S. Postel-Vinay

Circulating Cell-Free Tumor DNA Analysis of 50 Genes by Next-Generation Sequencing in the Prospective MOSCATO Trial

Purpose: Liquid biopsies based on circulating cell-free DNA (cfDNA) analysis are described as surrogate samples for molecular analysis. We evaluated the concordance between tumor DNA (tDNA) and cfDNA analysis on a large cohort of patients with advanced or metastatic solid tumor, eligible for phase I trial and with good performance status, enrolled in MOSCATO 01 trial (clinical trial NCT01566019). Experimental Design: Blood samples were collected at inclusion and cfDNA was extracted from plasma for 334 patients. Hotspot mutations were screened using next-generation sequencing for 50 cancer genes. Results: Among the 283 patients with tDNA–cfDNA pairs, 121 had mutation in both, 99 in tumor only, 5 in cfDNA only, and for 58 patients no mutation was detected, leading to a 55.0% estimated sensitivity [95% confidence interval (CI), 48.4%–61.6%] at the patient level. Among the 220 patients with mutations in tDNA, the sensitivity of cfDNA analysis was significantly linked to the number of metastatic sites, albumin level, tumor type, and number of lines of treatment. A sensitivity prediction score could be derived from clinical parameters. Sensitivity is 83% in patients with a high score (≥8). In addition, we analyzed cfDNA for 51 patients without available tissue sample. Mutations were detected for 22 patients, including 19 oncogenic variants and 8 actionable mutations. Conclusions: Detection of somatic mutations in cfDNA is feasible for prescreening phase I candidates with a satisfactory specificity; overall sensitivity can be improved by a sensitivity score allowing to select patients for whom cfDNA constitutes a reliable noninvasive surrogate to screen mutations. Clin Cancer Res; 22(12); 2960–8. ©2016 AACR.

Phase I trials in oncology: a new era has started

The landscape of oncology drug development is currently undergoing fascinating revolutions and there is an urge to re-challenge many paradigms of early phase trials, which are currently becoming out-of-date. Needless to say, the advent of the first molecularly targeted therapies at the beginning of this millenary has opened this changing time, but this was only the wind before the tempest. Now, oncology faces the arrival of multiple new comers, including not only novel targeted therapies, but also active immune therapies, antibody-drug conjugates, and probably soon adoptive cell transfer using chimeric artificial T-cell receptors (CARs). Paradoxically, the need for rapid and efficient drug development has never been as significant, as illustrated by the advent of Food and Drug Administration (FDA) breakthrough designations based on phase I trials results, as well as FDA conditional approvals based on phase I and II data. As a result, phase I investigators are challenged by the need to conciliate efficiency with the implementation of novel drug development models for novel drugs that present unusual levels of activity and ad hoc specificities. Which recent trials have presented the most successful drug development model? Examples include the ALK/ROS inhibitors crizotinib (PF-02341066, Pfizer) and ceritinib (LDK378, Novartis), as well as the programmed death 1/programmed death-ligand 1 (PD-1/PD-L1) inhibitors nivolumab (BMS936558, Bristol-Myers Squibb), pembrolizumab (MK-3475, Merck) and MPDL3280A (Genentech-Roche). For the first two trials, thorough patient selection has been the key of success [1, 2]. In both studies, several thousands of patients were screened in order to eventually select 82 and 130 of them that fulfilled the eligibility criteria of presenting an ALK-rearranged disease (at least for non-small cell lung cancer patients). Interestingly, both phase I trials were amended to allow recruitment of hundreds of additional patients (n = 550 for crizotinib [3, 4] and n = 246 [3] for ceritinib). The impressive activity results observed in these trials have triggered an accelerated approval by the FDA in August 2011 for crizotinib and in April 2014 for ceritinib, <5 years after the first patient was enrolled in the corresponding phase I study. This starkly contrasts with previous drug development schedules, which presented a time-lapse of ∼10 years between first-in-human studies and drug approval at the beginning of the century. Such successful stories strongly encourage the development of large-scale phase I trials enriched in molecularly selected patients—whenever a robust selection biomarker is available, thus implying the screening of several hundreds of patients and international collaborations. However, recent events have shadowed this impressive success of ceritinib. Indeed, despite the undisputable activity of the drug, uncertainties regarding the optimal dose and prandial conditions of administration were still unsolved at the end of that study [5, 6]. Although the drug was approved at the dose of 750 mg o.d., reviewers were concerned with the drug’s gastrointestinal (GI) tolerability and recommended investigating lower doses of 450 mg and 600 mg ceritinib taken with a meal, as a significant positive food effect was demonstrated for this drug and as this may improve the GI tolerability [6, 7]. How could have these issues been better addressed in phase I trials? One potential solution is the multiplication of expansion cohorts at doses below the maximum tolerated dose (MTD). In such design, dose levels that are known to be safe and at which pharmacodynamic parameters show a biological activity of the drug, can be expanded in parallel of the continuation of the dose escalation. This design, in which the patient could for example be its own control for food effect assessment in an expansion cohort, combines the advantage of collecting more data on PK data (including inter and intrapatient variability), as well as treating more patients when the drug is obviously active. This may help improving the ability of phase I trials of targeted agents to predict doses that will eventually be registered—as this ability has been reported to be poorer for targeted agents than for phase I trials evaluating conventional cytotoxic chemotherapies [8]. For immunotherapies, there is no doubt that the revolution is ongoing and that many drugs will flow through drug development processes to approval, considering the extremely promising levels of activity. One of the most recent examples of such successful drug development is the PD-L1 inhibitor MPDL3280A. Based on the results of the phase I trial only, this drug was granted a breakthrough therapy designation for urothelial bladder cancer. A large confirmatory phase II study has been mandated for confirmation (NCT02108652). However, the drug development model of immune therapies is often complex, as these agents challenge all previously established paradigms of cancer drug development even more deeply than targeted agents. As an illustration, after several phase I trials in monotherapy and combination, as well as phase II and phase III trials evaluating different schedules, it is still unclear what is the optimal dose of nivolumab to be administered [9–11]. Optimal patient selection is also a challenge, as there is no consensus on which technique, which antibody or which threshold should be used to assess PD-L1 positivity status. This obviously contrasts with a patient selection based on a specific mutation, as DNA sequencing is a robust and almost universal technique. If no consensus can be found soon on assessing PD-L1 positivity, this may become a worrying issue for a future large-scale administration of these drugs, as each compound currently has its own companion selection biomarker. More importantly, most phase I investigators ed ito ria ls editorials Annals of Oncology 26: 7–9, 2015 doi:10.1093/annonc/mdu513 Published online 30 October 2014