The Study On Key Molecules Of Gastric Cancer And Tumor Microenvironment

Based on the GLOBOCAN2008, about12.7million cancer cases and7.6million cancer deaths are estimated to have occurred in2008. In China, the gastriccancer (GC) incidence ranked the second in cancer, and third of the cancer mortality.Although, the early cancer screening techniques and the surgery guidedcomprehensive treatment strategies of GC developed rapidly recently, thereoccurrence and invasion after curative resection of GC still high. The root causesare cancer invasion and metastasis. Thus, the future research will focus on cancerinvasion and metastasis and comprehensive treatment strategy. Recently, more andmore researches demonstrated that the cancer invasion and metastasis is amulti-stage and multi-factor involved dynamic process. It not only includes thecancer cells themselves, but also the tumor microenvironment.Due to the personal judgment errors of traditional pathological morphology, it isurgent to develop the quantitative analysis technique based on the molecularpathology. In addition, tumor microenvironment is a complex community, involvingmany components with different functions. The existed techniques are farsatisfactory to realize the in situ multiplexed imaging of tumor microenvironment.Thus, the in situ multiplexed imaging technique is urgently needed.Quantum dots (QDs) are engineered nanoparticles with unique surface and sizeeffect so that it may have many advantages over fluorescent dyes, includingcontinuous emission spectrum, size-tunable, enhanced signal brightness andresistance to photobleaching. Recently, combined with the QDs based imagingtechnique, the molecular imaging, cellular imaging and in vivo imaging has progressed quickly. Due to the advantages of QDs in molecular imaging, the novelfluorescent nanoparticles have been widely used in the molecular target imaging.This study focused on the clinical dilemma of GC, and explored the tumorbiological behavior from the molecular pathology of cancer cells and tumormicroenvironment. It aimed to reveal the co-evolution theory of cancer cells andtumor microenvironment combined with the application advantages of QDs basedmolecular probe, so as to in depth understand the mechanisms of GC invasion andmetastasis.Two parts are included in this study:Part I: High Expression of Transform Growth Factor Beta1inGastric Cancer Confers Worse Outcome: Results of a Cohort Studyon184PatientsObjective: As a multifunctional cytokine, transform growth factors beta (TGFβ)regulates many biological processes by protein kinase receptors and drosophilamothers against decapentaplegic protein (SMAD) mediators, and plays importantroles in tumor growth, invasion and metastasis. Briefly, in healthy system, TGFβ actsas a tumor suppressor to inhibit cell proliferation, induce apoptosis and regulateautophagy. With tumor development, TGFβ will promote epithelial-mesenchymaltransition (EMT), tumor cell motility, invasion and metastasis.GC is one of the most common carcinomas and the third cause of cancer deathin the world. GC is divided into two types based on Lauren classification: diffusetype and intestinal type. Different with intestinal type GC, diffuse type GC ischaracterized by a diffusely infiltrating growth accompanied by stromal fibrosis anda poor prognosis. TGFβ1just is a powerful fibrotic stimulating factor, and contribute to create a favorable environment for the dissemination of GC. Compared withnormal gastric tissues, TGFβ1is significantly up-regulated in GC. Furthermore,TGFβ1over-expression of in GC tissues is correlated with GC prognosis, especiallyin diffused type GC. These studies may imply TGFβ1has different roles in differenttype GC. What is more, because of the complex secretion mechanism andbidirectional function of TGFβ1, previous studies may not accurately describe therelationship between TGFβ1expression and patients prognosis. Therefore, it isdesirable to explore the relationship between TGFβ1expression and prognosis ofpatients with GC in different clinicopathological subgroups. This study was mainlyto investigate TGFβ1expression and its clinical significance in differentclinicopathological subgroups of GC patients by immunohistochemistry method.Methods:1. Patients: A total of184GC patients receiving surgery-basedmulti-disciplinary treatment from December2002to February2011were collected.The patients’ paraffin specimens were constructed into three tissues microarrays(TMAs) which contained184tumor tissues and41peritumoral tissues (450cores,two cores for each tissues), developed by Shanghai Biochip Company Ltd.(Shanghai, China). In order to control the quality of TMAs, only tumor tissues at thecancer invasion front were selected, so that the typical biological behaviors of cancercould be revealed. TNM stage and pathologic classification were determinedaccording to the7th edition AJCC TNM system and Lauren classification,respectively. Up to May31,2012, the median follow-up of those GC patients was57.3(range:16.6-100.1) mo, with108(58.7%) GC-specific deaths occurred.2.Immunohistochemistry of TGFβ1: SP immunohistochemistry method was used tostain the TMAs. Briefly, after de-waxing, the TMAs were performed microwaveantigen retrieval for20min at moderate baking temperature in0.01mmol/L (pH=6.0)citrate buffer solution. After cooling at room temperature, TMAs were treated with 0.03%hydrogen peroxide methanol for10min to inactivate endogenous peroxidase.Then2%BSA was used to block TMAs to decrease background intensity. Everychip treated overnight at4℃with250μl rabbit anti-human TGFβ1monoclonalantibodyand then incubated with the corresponding secondary antibody (1:250dilution) for30min at37℃. Then treated with0.2%diaminobenzidine solution for2min to coloration. The TMAs were counterstained with hematoxylin anddifferentiated by hydrochloric acid alcohol.3. Result evaluation: The results ofimmunohistochemistry were analyzed by two experienced pathologists with over10years of experience in clinical tumor pathology. The scoring system combined cellsstaining intensity with cells positive rate as reported previously, and briefed in thefollowing: Cells staining intensity was defined as no stain, slight stain, medium stainand strong stain, and correspondingly scored as0,1,2and3, respectively. Cellspositive rate was defined as grade1with positive TGFβ1immunostaining in <10%of tumor cells stained positive, grade2with10%to25%of tumor cells stainedpositive, grade3with25%to50%of tumor cells stained positive, and grade4with>50%of tumor cells stained positive, and correspondingly scored as0,1,2and3,respectively. The final score was the sum of cells staining intensity score and cellspositive rate score. The sum score0-3was defined as low expression and4-6as highexpression.4. Statistical analysis: Statistical analysis was performed with SPSS17.0software. Cumulative survival was calculated by the Kaplan-Meier method andanalyzed by the Log-rank test. Correlation test was calculated by Pearson chi-square.Univariate and multivariate survival analysis were performed with the Coxproportional hazards method. P＜0.05was judged to be significant.Results:1. TGFβ1expression in GC: TGFβ1expressed in166(90.2%) GCtumors and33(80.5%) peritumoral tissues, as brownish fine granules in thecytoplasm and membrane of GC cells. Based on the above evaluation criteria,82 (44.6%) GC tissues and28(68.3%) peritumoral tissues had low expression, and102(55.4%) GC tissues and13(31.7%) peritumoral tissues had high expression. Theexpression of TGFβ1in GC tissues were significantly higher than in peritumoraltissues (χ2=7.554, P=0.006). In some patients, expression of TGFβ1in tumor stromaand cancer cell was also observed.2. The relationship between TGFβ1expressionand clinicopathological parameters: TGFβ1expression was higher in the old thanin the young (P=0.017) and higher in intestinal type GC than in diffuse type GC(P=0.015). There were no statistically significant differences in TGFβ1among otherclinicopathological parameters.3. The relationship between clinicopathologicalparameters and OS of GC patients stratified by TGFβ1expression: Highexpression of TGFβ1was related to worse OS of GC patients. Log-rank test showedthat diffuse type GC patients had a poor survival than intestinal type GC patients(P=0.018). To further evaluate the relationship between TGFβ1expression and OS,the clinicopathological parameters were divided into different subgroups. HighTGFβ1expression had a worse survival in young people, female, diffuse type GC,poor differentiation, and lymph nodes metastasis. Univariate analysis showed thatage, Lauren classification, pathological grading, serosal invasion, lymph nodesmetastasis and TGFβ1expression were risk factors. Multivariate Cox proportionalhazards analysis showed that age, pathological grading, serosal invasion and TGFβ1expression were independent risk factors.Conclusion: This study explored the relationship between TGFβ1expressionand GC prognosis. The results showed TGFβ1mainly expressed in the GCcytoplasm and cytomembrane, with significantly higher expression in GC tissuesthan peritumoral tissues. This finding was similar to previous reports [18,22,26,30].What deserves more attention was the finding that some positive staining was alsofound in GC nucleus and stromal cells in some patient. This may had special significance from the perspectives of tumor microenvironment.Although it has been demonstrated that TGFβ1high expression was related topoor prognosis (including our study), previous studies have not defined whetherTGFβ1expression was related to other clinicopathological parameters. Our studyonly showed TGFβ1expression was related to patients’ age and GC type among theclinicopathological parameters investigated. Interestingly, TGFβ1is involved withEMT, and fibrosis is the typical feature of diffuse type GC, but this study showedTGFβ1expression was higher in intestinal type GC than in diffuse type GC.Therefore, more studies are needed to further elucidate the bidirectional functionsand prognosis value of TGFβ1in GC. Furthermore, Cox’s proportional hazard modelshows that TGFβ1expression, age, clinicopathologicalical grading and serosalinvasion were independent risk factors.In conclusion, this study has demonstrated that TGFβ1high expression wasrelated to poor prognosis of GC patients, who tend to be young, female, distantgastric tumor with poor differentiation, diffuse type, and lymph nodes metastasis.The present study perhaps has some enlightenment for TGFβ1treatment, namely inthe above subgroups GC patients, the treatment aim at TGFβ1would be moreeffective. Part II: Tumor invasion unit in gastric cancer revealed byQDs-based in situ molecular imaging and multispectral analysisObjective: Gastric cancer (GC) is the fourth most common cancer worldwide,and the third leading cancer cause in China. Despite recent progresses in the earlydiagnosis and the surgery-centered multidisciplinary treatments for GC, the overallclinical outcome of such patients is still far from satisfactory, mainly due to the post-treatment occurrence and metastasis, via blood circulation, lymphatic channelsor direct cancer cells invasion and seeding. To tackle this problem, many effortsfocusing on cancer cells have been made. And eventually, the oncology communityhas come to the understanding that cancer is a disease of imbalance, i.e., not merelya disease of rogue cells but the body’s mismanagement of those cells, thefundamental importance of such theoretical changes is that we have to pay particularattention to tumor microenvironment, in addition to cancer cells, because tumormicroenvironment plays an important role via the co-evolution of tumor cells andstroma. Thus, it is urgent ro explore the co-evolution of tumor cells and stroma.In GC, tumor microenvironment is a complex and dynamic community, whichis undergoing constant evolutions during cancer invasion, involving tumor cellsescape from primary sites into vasculature (blood circulation and lymphatic channel),reside and adhere to endothelial cells, penetrate from vasculature into other organsand reside in them, accompanied with tumor neo-vessels growth. Many importantcomponents in tumor microenvironment work together to contribute to cancerinvasion. Major components in tumor microenvironment are inflammatory cells suchas macrophages, immune cells such as T and B lymphocytes, stromal cells such asfibroblasts, and neo-blood vessels of various stages of maturity. These players mustbe in an appropriate anatomic proximity and spatial vicinity with the tumor cells inorder to facilitate cancer invasion. And indeed, recent studies have shown that tumorcells, macrophages and tumor neo-vessels in close vicinity with one another form aunique structure called tumor microenvironment of metastasis (TMEM), or in moreeasily understandable terms, called ‘tumor invasion unit’. Therefore, thesimultaneous recognition and analysis of all the components in the tumor invasionunit is very important to understanding the new perspective of cancer invasion.There are few techniques that can simultaneously image multiple components in complex tumor microenvironment of the same tissue section. Thus, it is urgent todevelop a more holistic method to image the complex interactions of stromalcomponents in situ. Quantum dots, with its unique size and surface effects, haveshown great potential in biomedical application, especially in multiplexed imagingin situ. In this study, taking the advantages of established QDs-based multiplexedimaging in situ, we analyzed on the interactions between macrophages, tumorneo-vessels and cancer cells, and developed a computer-based algorithm of tumorinvasion units.Methods:1. Patients and specimens: Tissue sections (4m thickness) of90human GC cases were selected from the central database on GC established at ourcancer center, including30with detailed pathological information, and60withcomplete clinic-pathological and survival information available on the patients.Tumor tissues from the first set of30patients were used for a trial study, to explorethe correlation of tumor invasion unit with classical pathological features, in order totest if there was any relationship between tumor invasion unit unfavorablepathological features. Tumor tissues from the second set of60patients with detailedsurvival information were used for validation study, in order to further verify iftumor invasion unit could predict the overall survival (OS).2. QDs-based doublemolecular imaging: The QDs-based in situ molecular imaging procedures wereperformed with the following major steps: tissue slides de-paraffinizing→antigenretrieval→blocking→primary antibody for macrophages and CD105→washingand blocking→staining with QDs-525, QDs-585or QDs-655→washing→detectionand acquisition.3. Image capture and analysis: The QDs stained images werecaptured by Olympus DP72cameraunder CRi Nuance multispectral imaging system.The QDs-525, QDs-585, QDs-655were excited by UV light (330-385nm). Aspectral cube for each image, which contains the complete spectral information at10 nm wavelength intervals from420-720nm, was collected by the CRi Nuancemultispectral imaging system. And all the cubes were captured under the samecondition at×200magnification with the same settings for each image, so as toavoid the selection bias. The QDs fluorescence signal unmixing was processed bythe software package within the Nuance system. Then, after obtaining the images ofsignal unmixing, the macrophages were counted on each image and the totalmacrophages was documented as the counts for the patient for further analysis. Thesame calculation method was used for tumor neo-vessels counting. After processingQDs images, the images with double signals of both macrophages and tumorneo-vessels at the tumor nest area were acquired. As the tumor invasion unitconsisted of cancer cells, macrophages and tumor neo-vessels, a circle with thediameter of60m centered on macrophages was chosen, approximately three celldiameters across. With the computer-based algorithm, if there was any red signal inthis circle, it was counted as1, otherwise; it was counted as0. Then the total countsof five images for each patient was output as the result of tumor invasion units.4.Statistical analysis: Statistical analyses were performed with SPSS software version21.0. For the comparison of individual variables, Fisher’s exact test, t test andMann-Whitney Test were conducted as appropriate. The Kaplan-Meier survivalcurves were plotted to analyze the OS by different study parameters, with log ranktest to define the statistical differences between the subgroups. Two sided P <0.05was judged as statistically significant.Results:1. Two types of tumor invasion unit with dynamic changes: Basedon the spreading and layout patterns of neo-vessels and the macrophages infiltrationsteps, two forms of tumor invasion units in constant dynamic changes could berecognized. In form one, the tumor neo-vessel was seen in longitudinal section. Withthe special longitudinal spreading and irregular morphology, the tumor neo-vessels presented to be multi-angled accompanied with macrophages infiltration at eachangle. Macrophages undergoing intravasation could also be observed, half in andhalf out of the blood vessel. In form two the tumor neo-vessel is seen in cross section,with macrophages lodging the vessel with membrane processes, crossing the vesselwall, and in close vicinity to the vessel, thus revealing the dynamic interactionsbetween macrophages and neo-vessels to facilitate tumor invasion.2. Tumorinvasion unit was correlated with worse clinico-pathological features in30GCcases in the trial study: First, we selected a trial set of30GC patients with the threemost common pathological types, including well differentiated, poorly differentiatedand signet-ring cell carcinoma. These patients were classified into3groups bydifferent pathological types,10cases in each group. According to the quantitativeanalysis, the mean macrophages density was higher in the poorly differentiatedgroup (1544) than the well differentiated group (317) and the signet-ring cell group(1011), with statistical significance in three groups. The results of tumor neo-vesselsanalyses were the same as the macrophages. The poorly differentiated group had thehighest count but the well differentiated group had the lowest counts, with statisticalsignificances among three groups (P <0.001, for between-group comparisons). Andalso, the number of tumor invasion units was much higher in the poorlydifferentiated group (373) than the well differentiated group (82) or the signet-ringcell group (177), with statistical significance in three groups (P=0.000, for allbetween-group comparisons). Thus, in this trial set, compared with the analysisresults of macrophages and tumor neo-vessels, tumor invasion units had the similareffects. However, the P value of tumor invasion unit was the smallest, suggesting itsstronger correlation with worst histological types.3. The impact of tumor invasionunit on clinical outcome: Taking the median value of macrophages, neo-vessels andtumor invasion units as the cut-off value, these parameters were divided into high density groups if the individual value was above the cut-off value, and low densitygroups if the individual value was below the cut-off value. Their OS curves bymacrophages density, neo-vessels density and tumor invasion units showed that bothmacrophages and tumor invasion units were independent factors to impact onsurvival, while tumor neo-vessel density itself was not an independent factor forsurvival. ROC analysis also demonstrated that tumor invasion unit had the largestarea under the curve (63.2%), suggesting that tumor invasion unit had biggerpredicting power than macrophages for OS prediction.4. Typical examples ofimpacts of tumor invasion unit on clinical prognosis: To further validate thisobservation, we selected6cases for detailed analysis,3cases of poorlydifferentiated adenocarcinoma with identical TNM stages and clinical treatments,and another3cases of highly differentiated adenocarcinoma with identical TNMstages and clinical treatments. From these cases, it could be observed more clearlythat the higher the number of tumor invasion units, the shorter the OS, regardless ofthe tumor differentiation and histological types. In the3cases of poorlydifferentiated adenocarcinoma, the tumor invasion units were7,20and58per×200magnification field, respectively; and the corresponding OS were20.6,11.1and3.3months, respectively. In another3cases of well differentiated adenocarcinoma, thetumor invasion units were3,11, and24per×200magnification field, respectively;and the corresponding OS were42.6,27.6, and14.3months, respectively.Conclusion: This study demonstrated that the complex and constantinteractions between tumor cells and their microenvironment could have asignificant impact on the clinical outcomes of GC patients. Specifically, tumor cells,new blood vessels due to tumor angiogenesis and infiltrating macrophages in thetumor microenvironment could form a spatially very close entity called tumorinvasion unit that facilitates tumor invasion. The number of tumor invasion units in the tumor tissue is negatively correlated with poor tumor differentiation and worsesurvival. In addition, the tumor invasion unit has bigger predictive power of OS thansingle component counts of macrophages and tumor neo-vessels, respectively.Interestingly, we have found that the tumor neo-vessels density was not anindependent prognostic factor for OS. This may suggest that the effects of tumorneo-vessels on cancer invasion can be regulated by tumor macrophages infiltrationwith synergistic effect, even at early stages of cancer development. This could atleast account for the emerging concept of the tumor invasion unit. QDs-basedmolecular imaging and multispectral analysis could make a unique contribution inthis regard, as the current study demonstrated. QDs are engineered nanoparticleswith unique optical properties suitable for biomedical application. Compared withorganic dyes and fluorescent proteins, due to its unique size and surface feature, QDshave many advantages such as composition-tunable light emission, enhancedfluorescence brightness, strong resistance to photobleaching and chemicaldegradation. In addition, different colors of QDs can be simultaneously excited by asingle light source, with minimal spectral overlapping, which provides significantadvantages for multiplexed detection of target. This property is very suitable forinvestigating the complex interactions between tumor cells and their surroundingmicroenvironment at the architectural level. In this study, the infiltratingmacrophages and tumor angiogenesis were simultaneously labeled green and red,respectively. The blue auto-fluorescence also showed clear tumor backgroundstructure. Therefore, instead of producing artificial overlay images by conventionalimaging techniques, this QDs based molecular imaging technique could provideauthentic multicolor images to better reveal the complex interactions of differentcomponents in the tumor tissue.Cancer recurrence and metastasis is the root cause of treatment failure in GC, and tumor cells escape is the first and most important step towards distant metastasis.There are currently no reliable methodologies to predict the risk for metastaticdisease. Taking the advantages of QDs-based multiplexed molecular imagingtechnology, this study has simultaneously revealed the spatial distribution of tumorinvasion unit, consisting of tumor cells, macrophages and tumor neo-vessels anddeveloped a computer-based algorithm for tumor invasion unit analysis at molecularlevel. Both the novel approach and the results in this study suggested the greatpredictive power of tumor invasion unit for OS. A next step will be to validate thesefindings in a larger, independent and standardized GC database with known clinicaloutcome. If the results can be substantiated, the tumor invasion unit could be apowerful tool addition to the current approach for guiding personalized therapy andassessing prognosis, so as to better prevent over-treatment and under-treatment ofcancer patient.