Cognitive Brain Signatures of Youth With Early Onset and Relatives With Schizophrenia: Evidence From fMRI Meta-analyses

Abstract

Deficits in cognitive function are a major characteristic of schizophrenia. Many functional magnetic resonance imaging (fMRI) studies examine brain correlates of cognitive function in adults with schizophrenia, showing altered implication of associative areas such as the prefrontal cortex and temporal cortex. fMRI studies also examine brain representation of cognitive function in adolescents with early onset schizophrenia and those at risk of the disorder, yet results are often inconsistent. We compile and analyze data from eligible fMRI studies using quantitative meta-analyses to reveal concordant brain activity associated with adolescent relatives of patients with schizophrenia and those with early onset schizophrenia. Results show similar functional hubs of brain activity (eg, precuneus) yet in opposite hemispheres and clusters in ventrolateral rather than dorsolateral prefrontal cortices. Other areas of altered implication include the middle temporal gyrus, insula, and cerebellum. We discuss the findings in reference to the protracted maturation of the prefrontal cortex and possible effects due to the medication status of the two groups.

Schizophrenia is a severe mental illness marked by cognitive and psychotic symptoms, as well as debilitating negative symptoms. The disorder leads to significant difficulties in daily functioning, employability, and social relations.¹ Cognitive symptoms in schizophrenia have been linked with altered anatomical and functional brain indices, mainly in frontal and temporal cortices. Specifically, functional neuroimaging research has examined brain areas of adults with schizophrenia showing increased and decreased implication in prefrontal brain regions when solving cognitive tasks.^2–6 The majority of individuals at risk of developing schizophrenia do so during early adulthood, at around 18 to 25 years of age, and there is evidence of prodromal signs that occur prior to illness onset in childhood and adolescence.⁷ Target groups for identifying prodromal signs are adolescent relatives of individuals with schizophrenia and youth with early onset schizophrenia. Results from adult functional magnetic resonance imaging (fMRI) meta-analyses of working memory and executive function tasks of unaffected relatives compared to controls also show both hypo- and hyperactivation in prefrontal cortices.^8,9 Critically, the prefrontal cortex has protracted developmental patterns spanning into young adulthood.¹⁰ Therefore, abnormalities in brain responses of youth at risk of schizophrenia may show altered implication in this region. To investigate this hypothesis, we use quantitative fMRI meta-analyses of brain responses to cognitive tasks of youth with relatives affected by schizophrenia and youth with early onset schizophrenia.

Although biological markers differentiating patients with schizophrenia from normal controls and other patients have been elusive,¹ schizophrenia is associated with specific brain characteristics. Patients with schizophrenia have been shown to exhibit enlarged ventricles, reduced brain volume in various regions, including the frontal, parietal, occipital, insula, and anterior cingulate cortices, as well as the hippocampus.^11–14 Cortical white matter abnormalities and connectivity problems are common in those at risk for schizophrenia,^15–19 yet few have extensively studied the course of the illness.²⁰ A meta-analysis has also shown that individuals at risk of developing psychosis showed gyral reductions in frontal and temporal regions, as well as reductions in anterior cingulate, hippocampal/parahippocampal regions, and the left precuneus.²¹ Moreover, those who transitioned to psychosis showed gray matter reductions in inferior frontal and superior temporal gyri.²² Similarly, those who transition compared with those who do not may show volumetric decreases in the cingulate, insular, prefrontal and temporal cortices, and the cerebellum.^12,23

Meta-analyses of functional correlates of cognitive abilities in adult patients with schizophrenia have examined convergence across studies. The first meta-analysis was published in 2005 by Glahn and colleagues, who evaluated 12 studies that recorded brain responses to working memory processes in individuals with and without schizophrenia. Results showed both increased and decreased implications of the prefrontal cortex in persons with schizophrenia compared to control participants.⁴ A subsequent meta-analysis with 41 studies on executive function replicated this finding.⁵ A more recent meta-analysis examined gray matter volume and resting state across studies and reported reductions in the left insula and right inferior temporal gyral gray matter volume in relatives with schizophrenia compared to controls,¹⁴ supporting an earlier structural and functional meta-analysis on patients with first-episode psychosis.²¹ Taken together, these meta-analyses implicate a set of areas associated with working memory processing and executive function, including the temporal, cingulate, and insular cortices.^4,5 These brain areas are associated with the executive (ie, working memory) networks, also found in large quantitative meta-analyses of healthy adult participants performing cognitive tasks.^24,25

Importantly, cognitive abilities such as working memory improve gradually over childhood and adolescence.^26,27 Although we could assume that younger age groups possess similar brain systems as adults, empirical data are in partial support. Specifically, younger, healthy, typically developing individuals show conjunction with adults in posterior parts of the cortex (eg, parietal areas), and disjunction in prefrontal parts of the cortex (eg, dorsolateral prefrontal areas²⁸). Lack of prefrontal concordance across studies using children and adolescents may be associated with increased variability in activity location; for instance, different prefrontal locations and hemispheric asymmetry are identified by original studies.²⁸ Thus, if the prefrontal cortex is a key brain marker for adult cognitive symptoms in schizophrenia, it remains to be investigated whether it is a brain marker related to prodromal signs of schizophrenia in younger age groups at risk of schizophrenia, such as relatives of patients with schizophrenia.

Prodromal signs of the disorder may also be observed in relatives of patients with schizophrenia. A meta-analysis comparing adults with relatives with schizophrenia and healthy controls showed hyperactivation within the right frontal, and left parietal regions, yet hypoactivation within the bilateral frontal, left temporal, and right parietal cortex was also reported.⁸ fMRI studies have examined the brain correlates on cognitive abilities in youth with early onset schizophrenia and youth with a relative, usually a parent, with schizophrenia.^29–33 For example, some studies identify working memory deficits in association with the dorsolateral cortex,³² specifically in adolescents with schizophrenic parents.²⁹ Other studies show no behavioral difference in the performance between adolescents with schizophrenic parents and controls yet schizophrenic relatives did show increased, rather than reduced, dorsolateral prefrontal as well as striatal activity during correct vs. incorrect memory performance.³⁰ Adolescents with early onset schizophrenia show reduced working memory performance and reduced ventrolateral prefrontal activity,³⁴ whereas other studies show reduced activity in posterior brain regions, but not in prefrontal areas.^33,35

Since no single study is definitive, quantitative fMRI meta-analyses can identify overarching concordance across studies. Notably, the majority of quantitative meta-analyses in the literature use the activation likelihood estimation (ALE) method in software provided by GingerALE.³⁶ However, recent recommendations of the software developers suggest that a minimum number of 17 to 20 experiments are needed for sufficient power.³⁷ Although past studies with schizophrenia did not adhere to this recommendation, we decided to analyze our data with an alternative method of quantitative meta-analyses that uses foci as well as effect-size to estimate concordance. Specifically, effect-size signed differential mapping (ES-SDM)³⁸ adopts features of ALE and incorporates statistical values (t-statistic) of peak coordinates to increase statistical power. Moreover, to evaluate the robustness and replicability of suprathreshold clusters, ES-SDM offers options for a jackknife sensitivity procedure. Therefore, we used ES-SDM to examine brain responses to cognitive tasks in youth with schizophrenic relatives and youth with early onset schizophrenia.

Based on past literature, we expect altered brain responses mainly in posterior brain regions in youth affected by schizophrenia and reduced implication of the prefrontal cortex in (a) early onset schizophrenia and (b) relatives of schizophrenia patients. The coordinate results that we provide may serve as a topographical atlas for target brain regions in future neuroimaging investigations and theoretically inform psychological models of schizophrenia.

Methods

Study Selection

The literature was searched to identify fMRI articles that focused on developing individuals with schizophrenia or relatives of those with schizophrenia. Our search focused on broad terms associated with fMRI, schizophrenia, and youth/adolescent, rather than terms such as “early onset schizophrenia” in an effort to capture as many published articles as possible. We used both terms “youth” and “adolescents” to improve the likelihood of assessing as many published articles as possible. Specifically, the first search was performed by entering keywords “schizophrenia” AND “fMRI” AND “youth” into the web of knowledge database (http://www.webofknowledge.com) and Pubmed (https://www.ncbi.nlm.nih.gov/pubmed/), while the second search was performed by entering the keywords “schizophrenia” AND “fMRI” AND “adolescents.” Searches were performed in October 2019. The search yielded a total of 193 articles, which were screened for eligibility. We only considered published studies and associated published supplementary material. To identify eligible articles, we relied on a standard protocol³⁹ and removed studies that: (1) did not report fMRI foci in standard stereotactic coordinate space (either Talairach or Montreal Neurological Institute, MNI); (2) did not use whole-brain analysis; (3) used contrasts examining adults or nonhumans; and (4) used contrasts that did not include individuals with early onset schizophrenia or relatives of individuals with schizophrenia. The total number of eligible studies was 16, which reported between-group whole-brain results in MNI or Talairach space. Activations (hyperactivity) of schizophrenia-related individuals were extracted from “schizophrenia” minus “control” contrasts, whereas deactivations (hypoactivity) of schizophrenia-related individuals were extracted from “control” minus “schizophrenia.” See figure 1 for the flowchart of article selection for the meta-analyses.

Fig. 1.

Prisma flowchart showing steps taken for identifying articles included in the meta-analyses (template by Moher et al., 2009).⁴⁰

Software and Analysis

ES-SDM meta-analysis software from the Seed-based d Mapping project (http://www.sdmproject.com) was used to perform the meta-analyses.³⁸ Based on ALE, this analysis combines statistical parametric t-maps and peak coordinates of clusters from multiple studies to increase statistical power.³⁸ Effect-size brain maps and variances are derived from reported t-statistics. The full width at half maximum (FWHM) in SDM was set at the default (20 mm) to control for false positives.³⁸ To optimally balance sensitivity and specificity, resulting statistical maps were thresholded at P = .005 to control for family-wise error rate.³⁸ Jackknife sensitivity analyses were performed to assess the replicability of suprathreshold clusters. ES-SDM jackknife analyses reiterate the meta-analyses as many times as the number of studies and remove one study at a time outputting a percentage that signifies the strength of replicability. For illustration, ES-SDM values were overlaid onto an anatomical template normalized to MNI space using Mango image viewer software (http://rii.uthscsa.edu/mango/mango.html).

Results

Descriptive Results

Sixteen articles were eligible for inclusion in the meta-analyses. Table 1 summarizes participant demographics, task type, group type, and contrast type. Data from a total of 269 participants were included in the meta-analyses; 8 articles reported contrasts from early onset schizophrenia (118 participants; 86 males; mean age = 15.85; SD = 2.01), and 8 articles reported contrasts from schizophrenia relatives (151 participants; 85 males; mean age = 15.20; SD = 1.98).

Table 1.

Information on Source Datasets Included in the Meta-analyses

Article	N	Male	Hand (R)	Mean (SD); Range	Task Type	Foci	Group Type	Contrast Typea
Bakshi et al., 201193	19	12	NA	14.3(3.1); 8–20	n-back	2	SCZ-Rel	+
						2	SCZ-Rel	+
Barbour et al., 201294	19	14	NA	14.71; 10–20	Rating pictures	1	SCZ-Rel	+
						2	SCZ-Rel	−
Bittner et al., 201534	17	11	13	17.9; 15.1–19.9	Delayed discrimination	7	EOS	+
						2	EOS	−
						11	EOS	−
						3	EOS	−
Diwadkar et al., 2011a29	19	12	NA	14.16(2.85); 8–19	n-back	6	SCZ-Rel	−
Diwadkar et al., 2011ba31	18	5	NA	14(3.1); 8–19	Visual memory	3	SCZ-Rel	−
Diwadkar et al., 201230	19	12	NA	14.3; 8–19	Visual memory	2	SCZ-Rel	+
Hart et al., 201395	21	11	19	14.4(2.56); 9–18	Oddball	4	SCZ-Rel	+
						1	SCZ-Rel	+
						14	SCZ-Rel	+
						10	SCZ-Rel	−
						2	SCZ-Rel	−
						1	SCZ-Rel	−
						7	SCZ-Rel	−
Kyriakopoulos et al., 201296	25	14	24	16.1(1.5); 12–19	n-back	13	EOS	+
Pauly et al., 200833	12	12	All	17.5(0.7)	Visual memory	3	EOS	+
						8	EOS	+
						1	EOS	+
						7	EOS	−
						1	EOS	−
						5	EOS	−
Rajarethinam et al., 201197	15	7	All	15.9(3.1)	Comprehension	5	SCZ-Rel	−
Seiferth et al., 200998	12	12	All	17.8(1.4)	Discrimination	1	EOS	+
						4	EOS	+
						1	EOS	+
						3	EOS	−
						18	EOS	−
						3	EOS	−
						3	EOS	−
Thermenos et al., 200799	21	12	18	19.9(4); 13–28	Elaborative encoding	2	SCZ-Rel	+
Urben et al., 2016100	6	5	All	16.5(1.05)	Verbal fluency	15	EOS	+
Wagshal et al., 2014101	10	5	All	12.6(2.32); 8–16	Weather prediction	6	EOS	+
						5	EOS	+
						17	EOS	+
						9	EOS	−
						6	EOS	−
						1	EOS	−
White et al., 2011a35	14	12	NA	13.4(2.6)	Sternberg	1	EOS	−
White et al., 2011b102	22	15	17	15(2.8)	Sternberg	10	EOS	+
						11	EOS	+
						4	EOS	+
						4	EOS	+
						6	EOS	+
						3	EOS	−

Note: n = sample size; R = Right-handed; NA = not available; SCZ-Rel = Relative or relative of Schizophrenia patient; EOS = Early onset schizophrenia; ^aschizophrenia minus healthy controls group (+), healthy controls group minus schizophrenia (−).

ALE Maps

ES-SDM³⁸ adopts features of ALE and incorporates statistical values (t-statistic) of peak coordinates to estimate significant clusters. ALE maps illustrated on figure 2 show suprathreshold clusters associated with adolescents with early onset schizophrenia, with all coordinates listed on table 2. Specifically, analysis on adolescents with early onset schizophrenia revealed reliable hyperactivity compared to controls within the left precuneus (Brodmann area [BA] 23), left inferior frontal gyrus (BA 45), and right cerebellum, wheras hypoactivity included the left fusiform gyrus (BA 19) extending to the cerebellum and right precuneus (BA 5).

Fig. 2.

2-D and 3-D images showing suprathreshold clusters associated with hyper- and hypo-activation of adolescents with early onset schizophrenia.

Table 2.

Significant Regions of Activation for (A) EOS > HC and (B) EOS < HC (Thresholded at P < .005; FWHM 20 mm)

Region	BAa	x	y	z	SDM-Z	P	Voxelsb	Jackknife
(A) EOS > HC
L Precuneus	23	−4	−60	28	2.599	1.217e-4	433	100c
L Inferior frontal gyrus	45	−52	38	12	2.169	1.035e-3	290	57
R Cerebellum		10	−80	−22	2.482	2.270e-4	147	86c
(B) EOS < HC
L Fusiform gyrus /cerebellum	19	−38	−68	−12	−1.614	4.577e-4	709	77c
R Precuneus	5	4	−50	46	−1.462	1.076e-3	164	77c

Note: EOS = early onset schizophrenia patients; HC = healthy controls; BA = Brodmann area; ^apeak coordinates with overlapping BA areas (in brackets); ^b2 mm × 2 mm × 2 mm; SDM-Z = signed differential mapping z-score; Jackknife replicability is represented as percentage; L = Left; R = Right; foci represented in MNI space; ^cregions greater than 75% replicability.

ALE maps illustrated on figure 3 show suprathreshold clusters associated with adolescents who had a parent diagnosed with schizophrenia, with all coordinates listed on table 3. The meta-analysis on adolescents who had a parent with schizophrenia showed concordance (ie, agreement across studies) based on hyperactivations in the right precuneus (BA 7/19) and right superior longitudinal fasciculus III extending to the insula. Concordance based on hypo-activations was observed in the right cingulate (BA 23), left insula (BA 13/48), and right middle temporal gyrus (BA 22).

Fig. 3.

2-D and 3-D images showing suprathreshold clusters associated with hyper- and hypo-activation of relatives with schizophrenia compared to controls.

Table 3.

Significant Regions of Activation for (A) SCZ-Rel > HC and (B) SCZ-Rel < HC (Thresholded at P < .005; FWHM 20 mm)

Region	BAa	x	y	z	SDM-Z	P	Voxelsb	Jackknife
(A) SCZ-Rel > HC
R Precuneus	7/19	16	−74	36	1.523	2.017e-4	887	77c
R Longitudinal fasciculus /insula		28	16	6	1.115	3.373e-3	130	55
(B) SCZ-Rel < HC
R Posterior cingulate gyrus	23	2	−24	34	−1.364	2.083e-3	344	77c
L Insula	13(48)	−32	−24	4	−1.387	1.835e-3	183	77c
R Middle temporal gyrus	22	56	−44	8	−1.428	1.403e-3	149	77c

Note: SCZ-Rel = adolescent relatives of schizophrenia patients; HC = healthy controls; BA = Brodmann area; ^apeak coordinates with overlapping BA areas (in brackets); ^b2 mm × 2mm × 2mm; SDM-Z = signed differential mapping z-score; Jackknife replicability is represented as percentage; L = Left; R = Right; foci represented in MNI space; ^cregions greater than 75% replicability.

Jackknife Analyses

To test the robustness of significant clusters, we used ES-SDM jackknife sensitivity analyses that reiterate the meta-analyses as many times as the number of studies and remove one study at a time outputting a percentage that signifies the strength of replicability. Results show that all clusters were highly replicable with more than 75% replicability, with the exception of the left inferior frontal gyrus that shows increased implication for adolescents with early onset schizophrenia compared to healthy controls (replicability 57%), and the right longitudinal fasciculus that showed increased implication to adolescents with relatives with schizophrenia compared to healthy controls (replicability 55%).

Discussion

Schizophrenia symptoms consist of disabling abnormalities in cognitive abilities such as deficits in logical sequences and loose associations. Functional neuroimaging studies use event-related cognitive paradigms to evaluate brain correlates of these processes. We investigated concordance across studies that examined brain responses to cognitive abilities in youth with early onset schizophrenia and youth with a parent diagnosed with schizophrenia. We highlight four main findings. First, in the prefrontal cortex, we observe clusters in ventrolateral prefrontal regions, rather than dorsolateral prefrontal regions. Second, the precuneus showed an altered activity compared to control participants for both groups affected by schizophrenia. Third, an altered implication of the cerebellum was observed for youth with early onset schizophrenia, whereas the implication of the insula was altered only for the youth with parents with schizophrenia. Fourth, the posterior cingulate and middle temporal gyrus showed reduced implications for youth with parents with schizophrenia.

The prefrontal cortex is well known for its dysfunction in adults with schizophrenia.^2,41,42 Critically, the prefrontal cortex is a large cortical region making 12.51% of the total brain volume.⁴³ The results revealed a cluster within the left inferior frontal gyrus demonstrating increased implication in adolescents with early onset schizophrenia compared to controls, with weak reliability. The inferior frontal gyrus has been ascribed with many cognitive functions such as working memory,^28,44 inhibitory control,^45–47 negative priming,⁴⁸ and voluntary risky decision-making.^49,50 Thus, according to the hierarchical model of prefrontal function, the inferior frontal gyrus does not underlie a single cognitive process but instead it is common to many processes that have a generic common mental activity requirement for processing a few items in mind.⁵¹ Notably, no suprathreshold concordance in the dorsolateral prefrontal cortex is observed for any of the performed meta-analyses. Unlike the inferior frontal gyrus, the dorsolateral prefrontal cortex consists mainly of the middle frontal gyrus, which is associated with processing problems of higher abstraction and complexity⁵¹ and is one of the last areas to mature in humans.¹⁰ The absence of reduced (or increased) involvement of dorsolateral areas may be related to incomplete maturation of the prefrontal cortex during adolescence. This finding is consistent with past meta-analyses with children and adolescents, which suggests that increased variability in the implication of prefrontal regions results in discordant findings for these age groups.^28,52 This may be attributed to the fact that adolescent participants in studies on early onset schizophrenic were medicated. There is some evidence to suggest that antipsychotic medication may impact prefrontal cortical activity and cortical thickness, with some differences between first- and second-generation antipsychotics.^53,54 Hence, the differential activation observed between the medicated and unmedicated groups may be at least partly artefactual.

Differences in activation between groups affected by schizophrenia and control groups were observed within the insula and cerebellum. Although less widely discussed, the insula and the cerebellum play a key role in cognitive abilities such as working memory in children²⁸ and adults.²⁴ In the current findings, hemispheric asymmetry in the cerebellum and insula was observed for adolescents with schizophrenia and adolescents with relatives with schizophrenia, respectively. Specifically, the insula shows reduced implication in the left hemisphere and increased implication in the right hemisphere for adolescents with relatives with schizophrenia. Anatomically, the insula is located deep in the gyri that connect the frontal and temporal regions. Functionally, in addition to interoceptive feeling,⁵⁵ the insula has been associated with various mental functions such as working memory,⁴⁴ different types of inhibition,⁴⁷ emotional valence,⁵⁶ and reward anticipation,⁵⁷ to which the insula has its own topographic map.⁵⁸ Perhaps the altered insula implication in adolescent relatives of schizophrenia may relate to the differences in motivational aspects of cognitive processes.

The cerebellum has been associated with motor coordination and temporal sequencing of cognitive information,^28,44 an important aspect of cognitive function that requires visuomotor sequencing of various actions. Meta-analyses of adult studies addressing the role of the cerebellum in schizophrenia have discussed its dysfunction in tasks of working memory and emotion.^59,60 We observed increased implication of the right cerebellum and a decreased implication of the left cerebellum for youth with early onset schizophrenia. The fact that the cerebellum is observed in the early onset group rather than the youth with relatives with schizophrenia may mark it as a target area for future studies of this disorder. Notably, laterality differences in both the insula and cerebellum serve as reference points for future research.

Posterior brain regions in the precuneus (BA 19) and posterior cingulate (BA 23) were also found to differ between groups affected by schizophrenia and control groups. The precuneus, within the parietal cortex, is bordered medially by the posterior cingulate cortex and ventrally to the fusiform gyrus. BA 19 extends to the posterior parts of the precuneus and fusiform gyri, and BA 23 covers medial parts of the precuneus and the posterior cingulate cortex. The analysis revealed that BA 23 is increased for adolescents with early onset schizophrenia whereas this region is reduced for adolescents with relatives with schizophrenia. Medial parts of the precuneus and posterior cingulate gyrus have been associated with decision-making in reward processing⁶¹ and self-directed thinking associated with the default-mode network in the absence of a task in healthy individuals.⁶² Notably, these networks are negatively related in healthy adults (ie, as brain areas of one network increase in activity, brain areas of the other network decrease in activity^44,63), yet are positively related in healthy children,⁶⁴ suggesting that there is a transformation in the processing of self-referential thinking between childhood and adulthood,⁶⁵ with profound lateralization of effective connections.⁶⁶ Interestingly, the default-mode network in schizophrenia patients and relatives of patients with schizophrenia seems to be affected,^67,68 thereby establishing further evidence that the precuneus may be associated with internally driven thoughts in early onset schizophrenia and relatives of patients. An increased functional implication in this region for adolescents with early onset schizophrenia may be driven by medication,^53,54 as a side effect of overcompensation, similar perhaps to the insula and inferior frontal gyrus, discussed above.

Posterior parts of the cortex (ie, BA 19 in either hemisphere) are associated with visual-spatial processes as they diverge into either the ventral stream that leads to the temporal lobe or the dorsal stream that leads to the parietal lobe^69,70 into the “what” and “where” pathways, respectively.⁷¹ BA 19 in the parietal cortex, found in the right hemisphere, shows increased implication for adolescents with relatives with schizophrenia. BA 19 in the fusiform gyrus has reduced implication for adolescents with early onset schizophrenia. Because this area is not in the same hemisphere, it may be less likely for this effect to be attributed to medication. An alternative explanation for this reverse effect could be strategy differences based on the visual-spatial requirements of the task.

The right middle temporal gyrus showed reduced implication for adolescents with relatives with schizophrenia, compared to controls. In the left hemisphere, the middle temporal gyrus is associated with language and semantic processing^72–74; however, the function of the right middle temporal gyrus has been associated with processing saccades,⁷⁵ processing emotions and social cognition,^76–78 and moral judgments.⁷⁹ In schizophrenia, the right middle temporal gyrus is associated with retrospective and prospective self-referential thinking,⁸⁰ and volumetric studies show gray matter abnormalities in patients with schizophrenia⁸¹ and those with first-episode schizophrenia,⁸² which may explain reduce implication of this region.

Limitations

Brain regions identified in these meta-analyses show convergence across published studies that reported experiments on brain responses to various cognitive tasks in adolescents with and without association to schizophrenia (ie, adolescents with a relative with schizophrenia and diagnosis of early onset schizophrenia compared to healthy controls). Brain activation to cognitive tasks inherently involves varying degrees of attention and task difficulty that was not considered in the current study due to an insufficient number of studies by task type. We also note that we only considered peer-reviewed published studies and recognized that results may be prone to type II errors as meta-analyses are exposed to publication bias. Moreover, as with any fMRI meta-analyses, interpretation of findings is liable to reverse inference because secondary analyses cannot manipulate experimental conditions as in original experiments. Although it is common to observe variability in methodological choices of original experiments, meta-analyses have proven to be a valuable tool for integrating many datasets to establish converging patterns among data.⁸³

Recommendations for Future Studies

fMRI meta-analyses are useful at least in two fundamental ways: to identify overarching patterns of results reported in the literature and to identify methodological and reporting issues in the current literature that can improve for future meta-analyses. We identify several points that we experienced in performing these meta-analyses and make related recommendations for future research. First, it is important for future work to report whole-brain coordinates, at least as supplementary material, because a substantial number of studies report only region of interest analyses and had to be excluded from the meta-analyses. Second, it is useful to include behavioral indices associated with the performance of each group as this would allow future meta-analysis to examine coordinates attributed to dysfunction of specific cognitive processes. Although practically difficult, it would be immensely useful to examine cognitive function before medication onset, which would allow for the investigation of the unmedicated state of the disorder. To date, the meta-analyses performed in the current study and prior studies^8,9,84 have been performed using between subjects (eg, group affected by schizophrenia > control group). We recommend future studies to report coordinates of within-group results, if possible as supplementary material as this would allow for performing within-group meta-analyses to identify overarching patterns of concordance. Currently, this option is not possible because most studies report between-subject contrasts. Hence, our recommendation is for future studies to report foci corresponding to within-group contrasts as well as between-group contrasts to allow researchers to compare convergence of brain activation derived from both types of experimental design.⁸⁵ Overall, more research is needed with youth affected by schizophrenia and we feel optimistic that knowledge gained in this direction would help improve early treatment options for affected individuals.

Conclusions

Schizophrenia affects less than 1% of the population worldwide; however, its social costs are staggering as measured by Disability Adjusted Life Years Lost (DALY).⁸⁶ Specifically, the cost of schizophrenia to the U.S. economy amounts to about $62 billion per year, and the illness is associated with a reduced life expectancy of about 25 years as a result of various associated problems, including suicide, cancer, or heart disease.^87–89 Developing a clearer profile of biomarkers at an earlier point before frank illness onset, and before the neurological picture becomes more complicated due to medication, may facilitate a more accurate differential diagnosis, early intervention, and relapse prevention. Our research demonstrates different patterns when adolescents are relatives to patients with schizophrenia and those who have received a diagnosis for early onset of the disease and are on medication. Specifically, our key findings highlight the altered implication of the parietal cortex in both groups affected by schizophrenia; however, the insula and cerebellum are altered only for adolescents with relatives with schizophrenia and adolescents with early onset schizophrenia, respectively. These findings support the clinical notion that neural expressions of the disease are affected by a multiplicity of factors that may benefit from target investigations of marker brain regions. To date, identifying biomarkers for psychiatric illnesses including schizophrenia has been elusive; however, ongoing research in this field is being conducted.^90,91 Schizophrenia is considered to be a heterogeneous disorder⁹⁰ associated with limited diagnostic reliability.⁹² These issues will make identification of biomarkers an exceptionally challenging process. However, practically, our results provide a set of stereotaxic coordinates that can serve as a neurofunctional atlas for future region-of-interest studies. Considerably more research is needed in understanding brain correlates that may be aberrant in those at risk of converting, in order to eventually be able to accurately distinguish these participants from healthy controls and those with first-episode schizophrenia.

Funding

This work was in part supported by the Russian Foundation for Basic Research grants (ofi-m 17-29-02518) (Neuroimmunological status and basic cognitive-affective structures of the human brain in norm and in patients with schizophrenia).