Special Perspectives

Targeting Tyrosine Kinases in Cancer: The Second Wave

See allHide authors and affiliations

Science  26 May 2006:
Vol. 312, Issue 5777, pp. 1175-1178
DOI: 10.1126/science.1125951


One of the most exciting developments in cancer research in recent years has been the clinical validation of molecularly targeted drugs that inhibit the action of pathogenic tyrosine kinases. Treatment of appropriately selected patients with these drugs can alter the natural history of their disease and improve survival. The clinical validation of these “first-generation” tyrosine kinase inhibitors has been the prelude to a second wave of advances in molecular targeting that is expected to further change the way we classify and treat cancer. Efforts are now being directed at identifying the tumor subtypes and patients who will benefit the most from these drugs. In addition, new compounds that circumvent acquired resistance to the first-generation tyrosine kinase inhibitors are being tested in patients with refractory disease. Agents directed against new molecular targets are also being explored.

The rationale for targeting protein tyrosine kinases in human cancer is compelling. These enzymes regulate multiple cellular processes that contribute to tumor development and progression, including cell growth, differentiation, migration, and apoptosis. In model systems, perturbation of tyrosine kinase signaling can result in malignant transformation (1). The human genome encodes 90 proteins with tyrosine kinase domains (2), and many human tumors display aberrant activation of tyrosine kinases caused by genetic alterations. For tumors whose growth is driven by these activated kinases, targeted drugs can potentially inhibit or reverse malignant progression.

Clinical studies conducted over the past decade have established that tyrosine kinase inhibitors are safe and therapeutically active in selected populations of cancer patients, and several of these drugs are now part of the standard treatment regimen for specific tumor types. These include imatinib (directed against BCR-ABL and other kinases), trastuzumab (directed against the HER2/ErbB2 receptor), and cetuximab, erlotinib, and gefitinib [directed against the epidermal growth factor receptor (EGFR)]. Here, I provide a brief history of the first generation of tyrosine kinase inhibitors and discuss how the lessons learned from initial clinical experience is paving the way for the development of new and more effective drugs.

Imatinib (STI-571, Gleevec)

The discovery of imatinib is rooted in cytogenetic research performed more than 30 years ago on leukemic cells from patients with chronic myelogenous leukemia (CML). These cells display a characteristic reciprocal translocation between chromosomes 9 and 22 that generates the so-called “Philadelphia (Ph) chromosome.” At the molecular level, this translocation juxtaposes the coding sequences of the bcr gene and c-abl genes. The c-abl gene encodes a nonreceptor tyrosine kinase, and the genetic fusion creates an oncoprotein, BCR-ABL, with constitutively active tyrosine kinase activity. BCR-ABL powers the rapid clonal expansion of pluripotent hematopoietic stem cells that underlies CML. The uncontrolled kinase activity of BCR-ABL is sufficient to cause leukemia, making it an ideal therapeutic target.

Identification of the causative genetic lesion in CML stimulated a search for inhibitors of the ABL kinase. Screening of compound libraries for molecules that had tyrosine kinase inhibitory activity resulted in the identification of 2-phenylamino pyrimidines as promising agents. From this initial lead, imatinib (STI-571) was synthesized and was found to potently inhibit the proliferation of BCR-ABL–expressing cells and their ability to form tumors in mice (3). In the initial trial in patients with CML, imatinib had minimal side effects, and almost all treated patients showed complete hematologic responses, defined as normalization of hematological parameters in the peripheral blood (4). Cytogenetic responses, defined as a major reduction in Ph-chromosome–positive cells in the bone marrow, occurred in half of the patients, and several had complete cytogenetic remissions. Responses were also seen in patients with Ph-chromosome–positive acute lymphoblastic leukemia (ALL) and patients with CML in blast crisis. Several later studies confirmed the results of these trials, culminating in the demonstration that imatinib is superior to interferon-α–based treatments in terms of cytogenetic response and likelihood of progression to accelerated-phase or blast crisis CML (5).

Imatinib is a “promiscuous” tyrosine kinase inhibitor in that it blocks the activity of additional tyrosine kinases, including the c-Kit receptor and the platelet-derived growth factor receptor (PDGFR). A subset of gastrointestinal stromal tumors (GISTs) display mutations in the c-KIT gene and express permanently activated forms of the c-Kit receptor. Patients with such tumors, who are largely unresponsive to conventional chemotherapy, show an excellent response to imatinib (6). Interestingly, almost one-third of GISTs lacking mutations in c-Kit have intragenic mutations in the PDGFRA gene, resulting in constitutively active PDGFRα, a finding that potentially explains the clinical responses to imatinib in GISTs with wild-type c-Kit. Imatinib also exhibits robust clinical activity in several other cancers associated with PDGFR alterations (7). These include chronic myelomonocytic leukemia (CMML), which is characterized by the constitutively active TEL-PDGFRβ fusion tyrosine kinase; hypereosinophilic syndrome, which is characterized by the FIP1L1-PDGFRα fusion protein; and dermatofibrosarcoma protuberans, which is characterized by a t(17,22) chromosomal translocation leading to constitutive production of PDGF ligand and subsequent PDGFR activation. Together, these results suggest that activating mutations in genes encoding the molecular targets of imatinib are reliable biomarkers of “kinase dependence” and thus may predict which patients are most likely to benefit from the drug.

Trastuzumab (Herceptin)

HER2 is a member of the HER (erbB) family of transmembrane receptor tyrosine kinases, which also includes EGFR (HER1, erbB1), HER3 (erbB3), and HER4 (erbB4). Receptor-specific ligands bind to the ectodomains of EGFR, HER3, and HER4, resulting in the formation of homodimeric and heterodimeric kinase-active complexes to which HER2 is recruited as a preferred partner (8). Interestingly, HER2 does not directly interact with receptor ligands but can potently enhance signaling by HER2-containing heterodimers and/or increase the binding affinity of receptor ligands to EGFR and HER3/4. Overexpression of HER2 causes malignant transformation of mammary epithelial cells. Approximately 25% of invasive primary breast cancers exhibit HER2 gene amplification, and this molecular feature correlates with reduced patient survival.

Trastuzumab is a humanized monoclonal antibody that binds to the extracellular region of HER2 and inhibits the growth of HER2-overexpressing cells (9). Its antitumor activity has been ascribed to several distinct mechanisms, including down-regulation of HER2 from the cell surface membrane, blockade of metalloprotease-induced proteolytic cleavage of HER2, antibody-dependent cell-mediated cytotoxicity (ADCC), and down-regulation of angiogenic factors. In initial clinical studies in the 1990s, trastuzumab was shown to induce tumor regressions in patients with advanced metastatic breast cancer for whom all available lines of therapy had failed. In subsequent studies, trastuzumab was found to offer significant clinical benefit for patients with HER2-positive metastatic breast cancer (10) and to improve survival when combined with chemotherapy (11). The survival benefits seen with combination therapy were particularly impressive and suggested that the impact of trastuzumab on the course of the disease might be even stronger with earlier treatment. This has now been confirmed in five large studies that together have included more than 14,000 women. The results show that the postoperative administration of trastuzumab to women with HER2-overexpressing breast cancer, given either in combination with chemotherapy or sequentially after chemotherapy has been completed, reduces local and distant (metastasic) recurrences by about one-half (12, 13) and, as expected, improves patient survival by over one-third in those trials with sufficient follow-up (13).

It is worth emphasizing that if these pivotal tratuzumab trials had been conducted with breast cancer patients who had not been preselected according to their HER2 status, the therapeutic activity of the drug would likely have been missed, thus threatening its clinical development and approval. This is a recurring scenario in the clinical development of tyrosine kinase inhibitors.

Anti-EGFR Therapies

As noted above, the EGFR has a high degree of homology with HER2, and there are strong similarities in the development of drugs targeting the two receptors. Human tumors of epithelial origin express high levels of EGFR. For this and other reasons, this receptor tyrosine kinase was first proposed as a target for cancer therapy more than 20 years ago [summarized in (14)]. Intensive research performed over the ensuing years has resulted in the successful development of two distinct classes of anti-EGFR drugs that have recently received regulatory approval for the treatment of cancer. These are monoclonal antibodies directed against the extracellular domain of the receptor (anti-EGFR MAbs) and small-molecule inhibitors of the receptor's tyrosine kinase activity (TKIs).

The antibody-based drugs compete with ligand for binding to the extracellular domain of EGFR and, as with many other antireceptor antibodies, they induce receptor internalization. The small-molecule inhibitors, on the other hand, act intracellularly by competing with ATP for binding to the tyrosine kinase domain of EGFR, thereby abrogating the receptor's enzymatic activity. The small molecules also block the catalytic activity of EGFR mutants lacking the extracellular domain, and thus may prevent ligand-independent activation of EGF receptor kinase activity as well.

At the level of downstream receptor-dependent signaling pathways, anti-EGFR MAbs and small-molecule TKIs have many similar effects. Both strategies result in an efficient blockade of the major EGFR signal transduction pathways, including the mitogen-activated protein kinase (MAPK) and PI3K/Akt pathways and the Jak/Stat pathway. The mechanisms of action and the antitumor effects of the two drug classes do not completely overlap, however. For example, in addition to promoting receptor internalization, the anti-EGFR MAbs elicit ADCC, which renders tumor cells more vulnerable to attack by the immune system. Conversely, the small-molecule TKIs, but not the antibody-based drugs, are active against more than one ErbB receptor type. The two classes of drugs have additive growth inhibitory effects on cancer cells in vitro (15), a finding that has set the stage for ongoing clinical studies combining the two therapies.

Anti-EGFR MAbs have shown clinical activity in a variety of epithelial tumors and one of them, cetuximab, has been approved for the treatment of advanced colorectal cancer and, more recently, head-and-neck tumors. In the setting of colorectal cancer, cetuximab is active in patients whose illness has become refractory to chemotherapy (16). However, as with trastuzumab, cetuximab may be most valuable when used earlier in combination with chemotherapy, either as first-line therapy for metastatic disease or in the adjuvant setting. For example, up to 80% of patients respond to the combination of cetuximab and chemotherapy when given as the initial treatment in patients with metastatic colorectal carcinoma. In the setting of advanced head-and-neck tumors, a disease with high morbidity and for which there is a desperate need for new drugs, anti-EGFR MAbs are active in refractory patients previously treated with several lines of chemotherapy and radiation therapy (17). In a recent study, treatment of locoregionally advanced head-and-neck cancer with concomitant radiotherapy plus cetuximab was shown to improve locoregional control and reduce mortality without increasing the common toxic effects associated with radiotherapy of this cancer type (18). Head-and-neck cancer has been particularly resistant to systemic therapies, and it is noteworthy that cetuximab is the first new treatment approved for this cancer in 30 years.

EGFR TKIs are largely inactive in colorectal cancer; however, two of these drugs, gefitinib and erlotinib, have been approved for the treatment of non–small cell lung cancer (NSCLC). EGFR TKIs also improve survival in patients with advanced pancreatic cancer (19), and responses have been observed in head-and-neck cancer (20) and in glioblastoma (21).

I will discuss in some detail the development of the small-molecule TKIs in NSCLC, as this example best illustrates the progress being made in the field of molecularly targeted therapies. Initial studies of these drugs in patients with refractory disease revealed moderate but real response rates. As was the case with trastuzumab and cetuximab (see above), these agents were then combined with conventional chemotherapy in a series of large (phase III) studies [for review, see (20)]. Disappointingly, no additional survival benefit was evident when erlotinib or gefitinib was added to conventional chemotherapy regimens. The anti-EGFR TKIs do show varying degrees of antitumor activity when administered as single agents, however. In a phase III study comparing erlotinib with the best supportive care in NSCLC patients for whom first- or second-line chemotherapy had failed (22), those who received TKI showed a statistically significant improvement in response rate and overall survival (6.7 months versus 4.7 months). A similar response rate was observed in a phase III study of gefitinib in NSCLC, although in this case the results did not reach statistical significance (23). Although both studies demonstrated that TKIs have a modest level of activity in unselected patient populations, a small subset of patients in the studies appeared to benefit more substantially from the therapy. Patients with adenocarcinoma (particularly bronchioalveolar carcinoma), those who had never smoked, women, and Japanese patients had a higher response rate, which suggests that their particular tumors might be EGFR-dependent. These clinical findings were followed by the discovery that a subset of NSCLCs harbor somatic mutations in the EGFR gene, and the presence of these mutations correlates with a positive clinical response to gefitinib and erlotinib (24, 25). EGFR mutations affecting four exons of the gene have been described: substitutions for G719 in the nucleotide binding loop encoded by exon 18, in-frame deletions within exon 19, in-frame insertions within exon 20, and substitution for L858 or L861 in the activation loop encoded by exon 21.

The clinical benefit observed with anti-EGFR TKIs is not restricted to patients with tumors harboring EGFR gene mutations. Other potential markers of sensitivity/resistance to EGFR TKIs include the presence of EGFR gene amplification, expression levels of ErbB3, and possibly mutations in the RAS and HER2 genes. Increased EGFR copy number and EGFR mutations are not mutually exclusive events: About 65% of patients with EGFR amplification also have EGFR mutations (26). On the other hand, the presence of K-RAS mutations, which are frequent in smokers, correlates with resistance to EGFR inhibitors (27). Mutations in the HER2 gene have been observed in lung adenocarcinomas (28), although their ability to serve as markers of TKI response remains to be explored. Intriguingly, these mutations are small in-frame insertions located in exon 20 in a position analogous to the insertion mutations found in EGFR (codons 774-781). Another potential molecular predictor of response to EGFR TK inhibitors is ErbB-3 protein, which is expressed at high levels in gefitinib-sensitive NSCLC cell lines and in patients responsive to gefitinib therapy (29).

Acquired Resistance and Second-Generation Tyrosine Kinase Inhibitors

Despite the early successes with the tyrosine kinase inhibitors discussed above, the majority of responding patients will eventually develop resistance to the drugs. Resistance can be caused by amplification of the oncogenic protein kinase gene or other mechanisms, but in a high fraction of cases, resistance can be traced to the selection of cancer cells with secondary mutations in the gene encoding the targeted kinase. The resistance mutations often affect amino acids within the kinase catalytic domain, and they prevent or weaken interaction of this domain with the drug. Resistance mutations have been observed in the kinase domains of BCR-ABL, Kit, and the PDGFR in the tumor cells of patients treated with imatinib (30). Likewise, the tumors of patients who initially responded to gefitinib or erlotinib have been found to acquire secondary EGFR mutations that render them resistant to these agents (31).

In light of this new challenge, agents active against new mutations that arise during therapy with first-generation tyrosine kinase inhibitors are being rapidly developed (Fig. 1). Dasatinib (BMS-354825), for example, is a dual SRC-ABL inhibitor that binds ABL with less stringent conformational requirements than imatinib (32), and it has been shown to be active in patients with imatinib-resistant CML (33). In GISTs that have acquired mutations conferring resistance to imatinib, SU11248 (Sutent), a multitargeted tyrosine kinase inhibitor that blocks vascular endothelial growth factor receptor (VEGFR), PDGFR, and KIT activation, has shown substantial activity (34). In the case of acquired resistance to drugs targeting EGFR, there seems to be an added complexity because some of the resistance mutations are primary. Patients who initially respond to gefitinib or erlotinib may acquire secondary EGFR mutations, specifically the T790M mutation, which resides within the protein's catalytic ATP-binding site (31, 35). However, T790M mutations can also arise in tumors of previously untreated patients; the same is true of an exon 20 insertion mutation that confers resistance (36). It is encouraging that tumor cells with acquired or primary mutations conferring resistance to gefinitib and erlotinib appear to be sensitive to a series of irreversible EGFR inhibitors, a group of small-molecule TKI that covalently cross-link the receptor (37). Clinical studies of these agents are ongoing in patients with NSCLC.

Fig. 1.

First- and second-generation tyrosine kinase inhibitors for cancer treatment. Over time, tumors that respond to the first-generation inhibitors often develop resistance to the drugs, in some cases because of the appearance of secondary mutations in the gene encoding the targeted tyrosine kinase. Second-generation inhibitors have been developed to address this problem. The first-and second-generation drugs can be administered sequentially or as a combination therapy.

The molecular mechanisms underlying acquired resistance to trastuzumab are less well defined, although loss of PTEN function and absence or loss of the extracellular, antibody-binding domain of the receptor have been implicated. Lapatinib (GW572016) is a small-molecule dual inhibitor of EGFR and HER2 (38) that has shown activity in patients with HER2-overexpressing advanced breast cancer whose disease has progressed after treatment with trastuzumab (39). It has also shown activity when administered as a first-line therapy and seems to be synergistic when combined with trastuzumab (40, 41). Large phase III studies are underway both in patients with advanced breast cancer and in the early postoperative setting.

Kinase Dependence in Other Tumor Types: Therapeutic Opportunities

Genome sequence analyses across different tumor types have identified additional tumor-specific activating mutations in tyrosine kinase genes that may drive tumor growth. These include BRAF mutations in melanoma (42) and mutations in the Flt3 receptor tyrosine kinase in one-third of acute myeloid leukemia (AML) and a smaller group of ALL (43). Somatic mutations in the PI3KCA gene, which encodes the p110α catalytic subunit of phosphoinositol-3-kinase (PI3K), have also been identified in a wide range of cancers (44). BRAF mutant tumors are exquisitely sensitive to small-molecule inhibitors of MAPK kinase (MEK) (45) and offer a rational therapeutic strategy for this genetically defined tumor subtype. The finding of clinical responses to MEK inhibitors in patients with advanced melanoma is highly encouraging (46). Activation of the insulin-like growth factor receptor 1 (IGF-1R) may be involved in abrogating the antitumor effect of EGFR TKIs and trastuzumab (47). Targeting of the IGF-1R is being explored in the clinic with small-molecule inhibitors as well as monoclonal antibodies.

Finally, mutations in the PTEN (phosphatase and tensin homolog deleted in chromosome 10) tumor suppressor gene are also of interest in the context of kinase dependence. PTEN is a phosphatase that selectively dephosphorylates the lipid phosphatidylinositol (3,4,5) trisphosphate. Loss of PTEN function occurs in a wide range of human cancers (48) and is associated with increased PI3K activity. Such tumors are likely dependent on this pathway, as PTEN-null cells are highly sensitive to mTOR and PI3K inhibitors in model systems (49). On the basis of these preclinical studies, PTEN mutational status is being used to select patients for enrollment into trials testing the clinical activity of mTOR inhibitors and, in the near future, PI3K inhibitors.

Riding the Second Wave of Tyrosine Inhibitors: Lessons from the Past

How the clinical development of tyrosine kinase inhibition in cancer will ultimately progress is not easy to predict, as the field will at times be driven by unanticipated findings along the way. This unpredictability is exemplified by the discovery of secondary mutations in tyrosine kinase genes in the tumors of patients who developed resistance to imatinib and erlotinib. Although acquired resistance to anticancer agents is an expected consequence of prolonged treatment, it was somewhat surprising that tumor cells with a large repertoire of growth-promoting mechanisms resort instead to reactivation of the same kinase that caused their malignant transformation.

That said, some of the lessons learned from initial clinical experience will surely influence how these drugs will be developed. The activity of imatinib in CML and GIST, trastuzumab in breast cancer, and anti-EGFR agents in NSCLC has validated the concept that certain tumors are “oncogene dependent” (50). This underscores the need to direct anticancer drugs to the subset of tumors whose growth depends on the oncogenic signals targeted by the molecular therapies. In other words, random testing of imatinib across tumor types without focusing on Bcr-Abl– and c-Kit–dependent cancers, or of trastuzumab in non-HER2 amplified tumors, would have been a clinical development strategy destined to fail. The second-generation inhibitors that are effective against tumors with acquired drug resistance illustrate the potential of mutation-specific therapies, but they also raise the question of whether secondary resistance might be avoided or delayed by the combined administration of tyrosine kinase inhibitors up front instead of a sequential approach. An answer to this question may come from ongoing clinical trials that are testing combinations of agents directed at a single tyrosine kinase.

The majority of advanced solid tumors are genetically complex and, with rare exceptions, it is unlikely that a given tumor will be entirely dependent on one abnormally activated kinase or signaling pathway for its malignant behavior. There is also a considerable level of compensatory “cross talk” between receptors within one signaling pathway, as well as cross talk between distinct signaling pathways regulating cell proliferation, trafficking, and survival. As with conventional chemotherapeutic agents, which are often most effective when administered as combination therapies, rationally developed combinations of molecularly targeted agents are likely to be more potent than single-agent therapies. In terms of combining molecularly targeted agents with conventional chemotherapy, the negative results of the anti-EGFR TKI studies in NSCLC have greatly limited the enthusiasm to embark on large studies in unselected patient populations.

From a practical standpoint, it will be mandatory to have tumor tissue available from patients participating in clinical trials in order to study the molecular features that correlate with sensitivity or resistance to these molecularly targeted agents. The availability of tissue and serum from all patients may allow retrospective identification of a molecular profile or surrogate marker characteristic of responding tumors, even when the demonstration of activity is limited to a small group of patients. In turn, this profile or marker could be used prospectively for patient enrollment into subsequent studies with selected patients. Ideally, such predictive biomarkers would encompass not only specific gene mutations but also gene expression signatures, which provide information about the activation status of several oncogenic pathways (51). At the time of tumor progression, one could also consider assessment of newly acquired mutations to select the next line of therapy. In early studies, incorporation of biomarkers of drug effects and drug sensitivity in tumors will also be coupled with noninvasive molecular imaging, which could provide early indications of clinical benefit (52).

Future research on molecularly targeted therapies will focus on the identification of new drugs and drug targets, improved selection of tumors sensitive to these drugs, and the rational design and optimization of combination therapies. The new wave of discoveries will help transform oncology from its current state of empirically based patient management to one in which treatment decisions are based on mechanistic approaches that successfully integrate molecular biology, pathology, imaging, and clinical medicine.

References and Notes

View Abstract

Navigate This Article