Skip to main content
PMC home page
The American Journal of Pathology logoLink to The American Journal of Pathology
. 2009 Oct;175(4):1525–1535. doi: 10.2353/ajpath.2009.090295

Correlation of CXCL12 Expression and FoxP3+ Cell Infiltration with Human Papillomavirus Infection and Clinicopathological Progression of Cervical Cancer

Fatimah Jaafar *, Elda Righi †, Victoria Lindstrom †, Christine Linton †, Mahrokh Nohadani *, Susan Van Noorden *, Tyler Lloyd *, Joshua Poznansky *, Gordon Stamp *, Roberto Dina *, Dulcie V Coleman *, Mark C Poznansky †
PMCID: PMC2751549  PMID: 19808652

Abstract

Human cervical cancer is an immunogenic tumor with a defined pattern of histopathological and clinical progression. Tumor-infiltrating T cells contribute to immune control of this tumor; however, cervical cancer dysregulates this immune response both through its association with human papillomavirus (HPV) infection and by producing cytokines and chemokines. Animal tumor models have revealed associations between overproduction of the chemokine stromal cell-derived factor-1 (SDF-1 or CXCL12) and dysregulation of tumor-specific immunity. We therefore proposed that CXCL12 expression by cervical precancerous and cancerous lesions correlates with histopathological progression, loss of immune control of the tumor, and HPV infection. We found a significant association between cancer stage and CXCL12 expression for squamous and glandular lesions as well as with the HPV16+ (high-risk) status of the neoplastic lesions. Cancer progression was correlated with increasing levels of FoxP3 T-cell infiltration in the tumor. FoxP3 and CXCL12 expression significantly correlated for squamous and glandular neoplastic lesions. These observations were supported by enzyme-linked immunosorbent assay and Western blotting. In addition, we demonstrated CXCL12 expression by dyskaryotic cells in ThinPrep cervical smears. This study robustly links increased CXCL12 expression and FoxP3+-cell infiltration to HPV infection and progression of cervical cancer. It supports the detection of CXCL12 in cervical smears and biopsies as an additional biomarker for this disease.

Human cervical cancer is an immunogenic tumor, and several tumor-specific antigens have been identified in recent years.1,2,3 Persistent infection with high-risk human papillomavirus (HPV) (types 16 and 18) has been shown to be an important risk factor for malignant transformation of the cervical epithelium.4,5 Tumor immunity in this context results from the positive recognition of tumor antigens by immature dendritic cells, which subsequently migrate to draining lymph nodes and interact with and activate both helper CD4+ T cells and effector antigen-specific CD8+ T cells.1 Activated and targeted antigen-specific cytotoxic CD8+ T cells subsequently circulate through the peripheral blood to the tumor and then selectively transmigrate into the cancer itself and kill tumor cells. Immune control of cervical cancer is therefore critically dependent on the activation and migration of tumor-specific cytotoxic T cells into the tumor and the subsequent killing of neoplastic cells. The success of the immune response in destroying cervical cancer is thought to depend on the relationship between the kinetics of tumor growth and the efficacy of tumor-specific cytotoxic T cells (in balance with regulatory T cells) in infiltrating the tumor and killing neoplastic cells over time. It is well established that the magnitude of T-cell infiltration within certain forms of cancer correlates with a favorable prognosis.6

Tumors, including cervical cancer, evade immune recognition using a number of different mechanisms, including the down-regulation of major histocompatibility class I antibodies, the immunological ignorance to tumor antigens, the lack of costimulatory molecules and antigen loss, or the expression of inhibitory molecules.7,8 Several studies have also shown that tumors, including cervical cancer, overexpress specific chemokines, such as stromal cell-derived factor 1 (SDF-1 or CXCL12), that are thought to dysregulate the immune response to the tumor.9,10,11 Chemokines are a superfamily of 8- to 11-kDa proteins found in humans that have been shown to be critical in causing directional movement of immune cells in vitro and in determining the localization of leukocytes in models of inflammatory, immune-mediated diseases, and cancer.11,12,13 Chemokines have been shown to act as chemoattractants for leukocyte subpopulations including T cells and dendritic cells. Chemokines signal via Gαi protein-coupled receptors on the cell surface and subsequently induce directed cell movement in response to a gradient of the chemokines.14

The chemokine CXCL12 is a known T-cell chemoattractant that selectively binds the receptor CXCR4.9,10,11 Until recently CXCL12 was thought to have chemoattractant activity only for T cells. We recently demonstrated that the chemokine CXCL12 could serve as a bidirectional cue, attracting T cells at low concentrations and repelling at high concentrations in vitro and in vivo.15 Certain murine models of melanoma and ovarian cancer indicate that the effect of CXCL12 on the tumor immune response is dose dependent.16 Low levels of CXCL12 expression result in infiltration of the tumor by T cells, a delay in tumor growth, and the development of a long-lived tumor-specific T cell-mediated immune response. In contrast, high levels of CXCL12 expression result in reduced infiltration of the tumor by tumor antigen-specific T cells17 and the suppression of antitumor immune responses via various mechanisms including the intratumoral accumulation of FoxP3+ suppressor T cells.18,19

We hypothesized that secretion of CXCL12 by cervical tumors is correlated with progression of the tumor and loss of immune control. To explore this hypothesis we first examined whether clinicopathological progression of cervical cancer is associated with an increase in CXCL12 expression and a concomitant reduction in immune cell infiltration into the tumor as the tumor progresses from a preinvasive to an invasive state. CXCL12 expression and T-cell infiltration were studied in tissue sections of normal cervix, preinvasive, microinvasive, and invasive squamous cervical cancer, and preinvasive and invasive adenocarcinoma using immunohistochemical (IHC) staining methods in concert with quantitative digital image analysis, Western blotting, and enzyme-linked immunosorbent assay (ELISA). Because HPV is associated with cervical cancer, we also examined the relationship between HPV status and CXCL12 expression in normal, preinvasive, and invasive cervical cancer. In this way we demonstrated that CXCL12 is a robust biomarker for clinical progression that is correlated both with HPV infection and the accumulation of FoxP3+ T cells in the tumor.

Materials and Methods

Selection of Cases for Tissue Microarray

Cases of preinvasive, microinvasive, or invasive squamous cervical cancer and cases of preinvasive or invasive adenocarcinoma were selected from the computerized patient records of Hammersmith Hospital. The corresponding H&E-stained histological slides were retrieved from the hospital file and reviewed by a consultant pathologist (R.D.). On the basis of microscopic review, the slides were then assigned to one of seven categories, each reflecting a histopathologically defined stage in the development of cervical cancer, ie, CIN1 (with or without HPV changes), CIN2, CIN3, microinvasive squamous carcinoma (MI), invasive squamous cell carcinoma (INV.SCC), adenocarcinoma in situ (AIS), or invasive adenocarcinoma (ADENOCA). The paraffin-embedded, formalin-fixed tissue blocks were then retrieved from the histopathology archive together with control blocks containing normal ectocervical or endocervical tissue. A total of 107 cases were identified as suitable for tissue microarray (TMA). These included 83 positive cases comprising CIN1+/− HPV (n = 15), CIN2 (n = 15), CIN3 (n = 20), MI (n = 7), INV.SCC (n = 15), AIS (n = 6), ADENOCA (n = 5), and 24 control cases, ie, normal ectocervix/squamous epithelium, and endocervix/glandular epithelium (n = 15). From the 83 positive cases by TMA, 348 tissue cores were prepared, comprising CIN1+/− HPV (n = 30), CIN2 (n = 43), CIN3 (n = 99), MI (n = 27), INV.SCC (n = 116), AIS (n = 12), and ADENOCA (n = 21). In addition, 14 cores from normal ectocervix and 21 cores from normal endocervix were prepared by TMA. This study was performed with ethics approval from the Hammersmith and Queen Charlotte’s & Chelsea Research Ethics Committee (reference 05/ Q0406/170).

Manual TMA

A manual tissue microarrayer (Beecher Instruments Inc., Sun Prairie, WI) was used to prepare sections for histological and immunohistochemical staining and in situ hybridization. For this study, seven blocks were prepared (labeled B to H) and from 50 to 100 tissue cores were arrayed in each block. Representative areas were first selected from each donor tumor block by premarking sampling regions from the corresponding H&E-stained histological slide. The premarked H&E slide and the corresponding tissue block were then aligned, and the slide was moved out so that tissue cores of about 1.0 mm in diameter could be punched using the arraying device and then arrayed into a recipient block. Serial 2-μm sections of the resultant TMA blocks were cut using a rotary microtome (AS 325, Shandon, Runcorn, UK). The blade was wiped with 70% ethanol to avoid cross-contamination. For every fifth level, a section was stained with H&E stain and kept for comparison with the IHC sections.

IHC Staining Methods

Two methods of IHC staining were used for this study—the avidin-labeled biotin complex (ABC) IHC staining method and a double immunofluorescence (IF) staining method. The ABC method was used to investigate the distribution of a single specific antigen in each section, eg, CXCL12 (R&D Systems, Minneapolis, MN) or T-cell marker (CD3, CD4, CD8, or FoxP3) (Novocastra, Newcastle, UK, and eBioscience, Hatfield, UK); the IF method was used to study the distribution of both CXCL12 and CD3 antigens in the same section. In ABC IHC staining, after dewaxing and rehydration of sections, antigen retrieval was performed using the microwave method in 0.01 mol/l buffered sodium citrate solution (pH 6.0). After inactivation of endogenous peroxidase with 0.3% hydrogen peroxide in methanol (for the peroxidase detection method only) and blocking of nonspecific background staining with normal goat serum, sections were incubated with the primary antibody. After an overnight incubation, sections were then incubated with the appropriate biotinylated secondary antibody followed by a peroxidase- or alkaline phosphatase-labeled avidin biotin complex (Vector Laboratories, Peterborough, UK). The enzyme signal was detected with diaminobenzidine/H2O2 for peroxidase (Vector Laboratories) or Fast Red with naphthol phosphate substrate for alkaline phosphatase (AP) (Biogenex, Longfield, Kent, UK). Harris hematoxylin was used as a counterstain for peroxidase sections, whereas Meyer’s hemalum was used for AP sections. Sections were mounted using DPX (peroxidase) or aqueous mountant (AP) for examination with the light microscope. In double IF staining, sections were pretreated as described for the IHC method and then were incubated with a mixture of the two primary antibodies followed by application of a mixture of two suitable secondary antibodies with different fluorescent labels, ie, CY3 and Alexa Fluor. After incubation in the dark with double fluorescence-linked secondary antibodies, sections were mounted with SlowFade Gold antifade reagent with 4′-6-diamidino-2-phenylindole (Invitrogen, Paisley, UK). PBS was used in place of the primary antibody as a negative antibody control. Nuclear staining (blue) was performed with 4′-6-diamidino-2-phenylindole for the double IF staining method. Thymus sections were used as positive tissue controls. Stained slides were then viewed using the Nikon Eclipse 80i/Olympus BX40 fluorescence microscope. Red-colored images of CXCL12 expression and green-colored images of CD3 T cells were visualized using green and blue filters, respectively, and assessed manually in the first instance. Thereafter they were analyzed quantitatively using Image-Pro Plus (6.0 version).

Imaging and Quantitation of IF TMA Sections Using Image-Pro Plus

All images of IF- and H&E-stained sections were taken using a digital camera (Nikon DXM 1200F, Nikon, Kingston upon Thames, Surrey, UK) and then saved as JPEG or TIFF files. From the IF-stained sections, an average of five images per TMA core were taken at ×400 magnification. Each image had approximately the same area of squamous or glandular epithelium together with stroma directly underneath it. Expression of CXCL12 antigens and CD3+ T-cell count were analyzed using the image analysis program Image-Pro Plus. Two parameters were chosen to quantify CXCL12 expression from the IF-stained tissue sections, namely, CXCL12 intensity and CXCL12 percent area. Both parameters were evaluated using a prerecorded “macro” containing a set of commands that had been designed by the user (F.J.) to allow repeated performance of a certain set of measurement on the image to be analyzed. For our study, a set of commands called Macro 1 has been designed for the purpose of evaluating the intensity and percent area for each fluorescent color channel: ie, CY3, red (550 nm); Alexa Fluor, green (488 nm); and DAPI, blue (365 nm). The CD3+ T cells that were conjugated to the Alexa Fluor fluorochrome were identified on the images after elimination of the other two fluorescent color channels (red and blue) and counted manually at ×400 magnification in five randomly selected fields per tissue core. The CD3+ cell counts were recorded on Image PRO Plus. The data obtained from Image PRO Plus were used to provide the mean values of intensity and percent area of CXCL12 as well as the mean values of CD3 T-cell count. The mean value assigned to each tissue core was obtained from the mean values of all of the images analyzed for the tissue core. The results of digital evaluation of CXCL12 in the IF-stained sections were subsequently compared for intensity and localization of CXCL12 in ABC-stained sections.

Western Blotting and ELISA Measurement of Tissue CXCL12 Concentrations

Protein lysates were extracted from fresh snap-frozen human cervical tissues that had been obtained directly from colposcopic biopsies or the tissue bank at the Hammersmith Hospital: normal cervix (n = 9), CIN1 (n = 3), CIN2-3 (n = 15), and INV.SCC (n = 6). A frozen section was prepared from each sample and stained using H&E to ascertain the presence of normal or abnormal cervical epithelium before tissue lysis. In most of the samples a small amount of stroma was observed beneath the epithelial layer. Human recombinant CXCL12 (PeproTech, London, UK) was run at 100 ng/lane for every blot to serve as a positive control.

Protein assay of these lysates was then performed using a Bio-Rad protein assay kit. Proteins were separated in 12% SDS-polyacrylamide gel electrophoresis (Novex Mini Gel, Invitrogen), transferred onto a nitrocellulose membrane, and stained overnight with anti-SDF-1 antibody (R&D Systems), followed by detection using horseradish peroxidase-labeled anti-mouse antibody (Santa Cruz Biotechnology Inc., Santa Cruz, CA) and an enhanced chemoluminescence kit (Amersham Biosciences, Piscataway, NJ). The quantity of human CXCL12 present in tissue homogenates prepared from snap-frozen tissue obtained from the same source as above was determined by a specific ELISA Quantikine human CXCL12 immunoassay (R&D Systems). The specimens selected for ELISA comprised normal thymus (n = 1), normal cervix (n = 5), and INV.SCC (n = 5). Each tissue sample (100 mg) was cut into small pieces using forceps, scissors, and a disposable scalpel and put in a cryotube of 1.8 ml adding 1 to 1.5 ml of Dulbecco’s modified Eagle’s medium. The tissue pounded inside the cryotube was cut down using a disposable scalpel, and the tubes were vortexed to mix thoroughly. Cells were lysed by the freeze-thaw method (thaw at 37°C and immediately return to −80°C for three cycles), transferred into a 3-ml syringe fitted with a 0.22-μm filter (Millex-Ha Filter Unit 0.45 μm filter unit, Millipore Corporation, Carrightwohill, Cork, Ireland) and pushed through the filter to remove large pieces of cell debris. Extracts were analyzed using the human CXCL12 Quantikine ELISA kit and the absorbance of each well determined using a microplate reader at 450 nm (Spectra Fluor microplate reader, Tecan, Theale, Reading, UK). The absolute CXCL12 concentration was determined using a standard curve for CXCL12 according to the manufacturer’s recommendations.

HPV in Situ Hybridization

TMA tissue sections from 81 of the original 107 cases (Table 1) were used for this study. The cases included CIN1 (n = 10), CIN2 (n = 10), CIN3 (n = 18), MI (n = 5), INV.SCC (n = 6), AIS (n = 5), ADENOCA (n = 5), normal ectocervix (n = 6), and normal endocervix (n = 8). HPV DNA was detected using the Bond system (Leica Microsystems, Milton Keynes, Bucks, UK), a fully automated system for ISH staining. Two biotin-conjugated DNA probes (Vision BioSystems, Newcastle upon Tyne, UK) were used in the Bond system for ISH: one probe (designated HPV6,11) detected the low-risk HPV types 6 and 11; and the other probe (designated HPV16+) detected the high-risk HPV types 16, 18, 31, 33, and 51. Both DNA-negative and DNA-positive controls were processed at the same time as the cases. The protocol for the HPV DNA ISH process using a Bond polymer refine detection kit (Bond system, Leica Microsystems) was followed according to the manufacturer’s instructions. Control slides were checked under the microscope before the sections were dehydrated for permanent mounting. Positive cases were identified by the presence of dark brown diaminobenzidine/H2O2-stained nuclei. The HPV status of the TMA sections was correlated with the CXCL12% area on the serial sections from the same specimen as determined by IHC and Image PRO Plus as described earlier.

Table 1.

HPV Typing, CXCL12 % Area, and FoxP3† Cell Count

Group and HPV probes Normal ectocervix (n = 6) CIN1 (n = 10) CIN2/3 (n = 33)* INV.SCC (n = 14) Normal endocervix (n = 8) AIS (n = 5) ADENOCA (n = 5) CXCL12 % area FoxP3† cell count Mean ± SD
Group 1: 6,11 negative; 16+ negative 5 4 5 2 7 2 1 30.53 ± 19.22 (n = 26) 18.2 ± 22.4 (n = 26)
Group 2: 6,11 positive; 16+ negative 0 0 2 1 0 0 0 62.14 ± 9.33 (n = 3) 48 ± 5.7 (n = 2)
Group 3: 6,11 negative; 16+ positive 1 4 7 5 1 3 3 41.98 ± 17.30 (n = 24) 20 ± 13.7 (n = 22)
Group 4: 6,11 positive; 16+ positive 0 2 19 6 0 0 1 60.75 ± 16.64 (n = 28) 28.9 ± 20.5 (n = 23)
Open in a new tab

Low- and high-risk HPV typing of the study groups. These data show an increasing percentage of high risk HPV16+-positive cases as the neoplastic lesions progress: normal ectocervix = 17%, CIN1 = 60%, CIN2/3 = 79%, INV.SCC = 79%, normal endocervix = 13%, AIS = 60%, ADENOCA = 80%. Table 1 also shows HPV status versus the CXCL12 % area (mean ± SD) and FoxP3† cell counts (mean ± SD) for pooled squamous and glandular neoplastic lesions (including normal ectocervix and endocervix). A significant increase in CXCL12 % area between HPV-negative lesions (group 1) and other HPV-positive groups was detected (group 2, P = 0.0099; group 3, P = 0.0321; and group 4, P < 0.0001) (unpaired two-sample t-test with unequal variances and unequal sample size). A significant difference in CXCL12 % area was also detected between HPV-positive lesions in groups 2 and 3 (P = 0.0002). A study of HPV status versus FoxP3†-positive cell count revealed an increased presence of FoxP3†-positive cells in groups 2, 3, and 4 containing HPV-positive lesions compared with group 1 containing HPV-negative lesions. These differences did not reach statistical significance (group 1 versus group 2, P = 0.0757; versus group 3, P = 0.7441; and versus group 4, P = 0.0885). 

*

Includes microinvasive squamous carcinoma (MI). 

Staining of ThinPrep Cervical Smears for CXCL12 Expression

Slides from 12 cervical scrape liquid-based cytology specimens that had previously been reported as containing severely dyskaryotic cells (seven specimens) or reported as negative for malignancy (five specimens) were prepared by the ThinPrep method and fixed in alcohol. Slides were immersed in 0.01 mol/L buffered sodium citrate solution (pH 6.0) and microwaved for antigen retrieval. Endogenous peroxidase was inactivated with 0.3% hydrogen peroxide in methanol and nonspecific background staining was blocked by treatment with normal goat serum. Slides were then incubated with anti-human CXCL12 antibody (clone 79018: 1 in 100, R&D Systems). PBS replaced the primary antibody as a negative control. The second incubation was in biotinylated goat anti-mouse Ig (Vector Laboratories) followed by a peroxidase-labeled avidin/biotin complex. Diaminobenzidine/H2O2 was used for signal detection with Harris hematoxylin as nuclear counterstain. Slides were mounted in DPX.

Statistical Analysis

Specimens were categorized according to histological status (normal ectocervix, CIN1, CIN2 and CIN3, MI, and INV.SCC, and normal endocervix, AIS, and ADENOCA) and mean values for CXCL12% area and T-cell counts were determined for each category. Differences between categories were examined using analysis of variance after testing for normal distribution; log value used where appropriate (namely, CD4, CD8, and FoxP3 positivity). A linear contrast test was used to determine whether there was a trend for increasing severity across the categories, taking into account unequal variance. Linear regression was used to examine the correlation between variables, eg, CXCL12% area and FoxP3† T-cell count. A significance level of 5% was used in all of the tests. Statistical analysis was performed using SPSS software (version 15.0). To determine whether there was any significant difference between groups in terms of HPV status, the unpaired two sample t-test, with unequal variances and unequal sample size was used.

Results

Qualitative Analysis of ABC (AP) IHC Staining of TMA Sections for CXCL12

A total of 383 TMA tissue cores (prepared from 83 positive cases and 24 controls) stained by the ABC (AP) method using Fast Red chromogen were assessed manually by light microscopy for CXCL12 expression. The antigen was expressed in the epithelium and stroma of neoplastic squamous and glandular lesions. The location of the epithelial staining was clearly intracellular and well defined, whereas stromal staining was generally extracellular and diffuse. An increase in intensity and distribution of CXCL12 was noted as the lesions progressed (Figure 1, A–F and M–O). In squamous neoplastic lesions (Figure 1, A–F) the intensity of CXCL12 was most marked in CIN3, MI, and INV.SCC, and in CIN3, the antigen was present throughout the full thickness of the epithelium. In CIN1 and CIN2, antigen expression was confined to the basal layers of the epithelium. Normal squamous epithelium of the ectocervix was negative. In glandular lesions, the neoplastic epithelial cells of AIS showed intense CXCL12 staining of their luminal border; whereas the neoplastic cells of ADENOCA exhibited strong positive intracytoplasmic and luminal border staining for CXCL12 (Figure 1, N and O). Normal glandular epithelium of the endocervix was consistently negative for CXCL12 (Figure 1M).

Figure 1.

Figure 1

Open in a new tab

TMA sections of normal ectocervix, normal endocervix, and neoplastic cervical squamous and glandular lesions stained by ABC IHC AP Fast Red labeling method and double IF staining. Nuclei (blue) in G–L and P–R are stained with 4′−6-diamidino-2-phenylindole; nuclei (blue) in A–F and M–O are stained with Meyer’s hemalum. A–F: Sections of normal ectocervix, CIN1, CIN2, CIN3, MI, and INV.SCC stained for CXCL12 (positive epithelium in red). Normal ectocervical epithelium shows no staining for CXCL12. The area of the epithelium positive for CXCL12 is confined to the basal layers in CIN1 and CIN2 lesions, and it progresses in CIN3, MI, and INV.SCC in which it is increased and present throughout the full thickness of the neoplastic epithelium, as shown by double-headed arrows. G–L: Serial TMA sections corresponding to A–F stained with double IF for CXCL12 (red) and CD3 (green). CD3+ T cells (green rim staining) can be seen in the stroma of K and L (MI and INV.SCC). Green staining for CD3+ is more clearly shown after the red and blue staining of CXCL12 and nuclei, respectively, are eliminated (K, right, CD3+ cells indicated by arrows). M–O: Images of TMA sections of normal endocervix, AIS, and ADENOCA stained for CXCL12 using an ABC (AP) Fast Red labeling method. There is increasing intensity of CXCL12 (red staining) in abnormal glandular epithelial cells as the lesion progressed from AIS (N) to ADENOCA (O). Normal glandular epithelium was consistently negative (M). Corresponding sequential serial TMA sections (P–R) confirmed a progressive increase in CXCL12 staining with scanty CD3 T cell infiltration (green staining) through AIS (Q) and ADENOCA lesions (R). Q, right: Green staining for CD3+ after elimination of the red and blue staining of CXCL12 and nuclei, respectively (CD3+ cells are indicated by arrows). Scale bar = 50 μm. Asterisk denotes staining with anti-CD3.

Quantitative Analysis of CXCL12 and CD3 in IF-Stained TMA Sections

A total of 383 TMA tissue cores (prepared from 83 positive cases and 24 controls) stained by the double IF staining method were analyzed quantitatively for CXCL12% area and CD3 count using the image analysis programs described above (Figure 1, G–L and P–R). A significant increase in CXCL12% area with increasing severity of the cervical neoplastic lesions was observed for both squamous (P < 0.001) and glandular neoplastic lesions (P = 0.005) (Figure 2, A and C). A significant increase in CD3+ T-cell counts with increasing severity of the neoplastic lesions was also seen in squamous cervical neoplasia (P < 0.001) (Figure 2B) although this increase was not observed in glandular cervical neoplasia (Figure 2D). A significant correlation between CXCL12 percent area and CD3 cell count for was also noted for all squamous categories (P = 0.003; r2 0.119) (Figure 2E) but not for the glandular lesions (Figure 2F). The distribution and intensity of CXCL12 staining in squamous and glandular lesion as determined by the ABC (AP) staining method was almost identical with the distribution and intensity of CXCL12 demonstrated by IF staining; both showed an increase in distribution and intensity as the cervical lesions progressed through CIN1, CIN2, and CIN3 to invasive cancer (Figure 1).

Figure 2.

Figure 2

Open in a new tab

Data obtained as a result of quantitative image analysis of IF-stained TMA sections of normal ectocervix and endocervix and neoplastic squamous and glandular lesions with respect to CXCL12 percent area and CD3 cell count. The data are presented as boxplots (A–D) and as correlation graphs (E and F). A: Boxplots of CXCL12% area showing an increase in CXCL12% area with increasing severity of the squamous lesion (analysis of variance test P < 0.001 and linear contrast test P < 0.001; assuming unequal variances). B: Boxplots of CD3 cell count showing an increase in CD3 cell count with increasing severity of the squamous neoplastic lesion (analysis of variance test P = 0.001 and linear contrast test P < 0.001; assuming unequal variances). C: Boxplots of CXCL12% area in normal endocervix and glandular neoplastic lesions showing an increase in CXCL12% area with increasing severity of the glandular neoplastic lesions (analysis of variance test P < 0.001 and linear contrast test P = 0.005; assuming unequal variances). D: Boxplots of CD3 cell count in normal endocervix and glandular neoplastic lesions showing no significant difference in CD3 cell count with increasing severity of the neoplastic lesion (analysis of variance test P = 0.884 and linear contrast test P = 0.603; assuming unequal variances). E: Correlation of CXCL12% area with CD3 cell count for normal ectocervix and all neoplastic squamous lesions is statistically significant (CIN1, CIN2, CIN3, MI, and INV. SCC) (Pearson correlation test P < 0.005, r2 = 0.119). F: Correlation of CXCL12% area with CD3 cell count for normal endocervix and all glandular neoplastic lesions was not found to be significant (AIS and ADENOCA) (Pearson correlation test P = 0.36, r2 = 0.05).

Quantitative Analysis of ABC (Peroxidase)-Stained TMA Sections for CD4, CD8, and FoxP3

In view of the above findings we proceeded to examine subpopulations of T cells within the TMA tissue cores. CD4 and CD8 antigens were observed to be membrane bound, whereas the FoxP3 antigen was expressed intracellularly. Small mononuclear cells positive for these antigens were noted at all stages of development of cervical cancer and were predominantly located in the stroma (Figure 3, A–F). The antigen-positive cells were counted using a manual counting method (positive cells were counted in five randomly selected fields viewed at ×400 magnification), and the mean value was obtained for each category. A slight upward trend in mean CD4 and CD8 cell count (P = 0.01 and P = 0.005, respectively) was observed with increasing severity of the squamous neoplastic lesions (Figure 4, A and B), but this was not present in glandular neoplastic lesions (Figure 4, C and D). In contrast, there was a marked increase in the number of infiltrating FoxP3+ T cells as the squamous and glandular neoplastic lesions progressed from preinvasive to invasive carcinoma (P < 0.001 and P = 0.001, respectively) (Figure 4, E and F). Correlation graphs of FoxP3 cell counts and CXCL12% area showed a highly significant correlation for both squamous (P < 0.001; r2 = 0.191) and glandular (P < 0.001; r2 = 0.593) categories (Figure 4, G and H). No significant correlation was detected for CXCL12% area and CD4+ or CD8+ T-cell subpopulations (data not shown).

Figure 3.

Figure 3

Open in a new tab

Representative TMA sections of normal and abnormal ectocervix (ABC immunoperoxidase staining) showing infiltration of CD3+, CD8+, and FoxP3+ lymphocytes in normal ectocervix (A, C, and E) and invasive squamous carcinoma (B, D, and F). Cell surface staining of CD4+ and CD8+ T cells and intracellular staining of FoxP3+ T cells is shown. Scanty infiltration of positive cells is shown for normal ectocervix compared with heavier infiltration of these cell subpopulations in neoplastic epithelium. Scale bar = 50 μm.

Figure 4.

Figure 4

Open in a new tab

Quantitative assessment of CD4+, CD8+, and FoxP3+ cells in normal and dysplastic cervix. Boxplots of CD4+ (A) or CD8+ (B) in normal ectocervix and grades of dysplastic lesions including CIN 1, CIN2, CIN 3, MI, and INV.SCC. The CD4 cell count increase correlates with the severity of the neoplastic lesion (analysis of variance test P < 0.005 and linear contrast test P = 0.01; assuming unequal variances). An increase in CD8 T-cell infiltrate with MI and INV.SCC (analysis of variance test P = 0.005 and linear contrast test P = 0.005; assuming unequal variances). Boxplots of CD4+ (C) or CD8+ (D) in normal endocervix and grades of dysplastic lesions including AIS and ADENOCA. CD4 cell count increase correlates with the severity of the neoplastic lesion (analysis of variance test P < 0.005 and linear contrast test P = 0.01; assuming unequal variances). Increase of CD8 T-cell infiltrate with MI and INV.SCC (analysis of variance test P = 0.005 and linear contrast test P = 0.005; assuming unequal variances). Boxplots of FoxP3+ cells in ectocervix and associated dysplastic lesions and endocervix and AIS/ADENOCA are shown in E and F, respectively. An increase in FoxP3 cell count with increasing severity of the squamous neoplastic lesions is noted (analysis of variance test P < 0.001 and linear contrast test P < 0.001; assuming unequal variances). An increase in FoxP3 cell count with increasing severity of the glandular neoplastic lesions is also detected (analysis of variance test P = 0.006 and linear contrast test P = 0.001; assuming unequal variances). Correlation graphs of CXCL12% area with FoxP3+ cell counts for normal ectocervix and endocervix with neoplastic squamous (G) and glandular lesions (H) show highly significant correlations between the two variables measured (Pearson correlation test P < 0.001, r2 = 0.191 and P < 0.001, r2 = 0.593, respectively).

Quantitation of CXCL12 Expression by Western Blotting and ELISA

To confirm the expression of CXCL12 by tissues examined in this study both Western blots and ELISAs were performed on tissues. An increasing percentage of cases showed a positive CXCL12 α band in the 14-kDa molecular weight range with progression of cervical neoplasia. All normal cervical tissue samples were negative. As shown by the representative blot in Figure 5A, CXCL12 was absent in normal ectocervical tissue (lanes 1 and 4) and normal endocervical tissue (lane 6) but was present in samples from preinvasive and invasive cervical carcinoma (CIN2, CIN3, and INV.SCC). In one sample of INV.SCC (lane 5) a band corresponding to CXCL12 was not detected. However, a repeat sample from the same specimen showed a positive band (lane 10). In general, samples of INV.SCC had stronger bands of CXCL12 expression (lanes 3 and 8) than preinvasive cervical neoplasia samples (lanes 2 and 7). ELISA showed that the amount of CXCL12 in INV.SCC was three to four times greater than the amount in normal cervix. The mean value for CXCL12 for normal cervix (five cases) was 57.6 ± 9.4 ng/ml (range, 28.5 to 85.6 ng/ml) and the mean value for CXCL12 for INV.SCC (five cases) was 232 ± 51.4 ng/ml (range, 114 to 356 ng/ml) (P = 0.0018, Student’s t-test). In this way we demonstrated that CXCL12 was detectable by ELISA and on Western blots from dysplastic cervical lesions and that the chemokine was expressed at significantly higher levels in INV.SCC than in normal cervix.

Figure 5.

Figure 5

Open in a new tab

A: Western blot analysis of CXCL12 expression in cervical neoplastic samples from squamous and glandular epithelia. Samples were separated by SDS-polyacrylamide gel electrophoresis in lanes as indicated in the key box. Positive bands corresponding to CXCL12 are observed in samples from CIN3, INV.SCC and CIN2 but are absent in samples from normal squamous epithelium and normal glandular epithelium. Lanes labeled as in inserted key. Positive bands for β-actin are shown for all tissue-derived samples in lanes 1–8 and lane 10. B: Representative examples of in situ hybridization of TMA sections using two Bond probes (HPV6,11 and HPV16+). CIN1 showing positive nuclear staining (brown) for HPV6,11. C: CIN3 showing positive nuclear staining for HPV16+. D: Boxplots of CXCL12% area of HPV16+ probe-negative (neg) and HPV16+ probe-positive (pos) cases for pooled samples of normal ectocervical and squamous neoplastic lesions are shown. An independent sample t-test shows a significant difference between the two groups (P < 0.05, assuming equal variances). E: A significant difference is demonstrated in boxplots of CXCL12% area of HPV16+ probe-negative (neg) and HPV16+ probe-positive (pos) cases for pooled samples of normal endocervical and glandular neoplastic lesions (P < 0.005). The source of the data used for the boxplots in D and E were derived from Table 1. Scale bar = 50 μm.

Results of HPV in Situ Hybridization

Positive nuclear staining was seen mainly in the surface layers of the squamous epithelium (Figure 5, B and C) and characteristically was associated with koilocytosis. HPV typing was performed on tissue samples (TMA) from 81 cases where there was an adequate amount of tissue as described above. The prevalence of HPV antigen in the squamous and glandular lesions was as follows: 31 (38%) and 52 (64%) of the 81 TMA sections were positive for HPV6,11 and HPV16+ probes, respectively. Of normal cases 14% were positive for HPV16+ compared with 75% of the neoplastic cases. Table 1 reports HPV typing against histopathological staging of cervical cancer, CXCL12% area, and FoxP3 cell count. Progression of neoplastic lesions was associated with an increasing percentage of HPV16+-positive cases (normal ectocervix = 17%, CIN1 = 60%, CIN2/3 = 79%, INV.SCC = 79%, normal endocervix = 13%, AIS = 60%, and ADENOCA = 80%). CXCL12% area was significantly increased in cases that were positive for both HPV6,11 and HPV 16+ probes compared with cases that were negative for both probes (P < 0.0001). Furthermore, cases that were positive for only HPV6,11 demonstrated a higher CXCL12% area than those that were HPV16+ positive alone (P = 0.0002). A higher mean number of FoxP3+ cells were detected in cases that were positive for both HPV6,11 and HPV16+ (28.9) and in cases that were positive only for HPV6,11 (48.0) in comparison with HPV-negative samples (18.2), but these associations did not reach statistical significance possibly because of low sample number (18.2 versus 48, P = 0.0885 and 0.0757, respectively). Comparison of CXCL12 percent area in HPV16+ probe-positive and HPV16+ probe-negative cases showed a significant difference between the two groups for both pooled samples of normal ectocervix and squamous neoplastic lesions (P = 0.05) and pooled samples of normal endocervix and glandular neoplastic lesions (P = 0.001) (Figure 5, D and E). Significantly higher levels of CXCL12 staining were seen in all HPV-positive samples compared with all HPV-negative samples in this context (P < 0.0001).

Results of ABC IHC Staining of ThinPrep Smears for CXCL12

The slides were screened for CXCL12 localization as evidenced by brown staining of the cytoplasm of the epithelial cells. Intense brown staining was noted in the cytoplasm of dyskaryotic cells in the seven specimens known to be positive for malignancy (Figure 6, C–E) whereas no staining was seen in the normal mature squames or endocervical cell clusters in either the five negative or seven positive cases (Figure 6, A and B). Faint brown staining of very occasional discrete immature metaplastic cells was observed in three specimens, but the benign nature of the cells was obvious from their morphology.

Figure 6.

Figure 6

Open in a new tab

Representative samples prepared from ThinPrep slides of normal and abnormal cervices (Pap tests) and stained for CXCL12 using the ABC IHC immunoperoxidase staining method. Positively stained dyskaryotic (abnormal) cells appear brown. A: Normal squamous cells in a ThinPrep slide from a Pap test that was reported negative for malignancy. Note the absence of brown stain in the cytoplasm and small regular nuclei of the squamous epithelial cells (×200). B: Normal endocervical cells in the same ThinPrep slide. Note the absence of brown stain and the honeycomb pattern of the cells in the cluster, which have small regular rounded nuclei (×400). C: Four severely dyskaryotic (high-grade squamous intraepithelial lesions/CIN3) cells in a ThinPrep slide from a Pap test specimen that was reported as positive for malignancy. Note positive CXCL12 staining (dark brown) cytoplasm of the dyskaryotic cells. Normal squames in the same slide are negative for CXCL12 (×200). D: Two large dyskaryotic cells from same specimen as C showing positive CXCL12 staining. The large irregular nuclei (blue) characteristic of dyskaryotic cells are clearly seen (×400). E: Cluster of severely dyskaryotic cells in a ThinPrep slide from the same specimen. Note the strong positive CXCL12 cytoplasmic staining (brown) and enlarged irregular nucleus (blue) in each abnormal cell. The few normal squames in the picture are negative for CXCL12 (×400).

The distinction between dense CXCL12 staining of the cytoplasm dyskaryotic cells from positive cases versus faint or absent staining in normal epithelial or metaplastic cells was robust for all samples in this study.

Discussion

This study has shown that there was a significant increase in the expression of the chemokine CXCL12 as measured by IHC and ELISA in cervical epithelium as the neoplastic lesions progressed from preinvasive to invasive cancer. Our study also showed that CXCL12 was not expressed by normal cervical squamous or glandular epithelium. The difference in the intensity of staining between the normal cervical epithelium and low- and high-grade precancerous lesions and invasive cancer was verified by Western blot. These findings complement previous studies9,20,21 in which expression of CXCR4 in tissue sections from patients with clinically invasive cervical cancer was investigated with respect to the potential for metastatic spread of tumor. They found that CXCR4 expression could be detected in most primary tumors, and strong expression in the primary tumor was associated with metastatic spread to pelvic nodes, whereas normal cervical epithelium was negative for CXCR4. These authors also investigated the effect of CXCL12 and its receptor in human cervical cancer cell lines and showed that CXCL12 is a chemoattractant for cervical carcinoma cells and may promote proliferation. Majka et al9 concluded that CXCR4 expression is strongly associated with the malignant phenotype.

The focus of our study is the role that CXCL12 might play in dysregulating T-cell infiltration into cervical cancer as it progresses from preinvasive to invasive disease. CXCL12 has the potential to modulate efficacious antitumor responses in view of its properties as a powerful chemoattractant for T and pre-B lymphocytes and dendritic cells to the tumor site.16 However, the antitumor activity of CXCL12 is tightly regulated by local levels of CXCL12. Dunussi-Joannopoulos provided experimental evidence that low levels of CXCL12 are associated with antitumor response, whereas high levels may inhibit T-cell attraction.22 Our study showed that tumor progression as measured by clinicopathological stage of cervical neoplasia, was associated with CD3+ T-cell infiltration into the tumor. Furthermore, the most significant association was between advanced stage of tumor progression and high levels of FoxP3+ T cells in the tumor. Of particular significance was the correlation between CXCL12 expression and FoxP3 for both squamous and glandular cervical neoplastic lesions. Our results suggest that high levels of CXCL12 lead to retention or accumulation of FoxP3+ T cells in the progressing cervical cancer. A weak correlation was demonstrated between CD4+ and CD8+ T-cell infiltration and tumor progression, and no correlation was reported between CXCL12% area and other T-cell subpopulations, thus reinforcing the role of FoxP3+ T cells in this context and confirming a previously reported complex relationship between the T-cell infiltration and tumor progression.23

The role of the axis CXCL12/CXCR4 in tumor spread and progression has already been demonstrated in various cancer models.24,25,26 Besides, CXCR4 is the most constantly expressed receptor in cancer,10 and its level in tumor tissue has been correlated with clinical outcomes.27,28 In addition, CXCL12 expression has been directly correlated with known determinants of immune dysregulation, such as HPV infection itself and regulatory T cell infiltration.18,29 A direct relationship between the HPV infection and the onset of an immune suppressive environment has been suggested by a prospective study of HPV16-related lesions, in which regulatory T cell frequencies were significantly increased in women who had persistent HPV16 infection and HPV16-specific interleukin-2 producing T-helper cells.19 Furthermore, it has also been observed that, in patients with HPV6- and HPV11-derived genital condylomata, FoxP3+ regulatory T cells with a suppressive cytokine milieu accumulated in large warts.30 The WHIM syndrome (wart, hypogammaglobulinemia, infection, and myelokathexis syndrome), which is a syndrome characterized by disseminated HPV-induced warts, leukopenia due to bone marrow cell retention and other immune dysfunctions, has been correlated to a defect in CXCR4 receptor and CXCL12 signal transmission, thus confirming a major role of the cytokine in contributing to immune cell dysregulation and T cell-impaired trafficking in the context of HPV-derived lesions.31

Our study confirmed a significant relationship between HPV typing and CXCL12% area in both squamous and glandular neoplastic lesions, showing an increase of CXCL12% area in cases that were positive for both HPV6,11 and HPV16+ probes compared with cases that were negative for both probes. Interestingly, cases that were positive for only HPV6,11 demonstrated a higher CXCL12% area than those that were only HPV16+ positive. Despite the higher values reported between HPV status and FoxP3+ cell count in cases that were positive for both HPV6,11 and HPV16+ and in cases that were positive only for HPV6,11, the results did not reach significance because of low numbers of samples in certain groups due to a lack of available tissue. Further studies are clearly needed to confirm these provocative findings. In our study, the link between HPV, CXCL12, FoxP3+ T infiltrating cells, and tumor progression has been confirmed through multiple assays, histopathologically and immunologically, and proven both in squamous and glandular neoplastic cervical lesions. Thus, CXCL12 may be a good candidate as a marker of cervical cancer clinical progression through the well established multistep histological patterns of preinvasive, microinvasive, and invasive cancer.

Several large studies have shown that virtually all squamous cell carcinomas of the cervix and most of the adenocarcinomas are HPV-positive, and our results are consistent with these findings. More than 80% of the preinvasive and invasive lesions tested were positive for the most common high-risk HPV. Integration of viral DNA into the host genome is an essential step in the carcinogenic process and although integration can occur at a number of sites in the viral genome, integration in the region of the E2 gene causes loss of transcriptional control of E6 and E7 with subsequent loss of function of the p53 and the Rb pathways resulting in uncontrolled cell proliferation.32

Cervical cancer is the second most common cancer in women and is the cause of extensive morbidity and mortality worldwide. Invasive disease can be prevented by detection of the cancer at its earliest stages and the Pap test has proved to be a valuable method of screening for preinvasive disease, limited only by the fact that it is a very labor-intensive procedure and has a false-negative rate for high-grade neoplastic lesions of between 8 and 15%.33 A method of screening for cervical cancer that has the potential for full automation could be of value in refining the Pap test and expanding its use, especially in countries in which cervical cancer is rife, and cervical screening is most needed. Our preliminary research on IHC staining of ThinPrep slides for CXCL12 indicates that this chemokine is a potentially valuable marker that could be adapted to the automated analysis of cervical smears. In the preliminary histological and cytological studies presented here CXCL12 appears to be highly specific for dyskaryosis with clearly negative normal epithelium and a strong case can be made for further exploration of CXCL12 as a diagnostic or screening tool in clinical practice. Finally, the finding of statistically significant correlations between CXCL12 expression, FoxP3+ cell infiltration and HPV infection in the progression of cervical cancer provides insight into pathogenetic mechanisms by which a viral infection may contribute to dysplastic changes in cervical epithelial and glandular cells while initiating a process by which these cells evade the immune system as a result of effector immune cell suppression through the recruitment of suppressor T cells and the overexpression of an effector T-cell chemorepellent.

Acknowledgments

We thank Professor Kikkeri Naresh, Consultant at the Histopathology Department, Hammersmith Hospital, London, Haziq Jamil and Paul Bassett for their valuable advice and help with the statistics. We thank Dr. Jacob B. Poznansky for his help with coordinating the activities of the collaborating research teams.

Footnotes

Address reprint requests to Mark C. Poznansky, M.D., Ph.D., Associate Professor, Harvard Medical School, Infectious Diseases Medicine, Massachusetts General Hospital (East), 149 Building, 13th St., Charlestown, MA 02129. E-mail: mpoznansky@partners.org; or Dulcie V. Coleman, F.R.C.P., M.D., Professor, Imperial College, Department of Histopathology, Hammersmith Hospital, Ducane Road, London W12 ONNH, UK. E-mail: d.v.coleman@ic.ac.uk.

Supported by the U.S. Public Health Service (grant R01 AI49757), the Marsha Rivkin Center, and philanthropic funds (M.C.P. and E.R.), the Imperial College Trust Fund (D.V.C.), and the Government of Brunei, Darussalam (F.J.).

References

  1. Cannon MJ, Santin AD, O'Brien TJ. Immunological treatment of ovarian cancer. Curr Opin Obstet Gynecol. 2004;16:87–92. doi: 10.1097/00001703-200402000-00015. [DOI] [PubMed] [Google Scholar]
  2. Santin AD, Bellone S, Underwood LJ, O'Brien TJ, Ravaggi A, Pecorelli S, Cannon MJ. Novel immunotherapeutic strategies in gynecologic oncology. Dendritic cell-based immunotherapy for ovarian cancer. Minerva Ginecol. 2002;54:133–144. [PubMed] [Google Scholar]
  3. Waldmann TA. Effective cancer therapy through immunomodulation. Annu Rev Med. 2006;57:65–81. doi: 10.1146/annurev.med.56.082103.104549. [DOI] [PubMed] [Google Scholar]
  4. Muñoz N. Human papillomavirus and cancer: the epidemiological evidence. J Clin Virol. 2000;19:1–5. doi: 10.1016/s1386-6532(00)00125-6. [DOI] [PubMed] [Google Scholar]
  5. Walboomers JM, Jacobs MV, Manos MM, Bosch FX, Kummer JA, Shah KV, Snijders PJ, Peto J, Meijer CJ, Muñoz N. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol. 1999;189:12–19. doi: 10.1002/(SICI)1096-9896(199909)189:1<12::AID-PATH431>3.0.CO;2-F. [DOI] [PubMed] [Google Scholar]
  6. Castelli C, Maeurer MJ. The T-cell response in patients with cancer. Adv Cancer Res. 2002;86:149–193. doi: 10.1016/s0065-230x(02)86005-5. [DOI] [PubMed] [Google Scholar]
  7. Eiben GL, Velders MP, Kast WM. The cell-mediated immune response to human papillomavirus-induced cervical cancer: implications for immunotherapy. Adv Cancer Res. 2002;86:113–148. doi: 10.1016/s0065-230x(02)86004-3. [DOI] [PubMed] [Google Scholar]
  8. Sheu BC, Chang WC, Lin HH, Chow SN, Huang SC. Immune concept of human papillomaviruses and related antigens in local cancer milieu of human cervical neoplasia. J Obstet Gynaecol Res. 2007;33:103–113. doi: 10.1111/j.1447-0756.2007.00492.x. [DOI] [PubMed] [Google Scholar]
  9. Majka M, Drukala J, Lesko E, Wysoczynski M, Jenson AB, Ratajczak MZ. SDF-1 alone and in co-operation with HGF regulates biology of human cervical carcinoma cells. Folia Histochem Cytobiol. 2006;44:155–164. [PubMed] [Google Scholar]
  10. Balkwill F. The significance of cancer cell expression of the chemokine receptor CXCR4. Semin Cancer Biol. 2004;14:171–179. doi: 10.1016/j.semcancer.2003.10.003. [DOI] [PubMed] [Google Scholar]
  11. Kryczek I, Wei S, Keller E, Liu R, Zou W. Stroma-derived factor (SDF-1/CXCL12) and human tumor pathogenesis. Am J Physiol Cell Physiol. 2007;292:C987–C995. doi: 10.1152/ajpcell.00406.2006. [DOI] [PubMed] [Google Scholar]
  12. Baggiolini M. Chemokines and leukocyte traffic. Nature. 1998;392:565–568. doi: 10.1038/33340. [DOI] [PubMed] [Google Scholar]
  13. Singh S, Sadanandam A, Singh RK. Chemokines in tumor angiogenesis and metastasis. Cancer Metastasis Rev. 2007;26:453–467. doi: 10.1007/s10555-007-9068-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Poznansky MC, Olszak IT, Foxall R, Evans RH, Luster AD, Scadden DT. Active movement of T cells away from a chemokine. Nat Med. 2000;6:543–548. doi: 10.1038/75022. [DOI] [PubMed] [Google Scholar]
  15. Vianello F, Olszak IT, Poznansky MC. Fugetaxis: active movement of leukocytes away from a chemokinetic agent. J Mol Med. 2005;83:752–763. doi: 10.1007/s00109-005-0675-z. [DOI] [PubMed] [Google Scholar]
  16. Zou W, Machelon V, Coulomb-L'Hermin A, Borvak J, Nome F, Isaeva T, Wei S, Krzysiek R, Durand-Gasselin I, Gordon A, Pustilnik T, Curiel DT, Galanaud P, Capron F, Emilie D, Curiel TJ. Stromal-derived factor-1 in human tumors recruits and alters the function of plasmacytoid precursor dendritic cells. Nat Med. 2001;7:1339–1346. doi: 10.1038/nm1201-1339. [DOI] [PubMed] [Google Scholar]
  17. Vianello F, Papeta N, Chen T, Kraft P, White N, Hart WK, Kircher MF, Swart E, Rhee S, Palù G, Irimia D, Toner M, Weissleder R, Poznansky MC. Murine B16 melanomas expressing high levels of the chemokine stromal-derived factor-1/CXCL12 induce tumor-specific T cell chemorepulsion and escape from immune control. J Immunol. 2006;176:2902–2914. doi: 10.4049/jimmunol.176.5.2902. [DOI] [PubMed] [Google Scholar]
  18. Curiel TJ, Coukos G, Zou L, Alvarez X, Cheng P, Mottram P, Evdemon-Hogan M, Conejo-Garcia JR, Zhang L, Burow M, Zhu Y, Wei S, Kryczek I, Daniel B, Gordon A, Myers L, Lackner A, Disis ML, Knutson KL, Chen L, Zou W. Specific recruitment of regulatory T cells in ovarian carcinoma fosters immune privilege and predicts reduced survival. Nat Med. 2004;10:942–949. doi: 10.1038/nm1093. [DOI] [PubMed] [Google Scholar]
  19. Molling JW, de Gruijl TD, Glim J, Moreno M, Rozendaal L, Meijer CJ, van den Eertwegh AJ, Scheper RJ, von Blomberg ME, Bontkes HJ. CD4+CD25hi regulatory T-cell frequency correlates with persistence of human papillomavirus type 16 and T helper cell responses in patients with cervical intraepithelial neoplasia. Int J Cancer. 2007;121:1749–1755. doi: 10.1002/ijc.22894. [DOI] [PubMed] [Google Scholar]
  20. Yang YC, Lee ZY, Wu CC, Chen TC, Chang CL, Chen CP. CXCR4 expression is associated with pelvic lymph node metastasis in cervical adenocarcinoma. Int J Gynecol Cancer. 2007;17:676–686. doi: 10.1111/j.1525-1438.2007.00841.x. [DOI] [PubMed] [Google Scholar]
  21. Zhang JP, Lu WG, Ye F, Chen HZ, Zhou CY, Xie X. Study on CXCR4/SDF-1α axis in lymph node metastasis of cervical squamous cell carcinoma. Int J Gynecol Cancer. 2007;17:478–483. doi: 10.1111/j.1525-1438.2007.00786.x. [DOI] [PubMed] [Google Scholar]
  22. Dunussi-Joannopoulos K, Zuberek K, Runyon K. Efficacious immunomodulatory activity of the chemokine stromal cell-derived factor 1 (SDF-1): local secretion of SDF-1 at the tumor site serves as T-cell chemoattractant and mediates T-cell-dependent antitumor responses. Blood. 2002;100:1551–1558. [PubMed] [Google Scholar]
  23. Clarke B, Tinker AV, Lee CH, Subramanian S, van de Rijn M, Turbin D, Kalloger S, Han G, Ceballos K, Cadungog MG, Huntsman DG, Coukos G, Gilks CB. Intraepithelial T cells and prognosis in ovarian carcinoma: novel associations with stage, tumor type, and BRCA1 loss. Mod Pathol. 2009;22:393–402. doi: 10.1038/modpathol.2008.191. [DOI] [PubMed] [Google Scholar]
  24. Marchesi F, Monti P, Leone BE, Zerbi A, Vecchi A, Piemonti L, Mantovani A, Allavena P. Increased survival, proliferation, and migration in metastatic human pancreatic tumor cells expressing functional CXCR4. Cancer Res. 2004;64:8420–8427. doi: 10.1158/0008-5472.CAN-04-1343. [DOI] [PubMed] [Google Scholar]
  25. Müller A, Homey B, Soto H, Ge N, Catron D, Buchanan ME, McClanahan T, Murphy E, Yuan W, Wagner SN, Barrera JL, Mohar A, Verástegui E, Zlotnik A. Involvement of chemokine receptors in breast cancer metastasis. [see comment]. Nature. 2001;410:50–56. doi: 10.1038/35065016. [DOI] [PubMed] [Google Scholar]
  26. Scotton CJ, Wilson JL, Scott K, Stamp G, Wilbanks GD, Fricker S, Bridger G, Balkwill FR. Multiple actions of the chemokine CXCL12 on epithelial tumor cells in human ovarian cancer. Cancer Res. 2002;62:5930–5938. [PubMed] [Google Scholar]
  27. De Falco V, Guarino V, Avilla E, Castellone MD, Salerno P, Salvatore G, Faviana P, Basolo F, Santoro M, Melillo RM. Biological role and potential therapeutic targeting of the chemokine receptor CXCR4 in undifferentiated thyroid cancer. Cancer Res. 2007;67:11821–11829. doi: 10.1158/0008-5472.CAN-07-0899. [DOI] [PubMed] [Google Scholar]
  28. Kajiyama H, Shibata K, Terauchi M, Ino K, Nawa A, Kikkawa F. Involvement of SDF-1α/CXCR4 axis in the enhanced peritoneal metastasis of epithelial ovarian carcinoma. Int J Cancer. 2008;122:91–99. doi: 10.1002/ijc.23083. [DOI] [PubMed] [Google Scholar]
  29. Perrone G, Ruffini PA, Catalano V, Spino C, Santini D, Muretto P, Spoto C, Zingaretti C, Sisti V, Alessandroni P, Giordani P, Cicetti A, D'Emidio S, Morini S, Ruzzo A, Magnani M, Tonini G, Rabitti C, Graziano F. Intratumoural FOXP3-positive regulatory T cells are associated with adverse prognosis in radically resected gastric cancer. Eur J Cancer. 2008;44:1875–1882. doi: 10.1016/j.ejca.2008.05.017. [DOI] [PubMed] [Google Scholar]
  30. Cao Y, Zhao J, Lei Z, Shen S, Liu C, Li D, Liu J, Shen GX, Zhang GM, Feng ZH, Huang B. Local accumulation of FOXP3+ regulatory T cells: evidence for an immune evasion mechanism in patients with large condylomata acuminata. J Immunol. 2008;180:7681–7686. doi: 10.4049/jimmunol.180.11.7681. [DOI] [PubMed] [Google Scholar]
  31. Balabanian K, Levoye A, Klemm L, Lagane B, Hermine O, Harriague J, Baleux F, Arenzana-Seisdedos F, Bachelerie F. Leukocyte analysis from WHIM syndrome patients reveals a pivotal role for GRK3 in CXCR4 signaling. J Clin Invest. 2008;118:1074–1084. doi: 10.1172/JCI33187. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Tjalma WA, Van Waes TR, Van den Eeden LE, Bogers JJ. Role of human papillomavirus in the carcinogenesis of squamous cell carcinoma and adenocarcinoma of the cervix. Best Pract Res Clin Obstet Gynaecol. 2005;19:469–483. doi: 10.1016/j.bpobgyn.2005.02.002. [DOI] [PubMed] [Google Scholar]
  33. International Agency for Research on Cancer/World Health Organization Lyon, France: IARC Press; Handbooks of Cancer Prevention, vol 10. Cervical Cancer Screening. 2005:pp 469–483. [Google Scholar]

Articles from The American Journal of Pathology are provided here courtesy of American Society for Investigative Pathology

RESOURCES