Oncofusion

start / graph / risk-an-h3k9-axis-inhibitor-given-with-a-top2-poison-removes-a-repair-licence

Riskmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q32-brca1-hr-licensing

Giving a G9a/GLP inhibitor alongside etoposide or doxorubicin removes a DNA-repair licence at the moment TOP2 breaks are being made - and the sibling compound UNC-0638 is measured to block BRCA1/BARD1 arrival at a break, in children on curative therapy

also called G9a inhibitor plus etoposide repair liability · UNC0642 removes an HR licence · the prophylaxis inverts sign during chemotherapy

This graph proposes giving children a drug that strips a silencing tag off DNA, alongside chemotherapy, to stop a leukaemia-causing break from happening. Three separate papers now show that removing that same tag stops the cell's repair machinery from reaching a break - so the drug might prevent one break and leave every other break unrepaired.

The same thing, for a scientist

**The proposal at risk.** `mechanism-h3k9me1-2-ctcf-balance-gates-kmt2a-rearrangement` establishes that raising H3K9me1/2 across the *KMT2A* breakpoint cluster region lowers CTCF occupancy at the exon-11 element and makes the locus amplify and break, and that doxorubicin and etoposide enter that chain by lowering KDM3B and CTCF. The proposal that follows is **prophylactic G9a/GLP inhibition** - `drug-unc0642` - co-administered with anthracycline or epipodophyllotoxin to keep the mark low and the anchor occupied. Several risk entries already in this graph attack that proposal on off-target breadth, germline exposure, the treatment window, and whether the mark is even present. **This one attacks it on repair, and unlike the others it rests on a direct pharmacological measurement.** **The measurement.** In U2OS cells with induced breaks, *'inhibitors blocking H3K9 methylation, including **UNC-0638**, chaetocin, and **Bix-01294**, disturbed the immediate accumulation of H3K9me3, HP1-gamma, and BRCA1/BARD1 but not CTCF at the DSBs'* [E1, Supplementary Fig. S5E-G]. **UNC-0638 is the immediate chemical sibling of UNC0642** - the same G9a/GLP quinazoline series from the same chemical-probe programme. So the proposal's own pharmacology is reported to block BRCA1/BARD1 arrival at a double-strand break. **Two independent lines give the same sign by different mechanisms, which is what lifts this above a single-paper worry.** - **G9a-RPA, 2017.** Loss of G9a impairs DNA damage repair and **enhances the sensitivity of cancer cells to radiation and chemotherapeutics**; CK2 phosphorylates G9a at Ser211 on break induction, chromatin-enriched G9a binds RPA directly and promotes RPA and RAD51 loading, facilitating HR [E2]. The authors offer this as a rationale *for* G9a inhibitors - chemosensitisation is the intended effect in a tumour, and it is precisely the unintended effect in a child's normal progenitors. - **EHMT1/2 in ovarian carcinoma, 2019.** Disruption of EHMT1/2 **ablates both homologous recombination AND non-homologous end joining**, with large increases in gammaH2AX [E3]. If both pathways go, a TOP2 break has no accurate route and no fast route. **And there is a human state that resembles the proposed one.** Germline BRCA1 mutation plus topoisomerase II-directed therapy has produced therapy-related acute leukaemia with a TOP2-type breakpoint [E4], and 10 of 47 (21%) breast-cancer survivors with therapy-related leukaemia carry a deleterious germline repair-gene mutation [E5]. A drug that transiently phenocopies impaired BRCA1/BARD1 loading, given for the duration of etoposide exposure, is a pharmacological approach to a genotype with an observed leukaemogenic outcome. **GRADED HONESTLY, because the inference is long.** The cascade is **U2OS and HEK-293T**, with **enzyme-, laser- and phleomycin-induced** breaks; etoposide appears in that paper only in co-immunoprecipitations, not in the recruitment imaging or the repair assays. The HP1 paralogue is **HP1-gamma**, and this graph's H3K9 work is about HP1-alpha and about H3K9me1/2 rather than me3. [E2]'s cells are unnamed 'cancer cells' and its mechanism is an RPA contact rather than a mark. [E3] is adult ovarian carcinoma under a PARP inhibitor. [E4] is n = 2 adults, neither KMT2A-rearranged. **Nothing here is in a haematopoietic cell, an infant, or at the *KMT2A* locus.** The chain from these results to a child on an infant-ALL protocol has at least four unbridged steps, and the honest severity is **moderate-to-high in consequence and low in demonstrated probability**. **What it changes now, with no new data.** The prophylaxis proposal must not be written up, presented or costed without this liability stated next to it, and any preclinical plan for it must include a repair arm. That costs nothing and is the actionable half.

Go deeper
The one sentence, and why it is in a supplementary figure

'Consistently, inhibitors blocking H3K9 methylation, including UNC-0638, chaetocin, and Bix-01294, disturbed the immediate accumulation of H3K9me3, HP1-gamma, and BRCA1/BARD1 but not CTCF at the DSBs (Supplementary Fig. S5E-G).' 1

That sentence appears nowhere in the paper's title, abstract or keywords. No PubMed or Europe PMC query built from UNC0638, G9a, etoposide or BRCA1 would surface it, because it is a supporting panel in a paper about CTCF. It was found by fetching the full text and reading it - which is rule 80 operating in the direction the rule is usually stated about, and it is the reason this liability was not already in this graph.

Two honest deflations of the sentence:

  • Chaetocin is in the list and should not be. It is a nonspecific histone lysine methyltransferase inhibitor and a thioredoxin-reductase-1 inhibitor whose cytotoxicity is attributed to ROS (recorded already on gene-suv39h1). Its presence does not strengthen the trio; it means only two of the three arms are interpretable.
  • Both interpretable arms are G9a-directed. UNC-0638 and BIX-01294 are both G9a/GLP inhibitors. So the sentence establishes that blocking G9a blocks BRCA1/BARD1 loading. It does not establish that blocking SUV39H1 or SETDB1 would - and for this graph's proposal that is the relevant enzyme anyway, so the coincidence favours the risk rather than weakening it.
Three lines, three mechanisms, one sign
sourcesystemperturbationrepair outcome
1 2026U2OS, engineered breaksUNC-0638, BIX-01294H3K9me3, HP1-gamma and BRCA1/BARD1 accumulation at DSBs all disturbed; CTCF unaffected
2 2017cancer cells, DSBsG9a lossDNA damage repair impaired; sensitisation to radiation and chemotherapeutics; RPA and RAD51 loading reduced via a direct CK2-phospho-G9a-RPA contact
3 2019HGSOC lines + PDX, PARPiEHMT1/2 genetic or pharmacologic disruptionHR and NHEJ both ablated; large gammaH2AX increase; cells held in G1

The three mechanisms are different - a histone-mark licence, a direct RPA contact, and something the authors attribute to damage plus cell-cycle dysregulation. That is a weakness for mechanism and a strength for risk assessment. If three groups perturbing the same enzyme family in three systems by three routes all find repair worse, the direction is robust even though no single mechanism is confirmed.

The 2 framing deserves quoting because it inverts cleanly: the authors present their result as 'further rationale for the use of G9a inhibitors as a cancer therapeutic' - i.e. chemosensitisation is the selling point. In a tumour you want that. In the normal marrow of a child receiving etoposide, that is the harm.

What this does NOT establish, stated plainly
  • It does not show that UNC0642 causes leukaemia. No animal has been given a G9a inhibitor with a TOP2 poison and followed for leukaemia incidence, in any species.
  • It does not show the cascade runs in blood. Every mechanistic source is non-haematopoietic.
  • It does not show the liability exceeds the benefit, because the benefit has never been quantified either - the protective effect rests on cytogenetic endpoints in small groups in epithelium and mice.
  • It does not resolve the mark. This graph's rearrangement mechanism is about H3K9me1/2 and this cascade is reported for H3K9me2/me3; they overlap in G9a's product but are not the same claim.
  • It does not distinguish 'no repair' from 'different repair'. 3 reports both HR and NHEJ ablated; 1 reports HR specifically; and gene-kdm5a records a published line in which H3K4 methylation promotes RIF1-dependent NHEJ. A shift toward end joining at a translocation-prone locus would be worse than a uniform slowdown, and none of the sources distinguishes the two outcomes at a specific locus.

The defensible claim is narrow and still worth writing down: the compound class this graph proposes giving with a TOP2 poison has, in three systems, been shown to impair double-strand-break repair, and one of those results is on the recruitment step that the proposal's own target enzyme controls.

Facts and figures
kindessential-function
severityconsequence potentially severe (therapy-related leukaemia or increased genomic instability in a child receiving curative therapy); demonstrated probability in the relevant system LOW, because no evidence exists in any haematopoietic cell, any infant, or at the KMT2A locus
the proposal at riskprophylactic G9a/GLP inhibition (UNC0642 or a successor) co-administered with anthracycline or epipodophyllotoxin, to hold H3K9me1/2 low across the KMT2A breakpoint cluster region and keep CTCF bound at the exon-11 element, as follows from `mechanism-h3k9me1-2-ctcf-balance-gates-kmt2a-rearrangement`
the mechanism of the liabilityH3K9 methylation at a break is not only a silencing mark, it is a recruitment licence: SUV39H1/SETDB1 write it, HP1-gamma reads it, HP1-gamma's chromoshadow domain binds the BARD1 PxVxL motif at 569-573, and early BRCA1/BARD1 then recruits SMARCAD1 and EXO1/DNA2 for long-range resection. Blocking the writing step with a G9a/GLP inhibitor removes the licence without touching CTCF.
the timing argument which is the core of itThe proposal's whole point is to give the inhibitor DURING anthracycline or etoposide exposure, because that is when the locus is at risk. That is also, necessarily, when the genome carries its maximum burden of TOP2-mediated double-strand breaks. So the proposal schedules the repair-licence blockade to coincide exactly with the peak break load. There is no version of the proposal that separates the two in time without abandoning its rationale.
mitigationThree things, in increasing cost. (1) COSTS NOTHING: state this liability wherever the prophylaxis proposal is stated, and require any preclinical plan for it to carry a DNA-repair arm - RAD51 foci, DR-GFP, residual gammaH2AX - alongside its rearrangement endpoint. (2) CHEAP: run the direct test in a CD34+ progenitor or a B-lineage line, etoposide plus or minus UNC0642, repair readout. (3) DESIGN-LEVEL: if the liability is real, the only schedules that survive are ones that separate the inhibitor from the genotoxic exposure in time - which the proposal's own rationale forbids - or a delivery strategy restricted to a cell population not receiving the cytotoxic. Neither is currently specified.
cell typeNONE of the supporting evidence is in a haematopoietic cell. [E1] is U2OS osteosarcoma and HEK-293T; [E2] is unnamed cancer cells; [E3] is high-grade serous ovarian carcinoma lines and one PDX; [E4] and [E5] are adult humans after breast or ovarian cancer therapy. The population the risk is ABOUT is infants and young children with KMT2A-rearranged ALL on curative protocols containing etoposide, daunorubicin or doxorubicin.
cell stateDNA DAMAGE during S/G2. Two conditions compound here. The licence this risk is about operates on breaks in cycling cells, so the liability is largest in the proliferating compartments - which during induction chemotherapy includes normal haematopoietic progenitors and gut and hair epithelium as well as blasts. And [E3] reports that EHMT1/2 inhibition holds cells in G1, which would itself change how a co-administered cell-cycle-dependent cytotoxic behaves.
epigenetic readoutH3K9me2 and H3K9me3 at break-proximal chromatin, as a RECRUITMENT LICENCE rather than as a silencing mark. This is the distinction that makes the risk non-obvious: every other entry in this graph treats H3K9 methylation at this locus as a gate on rearrangement, and this entry treats the same mark at any break as a gate on repair. Both can be true simultaneously and they point in opposite directions for the same drug.
shown for fusionNO - nobody has looked, in any fusion. No experiment in any of the five sources involves a fusion protein, a leukaemia cell or the KMT2A locus, and no G9a-inhibitor-plus-genotoxic combination has been tested for repair competence in any haematopoietic system.
absence classnever-measured
affects
[
  "drug-unc0642",
  "drug-etoposide",
  "drug-doxorubicin"
]
why this is different from the risk entries already here
{
  "risk-a-kdm3b-or-g9a-intervention-moves-ctcf-at-most-anchors-in-the-genome": "off-target breadth - thousands of loci changed to protect one",
  "risk-the-h3k9me3-exoneration-of-g9a-inhibition-is-epithelial-only": "the exonerating measurement was in retinal pigment epithelium at n = 2",
  "other_existing_objections": "germline exposure at the wrong life stage; a closing treatment window; the mark may not be present at the locus at all; and the fusion-target-promoter arm of mark-h3k9me3 pointing the other way",
  "what_THIS_one_adds": "a direct pharmacological measurement on the proposal's own compound class - UNC-0638 blocking BRCA1/BARD1 accumulation at breaks - plus two independent functional lines and one human clinical analogue. The existing objections are about specificity and context; this one is about the drug doing a second thing, in the same cells, at the same time, that is harmful."
}
what would show this is NOT a problem
[
  "In a human CD34+ or B-lineage progenitor exposed to etoposide at a clinically achievable concentration, UNC0642 at its own effective concentration produces NO reduction in RAD51 foci, no reduction in DR-GFP HR efficiency and no increase in residual gammaH2AX at 24 h relative to etoposide alone. That is the direct test and it has never been run in any blood cell.",
  "OR: the G9a-dependence of BRCA1/BARD1 loading is shown to be catalysis-independent - a catalytically dead G9a rescues BRCA1/BARD1 accumulation - in which case a SET-domain inhibitor such as UNC0642 would not reproduce the knockdown phenotype and the liability would shrink to an off-target concern.",
  "OR: H3K9me2 (G9a's product) turns out not to be the mark the HP1-gamma step reads at a break, the relevant species being H3K9me3 written by SUV39H1 and SETDB1 - in which case a G9a-selective inhibitor might spare the licence. Note [E1] reports UNC-0638 and BIX-01294 disturbing H3K9me3 accumulation, which argues against this escape, but it was not tested with a SUV39H1-selective tool because none exists."
]
How this could bite us back
  • The proposal cannot be rescheduled out of the problem. Its rationale requires the inhibitor to be present exactly when the anthracycline or etoposide is, which is exactly when the genome carries its maximum TOP2 break load - so the repair-licence blockade and the peak break burden are the same window by construction.
  • The harm would be invisible in the proposal's own endpoint. The preclinical work behind the prophylaxis scores KMT2A locus rearrangement, not global repair fidelity or leukaemia incidence. A drug that halves rearrangement at this locus while impairing repair genome-wide would look like a success in every assay currently proposed.
  • The harm would also be invisible for years in a clinic. Therapy-related leukaemia after TOP2 exposure has a latency measured in years, and infant-ALL protocols already carry a therapy-related-malignancy risk. A trial powered for the intended benefit would be far too small and far too short to detect this.
  • And the reverse bite-back, which is the honest one: if this liability is real it also argues that H3K9 methylation at the KMT2A locus is PROTECTIVE during chemotherapy rather than permissive - which contradicts the mechanism the prophylaxis is built on. Both cannot be fully true at the same locus, and nobody has measured which holds there.
  • Finally, note that the KDM5A limb of the same cascade means pan-KDM5 inhibitors carry the same liability, and this graph has never counted them as a genotoxic-combination-relevant class.
Still open
  1. The direct test, never run in any blood cell: in human CD34+ progenitors or a B-lineage progenitor line, does UNC0642 at its effective concentration reduce RAD51 foci, reduce DR-GFP HR efficiency, or increase residual gammaH2AX at 24 h after a clinically achievable etoposide exposure, relative to etoposide alone?
  2. Is G9a's repair role catalytic? A catalytically dead G9a rescue experiment would decide whether a SET-domain inhibitor carries this liability at all, and it has not been done in either [E1]'s or [E2]'s system.
  3. Does H3K9-axis inhibition during a TOP2 exposure shift repair toward end joining rather than abolishing it? At a translocation-prone locus that would be the worse outcome, and no source here distinguishes the two.
  4. What is the base rate this risk should be measured against? Infant-ALL protocols already carry a therapy-related-malignancy burden, and no entry in this graph states it - without that number the incremental risk of the proposal cannot be sized even qualitatively.
Every source behind this entry (5)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In U2OS cells with induced double-strand breaks, inhibitors blocking H3K9 methylation - UNC-0638, chaetocin and Bix-01294 - disturbed the immediate accumulation of H3K9me3, HP1-gamma and BRCA1/BARD1 at the DSBs, but did not disturb CTCF. This sits inside the larger finding that CTCF recruits HP1-gamma, KAP1, SUV39H1, SETDB1 and KDM5A to breaks, that this repressive assembly licenses rapid BRCA1/BARD1 accumulation, and that failure of it abrogates EXO1/DNA2 recruitment, DNA end resection and homologous recombination.

    CTCF primes repressive chromatin for rapid BRCA1/BARD1 recruitment to drive homologous recombination repair of DNA double-strand breaks. · Hwang SY, Choi D, Choi MJ, Jo HR, Jeong JH, Kang MA, Lee JS · Nucleic Acids Research 2026
    PMID 42475683DOI
    What this evidence cannot show

    Confidence is deliberately MODERATE and not strong, although the paper's main findings are graded strong elsewhere in this beat. Reasons: the inhibitor panel is in Supplementary Fig. S5E-G with no quantification reproduced in the main text and none read here; no concentrations or exposure times for the three compounds are given in the sentence; the readout is immediate accumulation by imaging, not a repair-outcome measurement under drug; chaetocin is uninterpretable (nonspecific KMT inhibitor and thioredoxin-reductase-1 inhibitor) leaving two G9a-directed arms; and the system is U2OS osteosarcoma with FokI-, AsiSI-, laser- and phleomycin-induced breaks, with etoposide used only for co-immunoprecipitation. UNC-0638 is the chemical sibling of UNC0642, not the same compound. DOI taken from PubmedData/ArticleIdList only.

    Evidence that argues against this (1)
    • This graph's own H3K9 dispute runs the other way at fusion-target promoters: SUV39H1 behaves as a tumour suppressor in mouse MLL-AF9 AML, with over-expression lengthening latency and knockdown accelerating disease. If H3K9 methylation restrains the fusion programme, lowering it is dangerous for a second and unrelated reason - which does not rescue the prophylaxis, but does mean this entry is not the only objection and is not the strongest one in every context. Chu Y et al. 2020
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Loss of G9a impairs DNA damage repair and enhances the sensitivity of cancer cells to radiation and chemotherapeutics. In response to double-strand breaks, G9a is phosphorylated at serine 211 by casein kinase 2 and recruited to chromatin; chromatin-enriched G9a directly interacts with replication protein A and promotes loading of RPA and the RAD51 recombinase onto DSBs, facilitating homologous recombination and cell survival; the RPA-G9a interaction was confirmed to be critical for RPA foci formation and HR upon DNA damage. The authors present this as further rationale for using G9a inhibitors as cancer therapeutics.

    G9a coordinates with the RPA complex to promote DNA damage repair and cell survival. · Yang Q, Zhu Q, Lu X, Du Y, Cao L, et al. · Proceedings of the National Academy of Sciences 2017
    PMID 28698370DOI
    What this evidence cannot show

    Abstract read; full text not fetched in this beat, so the cell lines are unnamed here and no methods claim is made. The mechanism is a direct G9a-RPA protein contact, NOT a histone-mark licence - so this supports the DIRECTION of the risk and not [E1]'s mechanism, and crucially it leaves open whether a catalytic SET-domain inhibitor such as UNC0642 would reproduce a G9a-knockdown phenotype. 'Chemotherapeutics' is not resolved to TOP2 poisons in the abstract. No haematopoietic system.

    Evidence that argues against this (1)
    • If G9a's repair role is mediated by a protein-protein contact with RPA rather than by its methyltransferase activity, then UNC0642 - a substrate-competitive SET-domain inhibitor - might leave the contact intact and not impair repair at all. Nothing in this paper tests a catalytically dead G9a for RPA loading, and that single experiment would substantially resize this risk in either direction. Yang Q et al. 2017
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Genetic or pharmacologic disruption of EHMT1/2 (GLP/G9A) ablates BOTH homologous recombination and non-homologous end joining. In high-grade serous ovarian carcinoma, PARP-inhibitor-resistant cells display a global increase in H3K9me2 with over-expression of EHMT1/2, reproduced in a PARPi-resistant patient-derived xenograft; disrupting EHMT1/2 sensitises those cells to PARPi without increasing PARPi-induced apoptosis; functional DNA repair assays show that disruption ablates HR and NHEJ while immunofluorescence for phosphorylated H2AX shows large increases in DNA damage; and cell-cycle analysis shows EHMT1/2-inhibitor-treated cells remain in G1 while PARPi increases the S and G2 fraction.

    Histone methyltransferases EHMT1 and EHMT2 (GLP/G9A) maintain PARP inhibitor resistance in high-grade serous ovarian carcinoma. · Watson ZL, Yamamoto TM, McMellen A, Kim H, Hughes CJ, Wheeler LJ, et al. · Clinical Epigenetics 2019
    PMID 31775874DOI
    What this evidence cannot show

    Abstract read in full and the record fetched fresh from efetch; full text not read. Adult high-grade serous ovarian carcinoma, not haematopoietic and not paediatric; the genotoxic partner is a PARP inhibitor, not a TOP2 poison. The authors attribute sensitisation to DNA damage plus cell-cycle dysregulation and also report PI3K/AKT/mTOR transcriptional changes, so the repair phenotype is not cleanly attributable to a chromatin licence. The specific EHMT1/2 inhibitor used is not named in the abstract. IDENTIFIER NOTE: the DOI here was corrected in-run - a first draft carried 10.1186/s13148-019-0764-4 written from recall, which is wrong.

    Evidence that argues against this (1)
    • Ablating HR and NHEJ simultaneously plus arresting cells in G1 is a broad phenotype consistent with general toxicity rather than with a specific repair-licensing mechanism, and the authors offer cell-cycle dysregulation as a co-mechanism. A broad phenotype makes the mechanistic claim weaker and the safety concern stronger at the same time, and both readings are recorded rather than choosing the convenient one. Watson ZL et al. 2019
  4. E4weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    The human state this drug combination would transiently mimic has an observed leukaemogenic outcome: two hereditary breast-ovarian carcinoma patients with germline BRCA1 mutations developed therapy-related acute leukaemia following topoisomerase II-directed therapy - a t-APL with a RARA intron 2 breakpoint at chr17:40,347,487 corroborating breakpoint clustering after topoisomerase II inhibition, and a t-ALL - and the t-ALL case carried a BRCA1 mutation identical to a previously reported case after the same primary disease, which the authors argue makes chance unlikely.

    Characterization of therapy-related acute leukemia in hereditary breast-ovarian carcinoma patients: role of BRCA1 mutation and topoisomerase II-directed therapy. · Bagal B, Kumar R, Gaur T, Talreja V, Bonda A, Patkar N, Shetty D, Kowtal P, Subramanian PG, Gupta S, Sarin R, Hasan SK · Medical Oncology 2020
    PMID 32277283DOI
    What this evidence cannot show

    n = 2, adults, no comparison group, no denominator, neither leukaemia KMT2A-rearranged. Germline biallelic-pathway impairment over a lifetime is NOT equivalent to transient pharmacological blockade of an upstream chromatin step during a defined exposure - the analogy is directional and nothing more. Included because the risk entry would otherwise rest entirely on cell lines, and a reader deserves to know a human observation in this shape exists and how thin it is.

    Evidence that argues against this (1)
    • Germline BRCA1 carriers also develop therapy-related myeloid neoplasms after PARP inhibitors rather than TOP2 poisons, so the association is with impaired repair plus any genotoxic exposure rather than with TOP2 specifically - which broadens the concern but weakens the specific inference that a TOP2-poison combination is the uniquely dangerous one. Marmouset V et al. 2022
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Germline repair-gene impairment is over-represented among people who develop leukaemia after cytotoxic therapy: of 47 breast-cancer survivors with therapy-related leukaemia screened on a comprehensive susceptibility panel, 10 (21%) carried a deleterious inherited mutation - BRCA1 3, BRCA2 2, TP53 3, CHEK2 1, PALB2 1 - and 40 of 70 with an available family history (57%) had a close relative with breast, ovarian or pancreatic cancer.

    PMID 26641009DOI
    What this evidence cannot show

    Single-institution registry, n = 47 sequenced, adults. Rule 84 applies: the abstract supplies NO comparison group and no population base rate, so 21% cannot be converted to a relative risk and this beat did not supply the null. Three of the ten carriers have TP53, which is not an HR gene. The cytotoxic exposures are not resolved to TOP2 poisons.

    Evidence that argues against this (1)
    • With TP53 as common as BRCA1 in this series, the association is with cancer-susceptibility genotype broadly rather than with homologous recombination specifically, and a G9a inhibitor does not phenocopy TP53 loss. This weakens the mapping from the series to the proposed drug state. Churpek JE et al. 2016
Connected to