Blood Test Counting DLL3-Positive Tumor Cells Predicts Who Responds to Tarlatamab in Small Cell Lung Cancer
Counting DLL3-positive circulating tumor cells before treatment identified every patient who benefited from tarlatamab in a 20-patient small cell lung cancer cohort.
Summary
Tarlatamab, a bispecific antibody that directs T cells to DLL3-expressing tumors, works in many patients with small cell lung cancer (SCLC), but outcomes vary and no biomarker guides its use. Tissue staining for DLL3 has not predicted response. Researchers instead used a microfluidic chip to isolate circulating tumor cells (CTCs) from blood without bias for size or epithelial markers, then measured DLL3 on each cell. In 20 evaluable patients, those with at least 25% DLL3-positive CTCs all gained clinical benefit (partial response or stable disease), while most with lower fractions progressed. Sensitivity was 85% and specificity 100%. Resistance later arose either from DLL3 loss on CTCs or from signs of T-cell dysfunction. The cohort is small and single-center, so validation is needed.
Detailed Summary
Small cell lung cancer is aggressive and almost always relapses after first-line chemotherapy plus immunotherapy. Tarlatamab, a DLL3-targeting T-cell engager, is now a standard option for relapsed disease, yet more than half of patients progress within about 6 months, and the first two doses require hospital observation for cytokine release syndrome. Tissue immunohistochemistry shows DLL3 in nearly all biopsies and has not predicted who benefits, so there is no selection biomarker.
The team enriched CTCs from 10-20 mL of blood using the CTC-iChip, a negative-depletion microfluidic platform that removes blood cells without relying on EpCAM or cell size, which matters because SCLC cells are small and often EpCAM-low. CTCs were stained for DLL3, epithelial markers and leukocyte markers and scored by multispectral imaging. Of 32 patients treated with tarlatamab, 20 met prespecified criteria for response evaluation. Single-cell RNA sequencing of biopsies from 10 patients, plus a published 19-tumor dataset, was used to examine DLL3 heterogeneity.
Among 2,987 baseline CTCs, 58% were DLL3-only, 31% epithelial-marker-only, and 11% positive for both, so most DLL3-positive CTCs would be missed by epithelial-based capture. In biopsies, DLL3 RNA was detected in a median of 35% (cohort B) and 64% (cohort C) of cancer cells, independent of molecular subtype and prior treatment. In contrast, CTCs separated patients into DLL3Pos (n=11, at least 25% positive) and DLL3Low (n=9). All 11 DLL3Pos patients benefited (6 partial responses, 5 stable disease); among DLL3High patients (>75%), 6 of 8 had partial responses. Of the 9 DLL3Low patients, 7 progressed. Predicting clinical benefit gave 85% sensitivity and 100% specificity. One DLL3Low responder had an aberrant DLL3 splice variant affecting the cytoplasmic epitope of the diagnostic antibody, suggesting assays should use antibodies matched to the therapeutic target. Archival tissue IHC again failed to predict response.
The abstract also reports that necrotic CTC clusters accompany treatment-induced tumor lysis, and that acquired resistance involves either DLL3 loss with persistence of other targetable neuroendocrine epitopes, or retained DLL3 with systemic T-cell dysfunction markers. Full details of those analyses were not available in the provided text.
Implications: CTC DLL3 scoring could help select patients for tarlatamab and guide next-line choices at resistance. Caveats include the small single-institution cohort, a 25% cutoff taken from prior literature, and the need for prospective validation.
Key Findings
- Patients with at least 25% DLL3-positive CTCs (n=11) all achieved clinical benefit: 6 partial responses and 5 stable disease on tarlatamab.
- Baseline CTC DLL3 scoring predicted clinical benefit with 85% sensitivity and 100% specificity in 20 response-evaluable patients.
- 58% of DLL3-positive CTCs lacked epithelial markers, so EpCAM-based capture would miss most of them.
- Biopsy scRNA-seq showed mixed DLL3 expression in every tumor, independent of molecular subtype or prior therapy; tissue IHC did not predict response.
- Resistance involved DLL3 loss with other targetable neuroendocrine epitopes retained, or DLL3 retained alongside systemic T-cell dysfunction markers.
Methodology
Prospective single-institution cohort of 32 ES-SCLC patients on tarlatamab, with 20 meeting prespecified response-evaluable criteria (RECIST 1.1). CTCs were isolated by CTC-iChip negative depletion and scored for DLL3 by multispectral immunofluorescence at baseline and longitudinally. Biopsy scRNA-seq (10 patients plus a published 19-patient dataset) characterized DLL3 heterogeneity.
Study Limitations
The response-evaluable cohort was only 20 patients from one center, with wide confidence intervals and a 25% cutoff borrowed from prior studies, so findings need independent prospective validation. Matched biopsies for scRNA-seq were unavailable for the CTC cohort, and an aberrant splice variant in one patient showed the diagnostic antibody can misclassify DLL3 status. The CTC-iChip and multispectral imaging workflow is specialized and not widely available.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe: