The Radiologic Features of Cystic versus Noncystic Glioblastoma Multiforme as Significant Prognostic Factors

Seven cases were classified as cystic GBMs and fourteen were noncystic GBMs. The cystic GBM group consisted of three men and four women, and the noncystic GBM group included nine men and five women. Table 1 summarizes the results of the statistical differences between the cystic and noncystic GBM groups. There was a statistically significant difference in age (Mann-Whitney U test; p <0.05) between the cystic GBM group (mean age: 34 years, range: 2-18 years) and the noncystic GBM group (mean age: 52 years, range: 26-81 years). The median and range of the preoperative and postoperative KPS of the cystic GBM patient were 70, 30-80 and 70, 40-80, respectively. The median and range of the preoperative and postoperative KPSs of the noncystic GBM patient were 60, 30-80 and 65, 40-90, respectively.

Table 1
Clinical Characteristics of the Cystic and Noncystic GBMs

• Click for larger image

The tumor volume was calculated to be more than 70 cm3 in four of the seven cystic GBM patients (57%) and in four of the fourteen noncystic GBM patients (28%). Six cystic GBM patients (85%) had a well defined interface between the tumor and the normal brain tissue. The micrographs of the cystic GBM patients showed bizarre tumor cells with abundant eosinophilic cytoplasm and these cells were found in an edematous stroma. The pleomorphic tumor cells showed variations of size and shape. The large tumor cells had abundant eosinophilic cytoplasm with fibrillary processes (Fig. 2).

The other GBMs had poorly defined interfaces with tumor cells infiltrating into the brain parenchyma. The tumoral wall thickness was less than 2 cm in six of the seven cystic GBM patients (85%) and in three of the fourteen noncystic GBM patients (22%). The peritumoral edema was calculated to be less than 40 cm3 in five of the seven cystic GBM patients (71%) and in two of the fourteen noncystic GBM patients (14%). Two cystic GBM patients (28%) and eight noncystic GBM patients (56%) showed intratumoral hemorrhage. All the cystic GBMs and 71% of the noncystic GBM patients received tumor resection. The median and range of the resection rate of cystic GBM patients were 78% and 53~98%, respectively. The median and range of the resection rate of the noncystic GBM patient were 79% and 25~100%, respectively. Five cystic patients and six noncystic patients received combined radiotherapy and chemotherapy, respectively. The other two and four cystic and noncystic GBM patient underwent radiotherapy alone.

In summary, the cystic and noncystic GBMs differed significantly for peritumoral edema, the tumor interface and the tumor wall thickness (Fisher's exact test; p < 0.05), but there was no significant difference for tumor volume and intratumoral hemorrhage (Table 1). The postoperative cumulative survival outcomes for the GBMs groups are shown in Table 2. For those with cystic GBMs, the 6-month, 1-year, 2-year and 3-year cumulative survival rates were 100%, 100%, 67% and 50%, respectively. For the patients with noncystic GBMs, the 6-month, 1-year, 2-year and 3-year cumulative survival rates were 69%, 49%, 14% and 7%, respectively. For the patients with cystic GBMs, the median survival was 43.8 months (range: 21.8-74.1 months). For the noncystic GBM patients, the median survival time was 12.5 months (range: 0.1-49.1 months). The differences in median survival were statistically significant (log-rank test; p = 0.003) according to the Kaplan-Meier estimates (Fig. 3).

Fig. 3
A. Micrographs show the cystic GBM with the surrounding brain parenchyma. Bizarre tumor cells with abundant eosinophilic cytoplasm are found in an edematous stroma.
B. The pleomorphic tumor cells show variations of size and shape. The large tumor cells have abundant eosinophilic cytoplasm with fibrillary processes.

• Click for larger image

Table 2
Outcomes of the Cystic and Noncystic GBMs

• Click for larger image

Glioblastomas have historically been classified as primary or secondary. The terms primary and secondary glioblastomas were first used by the German neuropathologist Hans-Joachim Scherer (12). Primary GBMs have been termed de novo. This term refers to the lack of a precursor lesion resulting from a single-step malignant transformation. Secondary GBMs has been referred to as progressive GBM. They develop by malignant progression from a low-grade astrocytoma or anaplastic astrocytoma. The development time varies from months to years. Primary and secondary GBMs are distinct entities according to their different genetic pathways, the associated patient ages and the different responses to therapy. Based on genetic typing, some authors also refer to primary and secondary glioblastomas as glioblastoma types 2 and 1 (9, 12, 13, 14). Several authors have studied the difference between primary and secondary GBMs as related to the genetic pathways and the clinical and radiologic features. Kleihues and Ohgaki have classified primary and secondary GBMs by the different genetic types and they found that secondary GBMs tend to develop at a younger age. For the radiologic features, the existence and nonexistence of necrosis and cystic formation is an important factor to evaluate the subtypes of GBMs. In general, approximately 10% of the tumors of the CNS are accompanied by a cystic area (15). There is a growing interest in the cystic lesions that accompany brain tumors. A cyst is commonly found in low-grade astrocytomas and in most pilocytic astrocytomas. The cause of cyst formation remains controversial. One hypothesis has claimed that large cysts are malignant transformations of previously undiagnosed cystic low-grade glioma. Laws et al reported that the low grade astrocytomas that include cysts have a better prognosis (16). According to Satoshi et al, the prognosis for cystic GBMs was significantly better than that for noncystic GBMs. They reported that cystic GBM patients were significantly younger than the non cystic GBM patients (13). Other studies have concluded that the secondary GBM patients are younger than the primary GBM patients (12). These observations suggest that most of the cystic GBMs have clinical features that are like those of secondary GBMs. Accordingly, it has been postulated that the large cysts accompanying tumors develop by malignant transformation from undetected low-grade glioma. Other hypotheses to explain the presence of cysts in GBMs include necrotic degeneration of brain tissue or blood-brain barrier (BBB) disruption. Maldaun et al revealed that the cystic patients' clinical characteristics such as age and the clinical manifestations differ from the previously published data about GBMs and they were not the same as the features of secondary GBM (a long preoperative clinical history and younger patients). Accordingly, no clinical data has been found to support the hypothesis of malignant transformation (8, 9, 10, 17, 18). In addition, Lohle et al modified the second hypothesis in several of their studies. In the past literature, it was believed that tumor cyst formation was caused by degeneration or necrosis of tumor tissue. The postulated hypothesis of intratumoral cyst formation is secretion of fluid by the tumor cells and cystic degenerative change within the tumor. However, the more recent reports by Lohle et al have suggested that cyst formation is due to an edematous process caused by BBB disruption and it is not the result of necrosis, based on the chemical assessment. This hypothesis proposes that cyst formation is caused by exudation of blood plasma proteins and the accumulation of interstitial fluid in the brain parenchyma. Lohle et al revealed a high protein concentration in the cysts with a highly similar spectrum of proteins in the tumor cyst fluid and the blood plasma. Brain tumor cyst fluid analysis found that blood plasma proteins constituted a major fraction (92%) of the tumor cyst fluid proteins. Fifty-fold higher concentrations of proteins were present in the tumor cyst fluid compared with that of the cerebrospinal fluid. This data suggests there is increased barrier permeability due to BBB disruption followed by exudation of plasma proteins into the brain parenchyma with formation of edema and the transition of the edematous tissue into cystic tumors (11, 19, 20).

Most GBMs have massive edema in the peritumoral area. This edema is due to the infiltration of tumor cells into the white matter and the mass effect of the tumors themselves. Vasogenic edema is caused by exudation of blood plasma proteins and the accumulation of interstitial fluid in the brain parenchyma through a damaged BBB, and this is similar to that producing cyst formation. According to Satoshi et al, cystic GBMs had smaller peritumoral edema regions and sharper boundaries than did the noncystic GBMs. Sawaya et al revealed a significant correlation between tPA (tissue plasminogen activators) and the presence and amount of peritumoral edema (21). Our study revealed that the cystic GBM patients were significantly younger and they had more favorable survival outcomes than did the patients with noncystic GBMs. This is similar to Satoshi's results. They reported the prognosis for cystic GBM was significantly better than that for noncystic GBM and the cystic GBM patients were significantly younger than the non cystic GBM patients (13). These observations suggest that most of the cystic GBMs have clinical features that are similar to those of secondary GBMs. However, this is not good evidence that cystic GBMs are secondary GBMs. This hypothesis additionally could be tested by looking at the gene expression and the IDH1 mutation status.

This study has some limitations. First, it is designed to retrospectively collect data from the preoperative MR images and medical records. A small patient population was enrolled in this study. As mentioned earlier, it is difficult to distinguish cysts with necrosis, regardless of our several criteria. We also had some problems measuring the edema in some cases of infiltrating tumors due to the indistinct separation between the tumor and the surrounding edema.

In conclusion, the cystic GBMs had a well-defined tumor interface, a thin wall and minimal edema as compared with the noncystic GBMs. The cystic GBM patients were significantly younger than the noncystic GBM patients. In addition, our analysis demonstrated that the patients with cystic GBMs had more favorable survival outcomes than did the patients with noncystic GBMs.