Article Text

Download PDFPDF

Hypoxia-inducible factor-1α expression correlates with focal macrophage infiltration, angiogenesis and unfavourable prognosis in urothelial carcinoma
  1. C-Y Chai1,
  2. W-T Chen2,
  3. W-C Hung3,
  4. W-Y Kang1,2,
  5. Y-C Huang4,
  6. Y-C Su2,
  7. C-H Yang2
  1. 1
    Department of Pathology, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan
  2. 2
    Department of Pathology, Kaohsiung Medical University Chung-Ho Memorial Hospital, Kaohsiung, Taiwan
  3. 3
    Institute of Biomedical Sciences, National Sun Yat-Sen University, Kaohsiung, Taiwan
  4. 4
    Graduate Institute of Medicine, College of Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan
  1. C-Y Chai, Department of Pathology, Kaohsiung Medical University Chung-Ho Memorial Hospital, No. 100, Tzyou 1st Road, Kaohsiung City 807, Taiwan; cychai{at}


Background: Hypoxia inducible factor (HIF)-1α is a critical regulatory protein of cellular response to hypoxia and is closely related to angiogenic process.

Aims: To explore the potential role and the prognostic value of HIF-1α in urothelial carcinoma (UC).

Methods: Clinicopathological and follow-up data on 99 UC cases were reviewed and immunostained for HIF-1α, CD68, vascular endothelial growth factor (VEGF) and CD34 antigen. Tumour-associated macrophage (TAM) counts and HIF-1α expression were compared with clinicopathologic characteristics, overall survival (OS) and disease-free survival rates (DFS).

Results: High expression of HIF-1α was detected in 55 of 99 (55.6%) tumours. HIF-1α expression was correlated with tumour size, histological grade, tumour invasiveness and recurrence. VEGF and microvessel density (MVD) demonstrated their positive correlation with HIF-1α overexpression, supporting the correlation of HIF-1α up-regulation with tumour angiogenesis. Higher TAM infiltration was identified in high expression of HIF-1α cases rather than HIF-1α low expression cases (p = 0.002). Kaplan–Meier analysis found that HIF-1α overexpression and high TAM count was only associated with worse DFS (p = 0.009, p = 0.023) but was not associated with OS (p = 0.696, p = 0.141). Multivariate analyses indicated only tumour size (p = 0.038) to be an independently significant prognostic factor for OS, in addition, HIF-1α expression (p = 0.011), as well as histological grade (p = 0.038), and MVD (p = 0.004), to be independently significant prognostic factors for DFS.

Conclusions: Our results indicate that HIF-1α is a key regulator of the angiogenic cascade. We show that HIF-1α is an independent prognostic factor for disease-free survival.

Statistics from

Angiogenesis is essential for the growth, invasion, and metastasis of tumours.1 Tumours cannot grow in the absence of angiogenesis because of the lack of oxygen in the centre of the tumour, which results in apoptosis and necrosis. Regions of hypoxia are known to exist within many tumours, and are associated with resistance to chemotherapy and radiotherapy, poor prognosis, malignant progression, local invasion and distant metastasis as well as a more malignant phenotype.2 3 Hypoxia-inducible factor-1 (HIF-1) is a major regulator of cell adaptation to hypoxic stress and plays a critical role in tumourigenesis and angiogenesis.4 5 It is a heterodimer composed of two subunits: HIF-1α and HIF-1β. HIF-1α is the oxygen-regulated subunit that determines HIF-1 activity.6 Under non-hypoxic conditions, HIF-1α is subject to ubiquitination and proteasomal degradation. Thus, under hypoxic conditions, HIF-1 transcriptional activity increases rapidly due to HIF-1α protein overexpression.6 HIF-1 regulates many other genes that are involved in many aspects of tumour progression, including metabolic adaptation, apoptosis resistance, angiogenesis and erythropoiesis,7 which comprise the most important processes facilitating tumour cell survival, proliferation, progression and distant metastasis; these process can also explain the association of activated HIF-1α pathway with aggressive tumour behaviour and poor prognosis.8

Monocytes are recruited into tumours from the circulation along defined chemotactic gradients, and they then differentiate into tumour-associated macrophages (TAMs) and then accumulate in avascular and necrotic areas, where they are exposed to tumour hypoxia.9 A number of recent studies have shown that macrophages respond to the levels of hypoxia found in tumours by up-regulating such transcription factors as HIF-1 and -2, which in turn activate mitogenic, pro-invasive, proangiogenic, and prometastatic genes.10 11 Interestingly, recent studies have discovered that TAM accumulation in low oxygen conditions has been associated with angiogenesis and with the production of angiogenic factors such as vascular endothelial growth factor (VEGF), tumour necrosis factor (TNF)α, basic fibrobalst growth factor (bFGF) and CXCL8 (also known as human interleukin-8) in human cancers,12 consequently having unfavourable clinical outcomes.13 It is possible that, once macrophages reach hypoxic tumour sites, they may promote tumour growth and metastasis by releasing these factors to stimulate angiogenesis. However, to the best of our knowledge, HIF-1α expression associated with TAM infiltration during hypoxia has not been previously reported in urothelial carcinoma (UC). In this study, we intend to ascertain whether HIF-1α expression is correlated with angiogenesis through focal TAM infiltration or involved in TAM angiogenic activation by correlating its expression with tumour microvessel density (MVD) as a marker of angiogenesis in human UCs. We propose that HIF-1α may induce TAMs aggregation and the release of cytokines and angiogenetic factors, resulting in angiogenesis as TAMs also up-regulate VEGF and other proangiogenic factors in response to hypoxia. Our results would be helpful in clarifying the functional correlation between HIF-1α and TAM in tumour angiogenesis in this tumour during hypoxia. HIF-1α expression is related to the patient’s prognosis in lung, prostate and breast carcinoma.5 The traditional criteria of histological grade, stage and tumour size cannot accurately predict the probability of recurrence or progression of UCs. If the prognosis of an individual patient could be predicted earlier in the course of the disease, those with a high likelihood of recurrence or progression could be treated more aggressively. Hypoxia is common in all solid tumours and is also associated with radiation resistance and poor prognosis.14 Therefore, it is worth determining whether HIF-1α is an accurate and reliable prognostic factor for UC.


A total of 99 specimens of UC were obtained from the archives of the Department of Pathology, Kaohsiung Medical University Chung-Ho Memorial Hospital, Taiwan between 1995 and 2000. Subjects were 42 men and 57 women with a mean (SE) age of 64.5 (1.31) years (median 67, range 21–83). The median size of tumour was 3.0 cm (range 0.5–12.0 cm). Haematoxylin and eosin (H&E)-stained sections from each tumour sample were examined by an experienced pathologist to confirm histological diagnosis and assess tumour content. Histological diagnoses including the tumour invasiveness and histologic grade were according to the guidelines proposed by the World Health Organization.15 Clinical profiles were all obtained from Kaohsiung Medical University Chung-Ho Memorial Hospital medical records for analysis. Follow-up was available for 99 cases after surgical resection. During this period, 26 cases developed local recurrences, 13 had distant metastases and there were 12 disease-related deaths.

Immunohistochemical staining

Sections 4 μm thick, serially cut from formalin-fixed and paraffin-embedded tissue blocks, were de-paraffinised, rehydrated and autoclave-treated at 121°C for 10 min in 0.1 M citrate buffer (pH 6.0) to induce antigen retrieval. Endogenous peroxidase in the section was blocked by incubation in 3% hydrogen peroxide for 5 min. After blocking with 0.5% goat serum for 60 min, the sections were incubated with primary antibodies HIF-1α, VEGF (Santa Cruz Biotechnology, California, USA), CD68 and CD34 (DAKO, Glostrup, Denmark), applied at room temperature for 1 h. Then, the DAKO REAL EnVision Detection kit (DAKO) was applied for 30 min. Finally, sections were incubated in 3’3-diaminobenzidine for 5 min, followed by Mayer’s haematoxylin counterstaining and mounting. Negative controls were obtained by replacing the primary antibody with non-immune serum.

Immunostaining evaluation

HIF-1α protein immunoreactivity was present in the nuclei with or without cytoplasms of neoplastic cells. Using the semiquantitative scale described previously,5 the HIF-1α protein expressions were classified as follows: –, no staining; 1+, nuclear staining in less than 1% of cells; 2+, nuclear staining in 1–10% of cells with or without weak cytoplasmic staining; 3+, nuclear staining in 11–50% of cells with or without distinct cytoplasmic staining; and 4+, nuclear staining in more than 50% of cells with or without strong cytoplasmic staining. For further analysis, using a cut-off point to define two groups of low and high HIF-1α expression, –, 1+ or 2+ staining patterns were regarded as low expression, and 3+ or 4+ staining patterns were regarded as high expression. VEGF expression was observed in at least 10% of tumour cells to the intensity of cytoplasmic staining. VEGF staining was classified into the following two grades: negative and weak cytoplasmic staining was regarded as low expression, distinct and strong cytoplasmic staining was regarded as high expression. Each slide was scored independently and the results were examined by two blinded independent reviewers (CYC and WYK).

Microvessel and macrophage detection and quantification

CD68 was used to mark the macrophages within UC and CD34 was designed to label the microvessels. For CD34 staining, slides were first observed at low power magnification (×100) to identify areas with the highest density of microvessels (“hotspots”). Each hotspot was then evaluated at high power magnification (×200) and the number of stained microvessels per high power field was determined. Any brown-stained endothelial cell was considered to be a single countable microvessel. Large vessels with thick muscular walls were excluded from the counts. The presence of a clearly defined vessel lumen was not required to verify the structure as a microvessel. The final microvessel score was the average of the vessel counts obtained from these 10 fields. The quantification of TAMs or MVD was randomly shifted within each carcinoma area and carcinoma tissues except for areas of extensive necrosis, which were entirely evaluated in the selected sections. The mean for 10 fields adjacent to or within tumour cells at ×400 and ×200 high-power field (HPF) was counted for each slide and the values were regarded as TAMs density and MVD.

Statistical analysis

Correlations of clinicopathologic characteristics and HIF-1α expression were assessed using the χ2 test. The Student t test was used to assess the correlation between HIF-1α expression and MVD or TAMs density. Survival analysis was carried out using death from disease or recurrence/metastasis as the endpoints for OS and DFS. Associations of variables with survival analysis were tested using univariate and multivariate analyses by the Cox proportional hazard model. Survival curves were calculated by the Kaplan–Meier method and difference between survival curves was analysed by the log rank test. All statistical analyses were performed with the SPSSV. 12.0 statistical software program (SPSS Inc., Chicago, Illinois, USA). p Values of less than 0.05 were considered to be statistically significant.


Correlation between HIF-1α expression and clinicopathologic characteristics

HIF-1α immunohistochemical expression was observed to show a predominant nuclear staining in tumour cells with or without cytoplasmic staining in UC (fig 1A). Most solid human tumours have focal hypoxic areas that cause low oxygen tension and regularly contribute to tumour necrosis. HIF-1α is a primary determinant of HIF activity and high expression adjacent to the necrotic region (fig 1B). The correlations between expression of HIF-1α and the clinicopathologic characteristics described are shown in table 1. Of 99 cases, 55 (55.6%) were high expression, and 44 cases (44.4%) were low expression for HIF-1α. The high expression of HIF-1α was significantly positively correlated with tumour size, high histologic grade, tumour invasiveness, and recurrence (p = 0.001, p = 0.028, p = 0.002, and p = 0.011, respectively). In contrast, HIF-1α expression had no statistically significant correlation with gender, age, location and distant metastasis in UC (table 1).

Figure 1 Immunohistochemical staining for hypoxia-indicuble factor (HIF)-1α, vascular endothelial growth factor (VEGF), CD34 and CD68. A. High-grade carcinoma with tumour invasion showing strong nuclear and cytoplasmic staining for HIF-1α. B. Tumour cell showing high expression of HIF-1α adjacent to necrotic region. C. VEGF is low expressed in the non-invasive urothelial carcinoma (UC). D. High-grade carcinoma with tumour invasion showing strong cytoplasmic staining for VEGF. E. High-grade UC with tumour invasion showing dense blood vessels. F. Intratumour cells showing intense macrophage infiltration in tumour invasion UC. A–E, original magnification, ×200; F, original magnification, ×400.
Table 1 Correlation of hypoxia-inducible factor (HIF)-1α expression with known clinicopathologic characteristics in 99 urothelian carcinoma (UC) cases

VEGF expression is associated with tumour progression and HIF-1α overexpression in UCs

To further confirm that HIF-1α in UCs also correlates with angiogenesis, a parallel immunohistochemical staining of VEGF was performed on these 99 cases, in which VEGF expression was assessed as a major marker for angiogenesis. VEGF expression was closely associated to tumour invasiveness (p = 0.017, data not shown). Figure 1C,D shows immunostaining of VEGF signals in tumours at various grades and tumour invasiveness. Figure 1C shows weak staining in low grade and non-invasive UC. In contrast, fig 1D shows strong staining in the cytoplasm of tumour cells. As shown in fig 2, using the χ2 test, overexpression of VEGF in UC was found to correlate positively with HIF-1α high expression (p<0.001).

Figure 2 Correlation of vascular endothelial growth factor (VEGF) expression with hypoxia-inducible factor (HIF)-1α expression in urothelial carcinoma (UC) cases.

MVD is correlated with HIF-1α overexpression in UCs

Firstly, we found that MVD count was closely associated with histological grade and tumour invasiveness of UC (p = 0.05, p<0.001; data not shown). Secondly, to study the relationship between HIF-1α expression and vascularisation, we investigated the distribution of MVD in areas of HIF-1α expression. MVD was measured by CD34 immunostaining. According to our study, in HIF-1α-high expression UCs had significantly higher MVD than in HIF-1α-low expression UCs (20.35 (8.61)/HPF vs 12.36 (7.67)/HPF; p<0.001, fig 3). Moreover, fig 1E shows representative staining of microvessels in tumour invasive UC, demonstrating that strong HIF-1α expression is highly correlated with increased tumour MVD.

Figure 3 Correlation of microvessel density (MVD) counts with hypoxia-inducible factor (HIF)-1α expression in urothelial carcinoma (UC) cases. Error bars, standard deviation of the number (n) of samples analysed.

HIF-1α expression is associated with TAM infiltration in UCs

We examined whether macrophages were involved in tumour invasion of UC because they can infiltrate into tumour tissues and secrete a variety of cytokines and growth factors, and further determined whether they might be associated with HIF-1α expression. Our study indicated there is a higher propensity of TAM to infiltrate deep tumours (fig 1F), but only a limited distribution in superficial UCs (15.31 (11.83)/HPF versus 9.68 (10.54)/HPF; p = 0.023, data not shown). To assess the correlation between TAM and HIF-1α expression, CD68 immunostaining was performed in all cases. As shown in fig 4, more TAM infiltration was found in HIF-1α-high expression UCs (16.57 (11.99)/HPF) and was significantly higher than in HIF-1α-low expression UCs (9.51 (10.08)/HPF; p = 0.002).

Figure 4 Correlation of tumour-associated macrophage (TAM) infiltration with hypoxia-inducible factor (HIF)-1α expression in urothelial carcinoma (UC) cases. Error bars, standard deviation of the number (n) of samples analysed.

Prognostic relevance of HIF-1α expression and TAM in UCs

When survival of cases with high expression of HIF-1α was compared with survival of cases with low expression of HIF-1α, Kaplan–Meier analysis (log-rank test) found that overexpression of HIF-1α was only associated with worse DFS (p = 0.009) but was not associated with OS (p = 0.696) (fig 5A,C). We next examined whether the TAM count was of value in predicting clinical outcome or prognosis in UC cases. The median values (of the TAM count: 9.67/HPF) were compared with OS and DFS. TAM count was not associated with OS (p = 0.141) but was significantly associated with DFS (p = 0.023) with cases in the high category group having worse survival (fig 5B,D). For further analysis, we investigated whether HIF-1α expression was an independent prognostic variable. A Cox proportional hazards model was constructed using established prognostic factors and HIF-1α expression. The univariate analysis revealed that tumour size (p = 0.018), but not HIF-1α expression (p = 0.696), were independent prognostic factors for OS (table 2). In addition, cases with a high HIF-1α expression had a significantly (p = 0.012) worse DFS prognosis than those with a low HIF-1α expression, while univariate analyses also indicated the large tumour size (p = 0.015), histological grade (p = 0.027), and TAM count (p = 0.015) to be prognostic factors for DFS (table 2).

Figure 5 Kaplan–Meier survival curves of 99 urothelial carcinoma (UC) cases stratified for hypoxia-inducible factor (HIF)-1α expression and tumour-associated macrophage (TAM) count. A. Overall survival (OS) by HIF-1α expression. B. OS by TAM count. C. Disease-free survival rate (DFS) by HIF-1α expression. D. DFS by TAM count.
Table 2 Univariate analysis by the Cox proportional hazards regression analysis for overall survival (OS) and disease-free survival rate (DFS)

Finally, multivariate analyses indicated only tumour size (p = 0.038) to be independently significant prognostic factors for OS. In addition, multivariate analyses indicated HIF-1α expression (p = 0.011), as well as histological grade (p = 0.038), and MVD (p = 0.004), to be independently significant prognostic factors for DFS. However, a multivariate analysis showed no significant association between tumour invasiveness, TAM count, VEGF expression and either DFS or OS (table 3).

Table 3 Multivariate analysis by the Cox proportional hazards regression analysis for overall survival (OS) and disease-free survival rate (DFS)


HIF-1α overexpression is a marker of hypoxia because its intracellular degradation by the proteasome pathway is oxygen dependent.16 17 The major objective of this study is to demonstrate HIF-1α expression closely correlated with tumour-induced hypoxia and angiogenesis in UC. However, to the best of our knowledge, HIF-1α expression associated with TAM infiltration during hypoxia has not been previously reported in UC. Therefore, we hypothesised that hypoxia is a major stimulus angiogenesis and macrophage infiltration that is consequently up-regulated by HIF-1α. In addition, it is worth determining out whether HIF-1α expression and TAM count are accurate and reliable prognostic factors for UC.

According to our study, we correlated the HIF-1α expression with several conventional clinicopathological parameters. Firstly, higher HIF-1α expression was noted in the necrosis areas of the tumour and especially in tumours with large size. Secondly, HIF-1α expression was correlated strongly with higher histological grades and tumour invasiveness and also with higher probability of recurrence. In addition, our study revealed that HIF-1α overexpression is associated with increasing VEGF expression, one of its main downstream effectors, confirming angiogenesis as one of the proposed mechanisms by which HIF-1 activation stimulates tumour growth and leads to increasing VEGF-mediated tumour angiogenesis. VEGF is up-regulated in poorly vascularised areas, triggering the angiogenesis process, which is essential for tumour growth, progression and metastasis of solid tumours.18 19 It is now widely accepted that VEGF expression is mediated by HIF-1α during hypoxia. It has long been established that macrophages are powerful stimulators of angiogenesis via secretion of a number of direct and angiogenic factors.20 The specific mechanism for the induction of angiogenesis by TAMs in bladder cancer has been established.13 Results derived from our statistical analysis of MVD and TAM also provided some insight into the possible location and function of HIF-1α expression in oncogenesis. Firstly, MVD was significantly higher than an HIF-1α low expression area. Secondly, TAM was positively correlated with HIF-1α expression. We have also shown that increased macrophage infiltration into tumours was associated with increased VEGF expression and MVD (data not shown). We have shown that in UCs, macrophage migrate to areas distant from angiogenic hotspots, suggesting that macrophages infiltration in tumour hypoxia regions may be responsible for the induction of HIF-1α expression or by the release of macrophage chemoattractants. These include VEGF21 and endothelin-222 that are up-regulated by hypoxic tumour cells. These observations suggest that the correlation between HIF-1α expression and angiogenesis could be related either to stimulated angiogenic factors such as VEGF, from HIF-1α expressing tumour cells in hypoxic areas, or to the possibility that TAM infiltration and secretion factors directly stimulate angiogenesis.

Take-home messages

  • Hypoxia inducible factor (HIF)-1α expression is closely correlated with tumour-induced hypoxia and angiogenesis in urothelial carcinoma.

  • The correlation between HIF-1α expression and angiogenesis could be related either to stimulated angiogenic factors such as vascular endothelial growth factor (VEGF), from HIF-1α expressing tumour cells in hypoxic areas, or to the possibility that tumour-associated macrophages infiltration and secretion factors directly stimulate angiogenesis.

  • HIF-1α expression was correlated strongly with higher histological grades and tumour invasiveness and also with higher probability of recurrence in urothelial carcinoma.

  • HIF-1α may serve as a single reliable tool for the early and accurate prediction of tumour recurrence in urothelial carcinoma (UC).

HIF-1α has been related to unfavourable prognosis in women with cervical cancer and poor response to radiotherapy in patients with oropharyngeal cancer.23 24 Giatromanolaki et al have shown that HIF-1α overexpression is a common event in non-small cell lung carcinomas, up-regulating the angiogenic pathways and associated with poor prognosis.25 We also analysed the effect of positivity in HIF-1α staining on survival of cases in our cohort. The major finding in the present study was the dismal prognostic influence of HIF-1α overexpression on tumour progression in UC. Using the Log-rank test, we demonstrated that in UC, HIF-1α and TAM count were significantly associated with DFS but not with OS. These results are in agreement with a recent report by Vasilios E et al26 showing that HIF-1α overexpression had a marginal significance on progression-free survival in patients with superficial bladder UC. However, we have shown that overexpression of HIF-1α is correlated with tumour recurrence in UC. Although in the multivariate analysis HIF-1α was not of prognostic significance for tumour progression, it may serve as a single reliable tool for the early and accurate prediction of tumour recurrence.

Koga F et al demonstrated no significant association between TAM count and the prognosis in patients with invasive bladder cancer.27 Nevertheless, according to our study, we have demonstrated that in UC, increased TAM count is associated with reduced DFS. An interesting observation in our study was the fact that when HIF-1α was included with tumour size, histologic grade, tumour invasiveness and angiogenic factors in the multivariate model, tumour size had an independent prognostic effect on OS, and histologic grade, HIF-1α expression and MVD could retain their independent prognostic significance on DFS. Therefore, we inferred that tumour size is an important indicator of tumour aggressiveness; furthermore, this may be through promoted expression of HIF-1α in UCs. In addition, this is probably due to the close association of HIF-1α with MVD, suggesting that HIF-1α influences UC outcome via stimulation of the angiogenic process.

Since hypoxia is a hallmark feature of malignant tumours and hypoxic tumour cells are relatively resistant to radiotherapy and chemotherapy, these areas have become a target for novel forms of anticancer therapy. These include hypoxia-targeted gene therapy in which macrophages are armed with therapeutic genes that are activated by hypoxia-responsive promoter elements.28 We expect demonstration of the clinical association of HIF-1α with hypoxia may lead to exploration of its potential as a new prognostic indicator and possibly a target for gene therapy for UC.


The authors thank the Statistical Analysis Laboratory team, Department of Clinical Research, Kaohsiung Medical University Chung-Ho Memorial Hospital for their help.



  • Competing interests: None declared.

  • Funding: This work was supported by a grant 5N-01 from Kaohsiung Medical University Hospital and the National Sun Yat-Sen University-Kaohsiung Medical University Joint Research Center, Kaohsiung, Taiwan.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.