Our hypothesis is that HMGA2 is highly expressed at the border zone. Due to hypoxia protein expression shifts to the invasive front of the tumor and causes infiltration and malignancy (26, 27). For this reason, we were determined to analyze the HMGA2 expression in the border zone. Moreover, we strongly believe that there could be a link between the microstructure and molecular genetics at the tumor border and the fact that most recurrences arise directly adjacent to the resection cavity (28). The importance of the molecular setting of glioblastoma and its influence on treatment recommendation and prognosis estimation is known and with MGMT and IDH two of these markers are stablished diagnostic marker for malignant glioma (5, 29). In previous studies HMGA2 proved to be correlated with bad prognosis and shorter PFS and OS (30). So far, no study examined and compared the HMGA2 expression in different parts of glioblastoma and hence little information is known about the actual expression cluster of HMGA2 in glioblastoma.
Our results confirm that HMGA2 is significantly higher expressed in glioblastoma than in normal brain tissue and HMGA2 is higher expressed in the border zone than in the tumor center. This fact is interesting regarding separation of healthy and normal brain tissue to malign glioblastoma cells. Tumor cells might be distinguished by the overexpression of HMGA2 (31). Single glioblastoma cells at the border zone of glioblastoma might be identified by visualizing these new markers such as HMGA2. This could allow us a better understanding and more information of the spreading and invasive character of glioblastoma and a better labeling of the border zone. Should HMGA2 prove to be a highly relevant and specific marker of the border zone, future approaches of intraoperative visualization by imaging or histopathological methods may show clear differentiation from normal brain tissue. This could improve the surgical outcome. It is still unclear whether HMGA2 is expressed in all glioblastoma types and whether imaging HMGA2 truly represents the invasive nature of the tumor and provides additional beneficial information. To know the exact difference of HMGA2 expression between glioblastoma tissue und healthy brain tissue, more tests on normal and healthy brain tissue are needed.
It would be worth considering correlating HMGA2 expression with other known hypoxia markers. An interaction between the markers is quite conceivable and could show more insight into the dynamics of the border zone.
Also, high HMGA2 expression in the border zone shows the tendency to correlate with quicker progression and shorter overall survival (Fig. 3 and Fig 4). The overexpression of HMGA2 in the border zone could contribute to the high recurrence rate at the surrounding border zone. The directly adjacent areas appear to be a good host for recurrences, as approximately 80% of recurrences originate locally from the tumor border zone, although it has been demonstrated that at the time of diagnosis some tumor cells have already spread distally in the brain and can even be detected systemically in the bloodstream (32). These results emphasize the importance of the border zone of glioblastoma. Obviously, the sample size and study design are not suitable for accurate survival data and is not conclusive, especially since no significant results could be presented. In addition, except for a few clinical data like age and sex, we did not consider secondary diagnoses or diseases that may also play a role on overall survival. Comparison with large databases that also include molecular genetics would be one way to add a little more power to the survival analysis (33). The cells profile of the tumor border with its gene expression profile such as HMGA2 might also be the reason why gross total resection with an extensive resection of the border zone shows significantly better results (34). A thorough and good resection of the border zone might be crucial and have a bigger impact than the resection of the tumor center. First tests showed that supramarginal resections in non-eloquent areas showed a survival advantage with the same quality of life (35). In the future, tissue samples obtained intraoperatively should possibly also be taken explicitly from the border zone of the tumor and sent to further examination. Should the diagnostics of the border zone tissue show an aggressive genetic profile and overexpression of HMGA2, more aggressive and focused radiotherapy in the adjacent areas might be considered to prevent rapid progression (36). Targeted therapy of the border zone remains difficult because eloquent areas or important structures are often adjacent and surgical removal is limited. Also intensified radiotherapy can also have significant side effects. Before HMGA2 or other molecular markers have a decisive impact on therapy delivery, the relevance of such markers must be clear. The data obtained in this study, recommend further investigation with the goal of developing better therapeutic options for the border zone of glioblastoma.
There is reason to believe that the overexpression of HMGA2 in the border zone is due to a niche at the border zone (37). It is becoming clearer that there seems to be structural differences in the border zone of glioblastoma (38). A high HMGA2 expression in the border zone could be one element of a very lively and active border niche, where glioma stem cells, oligodendrocyte progenitor cells and microglia seem to be overly present (39). A thorough understanding of this microstructure and its effect on glioblastoma and normal brain tissue could be promising.
HMGA2 expression did not vary significantly when stratified by clinical parameters. In our results HMGA2 expression showed no correlation in gender, age or MGMT methylation status. In comparable studies for other tumor entities such as renal carcinoma, the results also did not show any significant differences in sex, age and tumor size (40). Data on HMGA2 expression in glioma, have also failed to proof a positive correlation with gender and age (41). MGMT methylation status did not affect HMGA2 expression in our results. This supports previous data that also showed no relationship between HMGA2 and MGMT methylation status (25). Larger cohorts are needed to explore possible correlation to clinical parameters, although so far there is no indication for such a link.
It could be reasonable if therapeutic strategies were used specially to treat the border zone and thus eliminating as much from the vital tumor border as possible. Targeted inhibition of HMGA2 in other tumor entities and glioblastoma already showed some beneficial results by reducing endothelial-mesenchymal transition and thus decreasing stemness and invasion (42).In terms of surgery, the elimination and resection are mostly limited to the location of the tumor and navigational systems (43). The location cannot be changed, thus gross total resection or even supramaximal resection is dependent on good navigational systems and intraoperative monitoring. Increasing the quality of these tools should be aimed at so that the resection level and depth in the border zone could be elevated (44). Taking advantage of molecular markers in the border zone and combining the visualization of these markers with navigational systems, could increase the resection level in the border zone (45). Intraoperative real-time histology using Rama-Spectroscopy, could provide more clarity on the resection of the border zone in the future (46). Imaging of markers specific for the border zone of the glioma would be a useful next step to improve resection outcome.