{
  "id": "surrogate-endpoint",
  "title": "Surrogate Endpoint",
  "lang": "en",
  "source_lang": "en",
  "summary": "A surrogate endpoint substitutes for a direct measure of clinical benefit in a trial, with its usefulness depending on evidence that treatment effects predict patient outcomes.",
  "category": "health",
  "subcategory": "medicine",
  "type": "concept",
  "aliases": [
    "Surrogate end point",
    "Surrogate outcome"
  ],
  "generated_at": "2026-11-04",
  "model": "pro/gpt-6.1-sol",
  "url": "https://aiwiki.page/en/surrogate-endpoint/",
  "translations": {
    "zh-Hans": {
      "url": "https://aiwiki.page/zh/surrogate-endpoint/",
      "json": "https://aiwiki.page/zh/surrogate-endpoint.json"
    }
  },
  "sections": [
    {
      "heading": "",
      "text": "A surrogate endpoint is a measurement used in a clinical trial in place of a direct measure of how a patient feels, functions, or survives. It is intended to predict the effect of treatment on a clinically meaningful outcome, often allowing researchers to assess treatment effects sooner or with fewer participants. A surrogate may be a laboratory measurement, an imaging finding, or another disease-related measure. Improvement in a surrogate does not, by itself, establish that a treatment benefits patients: that inference depends on evidence supporting the surrogate in the particular setting."
    },
    {
      "heading": "Relationship to biomarkers and clinical outcomes",
      "text": "A biomarker is a measured characteristic indicating a biological process, a disease process, or a response to an intervention. Biomarkers have many roles, including diagnosis, prognosis, treatment selection, and monitoring. Only some are suitable as surrogate endpoints.\n\nThe distinction concerns what the measurement is being used to establish:\n\n- A prognostic biomarker identifies people more likely to experience an outcome.\n- A response biomarker changes after treatment and indicates a biological response.\n- A surrogate endpoint is used to infer a treatment effect on a clinically meaningful outcome.\n\nA marker may serve the first two roles without being suitable for the third. For example, a measurement associated with a high risk of death does not necessarily become a valid surrogate merely because treatment lowers it.\n\nA clinical endpoint directly captures a disease event or patient outcome, such as a symptomatic stroke, hospitalization, or death. Measures of symptoms, functioning, and health-related quality of life can also assess clinical benefit directly. A patient-reported outcome is therefore not a surrogate simply because it is subjective or collected through a questionnaire.\n\n“Primary endpoint” and “surrogate endpoint” describe different properties. Primary identifies an endpoint’s designated role in the trial; surrogate identifies its relationship to the clinical benefit being inferred. A surrogate can be a primary or secondary endpoint."
    },
    {
      "heading": "Why surrogate endpoints are used",
      "text": "Some important outcomes develop slowly or occur infrequently. Establishing whether an intervention prevents death, organ failure, or disability may require large studies and years of follow-up. An earlier measurement can sometimes make drug development more feasible.\n\nPotential advantages include:\n\n- Shorter trials and earlier assessment of treatment activity.\n- Smaller sample sizes when the surrogate is frequent or measured continuously.\n- Earlier identification of treatments unlikely to be effective.\n- More feasible studies in rare diseases or conditions with long intervals before major complications.\n\nThese advantages depend on the measurement’s reliability and predictive value. A rapidly measured endpoint is not necessarily a trustworthy substitute for a patient outcome."
    },
    {
      "heading": "Validation: association is not enough",
      "text": "Surrogate validation concerns whether treatment effects on the surrogate predict treatment effects on the clinical outcome. This is stronger than showing that patients with favorable surrogate values tend to have better outcomes.\n\nEvidence commonly includes biological understanding, observational findings, and results from randomized controlled trials. Two relationships are particularly important:\n\nIndividual-level association. Within studies, patients with better surrogate measurements tend to have better clinical outcomes.\n\nTrial-level association. Across trials, treatments producing larger effects on the surrogate also tend to produce larger effects on the clinical outcome.\n\nA strong individual-level association alone is insufficient. It may reflect disease severity or other patient characteristics rather than a mechanism through which treatment produces benefit. Trial-level evidence is especially important when the purpose is to predict the benefit of a new intervention.\n\nStatistical approaches include analyses combining multiple trials, models of treatment-effect relationships, and causal methods examining pathways between treatment, surrogate, and outcome. These approaches have different assumptions; no single correlation coefficient establishes universal validity.\n\nValidation is context-specific. Relevant features include the disease, disease stage, patient population, treatment mechanism, measurement method, follow-up period, and clinical outcome being predicted. Evidence supporting a surrogate for one drug class or complication may not support its use for another."
    },
    {
      "heading": "Examples and their boundaries",
      "text": "Measurement | Use as a surrogate | Important qualification\nBlood pressure | Used in evaluating treatments for [[hypertension | hypertension]] and their effects on cardiovascular risk\n[[glycated-hemoglobin | Glycated hemoglobin]] (HbA1c) | Used in glucose-lowering trials in [[diabetes-mellitus\nHIV viral load | Used to assess virologic efficacy of [[antiretroviral-therapy | antiretroviral therapy]] for [[hiv-aids\nTumor response | Tumor shrinkage or disappearance may support evaluation of treatments for [[cancer | cancer]]\n\nThe same measurement may have different evidentiary strength in different settings. In oncology, progression-free survival measures time until defined disease progression or death. When progression is determined through imaging without corresponding symptoms, its relationship to direct patient benefit requires careful interpretation. It is not interchangeable with overall survival, and its status as a surrogate depends on the disease and treatment context."
    },
    {
      "heading": "How a surrogate can mislead",
      "text": "A treatment can improve a surrogate while providing no clinical benefit, or even causing harm. Several mechanisms can produce this discrepancy:\n\n- Incomplete representation of disease: the surrogate reflects only one pathway among several determining the outcome.\n- Effects outside that pathway: treatment improves the measured process but causes toxicity elsewhere.\n- Marker–mechanism mismatch: the measurement accompanies disease without being a cause of the outcome.\n- Limited durability: an early improvement does not persist long enough to affect later events.\n- Measurement limitations: assessment timing, missing data, or inconsistent interpretation distort the apparent treatment effect.\n\nThese issues illustrate the difference between association and causation. Even a marker involved in a causal pathway may fail to capture the treatment’s net effect.\n\nA historical example is the Cardiac Arrhythmia Suppression Trial. Encainide and flecainide suppressed ventricular premature beats after myocardial infarction, but increased mortality compared with placebo. Suppression of the rhythm abnormality did not establish improved survival."
    },
    {
      "heading": "Regulatory use and interpretation",
      "text": "Regulators distinguish between surrogates supported by substantial clinical evidence and endpoints considered reasonably likely to predict benefit but not yet fully validated. Terminology and requirements differ by jurisdiction.\n\nIn the United States, accelerated approval can be based on a surrogate endpoint reasonably likely to predict clinical benefit, or on certain intermediate clinical endpoints. Such approval requires further evidence to verify benefit. An intermediate clinical endpoint measures a clinical effect earlier than the ultimate outcome of interest; it is not necessarily a biomarker.\n\nRegulatory acceptance for a particular use does not establish that an endpoint is valid for every disease or treatment. If subsequent studies fail to verify benefit, the indication may be withdrawn or its use restricted.\n\nInterpreting a surrogate-based trial therefore involves separate questions: whether treatment changed the measured endpoint, whether that change reliably predicts clinical benefit, and whether the treatment’s overall benefits outweigh its harms. A statistically convincing result on a surrogate answers the first question, but does not automatically answer the others."
    }
  ],
  "links": [
    "clinical-trial",
    "biomarker",
    "clinical-endpoint",
    "stroke",
    "health-related-quality-of-life",
    "patient-reported-outcome",
    "drug-development",
    "randomized-controlled-trial",
    "hypertension",
    "glycated-hemoglobin",
    "diabetes-mellitus",
    "hiv-aids",
    "cancer",
    "causation",
    "placebo"
  ],
  "sources": [],
  "license": "https://creativecommons.org/licenses/by/4.0/"
}