Fax: (717) 531-5076
Serum epidermal growth factor receptor/HER-2 predicts poor survival in patients with metastatic breast cancer
Article first published online: 17 OCT 2006
Copyright © 2006 American Cancer Society
Volume 107, Issue 10, pages 2337–2345, 15 November 2006
How to Cite
Souder, C., Leitzel, K., Ali, S. M., Demers, L., Evans, D. B., Chaudri-Ross, H. A., Hackl, W., Hamer, P., Carney, W. and Lipton, A. (2006), Serum epidermal growth factor receptor/HER-2 predicts poor survival in patients with metastatic breast cancer. Cancer, 107: 2337–2345. doi: 10.1002/cncr.22255
- Issue published online: 8 NOV 2006
- Article first published online: 17 OCT 2006
- Manuscript Accepted: 17 AUG 2006
- Manuscript Received: 30 JUN 2006
- metastatic breast cancer;
- predictive factor;
- prognostic factor;
Epidermal growth factor receptor (EGFR, HER-1, and erbB1) is overexpressed in primary breast cancer and had been identified as a poor prognostic factor.
Pretreatment serum EGFR levels were quantified by using an enzyme-linked immunoadsorbent assay in a Phase III first-line trial of letrozole and tamoxifen and were correlated with patient outcomes.
Serum EGFR levels in a control group of 117 healthy, postmenopausal women measured 64.1 ± 13.3 ng/mL (mean ± standard deviation). Using a cutoff EGFR level of 44.1 ng/mL from the control group (5% nonparametric method), 53 of 535 patients (10%) had decreased serum levels of EGFR. Patients with decreased serum EGFR had no significant difference in objective response rate (ORR), clinical benefit rate (CBR), time to progression (TTP), or time to treatment failure (TTF); however, they did have significantly reduced survival compared with patients who had normal serum EGFR levels (median survival, 23.3 months vs. 30.9 months; P = .007). A combined analysis of pretreatment serum EGFR and HER-2 yielded no additional predictive information for ORR, CBR, TTP, or TTF compared to serum HER-2 alone. However, in the current analysis, a subgroup of patients who had decreased serum EGFR and normal serum HER-2 was identified (n = 39 of 535 patients; 7.3%) that had significantly reduced survival compared with patients who had normal serum levels of both EGFR and HER-2 (median survival, 23.5 months vs. 37.1 months; P = .005). In multivariate analysis, a decreased serum EGFR level remained a significant independent prognostic factor for decreased survival (hazards ratio, 1.58; P = .007).
In patients who had metastatic breast cancer, decreased serum EGFR/normal serum HER-2 predicted shorter survival compared with patients who had normal levels of serum EGFR/HER-2. This patient subgroup deserves further study to assess their response to and selection for anti-EGFR-directed therapies. Cancer 2006. © 2006 American Cancer Society.
The epidermal growth factor receptor (EGFR/ErbB1) is 1 of 4 members of the ErbB family of receptor tyrosine kinases that includes HER-2/neu/erbB2, HER-3/ErbB3, and HER-4/ErbB4.1–3 Similar to other growth factor receptors,4 EGFR has 3 major domains: an intracellular protein tyrosine kinase, an extracellular glycosylated ligand-binding domain, and a hydrophobic transmembrane anchor that connects the other 2 domains.5, 6 Activation of EGFR ultimately results in cell cycle progression through various mechanisms, including initiation of Src family kinases, phosphatidylinositol-3 kinase (PI3 kinase), and mitogen-activated protein kinase.7, 8
Numerous mechanisms of EGFR activation have been elucidated. Cohen et al. first described the role of epidermal growth factor (EGF) and transforming growth factor-α (TGF-α) in the direct ligand-dependent activation of EGFR9 that others subsequently showed was a result of homodimerization and heterodimerization of the extracellular domain, leading to autophosphorylation of the cytoplasmic tyrosine kinase.10 Other ligand-dependent activators include amphiregulin, heparin-binding EGF-like growth factor, and β-cellulin.11 More recently, ligand-independent mechanisms were identified and may be possible reasons for resistance to anti-EGFR therapy.8 Such mechanisms include extracellular integrin-induced EGFR activation12 and constitutive pathway activation downstream of EGFR, such as through Src13 or PI3 kinase mutation.14
Much attention has been given to tumor EGFR levels, as measured by immunohistochemistry (IHC), in relation to clinical outcome and therapeutic response. EGFR overexpression has been documented in gastric, esophageal, pancreatic, nonsmall cell lung, prostate, squamous cell, breast, ovarian, cervical, glioma, and bladder cancers.2, 15–24 Elevated tissue EGFR reportedly has prognostic significance in several solid tumors. Tissue EGFR overexpression in patients with squamous cell carcinoma of the head and neck23, 24 and in patients with epithelial ovarian cancer25 was correlated with a poor prognosis and disease progression. However, the prognostic significance of elevated primary tumor EGFR in patients with breast cancer is controversial. Some reports have described a correlation between EGFR level and clinical outcome,26, 27 whereas other reports have refuted this correlation.28 It has been suggested that confounding variables inherent to these studies, such as adjuvant therapy, are responsible at least in part for the discrepancies.29 In a summary report of 5232 patients with breast cancer, the authors concluded that no definitive correlation has been established to date between tissue EGFR levels and clinical outcomes, although they noted that most of the studies they examined did report a weak association between EGFR-positive breast cancers and reduced overall survival.20
Serum EGFR has been the topic of relatively few reports, and the results from those studies are conflicting. In 2 studies, no significant difference was reported in serum EGFR levels between a healthy control group and patients with lung cancer30 or patients with squamous cell carcinoma of the head and neck.31 Some investigators identified significantly elevated levels of serum EGFR both in patients with cervical carcinoma and in patients with carcinoma in situ compared with healthy controls,32 whereas other investigators reported significantly elevated levels of serum EGFR in patients who had lymph node-positive lung cancer compared with the levels in patients who had lymph node-negative lung cancer, but no significant difference was reported in serum EGFR levels between patients with lung cancer and a control group of patients with nonmalignant disease.33 Conversely, our group and others have identified significantly decreased serum EGFR levels in patients with pancreatic, gastric, colon, lung, renal, prostate, ovarian, and breast cancers.34–37
We previously reported that patients with metastatic breast cancer who had elevated pretreatment serum HER-2 levels had a decreased response to first-line hormone therapy and decreased survival.38 The objective of the current study was to determine the predictive and prognostic significance of serum EGFR levels in 535 women with advanced breast cancer who were enrolled in a Phase III, first-line, hormone therapy trial in patients with metastatic breast cancer.
MATERIALS AND METHODS
In the original clinical trial, 907 patients were assigned randomly to receive either letrozole at a dose of 2.5 mg given once daily (453 patients) or tamoxifen at a dose of 20 mg given once daily (454 patients), and the patient inclusion/exclusion criteria, clinical and radiologic assessments, and efficacy assessments that were used in this study were identical to those published previously.39 In the current study, pretreatment serum samples that were obtained within 14 days before the initiation of therapy were available from 535 patients for both HER-2 and EGFR levels. This project was approved by the Institutional Review Board of the Hershey Medical Center of Pennsylvania State University.
Briefly, postmenopausal women with histologically or cytologically confirmed breast cancer and with locally advanced (Stage IIIB according to the 1992 American Joint Committee on Cancer criteria), or women with locoregionally recurrent disease that was not amenable to treatment by surgery or radiotherapy, or women with metastatic disease were eligible for the study. Patients were required to have tumors with positive estrogen receptor status and/or positive progesterone receptor status or with the status of both receptors unknown. Patients who were treated previously with 1 regimen of chemotherapy for advanced disease were allowed to participate in the study provided that they had objective evidence of disease progression within the 3 months before study enrollment.
Clinical and Radiologic Assessments
A complete tumor assessment was performed at baseline, and areas that were identified as positive for disease at baseline were monitored every 3 months for response assessment according to International Union Against Cancer criteria.40 All patients were monitored for survival at least every 6 months after the termination of study treatment.
Primary Efficacy Assessments
In this study, the primary efficacy endpoint was time to progression (TTP). TTP was defined as the interval between the date of randomization and the earliest date of disease progression. An increase ≥25% in measurable lesions (a single lesion or the sum of products of all measurable lesions), an estimated increase ≥25% in existing assessable or nonmeasurable/nonassessable disease, or the appearance of new lesions were considered disease progression. Discontinuation of treatment with documented evidence of clinical deterioration because of breast cancer or death from breast cancer or from an unknown cause (with documented evidence of clinical deterioration caused by breast cancer) while a patient was receiving treatment or within 6 weeks of treatment discontinuation also constituted disease progression.
Secondary Efficacy Assessments
Secondary endpoints of the current study included the overall objective tumor response rate (ORR), the duration of overall response, the rate of clinical benefit, the duration of clinical benefit, the time to treatment failure (TTF), and overall survival. ORR was defined as the proportion of patients who had a complete response (CR) or a partial response (PR) confirmed by a second evaluation at least 1 month (but generally 3 months) later. Overall response was evaluated at 3-month intervals after the initiation of therapy. The rate of clinical benefit was defined as the proportion of patients who achieved a confirmed objective response (CR or PR) or who had stabilization or no change lasting for ≥24 weeks.
The TTF was defined as the interval between the date of randomization and the earliest date of disease progression, withdrawal of study treatment for any reason, withdrawal of consent, loss to follow-up, or death from any cause. Overall survival was defined as the time from the date of randomization to the date of death. It was assumed that patients who were lost to follow-up had died.
Blood was drawn by forearm venipuncture 14 days prior to the initiation of first-line hormone therapy. Those blood samples were centrifuged and stored at −70°C.
EGFR Enzyme-Linked Immunoadsorbent Assay
The Oncogene Science EGFR Microtiter enzyme-linked immunoadsorbent assay (ELISA) (Oncogene Science/Bayer Corporation, Cambridge, MA) was employed according to the manufacturer's protocol. Briefly, 100 μL of 1:50 diluted sample or EGFR standard were incubated in duplicate for 1.5 hours in wells that were precoated with antihuman EGFR mouse monoclonal capture antibody. After washing, 100 μL of alkaline-phosphatase labeled anti-EGFR detector antibody were added to each well. After a 30-minute incubation, the wells were washed, and 100 μL of substrate (paranitrophenylphosphate) were added to each well for 45 minutes. After reaction termination, optical density was measured at 405 nm. The manufacturer states that the EGFR Microtiter ELISA has a functional range of from 0.25 ng/mL to 300 ng/mL, an interassay coefficient of variability (CV) of 8.8%, and an intraassay CV of 6.6%. In this study, the manufacturer-provided internal controls yielded an interassay CV of 6.7% and an intraassay CV of 4.0%. The EGFR capture antibody specifically recognizes the 110-kilodalton (kD) extracellular domain fragment of EGFR and does not cross-react with HER-2/neu or inhibit EGF binding.
The HER-2 ELISA protocol and serum HER-2 results for this Phase III clinical trial have been reported previously.38
The time-to-event data (e.g. TTP, overall survival) were analyzed by using Cox proportional hazards regression, and the time-to-event statistics (e.g. median TTP) were estimated by using the Kaplan–Meier product-limit method. The overall objective response (confirmed CR or confirmed PR) was analyzed by using a logistic regression method. For the main analyses, 4 groups were formed according to HER-2 expression levels (elevated vs. normal) and EGFR expression levels (decreased vs. normal). The reference group was HER-2 normal/EGFR normal. Multivariate analyses examining the effects of several baseline covariates on the endpoints TTP, overall survival, and ORR were conducted. All analyses were completed with SAS software (version 8.2; SAS Institute, Cary, NC) on the Novartis UNIX platform (Novartis Pharma AG).
This multinational, double-blind, double-dummy, randomized, Phase III trial comparing letrozole with tamoxifen as first-line treatment for advanced breast cancer in postmenopausal women had shown previously that letrozole was superior to tamoxifen, as reported by Mouridsen et al.39 Subsequently, we reported that serum HER-2 was a significant, independent, adverse predictive and prognostic factor for first-line hormone therapy.38 We reported that patients who had elevated serum HER-2 had significantly decreased ORR, CBR, TTP, TTF, and overall survival compared with patients who had normal serum HER-2.
The treatment arms in this study were divided equally (274 of 535 patients received letrozole and 261 of 535 patients received tamoxifen). Patients were divided into 4 subgroups, depending on their serum levels of EGFR and HER-2. In the 4 patient subgroups (normal EGFR/normal HER-2, normal EGFR/elevated HER-2, decreased EGFR/normal HER-2, and decreased EGFR/elevated HER-2), there were no significant differences in assigned treatment, hormone receptor status, lung metastases, or prior adjuvant tamoxifen therapy. There was a significantly greater incidence of bone metastases (79%) in the decreased EGFR/elevated HER-2 group and a significantly lower incidence of liver metastases (7%) in the normal EGFR/normal HER-2 group. The decreased EGFR/normal HER-2 group had a significantly greater proportion of patients age ≥70 years (56%) and a significantly lower proportion of patients who had received prior adjuvant chemotherapy (8%) (Table 1).
|Characteristic||No. of patients (%)|
|Normal HER-2||Elevated HER-2|
|Decreased EGFR (n = 39)||Normal EGFR (n = 338)||Decreased EGFR (n = 14)||Normal EGFR (n = 144)|
|Femara assigned||22 (56)||167 (49)||9 (64)||76 (53)|
|Receptor status positive||27 (69)||213 (63)||11 (79)||100 (69)|
|Bone metastases present||21 (54)||155 (46)||11 (79)||87 (60)|
|Liver metastases present||4 (10)||25 (7)||2 (14)||34 (24)|
|Lung metastases present||8 (21)||92 (27)||2 (14)||34 (24)|
|Prior adjuvant tamoxifen||7 (18)||70 (21)||4 (29)||22 (15)|
|Age ≥70 y||22 (56)||104 (31)||6 (43)||37 (26)|
|Adjuvant chemotherapy||3 (8)||88 (26)||4 (29)||36 (25)|
|Therapeutic chemotherapy||7 (18)||29 (9)||3 (21)||14 (10)|
Serum EGFR levels were obtained from premenopausal (n = 78 women) and postmenopausal (n = 117 women) healthy female controls. Significantly lower serum EGFR levels were observed in the postmenopausal healthy female control group (mean serum EGFR, 64.1 ± 13.3 ng/mL) compared with the premenopausal healthy female control group (mean serum EGFR, 69.3 ± 9.2 ng/mL; P = .003). Therefore, only the postmenopausal female controls were used as the control group, because this Phase III trial included only postmenopausal women with metastatic breast cancer. In a report on menopausal status and healthy control serum EGFR levels, Baron et al. also reported that serum EGFR levels were significantly lower in postmenopausal women.41 The 117 healthy postmenopausal control group in our study had a mean± standard deviation serum EGFR level of 64.1 ± 13.3 ng/mL (range, 39.5–117.1 ng/mL). A lower limit of normal serum EGFR cutoff level of 44.1 ng/ml was established by using the 5% nonparametric method.
Serum EGFR Univariate Analysis
Using a serum EGFR lower limit of normal cutoff level of 44.1 ng/mL, as established from the control data (5% nonparametric method), 53 of 535 patients (10%) had significantly decreased serum EGFR. Patients who had significantly decreased serum EGFR had no significant difference in TTP (median, 6.5 months) compared with patients who had normal serum EGFR (median TTP, 8.7 months) (hazards ratio [HR], 1.11; P = .52). Similarly, decreased serum EGFR status did not correlate with a significant difference in TTF (median, 5.8 months in patients with decreased serum EGFR vs. 6.7 months in patients with normal serum EGFR) (HR, 1.15; P = .37). Furthermore, the ORR did not differ significantly between the 2 groups (ORR, 25% in patients with decreased serum EGFR vs. 25% in patients with normal serum EGFR) (odds ratio [OR], 0.88; P = .70). Similarly, patients who had decreased serum EGFR had no significant difference in CBR (36%) compared with patients who had normal serum EGFR (CBR, 44%) (OR, 0.7; P = .24). However, the 1 outcome that revealed a significant difference between patients with decreased serum EGFR and patients with normal serum EGFR was overall survival. Patients with decreased serum EGFR had a median survival of 23.3 months, whereas patients with normal serum EGFR had a median survival of 30.9 months (HR, 1.58; P = .007) (Table 2).
|Variable||Normal EGFR (n = 482)||Decreased EGFR (n = 53)|
|Deaths (%)||292 (61)||40 (75)|
|Median TTD (95% CI), mo||30.9 (28.6, 34.6)||23.3 (12.3, 32.0)|
|Hazards ratio (95% CI)||1.58 (1.14, 2.20)|
Serum EGFR/HER-2 Combined Analysis
It was reported previously that elevated serum HER-2 was a significant, independent, adverse predictive and prognostic factor to first-line hormone therapy.38 Combined analysis of pretreatment serum EGFR and HER-2 levels yielded no additional predictive information compared with serum HER-2 levels alone for clinical response to first-line hormone therapy as measured by ORR, CBR, TTF, and TTP (Fig. 1). The combined analysis of pretreatment serum EGFR and HER-2, however, did define a subset of patients who had decreased serum EGFR levels and normal serum HER-2 levels (Fig. 2, line 2) (n = 39 of 535 patients; 7.3%) and who had significantly reduced overall survival compared with patients who had normal serum levels of both EGFR and HER-2 (Fig. 2, line 1) (median survival, 23.5 months vs. 37.1 months) (HR, 1.77; P = .005) (Table 3) (Fig. 2). The overall survival of the decreased EGFR/normal HER-2 patient subgroup (Fig. 2, line 2) did not differ significantly from the elevated serum HER-2 subgroups (Fig. 2, lines 3 and 4).
|Statistic||Normal HER-2||Elevated HER-2|
|Decreased EGFR (n = 39)||Normal EGFR (n = 338)||Decreased EGFR (n = 14)||Normal EGFR (n = 144)|
|Deaths (%)||28 (72)||179 (53)||12 (86)||113 (78)|
|Median TTD (95% CI), mo||23.5 (16.2, 36.5)||37.1 (34.1, 47.8)||17.4 (6.8, 25.3)||20.8 (16.6, 25.7)|
|Hazards ratio (95% CI)*||1.77 (1.19, 2.65)||—||1.43 (1.17, 1.74)||1.46 (1.30, 1.65)|
In multivariate analysis, decreased serum EGFR remained a significant independent prognosticator of reduced survival (HR, 1.69; P = .003). Other significant variables for decreased survival included the presence of liver metastasis (HR, 2.33; P < .0001) and elevated serum HER-2 (HR, 1.9; P < .0001), whereas positive hormone receptor status lowered the risk of mortality (HR, 0.78; P = .03), as did adjuvant chemotherapy (HR, 0.76; P = .05) (Table 4).
|Variable||Parameter||Standard error||Chi-square||P||Hazards ratio||95% CI|
|Receptor status (positive/otherwise)||−.25005||.11604||4.644||.031*||0.78||0.62–0.98|
|Adjuvant tamoxifen (yes/no)||.08771||.14061||0.389||.533||1.09||0.83–1.44|
|Adjuvant chemotherapy (yes/no)||−.27256||.13625||4.002||.046*||0.76||0.58–1.00|
|Therapeutic chemotherapy (yes/no)||.23951||.17807||1.809||.179||1.27||0.90–1.80|
|Age ≥ 70 (yes/no)||.04131||.12382||0.111||.739||1.04||0.82–1.33|
|Bone metastases (yes/no)||.11490||.11570||0.990||.321||1.12||0.89–1.41|
|Liver metastases (yes/no)||.84411||.15436||29.905||<.0001*||2.33||1.72–3.15|
|Lung metastases (yes/no)||.07673||.13072||0.347||.556||1.08||0.84–1.39|
In this study, we observed that serum EGFR levels were reduced significantly in 10% of 535 postmenopausal women with metastatic breast cancer who were enrolled in a Phase III trial of first-line hormone therapy. This subset of patients with decreased levels of serum EGFR did not differ significantly in terms of ORR, CBR, TTP, or TTF on first-line hormone therapy. However, patients who had decreased serum EGFR had significantly reduced overall survival compared with patients who had normal serum EGFR in both univariate and multivariate analyses. In addition, combined analysis of serum EGFR and HER-2 identified a subgroup of patients with decreased serum EGFR/normal serum HER-2 who had significantly reduced survival compared with patients who had normal serum levels of both EGFR and HER-2. The use of serum EGFR for identification of this subset of patients with metastatic breast cancer (7%), therefore, is effective for stratification of overall mortality in this disease.
The results of this report reaffirm several previous studies that have reported decreased serum EGFR levels in a proportion of patients with metastatic breast cancer.34, 35 Similar to our results, Marx et al. reported that 69 of 265 patients (26%) who were enrolled in a Phase III, second-line hormone therapy trial for advanced breast cancer had significantly decreased serum EGFR levels.34 In addition, serum EGFR levels were reduced in patients with Stage III and IV epithelial ovarian cancer compared with a healthy age-matched female control group,36 and serum EGFR levels were used to distinguish healthy controls from patients who had ovarian cancer patients with 89% accuracy.37
To date, the data regarding serum EGFR levels and the prognosis for patients with various solid tumors are conflicting. Oh et al. concluded that there was no prognostic value associated with serum EGFR in cervical cancer,42 whereas Gregorc et al. identified an increasing serum EGFR level at 28 days after gefitinib treatment in patients with lung cancer as a significant indicator of disease progression and shorter progression-free survival.43 In breast cancer, our results are the first to our knowledge to demonstrate that patients with metastatic disease and decreased serum EGFR levels have reduced survival.
The source of circulating EGFR has yet to be established definitively. Several mechanisms have been suggested for the origin of serum EGFR: proteolytic cleavage of the extracellular domain of the EGFR glycoprotein and/or alternate transcription of primary messenger RNA.3 Potential explanations for decreased serum EGFR levels in a proportion of patients with metastatic breast cancer include: 1) Increased circulating TGF-α levels, which have been reported in serum from patients with breast cancer,44 may form complexes with circulating EGFR and, subsequently, may be cleared more rapidly from the circulation; and 2) an internal autocrine mechanism also may occur in which newly synthesized TGF-α binds to and activates EGFR, with the result that less EGFR would be available for proteolysis at the tumor cell surface. With either of these mechanisms, activation of the EGFR pathway may explain the decreased serum EGFR levels and the worse prognosis observed in these patients.
In this report, we identified a subgroup of patients who had normal serum HER-2 levels and decreased serum EGFR levels and had a significantly shorter survival compared with patients who had normal serum values of both HER-2 and EGFR. These results were very similar to those from a recent report of 807 patients with breast cancer in which the authors concluded that primary breast tumors with overexpression of both HER-2/neu and EGFR by IHC had the shortest survival.45 In our study, patients who had decreased serum EGFR levels deserve further study for evaluation and selection for treatment with anti-EGF receptor-targeted therapy.46 In addition, determination of serum EGFR and HER-2 levels may identify patient subgroups that will respond best to different combinations of hormone and anti-HER family growth factor receptor therapy: Patient Group 1 (normal serum EGFR/HER-2) (Fig. 2) should receive hormone therapy alone; Patient Group 2 (decreased EGFR/normal HER-2) may be treated optimally with a combination of hormone therapy and anti-EGFR therapy; Patient Group 3 (normal EGFR/elevated HER-2) would be treated best with hormone therapy and anti-HER-2 therapy (Herceptin); and, finally, Patient Group 4 (decreased EGFR/elevated HER-2) may be treated most effectively with hormone therapy and combined EGFR/HER-2 blockade, for example, with pan-HER family tyrosine kinase inhibitors. Therefore, patient-customized selection of targeted therapy may be required for optimal blockade of the estrogen receptor/HER family receptor crosstalk that is a likely cause of resistance to hormone therapy.46
- 18Role of receptors in the management of patients with breast cancer. Diagn Oncol. 1991; 1: 43–52., , .
- 34Serum EGFR in metastatic breast cancer patients. Proc Am Soc Clin Oncol. 2002; 21: 436a. Abstract., , , et al.
- 35Normal levels of serum EGFr and decreases in several cancers. Proc AACR. 2003; 43: 47. Abstract 240., , , et al.
- 40International Union Against Cancer. Manual of adult and pediatric medical oncology. In: MonfardiniS, BrunnerK, CrowtherD, et al., editors. Evaluation of the Cancer Patient and the Response to Treatment. Berlin: Springer; 1987: 22–38.