Dose escalation belies the therapeutic intent of phase I trials


April 14, 2019

Adapted from the text of a ‘Comments and Controversies’ piece rejected by the Journal of Clinical Oncology in April 2018

Not long ago in the Journal of Clinical Oncology, bioethicist Jonathan Kimmelman challenged the doctrine of therapeutic intent in phase I trials. Against ASCO’s “therapeutic position,” consistent normatively with the medical profession’s traditional alliance with the individual patient, Kimmelman¹ advocated a “research position” consistent with the professional bioethicist’s penchant for subordinating the individual person to “scientific advance” and other such collectivistic ends. In “reaffirming and clarifying” its original position, the ASCO response² did little beyond rebutting several evidentiary points raised by Dr. Kimmelman (as to tallies of adverse events and therapeutic outcomes in phase I trials) and recapitulating the severely constrained decision-space within which patients and their oncologists must actually weigh these facts when considering phase I trial enrollment. In the process, the parties to this exchange overlooked opportunities to elevate simultaneously the scientific and therapeutic stature of phase I trials.

Only individual persons of course can truly be said to possess ‘intent’. But if we must talk about the intent of a trial itself — abstracted from the intent of the various individuals (researchers, physicians, patients) involved — then we should do this by reference to the methods stipulated in the trial design. Thus, our attention in this matter should turn to the intent of the trial methodologist. Nevertheless, the exchange¹⁻² paid remarkably little attention to pertinent details of trial methodology. At no point was any specific phase I design feature referenced explicitly. Instead, only aspirational declarations were offered—and without citations: “Innovative phase I trial designs can and should limit the risks of patients receiving a dose … that is too low to be effective;”² “many [phase I cancer studies] use designs aimed at maximizing clinical benefit.”¹ To be sure, the original ASCO statement³ included numerous references to the methodologic literature (see its citations 6–14, 17 and 18), most advocating model-based dose-finding methods of the Continual Reassessment Method (CRM) family⁴. (These are the most widely used model-based dose-finding methods in actual phase I trials⁵⁻⁶.)

What would detailed scrutiny of actual phase I trial methodology have contributed? Anyone who reads many phase I trial reports often encounters the words, “intrapatient dose escalation was not allowed.” This awkward turn of phrase means nothing more than that dose-escalation designs eschew titration. Amusingly, in preferring so awkward a phrase over the far simpler ‘titration’, the literature oddly rehearses its own ill-considered consensus that “it can be difficult to present and interpret results of trials that allow intrapatient dose escalations”⁷. It cannot amuse the trial participant, however, to know that dose-escalation designs conjoin two highly problematic practices that simultaneously render trials less safe and less therapeutic.

The earliest-enrolling participants in dose-escalation trials suffer a well-known ‘first-mover disadvantage’ upon entering cohorts permanently assigned to doses likely to be subtherapeutic. Concomitantly, drug-naïve participants who have not yet demonstrated tolerance of those lower doses leapfrog into higher-dose cohorts. Appreciated in light of a realistic conception of drug tolerance as heterogeneous⁸, these practices seriously undermine the claim of therapeutic intent. They present at once a failure to maximize potential for benefit (in the former practice) and a failure to exercise reasonable precaution against harm (in the latter). Articulating a principle I term ‘precautionary coherence’ (PC), I have recently advanced a direct attack⁹ on this latter practice, which perforce addresses the former as well. In the course of rectifying these twin failures, I also demonstrate a ‘dose-survival curve’ dispelling the abovementioned consensus that it is “difficult to present and interpret results of trials that allow [titration].”

The PC principle lies entirely within the comprehension of any clinician or trial participant, as I demonstrate with a lay explainer¹⁰. Indeed, whereas coherence principles in dose finding¹¹ have classically represented an arcane defense of the conceptual integrity of dose-finding designs from ad hoc meddling by clinicians, PC represents precisely the opposite: the reassertion of a crucial clinical intuition, against biostatistics’ traditional regard for patients as ‘exchangeable’¹². The broad accessibility of the PC principle has two immediate implications.

Firstly, it opens up the recherché realm of dose-finding methodology to universal scrutiny. The further development of the discussion Dr. Kimmelman has initiated need not pay so much deference to the supposed expertise of dose-finding methodologists. The details of dose-finding designs lie open to critical scrutiny from clinical and ethical perspectives.

Secondly, since trial participants will readily understand the abovementioned first-mover disadvantage, informed consent must address the issue frankly. This becomes especially urgent when multiple trials may contend for the participation of any given patient. In this competitive landscape, patients have a right to methodological transparency relevant to their consideration of trials’ therapeutic intent and potential.

Whereas Kimmelman seems to advance his “research position” as normatively desirable, I concur with the ASCO position that phase I trials ought to have therapeutic intent. But as to the positive question of whether they actually do, I must side with Kimmelman — albeit not for the reasons he advances, which the ASCO reply² handily rebutted. Rather, I fear the near-universal adoption of dose-escalation designs in phase I trials belies their claim to therapeutic intent.

Beyond the mere “reaffirming and clarifying” of its position, therefore, I call on ASCO to particularize its recommendations on phase I trial methodology. Specifically, ASCO should stipulate:

Whatever our distinct ethical perspectives, we all share common goals such as ensuring the efficient functioning of the drug development process. We all want high-quality phase I trials with sound scientific rationale to enroll swiftly and yield their results without undue delay. Upending the perverse illogic of a first-mover disadvantage can only accelerate enrollment in these trials. (The Executive Director of a rare-cancer foundation not long ago acknowledged to me delaying notifying its community about new phase I trials, for concern they might still be enrolling cohorts at subtherapeutic doses!) Likewise, even ’moral strangers’¹³ standing at opposite ends of the individualism-collectivism axis can agree on the importance of improving the population-level efficacy of cancer therapeutics. By establishing dose individualization as an aim in phase I trials, we will obtain earlier efficacy signals, quicker progress to phase II for worthy candidates, and greater efficacy for our approved drugs.¹⁴


1. Kimmelman J: Is Participation in Cancer Phase I Trials Really Therapeutic? JCO 35:135–138, 2016

2. Weber JS, Levit LA, Adamson PC, et al: Reaffirming and Clarifying the American Society of Clinical Oncology’s Policy Statement on the Critical Role of Phase I Trials in Cancer Research and Treatment. JCO 35:139–140, 2016

3. Weber JS, Levit LA, Adamson PC, et al: American Society of Clinical Oncology Policy Statement Update: The Critical Role of Phase I Trials in Cancer Research and Treatment. JCO 33:278–284, 2015

4. O’Quigley J, Pepe M, Fisher L: Continual reassessment method: a practical design for phase 1 clinical trials in cancer. Biometrics 46:33–48, 1990

5. Rogatko A, Schoeneck D, Jonas W, et al: Translation of innovative designs into phase I trials. J Clin Oncol 25:4982–4986, 2007

6. Love SB, Brown S, Weir CJ, et al: Embracing model-based designs for dose-finding trials. British Journal of Cancer 117:332–339, 2017

7. Le Tourneau C, Lee JJ, Siu LL: Dose escalation methods in phase I cancer clinical trials. J Natl Cancer Inst 101:708–720, 2009

8. Norris DC: Dose Titration Algorithm Tuning (DTAT) should supersede ‘the’ Maximum Tolerated Dose (MTD) in oncology dose-finding trials. F1000Research 6:112, 2017

9. Norris DC: Precautionary Coherence Unravels Dose Escalation Designs. bioRxiv 240846, 2017

10. Norris DC: The conduct of most first-in-human oncology drug trials is conceptually incoherent, and unethical. [Internet], 2017[cited 2018 Mar 18] Available from:

11. Cheung YK: Coherence principles in dose-finding studies. Biometrika 92:863–873, 2005

12. Greenland S, Robins JM: Identifiability, exchangeability and confounding revisited. Epidemiol Perspect Innov 6:4, 2009

13. Engelhardt HT: The foundations of bioethics: liberty and life with moral diversity. Reason Pap 101–108, 1997

14. Norris DC: One-size-fits-all dosing in oncology wastes money, innovation and lives. Drug Discov Today 23:4–6, 2018