Siremadlin

Siremadlin (HDM201) is Novartis's oral MDM2 inhibitor, designed to reactivate p53 - the cell's main tumor-suppressor protein - in cancers that still carry a working copy of the p53 gene. The drug is in Phase 1/2 across multiple programs, with the lead opportunities in acute myeloid leukemia (AML, in combination with venetoclax and azacitidine) and p53 wild-type soft tissue sarcomas, particularly the well-differentiated/dedifferentiated liposarcoma (WDLPS/DDLPS) subset where MDM2 is amplified in >90% of cases [1,9]. No MDM2 inhibitor has ever been approved despite 20+ years of industry effort, so siremadlin is fighting both a scientific and a commercial graveyard [1].

Siremadlin is a novel small molecule that has never been approved anywhere. The node here is tied to NCT05180695, a Phase 1/2 combination with pazopanib in p53 wild-type advanced soft tissue sarcomas (n=47, active, not recruiting) run by Centre Léon Bérard as an investigator-led study [2]. Novartis itself is sponsoring the higher-value readouts: NCT05155709, a Phase 1 combination with venetoclax plus azacitidine in chemo-ineligible AML (completed, n=14) [3], and Phase 1b data with ribociclib in liposarcoma published in 2022 [4]. First-in-human monotherapy data in advanced TP53 wild-type solid tumors and acute leukemia were published in 2022 by Stein and colleagues, establishing dose, pharmacology, and the on-mechanism toxicity profile [1]. Translational PK/PD modeling supporting intermittent dosing schedules was published by Guerreiro et al. in 2021 [5]. No FDA breakthrough therapy, fast track, or orphan designations have been disclosed for siremadlin to date. Expected timing for a registrational readout is unclear - Novartis has not committed to a Phase 3 start publicly, which itself is informative given the program has been running since the late 2010s. Realistic next inflection: AML combination efficacy data and a go/no-go decision on registration-enabling design.

Think of p53 as the cell's panic button - when DNA gets damaged or something goes wrong, p53 either pauses the cell to let it repair or tells it to die. MDM2 is the off-switch for that panic button: it tags p53 for destruction by the proteasome (the cell's protein shredder) and exports it out of the nucleus where it can't do its job [6]. About half of human cancers solve the p53 problem by mutating the gene itself; the other half keep p53 functional but amplify or overexpress MDM2 to keep it suppressed. Block MDM2, free p53, and tumor cells that still have a working copy should hit the brakes or die. The genetic case is strong - Li-Fraumeni syndrome patients with germline p53 mutations get cancers everywhere, so p53 loss is clearly causal [7]. The patient population question is more favorable than it looks: in well-differentiated and dedifferentiated liposarcoma, MDM2 amplification is the defining genetic event in >90% of tumors, which is why the class keeps testing there [9]. In AML, ~90% of patients are TP53 wild-type at diagnosis (TP53 mutations cluster in complex-karyotype and secondary AML) [10]. The pharmacological case is weaker. Multiple MDM2 inhibitors - Roche's idasanutlin, Daiichi/Rain's milademetan, Kartos's navtemadlin - have run Phase 2/3 trials over the last decade and failed to convert biology into durable clinical benefit. The mechanism works on paper. The drugs haven't worked in patients.

NCT05180695 is the trial driving this node: a Phase 1/2 dose-escalation and expansion of siremadlin plus pazopanib (a VEGFR-targeted tyrosine kinase inhibitor approved for sarcoma) in p53 wild-type advanced or metastatic soft tissue sarcoma, n=47, sponsored by Centre Léon Bérard rather than Novartis directly [2]. Primary endpoint for dose escalation is maximum tolerated dose; the expansion looks for efficacy signal. Patient selection requires confirmed TP53 wild-type status, which is the right move - MDM2 inhibitors cannot work in p53-mutant tumors and prior class failures partly traced to inadequate selection. WDLPS/DDLPS is the natural enrichment indication given MDM2 amplification rates >90% [9]. The bigger Novartis-sponsored bet is NCT05155709 in AML with venetoclax/azacitidine (Phase 1, completed, n=14) [3], a logical combination because venetoclax-azacitidine is the current standard for chemo-ineligible AML (VIALE-A CR+CRi ~66%, median OS 14.7 months) and adding p53 reactivation could deepen responses [11]. Sample sizes are small, no randomized comparator, and no overall survival endpoint at this stage - these are signal-finding studies, not registration trials. Enrollment is closed or near-closed. The trial design is rational for a Phase 2 program but does not on its own answer the question that has killed every prior MDM2 inhibitor: does the efficacy signal beat the thrombocytopenia.

Our model puts the chance this drug is eventually approved at 7%. That figure starts from a historical baseline of about 13% for Phase 2 drugs in this area, then shifts up or down based on ten facts about the trial and its sponsor. Two things help the estimate: a non-randomized design and light or open-label blinding; two things hurt it: few secondary endpoints and weak or limited earlier-phase results. The remaining facts land near average and leave the number close to where the base rate put it.

Safety risk is concrete and on-mechanism: thrombocytopenia. Activating p53 in bone marrow megakaryocytes (the large precursor cells that produce platelets) kills them, platelets crash, you get bleeding risk and dose interruptions. Every MDM2 inhibitor has hit this. Siremadlin's first-in-human paper documented dose-limiting thrombocytopenia and neutropenia and required intermittent dosing schedules to manage it [1,5]. Efficacy risk is the bigger problem: a working mechanism that has never converted to approval despite multiple sponsors, indications, and combinations. Idasanutlin (Roche) failed the Phase 3 MIRROS trial in relapsed/refractory AML - the combination with cytarabine improved response rate over cytarabine alone but did not extend overall survival, the regulatory endpoint that matters [12]. Milademetan (Rain Oncology) failed the Phase 3 MANTRA trial in dedifferentiated liposarcoma in 2023 and the company subsequently restructured. Kartos's navtemadlin and Boehringer's BI-907828 (brigimadlin) remain in active development with potential readouts in 2026 - a positive class readout from either could validate the mechanism and accelerate Novartis's Phase 3 decision; another failure would likely end internal confidence in the target. The pattern suggests something structural - likely the therapeutic window between p53 activation in tumor versus marrow is too narrow for sustained dosing. Execution risk is moderate: the sarcoma trial is investigator-sponsored and small, and Novartis has not signaled a registrational Phase 3 design. Commercial risk if approved: payers will demand head-to-head benefit over venetoclax-azacitidine alone in AML, and sarcoma is a small market (US AML incidence ~20,000/year, of which roughly half are chemo-ineligible older adults; DDLPS incidence ~1,000-2,000/year) with pazopanib already generic.

Worth watching, but skeptically. The signal-worth-checking is the AML combination with venetoclax-azacitidine - that is where the mechanism has the best therapeutic logic and the standard of care has the most room for deepening. A composite complete response (CR+CRi) rate meaningfully above the ~66% venetoclax-azacitidine doublet historical from VIALE-A [11], paired with a tolerable platelet profile, would be the data point that distinguishes siremadlin from the rest of the class. The sarcoma readouts are smaller signal - pazopanib combinations rarely produce registrational results in unselected disease, and DDLPS is the indication where the prior class failure (milademetan MANTRA) was clearest. Novartis posted approximately $50.3B in 2024 net sales from continuing operations [8] and can absorb continued development cost, but the company has been disciplined about cutting underperforming oncology assets, so silence past 2026 should be read as bearish. Check back at ASH 2026 (American Society of Hematology annual meeting, December) for AML combination data; check back at ASCO 2026 (American Society of Clinical Oncology, June) and ESMO 2026 (European Society for Medical Oncology, October) for sarcoma combination data. A class-level readout from navtemadlin or BI-907828 in 2026 is a major binary catalyst for the entire program. If both siremadlin readouts pass quietly, no Phase 3 is announced, and a competitor reads out negatively, this becomes a slow phase-out. The model's 23.9% feels right. The class needs one drug to actually work before any of these will be worth more than option value.