Nonendocrine Pathways and Endocrine Resistance: Observations with Antiestrogens and Signal Transduction Inhibitors in Combination

Endocrine response pathways in breast cancer have often been described solely in terms of the intracellular pathways used by estrogens and the subsequent antagonistic effects that antiestrogens have on estrogen production or the estrogen receptor α (ER). However, we now know that the ER does not work in isolation from the remainder of the cell phenotype, and many other signaling elements, notably those comprising growth factor transduction cascades, have been identified that can influence or be influenced by ER signaling (1, 2). Such elements, therefore, have the potential to enhance the growth-promoting activity of estrogens in breast cancer cells. Moreover, when altered in expression, they could adversely influence therapeutic responses to anti-hormonal drugs and drive resistance to such agents (1, 2). The expectation is that the cotargeting of such elements with ER may more effectively limit the actions of estrogens; enhance quality and extend duration of response to antiestrogens; provide novel, efficient treatments for de novo and acquired antiestrogen-resistant states; and thereby significantly improve breast cancer survival (3).

This article focuses on the relationship between altered growth factor signal transduction and antiestrogen failure in breast cancer in vitro models. It reveals how we are beginning to successfully target the cellular actions of such signaling pathways to improve the actions of anti-hormonal drugs and block the aggressive disease progression that often accompanies loss of endocrine response.

It is now widely documented that the inappropriate activation of growth factor signaling cascades, either through an enhanced supply of growth factor ligands or via up-regulation and increased activation of their target growth factor receptors or their recruited downstream signaling elements, can readily promote antiestrogen failure in breast cancer cells. As such, this phenomenon has been described for the overexpression of multiple growth factors and their receptors, including heregulins acting through HER3 and HER4 (4, 5, 6), epidermal growth factor (EGF) and transforming growth factor (TGF)-α acting through the EGF receptor [EGFR (2, 7, 8)], and insulin-like growth factor (IGF)-I and -II acting through the type I IGF receptor [IGF-IR (9, 10)], with HER2 contributing to antiestrogen failure either directly when overexpressed (11, 12, 13) or indirectly through heterodimerization with other erbB receptor family members (8).

In our in vitro studies, the resistant sublines that we have developed by prolonged exposure of endocrine-responsive MCF-7 breast cancer cells to antiestrogens uniformly express increased amounts of EGFR mRNA and protein (2, 7, 8). Thus, whereas EGFR immunostaining of the parental MCF-7 cells demonstrates only modest levels of this receptor, tamoxifen- or fulvestrant-resistant cells show up to 40-fold higher levels of EGFR membrane staining. We have also noted marked parallel increases in HER2 immunostaining in antiestrogen-resistant cells, again localized at the plasma membrane (8, 14). Complementary data have been reported previously for EGFR by Yarden et al. (15), who demonstrated that in the absence of estrogen EGF has a much stronger proliferative effect, indicating an increased potential of such cells to use this ligand for growth. Treatment with the pure antiestrogen ICI 164,384 was reported to also increase EGF growth responses, further confirming that therapies depriving cells of their ER signaling enhance sensitivity to EGFR ligands. Suppression of ER signaling has been associated with increased EGFR in MCF-7, T47D, and BT474 cell lines (15, 16). Our phenotypic data monitoring EGFR and HER2 in breast cancer cell lines are supported by a battery of in vitro gene transfer studies demonstrating a causative association between these receptors and antiestrogen resistance (11, 12, 13, 17, 18) and also by the EGFR/HER2 profiles observed in additional acquired antiestrogen resistance models (19, 20, 21, 22, 23).

Consistent with the concept that overexpressed EGFR and HER2 play a role in the development of antiestrogen resistance, we have demonstrated that these receptors colocalize using immunofluorescence and that they are heterodimerized with an increased basal level of activation using immunoprecipitation (8). Because the tamoxifen-resistant variants also express several EGFR ligands, each of which can further increase activation of EGFR and HER2 and induce additional growth responses, it is likely that the new growth signal originates from an EGFR-driven autocrine regulatory loop (8). Further assessment of the importance of EGFR/HER2 signaling in antiestrogen-resistant cells was made possible after we developed immunocytochemical and Western blotting procedures for localizing the phosphorylated (activated) forms of extracellular signal-regulated kinase (ERK) 1/2 mitogen-activated protein kinase (MAPK), Akt, and protein kinase C (PKC) α and δ using phosphorylation state-specific antibodies (7, 8, 24). These signaling intermediates are pivotal components of the intracellular phosphorylation cascades from the plasma membrane to the nucleus recruited for EGFR/HER2-induced proliferative and survival signals (25, 26, 27, 28). Using such techniques, we have observed that each of these elements is considerably increased in our tamoxifen-resistant subline as compared with the parental MCF-7 cells and is further inducible by various ligands for the EGFR (8). Interestingly, complementary associations have been reported previously by other groups between acquisition of tamoxifen resistance by ER-positive breast cancer cells in vitro and increased ERK1/2 MAPK activation (13, 29, 30, 31) or Akt phosphorylation (28, 32). MAPK increases are also reported to contribute to the growth of ER-positive breast cancer cells during adaptation to long-term estrogen deprivation (23, 33). Similarly, overexpression of PKCα in ER-positive breast cancer cells appears able to promote hormone-independent growth (34).

Our tamoxifen-resistant variants, like their clinical counterparts, continue to express ER at a level equivalent to that observed in the parental cell line and retain sensitivity to fulvestrant challenge (35). It is now evident that several growth factor-induced protein kinases (e.g., ERK1/2 MAPK and Akt), in addition to their direct stimulation of proliferation and survival signals, are able to target and phosphorylate key regulatory sites on the ER protein, notably within the AF-1 domain. This results in ligand-independent receptor activation (28, 36, 37). This process appears to be able to enhance the activity of the tamoxifen-ER complex as a positive nuclear transcription factor, increasing its transcriptional and growth-promoting properties (36, 38). Certainly, our recent studies have revealed that tamoxifen-resistant cells show an increase in the basal phosphorylation of ER on serine 118 (39) and also serine 167 residues, putative target sites for phosphorylation by ERK1/2 MAPK (37) and Akt, respectively (28), with their growth promoted rather than inhibited by tamoxifen. Increases in serine 118 phosphorylation have also been detected in cells adapted to long-term estrogen deprivation (23, 40). Phosphorylation of ER in our tamoxifen-resistant cells again appears to be under the control of the EGFR. Thus, exposure of these cells to either EGF or TGF-α further enhances ER phosphorylation, whereas exposure to the EGFR-selective tyrosine kinase inhibitor gefitinib (Iressa) antagonizes basal and EGF ligand-primed phosphorylation of these sites (39).

Whereas we are currently studying the molecular detail linking ER phosphorylation to the growth of tamoxifen-resistant cells, it is already clear from our studies that the ER is involved in maintaining the new EGFR-driven growth regulatory loop (35). Thus, exposure of tamoxifen-resistant cells to the pure antiestrogen fulvestrant, at a dose that depletes ER protein by increasing sensitivity of this receptor to proteolytic attack and disrupting its nucleocytoplasmic shuttling, leads to a concomitant loss of activation of EGFR and HER2. There is an equivalent reduction in activation of the EGFR/HER2 downstream signaling components ERK1/2 MAPK and Akt. Importantly, the loss of ER and EGFR/HER2 signaling after fulvestrant treatment is associated with effective inhibition of the growth of the tamoxifen-resistant cells. Because fulvestrant does not decrease the total cellular levels of the EGFR, HER2, or ERK1/2 MAPK proteins in such cells, it appears likely that this antiestrogen influences the activity of the growth factor signaling pathway by limiting the availability of one or more of its ligands. Interestingly, our preliminary studies indicate that a key EGFR ligand targeted by fulvestrant in tamoxifen-resistant cells may be TGF-α (35). Such a concept is reinforced by “add-back” experiments in which exogenous TGF-α not only reactivates EGFR, HER2, ERK1/2 MAPK, and Akt signaling but also supports substantial tumor cell growth in the presence of fulvestrant. Strengthening the EGFR pathway is thus able to entirely circumvent the catastrophic effects of this antiestrogen on the ER protein in such cells. EGFR ligand-treated cells are, as a consequence, refractory to the growth-inhibitory effects of tamoxifen, fulvestrant, or estrogen withdrawal (i.e., “complete endocrine insensitivity”), data implying that the primary growth-regulatory role for ER in the tamoxifen-resistant cells is to maintain the efficiency of EGFR signaling (2).

Numerous studies have demonstrated that IGF-IR is a key receptor in the growth of breast cancer cells that shows a significant level of productive cross-talk with ER (41, 42). Thus, estrogens appear to favor synergistic interactions with IGFs, with these steroid hormones inducing, for example, expression of the IGF-IR, whereas IGFs prime activation of several protein kinases that are able to phosphorylate ER and enhance its transcriptional activity (41, 42, 43). As such, our recent studies have shown that the parental endocrine-responsive MCF-7 cells express considerable levels of total and activated IGF-IR, which are particularly prominent at the plasma membranes.¹ There are also appreciable levels of phosphorylated insulin receptor substrate-1, a major downstream element recruited to IGF-IR signaling that is involved in promoting tumor cell growth. Productive cross-talk between IGF-IR and ER signaling is apparent in the endocrine-responsive cells, with IGFs priming ER phosphorylation on serine 118 and serine 167. There is a parallel increase in growth, with the effect reduced by an IGF-IR-selective inhibitor, AG1024. Interestingly, whereas tamoxifen treatment of the endocrine-responsive cells reduces the levels of total and activated IGF-IR by approximately 80% as an integral part of the endocrine response mechanism (44), we have recently noted that IGF-IR level and activation are substantially recovered in the tamoxifen-resistant subline, again apparent at the plasma membranes.¹ Moreover, in the resistant cells, the IGF-IR inhibitor AG1024 reduces ER phosphorylation and partially diminishes tumor cell growth (∼40%). Thus, despite EGFR/HER2 signaling comprising the key growth mechanism in tamoxifen resistance, such cells also appear to show some requirement for IGF-IR activation. In this context, it is noteworthy that exposure of the tamoxifen-resistant cells to IGF-II not only leads to the predicted increase in phosphorylation of IGF-IR, but also to a reproducible secondary increase in EGFR activation, events that are blocked by AG1024. Indeed AG1024, as well as an IGF-II neutralizing antibody, partially reduces EGFR signaling in nonprimed cells. These preliminary data identify a previously unrecognized functional interaction in resistant cells between IGF-IR and EGFR signaling that is unidirectional because EGF-primed events appear unaffected by AG1024 challenge. Interestingly, IGF-IR is permissive for EGFR signaling in other cell systems (45), and EGFR blockade is reported to abrogate IGFR-driven MAPK signaling to promote apoptosis of mammary epithelial cells (46). IGF-IR transactivates EGFR via metalloprotease-dependent release of the EGFR ligands amphiregulin or heparin-binding-EGF in several tumor cell types (47). Furthermore, IGF-IR has been reported to activate EGFR via production of TGF-α in colon carcinoma cells (48). However, it remains controversial whether IGF-IR/EGFR cross-talk in breast cancer cells involves metalloprotease-mediated EGFR ligand release (46, 49). A further mechanism implicated in this cross-talk in breast cancer cells is physical association of IGF-IR with erbB receptors, where IGF-IR acts to directly phosphorylate of the erbB receptor (50). The molecular biology underlying this interaction within tamoxifen-resistant cells is currently under intense investigation by our group.

In an attempt to further dissect the relative importance of EGFR and IGF-IR signaling to tamoxifen resistance, we have sought to investigate the reversibility of the growth-inhibitory effects of AG1024 and gefitinib in the tamoxifen-resistant cells by addition of EGFR and IGF-IR ligands, respectively. We have observed that whereas the effects of AG1024 are readily overridden by EGF and TGF-α, the actions of gefitinib are only partially reversible by IGF-I or -II. These data suggest that IGF-IR signaling (like ER) may play a supportive role to the EGFR/HER2 axis in directing growth of tamoxifen-resistant cells. Although controversial, some importance for IGF-IR signaling in facilitating tamoxifen-resistant growth has been reported previously in vitro (51, 52). Furthermore, gene transfer studies to overexpress IGF-II, IGF-IR, and insulin receptor substrate-1 have demonstrated relationships with estrogen independence (9, 53, 54, 55). In addition, IGF-IR up-regulation has been reported to occur after long-term estrogen deprivation (10, 23).

Although development of endocrine resistance pathways clearly enables tumor growth in the absence of steroid hormone or presence of an antiestrogen, little is known about how this event might impact on other important biological features of the cancer that are central to disease progression and spread. Such knowledge is likely to be key to understanding the adverse patient prognosis associated with development of resistance. In this context, our research findings have clearly established that the noninvasive breast cancer cell line MCF-7 gains metastatic potential on development of tamoxifen resistance (56). Some estrogen-independent models (including those generated via heregulin transfection) have also been reported to gain invasive capacity (57, 58). The significant (4-fold) increase in motility and invasion apparent in our tamoxifen-resistant cells can be explained, in part, by their enhanced EGFR signaling. Thus, TGF-α-induced activation of the EGFR signaling pathway greatly enhances invasion and motility of tamoxifen-resistant cells, whereas EGFR inhibition using gefitinib reduces their basal motility and invasion by ∼80% and 50%, respectively. These data suggest that cell motility is modulated primarily by EGFR. Furthermore, gefitinib treatment results in marked inhibition of ERK1/2 MAPK activity, a signaling intermediate known to play a key role in regulating this cellular end point (59). Gefitinib has previously been shown to reduce invasive capacity and motility of several other cell types (60). However, it appears likely that EGFR is supplemented by additional mechanisms to promote enhanced invasion in tamoxifen resistance because only partial inhibition of invasion is achieved by EGFR blockade. ER is a key contributory factor because challenge of the tamoxifen-resistant cells with fulvestrant reduces their invasion by approximately 70%.

In total, the above in vitro data are highly supportive of the notion that altered growth factor signaling, notably EGFR/HER2/IGF-IR, may make a significant contribution to the development of antiestrogen resistance and insensitivity, with these components differentially impinging on additional features of disease progression. Such signaling (and its interaction with ER) is likely, therefore, to be promising as a target for novel therapeutic strategies to treat, delay, or even prevent the development of these conditions, hopefully improving patient survival. To assess the potential impact of anti-growth factor therapies and to provide essential preclinical data in support of clinical trials with such agents, we have systematically evaluated multiple signal transduction inhibitors including gefitinib, trastuzumab (Herceptin), and AG1024 in our antiestrogen-resistant models.

Gefitinib is a potent small molecule EGFR-selective tyrosine kinase inhibitor that has previously been demonstrated to be highly effective in blocking the growth of several EGFR-positive tumor cell lines in vitro (61, 62). It is a nonpeptide anilinoquinazoline that selectively inhibits ATP binding to the tyrosine kinase domain of EGFR and is currently demonstrating promise in preclinical and clinical studies examining a number of cancer types (61, 62, 63, 64). Gefitinib inhibits EGFR tyrosine kinase activity at concentrations at least 100-fold lower than those of many other kinases tested, notably including HER2 (64). In line with its action as a competitive inhibitor of ATP binding to EGFR, gefitinib has been shown to prevent autophosphorylation of EGFR in several cultured tumor cell lines, resulting in inhibition of the activation of key downstream signaling molecules perceived to be central to proliferation and cell survival (61, 62, 64). Significantly, in our breast cancer models of tamoxifen or fulvestrant resistance, gefitinib efficiently blocked EGFR autophosphorylation and activation of ERK1/2 MAPK and Akt under both basal and EGF ligand-primed conditions (2, 7, 8). In each instance, cell growth was markedly inhibited, contrasting the relative lack of effect of this drug on the growth of the parental endocrine-responsive MCF-7 cells. In agreement with these data, van Agthoven et al. (21) have demonstrated that an EGFR-blocking antibody can inhibit proliferation of antiestrogen-resistant ZR-75-1 cells. The increase in cellular expression of EGFR generated by antiestrogens thus appears to provide a promising molecular target for effective treatment of endocrine-resistant and endocrine-insensitive phases of the disease.

Extensive profiling of EGFR, its important tumor ligand TGF-α, and its downstream ERK1/2 MAPK signaling in breast tumor specimens indicates that this concept may prove relevant to clinical disease. Thus, increases in these elements have been associated with de novo antiestrogen resistance and estrogen receptor negativity, as well as with adverse clinicopathology, metastasis, and shortened relapse-free survival in breast cancer patients (24, 65, 66, 67, 68, 69, 70, 71). We have furthermore observed that elevated tumor epithelial expression of both TGF-α and EGFR and activation of ERK1/2 MAPK are detectable on tamoxifen relapse using highly sensitive immunocytochemical assays (72). Other groups have demonstrated that additional signaling molecules shown to be EGFR regulated in our tamoxifen-resistant model, notably PKCα and Akt, associate with tamoxifen failure and metastasis, respectively, in clinical disease (73, 74). Given the apparent clinical relevance of EGFR and its signaling, several Phase II studies have been initiated examining gefitinib monotherapy in tamoxifen-resistant breast cancer. Preliminary data indicate that the drug has an acceptable tolerability profile in patients, with tumor remissions and disease stabilization recorded (75).

Although the above in vitro data with gefitinib clearly demonstrate a central role for EGFR signaling in the development of antiestrogen resistance, it is equally evident that phosphorylation of HER2, the favored heterodimerization partner of the EGFR, is also relevant (8). Increased HER2 has invariably been clinically associated with an aggressive tumor phenotype and poorer survival (76) and, controversially, with tamoxifen resistance (65, 68, 77, 78, 79, 80). Recent neoadjuvant studies have shown that the antiproliferative effect of tamoxifen is inferior in HER2-positive/ER-positive tumors (80). Moreover, amplification of the c-erbB2 gene and increases in HER2 status have been reported on relapse with antiestrogen treatment (81, 82). We have accordingly examined the role of HER2 in antiestrogen-resistant cell growth using trastuzumab. This is a humanized HER2-directed antibody therapy that inhibits the growth of many HER2-positive cancer cell lines and promotes tumor remission in approximately one-third of clinical metastatic breast cancers overexpressing HER2 by gene amplification (83). We have noted that trastuzumab is effective in inhibiting the growth of our acquired tamoxifen-resistant subline, in marked contrast to its lack of significant effect on the parental endocrine-responsive cells (2, 8). Complementary data have been obtained by Kurokawa et al. (13) and Witters et al. (84, 85), who report efficacy of the HER2 tyrosine kinase inhibitor AG1478 and trastuzumab, respectively, in MCF-7 models of tamoxifen resistance derived by stable transfection of HER2 cDNA and in BT-474 cells that overexpress HER2. Our in vitro data indicate that (a) the role of HER2 in growth regulation is extremely limited in parental endocrine-sensitive cells and (b) autocrine activation of EGFR in tamoxifen-resistant cells recruits HER2 receptor as an essential heterodimerization partner to direct cell growth.

In contrast to the largely selective inhibitory effects of gefitinib or trastuzumab for tamoxifen-resistant cells, AG1024 appears highly growth inhibitory to both the parental MCF-7 and tamoxifen-resistant cell lines.¹ This sensitivity is reflective of the expression and activation of the IGF-IR detected in these cell types and of the coupling of IGF-IR to EGFR signaling in the tamoxifen-resistant subline. As stated above, previous model system studies have linked IGF-IR signaling with resistance to tamoxifen and to estrogen withdrawal, with growth of a tamoxifen-resistant cell line, MCF 7/5-23, also significantly inhibited by an IGF-IR monoclonal antibody, αIR-3 (52). Clinically, whereas Railo et al. (86) have associated increased IGF-IR expression with shortened disease-free survival in ER-negative patients, such data are controversial because IGF-IR in breast cancer appears in general to be a favorable prognostic indicator (87, 88). However, it is notable that there are few data examining IGF-IR expression in relation to antiestrogen resistance. Moreover, no studies have as yet examined activation of this receptor or indeed monitored its expression/activation during treatment and at the time of antiestrogen relapse. Such studies are ongoing in our own laboratories, using phospho-specific antibodies. Interestingly, higher levels of the key IGF-IR signaling element insulin receptor substrate-1 have been reported to predict for a higher incidence of recurrence in ERα-positive patients (89). Moreover, Akt is predictive for disease relapse (73) and associated with ER negativity (90), whereas elevated IGF-IR is associated with radioresistance and local recurrence after radiation and lumpectomy in clinical breast cancer (91). In total, these data imply some relationship between IGF-IR signaling and disease progression in the clinic.

In common with antiestrogen treatment of endocrine-responsive cells, responses to gefitinib, trastuzumab, or AG1024 in antiestrogen-resistant cells are incomplete, with a proportion of cells persisting during the responsive phase, and stable anti-growth factor-resistant sublines developing within 3–6 months.¹ These in vitro data are mirrored by clinical experience with trastuzumab in HER2-overexpressing breast cancers, where a proportion of patients are de novo resistant, and most responsive tumors acquire resistance within 12 months (92, 93). Similar data are emerging for gefitinib in breast cancer, where again resistance can be apparent de novo, whereas disease relapse can occur subsequent to the responsive phase (75, 94). No clinical data as yet exist examining the clinical response/resistance profile on IGF-IR inhibition in breast cancer, although examination of IGF-IR antisense in additional in vivo cancer models indicates an initial delayed tumor formation with decreased metastatic capacity and, ultimately, regrowth (95). Characterization of our dually tamoxifen- and gefitinib-resistant breast cancer cells has shown that although they continue to express EGFR, there is no evidence of activation of this receptor.¹ Moreover, these cells are unresponsive to exogenous EGF or TGF-α. Compared with the tamoxifen-resistant cells, our doubly resistant cells, although slow-growing, demonstrate a further increase in their invasive behavior. Our initial screen of other signaling pathways within this cell line has revealed appreciable IGF-II expression and increased IGF-IR phosphorylation. The cells are also readily stimulated to grow by IGF-I or -II and exhibit increased sensitivity to growth inhibition by the IGF-IR-selective agent AG1024 in comparison with their tamoxifen-resistant parental cells. Furthermore, they show increased basal levels of activated Akt-1 and PKCδ, signaling molecules that are again highly stimulated by IGF-I and IGF-II challenge and inhibited by AG1024. Our preliminary examination of a dually fulvestrant- and gefitinib-resistant cell line has similarly revealed that IGF-IR and its activation are increased and that the subsequent signaling can be further primed by exogenous IGFs. Taken together, these observations point to a central importance for IGF-IR signaling in the growth of gefitinib-resistant breast cancer cell lines. Interestingly, Bianco et al. (96) have recently observed an association between elevated Akt activity and de novo gefitinib resistance in vitro that in their cell lines appears to be permitted by loss of PTEN. There is also an early indication that elevated Akt may similarly be associated with de novo gefitinib resistance in clinical breast cancer (94). Moreover, IGF-IR, Akt, and PKCδ have also been linked to trastuzumab resistance in vitro (93, 97, 98, 99). Indeed, Lu et al. (93) demonstrated that the IGF-IR antibody αIR3, or IGF-BP3, abrogates trastuzumab-resistant growth, and they have tentatively suggested that trastuzumab antagonism by IGF-I signaling may occur via targeting of p27^KIP1 to the proteasome degradation machinery (100).

In light of the above signaling data and the undesirable invasive phenotype of our antiestrogen- and gefitinib-resistant cell lines, we have attempted to delay or prevent development of these resistant states by rationally combining drugs that anticipate and abrogate the recruited resistance mechanism.

Our initial studies involved cotreating anti-hormone-responsive MCF-7 cells with tamoxifen and gefitinib, where gefitinib is used in anticipation of the cells adopting EGFR signaling for their tamoxifen-resistant growth (101, 102). Significantly, gefitinib blocked the tamoxifen-induced increase in EGFR and its phosphorylation, together with the downstream activation of ERK1/2 MAPK and Akt. Indeed, not only did gefitinib substantially repress proliferative activity of the cotreated cells, but its presence also led to a progressive and profound loss in cell number due to a marked increase in apoptosis, with no viable cells remaining by week 12 in culture. As such, cotreatment efficiently improved the inhibitory activity of tamoxifen and prevented evolution of resistance. Similarly, fulvestrant and gefitinib cotreatment showed superior antitumor activity versus single-agent therapy (7, 101, 102). These studies suggest that the antiestrogen-induced increase in EGFR maintains a basal level of cell growth and is a key compensatory survival mechanism. The observations of Yarden et al. (15) are confirmatory, where blocking EGFR with a neutralizing antibody induced a 3-fold increase in apoptosis in estrogen-deprived BT474 cells. Data from Massarweh et al. (103) have similarly demonstrated that gefitinib is able to delay resistance to an estrogen deprivation strategy and significantly improve the inhibitory activity of tamoxifen in HER2-transfected MCF-7 cells in vivo. As such, clinical trials examining gefitinib cotreatment with anti-hormonal strategies are proposed in breast cancer.

More recently, we have begun to investigate the effectiveness of combining gefitinib and AG1024 in tamoxifen-resistant cells, where AG1024 is used in anticipation of the cells adopting IGF-IR signaling for their gefitinib-resistant growth. In this model, the combined targeting of EGFR and IGF-IR promotes a complete cell loss by 6 weeks, thus again effectively blocking evolution of therapeutic resistance and confirming that IGF-IR is a key compensatory mechanism acquired during gefitinib treatment. Cotargeting HER2 and IGF-I receptors using trastuzumab and a dominant negative IGF-IR expression strategy has also been reported to cause synergistic inhibition of growth in HER2-overexpressing, tamoxifen-resistant breast cancer cells (100).

Because our studies have been performed in breast cancer cell model systems, they are unlikely to reveal the sole mechanisms that limit therapeutic efficacy and drive resistance in the clinic. Nevertheless, they clearly demonstrate that alterations in growth factor signaling elements have considerable potential to influence cellular and growth responses to antiestrogens as well as anti-growth factors, allowing a proportion of cells to survive and develop resistance with an increasingly aggressive phenotype. As such, these data establish a strong rationale for (a) targeting of these elements in resistance and (b) use of “intelligent” combination therapies early during the responsive phase of drug treatment to delay or prevent development of resistant states. It is hoped that such novel therapeutic strategies will prove clinically applicable in the future and will significantly improve the survival of breast cancer patients.

Dr. Mina Bissell: You never showed a difference in tolerance to these drugs between a nonmalignant and a malignant cell. When you combine these therapies, what happens in nonmalignant cells? Do they die as well?

Dr. Nicholson: We have never looked at that. We have looked in other breast cancer cell lines, such as T47D, for all these combined effects, and in each instance we do see a substantial improvement in cell kill.

Dr. Bissell: We find in our three-dimensional model that combination therapy is really the thing that works, but when we use the nonmalignant cells with the same combination, then we can choose combinations that would affect one as opposed to the other, because after all you don’t want to kill the patient with the therapy.

Dr. Nicholson: I would absolutely agree. What we are using are model systems, and we are basically trying to develop some thoughts that might stimulate the pharmaceutical companies to get into clinical trials with these novel types of treatment. There is obviously no guarantee in the clinical situation that these are the pathways one needs to look at, but I think the models do illustrate the potential of the approach. The old approach of targeting one pathway is probably not sufficient in the majority of patients, especially as the disease becomes more advanced. When we get the cocktail of drugs right and when we use them as early as is feasible, the hope is that they will significantly improve patient outlook.

Dr. Myles Brown: To test these hypotheses, the clinical paradigm has to be shifted to biopsying people at relapse. Treatment based on initial biopsy and initial status of these pathways is insufficient if we want to make any progress at the resistant stage.

Dr. James Ingle: When we designed clinical trials 30 years ago, we required confirmation of the ER status before they went on trial. We found that only 30% had disease accessible and in sufficient quantity to do the biochemical assays. Now we can do it with a needle biopsy. We need to reexamine the data on these molecular markers. We assume in clinical practice that the HER2 status of the primary is the HER2 status forever. That maybe a terrible misassumption, and the same is true with ER status.

Dr. Jeffrey Green: Probably all protocols should include an array analysis, so that you can really consider what is the complex of the gene expression in relation to some of these variables. Has anyone tried to correlate the available array data with ER status, response, and some of these additional growth factor pathways? For instance, can you take the data that predict a poorer response and show that there is an abnormal regulation of some of these other growth factor pathways and so, with that information, begin multiple combination therapies at the very start?

Dr. Nicholson: We are currently using our model systems to study the patterns of gene expression that occur during therapy and on development of resistance, and we would hope that using this approach we will build up the very data you are asking for. Additionally, however, we also need to take the same approach for protein expression profiling.

Dr. Carlos Arteaga: The fact that these pathways are up-regulated in the tamoxifen-resistant cells makes a lot of sense from a signaling and cancer biology perspective, but it would be nice to have some hint of a validation in cancer tissues treated with the SERM (selective estrogen receptor modulator).

Dr. Matthew Ellis: We are doing what you might call dynamic gene expression studies in the neoadjuvant setting. I would underscore the importance of doing this in the advanced disease setting; in the neoadjuvant setting, we don’t really look at acquired resistance because the tumors are taken out at the moment of maximal response.

Dr. Brown: You may have a very good response to therapy or you may have none. Getting those tumor samples is obviously worth doing. Getting tumor at relapse is almost never done in solid tumors in clinical practice, whereas it is routine in heme malignancies. We really need to be doing that.

Dr. Richard Santen: Have you looked at your tamoxifen-resistant models to see whether you can get tumor regression, either in vitro or in vivo, by giving high doses of estrogen?

Dr. Nicholson: No. We do, however, know that if you block tumor cell growth with fulvestrant, then reintroduce estradiol, you can get a good growth response. In this instance it may be possible that if we gave higher doses of estradiol that they might be growth inhibitory. I will try that.