Commentary on: A call for better reporting of trials using surrogate primary endpoints.

Abstract

Drs. Manyara, Ciani, and Taylor make the important point that randomized controlled trials (RCTs) using a primary surrogate endpoint should be more transparent in their reporting details of biomarkers in trials. They suggest a clear statement concerning use of a surrogate primary endpoint and providing information on validity and limitations of the surrogate. They announce a new project to develop SPIRIT and CONSORT extensions specific to surrogate endpoints SPIRIT-SURROGATE and CONSORT-SURROGATE. There are no fully qualified and validated surrogate biomarkers for Alzheimer's disease (AD) RCTs and none that could serve as a primary RCT endpoint. As described by the US Food and Drug Administration (FDA) a surrogate endpoint is a clinical trial endpoint used as a substitute for a direct measure of how a patient feels, functions, or survives.1 A surrogate endpoint does not measure the clinical benefit of primary interest in and of itself, but rather is expected to predict clinical benefit. Epidemiologic, therapeutic, pathophysiologic, or other scientific data provide the evidentiary basis for establishing a biomarker as a surrogate. A surrogate must change in response to multiple therapies across multiple RCTs and must explain the change in clinical outcome as well as correlate with it.2 Designation as a surrogate requires substantial statistical evidence based on trial meta-analyses.3 Robert Califf, US Commissioner of Food and Drugs, makes the point that “the single most common and serious error in the evaluation of biomarkers is the assumption that a correlation between the measured level of a biomarker and a clinical outcome means that the biomarker constitutes a valid surrogate.”2 The primary outcome of the RCTs of the aducanumab trials was the Clinical Dementia Rating–Sum of Boxes (CDR-SB; NCT02484547; NCT02477800) and the primary endpoint was the drug–placebo difference on this outcome at week 76. Accelerated approval of aducanumab was based on amyloid plaque lowering as shown by amyloid positron emission tomography (PET) considered reasonably likely to predict clinical benefit.4 Reasonably likely surrogate endpoints are supported by a strong mechanistic and/or epidemiologic rationale, but the amount of clinical data available is not sufficient to support them as a validated surrogate endpoint (defined above). Reasonably likely surrogate endpoints are used to support the FDA's Accelerated Approval program, which is intended to provide patients with serious diseases more rapid access to promising therapies.5 Accelerated approval is followed by post-marketing studies to determine whether the reasonably likely surrogate endpoint, in fact, predicts the clinical benefit. The aducanumab trials would not have been included in reporting requirements for trials using surrogate primary endpoints; their primary endpoints were drug–placebo differences on clinical measures; amyloid lowering was not a primary or a secondary endpoint and was not included in the sequential testing rank order implemented (NCT02484547; NCT024778006). I concur with Drs. Manyara, Ciani, and Taylor that biomarkers should be described in greater detail in the SPIRIT and CONSORT criteria. Using the FDA biomarker lexicon, these measures would be described as risk/susceptibility, diagnosis, monitoring, pharmacodynamic/response, predictive, prognostic, or safety biomarkers.7, 8 Candidate surrogate biomarkers or biomarkers that might be considered reasonably likely to predict clinical benefit are pharmacodynamic biomarkers. Biomarkers that have been qualified for a specific context of use in the RCT should be described in the SPIRIT and CONSORT reporting framework.9 The SPIRIT-SURROGATE and CONSORT-SURROGATE initiative is important and might be more broadly conceived as a SPIRIT-BIOMARKER and CONSORT-BIOMARKER initiatives. Another important adjustment to the SPIRIT and CONSORT criteria is to require the reporting of ethnicity and race among trial participants. These data are sometimes not collected and often omitted from clinical trial reports.10 They are critically important to assess our success in increasing the diversity of clinical trial populations. JC is supported by NIGMS grant P20GM109025; NINDS grant U01NS093334; NIA grant R01AG053798; NIA grant P20AG068053; NIA grant P30AG072959; NIA grant R35AG71476; Alzheimer’s Disease Drug Discovery Foundation (ADDF); Ted and Maria Quirk Endowment; and the Joy Chambers-Grundy Endowment. J.C. has provided consultation to Acadia, Alkahest, AlphaCognition, AriBio, Biogen, Cassava, Cortexyme, Diadem, EIP Pharma, Eisai, GemVax, Genentech, Green Valley, Grifols, Janssen, Karuna, Lilly, Lundbeck, LSP, Merck, NervGen, Novo Nordisk, Oligomerix, Ono, Otsuka, PRODEO, Prothena, ReMYND, Resverlogix, Roche, Signant Health, Suven, and United Neuroscience pharmaceutical, assessment, and investment companies. Author disclosures are available in the supporting information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.