Introduction

Low-grade gliomas (LGGs) comprise 30–50% of brain tumors in children, occurring anywhere within the central nervous system (CNS). The commonest are astrocytic tumors with pilocytic astrocytoma being the most common [1, 2]. Cerebellar astrocytoma is the most prevalent site accounting for 15–25% of all CNS tumors in children [1, 2].

Treatment strategies for childhood LGG have changed considerably over the past few decades. Although observation is an option, particularly for patients with neurofibromatosis type 1 (NF1) and/or non-progressive tumors, surgical resection remains the gold standard and is considered curative, especially when complete resection is achieved [3,4,5]. The ability to achieve complete resection is largely dependent on tumor location, complete resection rates therefore varying between studies [6,7,8].

Chemotherapy is used both as salvage therapy for progressive disease and as first-line therapy, particularly given the efficacy and safety of low-toxicity regimens [9,10,11].

Radiotherapy is reserved for recurrent/progressive tumors due to the significant late sequalae [12]. Hence, radiation/progression-free survival (RPFS) is increasingly reported as a primary endpoint outcome measure [10, 13].

Ten-year overall survival (OS) for childhood LGG exceeds 90% [14]. Complete surgical resection is the most consistent positive prognostic factor for OS and progression-free survival (PFS), particularly for hemispheric tumors [15,16,17]. Other prognostic factors include NF1 status, tumor site, histology, and grade [18].

Given the high OS rate, treatment-related long-term sequalae must be considered when making treatment decisions [14]. This requires a timely complex multidisciplinary approach, with early detection and referral in the community, multimodality therapy, and long-term follow up. Therefore, managing childhood LGG can be challenging in resource-poor developing countries. However, Saudi Arabia is a resource-rich developing country and we hypothesized that the outcomes of childhood LGG would be similar to those reported by developed countries. We conducted a single institutional retrospective study to explore the demographics and clinical characteristics of childhood LGG to identify challenges faced in developing countries that may impact on outcomes.

Methods

This retrospective cohort study enrolled 60 consecutive children with LGG between January 2001 and June 2016 at the Princess Norah Oncology Center, King Abdulaziz Medical City-Jeddah, Saudi Arabia. All children aged 14 years or younger with LGG (grade I or II as per WHO revised classification system [1]) were eligible for the study. Radiological diagnosis of LGG was made only for chiasmic lesions that had characteristic radiological appearances, especially in the setting of NF1. Histology was available for all other tumors. Patients with progressive/recurrent LGG and patients with previous history of surgery or chemotherapy were eligible. Patients with missing data were excluded. The King Abdullah International Medical Research Center and Ethics Review Committee approved this study (KAIMRC ref. no. RJ16/012/J). Data collected included age, gender, date of diagnosis, NF1 status, presenting symptoms and duration, tumor location, histolopathology, initial therapy, surgery, radiotherapy and chemotherapy details, treatment-related toxicities, outcome, and long-term sequelae.

Definitions

Complete surgical resection: complete tumor removal as reported by the operating neurosurgeon and confirmed by postoperative neuroimaging. Subtotal/partial resection: 50–99% reduction in tumor volume, with < 50% reduction in tumor volume considered as biopsy only. Date of progression: date of the neuroimaging study that confirmed radiological/clinical progression. Relapse: reappearance or progression of disease at the original site or the appearance of new intra- or extra-axial tumor confirmed by neuroimaging. Complete response (CR): disappearance of all identifiable disease. Partial response (PR): 50% or greater reduction in maximal tumor volume (three-dimensional response criteria), with no new lesions or progression of any lesion. Minor response (MR): 25–50% reduction in tumor volume. Stable disease (SD): < 25% change in tumor volume and progressive disease (PD) as an increase of ≥ 25% in tumor volume in any site of residual tumor, or reappearance of tumor at any site. The Dodge anatomical classification system was used to define optic pathway tumors. [19]

Radiology review

All images were reviewed by a dedicated pediatric neuroradiologist. Responses were assessed according to the SIOP-LGG 2004 protocol criteria. Tumor volumes were estimated using the following formula: axial × coronal × sagittal × 0.5, with the largest diameter in each plane used [20].

Treatment approach

Biopsy was recommended for all patients except those with typical chiasmic tumors, especially in NF1 patients, where clinical diagnosis was acceptable. Following surgery, patients with no evidence of residual tumor were observed while patients with measurable residual tumor were observed, given chemotherapy, or treated with radiation therapy. Patients with unresectable tumors were either observed or given initial chemotherapy, radiotherapy, or had combined chemoradiotherapy. Patients with evidence of disease progression/recurrence were treated by surgery, chemotherapy, radiotherapy, or a combined approach.

Statistical analysis

Time-to-event endpoints were defined as follows: OS represented the time from diagnosis to death from any cause; PFS represented the time from diagnosis to disease progression or death from any cause; RPFS represented the time from diagnosis to disease progression, radiation therapy, or death from any cause, whichever occurred first. Data were analyzed using SPSS software (Chicago, IL). Descriptive statistics including numerical frequencies, percentages, and measures of central tendency were used to report demographic and clinical characteristics. Overall survival, PFS, and RPFS probabilities were estimated using the Kaplan-Meier method. The log-rank test was used to compare survival probabilities. P values < 0.05 were considered significant.

Results

Of the 60 patients initially identified, one had incomplete data. Thus, 59 patients were included in the final analysis (Fig. 1). The clinicopathologic characteristics are shown in Table 1. Median age at diagnosis was 6.0 years (0.5–14). Cerebellar tumors were most common (23.7%), among which pilocytic astrocytoma was the dominant pathology (92.9%). In contrast, cerebral tumors showed diverse pathology including ganglioglioma, PXA, oligodendroglioma, and pilocytic astrocytoma. All patients with NF1 had tumors in the HOP location (one anterior, one posterior, and four both). Histopathology was available in two patients and confirmed pilocytic astrocytoma.

Fig. 1
figure 1

Study population and treatment strategy. Sx surgery, XRT radiotherapy, CT chemotherapy, No prog no progression

Full size image
Table 1 Clinicopathological characteristics and survival outcomes of children with low-grade glioma
Full size table

The duration of symptoms prior to diagnosis was > 6 months in 22 (37.3%) patients and < 6 months in 37 (62.7%), including three asymptomatic patients. Headache was the commonest presenting symptom (62.7%), followed by vomiting (50.8%), visual symptoms (42.4%), ataxia/cerebellar manifestations (32.2%), seizures (28.8%), hemiparesis (18.6%), cranial nerve palsies (6.8%), torticolis (5.1%), irritability (3.4%), and decreased level of consciousness (3.4%). Other manifestations included tremor, loss of milestones, developmental delay, decline in school performance, behavioral and sleep changes, earache, and polyuria.

Treatment strategies

Seventeen (28.8%) patients were initially observed, 21 (35.6%) had initial surgical intervention, 8 (13.6%) had chemotherapy, 3 (5.1%) had radiotherapy, and 10 (16.9%) had initial combination therapy (Fig. 1). At a median follow-up of 3.5 years, 40 (67.8%) patients have not required therapy beyond the initial treatment.

Observation

Seventeen (28.8%) patients were observed initially with 13 being of hypothalamic and midline location. None were cerebellar, four were patients with NF1, and three patients with tuberous sclerosis complex. Twelve patients (71%) did not require further therapy. Five progressed, of which three had subsequent surgical resection and required no further therapy, one was given chemotherapy and had stable disease, and one (bithalamic) died without further therapy. None of the initial observation group required radiotherapy (Fig. 1).

Surgery

A total of 36 (61%) patients had surgical intervention during their course of treatment. Among these, complete resection was achieved in 8 (22.2%), subtotal/partial resection in 20 (55.6%), < 50% resection in 7 (19.4%), and 1 patient had missing resection details. Twenty-seven (75%) patients had residual disease: 22 (81.5%) had residual disease ≥ 3cm3 and 5 (18.5%) had residual disease < 3cm3. Twenty-three patients (39%) required ventriculo-peritoneal shunt insertion.

Twenty-three out of 25 cases with hemispheric (cerebellar and cerebral) disease had surgery. Complete surgical resection was achieved in 8 (34.8%), subtotal/partial resection in 13 (56.5%) < 50% resection in one patient, and 1 patient had missing surgical details. Surgery was attempted in 2 out of 14 cases with midline tumors, and 11 out of 20 cases with HOP tumors.

Chemotherapy

Eighteen patients (30.5%) received 32 chemotherapy regimens. Vincristine/carboplatinum was given 15 times in 14 patients, vinblastine was used 12 times in 10 patients, thioguanine-procarbazin-CCNU-vincristine was used in two patients, temozolamide in was used two patients, and vinorelabine was used in one patient. In addition, mTOR inhibitor therapy was used in one patient with SEGA. Carboplatinum allergy was recorded in 41.7% of patients who completed their carboplatinum course.

Among six patients with NF1, two had “upfront” chemotherapy due to symptomatic disease. These patients progressed requiring second-line therapy. One patient died at the time of progression with an intra-tumor bleed and brain herniation.

Radiotherapy

Nine (15.3%) patients received radiation therapy during their course of treatment. Three had bithalamic disease, two had PXA, three received radiation for operative residue, and one with HOP was given radiation therapy upon progression. Seven patients received radiation as primary/adjuvant therapy and two patients as salvage therapy at the time of recurrence/progression.

Survival analysis and prognostic factors

After a median follow-up of 42 months (range, 5–187 months), 5-year OS, and PFS were 90.6 ± 4.7 and 54.3 ± 8.4%, respectively (Fig. 2). Although anatomic site was not predictive of survival (p = 0.414), all patients with cerebellar location survived. Pathology was not predictive of survival (p = 0.76) or progression (p = 0.46). Only one patient with pilocytic astrocytoma died, of intra-tumor bleed and brain herniation, giving a 5-year OS of 94.7 ± 5.1% for these tumors. All four patients with metastatic disease survived. Differences in OS, PFS, and RPFS rates stratified by age, sex, anatomic site, duration from onset of symptoms, NF1 status, histology, WHO grade, and presence of VP shunt did not reach statistical significance (Table 1). However, 5-year PFS rates were 88.9 ± 10.5% for patients treated with initial combined therapy, 56.3 ± 15.6% for observation, 53.3 ± 14.0% for surgery, 22.9 ± 19.7% for chemotherapy, and 33.3 ± 27.2% for radiotherapy (p = 0.006) (Fig. 3).

Fig. 2
figure 2

Five-year overall survival and progression-free survival (n = 59). OS overall survival, PFS progression-free survival

Full size image
Fig. 3
figure 3

Progression-free survival by initial treatment modality

Full size image

The impact of initial therapy was analyzed also by tumor location. The 11 patients who were treated with initial chemotherapy or radiotherapy were all symptomatic with short duration of symptoms (< 6 months), suggesting aggressive behavior. The tumor was of midline/HOP location in all 11 patients and therefore was not amenable to initial surgery. Initial chemotherapy was used in 8 and radiotherapy in 3 patients (Table 2). These patients had significantly poor outcomes (Fig. 3). Surgery was the initial therapy in 21 patients. Of these, 19 were hemispheric tumors (14 cerebellar and 5 cerebral). Despite the majority (90%) of patients in the initial surgery group having tumors of hemispheric location, the 5-year PFS was 53.3 ± 14.0%.

Table 2 Distribution of patients by primary tumor site, histopathology, and initial therapy
Full size table

The extent of surgical resection predicted 5-year PFS: complete resection 87.5 ± 11.7%, subtotal/partial resection 27.9 ± 14.6%, and < 50% resection or no surgery 51.0 ± 12.6% (p = 0.047). Similarly, the extent of surgical resection significantly predicted 5-year RPFS (p = 0.013) (Fig. 4).

Fig. 4
figure 4

Radiation-free/progression-free survival by extent of surgical resection. RPFS radiation-free/progression-free survival. Complete surgical resection was defined by the operating neurosurgeon and confirmed by neuroimaging. Subtotal/partial resection was defined as 50–99% reduction in tumor volume. Biopsy was defined as less than 50% reduction in tumor volume

Full size image

Twenty-seven patients had residual disease. Of these, 16 were observed, 7 received adjuvant chemotherapy, and 4 received adjuvant radiation therapy. The corresponding 5-year PFS by treatment modality for patients with residual disease was 12.5 ± 11.4%, 66.7 ± 19.2%, and 100%, respectively (p = 0.033) (Fig. 5).

Fig. 5
figure 5

Progression-free survival by post-surgical adjuvant treatment modality for patients with residual disease

Full size image

Discussion

This study compared the clinical characteristics and outcomes of childhood LGG in a resource-rich developing country with those in developed countries, identifying challenges in the management of childhood LGG in Saudi Arabia.

Clinicopathologic characteristics of patients in this study were similar to published international data [6,7,8]. Despite the high prevalence of consanguinity among this cohort, the frequency of neurocutaneous syndromes was comparable to the 10.5% NF1 and 1.6% tuberous sclerosis complex incidence reported in a German series [7]. International cooperative groups reported cerebellar location as the most common tumor site [6,7,8, 21], whereas HOP location was most frequent in our study. Thalamic tumors were observed in 11.9% of our patients, similar to the incidence of 7.2–10.8% in the UKCCLG, CNS9702, and Johns Hopkins’s series [8, 15]. However, primary bithalamic occurrence was overrepresented in our study (57% of thalamic tumors and 6.8% of all LGG) compared to a large French series (bilateral occurrence in 13% of thalamic tumors) [22]. Two of the four patients with bithalamic tumors in our study died, one progressed, and one was lost to follow-up. This poor outcome is consistent with published data reflecting the limited treatment options in bithalamic tumors as complete resection is usually not possible and the role of chemotherapy and radiotherapy is questionable [23].

The 5-year OS in childhood LGG exceeds 90% in all studies, including ours, as summarized in Table 3. However, the 5-year PFS observed in our study was inferior to that reported by the UK and Childhood Oncology Group (COG) study groups, although comparable to that reported in Jordan and the HIT-LGG1996 study (Table 3) [6,7,8, 24]. Differences in initial treatment modality between our study and other studies could explain this. Complete surgical resection is the strongest predictor of both survival and progression [14]. This study reports 5-year PFS of 87.5 ± 11.7% for patients who achieved complete surgical resection. However, only 61% of our patients had surgical intervention (Fig. 6) and only eight patients achieved complete surgical resection (22.2% of all surgeries and 36.3% of surgeries at cerebral/cerebellar sites). In contrast, the COG study reported complete surgical resection in 64% of the entire patient cohort and 74% of those with cerebral/cerebellar location [6]. In the COG study, complete resection resulted in 5-year OS and PFS of 99 ± 0.5 and 94 ± 1.4%, respectively, compared to 90 ± 3.7 and 53 ± 5.7% for those with a residual tumor ≥ 1.5cm3. Thus, in this cohort, outcomes following complete surgical resection were only slightly inferior to those reported in international series, the challenge being to increase rates of complete surgical resection in Saudi Arabia.

Table 3 Survival outcomes of childhood low-grade glioma reported in studies from developed and developing countries
Full size table
Fig. 6
figure 6

Summary of reports of percentage of patients treated with surgery and percentage achieving complete resection in childhood LGG. SEER US Statistics Epidemiology and End Results report [21], COG Children’s Oncology Group [6], UKCCLG UK CCLG [8], HIT-LGG German-speaking country study [7], KHCC King Husein Cancer Center in Jordan [24], PNOC Princess Noorah Oncology Center (present study)

Full size image

In this study, patients who were observed or received upfront surgery had an intermediate prognosis (Fig. 3), whereas other groups reported best outcomes for patients receiving initial surgery [6]. The low rate of surgical intervention in our study (Fig. 6) may therefore explain the lower 5-year PFS we report. The variability in outcomes existed even when hemispheric location was taken into consideration. Despite 90% of patients in the initial surgery group having hemispheric tumors, the 5-year PFS of patients in the initial surgery group was poor compared to international results [5, 6, 16]. This highlights the challenges faced in areas with limited availability of advanced neurosurgical technology with significant implications for outcome.

In this report, one third of surgeries were performed at referral hospitals outside the reporting institute, these referring hospitals only achieving 8% complete resection. Further studies are needed in developing countries to identify the challenges that contribute to this low complete resection rate. Delayed presentation and access to care may be contributory. However, free cancer treatment is provided for all nationals in resource-rich developing countries such as Saudi Arabia, and in the present study, the duration of symptoms did not impact on outcome. Ensuring all centers have the technology and expertise to achieve gold standard rates of surgical intervention and resection is therefore an important objective. Alternatively, prompt referral of patients to “centers of excellence” with advanced neurosurgical technology may improve outcomes.

Adjuvant therapy for residual disease significantly improved outcome (Fig. 5) with 5-year PFS of 100% for patients who received radiotherapy and 66.7 ± 19.2% for those who received adjuvant chemotherapy compared to 12.5 ± 11.4% for those with observation only. Adult data from the EROC-MRC study showed improved PFS with immediate postoperative radiotherapy [25]. However, pediatric data is less clear with some reports showing no difference in PFS for immediate postoperative irradiation compared to chemotherapy or observation alone in children with residual LGG. [26, 27] Although radiotherapy appeared effective for postoperative residue in this study, this should be interpreted with caution as our study was limited by its retrospective observational design and small number of patients. Furthermore, of the 36 patients receiving surgery during their treatment course, 81.5% had a residual tumor volume > 3 cm3. A majority (59.3%) of these patients had no further adjuvant therapy. The poor outcome in these patients not receiving further adjuvant therapy may therefore be explained by the high percentage of patients with a large residual volume. This may explain also why adjuvant therapy was associated with improved outcome, adjuvant therapy possibly being unnecessary if post-surgery residue is < 3cm3. Further prospective trials are warranted to study the role of adjuvant therapy in children with LGG with postoperative residual tumor.

Current treatment approaches aim to avoid/delay the use of radiation therapy in children with LGG due to the significant long-term adverse effects [28,29,30]. Accordingly, studies of LGG now report radiation- and progression-free survival as primary endpoints [10]. In our study, 15.2% of patients received radiation therapy compared to 8.1–18% reported by major cooperative groups (Table 3) [6,7,8, 21]. The 5-year OS in this study was 71% for patients who received radiation therapy, compared to 95% for those who did not receive radiation therapy (p = 0.06). This is in keeping with the multivariate analysis of OS reported in the SEER data that identified the administration of radiation as being associated with the greatest risk of death (hazard ratio = 3.9) [21]. In our study, 5-year RPFS was 47.5 ± 8.7%: a recent Canadian study reporting 5-year radiation-free survival of 87.0% with 5-year PFS of 53.2% using weekly vinblastine [10] .

Limitations and strengths

Although retrospective, this is a consecutive patient series with clearly defined inclusion and exclusion criteria. All radiological images were reviewed by a specialized pediatric neuroradiologist and histopathology was reviewed by a specialized neuropathologist, with national and international referral if any doubt remained. We suggest that independent central pathology review is warranted in LGG, especially in the context of retrospective studies, in particular as disease classification and definitions have been and are likely to continue to be updated serially by the WHO. The small number of patients in each subgroup limits the conclusions that can be drawn. However, these data show clear trends in keeping with published data, underscoring the urgent need for national and international collaborative efforts.

Conclusion

This study explored the clinical characteristics and outcomes of childhood LGG in Saudi Arabia to identify challenges in developing countries that may impact on outcomes. First, a significant number of children had long-standing symptoms before referral, highlighting challenges related to delayed patient presentation and access to care. Evaluation of referral systems and community awareness are potential targets for improvement. Second, we observed fewer surgeries and lower rates of complete resection compared to international series. This needs to be addressed given the clear value of complete resection. Third, adjuvant therapy improved the outcome for patients with “bulky” residual disease, resulting in OS rates comparable to international results. Low toxicity chemotherapy regimens to address residual tumor may be justified to avoid or delay progression and the need for radiotherapy. Finally, the present study emphasizes the compelling need for regional/international collaboration to better identify prognostic factors in childhood LGG.