The necessity to include the
Sub-Saharan African population in
neuroimaging genetics research of
schizophrenia
Research master Brain and Cognitive sciences: Literature thesis
Institute for Interdisciplinary Studies,
University of Amsterdam
Author: Pien Jellema (10709703) Assessor (UvA): Esther van Duin Co-assessor: Marieke van der Pluijm
Contents
Abstract ... 3
Introduction ... 4
Strategies for unravelling underlying mechanisms ... 5
Relevance ... 5
Design and operationalization ... 5
1. Brain-behaviour associations in schizophrenia ... 7
2. Schizophrenic genetic variations and neuroimaging abnormalities ... 9
2a. Neuroimaging genetics research ... 9
2b. Methodological remark on included study populations of consortia ... 11
3. The SSA population in schizophrenia genetics research ... 12
3a. Population stratification ... 12
3b. Genetic differences of the SSA population ... 13
3c. Benefits and issues of difference in genetic architecture ... 13
4. Epidemiology and prognosis of schizophrenia in SSA ... 16
4a. Environmental and social factors ... 16
4b. Schizophrenia prognosis in SSA ... 17
Conclusion ... 17
Discussion ... 18
Acknowledgements ... 20
References ... 21
Box 1 Explanation of concepts
• Positive symptoms: Psychotic behaviour such as hallucinations and delusions (1,3)
• Negative symptoms: Social withdrawal, depressed mood, flat affect, speech difficulties (1,3,4) • Cognitive symptoms: Trouble with working memory, organizing and executive functioning (1,3) • Biomarker: A biological marker used as a measurable indicator in clinical research (13)
• Interdisciplinary research: Research which integrates knowledge of multiple study domains (23) • Neuroimaging genetics research: Research which associates neuroimaging and genetics findings (3) • Consortium: Collaborative research project (12)
• Sub-Saharan African (SSA) population: Native population to the Sub-Saharan area of the African continent (83) • Population stratification: Population admixture (61)
• Epidemiology: Overarching term for factors involved in incidence, prevention and distribution of a disease (47) • Prognosis: The expectation of symptom development of a disease (18)
• Polygenic disorder: A disorder caused by lots of common genetic variants (61) • Genetic risk loci: Location of a particular risk-gene on a chromosome (37)
• Human reference genome: Representative reference of the entire human genome (60)
• Genome-Wide Association Studies (GWAS): Genetic research tool to find associations between genome and traits (21)
• Heritability: The relative effect of genetics compared to environmental causal factors (15)
• Single Nucleotide Polymorphism (SNP): A single genetic base position which can be associated with a trait, and shed light on underlying cellular pathways specifically related to complex diseases, such as schizophrenia (21) • Linkage Disequilibrium (LD): A correlation between SNPs with less chance on recombination. SNPs within LD are
thus closer located on a genetic location and have a lower chance of being split up by means of cross-over. In this way, combinations of SNPs can survive generations (24)
• Polygenic Risk Score (PRS): PRS is used to predict certain phenotypic traits based on variation within enormous amounts of genetic variants (16)
• Epigenetics: Functional genome changes, not caused by genetic alterations but often by environmental factors (18)
• Copy number variation: Variation in copy numbers of genome sections which may alter across individuals or populations (21)
• Haplotype: the combined set of genes located on a single chromosome (62) • Principle component analysis: Method to calculate ancestry in study sample (3,37) • TagSNP: SNP used to tag a group of alleles associated with a certain trait (24) • Genome: Entire genetic material of an individual (22)
• Allele: Genetic variant (38)
• Genetic homogeneity: Low genetic admixture (14)
• Structural Magnetic Resonance Imaging (sMRI): Structural neuroimaging technique (12) • Functional MRI (fMRI): Technique to measure functional brain activity (28)
• Resting-state fMRI: Measurement of functional brain activity at rest (27) • Structural connectivity: Infrastructure of the brain (31)
• Diffusion Tensor Imaging (DTI): MRI technique to measure structural connectivity (31) • Functional connectivity: Functionality of brain infrastructure (27)
• Intracranial brain volume: Total brain volume inside the cranium (43) • Ventricular volume: Volume of ventricles (22)
Abstract
Schizophrenia is an extensively studied complex psychotic illness, due to the lack of efficient treatment and global prevalence. By means of interdisciplinary research, representative and large study samples, neuroimaging genetics consortia aim to unravel the underlying mechanism of schizophrenia. Despite the wide geographical spread of the members of these consortia, Sub-Saharan African (SSA) populations are largely underrepresented. This literature review aims to pool useful literature and expert opinions to better understand whether this is problematic and whether the incorporation of the SSA population could be used to neuroimaging genetics research of schizophrenia. First, brain-behaviour associations seemed to be predominantly found in frontal and temporal regions. Second, robust associations between genetic risk for schizophrenia and differences in structural and functional brain measures were reported. Third, extrapolation of results derived from European ancestry- to SSA populations may be problematic, due to differences in genetic architecture. The high genetic diversity across SSA populations and the homogeneous genetic nature within SSA subpopulations are suggested to enable the development of genetic prediction models and reliable gene-trait associations. The issues faced by the SSA population in genetic schizophrenia research could, however, complicate the use of these benefits. Finally, some differences seem to exist in schizophrenia epidemiology and prognosis in SSA. In conclusion, extrapolation of schizophrenia research outcomes derived from European ancestry study samples to SSA populations should be carefully considered due to the cross-population differences. However, future collaboration with the SSA population in schizophrenia research could provide deeper insight into the relation between neuroimaging and genetics in schizophrenia due to the unique use of the distinct genetic architecture of this population. Provided that the limitations of this implementation, practical issues, and aspects of transcultural psychiatric collaboration are taken into account.
Introduction
Schizophrenia is a psychotic illness, also known as the psychosis spectrum syndrome (1,2). Its symptomatology could be generally characterized by difficulty in distinguishing reality from fiction (2) but further includes a range of positive, negative and cognitive symptoms (box 1) (1,3,4). Worldwide, mental health and neurological disorders, including schizophrenia, are the leading cause of disability (5). The global prevalence of schizophrenia is 1% (1) and is expected to be consistent all over the world (6). However, both prevalence and incidence numbers are lacking in lower-income countries, such as the African region (6). Despite the relatively low prevalence number, extensive research has been conducted, mainly in Westernized countries (3,7). This trend of exclusion the Sub-Saharan African (SSA) region in clinical research (6) could be explained by its huge health and economic burden all across the globe (box 1) (8,9). Generally, the schizophrenic symptoms are treated with antipsychotic medication (10). However, there is a high variation in medication response (10). Consequently, the health
and economic burden for individuals with schizophrenia and the society in general could be mainly caused by the lack of a cure, persistence of symptoms, common onset in adolescence, and high indirect costs related to these aspects (8). The poor prognosis of schizophrenia can also be illustrated by high mortality rates (6,11,12). Concretely, 10% of the global schizophrenia population commits suicide (6). Therefore, considerable funds have been invested to find biomarkers of schizophrenia to create a more efficient and less expensive treatment (10). It is generally believed that the behavioural phenotypes of schizophrenia can be understood once the genetic and psychobiological bases are identified (13).
To date, the extensive research surrounding the underlying mechanism of schizophrenia has led to several insights. The majority of the studies agree that schizophrenia is highly heritable (79%) (14,15) and a polygenic disorder (box 1) (13), which means it is caused by the combination of a multitude of genetic variants (16). Genome-wide association studies (GWAS, box 1) are commonly used to reveal schizophrenia genetic risk loci and their relation to neurobiological expression (box 1) (17). Research has frequently shown that psychiatric disorders, could not only be polygenic but could also interact with environmental factors in a complex system (18). Additionally, schizophrenia symptomatology seems heterogeneous across its patient population (13). Altogether, these insights
Box 2: Abbreviations
SSA: Sub-Saharan Africa
GWAS: Genome-wide association studies
SNP: Single Nucleotide Polymorphism
LD: Linkage Disequilibrium PRS: Polygenic Risk Score
MRI: Magnetic Resonance Imaging sMRI: structural MRI
fMRI: functional MRI
DTI: Diffusion Tensor Imaging SES: Socio-economic status WEIRD: Western, Educated,
Industrialized, Rich, and Democratic IQ: Intelligence Quotient
Strategies for unravelling underlying mechanisms
Despite the insights into the general understanding of schizophrenia, a lot is still unknown regarding the underlying mechanism of this complex disorder. One strategy to unravel this mechanism is the finding of biomarkers (box 1) (10). However, the highly variable nature of schizophrenia makes it practically impossible to find a single responsible biomarker (10). Therefore, many studies have explored structural brain abnormalities (12,19), functional brain differences (20), cognitive differences (19) and genetic risk factors (21). Despite the high number of these studies, replicable results are often lacking, which could indicate a necessity to rethink the research methodologies currently employed (12,14,22).
An example of a strategy which could contribute to the understanding of underlying mechanisms and the pursuit of replicable results could be the integration of current knowledge from different study domains (7). This could be done with an interdisciplinary approach (box 1) (23). For example, integrated neuroimaging genetics research could find relationships between the diverse symptomatology and the polygenic nature of schizophrenia (box 1) (3,12,13). This method would evaluate genetic variation through phenotypes established with neuroimaging. Hence, neuroimaging research itself could bridge the gap between genes, the brain and behavioural phenotypes (3,12,13).
Another strategy which could help the pursuit of replicable results is the inclusion of a representative sample of the global clinical population (3,12). The prevalence of schizophrenia seems to be fairly similar across the globe (16). The inclusion of an ethnically diverse study population could thus contribute to finding more generalizable results to the global clinical population (3), and balancing out of population-specific environmental and genetic variation (24).
Lastly, more replicable results may be found by substantially increasing the size of the study samples (7,13). This could be accomplished by collaborating within a consortium (box 1) which includes large study populations and uses standardized protocols to find robust results (3,12). In the field of neuroimaging genetics, multiple large consortia have been initiated. The targets of such consortia on schizophrenia differ from finally understanding cortical anatomical variations (ENIGMA), to understanding the contribution of genetic markers to brain traits (GENUS), and to identifying the neuroimaging genetic substrates of behavioural alterations (IMAGEN) (3,12,13).
Combining interdisciplinary research, representative study samples, and large study populations could further be used to generate more insights into the underlying mechanism of schizophrenia and yield more replicable results (3,12,13). This is what the majority of large neuroimaging genetics consortia on schizophrenia aim to achieve.
Relevance
Despite the large number of included individuals within neuroimaging genetics consortia on schizophrenia, populations with non-European origins such as the SSA population (box 1) are largely underrepresented (3,6). Considering the possible variation in the genetic architecture and environmental factors between the SSA- and most-studied populations, it would be interesting to study whether this imbalance might be problematic for the extrapolation of results to the SSA population (14). Moreover, it would be valuable to find out whether incorporating the SSA population could be used to generate more insights into the underlying mechanism of schizophrenia, and yield more replicable results (3,12,13).
This literature review aims to pool useful sources to elucidate whether neuroimaging genetics research of schizophrenia can benefit from including the SSA population.It further clarifies the relation between schizophrenic clinical features, neuroimaging, genetics, and the SSA population itself. I hypothesize that including the SSA population in neuroimaging genetics research of schizophrenia could be used to generate more insights into the underlying mechanism of schizophrenia and yield more replicable results.
Design and operationalization
In order to find an answer, the main research question will be subdivided into four sub-questions in which information will be organized based on the perspective of authors and themes. Thereby, both literature and expert opinions will be combined. In order to understand how the incorporation of the SSA population could be useful to schizophrenia research, the current findings in this field of research must be considered first. The first sub-question will assess the relation between schizophrenia brain and behavioural abnormalities. Second, the relation between schizophrenic genetic variations and neuroimaging abnormalities will be evaluated. The position of the SSA population in schizophrenia research shall subsequently be analysed. The third sub-question examines the current position of the SSA population in schizophrenia genetics research. Finally, the epidemiology and prognosis of schizophrenia in SSA will be reviewed. Each sub-question will be discussed in a chapter separately. An
interdisciplinary discussion will then incorporate research findings from population science, genetics, neuroimaging, and psychiatry. The main findings of each chapter will be incorporated into a conceptual model (figure 9).
To study how neuroimaging genetics research might benefit from including the SSA population, genetics outcomes are operationalized as heritability, GWAS, single nucleotide polymorphism (SNP), linkage disequilibrium (LD), and polygenic risk score (PRS) outcomes. Epigenetics, haplotypes, and copy number variations (CNVs) are outside the scope of this literature review. All of these concepts are explained in box 1. Neuroimaging research methodologies included in this literature review are structural magnetic resonance imaging (sMRI), functional MRI (fMRI), functional connectivity analysis based on resting-state fMRI and structural connectivity analysis based on diffusion tensor imaging (DTI) (concepts explained in box 1.). Both global cortical measures and regional cortical and subcortical measures are included. However, brain measures related to ventricular and intracranial brain volumes are outside the scope (box 1).
For clarification purposes, this literature review will refer to the term ‘schizophrenia’ as the psychiatric diagnosis only, thereby respecting the discussion surrounding the term and its stigma (2,25). The schizophrenic population discussed in this literature review will imply the diagnosed population in the first place unless otherwise specified. Large consortium studies will be referred to as multicentre studies, specifically aiming to find more replicable results by substantially increasing the size of the study samples, consequently, the statistical power. An overview of the literature review design and operationalization is illustrated in figure 1.
Figure 1. Schematic overview of literature review design and operationalization. In order to find out whether
neuroimaging genetics research of schizophrenia can benefit from including the Sub-Saharan African (SSA) population, first the current literature will be reviewed on the relation between schizophrenia symptoms and genetics (sub-question 1) and the relation between schizophrenia-related genetics and neuroimaging (sub-question 2). The findings of this literature will be based on European study samples. Next, literature based on the SSA population will be reviewed, and its relation to genetics (sub-question 3) and schizophrenia prognosis and epidemiology (sub-question 4). The numbers in the figure correspond to the sub-questions. Tools used in reviewed neuroimaging and genetics literature are listed. Explanation of concepts and abbreviations can be found in boxes 1 and 2.
1. Brain-behaviour associations in schizophrenia
Historically, the relation between brain and behavioural abnormalities in schizophrenia has been most frequently studied by means of behavioural diagnostic markers (21). However, in order to understand what is currently known about brain-behaviour associations in schizophrenia, neuroimaging tools can be useful (3,12,13). To illustrate, neuroimaging studies could aim to find associations between neuropsychological measures or clinical features and MRI outcomes (box 1). This chapter will assess the recent literature on brain-behaviour associations in schizophrenia.
Based on brain differences between schizophrenic cases and healthy controls, a great amount can be learnt about clinical features, symptomatology and neuropsychological processes (26–28). The findings of recent studies on this topic are summarized in table 1; Brain differences are categorized in structural, functional and network differences. Overall, especially frontal and temporal cortical regions seemed to be deteriorated, however all over the brain reduced volumes and cortical thinning appeared to be found (12,26,29,30). This evidence supports the hypothesis that frontal and temporal regions could be target regions of underlying schizophrenia neurobiological mechanisms (12). Common schizophrenia symptoms, both positive and negative seemed to be associated with these structural differences (12,26,30). Auditory verbal hallucinations, in particular, could be localised in the temporal gyrus (figure 2) (26). This type of hallucinations could be considered as one of the most severe schizophrenia symptoms, as they appeared to be very common in the schizophrenic population (26). The thinning seen in the frontal regions of individuals with schizophrenia could be explained by problems with executive and cognitive functioning (28). These functions, moreover, were often associated with prefrontal cortex processes (28). Both on a functional and structural brain network level, abnormalities have been reported when comparing schizophrenic cases to controls (27,31).
These abnormalities have been related to the disconnection hypothesis, which describes distortions and disconnections within brain networks giving rise to common schizophrenic symptoms (27).
Within the schizophrenia population itself, clinical subgroups could be the result of differences in brain processes on a cellular level (10). These subgroups could differ in treatment response (10). Dopamine receptor blocking medication helped to inhibit positive symptoms, such as hallucinations, in a large proportion of the schizophrenic population (10). But another subgroup of the schizophrenic population was resistant to this medication. This finding suggests that there are subgroups within the schizophrenic population relying on different biological network disruptions (32). This might also explain the high variation in symptomatology (32).
In summary, the brain-behaviour associations in individuals with schizophrenia may be mainly found in frontal and temporal regions (12,26,29,30). These regions seemed to be primarily targeted by underlying neurobiological mechanisms of schizophrenia and associated with clinical features and common symptoms (12,26,29,30). Furthermore, schizophrenia symptomatology might be explained by the disconnection hypothesis, which was based on distortions and disconnections within brain networks of schizophrenic cases (27,31). Within the schizophrenia population itself, variance in clinical features could be established by distinct cellular brain processes (10,32), emphasizing the need for clinical subgroup identification based on biomarkers (10).
Figure 2. Thinner cortical thickness in the temporal gyrus in auditory verbal hallucination schizophrenia individuals compared to individuals without these hallucinations. P
values are indicated on the colour bar. Adapted from Cui et al. (2018) (26).
Table 1. Summary of recent studies on the topic of associations between behavioural- and brain abnormalities in individuals with schizophrenia. Brain differences are categorized in structural, functional and network differences.
Abbreviations and concepts can be found in boxes 1 and 2.
Authors Study sample Clinical features, symptom categories and
neuropsychological measures
Techniques Main findings brain abnormalities in schizophrenic cases
Structural differences van Erp et al.
2018, ENIGMA consortium Schizophrenic cases compared to healthy controls • Positive symptoms • Negative symptoms • Antipsychotic treatment • Age of schizophrenia onset
sMRI • Thinner cortex bilaterally
• Smaller surface area of the entire brain • Effects especially seen in frontal and
temporal regions
• Antipsychotic treatment, and more severe positive and negative symptoms associated with even thinner cortex • Earlier age of onset associated with
thinner insula cortex Koelkebeck et al. 2019 Schizophrenic cases compared to healthy controls • Comparison between schizophrenic cases from German and Japanese cultures
sMRI • Smaller grey matter volume in frontal and temporal regions
• Small regional grey matter differences cross-culturally
Cui et al.
2018 Schizophrenic cases with auditory verbal hallucinations compared to those without • Auditory verbal hallucinations sMRI
• Thinner temporal gyrus associated with auditory verbal hallucinations
Ducharme et al. 2014 Healthy children and adolescents • Negative symptoms
sMRI • Thinner ventromedial prefrontal cortex associated with negative symptoms Walton et al. 2018, ENIGMA consortium Schizophrenic cases • Negative symptoms sMRI
• Negative symptom severity associated with thinning of prefrontal regions, including the left medial orbitofrontal cortex
Functional differences Geisler et al.
2015 Schizophrenic cases compared to healthy controls • Executive and cognitive functioning • Working memory
sMRI, fMRI • Reduced regional cortical thickness associated with diminished executive- and cognitive functioning
• Differences in working memory associated with fronto-parietal regions Network differences Cheng et al. 2015 Schizophrenic cases compared to healthy controls • Symptom severity • Illness duration Resting-state fMRI
• Global alterations within the thalamo-cortical functional connectivity network • Network alterations associated with
symptom severity and illness duration Kelly et al. 2018, ENIGMA consortium Schizophrenic cases compared to healthy controls • Age at onset • Medication dosage
DTI • Poorer structural cortical network throughout the entire brain • No association between structural
network alterations and age at onset or medication dosage
2. Schizophrenic genetic variations and neuroimaging abnormalities
In this chapter, schizophrenic genetic variation and its relation with neuroimaging outcomes will be assessed. It is hypothesized that the brain abnormalities discussed previously are also associated with certain genetic variants, which will be evaluated in the current chapter.
First of all, the current research findings of neuroimaging genetics research of schizophrenia will be discussed. Psychobiological markers, such as genetic or neuroimaging markers, could enable a better understanding of the underlying issues involved in schizophrenia (3,12,13). For this, high statistical power is required to find reliable results explaining the heterogeneous nature of schizophrenia (Correspondence with Dr. G. Blokland, (33)). In order to significantly increase statistical power, a movement of consortia has started that includes large study samples (Correspondence with Dr. G. Blokland, (33–35)). These consortia combine data of multiple research groups that is acquired according to the same protocol (Correspondence with Dr. G. Blokland, (33–35)). By combining data from a variety of research groups, these consortia aim to include a more diverse and representative schizophrenia population (3,12,13).
Secondly, a methodological remark will be made on the included study population of neuroimaging genetics consortia on schizophrenia. Whether the current included study population of these consortia is representative of a worldwide schizophrenia population is drawn into question.
2a. Neuroimaging genetics research
The relation between schizophrenic clinical features and its polygenic nature can be found within neuroimaging genetics research (3,12,13). This field of research tries to find neuroimaging phenotypes that could be associated with schizophrenia-specific genetic variants (Correspondence with Dr. G. Blokland, (3,12,13). Based on the common disease, common variant hypothesis, it would be expected that such a globally common prevalent psychiatric disease as schizophrenia would be mediated by common genetic variants across populations (36). According to that hypothesis, several SNPs of interest have been found which could relate to disrupted cellular pathways in schizophrenia (Correspondence with Dr. G. Blokland, (21)). Together with PRS, this genetic data can be used to run associative analyses with neuroimaging data (Correspondence with Dr. G. Blokland, (3,13)). A visual overview of the currently identified schizophrenia genetic risk loci is given in figure 3.
Most frequently, the methodology of neuroimaging genetics research is used to match at-risk brain regions, such as the hippocampus (37) or subcortical volumes (33), to an entire GWAS data set. The inverted methodology is applied to match single risk genes or
high PRS for schizophrenia to structural voxel-wise or functional data (box 1) (34,35,38). A priori chosen at-risk brain regions have often been associated with common schizophrenia symptoms (37) or are often found to differ in size in schizophrenic cases compared to healthy controls (33). Moreover, cortical thickness could be a good measure to capture the cumulative effects of multiple environmental factors on cortical psychobiological characteristics (39,40).
The findings of recent studies on this topic are summarized in table 2, in which the results are categorized based on the associated consortium. Taken together, genetic risk for schizophrenia seemed to be associated with differences in structural brain measures such as decreased volume in fronto-temporal regions (41), hippocampal volume (37), enlarged volume of putamen (34), thinner global cortical thickness (42,43), and smaller total brain and thalamic
Figure 3. Locations of schizophrenia candidate genes. The chromosome
ideograms represent currently identified risk genes for schizophrenia. Numbers in superscript depict methodologies outside the scope of this literature review. Adapted from Liu et al. (2019) (94).
volumes (43), yet with few other subcortical volumes (33). Furthermore, genetic risk for schizophrenia also seemed to be associated with differences in functional brain measures, including higher functional activity in the orbito-frontal regions during an impulsivity test (38), and reduced functional brain activity to emotional stimuli and social impairment (35). Taken together, associative neuroimaging genetics studies seemed to find significant associations between genetic risk for schizophrenia and differences in global cortical measures or (sub)cortical regions, on both structural and functional levels (34,35,37,38,41–44). However, some non-significant associations were also reported (33).
Table 2. Summary of recent studies on the topic of associations between neuroimaging and genetics in schizophrenia.
Results are categorized based on the associated consortium. Abbreviations and concepts can be found in boxes 1 and 2.
Consortium Authors Study Sample Genetic concepts
Brain measures
Main findings brain abnormalities in schizophrenic cases GENUS Petryshen, Blokland, and The GENUS Consortium 2019 Schizophrenic cases, healthy controls and familial high-risk participants PRS, SNP,
GWAS sMRI, neuro-psychological assessment
• Schizophrenia PRS associated with decreased grey matter volume of fronto-temporal regions
• Schizophrenia PRS associated with impaired cognitive functions IMAGEN & ENIGMA Hass et al. 2013 Schizophrenic cases compared to healthy controls SNP, GWAS sMRI, neuro-psychological assessment • SNPson chromosomes 1, 2, 10 and 19p13.11 were associated with hippocampal volume in both
schizophrenic individuals and healthy controls
IMAGEN Luo et al. 2019 Healthy adolescents, schizophrenic adults and healthy siblings
GWAS sMRI • Predetermined schizophrenia risk gene SLC39A8 associated with enlarged grey matter volume in the putamen in healthy adolescents • This gene-neuroimaging association
is weakened in the schizophrenic case group
IMAGEN Heinrich et
al. 2013 Healthy adolescents GWAS fMRI, neuro-psychological assessment
• Healthy adolescent risk allele carriers showed higher functional activity in the orbito-frontal regions during an impulsivity test
• Risk allele carriers show more impulsivity in neuropsychological assessment
IMAGEN Velthorst et
al. 2018 Healthy adolescents PRS fMRI
• High PRS for schizophrenia
associated with severity of psychotic episodes
• High PRS for schizophrenia negatively associated with functional brain activity to emotional stimuli and social impairment
IMAGEN French et al. 2015
Healthy adolescents
PRS sMRI • Cannabis use in early adolescence associated with thinner cortical thickness in males with high PRS ENIGMA Franke et al.
2016 Schizophrenic cases compared to healthy controls GWAS, SNP, LD sMRI
• Common alleles associated with schizophrenia not significantly related to subcortical volumes
ENIGMA de Zwarte et
al. 2019 Schizophrenic cases, first degree relatives,
Heritability sMRI • First degree relatives showed smaller total brain and thalamic volumes, and reduced cortical thickness than healthy controls
2b. Methodological remark on included study populations of consortia
By combining data from a variety of research groups, neuroimaging genetics consortia aim to include a more diverse and representative schizophrenia population (3,12,13). However, one could question whether the included study population is representative of a worldwide schizophrenia population. As previously stated, the SSA population is often underrepresented in schizophrenia studies (6). This imbalance in study populations is commonly encountered in neuroscience research in general, as the majority of neuroscience research focuses on convenient small samples (23), or is based on Western, Educated, Industrialized, Rich, and Democratic (WEIRD) populations (45). However, to understand individual differences in traits such as schizophrenia symptoms, it is important to include multiple ethnic populations into the study sample. Otherwise, study outcomes of brain-behaviour mechanisms can be undermined by an underrepresented sample of the human population (23). Figure 4 shows the diversity of the study population of the ENIGMA consortium (46). Despite the wide geographical spread of consortium members, ENIGMA seems to focus on WEIRD populations. The SSA population, for example, is largely underrepresented in this figure.
The lack of data on schizophrenia in the SSA region is not uncommon (47) but should be carefully considered since psychiatric disorders could show variation in prevalence and response to drug treatment across ethnic populations (48–50). To illustrate, schizophrenia diagnosis seemed to be more than two times as frequent in an African-American population than in a European-American population (49,51). An inclusive sampling method in neuroimaging genetics research could therefore contribute to understanding these cross-population differences (23). Internal validity and efficiency of research, moreover, may improve by studying a usefully targeted population (23,45). Inclusion of population-specific environmental factors may further prevent extrapolation of unrepresentative data (23). This
argument could clearly be explained by the example of the heritability of intelligence quotient (IQ) which seemed to be confounded by socio-economic status (SES) (box 3) (45). Thus attempts of neuroimaging genetics consortia on schizophrenia which aimed to include a representative study population may be less successful than proposed (3,12,45).
This chapter aimed to assess schizophrenic genetic variation and its relation to neuroimaging outcomes. Based on the neuroimaging genetics consortia outcomes, robust results seemed to be found that explain the association between genetic risk for schizophrenia and differences in structural and functional brain measures (34,35,37,38,41–44). The high statistical power of these consortia, moreover, may contribute to a better understanding of the manifestation of schizophrenia in a more global population (3,12,13). Despite the wide geographical spread of consortium members, these consortia seem to focus on WEIRD populations (45,46). Therefore, extrapolation of neuroimaging genetics results on schizophrenia to other populations, such as the SSA population, should be carefully considered (23,45,48–51).
Figure 4. Diversity of collaborating research groups of the Schizophrenia consortium. Red dots indicate locations of
ENIGMA-Schizophrenia members. The majority of the ENIGMA schizophrenia members are located in Westernized countries and are absent in SSA regions. Adapted from Turner and van Erp (2019) (46).
Box 3: Example of confounding of environmental factor in heritability research
Often research claims that IQ is highly heritable (93). However, rarely subjects with low SES are included in these studies (45). A twin study which did compare subjects with low and high SES revealed that for twins with high SES, IQ was 70% heritable (45). However, in twins with low SES, IQ was merely 10% heritable. The difference in heritability found
between high and low SES was thought to be confounded by
environmental variability. For high SES twins, environmental variability was practically non-existent. Whereas for low SES twins, environmental variability seemed very high.
3. The SSA population in schizophrenia genetics research
The current chapter will examine the position of the SSA population in schizophrenia genetics research. The previous chapter described, among others, the focus of schizophrenia neuroimaging genetics consortia on WEIRD study populations. Additionally, it questioned whether the lack of data on other populations, such as the SSA population, might cause trouble for the extrapolation of results. This will be evaluated in the current chapter. In order to do so, first, the population stratification in schizophrenia genetic research will be evaluated (box 1). Second, the differences in the genetic architecture of the SSA population will be reviewed. Third, some benefits and issues of this difference in genetic architecture will be assessed. Due to the general lack of schizophrenia genetic research on the SSA population itself (52), the results of other African ancestry populations will be included in this chapter, as these populations find their origins in SSA.
3a. Population stratification
Genetic ancestry can be used as a mediator in schizophrenia research as it is a valuable biomarker which describes ethnic diversity within a study sample (Correspondence with Dr. G. Blokland, (3). Moreover, ancestry is a more accurate measure of study population stratification than race due to the diverse migration routes in human history (14,53). The methodology used to account for ancestry in schizophrenia research is explained in box 4.
The ancestral population stratification of large neuroimaging genetics consortia on schizophrenia (table 3), and GWAS studies (figure 5) shows an overall focus on European participants, more so than on any other ancestry population. This is a rather interesting pattern, as not all genetic variants of diseases within a European ancestry population reflect the situation of non-European ancestry populations (22). Moreover, genotype frequencies could differ amongst ancestral populations (54), and genetic origins of a certain trait could be located on different alleles across different populations (box 1) (17,55). Thus, European ancestry-specific genetic variation may not always directly translate to the other ancestry populations. The PRS of schizophrenia, which is based on European genetic data, could also be confounded by ancestry (16). Furthermore, research into antipsychotic medicine has rarely been performed in an African ancestry population (54). However, across ancestry populations differences seemed to exist in the functioning of the innate immune system which could mediate the reaction to antipsychotic medication (56). Overall, it seems clear that the majority of clinical research findings have been based on a European ancestry population, yet these findings may not directly translate to the African ancestry population.
uthors Consortium Total N (cases and controls) African ancestry (%) European ancestry (%) Other ancestries (%) Blokland et al. 2019 GENUS 10.801 7 70 23
Hass et al. 2013 IMAGEN & Mind Clinical Imaging Consortium
328 0 100 0
Heinrich et al. 2013
IMAGEN 1360 0 100 0
Franke et al. 2016 ENIGMA & Psychiatric
Genomics Consortium 88.026 0 100 0
Table 3.) Division of ancestry within total study populations of large neuroimaging genetics consortia on schizophrenia. Ancestry of included study populations is categorized in African-, European- and other ancestries. The
amount of each ancestry category is indicated in percentages (%). Overall, a lack of African ancestry in these studies is seen, and a focus on participants of European ancestry. The examples of ENIGMA studies in this table do not report ancestry at all.
Box 4: Methodology used to account for ancestry in schizophrenia research
Ancestry can be calculated by means of principal component analysis (box 1) (Correspondence with Dr. G. Blokland, (3,37)). When implemented in statistical analysis, population stratification of a genetic heterogeneous study population is often seen as problematic (37). In order to validate the study outcome based on an “ethnically homogeneous sample”, the majority of schizophrenia research thus focuses on the largest subsample of similar descent (22,37). This often comes down to only including data from European ancestry participants. Subsequently, some studies check whether the results would replicate when data of other ancestries are added (Correspondence with Dr. G. Blokland, (3)).
3b. Genetic differences of the SSA population
The SSA population is suggested to differ in genetic architecture from the European ancestry population (16,22,50,57), which explains why schizophrenia genetic results based on a European ancestry population might not directly extrapolate to the SSA population. The difference in genetic architecture could be explained by the ancestral origins of the oldest human population from the African continent, which seemed to contain the
highest genetic diversity (figure 6) (53,58). Ancestors of non-African populations dispersed through different migration routes over the world and appeared to have lost a great deal of their genetic diversity in bottlenecks (52). To illustrate, genomes of SSA populations might contain nearly one million more genetic variants per person compared to non-SSA populations (box 1) (57). Moreover, the SSA population has maintained larger and better structured subpopulations due to the isolated migration pattern of tribes (24). This could explain the more homogeneous genetic nature of these subpopulations (box 1) (24). Taken together, the differences in the genetic architecture of the SSA population might be best described by its high genetic diversity and the homogeneous genetic nature of SSA subpopulations (24,53,58).
3c. Benefits and issues of difference in genetic architecture
The difference in the genetic architecture of the SSA population may be beneficial to genetic schizophrenia research but might also be challenged by some issues. First, some benefits will be discussed. The high genetic diversity of the SSA population implies a more diverse LD pattern across subpopulations (box 1) (24). Moreover, the genetic homogeneity of SSA subpopulations suggests more genes in LD, including genetic risk factors (22). These two genetic characteristics of the SSA population can be useful to discover new genes, develop genetic prediction models, and find more reliable and detailed associations between genes and traits (14,57,59). To illustrate, it was shown that the inclusion of an African ancestry population had aided to identify an association amongst genetic variants of a non-communicable disease, which could not be found in a European study sample (22).
Figure 5. Distribution of ancestry of GWAS studies in the January 2019 GWAS catalog. Distributions of
included ancestry categories are indicated based on studies (left) and on individuals (right). More than half of all the GWAS studies and more than three-quarters of all the individuals included in GWAS studies are of European ancestry. Only 9.56% of GWAS studies- and 2.03% of individuals in all GWAS studies are of African descent. Adapted from Sirugo, Williams, and Tishkoff (2019) (24).
Secondly, some issues faced by the SSA population due to genetic schizophrenia research will be discussed. The most important issue is the lack of genetic schizophrenia research on the African ancestry population (52). In some fields of clinical genetic research, no studies have been performed at all that include the SSA population (52). This also seems to be the case for research on the heritability of schizophrenia (14,15), which merely included twin pairs of European ancestry, despite its large sample size (n = 31,524 twin pairs included in the study of Hilker et al. (2018)). This observation may raise some questions regarding the heritability of schizophrenia in the SSA population. The reference genome used in schizophrenia GWAS studies may also face some issues for the SSA population. In fact, it was recently found that the commonly used human reference genome sequence (GRCh38) might lack 10% of African ancestry genes (box 1, figure 7) (60). This result was based on the genomes of 910 pan-African individuals and suggests a need for higher genomic coverage for GWAS studies in the pan-African ancestry population (52). Increasing the coverage of the human reference genome sequence is not a new issue. The “1000
Figure 6. Human migration routes out of Africa and its effect on genetic diversity. (A) The oldest human
populations seem to originate from the African continent. Thus, non-African populations could have gone through a bottleneck when migrating out of Africa. This map only illustrates human migration routes towards the Asian continent. Dates are indicated in thousands of years (ka). Adapted from Bae, Douka, and Petraglia (2017) (58). (B) A schematic model presenting the loss of genetic diversity along with human migration. Each coloured dot indicates a distinct allele. Out of Africa migration tends to continue with a subset of the original genetic diversity. During these migration events, some alleles could be lost. This illustration indicates that the highest genetic diversity could reside in African populations. Adapted from Rosenberg and Kang (2015) (92).
A
human reference genome with data of more ancestries (Correspondence with Dr. G. Blokland, (62)) to create a new reference genome called “Eagle2” (62). The addition of these new genomes could cause a gain in knowledge regarding rare genetic variants (62). Ancestries that were added to the Eagle2 reference genome were African (n = 3.817), East Asian (n = 5.164), Latino (n = 7.144) and European subpopulations (n = 61.684). This means that merely 4,9% within the new set of added genomes, was of African ancestry. Thus, despite its efforts, the Eagle2 reference genome still contains a small percentage of African ancestry genomes.
Another issue might be the potential inadequacy of schizophrenia genetic risk predictors derived from a European study population to replicate to the SSA population (16,24). Differences in the predictive power of PRS (16) and tagSNPs (box 1) (24) could be due to the diversity in LD patterns across populations (24). This diversity in LD across populations likely plays a role in the transferability of causative variants (figure 8) (24). Besides, differences in population-specific allele frequency and variation could lead to more uncertainty in the predictive power of PRS across populations (16,24). To illustrate, the PRS of schizophrenia derived from a European study population seemed to be much less informative in the African ancestry population (16). In order to understand variance in genetic and neurobiological expression, GWAS research in the SSA population is needed (17). In sum, the benefits of the distinct genetic architecture of the SSA population are best described by its unique use to discover new genes, develop future genetic prediction models, and find more reliable and detailed associations between genes and traits (14,57,59). Whereas issues faced by the SSA populations in genetic schizophrenia research are the lack of research on this population (52), the misalignment of African ancestry genomes to reference genomes (60,62) and the potential inadequacy of schizophrenia genetic risk predictors derived from a European study population to replicate to this population (16,24).
This chapter has identified the position of the SSA population in schizophrenia genetics research. In summary, it seems that the majority of research findings have been based on a European ancestry population, thus these findings cannot not directly translate to the African ancestry population (16,22,54). This difficulty to extrapolate results across populations is caused by the differences in the genetic architecture of the SSA population (16,22,50,57), which includes high genetic diversity across SSA populations, and the homogeneous genetic nature within SSA subpopulations (24,53,58). It is also these factors that are suggested to be very useful and beneficial to genetic schizophrenia research (14,57,59). However, the issues faced by the SSA population in genetic schizophrenia research could complicate the use of these benefits (16,24,52,60,62).
Figure 7. Example of misalignment of the African ancestry genome (CAAPA) to the often used reference genome GRChr38, and close alignment to the Chinese genome. Only the blue, red
and yellow segments of the African ancestry genome (CAAPA) are well covered by the GRCh38 reference genome. The rest of these genomes are misaligned. On the contrary, the CAAPA and Chinese genomes are well aligned. Adapted from Sherman et al. (2019) (60).
Figure 8. Two alternative scenarios of the transferability of
causative variants depending on the role of linkage disequilibrium (LD) amongst populations. The upper scenario is based on the
hypothesis that there is a common causative variant, which can be tagged by different SNPs in different populations. The lower scenario illustrates the situation when there are multiple causative variants amongst populations. These can be tagged by different SNPs (solid lines), or by the same SNPs with weaker LD (dotted line). Adapted from Sirugo, Williams, and Tishkoff (2019) (24).
4. Epidemiology and prognosis of schizophrenia in SSA
When studying differences in schizophrenia between the SSA- and most-studied populations, other features related to epidemiology and prognosis are also of interest. Both epidemiology and prognosis of psychotic disorders have been found to vary across societies and geographical locations (47,48). Therefore, this chapter will first assess differences in environmental and social factors. Secondly, population-specific variation in schizophrenia prognosis will be reviewed, including current treatment practices in SSA.
4a. Environmental and social factors
There is growing evidence that environmental and social factors, play an important role in the onset of schizophrenia and are associated with brain and behaviour connections (23). These environmental factors might interplay with genetics and can be used as psychobiological predictors of schizophrenia (13,21,23). Additionally, individuals of multiple cultures can be exposed to different environmental factors (23). For example, childhood trauma, such as witnessing armed conflicts, sexual abuse and domestic violence, could cause toxic stress and seemed to be considerably more common in SSA regions (48,63). Childhood trauma is also suggested to be a high-risk factor for the onset of schizophrenia and multiple other psychiatric disorders (48,63). The impact of childhood trauma on the onset of schizophrenia may be due to its potential impact on gene expression (48) or its suggested effect on prefrontal cortex functioning (4,23).
Another environmental factor of interest could be substance abuse, as it may trigger the onset of schizophrenia (64). Especially abuse of cannabis and alcohol seemed to be associated with the onset of schizophrenia. In the SSA region, substance abuse is a major public health concern (65). For instance, younger populations in Kenyan secondary schools have been suggested to struggle with substance abuse (66). The substance abuse could have evolved from the urban transition caused by uncontrolled growth, rapid cultural changes, and high poverty rates in SSA (67). Due to this urban transition, people in SSA seem to face more exposures to stressful events that may drive them towards substance abuse (67).
Stress could also be involved in the pathogenesis of schizophrenia (54) and can present itself in several forms across populations (68,69). In Western countries, work-related stress is expected to be more common (68), whereas economic stress, due to poverty, seems to be more prevalent in SSA countries (70). With regard to African ancestry migrant populations, stress related to social disadvantages of being an ethnic minority seemed to be more common (54). Furthermore, this type of stress might be related to an increased risk of developing schizophrenia (54).
Social factors, including SES, could also affect the epidemiology of schizophrenia (6,23,71). For example, childhood SES could be used as a predictor of structural and functional brain measures related to schizophrenia (23). The size of one’s social network, moreover, seemed to be related to structural brain measures (23). Interestingly, the cultural diversity in social and community support systems are assumed to explain schizophrenia prognosis differences amongst populations (6,72). One suggestion following from this is that individuals with schizophrenia of collectivist cultures have a milder prognosis and fewer relapses than individuals of individualistic cultures (71,72). With regard to the SSA population, their culture could be broadly categorized as a collectivist culture due to the importance of social and community support (6,72). The protective aspect of the SSA culture, however, might only be protective of schizophrenia prognosis when the schizophrenic symptoms are not suspected of causing harm to the collective (Correspondence with Dr. R. Schwartz, (72)).
Lastly, the stigma surrounding mental health help-seeking behaviour in SSA populations could impact schizophrenia prognoses (73). Despite the expectation of equal prevalence of schizophrenia across the globe, it is often thought that psychiatric disorders are far less common in SSA regions due to its invisibility in these regions (72,74). However, it seems that the invisibility may be caused by the strong stigma surrounding mental health help seeking behaviour in SSA cultures rather than lower prevalence (73). The stigma could be associated with the fear of social exclusion (73). To illustrate, rarely any individual in a Somali immigrant population wanted to admit mental health issues, despite its severity in this population (75).
Taken together, childhood trauma (48,63), substance abuse (64), stress (54), childhood SES (23), social and community support systems (6,72), and stigma surrounding mental health help-seeking behaviour (73) could impact differences in schizophrenia epidemiology and prognosis in SSA populations compared to most-studied populations. These differences in environmental and social factors could emphasize the need to study schizophrenia in population-specific contexts (65). Moreover, it could reveal interesting opportunities for population-specific prevention strategies (63).
4b. Schizophrenia prognosis in SSA
Schizophrenia prognosis could show some population-specific variation (29,76) and could be affected by its treatment practices (11). In SSA, these treatment practices range from psychiatry to more traditional options (11,77). However, there is a lack of psychiatrists available in SSA regions. For example, in Zimbabwe, only 10 psychiatrists were available to help its population of 13 million people (11). This lack of psychiatric help is mainly due to overall economic shortages in SSA regions (77). Therefore, alternatives to psychiatry such as faith and traditional healers are common first mental help resources (Correspondence Dr. R. Schwarz, (77)). Traditional healers mainly base their help on spiritual beliefs and often ascribe schizophrenia symptomatology as caused by supernatural powers, demons, evil spirits or witchcraft (77). Moreover, SSA individuals with high levels of self-stigma were more likely to ascribe their severe mental health condition to similar ideas (73,78). Misdiagnosis is suggested to occur frequently in traditional healing practices and is often associated with a lack of education (47). To illustrate, the schizophrenia-like diagnosis of “Losing touch with reality”, also named “Nhlanyi” in xi-Tsonga, was thought to be successfully treated in 47% of all cases by traditional healers (5). In another example, merely 53.8% of individuals characterized as being “disturbed” by traditional healers were indeed diagnosed with a psychotic disorder (47). However, others suggest that schizophrenia-like concepts such as “madness” are similarly recognized and understood by traditional healers (79). Furthermore, it is not uncommon for traditional healers to refer individuals with signs of psychosis to Western psychiatric help, though be it often in a late stadium (Correspondence Dr. R. Schwarz, (79)). Regarding the expression of common schizophrenia symptoms, differences between SSA- and Western populations seemed unlikely (Correspondence Dr. R. Schwarz, (79)). Instead, these differences between both populations are suggested to generally lie in help-seeking behaviour and available anti-psychotics (Correspondence Dr. R. Schwarz, (77)). Small regional brain structures, moreover, might alter in volume across schizophrenia populations from different cultural backgrounds (29). In addition, African ancestry individuals with schizophrenia might show different structural and functional brain connectivity patterns (76). Taken together, the variation of schizophrenia prognosis in the SSA population includes differences in treatment practices, available anti-psychotics, and some potential differences in brain structures and functionality, but there are virtually no differences in expression of common schizophrenia symptoms (Correspondence Dr. R. Schwarz, (11,29,76,77)).
Thus, variation in schizophrenia epidemiology and prognosis between the SSA and most-studied populations seems to exist and could consist of differences in environmental and social factors that may be harmful or protective of schizophrenia pathogenesis (6,23,48,54,63,64,72,73).
Conclusion
This literature review has pooled useful sources to better understand whether neuroimaging genetics research of schizophrenia can benefit from including the SSA population, thereby it clarified the relation between schizophrenic clinical features, neuroimaging, genetics and the population of SSA itself (figure 9). The reviewed literature revealed that relations between brain and behavioural abnormalities seemed to be predominantly found in frontal and temporal regions (12,26,29,30) and might be explained by the disconnection hypothesis (27,31). Large neuroimaging genetics consortia report robust associations between genetic risk for schizophrenia and differences in structural and functional brain measures (34,35,37,38,41–44). Despite the wide geographical spread of consortium members, these consortia seem to focus on European ancestry populations (45,46). Extrapolation of these results to the SSA population may be problematic due to differences in the genetic architecture of the SSA population (16,22,50,57). Moreover, the high genetic diversity across SSA populations and the homogeneous genetic nature within SSA subpopulations (24,53,58) are suggested to be highly useful and beneficial to genetic schizophrenia research (14,57,59). The issues faced by the SSA population in genetic schizophrenia research, however, could complicate the use of these benefits (16,24,52,60,62). Differences in epidemiology and prognosis of schizophrenia in SSA seem to be due to several environmental and social factors, and alternative treatment practices (6,11,76,77,23,29,48,54,63,64,72,73), yet manifestation of common schizophrenia symptoms seems to be similar across SSA and Western populations (Correspondence Dr. R. Schwarz, (79)).
In conclusion, extrapolation of schizophrenia research outcomes derived from European ancestry study samples to SSA populations should be carefully considered due to the multiple differences amongst these populations (22–24). However, the inclusion of the SSA population in neuroimaging genetics research could be beneficial due to the unique use of its distinct genetic architecture (14,57,59).
Discussion
There are some limitations to the arguments of the reviewed literature. For instance, the enormous sample sizes included in consortium studies seem to be key to identifying true schizophrenia genetic risk variants (24). In order to do so, variants with smaller effect sizes are cancelled out to capture a global perspective (33). On the contrary, these smaller effects could obtain population-specific information (24). Thus, consensus should be achieved with regards to extrapolation of study results across diverse populations. Another issue related to enormous sample sizes (3,33), is the overfitting of data (18). Overfitting is an especially common problem in big data sets and could happen when a model returns a false positive predicting the outcome of new cases outside of the data set (80).
Based on the population differences in genetic variation, some genetic concepts may need some reconsideration (36). For instance, as a result of the common prevalence of schizophrenia across the world, its highly heritable character is also suggested to be common across populations (14,15). Research on the heritability of schizophrenia is mainly based on European twin registers and seems to lack replication in SSA populations (14). Heritability percentages, moreover, could be confounded by environmental factors (45). Furthermore, the African genome appears to be insufficiently covered by reference genomes (60). For all these reasons, the heritability of schizophrenia may need some rethinking in the context of the SSA population.
Another issue involved in cross-population comparisons is a difference in the impact of environmental factors
Figure 9. Conceptual model of research on Schizophrenia and its relation to the Sub-Saharan African (SSA) population.
Arrows reflect a relationship between presented concepts. More in-depth information is provided in the designated chapters of this paper. Harmful effects of environmental and social factors are indicated by (-) and protective effects by (+). Important concepts are underlined.
cross-population research, it is unknown whether these differences in environmental factors alter the expression of schizophrenia risk genes (18). Therefore, the inclusion of the SSA population in schizophrenia neuroimaging research could potentially shed light on this case (48), as this type of research can clarify the gene-environment interaction (18). Moreover, the relatively high amount of harmful environmental factors might also be implicated in the prevalence of schizophrenia in SSA (54). In contrast to current beliefs of equal schizophrenia prevalence across the globe (1), some African ancestry migrant populations seemed to report an increased risk of schizophrenia development (54). Differences in the severity of environmental factors (54) or inadequate genetic predictive power (16), could explain the increased risk for schizophrenia development in this specific population.
A limitation of the inclusion of literature on African ancestry populations due to the general lack of research on the SSA population itself (52) should be acknowledged. Namely, African ancestry populations do not only include SSA populations but also African ancestry migrant populations (14,54). For instance, variation in environmental factors and genetic architecture may explain
the differences between SSA populations and African ancestry migrant populations (50,53,54). The different genetic architecture of African ancestry migrant populations includes a loss of genetic diversity mainly caused by the transatlantic slave trade (Correspondence with Dr. R. Schwartz, (81)). Therefore, results derived by African ancestry migrant populations should be carefully extrapolated to SSA populations.
Some practical issues exist in the implementation of schizophrenia research in SSA, such as inadequate infrastructure, scarce availability of technology, limited properly trained staff, an economic burden, and lack of acknowledgement of mental health disorders (box 5) (52,73,77,82,83). These practical issues could explain the shortage of qualitative neuroimaging- or genetics results in SSA (52,82). Another issue stemming from the incorporation of the SSA population in schizophrenia research practices is the difference in cultural standards and values (11). One may wonder whether it will be truly beneficial for the SSA population itself to transfer towards a Westernized viewpoint of schizophrenia (11), a viewpoint where the ultimate goal is to find biomarkers that could increase schizophrenia treatment efficiency (10) rather than getting treated for being “disturbed” by a traditional healer (73,77). Therefore, some lessons can be learned from successful local approaches addressing mental health in SSA, which take cultural standards and values into consideration (box 6) (11,84). These approaches would actively aim to prevent a colonialist approach.
The economic burden is an issue that also returns in interdisciplinary research. Grant givers frequently rate this type of research as too risky, yet it is exactly these studies that can gain the biggest scientific
Box 5: Practical issues involved in schizophrenia research in SSA
Practically, there are some issues involved in including the SSA population in schizophrenia neuroimaging genetics research. First of all, there are practical issues regarding infrastructure and availability of technology (82). Especially regarding MRI accessibility, as not many MRI scanners are available across the African continent (82). To illustrate, only 84 MRI scanners are available for a population of over 350 million people in West-Africa (82). Besides, the majority of these MRI scanners have a much weaker magnetic field than scanners used for research in Western countries (3,29,82). This difference in the magnetic field complicates data comparison, as there are differences in image resolutions (3,82). GWAS research also requires some
specialised equipment (33). Blood samples regularly capture the highest concentration of DNA with the least amount of contamination (57). This method, however, is rather intrusive and logistically quite complicated to manage in SSA regions (57). Therefore, using another DNA extraction method based on saliva samples, for example, could be a practical adjustment (57). Furthermore, a constant and reliable power supply is needed for MRI research (90). With few exceptions, this may be problematic in the majority of SSA regions due to the frequent power cuts (82). Regarding the liquid helium itself, provision of supply is also not always available in many SSA countries (82).
Secondly, limited staff is trained with knowledge concerning MRI physics, MRI data analysis, its clinical applications, and quality checks, due to its lack of availability and familiarity in SSA (82). However, all of these aspects are crucial in generating reliable and useful data, which may be shared when research institutes collaborate (3). One practical suggestion to enforce staff training could be the adoption of MRI labs in SSA by established neuroimaging labs in the Western countries. This way, new neuroimaging labs in SSA would not have to reinvent the wheel themselves.
Thirdly, the lack of acknowledgement of mental health disorders across SSA could cause some practical problems (73,77). Though common schizophrenia symptoms seem to be often recognized by traditional healers, there is some variation in diagnosis (73,77). Moreover, very little psychiatric help is available (11). The lack of acknowledgement may explain the scarcity of schizophrenia research in SSA.
The last practical issue of schizophrenia research in SSA could be the economic burden, as there is very little budget available for clinical research on psychiatric disorders (77,82). However, many countries in SSA seem to experience sustained economic transformations (91), including higher living standards (82). A side effect of this economic transformation may be the amplification of exposure to risks of mental health disorders (67), thus causing a shift from an economic burden to a mental health burden (83). Therefore, the promotion of schizophrenia research in SSA may also be supported by an ethical argument, namely the reduction of patient suffering (11).
