World Lung 2025 – cell therapy makes a cautious start against DLL3

As the resurgence of DLL3 continues with promising data for antibody-drug conjugates, could this oncology target also see use for a Car-T therapy? Novartis thought so when it paid Legend $100m for rights to the anti-DLL3 Car-T project LB2102 two years ago, and just revealed human data have shown what it got for its money.

For now, the best that can be said about LB2102 is that it's a work in progress, with the results, presented at World Conference on Lung Cancer on Tuesday, amounting to a 15% response rate among 13 patients with relapsed/refractory SCLC. This is well below the 50% or higher ORRs that Ideaya and Zai Lab are boasting of for their respective anti-DLL3 ADCs.

On a similar cross-trial basis 15% also lags the 40% that Amgen's anti-DLL3 T-cell engager Imdelltra showed in Dellphi-301, a study that backed this first-in-class drug's accelerated approval. Such comparisons suggest that cell therapy companies are right in pursuing DLL3 with caution; OncologyPipeline reveals just one other DLL3-directed Car-T therapy in clinical development.

Some promise?

Still, the LB2102 data do reveal some glimmers of promise. For instance, there were only two cases of cytokine release, both at grade 1, and no ICANS up to the dose level tested (8 million Car-T cells/kg).

If nothing else this suggests scope for testing higher doses, in the hope that this might dial up efficacy. The phase 1 study plan includes two further doses, and the highest of these, 16 million cells/kg, is now enrolling. Researchers also told World Lung that patients' DLL3 expression remained high after treatment, suggesting low likelihood of relapse by antigen loss.

Indeed, both of the two partial responses seen are showing impressive durability; one patient relapsed at around 240 days after infusion, while the other is still in response at 150 days. There was apparently also a third response, lasting for eight months, but this patient was treated off protocol so couldn't be included in the formal dataset.

In total this presentation comprised 15 efficacy evaluable subjects, including two with neuroendocrine cancers, an emerging tumour for anti-DLL3 therapy. However, neither of these two patients responded.

Competition

While Amgen was the first company to bring to market an anti-DLL3 drug, Imdelltra, it was also first to take a Car-T therapy into the clinic. However, that project, AMG 119, the result of a 2015 deal with the now Gilead-owned Kite Pharma, was discontinued six years ago.

According to OncologyPipeline the only other anti-DLL3 Car-T therapy in clinical development is Chengdu Brilliant's BHP01, which entered phase 1 in April. Very little is known about this asset, which is claimed to feature an α-PD-L1/4-1BB modifying activity.

This is likely an attempt to improve on AMG 119, and for its part LB2102 is armoured with dominant-negative TGFBR2 – the same technology that AstraZeneca is using in its anti-GPC3 and Steap2 Cars, AZD7003 and AZD0754 respectively. When the Novartis/Legend deal was struck the possibility was also mentioned of employing the Swiss group's fast T-Charge manufacturing.

And there are other approaches to targeting DLL3 beyond MAbs and cell therapies. At World Lung Abdera Therapeutics presented the first human data with its actinium-based anti-DLL3 radiopharmaceutical ABD-147, from a phase 1 study in SCLC and NEC. However, its poster merely detailed pharmacokinetic results from six patients receiving the first dose, 1.6MBq.

There were said to be no dose-limiting toxicities, but there were two treatment-related adverse events, of grade 4 thrombocytopenia and grade 1 paresthesia (tingling sensation), plus two deaths deemed unrelated to ABD-147. The trial will now dose patients at 3.2MBq.

Car-T therapies against DLL3

Notes: NEC=neuroendocrine cancers; *dnTGFBR2-armoured, with possibility of future manufacturing using Novartis’s T-Charge tech; **described as α-PD-L1/4-1BB modifying; ^CD70-armoured. Source: OncologyPipeline.