Association of fibroblast growth factor receptor 1 gene amplification with poor survival in patients with esophageal squamous cell carcinoma

Purpose To investigate whether FGFR1 gene amplification is associated with clinicopathologic characteristics and its potential impact on survival in patients with resected esophageal squamous cell carcinoma (ESCC). Methods Five hundred fifty-six ESCC patients undergoing curative resection of ESCC were retrospectively studied. FGFR1 gene copy number was determined in microarrayed tumor samples using fluorescent in situ hybridization (FISH) analysis. FGFR1 gene amplification status was prespecified as copy number ≥ 6 or FGFR1/CEN 8 ratio ≥ 2.2. FGFR1 expression was evaluated by immunohistochemistry. Overall survival (OS) and disease-free survival (DFS) were analyzed using the Kaplan-Meier method followed by the log rank test. Correlation with survival was examined using multivariate Cox regression. Results FGFR1 amplification was identified in 67 (12.1%) patients; these patients had significantly shorter OS (50.0 vs 32.0 months; log rank; P<0.001) as well as shorter DFS (47.0 vs 28.0 months; log rank; P<0.001) than those without FGFR1 amplification. Under a Cox proportional hazard model, FGFR1 amplification was associated with significantly shorter OS (adjusted hazard ratio [AHR]=1.61; 95% CI, 1.10-2.43, P=0.004) and DFS (AHR=1.72; 95%CI, 1.15-2.48; P<0.001). Moreover, cases with high intratumoral FGFR1 expression showed significantly shorter OS and DFS than those with low FGFR1 expression. The frequency of FGFR1 amplification was significantly higher in heavy drinkers than in moderate and light drinkers. Conclusion FGFR1 amplification is an independent adverse prognostic factor in surgically resected ESCC. FGFR1 may be a promising therapeutic target in patients with ESCC.


Research Paper
Barrett's esophagus, and mainly affects Caucasian population [4]. The two main subtypes of esophageal cancer have considerably different epidemiological features but share the same dismal prognosis despite recent advances in imaging and surgical techniques [5].
Over the last decade, molecularly stratified therapy have afforded benefits to patients with many types of cancer but, unfortunately, the same is not true for ESCC [6]. Many genomic abnormalities have been described in ESCC and there is growing evidence of their biological significance [7,8]. Further delineation of genetic alterations may help uncover aberrant molecular pathways, novel biologic markers and tumorigenic pathways, and eventually allowing successful targeted therapy. Fibroblast growth factor receptor 1 (FGFR1), whose gene is located at 8p12, is a member of the FGFR family of receptor tyrosine kinases (FGFR1-4). FGFRs activation leads to downstream signaling via the PI3K/ AKT/RAS/MAPK pathways, which are essential to cell growth, survival, migration, and angiogenesis. FGFR1 amplification has been identified in breast cancer, head and neck squamous cell carcinoma, ovarian cancer, ESCC, bladder cancer and lung cancer [9][10][11][12]. Existing studies on the prognostic impact of FGFR1 amplification for ESCC have yielded conflicting results. Loga et al. investigated the prevalence of FGFR1 amplification in a tissue microarray containing 346 esophageal adenocarcinomas and 254 ESCCs using dual-labeling fluorescent in situ hybridization (FISH) analysis and found that FGFR1 amplification correlated with the histologic subtype of ESCC (9.4% vs. esophageal adenocarcinoma 1.6%, P<0.001) [13]. However, they failed to demonstrate an association between FGFR1 amplification and clinical outcome. Kwon et al. studied 180 patients with resected ESCC and found FGFR1 amplification in 21.4% (37/173) patients; they observed that FGFR1 amplification was an independent predictor of prolonged OS in these patients [14]. Kim et al. investigated 526 curatively resected ESCC using FISH for FGFR1 amplification and found that an association between high FGFR1 amplification (defined as an FGFR1/centromer 8 ratio ≥ 2.0, or average number of FGFR1 signals/tumor cell nucleus ≥ 6.0, or percentage of tumor cells containing ≥ 15 FGFR1 signals or large cluster in ≥ 10% with significantly shorter disease-free survival (DFS) and OS [15]. FGFR1 inhibition in cell lines and mouse models with FGFR1-amplified engrafted tumors suppressed tumor cell growth and induced apoptosis, suggesting that FGFR inhibitors may be an effective therapeutic option in SCCs with FGFR1 amplification [11,16,17]. Furthermore, FGFR1 inhibitors (e.g. dovitinib) have entered early stage clinical trials in patients with solid tumors [18,19].
In this study, we examined FGFR1 amplification status and analyzed the impact of FGFR1 amplification on the OS and DFS in 556 ESCC patients who received radical resection of curative intent at our institution.

Demographic and baseline characteristics
A total of 687 patients underwent radical resection for ESCC during the study period. 556 cases with undisputable survival data were included in data analysis in the current study ( Figure 1). The demographic and baseline characteristics of the study participants are shown in Table 1

FGFR1 amplification
Among a total of 556 cases, 67 (12.1%) were positive for FGFR1 amplification (Supplementary Figure  1). The median FGFR1 gene copy number in all patients was 3 (range, 2 to 17 copies per cell). The median FGFR1 gene copy number was 10 (range, 6 to 17) in ESCC patients with FGFR1 amplification and 2 (range, 2 to 5) in those without FGFR1 amplification. The median FGFR1/CEN8 ratio was 3.6 (range, 2.2 to 8.5) and 1.4 (range, 1.0 to 2.0) for ESCC patients with or without FGFR1 amplification, respectively. ESCC patients with and without FGFR1 amplification were comparable in the demographic and baseline variables except that the percentage of heavy drinkers was significantly higher in patients with FGFR1 amplification than those without FGFR1 amplification (86.6% vs 43.1%, P<0.001) ( Table  1).

Survival and FGFR1 amplification
The median follow-up duration for the study cohort was 47.0 months (range, 9.0 to 137.0 months). As expected, both of the median OS and DFS were longer in patients with stage I to II disease than in those with stage III disease (73.0 vs 32.0 months; P<0.001; 68.0 vs 28.0 months; P<0.001). Kaplan-Meier analysis showed that ESCC patients with FGFR1 amplification had significantly shorter OS (32.0 months; range, 11.0 to 116.0 months) than those without FGFR1 amplification (50.0 months; range, 9.0 to 137.0 months) (log rank; P<0.001) ( Figure 2

FGFR1 expression by IHC
In addition to gene copy number analysis, we also evaluated the correlation between FGFR1 amplification and FGFR1 expression by using IHC (Figure 3). The mean IHC score for FGFR1 was 34.1 ± 39.5 (range, 0 to 242) for the entire study sample. A cutoff IHC score of 62 was used to stratify ESCC patients into the high FGFR1 expression group (n=81) and the low FGFR1 expression group (n=475). ESCC patients with high FGFR1 expression and those with low FGFR1 expression were comparable in the demographic and baseline variables except that the percentage of heavy drinkers was significantly higher in patients with high FGFR1 expression than those with low FGFR1 expression (79.0% vs 43.2%, P<0.001) (Supplementary Table 1

FGFR1 amplification and patient response to adjuvant chemotherapy
We assessed the impact of neoadjuvant and adjuvant chemotherapy on postoperative survival according to histologic stage. Among 52 patients who received neoadjuvant chemotherapy, one was in stage II while the rest were in stage III. Therefore, we compared survival in subjects with stage III disease (n=51 and 68 for those with vs without neoadjuvant chemotherapy, respectively). No significant difference in OS (34.

Alcohol intake and FGFR1 amplification
Given the high percentage of heavy drinkers (48.4%) in the study sample, we examined the rate of FGFR1 amplification according to alcohol intake. We found higher rate of FGFR1 amplification in heavy drinkers (21.5%) than light (1.0%) or moderate drinkers (5.3%). Furthermore, the incidence of FGFR1 amplification was 26.1% in patients whose alcohol intake was more than 200 g/day, 25.0% in patients whose alcohol intake was 100.1 to 200 g/day, 16.4% in patients whose alcohol intake was 30.1 to 100 g/day, 5.3% in patients whose alcohol intake was 15.1 to 30 g/day, and 1.0% in patients whose alcohol intake was less than 15 g/day, suggesting increasing rate of FGFR1 amplification with increment of alcohol intake (P trend < 0.001) ( Figure 6).

DISCUSSION
FGFR1 is an important signaling molecule implicated in multiple cellular process including cellular growth and survival as well as angiogenesis. FGFR1 amplification has been documented in ESCC and other cancer types; however, it still remains controversial whether FGFR1 amplification adversely impacts on the clinical outcome of ESCC patients [13][14][15]. In this study, we investigated a large cohort of 556 ESCC patients who received radical resection of curative intent at our institution and whose FGFR1 amplification was known by FISH. We demonstrated that FGFR1 amplification was an independent adverse prognostic predictor of OS and DFS of ESCC patients. This finding was further supported by our immunohistochemical analysis showing that high intratumoral FGFR1 expression correlated with significantly shorter OS and DFS of ESCC patients. FGFR1 amplification status has been investigated in several previous studies using different approaches such as comparative genomic hybridization, single nucleotide polymorphism array, and FISH assay [20][21][22][23]. Unlike EGFR mutation in lung cancer, which is more prevalent in Asians than in Caucasians, the frequency of FGFR1 amplification in ESCC does not seem to be widely different by ethnicity. The frequency of FGFR1 amplification, as detected by FISH analysis in our cohort, was 12.1%, which was higher than 9.4% reported by Loga et al. for a Caucasian ESCC patients cohort, [13] but lower than 21.4% for a small Korean cohort reported by Kwon et al [14].
In head and neck squamous cell carcinoma, FGFR1 amplification was significantly associated with poor prognostic factors such as higher T stage, and visceral metastasis [24]. In lung squamous cell carcinoma, Kim et al. found that FGFR1 amplification was an adverse prognostic factor while Tran et al. reported the opposite outcomes [25,26]. Tumor heterogeneity, varied assaying methods, lack of consensus criteria for FGFR1 amplification and small sample size may contribute to the controversial results. Current evidence on the prognostic significance of FGFR1 amplification for ESCC also remains inclusive. Two studies of smaller cohort size yielded conflicting results. Loga et al. found no association between FGFR1 amplification and clinical outcome of ESCC patients while Kwon et al. showed that FGFR1 amplification was an independent predictor of prolonged OS of ESCC patients [13,14]. Kim et al. investigated 526 curatively resected ESCC and showed that high FGFR1 amplification was associated with significantly shorter DFS and OS [15]. Consistently, our study of 556 ESCC patients also showed that FGFR1 amplification was independently associated with worse OS and DFS. On the basis of previous studies, we adopted a criteria including both of gene copy number and FGFR1/CEN8 ratio in our study [23]. High FGFR1/CEN8 ratio has been used to screen patients of squamous cell lung cancer who may benefit from treatment with FGFR inhibitors [27,28]. Apart from genetic evidence, we further showed that high intratumoral FGFR1 expression was also associated with shorter OS and DFS, lending support to the proposition that FGFR1 amplification predicts poor clinical outcome of ESCC patients.
The genomic landscape of ESCC is highly complex [29]. Delineation of aberrant signaling pathways in ESCC has identified candidate molecular targets such as HER2, EGFR, VEGF, FGFR and PI3K. However, lack of reliable prognostic predictors hampers development of targeted therapies for ESCC. Our study has demonstrated that FGFR1 amplification is an independent adverse predictor of OS of ESCC patients, suggesting that FGFR1 may be a promising molecular therapeutic target [30]. Defining therapeutic targets for subgroups of ESCC patients is critical to advancing treatment of this disease. Currently, FGFR1 amplification may define one large subgroup, and efforts to target this population are already ongoing with some clinical trials already in progress. FGFR1 inhibitors that are currently in clinical development are shown in Supplementary Table 2. Dovitinib, a potent FGFR inhibitor, has demonstrated antitumor activity in heavily pretreated patients with FGFR pathway-amplified breast cancer but has not been studied in clinical trials for ESCC patients [31].   Patients who would benefit from postoperative adjuvant chemotherapy in ESCC are still undefined [32][33][34]. Our data showed the benefit of adjuvant chemotherapy in patients with stage III ESCC. Patients with FGFR1 amplification benefitted from adjuvant chemotherapy while patients without FGFR1 amplification did not. However, this result should be cautiously interpreted because of the small sample size and, in addition, FGFR1 amplification may be a local manifestation of global genomic instability. Concurrent genetic alterations in other regions of the genome associated with DNA damage/ repair may contribute to greater benefit from adjuvant chemotherapy [35]. Therefore, comprehensive evaluation of genomic alterations will be needed to develop optimized treatment decisions for patients with resected ESCC.
Alcohol intake is an established risk for many types of cancer including ESCC while how it may affect the gene is still unknown [21,36]. A novel finding of the    current study is significant association between FGFR1 amplification and alcohol intake. We also found that the incidence of FGFR1 amplification rose as the amount of alcohol intake increased. Our findings suggest that FGFR1 amplification may be a major oncogenic aberration in ESCC that is induced by alcohol abuse. Interestingly, among heavy drinkers with more than 100g per day, approximately 20% showed FGFR1 amplification. On the basis of this finding, ESCC patients with alcohol consumption more than 100 g/day may be targeted for screening for FGFR1 amplification.
The main limitations of this study reside in its retrospective nature and the fact that patients were selected from a single tertiary care institution. Also, clinical decisions were not likely uniform in such retrospective study, and could have produced biases. Future studies are required to validate our findings in an independent cohort, especially in a Caucasian population.
In this study, we demonstrated that FGFR1 amplification is an independent adverse prognostic factor in surgically resected ESCC, indicating that FGFR1 amplification may be a relevant therapeutic target in ESCC. FGFR1 amplification was positively associated with alcohol consumption, which strongly implies that FGFR1 amplification is an oncogenic aberration caused by alcohol abuse.

Patients
We retrospectively analyzed the clinicopathologic data of patients with pathologically proven primary thoracic ESCC who received radical resection of curative intent at Shandong Provincial Hospital affiliated to Shandong University, Jinan, China, between January 2004 and December 2009. Patients with incomplete survival data were excluded. Archived tumor specimens were obtained from the Tissue Bank of Shandong Provincial Hospital. ESCC was staged according to the American Joint Committee on Cancer (seventh edition) guidelines.
The study protocol was approved by the Ethics Committee of Shandong Provincial Hospital. Patient consent was not required because of the retrospective nature of this study. Acquisition of tissue specimens was approved by the Shandong Provincial Hospital Ethics Committee and performed in accordance with the established institutional and state guidelines.

Patient assessment
We retrieved data, including age, gender, comorbidities, smoking and drinking status, TNM stage, OS and DFS, from the medical records of eligible patients. Never-smokers were defined as those smoking fewer than 100 cigarettes in lifetime; former smokers were those who had stopped smoking for more than one year, and current smokers were those who currently smoke or quit smoking for less than one year [37]. Alcohol intake was categorized into light drinker (0-15g/day), moderate drinker (15.1-30g/ day) and heavy drinker (>30 g/day) [38]. Diabetes was defined as fasting blood glucose ≥ 7.0 mmol/L or 2-hour plasma glucose ≥ 11.1 mmol/L or previously diagnosed diabetes. Hypertension was defined as seated resting systolic blood pressure ≥ 140 mmHg or diastolic blood pressure ≥ 90 mmHg, or taking an anti-hypertensive medication [39].

Follow-up assessment
All patients were followed up every 3 months for the first 3 years, every 6 months for the following 3 years, and thereafter annually. Follow-up was done at the Out-Patient Department of Shandong Provincial Hospital or by phone calls or postal mails if the patients failed to appear or missed appointments. Survival was measured from the date of surgery to the date of death or the last follow-up. The last follow-up visit was carried out in June 2015. DFS was defined as the length of time from the date of surgery to the date of initial tumor relapse (local/ distant recurrence) or death as a result of any cause. OS was calculated from the time of surgery to death or the last follow-up date.

Gene copy number analysis
Paraffin-embedded tumor specimens were used to construct a tissue microarray with 2-mm-diameter cores. Two pathologists (Y.J.Q. and Q.X.W.) confirmed the diagnosis of ESCC by hematoxylin and eosin (H&E) staining. FISH assay was performed on the tissue microarrays by using a ZytoLight SPEC FGFR1/ CEN 8 Dual Color Probe (ZytoVision, Bremerhaven, Germany) according to the manufacturer's recommended protocol. All samples were independently analyzed by two evaluators (Y.J.Q. and Q.X.W.) who were blinded to clinical data using a fluorescence microscope (DM4000, Leica). At least 100 nuclei per sample were evaluated. FGFR1 amplification cases were defined as harboring a gene copy number ≥ 6 or FGFR1/CEN 8 ratio ≥ 2.2 [40]. Images were produced using the AxioCamMRm CCD camera and Axiovision v4.5 software suite.

Immunohistochemistry
Immunohistochemical analysis was performed using anti-FGFR1 antibody (GTX100264, GeneTex, U.S.A.) following conventional methods. FGFR1 expression levels were scored semi-quantitatively independently by two pathologists (Y.J.Q. and Q.X.W.) who were blinded to patient information. Only clear membranous staining of tumor cells was considered to be positive; cytoplasmic or granular staining was considered negative or trace.
The percentage of positive tumor cells per core (0% to 100%) was multiplied by the dominant staining pattern (1, negative or trace; 2, weak; 3, moderate; 4, intense). The overall score theoretically ranged from 0 to 400 [41]. The cutoff value for IHC scores was determined using the X-tile software (version 3.6.1) to stratify ESCC patients into the high and low FGFR1 expression groups [42]. Cases with discordant results underwent a consensus review by a third pathologist (Z.G.Y.).

Statistical analysis
Patient demographic and baseline characteristics are expressed as mean (SD) or percentage (%) and compared using Student's t-test and χ 2 test, where appropriate. The primary end point of OS and secondary end point of DFS are presented as median and 95% confidence interval (95%CI), and analyzed using the Kaplan-Meier method, followed by log rank test. Statistical significance was set at P<0.05. All statistical analyses were performed by using SPSS 19.0 software.

Author contributions
Dong Wang, Licheng Du and Zhou Wang designed this study and wrote this paper with assistance from Xiangyan Liu, Zhe Yang and Yejun Qin. Zhou Wang, Yang Jia, Xueqing Wang, Liang Qiao and Zhenchuan Liu helped collect clinical data. Dong Wang, Licheng Du, Huaxia Chen, Zhaohua Xiao and Zhenchuan Liu performed tissue analyses with assistance from Yejun Qin, Qiangxiu Wang and Zhigang Yao. Statistical analyses were performed by Mo Shi and Bin Shang with assistance from Zhou Wang and Huaxia Chen.