Why This Study Matters
TROP2 is expressed in approximately 90% of NSCLC tumor samples, yet the paper states that TROP2 expression by pathologist assessment using conventional immunohistochemistry with visual scoring does not predict response to TROP2-directed antibody–drug conjugates in NSCLC. Dato-DXd is a TROP2-directed ADC; in the phase 3 TROPION-Lung01 trial it improved progression-free survival versus docetaxel in previously treated advanced NSCLC. A way to select patients who are more likely to benefit has been an unmet need.
Study Design
The authors developed TROP2 quantitative continuous scoring (QCS), a fully supervised computational pathology approach that quantifies TROP2 in individual tumor cells and subcellular compartments from IHC-stained whole-slide images. TROP2 NMR is the ratio of membrane relative to membrane + cytoplasmic TROP2. Discovery used TROPION-PanTumor01 (a first-in-human phase 1 study); the cut-point was then optimized and validated retrospectively in TROPION-Lung01, a global, multicenter, randomized, open-label phase 3 study of Dato-DXd versus docetaxel. Preclinical experiments in NSCLC cell lines and xenografts supported the findings. A sample was considered TROP2 NMR-positive if ≥75% of tumor cells had a TROP2 NMR of ≤0.56 (Q75 quantile, cut-point ≤0.56).
180 NSCLC patients enrolled in TROPION-PanTumor01 (ITT), of whom 115 were biomarker-evaluable; 352 biomarker-evaluable in TROPION-Lung01; 221 in the NSQ, non-AGA subset (108 Dato-DXd; 113 docetaxel).
Patient Population
The primary intended-use population was patients with NSQ NSCLC without actionable genomic alterations. In the full TROPION-Lung01 biomarker-evaluable population (n = 352), 23.8% and 21.7% of patients had squamous histology and 14.0% and 16.1% had AGAs in the Dato-DXd and docetaxel arms, respectively; the NSQ, non-AGA subset was 62.8% of that population. Number of sites and randomization ratio of TROPION-Lung01 are not specified in the source text used for this article.
Primary Endpoint Results
The biomarker was optimized on PFS. In TROPION-PanTumor01 (n = 115), 73% (84 of 115) of samples were NMR-positive, and NMR positivity was associated with longer PFS (median 5.5 versus 2.9 months; HR 0.39, 95% CI 0.22–0.68, P = 0.00054) and higher ORR (22.6% versus 3.2%). In TROPION-Lung01 NSQ, non-AGA patients, PFS with Dato-DXd was median 7.2 versus 4.0 months (HR 0.40, 95% CI 0.25–0.64) for NMR-positive versus NMR-negative; with docetaxel it was median 4.1 versus 4.4 months (HR 0.94, 95% CI 0.59–1.49), interaction P = 0.0098.
Secondary and Subgroup Analyses
| Cohort / analysis | NMR-positive vs NMR-negative | Statistics |
|---|---|---|
| TROPION-PanTumor01 BEP (n = 115): PFS | Median 5.5 vs 2.9 months | HR 0.39, 95% CI 0.22–0.68, P = 0.00054 |
| TROPION-PanTumor01 BEP: ORR | 22.6% vs 3.2% | Not specified in source |
| TROPION-Lung01 NSQ, non-AGA (n = 221), Dato-DXd: PFS | Median 7.2 vs 4.0 months | HR 0.40, 95% CI 0.25–0.64 |
| Same, docetaxel: PFS | Median 4.1 vs 4.4 months | HR 0.94, 95% CI 0.59–1.49; interaction P = 0.0098 |
| Same, Dato-DXd: ORR | 36.8% vs 22.5% | Not specified in source |
| Same, Dato-DXd: OS | Median 17.0 vs 10.8 months | HR 0.45, 95% CI 0.29–0.74; interaction P = 0.014 |
| TROPION-Lung01 full BEP (n = 352), Dato-DXd: PFS | Median 6.9 vs 2.9 months | Not specified in source |
| Same, Dato-DXd: ORR | 32.7% vs 16.9% | Not specified in source |
Additional analyses: replicate adjacent tissue sections showed consistent associations with PFS, and in the phase 3 NEPTUNE and POSEIDON first-line datasets, TROP2 NMR did not identify patients with longer PFS in any treatment arm, which the authors interpret as indicating the biomarker is not prognostic. Subgroup-level statistics for squamous or AGA-positive patients are not specified in the source text used here.
Safety Profile
Safety by NMR status was assessed in 344 treated patients in the TROPION-Lung01 biomarker-evaluable population. Rates of all-grade and grade ≥3 treatment-related adverse events with Dato-DXd were similar regardless of TROP2 NMR status. Rates of all-grade treatment-related adverse events of special interest (stomatitis, ocular surface events and adjudicated drug-related interstitial lung disease) were higher in the NMR-positive than NMR-negative group, although most events were grade 1 or 2 in both groups.
| Safety finding (Dato-DXd) | NMR-positive | NMR-negative |
|---|---|---|
| Grade ≥3 stomatitis (the only grade ≥3 AESI with higher incidence in NMR-positive) | 7% | 3% |
| Other TRAE and AESI rates | Not specified in the text of source (reported in Extended Data Table 3) | Not specified in the text of source (reported in Extended Data Table 3) |
Interpretation and Broader Context
The authors conclude that TROP2 NMR may be the first TROP2 computational pathology biomarker predicting clinical response to Dato-DXd in NSCLC. Their proposed rationale is that cytoplasmic enrichment of TROP2 may reflect more active receptor trafficking and internalization; they note that TROP2 NMR is not a direct measurement of Dato-DXd internalization. Roche Diagnostics and AstraZeneca are developing the VENTANA TROP2 (EPR20043) RxDx Device, which received FDA Breakthrough Device Designation in April 2025. Prospective validation is planned within ongoing phase 3 trials including AVANZAR (NCT05687266), TROPION-Lung07 (NCT05555732), TROPION-Lung10 (NCT06357533) and TROPION-Lung17 (NCT07291037).
Analyses of samples from patients in TROPION-Lung01 were retrospective and exploratory, reflecting standard initial biomarker identification and development approaches, and further studies are required to validate TROP2 NMR prospectively as predictive for Dato-DXd in NSCLC.
— Garassino et al., Nature Medicine, 2026 (Discussion)
Limitations
The authors state that the TROPION-Lung01 analyses were retrospective and exploratory and that prospective validation is required. The biomarker cut-point was optimized in the same TROPION-Lung01 subset used to report the outcomes; bootstrap and cross-validation were used to assess overfitting. The authors also state that TROP2 NMR was developed using Dato-DXd datasets and that its applicability to other TROP2 ADCs or tumor types is unknown. Statistics for squamous and AGA-positive subgroups and several adverse-event rates are not specified in the source text used for this article. Funding: research funded by AstraZeneca; TROPION-PanTumor01 and TROPION-Lung01 were sponsored by Daiichi Sankyo, Inc. and cofunded with AstraZeneca.