Early phase dose-finding (EPDF) or dose-escalation trials, also referred to as phase I or phase I/II, are critical in clinical therapy development. Depending on the drug and endpoint of interest, the studies may be conducted in healthy volunteers or patients with the condition or disease. These studies involve interim dose decisions and may provide data on safety, adverse effects, pharmacokinetics (characterisation of a drug’s absorption, distribution, metabolism, and excretion), pharmacodynamics, biomarker activity, clinical activity and other information needed to choose a suitable dosage range and/or administration schedule to inform further studies. Results from these trials directly influence decisions on further development and whether the selected doses and schedules are sufficiently safe and have promising results on activity.

A clinical trial protocol is a vital document that details the study rationale, methods, organisation and ethical considerations.1 By providing the details to guide the conduct of a high-quality study, a well-written protocol is a shared central reference for the study teams2 3 and facilitates appraisal of its scientific, methodological, safety and ethical rigour by external reviewers. However, protocols can vary greatly in content and quality despite their importance.2 3 To address this, the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) 20132 statement was established to provide evidence-based guidance for the minimum essential content of clinical trial protocols and is widely endorsed as the international standard for trial protocols. Although the considerations of SPIRIT 2013 are largely applicable across many types of trials, some circumstances require additional items.2 Guidance on content specific to EPDF trials, including dose and schedule determination based on safety/tolerability either alone or with one or more pharmacokinetic or activity markers, is lacking. Examples of features unique to such trials include:

Starting dose and its justification.

How interim dose decisions will be undertaken (including clearly defined outcome measures and their assessment window, and analysis populations for interim adaptations).

How future recommended dose(s) will be selected.

Incomplete or unclear information on the design, conduct and analysis in EPDF trial protocols and reporting papers hinders the interpretability and reproducibility of the results from such studies, which may impact the overall clinical development timeline, lead to erroneous conclusions on safety and efficacy, and compromise the safety of trial participants.4

This is particularly relevant as a considerable number of early phase trials are sponsored and run by academic institutions or publicly funded organisations with funding from non-commercial sources, including Research Councils and medical charities (eg, Cancer Research UK, Wellcome Trust, US National Cancer Institute). In the UK, 159 out of 1157 (14%) phase I clinical trials, started in 2014–2018, had non-industry sponsors (data from ClinicalTrials.gov). This emphasises the importance of this research to public research institutions and industry alike. Based on results from ClinicalTrials.gov of trials in all countries, there are substantially more phase I trials than phase III trials (13 826 phase I vs 9501 phase III which started in 2014–2018). Data from pharmaceutical trials in the USA in 2004–2012 show that the estimated average cost of a phase I trial across all therapeutic areas ranged from US$1.4 to US$6.6 million5; such high costs reinforce the importance of managing resources efficiently. The attrition rate throughout the drug development process is high, and the success rate between phase I studies and marketing authorisation has been reported as between 9.8% and 13.8%,6 7 with failure being primarily attributable to either poor tolerability or lack of biological activity (79% of failed studies over the period 2016–2018).8 In this context, EPDF trial results must be assessed accurately to avoid poor dose selection, which will often lead to failed trials (phases II and phase III), delays in regulatory submissions, additional postmarketing commitments or dose changes postapproval due to excessive toxicities or lack of efficacy.9

The use of more complex dose-escalation designs such as model-assisted or model-based designs is rising: 1.6% (20/1235 phase I published cancer trials) in 1991–200610 to 6.4% (11/172) by 2012–2014.11 Such designs are more complex to implement12–14 and require the specification of more design features.15 Further transparency and reporting demands are needed in such protocols and trial reports to facilitate understanding of the design, ensure the methods and results are reproducible, and explain how dose decisions will be and have been made.16–18

More than 580 biomedical journals now require that trial reports conform to the CONsolidated Standards Of Reporting Randomised Trials (CONSORT) 2010 reporting guidelines for randomised parallel group clinical trials or an appropriate CONSORT extension to improve transparency, reproducibility, consistency and accuracy in reporting.19–21 A total of 153 journals, as well as a growing number of commercial and non-commercial funders, regulators, trial organisations and patient groups, have also endorsed SPIRIT.22 A systematic review based on more than 16 000 trials published in 2012 showed that journal endorsement of the CONSORT guidelines was associated with more completely reported randomised trials.23

Neither the original guidance, SPIRIT 2013 and CONSORT 2010, nor their extensions adequately cover the features of EPDF trials. The DosE-FIndiNg Extensions (DEFINE) study aims to enhance transparency, completeness, reproducibility and interpretation of EPDF trial protocols and their reporting of results, across all disease areas, and to build on the checklists outlined in the SPIRIT 2013 and CONSORT 2010 statements.