Good things come in 3s

DLL3, B7H3 and Her3 are compelling tumor antigens to target with cancer therapeutics. They have interesting patterns of expression in different cancer indications; thus, diverse therapies for attacking these targets have been developed. Each target and each therapeutic modality induce varying degrees of clinical efficacy, as well as causing toxicities. Importantly, effective combinatorial approaches that include these targets are emerging. Finally, while there are clear front runners among the pharmaceutical companies developing targeted therapeutics for these antigens there are also emerging biotechs aggressively pursuing these targets. The result is a complex clinical, commercial and competitive landscape.

Let’s dive in…

Wave 1: Delta-like ligand 3

DLL3 is a ligand for the cell development and differentiation protein called Notch, whose dysregulation is associated with tumor biology. Normal tissues and cells express little or no DLL3 except during fetal development. In contrast, DLL3 expression is notable in neuroendocrine tumors that can arise in various niches throughout the body. Expression can be very high in diverse neuroendocrine tumors including small cell lung cancer (SCLC), neuroendocrine prostate cancer (NEPC), neuroendocrine pancreatic cancer (NEP), Merkel cell carcinoma, adrenal cancer, the gut carcinoid tumors etc. The majority of clinical studies have been done in the more common of these cancers, SCLC and NEP.

SCLC is a large indication that is common wherever smoking is common. SCLC accounts for 14% of all lung cancers with about 250,000 new cases per year globally.

SCLC is a highly aggressive neuroendocrine tumor characterized by rapid growth and early, widespread metastasis. Over 70% of new patients are diagnosed with stage IV disease, also called extensive stage disease (ES-SCLC). For these patients, clinical onset means severe symptoms and a rapid decline in overall health. Typical treatment has been chemotherapy and radiotherapy with an initially high overall response rate (ORR) but then rapid recurrence followed by poor prognosis. The overall survival rate (OS) at 5 years post-diagnosis is about 3% for ES-SCLC.

The discovery that SCLC responds to immune checkpoint (PD-1 or PD-L1) inhibitors has advanced the standard of care somewhat. Two large and randomized clinical trials, IMpower-133 and CASPIAN, showing statistically significant improvements in outcomes when chemotherapy was combined with atezolizumab or durvalumab, respectively, in the first-line treatment of ES-SCLC. Based on data from studies done in second-line plus patients, combinations of chemotherapy with either pembrolizumab or nivolumab are indicated for the treatment of advanced refractory SCLC. However, across lines of therapy from first-line to third-line, the response rates and survival benefits achieved with chemo or with the chemo/immune checkpoint combinations are modest, with ORR typically under 25% and median progression free survival (PFS: alive without disease getting worse) of 4 months or less and median OS between 4 and 7 months.

The finding that SCLC was “checkpoint-responsive” tells us a little bit about the biology of this cancer. SCLC generally has a “hot” tumor microenvironment (TME) meaning that it is infiltrated by immune cells, including the T cells that are activated by immune checkpoint therapies. This leads us to Amgen’s T cell-engager therapeutic, tarlatamab. This therapeutic uses a bispecific antibody format to bind to DLL3 on SCLC cells and to the CD3 protein on T cells. CD3 binding activates the T cells to attack and kill any cell that expresses DLL3, thus killing SCLC cells. Tarlatamab triggered an ORR of ~40% in refractory (second-line or later) SCLC patients in the DeLLphi-301 clinical trial published in 2023. The results were remarkable for this very sick patient population, with a median OS of 14 months at the dose of 10 mg given every 2 weeks:

The data supported FDA approval, in May 2024, of tarlatamab for refractory ES-SCLC.

Toxicities were challenging enough to cause 13% of patients to reduce dose or skip doses and 3% to discontinue treatment. Because of added toxicity without additional benefit the 100 mg dose was dropped.

The DeLLphi-304 study was reported at ASCO earlier this month. This Phase 3 trial compared tarlatamab with chemotherapy as second-line treatment. Results were again promising, with a 35% ORR and a median OS of 13.6 months (versus 20% ORR and 8.3 months median OS in the chemo arm).

Amgen’s clinical development studies of tarlatamab include the DeLLphi-305 trial in ES-SCLC testing a first-line maintenance regimen following durvalumab plus chemo (Amgen markets the anti-PD-L1 checkpoint inhibitor durvalumab). Patients are enrolled while still in response to durvalumab plus chemo and then receive tarlatamab plus durvalumab. The primary endpoint is median OS versus durvalumab maintenance alone. The DeLLphi-306 study investigates the earlier setting of first-line treatment of limited-stage SCLC.

The tarlatamab results have of course triggered a stampede of development of DLL3-targeted therapies. Boehringer Ingelheim’s DLL3xCD3 bispecific therapy obrixtamig is in clinical development across several neuroendocrine cancers including SCLC. Merck’s DLL3xCD3 bispecific therapy gocatamig/MK-6070 is being developed in SCLC and is the subject of a joint commercialization agreement with Daiichi Sanyo (more on this when we get to B7H3-targeting, in Wave 2). Gocatamig is a “TriTAC” which just means it has 3 distinct domains. This brings up an interesting technical point, which is that these three bispecifics have three very different designs:

Side-bar: at Aleta Biotherapeutics we favor the type of design that is illustrated for gocatamig, ie. multiple domains linked together to create novel functionality. This type of design is Incorporated into all of our Engager programs (www.aletabio.com).

Before looking at newer programs it is worth stating that there has also been bad news for certain programs along the way, and it concerns toxicities. Programs across therapeutic modalities have been terminated due to unfavorable efficacy/toxicity profiles, as represented by these programs:

Those setbacks have not deterred the field but the results do suggest that each therapeutic and modality will have to be weighed and measured carefully. With that in mind, here is a snapshot of programs in development, with a few of our favorites highlighted on probability of success:

There are just a few comments here, as the large pharma programs are significantly ahead as covered earlier. First, I’m less in favor of the more complex programs like those targeting 4-1BB or CD28 or CD47: none of those targets have been successful in antibody formats. Second, Suzhou Zelgen’s ZG006 is, cleverly, targeting DLL3 with a biparatopic design, meaning two different binding sites on the DLL3 protein are bound. That is a decent hypothesis to explore and their early clinical data at ASCO 2025 looked interesting.

Now to antibody-drug conjugates (ADC) where we saw some early programs with difficult toxicities. It’s probably worth noting that ADCs as a class are becoming less prone to toxicities than they were even 5 years ago, in part due to new generations of linker technology and payloads, and in particular the use of Topo1 inhibitors as the payload on the ADC. These advances revolutionized the ADC field. Let’s check out the swells coming in:

A few comments here as well. First, the field is dominated by China companies; I think we are used to this phenomenon by now. That said, the clear leader in the ADC space is Zai Lab: they presented at ASCO 2025 and showed well in the ongoing Phase 1 dose escalation study in ES-SCLC. Most patients (90%) were refractory to checkpoint inhibitor therapy and half were second-line refractory – a tough population to treat. The ADC performed well with an ORR = 68% (only 28 patients though so a small sample size). They ran into significant toxicities at higher doses which will be something to watch in future. Second we see the appearance of bispecific ADC programs in the preclinical space – this will be problematic given the toxicity associated with DLL3 itself, before adding a second target.

Finally, let’s look a CAR-T cells. These are the patient’s own T cells that are engineered to express a binder to directly attack DLL3-positive cells. Recall that Amgen’s AMG 119 CAR-DLL3 program was terminated due to toxicity. Regardless, programs continue to emerge:

The one program here reporting clinical data is from Legend Biotech in collaboration with Novartis. LB2102 features DLL3-targeting and expression of a TGFBR2 dominant-negative protein on the CAR T cells surface to thwart TGFbeta-mediated immunosuppression. At ASCO 2025, Legend reported early results from a Phase 1 dose escalation study – the dose being the number of CAR T cells infused into the patient. The best overall responses included 1 partial response each at dose levels 3 and 4 (so 2/12 responses, ORR = 16.7%). However serious toxicities were seen with grade 3-4 toxicities dominating at the higher dose levels (dose levels 2 though 4). 4 patients (33%) had grade 4 adverse events. The toxicities may be due to the lymphodepletion regimen used, regardless, better response rates will be needed here.

To sum up: DLL3-targeted therapeutics – bispecifics, ADCs, CARs – constitute a rapidly growing and highly innovative drug development landscape for the treatment of highly neuroendocrine aggressive cancers. An awesome space to watch.

Stay tuned.

The Next Wave:

Targeting B7H3 – a drug development tsunami in the making?