Pentimento and restoration — the fusion overpaints the cell, and cleaning reveals whatever was underneath
also called pentimento · conservation ethics · overpainting and the ground
Painters change their minds, and the earlier version slowly shows through the newer paint. A fusion protein paints a leukemia over a normal blood cell, and when a drug strips that paint away, what appears is whatever the painter started with.
The same thing, for a scientist
Conservation science separates three layers in a painting: the *ground* (the prepared surface, laid before any image), the *original image*, and *overpaint* added later. Pentimento is the earlier layer becoming visible through the newer one as the upper paint thins with age. The mapping to a KMT2A-rearranged blast is unusually clean. The ground is the DNA methylation landscape of the cell of origin, mitotically heritable and laid down before transformation; the image is the normal developmental program; the overpaint is the fusion-driven transcriptional program. Menin inhibition is a cleaning operation: it thins the overpaint. The conservator's question — what is revealed, and is it the original or something else — is exactly the clinical question of where a blast goes when the differentiation block is released, and the analogy makes the non-obvious prediction that the answer is written in the ground rather than in the overpaint, and is therefore already measurable at diagnosis.
Why the ground, and not the image, is the prediction1 deeper
Infant B-ALL including MLL-rearranged disease has an exceptionally silent DNA mutational landscape, with the discriminating signal methylomic and transcriptomic rather than mutational 1. Methylation is mitotically heritable and, in leukemia generally, retains a usable record of the originating cell: methylome-based cell-of-origin modelling maps individual CLL patients onto the normal B-cell differentiation continuum and separates disease-specific aberrant methylation from inherited physiological programming 2. That is precisely the ground-versus-overpaint distinction, implemented statistically.
The transcriptome, by contrast, is the overpaint. It is what the fusion is currently doing, and it will change the moment the drug acts. Predicting destination from the thing you are about to remove is the error the analogy identifies.
The observation that fits, and the one that does not
Fits: myeloid relapses after lineage switch in MLL/AF4 leukemia share oncogene fusion breakpoints with their matched lymphoid presentations and originate from a range of differentiation stages, from immature progenitors through committed B-cell precursors 3. A range of grounds, a range of destinations.
Does not fit, or at least complicates: the same work shows lineage switching involves substantial chromatin-accessibility change and transcriptional rewiring recurrently associated with altered chromatin modifiers, notably CHD4, and that perturbing CHD4 alone induces myeloid gene expression in MLL/AF4 models 3. That is an active repainting, not a passive revealing. If the destination is decided by an acquired chromatin lesion rather than by the inherited ground, the pentimento frame is wrong in an interesting and specific way.
The restoration ethics, which map onto a real therapeutic choice
Modern conservation is governed by two principles that translate directly.
Reversibility. A conservator only applies interventions that a future conservator can undo. The therapeutic counterpart is the difference between an occupancy inhibitor and a degrader: the first is reversible, the second is not. The field has largely assumed the degrader is better because it is more complete. The conservation principle points the other way, and the argument is not frivolous — if the destination revealed by removing the fusion is sometimes worse (myeloid switch), reversibility is a safety feature.
Minimal intervention. Remove only as much overpaint as necessary. The counterpart is fractional target occupancy, which the graph already flags as the relevant medicinal-chemistry variable for a buffered program, and which connects to hyp-agency-revocation-hysteresis-pulsed-menin.
Where the analogy is decoration rather than structure
The Ship of Theseus question — how much can change before it is no longer the original — is aesthetically satisfying here and generates nothing testable. It is not included in the mapping. Similarly, forgery and attribution have superficial appeal (identifying a cell of origin from unselected detail) but collapse into the existing hyp-methylome-of-origin-sets-fusion-target-set without adding a prediction, and are therefore left out.
The load-bearing content of this analogy is exactly one idea: the ground survives the overpaint and therefore predicts what a cleaning reveals. Everything else is atmosphere, and is labelled as such.
| domain | art |
|---|---|
| source concept | pentimento, the ground-image-overpaint stratigraphy of a painting, and the conservation principles of reversibility and minimal intervention |
| deliberately excluded | Ship of Theseus and forgery/attribution mappings were considered and excluded because they generate no prediction beyond what hyp-methylome-of-origin-sets-fusion-target-set already contains. |
mapping
[
{
"bio": "the DNA methylation landscape of the cell of origin, mitotically heritable and laid down before transformation",
"analog": "the ground — the prepared surface, laid before any image"
},
{
"bio": "the normal developmental transcriptional program of that cell",
"analog": "the original image"
},
{
"bio": "the fusion-driven HOXA9/MEIS1 program",
"analog": "overpaint applied later by a different hand"
},
{
"bio": "menin inhibition thinning the fusion program",
"analog": "a cleaning operation that removes overpaint"
},
{
"bio": "the lineage a blast adopts when the differentiation block is released",
"analog": "the image revealed by the cleaning, which is determined by what lies beneath"
},
{
"bio": "occupancy inhibitor versus degrader",
"analog": "the conservation principle of reversibility — only do what a later hand can undo"
},
{
"bio": "fractional target occupancy in a buffered program",
"analog": "minimal intervention — remove only as much as necessary"
},
{
"bio": "CHD4-driven active chromatin reprogramming during lineage switch",
"analog": "not pentimento at all, but a second painter repainting — the observation that could falsify the frame"
}
]generated hypotheses
[ "hyp-pentimento-ground-predicts-post-drug-destination" ]
- If destination is set by the ground, then releasing the differentiation block in a cell whose ground is myeloid-primed produces a myeloid leukemia, and the drug has converted the disease rather than treated it.
- The reversibility argument cuts against the degrader strategy that much of the field, and parts of this graph, currently favour — and it does so on safety rather than efficacy grounds, which is harder to dismiss.
- A cleaning that goes too far destroys the painting. A therapy that removes the fusion program from a cell whose underlying developmental program is itself damaged may leave nothing viable, which is one account of the profound cytopenias seen with this class.
- Does a fetal haematopoietic methylation reference exist at the resolution a Methyl-COOM-style model would need for infant KMT2A-rearranged ALL, or would it have to be generated?
- Is lineage switch a revealing of the ground or an active repainting driven by CHD4 and related chromatin modifiers? The two produce the same phenotype and have opposite implications for prediction.
- If reversibility is genuinely a safety feature, is the field's preference for degraders over occupancy inhibitors in this disease pointed in the wrong direction?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Genome-wide methylome and transcriptome analysis of 69 uniformly treated infants with de novo MLL-rearranged and non-MLL-rearranged B-ALL shows an exceptionally silent DNA mutational landscape consistent with prenatal origin, with methylome signatures mapping onto chromatin states related to enhancer and transcriptional control in normal haematopoietic cells and cleanly separating MLL-rearranged from non-MLL-rearranged cases.
Integrative methylome-transcriptome analysis unravels cancer cell vulnerabilities in infant MLL-rearranged B cell acute lymphoblastic leukemia · Tejedor JR, Bueno C, Cobo I, et al. · Journal of Clinical Investigation 2021What this evidence cannot show
Establishes that methylation carries the discriminating signal at diagnosis; it does not itself perform cell-of-origin inference, and it says nothing about behaviour under menin inhibition.
Evidence that argues against this (1)
- The analogy's load-bearing premise is that the ground is laid once and then stays put. Paired diagnosis-remission-relapse profiling in paediatric T-ALL — whole-exome, targeted ultra-deep sequencing and DNA methylation array — found relapse-specific promoter hypomethylation together with acquired mutations in epigenetic modulators (SUZ12, WHSC1, SMARCA4). The methylome is remodelled under therapy over exactly the interval during which the diagnostic 'ground' is supposed to remain a valid record. A conservator's ground is inert; this one is not. Caveat: T-ALL rather than KMT2A-rearranged B-ALL, n=13, array-based. Kunz JB et al. 2015
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Methylome-based cell-of-origin modelling (Methyl-COOM) treats the cancer epigenome as the sum of the pattern present in the tumour-initiating cell plus tumour-acquired changes, and uses a linear regression model to map individual CLL patients onto the continuum of normal B-cell differentiation, separating disease-specific aberrant methylation from physiological B-cell programming; the authors propose it as generalisable to other malignancies.
Methylome-based cell-of-origin modeling (Methyl-COOM) identifies aberrant expression of immune regulatory molecules in CLL · Wierzbinska JA, Toth R, Ishaque N, et al. · Genome Medicine 2020What this evidence cannot show
Developed and validated in CLL, a mature B-cell malignancy with a well-mapped normal differentiation reference. Transferring it to fetal-origin infant leukemia requires a fetal haematopoietic methylation reference that may not exist at the needed resolution.
Evidence that argues against this (1)
- WHERE THE ANALOGY BREAKS, in the source domain. Conservation science does not in fact enjoy clean layer separation, which is the property the analogy imports. Gooch and Tumblin, working specifically on visualising pentimenti, report that X-ray interpretation is difficult because X-rays superimpose features from all paint layers and because image intensity depends jointly on paint-layer thickness and each pigment's opacity — so the earlier layer is recovered by user-guided inference, not read off. Methyl-COOM has the same shape: a regression that assumes the observed methylome decomposes into an inherited component plus tumour-acquired change, fitted and validated in CLL — an indolent mature B-cell malignancy with a well-characterised normal differentiation continuum to regress against. Infant KMT2A::AFF1 has neither. No published application of Methyl-COOM to an acute leukaemia was found. Verification note: the Gooch and Tumblin record was confirmed from the publisher's listing and abstract (title, authors, journal, volume/issue, year, DOI); full text was paywalled, so only the abstract's own statements are used. Gooch AA et al. 2007
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Myeloid relapses after lineage switch in MLL/AF4 leukemia shared oncogene fusion breakpoints with their matched lymphoid presentations and originated from a range of differentiation stages from immature progenitors through committed B-cell precursors; switching involved substantial chromatin-accessibility change and transcriptional rewiring including alternative splicing, recurrently associated with altered expression, splicing or mutation of chromatin modifiers notably CHD4, and perturbing CHD4 alone or with other mutated epigenetic modifiers induced myeloid gene expression in MLL/AF4 models.
Epigenetic regulator genes direct lineage switching in MLL/AF4 leukemia · Tirtakusuma R, Szoltysek K, Milne P, et al. · Blood 2022What this evidence cannot show
This is the principal evidence against the passive-revealing version of the analogy: it describes active reprogramming driven by acquired chromatin-modifier lesions, not a ground showing through.
Evidence that argues against this (1)
- Read against the analogy, this study says the destination is written at the time of switching, not at diagnosis: the myeloid phenotype was recurrently associated with altered expression, splicing or mutation of chromatin modifiers, notably CHD4, and perturbing CHD4 alone induced myeloid gene expression in MLL/AF4 models. That is not pentimento — an older layer showing through as the upper one thins — it is a new lesion acquired late. The analogy's distinctive, non-obvious prediction (the answer is already measurable at diagnosis) is the part that this result most directly undercuts. Tirtakusuma R et al. 2022
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Genome-wide analysis of infant B-ALL identified the cellular origin and prognostic markers of the disease, including that reciprocal AFF1::KMT2A expression in 45% of t(4;11) cases predicted 4-year event-free survival of 62.4% versus 11.7%.
Unraveling the cellular origin and clinical prognostic markers of infant B-cell acute lymphoblastic leukemia using genome-wide analysis · Agraz-Doblas A, Bueno C, Bashford-Rogers R, et al. · Haematologica 2019What this evidence cannot show
Cell-of-origin inference here is transcriptome-based rather than methylome-based, which is exactly the comparison the generated hypothesis proposes to make.
Evidence that argues against this (1)
- Note the tension inside the entry's own evidence: the strongest published prognostic discriminator cited here for infant B-ALL is expression of the reciprocal AFF1::KMT2A fusion — a genetic/transcriptional feature, not a methylomic one. Searched for a published head-to-head showing a diagnostic methylation cell-of-origin score outperforming transcriptomic or genetic markers for outcome or lineage destination in KMT2A-rearranged leukaemia; none found. So the analogy's comparative claim (the ground beats the overpaint as a predictor) has no supporting comparison anywhere in this disease, and the one head-to-head-ish datum in the entry points the other way. Recorded so the gap is known to have been checked. Q12-analogy-counter (search record) 2026
the 'ground' is not inert: relapse-specific promoter hypomethylation and acquired epigenetic-modifier mutations remodel the methylome over the interval the diagnostic score must span
the methylome as the ground beneath the fusion overpaint
- analogous to →Lineage switch in KMT2A-rearranged leukemia — the same clone changes what kind of blood cell it is
lineage switch as either pentimento or a second painter, which the analogy cannot itself decide
- related to →The differentiation block — what it actually is, operationally, in KMT2A::AFF1 leukaemia
the block as the overpaint that a cleaning removes
generated by the ground-image-overpaint stratigraphy