We evaluated the expression of COX2 in the yet largest cohort of 1048 vestibular schwannomas including 115 schwannomas of NF2 patients and compared the expression with demographic factors as well as tumor extension, MIB1 expression and the intake of COX2 responsive medication.
ASA and COX2
We were able to demonstrate that there was no significant difference in COX2 expression in vestibular schwannomas with the intake of ASA, NSAIDs, glucocorticoids or other immunosuppressive medications. The association of tumor growth in sVS and inflammatory histopathological characteristics has been shown in a few studies. An increase of the expression of COX2 was observed in vestibular schwannoma tissue [
12] and a vestibular schwannoma cell line [
7]. Prostaglandin levels, a product of COX2, can be decreased with ASA and result in reduced cell proliferation compared to a normal Schwann cell culture [
7]. A significant decrease in tumor progression after ASA administration was also recently demonstrated in a mouse model [
26]. Additionally, de Vries et al. revealed an increased rate of CD163 positive macrophages in ten fast-growing sVS compared to a group of ten slow-growing sVS [
6]. In a recent study, Carvalho and colleagues demonstrated a bidirectional regulation of COX2 expression in canine mammary tumor cells and CD68 positive macrophages, assuming a tolerogenic tumor microenvironment [
3]. Overall, there is growing evidence to support the importance of inflammation in vestibular schwannoma growth.
The role of immunosuppression in NF2 has been explored and has gained increasing interest in regard to the development of possible targeted treatment strategies. Merlin, the product of the deleted gene in NF2, is known to have an inhibitory effect on the mammalian target of rapamycin complex 1 (mTORC1), leading to increased mTORC1 activity in NF2 associated tumors [
16]. Growth inhibition by rapamycin, an immunosuppressant drug, was demonstrated using an in vitro cell model and a mouse model for NF2 [
10] followed by a phase II trial, where everolimus was applied to 10 NF2 patients, which failed to reach the primary endpoint of tumor volume reduction [
11]. However, the involvement of COX2 as a downstream target of mTORC1 has been described in tuberous sclerosis complex associated tumors and downregulation of COX2 showed antiproliferative effects [
20].
Three large retrospective studies have shown contradicting results regarding the effect of ASA intake on radiographic tumor growth of sporadic VS. One study demonstrated a clear reduction of tumor growth by volumetric measurement in 347 sVS [
17]. On the contrary Hunter et al. showed no effect of ASA intake on tumor growth in a series of 564 VS [
13]. A most recent study assessed 437 cases and found no impact on ASA as well [
21]. The Congress of Neurological Surgeons recently recommended the administration of aspirin for patients with vestibular schwannomas as an antiproliferative treatment [
32].
Our large dataset shows that the expression of COX2 is not significantly altered in vestibular schwannomas by regular ASA intake, besides the established regulatory impact of ASA on the expression of COX2 [
33,
34]. Interestingly, tumor tissue that was exposed to ASA treatment prior to surgery showed slightly higher expression rates but, as already mentioned, without statistical significance. The dosage of regular ASA intake can vary in other countries but the dosage for cardiovascular disease or prophylaxis of cardiovascular disease in Germany is 100 mg daily, which was the case for all 49 patients with regular ASA intake. Sporadic ASA intake due to headache was not considered as regular intake. The medication was given for a few months to several years prior to surgery. However, it is unknown how the dosage needs to be in vivo to exert a decrease of COX2 expression and a subsequent antiproliferative effect as observed in vitro [
7]. It is unlikely that the dosage that is used to sustain an antiplatelet effect will also be suitable to have an impact on COX2, especially since the antiproliferative regulation seems to be on a transcriptional level and not via an acetylation of the enzyme COX2 [
33,
34]. A prospective phase II trial that is currently recruiting patients for the investigation of aspirin efficacy regarding growth inhibition in vestibular schwannomas administers 325 mg twice daily (NCT03079999). This trial will very likely solve the question whether ASA should be recommended as an antiproliferative treatment option for VS patients. Nonetheless, our data does not support this principle.
It is necessary to stress, that our data showed that the expression of COX2 was significantly lower in NF2 patients, challenging the concept of ASA recommendation for this patient subgroup even more. However, it is unknown which level of COX2 expression can be regarded as possibly responsive to COX2 inhibition resulting in an antiproliferative effect.
Proliferation and inflammation
As a marker for tumor proliferation, MIB1 expression was quantified unbiased by computerized image analysis for all evaluated tumor samples and was found to be significantly associated with higher COX2 expression, suggesting a link between COX2 expression and tumor proliferation, which was previously described in a retrospective analysis of thirty VS [
12]. Now we can confirm a robust association in a large retrospective cohort.
Overall, the MIB1 index is usually quite low within this slow growing tumor entity, with only 4 cases with indices over 4% in this study cohort of 1048 cases. Some may argue that the tumor growth may not be properly represented by MIB1 indices in vestibular schwannoma. However, in a case series with 15 clinically aggressive vestibular schwannomas that all showed a radiological growth rate of > 15 mm/year, MIB1 expression was significantly increased compared to a control group [
14]. It has also been described that the risk of regrowth after VS resection is associated with an increased MIB1 expression [
9,
15].
It should be kept in mind that VS growth is not induced by the eukaryotic cell-cycle alone. There is growing evidence of inflammation as a significant factor in tumor growth and proliferation in VS [
6,
19,
26,
29]. It remains unclear if the widely spread COX2 expression in almost all vestibular schwannomas of this cohort, is due to tumor growth or an expression of inflammation. Certainly, a higher MIB1 expression may also be the result of increasing inflammation and macrophage invasion which also have proliferative properties.
It is established that Inflammation plays an important role in cancer and there is evidence of a protective effect of NSAID intake in certain cancers [
27,
30]. More specific COX2 expression was identified as a prognostic factor in a subset of breast cancer patients [
5]. In urethral bladder cancer COX2 expression was shown to be over 28-fold increased when compared to normal urethral mucosa and is associated with tumor invasion and recurrence [
1]. The observed differences in COX2 expression in this cohort are far from such an increase which is due to the methods applied but significant differences were found. This phenomenon needs further profound assessment but has the potential to reveal new therapeutic targets like tumor associated macrophages [
2,
24]. However, in this cohort regular intake of ASA, NSAIDs, glucocorticoids or other immunosuppressants did not show a decreased level of COX2 or MIB1 expression suggesting that the commonly used ASA dosages have no beneficial effect. On the contrary MIB1 indices were significantly higher in tissue samples that underwent regular glucocorticoid treatment, a finding that cannot be explained at this point.
Tumor extension
While the initial vestibular schwannoma is restricted to the internal acoustic meatus, tumor growth has to overcome the bony border of the canal, resulting in dilatation of the canal with further growth. It is possible that this step during tumor progression may differ from tumor growth into the cerebellopontine cistern where there is less mechanical resistance. Accordingly, the tumor microenvironment may also be different.
The odds ratio between T1 and T2 showed the highest rate for COX2 expression with a subsequent decrease of the odds ratio between T2/T3 and T3/T4. If this can be linked to the anatomical and mechanical aspects of the surrounding tissue during tumor growth is unclear. In this cohort there was no association between tumor extension and MIB1 expression. Suggestions have been made that the current tumor size is a factor influencing future tumor growth [
21]. However, in the study by MacKeith and colleagues, growth was defined as 2 mm increase in diameter, which does not represent the true percentual increase of the tumor volume and is obsolete for tumor volume and growth assessment [
21,
31].
Limitations
Overall, the retrospective design is the major limitation of the study. Especially the retrospective collection of clinical data can be problematic due to incomplete documentation. However, the regular intake of aspirin was well documented since it is crucial to discontinue substances that can compromise platelet function and coagulation before intracranial surgery. In all 49 patients regular aspirin intake was prescribed due to diagnosed cardiovascular disease or increased risk profile for cardiovascular disease. Thus, the subpopulation of patients with aspirin use was older. Due to the retrospective nature of the study this selection bias can not be completely neutralized. However we have addressed this issue by incorporating the factor age into the multivariate analysis which showed robust results. Since the dosage of regular ASA intake can vary in other countries, the dosage for cardiovascular disease or prophylaxis of cardiovascular disease in Germany is 100 mg daily which was the case for all 49 patients with regular ASA intake. The medication was given for a few months to several years prior to surgery. Sporadic ASA intake due to headache was not considered as regular intake.
Additionally, retrospective studies on benign tumor tissue are always prone to a surgical selection bias. Large tumors are more likely to be treated surgically without further follow up scans. Therefore small, clinically stable or non-growing tumors that did not need surgical treatment are underrepresented in this cohort where tissue markers were assessed. This bias cannot be controlled. Overall, within this large cohort of 1048 cases there is still a significant number of small T1 and T2 tumors that underwent surgical resection.
The tissue microarray method has known limitations mostly due to the analysis of a small portion of the tumor that may not be representative. Therefore, we extracted two 1000 μm tissue cylinders of representative tumor tissue for each tumor and avoided areas with non-tumorous tissue (stroma, vessels, etc.). The amount of extracted tissue is sufficient since there is usually no tumor tissue heterogeneity in vestibular schwannomas except for Antoni A and B growth patterns.
The immunohistochemical staining of COX2 may not be the optimal method to measure the effect of ASA on COX2 in vestibular schwannoma tissue. However, it has been known for quite some time that ASA has a far-reaching regulatory impact by inhibiting the expression of COX2 [
33,
34]. Thus, the immunohistochemical staining is adequate to measure the effect of ASA on COX2 levels.