Most of the drugs currently in development for personalized oncology fall into two classes, either molecularly targeted kinase inhibitors or ligands for cell surface receptors (see Figure 2
). Typically, kinase inhibitors are small molecules while cell surface receptors are mainly targeted by monoclonal antibodies. We decided to not discuss a third class, therapeutic cancer vaccines, as many of the big pharmaceutical companies (with some notable exceptions) are still in wait and see mode. Despite the approval of sipuleucel-T (Provenge) and all the excitement it has provided for the field, we need to see more real successes before significant resources will be invested in this approach.
Selective look at ‘new mode of action’ (MoA) currently in development.
4.1.1. Molecularly Targeted Kinases Inhibitors
With more than 500 distinct kinases identified and over 60 kinase inhibitors in clinical trials or FDA approved [5
], this important class of proteins is a rich area for therapeutic drug development. It is beyond the scope of this paper to review all kinase targets that are currently being exploited for drug discovery but there are some distinct trends that can be distilled.
There are three distinct classes among kinases that can be identified. First, there are kinase targets that escape normal regulatory mechanisms following genetic mutation or translocation. The constitutive activity of this class of kinase target, also called oncogene addiction, makes them essential for survival and/or proliferation of the cancer cell and thus susceptible to the appropriate kinase inhibitor. One of the most notable successes in this area was the discovery of the V600E mutation, located in the activation loop of BRAF [17
]. The targeted inhibition of this particular mutation has led to the discovery and subsequent approval of vemurafenib in melanoma. For a second class of kinase target, inhibition of the kinase results in a synthetic lethal phenotype when paired with another non-lethal mutation in the particular pathology of the tumor cell. Although rarely mutated in cancer, these kinases are preferentially required for the survival and/or proliferation of cancer cells and may be located in key signaling pathways downstream of transforming oncogenes. Examples include mitogen-activated protein kinase kinase 1 (MEK1) and 2 (MEK2), which are located in the critical mitogen-activated protein kinase (MAPK) pathway, mammalian target of rapamycin(mTOR), which is located in the phosphatidylinositol 3-kinases (PI3K) protein kinase B (Akt) signaling system, and the ribosomal S6 kinase (RSK) [18
]. While there are no synthetically lethal kinase inhibitors in the clinic, this concept of synthetic lethality has thus far been exploited with the anticipated utility of inhibitors of poly-ADP(ribose) polymerase (PARP) in BRCA1/2 deficient breast cancers [19
] A third class of kinase targets are expressed in the tumor or in surrounding tissues and are required for different stages of tumor formation and maintenance in the human host. Notable examples are the vascular endothelial growth factor (VEGFR) and fibroblast growth factor receptor (FGFR) kinases [21
], which are important in developing and sustaining tumor blood supply, as well as the pyruvate kinase which is required for the tumorigenic switch to aerobic glycolysis.
With the recent approvals of vemurafenib (targeting the V600E mutation in the serine/threonine-protein kinase B-Raf (BRAF)) and crizotinib (targeting the anaplastic lymphoma kinase (ALK) and echinoderm microtubule-associated protein-like 4 (EML4), ALK–EML4 fusion protein), both of which fall into the first category of kinase inhibitors, this class became the poster child of personalized medicine in cancer and we will hopefully find and define similar such targets (e.g., inhibitors of the V617F mutation in Janus kinase 2 (JAK-2) or of the D816V mutation in the proto-oncogene c-Kit (cKIT) are currently under clinical investigation) [8
]. As these kinds of mutations are readily identified by DNA sequencing they lend themselves to easier engineering of mechanistically relevant cellular assays that can be used during the course of inhibitor discovery and optimization. One problem that is inherent to this class of inhibitors is the duration of action as escape mutations can develop very readily [22
]. Even on this front we make progress with the recent approval of nilotinib that targets mutations that develop after imatinib therapy [24
] (and another such molecule, ponatinib, is in Phase II clinical trials [25
]). It is to be expected that there will be a lot of emphasis by drug developers on this class of molecules.
Kinase targets of the second and third classes are not directly transforming but instead are required for the survival, proliferation and/or tumor genesis of cancer cells. As such, they can be highly context dependent and much more difficult to investigate in preclinical experiments. Especially the mechanistic basis for selective cytotoxicity of many inhibitors in the second class is yet poorly understood. Typically, kinases that can be targeted through a synthetic-lethal interaction specific to cancer cells must be investigated and validated by a largely empirical process. Thus we may only be able to “personalize” these kinds of targets in large post approval studies which hopefully will allow us to identify genetic patient phenotypes that would reap the most benefit. Likewise, kinase targets required for tumor formation and maintenance are usually evaluated in animal tumor models that are suboptimal reflections of the disease evolution in humans, so better tumor models are needed to increase the success rate of these targets. Thus we can expect a much slower pace of development in these kinds of target classes.
4.1.2. Ligands for Cell Surface Receptors
Cell surface receptors provide an almost equally rich class of targets. Unlike kinases, which reside inside the cell, they are expressed on the cell surface where they are activated by other protein ligands, thus setting a signaling cascade in motion. Since protein–protein interactions are notoriously difficult to disrupt by small molecules the majority of the cell surface receptors are being targeted by monoclonal antibodies.
Also, in this class we can broadly distinguish between three classes of targets. First, there are growth factor receptors which fuel the cancer cells’ growth and proliferation. The most famous example and probably the first example of a targeted therapy for which there was a companion diagnostic available is trastuzumab (Herceptin), a monoclonal antibody which binds to the human epidermal growth factor 2 receptor (HER2) [26
]. The HER2
gene is amplified and overexpressed in 15–25% of breast cancers, and HER2-positive carcinomas are typically associated with high tumor grade and invasiveness. However, despite having made a large impact on treating HER-2 positive patients, the majority of cancers that initially respond to Herceptin begin to progress again within one year [27
]. Other examples in this class include epidermal growth factor receptor (EGFR), insulin-like growth factor receptor 1 (IGF-1R), hepatocyte growth factor receptor (HGFR), platelet-derived growth factor receptor (PDGFR), etc.
Cell surface targets in the second class are not necessarily directly expressed by the tumor cell but rather by the host environment feeding the tumor and are required for tumor maintenance and metastasis. The most famous example for this class is bevacizumab (Avastin) a monoclonal antibody that is approved for the treatment of non-small cell lung cancer, metastatic breast, colorectal, and kidney cancer. Bevacizumab binds to vascular endothelial growth factor (VEGF) and prevents it from interacting with receptors on endothelial cells, blocking a step that is necessary for the initiation of new blood vessel growth (angiogenesis) [28
]. Despite its commercial success bevacizumab has also serious side effects (e.g., serious bleeding events) and its approval in breast cancer has been recently revoked by the FDA (despite EMA’s decision to keep this indication on bevacizumab’s label). Other examples in this class include placental growth factor (PGF), fibroblast growth factor receptor (FGFR), and inhibitors of many different kinds of integrin receptors (αv,2,4
). The third class can be broadly summarized as immunomodulatory agents which do not directly interact with the tumor but stimulate the immune system to attack malignant cells. The most prominent example here is ipilimumab, a monoclonal antibody against cytotoxic T-lymphocyte protein 4, CTLA4, and which was recently approved for the treatment of late stage melanoma [29
]. CTLA4 is an inhibitory protein expressed on T-cells and blocking it is part of a strategy to inhibit immune system tolerance to tumors. As this strategy can be of general use for many tumor types, ipilimumab is currently tested in other cancer indications. Other targets in this class include CD19/20, IL-2R, CD40, CD56, etc.
It is more difficult to personalize therapies targeting cell surface receptors as most of these targets are expressed on normal cells as well. And since all of the pathways that are stimulated by growth factors are redundant/overlapping to a certain degree, interfering with one may only yield a modest effect. This could be the reason why we see resistance against trastuzumab in HER-2 positive tumors occurring so rapidly as back-up pathways may take over and continue to provide the positive signal for tumor growth. Thus it may be necessary to target more than one growth factor receptor simultaneously in order to choke off tumor growth signals. With the advent of protein engineering there is now an exciting array of multifunctional fusion proteins that can target two or three of these receptors simultaneously and some of them are already being explored in clinical trials. Hopefully, in the not too far future we will know whether this strategy will enhance the therapeutic benefit.
By definition, agents that target the host tumor environment and angiogenesis cannot be personalized as these targets are not transformed and thus their expression level, genetic make-up, etc. should be the same for every patient. Nevertheless, each individual cancer may spark a somewhat different microenvironment and we need to understand how to best characterize this environment and how it responds to these agents. Thus a personalization strategy then is more directed towards how we use the agents we already have in hand (and there are a few more in clinical trials) based on some yet to be defined biomarkers, and how we can incorporate them in an individualized treatment regimen to maximize their therapeutic effects.
With the immunotherapy approaches, the key challenge we face is that despite signs of them working, the therapeutic effect thus far has been fairly modest. We need to better define how they are working, how we can amplify the efficacy, and which patients will benefit the most from them. In this regard, retrospective identification of predictive biomarkers is likely to be an important strategy well after approval of the drug. Once identified, the predictive power of the biomarker can then be explored in a hypothesis driven prospective clinical trial.