A surrogate endpoint is an indirect endpoint used in place of a clinically directly relevant outcome. It is typically based on a biomarker or an intermediate event and is intended to predict the expected benefit for patients, for example overall survival, morbidity, or quality of life.
Why surrogate endpoints are used
Surrogate endpoints can accelerate studies because they can be measured earlier or occur more frequently than hard clinical endpoints. In oncology, for example, progression-free survival is often discussed as a surrogate for overall survival. In cardiology, laboratory or imaging parameters can serve as surrogates for clinical events.
Their use is particularly attractive in indications with long disease courses or rare events. This can reduce sample sizes and study duration, which is relevant for development programs and time to market. At the same time, however, the need for validation increases, as a surrogate does not automatically guarantee a patient-relevant benefit.
Validation, level of evidence, and practical rationale
A surrogate endpoint is only robust if it can be shown that changes in the surrogate reliably reflect changes in the clinical endpoint. Data from multiple studies or meta-analyses are often used for this purpose. Methodologically, a distinction is made between an individual association (patient level) and an association at the study level.
A key risk is that a therapy improves the surrogate without improving the clinical endpoint, or even causes harm in terms of safety. Examples show that individual biomarkers can be misleading if they are not causally linked to the clinical outcome.
For validation, criteria such as the Prentice criteria or the surrogate threshold effect are discussed. In practice, the evidence is usually gradual: some surrogates are established (e.g., viral load in certain indications), while others are only plausible and require additional confirmation through patient-relevant endpoints. When planning a study, it should be documented why the endpoint was selected, which data support the association, and which uncertainties remain.
Another practical point is harmonizing endpoint definitions across studies. If surrogate data are later to be included in a meta-analysis, time points, criteria, and analysis algorithms should be as consistent as possible; otherwise, comparability—and thus the strength of evidence for validation—decreases.
Methodological requirements in study design
If a surrogate endpoint is primary, measurement methodology, time points, evaluation criteria, and data quality must be defined with particular rigor. For imaging surrogates, standardized assessment procedures and, where applicable, blinded central reviews are important. For laboratory parameters, sample handling, analytics, and quality control are critical.
Operationally, it must also be clarified how surrogate measurements are reflected in monitoring and data management: which sources are considered source data, how deviations are documented, and how consistent coding is ensured. For biomarkers, stability, transport logistics, and laboratory qualification are also frequently critical points.
Cut-offs and units must also be controlled, especially when multiple laboratories or devices are used. For some biomarkers, reference standards or proficiency testing are relevant so that differences are not mistakenly interpreted as a treatment effect.
Relevance for clinical trials
From a CRO perspective, surrogate endpoints affect the entire operational chain: data collection, monitoring, data management, and statistical analysis. If a surrogate endpoint is intended to serve as the basis for marketing authorization, traceable definitions and robust data are essential because authorities and ethics committees critically assess clinical relevance.
From a regulatory perspective, surrogate endpoints are primarily accepted when they are well justified and recognized for the indication. In conditional approvals or accelerated procedures, a surrogate endpoint can enable earlier decisions, but it is often linked to obligations for post-authorisation studies. Mediconomics supports sponsors in defining endpoints, documenting the validation rationale, and setting up analyses so that benefit–risk assessments remain transparent.
For communication, it is important to clearly state the endpoint’s status: “surrogate” does not automatically mean “accepted.” In interactions with authorities, it should therefore be presented transparently how the endpoint has been used historically, what evidence for validation exists, and which additional data (e.g., overall survival, patient-reported outcomes) are being collected in parallel.
In many programs, a two-stage evidence concept is therefore chosen: the surrogate endpoint serves as the primary efficacy endpoint for an early decision, while patient-relevant endpoints are followed as key secondary endpoints or in a long-term extension phase. This allows early signals to be triangulated with later clinical data. For data interpretation, it is helpful to define in the SAP how discrepancies between the surrogate and the clinical outcome will be assessed and which additional analyses (e.g., landmark analyses or alternative progression criteria) are planned.
Frequently Asked Questions (FAQ)
Is a biomarker always a surrogate endpoint?
No. A biomarker can be collected exploratorily without serving as a surrogate for clinical benefit. Only when it is interpreted as a substitute endpoint and validated accordingly is it considered a surrogate endpoint.
Why do authorities sometimes accept surrogate endpoints?
When medical need is high and the link to clinical benefit is plausible and supported by data, surrogate endpoints can enable earlier decisions. Additional studies after approval are often required.
What risks are associated with surrogate endpoints?
The greatest risk is an incorrect conclusion: improvement in the surrogate without improvement in patient-relevant benefit. Therefore, validation, safety data, and a consistent evidence framework are crucial.
A common misconception is that a surrogate endpoint is automatically “more objective.” Biomarker measurements can also contain systematic errors, for example due to different sampling times, device changes, outlier rules, or non-harmonized laboratory reference ranges. Therefore, an endpoint strategy should always include a concept for quality control, site training, and traceable documentation in the Trial Master File.
Regulatory References
- ICH E9 (R1): Requirements for endpoint definition, statistical planning, and interpretation.
- ICH E6 (R3): Data quality, monitoring, and traceable documentation of endpoint collection.
- EMA reflection paper on surrogate endpoints: criteria for acceptance and evidence assessment of surrogate endpoints.