Antibody–drug conjugates have become one of the most important therapeutic classes in metastatic breast cancer, producing substantial improvements in response rate and progression-free survival across HER2-positive, HER2-low, hormone receptor–positive, and triple-negative disease. As these agents move into earlier treatment lines and increasingly compete with other highly active therapies, however, a more difficult question has emerged: how reliably does an improvement in progression-free survival predict an improvement in overall survival?
A meta-epidemiologic analysis published in the Journal of the National Cancer Institute on September 10, 2026, directly addresses this issue. The investigators examined randomized phase II–IV trials of antibody–drug conjugates in metastatic breast cancer that reported hazard ratios for both progression-free survival and overall survival. Their objective was not to reassess whether ADCs are effective, but to quantify how closely the magnitude of benefit observed for PFS tracks with the magnitude of benefit ultimately observed for OS.
The conclusion is clinically important: treatment effects on PFS were consistently larger than those observed for OS, and nearly one-quarter of trials demonstrated a statistically significant PFS advantage without a corresponding statistically significant overall-survival benefit.
This does not diminish the value of ADCs. It does, however, reinforce the need to distinguish between delaying progression and extending survival when interpreting increasingly complex metastatic breast cancer trials.
ADCs Have Changed the Therapeutic Landscape, But Also the Meaning of Trial Endpoints
Progression-free survival is attractive as a clinical-trial endpoint because it can be measured earlier than overall survival and is less directly affected by therapies patients receive after progression. For rapidly evolving drug classes such as ADCs, this allows trials to generate efficacy signals sooner and can accelerate regulatory decision-making.
The difficulty is that PFS and OS do not measure the same thing.
PFS evaluates how long a therapy delays progression or death while a patient remains within the treatment strategy being studied. OS captures the entire remainder of the patient’s clinical course, including crossover, subsequent therapies, treatment access, resistance biology, and the effectiveness of later-line options.
As metastatic breast cancer becomes increasingly treatable across multiple successive lines, those postprogression factors can substantially dilute, or occasionally obscure, the survival effect attributable to the randomized treatment.
The new JNCI analysis was designed to quantify that divergence specifically in the ADC era.
Twenty-One Randomized Trials Were Included
The investigators identified 21 randomized trials, including 18 phase III studies, evaluating nine different ADCs in metastatic breast cancer. Eligible studies had to report hazard ratios for both PFS and OS.
The primary statistical measure was the ratio of hazard ratios, defined as:
- rHR = HR for PFS / HR for OS
An rHR below 1 indicates that the relative treatment effect is larger for PFS than for OS. This method allows comparison of the magnitude of benefit across the two endpoints rather than simply asking whether each endpoint reached statistical significance.
The pooled rHR across the included studies was 0.79 (95% CI, 0.73–0.86), indicating that ADC treatment effects were systematically stronger for progression-free survival than for overall survival. That difference is not unexpected, but its consistency across trials is important.
One in Four Trials Improved PFS Without Demonstrating an OS Benefit
Perhaps the most clinically relevant result is the endpoint concordance analysis. Statistical significance for PFS and OS was concordant in 76% of trials, but 24% showed a significant PFS benefit without a statistically significant OS improvement.
In other words, approximately one in four ADC trials crossed the threshold for statistical significance for disease control but not for survival. This finding should not be interpreted as evidence that those treatments lacked clinical value. A meaningful delay in progression can preserve symptom control, postpone chemotherapy or another toxic therapy, improve quality of life, or extend time without disease worsening.
But it does challenge a common assumption in trial interpretation: that a positive PFS result can automatically be viewed as a reliable preview of eventual survival benefit. The data suggest that this relationship is real, but incomplete.
The PFS–OS Association Was Only Moderate
At the trial level, the association between treatment effects on PFS and OS was moderate, with an R² of 0.62 and a Spearman correlation coefficient of 0.67. This means that trials producing larger PFS improvements generally tended to produce larger OS improvements, but the relationship was far from perfect.
From a methodological perspective, that distinction matters when PFS is used as a surrogate endpoint. A valid surrogate should reliably predict the effect of treatment on the clinically definitive outcome. If the relationship is only moderate, then PFS may provide useful information about likely OS direction without fully predicting the magnitude of survival benefit.
The authors therefore characterize their findings as an exploratory framework for contextualizing PFS and OS in ADC trials rather than as final validation of PFS as a surrogate endpoint.
Triple-Negative Breast Cancer Showed the Largest Numerical Discrepancy
The analysis found that the numerical difference between PFS and OS treatment effects was particularly pronounced in triple-negative breast cancer, where the rHR was 0.72. That result is clinically plausible.
TNBC historically has had fewer effective postprogression treatment options than hormone receptor–positive or HER2-positive breast cancer, although this is changing rapidly with the introduction of ADCs, immunotherapy, PARP inhibitors, and biomarker-driven treatment.
The interpretation is not simply that ADCs are less capable of improving OS in TNBC. Rather, the analysis suggests that the magnitude of PFS benefit may overstate the magnitude of survival benefit when the two are compared proportionally. The reasons may be complex and can include aggressive disease biology, postprogression attrition, crossover, later-line therapy, and differences in trial design.
Because the available source provides the abstract rather than the complete manuscript, detailed subtype-specific analyses beyond those reported in the abstract cannot be reliably reconstructed.
Crossover Appears to Widen the PFS–OS Gap
Trials permitting crossover also demonstrated one of the largest numerical discrepancies between PFS and OS, with an rHR of 0.71. This is a particularly important observation in modern breast cancer trial design.
When patients in the control arm are allowed to receive the experimental ADC after progression, the difference in overall survival between randomized groups can become smaller even when the drug is highly effective. This is not necessarily evidence that the ADC provides little survival value. Instead, both groups may ultimately receive the active treatment, only at different points in the disease course.
Crossover therefore complicates OS interpretation because it changes the clinical question.
The comparison becomes less about:
- ADC versus no ADC
and more about:
- earlier ADC versus later ADC.
This distinction is increasingly important as highly active ADCs move into earlier lines of metastatic breast cancer.
PFS Can Remain Clinically Meaningful Even Without OS Separation
One risk of overemphasizing overall survival is treating PFS as if it has little intrinsic value. That would be an equally problematic interpretation.
A therapy that produces durable disease control can meaningfully delay symptoms, deterioration, additional chemotherapy, hospitalization, or the need to transition to a more toxic treatment strategy. In metastatic disease, the period during which a patient remains clinically stable and treatment remains effective is itself relevant.
The appropriate lesson from this meta-analysis is therefore not that PFS should be dismissed. It is that PFS and OS should be interpreted as related but distinct outcomes. The stronger the PFS effect, the more likely a favorable OS direction may be, but the absolute survival effect can still be altered by subsequent therapy, crossover, treatment sequencing, and disease subtype.
ADC Sequencing Makes OS Increasingly Difficult to Interpret
The issue becomes even more important as metastatic breast cancer gains multiple active ADCs. A patient may now receive one ADC followed by another, or an ADC may be moved from third line to second line and then into first-line therapy. In such settings, overall survival becomes a composite of several therapeutic sequences rather than the isolated effect of the randomized intervention.
This raises a difficult question for modern drug development: when a therapy clearly improves PFS earlier in the disease course but patients can receive the same or another highly effective ADC later, how much OS separation should realistically be expected?
A small or absent OS difference may indicate that early use does not improve lifetime survival. But it may also indicate that effective salvage therapy has narrowed the survival difference between treatment groups. This is why postprogression treatment reporting and crossover design are becoming increasingly important for the interpretation of contemporary ADC trials.
Regulatory and Clinical Decisions Should Avoid Endpoint Simplification
The findings also have implications for how new ADC indications are evaluated. PFS can support regulatory approval when treatment effects are large, consistent, and associated with meaningful clinical benefit. However, when a treatment is moved into an earlier setting where patients already have access to several effective subsequent therapies, reliance on PFS alone can make it difficult to determine whether earlier exposure genuinely improves long-term patient outcomes.
For clinicians, the decision is even broader than PFS versus OS.
Treatment selection also depends on toxicity, quality of life, treatment convenience, duration of therapy, biomarker profile, intracranial efficacy, resistance mechanisms, cost, and the consequences of using one ADC before another. The meta-analysis therefore reinforces a more mature interpretation of clinical evidence: no single endpoint should be read in isolation from the treatment sequence in which it was generated.
Important Limitations of the Current Evidence
The study is a meta-epidemiologic analysis, meaning that it evaluates relationships across trials rather than patient-level relationships between progression and survival. It therefore cannot determine how well an individual’s PFS predicts that individual’s OS.
The analysis also combines studies of nine ADCs conducted across different breast cancer subtypes, lines of therapy, comparator regimens, crossover rules, and clinical eras. Those differences may influence both the magnitude and interpretation of the PFS–OS relationship.
The authors themselves describe the findings as exploratory and emphasize the need for validation across diverse clinical settings.
The currently available source is the accepted-manuscript abstract rather than the complete article, so detailed subgroup methodology, individual trial contributions, the precise surrogate-threshold analysis, and additional sensitivity analyses are not available here and should not be inferred beyond what the abstract reports.
The Bottom Line
This JNCI meta-epidemiologic analysis provides an important caution for interpreting the rapidly expanding ADC literature in metastatic breast cancer.
Across 21 randomized trials evaluating nine ADCs, treatment effects were consistently larger for PFS than for OS, with a pooled ratio of hazard ratios of 0.79. Although endpoint direction was generally concordant, 24% of trials demonstrated a statistically significant PFS benefit without a statistically significant OS improvement. The relationship between the magnitude of PFS and OS benefit was moderate rather than strong, with an R² of 0.62.
The largest numerical discrepancies were observed in TNBC and in trials allowing crossover, both of which highlight how disease biology and postprogression treatment can influence the eventual survival signal.
The clinical message is not that PFS lacks value or that ADC efficacy has been overstated. It is that delaying progression and extending survival are not interchangeable outcomes, particularly in a treatment landscape where patients may receive multiple highly active therapies sequentially.
As ADCs move earlier and treatment sequencing becomes more complex, breast cancer trials will increasingly need to answer not only whether a drug controls disease longer, but whether the timing of that control ultimately changes the patient’s entire therapeutic course.
Reference
- Costa de Almeida LF, Cappellaro AP, Noronha MM, et al. Treatment Effect Discrepancy Between PFS and OS in Antibody-Drug Conjugate Trials for Metastatic Breast Cancer: A Meta-Epidemiologic Analysis. JNCI: Journal of the National Cancer Institute. Published September 10, 2026. doi:10.1093/jnci/djag324.
