Searching for Reliable Epidermal Growth Factor Receptor Response Predictors : Commentary re M. K. Nyati et al., Radiosensitization by Pan-ErbB Inhibitor CI-1033 in Vitro and in Vivo. Clin. Cancer Res., 10: 691–700, 2004.

Many years before the term “molecular targeting” was popularized in cancer therapeutics, the epidermal growth factor (EGF) and its receptor (EGFR or ErbB-1 or HER1) were recognized as important molecular regulators of cellular growth and differentiation (1, 2). Today, the EGFR represents one of our most promising molecular targets for cancer therapy (3, 4, 5, 6). The designers of anticancer molecules have fashioned a broad series of smart molecular inhibitors of EGFR signaling (Fig. 1), and, as of 1999–2000, the implementation of these new agents appeared ready to unfold in remarkable parallel to that of herceptin (3, 5, 7). After all, herceptin targeted a very close family member of EGFR, namely the HER2 (ErbB-2) receptor. Just as overexpression of HER2 in breast cancer helped predict a cohort of women most likely to respond to HER2 blockade, so too would overexpression of EGFR in a range of epithelial malignancies predict patients most likely to respond to EGFR blockade. Unfortunately, the EGFR story has proved far more elusive. The desired correlation between expression of the presumed molecular target (EGFR) and response to EGFR inhibitors has not been crisply borne out by clinical trials. Indeed, the broad series of maturing clinical data, as well as the expanding battery of preclinical data (cell line and xenograft), confirms that there is no clear relationship between EGFR expression (at least as measured by simple immunohistochemistry) and response to EGFR inhibition. The search for “true” EGFR response predictors is therefore intensifying, and many fundamental questions remain unanswered.

The report by Nyati et al. in this issue (8) touches on several active research questions in the EGFR investigative field. This study examines the impact of an irreversible pan-ErbB tyrosine kinase inhibitor (CI-1033) across a series of colon cancer cell lines in vitro and in vivo. Three themes that emerge from this work warrant brief comment. First, the authors suggest that the extent of suppression of tyrosine kinase activity by CI-1033, rather than the baseline activity level before treatment, may predict for ultimate treatment efficacy. Second, the authors confirm the capacity of CI-1033 to enhance radiation response (particularly in vivo) as has been identified for several other EGFR inhibitors (Refs. 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26; Table 1). Third, the authors suggest that selected downstream signaling molecules of the EGFR pathway (in this case phospho-extracellular signal-activated kinase [ERK] 1/2) may serve as candidate markers for predicting response to ErbB inhibition. Each of these three EGFR themes remains under intense investigation within basic laboratory, translational, and clinical research programs worldwide. As of early 2004, none of these postulates has yet been definitively established within the context of mature clinical trials. However, the rapidly enlarging spectrum of trials in progress may soon allow more definitive conclusions to be drawn.

A first EGFR research theme reflected by the Nyati article is whether the extent of suppression of EGFR signaling activity is important for ultimate clinical response. In short, the answer to this question remains unknown. To date, the most mature clinical trials data exists for so called mono-ErbB inhibitors (C225, ZD1839, OSI-774) as opposed to the dual- or pan-ErbB inhibitors (GW572016, CI-1033, and others). The mono-ErbB inhibitors are relatively EGFR-selective [particularly so for monoclonal antibodies (mAbs) over tyrosine kinase inhibitors (TKIs)], whereas the dual- or pan-ErbB inhibitors more readily influence multiple members of the ErbB receptor family. CI-1033 binds irreversibly to the receptor tyrosine kinase in distinction to several of the other EGFR-TKIs, which are reversible inhibitors. In light of the complex horizontal network of ErbB signaling systems (27), it may be that the broader downstream suppression afforded by the dual- or pan-ErbB inhibitors will provide more clinical response potency. On the other hand, clinical toxicity profiles may expand with more broad suppression of the ErbB receptor systems. In addition, it is highly likely that most advanced solid tumors derive their growth advantage from more than one aberrant molecular growth pathway. If this is true, no matter how effectively ErbB signaling pathways are blocked, tumors with alternate growth stimuli may retain essentially unperturbed growth dysregulation (28).

A second EGFR research theme reflected by the Nyati article is the potential capacity of EGFR inhibitors to enhance radiation treatment outcome. Despite an array of promising preclinical data (Refs. 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26; Table 1), there is as yet no mature clinical trial data to definitively answer this question. However, the first large-scale randomized trial to specifically address this EGFR/radiation question is anticipated to report outcome results in mid 2004. This Phase III trial randomized over 420 advanced head and neck (H&N) cancer patients to high-dose radiation plus/minus the anti-EGFR mAb C225 (Cetuximab, Erbitux) and completed enrollment in the spring of 2002 (29, 30). This represents a powerful study for the EGFR field. If the study confirms a survival advantage for the addition of C225 to radiation for advanced H&N cancer patients, this would provide strong confirmation that EGFR signaling blockade can augment radiation outcome. This would stimulate a series of EGFR/radiation studies across a spectrum of anatomical sites. However, if this trial is negative, the theme of combining EGFR inhibitors with cytotoxic agents would suffer a significant additional setback because the first large-scale Phase III trials that combined EGFR inhibitors with cytotoxic chemotherapy in advanced non-small cell lung cancer patients showed no survival advantage over that achieved with chemotherapy alone (31).

Finally, the issue of molecular predictors of response to EGFR inhibitors has remained elusive as described above. If simple quantitation of EGFR using immunohistochemistry correlated well with clinical response to EGFR inhibition, the selection of patients most likely to benefit from treatment would be very straightforward. Patients whose tumors achieved some designated threshold of EGFR expression would be selected for EGFR inhibitor therapy much in the same way that patients with estrogen receptor expression are selected for tamoxifen therapy. Because this correlation has clearly not proven true for EGFR (at least with simple EGFR immunohistochemical staining techniques), the ongoing search for predictive markers has intensified. Initially, the search has focused on a series of ErbB downstream molecules. Because ErbB receptor downstream signaling can engage RAS/mitogen-activated protein kinase pathways, PI3K pathways, and STAT signaling molecules, a broad variety of molecular predictors are being tested from the available clinical material. In addition to total and phosphorylated EGFR, these include total and phosphorylated forms of AKT, mitogen-activated protein kinase (MAPK), mitogen-activated protein/ERK (MEK), ERK, signal transducers and activators of transcription (STAT), PTEN, mTOR, and others (27, 32, 33). With an increasing appreciation of the inherent limitations of immunohistochemistry, more sophisticated pharmacogenomic approaches to explore potential EGFR correlations with clinical response are also under study (34, 35).

Now that several hundred clinical trials with various EGFR inhibitors are either complete or in progress, we are gaining leads regarding inhibitors that may “match up” best with certain anatomical tumor sites. The two major classes of EGFR inhibitor under current investigation (mAbs and TKIs) carry several simple distinctions. The mAbs are relatively large molecules that bind to the extracellular domain of the EGFR and exhibit a long half-life; thus weekly IV dosing is common. The TKIs are small, orally bioavailable molecules which interact with the ATP binding site at the cytoplasmic domain of the EGFR and exhibit a shorter half life; thus daily oral dosing is common. The large size of mAbs suggest they may not gain full access and penetration to all anatomical compartments (i.e., central nervous system), whereas this same characteristic may prevent traverse across the basement membrane of the alimentary tract, thus the minimal gastrointestinal toxicity observed with the mAb class of EGFR inhibitor. The small size of TKIs suggest they may have better access and penetration to anatomical compartments such as the central nervous system. However, this small size confers the capacity of TKIs to readily traverse basement membranes, thereby facilitating a primary dose-limiting toxicity, namely, that of diarrhea. These simple molecule size characteristics have contributed (in part) to the more intense examination and early promise of TKIs in brain tumors and mAbs in colorectal tumors for example.

Several promising EGFR research strategies are expanding rapidly in complementary directions. The search for reliable response predictors to EGFR inhibitor therapies represents perhaps the greatest current obstacle to realizing the full potential for EGFR inhibitors. At the cellular level, response profiles are now being characterized using genomic and proteomic approaches in an effort to establish unique “molecular fingerprints” for EGFR-responders versus nonresponders (36, 37, 38). At the tissue and organ level, advanced functional imaging technologies are exploring predictive capacity after “test exposure” to EGFR inhibition as a method to select patients (tumors) more likely to respond in the long run. Finally, the assumption that most advanced solid tumors derive their growth advantage from more than a single aberrant molecular growth pathway leads to the “combined molecular targeting approach” in which more than one class of inhibitor is applied simultaneously, much akin to classical multi-agent chemotherapy approaches. There are an expanding series of such combination-molecular-inhibitor approaches now underway in clinical trials (39, 40, 41, 42, 43) 

There is ample reason to believe that successful progress within one or more of the major EGFR research themes described above will afford significant advances for this field in the near future. As this occurs, the stepwise integration of specific EGFR therapies for a variety of anatomical tumor sites should follow in logical fashion. Whether administered as monotherapy in selected patients who are likely to respond, based on established molecular or imaging predictors, or in combination with radiation, chemotherapy, or other targeted molecular inhibitors, the EGFR inhibitors appear on the threshold of providing a valuable new tool in cancer therapeutics.