RESEARCH ARTICLE

Expression of basic fibroblast growth factor, CD31,and α-smooth muscle actin and esophageal cancerrecurrence after definitive chemoradiation

Yongshun Chen & Xiaohong Li & Haijun Yang & Yubing Xia &

Leiming Guo & Xiaoyuan Wu & Chunyu He & You Lu

Received: 12 January 2014 /Accepted: 17 April 2014# International Society of Oncology and BioMarkers (ISOBM) 2014

Abstract There is cumulative evidence that stromal reactionin cancer has an important diagnostic and prognostic signifi-cance. The aims of this study were to analyze the expressionof basic fibroblast growth factor (FGF-2), CD31, and α-smooth muscle actin (SMA) in esophageal cancer patientsand to establish their significance as indicators of diseaserecurrence after definitive chemoradiation (CRT). Protein ex-pressions of FGF-2, CD31, and SMA were evaluated byimmunohistochemistry and Western blot analysis in 70 pa-tients, 20 with esophageal squamous cell carcinoma (ESCC)and 50 with locally recurrent ESCC after definitive CRT.Twenty matched normal esophageal squamous epitheliumwere also studied as controls. Esophageal cancer tissuesshowed positive expression of FGF-2, CD31, and SMA; incontrast, FGF-2 expression was not detected and only littlestaining for CD31 and SMAwas noted in normal epithelium.Protein levels of FGF-2, CD31, and SMA were significantlyelevated in recurrent ESCC. Among the patients with locally

recurrent disease, expression of FGF-2 and SMAwas notablyhigh in whom the tumor recurred locally within 24 monthsafter definitive CRT. The 2- and 5-year local recurrence-freesurvival rate was 15.4 % and 0 in patients with high FGF-2expression, compared with 45.8 and 33.3 % in thosewho expressed low FGF-2, respectively (P=0.005). Ofpatients who expressed high SMA, the 2- and 5-year localrecurrence-free survival rate was 21.7 and 8.7 %, respectively,compared to those with low SMA expression which was 37.0and 22.2 %, respectively (P=0.016). Overexpression of FGF-2 and SMA is associated with local recurrence and reducedrecurrence-free survival after definitive CRT for ESCC. Thedata also suggest that targeting stromal cells may be an attrac-tive approach for esophageal cancer therapy strategies.

Keywords Basic fibroblast growth factor . CD31 .α-smoothmuscle actin . Chemoradiation . Esophageal cancer .

Recurrence

Introduction

There is increasing evidence that the tumor microenvironment,composed of noncancer cells and their stroma, has been recog-nized as a major factor influencing cancer development andprogression [1]. Tumor cells may alter the surrounding connec-tive tissues and modulate the metabolism of fibroblasts, thusresults in the production of a collagenous stroma, and tumorstroma provides a scaffold for tumor cells and vascular supplythat tumor requires for nourishment, gas exchange, and wastedisposal. Therefore, the interaction between tumor and stromalcells has been implicated in the regulation of tumor progres-sion, metastasis, angiogenesis, and the outcome of therapy.

Tumor stroma includes endothelial cells, immune cells, andfibroblasts. In epithelial neoplasms, one of the characteristics oftumor stroma is a pronounced fibrotic response, which consists

Y. Chen :Y. LuDivision of Thoracic Oncology, West China Hospital, Cancer Center,West China School of Clinical Medicine, Sichuan University, 37,Guoxue Lane, Chengdu 610041, China

Y. Chen (*) : L. Guo :X. Wu :C. HeDepartment of RadiationOncology, ZhengzhouUniversity AffiliatedCancer Hospital, Henan Cancer Hospital, 127, Dongming Road,Zhengzhou 450008, Chinae-mail: yongshun2007@163.com

X. LiDepartment of Pathology, Zhengzhou University Affiliated CancerHospital, Henan Cancer Hospital, Zhengzhou, China

H. YangDepartment of Pathology, Anyang Cancer Hospital, Anyang, China

Y. XiaDepartment of Medical Oncology, Kaifeng Cancer Hospital,Kaifeng, China

Tumor Biol.DOI 10.1007/s13277-014-1987-9

of myofibroblasts, excessive deposition of an extracellularmatrix, and neovascularization. Stromal myofibroblasts areconsidered to play a critical role in the complex process oftumor-stroma interactions; studies showed that myofibroblastsare located in the vicinity of esophageal cancer cells and areable to enhance tumor growth by secreting growth factors, suchas transforming growth factor-β, matrix degrading enzymes,such as matrix metalloproteinases, and angiogenic factors, suchas vascular endothelial growth factor [2, 3]. Basic fibroblastgrowth factor (FGF-2) is the prototypic member of a family ofrelated genes encoding heparin-binding proteins with growth,antiapoptotic, and differentiation-promoting activity; FGF-2has been used as a tumor fibroblast-specific marker in esoph-ageal cancer [4].

Activated fibroblasts in tumor stroma are commonly iden-tified by the expression ofα-smooth muscle actin (SMA), andmyofibroblasts represent an activated fibroblast phenotype[5]. Myofibroblasts produce an extracellular matrix enrichedin type III and V collagen, which is responsible for promotingthe progression of malignant tumors [6]. Angiogenesis is acritical process in tumor growth and development; it has beendemonstrated that neoangiogenesis is an independent adverseprognostic factor for many tumors, and CD31 expression iswidely used in quantifying tumor angiogenesis [7].

Worldwide, esophageal cancer is the eighth most commoncancer and the sixth most common cause of death from cancerwith 400,156 deaths in 2012 [8]. Squamous cell carcinomaaccounts for 95 % of all esophageal cancers in China [9].Definitive chemoradiation (CRT) is a curative treatment op-tion for patients with locally advanced esophageal squamouscell carcinoma (ESCC) who are not surgery candidates.Unfortunately, local failure remains a major concern, withpersistent or recurrent disease being reported in 46–68 % ofpatients, and most local failures occur in the gross tumorvolume; early local recurrence of the tumor was associatedwith advanced tumor stage (T3–4 disease) and large tumorsize (>8 cm) [10, 11]. It is not clear whether the alterations inESCC microenvironment can affect local tumor control afterdefinitive CRT. The aims of our study were therefore toanalyze the expression of FGF-2, CD31, and SMA in a cohortof ESCC patients and to determine whether differences in thelevels of these parameters may provide useful informationabout tumor recurrence.

Materials and methods

Case selection

For this study, we selected 70 patients with stage II–III thoracicESCC who underwent surgery at three centers for esophagealand gastric cancer in China between 2005 and 2008. Twentypatients with primary ESCC (16 men and 4 women, aged 42 to

74 years, median 59.5 years) and 50with locally recurrent ESCCafter definitive CRT (43 men and 7 women, aged 38 to 72 years,median 58.0 years) were investigated for FGF-2, CD31, andSMA levels. Twenty normal esophageal tissues which were5 cm apart from the edge of cancer lesion and without docu-mented microscopic invasive cancer were studied as controls.This study was approved by the Ethics Committee ofZhengzhou University, and informed consent for using esopha-geal tissues was obtained from the patients or their families.

In the 50 patients who developed recurrence within thegross tumor volume (GTV) after definitive concurrent CRT,the initial treatment consisted of four cycles of cisplatin25 mg/m2 infusion on days 1–3, 5-FU 500 mg/m2/day as acontinuous infusion on days 1 to 5, and a total radiation doseof 59.4 Gy delivered at 1.8 Gy/d, 5 days per week.Posttreatment assessment included physical exam, neck-thorax-abdomen CT scan, upper endoscopy plus biopsy, andbarium swallow. A complete clinical response was defined asno tumor at follow-up endoscopy with biopsy and CT scan 4–6 weeks after completion of initial therapy. The evaluationwas conducted every 3 months for the first year, 4 months thesecond year, and 6 months thereafter. The local recurrence hasbeen defined as cancer relapse at the primary site for morethan 3 months after the initial therapy, and histological biopsywas performed to prove recurrent tumor.

Immunohistochemical staining and evaluation

The following primary antibodies were used for immunohis-tochemical staining: mouse monoclonal anti-FGF-2 (dilution1:200), rabbit polyclonal anti-CD31 (dilution 1:100), andmouse monoclonal anti-SMA (dilution 1:400); all the anti-bodies were purchased from Santa Cruz Biotechnology (SantaCruz, CA, USA). Serial tissue sections (5-μm-thick) cut fromformalin-fixed, paraffin-embedded ESCC were placed on sa-linized slides for immunohistochemical analysis. Afterdeparaffinization and hydration through a series of xylenesand alcohols, tissue sections were incubated with 3 % hydro-gen peroxide in methanol for 30 min to block endogenousperoxidase activity. Primary antibody incubation was doneovernight at 4 °C. Specimens were next washed with PBS.They were incubated for 30 min with secondary antibodywith Envision Plus Dual Link-labeled polymer (DAKO,Carpinteria, CA) and then for 5 min with diaminobenzidinechromogen. Formalin-fixed, paraffin-embedded normalesophageal squamous epithelium was used as the control.

Scores of immunostaining were recorded based on assess-ment of the intensity and percentage of the stained area. Theexpression was quantified using a four-value intensity score(0, 1+, 2+, and 3+) and the percentage (0–100 %) of the extentof reactivity. Next, the expression score was obtained bymultiplying the intensity and reactivity extension values(range 0–300).

Tumor Biol.

Collagen staining

Collagen staining was done using Masson’s trichrome stain-ing technique, which was conducted according to the guide-line of the agent kit from Sigma (HT15, St. Louis, MO, USA).Briefly, 5-μm-thick sections cut from formalin-fixed, paraffin-embedded tissues were fixed in Bouin’s solution. After incu-bation in Weigert’s iron hematoxylin solution, the slides werestained with Biebrich scarlet-acid fuchsin and aniline blue anddehydrated in ethanol and xylene. Extensive washes in PBSwere done between each staining. With the Masson’strichrome stain, the smooth muscle cell cytoplasm was stainedred, while the collagenous fibrous tissue was stained blue.

Western blotting analysis

Ten sections (5-μm-thick) from the formalin-fixed, paraffin-embedded tissues were used for protein extraction as de-scribed by Wolff et al. [12]. Equal amounts of protein weresubmitted to a 15 % sodium dodecyl sulfate-polyacrylamidegel and transferred to PVDF membranes. After blocked inTBS-Tween 20 (TBST) containing 5 % skim milk for 1 h atroom temperature, the membranes were incubated with pri-mary antibodies overnight at 4 °C. Following incubating withhorseradish peroxidase-conjugated IgG secondary antibody,the immunoreactive bands were visualized by enhancedchemiluminescence system (Pierce, USA). The antibodiesused were anti-FGF-2 (dilution 1:500), anti-CD31 (dilution1:400), anti-SMA (dilution 1:800), and anti-β-actin.

Statistical analysis

The data were summarized using standard descriptive statis-tics; the data were expressed as mean±SE. Intergroup com-parisons were assessed with the Fisher exact test. The recur-rence interval was calculated from the last day of definitivechemoradiation to the time of the first documented localrelapse. Survival curves were estimated using the Kaplan-Meier method, and comparisons were made using the log-rank test. All data were analyzed using the SPSS 16.0 software(SPSS, Inc., Chicago, IL, USA) and two-sided P values <0.05were considered statistically significant.

Results

Protein expression of FGF-2, CD31, and SMA in humanesophageal tissues

Figure 1 illustrated the relative expression of FGF-2, CD31,and SMA in the specimens from patients with ESCC, recurrentESCC, and normal esophageal squamous epithelium as well.Using immunohistochemistry, esophageal cancer tissues

showed positive expression of FGF-2, CD31, and SMA; how-ever, FGF-2 expression was not detected, and only little stain-ing for CD31 and SMAwas noted in normal epithelium. Theimmunohistochemical score of FGF-2, CD31, and SMA was38.5±9.3, 51.2±26.7, and 43.3±13.5, respectively, in the pa-tients with primary ESCC. While the expression score of thesethreemarkers in recurrent ESCCwere 2.0–4.8-fold higher thanthat in primary ESCC, and the immunohistochemical score ofFGF-2, CD31, and SMA was 185.3±68.5, 103.3±28.1,and 140.9±36.9, respectively. For their expression levelsin tumor tissue, statistically significant differences were foundbetween recurrent ESCC and primary ESCC (Fig. 2). Whenthe tumor samples were submitted to Western blot analysis,the expression difference of bFGF, CD31, and SMA betweenthe two cohorts of patients was also detected (Fig. 3).

Among the 50 patients with locally recurrent disease afterdefinitive CRT, 35 patients in whom the tumor recurred local-ly within 24 months after initial therapy showed a FGF-2,CD31, and SMA expression score of 236.2±49.7, 125±30.3,and 173.7±41.6, respectively, and in the patients who had alocal recurrence more than 24months after definitive CRT, theexpression score of FGF-2, CD31, and SMAwas 148.3±33.5,85.6±21.2, and 96.3±28.1, respectively. The significant dif-ferences were observed in FGF-2 and SMA levels between thetwo subgroups of patients with recurrent ESCC (Fig. 2).Moreover, these differences were further confirmed byWestern blot analysis (Fig. 3).

Collagen staining in human esophageal tissues

To address the concern that the cancer-associated stroma itselfmight be indicative of disease recurrence, we also evaluatedcollagen deposits in the stromal area. As shown in Fig. 1, therewas very little collagen deposit in normal squamous epitheli-um. The extent of collagen deposits was usually high in tissuespecimens from recurrent ESCC, and more extensive collagendeposits were exhibited in tissues from the patients who had alocal recurrence within 24 months after definitive CRT, whichfollowed a trend similar to FGF-2 and SMA expression inesophageal cancer tissues.

Relationship between FGF-2, CD31, and SMA expressionand clinicopathologic features in recurrent ESCC

Among the 50 ESCC patients who developed local recurrenceafter definitive CRT, 70 % of the tumors were located in themiddle thoracic region; 31 (62 %) were classified as tumor-node-metastasis system stage III.We analyzed the correlationsbetween FGF-2, CD31, and SMA expression and clinicalhistopathologic parameters. The cases were arbitrarilysubdivided into the high-expression group and the low-expression group, according to immunohistochemical scoresat a cutoff point at the mean values of FGF-2, CD31, and

Tumor Biol.

Fig. 1 Microphotographs ofimmunohistochemical expressionof FGF-2, CD31, SMA, andcollagen staining in tissuespecimens of ESCC, recurrentESCC, and normal esophagealepithelium. For each histologictype and marker in esophagealcancers, both high and low levelsof magnification were shown

0

50

100

150

200

250

300

0

50

100

150

200

250

300

FGF-2 CD31

PrimaryESCC

Recurrence> 24ms

RecurrentESCC

Recurrence 24ms

Recurrence> 24ms

Recurrence 24ms

PrimaryESCC

RecurrentESCC

P=0.017 P=0.063P=0.000 P=0.047

0

50

100

150

200

250

300

SMA

Recurrence> 24ms

Recurrence 24ms

PrimaryESCC

RecurrentESCC

P=0.033P=0.009

a b

c

Fig. 2 Scores for immunohistochemical expression of FGF-2, CD31, and SMA in tissues obtained from ESCC patients and patients with locallyrecurrent ESCC after definitive chemoradiation. P values comparing different tumor tissue types were shown

Tumor Biol.

SMA. As shown in Table 1, protein expression of FGF-2,CD31, and SMA did not show associations with gender, age,

tumor location, the depth of penetration (T stage), nodal status(N stage), and clinical stage (P>0.05).

Actin

CD31

FGF-2

SMA

1 2 3 4

0

0.5

1

1.5

2

2.5

3

3.5

Primary ESCC

Recurrence > 24ms

P=0.002

P=0.039

P=0.056

P=0.007

Recurrence 24ms

Normal esophageal tissue

FGF-2 CD31 SMA

Rel

ativ

e de

nsit

y (r

atio

to A

ctin

)

b

aFig. 3 Western blot analyses ofFGF-2, CD31, and SMA proteinin normal esophageal epitheliumand ESCC tissues. The ratios ofFGF-2, CD31, and SMA to actinlevels in each experiment weredetermined densitometrically, andP values comparing differenttumor tissue types were shown. 1normal esophageal tissue, 2primary ESCC tumor tissue, 3tumor tissue from ESCC patientwho developed a local recurrencewithin 24 months after definitivechemoradiation, and 4 tumortissue from ESCC patient whohad a local recurrence more than24 months

Table 1 Clinicopathological fea-tures and expression of FGF-2,CD31, and SMA in the 50patients

Variables No. FGF-2 CD31 SMA

High Low P High Low P High Low P

Gender

Male 43 22 21 0.769 27 16 0.318 19 24 0.524

Female 7 4 3 3 4 4 3

Age (years)

<60 26 16 10 0.160 15 11 0.729 14 12 0.247

≥60 24 10 14 15 9 9 15

Tumor location

Upper thoracic 9 6 3 0.397 7 2 0.412 5 4 0.775

Middle thoracic 35 16 19 19 16 15 20

Lower thoracic 6 4 2 4 2 3 3

T stage

T1–2 16 10 6 0.308 9 7 0.710 9 7 0.318

T3-4 34 16 18 21 13 14 20

N stage

N0 24 12 12 0.786 16 8 0.355 12 12 0.586

N1 26 14 12 14 12 11 15

AJCC clinical stage

II 19 10 8 0.790 12 7 0.721 11 8 0.186

III 31 16 15 18 13 12 19

Tumor Biol.

Analysis of prognostic factors for local recurrence-freesurvival

At the time of last follow-up, eight (16 %) patients remainedalive, with a median follow-up of 35.6 months. The 3-, 5-, and10-year overall survival rate was 45.3, 26.3, and 12.0 %,respectively, for the 50 patients with locally recurrent ESCCafter definitive CRT; the median overall survival was33.0±3.8 months. The 2- and 5-year local recurrence-freesurvival rate was 31.6 and 16.0 %, with a median localrecurrence-free survival of 13.7 months.

The most common potential clinical factors affecting localtumor recurrence, including gender, age, tumor location, Tstage, N stage, and clinical stage, were summarized in Table 2.As expected, a significant correlation between tumor statusand outcome was seen. ESCC patients with T3–4 disease hada shorter time to local recurrence after definitive CRT, com-pared to those with T1–2 tumor (P=0.038).

Associations between levels of protein expression of FGF-2, CD31, and SMA in esophageal cancer and disease out-comes were then analyzed. Higher expression levels of FGF-2and SMA were related to a shorter local recurrence-free sur-vival. The 2- and 5-year local recurrence-free survival rate was15.4 % and 0 in the patients with higher FGF-2 expression,compared with 45.8 and 33.3 % in those who expressed lowerFGF-2, respectively (P=0.005). Of patients who expressed

higher SMA, the 2- and 5-year local recurrence-free survivalrate was 21.7 and 8.7 %, respectively, compared to thosewith lower SMA expression which was 37.0 and 22.2 %,respectively, and P=0.016 (Fig. 4).

Discussion

Local recurrence is the most common pattern of treat-ment failure after CRT in esophageal cancer, and ap-proximately 70 % of these local recurrences occur with-in 24 months after definitive CRT [13]. Tumors are acomplex tissue composed of carcinoma cells, varioustypes of stromal cells, and dense extracellular matrix;solid tumors may modulate their environment and keepstromal cells in a tumor-promoting and immunosuppres-sive state. Recent studies have indicated the relevanceof stromal cells to failure of systemic anticancer therapyand the development of therapeutic resistance [14]. Weshowed here that the stroma of ESCC contained FGF-2-positive cells, CD31-positive cells, and SMA-positivemyofibroblasts. Notably, protein expression levels ofFGF-2, CD31, and SMA were much higher in speci-mens from recurrent ESCC patients.

Ectopic expression of FGF-2 has been shown to transformnormal cells into a malignant phenotype, and inhibition ofFGF-2 expression can reverse the transformed phenotype[15]. FGF-2 has been shown recently to confer a broad spec-trum resistance to various cytotoxic agents in several cancertypes, through providing cancer cells with prosurvival andmitogenic signals and protection against apoptosis inducedby cytotoxic drugs [16, 17]. Barclay et al. studied the correla-tion between FGF-2 expression and clinicopathologic featuresand outcome for resected esophageal cancer [4]. FGF-2mRNA was not detected in normal esophageal epitheliumbut was overexpressed in 83 % of tumors, and patients withtumor overexpressing FGF-2 had a significantly increased riskfor tumor recurrence after radical esophagectomy. Themediandisease-free survival exceeded 36 months in patients withtumors underexpressing FGF-2, while the patients with tumoroverexpressing FGF-2 had a median disease-free survival ofonly 12 months. But they investigated the prognostic role ofFGF-2 gene expression in 48 patients with esophageal cancerwho underwent esophagectomy, and tumor histology wassquamous cell carcinoma in only 10 (21 %) patients.Twenty patients with primary ESCC and 50 with locallyrecurrent ESCC after definitive CRT formed the basis ofour study. The findings showed that FGF-2 proteinexpression was significantly increased in recurrent tu-mors. Interestingly, the FGF-2 protein level was higherin the patients who developed a local recurrence within24 months after definitive CRT than that in whom therewas no recurrence during the 24-month follow-up

Table 2 Clinicopathological features and analysis of time to localrecurrence

Variables No. of patients(%)

Median time to localrecurrence (months)

P value

Gender

Male 43 (86.0) 14.2±1.5 0.493

Female 7 (14.0) 9.7±3.0

Age (years)

<60 26 (52.0) 16.5±2.7 0.229

≥60 24 (48.0) 9.8±1.5

Tumor location

Upper thoracic 9 (18.0) 13.7±7.0 0.765

Middle thoracic 35 (70.0) 13.5±1.7

Lower thoracic 6 (12.0) 17.0±7.5

T stage

T1–2 16 (32.0) 21.2±6.0 0.038

T3–4 34 (68.0) 10.1±2.0

N stage

N0 24 (48.0) 14.6±4.1 0.895

N1 26 (52.0) 13.2±2.3

AJCC clinical stage

II 19 (38.0) 15.0±2.5 0.924

III 31 (62.0) 12.3±2.1

Tumor Biol.

period. Radiation mainly kills the tumor cells either byapoptosis or mitotic cell death [18]. These results sug-gest that FGF-2 overexpression may also play an im-portant role in the acquisition of radiation resistance inesophageal cancer, though the mechanisms need furtherinvestigation.

Tumor regrowth after local irradiation is also dependent onblood vessel formation. Functional vessels and viableendothelial cells are undesirable after tumor irradiationbecause they might support the survival and growth ofremaining cancer cell clonogens and thus promote therelapse. However, it has been demonstrated that a fraction ofendothelial cells in growing tumors may be derived viatransdifferentiation from cancer stem-like cells, and stem-like cells are more resistant to radiation than other neoplasticcells. Endothelial cells can also be recruited from preexistingvessels from nonirradiated adjacent tissue. Proliferation andsprouting of endothelial cells that have survived irradiationform new blood vessels; this can result in nutrient supply fortumor regrowth [19]. The extent of neovascularization wasassessed by immunostaining for CD31 in our study. CD31-

positive cells were detected in ESCC, and higher numbers ofCD31-positive cells could be seen in recurrent tumors.

It has been reported that myofibroblasts are present duringthe wound-healing process and also during tumor progressionand disappear with tissue reconstruction or tumor regression;myofibroblasts are differentiated host fibroblasts that expressSMA as cytoplasmic microfilaments. Evidences demonstratethat myofibroblasts modulate various aspect of tumor progres-sion. Myofibroblasts express various cytokines and growthfactors, suggesting a role in tumor angiogenesis. A coculturesystem also revealed that tumor-derived fibroblasts mightinhibit tumor cell death [20]. We found that SMA-positivemyofibroblasts were present in esophageal cancer tissues, andSMA expression was significantly increased in tissues obtain-ed from recurrent ESCC. Tsujino et al. assessed the value ofmyofibroblasts in the cancer stroma as an indicator of diseaserecurrence after colorectal cancer surgery. They found that thepatients with stage II and III cancers with high SMA expres-sion had a shorter disease-free survival rate [21]. These find-ings suggest that tumor cells exert an influence upon the hosttissue preparing the microenvironment for the cancer

FGF-2 low (n = 24)FGF-2 high (n = 26)

P = 0.005

CD31 low (n = 20)CD31 high (n = 30)

P = 0.666

SMA low (n = 27)SMA high (n = 23)

P = 0.016

ba

c

Fig. 4 Kaplan-Meier curves illustrating protein expression of FGF-2, CD31, and SMA in recurrent ESCC patients

Tumor Biol.

progression, and stromal fibroblasts may play a vital role inthis process.

Among the patients with locally recurrent disease afterdefinitive CRT, expression of FGF-2 and SMA and collagendeposits were notably high in whom the tumor recurred local-ly within 24 months after initial therapy, suggesting the abun-dance of myofibroblasts as an indicator of early tumor recur-rence. As far as correlations with clinical variables are con-cerned, we assessed in patients with recurrent ESCC whetherlow versus high expression levels of FGF-2, CD31, and SMAwere predictive of local recurrence-free survival time and ifthis survival time was actually far shorter in those with higherFGF-2 and SMA protein levels.

In summary, despite the retrospective nature of the study,our findings indicate that FGF-2 and SMA are frequentlyoverexpressed in recurrent ESCC, which play important rolesin the biology and progression of ESCC; they are attractivetargets for esophageal cancer therapeutic strategies. Our dataalso provide further evidence for the therapeutic concept of theinhibition of neovascularization in esophageal cancer.

Acknowledgments This work was supported by a grant-in-aid from theNational Natural Science Foundation of China (No: U1204816) andHenan Provincial Science and Technology Bureau.

Conflicts of interest None

References

1. Bremnes RM, DonnemT, Al-Saad S, Al-Shibli K, Andersen S, SireraR, et al. The role of tumor stroma in cancer progression and progno-sis: emphasis on carcinoma-associated fibroblasts and non-small celllung cancer. J Thorac Oncol. 2011;6:209–17.

2. Noma K, Smalley KS, Lioni M, Naomoto Y, Tanaka N, El-Deiry W,et al. The essential role of fibroblasts in esophageal squamous cellcarcinoma-induced angiogenesis. Gastroenterology. 2008;134:1981–93.

3. Xu Z, Wang S, Wu M, Zeng W, Wang X, Dong Z. TGFβ1 and HGFprotein secretion by esophageal squamous epithelial cells and stromalfibroblasts in oesophageal carcinogenesis. Oncol Lett. 2013;6:401–6.

4. Barclay C, Li AW, Geldenhuys L, Baguma-NibashekaM, Porter GA,Veugelers PJ, et al. Basic fibroblast growth factor (FGF-2) overex-pression is a risk factor for esophageal cancer recurrence and reducedsurvival, which is ameliorated by coexpression of the FGF-2 anti-sense gene. Clin Cancer Res. 2005;11:7683–91.

5. Erez N, Truitt M, Olson P, Hanahan D. Cancer-associated fibroblastsare activated in incipient neoplasia to orchestrate tumor-promotinginflammation in an NF-kappa B-dependent manner. Cancer Cell.2010;17:135–7.

6. Kalluri R, Zeisberg M. Fibroblasts in cancer. Nat Rev Cancer.2006;6:392–401.

7. Hoff PM, Machado KK. Role of angiogenesis in the pathogenesis ofcancer. Cancer Treat Rev. 2012;38:825–33.

8. GLOBOCAN 2012: Estimated cancer incidence, mortality and prev-alence worldwide in 2012. GLOBOCAN. IARC. 2013.

9. Chen W, He Y, Zheng R, Zhang S, Zeng H, Zou X, et al. Esophagealcancer incidence and mortality in China, 2009. J Thorac Dis. 2013;5:19–26.

10. Shridhar R, Almhanna K, Meredith KL, Biagioli MC, Chuong MD,Cruz A, et al. Radiation therapy and esophageal cancer. CancerControl. 2013;20:97–110.

11. Welsh J, Palmer MB, Ajani JA, Liao Z, Swisher SG, Hofstetter WL,et al. Failure patterns in patients with esophageal cancer treated withdefinitive chemoradiation. Cancer. 2012;118:2632–40.

12. Wolff C, Schott C, Porschewski P, Reischauer B, Becker KF.Successful protein extraction from over-fixed and long-term storedformalin-fixed tissues. PLoS ONE. 2011;6:e16353.

13. Chen Y, Lu Y,Wang Y, Yang H, Xia Y, ChenM, et al. Comparison ofsalvage chemoradiation versus salvage surgery for recurrent esopha-geal squamous cell carcinoma after definitive radiochemotherapy orradiotherapy alone. Dis Esophagus. 2014;27:134–40.

14. Meads MB, Gatenby RA, Dalton WS. Environment-mediated drugresistance: a major contributor to minimal residual disease. Nat RevCancer. 2009;9:665–74.

15. Murphy PR, Sato Y, Knee R. Phosphorothioate antisense oligonucle-otides against basic fibroblast growth factor inhibit anchorage-dependent and anchorage independent growth of a malignant glio-blastoma cell line. Mol Endocrinol. 1992;6:877–84.

16. Carmo CR, Lyons-Lewis J, Seckl MJ, Costa-Pereira AP. A novelrequirement for Janus kinases asmediators of drug resistance inducedby fibroblast growth factor-2 in human cancer cells. PLoS ONE.2011;6:e19861.

17. Miyake H, Hara I, Gohji K, Yoshimura K, Arakawa S, Kamidono S.Expression of basic fibroblast growth factor is associated with resis-tance to cisplatin in a human bladder cancer cell line. Cancer Lett.1998;30:121–6.

18. Baskar R, Lee KA, Yeo R, Yeoh KW. Cancer and radiationtherapy: current advances and future directions. Int J Med Sci.2012;9:193–9.

19. Kozin SV, Duda DG, Munn LL, Jain RK. Neovascularization afterirradiation: what is the source of newly formed vessels in recurringtumors? J Natl Cancer Inst. 2012;104:899–905.

20. Li H, Fan X, Houghton J. Tumor microenvironment: the role of thetumor stroma in cancer. J Cell Biochem. 2007;101:805–15.

21. Tsujino T, Seshimo I, Yamamoto H, Ngan CY, Ezumi K, Take- masaI, et al. Stromal myofibroblasts predict disease recurrence for colo-rectal cancer. Clin Cancer Res. 2007;13:2082–90.

Tumor Biol.