Elevated nuclear CDK6 is associated with an unfavorable prognosis in lung adenocarcinoma patients. · Li H, Zhang J, Xie Y · Int J Clin Exp Pathol 2017;10:9614-9620 2017
    PMID 31966839
    ▸What this evidence cannot show

    Graded weak: a small single-cohort IHC association in a solid tumour, in a low-visibility journal, and the PubMed record carries NO DOI (PMID only, PMC6965964). Compartment scoring by IHC depends on the antibody and on fixation. Cited for the existence of a compartment-resolved readout, not for the prognostic number. Abstract only.

    ▸Evidence that argues against this (1)

    • An IHC compartment score cannot distinguish the kinase-active cyclin D/CDK6 complex from the Hsp90/CDC37-bound pool, which the biochemistry shows are differently distributed; a nuclear signal is not by itself an active-CDK6 signal. Mahony D et al. 1998

• related to →CDK4 - the other half of the class name, and it is NOT the gene this disease needs

  paralogous G1 kinases at opposite ends of the KMT2A-r selectivity ranking (13 vs 16,052 of 17,728)

• treated by →CDK4/6 inhibition in KMT2A-rearranged leukemia — the class was already tried, and the trial never reported

  CDK6 is the gene that carries the class rationale; CDK4/6 inhibitors engage it

• ← inhibitsCDKN2A - one gene, two reviewed UniProt records with the same primary symbol, and a deletion that is three times rarer here than in the infant leukaemias next door

  p16INK4a (P42771) is the endogenous inhibitor of CDK6

• regulates →CDKN2A - one gene, two reviewed UniProt records with the same primary symbol, and a deletion that is three times rarer here than in the infant leukaemias next door

  CDK6 transcriptionally induces p16INK4a in a kinase-independent complex (murine BCR-ABL+ B-ALL)

• ← regulatesKMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)

  CDK6 mRNA falls on siRNA knockdown of MLL-AF4; CDK6 is a transcriptional output of the fusion

• causes →The G1 state in KMT2A-rearranged leukaemia — where CDK6 does a job that is not about the cell cycle at all

  G1 restriction-point kinase underlying that cell state

• ← related toBMI1, CDKN2A and CDK6 are one circuit, not three genes - and in KMT2A-rearranged disease it has a loop nobody has measured

  the kinase the brake grips, and which induces the brake back kinase-independently

• ← supportsLive DepMap on fourteen genes: CDK6's pan-essentiality is a lineage artefact of pooling, and not one gene of the homeodomain cofactor code is a dependency anywhere

  live lineage-split DepMap: -0.6321 all lines, -1.3767 in 57 leukaemia lines

• part of →BMI1, CDKN2A and CDK6 are one circuit, not three genes - and in KMT2A-rearranged disease it has a loop nobody has measured

  the kinase arm of the circuit, and its feedback node

• related to →Venetoclax in KMT2A-rearranged leukemia — the apoptotic partner every other drug needs

  same study reports menin-inhibitor synergy with both abemaciclib and venetoclax in MLL1-r and NPM1c AML

• related to →Revumenib (SNDX-5613, Revuforj) — first approved menin–KMT2A interface inhibitor

  menin inhibition attenuates CDK6 levels and synergises with abemaciclib in MLL1-rearranged AML

• ← related toSUPERSEDED IN PART, CYCLE 56 - the MI-2-2 response is ~9x fusion-selective (that stands), but 'does not touch HOXA or MEIS1' was a fact about a weak compound: VTP-50469 collapses MEIS1 -5.25 log2 in the same line

  one of the few canonical targets that does fall: -0.41 on the top probe, -1.74 on 235287_at, -0.76 in combination, and flat in wild-type

• ← related toInstrument check: the top thirty RS4;11 MLL1-N 2 kb bins recovered from a deposited .tdf are JMJD1C (259x the genome median), ZMYND11, BAHCC1, CDK6, MYB, RUNX2, IGF2BP2, FOXP1, SATB1 and ARID1B - canonical KMT2A::AFF1 targets, annotated only against the series' own RefGene windows

  CDK6's TSS window carries the tenth-strongest RS4;11 MLL1-N bin at 99.5-fold the genome median, and its MLL1-N share ratio is 0.222 against menin's 0.045

• ← related toUnder EPZ5676 the canonical KMT2A-fusion targets look like they lose H3K79me2 far more than the genome (p=0.017 and p=0.0004), and against a baseline-matched null the entire effect vanishes (p=0.34 and p=0.22) - eight of the ten are simply top-decile genes

  highest baseline of the ten and the most completely stripped: 0.33% retained

• part of →The soluble-nucleoplasm versus chromatin-bound axis - the one localisation readout that keeps a size dimension, and nobody has run it on a KMT2A fusion

  CDK6 occupies the FLT3 and PIM1 promoters in FLT3-ITD AML while the bulk of cellular CDK6 is cytoplasmic and catalytically inactive; neither pool has been measured in a KMT2A-rearranged cell

• related to →CDKN2A - one gene, two reviewed UniProt records with the same primary symbol, and a deletion that is three times rarer here than in the infant leukaemias next door

  p16INK4a inhibits CDK6 and the two sit on one axis with BMI1; BMI1 and CDK6 are both exactly 326 aa, so this axis has no length defence against an accession swap