Complement activation in astrocytomas: deposition of C4d and patient outcome

Astrocytomas are CNS tumors originating from astrocytic glial cells or their precursor cells. They can be divided into four grades according to WHO criteria by the appearance of cell atypia, mitosis, microvascular proliferation and necrosis [1]. Grade I astrocytomas are called pilocytic astrocytomas. They are considered benign as they have a clear borderline and are generally slow growing. The biological background of pilocytic astrocytomas differs from grade II-IV tumors. Grade II-IV astrocytomas have a diffuse growing manner and lack a clear borderline. They are therefore referred to as diffusely infiltrating astrocytomas. These tumors often renew and proceed into more malignant grades. Glioblastoma multiforme is a grade IV astrocytoma, and the most prevalent form of astrocytic tumors in adult patients. Microscopically, glioblastomas can be distinguished from grade II-III astrocytomas by the occurrence of necrosis and microvascular proliferation. Glioblastomas have a vivid microvascular proliferation rate and this results often in abnormal, even glomeruloid microvascular growth patterns. This can sometimes be seen in pilocytic astrocytomas, as well. Benign pilocytic astrocytomas may also reoccur, if a remnant has been left behind during operation. However, they never turn into grades II-IV.

Complement is an innate component of the immune system. It can be initiated by three different pathways, including the classical pathway, lectin pathway and the alternative pathway [2]. Degradation of the first component of activated complement system yields an active enzyme that continues the cascade by cleaving the next zymogen of the cascade into a functioning enzyme. The follow-through of complement leaves behind inactive fragments of complement components. C4d is created when the complement control protein factor I inactivates C4b by cleaving it into C4c and C4d [3]. C4d remains covalently bound to the activation point. There is no apparent biological function associated with C4d alone [4], but due to its covalent binding and relatively long half-life, it can be considered as an activation marker of the classical and lectin pathways of complement. Only the classical and lectin pathways are known to involve C4d, whilst in the alternative pathway there is no cleavage of the C4 component [5].

C4d has to date been most widely surveyed on allograft tissue rejections. In allograft rejection studies, C4d positivity has been mainly seen in the cytoplasm of endothelial cells. Studies concerning tumors have shown that neoplastic cells and their extracellular matrix can also stain positive for C4d. For example significant C4d immunostaining has been reported in tumor cells of papillary thyroid carcinomas [6] and around neoplastic follicular dendritic cells in follicular lymphomas [7]. In the central nervous system (CNS) C4d staining has been observed in both glial and endothelial cells, but yet to our knowledge no study has been made to widely examine the appearance of C4d in astrocytic tumor tissue.

In this study, we examined complement activation in grade I-IV astrocytomas utilizing C4d immunohistochemistry. The purpose was to compare the appearance of C4d staining between different grades of astrocytomas and between primary and secondary resections of these tumors. Also complement activation on microvessel endothelium of astrocytomas was examined. One aim was also to assess via C4d immunopositivity if the extent of complement activation correlates with patient survival. Yin et al. have shown that the complement system activates more easily in stressed endothelial cells than cells under normal physiological conditions [8]. We hypothesize that the pathological blood flow caused by the abnormal microvascular patterns in astrocytic tumors with vivid microvascular proliferation activate complement and cause them to stain more strongly by C4d than astrocytomas with less vivid microvascular proliferation.

C4d positive staining was seen in endothelial cell cytoplasm, as well as in astrocytoma cell cytoplasm and tissue extracellular matrix. Positive staining is demonstrated in Figures 1, 2, 3. C4d negative normal brain tissue is demonstrated in Figure 4. Of the 102 astrocytomas (grades I-IV), 76 (74.5%) stained C4d positive. The C4d positive tumor area fractions of astrocytomas were between 0–84.7%, the mean value being 6.7%. C4d positive endothelium area fractions were between 0–3.5%, with a mean value of 0.2%. It was a common occurrence that tumor cells and endothelial cells simultaneously stained strongly.

There was a difference in the C4d staining between pilocytic grade I and diffusely infiltrating grade II-IV astrocytomas. The C4d positive tumor area percentages were higher in pilocytic tumors of grade I than in grade II-IV diffusely infiltrating astrocytomas (p = 0.041, Mann–Whitney test). This statistically significant difference is demonstrated in the box-plot in Figure 5. The mean of C4d positive tumor area fractions in pilocytic astrocytomas was 18.4%. The second highest positive tumor area percentages were achieved by the glioblastomas with a mean value of 8.2%.

A significant correlation between the CD34 positive endothelial area fraction and C4d positive endothelial area fraction was found in the diffusely infiltrating grade II-IV astrocytomas (r = 0.381, p < 0.001, Pearson Correlation). Figures 3a and 3b demonstrate the simultaneous C4d and CD34 positive staining of capillaries. As seen in Table 2, the proportion of C4d positive endothelium to overall vascularity (C4d / CD34) appears to rise along with the grade of the tumor, although no significance was achieved.

A difference in the C4d positivity was found between primary and recurrent diffusely infiltrating grade II-IV astrocytomas. Compared to primary tumors, recurrent tumors had more C4d positive samples in the C4d EXT variable. In 46% of primary tumors the C4d EXT was over 0%, whereas 65% of recurrent tumors had an C4d EXT value over 0%. (p = 0.047, Chi-Square test.). Comparing the other C4d variables between primary and recurrent tumors, no statistically significant differences were found.

Survival

In the five year follow-up, the increasing intensity of C4d staining (C4d INT) in diffusely infiltrating astrocytomas of grades II-IV was associated with patients having a worsened outcome (p = 0.014, log-rank test). The mean lifetime after primary resection for patients in the C4d INT group 0 was 32.5 months (95% confidence interval 21.9-43.0), in group 1 the mean lifetime was 29.8 months (CI 21.2-38.3), in group 2 the mean lifetime was 14.1 months (CI 6.7-21.4) and in group 3 the mean lifetime was 9.7 months (CI 3.0-16.3). The patient survival in different C4d intensity groups are presented in Figure 6. C4d EXT did not correlate significantly with survival. In addition, Cox multivariate analysis showed that only WHO grade was an independent prognosticator (Exp(B) 5.666, p < 0.001, 95% confidence interval 3.069 – 10.459). Other parameters included to the multivariate analysis were C4d INT, C4d EXT and patient age.

The intensity of C4d staining and C4d positive tumor area fraction were found to correlate with the grade of grade II-IV diffusely infiltrating astrocytomas. The increasing intensity of C4d staining in diffusely infiltrating astrocytomas was also associated with a patient’s worsened outcome in univariate survival analysis. Compared to primary tumors, recurrent tumors had more C4d positive cases. We also found a correlation between the CD34 positive area fraction and C4d positive endothelial area fraction in diffusely infiltrating astrocytoma samples. Interestingly, our results showed that in pilocytic grade I astrocytomas, the C4d positive tumor area fractions were higher than in diffusely infiltrating astrocytomas of grade II-IV. This was even though in the diffusely infiltrating astrocytomas positivity linearly and significantly followed the WHO grade. Although C4d intensity was found to be a significant prognosticator in univariate survival analysis in the present study, it did not appear as an independent prognostic factor in the Cox multivariate survival analysis. Reason for this is probably the close association of the C4d intensity with the tumor grade, which is found commonly to be the best prognostic factor of the diffusely infiltrating astrocytomas [1]. In addition, we believe that our material of primary diffuse astrocytomas is perhaps too small for reliable Cox analysis.

Vivid microvascular proliferation can be seen in pilocytic astrocytomas and glioblastomas [1]. One simple reason to the high C4d expression in these tumors could be that the abnormally proliferative, leaky blood vessels elicit inflammatory changes resulting in complement activation. Furthermore, the extensive activation of the complement in abnormal blood vessels may consume the serum C4 component and this way explain the study finding of Gousias et al., in which they showed a reduced serum C4 in patients with grade II-IV glioma and an inverse correlation between serum C4 and glioma grade [10]. However, in the case of pilocytic astrocytomas, the high C4d positive tumor area fractions could also be associated with complement inhibiting complement regulatory proteins (CRPs). There is evidence that benign tumors express CRPs in a much lesser extent than malignant tumors [11, 12]. Thus complement activation could take place more readily in benign than malignant tissues due to a lower expression of complement inhibiting CRPs, leading to further emergence of C4d. Although, C4d could also accumulate on pilocytic cell surfaces, due to longer lifetime of the pilocytic cells compared to malignant astrocytic tumor cells.

In malignant astrocytomas CRPs have been connected to the ability to evade immune destruction. According to Mäenpää et al. malignant glioma cell lines are exceptionally resistant to complement mediated lysis, mainly because of the widespread expression of CRPs on their cell membranes [13]. Glioblastomas can also avoid complement-mediated killing by the active production of soluble complement inhibitors factor H and factor H-like protein 1 [14]. This may be also why glioblastomas stain more extensively than grade II and III diffusely infiltrating astrocytomas.

Hanahan and Weinberg have recently nominated the capability of cancer cells to evade immune destruction as a hallmark of cancer [15]. In the case of malignant astrocytic cells, CRPs could be one element that helps these cells evade immune destruction. Increased C4d intensity at the complement activation site may reflect thus the ineffectiveness of the immune destruction system against cancer cells, which leads to chronic activation of the complement system. Chronically activated complement can proceed towards inflammation [16]. Hanahan and Weinberg illustrate in their recent article that inflammation and genome instability are the enabling characteristics in the acquisition of core cancer hallmark capabilities, allowing cancer cells to survive, proliferate and disseminate [15]. Cytokines, free radicals, prostaglandins and growth factors are a part of the activated inflammatory complex. They can cause alterations in the creation of DNA mutations in tumor suppressor genes and failure in normal protein synthesis, causing defects in the function of the cell, leading to progression of cancer [17, 18]. Understandably all this can lead to worsened patient outcome.

The univariate survival study results, which showed that a worsened outcome is associated with intensified complement activation in grade II-IV astrocytomas, can thus reflect the effect of the pathological microenvironment provided by inflammation. Also, since the inflammatory microenvironment allows cancer cells to survive, proliferate and make the cell genome unstable, it is possible that the inflammatory microenvironment could also enhance the recurrence of tumors after primary resection. In our study, recurrent diffusely infiltrating astrocytomas had more C4d positive cases than primarily resected ones, supporting this supposition.

This study shows that the intensity of complement activation, detected by C4d positive staining, is associated with the grade of grade II-IV diffusely infiltrating astrocytomas and the primarity of resection. There is also an association between complement activation and the prognosis of patients suffering from diffusely infiltrating astrocytomas. The worsening of outcome can be connected to the microenvironment changes, which are caused by the inflammatory mediators evoked by complement activation.