全文

Introduction

Carcinoma of the uterine cervix is the fourth most common cancer in women, and the seventh overall, with estimated 528,000 new cases and 266,000 deaths worldwide in 2012, accounting for 7.5% of all female cancer deaths. Almost nine out of ten (87%) cervical cancer deaths occur in the less developed regions [1].

 

In Sweden, 421 new cases were recorded in 2011, accounting for 1.5% of all female cancers, and 139 deaths, which was 1.3% of all female cancer deaths. The incidence of cervical carcinoma has declined in most of Europe due to well organized screening programs for vaginal cytology [2]. In the developed world most patients present with early disease, either confined to the cervix or with limited extension beyond it (FIGO stages IB1–IIA). Radical, extended hysterectomy with node dissection and radical radiotherapy are the two options for these cases both giving 5-year survival rates of ~ 80–90%. For locally advanced disease (FIGO stages IIB–IVA) as well as for bulky stage IB (> 4 cm) radical radiotherapy has been used successful for nearly a century [3]. Pelvic radiotherapy offers a good chance of cure, but the maximum radiation dose is limited by normal tissue tolerance. Brachytherapy (BT) plays a major role in the therapeutic management of patients with cervical cancer in stages I–IV. The rapid dose fall-off allows a high central dose to the tumor, while sparing the risk organs [4]. The combination of external beam radiotherapy (EBRT) and intracavitary brachytherapy (ICBT) [5] and concomitant chemotherapy (cisplatin) has become standard of care for locally advanced disease [6].

 

However, the 5-year survival rate varies from 60% for patients with stage IIB disease to 20% for patients with stage IVA disease, and the identification of potential new targets and treatment strategies is urgently needed.

 

The Hedgehog signaling pathway is essential for numerous processes during embryonic development, including cell growth, cell differentiation, patterning and organogenesis. In normal adult tissues, this pathway is involved in stem cell population maintenance, tissue repair and regeneration [7], [8], [9] and [10]. On the other hand, in various types of cancer, uncontrolled activation of the Hedgehog signaling pathway has been observed [7], [11], [12] and [13].

 

As there is an emerging evidence that the Hedgehog signaling pathway may be involved in carcinogenesis of squamous cell carcinoma of the uterine cervix [14], and plays a crucial role in the development as well as in the progression of this disease to more aggressive and even therapy-resistant disease states, and since the Hedgehog signaling pathways can be considered a potential therapeutic target. We aimed to analyze the expression of five Hedgehog proteins in this disease before treatment with correlation to outcome. Hedgehog signaling pathway is associated with tumor hypoxia [15], and is also associated with a poor outcome in women with node negative cervical cancer treated with radiation [16]. A recent paper on cervical carcinoma has shown that the Hedgehog pathway plays a role in repopulation after chemoradiation and suggests that SMO may be a valid therapeutic target [17].

 

The Hedgehog ligand proteins activate a membrane receptor complex initiating a cytoplasmic signal transduction by activating GLI zincfinger transcription factors. The receptor complex is formed by Patched (PTCH) and Smoothened (SMO), where PTCH normally inhibits SMO. When Hedgehog ligands bind to PTCH, the repression of SMO by PTCH is released, allowing SMO to activate the GLI proteins [18]. The GLI proteins are large and multifunctional transcription factors, and there are three GLI proteins that behave differently with partially redundant functions. GLI1 and GLI2 can mediate Hedgehog signals and have been implicated in tumorigenesis.

 

GLI1 is known to function primarily as an oncogene if tumors arise due to overexpression. On the other hand, GLI2 and GLI3 function as oncogenes or tumor suppressor genes, depending on the type of mutation and cellular context [19].

 

In the present study, a consecutive series of 131 cervical cancer (FIGO stages I–IV) patients treated with external beam radiotherapy and brachytherapy (± concomitant chemotherapy) were evaluated to assess treatment efficacy, clinical and pathological predictive and prognostic factors in relation to the immunohistochemical expression of five Hedgehog proteins in the tumors. Only a few studies before have reported on the clinicopathological significance of the Hedgehog pathway in human cervical carcinomas, and especially in locally advanced carcinomas treated with chemoradiotherapy [17].

 

Material and methods

Patients

In all, 131 patients with invasive cervical carcinomas were included in the series (Table 1). All patients were treated with a combination of external beam pelvic irradiation and brachytherapy. A 4-field box technique was used to treat the pelvis with an external beam dose of 50 Gy in 2.0 Gy daily fractions, five times a week, delivered using 18 MV photons. The brachytherapy was given during the external treatment as 5 intrauterine treatments of 6.0 Gy each, using high-dose-rate equipment. In 47 patients (36%) concomitant chemotherapy (weekly cisplatin, 40 mg/m2) was administered. The mean age of the patients was 68 years (range 28–90 years). Tumor characteristics are presented in Table 1. In 107 tumors HPV- and KRAS-status were evaluable as additional background factors. Lymph node status was not evaluated at the FIGO-staging of the tumors. Tumor samples were collected between 1993 and 2007. All tissue samples were collected before any type of therapy was instituted. The study was approved by the Ethics Committee in Uppsala, Sweden.

 

 

 

Clinical samples

A consecutive series of tumor specimens were identified and a total of 131 cases of formalin-fixed and paraffin-embedded tumor blocks from punch biopsies were obtained from the Department of Pathology, Örebro University Hospital. Tissue microarray slides were employed for the purpose of effective analyses. For preparation of these slides, we punched two tissue columns from the original blocks and inserted them into new paraffin blocks (each containing 4 × 24 holes to accept the tissue columns). Then, serially sectioned slides were prepared. Each tissue microarray slide could hold 96 specimens, allowing us to analyze 96 specimens simultaneously with a minimum of variation during the staining process. Each specimen was round in shape and 1.0 mm in diameter, thereby providing a sufficient amount of tissue for histopathological analysis. In 122 to 126 cases it was technically possible to evaluate the immunohistochemical staining of the five Hedgehog proteins. All 131 samples were analyzed for at least one or more of the five proteins.

 

Immunohistochemistry

Sections (4 μM) were prepared on FLEX IHC Microscope Slides for immunohistochemistry using the DAKO EnVision technique, and DAKO Autostainer, according to the manufacturer's protocols. The sections were deparaffinized in xylene twice for 10 min, rehydrated in a descending series of ethanol (99%, 96%, and 70%), and followed by washes in distilled water. Antigen retrieval was achieved by heating the samples in low pH 6.1 (citrate buffer), using Heat Induced Epitope Retrieval (DAKO) for 20 min. Then the sections were washed in distilled water. Staining was performed using EnVision FLEX System for Autostainer Link and DAB EnVision according to the manufacturer's protocol. The slides were incubated with five primary antibodies (anti-PTCH, anti-SMO, anti-GLI1, anti-GLI2, and anti-GLI3). Antibody EnVision FLEX HRP for 20 min and for 30 min when manually methods were used. The immunostaining kits were purchased from DAKO Inc. (Glostrup, Denmark) and Abcam (Cambridge, UK). The slides were counterstained with Envision FLEX hematoxylin, dehydrated, and mounted with PERTEX for examination.

 

Evaluation of the immunohistochemistry

The percentage of positive tumor cells was calculated for every specimen examined. In all, more than 200 tumor cells were counted per patient specimen. GLI1/2 molecules were expressed in both cytoplasm and nucleus in contrast to the other Hedgehog signaling proteins, whose expressions were observed mainly in the cytoplasm. The nuclear expression was generally low, and therefore the cytoplasmic and nuclear staining were analyzed together. Each specimen was examined in high-power field (40 × objectives) and scored by one of the authors (LBM) (Fig. 1).

 

 

Mean and median values were calculated as well as the standard deviation. The median value was used as a cut-off point in some analyses to define predictive and prognostic groups with up- and down-regulation of the five Hedgehog proteins. A modified scoring system according to Sinicrope et al. [20] was also used. A quantification of positively stained cells was used resulting in five different scores from 0 to 4. Score 0 = < 5% stained cells, 1 = 5–25% stained cells, 2 = 26–50% stained cells, 3 = 51–75% stained cells, and 4 = > 75% stained cells. The intensity of the staining was not analyzed in this study. For PTCH two groups scoring 0 and 4 were compared with regard to clinical outcome. For SMO, GLI1 and GLI3 two groups scoring 0 and 3 were compared, and for GLI2 groups with scores 1 and 2 were analyzed.

 

Statistical methods

In the statistical analyses the Pearson chi-square, the t-test, ANOVA statistics, logistic regression analysis, Cox proportional hazard regression analysis, Kaplan–Meier technique, and Gehan's Wilcoxon test for the differences between proportions, mean values, median values, binary outcome, and survival curves. A P value < 0.05 (double-sided) was regarded as statistically significant. The Statistica (version 12, 2013) software package (StatSoft, Inc., Tulsa, OK 74104, USA) was used in the statistical analyses.

 

Results

A number of statistically significant clinicopathological prognostic background factors were analyzed with regard to recurrence-free survival rate (Table 2). Age, FIGO-stage, histology, and KRAS-mutation were the most significant factors in univariate, but also independent and significant in Cox proportional multivariate analyses.

 

 

Positive staining (median values) for the five Hedgehog proteins was recorded in 8% to 37% of the tumor cells evaluated. The highest frequency was noted for SMO and the lowest for GLI1. The distributions of positive staining were skewed and for SMO typically bimodal.

 

Low expression of PTCH (< 5%) was more frequent in adenocarcinomas (90%) than in squamous cell carcinomas (56%), and this difference was statistically significant (Pearson chi-square; P = 0.046). However, the expression of the other four proteins (SMO, GLI13) was not associated with the type of histology. Degree of differentiation of the tumor, DNA-ploidy, S-phase fraction, malignancy grading score (MGS), as well as tumor stage (FIGO) and tumor size (largest diameter) were not associated with the immunohistochemical staining of the five Hedgehog proteins in this study.

 

Among tumors with HPV16 present 86% showed low expression of PTCH (< median) compared to 14% of tumors with other HPV-types present. This difference was of borderline significance (Pearson chi-square; P = 0.098). All HPV16-positive tumors (n = 75) were associated with low expression of GLI2 (< median value), while tumors with other HPV-types (n = 32) had high expression of GLI2 (> median value). Expressions of SMO, GLI1 and GLI3 were not associated with type of HPV-strain.

 

There was a statistically significant (Pearson chi-square; P = 0.042) association between low SMO-expression (< median value) and KRAS-mutation (17% mutation in low SMO-tumors, and 4% in high SMO-tumors). There was also a statistically highly significant (Pearson chi-square; P = 0.004) difference in KRAS-mutation frequency in tumors expressing GLI2 in less than 25% of the cells (27% mutation) compared with tumors expressing GLI2 in more than 25% of the cells (no mutations). Expressions of PTCH, GLI1 and GLI3 showed no association with KRAS-mutation.

 

The rate of primary residual tumor or early local recurrences was 17% if expression of PTCH was low (< 5%) and 0% if PTCH was highly expressed (> 75%). This difference was of borderline significance (Pearson chi-square; P = 0.070). Distant recurrences were also more frequent in tumors with low PTCH expression (27%) than in tumors with high PTCH expression (6%). This difference was also of borderline significance (Pearson chi-square; P = 0.068). Tumors with overexpressed SMO had a higher frequency of tumor persistence or local recurrences (33%) than tumors with low SMO expression (7%). This difference was statistically significant (Pearson chi-square; P = 0.032). A concomitant high expression (> median) of three or more of the Hedgehog proteins was not associated (Pearson chi-square; P = 0.671) with treatment outcome or local recurrences.

 

Expressions of GLI1 and GLI3 were not associated with tumor persistence after treatment, local recurrences or overall recurrences when the median values were used as the cut-off level. On the other hand, for GLI2 expression there was a statistically significant difference with regard to overall (Gehan's Wilcoxon test; P = 0.004) and distant (Gehan's Wilcoxon test; P = 0.015) relapse rates for groups with expression of GLI2 in the range of 5–25% and with higher expression.

 

Loco-regional recurrences showed the same pattern of association with the Hedgehog proteins analyzed as persistent tumors and local recurrences. Thus, there is a higher frequency in SMO-positive tumors (16%) compared with SMO-negative tumors (8%), but not statistically significant (Pearson chi-square; P = 0.169).

 

Distant metastases showed an association with low expression of PTCH and especially low expression of GLI2. If a grouping with low PTCH expression (< 5%) and high expression (> 75%) was used a low expression rate was associated with (Gehan's Wilcoxon test; P = 0.064) higher frequencies of persistent or local relapses as well as of distant metastases. With regard to SMO expression a grouping of high (> 50%) and low expression (< 5%) was significantly (Gehan's Wilcoxon test; P = 0.025) associated with persistent tumor or local recurrences. A difference was also recorded for distant metastases, but not statistically significant. For GLI2 there was a highly significant difference in both overall and distant relapses when the group with 5–25% expression was compared with the group expressing GLI2 in more than 25% of the tumor cells.

 

Patients with tumors expressing PTCH in more than 75% of the cells had significantly (Gehan's Wilcoxon test; P = 0.023) better recurrence-free survival than patients with tumors expressing PTCH in less than 5% of the cells (Fig. 2). The opposite situation was true for SMO expression, where tumors with low expression (< 5%) had a significantly (Gehan's Wilcoxon test; P = 0.033) better prognosis than tumors with high expression (Fig. 3). For GLI1 and GLI3 no differences were detected with regard to survival rate. Regarding GLI2 a significantly (Gehan's Wilcoxon test; P = 0.005) better prognosis was noted when the protein was expressed in the range of 25–50% of the cells compared with an expression in a lower range of 5–25% (Fig. 4). In proportional multivariate Cox regression analyses expression of PTCH (P = 0.030) and SMO (P = 0.046) was significant and independent prognostic factor after correction for tumor stage. Expression of GLI2 (P = 0.002) was significant and independent after correction for both tumor stage and type of histology.

 

Discussion

The median value of the percentage of positive tumor cells used to define negative (down-regulated) or positive expression (up-regulated) of the Hedgehog proteins was not the most appropriate way to find groups with different predictive or prognostic value with regard to the clinical outcome parameters (primary cure of the tumor, recurrences, recurrence-free survival, cancer-specific survival and overall survival rate). This was the case for all five proteins analyzed in this study. When instead a modified scoring method (only proportion of stained epithelial cells) according to Sinicrope et al. [20] was used to define five groups with regard to the proportion of staining of the tumor cells important and significant predictive and prognostic impact was found. Furthermore, the most appropriate method was to use individual and different cut-off points for the Hedgehog proteins. For PTCH two groups could be defined with less than 5% positive cells (low) and those with more than 75% positive cells (high). For SMO two groups could also be defined also with less than 5% positive cells and another one with more than 50% positive cells. For GLI1 and GLI3 no predictive or prognostic information could be found regardless of cut-off points and group definitions. For GLI2 two groups could be defined with 5–25% positive cells and another group with higher percentage positive cells. These two GLI2-groups had a statistically highly significant predictive and prognostic value with regard to tumor recurrences and recurrence-free survival rate. Since we know that the three GLI-proteins have different mechanisms of actions [21] it could be reasonable to use different cut-off levels and to understand their various prognostic impacts when analyzed in infiltrative and advanced cervical carcinomas treated with chemoradiotherapy. GLI2 is the primary activator and transcriptional effector of Hedgehog signaling and GLI1 is a secondary target, downstream of GLI2. GLI2 has a strong nuclear expression persistent in all tumor cases. GLI2 seems to play a more important role than GLI1 and GLI3. Nuclear expression of GLI3 is very rare and its role as a transcription factor is probably minimal. Structure–function analyses have shown that C-terminal sequences are required for positive inducing activity and cytoplasmic localization, whereas N-terminal sequences determine dominant negative function and nuclear localization [22]. There are two important regulators of the GLI proteins, Sufu that forms complex with GLI proteins and prevents its nuclear accumulation, and Kif7 that promotes activation through dissociation of Sufu–GLI2 complexes to allow translocation of GLI2 into the nucleus [23]. The strong predictive and prognostic value of GLI2 in our series, in contrast to GLI1 and GLI3, supports this assumption. Of interest was that GLI2-expression was not related to histology, FIGO-stage or tumor size, despite the fact that these factors are known to be important predictive and prognostic factors in cervical carcinoma.

 

Low expression of PTCH was significantly more common in adenocarcinomas than in squamous cell carcinomas. Clinically, adenocarcinomas had a highly significantly worse prognosis in this series of cervical carcinomas after chemoradiotherapy. However, expressions of the other Hedgehog proteins were not associated with the type of histology.

 

Low expression of PTCH and GLI2 was common in tumors being positive for HPV16. Xuan et al. [24] found a relationship between Sonic Hedgehog expression and HPV16 infection. An association between HPV-infection and the Hedgehog signaling pathway has also been described before in the literature [25] and [26].

 

KRAS-mutation in cervical cancer was found in a prior study to be an important risk factor for tumor recurrences at distant sites after chemoradiotherapy [27]. In the present series all tumors with KRAS-mutation showed low SMO- and GLI2-expressions. Mills et al. [28] demonstrated that mutant KRAS initiated a cascade with binding of GLI1 to the IL-6 promoter regulating this cytokine by inducing the expression of Sonic hedgehog in cancer cells. Thus, GLI1 acts as a downstream effector of oncogenic KRAS in the tumor microenvironment in carcinogenesis. However, in our study we could not find any association between KRAS-mutation and GLI1-expression. On the other hand, low expression of GLI2 (< 25%) was associated with a high rate of KRAS-mutation (27%) compared to no mutations in tumors with higher GLI2 expression. Thus, low expression of GLI2, high frequency of KRAS-mutation, increased frequency of both overall and distant recurrences and were associated with each other in this study.

 

Tumor hypoxia and Hedgehog signaling pathway are associated in pancreatic carcinoma. Chaudary et al. [29] found an association between low SMO expression and HP5 (pO2 < 5 mm Hg) in cervical tumor tissue. However, hypoxia was not associated with clinical outcome in their study. In another study, using human cervix cancer xenografts, the same authors found that inhibition of Hedgehog gene activity during tumor growth reduced tumor size and also reduced lymphatic metastasis [30]. Since cervical carcinoma is a tumor known to contain significant hypoxic regions, especially in the central part of bulky tumors, it seems reasonable to assume that the Hedgehog signaling pathway could be of importance also in this disease. This hypothesis is supported in this study by the fact that persistent tumor after treatment with chemoradiotherapy and local recurrences had a significant association with some of the Hedgehog proteins (PTCH and SMO) analyzed.

 

The Hedgehog signaling pathway seems to be of importance in cervical carcinoma as well as in precursor lesions. The most important role is probably played in carcinogenesis and in interaction with HPV. In advanced invasive carcinomas treated with radiotherapy its role may be different but still of importance. A predictive and prognostic impact was found for PTCH, SMO, and GLI2 with regard to residual carcinoma after therapy, local and regional recurrences, and for PTCH and GLI2 also distant relapses. Expression of GLI2 has also been shown to be associated with survival in breast cancer [31]. The problem of tumor hypoxia in locally bulky tumor lesions may also be involved in this context. Psyrri et al. found in head and neck cancer a modulating effect of the Hedgehog pathway on EGFR and the downstream PIK3CA pathway, which is the most frequently mutated oncogene in squamous cell head and neck carcinoma [32]. Experimental studies on tumor cell lines support the concept of blocking the pathway signaling to reduce tumor size, invasion, epithelial to mesenchymal transition, and lymphatic tumor spread [25] and [26].

 

Thus, the Hedgehog signaling pathway and its proteins seem to play an important role in cervical carcinogenesis, but also in established invasive cervical carcinoma together with HPV-infection and KRAS-mutation, and blockage of this pathway may be a potential treatment option in the future [26] and [27]. The Hedgehog proteins (e.g. PTCH, SMO, and GLI2) and their expression might also be used as predictive and prognostic factors to improve treatment planning and hopefully outcome of the therapy in cervical carcinoma.

 

Funding sources

This work was supported by the Research Foundation at the Department of Oncology, and the Research Funds of the University Hospital, Örebro, and the Foundation for Research in Gynecological Cancer, Örebro, Sweden.

 

Conflict of interest statement

The authors declare no conflicts of interest in this study.

 

Acknowledgments

We are grateful to and wish to thank Berit Bärmark, Monica Sievert, Mats Karlsson and Lisbeth Lindvall for their skillful technical assistance and support of this study.

 

Appendix A. Supplementary data

The following is the Supplementary data to this article.