Intended for healthcare professionals

Research Methods & Reporting

Enhancing quality and impact of early phase dose-finding clinical trial protocols: SPIRIT Dose-finding Extension (SPIRIT-DEFINE) guidance

BMJ 2023; 383 doi: (Published 20 October 2023) Cite this as: BMJ 2023;383:e076386

Linked Research Methods & Reporting

Enhancing reporting quality and impact of early phase dose-finding clinical trials

  1. Christina Yap, professor of clinical trials biostatistics1,
  2. Jan Rekowski, principal statistician1,
  3. Moreno Ursino, research fellow and associate professor2 3 4 5,
  4. Olga Solovyeva, trial methodologist1,
  5. Dhrusti Patel, research assistant1,
  6. Munyaradzi Dimairo, senior research fellow6,
  7. Christopher J Weir, professor of medical statistics and clinical trials7,
  8. An-Wen Chan, Phelan senior scientist8,
  9. Thomas Jaki, professor of statistics and chair for computational statistics9 10,
  10. Adrian Mander, professor of medical statistics11,
  11. Thomas R Jeffry Evans, clinical director12,
  12. Richard Peck, honorary professor and global head of clinical pharmacology13 14,
  13. Kathryn S Hayward, associate professor and senior research fellow15 16,
  14. Melanie Calvert, director and centre director and academic lead17 18 19 20 21,
  15. Khadija Rerhou Rantell, senior statistical adviser22,
  16. Shing Lee, associate professor of biostatistics23,
  17. Andrew Kightley, patient and public involvement and engagement lead24,
  18. Sally Hopewell, professor of clinical trials and evidence synthesis25,
  19. Deborah Ashby, professor of medical statistics and clinical trials26,
  20. Elizabeth Garrett-Mayer, vice president27,
  21. John Isaacs, professor of clinical rheumatology and honorary consultant rheumatologist28 29,
  22. Robert Golub, professor of medicine (general internal medicine) and preventive medicine30,
  23. Olga Kholmanskikh, clinical assessor and oncology working party member31,
  24. Dawn P Richards, director of patient and public engagement32,
  25. Oliver Boix, expert statistician oncology early clinical development33,
  26. James Matcham, vice president34,
  27. Lesley Seymour, director35,
  28. S Percy Ivy, associate branch chief36,
  29. Lynley V Marshall, honorary appointment and consultant1 37,
  30. Antoine Hommais, project manager38,
  31. Rong Liu, senior director of biostatistics39,
  32. Yoshiya Tanaka, professor and chairman40,
  33. Jordan Berlin, associate director for clinical research41,
  34. Aude Espinasse, clinical trials programme manager1,
  35. Johann de Bono, regius professor of cancer research and professor in experimental cancer medicine1 37
  1. 1Institute of Cancer Research, London SM2 5NG, UK
  2. 2ReCAP/F CRIN, INSERM, Paris, France
  3. 3Unit of Clinical Epidemiology, University Hospital Centre Robert Debré, Reims, France
  4. 4INSERM Centre de Recherche des Cordeliers, Sorbonne University, Paris, France
  5. 5Health data and model driven approaches for Knowledge Acquisition team, Centre Inria, Paris, France
  6. 6Division of Population Health, Sheffield Centre for Health and Related Research, University of Sheffield, Sheffield, UK
  7. 7Edinburgh Clinical Trials Unit, Usher Institute, University of Edinburgh, Edinburgh, UK
  8. 8Department of Medicine, Women’s College Research Institute, University of Toronto, Toronto, Canada
  9. 9MRC Biostatistics Unit, Cambridge University, Cambridge, UK
  10. 10Computational Statistics Group, University of Regensburg, Regensburg, Germany
  11. 11Centre For Trials Research, Cardiff University, Cardiff, UK
  12. 12Institute of Cancer Sciences, CR-UK Beatson Institute, University of Glasgow, Glasgow, UK
  13. 13Department of Pharmacology and Therapeutics, University of Liverpool, Liverpool, UK
  14. 14Hoffmann-La Roche, Basel, Switzerland
  15. 15Departments of Physiotherapy, and Medicine (Royal Melbourne Hospital), University of Melbourne, Parkville, VIC, Australia
  16. 16Florey Institute of Neuroscience and Mental Health, University of Melbourne, Parkville, VIC, Australia
  17. 17Centre for Patient Reported Outcomes Research, Institute of Applied Health Research, University of Birmingham, Birmingham, UK
  18. 18Birmingham Health Partners Centre for Regulatory Science and Innovation, University of Birmingham, Birmingham, UK
  19. 19National Institute for Health and Care Research Applied Research Collaboration West Midlands, University of Birmingham, Birmingham, UK
  20. 20National Institute for Health and Care Research Blood and Transplant Research Unit in Precision Transplant and Cellular Therapeutics, University of Birmingham, Birmingham, UK
  21. 21National Institute for Health and Care Research Birmingham Biomedical Research Centre, NIHR Birmingham Biomedical Research Centre, Institute of Translational Medicine, University Hospital NHS Foundation Trust, Birmingham, UK
  22. 22Medicines and Healthcare products Regulatory Agency, London, UK
  23. 23Columbia University Mailman School of Public Health, New York, NY, USA
  24. 24Lichfield, UK
  25. 25Oxford Clinical Research Unit, NDORMS, University of Oxford, Oxford, UK
  26. 26School of Public Health, Imperial College London, St Mary’s Hospital, London, UK
  27. 27Center for Research and Analytics, American Society of Clinical Oncology, Alexandria, VA, USA
  28. 28Translational and Clinical Research Institute, Newcastle University, Newcastle upon Tyne, UK
  29. 29Musculoskeletal Unit, Newcastle upon Tyne Hospitals NHS Foundation Trust, Freeman Hospital, Newcastle upon Tyne, UK
  30. 30Department of Medicine, Northwestern University Feinberg School of Medicine, Evanston, IL, USA
  31. 31Federal Agency for Medicines and Health Products, Brussels, Belgium
  32. 32Clinical Trials Ontario, MaRS Centre, Toronto, ON, Canada
  33. 33Bayer, Berlin, Germany
  34. 34Strategic Consulting, Cytel (Australia), Perth, WA, Australia
  35. 35Investigational New Drug Programme, Canadian Cancer Trials Group, Cancer Research Institute, Queen’s University, Kingston, ON, Canada
  36. 36Investigational Drug Branch, Cancer Therapy Evaluation Program, Division of Cancer Treatment and Diagnosis, National Institute of Health, Bethesda, MD, USA
  37. 37Royal Marsden NHS Foundation Trust, London, UK
  38. 38Department of Clinical Research, National Cancer Institute, Boulogne-Billancourt, France
  39. 39Bristol Myers Squibb, New York, NY, USA
  40. 40First Department of Internal Medicine, University of Occupational and Environmental Health, Kitakyushu, Japan
  41. 41Vanderbilt-Ingram Cancer Center, Nashville, TN, USA
  1. Correspondence to: C Yap christina.yap{at}
  • Accepted 5 August 2023

SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) 2013 provides guidance for clinical trial protocol writing. However, neither the original guidance nor its extensions adequately cover the features of early phase dose-finding trials. The SPIRIT Dose-finding Extension (DEFINE) statement is a new guideline that provides recommendations for essential items that should be provided in the protocols of these trials. It details extensions to the SPIRIT 2013 guidance, incorporating 17 new items and modifying 15 existing items. The purpose of this guideline is to promote transparency, completeness, reproducibility of methods, and interpretation of early phase dose-finding trial protocols. It is envisioned that the resulting improvements in the design and conduct of early phase clinical trials will ultimately reduce research inefficiencies and inconsistencies, driving transformational advances in clinical care.

Developing an intervention is a lengthy process pursued in stages where decisions are based on balance of benefits and risks or harms of the intervention under investigation. Lack of efficacy or evidence of harm due to adverse safety profiles are common reasons for phase 2 and phase 3 trials to be unsuccessful.12 Phase 3 trial failures can reflect incorrect decisions made at earlier stages, including in early phase dose-finding (EPDF) trials (commonly known as phase 1, phase 1/2, or first-in-human trials). Reasons why interventions do not progress or succeed in later stages of clinical development include misleading preclinical studies, inadequate participant selection, inefficient trial design, suboptimal biomarker or outcome choices, and poor dose selection. The same reasons can also contribute to early discontinuation of promising interventions.

EPDF trials typically evaluate new interventions that can be used in different doses and can be pharmacological (chemical or biological—eg, drugs, vaccines, cell therapies, gene therapies), non-pharmacological (eg, radiotherapy, devices, rehabilitation, digital therapies), or a combination of both. They usually include a small number of healthy volunteers or participants with the disease under investigation. Either based on safety outcomes alone or increasingly jointly with outcomes of activity, EPDF trials aim to recommend a tolerated dose range for further study. In this article, a broad definition of dose is used because terms such as “dose finding,” “dose level,” “dose escalation,” and “dose expansion” are widely understood. Here, dose might refer not only to the amount of dose but can also comprise frequency, intensity, or duration of an intervention, for example.3 The term could therefore be regarded as synonymous to dosage, or dosing regimen, or unit dose, and it can apply to interventions given alone or in combination (see the glossary in box 1 for details).

Box 1



A measure of the physiological response that an intervention produces.

Algorithm based (rule based) design

A trial design that uses a simple set of predefined algorithms or rules to guide the decision making process for dose escalation or de-escalation. Examples include traditional 3+3, accelerated titration, and pharmacologically guided dose escalation designs.45

Biomarker substudy

A part of a clinical trial that investigates biomarkers, which are “a defined characteristic that is measured as an indicator of normal biological processes, pathogenic processes, or biological responses to an exposure or intervention, including therapeutic interventions. Biomarkers could include molecular, histological, radiographic, or physiological characteristics. A biomarker is not a measure of how an individual feels, functions, or survives.”6

Clinical benefit

A favourable effect on a meaningful aspect of how a participant feels, functions, or survives as a result of an intervention.7

Delphi survey

A series of questionnaires used sequentially to gather diverse opinions that allow experts to develop ideas about potential future developments around an issue. The questionnaires are developed throughout the process in relation to the responses given by participants.


In this article, dose is defined broadly and can be considered synonymous with dosage or dosing regimen (dose or schedule), or a unit dose. The unit dose is the amount or intensity of an intervention (eg, drug quantity, radiotherapy, exercise level), or the extent to which a participant might be exposed to an intervention on a single occasion. Information on dosage should include aspects of the intervention that describe how many times it was delivered and for how long—such as the number of sessions; their schedule; and their duration, intensity, or dose.3

Dose escalation or de-escalation

An incremental increase or decrease (or up-titration or down-titration) in the strength of any intervention (eg, a drug or exercise intensity level) to improve its tolerability or maximise its pharmacological or clinical effect.

Dose limiting criteria

Effects or markers that are presumably related to the intervention and that either are considered unacceptable or show the desired level of effect has been achieved and a further increase in dose is not required.8

Dose limiting toxicity

Side effects of an intervention that are serious enough to prevent an increase in the dose of that intervention.5

Dosing regimen or dosage

See dose.

Early phase dose-finding trial

An early phase trial where different doses of the investigated intervention are given to groups of participants, with interim assessments of the safety/tolerability (and other markers such as activity) of the intervention.

Estimand framework

Estimands provide a structural framework to define the target of estimation for a particular clinical trial objective.910 They require to specify the treatment condition of interest, the population targeted by the clinical question, the variable of interest or endpoint used to answer that question, the handling strategies for intercurrent events (ie, events occurring after treatment initiation that affect either the interpretation or the existence of the measurements associated with the clinical question), and a population level summary of the variable or endpoint.

Expansion cohort or dose expansion

A part of a dose escalation clinical trial that aims to accrue additional participants after an initial dose escalation part with different or targeted eligibility criteria to collect additional information on safety or activity.11


Can refer to an intervention group or arm, or specifically defined subgroups of the targeted participant population based on, for example, participant or disease characteristics.


The totality of possible adverse consequences of an intervention or treatment; they are the direct opposite of benefits, against which they must be compared.12 Harms can comprise of adverse events, adverse (drug) reactions, toxicities, treatment emergent adverse events, or those that are intolerable by participants.1213 They can also include tolerability assessment using patient reported outcomes as complementary to investigators’ reporting.1415

Interim analysis or review

A statistical analysis or review of accumulating data from an ongoing trial (interim data) to inform trial adaptations (before the final analysis), which might or might not involve treatment group comparisons.16

Model assisted design

A trial design that combines a clearly predetermined algorithm to guide the dose escalation or de-escalation as in rule based designs, and an underlying statistical model, as in model based designs.17 Examples include the modified toxicity probability interval design18 and the bayesian optimal interval design.19

Model based design

A trial design that assumes a relation between the dose of the intervention given to the participant and the likelihood of the participant experiencing an effect (such as toxicity or activity) and uses a parametric model to estimate that association. Examples include the continual reassessment method,20 escalation with overdose control,21 and the efficacy-toxicity trade-off based design.22

Multiple ascending dose

A trial design where a small number of participants (healthy volunteers or participants) receive several doses of an intervention over time to assess safety or tolerability and pharmacokinetic and pharmacodynamic profiles. Doses can remain the same or increase within a participant. The dose level is subsequently escalated for further participants according to the protocol, assuming that strict safety, effect, or pharmacokinetic criteria are met.

Operating characteristics

Characteristics that relate to the statistical behaviour or performance of the trial design in answering research questions. These might include the probability of correctly selecting the correct dose, statistical power, false positive error rate, bias in estimation of treatment effect, or probability of each adaptation taking place.1623


Described as what a drug does to the body; pharmacodynamics refer to how the drug works and how it affects the body.


Described as what the body does to a drug; pharmacokinetics refer to the movement of the drug into, through, and out of the body. It includes the analysis of chemical metabolism and the measurement or modelling of a substance from the moment that it is used up to the point when it is completely eliminated from the body.

Prespecified decision making criteria

Planned or prespecified rules to guide decisions, describing whether, how, and when the proposed trial adaptations will be used during the trial. The criteria involve prespecifying a set of actions guiding how decisions about implementing the trial adaptations are made given interim observed data (decision rules). They also involve prespecifying limits or parameters to trigger trial adaptations (decision boundaries), for example, stopping boundaries that relate to prespecified limits regarding decisions to stop the trial or any treatment arms early.

Single ascending dose

A trial design in which a small number of participants receive one dose of a therapeutic intervention at a given dose level to assess safety or tolerability and characterise the pharmacodynamics and pharmacokinetics of the intervention. Single ascending dose trials are often conducted in a small number of healthy volunteers, although some trials recruit participants with a disease of interest. The dose is subsequently escalated for further participants according to the protocol, assuming that strict safety, effect, or pharmacokinetic criteria are met.

Transition points

The points or parts in a clinical trial when the decision can be made to proceed to the next stage or phase, such as from dose escalation to dose expansion, from phase 1 to phase 2, or from a single ascending dose to multiple ascending dose.

Trial (design) adaptations

Prespecified changes or modifications (defined in advance) that can be made to various aspects of a trial while it is ongoing without undermining the trial’s validity and integrity.24 These prespecified modifications are driven by accruing interim data.25 Examples include adjusting the doses; changing the predetermined sample size; stopping the trial early for efficacy, futility, or safety; and switching the allocated treatment of participants owing to a lack of benefit or safety issues.16


To ensure the safety of trial participants in EPDF trials, decisions regarding dose escalation or de-escalation are made based on interim data. Different dose escalation approaches have been described in the literature, for example, algorithm based (also called rule based), model assisted, and model based designs.2627 The use of model assisted and model based designs, which have been reported to be more efficient but also more complex than algorithm based designs,428 rose from 1.6% (20/1235) of phase 1 cancer trials published in 1991-200629 to 8.6% (68/788) in 2014-19.4 Most recent data confirm this trend with the rate of advanced designs in cancer trials reported to be 19% (11/58) based on protocols posted on in 2017-23.30 The complexity of these designs is reflected in a more multifaceted implementation and the requirement to specify more details on design features,313233 which mandates more detailed protocols for EPDF trials to improve precision and transparency, and to facilitate understanding of trial design and decision making processes.

A trial protocol is a crucial document that outlines how a clinical trial will be conducted, ensuring the safety of patients and the integrity of data. It provides details on objectives, design, methodology, statistical analyses, and trial implementation. The protocol serves as the shared central reference for a trial team and is evaluated by external reviewers. Despite the importance of trial protocols, their content and quality vary considerably.34 To resolve this problem, the SPIRIT 2013 (Standard Protocol Items: Recommendations for Interventional Trials) statement3536 was established to provide evidence based guidance for the essential content of a trial protocol. Protocols underpinning EPDF trials require more transparency to facilitate a better understanding of the trial design and how dose decisions would be made.37 Inadequate or unclear information on design, conduct, and analysis in EPDF protocols hinders interpretability and reproducibility. It might also lead to unnecessary amendments and associated costs, as well as inadequate or biased reporting resulting in erroneous conclusions on safety and efficacy. The overall quality of EPDF protocols from in 2017-23 was reported to be substantially variable and poor, with insufficient reporting in many applicable SPIRIT 2013 items.30 For example, sections on ethics and dissemination strategy were frequently found to be dealt with insufficiently. Although SPIRIT 2013 largely applies to many types of trial designs, trials that use specialised designs might require additional protocol considerations. Several SPIRIT extensions have been proposed to improve its usefulness for specialised topics.383940414243 Neither the SPIRIT 2013 statement nor any of its extensions, however, sufficiently cover the needs of EPDF trials—although, globally, more phase 1 trials (n=18 716) than phase 3 trials (n=10 451) were registered on and first posted between 2018 and 2022. The number of phase 1 trials might even be an underestimate, because researchers are not required to register them on Because no consensus driven protocol guidance exists for EPDF trials,45 extension of the SPIRIT 2013 guidance to EPDF trials is urgently needed.

Summary points

  • Early phase dose-finding clinical trials are essential for clinical development as they provide the groundwork for further development and guide subsequent trials

  • SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) 2013 focused on randomised trials, and the new SPIRIT Dose-finding Extension (DEFINE) guideline has been extended to broaden its applicability to early phase dose-finding trials with interim strategies for dose escalation or de-escalation

  • After an international consensus guideline development process using the EQUATOR (Enhancing QUAlity and Transparency Of health Research) methodological framework, 32 items specific to early phase dose-finding were recommended for inclusion in clinical trial protocols

  • Inclusion of these SPIRIT-DEFINE items in clinical trial protocols could enhance transparency, completeness, reproducibility of methods, and trial usefulness in early phase dose-finding trials


The SPIRIT Dose-finding Extension (DEFINE) was conceptualised, designed, and conducted between January 2022 and July 2023 in concordance with the EQUATOR (Enhancing QUAlity and Transparency Of health Research) network’s methodological framework for guideline development.46 The study was led by the principal investigator (CY) and the DEFINE executive committee, who met online once or twice every three months before the international consensus meeting and once after. The DEFINE research team at the Institute of Cancer Research met weekly. Frequent email correspondences and one-to-one or small group meetings between the principal investigator and key members of the executive committee were arranged for any discussions whenever needed. SPIRIT-DEFINE was approved for sponsorship by the Institute of Cancer Research’s Committee for Clinical Research (reference No CCR5460). The UK Health Research Authority confirmed that no approval for research ethics was necessary. All participants gave their informed consent to participate in the Delphi survey and consensus meeting.

Generation of candidate protocol items

An initial SPIRIT-DEFINE checklist was drafted based on SPIRIT 2013,35 with additional protocol related candidate items taken from the companion guidance for trial reports of EPDF trials, CONSORT-DEFINE (CONsolidated Standards Of Reporting Trials Dose-finding Extension).3747 We used the multidisciplinary executive committee’s expert opinions and unpublished or grey literature including regulatory and industry advice documents to further refine the checklist as described.4547 Major international stakeholder groups were consulted, and their protocol or guidance templates included (when available), to inform the generation and wording of the candidate items and the structuring of the eventual checklist. These groups included phase 1 units accredited by the Medicines and Healthcare products Regulatory Agency, funders, pharmaceutical companies, contract research organisations, and research ethics committees.4547

International Delphi process

We solicited feedback on the draft candidate items for the SPIRIT-DEFINE checklist from a broad stakeholder group using a Delphi survey (fig 1). A comprehensive outline of the recruitment procedure for the Delphi survey is provided in the section titled “The Delphi process” within the DEFINE development process paper.47 The Delphi process adhered to established methodological guidance.484950 A total of 206 participants from 24 countries voted in round one (March to May 2022), and 151 participants voted in round two (May to June 2022). Before voting for round two, participants were presented with the distribution of round one ratings for each item as well as their own prior ratings.

Fig 1
Fig 1

Development process of items in the SPIRIT-DEFINE (Standard Protocol Items: Recommendations for Interventional Trials Dose-finding Extension) checklist

According to a predetermined rule, items voted as not important (scores 1-3) by at least 80% of respondents in round one were eliminated between rounds subject to confirmation by the executive committee. Items voted as critically important (scores 7-9) by at least 70% of respondents in round one were considered to have reached consensus and were automatically included in the SPIRIT-DEFINE checklist45 (fig S1 in web appendix 1).

In these two rounds of the Delphi poll, 36 SPIRIT-DEFINE candidate items were reviewed, 26 items satisfied the criterion to be included in the checklist, and 10 items qualified to be discussed at the consensus meeting. The process, decision criteria, and voting results of the SPIRIT-DEFINE candidate items are described in figure S1 and table S1 in web appendix 1. Additional information on the Delphi method, including qualitative and quantitative analyses and the outcomes of rounds one and two, is provided elsewhere.47

International consensus meeting

A total of 32 international delegates from academic, commercial, and regulatory sectors and two patient and public involvement and engagement partners attended the online consensus meeting on 11-12 October 2022 (tables S2 and S3 in web appendix 1 list the affiliations or roles of participants). The Delphi survey findings were presented alongside supporting evidence, written comments from participants, and examples from protocols for each candidate item to be reviewed at the consensus meeting. After the presentation, members were invited to discuss each item, before voting anonymously. Voting options for the candidate items were to include or discard the item in the checklist, with the threshold for inclusion being ≥70% and exclusion being <50%, with the rest left for further deliberation by the DEFINE executive committee (fig S1 in web appendix 1).

Of 10 candidate items, four were recommended for inclusion in the SPIRIT-DEFINE checklist and five were rejected. One item was left for further deliberation at the subsequent executive committee meeting, at which it was rejected (fig 1; table S1 in web appendix 1).

Final consultation and piloting of the checklist

After the consensus meeting, the DEFINE executive committee and consensus participants refined the language of the items and their related explanations. During the pilot testing phase of the checklist (December 2022 to January 2023), eight multidisciplinary trialists evaluated the SPIRIT-DEFINE checklist by applying it to actual trial protocols of planned or existing trials and noting areas for improvement. The feedback gathered further shaped the final version of the guideline, with the DEFINE executive committee and consensus meeting participants agreeing on the final wording.


Figure 1 presents the development journey of the SPIRIT-DEFINE checklist items from the Delphi survey to the consensus meeting, to refinement of the checklist after the final consultation and pilot testing. The final SPIRIT-DEFINE guidance recommends that, in conjunction with the existing SPIRIT 2013 items, 32 EPDF specific items (17 new and 15 modified) should be included prospectively in EPDF trial protocols. Table 1 presents the items of the SPIRIT 2013 checklist as well as new and modified items for the SPIRIT-DEFINE extension. The downloadable version of the SPIRIT-DEFINE checklist is available in web appendix 2.

Table 1

Recommended checklist items to consider in EPDF clinical trial protocols from SPIRIT 2013 and SPIRIT-DEFINE checklists

View this table:

It is useful to note that terminology and definitions associated with EPDF trials can vary, for instance, for different interventions and disease areas. Key terms used throughout this article are provided in the glossary (box 1).

To enable readers to comprehend the strategies for dose escalation or de-escalation and trial design adaptations and to ensure that the procedures and findings can be reproduced, aspects of the SPIRIT-DEFINE checklist specific to EPDF trials include a detailed elaboration of the trial design (eg, adaptive features, timing of interim analyses, planned dose range with starting dose, dose allocation method, interim decision making criteria, any expansion cohorts, operating characteristics, and dose transition pathways). Specification of planned opportunities for adaptations and their scope is essential to preserve the integrity of adaptive designs and for regulatory assessments.16 All these aspects influence the statistical methods for design and analysis; hence this extension recommends providing comprehensive information on statistical methods covering these adaptive features and requiring clear definitions of analysis populations and plans for handling intercurrent events that occur after treatment initiation.51 Both analysis populations and intercurrent events relate to the estimands framework, which provides guidance on defining the treatment effect under investigation in a clinical trial (for details, see the International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) E9 (R1) addendum on estimands910).

In more detail, the new and modified items specific to EPDF trials are listed in box 2.

Box 2

Overview of new and modified items in the SPIRIT-DEFINE checklist

Administrative information (one modified item)

  • Identifying the early phase dose-finding design in the title of the protocol.

Introduction (12 new items, three modified items)

  • Incorporating non-clinical or preclinical research informing an EPDF trial52 and any planned biomarker substudies in the background and rationale section.53

  • Highlighting key objectives for early phase dose-finding trials in the objectives section.

  • Expanding the trial design section to include adaptive features,1654 starting doses, and range of dose levels with rationale, skipping of doses, planned cohort size, dose allocation method, and any expansion cohorts.3752555657

Methods: participants, interventions, and outcomes (five modified items)

  • Providing enhanced intervention details3 including reporting them for each dose level and describing prespecified criteria for dose discontinuations, modifications, or delays.52

  • Extending the description of outcomes to any outcomes that will be used to inform planned adaptations.16

  • Including clinical and statistical assumptions supporting the planned sample size and operating characteristics, which relate to the statistical behaviour or performance of the trial design2351 (see box 1 for details).

Methods: assignment of interventions (for controlled trials) (one new item)

  • Detailing any rule or algorithm to update the allocation strategy.16

Methods: data collection, management, and analysis (two new items, three modified items)

  • Providing increased details regarding statistical methods to cover adaptive features, analysis populations, as well as handling of missing data and intercurrent events that occur after treatment initiation.1651

Methods: data monitoring (three modified items)

  • Providing increased details regarding the interim decision making process16 and reporting of harms (eg, toxicities, adverse events).

Dissemination policy (one new item)

  • Including plans for sharing results while the trial is still ongoing.

Appendices (one new item)

  • Adding a new section to cover dose transition pathways or dose decision paths.183358

  • SPIRIT=Standard Protocol Items: Recommendations for Interventional Trials; DEFINE=Dose-finding Extension.


Authors should state where information on recommended items can be accessed if not in the protocol (eg, in a data management plan, statistical analysis plan, or other trial specific documents). Authors should provide explanations for items where details cannot be provided.

For items that remained unchanged, we refer users to the SPIRIT 2013 statement paper35 and its explanation and elaboration document.36 Detailed explanations of new and modified SPIRIT-DEFINE items in table 1, along with examples from oncology and non-oncology settings, will be presented in a further publication by the authors. Here, we provide general comments and a brief overview of the items that may be less self-explanatory.

For item 8a.3, the protocol should include a description of the underlying statistical methods used to set up and implement the adaptive trial design. For dose adaptations based on model based designs,59 authors should provide details and explanations of the statistical methods, including model assumptions, the choice of model parameters, and the mathematical form of the model, if applicable. For model based and model assisted dose-finding designs,2759 researchers should provide the rationale for choosing a target risk or toxicity rate or acceptable range,60 the details on the dose transformation (including the full skeleton and its elicitation), and bayesian prior distributions chosen, if applicable.51 For rule based designs, such as 3+3 or Rolling 6,61 the rationale for their use should be outlined. For other adaptations, such as early stopping for futility, the underlying statistical methods (eg, conditional power, predictive power, or posterior probability of treatment effect) should be clearly specified.1651

For item 20c.2, authors should describe methods to be used to handle missing data, and detail strategies for handling intercurrent events—that is, events (such as dosing delays, reductions, or interruptions) occurring after treatment initiation that might affect either the interpretation or the existence of the measurements associated with the clinical question of interest. Such events are not limited to those connected to treatment but might also include withdrawal of consent or deaths unrelated to treatment or disease. Different strategies might be used for different types of intercurrent events,51 and sensitivity analyses can be planned to assess the effect of the chosen strategies on the trial results.

Researchers should clearly specify the rationale for the starting dose and choice of the method, for example, according to current regulatory guidelines5262 (item 8a.5), as well as the trial adaptation process and stopping rules (item 8a.4). Dose transition pathways or dose decision paths can take the form of a decision table or a flow diagram (item 34) to map out in advance how a proposed design would recommend doses (escalate, de-escalate, stay, or stop) based on participants’ key outcomes (eg, what the next dose would be if a certain number of participants in a cohort experience a clinically significant adverse event).183358 For instance, if two participants experienced no clinically significant adverse events, a design might recommend escalating to the next higher dose, but if both participants experienced clinically significant adverse events, the same design might recommend de-escalating to a lower dose. The exact content and form of dose transition pathways can vary depending on the specific features of the trial design, and no standard format exists.


Owing to their importance and impact on later clinical development, EPDF trials should always be conducted to the same rigorous standards as their late phase counterparts including phase 2 and phase 3 randomised clinical trials. Moreover, although there are more EPDF trials than late phase trials, insufficient guidance has been available to date on the essential information that an EPDF protocol should provide to ensure accurate, reproducible, and transparent trial conduct.

SPIRIT-DEFINE is aimed at extending the SPIRIT 2013 statement, proposing or modifying items tailored to the specific features of EPDF trials across all disease areas. A total of 17 new items have been proposed, and 15 SPIRIT 2013 items have been modified or refined to fit EPDF settings.

SPIRIT-DEFINE, like other SPIRIT extensions, is developed through an international consensus driven process using the EQUATOR methodological framework. The key difference is that SPIRIT-DEFINE focuses on the distinctive features of EPDF trial protocols.

Application of SPIRIT-DEFINE

Like SPIRIT 2013, the SPIRIT-DEFINE guidance is not intended to dictate trial design or conduct. It is anticipated to serve as a useful resource to trialists, journal editors, peer reviewers, funders, regulators, and research ethics committees to promote best practice in designing protocols for EPDF trials and to facilitate protocol appraisal. We also envision that it will enable both trial participants and the public to be more confident in EPDF trial design. It proposes minimum requirements that EPDF trial protocols should address, not necessarily in the order as presented in the checklist, with authors reporting additional information to enhance the quality of trial protocols. SPIRIT-DEFINE covers general trial protocol principles applicable to a wide range of EPDF trials, regardless of disease setting (oncology or non-oncology) or participant population (eg, adults or paediatric groups, patients or healthy volunteers, populations with impaired hepatic or renal function). Its primary focus is on early phase clinical trials, in which interim dosing adaptations are taken using accumulating trial data to either escalate, de-escalate, stay at the current dose, or stop the trial early. Nonetheless, some aspects of this guidance might apply and benefit the reporting quality of other types of trial protocols including early phase trials with only one dose or later phase dose-finding trials with dose escalation or de-escalation parts.

Key strengths and limitations

There are noteworthy strengths and limitations. Box 3 describes how using the SPIRIT-DEFINE guideline can improve transparency, completeness, reproducibility of methods, and interpretation of EPDF protocols.

Box 3

Advantages of the SPIRIT-DEFINE checklist

The SPIRIT-DEFINE checklist can improve:


The impact of the guidance will vary depending on its adoption across different channels (journals, regulators, and ethics committees are the expected routes). By promoting full reporting of relevant protocol details in regulatory submissions, ethics applications, and protocol publications, the guidance will significantly enhance transparency.


By using the checklist of recommended SPRIIT-DEFINE items in an EPDF protocol, it enables researchers to develop comprehensive, robust, detailed, and well structured protocols, providing essential contents on the trial design, conduct, and analytical approaches. This checklist enhances clarity, aids understanding of the planned approaches, and could reduce delays, for example, owing to protocol amendments. SPIRIT-DEFINE is primarily intended to guide the planning and writing of a trial protocol before a trial begins. However, this guidance can also be useful in reviewing and enhancing the completeness of protocols for ongoing trials. For instance, researchers can clarify outcome measures or how missing data will be handled if they have not been clearly defined. The SPIRIT-DEFINE guidelines can guide revision of these definitions to improve data collection and analysis for the remainder of the trial. Any changes to the protocol should be noted as amendments, and should be reported to maintain the scientific integrity of the trial.

Reproducibility of methods

Reproducibility is a cornerstone of scientific research. By using the SPIRIT-DEFINE guidelines, researchers can increase the reproducibility of their trials, enhancing the reliability and trustworthiness of their findings. For instance, by requiring a clear and explicit description of the trial design with escalation and de-escalation strategies and any other adaptive features (including providing essential information on model specifications for a model based dose escalation design), readers can better understand how the design would work and replicate the assessment of the design's performance and analytical methods.


With a full description of relevant features in the protocol guided by the checklist, a proper critical appraisal of the protocol's strengths, limitations, and any potential sources of bias is possible, assisting in the interpretation of the trial's results. Also, the subsequent trial conduct can be better interpreted if what was prespecified in the protocol is fully reported.

  • SPIRIT=Standard Protocol Items: Recommendations for Interventional Trials; DEFINE=Dose-finding Extension.


The SPIRIT-DEFINE guidance was shaped by experts in different fields including trialists, clinicians, statisticians, regulators, ethics committee members, journal editors, and funders. Throughout the development process, we collaborated effectively with stakeholders and the public, including two patient partners who brought their perspectives to the consensus meeting and made important contributions to the guidance document. This SPIRIT-DEFINE effort also benefited from the contemporaneous CONSORT-DEFINE development.4763 Aligning CONSORT-DEFINE and SPIRIT-DEFINE involved continuous exchange of information and evaluation of the pertinence of proposed items resulting in items being shared by both statements, with these being rephrased to fit the purposes of each guideline.

To increase the accuracy and usability of the SPIRIT-DEFINE guidance, we engaged and involved an international group of multidisciplinary stakeholders (table S2 in web appendix 1 shows the roles and affiliations of consensus meeting participants). However, as with any survey, our results are subject to non-response bias. Respondents were self-selected, as only interested individuals participated in the Delphi survey, and the demographics of those who did not participate could not be determined. Consensus participants were specifically approached to reflect the multidisciplinary expertise and professional roles relevant to the design, conduct, and reporting of EPDF trials. Nevertheless, smaller groups (eg, groups outside Europe, North America, and Asia) holding different views were potentially under-represented during the Delphi process, at the consensus meeting, and on the DEFINE executive committee. However, the utilised systematic, evidence based approach to develop these guidelines, including rigorous review of reporting practices in EPDF trials by stakeholders, will have helped mitigate this potential bias.

Another limitation reflects the complexity of EPDF trials compared with randomised parallel group trials. The SPIRIT-DEFINE extension contains several new or modified items that might challenge adherence to the checklist. To guarantee the visibility of certain components, we intentionally kept them separate as independent items rather than combining them. For example, SPIRIT item 8 (trial design for a randomised parallel group trial) was modified to become SPIRIT-DEFINE item 8a.1, and 10 new items (8a.2-8a.11) corresponding to different features of EPDF trial designs (and can be considered as sub-items of item 8) were added to the checklist as separate items rather than combining them into one composite item.

Enhancing the uptake and relevance of SPIRIT-DEFINE

Wide dissemination of the SPIRIT-DEFINE guidance is essential to increasing its appropriate uptake, and this will be done as previously outlined,45 including but not limited to journals currently known to endorse SPIRIT through the EQUATOR Network. We are preparing an explanation and elaboration document to provide in-depth details and examples in different settings, to assist reviewers, editors, and readers who require additional information or clarity about specific items.

Finally, the design of EPDF trials is a rapidly evolving field, particularly with the increasing use of seamless phases as well as innovative approaches such as basket, umbrella, and platform trials that all pursue multiple objectives in increasingly efficient ways with faster go or no-go decisions. As newer trial designs emerge, additional considerations might be needed to facilitate transparency, reproducibility, minimise potential biases, and ensure the veracity of the findings of EPDF trials. Thus, the DEFINE executive committee will continue to monitor and assess the need for updates to both the SPIRIT-DEFINE and CONSORT-DEFINE63 guidelines.


The SPIRIT-DEFINE guideline provides recommendations for essential items to be considered and included in clinical trial protocols to improve completeness and reporting quality for EPDF trials. We strongly recommend that stakeholders and reviewers adopt the SPIRIT-DEFINE checklist to enable the delivery of high quality, transformative, EPDF trials that impact clinical care.


We gratefully acknowledge the additional contributions made by the DEFINE research team, DEFINE executive committee and collaborators/advisers, Delphi participants, and international consensus meeting participants (web appendix 3).


  • Contributors: CY, OS, JR had full access to the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. CY, OS, MU, AM, AK, A-WC, TRJE, JdB, LS, KRR, SH, SPI, SL, TJ, MD, CJW, and MC were involved with the concept and design of the study. All the authors contributed to the acquisition, analysis, and interpretation of data; revised the manuscript critically for important intellectual content; and participated in the DEFINE consensus meeting. CY, JR, MU, and OS drafted the manuscript. JR conducted the statistical analysis. DP, OS, and AE gave administrative, technical, and material support. CY is the guarantor. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

  • Funding: The SPIRIT-DEFINE study obtained no external funding. The principal investigator (CY) used internal staff resources, together with additional resources from external partners, to conduct this study. The SPIRIT-DEFINE study is a component of the DEFINE project, which also developed the MRC/NIHR funded CONSORT-DEFINE guidance. ICR-CTSU receives programmatic infrastructure funding from Cancer Research UK (C1491/A25351), which has contributed to accelerating the advancement and successful completion of this work. TJ received funding from the UK Medical Research Council (MC_UU_00002/14). DA acknowledges support from the National Institute for Health and Care Research (NIHR) Imperial Biomedical Research Centre. KSH received funding from the National Health and Medical Research Council of Australia (grants 2016420, 2015705) and acknowledges support from the Heart Foundation of Australia (grant 106607). LVM is funded by the Oak Foundation via the Royal Marsden Cancer Charity and acknowledges further funding support from the UK’s Experimental Cancer Medicines Centre Paediatric Network grant and the NIHR Royal Marsden/Institute of Cancer Biomedical Research Centre. JdB received programmatic funding support from Cancer Research UK and the UK NIHR, who together support the Experimental Cancer Medicine Centre at the Royal Marsden and ICR, as well as from the Medical Research Council. The funders and sponsor had no role in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; the preparation, review, or approval of the manuscript; or the decision to submit the manuscript for publication. This article reflects the views of the authors, the Delphi participants, and the Consensus Meeting participants, and may not represent the views of the broader stakeholder groups, the authors' institutions, or other affiliations.

  • Competing interests: All authors have completed the ICMJE uniform disclosure form at and declare: no support from any organisation for the submitted work. JdB has served on advisory boards and received fees from companies including Amgen, Astra Zeneca, Astellas, Bayer, Bioxcel Therapeutics, Daiichi, Genentech/Roche, GSK, Harpoon, ImCheck Therapeutics, Janssen, Merck Serono, Merck Sharp and Dohme, Pfizer, and Sanofi Aventis; is an employee of the Institute of Cancer Research (ICR), which have received funding or other support for his research work from AZ, Astellas, Bayer, Cellcentric, Daiichi, Genentech, Genmab, GSK, Janssen, Merck Serono, Merck Sharp & Dohme (MSD), Menarini/Silicon Biosystems, Orion, Sanofi Aventis, Sierra Oncology, Taiho, Pfizer, and Vertex (the ICR has a commercial interest in abiraterone and poly (ADP-ribose) polymerase (PARP) inhibition in DNA repair defective cancers and PI3K/AKT pathway inhibitors (no personal income)); was named as an inventor, with no financial interest, for patent 8 822 438, submitted by Janssen that covers the use of abiraterone acetate with corticosteroids; has been the chief investigator/principal investigator of many industry sponsored clinical trials; and is a National Institute for Health Research (NIHR) senior investigator. The views expressed in this article are those of the author(s) and not necessarily those of the NHS, the NIHR, or the Department of Health. AM is employed by GSK. TRJE has received honorariums for consultancies (payable to the employing institution) from Ascelia, Astra Zeneca, Bayer, Bicycle Therapeutics, Bristol Myers Squibb, Celgene Eisai, Karus Therapeutics, Medivir, MSD, Otsuka, Roche, and Seagen; honorariums for speaker’s fees (payable to employing institution) from Astra Zeneca, Ascelia, Bayer, Bicycle Therapeutics, Bristol Myers Squibb, Celgene, Eisai, Nucana, Otsuka, Medivir, MSD, Roche, Seagen, and United Medical; support of costs of commercial clinical trials (payable to employing institution) from Astra Zeneca, Basilea, Bayer, Celgene, MiNa Therapeutics, Roche, Pfizer, Sierra, Lilly, Eisai, Glaxo Smith Kline, Novartis, Bicycle Therapeutics, Johnson and Johnson, CytomX, Vertex, Plexxikon, Boehringer, Athinex, Adaptimmune, Bristol Myers Squibb, MSD, Medivir, Versatem, Nucana, Immnuocore, Berg, Beigene, Iovance, Modulate, BiolinerX, Merck Serono, Nurix Therapeutics, T3P, Janssen Clovis, Sanofi-Aventis, Halozyme, Starpharma, Union Chimique Belge, Sapience, Seagen, Avacta, and Codiak; and funding from Cancer Research UK, Chief Scientist’s Office Scotland, and Medical Research Council UK; is the editor-in-chief of the British Journal of Cancer and has an honorary clinical contract with the NHS Greater Glasgow and Clyde Health Board. RP is an employee and a stockholder in F Hoffmann la Roche, and a family member is also an employee and a stockholder of F Hoffmann la Roche. KSH declares grant funding (payable to the employing institution) received by the Medical Research Future Fund (grant 2007425), National Health and Medical Research Council of Australia (grants 2016420 and 2015705), and Heart Foundation of Australia (grant 106607). SH and A-WC are members of the SPIRIT-CONSORT executive group and are leading the current update of the SPIRIT 2013 and CONSORT 2010 reporting guidelines, funded by the UK Medical Research Council NIHR Better Methods, Better Research (MR/W020483/1). MU acted as consultant for eXYSTAT, Saryga, PTC Therapeutics International, and ImCheck Therapeutics. MC is director of the Birmingham Health Partners Centre for Regulatory Science and Innovation, director of the Centre for Patient Reported Outcomes Research, and is a NIHR senior investigator. MJC has received funding from the NIHR, UK Research and Innovation (UKRI), NIHR Birmingham Biomedical Research Centre, NIHR Surgical Reconstruction and Microbiology Research Centre, NIHR, Applied Research Collaboration West Midlands, UK SPINE, Research England, European Regional Development Fund DEMAND Hub at the University of Birmingham and University Hospitals Birmingham NHS Foundation Trust, and NIHR Birmingham-Oxford Blood and Transplant Research Unit in Precision Transplant and Cellular Therapeutics; funding from Health Data Research UK, Innovate UK (part of UKRI), Macmillan Cancer Support, UCB Pharma, Janssen, GSK, Gilead Sciences, European Commission, European Federation of Pharmaceutical Industries and Associations, and the Brain Tumour Charity; and personal fees from Aparito, CIS Oncology, Takeda Pharmaceuticals, Merck, Daiichi Sankyo, Glaukos, GSK, the Patient-Centered Outcomes Research Institute, Genentech, and Vertex Pharmaceuticals outside the submitted work; has received lecture fees from the University of Maastricht, Maastricht, Netherlands; a family member owns shares in GSK. DPR is the volunteer vice president of the Canadian Arthritis Patient Alliance, a patient led and run organisation that derives most funding from independent grants from pharmaceutical companies. OB is an employee of Bayer AG. JM is an employee of Cytel (Australia). LS declares grant funding from AstraZeneca, Bayer, Pfizer, Merck, Roche, REPARE, Treadwell, and Janssen; has provided expert testimony for CADTH Health Canada; and declares AstraZeneca stock/options ownership. LM received honorariums for speaker fees from Bayer and as co-organiser, chair, and speaker at two educational preceptorships (online webinars); and advisory board/consultancy honorariums from Tesaro, BMS, and Illumina; is also a member of external data monitoring committees for early phase clinical trials run between Eisai and Merck. RL is an employee and stockholder of Bristol Myers Squibb. JB declares consultancy fees from Mirati, Insmed, Oxford BioTherapeutics, Biosapien, EMD Serono, Ipsen, Merck Sharp and Dohme, Perus, BMS, and Bexion; declares grant funding from Abbvie, Astellas, Atreca, Bayer, Dragonfly, I-Mab, Lilly, Incyte, EMD Serono, Pfizer, BMS, Transcenta Therapeutics, Tyra, Totus, Sumitomo Dainippon Pharma Oncology, 23 and me, Parthenon, and Hibercell; and sits on data safety monitoring committees for Astra Zeneca, Novocure, and Boehringer-Ingelheim. All other authors declare no conflict of interest.

  • Patient and public involvement: The DEFINE study’s lead for patient and public involvement and engagement (PPIE) (AK) was closely involved in the development of the project, and actively contributed to the development of the protocol and each of the development stages. We also engaged with several PPIE partners from both oncology and non-oncology fields for them to feed back on the checklists and to ensure that patient voices were heard. As part of our dissemination plan, we will be producing PPIE led lay publications to chart the development of both the SPIRIT-DEFINE and CONSORT-DEFINE guidelines.

  • Provenance and peer review: Not commissioned; externally peer reviewed.