Long-term survival outcomes of pineal region gliomas

This project has been approved by the Ethics Committee of HUH. This is a retrospective study of all consecutive histologically confirmed gliomas operated in HUH, between 1997 and 2015. Pre- and postoperative functional status measured by the modified Rankin Scale (mRS), and all the therapeutic modalities for the treatment were retrieved from hospital records. Radiological information of the gliomas was acquired from IMPAX 6.7.0.4511 (Agfa, Mortsel, Belgium). Final status of the patients was analyzed in July 2018. The extent of the surgical resection of the tumors was evaluated in the postoperative magnetic resonance imaging (MRI) studies. The term gross total resection (GTR) refers to the absence of any residual lesion in the postoperative T1WI, T2WI, or FLAIR MRI sequences. Subtotal resection (STR) means just a small residual lesion usually very attached to the neurovascular structures; and partial resection (PR) represents less than 95% of tumor removal. The American Academy of Pediatrics recommends 21 year-old as the upper age limit for the evaluation of pediatric patients [11, 12]. However, based on the distribution of our study population, we classified the patients as 5-year-old or younger and 20-year-old or older patients.

R programming environment (© R Core Team, Boston, USA) was used for statistical analysis. We performed a general description of the study population categorized into 2 groups: the low-grade I-II nondiffuse glioma and the high-grade diffuse grade II-IV glioma groups. Initial uni- and bivariate analysis of the data was performed for identification of factors predicting clinical outcome. Fisher, Wilcoxon, or Spearman correlation tests were utilized wherever appropriate. Univariate survival analysis related with the diagnosis and with the extent of resection was performed by the likelihood ratio Cox model. The multifactorial survival analysis was determined by Cox survival model. The raw p-value cutoff for significance in all tests was set at 0.05. Adjustment of the p-value was performed with Benjamini–Hochberg procedure (α = 0.1).

The comparative analysis between the diffuse and nondiffuse gliomas is presented in Table 2. The clinical presentation of the patients was related with acute or progressive hydrocephalus symptoms (88%), visual and ocular detriment (41%), memory disturbances (24%), and motor deficits (12%). Detailed information of the patients is attached as appendix of this paper. The supracerebellar infratentorial approach was used in most of the patients. However, other approaches complemented multiple surgeries in two giant-PA patients. Around 50% of the patients underwent preliminary shunt for the treatment of hydrocephalus. An average of two additional shunt-related procedures per patient were observed in these patients. Two PA patients had five and 10 shunt-related surgeries. Some procedural complications were present in two patients of the diffuse glioma group (shunt infection, infarction and pneumocephalus); and in 5 patients of the nondiffuse glioma group (wound infections, meningitis, and pneumocephalus).

The immediate postoperative evaluation showed a three-year-old WHO grade II glioma patient with poor neurological response after surgery due to tension pneumocephalus that required surgical decompression. Two diffuse glioma patients showed a new Parinaud´s syndrome and new double vision, respectively. Other two diffuse glioma patients did not present new neurological deficits. The immediate evaluation of the non-diffuse gliomas revealed four patients with new postoperative disfunctions (Parinaud´s syndrome, double vision with mild hemiparesis, and two patients with only new mild hemiparesis). At the last clinical evaluation, the diffuse glioma patients deteriorated progressively and died. One PA patient who underwent multiple surgeries harbored eye movement disorder associated with postoperative epilepsy controlled by medication. Another two-year-old PA patient who underwent primary radiotherapy and brachytherapy developed psychomotor developmental delay, right hand ataxia, and multiple meningiomas. The patient died 27 years after initial treatment with multiple surgeries for the tumor, for the shunt disfunctions, and for the treatment of the meningiomas. One patient had walking limitations after hip surgery unrelated with the neurosurgical procedure. All the other nine non-diffuse glioma patients harbored very minimal symptoms such as slight double vision (2 cases), sporadic headache (4 cases), or did not report symptoms (3 cases).

Two cases had early death within six months after surgery. The WHO grade II glioma patient died three months after surgery with an extensive delayed postoperative infarction in the territory of both posterior cerebral arteries. The patient underwent complete microsurgical resection of a large lesion (35 × 26 × 38 mm). The immediate postoperative imaging did not show signs of ischemia, but tension pneumocephalus that required neurosurgical decompression. The patient remained stable but with poor functional status. Two days later, MRI studies showed signs of infarction that increased in size for the following imaging. Another patient with a GBM that underwent a subtotal resection presented a rapid recurrence of the small lesion with diencephalic compromise in the MRI studies and died 5 months after surgery.

The quantitative evaluation of the pre- and postoperative functional status of the patients demonstrated overall improvement at the immediate postoperative mRS measured at hospital discharge of the patient (Wilcoxon test, p = 0.0008) and last clinical FU (Wilcoxon test, p = 0.03) (Fig. 1). One patient with a mRS-2 at the last clinical evaluation underwent hip surgery before glioma diagnosis. The tumor size did not associate a strong correlation with the preoperative (Spearman test, R = -0.25, p = 0.34) nor postoperative (Spearman test, R = 0.15, p = 0.8) functional status of the patients.

The overall mortality, similar to the disease-related mortality of this series, was established as 35.3% at a median (interquartile range) FU of 83 (52 – 148) months. The overall mortality for nondiffuse low grade gliomas was eight percent at a median (interquartile range) FU of 138 (78 – 175) months with a unique PA patient dead at 324 months from the initial treatment. All diffuse grade II-IV glioma patients died at a median (interquartile range) FU of 9 (5–12) months. The 5-year mortality was 100% for diffuse gliomas and 0% for nondiffuse gliomas (Fig. 2).

Although the number of treated patients was small, the extent of the surgical resection did not impact on the disease specific survival rates of nondiffuse (p = 1), diffuse (p = 0.2), nor all gliomas (p = 0.6) under the likelihood ratio Cox model (Fig. 3). Moreover, the extent of the resection was similar among the different histological types of gliomas, and among the diffuse and nondiffuse gliomas (Table 2). Overall, 15 of 17 patients underwent GTR (six PA, one ependymoma, one grade II glioma, one grade IV glioma) or STR (three PA, one ependymoma, one grade III glioma, one grade IV glioma), and only two patients (the cystic component of a giant PA, and one grade III glioma) underwent PR. On the other hand, the evidence of tumor at the last MRI of patients who underwent incomplete resection was significantly higher than those who underwent GTR (Fisher test, p = 0.02). The extent of resection did not correlate with tumor size (Wilcoxon test, p > 0.05). However, tumors with median (interquartile range) large diameters of 30 (23 – 35) mm tended to GTR compared to those with median (interquartile range) large diameters of 35 (26 – 44) mm that underwent incomplete removal. None of the non-diffuse gliomas presented progression of the disease after surgery. The intracranial progression of the diffuse gliomas was as follows: The grade II glioma did not present recurrence. The two grade III gliomas presented local recurrences without distal recurrences. One grade IV glioma presented local recurrence without distal recurrence after subtotal resection; and the other grade IV glioma patient did not present local recurrence, but multiple foci of distal recurrence in the posterior fossa after complete resection of the tumor.

The best predictor of mortality in univariate Cox model analysis, the WHO grade of the tumor (p < 0.001), was the single significant variable (p = 0.02) after multivariate Cox model survival analysis with extent of resection, age, tumor volume, and preoperative mRS (Fig. 4). We could not properly incorporated adjuvant radiochemotherapy for multivariate analysis of mortality since adjuvant radiochemotherapy protocols of the gliomas, particularly the diffuse ones, widely varied between the patients (Table 1).

Details of pineal region gliomas in the pediatric population are presented in Table 3. Four children under the age of five (one female and three males) harbored gliomas (three PA and one grade II diffuse glioma) of the pineal region. The tumor size in children looked significantly higher compared to adults and might be focus of further research.