Prognostic impact of carbonic anhydrase IX expression in human renal cell carcinoma

Authors

  • Johanna Sandlund,

    1. Department of Radiation Sciences, Oncology, Umeå University, Umeå, Sweden, *Department of Urology, University Medical Centre, Nijmegen, Nijmegen, the Netherlands, †Department of Medical Biosciences, Clinical Chemistry, and ‡Department of Surgical and Perioperative Sciences, Urology and Andrology, Umeå University, Umeå, Sweden
    Search for more papers by this author
  • Egbert Oosterwijk,

    1. Department of Radiation Sciences, Oncology, Umeå University, Umeå, Sweden, *Department of Urology, University Medical Centre, Nijmegen, Nijmegen, the Netherlands, †Department of Medical Biosciences, Clinical Chemistry, and ‡Department of Surgical and Perioperative Sciences, Urology and Andrology, Umeå University, Umeå, Sweden
    Search for more papers by this author
  • Kjell Grankvist,

    1. Department of Radiation Sciences, Oncology, Umeå University, Umeå, Sweden, *Department of Urology, University Medical Centre, Nijmegen, Nijmegen, the Netherlands, †Department of Medical Biosciences, Clinical Chemistry, and ‡Department of Surgical and Perioperative Sciences, Urology and Andrology, Umeå University, Umeå, Sweden
    Search for more papers by this author
  • Jeannette Oosterwijk-Wakka,

    1. Department of Radiation Sciences, Oncology, Umeå University, Umeå, Sweden, *Department of Urology, University Medical Centre, Nijmegen, Nijmegen, the Netherlands, †Department of Medical Biosciences, Clinical Chemistry, and ‡Department of Surgical and Perioperative Sciences, Urology and Andrology, Umeå University, Umeå, Sweden
    Search for more papers by this author
  • Börje Ljungberg,

    1. Department of Radiation Sciences, Oncology, Umeå University, Umeå, Sweden, *Department of Urology, University Medical Centre, Nijmegen, Nijmegen, the Netherlands, †Department of Medical Biosciences, Clinical Chemistry, and ‡Department of Surgical and Perioperative Sciences, Urology and Andrology, Umeå University, Umeå, Sweden
    Search for more papers by this author
  • Torgny Rasmuson

    1. Department of Radiation Sciences, Oncology, Umeå University, Umeå, Sweden, *Department of Urology, University Medical Centre, Nijmegen, Nijmegen, the Netherlands, †Department of Medical Biosciences, Clinical Chemistry, and ‡Department of Surgical and Perioperative Sciences, Urology and Andrology, Umeå University, Umeå, Sweden
    Search for more papers by this author

Torgny Rasmuson, Department of Radiation Sciences, Oncology, Umeå University, SE-901 85 Umeå, Sweden. e-mail: torgny.rasmuson@onkologi.umu.se

Abstract

OBJECTIVE

To evaluate the prognostic information of carbonic anhydrase (CA) IX expression in patients with renal cell carcinoma (RCC), as increased expression of CA IX is correlated with a worse prognosis in several malignancies.

PATIENTS AND METHODS

CA IX expression was assessed in RCC tumours from 228 patients, using a tissue microarray technique on archival material. The expression was related to RCC cell type, Tumour-Node-Metastasis (TNM) stage, nuclear grade and survival.

RESULTS

CA IX expression was significantly higher (P < 0.001) in 183 conventional than in 31 papillary RCC and 14 chromophobe RCC. For conventional RCC there was no correlation of CA IX expression with TNM stage or nuclear grade. To evaluate the prognostic information conventional RCC tumours were subdivided arbitrarily into three groups according to the CA IX expression, of 0–10%, 11–90% and 91–100% expression, respectively. Patients with tumours with 0–10% expression had a less favourable prognosis than those with 11–90% and 91–100% expression (P = 0.012, and 0.001), respectively. A multivariate analysis of prognostic factors for patients with conventional RCC showed that TNM stage, nuclear grade and CA IX were independent predictors of prognosis.

CONCLUSION

These results show that CA IX expression is higher in conventional than other RCC cell types; furthermore, patients with conventional RCC with low CA IX expression had a less favourable prognosis.

Abbreviations
CA

carbonic anhydrase

HIF-1α

hypoxia-inducible factor-1α

VHL

von Hippel-Lindau

(c)(p)(ch)RCC

(conventional) (papillary) (chromophobe) RCC

TMA

tissue microarray.

INTRODUCTION

RCC is an aggressive tumour, with a third of patients having metastases at the time of diagnosis [1] and >40% dying from their disease [2]. The development of hypoxia in solid tumours is associated with tumour aggressiveness, metastases, and recurrence after treatment [3]. To survive in a hypoxic environment organisms have developed responses like glycolysis, angiogenesis, vasodilatation and erythropoesis [4]. Likewise, hypoxia represents a clear difference between RCC and normal tissues and therefore is potentially useful as a target in RCC treatment [5].

Carbonic anhydrase (CA) IX is a transmembrane enzyme [6] that catalyses the reversible hydration of carbon dioxide into carbonic acid [7]. In tumour progression CA IX might maintain a normal pH in tumour cells under hypoxia, allowing tumours to adapt to a hypoxic microenvironment, which might permit continued tumour cell proliferation distant from blood vessels [8–10]. CA IX is strongly induced by hypoxia [11] via the hypoxia-inducible factor-1α (HIF-1α) [12].

In conventional/clear-cell RCC (cRCC), mutations in the von Hippel-Lindau (VHL) gene result in an up-regulation of CA IX [13,14], and via hypoxia-induced pathway and accumulation of HIF-1α[5,12,13]. CA IX is not expressed in normal human kidney tissue [15], but is generally over-expressed in RCC and particularly in cRCC [16–19]. Chromophobe RCC (chRCC) has been reported not to express CA IX [18]. In papillary RCC (pRCC), CA IX shows a perinecrotic distribution, while in cRCC the expression is more uniform [11].

In several malignancies, e.g. cervix/uterine, lung and breast cancer, increased CA IX expression is correlated with a worse prognosis [20–22]. Contrary to this, in patients with cRCC, high CA IX expression is associated with an improved prognosis [23–25].

The aim of the present study was to assess whether CA IX expression had prognostic information in RCC using immunohistochemistry on tissue microarray (TMA) sections containing many archived RCC specimens.

PATIENTS AND METHODS

The study includes 228 patients (131 men, 97 women; median age 66 years, range 25–87) with histopathologically verified RCC. All patients had a nephrectomy between 1982 and 1997 at the Department of Urology, Umeå University Hospital, Sweden. The RCC types were assessed according to the Heidelberg classification system [26]. The details of the patients are shown in Table 1.

Table 1. 
Clinical characteristics of the patients with RCC
CharacteristiccRCCpRCCchRCCTotal
Total1833114228
Male10618 7131
Female 7713 7 97
TNM stage
 I 4813 3 64
 II 20 4 5 29
 III 55 9 5 69
 IV 60 5 1 66
Nuclear grade
 1  2 3  5
 2 3310 5 48
 3 9216 8 116
 4 56 2 1 59
Survival:
 died from disease10015 2 117
 died from other disease 4510 7 62
 alive 38 6 5 49

The patients had a physical examination, chest radiography, and CT of the abdomen. When vena cava tumour thrombus invasion was suspected, cavography or MRI was used. Patients with skeletal-associated pain or elevated serum alkaline phosphatase were assessed with bone scintigraphy. Tumours were staged according to the TNM classification system 2002 [27], with nuclear grading according to Skinner et al.[28]. Tumour size was measured at the maximum diameter on the surgical specimen or by CT. The median (range) tumour diameter was 7.5 (0.6–25.0) cm. The patients were followed according to a programme including regular clinical and radiological examinations.

All tumour samples were obtained after permission from the patients and during the last few years with informed and signed consent. The study was approved by the ethics committee of Umeå University. Tumour samples were obtained from the surgical specimen, formalin-fixed, and paraffin-embedded for routine morphological examination and immunohistochemical staining. Representative paraffin tumour blocks, selected by primary evaluation of haematoxylin/eosin-stained slides, were chosen for TMA preparation. Two tissue cores were taken from each tumour with a sample needle (0.6 mm diameter) and placed in a new recipient paraffin block occupied by 98 tissue cores (Beecher Instruments, Sun Prairie, WI, USA). For microscopic evaluation, 4-µm paraffin sections were processed using standard procedures with de-paraffinisation and rehydration. Antigen was retrieved by microwave treatment for 10 min in 0.1 mm citrate buffer at pH 6.0. After allowing the slides to cool in the buffer for 1–2 h, endogenous peroxidase was blocked for 30 min in PBS with 3% H2O2. The slides were then incubated with monoclonal antibody M75 ON at 4 °C. The slides were incubated for 30 min with B*-antimouse 1 : 100 (Vector Laboratories, Burlinghame, MA, USA) in 0.5% BSA/PBS. Finally the slides were incubated for 30 min with avidin-biotin complex and developed for 10 min in 0.05% diaminobenzidine tetrahydrochloride (Fluka, Germany) in 0.05 mm Tris pH 7.8. After extensive washing under tap water, slides were counterstained with haematoxylin, dehydrated, and mounted with permanent mountant; slides were washed in PBS between each step.

CA IX expression was evaluated by three of the authors (J.S., E.O. and J.O.W) unaware of the clinical and pathological variables. After individual assessments a consensus was reached. The extent of staining was recorded as a percentage, in 5% intervals, of the tumour cells that had positive CA IX expression. For each cRCC the mean value of the two biopsies was used in further analysis. The staining intensity was not taken into account, i.e. cells were scored as positive or negative. CA IX staining was assessed at × 10 and at × 25/40 in cases where scoring was more difficult.

TNM stages I and II, and nuclear grades 1 and 2, respectively, were evaluated together for statistical analyses, as there were relatively few observations in stage II and grade 1. The results were analysed statistically using the Mann–Whitney U- and Kruskal–Wallis tests. Correlations between variables were tested as two-sided and according to the Spearman correlation test. Survival was assessed and illustrated using the Kaplan-Meier method, and survival time was compared using the log-rank test. Multivariate regression analysis was used with the Cox proportional-hazards method; in all tests P < 0.05 was considered to indicate statistical significance.

RESULTS

CA IX staining was evaluated in 228 patients with RCC; Fig. 1 shows the immunohistochemical staining of cRCC tumours with high and low CA IX expression. The median (range) CA IX expression in all tumour types together was 90 (0–100)%; the distribution varied among the different RCC types (Fig. 2). CA IX expression in cRCC was significantly higher than in pRCC and chRCC (both P < 0.001), but there was no difference between pRCC and chRCC. The median expression was 95% in cRCC, 2.5% in pRCC and 0% in chRCC. No CA IX expression was found in 45 (38%) tumours, of which 21 were cRCC (12% of all cRCC), 15 pRCC (48% of all pRCC), and nine of 14 chRCC. There was no correlation between CA IX expression, TNM stage, nuclear grade, age, gender, storage time (in 3-year intervals), or tumour diameter in any of the three subtypes.

Figure 1.

Immunohistochemical staining for CA IX in cRCC, showing high (A) and low (B) expression, respectively.

Figure 2.

A boxplot showing the CA IX expression in 183 cRCC, 31 pRCC and 14 chRCC.

The median (range) follow-up from diagnosis for surviving patients was 152 (73–263) months. Survival was mainly analysed for the 183 patients with cRCC tumours, because there were few patients with pRCC or chRCC. The cRCCs were divided arbitrarily into three groups (Table 2), according to the CA IX expression (0–10%, 11–90% and 91–100%). Patients with cRCC in the lower group had a significantly less favourable prognosis than those with higher expression (P = 0.012, and 0.001, respectively) (Fig. 3). When the patients were subdivided according to stage, CA IX expression had no effect on prognosis for those with stage IV, but only for patients with stage I-III disease (P = 0.01, log-rank test; Fig. 3).

Table 2.  Disease-specific survival for patients with cRCC as a function of CAIX expression Thumbnail image of
Figure 3.

Kaplan-Meier plot showing tumour-specific survival in relation to CA IX expression (0–10%, 11–90%, 91–100%) in patients with cRCC.

To evaluate the prognostic effect of CA IX expression for pRCC, patients were subdivided by the median of 2.5. Disease-specific mortality for patients with CA IX expression below the median was similar to that of patients above the median, at 53% and 50%, respectively.

To assess the prognostic effect of CA IX expression in patients with cRCC, we used multivariate analysis including disease-specific survival, age, gender, nuclear grade and TNM stage. After the final step of the evaluation, patients with low CA IX expression (0–10%) had a significantly worse prognosis than those with higher stage and nuclear grade (Table 3).

Table 3. 
Multivariate analysis of prognostic factors in 183 patients with cRCC
FactorRelative risk (95% CI)P
Age, years
 ≤661.0 
 >660.93 (0.62–1.40)0.740
Gender
 male1.0 
 female0.92 (0.61–1.39)0.710
Stage
 I–II1.0 
 III–IV8.36 (4.40–15.91)<0.001
Grade
 1–21.0 
 3–45.69 (2.22–14.58)<0.001
CA IX, %
 0–102.18 (1.18–4.02)0.013
 11–901.0 
 91–1000.83 (0.52–1.33)0.440

DISCUSSION

The CAs comprise a family of zinc metallo-enzymes, the isoenzymes of which are involved in processes like regulating acid-base balance, gas exchange, ion transport, carbon fixation and mucosal protection [29]. To continue tumour cell proliferation during hypoxic conditions, CA IX is thought to optimize conditions by maintaining a normal pH in tumour cells, and hence facilitate continuous growth distant from blood vessels [8–10].

In the present study there was a striking difference in CA IX expression among different tumour cell types; cRCC tumours had very high CA IX expression (median 95%) with few negative cases, while normal human kidney did not express CA IX [15]. This is in line with previous studies [16–19]. In the present samples, pRCCs and chRCCs had lower CA IX expression, as shown also by others [11,18]. This highlights the molecular mechanisms involved in these different entities. CA IX expression is a direct consequence of a VHL mutation, found in >75% of sporadic cRCC, independent of hypoxia. By contrast, CA IX expression in pRCC and chRCC is solely hypoxia-related.

There was no correlation between CA IX expression and stage or grade in patients with RCC. High CA IX expression has been found to be associated with higher grade and tumour necrosis in breast cancer [22], and with more advanced T stage in non-small-cell lung cancer [21]. The difference is also a reflection of the molecular mechanisms responsible for CA IX expression; high CA IX expression in other than cRCC tumours is hypoxia-related, and generally more prevalent in high-stage and high-grade tumours.

Patients with cRCC tumours with low CA IX expression had a less favourable prognosis than those with higher expression. A similar relationship between CA IX expression and prognosis was previously reported by others [23–25]. According to our results this prognostic difference is only found in patients with stage I-III disease (M0), a finding that contrasts with the results presented by Bui et al.[23]; they found a prognostic difference only in patients with stage IV disease (M1), but the reason for this difference is unknown.

For patients with pRCC and chRCC, CA IX expression had no apparent influence on the prognosis. This indicates a difference between the VHL-associated cRCC subtype and the two other RCC types. Whether this difference can be explained by VHL mutation status alone is uncertain.

In tumours such as cervix/uterine, lung and breast carcinoma, as well as pRCC, an increased CA IX expression has been associated with an unfavourable prognosis [20–22]. In cRCC high CA IX expression was shown to correlate with improved prognosis [23–25]. CA IX is strongly induced by hypoxia [11] via HIF-1α[12], and in cRCC via mutations to the VHL gene [13,14].

Our results are in line with previous studies on the angiogenic variables endoglin and CD31, where patients with cRCC and high endoglin and CD31 expression had a more favourable prognosis [30,31]. The generally higher CA IX, endoglin and CD31 expression in less aggressive cRCCs could reflect the fact that other pathways apart from the HIF pathway might be activated in these cRCCs.

The strength of the TMA technique is that it allows simultaneous analysis of very many tumours under standardized laboratory and evaluation conditions. Limitations of the technique are the restricted sample volume and a possibility that tumour heterogeneity might affect the representativity of the samples. We analysed two samples from each tumour and there was very little difference between the samples from each tumour.

In summary, we show that CA IX expression is higher in cRCC than in other RCC tumour cell types. Furthermore, patients with cRCC and low CA IX expression had a less favourable prognosis. Particularly in view of new therapeutic possibilities, evaluating molecular markers such as CA IX is important, as these will eventually enhance the ability to predict individual tumour behaviour and to stratify patients into more sophisticated risk categories. This should provide crucial information on the treatment and prognosis of patients with RCC, and allow a more rationale design of clinical management.

ACKNOWLEDGEMENT

This study was supported by grants from Lions Research Foundation, the Department of Oncology, the Medical Faculty, Umeå University, and the Swedish Cancer Society. The authors are grateful to Björn Tavelin for excellent statistical analysis.

CONFLICT OF INTEREST

None declared.

Ancillary