Tumor Genetic Testing for Patient Selection in Phase I Clinical Trials: The Case of PI3K Inhibitors

Much progress has been made in the development of targeted therapies in cancer during the last few years. As a result of whole-scale sequencing of cancer genomes, a comprehensive view of somatic mutational landscapes in cancer and potential therapeutic targets has emerged. Important findings of genome-scale sequencing include the discovery of mutations in BRAF and phosphatidylinositol 3-kinase (PI3K), just to cite some notable examples. The identification of driver genetic mutations has opened the gates to the development of targeted therapies that are specifically aimed at these mutations. This approach seems to be successful in some circumstances. For instance, an inhibitor that blocks a specific mutant of the serine/threonine kinase B-RAF (V600E-B-RAF) has recently demonstrated dramatic efficacy in treating metastatic melanoma Likewise, an inhibitor of the anaplastic lymphoma kinase (ALK) proved highly effective against a subset of non–small-cell lung cancers that had been prescreened for an oncogenic EML4-ALK translocation. Also, a Janus kinase 2 (JAK2) inhibitor has proved beneficial against myelofibrosis harboring an activating JAK2 mutation. In B cell hematological malignancies, which selectively express the class I PI3K isoform, a PI3K -specific inhibitor has shown remarkable activity in refractory tumors. In basal-cell carcinomas and in medulloblastomas harboring mutations in hedgehog pathways genes, hedgehog pathway inhibitors have also shown activity in patients with advanced disease. These results, together with prior identification of sensitivity to epidermal growth factor receptor (EGFR) agents in tumors with EGFR mutations, have resulted in a growing need to determine the presence of actionable mutations in tumors. This is not an easy task. To begin with, since many of these mutations are uncommon, there is a need to screen a large number of tumors in order to identify the presence of therapeutically relevant mutations. For example, even a common cancer such as non–small-cell lung cancer requires testing a large number of cases to identify those with EGFR mutations (10% of all cases), EML4-ALK rearrangements (4%), or MET, HER2, and B-RAF abnormalities (1% to 2% each). In the clinical trial of crizotinib for EML4-ALK-translocated lung cancer, 1,500 patients were prescreened to identify only 82 cases with the translocation The challenge of screening for a large number of mutations that are present only in a small number of patients may be eased by high-throughput genotyping platforms that perform well with nucleic acid derived from formalin fixation and paraffin-embedding tissue and can assess over a hundred mutations (and growing) with potential therapeutic implications. These early successes and the need to select patients with given mutations in order to observe clinical benefit is impacting on the way patients are being selected for early clinical trials with targeted agents. In a target dependent fashion we need to decide how early and in which situations should tumor gene mutational analysis be incorporated in the process. In this regard, the study by Janku et al in the article that accompanies this editorial illustrates some of the issues that the field is facing in targeting the PI3K gene, a frequently mutated gene in a variety of tumors. In the present study, the presence of PIK3CA mutations was prospectively analyzed in 140 patients referred to the phase I program at MD Anderson. In patients with breast, cervical, endometrial and ovarian cancer, they detected 25 tumors (17%) with PIK3CA mutations in the helical and kinase domain. In 23 patients harboring PIK3CA mutations who were then treated with inhibitors of the PI3K pathway, Response Evaluation Criteria in Solid Tumors Group–defined responses were observed in seven (30%). In comparison, only seven responses were observed in 70 patients with tumors without PIK3CA mutations (10% response rate). The authors conclude that the presence of PIK3CA mutations is associated with increased sensitivity to regimens that include PI3K/mammalian target of rapamycin (mTOR)/AKT inhibitors. Janku et al should be commended for their genotyping efforts and commitment to enrich their clinical trial population with patients with PI3K mutations. However, the interpretation of their results is challenging for a number of reasons. Laboratory studies to date have not conclusively shown that tumor cells harboring PI3K mutations have increased sensitivity to PI3K/mTOR and pan-PI3K inhibitors. We cannot rule out at this time what some of the more isotype specific compounds may display differential sensitivity, depending on which PI3K isoform is involved. The lack of a strong correlation between PI3K mutations and sensitivity to mTOR and pan-PI3K inhibitors contrasts with the tight correlation between the high sensitivity to B-RAF, ALK and EGFR inhibitors in cell lines harboring mutations in these genes. In addition to the laboratory findings not supporting a strong correlation, the lack of a stronger correlation between PI3K mutations and response to therapy could also be a result of additional reasons. As acknowledged by Janku et al, some of the JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L S