Cardiac troponin release on MCL-1 inhibition - reported for two structurally distinct compounds and described by one developer as a likely CLASS effect, which makes it an essential-function ceiling rather than a chemistry problem
also called MCL-1 inhibitor cardiotoxicity · troponin elevation BH3 mimetic · AZD5991 troponin · ABBV-467 troponin
Two different drugs that block MCL-1 both made a blood marker of heart-muscle damage go up in patients, even when nothing else about the heart looked wrong. Because the two drugs are chemically unrelated, the likeliest explanation is that heart muscle needs MCL-1 - which would mean the problem belongs to the target, not to the drug.
The same thing, for a scientist
Two first-in-human studies of structurally distinct selective MCL-1 inhibitors report cardiac troponin elevation. **ABBV-467**: increases in plasma cardiac troponin in **4 of 8 patients** with relapsed/refractory multiple myeloma, without other corresponding cardiac findings, with the authors concluding that the selectivity of ABBV-467 suggests treatment-induced troponin release is a consequence of MCL-1 inhibition and **may represent a class effect of MCL-1 inhibitors in human patients** [E1]. **AZD5991**: asymptomatic elevations of troponin I or T in 8 of 78 patients (10.3%), with a post hoc analysis finding elevated troponin T in 14/31 patients **before** any dose and in 54/65 after any dose at or after cycle 1; four deaths from adverse events of which only tumour lysis syndrome was drug-related; three objective responses, all in myelodysplastic syndrome; and an overall conclusion of high laboratory troponin elevation with a **low overall response rate** [E2]. This risk entry exists because `gene-mcl1` records MCL-1 as a real, KMT2A-linked, infant-relevant dependency, and the honest counterweight is that the target has twice failed to convert in humans with a signal pointing at the target itself.
Why 'class effect' is the load-bearing word
A toxicity attributable to a compound can be engineered around: change the scaffold, change the exposure, change the schedule. A toxicity attributable to the target cannot. The ABBV-467 authors make the target argument explicitly on the basis of their compound's selectivity 1, and the independent AZD5991 experience is consistent with it 2.
The biological prior supports them: MCL-1 is required for cardiomyocyte survival and mitochondrial function in mouse genetics, and cardiomyocytes are post-mitotic cells with no capacity to replace losses. This entry does not cite that mouse literature because this beat did not retrieve and read it - which is itself worth stating, because the mouse cardiac-MCL1 papers are the evidence that would settle whether this is an essential-function ceiling, and reading them is the single highest-value follow-up in this entry.
What is genuinely uncertain is whether asymptomatic troponin release predicts clinical cardiac injury. In AZD5991 the elevations were asymptomatic; in ABBV-467 there were no other corresponding cardiac findings 12. A biomarker that moves without a clinical event is a signal, not an outcome - but in a drug given to an infant for years of follow-up, a signal of post-mitotic cell loss is not something to discount.
The complication nobody should skip: baseline troponin
The AZD5991 post hoc analysis found elevated troponin T in 14 of 31 patients before any dose of the drug 2. Relapsed/refractory haematological malignancy patients are frequently anaemic, frequently anthracycline-exposed, and frequently in renal impairment - all of which raise troponin independent of any study drug.
So the AZD5991 troponin data, read carefully, are weaker evidence of drug attribution than the headline suggests, and the entry's confidence rests more on ABBV-467, where 4 of 8 is a small n but a clean population contrast was drawn by the authors themselves 1. Two weak-to-moderate signals in the same direction from unrelated chemistry is the actual state of the evidence, not one strong signal.
What this means for the infant KMT2A-rearranged proposal specifically
This graph's MCL-1 interest comes from infant KMT2A-rearranged ALL, where the effective preclinical regimen is MCL-1 inhibitor plus venetoclax plus dexamethasone - and where the MCL-1-plus-venetoclax doublet alone killed 8 of 12 mice from tumour lysis syndrome. Adding a possible on-target cardiac ceiling to that gives a target whose:
- single-agent in vivo activity is nil,
- obvious doublet is acutely lethal in mice,
- working triplet requires a steroid in a population where steroid exposure is already a concern, and
- class carries a human cardiac biomarker signal the developers attribute to the target.
That is not an argument for abandoning MCL-1. It is an argument that the first clinical question in this disease is a therapeutic-index question, not an efficacy question, and that a paediatric protocol would need cardiac biomarker monitoring built in from the first patient.
| kind | essential-function |
|---|---|
| developer conclusion | ABBV-467 authors state troponin release may represent a class effect of MCL-1 inhibitors in human patients |
| severity | unresolved - troponin elevation was asymptomatic and without other cardiac findings in both studies; clinical significance unknown |
| confounder | 14/31 AZD5991 patients had elevated troponin T BEFORE any dose, which substantially weakens drug attribution in that study |
| contrary preclinical | MIK665 in paediatric AML models reported minimal impact on cardiac tissue pathophysiology (mouse histopathology, not a human biomarker) |
| second distinct risk | tumour lysis syndrome - 8/12 mice died on the MCL1-plus-venetoclax doublet in infant KMT2A-r ALL models; abolished by adding dexamethasone |
| mitigation | cardiac biomarker monitoring from the first patient; intermittent rather than continuous exposure; combination-first design with tumour-lysis prophylaxis and ramp-up |
| reading queue | the mouse cardiomyocyte MCL1 genetics literature was NOT retrieved by this beat and is the evidence that would establish whether this is a true essential-function ceiling |
affects
[ "gene-mcl1", "target-mcl1-with-bcl2-and-dexamethasone-infant-kmt2a-all" ]
agents with the signal
[ "ABBV-467 (4/8 patients, multiple myeloma)", "AZD5991 (8/78, 10.3%, haematological malignancies)" ]
- This is the essential-function case, not an off-target case. If MCL-1 is required by cardiomyocytes, no amount of compound optimisation removes the toxicity, and every MCL-1 programme inherits it.
- Cardiomyocytes are post-mitotic. Losses are not replaced, so a cumulative toxicity in a one-year-old with a normal life expectancy after cure is a different proposition from the same exposure in a heavily pretreated adult with months to live - and the adults are the only people in whom it has been measured.
- The population that would receive this has already had anthracyclines, which is the classic cause of late cardiac failure in childhood cancer survivors. Adding a second, mechanistically distinct cardiac insult to that population is a late-effects decision as much as an oncology one.
- The acute risk is separate and immediate: the MCL1-plus-venetoclax doublet killed 8 of 12 mice from tumour lysis syndrome in infant KMT2A-rearranged models. Efficacy and acute lethality may be inseparable at the doses that work.
- Does asymptomatic troponin release on MCL-1 inhibition progress to clinical cardiac dysfunction with longer exposure? Both studies are short dose-escalations in adults with limited survival, which is the worst possible setting for detecting a slow cardiac outcome.
- Is cardiomyocyte MCL-1 dependence established in mouse genetics at a level that makes this an unavoidable ceiling? That literature exists and this beat did not read it; reading it is the decisive step.
- Can intermittent dosing separate leukaemic killing from cardiomyocyte loss? Leukaemic blasts divide and cardiomyocytes do not, which is a real biological handle on scheduling that nobody appears to have exploited.
- Should a paediatric MCL-1 protocol include anthracycline-exposure stratification? Infants with KMT2A-rearranged leukaemia receive anthracyclines, and prior cardiac injury is exactly the covariate that confounded the adult troponin data.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In a first-in-human dose-escalation study of the highly potent and selective MCL-1 inhibitor ABBV-467 in patients with relapsed/refractory multiple myeloma (NCT04178902), treatment resulted in disease control in 1 patient and caused increases in plasma cardiac troponin levels in 4 of 8 patients, without other corresponding cardiac findings; the authors conclude that the selectivity of ABBV-467 suggests treatment-induced troponin release is a consequence of MCL-1 inhibition and therefore may represent a class effect of MCL-1 inhibitors in human patients.
Selective MCL-1 inhibitor ABBV-467 is efficacious in tumor models but is associated with cardiac troponin increases in patients. · Yuda J, Will C, Phillips DC, ..., Souers AJ · Commun Med (Lond) 2023What this evidence cannot show
Abstract read this session. **n = 8 patients**, all adults with relapsed/refractory multiple myeloma - not leukaemia, not paediatric, not KMT2A-rearranged. Troponin rose without other cardiac findings, so this is a biomarker signal of unknown clinical consequence rather than a documented cardiac event. The class-effect conclusion is an inference the authors draw from their compound's selectivity, not a demonstrated cross-compound comparison. The trial registry identifier NCT04178902 appears in the abstract and was not independently verified against ClinicalTrials.gov by this beat.
Evidence that argues against this (1)
- A preclinical study of the MCL1 inhibitor MIK665 in paediatric AML models reported minimal impact on cardiac tissue pathophysiology, so the cardiac signal is not universal across compounds and species - although a mouse histopathology negative is weak evidence against a human biomarker positive, and the two studies measure different things. Connerty P et al. 2026
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In a phase 1 first-in-human study of AZD5991 (monotherapy n=61, venetoclax combination n=17) in relapsed/refractory haematological malignancies, the most common adverse events were diarrhoea (59.0%), nausea (55.1%) and vomiting (47.4%); four deaths occurred due to adverse events (cardiac arrest, sepsis, tumour lysis syndrome, acute respiratory failure) of which only tumour lysis syndrome was related to AZD5991; dose-limiting toxicities occurred in five patients; three patients with myelodysplastic syndrome achieved an objective response; asymptomatic elevations of troponin I or T occurred in 8 (10.3%) patients; a post hoc retrospective analysis found elevated troponin T in 14/31 patients before any AZD5991 dose and in 54/65 patients after any dose at or after cycle 1; no associations were found between elevated troponin and cardiovascular risk factors; the authors conclude that treatment was associated with a high incidence of laboratory troponin elevation and a low overall response rate.
A Phase 1 First-in-Human Study of the MCL-1 Inhibitor AZD5991 in Patients with Relapsed/Refractory Hematologic Malignancies. · Desai P, Lonial S, Cashen A, ..., Blachly JS · Clin Cancer Res 2024What this evidence cannot show
Abstract read this session. Adults with relapsed/refractory disease; no paediatric, infant or KMT2A-rearranged subgroup is reported in the abstract. **The baseline finding substantially weakens drug attribution**: 14 of 31 patients had elevated troponin T before any dose, and this population is routinely anthracycline-exposed, anaemic and renally impaired, all of which raise troponin. The three responses were all in myelodysplastic syndrome, so the low response rate is not specifically an acute-leukaemia result. One death was cardiac arrest, judged not drug-related; that judgement was made by investigators and is not independently verifiable from the abstract.
Evidence that argues against this (1)
- A low response rate in heavily pretreated adults given a single agent is a weak test of the target in a disease where the preclinical evidence says MCL-1 inhibition does nothing as monotherapy and works only in a triple combination with venetoclax and a glucocorticoid. The clinical failure so far is a failure of the monotherapy design as much as of the target. Karsa M et al. 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In preclinical models of infant KMT2A-rearranged ALL, twice-weekly S63845 at 12.5 mg/kg or higher combined with venetoclax induced lethal tumour lysis syndrome in 67% of treated mice (8 of 12), with histopathological evidence of debris accumulation in kidney and lung, whereas tumour lysis syndrome was not observed with either monotherapy or with concurrent dexamethasone.
BCL-2 and MCL1 co-inhibition enhances glucocorticoid therapy in preclinical models of infant KMT2A-rearranged acute lymphoblastic leukaemia. · Karsa M, Karsa A, Spurling D, ..., Somers K · Br J Haematol 2026What this evidence cannot show
Full text read this session from Europe PMC (PMC13461956; body present). The PubMed record returns an empty abstract, so an abstract-only reading of this paper yields nothing rather than something wrong. Twelve mice is a small experiment, and tumour lysis syndrome in immunodeficient mice bearing high-burden ALL xenografts is not straightforwardly translatable to a child receiving stepwise dose ramp-up and tumour-lysis prophylaxis, which is standard with venetoclax. Cited here as an acute-toxicity risk distinct from the cardiac one, not as evidence about troponin.
Evidence that argues against this (1)
- Tumour lysis syndrome is a consequence of effective and rapid leukaemic cell killing and is routinely managed clinically with ramp-up dosing, hydration and rasburicase; in the same study, adding dexamethasone abolished it. It is therefore a manageable schedule problem rather than a target ceiling, unlike the cardiac signal. Desai P et al. 2024
on-target cardiomyocyte consequence of MCL-1 inhibition
- confers resistance to →MCL1 - named in 34 entries of this graph with no entry of its own until now; a mitochondrial survival protein high in infant KMT2A-rearranged ALL, druggable, and so far undruggable in patients
on-target cardiac biomarker signal reported for two structurally distinct MCL-1 inhibitors
- confers resistance to →MCL1 plus BCL-2 plus dexamethasone: a triple combination that matched induction chemotherapy in infant PDX models
class ceiling and acute tumour-lysis risk bounding the infant combination proposal
the cell the risk is about: plasma cardiac troponin is cardiomyocyte content that has left cardiomyocytes, raised in 4 of 8 patients on a selective MCL-1 inhibitor and called a probable class effect by the developer