In the present study we explored the prognostic significance of several volumetric parameters for the progression-free and the overall survival in glioblastoma patients. All patients were treated with the same radiochemotherapy scheme, in order to ensure the uniformity of the population. Concerning the efficacy and toxicity of the radiochemotherapy scheme our results are in accordance with current literature [22
As in most other studies dealing with glioblastoma patients [29
], we confirmed the importance of the performance status as a factor influencing both survival parameters. It is of note, however, that the extent of resection was not found to be statistically significant. This may be explained by the subjectivity of this particular assessment (surgeons' impression and qualitative rough estimation of postoperative CT images) and dictates the need for the application of volumetry with early (within 48 h) postoperative MRIs. We can further assume, in this context, that the extent of resection may not be significant, when it is expressed as a percentage of the initial volume, but it is probably the absolute volume of the tumor remnant, which is the determinant factor affecting the prognosis.
The volumetric parameters, which were assessed in this study, were the pre-radiochemotherapy net-enhancing tumor volume and the preoperative enhancing tumor, necrosis and T2-abnormality. We found that pre-radiochemotherapy net-enhancing tumor volume and preoperative volume of necrosis are significant predictors of outcome. Our results are consistent with prior reports on the significance of volumetric parameters of glioblastomas.
In a recent study reported by Saraswathy et al. [33
] the investigators evaluated the prognostic importance of MR markers (anatomic, perfusion, diffusion and metabolic) in pre-treatment (radiochemotherapy) scans in patients with GBMs. They found that all of the above parameters were associated with survival. In particular, the volume of contrast enhancing lesion was inversely correlated with survival.
In another study by Keles et al. [34
] the authors concluded that the volume of residual disease (VRD) at the beginning of chemotherapy was a significant predictor for both survival time and time to progression in patients with recurrent glioblastomas. These patients, however, were neither chemotherapy-naive, nor newly diagnosed. In a different volumetric study of the same group [35
] evaluating the effect of extent of tumor resection and VRD on survival, the authors reported that both of these parameters influenced significantly TTP and OS in patients with glioblastoma. This patient group is not uniform, though, as they did not all receive chemotherapy and the protocols are not noted. In addition, the volumetric measurements were not only performed in MR scans but CT images, so the results are not really comparable.
On the contrary, in the multivariate analysis of Tralins et al., investigated the use of 18 F-FDG PET for the guidance of radiation dose escalation in GBM patients, it was found that only the volume of uptake and not the volumes measured on MR images, including the T1-weighted gadolinium enhancement, were of prognostic significance for survival or time to progression [36
]. That disagreement with our results could be explained by the different time of MRI acquisition, since in the above study it was performed during the course of radiotherapy.
In a recent study of Cao et al. [37
] the authors concluded that only the vascular leakage volume measured 1-2 weeks prior radiation therapy in dynamic contrast enhanced T2*-weighted images is of predictive value for survival and not the volume of contrast enhanced lesion measured in T1 weighted (nondynamic) images. However, in this study only patients with residual tumor volume > 4 cm3
were included and not all of the patients received chemotherapy and that could explain the discrepancy with our results.
At this point, we have to note that the prognostic value of contrast enhancement has certain limitations. This has been illustrated in a recent study by Piroth et al. [38
], correlating the volumetric findings of MRI with those of positron emission tomography (PET) using O-(2-[(18)F]fluoroethyl)-L
-tyrosine (FET) and concluding that the latter has a stronger prognostic impact. To summarize, although our study shows that pre-RCT net-enhancing tumor has a prognostic value, the results from the literature are rather contradictory in this context.
Concerning the influence of preoperative enhancing tumor volume, although it was found, in univariate analysis, to exert a negative impact on OS and PFS, this effect disappeared when adjusted for other significant predictors in the multivariate analysis. Our results are almost identical with the results of Weber et al. [39
] in a study evaluating the prognostic factors in cerebellar GBM. Similarly, in a large retrospective study from Lacroix et al. [29
] the multivariate analysis of 416 patients with GBM did not show any correlation of preoperative tumor volume with survival. There are numerous other studies [40
] supporting the lack of prognostic importance of the preoperative enhancing tumor volume. On the contrary, Xue et al. concluded [43
] that the accurate preoperative measurement of tumor volume with computer-based three-dimensional reconstruction is an important prognostic factor in high-grade gliomas. The description of the regions of interest that were delineated is not mentioned (contrast enhancement, margins, edema etc.), and additionally, the population of the study is not uniform, comprising anaplastic astrocytomas and glioblastomas altogether.
The most logical explanation for the difference in prognostic significance between pre and postoperative enhancing volume is the variation of the extent of resection. Since extent of the resection is not primarily correlated with the volume of the tumor, but with other factors as well (e.g. location, infiltration of critical areas), there is no distinct correlation between those two volumes. We also have to take into consideration that a variable period of time usually intervenes between the operation and the initiation of treatment, during which tumor re-growth may occur, thus rendering the pretreatment volume the most decisive factor for survival.
Necrosis is the imaging hallmark of GBM [44
] and is believed to indicate rapid growth and malignant behavior [45
]. We found that the volume of preoperative necrosis is a significant independent prognostic factor that negatively affects progression free survival, confirming several earlier studies pointing out this negative impact [29
]. In this context, it was interesting to observe that tumors with a high rate of MGMT protein positive tumor cells were significantly less necrotic, a finding that may be related to the recently shown decreased tumorigenicity of MGMT expressing cells in preclinical models [46
]. However, there are recent volumetric studies that failed to show any correlation between the absolute volume of necrosis and survival [47
]. Our finding suggests that large volume of necrosis is indicative of a more aggressive phenotype, which is also in accordance with the well established pathological view that large necroses are associated with sinister prognosis. Moreover, the necrotic core is associated with hypoxia, which has been shown to be a factor of poor response to radio or chemotherapy, possibly due to up-regulation of vascular endothelial growth factor (VEGF) expression that stimulates angiogenesis [50
]. To our knowledge there is no volumetric study correlating the absolute volume of necrosis measured in preoperative MR scan in glioblastoma patients with PFS. Interestingly there was no correlation with overall survival. This could be partially explained by the different therapies following recurrence, as many of them contained anti-VEGF agents, which would be more beneficial for tumors with high VEGF levels [55
The last preoperative volume measured was the T2 abnormality (high intensity signal in T2-weighted MR sequence). That volume includes the necrotic core, the enhancing tumor and a perimetric zone, consisting of vasogenic edema and tumor cells [56
]. In our study there was no association between that volume and PFS or OS. This is in accordance with the results of Crawford et al. [47
] and Li et al. [49
]. That result seems straightforward since that volume includes a great amount of edema, which is correlated with the dose of the prescribed corticosteroids and, consequently, is not a reliable measure of tumor burden.
Our study, certainly, points out that the precise volume determination of anatomic parameters is still essential in brain tumor research. It is simple, accurate, cost-effective and easily applicable from most oncological specialties. There are also several other imaging modalities (perfusion and diffusion weighted MRI, proton MR spectroscopic imaging, PET-CT) which can play a more significant role, maybe more important than the volumetry of anatomic lesions [33
]. More specifically, the use of amino acid tracers in PET has been shown in recent studies to be superior compared to MRI both in planning volumetric resection [61
], as well as in predicting the outcome of glioblastoma patients prior and after RCT [38
]. In addition, the significant correlation of the pretreatment volume of enhancement with overall survival, leads us to conclude that shortening (no longer than 6 weeks) of the interval between surgery and initiation of radiochemotherapy could be of great importance to survival, since it overcomes the problem of tumor regrowth [63
]. This interval however, in the light of data presented by Blumenthal et al. [64
], should not be earlier than 4 weeks allowing enough time for recovery of the brain from the surgical injury and edema.
There are certainly some drawbacks that we need to mention. As it has been pointed out in our previous study [23
], the lack of homogeneity in the MRI scans, as well as the process of digitization, may cause several variations in tumor delineation and the subsequently determined volumes. In addition, as the MRI was not performed during the optional time window of the first 48 h after surgery, visualization of the enhancing lesion was probably affected by postoperative changes. Moreover, our analysis was based on a small sample size, which may invoke the usual statistical uncertainties.
In this study, we did not observe any association of MGMT related parameters with patient outcome. It should be noticed, however, that the MLPA method used to determine MGMT
promoter methylation yielded evaluable results in only half the samples available for this investigation; the method works on paraffin tissue extracts but it requires relatively preserved DNA, which we only could obtain in a limited number of cases, as evaluated by a multiplex control DNA PCR assay. In comparison to the usually applied MSP-PCR, MLPA has the advantage of providing information on the methylation status of multiple sites in the MGMT
regulatory region in a semi-quantitative manner [27
]. The degree of this epigenetic change seems to be of predictive value, since tumors with incomplete MGMT
promoter methylation are reported to fail on temozolomide, while those with complete methylation show a significant trend to respond to this treatment [65
]. The rate of tumors with complete methylation in our series was comparable to this report, but statistics could not be performed since the absolute number of cases was low. These results, as well as the lack of concordance among MGMT
promoter methylation status, mRNA and protein expression results, once again reflect the problems encountered when assessing MGMT status on routine histologic material, as already reported in previous comparative studies [66