HER3-DXd

HER3-DXd (patritumab deruxtecan) is Daiichi Sankyo and Merck's HER3-targeting antibody-drug conjugate (ADC), built on the same deruxtecan (DXd) topoisomerase I inhibitor payload that drives Enhertu's commercial success. Merck paid $4B upfront plus $1.5B in continuation payments in October 2023 for ex-Japan rights to three Daiichi DXd ADCs - HER3-DXd, datopotamab deruxtecan (Dato-DXd), and ifinatamab deruxtecan - with up to $22B total including milestones; Daiichi retains Japan rights and manufacturing control [11]. The drug is being tested across solid tumors - KEYMAKER-U01 substudy (NCT06731907) pairs it with pembrolizumab in advanced NSCLC, while the key Phase 3 study (NCT07060807, n=1000) is enrolling in breast cancer [1][5]. TUXEDO-3 Phase 2 data published in Lancet Oncology in 2025 reported activity in patients with active brain metastases from both breast cancer and NSCLC, an unmet need where most systemic therapies fail [2][3]. HER3-DXd carries baggage: the FDA issued a Complete Response Letter (CRL) in June 2024 for the EGFR-mutant NSCLC accelerated approval submission over third-party manufacturing inspection issues, and the BLA (Biologics License Application - the FDA filing for a biologic drug) was withdrawn [4]. With Merck's $65B revenue base and a deep ADC bench, this asset matters strategically as part of the company's post-Keytruda thesis even if any single indication ends up modest commercially.

HER3-DXd is investigational with no approvals anywhere in the world. Daiichi Sankyo and Merck withdrew the BLA (Biologics License Application) in EGFR-mutant NSCLC after the FDA issued a Complete Response Letter in mid-2024; the CRL was driven by inspection findings at a third-party manufacturing facility, not by efficacy or safety data [4]. The companies are now running parallel Phase 2 and Phase 3 programs across multiple tumor types. KEYMAKER-U01 (NCT06731907) is a Phase 2 umbrella with pembrolizumab combinations in advanced NSCLC, n=90, ORR (objective response rate) primary endpoint [1]. The breast cancer Phase 3 (NCT07060807) is recruiting 1,000 patients with PFS (progression-free survival) as the primary endpoint and is the most consequential program in the franchise [5]. HERTHENA-PanTumor01 covered previously-treated advanced solid tumors as a basket Phase 2 (a single trial with multiple disease cohorts sharing one treatment) [7]. HR+/HER2- (hormone receptor positive, HER2 non-amplified) metastatic breast cancer Phase 2 data was published in Nature Medicine in 2025 [6]. No active FDA designations (breakthrough, fast track, orphan) remain on the molecule after the CRL. Expected next major readout: Phase 3 breast cancer PFS interim, plausibly 2026 or 2027 given current enrollment status and event-driven analyses.

HER3 (also called ERBB3) is a cell surface receptor in the same family as EGFR and HER2. Think of it as a docking station: HER3 binds neuregulin ligands and then partners with HER2 to switch on the PI3K/AKT growth signaling pathway inside the cell [8]. HER3 has barely any kinase activity on its own; its job is to recruit signaling partners. It's overexpressed in breast, lung, colorectal, and gastric cancers, and HER3 upregulation is a known escape route when EGFR or HER2 inhibitors stop working. The ADC works like a guided missile: the patritumab antibody finds HER3-expressing tumor cells, the conjugate gets internalized, and the cell releases deruxtecan (DXd) - a topoisomerase I inhibitor that breaks DNA during replication and kills the cell. Critically, DXd is membrane-permeable after intracellular release, so it diffuses into neighboring HER3-low or HER3-negative cells and kills them too. This 'bystander effect' partially mitigates (though does not eliminate) the risk that heterogeneous HER3 expression - patches of high and low expressers within the same tumor - would otherwise blunt efficacy. The same DXd payload is in Enhertu (trastuzumab deruxtecan), which AstraZeneca and Daiichi Sankyo built into a multibillion-dollar HER2-directed franchise. That's the cleanest commercial validation the payload chemistry could have. HER3 itself, however, is not yet validated as a drug target: no HER3-directed therapy has ever been approved, and the genetic case for HER3 dependence is weaker than for HER2. There are no recurrent activating HER3 mutations driving common tumors; it's a passenger more often than a driver. The bystander payload mechanic is part of why a non-driver target can still produce real responses.

The node references KEYMAKER-U01 (NCT06731907), a Phase 2 umbrella substudy testing HER3-DXd with pembrolizumab in advanced NSCLC, n=90, ORR primary endpoint [1]. Umbrella studies like this trade rigor for speed: small biomarker-defined arms looking for signal, no comparator. Useful for go/no-go decisions, not for approval. The franchise's actual approval bet sits in NCT07060807, a Phase 3 in breast cancer enrolling 1,000 patients with PFS as the primary endpoint [5]. The Phase 3 patient selection (HER3 expression cutoff, line of therapy, comparator arm) is the key unknown - the HR+/HER2- Phase 2 data published in Nature Medicine 2025 informs that selection logic [6]. That trial (ICARUS-BREAST01, n=99 post-CDK4/6 inhibitor and one prior chemo line) reported a confirmed ORR of 53.5% (90% CI 44.8-62.1%), with median PFS of 12.8 months in HER2-low patients and 9.1 months in HER2-0 patients [6]. Those are strong Phase 2 numbers in a population running out of options, and they explain why the Phase 3 PFS bet looks reasonable. HERTHENA-Lung01 in previously-treated EGFR-mutated NSCLC - the trial that supported the withdrawn BLA - reported a confirmed ORR of 29.8% (95% CI 23.9-36.2%) [12]. TUXEDO-3 enrolled patients with active brain metastases, an underserved population where CNS penetration of an ADC is a real biological question; both the breast and NSCLC arms reported activity in Lancet Oncology 2025 [2][3]. HERTHENA-PanTumor01 (basket Phase 2 in previously-treated solid tumors) provides broader signal-finding context [7]. The gastrointestinal Phase 1 (NCT06596694) and KEYMAKER-U06 GEJ adenocarcinoma substudy (NCT06469944) extend the umbrella. The strategic risk in this layout: lots of Phase 2 surface area, one Phase 3 swing in breast.

Our model gives this drug a 5% chance of eventually being approved. It starts from a baseline of about 13% - the historical approval rate for Phase 2 drugs in this area - then adjusts up or down based on ten facts about the trial and the company behind it. The sponsor's strong track record pushes the estimate up, but weak earlier-phase results, a randomized design, and the use of a control arm pull it back down. The remaining factors are close to average for this stage, so they leave the final number well below where it started.

Efficacy risk: HER3 expression is common but its causal role in driving any tumor's growth is weaker than HER2's role in HER2+ breast or EGFR's in EGFR-mutant lung. Phase 2 readouts have shown activity - 29.8% ORR in HERTHENA-Lung01 [12] and 53.5% ORR in ICARUS-BREAST01 [6] - but the lung number was not enough to clear an accelerated approval bar, and breast remains to be confirmed in Phase 3. Without a sharp predictive biomarker - and HER3 expression has not yet proven predictive in registrational settings - patient selection is fuzzier than it should be [7]. Safety risk: DXd-class ADCs carry an ILD (interstitial lung disease, a form of drug-induced lung scarring) signal that the Enhertu label spells out and that has caused trial deaths across the class. In HERTHENA-Lung01 specifically, 5.3% of patients (12/225 at the 5.6 mg/kg dose) had adjudicated treatment-related ILD, including one Grade 5 (fatal) event [12]. That rate is on par with Enhertu's and is high enough to weigh on prescriber comfort. Execution risk: The 2024 FDA CRL on the NSCLC BLA, driven by third-party manufacturing inspection findings, is a real red flag for franchise regulatory operations [4]. The parallel Dato-DXd CRL deepens that concern. Because Daiichi controls manufacturing under the 2023 collaboration, this risk is borne jointly but originates on Daiichi's side [11]. Commercial risk: ADC competition is intense - Enhertu, Trodelvy, Dato-DXd, and emerging HER3 programs from Hummingbird, Bolt, and others. Even with approval, HER3-DXd would face payer pressure to show clear advantage over established options, and pricing will be high enough that differentiation needs to be unambiguous. The HR+/HER2- metastatic breast cancer addressable market is substantial (roughly 40,000+ U.S. patients per year reaching late-line treatment), and EGFR-mutant NSCLC after TKI failure is a similarly meaningful slice (~15,000+ U.S. patients per year), but both indications are crowded.

Worth watching, but the signal-to-noise ratio is low until the breast cancer Phase 3 reads out. The most informative near-term data already exists: TUXEDO-3 brain metastases activity in Lancet Oncology 2025 [2][3] gives a read on CNS efficacy, the kind of unmet need that wins durable market share. ICARUS-BREAST01 in HR+/HER2- breast cancer reported 53.5% ORR and 12.8-month median PFS in HER2-low patients [6] - strong enough numbers to justify the Phase 3 commit, and the closest thing to a predictive-biomarker story this asset has produced. For Merck specifically, HER3-DXd is one of three assets in the October 2023 Daiichi collaboration ($4B upfront, $1.5B continuation, up to $22B total including milestones; Daiichi retains Japan rights and manufacturing) [11]. That structure matters: Merck's exposure is diversified across HER3-DXd, Dato-DXd, and ifinatamab deruxtecan, so a single-indication setback hurts but doesn't break the thesis. Daiichi's stock is more directly exposed to single-asset readouts because the same three assets are a much larger share of its pipeline. The company's $65B revenue base partially insulates it from Keytruda's eventual loss of exclusivity in 2028, and ADCs are a load-bearing piece of the post-Keytruda thesis [10]. Both 2024 CRLs (HER3-DXd in NSCLC, Dato-DXd in NSCLC) hit Merck financially through this deal - that's the read on why 'manufacturing execution risk' is not abstract here. Check back on: (1) Phase 3 breast cancer interim PFS, likely 2026-2027; (2) any updated CMC (Chemistry, Manufacturing, and Controls - the FDA's term for manufacturing quality documentation) resolution that could re-open an NSCLC accelerated approval pathway; (3) head-to-head positioning against Dato-DXd in shared indications. A first signal-grade data point would be Phase 3 breast cancer PFS with hazard ratio under 0.6 in a defined HER3-high population - that would change the asset's commercial trajectory.