Introduction
The incidence of esophageal carcinoma (EC) and the histological type varies widely with geographical location [
1]. Notably, esophageal squamous cell carcinoma (ESCC) is the predominant histological type in Asian areas with a high risk of cancer development in the esophagus, in contrast to Western countries (Northern Europeans and Caucasians in the USA) [
1]. In the United States and North and Western Europe, where esophageal adenocarcinoma is the predominant histological type, the incidence of ESCC has been decreasing in recent decades, whilst the incidence of esophageal adenocarcinomas, especially of the gastroesophageal junction, has been increasing [
2,
3].
Surgery remains the mainstay of therapy for esophageal carcinoma patients. However, ESCC diagnosis is associated with a high mortality owing to its aggressive behavior and the 5-year survival for non-metastatic disease ranges from 20% to 40% [
2]. Patients with metastatic disease who are treated with palliative chemotherapy using a combination of 5-Fluorouracil (5-FU) and Cisplatin have a median survival of less than one year and response rates of 25% to 45% in phase II and III trials [
4,
5]. Salvage options for patients with refractory EC are very limited [
6]. Unfortunately, there are no clinically approved targeted therapies for the treatment of esophageal carcinoma.
Human epidermal growth factor receptor 2 (HER-2) is overexpressed in breast and other types of human cancers and has been successfully developed as a therapeutic target [
6,
7]. Antibody-based therapy with Herceptin (Trastuzumab) has been used for the clinical treatment of HER-2-positive breast cancer [
6,
7]. Recently, Trastuzumab therapeutic efficacy has also been observed in HER-2-overexpressing gastric cancer [
7].
Previous studies in patients with ESCC indicated that the frequency of HER-2 protein expression ranged from 0 to 56% [
8,
9], whilst the frequency of HER-2 gene amplification ranged from 5 to 35% [
8,
9]. These data suggested that a proportion of ESCC patients could be candidates for Trastuzumab targeted therapy. Results from early clinical trials demonstrated a correlation between HER-2 expression and gene amplification with Trastuzumab therapeutic efficacy in patients with esophageal adenocarcinomas [
10]. However, over expression and amplification of HER-2 appeared to be fundamentally different within esophageal squamous cell carcinoma, with a tendency of lower positive rates and lower level amplification, compared to adenocarcinoma [
11]. Accordingly, it is not possible to imply Trastuzumab efficacy within esophageal squamous cell carcinoma, and as such, these studies aimed to address this preclinically. Therefore, we established xenograft mouse models derived from patient’s ESCC tissues and used these novel, clinically relevant ESCC models to explore the anti-tumor efficacy of Trastuzumab targeted therapy.
Materials and methods
Patients and tissue samples
ESCC tissues from 54 treatment-naïve patients were obtained intraoperatively during esophagetomy resection at Zhongshan Hospital (Shanghai, China) from March 2010 to June 2010. ESCC histology was confirmed by pathological analysis. Prior written informed consent was obtained from all patients and the study protocol was approved by the ethics committee at Zhongshan hospital, Fudan University. Freshly harvested ESCC specimen was separated into three parts. The first part was placed into medium containing antibiotics immediately after surgical resection under sterile conditions and transported to the animal facility within two hours for implantation into immunodeficient mice. The second part of the specimen was snap frozen immediately in liquid nitrogen for DNA/RNA extraction. The third part of the specimen was fixed in formalin and embedded into paraffin (FFPE) for pathological and IHC analysis.
Development of patient-derived esophageal carcinoma xenograft (PDECX) mouse models
Female nude (nu/nu) (8-10-week-old) and severe combined immune deficient (SCID) mice (Vital River, Beijing, China) were used for xenograft model generation. All experiments with immunodeficient mice were performed in accordance with the guidelines approved by IACUC. The PDECX mouse models were established with fresh ESCC tissues surgically removed from EC patients. Briefly, surgically removed patient’s ESCC tissues (F0 tissue) were cut into fragments of approximately 15 mm3 and implanted via Trocar needle subcutaneously into female SCID mice within two hours after the surgery. The implanted mice were observed daily until 90 days. Tumors were measured once a week by caliper to determine its subcutaneous growth. The xenografted ESCC tumors (~500 mm3) were harvested from the patient ESCC-bearing mice and were further implanted in female nude mice for expansion. After three consecutive mouse-to-mouse passages, the xenograft was considered to be stabilized and was submitted to the process of model characterization, including histopathologic analysis, HER-2 expression by FISH and IHC assays, and mutation detection for EGFR, K-ras, B-raf and PIK3CA genes. The xenograft tumor specimens in each passage of tumor-bearing mice were harvested and divided into three parts as same as the patient tumor samples. Fresh tumor fragments at passage 3–5 were frozen with 20% FCS in liquid nitrogen for further model recovery. The PDECX models were maintained in nude mice and were used for anti-tumor efficacy studies.
H&E and immunohistochemical staining
Tissues from all PDECX mouse models and their corresponding ESCC patients were harvested and fixed in 10% buffered formalin within 30 minutes after resection. Tissues were processed following the routine procedure after 24 hours fixation. Sections were stained with Hematoxylin and eosin and reviewed by a pathologist to confirm the ESCC diagnosis. Dual cores from each case were made on tissue microarray (TMA) for IHC and FISH analysis. For the IHC, all incubations were at room temperature and washes were performed with TBST. Tissues sectioned at 4 μm onto slides were dewaxed and rehydrated. Antigen retrieval was performed in a pressure cooker at 110°C for 5 minutes in retrieval buffer (S2367, DAKO). Endogenous peroxidase activity was blocked with 3% hydrogen peroxide (S2023, DAKO). Sections were then incubated with anti-HER-2 antibody (HerceptestTM, DAKO) for 30 minutes. Immunocomplexes were detected by incubation with DAKO K5204 for 30 minutes and were visualized with diaminobenzidine (K3468, DAKO) for 10 minutes. Normal IgG from the same species of primary antibody diluted to match the concentration of the primary antibody was used as the negative control. For the HER-2 negative cases, the experiment was repeated on the whole section in order to exclude heterogeneity.
In the present study, the IHC scoring on EC followed Hoffman’s criteria on human gastric cancer [
12]: no staining or <10% tumor cell positive staining were considered as 0/negative; faintly or barely perceptible staining on >10% tumor cell membrane were considered as 1+; weak to moderate positive staining on >10% tumor cells were considered as 2+; since tumor tissue on TMA mimicked the biopsy samples, cohesive moderate to strong staining on the membrane was scored as 3 + .
HER-2 FISH assay
HER-2/CEP17 probes for Bysis (Cat #30-161060) was used and FISH assay was performed according to the manufacturer’s instructions. In brief, TMA sections were dewaxed and dehydrated. Tissue samples were pretreated following the manual of SpotLight Tissue pretreatment kit (Invitrogen, Carlsbad, CA) (Cat #00-8401). After pretreatment, 10 μl HER-2/CEP17 probes were applied onto each TMA slide, which was subsequently covered with a coverslip and circled with rubber cement (Mpbio, Solon, OH) (Cat #11FIXO0125) around the edge of coverslip. Sections together with probes were codenaturated at 75°C for 4 minutes on OmniSlide In-Situ hybridization System, followed by incubation at 37°C overnight. Probes were washed sequentially with 0.3%NP40/1 × SSC at 75.5°C for 3 min, 0.3%NP40/2 × SSC at 75.5°C for 2 minutes and 2 × SSC at room temperature for 2 min, with twice for each wash. The slides were then dehydrated and mounted with 0.3 μg/ml DAPI mounting medium (Vector, Burlingame, CA) (Cat #H-1200). In each case, 100 tumor nuclei were evaluated based on Hoffmann criteria [
12] and other investigator [
9]. An absolute HER-2 gene copy number lower than four or an HER-2/Chr17 ratio of less than 1.8 was considered as negative for HER2 gene amplification. Cases with average HER-2 gene copy number >6 or a HER-2/CEN17 fluorescence ratio > or = 2 were considered as positive for gene amplification. Cases with HER-2 copy number between four and six or an HER-2/Chr17 ratio between 1.8 and 2.0 were considered as HER-2 equivocal.
Anti-tumor efficacy of Trastuzumab treatment
Tumor growth curves were generated by kinetic measurement of tumor volumes. For therapeutic experiments, tumor-bearing mice with a tumor range of 100 to 200 mm3 were randomly divided into the vehicle control or Trastuzumab treatment groups (8 animals per group). The mice in the treatment group were treated with monotherapy of Trastuzumab (Roche, China, 4 mg/kg/twice a week intraperitoneally) or combination of Trastuzumab with AKT inhibitor AZD5363 (120 mg/kg/twice daily, orally). Subcutaneous tumors in nude mice and mice body weight were measured twice a week. Tumor volumes were calculated by measuring two perpendicular diameters with calipers. Tumor volumes (V) were calculated by the formula: V = (length + [width]2)/2. Percentage of tumor growth inhibition (%TGI = 1 – [changes of tumor volume in treatment group/ changes of tumor volume in control group] × 100) was used for the evaluation of anti-tumor efficacy. For tumor regression, in which the tumor volume after treatment was smaller than the tumor volume at the beginning, the following equation was used: regression% = 100 × (T0-Ti)/T0, Ti and T0 was the tumor volume in the same group but measured at the different time points. T0 is the tumor volume on the day before first dosing of the compound, and Ti is the tumor volume at the last measurement day after compound treatment.
Screening of gene mutations
Formalin-fixed paraffin-embedded tumor blocks were reviewed for quality and tumor content. A single representative block, from PDECX mouse models and the corresponding patient ESCC samples, containing at least 70% of neoplastic cells, was analysed for gene mutations. Genomic DNA was extracted using the QIAamp Mini kit (Qiagen, Valencia, CA) according to the manufacturer’s instructions. Hotspots in exon 18, 19, 20, 21 of EGFR gene, exon 2 and 3 of K-ras gene, exon 11 and 15 of B-raf gene, and exon 9 and 20 of PIK3CA gene were screened by scorpions amplification refractory mutation system (ARMS) and mutant-enriched liquid chip PCR method [
13]. The former method and detection was supported by Amoy diagnostics Co. Ltd, Fujian, PR China, the latter method was completed by SurExam Bio-Tech Co. Ltd., Guangzhou Technology Innovation Base (Science City, Guangzhou, PR China). Positive results in both methods were further confirmed by PCR-direct sequencing. The primers for exon 9 of PIK3CA gene were GTCTTAGATTGGTTCTTTCCTG for forward and GCATTTAATGTGCCAACTACC for reverse, for exon 20 of PIK3CA gene were TTTGTCTACGAAAGCCTCTCTA for forward and CCATCACTTTTTCCTTCTCCAT for reverse.
Western blot (Wb)
Total proteins were extracted with mammalian protein extraction reagent (Beyotime, P0015, China). 80 μg of total protein was electrophoresed in 4%-12% Criterion XT Bis-Tris gels (Bio-Rad) and transferred to polyvinyl difluoride (PVDF) membranes (Millipore) at 340 mA for 75 min. Membranes were blocked with 5% non-fat milk in TBST buffer and probed with specific antibodies. All primary antibodies were used at a final concentration of 1 μg/ml. The blots were then visualized with a chemiluminescent detection system as described by the manufacturer. Antibodies against HER-2 (#06-562) was purchased from Upstate Biotechnology; p-Mtor-Ser 2448 (No. 2971) from DAKO Biotechnology, and mTor (No 2972), AKT (No 9272), p44/42 MAP Kinase (No 9102), S6 (No 2317), 4EBP1 (No 9452) and p-AKT-Ser 473 (M3628), Phospho-p44/42-MAPK-Thr202/Tyr 204 (No 4370), p-S6-Ser 240/244 (No 2215), p-4EBP1-Thr 70 (No 9455), and anti-GAPDH were purchased from CST Biotechnology.
Statistical analysis
Statistical significance was evaluated using a one-tailed, two sample t test. P values of less than 0.05 were considered to be significant.
Discussion
It is well established that patient-derived xenograft mouse models can better represent human tumors through increased diversity of molecular lesions and the preservation of three-dimensional tumor-stromal cell components and interactions [
15]. Although some patient-derived xenograft models, e.g., NSCLC [
16,
17], melanoma [
18], colon cancer [
19], breast cancer [
20], HCC [
21,
22], GIST [
23], have been established and used for evaluation of anti-tumor compounds [
24], the availability of patient-derived EC xenograft models is limited. Previous studies have however, developed xenograft EC models by implanting human EC cell lines into mice [
25,
26] and the first objective of this study was to develop novel PDECX models. After successfully establishing 25 PDECX mouse models through implantation of 54 patient EC samples, 5 PDECX mouse models were selected for evaluation of Trastuzumab anti-tumor efficacy based on HER-2 expression levels. Although the five models displayed varying growth rates (Figure
1), this was not related to differentiation and HER-2 positivity (Table
1). Indeed, the varying growth rates of patient-derived models are likely related (at least in part), to the requirement to directly implant tumor fragments, rather than standardized defined numbers of tumor cells.
Trastuzumab anti-tumor efficacy correlates well with HER-2 expression in patients with breast [
6,
27‐
29] and gastric cancer [
7,
24,
30]. Esophageal carcinoma represents another disease with a high frequency of tumor HER-2 expression and gene amplification [
8,
9,
11]. In this setting, the frequency of patients with tumor HER-2 positive expression ranges from 0 to 56% [
8,
9], whilst the frequency of HER-2 gene amplification varies from 5% to 35% [
8,
9,
11]. In addition, the predominant histological subtype of esophageal carcinoma differs between Eastern and Western countries, and adenocarcinoma has been shown to exhibit higher rates of HER-2 protein expression positivity and gene amplification compared to squamous cell carcinoma [
11]. Accordingly, most current preclinical and clinical studies are focused on esophageal adenocarcinoma and literature data supports a correlation between HER-2 gene expression (or amplification) and Trastuzumab efficacy in esophageal adenocarcinoma (preclinical and clinical phase I/II data [
10,
31] however, such data is rare in esophageal squamous cell carcinoma [
32]. In order to address the therapeutic efficacy of Trastuzumab in esophageal squamous cell carcinoma preclinical studies, we characterized 5 PDECX mouse models for HER-2 gene amplification and protein expression. In this study, we did not observe strong positive staining for HER-2 expression (3+) or gene amplification (FISH score ≥ 6) in the 25 established PDECX mouse models. Two PDECX mouse models (EC039 and 044) were positive (IHC 2+) for HER-2 expression and had relatively high
HER-2 copy number (~3.9).
HER-2 gene copy number and protein expression were highly consistent between the 5 PDECX mouse models and their corresponding patient’s EC tissues (Figure
2), suggesting that the PDECX mouse models better represent clinical EC disease, compared to standard cell line-derived xenografts.
Clinically, approximately 20% of esophageal adenocarcinoma patients who were HER-2 IHC 2+ and FISH negative demonstrated a complete clinical response [
10]. Mimura
et al. have also demonstrated that HER-2 expressing esophageal SCC cells could be killed by Trastuzumab-mediated ADCC and that anti-proliferative activity reflected the degree of HER-2 expression detected by flow cytometry, but not by HercepTest nor fluorescence in situ hybridization analysis. To determine if HER-2 positive (IHC 2+) and FISH negative esophageal squamous cell carcinoma could be responsive to Trastuzumab monotherapy, we tested Trastuzumab anti-tumor efficacy in 5 established PDECX mouse models with varying degrees of HER-2 expression and gene copy number. Unsurprisingly, there was no tumor regression, and no or marginal response to Trastuzumab treatment in the HER-2 negative PDECX models. Our data demonstrated that Trastuzumab induced complete tumor regression in one of the HER-2 positive models (IHC 2+) (Figure
3 and Table
1), but generated no significant anti-tumor response in the other. Whilst we believe that this potent anti-tumor effect is a direct consequence of anti-proliferative activity, we note that Trastuzumab has been observed to exert both mild anti-angiogenic and anti-metastatic effects [
33] however, we were unable to assess changes in these phenotypes within the studies documented here.
To explore the underlying mechanism behind these contrasting Trastuzumab responses, additional genetic changes were analyzed. It is well known that some key gene mutations including EGFR, K-ras, B-raf, PIK3CA, often affect the efficacy of molecular-targeted therapies [
34,
35]. In clinical studies of metastatic colon cancer, patients with wild type K-ras, B-raf and PIK3CA genes gain benefit from cetuximab therapy[
34], whilst lung cancer patients with mutations in exons 19 and 21 of the EGFR gene benefit from Iressa or Tarceva therapy. Furthermore, mutations in ESCC patients have been recently reported; 16% of ESCC patients (5/30 cases) harbor K-ras gene mutations [
36], 14% of ESCC patients (7/50 cases) contain EGFR mutation [
37], and 2.2-11.8% of ESCC patients have gene mutations in exon 9 of the PIK3CA gene [
38,
39]. These findings suggest that gene mutation(s) may play a key role in resistance to Trastuzumab treatment. After screening for EGFR, K-ras, B-raf and PIK3CA gene mutations in the 5 PDECX mouse models and their corresponding primary tumors, we found that model EC039 (and the corresponding patient tissue) had a helical domain mutation (G1624A, E542K) in exon 9 of the PIK3CA gene (Figure
4 and Table
1). This mutation is known to activate downstream signaling of the AKT pathway [
40,
41]. As such, this result likely explains why the HER-2 positive (IHC 2+) model EC039 was unresponsive to Trastuzumab treatment, whilst model EC044 (HER-2 positive (IHC 2+)) displayed tumor regression. PIK3CA activating mutations have been reported to associate with shorter time to progression (TTP) in Trastuzumab-treated breast cancer [
42]. More recently, particular attention has been paid to the role of the PIK3CA gene in Trastuzumab resistance. Mutations in exon 9 and 20 of the PIK3CA gene have been observed in 8% to 40% of breast cancers [
43‐
45]. Furthermore, an association between PIK3CA mutation and resistance to Trastuzumab therapy in HER-2-amplified breast cancer cell lines has also been reported [
46]. Berns
et al. have also suggested that oncogenic mutations of PIK3CA may render breast cancers more resistant to treatment with Trastuzumab [
47], and Eichhorn
et al. have further shown that mutational activation of PIK3CA can similarly render cells more resistant to the recently approved anti-HER2 agent Lapatinib. In this situation however, Lapatinib resistance caused by the dominant activating mutations in PIK3CA can be effectively reversed by treatment with the PI3K/mTOR inhibitor, NVP-BEZ235 [
48]. Moreover, the AKT inhibitor AZD5363 increases the efficacy of lapatinib and/or trastuzumab in xenografts derived from HER2+ breast cancer cell lines harboring PIK3CA mutation [
14]. In our study, we observed strong AKT pathway activation in the PIK3CA mutation model EC039 (Figure
5A) compared to wild type model EC044, suggesting that activated AKT pathway signaling likely contributes to Trastuzumab resistance in our efficacy study. To our knowledge, we are the first to describe a PIK3CA gene mutation which causes resistance to Trastuzumab therapy in a preclinical PDECX mouse model, and which can be partly resensitised using a combination of Trastuzumab and AKT inhibitor (Figure
5B). The inability to achieve full tumor regression in this combination study may be a consequence of only blocking PI3K pathway signaling downstream through AKT, thus still allowing PDK1, PKC and Rac signaling to occur.
In summary, we have established novel xenograft models derived from patient esophageal squamous cell carcinoma, and have used these clinically relevant animal models to investigate Trastuzumab efficacy in HER-2 expressing ESCC. Our results demonstrate that a non-amplified, HER-2 IHC 2+ model responds completely to Trastuzumab targeted therapy, and importantly, that PIK3CA gene mutation negatively associates with Trastuzumab anti-tumor efficacy in a xenograft model of ESCC. Moreover, resensitization to Trastuzumab therapy is achieved when treatment is combined with an AKT inhibitor to suppress PIK3CA pathway signaling. Taken together, this study illustrates that Trastuzumab can induce regression in a HER-2 positive tumor, and that PIK3CA mutation (in a HER-2 positive background) may be a potential resistance mechanism to Trastuzumab treatment in pre-clinical patient-derived EC xenograft models. We believe that this pre-clinical observation and hypothesis warrants further investigation in a clinical study.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
WX performed the mouse model studies, ZJ and JQ participated study design and drafted the manuscript. PG revised the manuscript. ZR and XS collected the fresh tissue specimens. LJ and XL performed PDECX model establishment and efficacy studies. ZL and HD participated in immunochemistry work. SX and YX completed the FISH studies, whilst ZG and TY participated in the sequence alignment. LS performed the statistical analysis and HJ checked the results. LY and XC finished the PCR experiments. HY and GD conceived of the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.