Association of hypoxia-inducible factors 1α and 2α with activated angiogenic pathways and prognosis in patients with endometrial carcinoma

Authors

  • Efthimios Sivridis Ph.D.,

    1. Departments of Pathology and Radiotherapy/Oncology, Democritus University of Thrace, Alexandroupolis, Greece
    Search for more papers by this author
  • Alexandra Giatromanolaki M.D.,

    1. Departments of Pathology and Radiotherapy/Oncology, Democritus University of Thrace, Alexandroupolis, Greece
    Search for more papers by this author
  • Kevin C. Gatter D.Phil.,

    1. Nuffield Department of Clinical Laboratory Sciences, John Radcliffe Hospital, Oxford, United Kingdom
    Search for more papers by this author
  • Adrian L. Harris D.Phil.,

    1. Imperial Cancer Research Fund, Molecular Oncology Laboratories, Institute of Molecular Medicine, John Radcliffe Hospital, Oxford, United Kingdom
    Search for more papers by this author
  • Michael I. Koukourakis M.D.,

    Corresponding author
    1. Departments of Pathology and Radiotherapy/Oncology, Democritus University of Thrace, Alexandroupolis, Greece
    • Tumor and Angiogenesis Research Group, P.O. Box 12, Alexandroupolis 68100, Greece
    Search for more papers by this author
    • Fax: +30-551-31522

  • for the Tumor and Angiogenesis Research Group


Abstract

BACKGROUND

Hypoxia-inducible factor 1α (HIF-1α) and HIF-2α are essential regulatory proteins for the adaptation of tumor cells to hypoxia, and they stimulate angiogenesis through activation of the vascular endothelial growth factor (VEGF) gene.

METHODS

HIF-1α and HIF-2α proteins were studied immunohistochemically in a group of 81 patients with Stage I endometrial adenocarcinoma of the endometrioid cell type. The results were correlated with intratumoral angiogenesis, the expression of the angiogenic factors VEGF and thymidine phosphorylase (TP), and the VEGF/receptor (VEGF/KDR) complex. Relations also were sought with estrogen receptor (ER) and progesterone receptor (PR), with the apoptosis-related proteins bcl-2 and p53, with several histopathologic parameters, and with patient prognosis. In addition, a sample of 25 normal endometria at various phases of the menstrual cycle was studied for the presence of HIF-1α and HIF-2α.

RESULTS

HIF-1α expression was detected in 49% of endometrial carcinomas. The expression was cytoplasmic or mixed nuclear/cytoplasmic. HIF-1α expression was associated with up-regulation of the VEGF pathway and with increased standard microvessel density (sMVD) and activated VEGF/KDR microvessel density (aMVD). It also was associated with a poor prognosis in both univariate and multivariate analyses. HIF-2α protein showed a pattern of expression similar to the pattern seen in HIF-1α, but expression of HIF-2α protein occurred in only 17% of endometrial carcinomas, and it was associated with increased TP reactivity. There also was a relation of HIF-1α expression with well-differentiated endometrial neoplasms, and there was a marginal association of HIF-1α and HIF-2α with ER expression. With reference to normally cycling tissues, HIF-1α nuclear/cytoplasmic expression was particularly strong in the samples of early proliferative phase endometrium compared with HIF-2α protein expression, which showed a constant reaction throughout the menstrual cycle.

CONCLUSIONS

The up-regulation of HIF-1α and, to a lesser extent, of HIF-2α is a common event in Stage I endometrial adenocarcinomas. In these tumors, HIF-1α expression is related to increased angiogenesis, through activation of the VEGF angiogenic pathway, and to an unfavorable prognosis. HIF-2α accumulation is associated with increased expression of the angiogenic factor TP. Cancer 2002;95:1055–63. © 2002 American Cancer Society.

DOI 10.1002/cncr.10774

Hypoxia-inducible factor 1α (HIF-1α) and HIF-2α are important regulatory proteins of cellular response to hypoxic stimuli.1, 2 Low levels (or, rather, constant levels) of both molecules are maintained intracellularly as a result of their continuous degradation by the proteasome pathway, an oxygen dependent function.3 Under hypoxic conditions, degradation is inhibited, and HIF-α proteins are accumulated. After heterodimerization with the protein HIF-1β, a nuclear receptor translocator that is expressed constitutively and remains unaffected by hypoxia, HIF-α binds to DNA at the hypoxia response elements (HREs),4 activating the vascular endothelial growth factor (VEGF) gene.

The recent development of monoclonal antibodies that recognize HIF-1α and HIF-2α widen the field of translational research. Because hypoxia is a major obstacle to tumor radiotherapy and often in chemotherapy, and because tumor cell invasion and migration are reinforced by hypoxic stimuli, the immunohistochemical assessment of HIF-α emerges as a promising tool to predict the outcome of patients after surgery and radiotherapy. Indeed, recent studies have related the expression of HIF-1α or HIF-2α with resistance to radiotherapy and with a poor postoperative outcome.5, 6 The up-regulation of certain angiogenic pathways also has been described in HIF-α-expressing tumors, such as nonsmall cell carcinomas of lung.7

In the current study, we investigated the expression of HIF-1α and HIF-2α in a homogenous group of 81 patients with Stage I endometrial adenocarcinomas of the endometrioid cell type. The results were correlated with intratumoral angiogenesis, with the expression of the angiogenic factors VEGF and thymidine phosphorylase (TP), and with the expression of the ligand/receptor complex VEGF/KDR. Furthermore, the expression levels of HIF-1α and HIF-2α were related with the steroid receptors estrogen receptor (ER) and progesterone receptor (PR), with the apoptosis-related proteins bcl-2 and p53, and with several histologic parameters and prognosis.

MATERIALS AND METHODS

Formalin fixed, paraffin embedded tissues from 25 normal endometria in various phases of the menstrual cycle and 81 endometrial adenocarcinomas of the endometrioid cell type were retrieved from the files of the Department of Pathology, Democritus University of Thrace. The histologically normal endometria were obtained from premenopausal women, age 34–40 years, who had undergone hysterectomy for nonendometrial disease, most frequently uterine leiomyomas. In all patients, the date of the last menstrual period was available. None of the patients had received hormone therapy. The normal endometria had been dated on hematoxylin and eosin-stained sections by using the histologic criteria of Noyes et al.8

All patients with endometrial carcinoma underwent surgery with total abdominal hysterectomy and bilateral salpingo-oophorectomy. Whole pelvic radiotherapy followed by a vaginal intracavitary booster radiation dose, was given to all patients with Stage I disease with deep (> 50%) myometrial invasion. No lymph node sampling of the iliac lymph nodes was performed, and staging was based on preoperative lymphangiography or pelvic and abdominal computed tomography (CT) scans. Histologic typing and grading of the endometrial tumors (Grade 1 vs. Grade 2–3) and the depth of myometrial invasion were assessed on hematoxylin and eosin-stained sections using standard criteria.9, 10 Lymphatic-vascular space invasion was recorded as present if tumor cells were seen within a space with a definite and clearly identifiable endothelial lining. The follow-up of patients who remained alive ranged from 22 months to 182 months, with a median of 70 months. At the time of analysis, 10 patients (12.3%) were dead with disease.

Immunohistochemistry for HIF-1α and HIF-2α Expression

The HIF-1α and HIF-2α proteins were detected using the ESEE 122 immunoglobulin (Ig) G1 (IgG1) monoclonal antibody (dilution, 1:20) and the EP190b IgG1 monoclonal antibody (neat), as described previously.11 Sections were deparaffinized, and peroxidase was quenched with methanol and H2O2 3% for 15 minutes. Microwaving for antigen retrieval was used (three times for 5 minutes each). The primary antibodies were applied for 90 minutes. After washing with Tris-buffered saline (TBS), sections were incubated with a secondary antirabbit-antimouse antibody (Kwik Biotinylated secondary antibody 0.69A; Shandon-Upshaw) for 15 minutes and washed in TBS. Kwik Streptavidin peroxidase reagent (039A; Shandon-Upshaw) was applied for 15 minutes, and sections were washed again in TBS. The color was developed by a 15-minute incubation with 3,3′diaminobenzidine tetrahydrochloride solution, and sections were counterstained weakly with hematoxylin.

Breast carcinoma tissue sections with strong nuclear HIF-1α and HIF-2α expression were used as positive controls. Normal mouse IgG was substituted for primary antibody at the same concentration as a negative control.

The assessment of HIF-α expression was based on previously described guidelines.11 HIF-1α and HIF-2α expression was both cytoplasmic and nuclear. Cytoplasmic staining was scored as absent, weak, or strong, and nuclear expression, when present, was accompanied by varying degrees of cytoplasmic reactivity, although pure nuclear expression was observed occasionally. The extent of staining also varied among tumors. The percentages of tumor cells with strong HIF-α cytoplasmic and/or nuclear reactivity were recorded in all optical fields. Tumors also were scored according to a two-scale system: high reactivity, denoting tumors with strong cytoplasmic and/or nuclear reactivity, and low reactivity, denoting tumors with absent or weak cytoplasmic reactivity and an absence of nuclear reactivity. In this way, the assessment of HIF association with other parameters was performed using HIF as a categorical variable (low reactivity vs. high reactivity) and as a continuous variable (the percentage of tumor cells with strong HIF expression within a sample).

Immunohistochemistry for Other Antigens

Sections were cut at 3 μm and were stained immunohistochemically with the following techniques: 1) a standard streptavidin-biotin method for the detection of ER, PR, p53, and bcl-211–13 and 2) the alkaline phosphatase/antialkaline phosphatase method for microvessel staining and for the detection of TP, VEGF, and the VEGF/KDR complex.14–17 Details of the primary antibodies, the working dilutions, and the antigen retrieval methods used are given in Table 1.

Table 1. Details of the Antibodies, Dilutions, and Antigen-Retrieval Methods Used in this Study
Primary antibodyDilution/incubation timeAntigen retrievalSpecificitySourceReference
  • MW: microwave heating; HJF: hypoxia-inducible-factor; VEGF: vascular endothelial growth factor; VEGF/KDR. VEGF ligand-receptor complex; TP: thymidine phosphorylase; ER: estrogen receptor; PR: progesterone receptor.

  • a

    At room temperature.

ESEE 1221:20 (90 minutes)aMWHIF-1αOxford University Cambridge, UKTalks et al.11
EP 190bNeat (90 minutes)aMWHIF-2αOxford UniversityTalks et al.11
JC70 (CD31)1:50 (30 minutes)aProteaseEndotheliumDako, Glostrope DenmarkGiatromanolaki et al.16
VG11:4 (90 minutes)aMWVEGFOxford UniversityGiatromanolaki et al.13
11B51:3 (60 minutes)aMWVEGF/KDRTexas UniversityTarley et al.14
P-GF.44CNeat (30 minutes)aTPOxford UniversityGiatromanolaki et al.16
ID51:20 (75 minutes)aMWERImmunon, Shandon, PASivridis et al.12
IA61:20 (75 minutes)aMWPRImmunonSivridis et al.12
DO-71:30 overnight at 4 °CMWp53DakoFox et al.17
1241:80 overnight at 4 °CMWbcl-2DakoFox et al.17

Known positive controls were included in each staining run. Omission of the primary antibody and substitution by nonspecific Ig at the same concentration were used as negative controls.

Staining Patterns and Evaluation

Positivity was indicated 1) as a distinct, brown, nuclear staining for ER, PR, endothelial cells, and p53; 2) as a cytoplasmic or mixed nuclear/cytoplasmic staining for TP; and 3) as cytoplasmic staining for VEGF, the VEGF/KDR complex, and bcl-2. The assessment of the immunohistochemically stained sections was based on the following guidelines: The percentage of tumor cells that expressed the various antigens under investigation was assessed semiquantitatively at ×200 magnification. The counts were performed on the entire tumor area. Thus, by using appropriate cut-off points, the molecular reactivity of each tumor could be typed as low or high and as negative or positive, and continuous variable analysis became possible.

The median value was used as a cut-off point to define tumors with high and low reactivity for VEGF and TP expression. However, as reported previously, TP reacts not only with tumor cells but also with stromal fibroblasts and myometrial cells confronting the invading tumor edge.18 At this site specifically, TP stromal reactivity was considered as high if > 50% of stromal/myometrial cells showed a strong staining reaction. A cut-off point of 10% was used to define groups with high or low ER, PR, bcl-2, and p53 reactivity, as indicated previously.13, 19

Tumor angiogenesis was assessed by microvessel counting. The procedure was similar for both standard microvessel density (sMVD) and activated microvessel density (aMVD; i.e., expressing the VEGF/KDR complex).16 Three areas of high vascular density (hot spots) were selected at the invading tumor front; the final microvessel score was the mean of the vessel counts obtained from these fields. Only blood vessels with a clearly defined lumen or a linear vessel shape, but not single endothelial cells, were taken into account.

In all tumors, assessment was performed independently by two pathologists (A.G. and E.S.) who were blinded to the patient data. Discrepancies were resolved on the conference microscope.

STATISTICAL ANALYSIS

Statistical analysis was carried out and graphs were constructed using the Instat 3.1 and GraphPad Prism 2.01 software packages, respectively (GraphPad Software Inc.). A Fisher exact test or an unpaired, two-tailed t test was used for testing correlations between noncontinuous categorical tumor variables (contingency tables) and continuous categorical tumor variables (comparison of the mean values from two sets of data), respectively. Linear regression analysis was used to assess correlation between continuous variables. Survival curves were plotted using the method of Kaplan and Meier, and the log-rank test was used to determine statistical differences between life tables. The end points were overall survival from the day of surgery. A Cox proportional hazards model was used to assess the effect of tumor variables on overall survival. A P value < 0.05 was considered significant.

RESULTS

Expression of HIF-α in Normal Endometrium

HIF-1α was expressed strongly in the nuclei and cytoplasm of both glandular and stromal endometrial cells (Fig. 1) in early-phase proliferative endometrium. In middle-to-late phase proliferative endometrium and in endometrium during the secretory phase of the menstrual cycle, a progressive reduction in HIF-1α expression was noted. In contrast, HIF-2α protein was expressed strongly in the cytoplasm and nuclei of the endometrial glandular and stromal cells throughout the menstrual cycle.

Figure 1.

Normal proliferative endometrium showing mixed expression of nuclear/cytoplasmic hypoxia-inducible factor 1α in glandular and stromal endometrial cells.

Expression of HIF-α in Endometrial Tumors

The expression of HIF-α in endometrial tumor cells was a mixture of nuclear and cytoplasmic expression (Fig. 2A,B). The intensity of cytoplasmic staining ranged from negative through weak to strong, whereas the extent varied from 0% to 100% of the total neoplastic cell population. Nuclear expression invariably was accompanied by strong cytoplasmic reactivity. The median and mean percentages of tumor cells that had strong HIF-1α reactivity were 20% and 35%, respectively. The median and mean percentages of cells with strong HIF-2α reactivity were 0% and 10%, respectively.

Figure 2.

Endometrial carcinomas showing mixed nuclear/cytoplasmic expression of hypoxia-inducible factor 1α (HIF-1α) (A) and HIF-2α (B).

Strong HIF-1α cytoplasmic reactivity, with or without nuclear reactivity, was noted in 40 of 81 tumors (49%), and these were considered to have high HIF-1α reactivity. Strong HIF-2α cytoplasmic reactivity, with or without nuclear reactivity, was noted in 14 of 81 tumors (17%), and these were considered to have high HIF-2α reactivity. Occasionally, strong HIF-α reactivity was noted in stromal fibroblasts or endothelial cells at the invading tumor front or in inner tumor areas. Strong HIF-α expression also was noted in myometrial vessels, whereas smooth muscle cells and fibroblasts were negative for HIF-α expression.

Correlation of HIF-α with Histopathologic Variables

High HIF-1α reactivity was associated significantly with low-grade tumors. Thirty-seven of 40 tumors (92%) that expressed high HIF-1α reactivity were Grade 1 tumors compared with 28 of 41 tumors (68%) that were Grade 2 and 3 (P = 0.01). There was no association of HIF-1α reactivity with the depth of myometrial invasion or with invasion of the lymphovascular spaces. Similarly, no association was noted between HIF-2α expression and the histologic variables studied (P > 0.38; data not shown).

Correlation of HIF-α with Angiogenesis

Using linear regression analysis, a significant correlation of HIF1α with sMVD (P = 0.03; correlation coefficient r2 = 0.06), aMVD (P = 0.006; r2 = 0.09), and VEGF expression (P = 0.002; r2 = 0.12) was noted. HIF-2α expression was related significantly to TP expression (P = 0.004; r2 = 0.10). A schematic representation of HIF-1α linear regression analysis with VEGF and aMVD is shown in Figure 3.

Figure 3.

Linear regression analysis of hypoxia-inducible factor 1α (HIF1α) expression with vascular endothelial growth factor (VEGF) (A) and activated microvessel density (aMVD) (B). r2: correlation coefficient.

Table 2 shows the association of HIF-α expression with angiogenesis parameters used as categorical variables. HIF-1α was associated significantly with high VEGF expression and high aMVD, whereas HIF-2α expression was linked with TP expression in tumor cells. No association of HIF-α with stromal TP expression was noted.

Table 2. Association of the Expression of Hypoxia-Inducible Factors 1α and 2α with Angiogenesis Parameters
ParameterHIF-1αHIF-2α
LowHighP valueLowHighP value
  1. HIF: hypoxia-inducible factor; sMVD: standard microvessel density; aMVD: activated microvessel density; VEGF: vascular endothelial growth factor; TP: thymidine phosphorylase.

sMVD      
 Low24170.183150.24
 High1723269
aMVD      
 Low30180.013580.99
 High1122276
VEGF      
 Low30160.0033560.38
 High1124278
TP (tumor cells)      
 Low29290.435170.01
 High711117
TP (stromal cells)      
 Low28280.604790.50
 High812155

Using the median sMVD as a cut-off point, there was no significant association of HIF-α with the degree of intratumoral vasculature. Figure 4A,B shows the percentage of tumors with strong HIF-α reactivity according to the MVD (MVD < 10, 11–20, 21–30, 31–40, and > 40). This rose as the MVD increased, whereas a P value of marginal significance (P = 0.07) was noted using a cut-off point of 30 microvessels per ×200 optical field.

Figure 4.

Correlation between microvessel density and reactivity for hypoxia-inducible factor 1α (HIF1a) (A) and HIF2a (B) in patients with endometrial carcinoma.

Correlation of HIF-α with Other Molecular Variables

A marginal association of HIF-1α and HIF-2α expression with ER positivity was observed (P = 0.06). Bcl-2 expression was somewhat more frequent in tumors that lacked HIF-2α reactivity, but neither this protein nor p53 nuclear accumulation was associated significantly with HIF-α expression (Table 3).

Table 3. Association of the expression of Hypoxia-Inducible Factors 1α and 2α with Molecular Variables
ParameterHIF-1αHIF-2α
LowHighP valueLowHighP value
  1. HIF: hypoxia-inducible factor; ER: estrogen receptor; PR: progesterone receptor.

ER      
 Negative36280.065680.06
 Positive512116
PR      
 Negative36290.1055100.46
 Positive511124
p53      
 Negative38370.9961140.57
 Positive3360
bcl-2      
 Negative24170.183150.24
 Positive1723269

Survival Analysis

The Kaplan–Meier survival curves stratified for HIF-1α and HIF-2α are shown in Figure 5A,B. There was a significant association between HIF-1α over-expression and a poor prognosis (P = 0.03).

Figure 5.

Kaplan–Meier survival curves stratified for hypoxia-inducible factor 1α (HIF-1α) reactivity (A) and HIF-2α reactivity (B) in patients (pts) with endometrial carcinoma.

To investigate the prognostic significance of HIF-α relative to standard histopathologic variables, the proteins HIF-1α and HIF-2α (separately and combined) were included in multivariate models together with the histologic grade, the depth of myometrial invasion, and the presence or absence of vascular invasion (Table 4). HIF-1α emerged as an independent prognostic factor, and HIF-1α maintained this property after the inclusion of postoperative radiotherapy in the model (data not shown). In a multivariate model that included all parameters from the current study (MVD and angiogenic factors), HIF-α was of no independent prognostic value (data not shown).

Table 4. Multvariate Analysis of Hypoxia-Inducible Factors 1α and 2α with Standard Histopathologic Variables
ParameterModel 1Model 2Model 3
P valueT ratioP valueT ratioP valueT ratio
  1. HIF: hypoxia-inducible factor.

HIF-1α0.012.50.0082.7
HIF-2α0.540.60.251.1
Histology grade0.141.40.121.50.540.6
Depth of invasion0.191.30.201.20.430.8
Vascular invasion0.261.10.251.10.510.6

DISCUSSION

The over-expression of HIF-1α and HIF-2α is a marker of hypoxia because, their intracellular degradation and, thus, their stabilization by the proteasome pathway, are oxygen dependent.3 In some circumstances, HIF accumulation may occur as a result of genetic alterations, such as loss or simple decrease of von Hippell–Lindau protein,20 a molecule that also is essential for HIF degradation. Other known causes of increased HIF-α transcriptional activity include activation of the mitogen-activated protein kinase21 and insulin-like growth factor 1 pathway.22, 23 In any case, the activated HIF pathway triggers biologic events that are associated intimately with aggressive tumor behavior. Thus, angiogenesis and glycolysis (two of the most important processes facilitating tumor cell survival, proliferation, and progression) are regulated directly by HIF-α binding to HREs of relevant genes. Furthermore, HIF-α-expressing tumors are expected to be resistant to radiation therapy, because these factors reflect tumor hypoxia and an enhanced transcription of proteins that favors tumor cell survival.24–26

The finding that HIF-α is over-expressed in response to cellular hypoxia was confirmed recently in clinicopathologic studies that used simple immunohistochemical techniques on paraffin embedded material.11, 27 Thus, for example, we showed that expression of HIF-α may be the driving force in up-regulating angiogenic pathways in nonsmall cell lung carcinomas.7 In those tumors, there was a strong association of HIF accumulation with MVD and with the expression of multiple angiogenic factors, including VEGF, TP, basic fibroblast growth factor (bFGF), and the angiogenic factor receptors KDR and bek-bFGF. The association of HIF-α accumulation with a poor postoperative outcome was confirmed in patients with cervical carcinoma and nonsmall cell lung carcinoma.7, 28 The predictive role of HIF-1α in response to radiotherapy was demonstrated in patients with oropharyngeal carcinoma.5 We also demonstrated a role for HIF-1α in defining resistance of early esophageal carcinoma to radiotherapy and photodynamic therapy.6

In the current study, several of these observations were extended to include patients with Stage I endometrioid carcinoma. Thus, HIF-1α over-expression was associated with up-regulation of the VEGF pathway, with increased MVD, and with a poor prognosis. HIF-2α was linked to overexpression of the angiogenic factor TP.

In addition, HIF-1α over-expression was related significantly with well-differentiated endometrial neoplasms and with a rich ER status, although the latter relation was only marginal. A similar association between HIF-1α and ER also was noted in patients with breast carcinoma.29 Several other growth factors and growth factor receptors, such as epidermal growth factor and insulin-like growth factor, up-regulate HIF-1α protein expression.30, 31 It is possible that ER-stimulated pathways induce HIF-1α expression, and it has been shown that estradiol stimulates VEGF in ER positive smooth muscle cell lines.32

The finding that HIF-1α frequently occurs in Grade 1 tumors indicates that even highly differentiated, ER positive endometrial tumors may be hypoxic, like apparently normal endometrial tissues. This is particularly evident in the early proliferative phase of the menstrual cycle, when HIF-1α over-expression is connected with new blood vessel formation and regeneration of the denuded endometrium, apparently through the VEGF angiogenic pathway. Restoration of the vascular supply results in reoxygenation of the tissue and reduction of the HIF-1α protein, a condition that lasts up to the late secretory phase. At the same time, the consistent and even expression of HIF-2α protein throughout the normal menstrual cycle is indicative of a different mechanism.

Furthermore, the association of good differentiation with HIF-1α but a poor prognosis for patients with HIF shows the potential importance of new markers in detecting heterogeneous behavior patterns in groups of patients classified by conventional criteria. These patients also may be resistant to radiation or may be candidates for therapy with hypoxic-activated drugs, such as tirapazamine and cisplatinum combinations.

With reference to the apoptosis-related proteins p53 and bcl-2, previous studies connected HIF-α accumulation directly with the mutant protein p53 and inversely with the protein bcl-2.7, 27 A trend for the latter relation was noted in the current study but did not reach the level of statistical significance. The similar absence of any apparent association between HIF-α and p53 may be a consequence of the very low rate of p53 mutations in the early stage of endometrioid carcinoma.13

The results showing that accumulated HIF-1α is associated with a poor prognosis help to elucidate the pathways that up-regulate angiogenesis: Our previous studies in patients with endometrial carcinoma demonstrated that high MVD, VEGF-activated angiogenic pathways, and TP expression were important prognostic factors.16, 33, 34 Our current studies suggest that HIF accumulation is a key step in this process. The fact that HIF-1α loses its independent prognostic value after the inclusion of angiogenic factors in the model suggests that HIF-α influences tumor behavior by triggering the angiogenic cascade. Such an initiative may be suppressed by endogenous inhibitors or may be accentuated further by HIF independent pathways, reducing the prognostic significance of HIFs. It should be clarified, however, that lymph node staging based on CT scan, as in our series, cannot replace surgical sampling of lymph nodes and, in these patients, the results of prognostic analysis should be interpreted with caution.

An interesting observation that emerged from this study was the occasional expression of HIF-α by the tumoral vasculature. This is analogous to our previous observation in nonsmall cell lung carcinomas.7 It is true that, under hypoxic conditions, both proteins can be up-regulated by normal cells, and HIF-2α (endothelial PAS domain protein-1), in particular, was described originally in the normal endothelium.35 In this context, hypoxia-mediated up-regulation of HIF-α is not improbable despite the immediate contact of vessels to erythrocytes. The blood flow in tumors is prone to circulatory stasis and, hence, oxygen depletion, as a result of vascular collapse after increased interstitial pressure and compression by the proliferating malignant cells.36, 37

It is concluded that HIF-α over-expression, as assessed by simple immunohistochemical techniques, is a frequent event in Stage I endometrioid carcinomas. HIF-1α, in particular, is related directly to angiogenic pathways and to a poor prognosis. The cyclic expression of a HIF-1α and the constitutive expression of HIF-2α demonstrated during the normal menstrual cycle suggest an important role for HIF-α in endometrial physiology.

Ancillary