DeVita. Cancer

190

Cancer Therapeutics

study in either investigation. It is also important to consider the contribution of tumor heterogeneity, the potential for suboptimal modulation of targeted pathways, and upregulation of compen- satory pathways to the modest or minimal single-agent activity demonstrated for molecularly targeted drugs in these trials. 10 The current generation of therapeutic precision oncology trials 20 and recent reports of institutional experience with NGS-selected treatment have attempted to address several of these critical consid- erations. Among the most important is the strength of the treatment algorithm used for drug selection. 83 The selection algorithm needs to define how mutations are prioritized for defined therapies in the case of single or multiple targetable alterations (Table 15.1), and specific prioritization rules must be developed prior to trial initia- tion. Contemporary investigational studies also generally exclude patient entry based on known targets for FDA-approved drugs or indications and prohibit “matching” drugs and targets where a solid base of clinical trial data already indicates that the use of the agent in a specific disease context is unlikely to be effective (e.g., sin- gle-agent BRAF inhibitors for patients with BRAF v600E -mutant ad- vanced colon cancer). In a recent example of the use of molecular profiling for the selection of a targeted drug, in a single-institution series of patients with metastatic adenocarcinoma of the lung where tumors underwent NGS with a multigene panel, 69 of the 478 patients whose mutational profiles were not associated with standard-of-care treatment options received matched therapy; clini- cal benefit was demonstrated in 52% of these individuals. 84 One of the major changes accompanying the implementation of molecular characterization assays to guide the choice of oncologic therapy has been the application of novel clinical trial designs to op- timize this approach. 85,86 The overarching concept is to coordinate clinical studies so that more than a single treatment or disease can be evaluated across a shared operational infrastructure, often focused on mechanism-based therapeutic methodologies. Such “Master Pro- tocols” offer the potential to examine several approaches in parallel using a collective diagnostic or therapeutic platform that, although complex to organize and operate, offers the potential to accelerate the entire clinical trials process. 87 As shown in Figure 15.3, one type of master protocol design, designated “basket” trials, generally fo- cuses on a variety of malignant histologies and uses a single drug that targets a single mutation. Such studies provide access to an investi- gational drug for a broad range of cancers that may carry a specific mutation at low frequency. Although basket trials generally examine one drug/mutation pair at a time, using a broad screening strategy may demonstrate clinical activity in patients carrying specific muta- tions, leading to future clinical trials or possible FDA approval for the combination of drug and biomarker. Several recent, successful, histology-agnostic basket trials have demonstrated substantive clinical benefit for pan-HER2 inhibitors in patients carrying HER2 mutations in a wide range of solid tumors beyond breast cancer, for larotrectinib in patients with various tumors expressing TRK fusion genes, 88 and for patients with mismatch repair–deficient tumors treated with a PD-1 inhibitor. 32 These studies have clearly demonstrated that the molecu- lar background, as well as the histologic context, of solid tumors needs to be considered in the overall paradigm of cancer drug development. A different type of master protocol can be described as an “um- brella” trial (see Fig. 15.3). Umbrella studies can take several differ- ent forms. Trials may be disease based, such as the Lung-MAP study sponsored by the National Cancer Institute (NCI), supported by the Foundation for the National Institutes of Health (NIH), and directed by the Southwest Oncology Group (SWOG) clinical trials group, in which patients with the same tumor type (originally advanced squa- mous cell lung cancer in this case) are treated with multiple drugs targeting multiple mutations. Such studies can be randomized or nonrandomized and frequently use a rules-based treatment assign- ment. The trial is managed as series of substudies with an overarching administrative structure for biomarker assessment and data acquisi- tion; this allows for the addition of new substudies as accrual goals (or futility boundaries) are reached. Disease-focused umbrella trials also frequently use shared control arms and adaptive statistical end points. With the initial research use of NGS, and continuing to the present, investigators have identified genotypes in single patients that appeared to explain an “exceptional” response or provided mo- lecular insight into an unusual disease that suggested a novel treat- ment program (Fig. 15.3). 77 These and other such gratifying results stimulated early attempts to define therapy based on a range of mo- lecular tumor characteristics beyond those applied in routine clini- cal practice. 78,79 Although molecular aberrations could be detected in a substantial percentage of patients enrolled on these trials, the degree to which the mutations detected in their tumors could be considered to underlie the growth characteristics of their malignan- cies, as well as the strength of the data used to “match” treatment to tumor, was limited. The nonrandomized nature of these initial clinical studies, as well as the definitions used to define objective clinical responses, also limited interpretation of their results. 9 A second phase of precision oncology trials (e.g., the BATTLE and SHIVA studies), some of which have been reported, 80,81 has provided important insights clarifying the feasibility, limitations, and requirements for optimizing the principles of precision med- icine in clinical oncology. 10,82 The BATTLE study demonstrated the feasibility of mandating tumor biopsies for molecular charac- terization as an entry criterion for patients with advanced-stage non–small-cell lung cancer as well as the potential to apply adap- tive randomization strategies in the era of targeted therapy. The SHIVA trial provided important lessons that have guided the con- duct of subsequent, histology “agnostic” precision oncology trials; this trial was a prospective, randomized study in which no differ- ence in progression-free survival was demonstrated between the experimental arm (which used a therapy targeting one of three molecular pathways) and physician’s choice of therapy for patients with advanced solid tumors. However, for both of these trials, the level of evidence used to assign patients to a limited number of spe- cific pathway-driven treatments was based on potentially action- able biomarkers (or mutational variants of unclear significance) rather than on strong evidence demonstrating that the specific mutation(s) of interest (or other molecular characteristics) that were detected could drive tumor cell proliferation. Furthermore, only a limited range of drugs was available for the targets under Copyright © 2018 Wolters Kluwer Health, Inc. Unauthorized reprodu tion of the article is prohibited. In general, however, the successful transition from a molecular assay to a predictive biomarker remains difficult. Problems that have slowed the field, but that have been appreciated to a greater degree recently, include the requirement for standard operating procedures for both sample acquisition and assay validation, the need for better statistical understanding of the sources of variability in the assay, the trial design features required to qualify the assay in a clinical setting, and the necessity for a deep understanding of how the test will be used in the appropriate clinical context. 70 If these issues are carefully considered at the earliest stages of assay development, the predictive tests needed to guide novel treatment programs and advance the field of precision oncology, including immunotherapy, can be de- veloped more effectively, and at an accelerated pace, facilitating the successful application of targeted therapeutics for cancer. 71 Application of the principles of precision medicine to oncology has progressed through several interrelated stages over the past two de- cades. Beginning with the approval of trastuzumab for patients with HER2 -expressing breast cancers (defined by immunohistochemis- try or fluorescence in situ hybridization) 72 and the demonstration of the efficacy of imatinib for patients with chronic myelogenous leukemia expressing the BCR-ABL oncogene, 3 determination of therapeutic choice over this time frame by measurement of the expression of individual gene variants has fostered dramatic changes in the treatment of patients with adenocarcinoma of the lung carry- ing EGFR, ALK, and ROS1 mutations 73–75 and patients with mela- noma expressing mutant V600 BRAF species. 76 PRECISION ONCOLOGY CLINICAL TRIALS ANDTRIAL DESIGNS