Membrane share of TROP2 identified longer PFS on Dato-DXd in nonsquamous NSCLC

Median PFS of 7.2 against 4.0 months on Dato-DXd, split by NMR status

Within the Dato-DXd arm of TROPION-Lung01, patients whose tumour samples scored TROP2 NMR-positive had a median progression-free survival of 7.2 months, and those scoring NMR-negative had 4.0 months, a hazard ratio of 0.40. The abstract gives these figures for TROPION-Lung01 without restating which of the trial’s populations they were measured in.

The figures compare patients inside one treatment arm by biomarker status. They show how the score sorts outcomes on Dato-DXd. They are not a comparison of Dato-DXd with docetaxel within the NMR-positive group, and the abstract carries no such between-arm figure.

Little difference by NMR status on docetaxel

Among patients given docetaxel in the same trial, median progression-free survival was 4.1 months with NMR-positive samples and 4.4 months with NMR-negative samples, a hazard ratio of 0.94.

The treatment-by-biomarker interaction test measures exactly this contrast between the two arms, and it gave P = 0.0098. A score that reflected prognosis alone would be expected to sort outcomes on both drugs. This one sorted them on Dato-DXd only, and the authors read the result as indicating that TROP2 NMR may be a predictive biomarker for Dato-DXd.

The membrane share of TROP2, scored cell by cell

QCS, the method behind the score, is a fully supervised computational pathology approach that quantifies TROP2 in individual NSCLC cells and in their subcellular components on immunohistochemistry-stained slides. From the membrane and cytoplasmic signals the authors derived NMR, the ratio of membrane TROP2 to membrane plus cytoplasmic TROP2. The score therefore measures how much of a cell’s TROP2 sits at its surface, rather than how much TROP2 the cell carries overall.

The authors start from a known limitation of the current method.

“However, TROP2 expression by pathologist assessment using conventional immunohistochemistry with visual scoring does not predict response to TROP2-directed antibody–drug conjugates in NSCLC.”Garassino et al., Nature Medicine (2026)

One trial to build the score, a second to set and test the cut-point

NMR was first developed in the biomarker-evaluable population of the TROPION-PanTumor01 study, 115 patients with advanced NSCLC. The cut-point was then optimized and validated in TROPION-Lung01, in the 221 patients with nonsquamous NSCLC and no actionable genomic alteration whom the authors name as the primary intended-use population, drawn from a biomarker-evaluable population of 352.

As the abstract describes it, both the optimization and the validation of the cut-point took place in TROPION-Lung01, and the analysis is confined to biomarker-evaluable patients. Both points bear on how the cut-point will perform in an independent cohort, which is the test a predictive biomarker generally has to pass before it is used to select treatment.