Abstract
Background: The role of completion surgery after concurrent radiochemotherapy (CCRC) for advanced cervical cancer remains controversial. Individual predictive factors of CCRC response and survival are mandatory for treatment adaptation and to determine a population who would take interest in completion surgery after CCRC. The aim of this study was to evaluate the ability of biomarkers to predict the response to CCRC. Patients and Methods: Between 1996 and 2008, in 58 patients with advanced cervical cancer for whom pre-therapeutic cone biopsy was available, we tested several biomarkers (ALDH1, CD44, CD24, IDO, Ki67, P63, CK7, p-Stat3, Foxp3 and IDO). Results: Residual disease was found in 49.1% of cases (n=26). We found a significant association between progression-free survival and residual disease on completion hysterectomy (p=0.044). Univariate analysis of the different factors showed that negativity for cytokeratin 7 expression was a strong predictor for the presence of residual tumor (p=0.001). Conclusion: These results are encouraging and CK7 could be used as a predictive factor of response to CCRC.
The gold standard to treat patients with locally advanced stages (IB2–IV) of cervical cancer is concurrent radiochemotherapy (CCRC). A multi-center study showed that survival was significantly better for patients receiving CCRC [hazard ratio (HR)=0.77, 95% confidence interval (CI)=0.60-0.98; p=0.037] compared to those receiving radiotherapy-alone (1). Moreover, a recent meta-analysis confirmed that CCRC improves 5-year survival by 6% (HR=0.81, p<0.001) compared to the same radiotherapy regimen (2). In a multivariate analysis, Ferrandina et al. showed that residual tumor and disease stage were the two most relevant prognostic factors for disease-free (DFS) and overall survival (OS) (3). However, despite advances in imaging techniques including Computed Tomography (CT), Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET), it remains difficult to detect residual cervical disease in order to identify good candidates for completion surgery. The role of completion surgery remains controversial and a recent randomized study failed to demonstrate any benefit of completion surgery due to premature closure because of poor accrual (4). However, Houvenaeghel et al. found that combining CCRC with curative surgery increased survival by reducing the risk of local relapse (5).
Individual predictive factors of response to CCRC and survival are mandatory to adapt treatment in case of failure and to determine a population of good responders who would not take interest in completion surgery after CCRC. Indeed this type of surgery has a high rate of morbidity, whereas the therapeutic value of such surgery remains debated (6). Moreover, studies on the precise molecular characteristics of cervical cancer are still needed for a better understanding and characterization of cervical cancer prognosis and therapeutic management.
In analogy with breast cancer research, some biomarkers could be tested in cervical cancer. Firstly, the implication of tumoral stem cells in the mechanism of recurrence, metastasis and treatment failure is well-known (7). They are characterized by CD44 positivity and CD24 negativity. As a stem cell marker, ALDH1 is an independent marker of poor prognosis in inflammatory breast cancer (8). Moreover Chen et al. showed the implication of ALDH1-positive cells in resistance to radiation therapy in oral cancer (9). Secondly, the tumor needs to escape the immune response in order to grow and to progress (10). Immunosuppression could be mediated through indoleamine-2,3-dioxygenase (IDO) (T-cell immunity) and FOXP3 (modulation of T-regulatory cells). Cervical cancer, as glassy cell carcinoma, belonging to the group of muco-epidermoid tumors is known to have a negative outcome (11). Most epidermoid carcinomas start from the junction between the endocervical and exocervical mucosae. The presence of glandular differentiation in an epidermoid carcinoma has been considered as a marker of the worst evolution. In this sense, occult glandular differentiation could explain some recurrences (11). Cytokeratin 7 (CK7) is used to mark this minority glandular component amongst squamous cell carcinoma cells (12). By contrast, P63 marker of basal cells in malpighian epithelium could be related to differentiation. Finally, Ki67 and pSTAT3 are implicated in cell replication and proliferation and could be related to outcome (13-16).
Therefore, the aim of this study was to evaluate the ability of these biomarkers, analyzed on initial cone biopsy, to predict the response to CCRC.
Patients and Methods
Between 1996 and 2008, 58 patients with advanced cervical cancer (IB2–IV FIGO stage), for whom pre-therapeutic biopsies were available, were followed-up at the Paoli Calmettes Cancer Institute. Clinical and histological data were collected. All patients underwent CCRC followed six to eight weeks later by completion simple hysterectomy associated with pelvic node sampling. CCRC consisted of a 45-Gy external-beam radiation therapy using a 4-field box to the whole pelvis, associated with weekly cisplatin-based chemotherapy. A low-dose brachytherapy was delivered with a dose of 15 Gy prescribed on the reference isodose encompassing the clinical target volume. A lateral-pelvic boost of 10 Gy was delivered with an anterior-posterior field for involved parametrium when it was judged necessary by the physician.
Immunohistochemistry. The immunohistochemical analyses were performed on the tissue obtained at the diagnostic biopsy; they were fixed in 10% buffered formalin and embedded in paraffin. From selected tissue blocks, a single section was cut and stained with hematoxylin and eosin (H&E). The stained section was used to assess the quality of the tissue material, to confirm the pathological diagnosis and to quantify lymphocytic stroma. Further sections were used for immunohistochemistry.
After de-paraffinization and rehydration, the slides were rinsed and immersed in target retrieval solution (pH 6.0; Dako S2031 or pH 9.0; Dako S2367; Dako, Glostrup, Denmark). The antibodies used were to CD44 (diluted 1/400), CD24 (1/300), Ki67 (1/50) (Neomarkers, Fremont, CA, USA); ALDH (1/50)(BD Sciences, Lexington USA, IDO (1/20) (17), FOXP3 (1/25) (E Biosciences, San Diego, CA, USA); P63 (1/50), CK7 (1/500) (Dako); p-STAT3 (1/200) (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Next, the post-primary blocker agent was applied for 30 min and the slides were then treated with a polyvalent secondary antibody for 30 min, followed by incubation with cerium-diaminobenzidine (DAB) (Dako K5001) and counterstaining with hematoxylin.
Evaluation of the immunostaining. CD44+CD24− profile was considered positive if strong complete membranous CD44 staining without any CD24 staining was observed. Expression of ALDH1, P63, CK7 and p-STAT3 was considered positive if any degree of cytoplasmic staining was present in the tumor cells. Expression of IDO was evaluated in tumor cells and in stroma, and considered positive if any degree of cytoplasmic staining was present. Expression of FOXP3 was considered positive if any degree of cytoplasmic staining was present in lymphocytic stroma. Due to the high level of positivity for Ki67 expression in tumor cells, positivity was considered when more than 80% of tumor cells were positive.
Slides were read by two experienced pathologists and the assessment of antibody's staining was performed once for all cases.
Statistical analysis. The analysis was performed with SPSSv16.0.Data were analyzed using the Chi-square test. Differences were considered significant when p<0.05. Spearman test was used to establish correlation coefficients. We used the Kaplan–Meier product limit method to describe survival and the log-rank test to assess differences. Overall survival time was calculated in months from the date of surgery to death, or the date of last follow-up for surviving patients and disease-free survival time from the date of diagnosis to recurrence.
Results
Patients' characteristics. Among the 58 patients, 81% had epidermoid cancer (n=47), 15.5% adenocarcinoma (n=9) and 3.4% mixed or muco-epidermoid tumors (n=2). The distribution by stage according to the TNM classification was as follows: T1 (FIGO stage 1B2) in 32.8% of cases (n=19), T2 in 41.4% (n=24), T3 in 13.8% (n=8) and T4 in 12.1% (n=7). Residual tumor, defined as the living residual cancer lesions discovered on the hysterectomy of completion hysterectomy, was found in 49.1% of cases (n=26) (mean size of 7.5 mm). It was associated in 15.7% cases with pelvic lymph node involvement.
Survival. The median follow-up was 46 months (range=6-160 months). Nineteen patients experienced disease relapse, including 13 local recurrences and eight distant recurrences. We found a significant negative association between progression-free survival (PFS) and the presence of residual disease on completion hysterectomy (p=0.044, Figure 1).
Correlation between biomarkers. We established correlation coefficients between markers of differentiation and stem cells. There was a significant inverse correlation between CK7 and P63 (p=0.026). There was also an inverse correlation between P63 and ALDH1 (p=0.012), and P63 and CD44+CD24− profile (p=0.038). However, ALDH1 and CK7 were significantly correlated (p=0.00061).
For the markers implicated in immune response, there was a correlation between levels of IDO in tumor cells and the density of the lymphocytic stroma (p=0.0042). Epithelial IDO and FOXP3 were positively correlated but the significance of the correlation was borderline (p=0.06).
For the markers of proliferation and differentiation, the correlations between Ki67 and PSTAT3, and between Ki67 and P63 were highly significant (p<0.001 for both).
Progression-free survival according to residual disease on hysterectomy after concurrent radiochemotherapy.
Prognostic factors for residual disease on completion surgery. Univariate analysis of the different factors showed that the absence of expression of CK7 was a strong predictor for the presence of residual tumor: 82% of cases with residual tumor were CK7-negative; by contrast, only 17% of cases without residual disease were CK7-negative, p=0.001). There was no significant association between the tumor stage and the presence of residual disease or CK7 expression. The other markers did not appear to have a significant correlation with the presence of a residual tumor.
CK7 is expressed differently depending on histological type. Among the 58 tumors, 47 were epidermoid carcinomas and 62% (29 tumors) expressed CK7. For the other histological types (adenocarcinomas and mixed or muco-epidermoid tumors), we found expression of CK7 in all cases.
The analysis of the relation between survival and biomarkers did not reveal any significant differences. For the association between biomarkers and survival, we found significant differences but did find a trend for a better OS (p=0.07, Figure 2) in the case of expression of CK7.
Discussion
Despite cervical swab screening for early lesions, cervical cancer is often diagnosed in the advanced stage in 50% of cases. The incidence is 10/100000 in France, and the disease is responsible for 1,834 deaths every year. The prognostic factors are the size of the tumor, the FIGO stage and lymph node involvement. The presence of residual tumor seems to depend on the characteristics of the tumor itself. In addition, according to Houvenaeghel et al., completion surgery leads to an improvement in local control of advanced cervical cancers (18, 19) and therefore has an impact on overall and recurrence-free survival (20). However, surgery performed after radiochemotherapy remains risky and results in severe comorbidities, notably with regard to the urinary system (6). In the current French Standard Options Recommendations (September 1999), completion surgery is an option.
Overall survival according to the presence of cytokeratin 7 (CK7) on pre-therapeutic biopsy.
The interest of this work lies in the fact that it explores predictive factors of residual tumor that will make it possible to limit indications for surgery to those who would have only partially responded to neoadjuvant treatments.
This study in a population of 58 patients shows that the population in question is representative of the population of patients with advanced cervical cancer in France in terms of prevalence according to stage. The prognosis according to the different known clinical factors was also representative in this population. In this series, tumor size, FIGO stage, lymph node involvement and the presence of residual disease appear to have an impact on the overall and recurrence-free survival curves (results not shown). In our series, as in the literature, residual tumor and lymph node involvement were significant prognostic factors (21).
Immunochemistry and response to therapy. FOXP3 is highly expressed by T-regulatory lymphocytes, which are involved in the mechanisms by which the tumor evades the immune system. According to Ladoire et al., the expression of FOXP3 by tumor cells is a predictor of improved survival in a sub-population of patients with HER2-positive breast cancer treated with neoadjuvant therapy (22). In cervical cancer, levels of FOXP3 seem increase in high grade precancerous lesions and in microinvasive lesions, as well as in metastatic lymph nodes (21). This evolution seems to be linked to that of IDO (23). In our series, FOXP3 was mainly expressed in the lymphoid cells and we did not find any correlation between response-to-treatment and expression of FOXP3, unlike the study by Jordanova et al. (24), These authors showed that a high concentration of T-regulatory cells was a predictor of a poor prognosis and that a low CD8/T-regulatory cell ratio was a significant and independent predictor of a poor prognosis. Our series, however, was not comparable: our patients had more advanced cervical cancer, the treatments were different (surgery for Jordanova et al. and radiochemotherapy followed by surgery in our series) and the brands for the antibody to markers used were different. It was possible to distinguish between IDO expression in the tumor cells and in the stroma cells. Like Nakamura et al. (21), we found that there was a front line of tumor invasion that strongly expressed IDO. This finding could be included in the hypothesis made by Yoshida et al. (25) that IDO expression in cells contributes to tumor progression. No prognostic value was found for IDO expression in the tumor cells or stroma in this series. In the same way, there was no significant relationship with regard to response to treatment.
However, there was a correlation between IDO and FOXP3 in this series, which reinforces the validity of our results with regard to the literature (21). The findings of these studies are thus in keeping with the idea of a balanced immune system, which explains the immunological failure that allows the tumor to develop (26, 27).
Proliferation and therapeutic response. As P63 is the marker of basal cells in the malpighian epithelium and the growth zone, its quantitative correlation with Ki67 validates our findings. The study of Pillai et al. found that Ki67 was a predictor of response to radiotherapy, but the patients only received one course of radiotherapy and were then monitored to determine the presence of a residual lesion in cases of disease evolution (15). Studying the expression of Ki67 and RhoC in epidermoid cancer of the cervix, the authors showed that RhoC was a potential marker of invasion and a predictor of metastatic potential (28). In our series, pSTAT3 had no prognostic value. This transcription factor did not correlate with the presence of residual tumor either.
Differentiation, stem cells and therapeutic response. CK7 is classically expressed by the glandular component. It, therefore, marks the so-called reserve cells of the exo/endocervix junction zone (12, 29). Thus a CK7-positive reserve cell could, in its normal state, correspond to a progenitor cell able to differentiate into an epidermoid or glandular cell. All adenocarcinomas therefore express CK7, but for malpighian carcinoma, there are two categories, the expression of CK7 could be positive or negative. It has been correlated with histological response in our series. Indeed, in our series, the absence of expression of CK7 (40.5%) seems to be associated with greater resistance to radiochemotherapy. However, CK7 was not a significant prognostic factor for OS or PFS since other factors such as lymph node involvement and initial tumor size were also strong prognostic factors in this evaluation. This result is all the more surprising since CK7 expression correlated with the expression of ALDH1, a marker of stem cells, which is classically associated with resistance to therapy. However, other studies concerning ALDH1 expression seem to diverge with regard to its favorable or unfavorable prognostic value. Indeed, in ovarian cancer, the presence of ALDH1 is associated with a favourable prognosis (30). The double marker CD44+CD24−, which is characteristic of stem cells, does not appear to be a prognostic factor or predictor of the response to radiochemotherapy. In addition, we found no correlation between ALDH1 levels and the CD44+CD24− ratio, suggesting that these markers identify different populations of stem cells.
Conclusion
In this study, we found that the absence of CK7 expression on initial biopsy is associated with the presence of residual disease on completion hysterectomy after CCRC. Moreover, the presence of residual disease is associated with poorer survival. These results are encouraging and CK7 could be used as a predictive factor of response to CCRC. CK7 could also be used to determine a population of good responders who would not undergo completion surgery after CCRC.
Footnotes
-
Conflicts of Interest Statement
None of the Authors had a potential conflict of interest.
- Received November 19, 2013.
- Revision received December 9, 2013.
- Accepted December 10, 2013.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved