Cervical cancer is the second most common cancer in women worldwide and is the leading cause of cancer-related death in women in developing countries.1 Despite the public health importance of cervical cancer, factors that influence the outcome of these patients are not completely understood. Understanding the mechanisms of progression of cervical cancer is important for improving prognosis as well as response to therapy and eventually improves the quality of life of patients.
Tumor oxygenation has been recognized as an important predictive parameter for survival in cervical cancer. Hypoxia’s influence on the effectiveness of radiotherapy had long been recognized (“oxygen enhancement effect”). A worsening of prognosis for patients with hypoxic tumors was also shown to exist in cases treated through surgery alone.2 The most important protein regulating the molecular response on hypoxia is hypoxia-inducible factor-1 (HIF-1), a heterodimer consisting of 2 subunits, HIF-1α and HIF-1β. As a transcription factor, HIF-1α can transactivate more than 70 target genes and is a master regulator of erythropoiesis,3 angiogenesis,4 cell metabolism, as well as genomic stability.5 In cervical cancer, multiple studies6–11 showed that patients with strong HIF-1α expression had a significantly shorter disease-free survival (DFS) and overall survival (OS), whereas others12–15 could not confirm this, although associations between HIF-1α expression and clinicopathological parameters (such as histology type, International Federation of Gynecology and Obstetrics [FIGO] stage, tumor grade, tumor size, lymph node metastasis [LNM], and anemia) are conflicting.6,10,12,13,16
In this study, we conducted a clinicopathologic study in patients with earlier-stage cervical cancer treated through surgery and performed a meta-analysis among patients with all stages of cervical cancer to estimate the prognostic importance of elevated tissue-based HIF-1α levels for DFS and OS. In addition, the relationship between HIF-1α expression and clinicopathological parameters (which may be predictors of poor clinical outcome of cervical cancer) was also examined.
MATERIALS AND METHODS
Patients and Tissues
Patients with primary invasive cervical cancer treated through radical hysterectomy and pelvic lymph node dissection at the Department of Obstetrics and Gynecology, Sun Yat-sen Memorial Hospital, Sun Yat-sen University, between January 2006 and December 2007 were investigated and followed up in this study. Patients who underwent definitive surgery with FIGO stage IA-IIB carcinoma of the cervix for whose tissue blocks were available before chemotherapy and/or radiotherapy were included in the study. During the observation period, 9 patients were lost and 74 patients were identified and included finally. The followed-up rate is 89.2%. Overall, the median observation time was 53 months (range, 11–68 months). The median age of the patients was 45 years (range, 28–66 years). Among these 74 patients, 14 patients experienced relapse. Of these 14 patients, 13 died because of cervical cancer.
Formalin-fixed, paraffin-embedded surgical or biopsy specimens of 74 patients with cervical cancer were examined. All the specimens were taken before chemotherapy. Diagnosis was established preoperatively through colposcope biopsy or cone excision. The study was approved by the institutional review board of Sun Yat-Sen Memorial Hospital, Sun Yat-sen University.
Tissue was cut into 4-μm section, placed on polylysine-coated slides, deparaffinized in xylene, and rehydrated using graded ethanol. Then, the slides were incubated in 10-mM citrate buffer at 121°C for 5 minutes and quenched for endogenous peroxidase activity in 0.3% hydrogen peroxide for 10 minutes. For the immunohistochemical detection of HIF-1α, the specimens were incubated overnight at 4°C with a monoclonal anti–HIF-1α antibody (Clone MAb H1α67, NB 100–105; Novus Biologicals)7 in a dilution of 1:80. Immunostaining was performed using Real Envision Detection kit (Gene Tech Co Ltd, Shanghai, China). Subsequently, the sections were counterstained with hematoxylin and mounted in a nonaqueous mounting medium.
The primary antibody was omitted for negative controls. As positive control, a specimen of endometrial carcinoma with strong expression of HIF-1α was used.17 The expression of HIF-1α was determined by 2 independent observers without knowing the information of patient’s outcome. The observers assessed semiquantitatively the percentage of stained tumor cells and the staining intensity. The percentage of positive cells was rated as follows: 1 point, less than or equal to 10%; 2 points, 11% to 30%; 3 points, 31% to 50%; and 4 points, greater than or equal to 51%. The staining intensity was rated as follows: 1 point, weak intensity; 2 points, moderate intensity; and 3 points, strong intensity. Points for the percentage of positive cells and staining intensity were added, and the specimens were attributed to 4 groups according to their overall score: negative, less than or equal to 2 points; weak expression, 3 to 4 points; moderate expression, 5 points; and strong expression, 6 to 7 points. In cases with disagreement, a consensus was reached after reviewing the slides together.
The analyses between HIF-1α expression and clinicopathologic parameters were assessed using the Pearson χ2 test or the Fisher exact test. Disease-free survival was defined from the primary surgery until the first evidence of progression of disease. Overall survival was defined as the period from the primary surgery until the death of the patient. Kaplan-Meier survival curves and Cox regression analysis were used to assess outcome. Small groups were pooled (grades 1 and 2; HIF-1α negative and weak expression).
The Cox proportional-hazards model was used for the multivariate analysis. The HIF-1α expression, FIGO stage, tumor grade, tumor size, depth of cervical invasion (DOI), lymphovascular space invasion (LVI), and LNM were entered into Cox regression. For all tests, a 2-tailed P ≤ 0.05 was considered significant. All these statistical analyses were performed with the Statistical Package for the Social Sciences (SPSS 15.0; SPSS Inc, Chicago, IL).
To further study these relationships, a meta-analysis of all relevant association studies was conducted. Detailed methods of meta-analysis were described in our previous study.18 Medline, PubMed, Embase, and Web of Science were searched by using combinations of the following keywords: “cervical cancer,” “HIF-1α,” “prognosis,” “prognostic,” “marker,” and “survival” (last update on January 1, 2013). Included studies were divided into 2 groups for analysis: those with data regarding DFS and those regarding OS. For the quantitative aggregation of the survival results, hazard ratios (HRs) and their 95% confidence intervals (CIs) were combined to give the effective value. If a given study provided results obtained from univariate and multivariate analyses, the latter was selected. When these statistical variables were not given explicitly in an article, they were estimated from available data using methods reported by Tierney et al.19 For the meta-analysis of the relation between HIF-1α overexpression and clinicopathological parameters, odds ratios (ORs) and their 95% CIs were combined to give the effective value. Heterogeneity assumption was checked by Q test and quantified using I2. Both fixed-effects (Mantel-Haenszel method) and random-effects (DerSimonian-Laird method) models were performed. Subgroup analyses were performed through treatment (chemoradiation or surgery). The Begg funnel plot and the Egger test were performed to access the publication bias of literatures. All of the statistical tests used in this meta-analysis were performed using STATA version 10.0 (StataCorp LP, College Station, TX).
HIF-1α Expression and Clinicopathologic Characteristics
Immunostaining was predominantly nuclear, but cytoplasmic staining was seen occasionally. Hypoxia-inducible factor-1α was expressed in 94.6% (70/74) of the patients (Supplement Figure 1, available at http://links.lww.com/IGC/A223). Clinical and histopathological characteristics are shown in Table 1. There were significant associations between FIGO stage (P = 0.024), tumor size (P = 0.003), anemia (defined as hemoglobin concentration <110 g/L, P = 0.010), and HIF-1α expression, respectively. Furthermore, hemoglobin concentration analyzed as a continuous variable was found to be significantly related to HIF-1α expression (Spearman correlation test, P = 0.002, r = −0.358). However, no significant associations were found between HIF-1α expression and age, histology type, tumor grade, DOI, LVI, or LNM.
The Kaplan-Meier curves showed a significantly worse DFS and OS for patients with increased HIF-1α expression (Fig. 1). Furthermore, log-rank tests revealed a significant correlation between HIF-1α expression, FIGO stages, tumor grade, tumor size, and DFS (P = 0.007, P = 0.028, P = 0.043, and P = 0.022) and OS (P = 0.023, P = 0.045, P = 0.021, and P = 0.043), respectively. The 3-year DFS rate for all patients was 83.8%: 94.3% for patients with absent/weak HIF-1α expression, 78.3% for patients with moderate HIF-1α expression, and 68.8% for patients with strong HIF-1α expression (Table 1).
Multivariate Cox regression analyses are summarized in Table 2. The multivariate analysis confirmed HIF-1α overexpression to be an independent predictor for poor DFS (P = 0.030; HR, 2.67; 95% CI, 1.10–6.47) and OS (P = 0.038; HR, 2.57; 95% CI, 1.06–6.23). In addition, high tumor grade was also an independent factor for poor DFS (P = 0.012; HR, 5.56; 95% CI, 1.47–21.13) and OS (P = 0.012; HR, 6.23; 95% CI, 1.49–25.97).
In our study, a total of 61 patients underwent surgery directly and 13 patients underwent neoadjuvant chemotherapy, receiving surgery after tumor reduction. After excluding these 13 patients, the results of survival analysis remained practically unchanged (data not shown). The margin status, which may be related to the patients’ prognosis, was all negative.
A total of 12 eligible studies were identified (Supplement Figure 2, available at http://links.lww.com/IGC/A223). Of these studies, 2 studies11,14 were eliminated because of inadequate data for the meta-analysis. In the other 2 studies,8,20 patients involved were overlapping; therefore, only the latest study8 was included. Thus, including our own data, the meta-analysis included 10 studies6–10,12,13,15,16 (composed of 617 patients) (For an overview of the studies identified and their main characteristics, see Table 3). Overall, DFS and OS were obtained in 7 studies.
The main meta-analysis results are given in Table 4. The meta-analysis of 7 studies on the prognostic value of HIF-1α expression showed that high HIF-1α levels were associated with poor DFS (HR, 1.98; 95% CI, 1.22-3.21; P = 0.006; Fig. 2), although there was moderate evidence for heterogeneity (I2 = 67.9%, Ph = 0.005). Restricting the analysis to studies wherein patients were treated through chemoradiation for the determination of HIF-1α expression did not alter the heterogeneity tests results (I2 = 51.5%, Ph = 0.083) (Table 4). However, further analysis showed that heterogeneity was partly caused by the results of the study by Hutchison et al.13 When this study was excluded from the meta-analysis, the heterogeneity decreased to low (I2 = 0%, Ph = 0.432). The results remained practically unchanged in the overall studies analysis but changed in the chemoradiation subgroup (HR, 1.53; 95% CI, 0.99-2.38; P = 0.057; change to HR, 1.94; 95% CI, 1.31–2.89; P = 0.001). Subgroup analysis indicated that HIF-1α overexpression was also significantly related with poor DFS in patients with cervical cancer treated through surgery (HR, 3.72; 95% CI, 2.00-6.93; P < 0.001) without significant heterogeneity in the data (I2 = 3.4%, Ph = 0.309).
The meta-analysis of the 7 studies on the prognostic value of HIF-1α expression showed that high HIF-1α levels were associated with poor OS (HR, 2.58; 95% CI, 1.86–3.56; P < 0.001; Fig. 2) without heterogeneity (I2 = 0%, Ph = 0.942). The subgroup analysis indicated that HIF-1α overexpression was significantly related with OS in patients with cervical cancer treated through either chemoradiation (HR, 2.34; 95% CI, 1.54–3.55; P < 0.000) or surgery (HR, 2.98; 95% CI, 1.78–4.98; P < 0.000) without significant heterogeneity.
When the data of HIF-1α expression and clinicopathological parameters were pooled, there were significant associations between high HIF-1α expression and large tumor size (OR, 2.04; 95% CI, 1.24–3.34; P = 0.005) and anemia (OR, 2.04; 95% CI, 1.07–3.88; P = 0.030), respectively (Table 4).
Neither the funnel plot nor the Egger test revealed any evidence of publication bias except the analysis on DFS studies (Egger test: P = 0.010) (figure and other data not shown). After excluding the study of Hutchison et al13, the publication bias disappeared (P = 0.495).
Both the present clinicopathologic study and the meta-analysis showed that HIF-1α overexpression might predict poor DFS and OS for cervical cancer. In addition, there were significant associations between HIF-1α expression and tumor size and anemia, respectively. To our knowledge, this clinicopathologic study is the first study reported in a developing country, which may be much applicable to the patients from these countries.
Our study revealed that overexpression of HIF-1α is a predictor of poor prognosis in a patient with cervical cancer who received surgery, which agrees with the finding in the researches of Birner et al7 and Fujimoto et al.9 Previous studies have revealed that HIF-1α could promote some critical aspects of cancer biology.21,22 First, HIF-1α influences a number of target genes that play roles in tumor progression. Moreover, HIF-1α regulates many signaling pathways, such as PI3K/AKT/mTOR,23 Notch,24 and Myc25 signaling pathways, to mediate tumor proliferation, differentiation, migration, and invasion. Last but not the least, it has been clearly established that human papillomavirus (HPV) infection represents as the causative agent of cervical cancer. Furthermore, research has shown that overexpression of HPV type 16 oncoproteins promotes angiogenesis via enhancing HIF-1α and vascular endothelial growth factor (VEGF) expression.26 These may explain why patients with HIF-1α overexpressing cervical cancers are at increased risk for mortality.
This study revealed correlations between HIF-1α overexpression and anemia. Similar conclusions were noted by Burri et al16 but cannot be confirmed by Ishikawa et al.10 Animal studies have revealed that tumor hypoxia was asso ciated with anemia.27,28 The study of Takubo et al29 has shown that HIF-1α l precisely regulated the cell cycle quiescence, which was maintained by hematopoietic stem cells. These suggest that HIF-1α expression and hemoglobin production may interact with each other through some certain mechanisms.
Our study reveals an association between HIF-1α overexpression and large tumor size in cervical cancer. This association has not been apparent in the previous individual studies. However, this association has been shown in other carcinomas, such as intrahepatic cholangiocarcinoma, gastric adenocarcinoma, pancreatic ductal adenocarcinoma, and the like. The previous studies have reported that proliferation of cancer cells was limited by the diffusion of oxygen. Hypoxia-inducible factor-1α could control oxygen delivery and metabolic adaptation to hypoxia.30 These clinical observations may be explained partially by the fact that expression of HIF-1α may allow cancer cells to efficiently obtain energy for growth under hypoxia caused by large tumor volume.
In the present study, sources of heterogeneity and publication bias were explored. Regarding DFS, the result of the study of Hutchison et al13 differed from other studies. It could be that hypoxia is not predictive of outcome in the series of patients studied as the authors described. This may be caused by clinical heterogeneity, which may be derived from the different patients (with different age, clinical stage, ethnicity, physical condition, etc).
Some limitations should be acknowledged. First, in our clinicopathologic study, the follow-up period of some patients did not reach 5 years. However, the recurrences or deaths almost appeared in 3 years after surgery. Hence, the follow-up period may be adequate to some extent. Second, the sample size of our clinicopathologic study was relatively small, which might influence the validity of our analysis. However, the meta-analysis pooled series of studies and had higher statistical power to make up for this disadvantage to some extent. Given the limited number of the studies included in the subgroup analyses through treatment, the results of the subgroup analyses should be interpreted with caution. Additional studies involving larger samples are warranted. Third, immunohistochemistry was a semiquantitative method, and this may influence the precision of the result. Moreover, in the meta-analysis, differences in primary antibodies, immunohistochemistry staining protocols, evaluation standards, and cutoff values for high HIF-1α expression may also contribute to heterogeneity. Thus, further multicenter researches using standardized and quantitative methods are encouraged. Fourth, it has been clearly established that HPV infection represents as the causative agent of cervical cancer. However, in the present study, only half of the enrolled patients have the information of HPV infection. Therefore, we had not analyzed the correlation of HPV infection and HIF-1α protein levels. Additional studies concerning HPV infection are warranted.
Despite its limitations, our study had some advantages. First, analyzing HIF-1α expression as well as DFS, OS, and the clinicopathological parameters in cervical cancer made our analysis more extensive and valid. Second, to our knowledge, this is the first meta-analysis on the association between HIF-1α expression and prognosis in cervical cancer. Furthermore, all of the results from the random-effects model were similar with those from the fixed-effects model, which indicated that the statistic results were robust.
In conclusion, HIF-1α overexpression is associated with poor survival of cervical cancer; therefore, it might serve as one of the prognostic markers in cervical cancer and might be a potential target in cancer therapy.
The authors thank their colleagues from the Department of Pathology for contributing to this study and for their continuous support.
1. World Health Organization. WHO Guidance Note: Comprehensive Cervical Cancer Prevention And Control: A Healthier Future For Girls And Women
. Geneva, Switzerland: WHO Press, World Health Organization; 2013.
2. Hockel M, Schlenger K, Aral B, et al. Association between tumor hypoxia and malignant progression in advanced cancer of the uterine cervix. Cancer Res
. 1996; 56: 4509–4515.
3. Gopfert T, Gess B, Eckardt KU, et al. Hypoxia signalling in the control of erythropoietin gene expression in rat hepatocytes. J Cell Physiol
. 1996; 168: 354–361.
4. Maxwell PH, Dachs GU, Gleadle JM, et al. Hypoxia-inducible factor-1 modulates gene expression in solid tumors and influences both angiogenesis and tumor growth. Proc Natl Acad Sci U S A
. 1997; 94: 8104–8109.
5. Semenza GL. Targeting HIF-1 for cancer therapy. Nat Rev Cancer
. 2003; 3: 721–732.
6. Bachtiary B, Schindl M, Potter R, et al. Overexpression of hypoxia-inducible factor 1alpha indicates diminished response to radiotherapy and unfavorable prognosis in patients receiving radical radiotherapy for cervical cancer. Clin Cancer Res
. 2003; 9: 2234–2240.
7. Birner P, Schindl M, Obermair A, et al. Overexpression of hypoxia-inducible factor 1alpha is a marker for an unfavorable prognosis in early-stage invasive cervical cancer. Cancer Res
. 2000; 60: 4693–4696.
8. Dellas K, Bache M, Pigorsch SU, et al. Prognostic impact of HIF-1alpha expression in patients with definitive radiotherapy for cervical cancer. Strahlenther Onkol
. 2008; 184: 169–174.
9. Fujimoto J, Alam SM, Jahan I, et al. Plausible linkage of hypoxia inducible factor-1alpha in uterine cervical cancer. Cancer Sci
. 2006; 97: 861–867.
10. Ishikawa H, Sakurai H, Hasegawa M, et al. Expression of hypoxic-inducible factor 1alpha predicts metastasis-free survival after radiation therapy alone in stage IIIB cervical squamous cell carcinoma. Int J Radiat Oncol Biol Phys
. 2004; 60: 513–521.
11. Markowska J, Grabowski JP, Tomaszewska K, et al. Significance of hypoxia in uterine cervical cancer. Multicentre study. Eur J Gynaecol Oncol
. 2007; 28: 386–388.
12. Haugland HK, Vukovic V, Pintilie M, et al. Expression of hypoxia-inducible factor-1alpha in cervical carcinomas: correlation with tumor oxygenation. Int J Radiat Oncol Biol Phys
. 2002; 53: 854–861.
13. Hutchison GJ, Valentine HR, Loncaster JA, et al. Hypoxia-inducible factor 1alpha expression as an intrinsic marker of hypoxia: correlation with tumor oxygen, pimonidazole measurements, and outcome in locally advanced carcinoma of the cervix. Clin Cancer Res
. 2004; 10: 8405–8412.
14. Mayer A, Wree A, Hockel M, et al. Lack of correlation between expression of HIF-1alpha protein and oxygenation status in identical tissue areas of squamous cell carcinomas of the uterine cervix. Cancer Res
. 2004; 64: 5876–5881.
15. Niibe Y, Watanabe J, Tsunoda S, et al. Concomitant expression of HER2 and HIF-1alpha is a predictor of poor prognosis in uterine cervical carcinoma treated with concurrent chemoradiotherapy: prospective analysis (KGROG0501). Eur J Gynaecol Oncol
. 2010; 31: 491–496.
16. Burri P, Djonov V, Aebersold DM, et al. Significant correlation of hypoxia-inducible factor-1alpha with treatment outcome in cervical cancer treated with radical radiotherapy. Int J Radiat Oncol Biol Phys
. 2003; 56: 494–501.
17. Seeber LM, Horree N, van der Groep P, et al. Necrosis related HIF-1alpha expression predicts prognosis in patients with endometrioid endometrial carcinoma. BMC Cancer
. 2010; 10: 307.
18. Huang M, Chen Q, Xiao J, et al. Prognostic significance of cyclooxygenase-2 in cervical cancer: a meta-analysis. Int J Cancer
. 2013; 132: 363–373.
19. Tierney JF, Stewart LA, Ghersi D, et al. Practical methods for incorporating summary time-to-event data into meta-analysis. Trials
. 2007; 8: 16.
20. Bache M, Holzapfel D, Kappler M, et al. Survivin protein expression and hypoxia in advanced cervical carcinoma of patients treated by radiotherapy. Gynecol Oncol
. 2007; 104: 139–144.
21. Schwab LP, Peacock DL, Majumdar D, et al. Hypoxia inducible factor-1alpha promotes primary tumor growth and tumor-initiating cell activity in breast cancer. Breast Cancer Res
. 2012; 14: R6.
22. Yuan WC, Lee YR, Huang SF, et al. A Cullin3-KLHL20 Ubiquitin ligase-dependent pathway targets PML to potentiate HIF-1 signaling and prostate cancer progression. Cancer Cell
. 2011; 20: 214–228.
23. Karar J, Maity A. PI3K/AKT/mTOR pathway in angiogenesis. Front Mol Neurosci
. 2011; 4: 51.
24. Sahlgren C, Gustafsson MV, Jin S, et al. Notch signaling mediates hypoxia-induced tumor cell migration and invasion. Proc Natl Acad Sci U S A
. 2008; 105: 6392–6397.
25. Zhang H, Gao P, Fukuda R, et al. HIF-1 inhibits mitochondrial biogenesis and cellular respiration in VHL-deficient renal cell carcinoma by repression of C-MYC activity. Cancer Cell
. 2007; 11: 407–420.
26. Li G, He L, Zhang E, et al. Overexpression of human papillomavirus (HPV) type 16 oncoproteins promotes angiogenesis via enhancing HIF-1alpha and VEGF expression in non-small cell lung cancer cells. Cancer Lett
. 2011; 311: 160–170.
27. Kelleher DK, Matthiensen U, Thews O, et al. Tumor oxygenation in anemic rats: effects of erythropoietin treatment versus red blood cell transfusion. Acta Oncol
. 1995; 34: 379–384.
28. Kelleher DK, Mattheinsen U, Thews O, et al. Blood flow, oxygenation, and bioenergetic status of tumors after erythropoietin treatment in normal and anemic rats. Cancer Res
. 1996; 56: 4728–4734.
29. Takubo K, Goda N, Yamada W, et al. Regulation of the HIF-1alpha level is essential for hematopoietic stem cells. Cell Stem Cell
. 2010; 7: 391–402.
30. Brahimi-Horn MC, Chiche J, Pouyssegur J. Hypoxia signalling controls metabolic demand. Curr Opin Cell Biol
. 2007; 19: 223–229.
© 2014 by the International Gynecologic Cancer Society and the European Society of Gynaecological Oncology.