Carbonic anhydrase IX is associated to poor prognosis

While the expression of CA IX in brain tumours had been studied earlier (Ivanov et al. 2001, Proescholdt et al. 2005), the analysis of CA IX in a large series of malignant astrocytomas and correlation to clinical features was lacking. In the present study (study II), the expression of CA IX was observed in 78% of astrocytic tumours. As with the previous reports, positive areas were often located in close proximity to necrotic regions of grade IV astrocytomas, and a significant association was found with the presence of necrosis. Also, some cytoplasmic staining was seen in the neoplastic cells of the infiltrative zone. In glioblastomas, these pseudopalisades represent a wave of tumour cells actively migrating away from the central hypoxia. Thus, it seemed possible that activity of CA IX is essential to tumour cells struggling under high metabolic stress.

Importantly, the cytoplasmic expression of CA IX had a strong association with increasing tumour WHO grade; 65% of grade II astrocytomas were CA IX-positive, whereas the positivity increased along with the grade, being 73% in grade III astrocytomas and 82% in glioblastomas. The correlation of CA IX with prognosis was then studied in survival analysis, which importantly showed that increasing cytoplasmic CA IX intensity correlated significantly to poor prognosis of the patients in univariate analysis, and this difference was significant for both grade II and grade IV astrocytomas, as well. Finally, Cox multivariate survival analysis showed that CA IX intensity, along with patient age and WHO tumour grade, was an independent prognostic factor. These findings supported the essential role of CA IX in tumourigenesis and malignancy of diffusely infiltrating astrocytomas. CA IX could be clinically used to assess the prognosis of astrocytoma patients.

The correlation of CA IX with prognosis has been confirmed by others. Korkolopoulou et al. (2007) showed that increasing CA IX immunopositivity was associated with a shortened survival in univariate analysis, and when tested in multivariate analysis, they reported similar independent prognosticators; CA IX, tumour grade and patient age were the only parameters entered into the model. In accordance with the present study, perinecrotic distribution of CA IX immunostaining was detected and intensity increased in parallel with the extent of necrosis and histological grade.

Furthermore, in a phase II trial, patients with recurrent malignant astrocytomas treated with bevacizumab and irinotecan (a combination of VEGF -neutralizing antibodies) were retrospectively

evaluated (Sathornsumetee et al. 2008). Again, the high expression of CA IX was associated with poor survival outcome in the patient cohort (Figure 6). Another hypoxia-regulated marker, VEGF, was associated with radiographic response but not with survival. Additionally, when both CA IX and HIF-2 were simultaneously included in a Cox model as separate factors, only CA IX remained as a statistically significant factor.

In addition to the studies mentioned above, Flynn et al. (2008) investigated hypoxia-regulated protein expression, including CA IX, in another cohort of patients with glioblastomas. They found no significant correlation between CA IX expression and patient survival or tumour grade when low-grade astrocytomas were included in the analysis. Though being statistically insignificant, the patients with CA IX-positive tumours seemed to have a trend towards a worse prognosis. The discrepancy between this and the three studies mentioned above could be due to different immunostaining method used. In this study and also in the study by Korkolopoulou et al. (2007) the M75 antibody is used, which represents an extensively used and reliable CA IX antibody which has no cross-reactivity with other CAs (Saarnio et al. 1998b). Another important issue is the smaller number of patients in the study of Flynn et al. (2008). The latest finding of Yoo et al. (2010) showed that CA IX expression was a predictive factor for poor survival and, again, correlated positively with increasing WHO grade. Thus, the initial finding of the study II seems to be steadily supported by others.

Figure 6. Kaplan-Meier survival curves stratified by biomarker status. (A) Survival stratified by carbonic anhydrase 9 (CA9) status. (B) Survival stratified by CA9 and hypoxia-inducible factor-2α (HIF-2) status. Figure: Sathornsumetee et al. (2008). Reprinted with permission. © 2008 American Society of Clinical Oncology. All rights reserved

For several years, CA IX has been linked to tumour hypoxia and necrosis, and several investigations have been conducted to assess this issue in detail (Ivanov et al. 2001, Proescholdt et al. 2005). Furthermore, CA9 has been reported to be one of the most consistently hypoxia-inducible genes in vitro in human glioblastoma cell lines (Said et al. 2007b). Indeed, CA IX is strongly regulated by hypoxia, and thus, could be used as an additional tool when defining the WHO grade.

The small biopsies of gliomas are often a challenge to the neuropathological diagnosis of glioblastoma, especially when the necrotic areas are absent. In these cases, CA IX expression could be useful to aid clinical diagnostics, because it significantly associates with increasing WHO grade and necrosis. However, the regulation of CA IX seems to be more complex and the immunopositivity has also been reported in tumour cells located near the blood vessels (Proescholdt et al. 2005). Correspondingly, it has been suggested that CA IX induction in gliomas may involve hypoxia-independent mechanisms, and acidosis also induces CA IX independently of pericellular hypoxia in glioblastoma cell lines (Ihnatko et al. 2006). Finally, in vitro studies on glioblastoma cells have suggested that CA IX expression can result from low oxygen concentrations or constitutive, oncogene-related overexpression (Said et al. 2007a).