Main

The HER-2 gene is located on chromosomal region 17q11.2–q12 and the EGFR gene is located on 7p12. These encode 185 kDa (p185) and 170 kDa plasma membrane glycoproteins, respectively. They share approximately 50% overall homology and are composed of an N-terminus extracellular ligand-binding domain, a transmembrane lipophilic segment, and a C-terminus intracellular region containing a tyrosine kinase domain.1 High-affinity ligand binding causes receptor dimerization, which results in the activation of an intrinsic protein tyrosine kinase activity and tyrosine autophosphorylation. These events activate a cascade of biochemical and physiological responses that are relayed to transcription factors, resulting in changes in gene and protein expression. Thus, these proteins are categorized as type I receptor tyrosine kinases (RTK).

Recently, these type I RTKs have attracted great deal of attention, due to the development of several clinical therapies targeting them. Antibodies against the external domains of these receptors, or small molecule inhibitors targeting their kinase activities are in development and/or approved for therapy and have garnered much attention for their dramatic effectiveness in some patients. The humanized monoclonal antibody against the external domain of HER-2 (trastuzumab or HerceptinTM, Genentech Inc., South San Francisco, CA, USA) is the most successful example. This monoclonal antibody-based therapy has been approved for use in breast cancer patients and has demonstrated an ability to extend median survival time in metastatic breast cancer patients exhibiting HER-2 overexpression.2 Similarly, IMC-C225 (cetuximab or ErbituxTM, ImClone Systems Inc., Branchburg, NJ, USA), a monoclonal antibody targeting EGFR, and among the various small molecule inhibitors of tyrosine kinases ZD1839 (gefitinib, IressaTM, AstraZeneca, Macclesfield, UK) are now approved for use in patients with colorectal cancers (FDA News, February 12, 2004) and non-small cell lung cancer,3 respectively.

It has been established that in breast cancer, HER-2 overexpression is principally the result of gene amplification. The US Food and Drug Administration approves the use of trastuzumab therapy in cases where HER-2 overexpression is detected by immunohistochemistry (IHC) or gene amplification is detected by fluorescent in situ hybridization (FISH). In addition to breast, lung, and colorectal carcinomas, many types of epithelial malignancies reportedly display increased HER-2 and/or EGFR expression on their surface membranes, sometimes accompanied by gene amplification. We have examined the expression of HER-2 in gastric4, 5, 6 and lung cancers,7 and of EGFR in gastric cancers8 using IHC, as well as gene amplification using FISH. We have shown that the major mechanism of protein overexpression in these cancers is gene amplification. In particular, in gastric and lung carcinomas, most tumor cells exhibiting HER-2 amplification exhibited signals localized to one or two clusters, indicating that the amplified gene was present in homogeneously staining regions (HSRs), similar to those observed in breast cancers. Thus, we have proposed that trastuzumab should be considered when designing adjuvant therapy for patients with gastric and lung cancers.4, 5, 6, 7

Colorectal carcinoma is the second leading cause of cancer death in the Western world. There have been many studies examining the overexpression HER-2 and EGFR in colorectal cancers; however, there are numerous discrepancies in the frequencies reported from IHC studies. There have been few studies examining gene amplification of HER-2 and EGFR in colorectal carcinomas. This encouraged us to examine colorectal cancer for aberrations in HER-2 and EGFR expression and gene copy number using combined IHC and FISH techniques. Our aim was to provide some rationale for the introduction of new adjuvant therapies for colorectal cancer patients. We sought first to determine the exact frequencies of HER-2 and EGFR abnormalities in colorectal cancers; second, to clarify the relationship between protein overexpression and gene amplification of HER-2 and EGFR and; third, to examine possible genetic heterogeneity of HER-2 and EGFR within and between primary and metastatic tumors.

Materials and methods

Tissue Samples

We examined 244 primary colorectal carcinomas and concurrently excised nodal metastases obtained from consecutive surgeries performed at the Department of Surgery, Yamanashi Medical University between 1994 and 2002. The patients consisted of 150 men and 94 women with a median age of 65.0 years (mean, 64.7; range, 32–91). The condition of the patients was assessed according to the system for staging primary tumor/regional lymph nodes/distant metastasis (TNM) described in the AJCC Cancer Staging Manual.9 The World Health Organization Classification of Tumors10 was used to determine histological classification. The 244 patients were classified into the TNM stages as follows: Stage 0, 8 patients; Stage I, 47 patients; Stage II, 76 patients; Stage III, 85 patients; and Stage IV, 28 patients. Histologically, the specimens were divided into 102 well-differentiated, 120 moderately differentiated, five poorly differentiated adenocarcinomas, and 16 mucinous carcinomas and one signet-ring cell carcinoma. This laboratory study was approved by the Institutional Review Board at the University of Yamanashi, and written informed consent was obtained from all patients.

IHC

Resected colon samples were immediately immersed in 20% buffered neutral formalin, fixed overnight, and embedded in paraffin according to standard procedures. Since immunoreactivity diminishes with time in formalin-fixed, paraffin-embedded sections stored on glass slides at room temperature, all sections (4 μm) were stained within 6 weeks of being cut. Serial sections (4 μm) that had been cut from representative formalin-fixed, paraffin-embedded cancer tissues and placed onto silanated glass slides (Matsunami, Tokyo, Japan) were used for hematoxylin–eosin staining, IHC detection of the HER-2 and EGFR, and FISH analysis. IHC detection of HER-2 and EGFR were carried out on all the primary tumors, as well as metastatic tumors of the lymph nodes in those cases where protein overexpression was detected in the primary tumors. A polyclonal antibody (Nichirei, Tokyo, Japan; dilution, 1:100) against the internal domain of the human HER-2, and a monoclonal antibody against the external domain of human EGFR (Novocastra Lab, Newcastle, UK; Working dilution, 1:20) were used. The specificities and sensitivities of the antibodies against HER-2 and EGFR were verified previously.4, 5, 6, 7, 11, 12 For EGFR detection, a high-temperature antigen unmasking technique was used: that is, autoclave the section in 0.01 M citrate buffer (pH 7.0) at 121°C for 10 min. Antibodies were visualized by avidin–biotin binding to peroxidase-conjugated secondary antibodies. In each analysis, a gastric cancer section that had been previously confirmed to overexpress HER-2 or EGFR5, 8 was included as a positive control.

HER-2 and EGFR positivity in the IHC analyses were reviewed by three pathologists (TT, SS, AO), who were unaware of the gene amplification. The intensity of reactivity was scored using a four-tier system: negative, no discernible staining or background type staining: 1+, definite cytoplasmic staining and/or equivocal discontinuous membrane staining: 2+, unequivocal membrane staining with moderate intensity: 3+ strong and complete plasma membrane staining. Samples exhibiting 2+ or 3+ immunostaining were considered positive. For both HER-2 and EGFR, the extent (%) of positive staining cells was measured in the representative large section in each tumor.

FISH

FISH analysis was undertaken for all 2+ and 3+ staining primary tumors and their metastatic lymph nodes and for 34 1+ staining samples. In addition, 30 tumors scoring negative for EGFR, and 30 tumors scoring negative for HER-2 were selected at random. Gene amplification of HER-2 (chromosomal locus 17q11.2–q12) and EGFR (7p12) were determined using fluorescently labeled DNA probe sets purchased from Vysis, Inc. (Downers Grove, IL, USA): LSI HER-2/CEP 17™ for HER-2 and LSI EGFR/CEP 7™ for EGFR. When necessary, LSI D7S486/CEP 7™ (Vysis) was also used to examine gene amplification of 7q31. Each probe sets contains a SpectrumOrangeTM-labeled gene (locus)-specific probe and a SpectrumGreenTM-labeled centromeric probe hybridized to centromeric region of the chromosome as the control to normalize copy number for each chromosome.

FISH was performed using standard methods, with a modification to incorporate an intermittent, short-term microwave treatment during the initial period of hybridization.13 In brief, 4 μm sections were deparaffinized by five successive, 3-min washes in xylene followed by five washes in ethanol and heating in a 0.01 M citrate buffer (pH 6.0) using a microwave processor (MI-77, Azumaya Company, Tokyo, Japan) for 10 min. After treatment in 0.2% pepsin/0.01 N HCl for 10 min at 37°C, the samples were aged in 0.1% NP-40/2 × SSC for 10 min at 37°C and their DNA was denatured by treatment in 70% formamide/2 × SSC for 5 min at 85°C. A measure of 10 μl of the probe solution was then placed on a glass slide with a coverslip. The sample slides in the hybridization mixture were then put in a microwave processor and were irradiated for 3 s at 2 s intervals (2.45 GHz, 300 W) with the temperature sensor set at 42°C, and then hybridized at 42°C overnight. Posthybridization washing was carried out according to the manufacturer's protocol. The tissue sections were counterstained with 4′,6-diamidine-2′-phenylindole dihydrochloride and p-phenylenediamine in phosphate-buffered saline and glycerol (DAPI II) (Vysis) and examined with a fluorescence microscope (Olympus, Tokyo, Japan) equipped with the Triple Bandpass Filter set (Vysis) for DAPI II, SpectrumOrange and SpectrumGreen, and filter sets specific to SpectrumOrange and SpectrumGreen. As positive controls, gastric cancer tissues that had been previously confirmed to have amplification of HER-2 or EGFR were used.5, 8

A cell was scored as positive for amplification when a definite cluster or more than 10 orange signals were found, as described previously.4, 5, 8 FISH images were taken using a photographic camera and recorded on film slides.

Statistical Analysis

Agreement among observers in interpretation of IHC specimens, and the associations between protein overexpression and amplification of EGFR or HER-2 were tested by kappa (κ) statistics.14 In accordance with the criteria of Landis and Koch,15 the κ-values were divided into several scales to evaluate the strength of the agreement: κ<0.00, poor; 0.00<κ<0.20, slight; 0.21<κ<0.40, fair; 0.41<κ<0.60, moderate; 0.61<κ<0.80, substantial; and 0.81<κ<1.00, nearly perfect. A χ2-test for independence was used to examine the correlation between overexpression of HER-2 and EGFR, and between overexpression of HER-2 or EGFR and the several histological classifications. Spearman's correlation coefficient by rank was used to analyze the association of expression of HER-2 or EGFR with the various pathological stages.

Results

IHC

Clinicopathological data and immunostaining results for the 27 cases exhibiting overexpression of HER-2 or EGFR are summarized in Table 1. Overexpression of HER-2 was found in eight (3%) of the 244 colorectal carcinomas analyzed (Figure 1a). The remaining 236 cases were all negative, with no tumors staining 1+ or higher. Overall interobserver agreement was substantial (κ=0.64, 95% confidence interval (CI), 0.58–0.69), and in particular agreement for the 3+ staining cases was nearly perfect (κ=0.89, 95% CI, 0.81–0.96). Overexpression of EGFR was found in 19 (8%) of the 244 tumors, and 1+ staining was found in 105 of the tumors. Although interobserver agreement was substantial (κ=0.80; 95% CI, 0.72–0.87) and moderate (κ=0.50; 95% CI, 0.42–0.57) for 1+ and 2+ staining respectively, it was nearly perfect for negative (κ=0.86; 95% CI, 0.79–0.94) and 3+staining (κ=0.93; 95% CI, 0.85–1.00).

Table 1 Pathological classification and result of IHC and Fish analysis of HER-2(a) and EGFR(b)
Full size table
Figure 1
figure 1

IHC of cells overexpressing HER-2 (a) and EGFR (b, c, and d). (a) Plasma membrane of adenocarcinoma cells exhibiting 3+staining (Case 1; × 125); (b) single cancer tubule composed of EGFR-positive and -negative cells (Case 15; × 250); (c) EGFR staining was confined to SSCB that emerge from the base or lateral edge of malignant glands (Case 18; × 125); (d) Poorly differentiated adenocarcinoma composed of EGFR-positive and negative cells (Case 11; × 125).

Full size image

In tumors exhibiting HER-2 overexpression, more than 80% of the cells in the specimens were positive for HER-2 staining, with the sole exception of Case 8, in which only 30–40% of cells showed positive staining. In contrast, in tumors exhibiting EGFR overexpression, high EGFR-expressing cells were localized into separate zones or foci (Figure 1b), and in all cases, except Case 10, positive cells comprised no more than 50% of the total cells. In 12 of the cases, positive cells comprised less than 5% of the total. In six of the 10 tumors containing 2+ staining cancer cells, positive staining was confined to small solid clusters or buds (SSCB) that emerge from the base or lateral edge of malignant glands (Figure 1c). In Case 11, overexpression was found predominantly in poorly differentiated cells (Figure 1d). In the rest of the tumors, no morphological distinction between positive and negative cancer cells were observed. There was no significant correlation between protein overexpression and histological type. We did not encounter a case in which both HER-2 and EGFR were concomitantly overexpressed. Statistically, no correlation was found between overexpression of these two oncoproteins.

FISH

By FISH analysis, all HER-2-overexpressing tumors also exhibited HER-2 amplification. HER-2 amplification was not found in any of the 30 negative controls. Thus, overexpression and amplification of HER-2 completely coincided. With regard to EGFR, nine of the 10 3+ staining tumors and two of the nine 2+ staining tumors were found to contain cancer cells exhibiting EGFR amplification. However, no gene amplification was found among any of the immunostained SSCB samples. Also, no gene amplification was observed among the 34 tumors exhibiting 1+ staining or the 30 nonimmunostaining control tumors. The association between protein overexpression and amplification of EGFR was substantial (κ=0.65; 95% CI, 0.48–0.88).

By FISH analysis, three major different amplification patterns were observed: one type consisted of one or two distinct clusters of orange signals (‘cluster type’; Figure 2a), and the second type as more than 10 scattered orange signals (‘scattered type’; Figure 2b). In both of these first two types, fewer than five green signals were observed. In the third type, more than 10 loosely clustered orange signals were found, accompanied with a similar number of green signals (‘accompanied type’; Figure 2c, d). HER-2 amplification was of the cluster type (first type) in all the positive tumors, although cluster-type appeared together with scattered-type in Case 5 (Figure 2e). Amplification of the EGFR was found to be of the scattered-type in nine tumors, although scattered-type was combined with cluster-type presentation in Case 9. Accompanied-type was observed in Cases 19 and 22. In both tumors, FISH analysis was repeated using the LSI D7S486/CEP 7™ probe, which revealed a copy number of 1–4 for 7q31 with no specific gene amplification (Figure 2f).

Figure 2
figure 2

Dual-color FISH analysis of HER-2/centromere 17 (a and e), EGFR/centromere 7 (b, c, and g) and 7q31/centromere 7 (f). (a) One or two clusters of orange signals of HER-2 are found in cancer nuclei (Case 1; × 625). (b) Same field of Figure 1b. FISH on a serial section shows 3+ staining cancer cells have multiple scattered orange signals of the EGFR gene (Case 15; × 250). (c and d) The same field was observed by the triple band filter (c) and the SpectrumGreen™-specific filter (d). Coamplified EGFR and centromere 7 are found to form loose clusters in the cancer nuclei (Case 19; × 625). (e) Cancer tubules are composed of cells with clustered signals and scattered signals of HER-2 (Case 5; × 250). (f) Cancer nuclei exhibited amplification of centromere 7 region (green signals), but no amplification of 7q31 (orange signals) (Case 19; × 625). (g) Amplification-positive and -negative cells were found in a metastatic lymph node (g) (Case 11; × 125).

Full size image

When tissue sections used for FISH were compared with serial sections used for immunostaining, the regions of cancer cells displaying gene amplification and protein overexpression completely overlapped, for both HER-2- and EGFR-positive tumors. Namely, HER-2 amplification was found more than 80% of the cells, being contras with EGFR amplification which was less than 50% in most tumors. This colocalization could sometimes be confirmed on a cell-by-cell basis, as shown in Figures 1b and 2b.

Six of the eight cases exhibiting HER-2 overexpression in the primary tumors had lymph node metastases and/or liver metastasis. All the metastatic nodes (13) and the liver metastasis (one) examined contained predominantly HER-2 overexpressing cells, with accompanying gene amplification. Among the 19 cases exhibiting overexpression of EGFR in the primary tumors, 12 had lymph nodes metastases and only three had EGFR-overexpressing cancer cells in the metastatic nodes, including Case 19, in which half of the metastatic nodes scored negative for overexpression. FISH analysis of EGFR-overexpressing tumors in lymph nodes revealed a mosaic pattern of amplified and nonamplified cells in Cases 11, 12, and 19 (Figure 2g). The types of amplification in the primary and metastatic tumors were the same, except in Case 5, where metastatic tumors lacked a cluster-type FISH pattern, and in Case 19, in which both accompanied-type and scattered-type patterns were observed in separate metastatic foci. Overexpression of HER-2 was not correlated with pathological stage. There was also no correlation between overall overexpression of EGFR and pathological stage; however, there was a positive correlation when only 11 overexpressing cases exhibiting gene amplification were included (P<0.05).

Discussion

Overexpression of HER-2 was found in 3% (eight of 244) of colorectal cancers, and in all of these, amplification of the HER-2 was observed. Overexpression of EGFR was found in 19 of the 244 (8%) cases of colorectal cancer, and gene amplification was observed in 11 of these. In previous IHC studies, there was considerable discrepancy in the frequencies and distribution of HER-2 and EGFR overexpression in colorectal carcinomas, as described below. These discrepancies were probably due to differences in tissue-fixation methods, the antibodies used, and the detection techniques and criteria for evaluating the results. Previous IHC studies examining HER-2 expression reported different percentages of positive cells, ranging from 0 to 100%.16, 17, 18, 19, 20, 21, 22, 23, 24 The major cause of these discrepancies seems to be the different criteria for evaluating the results. Studies that scored intracytoplasmic staining as positive reported larger frequencies, whereas studies that reported only membrane staining as positive was scored lower or zero frequency.16, 17, 18, 19, 22, 23, 24 The results of the current study appear similar to those of Osako et al,22 who used the same antibody as in our study. They observed membranous staining in only 3% of colorectal carcinomas (three of 100), and demonstrated gene amplification by Southern blot analysis in one of the three.22 The results of a recent study by McKay et al23 also seemed similar to ours; only eight out of 248 (3%) displayed strong immunoreactivity in the membrane of tumor cells, although 82% of the tumors expressed equivalent levels in adjacent normal mucosa. Most recently, Nathanson et al16 reported an incidence of 4% (five of 139) HER-2 overexpression by IHC, which showed extreme high concordance (κ=0.85) with gene amplification detected by FISH.

In contrast, Kapitanovic et al24 observed HER-2 immunoreactivity in normal colon tissue, especially adjacent to carcinoma, hyperplastic polyps, adenomas, and adenocarcinomas at different frequencies, intensities and extents in each case, and overall positivity in adenocarcinomas was 100%. Cytoplasmic and membrane staining seen with IHC was verified by the fact that p185 (HER-2) and phosphotyrosine were detected by Western blot analysis in both the membrane and cytosolic fractions. At the present time, the significance of cytosolic p185 is unclear, and in our own study no significant cytoplasmic staining was observed. Our study showed that overexpression of HER-2 in the cellular membrane in colorectal carcinomas is probably the result of HER-2 amplification, mostly occurring in cell clusters. This mechanism of increased HER-2 expression is similar to that known to occur in breast cancers.25

There also have been discrepancies in the reported frequencies of cells positive for EGFR and their distribution in previous IHC studies. The reason seems to be that antibodies that reliably detect EGFR expression in fixed and paraffin-embedded pathological material have only recently become available,26 as well as differences in scoring as described for HER-2 above.27 Recently, Goldstein and Armin28 noted that exclusive membrane immunoreactivity usually occurred in two morphological patterns. One was in undifferentiated cells, which were comprised of discontinuous, isolated single or small clusters of malignant cells in the stroma. The second was in SSCB that emerged from the base or lateral edge of malignant glands and that frequently were angulated. They speculated that tumor cells in the former type progress into those in the latter type.28 In our present study, we found six tumors with SSCB overexpressing EGFR, and also single tumor (Case 11) with a predominant component of undifferentiated desmoplastic carcinoma overexpressing EGFR. This tumor was associated with EGFR gene amplification; however, the six tumors with SSCB had no gene amplification, thus we found no genetic evidence to link the two types of EGFR-overexpressing tumors. The mechanism by which EGFR is overexpressed in the SSCB is unknown at the present time.

Excluding the five tumors with SSCB and the other two 2+ staining tumors, all the tumors exhibiting overexpression had amplification of the EGFR gene, mostly of the scattered-type. Furthermore, when IHC and FISH were compared in adjacent sections in these tumors, the coincidence of gene amplification and overexpression was confirmed almost on a cell-by-cell basis. The monoclonal antibody used is specific to the extracellular domain of EGFR. Expression of highly oncogenic variant III of EGFR, which is absent of the extracellular domain, may not be detected with this antibody.29 The DNA probe for EGFR spans the entire gene would likely detect any amplification, even the mutated form of the receptor. However, because the FISH analysis could not detect any amplified cells in 64 tumors without EGFR overexpression, it is very unlikely that variant III of EGFR is overexpressed by gene amplification in colorectal cancers. In most colorectal cancers, overexpression of EGFR caused by gene amplification comprised only a small portion of the cells in these tumors. Previous Southern blot studies as well as a study using quantitative real-time polymerase chain reaction have been unable to demonstrate EGFR gene amplification in colorectal cancers.30, 31 This could be explained by sampling errors or dilution on any positive signals when extracts from entire populations of cells are analyzed.

In mammalian cells, highly amplified DNA is found within two distinct structures: homogeneously staining regions (HSRs), which are located within a chromosomal site in the form of expanded chromosomal regions, and double minute chromosomes (DM), which are centromere-free circular structures.32 It is generally accepted that clustered signals found by FISH correspond to amplified signals in HSR, and scattered signals correspond to DM, excluding a few exceptional cases. In the present study, amplification of HER-2 was detected as clustered signals, and that of EGFR was detected as scattered signals in most cancer cells. Thus, it appears that the principal amplification pattern of HER-2 is HSR, and that of EGFR is DM. More direct evidence of amplified EGFR genes occurring in DMs of colon cancer cells was provided by Dolf et al,33 who used in situ hybridization of metaphase spreads in the colon cancer cell line DiFi, which had been established from the colon cancer of a patient with Gardner syndrome. It has been shown that in several cell lines amplified genes in the DM chromosomes were integrated into HSRs, whereas the converse breakdown of HSR to generate DM chromosomes has not been observed.34 In Cases 5 and 9, HSR and DM were found to coexist. This could represent the integration of DM to HSR in vivo. However, according to Dolf et al, no signs of stable reintegration of amplified EGFR have been observed in DiFi cells over the course of 3 years of cell culture.33

We are not aware of any previous reports of coamplification-type gene amplification. Usually, a centromeric probe is used as a reference probe to assist in distinguishing gene amplification from chromosomal polysomy in dual-color FISH. Increased gene number accompanied by an increase in the copy number of the chromosome on which the gene resides is considered a feature of polysomy. In Cases 19 and 22, the copy number of the sequence located on the long arm (7q21) of the same chromosome as the EGFR was not increased. Thus, this coamplification does not reflect the polysomy of chromosome 7. Instead, it may represent a large amplicon that includes 7p12 (locus of EGFR gene) and the chromosome 7 centromeric region. A recent comparative genomic hybridization (CGH) study of breast cancer samples showed that an amplicon including the HER-2 locus (17q11.2–q12) extended to 17p,35 while another CGH study reported that discontinuous ‘zebra’ patterns of allelic imbalance at 17q may be caused by gene amplification.36 In the present study, it was not clear whether the amplicons in coamplification-type gene amplification contains centromeric regions continuously or discontinuously, and if the latter is case, it is possible that other genes that play a role in colon cancer development and progression may be coamplified.

Breast cancers have attracted much attention because 20–30% of them exhibit HER-2 overexpression with gene amplification,37 and use of trastuzumab has increased the median survival time in metastatic breast cancer patients overexpressing HER-2 2. Cancers of head and neck, and non-small-cell lung cancers also exhibit a high frequency of EGFR overexpression ranging 34–62% with and without gene amplification.38, 39 Recent clinical studies reported that cetuximab or gefitinib showed some clinical benefit in patients of cancers of head and neck, and non-small-cell lung cancers.40 The present study demonstrated that HER-2 and EGFR abnormalities were found in only a minority of colorectal cancer cases. However, the small prevalences do not preclude patients of colorectal cancer from getting benefit from the receptor inhibitory therapies, because although small in number, high amplification/high overexpression of HER-2 or EGFR appears to define a distinct genotype/phenotype of colorectal cancer with properties that could respond well to the new adjuvant therapies.

Since possible targets of these adjuvant chemotherapies are metastatic tumors, it is relevant to characterize any genetic heterogeneity between primary and metastatic tumors, and more importantly between different metastatic tumors. Our findings showed that cancer cells overexpressing HER-2 were found both in the primary and metastatic tumors, and therefore examination of primary tumors could reliably predict the HER-2 status of the metastatic foci. These results also suggest that HER-2 could be a promising target for adjuvant therapy in colorectal cancers. In contrast, cancer cells with amplification and high-level overexpression of EGFR were a minor population in most primary tumors and displayed marked heterogeneity in metastatic sites, probably because of the instability of the amplified genes. This suggests that treatment that targets this molecule may encounter difficulties due to the emergence of resistant subclones lacking this alteration. It will be important to evaluate the gene and protein status of HER-2 and EGFR to design future adjuvant therapies for patients with colorectal cancers.