Targeting the function of the HER2 oncogene in human cancer therapeutics

Abstract

The year 2007 marks exactly two decades since HER2 was functionally implicated in the pathogenesis of human breast cancer (Slamon et al. 1987). This finding established the HER2 oncogene hypothesis for the development of some human cancers. An abundance of experimental evidence compiled over the past two decades now solidly supports the HER2 oncogene hypothesis. A direct consequence of this hypothesis was the promise that inhibitors of oncogenic HER2 would be highly effective treatments for HER2-driven cancers. This treatment hypothesis has led to the development and widespread use of anti-HER2 antibodies (trastuzumab) in clinical management resulting in significantly improved clinical anti-tumor efficacies that have transformed the clinical practice of oncology. In the shadows of this irrefutable clinical success, scientific studies have not yet been able to mechanistically validate that trastuzumab inhibits oncogenic HER2 function and it remains possible that the current clinical advances are a consequence of the oncogene hypothesis but not a translation of it. These looming scientific uncertainties suggest that the full promise of the treatment hypothesis may not yet have been realized. The coming decade will see a second generation of HER2 targeting agents brought into clinical testing and a renewed attempt to treat HER2-driven cancers through the inactivation HER2. Here I review the development of treatments that target HER2 in the context of the HER2 oncogene hypothesis, and where we stand with regards to the clinical translation of the HER2 oncogene hypothesis.

INTRODUCTION

HER2 belongs to the Human Epidermal Growth Factor Receptor (HER) family of tyrosine kinases conisting of EGFR (HER1, erbB1), HER2 (erbB2, HER2/neu), HER3 (erbB3), and HER4 (erbB4). The importance of HER2 in cancer was realized in the early 1980s when a mutationaly activated form of its rodent homologue neu was identified in a search for oncogenes in a carcinogen induced rat tumorigenesis model (Shih et al. 1981). Its human homologue HER2 was simultaneously cloned and found to be amplified in a breast cancer cell line (King et al. 1985). The relevance of HER2 to human cancer was established when it was discovered that approximately 25–30% of breast cancers have amplification and overexpression of HER2 and these cancers have worse biologic behavior and prognosis (Slamon et al. 1989). This finding established the HER2 oncogene hypothesis that overexpression of HER2 is etiologically linked with tumorigenesis in some human cancers. A substantial body of experimental evidence over the past two decades has come to solidly support this hypothesis. In numerous in vitro and transgenic models, HER2 overexpression by itself is potently transforming. In addition, analysis of human breast cancers has shown that amplification of the HER2 locus is an early event in human carcinogenesis and along with the experimental evidence confirming its potently transforming functions, this makes a highly compelling case implicating HER2 overexpression in the genesis of these human cancers. The signaling functions of HER2, the body of evidence confirming the transforming functions of HER2, the numerous proposed mechanisms mediating its transforming functions, and the data establishing the relevance of these findings to human cancer pathogenesis were discussed in depth previously (Moasser 2007). A direct consequence of the HER2 oncogene hypothesis of human cancer was that inhibitors of oncogenic HER2 would be highly effective treatment for HER2 driven cancers. Here I will review where we stand with regards to testing this treatment hypothesis and where we currently stand with regards to the therapeutic implications of the HER2 oncogene hypothesis.

TUMOR DEPENDENCE ON HER2

The tumorigenic potential of HER2 is solidly supported by experimental models (Moasser 2007). This by itself proposes HER2 as a possible target for anti-cancer drugs. However its suitability as a drug target is substantially strengthened by experiments demonstrating that HER2-driven tumors are dependent on HER2 function. This dependency, recently labelled oncogene addiction, identifies oncogenes that are high value targets for drug development (Hynes and Lane 2001).

HER2-dependency of HER2 amplified human cancers

Experimental models of HER2 overexpressing cancer cells using antisense, ribozyme, or siRNA methodologies consistently show that HER2 knockdown induces apoptosis in cell culture, or tumor regression in vivo, in the absence of HER2 expression, while tumor types that do not overexpress HER2 are not sensitive to HER2 knockdown (Colomer et al. 1994; Juhl et al. 1997; Roh et al. 2000; Faltus et al. 2004; Choudhury et al. 2004). Similar results are seen with kinase-dead HER2 and intracellular single chain anti-HER2 antibodies (Messerle et al. 1994; Beerli et al. 1994; Deshane et al. 1996).

HER2-dependency in experimental models

Engineered models of HER2-driven transformation using tetracycline inducible systems confirm that HER2 induced tumors require HER2 for continued tumorigenic growth and survival. This has been demonstrated in an HER2-transformed NIH3T3 tumor model wherein tumors regress upon withdrawal of the HER2 oncogene (Baasner et al. 1996; Schiffer et al. 2003). This has also been corroborated in a tet-inducible transgenic models. Tetracycline induced expression of activated HER2 in squamous epithelia of mice results in severe hyperplastic abnormalities of squamous epithelial tissues, which reverse upon withdrawal of the HER2 transgene expression (Xie et al. 1999). Tumors in MMTV-neuT mice are also dependent on continued oncogene expression. In the MMTV-rtTA/TetO-NeuNT bitransgenic variant of this model regulated by doxycycline, when expression of the neuT oncogene is induced in the mammary tissue of adult mice, this leads to the formation of multiple mammary tumors and lung metastases, and the entire primary tumor and metastatic disease fully regresses when neuT expression is withdrawn (Moody et al. 2002).

Although each of these models is subject to criticisms relating to their simplicity, when taken in aggregate, they are highly consistent and collectively make a highly compelling case that HER2 induced tumors are addicted to HER2. This has made HER2 one of the most sought after targets in cancer drug development.

Potential to escape from HER2 addiction

The inducible models do suggest the possibility of HER2-independent tumor recurrence that occurs after a period of complete regression. Tumors that are induced in the MMTV-rtTA/TetO-NeuNT model and which completely regress upon oncogene withdrawal, eventually recur after prolonged dormancy without induction of the neuT oncogene (Moody et al. 2002). This is associated with induction of the transcriptional repressor snail, and suggests that second hits may provide escape routes for residual tumor cells leading to tumor recurrence and progression driven by alternative pathways (Moody et al. 2005). Tetracycline regulated NIH3T3-HER2 tumors cells that regress following withdrawal of oncogene expression similarly recur following a period of remission despite absence of oncogene expression, although the molecular features associated with HER2 independent recurrence in this model are not yet described (Schiffer et al. 2003). The direct relevance of these models of recurrence to human tumors is not yet known and awaits analysis of human tumors that have recurred following a complete remission induced by HER2 targeted therapies.

INHIBITION OF HER2 FOR CANCER THERAPY

The evidence to support the HER2 oncogene hypothesis that HER2 initiates and drives the progression of HER2 overexpressing cancers is almost unimpeachable at this point. The direct consequence of this hypothesis is the treatment hypothesis that inactivation of HER2 could be highly effective therapy for patients with HER2 overexpressing cancers. Due to the large number of patients with this type of cancer, testing of the HER2 treatment hypothesis has been one of the most actively pursued programs in the cancer therapeutic arena. Testing in human subjects requires the development of safe and effective therapies that inactivate HER2 in patients tumors and in the best scenario are predicted to produce complete remissions and recapitulate results from preclinical models. Correlative scientific studies of these therapeutic agents in preclinical models and in patients are essential in order to determine the validity of the treatment hypothesis. These efforts have led to the development of the anti-HER2 mAb trastuzumab which has made significant clinical impact including a reduction in mortality from HER2 overexpressing disease. But mechanistic studies have been conflicting and suggest that the treatment hypothesis may not yet have been effectively tested. If the treatment hypothesis is correct and inhibiting oncogenic HER2 function would result in complete tumor regression, the clinical impact is predicted to be much larger than currently realized and we may have merely seen the tip of the iceberg. The available data with regards to several anti-HER2 targeted therapies are reviewed below. The two modalities for which there is considerable data are antibody therapies and small molecule kinase inhibitors. These are discussed separately below.

Biologic effects of panels of anti-HER2 antibodies

Almost immediately following the discovery that the oncogene HER2 was amplified in many breast and ovarian cancers and linked with disease biology, efforts began to develop inhibitors of this oncogene. The technology to develop mouse monoclonal antibodies (mAbs) had become available at this time and since HER2 functions as a growth factor receptor, it was a highly rationale hypothesis at the time that a mAb that binds the extracellular domain of HER2 and interferes with ligand activation would inhibit tumorigenic HER2 function. The proof-of-principle experiment was initially conducted in the neuT transformation model. In this model it was found that anti-Neu mAbs downregulate NeuT expression, supress cell growth, inhibit transformation and tumor growth in mice (Drebin et al. 1985; Drebin et al. 1988). This suggested that HER2 overexpressing human cancers could also potentially be treated with mAbs. Over one hundred mAbs were developed by numerous groups against the extracellular domain of human HER2. The effects of these mAbs on HER2 overexpressing human cancers turned out to be much more complicated than predicted from the more simplistic neuT model. The activities of some of these panels of mAbs against HER2 overexpressing tumor cell lines have been characterized and published and are summarized in Table 1. The results from these studies reveal that anti-HER2 mAbs can produce highly diverse outcomes. These include both growth inhibitory or growth stimulatory effects, differentiating effects, and pro-apoptotic effects. Some mAbs induce HER2 phosphorylation and others do not, some induce HER2 downregulation and others do not, some inhibit tumorigenic growth in vivo and others do not. The results of all these studies taken together do not formulate a clear picture of the mechanism by which an anti-HER2 mAb can inhibit tumor growth. Specifically, cell growth inhibition or tumor growth inhibition does not correlate with the mAb ability to downregulate HER2. In addition, anti-HER2 mAbs downregulate mutationally activated HER2 much more effectively than wildtype HER2, reproducing the effects seen with anti-Neu mAbs in the NeuT model (van et al. 1990). Adding complexity to the picture is that even growth inhibition in vitro does not correlate with tumor growth inhibition in vivo such that some mAbs are growth stimulatory in cell culture models yet inhibit tumor growth in mice (Stancovski et al. 1991; Harwerth et al. 1993). The mechanistic principles underlying the diversity of findings from anti-HER2 mAbs remain unclear to this day. But the compelling data regarding the role of HER2 in human tumorigenesis and evidence of anti-tumor efficacy of some anti-HER2 mAbs in preclinical models drove the clinical development of at least one such agent.

TABLE 1.

Published monoclonal anti-HER2 antibodies

Reference	N	Biochemical findings (in Her2++ tumors)	Biological findings (in Her2++ tumors)	Notes
(Stancovski et al. 1991; Bacus et al. 1992; Hurwitz et al. 1995; Klapper et al. 1997; Yarden 1990)	panel of 17	most induce phosphorylation, some induce endocytic degradation	some are growth inhibitory and differentiating, some are growth stimulatory	growth inhibition correlates with homodimerization and endocytic degradation
(van et al. 1990)	panel of 4	downregulation of ErbB2 is poor, downregulation of mutant ErbB2 is strong	only mutant HER2 was growth inhibited (soft agar)	engineered TM domain mutation
(Harwerth et al. 1993; Harwerth et al. 1992)	panel of 4	2 of 4 induce phosphorylation, 3 of 4 induce downregulation	1 of 4 inhibits cell growth 2 of 4 inhibit tumor growth	no correlation between HER2 turnover and growth inhibition
(Xu et al. 1993)	panel of 11	some internalize and downregulate HER2	2 inhibit monolayer growth of SkBr3, 7 inhibit growth in agar of SkBr3	no correlation between HER2 downregulation and growth inhibition, Fab fragment also inhibits growth
(Kita et al. 1996)	panel of 12	most induce phosphorylation	growth data not reported, one mAb induces apoptosis
(McKenzie et al. 1989)	panel of 10	not reported	not reported
(Tagliabue et al. 1991; Srinivas et al. 1993)	panel of 2	they induce phosphorylation and internalize	cause growth inhibition (Calu-3)	internalization requires bivalency and Fab is inactive
(Neve et al. 2001)	panel of 2	they internalize (both as scFv or IgG)	no growth inhibition	internalization does not require bivalency
(De et al. 2005)	one mAb	induces ErbB2 phosphorylation and downregulation	growth inhibition in vitro and in vivo
(Belimezi et al. 2006)	one mAb	internalizes	inhibits growth
(Hudziak et al. 1989; Sarup et al. 1991)	panel of 11	4D5 downregulates HER2	some inhibit growth in vitro. 4D5 inhibits growth in vivo

Development of trastuzumab (Herceptin®)

Of the over one hundred anti-HER2 mAbs generated in the 80s and 90s, one was developed for clinical testing. The mAb 4D5 was selected among a panel of mouse anti-HER2 antibodies at Genentech, Inc. (South San Francisco, CA) for development because of its anti-tumor effects in vitro and in mouse models (Hudziak et al. 1989; Shepard et al. 1991). Mouse mAb 4D5 was humanized for clinical use yielding several humanized variants. Some of these engineered variants lost in vitro anti-proliferative efficacy despite high affinity binding to HER2, but others retained anti-proliferative efficacy and one such clone was selected for further clinical development (named trastuzumab, Herceptin®) (Carter et al. 1992). The constant regions of these humanized mAbs were engineered for optimum participation in antibody dependent cellular cytotoxicity (ADCC) or complement dependent cytotoxicity (CDC) and indeed trastuzumab is much more efficient than its murine counterpart in mediating ADCC (Carter et al. 1992; Lewis et al. 1993; Tokuda et al. 1996). Trastuzumab has reduced cell culture anti-proliferative activity compared with the murine 4D5 but equally potent anti-tumor efficacy in mouse xenograft models (Tokuda et al. 1996; Baselga et al. 1998; Clynes et al. 2000).

Clinical anti-tumor activity of trastuzumab

The clinical anti-tumor activity of trastuzumab has now been extensively characterized in numerous clinical studies spanning the past decade and a half. Initial difficulties in identifying the subset of patients with HER2 overexpressing tumors by clinically available immunohistochemical methods were finally overcome by clinical implementation of a fluorescence in situ hybridization (FISH) assay to detect HER2 gene amplification and it is now evident that trastuzumab induces tumor regression in approximately 30–35% of patients with HER2 amplified metastatic breast cancer if used as upfront therapy (Vogel et al. 2002; Mass et al. 2005) and has much less activity if used after other chemotherapies (Baselga et al. 1996). In patients with metastatic disease, trastuzumab is not curative and disease progression resumes after a median duration of approximately 5 months despite continuous trastuzumab therapy (Vogel et al. 2002). The most beneficial clinical use of trastuzumab has been in combination with various cytotoxic chemotherapies. The addition of trastuzumab to multiple chemotherapy regimens significantly increases their anti-tumor efficacy (Slamon et al. 2001; Marty et al. 2005; Burstein et al. 2003). The biggest impact of trastuzumab has been in the treatment of patients with potentially curable early stage breast cancer. In early stage HER2 amplified breast cancer patients who receive chemotherapy after surgical resection, the addition of trastuzumab to their chemotherapy regimens significantly prolongs disease-free survival and reduces the chances of disease recurrence (Figure 1) (Piccart-Gebhart et al. 2005; Romond et al. 2005). Although these adjuvant therapy studies are still in their early years of follow-up, the powerful effects seen in the early follow-up period is widely believed to translate to a significant reduction in mortality from HER2 amplified breast cancer and the use of trastuzumab has rapidly become the standard management of early stage breast cancer patients. The clinical antitumor activity of trastuzumab is limited to tumors with HER2 overexpression and trastuzumab has no significant clinical activity against breast cancers without HER2 overexpression (Vogel et al. 2002; Mass et al. 2005). At this time its single agent activity appears to be limited to breast cancers and it has much less clinical anti-tumor activity against ovarian or endometrial cancers with HER2 overexpression (Fleming et al. 2003; Bookman et al. 2003) and continues to be investigated in other types of cancer.

Figure 1.

Outcome results with a median follow-up time of 2.0 years in 3351 patients with early stage breast cancer treated with chemotherapy and randomized to the addition of trastuzumab or control. Panel A shows the proportion of patients in each arm that remain cancer-free at the indicated years of follow-up. Many more patients from the trastuzumab-treated arm (solid line) are cancer-free compared with the control arm (broken line). Panel B shows the proportion of patients in each arm that are alive at the indicated years. More patients from the trastuzumab-treated arm are alive compared with the control arm. The effect of trastuzumab on disease-free and overall survival is highly statistically significant. (Romond et al. 2005) Copyright 2005 Massachusetts Medical Society, All rights reserved.

Implications of trastuzumab to the HER2 oncogene hypothesis

These improvements in the clinical management of patients with HER2 amplified breast cancer afforded by trastuzumab are a direct consequence of the HER2 oncogene hypothesis of breast cancer initially proposed two decades ago and are a testament to the potential of scientific research to impact human health and disease mortality. But while the success of trastuzumab is a consequence of the HER2 oncogene hypothesis, it is not sufficient to validate it. Validation of the oncogene hypothesis requires mechanistic evidence that trastuzumab treats patients through inactivation of tumor HER2. This evidence is currently lacking and more than a decades work in trying to determine the mechanism of action of trastuzumab has produced largely conflicting and inconclusive results and a compelling mechanistic model of how and whether trastuzumab inhibits oncogenic HER2 function has not emerged.

Mechanism of action of trastuzumab – HER2 downregulation

Extensive studies over the past decade have attempted to determine the molecular mechanism underlying the clinical anti-tumor activity of trastuzumab. The simplest hypothesis derives from the previously established anti-NeuT mAb and anti-HER2 4D5 mAb data showing that these mAbs induce the degradation of the targeted surface receptors NeuT or HER2 (discussed above). Although this seems like a fairly simple hypothesis to test conclusively, extensive analyses by numerous investigators studying the effect of trastuzumab on tumor cell HER2 expression have produced conflicting results, even in similar cell types and assays. While some studies show that trastuzumab downregulates HER2 in HER2 overexpressing tumor cells (Cuello et al. 2001; Marches and Uhr 2004; Citri et al. 2002; Lee et al. 2002), other studies clearly show that it does not (Longva et al. 2005; Austin et al. 2004). Part of the complexity in this field was resolved when it was determined that trastuzumab binds and internalizes with surface HER2 but re-emerges with HER2 at the surface, merely accompanying HER2 passively along its normal endocytic recycling route (Austin et al. 2004). The most compelling evidence at this point seems to support the position that trastuzumab does not induce the downregulation of HER2 in tumor cells. Consistent with this, three clinical studies show no reduction in tumor HER2 expression in patients undergoing treatment with trastuzumab (Arnould et al. 2006; Mohsin et al. 2005; Gennari et al. 2004). Therefore it appears unlikely that the clinical antitumor activity of trastuzumab is mediated through downregulation of tumor HER2.

Mechanism of action of trastuzumab – HER2 signaling

The principal prevailing hypothesis that rationalized the development of trastuzumab and other anti-HER2 mAbs for most of the nineties was that it inhibits the activation of HER2 by yet undiscovered ligands. However the hypothesized HER2 ligand was never discovered, and biochemical screens, post-genome computional studies, and crystal structure revelations have made it clear that HER2 has no physiologic ligand, and that its ligand responsive functions are mediated through heterodimerization with its ligand-activated HER family partners (Sliwkowski 2003). In fact the extracellular domain of HER2 constitutively exists in an active conformation resembling the ligand-bound state of the other HER family proteins, precluding any potential activating role for ligands (Garrett et al. 2003; Cho et al. 2003). Therefore the hypothesis that trastuzumab inhibits direct ligand binding and activation of HER2 is all but dismissed at this point. An alternative hypothesis that has been proposed is that trastuzumab inhibits the interaction of HER2 with its HER family partners or possibly other interacting proteins. But convincing evidence in support of this hypothesis has not yet emerged. In pulldown assays trastuzumab does not inhibit HER2–HER3 interaction (Agus et al. 2002), and in fluoresence resonance energy transfer (FRET) based assays trastuzumab also does not inhibit HER2 interaction with EGFR or HER3 (Diermeier et al. 2005). In a different model using truncated HER proteins fusing them to β-galactosidase fragments in an enzyme complementation assay, trastuzumab was reported to inhibit EGFR-HER2 interaction but not HER2–HER3 interactions (Wehrman et al. 2006). The artificial nature of the truncated receptors used in the latter study makes it less reliable, specially in light of FRET evidence to the contrary.

Mechanism of action of trastuzumab – Inhibition of HER2 cleavage

Trastuzumab binding inhibits the proteolytic cleavage and shedding of HER2 by ADAM proteases (Molina et al. 2001; Liu et al. 2006). This may in part inhibit the invasive properties of HER2 transformed cells since truncated HER2 induces a more invasive morphology and is associated with increased kinase activity, increased transforming efficiency, and is increased in patients with metastatic disease (Egeblad et al. 2001; Segatto et al. 1988; Molina et al. 2002; Christianson et al. 1998). Therefore trastuzumab may inhibit this aspect of HER2 function although the transforming function of HER2 is not known to require truncation and many HER2 overexpressing breast cancers do not have significant truncation of HER2.

Mechanism of action of trastuzumab – other findings

While the therapeutically important effect of trastuzumab on the function of its direct target HER2 remains to be defined, numerous reports have emerged describing the effects of trastuzumab on downstream signaling pathways. The anti-proliferative effect of mAb 4D5 or trastuzumab in cell culture models is associated with the induction of p27 and G1 block (Lane et al. 2000; Le et al. 2005; Marches and Uhr 2004). Trastuzumab affects the expression of tumor angiogenic factors and exhibits certain anti-angiogenic properties in mouse models (Izumi et al. 2002). Trastuzumab suppresses Akt signaling in some tumor cell types but not others (Longva et al. 2005; Yakes et al. 2002; Normanno et al. 2002), increases plasma PTEN localization and activity in cells (Nagata et al. 2004; Longva et al. 2005), and its anti-proliferative and anti-tumor effects are attenuated by PTEN knockdown (Nagata et al. 2004; Fujita et al. 2006). Consistent with a functional role for PTEN in clinical anti-tumor efficacy, tumors with reduced or absent PTEN are relatively resistant to trastuzumab-containing chemotherapy regimens (Nagata et al. 2004; Fujita et al. 2006). Although these data sets are complicated by the concomitant use of cytotoxic chemotherapy regimens, they are the only currently existing evidence linking intracellular signaling with the clinical antitumor activity of trastuzumab. An association between trastuzumab resistance and PTEN loss by itself does not necessarily imply that trastuzumab inhibits tumors through direct effects on tumor signaling, since PTEN loss has also been shown to mediate immuno-resistance (Parsa et al. 2006).

Mechanism of action of trastuzumab – Immunological targeting

An increasing body of evidence suggests that the in vivo anti-tumor effects of the anti-HER2 mAb 4D5 and the humanized trastuzumab may be mediated, at least in part, if not entirely, through immunological targeting mechanisms. mAb 4D5 activates ADCC in vitro (Carter et al. 1992; Clynes et al. 2000). This activity was greatly enhanced during the process to engineer the humanized version and trastuzumab is indeed highly efficient at activating ADCC in vitro (Carter et al. 1992; Cooley et al. 1999; Carson et al. 2001; Repka et al. 2003; Kubo et al. 2003). Mouse genetic models that experimentally manipulate Fc receptor function positively or negatively clearly demonstrate the role of host immunologic mechanisms in the anti-tumor efficacy of these agents. The antitumor activity of both mAb 4D5 and trastuzumab are almost entirely abolished in the loss-of-function FcRγ−/− mice, while the antitumor activity of subtherapeutic doses of mAb 4D5 and trastuzumab are greatly enhanced in the gain-of-function FcRII−/− mouse model (Clynes et al. 2000). Furthermore, when the Fc region of the mAb 4D5 is mutated at a single position to eliminate engagement of host Fcγ receptors, the mutant mAb 4D5 retains all its in vitro anti-proliferative activity but loses its in vitro ADCC activity and loses its in vivo anti-tumor efficacy (Clynes et al. 2000). This model has provided highly compelling evidence that the antitumor activity of trastuzumab is mediated, in large part, through immunological targeting of tumor cells.

Investigators have also recently begun to look for clinical evidence of immunological targeting by trastuzumab. In a clinical study of a trastuzumab-containing chemotherapy regimen compared with case-matched controls, trastuzumab treatment was associated with significantly increased numbers of activated cytolytic natural killer (NK) cells within tumors (Arnould et al. 2006). In a second study of trastuzumab monotherapy, the treatment induced varying degress of tumor infiltration with lymphoid cells and the patients that responded to trastuzumab had the highest degree of tumor lymphocyte infiltration and higher ADCC activity measured ex-vivo (Gennari et al. 2004). The evidence that trastuzumab works through an ADCC mediated mechanism has led some investigators to determine whether this activity can be further enhanced by additional immunostimulatory approaches. Consistent with this hypothesis, IL-2 enhances the in vitro ADCC activity of trastuzumab (Carson et al. 2001; Kubo et al. 2003). However a phase I trial of this combination modality found no evidence of increased clinical activity associated with IL-2 induced NK cell expansion (Repka et al. 2003). In another approach, a bi-specific antibody that targets HER2 as well as CD3 antigen on T-cells was developed in order to recruit T-cells to tumor cells. This agent showed promising evidence of activity in its phase I study and further studies are underway to determine its anti-tumor efficacy (Kiewe et al. 2006). In another approach, the in vitro ADCC activity of trastuzumab appears to be augmented by defucosylation, and the fucosyl-negative version of trastuzumab has been proposed for clinical testing (Suzuki et al. 2007).

These preclinical experimental models and clinical observational studies have provided compelling evidence that much of the anti-tumor activity of trastuzumab is mediated through immunological targeting mechanisms. Although these data sets are still relatively few, the failure of the signaling field to confirm that the clinical activities of trastuzumab are mediated through inhibition of HER2 signaling has led to more and more attention being focused on the pursuit of the immunological targeting hypothesis and additional studies to further explore this hypothesis are underway.

Targeting HER2 with pertuzumab

A second anti-HER2 mAb is currently undergoing clinical testing. From the initial panel of anti-HER2 mAbs developed at Genentech, Inc., the mAb 2C4 was also selected for further characterization and development and has properties distinct from mAb 4D5. mAb 2C4 has been modified for human clinical use by recombinant engineering to generate the humanized version pertuzumab (Adams et al. 2006). Pertuzumab is currently undergoing clinical trials and its clinical development thus far has been oriented towards diseases other than HER2 overexpressing cancers. It’s antitumor activity in HER2 overexpressing breast cancer is not yet known, but it has little activity in breast cancers without HER2 overexpression and in unselected ovarian cancers (Cortes et al. 2005; Fleming et al. 2005).

The activities of 2C4 or pertuzumab on cell signaling pathways have only been reported by selected investigators since these agents are not yet available for study to the wider scientific community. Compared with mAb 4D5 or trastuzumab, mAb 2C4 has much less anti-proliferative activity in vitro (Hudziak et al. 1989; Takai et al. 2005) but has in vivo antitumor activity in a number of tumor types including tumors without HER2 overexpression (Agus et al. 2002; Takai et al. 2005). 2C4 is reported to inhibit heregulin mediated HER2 receptor complex formation, phosphorylation, and MAPK and Akt activation in breast cancer cells without HER2 overexpression (Agus et al. 2002). Determination of the effects of pertuzumab on HER2 overexpressing cancers awaits much more preclinical and clinical studies.

Targeting HER2 with HER kinase inhibitors

The technology to develop selective tyrosine kinase inhibitors (TKIs) for human use succeeded antibody therapeutics by almost a decade. These agents, at least in theory, have certain advantages over antibody therapies for the treatment of HER2 amplified cancers. Antibody therapies are cell impermeable agents that bind the extracellular domain of HER2 and to this day it remains unclear whether or how this binding activity can suppress oncogenic HER2 function, although they may induce the clinically important immunological targeting of HER2 overexpressing cancer cells. TKIs are cell permeable agents and can potentially inhibit the ligand dependent and independent kinase activity of HER2 residing within the intracellular domain. This strategy has a solid rational basis since kinase activity is essential for the oncogenic function of neu or HER2 (Weiner et al. 1989; Qian et al. 1994). At least in theory, these agents offer the opportunity to inactivate HER2 kinase function in patients with HER2 overexpressing cancer and for the first time directly test the HER2 oncogene hypothesis in patients. However TKIs don’t have the singular target specificity of antibodies and their off-target effects potentially limits their therapeutic index compared with antibodies.

The development of HER-selective TKIs

Synthetic and natural product inhibitors of HER kinases from diverse structures were initially studied in the early 1990s, but their limited potencies and specificities precluded their usefullness as anti-tumor agents. The field was revolutionized by the discovery of modified quinazoline compounds as highly specific and potent inhibitors of the epidermal growth factor receptor (Ward et al. 1994; Rewcastle et al. 1995). Extensive structure-activity relationships were determined and numerous improved quinazoline compounds have subsequently been developed with varying selectivity characteristics for individual HER family members. In addition to quinazolines, several other structures have now been found to potently and selectively inhibit HER kinases. Table 2 lists a number of HER TKIs that have been publicly disclosed, and for which preclinical data has been presented. In addition to these, numerous other agents are in development which are not yet disclosed at the time of this review. Almost all of these agents are ATP analogs and inhibit kinase activity by binding within the ATP pocket of the catalytic domain (Stamos et al. 2002; Wood et al. 2004). Some of these compounds bind reversibly within the ATP pocket and are competitive with ATP while others bind irreversibly and are non-competitive with ATP.

TABLE 2.

List of disclosed HER family TKIs in preclinical and clinical development.

Developer	Compound	Class	Type	Selectivity			Activity demonstrated against these HER2+++ tumor models		References
				in vitro IC50 (nM)
				EGFR	HER 2	HER 4	in vitro	in vivo
AstraZeneca	ZD1839 gefitinib	quinazoline	reversible	27	370 0		BT474, SKBr3, SKOv3, MCF-7HER2	BT474, MCF-7HER2, GLM-1, GLM-4	(Wakeling et al. 2002; Moasser et al. 2001; Moulder et al. 2001; Ghossein and Bhattacharya 2001; Warburton et al. 2004; Yokoyama et al. 2006)
OSI-Genentech-Roche	OSI-774 erlotinib	quinazoline	reversible	2	350			KPL-4, Calu-3	(Moyer et al. 1997; Akita and Sliwkowski 2003; Friess et al. 2005)
GlaxoSmithKline	GW572016 lapatinib	thio-quinazoline	reversible	11	9	367	BT474, Calu3, N87, HB4aHER2, SUM225, H16N2HER2, SUM190, SKBr3, UACC893, UACC812	BT474	(Rusnak et al. 2001; Konecny et al. 2006)
ParkeDavis-Pfizer	CI-1033 PD183805 canertinib	quinazoline	irreversible	0.8	19	7			(Rabindran 2005; Slichenmyer et al. 2001)
OSI-Pfizer	CP-724714	quinazoline	reversible	4300	8		SKBr3, another panel of 9	FREV664E-HER2	(Jani et al. 2004b; Jani et al. 2004a; Finn et al. 2004)
OSI-Pfizer	CP-654577	quinazoline	reversible	670	11		BT474, SKBr3	FREV664E-HER2	(Rabindran 2005; Barbacci et al. 2003)
Wyeth-Ayerst	CL-387785 EKI-785	quinazoline	irreversible	0.4			SKBr3, 3T3HER2		(Discafani et al. 1999; Torrance et al. 2000)
Academic	AG1478	quinazoline	reversible	3	200 0		SKBr3, BT474	MMTV-neu	(Levitzki and Gazit 1995; Kurokawa et al. 2000; Lenferink et al. 2001; Lenferink et al. 2000)
Arry Biopharma	Arry-334543	quinazoline	reversible	7	2	10	BT474	BT474, Calu3, MDA-453	(Miknis et al. 2005; Pheneger et al. 2005)
Boehringer Ingelheim	BIBW-2992	quinazoline	irreversible	0.5	14		BT474, N87	MDA-453, N87, SKOv3	(Solca et al. 2006)
AVEO Pharm.-Misubishi	AV-412 MP-412	quinazoline	irreversible	1	18			BT474, KPL-4, BH248	(Robinson and Lin 2006; Fujii et al. 2005)
Novartis	AEE-788	pyrrolopyrimidine	reversible	6	6	160	BT474, SKBr3	HC11neuT	(Traxler et al. 2004)
Novartis	CGP-59326A	pyrrolopyrimidine	reversible	70	>10 uM		SKBr3, ZR75-30		(Lydon et al. 1998)
Novartis	PKI-166 CGP-75166	pyrrolopyrimidine	reversible	1	11	430	BT474, SKBr3	HC11neuT	(Traxler et al. 2001; Brandt et al. 2001)
Wyeth-Ayerst	EKB-569 pelitinib	cyanoquinoline	irreversible	39	120 0		SKBr3		(Wissner et al. 2003; Torrance et al. 2000)
Wyeth-Ayerst	HKI-272	cyanoquinoline	irreversible	92	59		3T3neu, SKBr3, BT474	3T3neu,BT474, SKOv3	(Rabindran et al. 2004; Tsou et al. 2005)
Wyeth-Ayerst	HKI-357	cyanoquinoline	irreversible	33	34		SKBr3	3T3neu,BT474, SKOv3	(Kwak et al. 2005; Tsou et al. 2005)
Bristol Myers Squibb	BMS-599626	pyrrolotriazine	reversible	22	32	190	BT474,N87,KPL4, HCC1419,HCC1954, Sal2	KPL4, Sal2	(Wong et al. 2006)
Takeda	TAK-165 mubritinib	triazole		>25u M	6		BT474	BT474	(Naito et al. 2002; Yoshida et al. 2002; 2006)

Although HER family kinases are highly homologous, many TKIs show selectivity among the individual family members when assayed by in vitro kinase assays using purified EGFR,HER2, or HER4 kinases, and these are listed in Table 2. However the biological relevance of these in vitro characteristics is not evident in cell based assays. The EGFR selective agent gefitinib inhibits the phosphorylation of all of the HER proteins in cells, as does the HER2-selective agent CP-654577 (Moasser et al. 2001; Normanno et al. 2002; Shepherd et al. 2001). In cell proliferation assays, HER2 overexpressing tumors are particularly sensitive to the highly EGFR-selective TKIs gefitinib, AG1478, CGP-59326A, and EKB-659 (see Table 2). In fact in an engineered HER2 overexpressing tumor model, the level of HER2 overexpression directly correlates with sensitivity to the EGFR-selective TKI AG1478 (Emlet et al. 2006). The mechanisms underlying the activities of EGFR-selective TKIs against HER2 signaling and HER2-driven growth are not currently understood, and may be due to the direct inhibition of HER2 kinase by their weaker anti-HER2 activity, high intracellular concentrations of these agents precluding their target selectivities, or a necessary role for EGFR in HER2-driven growth. The latter appears unlikely since at least in fibroblast models, HER2 is transforming even in the absence of EGFR expression (Chazin et al. 1992). These uncertainties with respect to the biologic importance of target-specificity are somewhat academic and virtually all HER TKIs show activity against HER2-driven tumors in mouse models regardless of their in vitro target selectivities (see Table 2) and they are all potentially candidate agents to test the validity of the HER2 treatment hypothesis in patients with HER2 overexpressing cancers.

Clinical anti-tumor activity of HER TKIs – Current data

Numerous TKIs are currently in various stages of preclinical and clinical development. The clinical development of each of these is prioritized towards specific cancer subtypes by their sponsors, therefore the activities of some of them against HER2 driven cancer may not be tested early in their course of development. But a limited amount of clinical data is now available to give a preliminary impression of the anti-tumor efficacy of this class of TKIs in patients with HER2 overexpressing cancers. The currently reported phase II efficacy data is summarized in Table 3. In addition to the drugs listed here, numerous multi-targeting TKIs that target other kinase families in addition to HER family kinases are in preclinical and clinical testing and although such drugs may be found to be active in the treatment of HER2-driven cancers, their multi-targeting characteristics make them less suited to test the HER2 oncogene hypothesis. Completed phase II studies of gefitinib and erlotinib have been reported in breast cancer patients. Although these were not conducted specifically in patients with HER2 overexpressing cancer, the cohorts include patients with HER2 overexpressing disease. Overall response rates of 0–10% were seen in these studies. The most informative data currently comes from clinical studies of lapatinib. Lapatinib has specifically been developed for the treatment of HER2 overexpressing breast cancer and numerous clinical efficacy studies as well as correlative scientific studies have been conducted and are ongoing to determine the activity of this agent in patients with HER2 overexpressing breast cancer. The phase II efficacy studies completed and confirmed to date show a response rate of 4–8% in patients with HER2 overexpressing breast cancer. Two additional ongoing but unconfirmed studies have reported higher response rates in the 24–30% range. A number of additional phase II studies are underway to test the efficacy of other investigational HER TKIs in patients with HER2 overrexpressing breast cancer and there will be much more data emerging in the next few years . The data so far shows evidence of weak clinical anti-tumor activity in this disease. In addition to the studies reviewed here, numerous clinical studies are underway to determine whether the addition of TKIs to cytotoxic chemotherapeutics, to trastuzumab, or to hormonal therapies produces new combinations with increased clinical activity and prolongation of survival in patients. These studies may lead to better treatment options for patients, but they are not a direct test of the HER2 oncogene hypothesis and a detailed discussion of these studies is beyond the scope of this review.

TABLE 3.

Results of currently presented phase II efficacy studies of HER TKIs in HER2 overexpressing cancers

Reference	Drug	phase	selection criteria	N	Response Rate	comments
(Baselga et al. 2005)	gefitinib/ZD1839	phase II	advanced breast ca	31	3%
(Robertson et al. 2003)	gefitinib/ZD1839	phase II	advanced breast ca - TAM-resistant	19	10%
(Albain et al. 2002)	gefitinib/ZD1839	phase II	advanced breast ca	63	2%	28 HER2+++ patients
(Winer et al. 2002)	erlotinib/OSI-774	phase II	advanced breast ca	69	3%	many HER2+++ patients
(Tan et al. 2004)	erlotinib/OSI-774	phase II	advanced breast ca	18	0%	2 HER2+++ patients
(Blackwell et al. 2006; Blackwell et al. 2004)	lapatinib/GW572016	phase II	HER2+++ breast ca - after trastuzumab	81	8%
(Burstein et al. 2004)	lapatinib/GW572016	phase II	HER2+++ breast ca - after chemo	140	4%
(Iwata et al. 2006)	lapatinib/GW572016	phase II	HER2+++ breast ca - after chemo	45	24%	(preliminary)
(Gomez et al. 2005)	lapatinib/GW572016	phase II	HER2+++ breast ca - first line	40	30%	(preliminary)
(Campos et al. 2005)	canertinib/CI-1033	phase II	ovarian ca - after chemo	105	0%	2 HER2+++ patients

Evidence that HER TKIs inhibit HER2 function in patients

The HER2 oncogene hypothesis would predict that the vast majority of HER2-driven tumors would initially respond to treatments that suppress HER2 kinase function. Correlative studies of tumor material from patients on treatment are essential for understanding whether HER2 function and signaling has effectively been suppressed by these treatments. Such correlative studies require purely research interventions in consenting patients by performing biopsies of their tumors just before and while on therapy and these studies are extremely difficult to accomplish for a myriad of practical and ethical reasons. At least two groups thus far have succeeded in generating clinical scientific data in patients on HER TKIs. In a phase I clinical study of lapatinib, tumor biopsies were obtained before and while on therapy to determine the suppression of tumor EGFR/HER2 signaling by immunohistochemical staining (Spector et al. 2005). This study showed mixed results with variable degrees of suppression of targets, partly because this was a phase I dose escalation study involving patients with different types of cancers, including cancers not known to be HER2-dependent, and starting dose levels that are likely less effective at target suppression. The data does however show a reduction in EGFR and HER2 phosphorylation in most patients, and a reduction in MAP kinase signaling. A reduction in Akt signaling was less evident in this data set. In a phase II study of gefitinib in breast cancer patients, skin biopsies and tumor biopsies were obtained in many patients before and while on therapy for immunohistochemical analysis of target suppression (Baselga et al. 2005). This study showed effective suppression of EGFR phosphorylation and MAPK signaling in skin and tumor from drug therapy, but no suppression of Akt signaling. HER2 phosphorylation was not assayed on this study and the 3 patients with HER2 overexpressing tumors that were enrolled on this study did not have on-treatment tumor biopsies for analysis. It should be noted that the use of immunohistochemical techniques on paraffin embedded tissues using phospho-specific antibodies is fraught with technical problems that limits their reliability, and until newer techniques emerge, such studies should be interpreted with caution (Baker et al. 2005). Despite the technical problems associated with phosphoprotein immunostaining and the fact that these two studies were not designed to specifically determine target inactivation in HER2 overexpressing tumors at maximal drug dosage, they do appear to suggest that the drugs do reach their tumor targets and at least partially inactivate them. Tumor biodistribution does not seem to be a limiting step since at least with regards to gefitinib, tumor and tissue concentrations have been measured and are much higher than serum concentrations and well above concentrations that fully suppress EGFR and HER2 signaling in cell culture models (McKillop et al. 2005).

Evidence that TKIs do not fully inhibit oncogenic HER2 functions

Significant mechanistic insight into the effective suppression of oncogenic HER2 signaling by TKIs was recently afforded by the analysis of steady state HER3 and downstream Akt signaling. While TKI treatment effectively suppresses EGFR and HER2 autophosphorylation and downstream MAP kinase signaling in HER2 amplified tumors, HER3 appears to escape TKI therapy at conventional doses and concentrations (Sergina et al. 2007). This is due to Akt-driven negative feebdack signaling which restores HER3 signaling activity despite significant suppression of HER2 kinase function thereby maintaining downstream Akt signaling and numerous Akt-driven pathways important for tumor survival (Sergina et al. 2007). This feedback loop essentially buffers HER3 signaling against the incomplete loss of HER2 kinase function and underscores the tumor cells critical need to maintain Akt signaling and numerous Akt-driven pathways important for tumor survival. The fact that HER3 signaling is buffered against an incomplete inactivation of HER2 kinase raises the bar for drug development because it suggests that proper testing of the HER2 oncogene hypothesis will require drugs that can completely inactivate HER2 kinase function. Testing this principle in cell culture models with much higher concentrations of TKI or with the addition of anti-HER3 siRNA approaches shows that HER2-overexpressing tumor cells will undergo apoptosis if HER2 function, including its transactivation of HER3 and Akt signaling, is interrupted for 48 hours or longer (Sergina et al. 2007). This reaffirms the oncogene-addicted nature of HER2-driven tumors and suggests that if HER2 can be effectively inactivated in patients tumors, this could produce highly significant and rapid anti-tumor responses. But the doses required to completely inactivate HER2 and effectively suppress HER3 signaling will likely produce significant toxicities in patients due to their off-target effects and may not be safely achievable. The effective suppression of oncogenic HER2 function in patients may require drugs that are much more potent than the current generation of compounds and at the same time are highly selective with a wide therapeutic index.

Other approaches to treat HER2 overexpressing cancers

Numerous other pharmacologic approaches are being pursued to develop effective therapies for patients with HER2 overexpressing cancers. These include anti-HER2 antibodies conjugated to a number of cellular toxins or anti-HER2 antibodies placed on chemotherapy containing immunoliposomes that can deliver cytotoxics more effectively to HER2 overexpressing cancer cells. A number of agents can interfere with cellular mechanisms that regulate gene expression or protein expression and these can reduce HER2 expression in tumor cells. These include histone deacetylase inhibitors, HSP90 inhibitors, and COX-2 inhibitors. While these promising treatment approaches are a consequence of the HER2 oncogene hypothesis, their non-specific mechanisms of action limits their usefullness in validating the hypothesis and are reviewed here. Readers are referred to several excellent recent reviews for a comprehensive coverage of these HER2 targeting approaches (Kumar and Pegram 2006; Menard et al. 2003; Chen et al. 2003; Brand et al. 2006). Numerous immunological modalities are also being pursued to enable host defense mechanisms to target HER2 overexpressing cancer cells, and these are also beyond the scope of this review. In addition, numerous approaches to specifically suppress HER2 expression have been postulated over the years and continue to be developed. These include antisense, ribozyme, and siRNA approaches to inhibit HER2 expression. These modalities have not yet translated to clinically effective products that can test the validity of the HER2 oncogene hypothesis in patients. Future development of these technologies, in particular siRNA approaches, may lead to a new class of agents that can specifically and effectively inactivate oncogenic HER2 function.

Resistance to inhibitors of HER2

Numerous mutational events downstream of growth factor receptors have now been described in cancers. These include activating mutations in BRAF, Ras, PIK3CA, and inactivating mutations or deletions of PTEN. Since these genes function downstream of HER2 and since each of these mutations induces constitutive signaling activity, at least in theory these mutations can uncouple downstream pathways from HER2 rendering tumor growth independent of HER2 and resistant to inhibitors of HER2. In breast cancers Ras and BRAF mutations are rare, but HER2 overexpression occurs frequently with PIK3CA mutations, but rarely with PTEN mutation (Saal et al. 2005). There is currently no data to determine whether the co-existence of PIK3CA mutation confers resistance to TKIs in HER2 overexpressing breast cancers. Of note, the commonly used BT474 cell line commonly used as a model of HER2 amplified breast cancer, and which is sensitive to TKIs and trastuzumab, harbors an unsual mutation in exon I of PIK3CA ((Saal et al. 2005) and cosmic database). PTEN mutations are rare in breast cancers, but reduced PTEN expression may have biological significance and is often seen in breast cancers (Bose et al. 2002). PTEN loss has been shown to be induce TKI resistance in an EGFR amplified breast cancer cell line (She et al. 2003; Bianco et al. 2003), but the analogous effect has not been shown in a HER2 amplified model. Future studies will establish whether PIK3CA and PTEN are important determinants of TKI sensitivity in HER2 overexpressing tumors. However the analysis of clinical resistance will only be meaningful if drugs have been developed that fully inactivate HER2 function. As discussed above in this review, trastuzumab does not seem to inactivate HER2 and its mechanism of action remains unclear, and TKIs studied to date seem to be partial inhibitors of oncogenic HER2 signaling in vivo.

Current status of the HER2 oncogene hypothesis

The preliminary data with lapatinib is a promising hint that at least a minority of HER2 overexpressing tumors may be dependent on HER2 kinase function. But why the majority of patients fail to respond to this therapy remains to be determined. It is possible that the HER2 oncogene hypothesis is wrong, and that despite the abundant and highly compelling evidence from experimental models that HER2 overexpressing breast cancers are HER2-driven and HER2-dependent, that this hypothesis does not hold true for patients with breast cancer and the experimental models are too simplistic to predict the behavior of naturally occuring cancers. However the HER2 oncogene hypothesis can only be discounted if clinical studies show lack of anti-tumor efficacy in spite of effective inactivation of tumor HER2 function and this has certainly not been shown. The alternative possibility is that current therapies do not effectively suppress oncogenic HER2 function in tumors. The recent revelation that current TKIs are not effectively suppressing HER3/PI3K/Akt signaling strongly supports this position, and opens the way for a future generation of TKIs.

Therefore it appears almost surely the case that the HER2 oncogene hypothesis has not yet been effectively tested in patients. Much more data will be forthcoming in the coming years with regards to numerous other HER TKIs with distinct structural and biochemical properties, including their anti-tumor efficacies and their biochemical target effects in patients tumors. If these agents are unable to completely inactivate HER2, then we may have to await a newer generation of more potent agents. Alternatively, HER2 oncogenic function can be inactivated by the combination of a HER TKI and an inhibitor of the HER3/PI3k/Akt signaling pathway. Inhibitors of this pathway are also in preclinical development and the combination treatment hypothesis will likely also be tested in the coming years as these agents enter clinical testing phases.

The scientific evidence implicates a role for HER2 in breast cancer similar to bcr-abl in chronic myelogenous leukemia (CML). Since effective inhibition of bcr-abl produces complete remissions in nearly all patients in chronic phase CML, it is imperative that the analogous treatment hypothesis regarding HER2 be effectively tested in breast cancer. The potential inherent in this hypothesis is enormous and includes the scenario of the first instance of an epithelial cancer being eradicated even in advanced stages. Such an outcome would have historic significance.