Research ArticleImmune Cell Infiltration and Identifying Genes of PrognosticValue in the Papillary Renal Cell CarcinomaMicroenvironment by Bioinformatics Analysis

Ting Liu ,1 Man Zhang ,2 and Deming Sun 2

1Department of Nephrology, The People’s Hospital of China Three Gorges University; The First People’s Hospital of Yichang,Yichang 443000, Hubei, China2Department of Urology, The People’s Hospital of China Three Gorges University; The First People’s Hospital of Yichang,Yichang 443000, Hubei, China

Correspondence should be addressed to Deming Sun; sundem1450@126.com

Ting Liu and Man Zhang contributed equally to this work.

Received 16 February 2020; Revised 21 June 2020; Accepted 7 July 2020; Published 26 July 2020

Academic Editor: Lucia Lopalco

Copyright © 2020 Ting Liu et al. This is an open access article distributed under the Creative Commons Attribution License, whichpermits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Papillary renal cell carcinoma (PRCC) is one of the most common histological subtypes of renal cell carcinoma. Type 1 and type 2PRCC are reported to be clinically and biologically distinct. However, little is known about immune infiltration and the expressionpatterns of immune-related genes in these two histologic subtypes, thereby limiting the development of immunotherapy for PRCC.Thus, we analyzed the expression of 22 immune cells in type 1 and type 2 PRCC tissues by combining The Cancer Genome Atlas(TCGA) database with the ESTIMATE and CIBERSORT algorithms. Subsequently, we extracted a list of differentially expressedgenes associated with the immune microenvironment. Multichip mRNA microarray data sets for PRCC were downloaded fromthe Gene Expression Omnibus (GEO) to further validate our findings. We found that the immune scores and stromal scoreswere associated with overall survival in patients with type 2 PRCC rather than type 1 PRCC. Tumor-infiltrating M1 and M2macrophages could predict the clinical outcome by reflecting the host’s immune capacity against type 2 PRCC. Furthermore,CCL19/CCR7, CXCL12/CXCR4, and CCL20/CCR6 were shown to be potential new targets for tumor gene therapy in type 2PRCC. Our findings provide valuable resources for improving immunotherapy for PRCC.

1. Introduction

Papillary renal cell carcinoma (PRCC) is the second mostcommon histological subtype of renal cell carcinoma afterthe clear cell subtype, where it accounts for 15% to 20% ofkidney cancers [1]. Type 1 and type 2 PRCC are shown tobe different types of renal cancer with distinct histologic fea-tures and diverse clinical prognosis. Compared with type 1PRCC, type 2 PRCC is more heterogeneous and aggressivewhich is also associated with higher mortality. Successfultreatment is challenging after metastasis occurs because thetargeted drugs are generally less robust [2]. The drug resis-tance mechanism remains unclear, and there is no standardtherapy for metastatic PRCC patients at present. PRCC has

attracted increasing attention recently, especially in the fieldof immunotherapy.

The tumor microenvironment clearly affects the survivaland development of tumor cells, and it generally has twomain components, where the cellular component is charac-terized by the infiltration of different cells such as inflamma-tory cells, immune cells, mesenchymal stem cells, andfibroblast cells, whereas the noncellular component mainlycomprises cytokines and chemokines [3, 4]. The componentsof the tumor microenvironment are highly complex, andeach has different effects on cell proliferation, tumor distantmetastasis, and drug resistance [5, 6]. In the microenviron-ment system, stromal and immune cells are two importantnontumor components, and they are suitable targets for

HindawiBioMed Research InternationalVolume 2020, Article ID 5019746, 12 pageshttps://doi.org/10.1155/2020/5019746

immunotherapy and prognosis evaluation in various cancerssuch as melanoma [7], non-small cell lung cancer [8], andprostatic cancer [9]. In addition, previous studies have shownthat the tumor microenvironment can have significantimpacts on the expression of genes in tumor tissues [10].Further investigations of genetic changes in specific tumormicroenvironments may facilitate targeted therapy.

Two algorithms called the MAlignant Tumor tissue usingExpression data (ESTIMATE) and the Cell-type Identifi-cation By Estimating Relative Subsets Of known RNATranscripts (CIBERSORT) were developed recently. TheESTIMATE algorithm was designed by Yoshihara et al.,and it can be used for calculating immune and stromal scoresby analyzing the specific gene expression characteristics ofimmune cells and stromal cells and for predicting the infiltra-tion of nontumor cells in tumor tissue [11]. CIBERSORT is agene expression-based deconvolution algorithm invented byNewman et al. that utilizes a matrix of 547 barcode genesfor characterizing immune cell components in order to iden-tify the diversity and full range of infiltrating immune cells[12]. These two useful algorithms have been applied inmicroenvironment research for various cancers [13–15],thereby demonstrating the practical utility of big data-basedarithmetic.

Little is known about the immune cell infiltration profileand genetic basis of PRCC, thereby limiting the developmentof immune treatments for PRCC. Thus, in the present study,we conducted a comprehensive assessment of tumormicroenvironment-related cells and genes, as well as deter-mined the effects of immune signatures on the outcomes oftype 1 and type 2 PRCC patients. We analyzed the expressionpatterns of 22 immune cells in these two tumor entities bycombining The Cancer Genome Atlas (TCGA) database withthe ESTIMATE and CIBERSORT algorithms. Subsequently,we extracted a list of differentially expressed genes (DEGs)associated with the immune microenvironment. These corre-lations were validated using multichip mRNA microarraydata sets for PRCC downloaded from the Gene ExpressionOmnibus (GEO).

2. Materials and Methods

2.1. Data Acquisition. The gene expression profiles andcorresponding clinical information were downloaded fromthe TCGA website for kidney renal papillary cell carcinomaprojects (TCGA-KIRP) (available online: https://portal.gdc.cancer.gov/). Immune scores and stromal scores werecalculated using the ESTIMATE algorithm (availableonline: https://bioinformatics.mdanderson.org/estimate/).The CIBERSORT algorithm (available online: https://cibersort.stanford.edu/) was applied to analyze the normal-ized gene expression matrix and to calculate the relative pro-portions of 22 infiltrating immune cells. Validation wasconducted using expression profiles GSE7023 and GSE2748downloaded from GEO (Affymetrix Human Genome U133Plus 2.0 platform) for 12 paracarcinoma tissues and 33 type1 and 31 type 2 PRCC tissues (available online: https://www.ncbi.nlm.nih.gov/geo/).

2.2. Processing and Aggregation of Raw Data. R (version3.6.0) package “Limma” [16] was used to process and mergedata downloaded from TCGA in order to obtain stromal andimmune cell scores, as well as the normalized gene expressionmatrix in the tumor microenvironment. The SVA package[17] with the ComBat function was used to normalize themultichip GEO data set to remove batch effects and obtaina standardized gene expression matrix. Normalized geneexpression data from TCGA and GEO were uploaded tothe CIBERSORT website and analyzed automatically usingthe default signature matrix with 1000 permutations. Subse-quently, we selected the filtered cases with P values < 0.05for further analysis because CIBERSORT generates a P valuefor the deconvolution of each sample using Monte Carlosampling [18].

2.3. Different Immune Cell Expression Patterns in PRCC andNormal Tissues. The data obtained after filtration were usedto analyze the expression patterns of 22 immune cell propor-tions in PRCC and normal tissues, where P < 0:05 indicated astatistically significant difference. Correlation analysis wasperformed based on 22 immune cells in order to determinethe internal relationships among each of the immune cells.

2.4. Identification of DEGs in PRCC ImmuneMicroenvironment. The data were rearranged according tohigh and low immune scores in order to identify DEGsassociated with the immune microenvironment. The cutoffcriteria were specified as follows: fold change > 1:5 andadjusted P value < 0.05.

2.5. Functional Enrichment Analysis of DEGs and Screeningfor Key Modules. The Database for Annotation, Visualizationand Integrated Discovery (DAVID) is used widely for biologi-cal information analysis (available online: https://david.ncifcrf.gov/). We utilized DAVID to perform Gene Ontology (GO)and Kyoto Encyclopedia of Genes and Genomes (KEGG)pathway enrichment analysis for DEGs. An adjusted P < 0:05indicated a significant difference. In order to identify the keymodules among the DEGs, a protein-protein interaction(PPI) network was generated using the Search Tool for theRetrieval of Interacting Genes (STRING) (version 10.5)online database (available online: https://string-db.org/). Aninteraction score > 0:95 was considered meaningful. Cytos-cape software (version 3.7.0) [19] was used to visualize inter-acting networks of DEGs. Subsequently, we used MolecularComplex Detection (MCODE) (version 1.4.2) to identify keymodules in the established network. The screening criteriacomprised: max depth = 100, degree cutoff = 2, k‐score = 2,MCODE score > 5, and node score cutoff = 0:2.

2.6. Prognostic Value of Microenvironment in PRCC. Clinicalinformation was downloaded from TCGA. Patients withany missing data were excluded. Corresponding survivalanalyses were conducted for PRCC patients using theKaplan–Meier method. The log-rank test was conducted,and a P value < 0.05 indicated a significant difference.

2 BioMed Research International

3. Results and Discussion

3.1. Relationships between Stromal/Immune Scores andPrognosis in PRCC Patients. The gene expression matrixand related clinical information for 136 PRCC tissues weredownloaded from TCGA. In order to evaluate the effects ofstromal/immune scores on the overall survival rates of PRCCpatients, we divided 136 PRCC cases into two groups (highvs. low score groups) according to the stromal or immunescores. Kaplan–Meier survival curves indicated that the stro-mal score was not related to overall survival, but a lowimmune score was associated with worse overall survival(Figure 1(a)). Additionally, after dividing tumor tissues intotype 1 and type 2 PRCC, the result showed that immunescores and stromal scores were associated with overall sur-vival in patients with type 2 PRCC rather than type 1 PRCC(Figures 1(b) and 1(c)). Wilcoxon’s signed-rank test was usedto further investigate the effects of the stromal and immunescores on clinical characteristics, such as the tumor stage, dis-tant metastasis, lymph node, and topography. The resultsindicated that both the stromal and immune scores had no

meaningful effects on clinical characteristics in patients withtype 1 PRCC (Figure 2(a)). Notably, immune scores ratherthan stromal scores were associated with tumor stage, distantmetastasis, and topography in type 2 tumor entity, indicatingthat low immune scores reflected an aggressive clinicalcourse in type 2 PRCC patients (Figure 2(b)).

3.2. Infiltration by 22 Immune Cell Types in PRCC andNormal Tissues. During the calculation process, CIBER-SORT allocates a P value for deconvolution to each samplein order to estimate the reliability of each result. Caseswith a P value < 0.05 were considered more credible. There-fore, the filtered files with P values < 0.05 were eligible forfurther analyses. After screening, we selected 95 TCGA cases(6 normal tissues, 38 type 1 PRCC, and 51 type 2 PRCC tis-sues) and 76 GEO cases (12 normal tissues, 33 type 1 PRCC,and 31 type 2 PRCC tissues) to investigate the profiles of 22infiltrating immune cells. The results showed that each typeof immune cell exhibited different infiltration properties intype 1 and type 2 PRCC tissues. Compared with normal tis-sues, PRCC tissues generally contained a lower proportion of

(a)

0 5 10 15Time (year)

Ove

rall

surv

ival

0 5 10 150.0

0.2

0.4

0.6

0.8

1.0O

vera

ll su

rviv

al

0.0

0.2

0.4

0.6

0.8

1.0

Time (year)

PRCC

Stromal score (P = 0.23) Immune score (P = 0.02)

(b) (c)Time (year) Time (year)

0 1 2 3 4 50 2 4 6 8 10 12

HighLow

Type 1 PRCC

Stromal score (P = 0.85) Immune score (P = 0.51)

Ove

rall

surv

ival

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

Time (year) Time (year)0 2 4 6 8 10 0 2 4 6 8 10

Type 2 PRCC

Stromal score (P = 0.03) Immune score (P = 0.02)

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

Figure 1: Relationships between stromal/immune scores and prognosis in PRCC patients. (a) Effects of stromal scores and immune scores onoverall survival rate in PRCC patients. Effects of stromal/immune scores on overall survival rate in (b) type 1 and (c) type 2 PRCC patients.

3BioMed Research International

M1 macrophages, naive B cells, plasma cells, and dendriticcells, whereas the fractions of the M2 macrophage and mono-cyte were relatively higher (Figure 3(a)). Importantly, type 2was characterized by M2 macrophage infiltration and M1macrophage deficiency which differed significantly from type1 PRCC (Figure 3(b)), revealing a potential role of M2 andM1macrophages in PRCC. To validate these preliminary con-clusions, we performed the same analysis using GEO data, andthe detailed results are presented in Supplementary Figs. S1aand S1b. Surprisingly, the results obtained based on the twodatabases were almost identical. In fact, the distributions ofcells exhibited obvious group-biased clustering according toprincipal component analysis (PCA) in different tissues

(Figure 3(c)), which showed that the gene expression patternof type 1 PRCC was different from that of type 2 PRCC. Addi-tionally, correlation analysis revealed that the relationshipsbetween immune cell subpopulations were low to moderate(Figure 3(d)).

3.3. Associations between Key Immune Cell Types and OverallSurvival in PRCC Patients. Immune cells that differed signif-icantly between PRCC and normal tissues were selected toperform survival analyses. Clinical data for 89 PRCC patientswere downloaded from TCGA. The results showed that highproportions of M2 macrophages were associated with aworse survival rate in patients with type 2 PRCC, whereas

−1500

−1000

−500

0

500

−1500

−1000

−500

0

500

−1500

−1000

−500

0

500

−1000

0

1000

2000

3000

−1000

0

1000

2000

3000

−1000

0

1000

2000

3000

−1000

0

1000

2000

3000

−1500

−1000

−500

0

500

P = 0.48 P = 0.31

M0 M1

M0 M1

N0 N1 N2

N0 N1 N2

Stage I IV

P = 0.28

II III

Stage I IVII III

P = 0.43

T1 T2 T3

T1 T2 T3

P = 0.91 P = 0.48 P = 0.21 P = 0.18

Imm

une s

core

Stro

mal

scor

e

Type

1 P

RCC

(a)

−1000

−500

0

500

1000

–1000

–500

0

500

1000

3000

–1000

0

1000

2000

3000P = 0.28

T1 T2 T3−1000

0

1000

2000

3000

−1000

0

1000

2000

3000

−1000

0

1000

2000

3000

−1000

0

1000

2000

T1 T2 T3

–1000

–500

0

500

1000

M0 M1 N0 N1 N2 Stage I IVII III

M0 M1 N0 N1 N2 Stage I IVII III

P = 0.01

P = 0.12

P < 0.01 P < 0.01P = 0.21

P = 0.32 P = 0.09

Imm

une s

core

Stro

mal

scor

e

Type

2 P

RCC

(b)

Figure 2: Relationship between stromal/immune scores and clinical characteristics. Associations between stromal/immune scores and distantmetastasis, lymph nodes, clinical stage, and topography in (a) type 1 PRCC patients and (b) type 2 PRCC patients. M: distant metastasis; N:lymph node; T: topography.

4 BioMed Research International

M2 macrophage CD8+ T cell Resting mast cellPlasma cellMonocyteNeutrophilResting dendritic cell Naive B cell Memory B cell Follicular helper T cell Memory CD4+ T cell Activated dendritic cell Activated mast cell EosinophilNaive CD4+ T cell Resting NK cellGamma delta T cellM1 macrophage Regulatory T cell Activated NK cellResting memory CD4+ T cellM0 macrophage

Type

TypeNormal

0

0.3

0.6

Type 1 PRCC Type 2 PRCC

(a)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Frac

tion

P = 0.43

P = 0.39

P = 0.87P = 0.07

P = 0.21

P = 0.81

P = 0.54P = 0.33

P = 0.42

P = 0.64P = 0.21

NormalType 1 PRCC Type 2 PRCC

Neu

trop

hil

Mas

t cel

l

Den

driti

c cel

l

M2

mac

roph

age

M1

mac

roph

age

M0

mac

roph

age

Mon

ocyt

e

Act

ivat

ed N

K ce

ll

Gam

ma d

elta

T ce

ll

Regu

lato

ry T

cell

Folli

cula

r hel

per T

cell

Mem

ory

CD4+

T ce

ll

Nai

ve C

D4+

T ce

ll

CD8+

T ce

ll

Plas

ma c

ell

Mem

ory

B ce

ll

Nai

ve B

cell

⁎⁎⁎⁎⁎⁎

⁎⁎⁎

⁎⁎

⁎⁎

⁎⁎⁎

⁎⁎⁎

⁎

⁎

⁎⁎⁎

⁎⁎⁎

⁎⁎⁎

⁎⁎⁎

(b)

–4

–2

0

2

4

6

–7.5 –5.0 –2.5 0.0 2.5PCA1

PCA

2

NormalType 1 PRCCType 2 PRCC

−2.5

0.0

2.5

–2.5 0.0 2.5 5.0

PCA1

PCA

2

Normal

(c)

Figure 3: Continued.

5BioMed Research International

M1 macrophages had the opposite association. However, nosignificant associations between immune cell types andoverall survival were found in patients with type 1 PRCC(Figure 4).

We further evaluated the relationships among survival-related immune cells and the clinicopathological characteris-tic in patients with type 2 PRCC. Subgroup analysis indicatedthat the proportion of M1 macrophages exhibited a decreas-ing trend with the development of topography and distantmetastasis in patients with type 2 PRCC (Figure 5(a)). Bycontrast, an increasing trend in the M2 macrophage fractionwas associated with tumor progression (Figure 5(b)).

3.4. Identification of DEGs and Key Genes in PRCC ImmuneMicroenvironment. As mentioned above, higher immunescores were related to better overall survival in patients withtype 2 PRCC. To explore the different gene expressionprofiles in the type 2 PRCC immune microenvironment, weordered the data according to the high and low immunescores. In the high immune score group, 122 overlappingDEGs were identified in TCGA and GEO databases, with108 significantly upregulated genes and 14 downregulatedgenes (Figure 6(a)). GO and KEGG enrichment analyseswere performed to determine the functions and pathwaysassociated with these DEGs. The results showed that immunecell adhesion or activation and cytokine/chemokine bindingfunction were the most significant biological process andmolecular function identified for these genes (Figure 6(b)).

In addition, cytokine-cytokine receptor interaction, chemo-kine signaling pathway, and Toll-like receptor pathwaywere the top pathways determined by KEGG analysis(Figure 6(c)). Subsequently, we performed PPI network anal-ysis to analyze the potential interplay among the DEGs usingthe online STRING tool. The PPI network obtained for theDEGs is illustrated in Supplementary Fig. S2. We obtainedthree meaningful modules using Cytoscape software, wheremodule 1 contained 23 hub genes (e.g., CCL20, CCl19, andCXCL12), and there were 21 hub genes and 10 hub genesin module 2 and module 3 (Figure 6(d)), respectively. Thesegenes were closely related to each other, revealing a compli-cated regulatory network in type 2 PRCC. Additionally, weprepared Kaplan–Meier survival curves based on 51 type 2PRCC samples downloaded from the TCGA database to ana-lyze the effects of the hub genes on overall survival. We foundthat high expression levels of CCL19 were associated with bet-ter overall survival in type 2 PRCC patients, whereas CXCL12and CCL20 showed opposite outcomes (Figure 6(e)).

4. Discussion

Emerging evidence indicates that patients in the same TNMstage might exhibit differences in their overall survivalbecause the TNM staging system focuses only on the traitsof the tumor itself, and it fails to fully consider the immunecharacteristics of tumor tissues [20]. The purity of a tumorand infiltrating immune cells may have crucial effects on

–1

–0.8

–0.6

–0.4

–0.2

0

0.2

0.4

0.6

0.8

1 1

0.27

0.08

0.14

−0.36

−0.38

−0.1

−0.3

−0.32

−0.07

−0.34

−0.17

−0.21

−0.18

0.09

−0.07

−0.04

0

−0.13

−0.13

−0.15

−0.2

0.27

1

−0.05

0

0.18

−0.18

−0.01

−0.12

0.07

0.04

−0.09

0.09

−0.08

−0.11

−0.02

−0.05

0.3

−0.05

0.02

−0.06

−0.06

−0.09

0.08

−0.05

1

0.18

−0.28

0.04

0.11

−0.04

−0.15

−0.08

−0.15

−0.03

−0.17

−0.01

−0.03

0.01

−0.03

−0.04

−0.07

−0.04

−0.08

−0.12

0.14

0

0.18

1

−0.45

−0.16

−0.14

−0.16

−0.21

−0.23

−0.42

−0.32

−0.47

−0.45

−0.03

−0.2

−0.05

−0.18

−0.07

−0.36

−0.44

−0.39

−0.36

0.18

−0.28

−0.45

1

0.2

0.09

0

0.13

0.17

0.34

0.37

0.34

0.12

−0.15

−0.12

0.09

−0.01

0.08

0.07

0.15

0.06

−0.38

−0.18

0.04

−0.16

0.2

1

0.03

0.28

0.05

−0.17

−0.21

−0.19

−0.03

−0.15

−0.06

0.11

0.18

−0.07

0.2

−0.09

−0.12

0.04

−0.1

−0.01

0.11

−0.14

0.09

0.03

1

0.14

0.05

−0.11

−0.15

−0.01

−0.02

0.06

−0.02

0.01

0.02

−0.05

0

−0.05

0.03

0.18

−0.3

−0.12

−0.04

−0.16

0

0.28

0.14

1

0.33

−0.39

−0.37

−0.39

−0.14

−0.12

0.05

−0.23

−0.07

−0.09

0.09

0.1

−0.16

0.2

−0.32

0.07

−0.15

−0.21

0.13

0.05

0.05

0.33

1

−0.16

−0.03

−0.07

0.03

0.16

−0.02

−0.07

0.05

−0.11

−0.01

0.01

−0.02

0.3

−0.07

0.04

−0.08

−0.23

0.17

−0.17

−0.11

−0.39

−0.16

1

0.56

0.57

0.12

0.08

0

0.35

−0.05

0.15

−0.09

−0.07

0.05

−0.07

−0.34

−0.09

−0.15

−0.42

0.34

−0.21

−0.15

−0.37

−0.03

0.56

1

0.68

0.4

0.43

−0.04

0.15

−0.08

0.03

−0.04

0.09

0.16

−0.08

−0.17

0.09

−0.03

−0.32

0.37

−0.19

−0.01

−0.39

−0.07

0.57

0.68

1

0.21

0.29

−0.04

0.19

0.05

0.03

−0.08

0

0.12

−0.07

−0.21

−0.08

−0.17

−0.47

0.34

−0.03

−0.02

−0.14

0.03

0.12

0.4

0.21

1

0.4

0.01

0.08

−0.08

0.23

0.1

0.19

0.21

0.02

−0.18

−0.11

−0.01

−0.45

0.12

−0.15

0.06

−0.12

0.16

0.08

0.43

0.29

0.4

1

0.03

0.18

−0.03

0.04

−0.16

0.28

0.17

0.11

0.09

−0.02

−0.03

−0.03

−0.15

−0.06

−0.02

0.05

−0.02

0

−0.04

−0.04

0.01

0.03

1

0.11

−0.02

−0.03

−0.03

0.12

−0.07

0.12

−0.07

−0.05

0.01

−0.2

−0.12

0.11

0.01

−0.23

−0.07

0.35

0.15

0.19

0.08

0.18

0.11

1

0.54

0.29

−0.07

0.15

0.11

0.04

−0.04

0.3

−0.03

−0.05

0.09

0.18

0.02

−0.07

0.05

−0.05

−0.08

0.05

−0.08

−0.03

−0.02

0.54

1

−0.04

−0.02

0.14

−0.04

0.02

0

−0.05

−0.04

−0.18

−0.01

−0.07

−0.05

−0.09

−0.11

0.15

0.03

0.03

0.23

0.04

−0.03

0.29

−0.04

1

0.34

0.26

0.15

0.02

−0.13

0.02

−0.07

−0.07

0.08

0.2

0

0.09

−0.01

−0.09

−0.04

−0.08

0.1

−0.16

−0.03

−0.07

−0.02

0.34

1

0.08

−0.04

0.02

−0.13

−0.06

−0.04

−0.36

0.07

−0.09

−0.05

0.1

0.01

−0.07

0.09

0

0.19

0.28

0.12

0.15

0.14

0.26

0.08

1

0.28

0.21

−0.15

−0.06

−0.08

−0.44

0.15

−0.12

0.03

−0.16

−0.02

0.05

0.16

0.12

0.21

0.17

−0.07

0.11

−0.04

0.15

−0.04

0.28

1

0.35

−0.2

−0.09

−0.12

−0.39

0.06

0.04

0.18

0.2

0.3

−0.07

−0.08

−0.07

0.02

0.11

0.12

0.04

0.02

0.02

0.02

0.21

0.35

1

Plas

ma c

ell

Nai

ve B

cell

Activ

ated

den

driti

c cel

lM

emor

y B

cell

Eosin

ophi

lAc

tivat

ed m

emor

y CD

4+ T

cell

Nai

ve C

D4+

T ce

llM

1 m

acro

phag

eFo

llicu

lar h

elper

T ce

llRe

gulat

ory

T ce

llCD

8+ T

cell

Gam

ma d

elta T

cell

Mon

ocyt

eRe

sting

mem

ory

CD4+

T ce

llN

eutro

phil

Activ

ated

NK

cell

Resti

ng m

ast c

ell

M2

mac

roph

age

Resti

ng N

K ce

llAc

tivat

ed m

ast c

ell

M0

mac

roph

age

Resti

ng d

endr

itic c

ell

(d)

Figure 3: Profiles of infiltrating immune cells in normal and PRCC tissues. (a) Heat map based on proportions of 22 immune cell types innormal, type 1, and type 2 PRCC tissues. (b) Violin plot showing differences in the expression of 22 immune cell types in normal (blue),type 1 (red), and type 2 (green) PRCC groups. (c) PCA analyses indicating group-biased clustering of immune cells from 76 GEO casesand 95 TCGA cases. (d) Correlation analysis based on 22 immune cell subpopulations. ∗P < 0:05, ∗∗P < 0:01, and ∗∗∗P < 0:001.

6 BioMed Research International

the clinical outcomes of patients with kidney cancer [21, 22],thereby suggesting that the Immunoscore system couldcomplement the prognostic judgment system for tumors.Recently, the application of immunotherapies such as

PD-1/PD-L1 inhibitors or tumor vaccines has highlightedthe importance of tumor immune environment traits forPRCC [23, 24]. However, the tumor microenvironment forPRCC, especially for the different pathological subtypes,

0 2 4 6 8 10

Naive B cell (P = 0.52)

Time (year)0 2 4 6 8 10

Plasma cell (P = 0.63)

Time (year)0 2 4 6 8 10

Monocyte (P = 0.12)

Time (year)

0 2 4 6 8 10

M1 macrophage (P = 0.03)

Time (year)0 2 4 6 8 10

M2 Macrophage (P = 0.07)

Time (year)0 2 4 6 8 10

Dendritic cell (P = 0.37)

Time (year)

0 2 4 6 8 10

Naive B cell (P = 0.58)

Time (year)0 2 4 6 8 10

Plasma cell (P = 0.47)

Time (year)0 2 4 6 8 10

Monocyte (P = 0.59)

Time (year)

0 2 4 6 8 10

M1 macrophage (P = 0.035)

Time (year)0 2 4 6 8 10

M2 macrophage (P = 0.01)

Time (year)0 2 4 6 8 10

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

0.0

0.2

0.4

0.6

0.8

1.0

Dendritic cell (P = 0.32)

Time (year)

HighLow

Ove

rall

surv

ival

Ove

rall

surv

ival

Ove

rall

surv

ival

Ove

rall

surv

ival

Type

1 P

RCC

Type

2 P

RCC

Figure 4: Kaplan–Meier survival curves obtained to investigate the impacts of key immune cell types on overall survival in type 1 and type 2PRCC patients. A P value < 0.05 was considered to indicate a significant difference based on the log-rank test.

7BioMed Research International

remains poorly understood, and there is a lack of immunebiomarkers for predicting the prognosis or therapeutic effectfor patients.

This study is the first to determine the relationshipsbetween immune/stromal cells and clinical characteristicsin PRCC patients. We found that a higher Immunoscorewas associated with better overall survival in patients withtype 2 PRCC, thereby suggesting that local immune infiltra-tion probably has potential antitumor effects in type 2 PRCC.However, we did not observe this association in type 1 PRCC,revealing that type 1 and type 2 PRCC are distinctly differentkidney tumors with specific gene expression patterns. Addi-tionally, type 2 was shown to be a more aggressive PRCC sub-type than type 1, and our findings have more importantclinical significance for type 2 PRCC. Previous studies havehighlighted the important roles of immune microenviromentin tumor metastasis and immunotherapy. High Immuno-score was believed to be associated with a lower number ofmetastases and a better clinical prognosis in colorectal cancerpatients [25]. Advanced urothelial carcinoma patients withhigh Immunoscores could obtain greater benefits fromimmunotherapy, where the predictive accuracy of the micro-environmental score was comparable with that of the tumormutation burden [26]. It seems that tumor tissues with asmall amount of immune cell infiltration are more likely toachieve tumor immune escape [27]. Thus, a high Immuno-score can indicate a strong antitumor immune responseand a better immunotherapeutic effect.

Immune cell subpopulations can exhibit diverse cellularinfiltration patterns in different tumors. Therefore, determin-ing the distributions and amounts of different infiltratingimmune cells in tumors and evaluating the function of eachcell subset are major priorities [28]. In this study, data fromTCGA and GEO were combined to establish the immune cellinfiltration profiles in PRCC tissues. Interestingly, we foundthat M2 macrophages had higher infiltration levels in type2 PRCC tissues than paracarcinoma and type 1 tissues, andthis variation was closely associated with distant metastasisand worse survival, whereas the opposite trend was deter-mined for M1 macrophages. These findings suggested thatmacrophages might be a promising breakthrough of immu-notherapy for type 2 PRCC. In fact, macrophage phenotypesare critical mediators of immune regulation and tumormetastasis [29], but these “polarized states” are debatablefor macrophages [30]. The M1 phenotype (antitumoral)and M2 phenotype (protumoral) may simply represent theextremes in a series of macrophage functional states ratherthan different actual cell subsets [30]. Our results highlightthe importance of the functional status of macrophagesubpopulations and the potential capacity for using tumor-associated macrophages as biomarkers for type 2 PRCC. Itshould be noted that multiple immune escape mechanismsexist in tumors to inhibit CD8+ T cell attacks. A high levelof CD4+/CD8+ T cells is linked with an extended prognosisin various tumors [31–34], but we failed to establish thiscorrelation in the present study, thereby implying that

Stag

e I

Stag

e II

Stag

e III

Stag

e IV

M1 macrophage

T1 T2 T3 N0 N1 N20.0

0.2

0.4

0.6

0.8

0.0

0.2

0.4

0.6

0.8

0.0

0.2

0.4

0.6

0.8

M0 M10.00.10.20.30.40.50.60.7

Frac

tion

P = 0.25 P = 0.12 P < 0.01 P < 0.01

(a)

Stag

e I

Stag

e II

Stag

e III

Stag

e IV

M2 macrophage

T1 T2 T3 N0 N1 N2 M0 M10.00

0.05

0.10

0.15

0.00

0.05

0.10

0.15

0.00

0.05

0.10

0.15

0.00

0.05

0.10

0.15

Frac

tion

P = 0.23 P = 0.08 P < 0.01 P < 0.01

(b)

Figure 5: Relationships between survival-related immune cells and clinicopathological characteristics in patients with type 2 PRCC. (a)Associations between M1 macrophages and clinical stage, topography, lymph nodes, and distant metastasis. (b) Associations between M2macrophages and clinicopathological characteristics. T: topography; N: lymph node; M: distant metastasis.

8 BioMed Research International

macrophages might play more important roles in PRCCrather than CD4+/CD8+ cells [35].

High-throughput sequencing allows researchers toobtain a better understanding of the DEGs in tumor micro-

environments. Previous studies determined the upregulationof genes related to immune cell activation (e.g., CD3D, CD8,CXCR3, CCL5, CXCL9, and CXCL10) in cancer patientswith a better prognosis [36–38], thereby highlighting the

108 224406

TCGA GEO TCGA GEO

76 14 39

CXCL1, CXCL2, CXCL11, CXCL12,CCL19, CXCR4, CCR1, CCR2, CCR5...

INPP5J, ELFN1, PSD2, KCNB2,PLK5, PURG, KCNH3, CASKIN1...

Upregulated genes Downregulated genes

(a)

CCBP

MF

Leukocyte activationLeukocyte cell−cell adhesion

Leukocyte differentiationLymphocyte activation

T cell activation

MHC protein complexTertiary granule membrane

Specific granuleTertiary granule

Vesicle lumen

MHC class II receptor activityImmunoglobulin binding

Chemokine receptor bindingCytokine activity

Cytokine receptor binding

Gene ratioCount

510

1520

P value

0.001

0.005

0.01

0.050.7 0.8 0.9

GO enrichment

(b)

Th1 and Th2 cell differentiation

Chemokine signaling pathway

Cytokine−cytokine receptor interaction

0 1 2 3 4 5–Log 10 (P_value)

IL21, CXCL10, CSF3R, CCL19TNFSF18, CSF1R, CXCL12...

ITK, CXCL10, CCL19CXCL12, CCR5, CCL4...

VCAM1, CCL19, LY96LTA, CCL4, CXCL12...CD79A, CD22, CR2FCGR2B, BTK, CD72...

TLR6, TLR7, CD80TLR8, CCL4, LY96...

CXCR4, ITK, CXCL12JAM2, RHOH, VCAM1...

Toll-like receptor signaling pathway

B cell receptor signaling pathway

NF-kappa B signaling pathway

Rap1 signaling pathwayHGF, RASGRP2, IGF1FPR1, FGF10, CSF1R...

Core genes Pathway

(c)

Module 1 Module 2 Module 3

CCL20

CXCL2

CCR2CXCL9

CCL4L1

CCL4

FPR1

CXCL11

CXCL12

CXCR4

CXCR3

CXCL1

C3AR1

CXCR2

CCR5

CXCL3

CCR1CCR7

CXCL16

C3

CCL19CCL13

CXCR1

CD86CD28

CD3D

HLA-DRB5

PTPRC

CD80

HLA-DRA

HLA-DQA2

HLA-DPB1CD3G

HLA-DQB1

HLA-DQB2

HLA-DPA1HLA-DQA1

ITK

GRAP2

LCKHLA-DRB1

LCP2CD3E

CD4

ITGB2

CD33

CYBAITGAM

ITGAL

CD53

CYBBFCER1GLILRB2

ITGAX

(d)

0 5 10 15

CCL19 (P = 0.01)

Time (year)

CCL20 (P = 0.046)

Time (year)

CXCL12 (P = 0.034)

Time (year)

Ove

rall

surv

ival

0 5 10 15

High expressionLow expression

0 5 10 150.0

0.2

0.4

0.6

0.8

1.0

Ove

rall

surv

ival

0.0

0.2

0.4

0.6

0.8

1.0

Ove

rall

surv

ival

0.0

0.2

0.4

0.6

0.8

1.0

(e)

Figure 6: Functional enrichment analysis and selection of key modules for DEGs in type 2 PRCC tissues. (a) Overlapping genes identified inTCGA and GEO databases. Significant (b) GO terms and (c) KEGG pathways for the upregulated genes identified in the high immune scoregroup of type 2 PRCC. (d) Three meaningful modules in the PPI network. (e) Kaplan–Meier curves were prepared according to the high andlow expression levels of core genes in type 2 PRCC.

9BioMed Research International

impacts of immune-related genes on tumor progression. Wescreened 108 upregulated DEGs in a high Immunoscoremicroenvironment in type 2 PRCC tissues. Functional enrich-ment analysis indicated that these genes are mainly responsi-ble for regulating cell-cell adhesion and cytokine/chemokinesignaling pathway activation. Variations in the cell-cell adhe-sion capacity are an important component of cancer cell inva-sion and metastasis [39, 40]. In addition, tumor-infiltratingimmune cells regulate cytotoxic effects and cancer-relatedinflammation by secreting various cytokines and chemokines[41]. Thus, the genes we identified are closely related to thedevelopment of type 2 PRCC.

The core genes were associated with immune cell activa-tion functions, including inflammation and immunity genes(e.g., CD3D, CD4, CD28, and HLA), chemokine genes (e.g.,CCL20, CXCL2, CCL19, and CXCL1), and chemokine rela-tive receptors (e.g., CXCR4, CCR6, and CCR7). It has beensuggested that the expression of CD3 and CD8 in the centerof tumors and invasive marginal areas can predict postoper-ative survival [42]. Correlations between immunity genesand overall survival were not found in our study, but weobserved that type 2 PRCC patients with high expressionlevels of chemokine CXCL12 or CCL20 had a low survivalrate, whereas the high level expression of CCL19 had antitu-mor effects. CXCL12 and CCL20 were shown to be stimula-tory chemokines related to tumor neovascularization [43].In addition, the CXCL12/CXCR4/CXCR7 and CCL20/CCR6chemokine axes are known to promote migration and theinvasiveness of cancer [44, 45]. By contrast, CCL19/CCR7executes antitumor effects via the induction of the Th1-polarized T cell immunity response [46] and an epithelial-to-mesenchymal transition [47]. CCL19 is also an adjuvantfor DNA vaccination [46], and it can improve the body’sresistance to tumor cells. Therefore, chemokines and theirrelated receptors are important elements that affect clinicaloutcomes, and different types of chemokines may havediverse effects on type 2 PPRC progression.

5. Conclusions

In the present study, we determined the detailed immunelandscape for type 1 and type 2 PRCC by elucidating distinctimmune infiltration patterns and assessing the prognosticvalue of different immune-infiltrating cells. We highlightedthe important functions and prognostic value of M1 macro-phages, M2 macrophages, and chemokines CCL19, CCL20,and CXCL20 in type 2 PRCC. These findings enhance ourunderstanding of immune infiltration and the expressionpatterns of immune-related genes in PCRR, which may bevaluable for improving immunotherapy approaches.

Data Availability

The data used to obtain our results are available from GEO(https://www.ncbi.nlm.nih.gov/geo/) and TCGA (https://portal.gdc.cancer.gov/), as well as the ESTIMATE (https://bioinformatics.mdanderson.org/estimate/) and CIBERSORT(https://cibersort.stanford.edu/) algorithm websites.

Conflicts of Interest

The authors declare that there is no conflict of interestregarding the publication of this paper.

Authors’ Contributions

Ting Liu and Man Zhang contributed equally to this work.

Acknowledgments

We thank International Science Editing (http://www.internationalscienceediting.com) for editing this manuscript.

Supplementary Materials

Figure S1: immune cell infiltration profiles determined innormal and PRCC tissues using GEO data. Figure S2:protein-protein interaction network of differentially expressedgenes in type 2 PRCC tissues. (Supplementary Materials)

References

[1] M. Chevarie-Davis, Y. Riazalhosseini, M. Arseneault et al.,“The morphologic and immunohistochemical spectrum ofpapillary renal Cell Carcinoma,” The American Journal ofSurgical Pathology, vol. 38, no. 7, pp. 887–894, 2014.

[2] G. Courthod, M. Tucci, M. Di Maio, and G. V. Scagliotti,“Papillary renal cell carcinoma: a review of the current thera-peutic landscape,” Critical Reviews in Oncology/Hematology,vol. 96, no. 1, pp. 100–112, 2015.

[3] D. Hanahan and R. A. Weinberg, “The hallmarks of cancer,”Cell, vol. 100, no. 1, pp. 57–70, 2000.

[4] D. Hanahan and L. M. Coussens, “Accessories to the crime:functions of cells recruited to the tumor microenvironment,”Cancer Cell, vol. 21, no. 3, pp. 309–322, 2012.

[5] A. Cretu and P. C. Brooks, “Impact of the non-cellular tumormicroenvironment on metastasis: potential therapeutic andimaging opportunities,” Journal of Cellular Physiology,vol. 213, no. 2, pp. 391–402, 2007.

[6] L. Kopfstein and G. Christofori, “Metastasis: cell-autonomousmechanisms versus contributions by the tumor microenviron-ment,” Cellular and Molecular Life Sciences, vol. 63, no. 4,pp. 449–468, 2006.

[7] A. Ladányi, “Prognostic and predictive significance of immunecells infiltrating cutaneous melanoma,” Pigment Cell & Mela-noma Research, vol. 28, no. 5, pp. 490–500, 2015.

[8] J. S. Seo, A. Kim, J. Y. Shin, and Y. T. Kim, “Comprehensiveanalysis of the tumor immune micro-environment in non-small cell lung cancer for efficacy of checkpoint inhibitor,”Scientific reports, vol. 8, no. 1, article 14576, 2018.

[9] S. L. Shiao, G. C. Y. Chu, and L. W. K. Chung, “Regulation ofprostate cancer progression by the tumor microenvironment,”Cancer Letters, vol. 380, no. 1, pp. 340–348, 2016.

[10] X. J. Ma, S. Dahiya, E. Richardson, M. Erlander, and D. C. Sgroi,“Gene expression profiling of the tumor microenvironmentduring breast cancer progression,” Breast Cancer Research,vol. 11, no. 1, article R7, 2009.

[11] K. Yoshihara, M. Shahmoradgoli, E. Martínez et al., “Inferringtumour purity and stromal and immune cell admixture from

10 BioMed Research International

expression data,” Nature Communications, vol. 4, no. 1,pp. 1–11, 2013.

[12] A. M. Newman, C. L. Liu, M. R. Green et al., “Robust enumer-ation of cell subsets from tissue expression profiles,” NatureMethods, vol. 12, no. 5, pp. 453–457, 2015.

[13] W. X.Wen, J. S. S. Soo, P. Y. Kwan et al., “Germline APOBEC3Bdeletion is associated with breast cancer risk in an Asian multi-ethnic cohort and with immune cell presentation,” BreastCancer Research, vol. 18, no. 1, p. 56, 2016.

[14] N. Shah, P. Wang, J. Wongvipat et al., “Regulation of theglucocorticoid receptor via a BET-dependent enhancer drivesantiandrogen resistance in prostate cancer,” eLife, vol. 6, articlee27861, 2017.

[15] W. H. Fridman, L. Zitvogel, C. Sautès–Fridman, andG. Kroemer, “The immune contexture in cancer prognosisand treatment,” Nature Reviews Clinical oncology, vol. 14,no. 12, pp. 717–734, 2017.

[16] C. W. Law and M. Alhamdoosh, “RNA-seq analysis is easy as1-2-3 with limma, Glimma and edgeR,” F1000Research, vol. 5,article 1408, 2016.

[17] J. T. Leek,W. E. Johnson, H. S. Parker, A. E. Jaffe, and J. D. Storey,“The sva package for removing batch effects and other unwantedvariation in high-throughput experiments,” Bioinformatics,vol. 28, no. 6, pp. 882-883, 2012.

[18] D. Van Ravenzwaaij, P. Cassey, and S. D. Brown, “A simpleintroduction to Markov chain Monte–Carlo sampling,” Psy-chonomic Bulletin & Review, vol. 25, no. 1, pp. 143–154, 2018.

[19] P. Shannon, A. Markiel, O. Ozier et al., “Cytoscape: a softwareenvironment for integrated models of biomolecular interac-tion networks,” Genome Research, vol. 13, no. 11, pp. 2498–2504, 2003.

[20] B. Mlecnik, G. Bindea, F. Pagès, and J. Galon, “Tumor immu-nosurveillance in human cancers,” Cancer Metastasis Reviews,vol. 30, no. 1, pp. 5–12, 2011.

[21] S. Chevrier, J. H. Levine, V. Zanotelli et al., “An immune atlasof clear cell renal cell carcinoma,” Cell, vol. 169, no. 4, pp. 736–749.e18, 2017.

[22] H. Matsushita, Y. Sato, T. Karasaki et al., “Neoantigen load,antigen presentation machinery, and immune signaturesdetermine prognosis in clear cell renal cell carcinoma,” CancerImmunology Research, vol. 4, no. 5, pp. 463–471, 2016.

[23] R. R. McKay, D. Bossé, W. Xie et al., “The clinical activity ofPD-1/PD-L1 inhibitors in metastatic non–clear cell renal cellCarcinoma,” Carcinoma, vol. 6, no. 7, pp. 758–765, 2018.

[24] R. T. Zhang, S. D. Bines, C. Ruby, and H. L. Kaufman,“TroVax® vaccine therapy for renal cell carcinoma,” Immuno-therapy, vol. 4, no. 1, pp. 27–42, 2012.

[25] M. Van den Eynde, B. Mlecnik, G. Bindea et al., “The linkbetween the multiverse of immune microenvironments inmetastases and the survival of colorectal cancer patients,”Cancer Cell, vol. 34, no. 6, pp. 1012–1026.e3, 2018.

[26] E. Becht, N. A. Giraldo, L. Lacroix et al., “Estimating thepopulation abundance of tissue-infiltrating immune and stro-mal cell populations using gene expression,” Genome Biology,vol. 17, no. 1, p. 218, 2016.

[27] A. Ribas, “Adaptive immune resistance: how cancer protectsfrom immune attack,” Cancer Discovery, vol. 5, no. 9,pp. 915–919, 2015.

[28] W. H. Fridman, F. Pagès, C. Sautès-Fridman, and J. Galon,“The immune contexture in human tumours: impact on

clinical outcome,” Nature Reviews Cancer, vol. 12, no. 4,pp. 298–306, 2012.

[29] Y. Lin, J. Xu, and H. Lan, “Tumor-associated macrophages intumor metastasis: biological roles and clinical therapeuticapplications,” Journal of Hematology & Oncology, vol. 12,no. 1, p. 76, 2019.

[30] R. Noy and J. W. Pollard, “Tumor-associated macrophages:from mechanisms to therapy,” Immunity, vol. 41, no. 1,pp. 49–61, 2014.

[31] L. A. Emens, S. C. Silverstein, S. Khleif, F. M. Marincola, andJ. Galon, “Toward integrative cancer immunotherapy: target-ing the tumor microenvironment,” Medicine, vol. 70, 2012.

[32] H. Angell and J. Galon, “From the immune contexture to theImmunoscore: the role of prognostic and predictive immunemarkers in cancer,” Current Opinion in Immunology, vol. 25,no. 2, pp. 261–267, 2013.

[33] B. Youngblood, J. S. Hale, H. T. Kissick et al., “Effector CD8 Tcells dedifferentiate into long-lived memory cells,” Nature,vol. 552, no. 7685, pp. 404–409, 2017.

[34] J. Xiao, W. Xi, Y. Zhu, H. Wang, X. Hu, and J. Guo, “Check-point molecule PD-1-assisted CD8+ T lymphocyte count intumor microenvironment predicts overall survival of patientswith metastatic renal cell carcinoma treated with tyrosinekinase inhibitors,” Cancer Management and Research,vol. 10, pp. 3419–3431, 2018.

[35] T. Chanmee, P. Ontong, K. Konno, and N. Itano, “Tumor-associated macrophages as major players in the tumor micro-environment,” Cancers, vol. 6, no. 3, pp. 1670–1690, 2014.

[36] Z. Jiang, Y. Xu, and S. Cai, “CXCL10 expression and prognos-tic significance in stage II and III colorectal cancer,”MolecularBiology Reports, vol. 37, no. 6, pp. 3029–3036, 2010.

[37] C. Curtis, S. P. Shah, S. F. Chin et al., “The genomic and tran-scriptomic architecture of 2,000 breast tumours reveals novelsubgroups,” Nature, vol. 486, no. 7403, pp. 346–352, 2012.

[38] R. G. W. Verhaak, P. Tamayo, J. Y. Yang et al., “Prognosticallyrelevant gene signatures of high-grade serous ovarian carci-noma,” The Journal of Clinical Investigation, vol. 123, no. 1,2012.

[39] V. Gkretsi and T. Stylianopoulos, “Cell adhesion and matrixstiffness: coordinating cancer cell invasion and metastasis,”Frontiers in Oncology, vol. 8, p. 145, 2018.

[40] P. Friedl and R. Mayor, “Tuning collective cell migration bycell-cell junction regulation,” Cold Spring Harbor Perspectivesin Biology, vol. 9, no. 4, article a029199, 2017.

[41] J. V. Fernandes, R. N. O. Cobucci, C. A. N. Jatobá, T. A. A. deMedeiros Fernandes, J.W. V. de Azevedo, and J. M. G. de Araújo,“The role of the mediators of inflammation in cancer devel-opment,” Pathology & Oncology Research, vol. 21, no. 3,pp. 527–534, 2015.

[42] J. Galon, A. Costes, F. Sanchez-Cabo et al., “Type, density, andlocation of immune cells within human colorectal tumors pre-dict clinical outcome,” Science, vol. 313, no. 5795, pp. 1960–1964, 2006.

[43] R. Janssens, S. Struyf, and P. Proost, “Pathological roles of thehomeostatic chemokine CXCL12,” Cytokine & Growth FactorReviews, vol. 44, pp. 51–68, 2018.

[44] A. Nazari, H. Khorramdelazad, and G. Hassanshahi, “Bio-logical/pathological functions of the CXCL12/CXCR4/CXCR7axes in the pathogenesis of bladder cancer,” InternationalJournal of Clinical Oncology, vol. 22, no. 6, pp. 991–1000,2017.

11BioMed Research International

[45] A. Muscella, C. Vetrugno, and S. Marsigliante, “CCL20promotes migration and invasiveness of human cancerous breastepithelial cells in primary culture,” Molecular Carcinogenesis,vol. 56, no. 11, pp. 2461–2473, 2017.

[46] J. Westermann, T. Nguyen-Hoai, G. Baldenhofer et al.,“CCL19 (ELC) as an adjuvant for DNA vaccination: inductionof a TH1-type T-cell response and enhancement of antitumorimmunity,” Cancer Gene Therapy, vol. 14, no. 6, pp. 523–532,2007.

[47] S. Cheng, J. Guo, Q. Yang, and X. Yang, “Crk-like adapter proteinregulates CCL19/CCR7-mediated epithelial-to-mesenchymaltransition via ERK signaling pathway in epithelial ovarian carci-nomas,” Medical Oncology, vol. 32, no. 3, p. 47, 2015.

12 BioMed Research International