Necrosis related HIF-1α expression predicts prognosis in patients with endometrioid endometrial carcinoma

Endometrial cancer is the most common malignant tumour of the female genital tract. The American Cancer Society estimated that 42160 women have been diagnosed with, and 7780 women have died of cancer of the female genital tract in 2009 in the US[1]. In the endometrium different subtypes of cancer can develop. Endometrioid endometrial carcinoma (EEC), or Type 1 cancers, are oestrogen dependent, often develop in a background of atypical complex hyperplasia and account for over 75% of cases. EEC patients generally have a good prognosis, with low mortality for stage 1 disease (5-year survival around 87%)[2]. However, survival drops in higher stage of disease. Additional prognostic factors could help to decide the need for adjuvant treatment and to identify new treatment strategies.

Solid tumours outgrow their own vasculature beyond the size of several mm³, resulting in hypoxia. Regions of necrosis are believed to demarcate regions of severe, chronic hypoxia[3]. Hypoxia is an important issue in carcinogenesis because it renders a more aggressive phenotype with increased invasiveness and proliferation, formation of metastases and poorer survival[4‐8]. Besides, hypoxic malignant cells are more resistant to radiotherapy and chemotherapy[9‐11]. In reaction to hypoxia, cells will alter their metabolism and activate certain survival genes. Hypoxia inducible factor 1 (HIF-1) plays an essential role in the adaptive cellular response to hypoxia[12, 13]. HIF-1 is a transcription factor composed of the subunits HIF-1α and HIF-1β, which are basic helix-loop-helix DNA binding proteins. HIF-1β is constitutively expressed at the protein level. The activity of HIF-1 is predominantly regulated at the post-translational level by regulating HIF-1α protein stability. Under normoxia, HIF-1α is hydroxylated by prolyl hydroxylases in the oxygen dependent degradation domain. Hydroxylated HIF-1α is recognized by the Von Hippel Lindau protein, ubiquitinated, and destined for degradation by the proteasome. This process is inhibited during hypoxia[14], where stabilized HIF-1α heterodimerizes with HIF-1β to transactivate target genes after nuclear translocation by binding to the consensus Hypoxia Responsive Element (HRE) 5'-RCGTG-3' in promoters and enhancers[15]. Among these are growth factors, glucose transporters, glycolytic enzymes, and genes involved in gluconeogenesis, high-energy phosphate metabolism, erythropoiesis, haem metabolism, iron transport, vasomotor regulation and nitric oxide synthesis[13, 15‐17]. Protein products of these downstream genes help the cell to survive the hypoxic stress by increasing oxygen delivery (angiogenesis) and by switching to anaerobic glycolysis.

Previous breast cancer studies[6, 7] showed that in hypoxic conditions, HIF-1α expression is seen perinecrotically, with induction of its target genes carbonic anhydrase 9 (CAIX), that plays a role in pH regulation[18] and Glut-1, a transmembrane glucose transporter [19, 20]. In normoxic conditions, HIF-1α expression can also be induced by other mechanisms, but was to a much lesser extent associated with downstream activation. CAIX and Glut-1 thereby serve to identify functional HIF-1α[7]. In breast cancer, we showed that patients with a diffuse (non-hypoxia associated) HIF-1α staining pattern had a relatively better prognosis[7]. Elevated levels of CAIX are predictive of hypoxia in various types of cancer and are related to poor prognosis[21, 22]. The prognostic value of CAIX and Glut-1 has not been studied in endometrial cancer. We previously suggested that HIF-1α plays an important role in endometrial carcinogenesis[23]. In postmenopausal woman, HIF-1α was increasingly overexpressed from inactive endometrium through hyperplasia to endometrioid carcinoma, paralleled by activation of its downstream genes. HIF-1α overexpression has further been shown to be correlated with a poorer survival in cancers of the brain[24], cervix[25] and ovary[26]. As HIF-1α is related to poor clinical outcome in some tumours, HIF-1α expression could be used to identify patients who are at risk of developing recurrent disease and who may benefit from adjuvant therapy. Moreover, targeting HIF-1α could be an attractive therapeutic strategy with the potential for disrupting multiple pathways crucial for tumour growth. In endometrial cancer, there are some conflicting data concerning the prognostic relevance of HIF-1α, probably due to varying methodology and patient groups[27‐30]. Whether the different HIF-1α expression patterns have different prognostic implications in endometrial cancer is yet unknown.

In accordance with the need to decrease energy usage under low oxygen conditions, hypoxia induces cell-cycle arrest[31]. Cell-cycle arrest is regulated by complex interactions between cyclins, cyclin-dependent kinases (CDKs), and cyclin-dependent kinase inhibitors (CKIs)[32]. p27^kip1 is a CKI that regulates progression from G₁ to S phase by inhibiting a variety of cyclin-CDK complexes. It has been shown that p27^kip1 expression is strongly reduced in endometrial cancer[33, 34]. In a recent study on cell cycle regulation during endometrial carcinogenesis[23], p27^kip1 protein was re-expressed in necrotic (i.e. hypoxic) areas of endometrial carcinomas. We showed that p27^kip1 re-expression by hypoxia was HIF-1α-dependent and led to cell cycle arrest. This possibly contributes to survival of cancer cells in hypoxic parts of the tumour[35].

Several studies demonstrated that loss of the p27^kip1 protein, as assessed by immunohistochemistry, is a negative prognostic marker in malignancies[36, 37]. However, the prognostic importance of HIF-1α induced re-expression of p27^kip1 is not known. Loss of p27^kip1 expression in endometrial carcinoma did not seem to correlate with worse prognosis in previous studies[38, 39].

Understanding the mechanisms of carcinogenesis and progression of endometrial cancer is important as these insights might lead to improved diagnostic tools for the pathologist, improved prediction of prognosis and response to therapy, and eventually better biology-based disease management in the individual patient. The aim of this study was therefore to (1) re-evaluate the prognostic value of HIF-1α with emphasis on expression patterns and investigate the additional effect of p27^kip1 expression on predicting survival in EEC and (2) to determine relationships with the other clinicopathologic markers in the endometrioid type of endometrial cancer.

Hypoxia and its key regulator HIF-1α have been shown to play an important role in endometrial carcinogenesis[23], but contradictory results have been published as to the prognostic value of HIF-1α overexpression in endometrial carcinoma[27‐30], while expression patterns have been ignored. The aim of this study was therefore to re-evaluate the prognostic value of HIF-1α overexpression in a representative group of patients with endometrioid endometrial cancer, with emphasis on expression patterns. Also, as p27^kip1 is re-expressed in hypoxic regions of EEC[35], we investigated the additional effect of p27^kip1 expression on predicting survival in EEC. HIF-1α overexpression, especially the perinecrotic type, and necrosis appeared to be independent indicators of poor prognosis. High p27^kip1 (>50% positive cells) expression was an additional prognostic factor in the subgroup of patients with perinecrotic type of HIF-1α expression.

Our results on the prognostic value of HIF-1α overexpression are in line with those of Sivridis et al.[27] where HIF-1α was associated with a shorter overall survival in stage 1 endometrial cancer. However, others did not find a significant prognostic impact of HIF-1α overexpression[28‐30]. Immunohistochemical HIF-1α studies are difficult to compare because of a variation in the definition of HIF-1α positivity. Previous studies did not consider the different expression patterns throughout the tumours (diffuse versus perinecrotic) that have been shown in other cancers to be prognostically crucial[7]. Another difference between the present and previous studies is the cut off value for HIF-1α expression. In the present study, the cut off value for prognostic value of HIF-1α was 35%, much higher than in other studies where the cut off varied between 1% and 5%[6, 41]. Our results indicate that also in endometrial cancer, the pattern of HIF-1α expression is more important for the prognosis than percentage and intensity of HIF-1α expressing cells in general. This significance of expression pattern could be explained by the fact that perinecrotic HIF-1α expression is thought to be hypoxia driven, whereas diffuse HIF-1α expression may rather be due to non-hypoxic stimuli[6, 7]. In a previous study from our group[23] we showed that diffuse HIF-1α expression was associated with the highest microvessel density (MVD); perinecrotic and mixed patterns were associated with an intermediate MVD (p < 0.05). This indicates that diffuse HIF-1α expression is more likely to be due to non-hypoxic stimuli than to non-necrosis associated hypoxia. We showed that perinecrotic HIF-1α expression is more often accompanied by activation of its downstream factors Glut-1 and CAIX, indicating it to be more active than diffuse HIF-1α. More activation of HIF-1α and its target genes would give more tumour cells a survival advantage in a hypoxic environment. Perinecrotic HIF-1α expression retained its prognostic significance in the subgroup of low stage, while significance was lost in stage III/IV patients, indicating that HIF-1α expression seems to be especially important in low stage patients. However, sample size of the subgroups (especially the high stages) was limiting. Therefore, no definitive conclusions on the prognostic value of HIF-1α in high stage patients can be made. Loss of p27^kip1 expression has been reported for a number of human tumour types and has been correlated with poor prognosis and tumour aggressiveness[36, 42]. Only one study showed a correlation between p27^kip1 expression and prognosis in endometrial cancer[43]. However, we found no prognostic influence of global p27^kip1 expression in endometrioid endometrial carcinoma. This is in line with the outcome of others[38, 39, 44]. Since we found hypoxia to be able to induce re-expression of p27^kip1 in a HIF-1α dependent way in a previous study[35], we evaluated the prognostic value of p27^kip1 in relation to HIF-1α in the present study. In the subgroup of patients with perinecrotic HIF-1α expression, high p27^kip1 expression (>50% positive cells) was associated with a shorter DFS and OS. This fits with our previously proposed model that hypoxia induced re-expression of p27^kip1 may result in dormancy of hypoxic cells, which in combination with HIF-1α induced expression of genes regulating supply of energy, growth factors and other survival factors, may promote cellular survival and adaptation of sub clones within the tumour that may contribute to metastatic disease and poor clinical outcome. In our previous study however, only central/perinecrotic p27^kip1 expression was associated with HIF-1α induced re-expression[35]. In the present study, central/perinecrotic p27^kip1 expression was still associated with perinecrotic HIF-1α expression; however this subgroup was too small to correlate with survival data.

Interestingly, all patients lacking tumour necrosis survived, and none of these patients developed a locoregional or distant recurrence. The prognostic value of necrosis is in concordance with previous results by Scholten et al.[45] who showed that necrosis is a significant prognostic factor in Stage I-III endometrioid endometrial carcinoma. Although necrosis is a strong independent prognostic factor for endometrial carcinoma, its clinical use may be limited because of a moderate reproducibility[45‐47]. Further, excluding the presence of necrosis requires adequate sampling. More objective biomarkers like perinecrotic HIF-1α expression could result in a higher reproducibility.

We conclude that in endometrioid endometrial carcinoma, absence of necrosis is a favourable prognostic factor. Tumours with a perinecrotic HIF-1 expression are more aggressive than tumours with purely diffuse HIF-1 expression. Furthermore, the combination of perinecrotic HIF expression and high p27^kip1 expression is indicative for an unfavourable prognosis. More aggressive adjuvant treatment might be necessary to improve the outcome of these patients. To this end, targeting the HIF pathway might provide an attractive strategy to treat hypoxic and highly angiogenic tumours. Thus far, selective HIF-1 inhibitors have not been identified. A number of non-selective inhibitors, which indirectly target signaling pathways upstream or downstream of HIF-1 are known to decrease the key regulating HIF-1α protein levels. Preliminary results show encouraging results for these single agent treatments, however severe toxicities have been encountered. Lack of specificity increases the difficulty in attributing any anti-tumorigenic effects of these drugs specifically to inhibition of HIF-1. Further studies are needed to show whether adaptation of therapy protocols might be of benefit for endometrial cancer patients with perinecrotic HIF-1α overexpression.