Surgical outcomes of pediatric brain tumors in Sub-Saharan Africa: A systematic review

Background Pediatric Brain Tumors (PBT) are a common cause of cancer-related mortality globally. Contrary to high-income countries (HIC), survival rates in low-and-middle income countries (LMIC) remains low despite advances in neurosurgical care and diagnostics over the past decades. The aim of this systematic review was to investigate the surgical outcomes for PBT in Sub-Saharan Africa, and the distribution of PBT types. Methods A systematic review was conducted on PubMed, for all available literature on the surgical outcomes of PBT in Sub-Saharan Africa, published before May 3, 2022. Two reviewers performed abstract, full text screening and data collection independently, resolving any conflicts by consensus. Results The search yielded 256 studies, of which 22 met the inclusion criteria, amounting to a total of 243 patients. Nigeria was the country with most data. Only subgroups of patients could be extracted from 12 studies, and variables of interest in 6 studies had inconsistent sample sizes. The age centered around 9 years, and there were approximately equal number of girls and boys. The most common tumor was medulloblastoma, followed by craniopharyngioma and astrocytoma. There was large heterogeneity in the reporting of outcomes, and a trend was difficult to discern, considering the large number of different tumor types and different extents of resection. Discussion and conclusion Data is insufficient and inconsistent, precluding statistical conclusions. There is a need for more studies in the field.


Background
Pediatric brain tumors (PBT) are the most common cause of pediatric cancer related mortality worldwide (Adel Fahmideh and Scheurer, 2021;Pollack, 1994). Yet, advances in diagnostics and treatment in high-income countries (HIC) over the past decades have resulted in relatively high survival rates (Allemani et al., 2018). Unfortunately, this is not the case in low and middle-income countries (LMIC) (Allemani et al., 2018), where 90% of the pediatric population lives (Bhakta et al., 2019). Often, prevalence and incidence of PBT in these countries appear low (Stoeter et al., 2021), which has been suggested to be due to underdiagnosis (Stagno et al., 2014), and lack of surgical treatment (Dewan et al., 2018). The first line treatment for most PBT is neurosurgical resection of the tumor (Pollack, 1994), and the extent of the brain tumor resection may depend on the tumor type (Grewal et al., 2020). However, data show a deficit in neurosurgeons compared to neurosurgical burden, especially in Sub-Saharan Africa (Dewan et al., 2018). The research material on PBT in Sub-Saharan Africa is scarce (Uche et al., 2021), but does provide important information on the current situation. One systematic review has been performed on abandonment of treatment for children with PBT in LMIC (Seah et al., 2019), and several attempts at mapping PBT in LMIC have been done (Allemani et al., 2018;Bhakta et al., 2019). Yet, a systematic review on the existing studies on PBT surgical outcomes in the Sub-Saharan African region is lacking.
The overall aim of this systematic review was to investigate the outcomes following surgery for PBT in Sub-Saharan Africa. More specifically, the following questions were posed: a. What are the surgical outcomes for PBT in the Sub-Saharan African region, such as survival rates, postoperative recovery, and postoperative complications? b. What types of PBT are included in the studies on surgical outcomes for PBT in the Sub-Saharan African region?

Search strategy and data selection
A systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Page et al., 2021). The search was performed on PubMed for all available literature on the surgical outcomes of PBT in Sub-Saharan Africa published before May 3, 2022 (Table 1). Two authors, VH and PL, performed abstract and full text screening independently, resolving any conflicts by consensus.

Eligibility criteria and study selection
Studies were included if they report outcomes after surgical intervention on patients aged 0-18 years with a histologically specified primary neoplastic intracranial brain tumor according to the 2021 5th edition of the World Health Organization (WHO) Classification of Tumors of the Central Nervous System (Central Nervous System Tumours, 2021). There was no restriction on the study design or language of publication. Studies with unavailable full text were excluded and listed in Supplementary File 1.

Data extraction and risk of bias assessment
Study and patient characteristics, including name of first author, year and country of publication, study type, number and age of patients, and tumor types were collected. Surgical outcomes, including surgical extent, survival/mortality data, length of hospital stay, quality-of-life, postoperative improvement and postoperative complications were collected. All data acquisition was performed independently by both screeners and cleared of conflicts. Data was collected on sample size for each variable of interest. Inconsistencies in the patient number between variables was used as a proxy for the relevant criteria in the Joanna Briggs Institute (JBI) Critical Appraisal Checklists for Case Reports and Case Series (Aromataris and Munn, 2020). Several inconsistences indicated more risk of bias. Furthermore, studies with large differences between number of patients of entire studies and relevant patients in this review also indicate a risk of bias, as well as the relevant sample size per se.

Data analysis
All data were first presented descriptively. Whenever possible, metaanalysis was performed in R (version 4.1.0) (Core Team, 2020) using the meta package (Balduzzi et al., 2019). Data were pooled with a random-effects model using the DerSimonian-Laird estimator for between-study variance (DerSimonian and Laird, 1986). Proportions were visualized using a forest plot. The I 2 was used to quantify the between-study heterogeneity. Publication bias was assessed using Egger's (Egger et al., 1997) and Begg's (Begg and Mazumdar, 1994) tests, and funnel plots were drawn. No meta-regression was performed.

Abbreviations:
HIC high-income country LMIC low-and middle-income country PBT pediatric brain tumor studies were published in Nigeria, amounting in 185 patients. Three studies came from Kenya (32 patients), two from South Africa (12 patients) and Ghana (3 patients), and one from each of Uganda, Tanzania, Sudan, and Rhodesia ( Fig. 1). Importantly, 12 studies contained only a subgroup of patients that met the inclusion criteria, effectively resulting in significantly fewer relevant patients than in the original studies (243 vs. 701 patients). Furthermore, there were 6 studies that had inconsistent number of patients for the relevant variables. The fraction of relevant patients, inconsistent number of patients and total patient number are indicators for the risk of bias.

Patient characteristics
Among the 22 studies, the average reported ages ranged from 1 month to 16 years (Table 3), and the median thereof was 9 years. Among case reports, boys were more common (6 boys vs 3 girls), while girls were slightly more common in studies with more than three relevant patients (unweighted average 53%, weighted average 51%). The five most common tumor types were medulloblastoma (11 studies, 61 patients), craniopharyngioma (5 studies, 57 patients), astrocytoma (7 studies, 46 patients), meningioma (9 studies, 24 patients), and ependymoma (5 studies, 14 patients). The two most common reported extents of resection were [gross] total (8 studies, 93 patients), and subtotal (7 studies, 90 patients).

Patient outcomes
Survival and mortality differed widely between studies, and the data were reported using variable formats (Table 3). Even within studies, there was large heterogeneity. The four largest studies (n ¼ 37-76 patients) included several tumor types each, preventing the reader from drawing tumor-specific conclusions related to surgical resection. Adeloye et al. investigated only craniopharyngioma in nineteen patients, for which they found that almost one-third of patients died within the first six months; however, the extent of resection was not specified, preventing comparison to other studies. Only four studies reported length of stay, but three of them were case studies. There were no trends identified in the reporting of the post-operative course and complications (Table 4). The only outcome that could be analyzed with a meta-analysis was 1-year survival, for the studies of Ndubuisi and both studies by Uche. However, since these three studies all have several different histological tumor types and different extents of resection, we refrain from presenting the results here since the clinical meaning is limited. Instead, the results are instead presented in Supplementary File 2 for the interested reader. Overall, considering the large amount of case reports and sub-setting of larger studies, the certainty of evidence from this compilation of data is low.

Discussion
This is the first systematic review aimed at summarizing the outcomes after pediatric brain tumor surgery in Sub-Saharan Africa. Nigerian institutions have published most data on the topic. Among the included studies, commonly only subgroups of patients met the inclusion criteria, and there were a handful of studies that had inconsistent sample sizes among variables of interest. Furthermore, there were 9 (41%) case reports among the included studies. Among the remaining studies, only six (46%) included more than 10 patients. All these flaws considerably biased our potential findings due to the subjective patient recruitment. Additionally, surgical extent differed for certain PBT types, such as craniopharyngioma (Table 3), which may be resected totally, or sub-totally, the latter often followed by radiotherapy (Grewal et al., 2020). This may have affected heterogeneity in the reporting of outcomes. Overall, it was difficult to identify a general tendency among the surgical outcome of pediatric brain tumors in Sub-Saharan Africa (Tables 3-4).
Among the 46 countries that constitute the Sub-Saharan African region, only 8 countries (Nigeria, Ghana, Sudan, Uganda, Tanzania, Kenya, Rhodesia and South Africa) were represented among study locations (Fig. 2). This further highlights the scarcity of data on PBT and associated surgical outcomes in this region, which is consistent with previous studies that suggest a low access to high quality pediatric cancer registries (Bhakta et al., 2019) and research data (Stefan, 2015) in Sub-Saharan Africa. When reported, post-operative mortality in general was higher compared to those reported in HIC (Neervoort et al., 2010;Foster et al., 2021). In chart reviews, with more than 10 patients, 1-year post-operative mortality of PBT ranged from 33 to 43% (Table 3), while overall data on surgical mortality of PBT in HIC ranged from 0 to 20% (Neervoort et al., 2010). Previous studies have suggested that the higher overall PBT mortality rates in Sub-Saharan African countries and other LMIC, are the result of underdiagnosis (Stagno et al., 2014), and lack of access to neurosurgical care (Dewan et al., 2018;Park et al., 2021).
The three most common PBT types were medulloblastoma, craniopharyngioma and astrocytoma (Table 2). Although the scarcity of data prevents conclusions to be drawn regarding distribution of specific pediatric tumor types in this region, a cautious comparison to HIC (Pollack, 1994) indicate some similarities in distribution of medulloblastoma and astrocytoma. However, craniopharyngioma, a slow-growing, benign tumor, does not appear as common in HIC (Pollack, 1994), which may indicate some geographical differences or may be due to delayed presentation. Moreover, results are consistent with previous studies that have shown a large variation in PBT presentation between settings in LMIC (Asirvatham et al., 2011;El-Gaidi, 2011;Jibrin et al., 2018).

Conclusion
Data on outcomes after pediatric brain tumor surgery in Sub-Saharan Africa is insufficient and inconsistent, preventing any statistical conclusions to be drawn. There is an overrepresentation of Nigerian studies, while several countries are not represented. This systematic review highlights the need for more studies in the field.