Clinical stakes of early clinical development in oncology


1,2Jean-Charles SORIA, MD, PhD
Drug Development Department
(DITEP), Gustave Roussy
Cancer Campus, and
South-Paris University
Villejuif, FRanCE

Clinical stakes of early clinical development in oncologyy

Interview with S. Postel-Vinay
and J. C. Soria,

Over the last decade, early phase trials in oncology have switched from the “one-size fits all” approach of the cytotoxic chemotherapy area, to a “personalized and precision medicine” strategy based on molecular selection for patients receiving novel molecularly targeted agents (MTAs). This patient-centered approach has revolutionized the role of phase 1 trials, which are now not only dose-finding studies, but also hypothesis-testing trials aiming at making the proof of concept of a given biological rationale. As such, many paradigms of the traditional drug development model have been challenged by novel stakes including molecular enrichment, early development of companion biomarkers, consideration of late or cumulative toxicities, appropriate recommended phase 2 dose definition and customized phase 1 trial designs for MTAs. Consequently, the role of early phase trials in accelerating the drug development and making it more efficient is increasingly important, and the smart design of phase 1 trials can now strongly influence the “go/no-go” decision for further drug development. In this issue of Medicographia, I will discuss the current stakes of early clinical drug development in oncology, and give tracks for improving and fostering the success of this process.

Medicographia. 2015;37:325-329 (see French abstract on page 329)

What is your experience in the Gustave Roussy Cancer Center?

After a postdoctoral fellowship at the University of Texas MD anderson Cancer Center, I (Jean-Charles Soria) was appointed assistant Professor of Medicine at Gustave Roussy Cancer Center in november 2001. I initially focused on lung cancer, but in 2006, was asked to lead the phase 1 unit. In 2006, I also became Professor of Medicine at South Paris University and was appointed Head of the Early Drug Development multidisciplinary committee at Gustave Roussy in 2007. In 2008, we created a dedicated inpatient and outpatient facility for phase 1, for which I acted as chief of service. In 2010, with Professor andré, I participated in the foundation of a new InSERM research unit (U981) to identify molecular predictors of efficacy in anticancer treatments. The growth of phase 1 activity (Figure 1, page 326) led in September 2013 to the establishment of a Drug Development Department (Département d’Innovations Thérapeutiques et d’Essais Précoces [DITEP]) encompassing over 60 phase 1 trials. Of these, at least 10 per year are dedicated to hematological malignancies and 6 to the combination of radiotherapy with molecularly targeted agents (MTas). The DITEP at Gustave Roussy is one of the top three facilities in Europe for anticancer early drug development.

What are the clinical stakes of the early clinical drug development in oncology?

Currently, the most important clinical stakes of the early clinical drug development in oncology are to smartly adapt the phase1 trials to novel MTas and to favor hypothesistesting trials, ie, based on a specific preclinical rationale, while keeping the costs and duration of early phase studies financially and ethically suitable.

Indeed, the advent of MTas over the last 15 years has revolutionized the early phase trials landscape. If the primary end point of phase 1 trials should remain the determination of the recommended phase 2 dose (RP2D) and description of the drug safety, early phase trials now also play a pivotal role in multiple other fields that will be key for the success of later drug development. notably, the “one-size fits all” approach, which does not take into account the patient-to-patient molecular variability of the tumor, is not acceptable anymore in the context of MTas. Moreover, as illustrated by the strikingly high attrition rate of the number of drugs developed between phase 1 and phase 2 trials, and by the proportion of drugs that still fail late in their development,1 there is an urge to improve early phase trial efficiency.

MTas, which now represent the vast majority of drugs that come into early phase trials, have challenged many paradigms that were established in the era of conventional cytotoxic chemotherapy (CCC) and brought new challenges, such as the need for selecting the right patients for molecular enrichment, the necessity to have access to the tumor for molecular analysis, the difficulty in determining the optimal dose and schedule, the importance of early development of companion biomarkers, the pitfall of the feasibility of drug combinations, and the dilemma of the choice of the most appropriate method and criteria for assessing tumor response.

Overall, I think that phase 1 trials face new stakes that have now been well identified, but have not yet been appropriately addressed in the current clinical practice. This relative delay in modifying phase 1 trials designs and the way such studies are carried out, is currently impacting on the success of some drugs’ development. Therefore, it is now crucial to optimally design all phase 1 trials as true “hypothesis-driven” and “hypothesis-testing” trials, by considering the above-mentioned points, prior to starting the clinical evaluation.

Figure 1
Figure 1. Adult patients enrolled in phase 1 trials at the Gustave Roussy Cancer

Figure illustrating the yearly number of new patients enrolled in phase 1 trials at Gustave Roussy.

What makes early clinical drug development in oncology efficient?

First of all, and prior to any first-in-human phase 1 trial, very robust preclinical data must be obtained in relevant and appropriate preclinical models. at the clinical stage, the thorough consideration of all the challenges and stakes previously mentioned will make early clinical drug development efficient. The right patient selection, the early development of companion biomarkers (including predictive biomarkers, pharmacokinetic [PK] and pharmacodynamic [PD] biomarkers as well as other intermediate or molecular biomarkers), the search for noninvasive access to tumor material and the appropriate choice of the response evaluation method and criteria are all key elements of successful early drug development.2 Furthermore, other important aspects should be considered, such as the choice of an appropriate design, which will favor both a rapid determination of the maximum tolerated dose (MTD) and RP2D (to minimize the number of patients included at low and potentially inefficient doses), and favor the inclusion of a sufficient number of patients at doses known to be safe and at which the target is modulated, so that more information is obtained about interpatient variability. Moreover, molecular enrichment (ie, the inclusion of patients whose tumor presents the molecular aberration of interest) should be favored, as this proportion currently remains insufficient.3 Phase 1 trials should also incorporate clear and early “go/no-go” decision rules, so that the evaluation of inefficient or too toxic compounds is terminated early. For example, molecules that have not delivered any signs of activity at the end of the phase 1 trial, despite an appropriate patient selection, should not be further investigated, even in combination.

Finally, mechanisms of resistance should be studied as early as at the phase 1 trials, so that preclinical research about how to overcome and bypass them can be started while later phase trials are being performed.

How do targeted therapies change the situation?

Targeted therapies have dramatically changed the situation for several aspects. First, patient selection has become crucial. Indeed, most phase 1 trials are now based on robust preclinical data, and as such, should be designed as “hypothesis-driven” and “proof-of-concept” trials rather than only dose-finding studies. This is well illustrated by the recent study published by Tsimberidou et al, which reported that the likelihood of experiencing some degree of response was almost four times higher in patients receiving a drug matched to their tumor’s molecular aberration, than in patients receiving an unmatched drug (29% vs 8%, respectively).4 This implies the parallel development of several companion biomarkers, and notably predictive biomarkers, which will allow selection of the right patient population. Phase 1 trials play a key role in the development of such biomarkers, as they will bring the first clinical data on assays that will require further validation in later phase trials prior to approval for clinical use. Beyond the development of an appropriate selection biomarker, tumor accessibility is a key stake of hypothesis-driven trials in the context of personalized medicine.

If archived tumor material is usually available for almost all patients, it does not represent the optimal material to use for patient selection: indeed, the molecular profile of the tumor can vary over time and frozen material, which is required for some molecular diagnostics, is usually not available. a first option is then to rebiopsy the patient’s tumor prior to starting the phase 1 trial (and ideally also when resistance occurs), but this is not always feasible or simply accepted by the patient. Therefore, liquid biopsies, including the analysis of circulating tumoral Dna and circulating tumoral cells, represent a very promising option for overcoming this recurrent issue. Besides the patient selection challenge, the determination of the dose that should be used in phase 2 trials is a real stake of current early phase trials evaluating MTas. Indeed, the MTD is not always reached with such agents, and the determination of the RP2D is consequently sometimes based on alternative parameters, such as the maximum feasible dose, or PD and PK data (optimal biological dose). This highlights the importance of having real-time access to PK and PD data as these can be key, not only for the dose-escalation process, but also for the dose-recommendation process. also, the method of determination of the RP2D should evolve to better correspond to the MTas’ schedule of administration and pattern of toxicities. Indeed, as these agents are due to be administered for a prolonged period of time, late toxicities (ie, occurring after the dose-limiting toxicity period), cumulative toxicities, and prolonged moderate toxicities should deserve more attention.5 as a matter of fact, such toxicities seriously threaten the quality of life of the patient and can impact on patient’s observance at later stages of drug development. This challenge is well-illustrated by the recent publication of Fontes-Jardim et al,6 which notably reports the poorer ability of phase 1 trials of MTas to predict the dose that will be used in phase 3 trials or the dose at which the drug will be approved, as compared with phase 1 trials evaluating CCC. To further investigate this issue, guidelines and recommendations have very recently been elaborated by a European Organisation for Research and Treatment of Cancer (EORTC)– led task force of international phase 1 experts, and which are now publicly available.7 One major point of these recommendations is that the RP2D should be determined based on achieving >75% of the intended relative dose-intensity, rather than being based on a certain proportion of severe toxicities.

Another challenge that phase 1 trials are now facing is the choice of the optimal technique and criteria for tumor response assessment. Indeed, if Response Evaluation Criteria In Solid Tumors (RECIST) criteria and computed tomography scans have proven useful and efficient for most—if not all— CCC, MTas sometimes cause true responses, which are either delayed or not evaluable by RECIST criteria. Examples include pseudoprogressions or prolonged stable diseases observed with gastrointestinal stromal tumors (GISTs) treated with c-kit inhibitors, which are better evaluated by Choi criteria, changes in tumor perfusion and vasculature under antiangiogenic agents, which are better evaluated by contrast ultrasound or perfusion magnetic resonance imaging, or more recently, pseudoprogressions or delayed responses observed under immunotherapies, which have led to the development of specific criteria, namely the immune-related Response Criteria (irRC).8 Therefore, the response evaluation method and criteria should be thoroughly chosen in phase 1 trials, as the detection of activity of the drug will be key for the “go/no-go” decision of the drug development in later phase trials.

Other stakes of early clinical drug development include the smart and relevant development of drug combinations: indeed, combining an MTa with a CCC or with another MTa has proven more challenging than expected, both in terms of toxicity management and determination of the optimal administration schedule or dosage. another upcoming challenge will be a smart combination with immune checkpoint inhibitors. Moreover, efficacy results of combination trials have sometimes been disappointing, and a lot remains to be learned in this field.

In a nutshell, MTas do considerably change the landscape of phase 1 trials and most of the required amendments to the current uses and practice have now been well identified. This has been followed by the elaboration of guidelines or recommendations that are unfortunately still inconstantly followed, but have encouragingly been increasingly implemented in recent phase 1 trials.

Does phase 1 trial design have to be amended?

Yes, phase 1 trial design should be amended in order to better address the above-mentioned points and better correspond to the MTa landscape. although important evolutions are required, the skeleton of phase 1 trials should remain unchanged: the primary end point should still be the evaluation of the drug safety profile and RP2D; phase 1 trials should also still encompass one dose–escalation and one dose–expansion phases (although the design of each of them could be modified); and finally, the dose-expansion phase should still focus on fine-tuning the recommended dose, rather than on evaluating efficacy.

However, major changes could be made to improve current phase 1 trials.9 First, model-based designs are still too rarely used, mainly because the “3+3” design is easy to work with, and has been doing well over years. The “accelerated titration,” « modified toxicity probability interval (MTPI), » or “Bayesien” designs may be more appropriate. Also, some provisions should be made to facilitate both the enrolment of patients whose tumor harbors the molecular alteration of interest, as well as the enrolment of patients at doses known to be safe and that display both a PD effect and the desired PK profile. In order to best allow this enrichment in molecularly selected patients, a smart collaboration between participating centers should be favored, rather than systematic competitive enrolment of molecularly unselected patients.

Figure 2
Figure2. The new trend in oncology drug development.

Second, the traditional phase 1 eligibility criteria should be revisited and customized for each phase 1 trial rather than reproduced from one trial to another. For example, systematically excluding patients having already received compounds targeting the same pathway, patients with brain metastasis, or patients with moderate renal or liver dysfunction, is not relevant, and eligibility criteria should be thoroughly defined to best match the agent evaluated.9

Third, recommendations that have recently been established to best define the RP2D in phase 1 trials of MTas, should be followed whenever possible.7 This includes notably revisiting all toxicities (whatever their cycle of occurrence) described during the trial prior to determining the next dose, when completing a given dose level, as well as targeting the delivery of >75% of the intended dose intensity when defining the RP2D.

Altogether, such design modifications should foster early drug development in oncology and make this process more efficient for MTas, not only at the phase 1 step, but also for later steps of the compound development.

Finally, it is very important to realize that modern phase 1, in which the tested compound demonstrates clear signs of clinical activity, can be expanded up to hundreds of patients. In that regard, phase 1 trials testing new anaplastic lymphoma kinase (aLK) inhibitors (ie, crizotinib and ceridinib) have both enrolled over 250 patients. Phase 1 trials testing programmed cell death (PD-1) or programmed death ligand 1 (PD-L1) antibodies have enrolled a minimum of 200 patients (nivolumab) and asmany as 1100 patients (pembrolizumab). These changes are illustrated in Figure 2.

How may phase 1 accelerate regulatory approvals of new expected anticancer drugs?

Recent successful phase 1 trials illustrate very well how early phase trials can accelerate regulatory approvals of new anticancer drugs. Such examples include phase 1 studies evaluating olaparib (aZD2281, astra Zeneca; a poly aDP-ribose polymerase [PaRP] 1/2 inhibitor), vemurafenib (PLX4032, Plexxikon; a B-raf inhibitor), crizotinib (PF-02341066, Pfizer; an aLK inhibitor), and the hedgehog inhibitor vismodegib (GDC-0449, Roche). In all these studies, the role of predictive biomarkers for response has been pivotal in the hypothesis-testing study design and provisions have been made to favor the inclusion of molecularly selected patients whose tumors presented the molecular alteration of interest. Consequently, the proof of concept and activity of the drug detected in phase 1 trials has, in some cases, even been sufficient to justify the procurement of a conditional approval from the US Food and Drug administration (FDa), based on the results of the phase 1 only. Crizotinib and Vismodegib are the best examples of success of this strategy, with accelerated timelines for approval, ie, approximately 5 years only between the beginning of the phase 1 trials and the regulatory approval.

We therefore believe that enrichment strategies, as early as phase 1 trials, are definitely key in accelerating drug development in oncology. To favor this, phase 1 centers need a very large panel of readily available phase 1 and phase 2 trials (or networks of partners to which the patients could be referred) to optimally deal with the diversity of available targets.

1. DiMasi Ja, Grabowski HG. Economics of new oncology drug development. J Clin Oncol. 2007;25(2):209-216.
2. Yap Ta, Sandhu SK, Workman P, de Bono JS. Envisioning the future of early anticancer drug development. Nat Rev Cancer. 2010;10(7):514-523.
3. Gutierrez ME, Kummar S, Giaccone G. next generation oncology drug development: opportunities and challenges. Nat Rev Clin Oncol. 2009;6(5):259-265.
4. Tsimberidou aM, Iskander nG, Hong DS, Wheler JJ. Personalized medicine in a phase I clinical trials program: the MD anderson Cancer Center initiative. Clin Cancer Res. 2012;18(22):6373-6383.
5. Postel-Vinay S, Gomez-Roca C, Molife LR, anghan B. Phase I trials of molecularly targeted agents: should we pay more attention to late toxicities? J Clin Oncol. 2011;29(13):1728-1735.
6. Jardim DL, Hess KR, Lorusso P, Kurzrock R, Hong DS. Predictive value of phase I trials for safety in later trials and final approved dose: analysis of 61 approved cancer drugs. Clin Cancer Res. 2014;20(2):281-288.
7. Postel-Vinay S, Collette L, Paoletti X, Rizzo E. Towards new methods for the determination of dose limiting toxicities and the assessment of the recommended dose for further studies of molecularly targeted agents—dose-limiting toxicity and toxicity assessment recommendation group for early trials of targeted therapies, an European Organisation for Research and Treatment of Cancer-led study. Eur J Cancer. 2014;50(12):2040-2049.
8. Wolchok JD, Hoos a, O’Day S, et al. Guidelines for the evaluation of immune therapy activity in solid tumors: immune-related response criteria. Clin Cancer Res. 2009;15(23):7412-7420.
9. Hollebecque a, Postel-Vinay S, Verweij J. Modifying phase I methodology to facilitate enrolment of molecularly selected patients. Eur J Cancer. 2013;49(7): 1515-1520.

Keywords: molecularly targeted drug; oncology; targeted therapy