T-DXd OS benefit: DESTINY-Breast04 subgroup analysis

HER2-Low Metastatic Breast Cancer and the DESTINY-Breast04 Benchmark

Trastuzumab deruxtecan (T-DXd) is now confirmed as a standard of care after prior chemotherapy in patients with HER2-low metastatic breast cancer, a designation that encompasses tumours scoring IHC 1+ or IHC 2+/ISH− on HER2 testing. The DESTINY-Breast04 trial provided the pivotal evidence base, and its subgroup forest plot data allow a granular examination of how overall survival (OS) benefit distributes across clinically meaningful patient characteristics.

The trial’s subgroup analysis spans eight patient characteristics: HER2 IHC status, number of prior chemotherapy lines, age, race, geographic region, ECOG performance status, visceral disease at baseline, and prior CDK4/6 inhibitor use. Across all of these, T-DXd consistently demonstrated an OS benefit over treatment of physician’s choice (TPC), but the magnitude and statistical certainty of that benefit vary considerably — a distinction that carries real weight in clinical decision-making. According to The New England Journal of Medicine, subgroup analyses from randomised trials remain among the most informative tools for personalising oncology treatment when interpreted with appropriate caution.

Understanding which subgroups derive the greatest and least benefit from T-DXd is essential for clinicians making treatment sequencing decisions. The DESTINY-Breast04 forest plot provides hazard ratios and 95% confidence intervals for each subgroup, enabling a structured comparison. Regulatory bodies including the FDA and the EMA have both evaluated this evidence in the context of T-DXd’s approval for HER2-low disease.

Overall Survival Hazard Ratios Across Eight Patient Subgroups

Across all eight subgroups examined in DESTINY-Breast04, T-DXd produced hazard ratios below 1.0 for overall survival, indicating a consistent directional benefit over TPC. The magnitude of that benefit, however, ranges from an HR of 0.35 — representing a 65% reduction in the risk of death — to an HR of 0.78, where the upper confidence interval bound crosses 1 and statistical significance is less certain.

The HER2 IHC subgroup comparison shows that IHC 1+ patients had an HR of 0.65 (95% CI 0.49–0.86), while IHC 2+/ISH− patients had an HR of 0.72 (95% CI 0.51–1.01). Both subgroups benefit from T-DXd, but the upper confidence interval bound for IHC 2+/ISH− patients barely touches 1.0, warranting careful interpretation. Accurate HER2 IHC testing is therefore not merely a diagnostic formality but a direct determinant of the strength of the OS benefit that can be anticipated.

Geographic region and race subgroups showed notably consistent HRs. White patients had an HR of 0.68 (95% CI 0.50–0.93), Asian patients had an HR of 0.68 (95% CI 0.48–0.96), and patients classified as Other race had an HR of 0.55 (95% CI 0.28–1.07). The Other race subgroup showed the numerically lowest HR, but the wide confidence interval — likely reflecting a smaller sample — prevents definitive conclusions. Across geographic regions, Europe and Israel had an HR of 0.67 (95% CI 0.49–0.91), Asia had an HR of 0.69 (95% CI 0.49–0.98), and North America had an HR of 0.66 (95% CI 0.38–1.13).

The Subgroups That Benefit Most: Visceral Disease, ECOG Status, and Prior Lines

Four patient subgroups in DESTINY-Breast04 stand out as deriving the greatest overall survival benefit from T-DXd, with hazard ratios between 0.35 and 0.63 and confidence intervals that remain entirely below 1.0 (or very close to it with clinical significance).

No Visceral Disease at Baseline

The most striking subgroup finding in the entire DESTINY-Breast04 forest plot is the absence of visceral disease at baseline. Patients in this category had a median OS that was Not Estimable (NE) in the T-DXd arm — meaning the median survival had not yet been reached at the time of data cut — versus 15.7 months (95% CI 12.9–20.6) in the TPC arm. The HR was 0.35 (95% CI 0.18–0.70). This represents a 65% reduction in the risk of death, a magnitude of benefit that substantially exceeds any other subgroup in the analysis.

“Patients without visceral disease at baseline had a median OS that was Not Estimable with T-DXd versus 15.7 months with TPC — an HR of 0.35 that represents the most substantial overall survival improvement in the entire DESTINY-Breast04 subgroup analysis.”

ECOG Performance Status 0

Patients with an ECOG performance status of 0 — indicating fully active functional capacity — had a median OS of 25.9 months (95% CI 23.0–29.3) with T-DXd versus 19.4 months (95% CI 15.1–22.8) with TPC, yielding an HR of 0.62 (95% CI 0.46–0.83). This compares favourably with ECOG PS 1 patients, who had an HR of 0.74 (95% CI 0.54–1.01), where the upper bound of the CI approaches 1. Better baseline functional status thus translates directly into a more pronounced and statistically certain survival benefit.

One Prior Line of Chemotherapy

Patients who had received only one prior line of chemotherapy had a median OS of 25.5 months (95% CI 23.4–28.9) with T-DXd versus 18.2 months (95% CI 15.6–22.5) with TPC (HR 0.62; 95% CI 0.46–0.83). This is notably better than the ≥2 prior lines subgroup (HR 0.78; 95% CI 0.57–1.07), where the upper CI bound crosses 1. Earlier use of T-DXd in the treatment sequence — before the tumour has been exposed to multiple lines of cytotoxic therapy — appears to preserve a greater OS benefit.

No Prior CDK4/6 Inhibitor Use

Patients who had not previously received a CDK4/6 inhibitor had a median OS of 30.3 months (95% CI 23.0–35.1) with T-DXd versus 22.4 months (95% CI 15.6–27.2) with TPC (HR 0.63; 95% CI 0.41–0.99). This represents the longest absolute median OS in the T-DXd arm across all subgroups. Patients who had received prior CDK4/6 inhibitors still benefited (HR 0.71; 95% CI 0.54–0.94), but the absolute OS gain was smaller, suggesting that prior CDK4/6 inhibitor exposure modestly attenuates the T-DXd benefit.

Where the Benefit Is Less Certain: Age, Prior Lines, and IHC Status

Three subgroups in DESTINY-Breast04 showed numerically higher hazard ratios and confidence intervals whose upper bounds crossed or closely approached 1.0, indicating that the OS benefit — while directionally consistent — is statistically less certain in these populations.

Patients Aged ≥65 Years

Older patients (≥65 years) had a median OS of 24.4 months (95% CI 18.4–28.0) with T-DXd versus 19.5 months (95% CI 11.1–30.2) with TPC, yielding an HR of 0.77 (95% CI 0.50–1.19). The upper bound of the CI exceeds 1.0, meaning the trial cannot exclude the possibility of no OS benefit in this age group. Notably, the TPC arm’s confidence interval is exceptionally wide (11.1–30.2 months), reflecting greater outcome variability in older patients receiving standard chemotherapy. The contrast with the <65 years subgroup (HR 0.64; 95% CI 0.50–0.82, CI entirely below 1) is clinically meaningful. According to ASCO guidelines on geriatric oncology, treatment decisions in patients aged ≥65 should incorporate comprehensive geriatric assessment alongside trial subgroup data.

Two or More Prior Lines of Chemotherapy

Patients with ≥2 prior chemotherapy lines had a median OS of 18.1 months (95% CI 16.1–21.5) with T-DXd versus 14.0 months (95% CI 10.8–19.1) with TPC (HR 0.78; 95% CI 0.57–1.07). The 4.1-month absolute difference is clinically meaningful, but the upper CI bound crossing 1.0 means this benefit does not meet conventional thresholds for statistical significance. This contrasts sharply with the 1 prior line subgroup (HR 0.62; CI entirely below 1), reinforcing the principle that earlier deployment of T-DXd in the treatment sequence preserves more of its survival benefit.

IHC 2+/ISH− Patients

IHC 2+/ISH− patients had a median OS of 23.6 months (95% CI 20.0–26.0) with T-DXd versus 17.1 months (95% CI 13.1–21.7) with TPC (HR 0.72; 95% CI 0.51–1.01). The upper CI bound of 1.01 is only marginally above 1, and the 6.5-month absolute OS difference is clinically substantial. The IHC 1+ subgroup had a slightly lower HR of 0.65 (95% CI 0.49–0.86), with a CI entirely below 1. This difference may reflect biological heterogeneity within the HER2-low spectrum — IHC 2+/ISH− tumours may have lower HER2 protein density, potentially affecting the drug-to-antibody ratio and intracellular payload delivery of T-DXd. This is an area of active investigation, with Nature publishing ongoing research into HER2 expression heterogeneity and antibody-drug conjugate efficacy.

Clinical Implications for Patient Selection and Treatment Stratification

The DESTINY-Breast04 subgroup data translate directly into actionable guidance for treatment stratification in HER2-low metastatic breast cancer. T-DXd is confirmed as a standard of care after prior chemotherapy across the broad HER2-low population, but the subgroup analysis identifies specific patient characteristics that should inform the urgency and confidence with which T-DXd is recommended.

Prioritise T-DXd for Patients With Optimal Baseline Characteristics

Patients with ECOG PS 0, one prior line of chemotherapy, no visceral disease, and no prior CDK4/6 inhibitor use represent the population most likely to derive substantial, statistically certain OS benefit from T-DXd. Clinicians should prioritise this agent for such patients, ideally before the accumulation of additional treatment lines that appear to attenuate the benefit. The 30.3-month median OS observed in CDK4/6 inhibitor-naive patients is particularly noteworthy, as it suggests that T-DXd retains meaningful activity even in patients who have not yet been exposed to targeted endocrine therapy combinations.

Individualised Assessment for Higher-Risk Subgroups

For patients aged ≥65 years or those with ≥2 prior chemotherapy lines, the directional OS benefit is present but statistically uncertain. Clinicians should not withhold T-DXd from these patients on the basis of subgroup data alone, but should incorporate a comprehensive clinical assessment — including functional status, comorbidities, and patient preferences — when weighing the expected benefit against the known toxicity profile of T-DXd, including interstitial lung disease risk. This approach aligns with the ESMO framework for treatment decision-making in metastatic breast cancer.

HER2 IHC Testing Accuracy Is Non-Negotiable

The slightly lower HR in IHC 1+ patients (0.65) compared to IHC 2+/ISH− patients (0.72) underscores the importance of accurate and reproducible HER2 IHC testing. Both subgroups benefit, but the precision of the scoring directly determines which patients are identified as HER2-low candidates. Misclassification of IHC 0 tumours as IHC 1+ — or vice versa — has direct therapeutic consequences. Standardisation of HER2 testing protocols across pathology laboratories remains an active area of focus for regulatory bodies and professional societies. The PatSnap resources library contains further analysis of HER2 testing standardisation efforts and their patent landscape implications.

Broad Applicability Across Demographics

Despite minor numerical variation, T-DXd demonstrated consistent OS benefit across all racial and geographic subgroups examined. White and Asian patients had identical HRs of 0.68, and all three geographic regions (Asia, Europe/Israel, North America) showed HRs between 0.66 and 0.69 — a range so narrow as to be clinically indistinguishable. This consistency supports the broad applicability of T-DXd across diverse patient populations globally, consistent with PatSnap’s ongoing analysis of oncology drug approvals across major regulatory jurisdictions.

Treatment Sequencing and CDK4/6 Inhibitor Context

The finding that patients without prior CDK4/6 inhibitor use derived a greater T-DXd OS benefit (HR 0.63 vs 0.71 for those with prior use) has implications for treatment sequencing in hormone receptor-positive HER2-low disease — the predominant subtype in the trial. As CDK4/6 inhibitors become near-universal in the first-line setting for HR+/HER2-low metastatic breast cancer, the majority of patients receiving T-DXd in clinical practice will have had prior CDK4/6 inhibitor exposure. The DESTINY-Breast04 data confirm that T-DXd remains effective in this context, but the sequencing question — and the potential role of T-DXd earlier in the treatment algorithm — warrants continued investigation in prospective trials.