Enhertu HER2-Mutant NSCLC Trial Meets PFS Endpoint but Overall Survival Remains Unsettled

Daiichi Sankyo and AstraZeneca’s Enhertu (trastuzumab deruxtecan) reduced the risk of disease progression versus pembrolizumab plus platinum-pemetrexed in treatment-naïve HER2-mutant non-small cell lung cancer (NSCLC), the Phase 3 DESTINY-Lung04 trial presented at the 2026 World Conference on Lung Cancer (WCLC) in Seoul showed. Median progression-free survival (PFS) was 14.3 months versus 8.3 months (hazard ratio 0.63; p<0.0001), and 70.0% of patients responded versus 44.5%. The Enhertu HER2-mutant NSCLC results nonetheless leave the first-line filing contested, because overall survival was immature and trended against the antibody-drug conjugate, and adjudicated interstitial lung disease (ILD) reached 20.8%.

Julia Rotow of the Dana-Farber Cancer Institute presented the data in the presidential symposium. DESTINY-Lung04 randomised 454 patients, 227 per arm, to T-DXd 5.4 mg/kg every three weeks or to investigator’s choice of cisplatin or carboplatin plus pemetrexed and pembrolizumab. At the 9 June 2026 data cutoff, median follow-up was 21.6 months in the T-DXd arm and 20.4 months in the control arm.

Progression-free survival: a clear win

The PFS benefit was consistent across the prespecified subgroups, including brain metastases, liver metastases, smoking status, exon 19 versus exon 20 mutations, and de novo versus recurrent disease. Duration of response also favoured T-DXd, at 13.4 months versus 9.7 months. Complete response rates were identical at 1.8% in both arms.

Ad

Time to second progression or death (PFS2) was 22.7 months on T-DXd versus 17.3 months on control (HR 0.80; 95% CI 0.62-1.02). The confidence interval crosses 1, so the sponsors’ argument that the first-line PFS gain is not fully erased by later therapy is not yet statistically conclusive.

Overall survival: unsettled

At the planned PFS analysis the overall-survival look was 46.9% mature and not formally tested. Median overall survival was 29.3 months on T-DXd versus 33.1 months on control (HR 1.15; 95% CI 0.88-1.52). There were 115 deaths in the T-DXd arm (50.7%) and 98 in the control arm (43.2%). The plenary slide stated: “No OS benefit was observed with T-DXd; subsequent treatments may have impacted results.” The Kaplan-Meier curves separate in favour of control from about month 20 and do not close.

Both sponsors framed the result the same way, saying interpretation “may be limited” by imbalanced subsequent therapy. Among patients who received any next treatment, HER2-directed agents were used in 39% after T-DXd versus 72% after control. Control patients were more often rescued with a HER2 drug, which can flatten or reverse an overall-survival curve even when first-line PFS is real. That does not dispose of a point estimate above 1, and the confidence interval still spans both harm and modest benefit.

Interstitial lung disease: the access-limiting risk

Adjudicated ILD or pneumonitis occurred in 20.8% of patients on T-DXd versus 2.3% on control. The grade split on T-DXd was 3.1% grade 1 (7 patients), 13.3% grade 2 (30 patients), 2.2% grade 3 (5 patients), 0.4% grade 4 (1 patient), and 1.8% grade 5 (4 patients). Most events, 78.7%, were grade 1-2 and 66% resolved. AstraZeneca told Fierce Pharma that nearly all grade 5 ILD cases had delayed steroid initiation or suboptimal dosing.

Grade 3 or higher treatment-related adverse events were otherwise comparable, at 34.1% on T-DXd versus 33.6% on control, despite a longer median exposure on T-DXd (12.3 months versus 7.1 months). Grade 3 or higher neutropenia was 11.1% versus 14.1%. For access and medical-governance teams the operative question is not adverse-event parity. It is whether first-line community and regional centres can run ILD monitoring tightly enough that a 1.8% fatal ILD rate is judged acceptable next to oral TKIs.

Enhertu HER2-mutant NSCLC: the competitive and regulatory picture

The competitive context shifted while DESTINY-Lung04 was running. Enhertu holds a 2022 accelerated approval in previously treated HER2-mutant NSCLC that has not been converted to full approval. In 2026 the FDA granted first-line accelerated approvals to two oral HER2 TKIs: Boehringer Ingelheim’s Hernexeos (zongertinib) and Bayer’s Hyrnuo (sevabertinib), on single-arm response rates of about 76% and 75%. Zongertinib’s first-line median PFS in Beamion LUNG-1 was 14.4 months, numerically matching DESTINY-Lung04, without the ILD typical of an antibody-drug conjugate.

The cross-trial comparison is crude: DESTINY-Lung04 is randomised against active standard of care, while the TKI labels rest on single-arm cohorts. Still, the practical message for guidelines and HTA bodies is blunt. An oral TKI posted a similar first-line PFS point estimate with higher complete-response rates and without ADC-typical ILD. DESTINY-Lung04’s unique asset is the head-to-head PFS win against pembro-chemo. Its unique liabilities are the survival direction, fatal ILD, and intravenous administration.

The companies plan to share the package with the FDA and other regulators. A plausible outcome range is a first-line label on PFS with prominent ILD warnings and a requirement for mature overall survival, or a request to wait for the next survival look. Conversion of the 2L accelerated approval remains a separate, unresolved thread.

What it means for access and reimbursement

For payers, coverage of second-line Enhertu is not automatically coverage of first-line Enhertu. Expect step-through the oral TKIs, ILD-monitoring requirements, and requests for survival updates. The evidence story should lead with the randomised PFS win and the 70% response rate, not with “first HER2-directed first-line option”, which the TKI labels already contest.

Diagnostics carry the cleanest system-level argument in markets where reflex next-generation sequencing is incomplete: first-line use of any HER2-directed agent raises the cost of missing HER2 testing at diagnosis. The live clinical question is sequencing, TKI then ADC versus ADC then TKI, rather than ADC versus pembro-chemo.

Source: IASLC WCLC 2026 (Seoul), abstract PL03.08; ClinicalTrials.gov NCT05048797.