Visual functions in children with craniopharyngioma at diagnosis: A systematic review

Childhood craniopharyngioma is a rare and slow growing brain tumour, often located in the sellar and suprasellar region. It commonly manifests with visual impairment, increased intracranial pressure and hypothalamic and/or pituitary deficiencies. Visual impairment in childhood adversely affects a child’s daily functioning and quality of life. We systematically reviewed the literature to provide an extensive overview of the visual function in children with craniopharyngioma at diagnosis in order to estimate the diversity, magnitude and relevance of the problem of visual impairment. Of the 543 potentially relevant articles, 84 studies met our inclusion criteria. Visual impairment at diagnosis was reported in 1041 of 2071 children (50.3%), decreased visual acuity was reported in 546 of 1321 children (41.3%) and visual field defects were reported in 426 of 1111 children (38.3%). Other ophthalmological findings described were fundoscopic (32.5%) and orthoptic abnormalities (12.5%). Variations in ophthalmological testing methods and ophthalmological definitions precluded a meta-analysis. The results of this review confirm the importance of ophthalmological examination in children with craniopharyngioma at diagnosis in order to detect visual impairment and provide adequate support. Future studies should focus on long-term visual follow-up of childhood craniopharyngioma in response to different treatment strategies to provide insight in risks and ways to prevent further loss of vision.


Introduction
Childhood craniopharyngioma (CP) is a rare and slow growing epithelial brain tumour (World Health Organization grade I) [1]. It is thought to arise from embryonic remnants of Rathke's pouch, located along the craniopharyngeal duct. CP is commonly located in the sellar and/or suprasellar region of the brain [2,3]. The incidence of CP is 0. 5 million persons per year, with a bimodal distribution in children (5-14 years) and adults (50-74 years) [2][3][4][5].
Despite the benign histological grade I classification, CP often recurs and may cause severe morbidity due to its close anatomic relation with important visual and endocrinological structures. Affected children commonly present with visual impairment, increased intracranial pressure (ICP) and hypothalamic and/or pituitary deficiencies [2,3,6]. Impaired visual function is a primary manifestation in 62-84% of all children diagnosed with a CP [3]. Nevertheless, it often takes years after the onset of symptoms before children actually get diagnosed [4,6].
Craniopharyngioma mainly causes visual impairment by direct infiltration and/or compression of the visual pathway. Damage of the visual pathway commonly manifests as decreased visual acuity (VA), visual field (VF) defects, typically bitemporal hemianopia, and/ or abnormal pupillary reponses [5][6][7][8][9][10][11]. Increased intracranial pressure as a result of obstructive mass effect of the tumour can lead to papilledema with subsequent optic atrophy and permanent vision loss as potential complications [8,12,13]. In addition, therapeutic interventions for CP such as tumour resection or post-surgical radiation therapy can lead to further visual loss. In particular, gross total tumour resection has a high risk of visual loss as a result of direct damage to visual structures or disruption of its vascularisation [2-4, 9, 14-16].
As described above, CP and its therapy commonly causes severe and permanent visual impairment, as well as hypothalamic-pituitary dysfunction. These tumour sequelae have a major impact on a child's health and quality of life [12,13,[17][18][19]. Therefore, early detection of visual impairment together with adequate treatment and support is of major relevance as it may reduce irreversible visual sequelae and improve long-term visual outcome and quality of life [12,13,20].
Early detection of visual abnormalities requires timely referral to an ophthalmologist for ophthalmological examination. Previous studies have already demonstrated the importance of ophthalmological examination in children with a brain tumour at time of diagnosis and during follow-up [12,13,17,18,20].
Several nonsystematic reviews have summarized ophthalmological findings in children and adults with CP. However, an extensive overview in subtopics like VA, VF, fundoscopy and orthoptic examination has never been published. With this systematic review we aim to provide a broad overview of the visual function in children with CP at diagnosis in order to estimate the diversity, magnitude and relevance of the problem of visual impairment in children with CP.

Protocol and registration
A review protocol was developed based on the Preferred Reporting Items of Systematic Reviews and Meta-Analyses (PRISMA) statement [21]. The systematic review was prospectively registered in the international prospective register of systematic reviews (PROSPERO) on April 23, 2020 (ID: 150419). In accordance with Dutch guidelines, no institutional ethical review board approval was required.

Information sources and search strategy
We conducted a systematic search in the Cochrane Library, Embase and PubMed in order to identify all eligible studies. The electronic databases were last searched on October 2, 2019 for a combination of the following key search terms and/or their synonyms: 'craniopharyngioma', 'vision', 'visual acuity', 'visual fields', 'optic chiasm', 'optical coherence tomography', fundoscopic abnormalities (e.g. 'papilledema') and orthoptic abnormalities (e.g. 'diplopia').
Finally, 84 studies met the eligibility criteria and were included in our review. A detailed overview of the selection process for included studies and reasons for exclusion after full-text screening is shown in Fig 1.

Study characteristics
Characteristics of the 84 included studies are shown in detail in Tables 2 and 3. Altogether, the studies included a total of 3531 children with CP with sample sizes ranging from 2 to 411 Table 1. Detailed description of risk of bias assessment using Newcastle-Ottawa Scale (NOS).

Domains Predefined criteria Maximum number of stars per domain
Selection Representativeness of exposed cohort (children with CP and visual impairment) � Cohort truly representative of the average child with a primary CP aged 0-18 or 0-21 years in the community together with a description of key characteristics (age, gender, tumour type etc.) • Selected group of children with CP (e.g. only giant CP) • No description of key characteristics ���� Selection of non-exposed cohort (children with CP without visual impairment) � Cohort drawn from the same community as the exposed cohort • Cohort drawn from a different source • No description of the derivation of the non-exposed cohort
There was a large variety between research aims of the included studies. In the majority of studies, visual function in children with CP at diagnosis was not a primary outcome measure, but details about visual function at diagnosis could be obtained from tables containing baseline characteristics. Precisely 10 studies reported primarily visual function and/or long term visual ouctomes in children with CP [6,7,10,39,52,53,72,83,102,107].  Table 4 shows the results of the risk of bias assessment by the Newcastle-Ottawa Scale (NOS) for the included studies. Overall, total scores ranged between 1 to 6 stars of a possible 7 stars. Eleven studies were awarded 1-3 stars, 72 studies 4-5 stars and one study 6 stars. These scores indicate that most of the included studies were of moderate quality. Quality was predominately limited by missing or incomplete information about VA testing and/or VF testing [14, 19, [14] and Tan (2017) [42] were the only two studies we could rate for comparability, since they included two and three cohorts, respectively.
Craniopharyngioma was located third ventricular or extended to the third ventricle in 110 children (5.8%). Retrochiasmatic CP was reported in 94 children (5.0%), with a sellar and suprasellar component in 2 patients in a study by Leng (2012) [58]. Caldarelli (2005) reported 14 retrochiasmatic or third ventricular CP [90]. Twenty-four children (1.5%) had a CP located in or with extension to the posterior cranial fossa, namely 2 CP were located at the temporal and posterior cranial fossa [101], 1 suprasellar CP with extension to the posterior cranial fossa and third ventricle [79], 15 sellar CP with extension to the posterior cranial fossa [46] and 5 sellar CP with extension to the cerebellopontine angle and the posterior cranial fossa, as well as infraclivus extension in 2 of 5 patients [75].

Visual impairment
Of the 84 studies eligible for data extraction, in 56 studies authors provided the total number of patients in whom visual function was impaired (Table 3). For these studies, visual impairment was described in 1041 of 2071 children (50.3%) with CP at diagnosis. Authors used different terms to describe visual impairment, for instance 'visual impairment', 'visual defects', 'vision loss' and 'visual complaints'. If a definition for impaired visual function was provided by the authors, this is shown in Table 3. Twenty-eight studies did not mention the total number of children with visual impairment in general, nevertheless data about one or more subdomains of visual function (visual acuity, visual field, fundoscopy or orthoptic examination) was available for these studies. Sixty-two studies reported about visual disturbance as an anamnestic symptom at diagnosis in 1135 of 2267 (50.0%) children with a CP.

Visual acuity
We identified 53 studies describing VA in children with CP at diagnosis (Table 3). Authors used different definitions and grading systems to describe VA. Four authors described the applied VA testing method, namely by Snellen test [9,52], LogMAR charts [72] or preferential looking charts [10,72]. Seven studies explicitly reported about best corrected VA (BCVA) instead of VA [6,7,60,72,83,84,95]. Authors of the other 46 studies did not describe whether they used BCVA or uncorrected VA (UCVA). The VA testing methods and definitions are shown in Table 3.
Multiple studies described VA by using VA scales. In a study by Chen (2003) 6 patients had a VA < 6/12 (35.3%) and 10 patients had a VA � 6/12 (58.8%) [6]. In two studies, 10 of 55 patients had a VA of � 20/200 in one or both eyes (18.2%) [60,84]. Kiran (2008) [79] reported VA of 6/9 in one of 2 patients and Kramer (1960) [38] VA of 6/60 in one of 6 patients. Tamasauskas (2014) described two children with a VA of 5/10 and 1/1000 [105]. Two children with CP in a study by Anderson (1989) had respectively a VA of 20/25 and counting fingers (CF) at 1 foot, and a VA of 20/300 and CF at 2 feet [53]. Finally, Synowitz (1977) presented VA data of 3 CP patients: one patient had no VA defects; one patient had only PL and the last patient could CF with his right eye at 1 m and could see hand movements with his left eye at 50 cm [69].
In summary, different grading systems and testing methods were used to report about decreased VA in 41.3% of children, with no specification of VA reduction in 43.9%. Blindness in one or both eyes was reported in 13.8% of children.
Despite the fact that 8 studies did not specify the VF defects in 30.3% of children with CP, uni-and/or bitemporal hemianopia is the most frequent VF defect in 132 of 542 children (24.4%).
The overall findings in children with orthoptic abnormalities (12.5%) showed diplopia in 14.5%, strabismus in 16.5% and nystagmus in 6.6% of the children.

Other vision related abnormalities
Apart from the abovementioned ophthalmological findings, some studies have described other vision related abnormalities as well (Table 3). Colour vision defects were reported by Mohd-Ilham (2019) in 4 of 11 patients (18.2%) [7]. Nielsen (2012) described ophthalmoplegia in 5 of 31 patients (16.1%) [103]. Right amaurosis was reported by d'Avella (2019) [91] in 1 of 8 patients (12.5%) and Chamlin (1955) [52] reported ptosis in 1 of 18 patients (0.05%). These vision related abnormalities were not the main focus of our study and were not analysed and/ or reported in any of the other studies included in our systematic review.

Discussion
Our review was designed to provide a detailed overview of the currently available evidence about visual function in children with CP at diagnosis. To the best of our knowledge, this is the first review that systematically describes the visual function in subtoptics like VA, VF, fundoscopy and orthoptic examination. We included 84 studies, with 56 studies explicitely providing data about visual impairment in general, and 55 studies providing specific data about VA and/ or VF. We found a high rate of visual impairment in children with CP at time of diagnosis (50.3%). Considerable rates were also reported for decreased VA (41.3%) and VF loss (38.3%). Papilledema (25.8%) and optic nerve atrophy (44.8%) were common fundoscopic findings in our review. The most common abnormalities in orthoptic examination (12.5%) were strabismus, diplopia and cranial nerve deficits. These findings are in agreement with several nonsystematic reviews of Bogusz (2018), who concluded that more than 50% of children with CP had visual impairment at diagnosis [110], and with Müller (2008) who described visual impairment, VF defects, papilledema and optic atrophy in respectively 62-84%, 36%, 20-35% and 35-45% of children with CP [4]. Drapeau (2019) described even higher rates for decreased VA and VF defects, namely in 70-80% of children with CP. In particular, Drapeau (2019) reported bitemporal hemianopia and papilledema in respectively 50% and 20% of children with CP [111].
The presented data in our review supports the importance of awareness in doctors for the fact that CP commonly induce visual impairment in children, as well as the importance of ophthalmological examination at diagnosis. Visual impairment due to damage of the optic nerves, optic chiasm and visual pathways often results in lifelong effects for children and their family, by affecting domains including childhood development, education, employment and self-perception [12,[112][113][114]. Visual problems may be reversible in early stages of visual impairment. Therefore, timely monitoring of visual function and early detection of visual impairment in children with CP is of major importance to preserve visual function and provide adequate treatment [12,13,17,18,20]. In children with irreversible visual impairment, timely referral for visual rehabilitation may reduce the adverse effects of visual impairment on health and/or vision related quality of life [115,116]. The impact of an impaired visual function on quality of life has also been reported in children with visual impairment with ophthalmological origin, for example, children who suffer from glaucoma and cataract [117,118].
Moreover, visual impairment has been reported as one of the factors that may lower the level of physical activity [119]. Especially in children with cranioharyngioma in whom hypothalamic damage can be severe, resulting in endocrine deficiencies and obesity, physical activity is crucial (40-50%) [119][120][121][122][123]. Both visual impairment and severe obesity negatively affect quality of life in childhood CP survivors [119,124,125].

Limitations of the included studies
Overall there was moderate-quality evidence for the presence of visual impairment in children with CP at diagnosis in our review. Although there were serious limitations to the data due to e.g. retrospective design of the studies, moderate risk of bias for some included studies and potential publication/reporting bias, the overall quality of evidence was raised by the number of included studies, study sizes, availability of confirmatory evidence and representativeness of study patients. Nevertheless, there are some issues that need to be discussed. First, different terminology was used to describe tumour locations and no concrete insights were given in the relationship between tumour location and visual loss. Therefore, we were not able to relate a more suprasellar tumour location involving the optic chiasm with the type and degree of visual loss. Second, no standardized ophthalmological examination was performed in a large proportion of included studies, and if performed, there was no uniformity in testing methods between the studies. Visual acuity and VF were described with different definitions and cut-off values per study, which makes grouping of results difficult. In the absence of standardized ophthalmological examination, it could be questioned whether we can presume that those children reported without visual impairment really have a normal visual function. In addition, performing reliable VA and VF testing in young and non-cooperative children is often very complicated [126,127]. Therefore, it is likely that these data were frequently missing in the included studies. Both of these issues, no standardized ophthalmological examination and difficulties with reliable VA and VF testing, could be reasons for underreporting of visual impairment in our review. Additionally, we were not able to compare the feasibility of different VA and VF testing methods for different age groups, because only a few studies provided information about the used tests and in these studies the authors often did not specify which testing methods were used for the different age groups. Furthermore, study authors reported the number of children per abnormality found by either fundoscopy or orthoptic testing. For this reason, the exact number of children with fundoscopic or orthoptic abnormalities is unclear, because one child might have more than one fundoscopic or orthoptic abnormality. Finally, authors used different cut-off values for the age limit of children. We initially planned to include studies only when patients were aged between 0 and 18 years. However, during study selection we encountered multiple studies still referring to patients as 'children' when aged < 24 years. We decided to include those studies as well, aiming to provide an extensive overview of visual function in children with CP at diagnosis. Nevertheless, heterogenity in age ranges for children across studies may lower their comparibility.

Strengths and limitations of this systematic review
The findings of the present systematic review should be interpreted by its strengths and limitations. We planned this review a priori and registered our review in PROSPERO with clearlydefined selection criteria. We conducted a comprehensive literature search and reviewed all reference lists of included studies. Two reviewers independently of each other performed the literature screening, data extraction and risk of bias assessment. In this way we retrieved and summarized the visual data of 3531 children with newly diagnosed CP. We also encountered some possible limitations for the methodology of our review. By screening references of included studies, we identified a relatively high number of additional studies eligible for inclusion (n = 15). Therefore, it might be possible that in our search we missed studies, though our cross reference search would in that case have identified those articles. Furthermore, full-text articles of 117 potential relevant abstracts found by our search in the electronic databases were not available despite searching the Utrecht University Library, Sci-Hub and contacting the corresponding author by mail and/or ResearchGate. Possible reasons for this could be that the abstracts are dated (we did not use a publication date filter) or that the full text did not exist. Lastly, we did not review visual function at follow-up as we initially planned for in our PROS-PERO registration. This was because many authors within our study selection, described visual follow-up data only in patients who received tumour treatment.

Recommendations
For future research, it is relevant to investigate the visual function at diagnosis and during long-term follow-up of childhood CP in response to surgery, radiotherapy and other treatment strategies, first by systematically reviewing the literature as well as in prospective collaborative studies. This will provide insight in risks and benefits of treatment regarding vision in children with CP, for professionals, patients and their caregivers. Furthermore, future studies should focus on reliable ophthalmological testing methods for young and non-cooperative children. As we have shown, the majority of studies did not report the methods they used for ophthalmological testing. Therefore, unfortunately, we were not able to compare feasibility of testing methods for different age groups due to lacking data. For future studies, it is important that all studies must use the correct testing methods for VA and VF and report these as such in the paper. Additionally, optical coherence tomography with analysis of the retinal layers might be applied as objective testing method in addition to VA and VF testing [84,[128][129][130].

Conclusion
Children diagnosed with CP have at least 50% risk of visual impairment at diagnosis regarding VA, VF, fundoscopy and/or orthoptic examination. Complete structured evaluation of visual function at diagnosis should be performed routinely in all children diagnosed with craniopharyngioma. However, large, well designed studies with standardized ophthalmological examination and uniform reporting with grading are needed to gain more insight in the visual function of these patients at diagnosis, after therapeutic interventions and during follow-up.