Epidermal growth factor receptor gene copy number in 101 advanced colorectal cancer patients treated with chemotherapy plus cetuximab

Background Responsiveness to Cetuximab alone can be mediated by an increase of Epidermal Growth factor Receptor (EGFR) Gene Copy Number (GCN). Aim of this study was to assess the role of EGFR-GCN in advanced colorectal cancer (CRC) patients receiving chemotherapy plus Cetuximab. Methods One hundred and one advanced CRC patients (43 untreated- and 58 pre-treated) were retrospectively studied by fluorescence in situ hybridization (FISH) to assess EGFR-GCN and by immunohistochemistry (IHC) to determine EGFR expression. Sixty-one out of 101 patients were evaluated also for k-ras status by direct sequencing. Clinical end-points were response rate (RR), progression-free survival (PFS) and overall survival (OS). Results Increased EGFR-GCN was found in 60/101 (59%) tumor samples. There was no correlation between intensity of EGFR-IHC and EGFR-GCN (p = 0.43). Patients receiving chemotherapy plus Cetuximab as first line treatment had a RR of 70% (30/43) while it was 18% (10/56) in the group with previous lines of therapy (p < 0.0001). RR was observed in 29/60 (48%) of patients with increased EGFR-GCN and in 6/28 (21%) in those without (p = 0.02). At multivariate analyses, number of chemotherapy lines and increased EGFR-GCN were predictive of response; EGFR-IHC score, increased EGFR-GCN and number of chemotherapy lines were significantly associated with a significant better PFS. Response to therapy was the only prognostic predictive factor for OS. In the 60 patients analyzed for k-ras mutations, number of chemotherapy lines, increased EGFR-GCN and k-ras wild type status predicted a better PFS. Conclusion In metastatic CRC patients treated with chemotherapy plus Cetuximab number of chemotherapy lines and increased EGFR-GCN were significantly associated with a better clinical outcome, independent of k-ras status.


Introduction
Treatment of advanced colorectal cancer (CRC) patients in the last ten years rapidly moved from a single agent 5fluorouracil (5-FU), modulated by Folinic Acid (FA), to combination chemotherapy including oxaliplatin (L-OHP) and irinotecan . The addiction of monoclonal antibodies directed to the vascular endothelial growth factor (VEGF), or to the epidermal growth factor receptor (EGFR) to a regimen with CPT-11-FA-5-FU increased progression free-survival (PFS) and overall survival (OS) in randomized phase III trials [1,2]. EGFR, whose locus is on the short arm of chromosome 7, is a transmembrane glycoprotein, with an intracellular tyrosine kinase domain. Binding of ligand to the EGFR domain induces receptor homodimerization or heterodimerization with other HER family members, which results in a transphophorilation of tyrosin-kinase and subsequent activation of a complex downstream signalling network [3]. EGFR activation appears to promote tumor growth and progression by controlling transcription, cell-cycle progression, apoptosis and differentiation [4]. Cetuximab is a MoAb active against the ligand binding site of EGFR with high specificity and higher affinity for EGF receptor than the natural ligands TGF-α and EGF. Preclinical models have demonstrated antitumor activity of Cetuximab by several mechanisms including inhibition of tumor cell proliferation, angiogenesis, invasion and potentiation of apoptosis; it seems also to mediate antibody-dependent cellular cytotoxicity [5]. In preclinical studies Cetuximab was able to overcome resistance to CPT-11 and to radiotherapy in colorectal cancer models [6,7]. Cetuximab is active either as a single agent and in combination with chemotherapy. Jonker et al. showed that Cetuximab increased PFS and OS when compared to best supportive care (BSC) in 572 patients previously treated with chemotherapy [8]. Cetuximab plus CPT-11 increased RR and PFS but not survival when compared to Cetuximab alone in the Bond trial [9] and to CPT-11 alone in the EPIC trial [10].
Phase II trials in untreated patients showed a high activity of the combination of Cetuximab plus doublets [11,12] or triplets [13]. A recent meta-analysis combining the OPUS and CRYSTAL trials showed an increase of overall survival adding Cetuximab to FOLFIRI (5-FU-FA-CPT-11) and FOLFOX 4 (5-FU-FA-L-OHP) [14]. When Pamitumumab, a fully humanized anti-EGFR anitibody, was added to FOLFOX 4 in first line treatment or to FOL-FIRI in second line treatment, it significantly increased RR and PFS [15,16]. The clinical relevance of these information indicate that chemotherapy plus an anti-EGFR antibody can be now considered as one standard option for patients with advanced CRC in first or second line of treatment. However these benefits are limited to a minority of patients and the identification of markers predictive of activity/resistance is clearly needed. EGFR expression, detected by immunohistochemistry (IHC), it does not represent a good predictive marker of response [17].
Moroni et al [18] were the first authors who evaluated the EGFR-gene copy number (GCN) in 31 selected patients with metastatic CRC treated with Cetuximab or Panitumumab. Eight out of nine patients who obtained a partial response had an increased EGFR gene copy number (GCN). By contrast, only one out of the twenty-one non-responders had an increased EGFR-GCN (p < 0.0001). However, there is no consensus on the predictive role of increased EGFR-GCN due to difficulty in reproducibility of the method of analysis, the limited number of patients evaluated and their heterogenic features. Lievre et al were the first who identified the mutation status of k-ras as the strongest predictive factor for resistance to anti-EGFR antibody showing that patients with mutated k-ras are genetically resistant to these agents. Therefore, the approved use of Cetuximab and Panitumumab is limited to patients with a wild-type k-ras status, because benefits in RR PFS and OS are limited only to k-ras wild-type patients.
The aim of the present study was to support further evidence of the predictive role of EGFR-GCN in terms of RR, PFS and OS in a retrospective series of 101 patients affected by advanced CRC and treated with chemotherapy plus Cetuximab. The role of kras status was also evaluated in a subset of 61 out of these 101 patients.

Patients eligibility
One hundred-one consecutive patients with pathologically confirmed metastatic CRC screened for EGFR immunostaining were retrospectively evaluated. Patients treated with Cetuximab as first line therapy had been previously included in controlled clinical trials, 2042 GOIM [12] and POCHER study [13] respectively. Pretreated patients received Cetuximab with CPT-11 alone or with FOLFIRI. Only one patients with pelvic recurrence and lung metastases was treated with Cetuximab and received a single course of radiotherapy.
Eligibility criteria included: age ≥ 18 years, Eastern Cooperative Oncology Group performance status of 0,1-2; life expectancy of at least 3 months; normal hematopoietic, hepatic, and renal functions; no history of brain metastases and no prior treatment with EGFR-targeting agents. Patients gave written informed consent before treatment.

Dosage and Drug Administration
Cetuximab was delivered with the same dosage and schedule both as single agent or in combination: a 2-hour intravenous infusion at 400 mg/m 2 followed by weekly 1hour infusion of 250 mg/m 2 .
Toxicity was graded according to the National Cancer Institute Common Toxicity Criteria (version 2.0).

Pretreatment and Follow-Up Studies
History, physical examinations, and a safety assessment were performed pre-treatment and weekly thereafter. Electrolytes, serum chemistries, liver and kidney function examinations were performed at baseline, every 2 weeks and at the end of treatment.
Tumors were measured pre-treatment and every 6 weeks and tumor response was assessed with CT scan according to the RECIST criteria [19].

EGFR Immunohistochemistry
Immunohistochemical stains were performed on 5 μ paraffin embedded tissue sections. Sections were deparaffinized and rehydrated in a series of alcohols and xylene according to established procedures. The sections were immunostained for EGFR using DAKO EGFR PharmDX kit (Dako, Milan, Italy). Antigen retrieval was performed using proteinase K for 5 min. Sections were then visualized with 3,3'-diaminobenzidine (DAB) as chromogenic substrate and counterstained with Mayer's haematoxylin. Negative controls included replacement of the primary antibody with non-reacting antibodies.
EGFR expression is defined positive as any membrane staining above background level was visualized. Moreover, an intensity score was applied as follows: negative no reaction; 1+ if the neoplastic cells displayed an incomplete, weak plasmamembrane/cytoplasmic; 2+ if neoplastic cells displayed a complete plasmamembrane immunostaining with a moderate intensity; and 3+ if neoplastic cells displayed a complete plasmamembrane strong immunostaining. Evaluation of the immunohistochemical results was performed independently and in blinded manner by two investigators (MM, MGD).
Two hundred nuclei per specimen were observed using a fluorescence microscope with a 100× lens using an Olympus BX 61 fluorescence microscope equipped with a 100 watt mercury lamp and with the Triple Bandpass Filter set (Vysis) for DAPI, SpectrumOrange and Spec-trumGreen. Fluorochrome signals were captured individually and images were generated via a computer with Quips genetic workstations and imaging software (Vysis).
EGFR gene was visualized as a red signal and the CEP 7 was visualized as a green signal. EGFR gene status was scored as the average number of EGFR red signals per nucleus and as the ratio between EGFR red signals and CEP7 green signals. Centromeric enumeration probe CEP7 was used as a control to determine copy number of chromosome 7, to adjust for the effects of aneuploid chromosome 7 when the EGFR gene copy numbers were counted Only nuclei with unambiguous chromosome 7 centromeric hybridization signals were scored for the EGFR signal numbers.
Polysomy of EGFR gene consisted of an increase of EGFR red signals (≥ three signals per nucleus) paralleled by the same increase of chromosome 7 (on which the EGFR gene is located) as measured by the number of CEP7 green signals per nucleus in at least 50% of neoplastic cells. Samples with a ratio EGFR gene/CEP7 ≥ 2.0 were esteemed amplified whereas samples displaying a CEP7 ≥ 3 were defined polysomic.
The PCR products were purified using Nucleospin Extract II Purification kit (M-Medical). Cycle sequencing was performed using BigDye Terminator v 3.1 kit (Applied Biosystems, Monza, Italy), and analyzed with a ABI 3130 capillary electrophoresis system (Applied Biosystems).
The presence of an heterozygous k-ras mutation in the tumor was defined as the appearance of a mutant peak with an height of at least one-third of that of the wild type. All sequencing analyses were performed at least twice on two independents PCRs.
In all the 61 CRC patients analysed, DNA was extracted from the primary tumour.

Statistical Analysis
Descriptive statistics were used to summarize pertinent study information. The association between variables was tested by the Pearson Chi-Square test or the Fisher's Exact test. Logistic regression multivariate analysis was used to assess the impact of the following variables on the response rate: number of lines, EGFR IHC score, GCN, number of metastatic lines, liver metastases, primary tumor site. Results are reported as odd ratio (OR) with 95% CI. PFS and OS were calculated by the Kaplan-Meier product-limit method from the date of the first day of treatment until progression of disease or death for any cause or for disease. If a patient had not progressed/died, survival or progression was censored at the time of the last visit. The log-rank test was used to assess differences between subgroups. Significance was defined at the p < 0.05 level [20]. The Hazard risk and the confidence limits were estimated for each variable using the Cox univariate model and adopting the most suitable prognostic category as referent group [21]. A multivariate Cox proportional hazard model was also developed using stepwise regression (forward selection) with predictive variables which were significant in the univariate analyses. Enter limit and remove limit were p = 0.10 and p = 0.15 respectively. The SPSS (13.0) statistical program was used for analysis.

EGFR Protein Expression and EGFR-GCN
EGFR protein was over-expressed in 90 out of 101 patients (89%), 22 with a 1+ staining score, 40 with a 2+ score and 28 with a 3+ score.

Response to Chemotherapy plus Cetuximab
From February 2004 to May 2007 101 CRC patients (62 male, 39 females; median age 63 years, range 26-80) were screened for EGFR tumor expression and treated with Cetuximab (Table 2). Forty-three patients were treated with chemotherapy plus Cetuximab as first line. Fiftyeight patients received Cetuximab as second or more lines of chemotherapy with a median number of two (range 1-5) and a median interval of 18 months (range 1-60) between starting of chemotherapy and Cetuximab. Only 12 patients received Cetuximab as monochemotherapy.
Ninety-nine out of 101 patients were evaluable for response: 40 patients (40%, CI 31-50) had a partial response, 31 (31%) had a stable disease, 20 patients (20%) had a progression and eight (8%) had non-measurable disease. In those patients who received Cetuximab as first line treatment we observed a RR of 70% (30/43) while it reached 18% (10/56) in the group with 2 nd or further line of therapy (p < 0.0001).
Multivariate regression analysis showed that patients treated as first line had a better chance of response than pretreated patients [HR 13.90 (4.41-43.83) p < 0.0001] and those with increased EGFR-GCN better than nonincreased [HR 6.27 (1.72-22.89) p < 0.005].

Relation between EGFR-GCN and Protein Expression with PFS and OS
At time analysis was done 65 patients (64%) progressed and only 19 patients (19%) deceased. Median follow-up for all patients was 12 months (range 1-34).
In the group of patients as first line treatment median PFS was 12 months (95% CI 9-15) versus a median PFS of 6 months (95% CI 4-9) for the group of patients who received Cetuximab as a II or more line therapy, p = 0.01.
As illustrated in Table 3, Cox model analysis showed IHC EGFR score 2-3 increased EGFR-GCN and first line chemotherapy significantly associated with a better PFS. When patients were divided into four groups, according to line of therapy and EGFR-GCN, a statistically significant difference for PFS was observed, with first-line patients/increased EGFR-GCN having the best PFS and pre-treated/non-increased EGFR-GCN the worst (p < 0.0001) (Figure 1).
At multivariate analysis response to therapy was the only prognostic predictive factor for OS. No difference in OS was observed among the four groups of patients (data not shown).

K-ras analysis and EGFR-GCN
k-ras analysis was performed in 61/101(59.4%) patients. There were no differences in clinical data, in patients with or without K-ras analysis, such as median age, EGFR expression, EGFR-GCN, site of primary tumor, incidence of liver metastasis, response to treatment and line of chemotherapy treatment (data not shown). There was no correlation between patients with increased EGFR-GCN  and wild-type status 23/35 (65%) and patients with nonincreased EGFR-GCN and wild-type status 11/18 (61%). K-ras mutations were found in 23/61 (37.7%). There was no correlation between k-ras status and response to treatment with 18/38 objective response (47.4%) in k-ras wt patients and 10/22 (45,5%) in k-ras mutation, with an OR = 1.08 [(CI 95% 0.38-3.10), p = 0.89]. One patient was not evaluable for response. In the 61 patients analyzed for k-ras increased EGFR-GCN maintained its predictive role for PFS together with EGFR-score 2-3 and k-ras status wild-type (Table 3).

Discussion
The availability of anti-EGFR antibodies in advanced colorectal cancer and the need to increase their limited efficiency in unselected patients are factors requiring the development and validation of laboratory tests which could predict who might benefit from this treatment in terms of activity and efficacy.
We studied a population of patients mainly treated with chemotherapy plus Cetuximab, 42% of whom treated as first line treatment. We demonstrated that: a) there is no correlation between EGFR immunostaining and EGFR-GCN; b) increased EGFR-GCN, first line chemotherapy, EGFR score 2-3 were predictive factors for PFS; c) these data were confirmed as independent from k-ras status.
We are aware that interpretation of these results seem to be difficult because of a mixed sample of patients treated as first or further lines of chemotherapy and mostly having received chemotherapy plus Cetuximab and not Cetuximab alone.
However the increase of RR due to Cetuximab addiction in CPT-11 containing regimens in first or second line of therapy is in the same range. In the Crystal trial RR was 59.3% with FOLFIRI plus Cetuximab versus 43.2% in FOLFIRI alone (wild type patients) [2]. In the EPIC trial RR was 4.2% with CPT-11 and 16.4% with CPT-11 plus Cetuximab [10]. In The Peeters trial [16] in second line of treatment RR was 10% with FOLFIRI alone and 35% FOLFIRI plus Panitumumab. So it could be reasonable to analyse patients in first or more line of chemotherapy together.
The role of increased EGFR-GCN and of number of chemotherapy lines as prognostic factors in the Kaplan-Meier curves for PFS are clearly shown in Figure 1 where four distinct groups of patients can be separated according to the line of chemotherapy and EGFR-GCN. Patients treated as first line and with an increased EGFR-GCN had the best PFS but a significant difference in PFS was also found in patients treated with Cetuximab plus chemotherapy as second or more line with or without an increased EGFR-GCN (p = 0.03).
In our population 89/101 patients were treated with combination of chemotherapy plus Cetuximab. Literature concerning the role of EGFR-GCN in patients treated with chemotherapy alone is very limited. Our data indicate that, in the context of combination of chemotherapy     [26]. In our series a sample was defined polysomic for the EGFR gene when at least 50% of examined neoplastic cells had ≥ 3 signals per nucleus paralleled by the same increase of chromosome 7 on which the EGFR gene is located. Therefore, much of the data from literature is similar although an international consensus on the definition of cut-off points is needed.
Is the information coming from GCN useful in clinical practice for the patient? EGFR-GCN is indicative of a subgroup of patients who will most likely benefit from this combination but currently FISH results do not enable us to discriminate the responsive patient.
The molecular picture of colorectal cancer seems to be so complex that it is difficult to identify a single molecular marker to assess responsiveness or a better outcome. K-ras mutation has emerged as the strongest predictive factor for resistance to anti-EGFR moAbs [27] but its role as a marker of survival has been demonstrated only in wild-type and not in the mutated patients when treated with panitumumab [28]. Recent data from Crystal and OPUS studies showed that addiction of Cetuximab to FOLFIRI (183 patients) versus FOLFIRI alone (214 patient) was associated with a non statistically significant increase in RR, PFS, and OS in k-ras mutant patients [14]. Etienne-Grimaldi et al showed that advanced CRC patients treated with 5-FU without anti-EGFR moAb had the same response potentiality and the same survival rates independent from k-ras mutational status [29].
Another point to be addressed is the role of IHC for EGFR. To date, it has been demonstrated that the tumor EGFR expression detected by IHC does not represent a good predictive marker of response to Cetuximab. The analysis of our results showed that the intensity of EGFR tumor expression (IHC score 2-3 vs 0-1) was significantly related to a prolonged PFS (Table 3). At our knowledge, this is the first study in which EGFR overexpression (score 2+/3+) detected by IHC appears to be relevant in predicting PFS demonstrating that patients bearing advanced CRC strongly positive for EGFR may benefit from therapy with MoAbs. Up to now, we have no explanation for this result which is contrary to that reported in the literature and needs to be confirmed in a larger and more homogeneous series.
In conclusion, in our advanced CRC population treated with Cetuximab plus chemotherapy an increased EGFR-GCN conferred a treatment advantage in untreated and pretreated patients. This effect was maintained in the subset of k-ras evaluated patients. Integration of this information with that coming from other molecular pathways could lead to a personalized "targeted" therapy for these patients.