Vogelstein has stated that if the mutations observed in proto-oncogenes are what drives the neoplastic process, they should then be considered among the best markers for detecting specific tumors.1 Although this concept has become axiomatic, it has not held true for the epidermal growth factor receptor (EGFR/human epidermal growth factor receptor 1 [HER1]) or its related HER family members. The essence of this paradox as it applies to human epidermal growth factor receptor 2 (HER2) recently was reviewed by Allison.2 EGFR expression, similarly, has proven to be an unreliable positive predictive marker for “EGFR-driven” malignancies. Extensive analyses suggest that more than just technical issues confound the clinical utility of these 2 receptor biomarkers.
As a case in point, the US Food and Drug Administration has approved panitumumab and cetuximab, 2 EGFR-directed monoclonal antibodies, for the treatment of patients with advanced colorectal cancer (CRC). Paradoxically, tumor EGFR expression does not predict responsiveness to either of these antibodies, whereas mutations in genes encoding downstream effectors (ie, PTEN, BRAF, and KRAS) are negative predictors of efficacy.3 The failure of 2 recent phase 3 cetuximab trials in patients with CRC selected for both EGFR and wt K-ras expression (the N0147 and MRC COIN4 trials) illustrates the urgent need for improved methods with which to select patients for treatment with these drugs.
In this regard, we previously reported the expression of 3 alternate EGFR isoforms, which arise from naturally occurring alternative transcripts.5 One of these transcripts encodes a 90/110-kilodalton cell surface protein designated soluble epidermal growth factor receptor (sEGFR), which is expressed in normal human tissues and tumors, and also is the source of the major circulating isoform of sEGFR in human blood.6
It has been speculated that this circulating sEGFR isoform may serve as a sink or even as an alternate target for antibody-based, EGFR-directed therapeutics,7 in a manner similar to trastuzumab binding to shed sHER2 in blood.8 This proposal is supported by sequence inspection of sEGFR, which contains the epitope (subdomain III) recognized by both cetuximab and panitumumab. As shown in Figure 1,9 we directly demonstrated that both cetuximab and panitumumab recognize sEGFR using antibody concentrations far below therapeutic dosages. Because serum sEGFR concentrations range between 0.0125 and 5 mg/mL (approximately 100 to 50,000 fmol/mL) in healthy adults and may be even lower (<10 fmol/mL) in some patients with cancer,10, 11 we propose that this interaction likely occurs in vivo in patients treated with either cetuximab or panitumumab. Given that trough and peak cetuximab treatment concentrations range from 41 to 235 mg/mL (241,176 to 1,382,353 fmol/mL of immunoglobulin G), circulating cetuximab concentrations would be predicted to range from a 140,000-fold to only 5-fold molar excess relative to circulating sEGFR concentrations in treated cancer patients. Because serum sEGFR arises from a cell surface precursor that is ubiquitously expressed in normal tissues,12, 13 the effective concentration of these antibodies for the presumed (single) therapeutic target (ie, EGFR) may be even lower in most patients. In contrast, some patients with cancer exhibit baseline serum sEGFR concentrations <10 fmol/mL11; these patients would be predicted to receive a substantially greater effective dose of cetuximab. Although such antibody-sEGFR interactions may be relevant in calculating the effective dose of these therapeutic antibodies in patients with cancer, we would not predict a similar interaction between sEGFR and small-molecule inhibitors such as gefitinib or erlotinib; previous studies examining serial serum sEGFR concentrations in patients with breast cancer who were treated with gefitinib are consonant with this prediction.14
Taken together, these observations suggest that serum sEGFR may be an unanticipated “first target” of EGFR-directed antibodies in humans, in whom it may interfere with accurate pharmacokinetic/dynamic mea-surements, and also may moderate therapeutic efficacy. Moreover, the coexpression of sEGFR with EGFR on the surface of tumor cells may contribute to the lack of concordance noted between tumor “EGFR expression” (as it is currently assayed) and responsiveness to cetuximab or panitumumab. However, it is clear that tumor cells are not likely to be the major source of circulating sEGFR (unlike circulating sHER2) because several studies have shown that serum sEGFR concentrations actually increase after tumor resection and/or chemotherapy.15-17
Although the study of soluble EGFR/HER isoforms is in its infancy, we propose that the expression of these intrinsic regulators of EGFR/HER signaling adds a new level of complexity to our understanding of EGFR signal transduction that is perhaps analogous to the role of insulin-like growth factor-binding proteins in insulin/insulin-like growth factor 1 family signaling. Despite our limited understanding of the function(s) of these naturally occurring EGFR isoforms in blood and other normal tissues, the potential of these proteins to interfere with the measurement of EGFR in human tumors/tissues, as well as the therapeutic targeting of EGFR in patients with cancer, should be considered in the design of future clinical trials using EGFR-directed immunotherapeutic drugs.