DLL3 in SCLC: Laboratory Insights and Molecular Regulation

DLL3 is a protein found on the surface of small cell lung cancer (SCLC) cells but largely absent from healthy adult tissues, making it one of the most promising molecular targets in an otherwise difficult-to-treat disease. Its regulation is tightly linked to the neuroendocrine identity of SCLC cells, driven by a transcription factor called ASCL1 and connected to the Notch signaling pathway that governs cell fate. The laboratory work unpacking how DLL3 behaves at the molecular level has directly shaped an expanding wave of experimental therapies, from bispecific antibodies to radioimmunotherapy, and understanding that biology helps explain both the promise and the pitfalls of targeting it.

What DLL3 Actually Does in SCLC Cells

DLL3 belongs to the Delta-like family of Notch ligands, but it behaves differently from its relatives. Most Notch ligands sit on a cell’s surface and activate Notch receptors on neighboring cells, sending signals that influence growth, differentiation, and identity. DLL3 does roughly the opposite: it acts as an inhibitor, working inside the same cell that produces it rather than signaling to neighbors. In developmental biology, this was first understood in the context of spinal segmentation, where DLL3 was shown to interact with Notch1 inside the cell’s internal compartments and route the receptor toward degradation before it could ever reach the surface. That mechanism, called cis-inhibition, prevents Notch signaling from turning on within that cell.¹

In SCLC, this cis-inhibition takes on a different significance. SCLC cells are neuroendocrine in character, meaning they share features with hormone-secreting nerve cells. Active Notch signaling tends to push cells away from that neuroendocrine identity. By suppressing Notch from within, DLL3 helps lock SCLC cells into their neuroendocrine state. Experiments in a mouse model of SCLC confirmed this directly: when researchers forced DLL3 expression in SCLC cells that had active Notch signaling (marked by a fluorescent reporter tied to the Notch target gene Hes1), the proportion of Notch-active cells dropped, providing clear evidence that DLL3 suppresses Notch signaling in cis within this cancer context.²

How SCLC Cells Turn On DLL3 Production

The transcription factor ASCL1 is the central driver of DLL3 expression. ASCL1 is essential for the development of neuroendocrine cells in the lungs and is one of the defining molecular features of the most common SCLC subtype. It binds directly to the regulatory regions of the DLL3 gene, switching it on as part of a broader neuroendocrine gene program.³ ASCL1 also controls multiple other genes in the Notch pathway, so DLL3 is not an isolated output but part of a coordinated regulatory circuit in which ASCL1 simultaneously activates neuroendocrine identity genes and dampens the Notch signals that would oppose that identity.⁴

This connection to ASCL1 explains a pattern clinicians observe: SCLC tumors dominated by the ASCL1-high subtype tend to have the highest DLL3 levels. Other SCLC subtypes, driven by different transcription factors like NEUROD1 or POU2F3, can still express DLL3 but often at lower or more variable levels. The practical consequence is that the molecular subtype of a patient’s tumor may influence how much DLL3 is available as a therapeutic target.

Surface Expression and Why It Matters for Targeting

A critical distinction for drug development is where DLL3 protein ends up. In normal adult tissues, DLL3 expression is minimal and, where it exists at all, confined to the inside of cells. Normal neurons, pituitary cells, and testicular cells produce small amounts of DLL3, but the protein stays in the cytoplasm and never reaches the outer cell membrane.⁵ In SCLC and other high-grade neuroendocrine tumors, by contrast, DLL3 is aberrantly displayed on the cell surface.⁶

This difference, cytoplasmic in normal tissue versus surface-exposed in cancer, is what makes DLL3 such an attractive drug target. Antibodies and other biologics can only reach proteins displayed on the outside of cells, so a drug designed to latch onto DLL3 would preferentially attack tumor cells while mostly sparing healthy tissue. The contrast is unusually clean compared to many other cancer targets, where the protein is also found on the surface of normal cells and targeting it causes collateral damage.

Large-scale studies of SCLC tumor samples consistently find high rates of DLL3 positivity. In one international real-world analysis, DLL3 expression was detected in the vast majority of SCLC specimens. A separate study that specifically scored staining intensity found DLL3 positivity in roughly 94% of SCLC samples, with somewhat lower rates in large cell neuroendocrine carcinoma (about 80%) and tumors with mixed histology (about 63%).⁷

The Antibody-Drug Conjugate That Failed and What It Taught the Field

The first major attempt to exploit DLL3 therapeutically was rovalpituzumab tesirine, commonly called Rova-T. This was an antibody-drug conjugate: an antibody that binds DLL3 on the tumor surface, carrying a potent cell-killing chemical payload that gets released once the drug is internalized. Early-phase studies generated excitement, with some heavily pretreated SCLC patients showing tumor shrinkage. But when Rova-T was tested in larger phase III trials, it failed to improve survival, and its development was halted.⁸⁹

Rova-T’s failure was not necessarily a failure of DLL3 as a target. The drug’s payload was a pyrrolobenzodiazepine dimer, a class of DNA-damaging agent associated with significant toxicity. It caused fluid buildup, skin reactions, and liver injury at rates that narrowed the therapeutic window. The lesson was that the antibody found its target reliably enough, but the cargo it carried was too toxic for an acceptable safety profile. Subsequent therapeutic strategies have moved toward approaches that harness the immune system rather than relying on a chemical warhead.

Bispecific T-Cell Engagers

The most clinically advanced successor to Rova-T is the bispecific T-cell engager (BiTE) approach. These are engineered molecules with two binding arms: one grabs DLL3 on the tumor cell, and the other grabs CD3 on a T cell, physically bridging the immune cell and the cancer cell so the T cell can kill the tumor directly. The concept bypasses the need for a chemical payload entirely, instead redirecting the patient’s own immune system.

Preclinical work with AMG 757 (later developed into tarlatamab) showed striking results. The drug killed SCLC cell lines even when those lines expressed very low amounts of DLL3, fewer than a thousand molecules per cell. In mouse models using patient-derived SCLC tumors, AMG 757 triggered complete tumor regression in the majority of animals. In orthotopic models, where tumors grew in the lungs or spread to the liver as they would in a patient, treatment drove tumor signals down to near the limit of detection. The drug’s activity was accompanied by measurable T-cell infiltration into tumors and upregulation of T-cell activation markers, confirming that the immune cells were being recruited and switched on.¹⁰¹¹

These preclinical findings translated into clinical trials. Tarlatamab became the first DLL3-targeted therapy to receive regulatory approval for previously treated SCLC, vindicating the target after Rova-T’s failure. The mechanism also brings its own side-effect profile, discussed in a later section, but the efficacy signal was strong enough to change the treatment landscape.

CAR-T Cells and Armored Variants

Another immune-based strategy under investigation is chimeric antigen receptor T-cell therapy, or CAR-T. Here, a patient’s T cells are removed, genetically engineered to recognize DLL3, and infused back. The approach has transformed treatment for certain blood cancers, but solid tumors like SCLC present a harder challenge because the tumor microenvironment actively suppresses immune cells that manage to infiltrate.

To counter that suppression, researchers have developed “armored” CAR-T cells that secrete immune-boosting signals. One line of work showed that DLL3-targeting CAR-T cells engineered to release the inflammatory cytokine IL-18 were substantially more potent than standard DLL3 CAR-T cells in both patient-derived tumor models and mouse SCLC models.¹² The IL-18 secretion appeared to help the engineered T cells sustain their attack in the hostile tumor environment. This remains preclinical work, but it illustrates how DLL3’s tumor-restricted surface expression makes it a viable handle for multiple immune-engineering approaches.

Radioimmunotherapy Approaches

A different strategy uses DLL3-targeting antibodies not to recruit immune cells but to deliver radiation directly to tumors. In radioimmunotherapy, a radioactive isotope is attached to a DLL3-binding antibody, which homes to the tumor and irradiates it from within. Work in neuroendocrine prostate cancer models demonstrated that a lutetium-177-labeled anti-DLL3 antibody achieved complete responses in tumor-bearing mice at multiple dose levels, with curative responses at higher doses and minimal uptake in DLL3-negative tumors.¹³

A head-to-head comparison in SCLC patient-derived xenografts tested DLL3-targeting radioimmunoconjugates using two different isotopes (lutetium-177 and actinium-225) against a non-radioactive antibody-drug conjugate benchmark. The antibody-drug conjugate performed well, with full suppression in most mice, while the radioactive versions controlled tumor growth for about three weeks before regrowth appeared and extended survival, particularly with actinium-225. The results highlight that radioimmunotherapy is a viable approach but may work best in combination or in settings where other strategies have fallen short.¹⁴

Combining DLL3-Targeted Therapy with Checkpoint Inhibitors

SCLC has historically responded poorly to immunotherapy with PD-1 or PD-L1 inhibitors used alone. Preclinical evidence suggests that DLL3-targeted agents and checkpoint inhibitors may work better together than either does independently. In one study, adding a PD-1 blocking antibody to a DLL3-targeting bispecific antibody dramatically enhanced anti-tumor activity and prolonged survival in mouse models of SCLC.¹⁵ The logic is straightforward: the bispecific antibody recruits T cells to the tumor, and the checkpoint inhibitor removes one of the brakes that would otherwise slow those T cells down once they arrive.

Similar synergy was observed with Rova-T. Despite that drug’s failure as a single agent, combining sub-effective doses of Rova-T with anti-PD-1 therapy in a mouse SCLC model that retained a functioning immune system produced better anti-tumor effects than either treatment alone.¹⁶ The implication is that DLL3-targeted agents may serve partly as immune primers, making tumors more visible to the immune system and thereby enhancing the effect of checkpoint blockade.

Monitoring DLL3 Through Liquid Biopsy

Measuring DLL3 on tumor tissue requires an invasive biopsy, which is not always practical in SCLC because the disease often presents in locations that are difficult to sample. Liquid biopsy, which analyzes circulating tumor cells (CTCs) from a blood draw, offers an alternative. Studies have shown that DLL3-positive CTCs can be detected in the blood of most SCLC patients before treatment, with one study finding them in about 74% of patients at baseline. After one cycle of chemotherapy, both the detection rate and the number of DLL3-positive CTCs dropped significantly, then rose again when the disease progressed. In that study, the presence of DLL3-positive CTCs at baseline was associated with worse progression-free survival, and their detection at disease progression was associated with worse overall survival.¹⁷

Earlier work on CTCs in the context of Rova-T treatment found that patients whose baseline CTCs expressed DLL3 showed significant decreases in CTC counts after receiving the drug, suggesting that Rova-T was specifically killing the DLL3-expressing circulating cells.¹⁸ The broader vision is that liquid biopsy could eventually serve as a companion diagnostic for DLL3-targeted therapies, allowing clinicians to assess DLL3 status without a tissue biopsy and to track treatment response in real time.¹⁹ The challenge is standardization: different assay platforms capture different subsets of CTCs, and some DLL3-positive cells are missed by conventional capture methods that rely on other markers.

DLL3 Beyond the Lung

DLL3 is not exclusive to SCLC. Because its expression is driven by neuroendocrine differentiation pathways, it shows up in high-grade neuroendocrine cancers arising in other organs. In neuroendocrine prostate cancer, a particularly aggressive variant of castration-resistant disease, DLL3 was found in roughly three-quarters of cases but was essentially absent in localized prostate cancer and benign prostate tissue. Its expression correlated with loss of the RB1 tumor suppressor and with other neuroendocrine markers.²⁰

Beyond the prostate, DLL3 has been detected in neuroendocrine carcinomas originating from the bladder, uterine cervix, skin, and head and neck region.⁷ DLL3-targeted therapies developed for SCLC have already shown clinical activity in some of these extrapulmonary neuroendocrine cancers.²¹ The shared biology of neuroendocrine differentiation means that drugs validated in SCLC may find second lives in cancers that are individually rare but collectively represent a meaningful patient population. Accurate pathological assessment and standardized scoring of DLL3 expression across tumor types will be essential for identifying which patients are most likely to benefit.²²

The Resistance Problem

Like nearly all targeted cancer therapies, DLL3-directed treatments face the prospect of resistance. The concern specific to DLL3 is that therapeutic pressure could select for tumor cells that have lost or reduced DLL3 expression. SCLC is already known for its plasticity; tumors can shift between neuroendocrine and non-neuroendocrine states during treatment. Because DLL3 expression is tied to the neuroendocrine program, a tumor that transitions away from that identity could become invisible to DLL3-targeted drugs. Research has flagged this as a realistic escape route: therapy may select for, or in some contexts drive, a DLL3-low or non-neuroendocrine state.²³

This creates a somewhat paradoxical situation: the same lineage commitment that makes DLL3 such a reliable target in the first place is the thing that can be lost under treatment pressure. Monitoring DLL3 status over time, potentially through liquid biopsy as discussed above, could help catch this shift early. It also argues for combination strategies that attack the tumor through multiple pathways simultaneously, making it harder for a single escape mechanism to render treatment ineffective.

Safety Profile of the Bispecific Approach

With tarlatamab now in clinical use, its side-effect profile is becoming well characterized and illustrates the trade-offs of immune-redirecting therapies. In the phase II DeLLphi-301 trial, cytokine release syndrome (CRS) was the most common adverse event, occurring in about half of treated patients. CRS happens when the bridged T cells release a burst of inflammatory signals; it typically manifests as fever, low blood pressure, and fatigue, and ranges from mild to life-threatening. Pyrexia, decreased appetite, taste changes, and anemia were also common. About 10% of patients experienced immune effector cell-associated neurotoxicity syndrome (ICANS), a neurological side effect linked to immune activation in or near the brain.²⁴

In a subset of patients who entered the trial with higher baseline risk factors, the rates of both CRS and ICANS were notably elevated. About 62% experienced CRS, with severe (grade 3 or higher) CRS in roughly 15%. ICANS occurred in close to half of these higher-risk patients, with severe ICANS in about 14%.²⁵ These rates are manageable with established protocols for CRS and ICANS from the blood cancer field, but they underscore that DLL3-targeting immunotherapies are far from side-effect-free and require experienced clinical teams. The fact that DLL3 is expressed at low levels in some normal neurons has raised theoretical concerns about on-target, off-tumor neurotoxicity, though the cytoplasmic (not surface) location of DLL3 in normal neural tissue appears to limit this in practice.