Pilocytic astrocytoma (PA) is the most frequent brain tumor in children and adolescents [1,2]. It mostly occurs under age 20 and the preferential location is the cerebellum and hy-pothalamic-optic pathways [3,4,5]. Neurofibromatosis type 1 (NF-1) is associated with PA in 15-20% of the cases [6]. PA is considered a benign neoplasm graded I by the World Health Organization classification [7,8]. Although it is a relatively frequent brain tumor there are few large series reported in the literature and most of them concern retrospective studies of cases proceeding from several institutions [3,4,5,9,10,11,12,13]. Several aspects concerning its biology remain obscure or are matter of controversy [14,15,16,17,18,19]. The influence of clinical variables on prognosis is not well understood. Therapeutic strategies for residual tumors after surgery are not uniform [20,21,22,23,24,25,26]. The use of adjuvant therapy in these cases is largely controversial [27,28,29,30]. Also, it is not known whether there is a biological difference in its behaviour according to geographical areas or different ethnic populations. In a recent study, Tihan et al. [5] found some differences among patients proceeding from four Medical Centers in three distinct countries. They point out that it is important to determine whether PA from different regions is similar enough to be included in a single cohort for clinical trials.

Our purpose is to present the clinical, laboratorial, and therapeutic aspects of a cohort of Brazilian patients studied in a single institution, and to compare the findings with the most relevant series of the literature.

We studied all patients with PA diagnosed between 1984 and 2014 at the Santa Casa of São Paulo School of Medicine. Clinical, radiological, therapeutic and follow-up data were collected. Inclusion criteria were patients under age 16 with tumors located in the brain, followed-up for at least 6 months, whose medical records contained appropriated data for pertinent analysis. Pathological diagnosis was reviewed by one of us (SR). Clinical factors which were correlated with prognosis included interval between onset of symptoms and diagnosis, age at diagnosis, gender, ethnic background, location of the tumor, image characteristics, extension of resection, and adjuvant therapy. MRI was performed one month after surgery and each 6 months in the 3 first years of follow-up and then annually. Progression was defined as an increase of the size of the residual tumor or appearance of new clinical neurological symptoms or signs related to the tumor.

Chemotherapy included carboplatin and vincristine according to the protocol proposed by Packer et al [27,28]: Carboplatin was administered at an intravenous dose of 180mg/m2 for four consecutive weeks followed by three week-rest period and then renewed for three more weeks. Vincristine was used at a dose of 1.5mg/m2/week for 10 weeks concurrent with carboplatin. The maintenance regimen consisted of four doses of carboplatin at 175mg/m2/ week and vincristine at 1.5mg/m2/week administered weekly for the first three weeks of each four-week cycle. There was 2-week rest between each maintenance cycle. . The regimen was continued for 12 cycles. Radiation therapy was delivered in conformational fractioned scheme (range dose: 40.5-54 Gy), in approximately 1,8 Gy daily, during five weeks.

Statistical analysis included Mann-Whitney and Krus-kal-Wallis tests was performed with SPSS (Statistical Package for the Social Science), version 13.0. For analysis of progression free survival (PFS) and overall survival (OS), the Kaplan-Meier method was performed. The study was approved by the institutional Ethics Committee.

The epidemiological and clinical findings in this series do not differ from those found in large series from different countries [4,9,10,12,13]. Two-thirds of the cases occur between 0 and eight years of age and 20% of patients was younger than 36 months at time of the surgery. Sex preponderance varies widely [3,4,10]. In the present series females accounted for 54% of the cases (1:1.2). Cerebellar tumors were the most frequent (52%). There was no preponderance as to the ethnic background. From these findings, it seems that there are no essential differences in the properties of PA in different geographic regions of the globe; an issue recently raised by Tihan and co-workers [5].

Tumor recurrence was more common in younger children. In our study ten patients were younger than 36 months at the time of surgery and 7 had tumor recurrence. Probably, this finding is due the fact that in most of the cases the resection was not total. These results were encountered in other reports [5,13].

As it seems now a common sense, the extent of resection is the main positive predictive factor [3,4,9,12]. Wisoff et al. analysed prospectively, for five years, a cohort of 518 children with low grade gliomas (76% PA). The progression free survival was 94% and 61% after GTR and PR, respectively. These patients had no adjuvant therapy [31]. In the present study, no tumor recurrence was observed in all patients but one who had GTR. In one case, the tumor recurred 8 years after surgery. This poses the issue of when a patient with GTR may be considered cured and for how long he or she must be followed-up.

Nowadays, the major issue concerning PA is related to the management of patients with residual tumor after surgery or of those whose tumor cannot be operated on. This issue comes from the fact that the biological behaviour of PA is unpredictable. Some tumors with or without adjuvant therapy may remain quiescent for decades whereas others show a more aggressive behaviour with several recurrences or dissemination in spite of adjuvant therapy.

The vast majority of PA seem to be very indolent in their biological behaviour. There were only two deaths (4%) due to tumor progression. Recurrence occurred in 14 patients (1 GTR, 10 PR, 2 biopsies and 1 not operated). From these, the tumor remained stable after reoperation associated or not to adjuvant therapy in eight patients, three of which were followed- up for more than 10 years, without recurrences. Only in 4 patients there was more than one tumor recurrence (two recurrences in 3 patients and three in one). Six patients experienced recurrence between 4 and 8 years after treatment whereas in the remaining 8, recurrences occurred within the two first years.

The interval between treatment and recurrence is unpredictable. For these reasons, at the current stage of our knowledge about the biology of PA, we think that these tumors should be treated very cautiously.

Total resection should be performed whenever possible. When operation is not possible for whatever reason or in cases with residual tumors after partial resections, a “wait and see” strategy with a careful clinical and radiological follow-up should be the rule. In case of recurrence, a new surgery should be performed. In case of new progression, chemotherapy in children below age 4 or 5 should be performed. For older children, radiation therapy may be considered in spite of the absence of evidence in the literature of the superiority of this method. In the present series, there were no statistical differences between both methods regarding PFS. It is worth of mentioning that five out seven of our patients submitted to chemotherapy experienced a sustained reduction of the tumor volume. The algorithm for therapeutic strategies recommended by the authors based on the data of this cohort is shown in figure 3.

The ideal method of treatment of PA will be attained only when their biology is fully understood or at least when predictors factors for PFS are uncovered. The epidemiological and clinical data so far analysed-and the present study is an example- failed to clarify this issue. Also, the studies focused on the neuropathology of PA have failed in demonstrating any undeniable predictor morphological feature [3,4,5,14,15,16].

In the last few years many studies on the molecular genetics of these tumors have emerged. The mitogen-activated protein kinase (MAPK) pathway has a central role in the tumorigenesis of PA [18]. The tandem duplication of chromosome 7q34 with gene fusion KIAA1549: BRAF was found in 55% of PAs without NF1 [32,33]. There are 5 known subtypes fusion, detected by PCR method: KIAA1549-exon- 16/BRAF-exon-9, KIAA1549-exon-15/BRAF-exon-9, KIAA1549-exon-16/BRAF-exon11, KIAA1549- exon-18/ BRAF-exon-10 and KIAA1549-exon-19/BRAF-exon-9[33]. Although these molecular studies open an important path towards the discovery of drugs targeted against PAs, they have, up to know, as the clinical ones, failed in finding any predictor factor [5,13,19,31,32,33].