Endpoint surrogacy in oncology Phase 3 randomised controlled trials

Jianrong Zhang, Meagan Pilar, Xiaofei Wang, Jingxia Liu, Herbert Pang, Ross C. Brownson, Graham A. Colditz, Wenhua Liang, Jianxing HeBritish 2020 · DOI · OpenAlex

Abstract

Endpoint surrogacy is an important concept in oncology trials. Using a surrogate endpoint like progression-free survival as the primary endpoint—instead of overall survival—would lead to a potential faster drug approval and therefore more cancer patients with an earlier opportunity to receive the newly approved drugs.

Introduction

In the landscape of clinical oncology, the definitive measure of a drug's efficacy is Overall Survival (OS), representing the total time from randomization until death from any cause. While OS is the "gold standard" for clinical benefit, its assessment often requires prolonged follow-up periods and large patient cohorts, which can delay the regulatory approval of potentially life-saving therapies. To address this, researchers frequently employ surrogate endpoints, most notably Progression-Free Survival (PFS)—defined as the time from randomization to objective tumor progression or death.

The utilization of surrogate endpoints is intended to expedite drug development. However, the validity of these surrogates—their ability to accurately predict the ultimate clinical benefit (OS)—remains a subject of intense debate. If a surrogate does not reliably correlate with survival, there is a risk of approving treatments that offer transient tumor control without extending life.

Methodology and the Meta-Analytic Approach

A recent editorial reflects on a comprehensive systematic review conducted by Belin et al., which scrutinized 91 validation studies across 24 distinct cancer types. The gold standard for validating a surrogate is the meta-analytic approach. This method evaluates the correlation between the treatment effect on the surrogate (e.g., the Hazard Ratio for PFS) and the treatment effect on the final outcome (the Hazard Ratio for OS) across multiple trials.

The strength of this relationship is typically quantified using the coefficient of determination ($R^2$). In regulatory and academic circles, an $R^2 \ge 0.6$ is often cited as the threshold for "good" surrogacy, indicating that at least 60% of the variance in the OS treatment effect can be explained by the variance in the surrogate treatment effect.

Results: The Inconsistency of Surrogacy

The findings from the analysis of 91 validation studies reveal a concerning lack of consistency in surrogate performance:

  • Low Validation Success: Only 52% of the reviewed studies concluded that PFS served as a valid surrogate for OS.
  • Weak Correlation Strength: Among the studies that did validate PFS, only about half (51%) achieved the $R^2 \ge 0.6$ threshold for "good" surrogacy.
  • Methodological Heterogeneity: There is significant variation in how surrogacy is calculated and reported. Researchers utilize diverse metrics, including hazard ratios, differences in median survival, and milestone survival rates, complicating the comparison of results across different studies.

Discussion: Indication-Specific Validation and the Immunotherapy Gap

One of the most critical insights is that surrogacy is not a universal attribute. A surrogate may be highly predictive in one disease context (e.g., colorectal cancer) but fail in another (e.g., lung cancer). This is particularly evident in the era of immunotherapy. Because immune-checkpoint inhibitors function differently than traditional cytotoxic chemotherapy—often resulting in delayed responses or "pseudoprogression"—traditional tumor-growth metrics like PFS may not align with survival outcomes in the same manner.

Furthermore, the "Methodological Wild West" of surrogacy validation highlights a need for standardization. The lack of consensus on whether to evaluate surrogacy at the patient level or the trial level, and what $R^2$ threshold constitutes "validity," creates significant hurdles for policy-level decision-making.

Conclusion and Future Directions

To improve the reliability of surrogate endpoints, the oncology community must move toward more standardized and rigorous validation frameworks. The authors advocate for several key initiatives:

  1. Increased Data Transparency: Access to raw, individual patient-level data from both pharmaceutical and government-sponsored trials is essential for robust meta-analyses.
  2. Adoption of ReSEEM Guidelines: Harmonizing the reporting of surrogacy studies through the ReSEEM (Reporting of Surrogate Endpoint Evaluation Metrics) guidelines will ensure future studies are comparable and rigorous.
  3. Exploration of Novel Metrics: As treatments evolve, so must our endpoints. Metrics such as metastasis-free survival, milestone survival, and advanced statistical techniques like landmark analysis or restricted mean survival time (RMST) may offer more accurate reflections of clinical benefit.

Ultimately, while surrogate endpoints are vital for the rapid delivery of new medicines, their use must be grounded in rigorous, indication-specific validation to ensure that "faster" does not come at the cost of "effective."

Summary updated

Abstract

Endpoint surrogacy is an important concept in oncology trials. Using a surrogate endpoint like progression-free survival as the primary endpoint-instead of overall survival-would lead to a potential faster drug approval and therefore more cancer patients with an earlier opportunity to receive the newly approved drugs.
BibTeX
@inproceedings{zhang2020endpoint,
  author = {Jianrong Zhang and Meagan Pilar and Xiaofei Wang and Jingxia Liu and Herbert Pang and Ross C. Brownson and Graham A. Colditz and Wenhua Liang and Jianxing He},
  title = {{Endpoint surrogacy in oncology Phase 3 randomised controlled trials}},
  booktitle = {British},
  year = {2020},
  doi = {10.1038/s41416-020-0896-5},
}

Loading related content...

Figure