Somatic Mutations of Epidermal Growth Factor Receptor in Bile Duct and Gallbladder Carcinoma

Materials and Methods

Patients and tissues. The specimens for this study were obtained from paraffin-embedded samples of 40 patients with bile duct and gallbladder carcinomas: 15 of them were intrahepatic cholangiocarcinomas, 14 were extrahepatic cholangiocarcinomas, and 11 were gallbladder carcinomas. There were 23 males and 17 females with a median age of 66 years. Patient characteristics, including sex, age, stage, and histologic grading, are summarized in Table 1 . The patients were of Italian origin. Tumor specimens were obtained from the patients before any systemic treatment was administered.

Histologic type was determined according to WHO criteria (23). Patient stage at the time of diagnosis was determined according the tumor-node-metastasis staging system (24).

Samples of peripheral blood were obtained from healthy volunteers after informed consent was given.

Isolation of genomic DNA and mutational analysis of EGFR. Genomic DNA was extracted from deparaffinized samples, with the use of the QIAamp DNA Mini kit (Qiagen, Milan, Italy) following the manufacturer's instructions. To reduce contamination with normal cells, the tumor portion was obtained by laser microdissection (VSL-337ND-S, Spectra-Physics, Mountain View, CA). The kinase domain of EGFR coding sequence, from exons 18 to 21, was amplified by using primers and nested PCR conditions as previously described by Lynch et al. (12). The PCR products were then purified by QIAquick PCR purification kit (Qiagen), and sense and antisense sequences were obtained by using forward and reverse internal primers, respectively. Each exon was sequenced using the BigDye Terminator Cycle sequence following the PE Applied Biosystems strategy and Applied Biosystems ABI PRISM3100 DNA Sequencer (Applied Biosystems, Foster City, CA). All mutations were confirmed, performing two independent PCR amplifications. In one case in which different mutated sequence were found by independent PCR amplifications, the coexistence of several mutants was confirmed by subcloning PCR products into PCR-4 vector using TOPO TA-cloning kit (Invitrogen, Milan, Italy) and further sequencing of 40 clones for each PCR product.

EGFR activation analysis. The activation of EGFR downstream signaling was evaluated by immunohistochemistry detection of phosphorylated MAPK and phosphorylated Akt. A rabbit polyclonal antibody against phosphorylated MAPK at Thr²⁰²/Tyr²⁰⁴ (Cell Signaling Technology, Beverly, MA) and a rabbit polyclonal antibody against phosphorylated Akt at Ser⁴⁷³ (Cell Signaling Technology) were used. Positive immunostaining was attributed to nuclear staining for phosphorylated MAPK and cytoplasmic with a faint membranous staining for phosphorylated Akt.

Statistical analysis. The variables measured in the study were investigated for association by using the Fisher's exact test or χ² test as appropriate. P < 0.05 was considered significant.

Results

EGFR mutational analysis. The EGFR genomic sequence coding for the tyrosine kinase domain was analyzed in a series of 40 bile duct and gallbladder carcinomas. Exons 18, 19, 20, and 21 were directly sequenced after PCR amplification. The details of the EGFR mutation spectrum are shown in Table 2 . Six patients (15%) had point mutations in the EGFR tyrosine kinase domain (Fig. 1 ): two patients with extrahepatic cholangiocarcinoma (cases 2 and 6), one patient with gallbladder carcinoma (case 4), and three patients with intrahepatic cholangiocarcinoma (cases 1, 3, and 5).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Electropherograms of tyrosine kinase domain sequence (exons 18-21) of the EGFR gene. Mutations found in 6 of 40 patients with bile duct and gallbladder carcinoma are shown and compared with the corresponding wild-type sequence. Cases 1 to 5 had single missense point mutations; case 6 had multiple mutation detected by independent PCR (#1 to #3). The exon 20 variant is also shown in a heterozygous status in case 3.

Mutations were classified as heterozygous when the wild-type sequence was amplified on the second allele, or homozygous/hemizygous when only the mutated sequence was amplified, suggesting either a biallelic mutation or a monoallelic mutation associated to the loss of the second allele.

No mutation was detected in exon 18 in all tested tumors.

One homozygous/hemizygous point mutation was found in exon 19; it consisted of a transition from A to G, resulting in amino acid substitution from lysine to arginine at codon 757 (K757R).

In exon 20, one silent point mutation or variant, already described by Nagahara et al. in colorectal carcinoma (20), was found at codon 787 (Gln; CAG-to-CAA) in 36 of 40 (90%) patients. This variant was present both in homozygous/hemizygous (78%) and heterozygous status (22%). In addition, in two cases missense point mutations of exon 20 were found: one patient had one mutation involving the codon 775 (TGC-to-TAC) with amino acid substitution from cysteine to tyrosine (C775Y), and the second had one mutation at codon 790 (ACG-to-ATG; Thr-to-Met). This last mutation was previously described in lung adenocarcinoma by Kobayashi et al. (25). They showed that T790M mutation was associated to emerging resistance during gefitinib treatment.

Mutated sequences of exon 21 were found in three samples. One patient had a homozygous/hemizygous missense point mutation (GCC-to-ACG; Ala-to-Thr) at codon 864 (A864T). The second patient had one (GAA-to-AAA; Glu-to-Lys) at codon 872 (E872K). The third patient had different mutations obtained with independent PCR: a heterozygous point mutation with amino acid substitution from glycine to serine (GGC-to-AGC) at codon 882 (G882S), a second heterozygous point mutation with amino acid substitution from valine to isoleucine (GTA-to-ATA) at codon 843 (V843I), and a third heterozygous silent mutation at codon 858, which is the most frequently missense mutated codon (L858R) in lung cancer (11–15, 18, 21, 22). The presence of different mutated sequences was confirmed by PCR products subcloning into PCR-4 vector and was found not to be due to microsatellite instability, as shown by the analysis of a marker panel (BAT25, BAT26, D2S123, D2S346, and D17S250) according to Bethesda guidelines (data not shown; ref. 26).

We investigated the somatic origin of the mutations by performing PCR analysis and sequencing of DNA extracted from nontumor tissues of patients. Only exon 20 variant at codon 787 was also observed in the corresponding nonneoplastic tissues of the patients. We confirm the high prevalence of EGFR exon 20 variant in Italian subjects also in healthy donors, as five of six sequences of exon 20 of EGFR in mononuclear cells obtained from healthy volunteers harbored the silent G to A. The distribution of EGFR gene mutations was significantly different between males and females as also described in lung adenocarcinoma (27–29). They were present in 1 of 23 (4.3%) males and in 5 of 17 (29.4%) females (P < 0.05). No correlation was found between mutations of the EGFR gene and extrahepatic or intrahepatic origin of the tumor nor with other clinical and pathologic findings.

EGFR activation analysis. To evaluate the phosphorylation status of EGFR downstream transducers MAPK and Akt, immunohistochemistry was done on all tumor samples. The results showed that of the 40 specimens, 16 (40%) show MAPK phosphorylation, and 22 (55%) show Akt phosphorylation (Table 3 ). Five of six (83.3%) cases with EGFR mutations show MAPK phosphorylation (P < 0.05). The same proportion of EGFR-mutated cases had Akt phosphorylation without statistical significance (P > 0.05).

Discussion

Chronic biliary inflammation and cholestasis, leading to the production of proinflammatory cytokines and reactive oxygen species, cause protracted cellular stress and irreversible DNA damage. This in turn leads to malignant transformation of cholangiocytes (4). Moreover, the presence of biliary acids in chronic cholestasis functionally transactivates EGFR via transforming growth factor-α (10), leading to activation of downstream pathways that promote cell survival, proliferation, and inhibition of apoptosis, which represent major processes characteristics of cholangiocarcinogenesis (30). The discovery of acquired somatic mutations in the EGFR gene that modify the activity of tyrosine kinase domain could be crucial in gaining a better understanding of molecular pathogenesis of this highly aggressive tumor.

EGFR activation regulates a number of cellular functions and has been strongly implicated in carcinogenesis, thus making EGFR tyrosine kinase a promising therapeutic target in solid tumors (6, 7). However, it is necessary to carefully evaluate whether a selective inhibition of pathways involved in carcinogenesis can produce sufficient effect on tumor growth. Previous studies have shown that the activity of small-molecule inhibitors, such as gefitinib or erlotinib, in the treatment of non–small cell lung carcinoma is associated with the existence of specific mutations in the tyrosine kinase domain of EGFR in tumors (11–15, 18, 21). The prevalence of such mutations is about 20% but is higher in patients of East Asian ethnicity (30%) than in other ethnicity (10%). They are statistically more frequently found in females than males, nonsmokers than smokers, and in adenocarcinoma than other histologies. Most tumors that responded to gefitinib or erlotinib displayed mutations in the kinase domain of the EGFR gene (11–13, 29).

In cholangiocarcinoma, somatic mutations of the EGFR gene have not been reported to date. As with lung carcinoma, mutations in the tyrosine kinase domain of the EGFR gene may be associated with clinical efficacy of EGFR inhibitors.

In the present study, we found that 6 of 40 (15%) biliary tree and gallbladder carcinoma have EGFR gene mutations in the sequence coding for the tyrosine kinase domain. All of the mutations were somatic acquired point mutations, and most were found within exon 21.

Four of eight point mutations found seemed to be homozygous as previously described for other tumor types (14, 19) and might be considered biallelic. Alternatively, they could be considered hemizygous. In these cases, there may be the contemporary presence of monoallelic point mutations with deletions of the second allele involving the intronic regions to which primers used for PCR amplification anneal. The technique applied allows the identification of all deletions previously described (12) in lung tumors; however, other possible deletions not detected by these conditions may exist.

In one case, different mutations were simultaneously present in the same tissue suggesting the existence of a heterogeneous population of cancer cells. In our study, tumor cells were obtained from tissues by laser microdissection, which facilitates precise discrimination between tumor cells and analysis groups of fewer cells. This technique probably contributes to highlight even minimal differences between tumor clones thus revealing cancer heterogeneity.

In five cases, we detected mutations that have not been previously described in other type of tumors.

No in-frame deletions within the exon 19, frequently observed in non–small cell lung cancer and in head and neck cancer (19), were found. In our study, codon 858, which represents the most common site of mutation in non–small cell lung carcinoma, showed only one silent nucleotide substitution in one case. In two cases, we confirmed previously described mutations of the tyrosine kinase domain. The V843I mutation determined in case 6 has been detected by Shih et al. (31) in a patient with lung carcinoma that experienced a partial response to gefitinib treatment. The T790M mutation determined in case 3 was previously detected in some cases of lung carcinoma that recur after an initial response to gefitinib (25). These findings suggest that the detection of EGFR mutations in bile duct carcinoma does not imply that these tumors are sensitive to the clinical available molecular inhibitors. However, functional analysis of the T790M mutation revealed that other small-molecule inhibitors could actively block EGFR phosphorylation (32).

Previous studies showed the association between EGFR signal transducers activation and better outcome in lung cancer patients treated with gefitinib (33). All the mutations that we found in bile duct carcinoma led to activation of one or both of the EGFR signal transduction pathways analyzed. The percentage of cases with activation of EGFR downstream pathways was higher in EGFR-mutated compared with EGFR wild-type tumors. This observation sustains the possibility to obtain objective response to small-molecule inhibitors in bile duct carcinoma.