ORIGINAL PAPER
Elevated DLL4expression is correlated with VEGF and predicts poor prognosis of nasopharyngeal carcinoma
Jia-Xing Zhang •Man-Bo Cai •Xiao-Pai Wang •Li-Ping Duan •Qiong Shao •Zhu-Ting Tong •Ding-Zhun Liao •Yang-Yang Li •Ma-Yan Huang •Yi-Xin Zeng •Jian-Yong Shao
Received:24September 2012/Accepted:26September 2012/Published online:30December 2012ÓSpringer Science+Business Media New York 2012
reactive materials studies
Abstract Delta-like ligand 4(DLL4),one of the transmembranous Notch ligands,is upregulated at the site of tumor growth,particularly during tumor angiogenesis.Expression pattern of DLL4in nasopharyngeal carcinoma (NPC)and the clinical and prognostic significance remain unclear.In this study,immunohistochemical analysis (IHC)was used to examine the protein level of DLL4in NPC tissues from two independent cohorts.In the testing cohort (311cases),we applied the X-tile program software able to assess the optimal cutoff points for biomarkers in order to accurately classify patients according to clinical outcome.
In the validation cohort (113cases),the cutoff score derived from X-title analysis was investigated to determine the association of DLL4expression with disease-specific survival (DFS).Our results showed that high expression of DLL4was observed in 134of 313(42.8%)in the testing cohort and 58of 113(43.6%)in the validation cohort.High expression of DLL4independently predicted poorer disease-specific survival,as evidenced by univariate and multivariate analysis (P \0.05).Moreover,DLL4expres-sion was significantly elevated in distant NPC metastases relative to primary NPC tumors (P =0.001).Importantly,we found a significant positive relationship between DLL4and vascular endothelial growth factor (VEGF)(P \0.001).Patients with dual elevated DLL4and VEGF expression displayed a significant overall survival disadvantage com-pared to those with dual low expression (P \0.05).These findings provide evidence that high expression of DLL4serves as an independent predictor of poor prognosis in NPC patients.
Keywords DLL4ÁNasopharyngeal carcinoma ÁImmunohistochemistry ÁPrognosis
Background
Nasopharyngeal carcinoma (NPC),which exhibits a high prevalence in southern China and Southeastern Asia,dis-plays the highest rate of metastasis compared with other head and neck cancer
s [1–4].The majority of NPC patients exhibit regional lymph node (LN)or distant organ metas-tases at the time of diagnosis [5–7].However,the molec-ular mechanisms underlying NPC development and progression remain poorly understood.Thus,a substantial amount of research on NPC has focused on the discovery
Jia-Xing Zhang,Man-Bo Cai and Xiao-Pai Wang:These authors made equal contributions and are joint first authors.
J.-X.Zhang ÁY.-X.Zeng ÁJ.-Y.Shao (&)
State Key Laboratory of Oncology in South China,Sun Yat-sen University Cancer Center,Guangzhou 510060,China e-mail:
J.-X.Zhang ÁX.-P.Wang ÁQ.Shao ÁY.-Y.Li ÁJ.-Y.Shao Department of Molecular Diagnostics,Sun Yat-sen University Cancer Center,Guangzhou 510060,China
M.-B.Cai ÁY.-X.Zeng
Department of Experiment Research,Sun Yat-sen University Cancer Center,Guangzhou 510060,China
L.-P.Duan
Institute of Scientific and Technical Information of China (ISTIC),100038Beijing,China
Z.-T.Tong
Department of Radiotherapy,The First Affiliated Hospital,An Hui Medical University,Hefei,China
D.-Z.Liao ÁM.-Y.Huang
Department of Pathology,Sun Yat-sen University Cancer Center,Guangzhou 510060,China
Med Oncol (2013)30:390
DOI 10.1007/s12032-012-0390-x
of promising molecular and genetic alterations responsible for the invasive potential and clinical progression of this malignancy[8–11].However,current investigation and identification of novel molecular biomarkers in NPC that provide reliable prognostic value is limited.Thus,in this study,we aimed to identify an additional novel molecular marker able to not only predict the clinical prognosis but also provide optimal therapeutic targeting to benefit NPC patients.
Angiogenesis is an essential step for tumor growth and progression and plays a vital role in tumor invasion and metastasis[12].One signaling pathway that may contribute to this process is the Notch signaling pathway[13,14].The mammalian Notch family consists of four receptors,Notch 1–4,andfive ligands,Jagged1and2,and Delta-like ligand 1,3,and4(DLL1,-3,-4).DLL4,a ligand for Notch receptors1,3,and4,is upregulated by vascular endothelial growth factor(VEGF)and hypoxia,and its activation triggers a negative feedback that restrains the excessive VEGF-effects of vascular sprouting and angiogenesis[15–19].Overall,DLL4functions as a negative regulator of tumor angiogenesis by reducing the number of blood ves-sels;yet surprisingly,it enhances the vasculature structure and function within a tumor by inducing larger vessels, thereby increasing vessel perfusion and tumor oxygenation, resulting in increased tumor growth and progression in vivo [20–22].Recently,DLL4has been shown to mediate tumor resistance to bevacizumab in vivo[23].Activation of the Notch/DLL4signaling pathway has previously been described in several human malignancies and has been linked to poor patient prognosis[22,24–30].Recently reported immunohistochemistry(IHC)studies have dem-onstrated that DLL4protein expression is not restricted to endothelial cells,but is widespread in neoplastic as well as in reactive tissues[22,24,29].Moreover,in breast and colon cancers,DLL4was also reported to possess a posi-tively associated expression level with VEGF and hypoxia markers[28,31].However,the characterization and clin-ical significance of DLL4protein expression in NPC has not been fully elucidated.
We have previously reported that elevated expression of VEGF protein is a strong independent predictor of survival in advanced NPC[32].However,by now,no study was conducted to explore the expression pattern of DLL4in NPC.In this study,wefirst constructed two independent NPC tissue sample cohorts from two institutions.Based on our previous work,we intend to analyze the expression level of NPC as well as its relationship regarding NPC patients’prognosis.In the testing cohort(311cases),we applied the X-tile program software able to assess the optimal cutoff points for biomarkers in order to accurately classify patients according to clinical outcome.In the validation cohort(113cases),the cutoff score derived from X-title analysis was investigated to determine the associa-tion of DLL4expression with disease-specific survival (DFS).As far as we concerned,our study is thefirst research to elucidate the expression pattern of DLL4in NPC.Our work demonstrated that DLL4-associated VEGF high expression could potentially serve as a molecular target for NPC therapy.
Materials and methods
Patients and tissue microarray
For this retrospective study,formalin-fixed,paraffin-embedded specimens of NPC were collected between1997 and2000from Sun Yan-Sun University Cancer Center (SYSUCC)(Guangzhou,China).Of
the311patients,were 235male and76female,with a median age of47years (range,16–78years).The disease stages for all patients were classified according to the1992NPC staging system of China[33]:2patients were stage I,54patients were stage II,164patients were stage III,and91patients were stage IV.Statistical analysis showed that the agreement rate was72%.Besides,we collected31paired archival tissue samples of liver/pulmonary metastases from NPC cases between January2009and December2009in SYSUCC.All NPC were obtained prior to treatment with standard curative radiotherapy with or without chemo-therapy.The average duration of follow-up in this cohort was63months(median,68months;range,1–120months).
In parallel,we assessed another randomly selected, independent validation cohort of133NPC patients between March2003and February2004from the Department of Pathology of Hunan Provincial Cancer Hospital(Changsha,China).This cohort included97males and36females,with a median age of47years(rang from 15to82years):20patients were stage II,68were stage III, and45were stage IV.The average duration of follow-up in this cohort was41months(median,41months;range 6–79months).
All of the clinical materials were obtained from patients who provided informed consent in accordance with insti-tutional review board standards and guidelines.Paraffin-embedded specimens from the previously constructed tissue microarray(TMA)and procedures for the TMA construction were reported
previously[32]. Immunohistochemistry(IHC)
IHC analysis was performed to examine DLL4and VEGF expression in NPC tumor tissue.Primary antibodies against DLL4(1:300dilution,Abcam,USA)and VEGF(1:200
Page2of10Med Oncol(2013)30:390
dilution,Santa Cruz,USA)were used in this study.The staining protocol utilized in this study was described pre-viously[32].The IHC results were evaluated and scored independently by two pathologists with no prior knowledge of the clinicopathological outcomes of the patients.Semi-quantitative estimates were achieved using a composite score comprised of the sum of the staining intensity values and the relative abundance of positive cells.The intensities were graded as0(negative),1(weakly positive),2(mod-erately positive),and3(strongly positive).The abundance of positive cells was graded from0to4(0\5%positive cells;1,5–25%;2,26–50%;3,51–75%;4,76–100%).
Immunoblot analysis
Cells were harvested and lysed with RIPA buffer(Upstate, USA).Equal amounts of denatured proteins were separated by SDS-PAGE and then transferred electrophoretically to PVDF membranes(Amersha
m,UK)for immunoblot analysis.The antibody used for immunoblot analysis against DLL4was(1:500dilution,Abcam,USA)an anti-GAPDH antibody(1:3000dilution,Santa Cruz,USA)and was used as the loading control.All protein bands were detected using an enhanced chemiluminescent(ECL) Western blot kit(Amersham,UK).
Selection of cutoff point score
We applied X-tile plots to assess DLL4expression levels and to optimize the cutoff points based on patient survival outcomes.X-title plots provide a single,global assessment of dividing a population into low-and high-level marker expression in every possible way.X-title data were pre-sented in a right triangular grid where each point represents a different cut-point.The intensity of the color of each cutoff point represents the strength of the association.The X-title software allows the user to move a cursor across the grid and provide an‘‘on-the-fly’’histogram of the resulting population subsets along with an associated Kaplan–Meier curve.The X-title software provides a method of dividing a single cohort into training and validation subsets for P value estimation.In addition,the software can perform standard Monte Carlo ,cross-validation)to produce corrected P values to assess statistical significance of data assessed by multiple cut-points.The X-tile program divided the testing cohorts(311cases)randomly into a matched training and validation set as a method for selecting optima
l cut-points,respectively[34].Statistical significance was assessed using the cutoff score derived from the testing cohort and a separate validation set using a standard log-rank method,with P values obtained from a statistical table.The X-tile plots allowed for the determi-nation of an optimal cutoff value while correcting for the use of minimum P statistics by Miller–Siegmund P value correction[35].
Clinical outcome assessment
The two cohorts of patients in this RCT were all followed up in accordance with a strict protocol.After completion of therapy,patients were observed at3-month intervals during thefirst3years and at6-month intervals thereafter.For deceased patients,we classified the underlying cause of death according to data available on the death certificate. Local relapse was defined as recurrence of NPC confirmed by biopsy,and distant metastasis was defined as evidence of distant metastasis by chest X-ray/computed tomography, abdominal ultrasound,and bone scan.
Statistical analysis
For survival analysis,the optimal cutoff point for DLL4 expression was obtained using X-tile software version3.6.1 (Yale University School of Medicine,New Haven,CT, USA).The statistical significance of the correlation between DLL4expression level and patient survival was estimated using the Mantel–Co
x log-rank test.Monte Carlo simulations were used to adjust for multiple observations in optimal cutoff point selection.Receiver operating charac-teristic(ROC)curve analysis was conducted to evaluate the predictive value of the parameters.The chi-square test or Fisher’s exact test was employed to evaluate the relation-ship between DLL4expression and clinicopathological variables.The multivariate Cox proportional hazards model was utilized to estimate the hazard ratios and95% confidence intervals for patient outcome.The relationships between the DLL4expression levels and DFS were determined by Kaplan–Meier analysis.Log-rank tests were performed to determine the difference in survival proba-bilities.All P values quoted were two-sided,and P\0.05 was considered a statistically significant.Statistical analy-sis was performed using SPSS16.0software;SPSS,Chi-cago,IL,USA).
Results
DLL4expression patterns in nasopharyngeal cell lines and tissues and X-title analysis
Western blot analysis demonstrated that in primary NPCs, 6/8(75%)of cases exhibited upregulated DLL4expres-sion compared to adjacent non-neoplastic nasopharyngeal tissues(Fig.1a).For DLL4IHC staining,positive staining was observed primarily in the cell membrane and cyto-plasm(Fig.1b,c,d).DLL4exhibited strong IHC staining
Med Oncol(2013)30:390Page3of10
in NPC nests,whereas the paired normal epithelium exhibited weak IHC staining.Positive DLL4staining in tumor vessels of NPC and human colon cancer (positive control)was also observed (Fig.1e,f),while no positive signal was observed in vessels adjacent to normal colon mucosa (negative control)(Fig.1g).
To assess statistical significance and,therefore,avoid arbitrary cutoff point selection,the X-tile program was employed to determine cutoff scores for DLL4expression.According to the X-tile plots,we divided the testing cohort into low and high populations based on the cutoff point of over four.The optimal cutoff point determined by the testing cohort was applied to the validation cohort,which in turn identified the highly statistically significant cutoff scores (Fig.2).In the testing cohort,high expression of DLL4was observed in 134of 311(43.1%)of NPC cases.The 5-year DSS rates differed substantially and statistically significantly between low-and high-level DLL4expression of NPC patients in the testing cohort (68.0.vs.50.2%,Fig.2a;P =0.009),whereas in the validation cohort,high DLL4expression was observed in 58of 133(43.6%)NPC.The 5-year DSS rates differed substantially and
statistically significantly between low-and high-level DLL4expression of NPC patients in overall cases (65.1vs.34.8%,Fig.2b;P =0.004).DLL4expression and clinical features
The correlation between expressions of DLL4in NPC with respect to clinicopathologic features is shown in Table 1.In the testing cohort,we detected a positive association between DLL4expression level and local relapse and distant metastasis.In the validation cohort,we observed a positive relationship between DLL4expression and distant metastasis only.However,we failed to detect a relationship between DLL4and other patient characteristics,including age,gender,tumor stage,node stage,and TNM stage.To evaluate the clinical relevance of our hypothesis that DLL4regulates NPC metastasis,we collected paired archival tissue samples of liver/pulmonary metastases from NPC cases for comparison with primary NPC tissues (n =31).IHC staining demonstrated that the protein levels of DLL4were significantly higher in the distant metastatic lesions than the matched primary lesions (Fig.3a,
b).
Fig.1DLL4expression in human NPC and normal
adjacent epithelium.a Western blot analysis of DLL4
expression in representative NPC (T)and non-neoplastic nasopharyngeal mucosal tissue (N).Equal loading of protein was determined by GAPDH.b DLL4showed cytoplasmic overexpression in NPC tumor tissue than normal
nasopharyngeal epithelium;c high-level expression of DLL4in a NPC tumor;d low-level expression of DLL4in NPC tumor;e high-level expression of DLL4was observed in the vessel of NPC.f Strong DLL4staining in the vessel of colon cancer.g No DLL4staining in the vessels of normal colon mucosa.The vessels in each image are indicated by the arrow
Page 4of 10Med Oncol (2013)30:390
DLL4expression and survival analysis
To evaluate the prognostic value between DLL4expression levels and clinicopathologic features,ROC
curves were plotted to test patient survival status.ROC curve analysis confirmed the predictive value of DLL4regarding NPC-specific survival in the validation cohort [area under the curve (AUC)=0.612,P =0.026,Fig.4a].When these findings were further investigated in all cases,DLL4was found to be a promising predictor for NPC patient survival status (AUC =0.591,P =0.001,Fig.4b).Univariate analysis and multivariate Cox regression analysis
Expression of DLL4was found to be a significant prog-nostic indicator of poor DSS [hazard ratio (HR),2.076;95%confidence interval (CI),1.248–3.452,P =0.005;
Table 2]in the validation cohort,as well as in overall case (HR,1.886;95%CI,1.444–2.464;P \0.001;Table 2).Regarding other parameters,node stage and overall stage expression levels were determined to be positive significant prognostic factors for patient survival in both cohorts.After multivariate analysis,the expression level of DLL4was found to be a significant independent prognostic factor for poor DSS (HR,1.821;95%CI,1.081–3.068,P =0.024;Table 2)in the validation cohort,as well as in overall case (HR,1.809;95%CI,1.380–2.370;P \0.001;Table 2).
Correlation between the expression of DLL4and VEGF in NPC tissues
Our previous study revealed that VEGF protein levels are strong independent predictors of survival in
advanced NPC cases [32].In this study,we further evaluated the
potential
Fig.2X-tile plots of the prognostic marker of DLL4in the NPC cohorts.X-tile analysis was carried out on patient data from the training cohort,equally subdivided into training and validation subsets.X-tile plots of training sets are displayed in the left panels ,with matched validation sets in the smaller inset .The plot showed the chi-square log-rank values created when the cohort was divided into two populations.The cut-point highlighted by the red /blue circle in the left panels was demonstrated on a histogram of the entire cohort (middle panels )and a Kaplan–Meier plot (right panels ).P values were defined by using the cut-point derived from a training subset to parse a separate validation subset.a DLL4expression was divided at the optimal cut-point,as defined by the most significant on the plot (with positive staining of DLL4;P =0.009).b The optimal cut-point for DLL4expression determined by X-tile plot of the testing cohort was applied to the validation cohort and reached high statistical signifi-cance (P =0.004)
Med Oncol (2013)30:390Page 5of 10

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。