Identification of comprehensive genetic factors, pathways, and shared genetic architecture of putamen volume in adolescent cohort

Abstract

The putamen plays a key role in motor control, learning, and cognition, with abnormal putamen volumes associated with neuropsychiatric disorders. Using data from the ABCD study, we performed a genome wide association study (GWAS) of putamen volumes, followed by replication and pathway enrichment analyses. We next evaluated the shared genetic architecture of putamen volume and neuropsychiatric disorders—including depression, schizophrenia, Parkinson’s, ADHD, bipolar, and OCD— using SNP associations from this and prior GWASs. We identified 142 genome-wide significant SNP associations in White participants. Most identified SNPs were in gene regulatory regions and in the neuronal growth-linked genes, DCC and DSCAM. Sixteen of the seventeen most significant associations observed in White participants were also observed in non-White participants. Seventeen known SNPs from prior GWASs of putamen volume were also replicated in our ABCD analysis, including five of the eight most significant GWAS SNPs. There was considerable genetic heterogeneity between White and non-White participants in putamen-linked SNPs with significant differences between the minor allele frequencies across the two groups (Wilcoxon rank-sum test Exact prob < 0.0001). We identified a key pathway (REACTOME_DSCAM_INTERACTIONS) associated with putamen volume that involves DSCAM gene, netrin-1 protein and/or DCC gene. In addition, 28 unique SNPs from prior GWASs of neuropsychiatric disorders were strongly associated with putamen volume at Bonferroni-corrected significance, while 41 SNPs shared by at least three disorders were associated with putamen volume at a 0.05 threshold. Our findings provide deeper insights into the shared genetic architecture and cross-population differences in genetic associations of putamen volume.

Introduction

The putamen, a structure in the basal ganglia, is crucial for motor control, learning, and cognitive functions. Abnormal putamen volumes have been associated with several neuropsychiatric disorders [1], including Parkinson’s disease [2], Huntington’s disease [3], schizophrenia [4], obsessive-compulsive disorder (OCD) [5], bipolar disorder [6], and attention impairment in attention deficit hyperactivity disorder (ADHD) [7]. Despite the potential importance of putamen volumes in neuropsychiatric disorders, its genetic underpinning, particularly in youth, remains poorly characterized.

In several neuropsychiatric and neurodegenerative disorders, including Parkinson’s [2], Huntington’s [3], Alzheimer’s disease [8], ADHD [7], and major depression [9], putamen volumes are typically reduced. In contrast, enlargement of the putamen has been observed in bipolar disorder [6], schizophrenia [4], autism [10], and OCD [5]. Dopamine dysregulation and neuroinflammation may give rise to these structural changes in conditions like Schizophrenia and OCD [11]. In major depression, reductions in putamen volume have been associated with impaired reward sensitivity and motivational deficits. Smaller putamen volumes, for example, correlate with symptoms of anhedonia in major depression patients [12,13]. The increased putamen volume observed in schizophrenia and bipolar disorder, particularly in patients treated with antipsychotic or mood-stabilizing medications, is believed to result from drug-induced neuroplastic changes, leading to structural enlargements [14,15]. Genetic predispositions and environmental stressors also contribute to alterations in putamen volume [16,17].

Putamen volume heritability is estimated between 71–79% [18], with twin studies confirming genetic influence [19]. Genome-wide association studies have identified loci, such as KTN1 [18], DCC [17,20], BCL2L1 [18, 20], linked to putamen volume, suggesting that neural growth and plasticity of the putamen are influenced by genetic variations. There may also be genetic overlap between putamen volume and neurodegenerative diseases like Huntington’s disease [21,22] and Parkinson’s disease [23], where putamen atrophy is pronounced, and neuropsychiatric diseases like schizophrenia, where dopamine-related genes are associated with the disease [24, 25]. There may also be a shared genetic architecture between putamen volume and “broad” depression, including major depression [26] and depressive symptoms.

Most large-scale genetic studies of putamen volume, including the aforementioned, relied on adult or full-lifespan datasets that included adolescent participants but were not specifically focused on adolescent populations [17,18, 20, 27–29], except for one study that examined gray matter [30] rather than total putamen volume. The genetic underpinning of putamen volume may differ depending on the age of participants [31]. Brain regions follow distinct maturational trajectories, with subcortical structures such as the putamen reaching peak volume earlier in development than cortical regions. As a result, the genetic factors driving putamen volume during active development may differ from those involved in the gradual volume loss observed during aging. Changes in putamen volume in older individuals may not occur for the same reasons as they do in the younger populations. Moreover, many neuropsychiatric disorders emerge during adolescence (including ones associated with altered putamen volume). Thus, studying the relationship between genetic variants and putamen volume among youth may be particularly important to gain insights into disease course and etiology. Using the Adolescent Brain and Cognitive Development (ABCD) dataset, we performed a main-effect GWAS of putamen volume, followed by evaluation of the generalizability of GWAS discovery associations and pathways enrichment scoring analysis. Second, we examined whether single nucleotide polymorphisms (SNPs) identified in prior GWASs of putamen volumes would overlap with our discovery SNPs or replicate in the ABCD dataset. Third, we examined whether SNPs associated with neuropsychiatric disorders – depression, schizophrenia, Parkinson’s disease, ADHD, bipolar disorder, and OCD – from prior GWASs would associate with putamen volume in ABCD dataset and identified their shared genetic architecture. We hypothesized that our GWAS analysis of the ABCD dataset cohort would (i) identify genetic variants associated with putamen volume (“discovery SNPs”), (ii) that genetic variants associated with putamen volumes from prior GWASs would associate in the ABCD dataset (“known SNPs”), and (iii) that genetic variants associated with neuropsychiatric disorders from prior GWASs would associate with putamen volume in the ABCD dataset and these genetic variants would share across the neuropsychiatric disorders (“shared genetic architecture”). Lastly, we explored the overlap between discovery SNPs, known SNPs, and SNPs associated with neuropsychiatric disorders from prior GWASs. Our expectation was that we would identify novel genetic associations with putamen volume, observe overlap with known SNPs, and reveal a shared genetic architecture between psychiatric disorders and putamen volume.

Material and methods

Dataset

We used phenotypic, genetic and brain imaging (MRI) data from the Adolescent Brain Cognitive Development (ABCD) Study. ABCD is a large-scale, longitudinal research project aimed at understanding brain development and its impact on cognitive, social, emotional, and academic outcomes in adolescents. Launched in 2016, it follows over 11,000 children across the U.S., beginning at ages 9-10, to track their development into young adulthood. The study employs neuroimaging, behavioral assessments, and biospecimen analyses to explore the influence of genetic, environmental, and lifestyle factors on the developing brain to understand adolescent brain maturation and mental health trajectories [32, 33]. While the study population comprised mostly of White and Black participants, there were also participants from other self-identified racial/ethnic groups, including Natives and Islanders (Native American, Alaska Native, Native Hawaiian, Guamanian, Samoan, and Other Pacific Islander), and Asians and Other Races (Chinese, Japanese, Korean, Filipino, Vietnamese, Other Asian, Asian Indians, and Other Races) (Supplementary Figure 1). In this study, we partitioned the data into two groups, i.e., White and non-White participants, using self‑identified race/ethnicity from the dataset [34]. We used the data from White participants, which was the largest group, for the discovery GWAS analysis, and aggregated data from non-White participants to evaluate the generalizability(i.e., replication of associations) in this distinct and genetically diverse group of participants.

Genotypic dataset

We obtained imputed genotyping data for single nucleotide polymorphisms (SNPs) from ABCD release 5. To ensure robust results, we performed quality control (QC) analyses separately for White and non-White participants, respectively, discarding all multiallelic variants. We applied stringent QC procedures on the imputed SNP dataset that were designed to filter out low-quality data and involved several criteria [35]. Specifically, we excluded individuals with more than 5% missing genotype data, and SNPs with more than 5% missing data were similarly removed. We also applied a minimum minor allele frequency (MAF) threshold of 1%, ensuring that only common variants were included in the analysis. SNPs that significantly deviated from Hardy-Weinberg equilibrium (HWE) at a P-value threshold of less than 1e-6 were excluded, as such deviations may indicate genotyping errors. As part of our QC process, we checked for sex misclassification by comparing the genetic sex inferred from the genotype data to the recorded sex, and we assessed the genetic relatedness of individuals based on identity by descent (IBD), excluding any misclassified or related participants. In addition, in order to address heterogeneity in ancestry and to carefully model population structure within each group, we estimated principal components (PCs) [36] to account for population structure and potential confounding due to ancestry. The PCs were calculated based on linkage disequilibrium (LD)-pruned genotype SNPs within each racial group and followed standard practice of using scree plots to determine the optimal number of PCs, yielding 4 PCs each in both White and non-White participants (Supplementary Figures 2 and 3). The QC filters, procedures, and population stratification resulted in high-quality genetic data consisting of 8.28 million SNPs from 6200 White participants and 12.06 million SNPs from 3130 non-White participants.

Phenotypic/demographic dataset

The study involved demographic variables age, sex (M/F), and race/ethnicity. The White participants included 2896 females (46.71%) and 3304 males (53.29%), while the non-White participants included 1240 females (39.62%) and 1890 males (60.38%). The mean age of White and non-White participants was 118.85 (SD = 7.49) and 118.71 (SD = 7.35) months, respectively.

Imaging dataset

The imaging component of the study was designed by the ABCD Data Analysis and Informatics Center (DAIC) in collaboration with the ABCD Imaging Acquisition Workgroup. Imaging techniques and assessments were carefully chosen, optimized, and standardized across all sites of data collection [33]. Structural MRI scans were segmented into cortical and subcortical regions, which were then utilized to define individualized putamen regions for each participant. We used the total putamen volume as the main outcome variable. We also included brain imaging data on total brain volume in our analysis to adjust the genetic associations of putamen volume for this variable.

Statistical analysis

We assessed the distribution of the outcome variable (total putamen volume) for deviations from normality and determined that no transformation was necessary to achieve approximate normality within each population group. The distributions are shown in Supplementary Figures 4 and 5, which ensures that the assumption of normality for the outcome variable is met for the forthcoming analysis.

Main-effect discovery GWAS

We performed a discovery GWAS analysis to evaluate the SNP main effect (ME) for the total putamen volume using 8.28 million SNPs from 6200 White participants. We modeled these SNP main effect GWAS analyses under an additive model using linear regression, adjusting for age, sex, brain volume, site, and ancestry PCs to control for population structure within each racial group, as shown below:

$$\mathit{Total}\mathit{Putamen}\mathit{Volume}~\mathit{SNP}+\mathit{Age}+\mathit{Sex}+\mathit{Brain}\mathit{Volume}+\mathit{site}+\mathit{ancestry}\mathit{PCs}$$

The genome-wide significance of ME associations was evaluated at the Bonferroni correction, i.e., 6.04e-9 for White participants.

Evaluation of the generalizability of GWAS associations

Within the distinct and genetically diverse non-White participants, we examined the SNPs that showed Bonferroni-corrected significant associations with putamen volume in the discovery GWAS analysis conducted on White participants. For each SNP identified in the discovery GWAS analysis, we evaluated the same linear regression association model in non-White participants, i.e., Total Putamen Volume ~ SNP + Age + Sex + Brain Volume + site + ancestry PCs, and determined the association p-values at the 0.05 significance level.

Gene pathway scoring analysis

We used the association results (i.e., p-values) from the discovery GWAS to perform pathway enrichment scoring, evaluating the scores of 1077 predefined pathways from the KEGG, BioCarta, and Reactome databases. Genome-wide SNPs were mapped to genes, and the association scores for all genes in a pathway were calculated, as implemented in the Pathway Scoring Algorithm [37]. Briefly, the algorithm computes gene scores by aggregating SNP p-values from a GWAS, either by summing or taking the maximum of the corresponding chi-squared association statistics, while accounting for their LD-induced correlation structure through efficient numerical and analytic methods. For pathways, gene sets are obtained from external databases, and pathway scores are derived by integrating the scores of constituent genes using chi-squared–based or empirical procedures. To avoid inflation due to correlated signals, genes that are in strong LD are dynamically fused into single units before pathway scoring, thereby preventing double-counting of shared association signals at the pathway level.

Evaluation of the overlap and replication of known associations from prior studies

We searched the NHGRI GWAS catalog [38] for all candidate SNPs that have been associated with putamen volume thus far. First, we explored the overlap between these SNPs and our discovery GWAS SNPs, then evaluated their associations in the ABCD dataset for White and non-White participants separately using the aforementioned linear regression model on total putamen volume adjusted by age, sex, brain volume, site, and the ancestry PCs for each resulting SNP. We considered the replication statistically significant if the p-values were at the Bonferroni corrected threshold level (i.e., 7.94e-4 for 63 SNPs) in at least one population group, i.e., in White, non-White, or both participants.

Shared genetic architecture of putamen volume and related psychiatric disorders

Given that abnormal putamen volume has been linked to depression [9, 26], schizophrenia [4], Parkinson’s disease [2], ADHD [7], bipolar disorder [6], and OCD [5], we searched for SNPs associated with these neuropsychiatric disorders and evaluated whether these SNPs were also associated with total putamen volume in the ABCD dataset. We examined this for White and non-White participants separately using linear regression, adjusting for age, sex, brain volume, site, and ancestry PCs, in a similar manner. We performed analyses and organized the outcomes in two ways to test: 1) whether a SNP associated with each neuropsychiatric disorder, as identified from the NHGRI GWAS catalog, was also associated with putamen volume at a stringent Bonferroni-corrected threshold, and 2) whether a SNP associated with two or more neuropsychiatric disorders, as identified from the NHGRI GWAS catalog, was also associated with putamen volume at a threshold level of 0.05 in White and/or non-White participants in the ABCD dataset. We also explored whether any of these known SNPs intersected with our discovery SNPs.

Lastly, we also evaluated the independence of the associated SNPs in both the discovery GWAS results and the shared genetic architecture, based on LD patterns in White and non-White populations, using a LDLink [39].

All data processing and analyses, unless specified otherwise, were performed using STATA SE 17.0 (StataCorp LP, College Station, TX, USA), PLINK v1.9 [40], or R version 4.3.1 with RStudio, 2023.06.1 Build 524, © 2009–2023.

Results

Main-effect discovery GWAS

Fig. 1A and B display the Manhattan and QQ plots of SNP p-values from the GWAS analyses of total putamen volume among White participants in the ABCD study. As depicted in these plots, we observed very strong GWAS signals on chromosomes 14, 18, and 21, which included 142 genome-wide significant SNP associations. A list of the 50 most significant genome-wide SNPs from White participants is shown in Table 1, with all SNPs listed in Supplementary Table 1. The pie chart showing the proportions of functional contexts for all genome-wide significant SNPs is presented in Fig. 1C. Of these, 112 and 13 intronic variants were from the protein-coding genes DCC (chromosome 18) and DSCAM (chromosome 21), respectively. Seven SNPs were from the long non-coding RNA (lncRNA) LOC124903320 (chromosome 14). Among the most significant SNPs, five (rs8017172, rs1953350, rs1953351, rs1959089, and rs1953352) were regulatory region variants, located in enhancer regions and CCCTC-binding factor (CTCF) protein binding sites. All genome-wide significant SNPs from the DSCAM gene were in strong LD (R² = 0.848–1). The SNPs from the LOC124903320 gene were also in LD with R² = 0.76–1. The SNPs from the DCC gene showed strong LD in multiple groups, covering all SNPs with R² = 0.80–1. Chromosome-wise overall LD patterns for each pair of discovery SNPs are provided in Supplementary Table 2. While LOC124903320 gene is an uncharacterized locus (https://www.ncbi.nlm.nih.gov/gene/124903320), published literature suggests that both DCC and DSCAM have been linked with neuronal growth and human central and peripheral nervous system development [41–47].

Fig. 1. Genome-wide main-effect association analysis summary in White participants.

A: Manhattan plot of GWAS association results. B: QQ plot of GWAS association results. C: Pie chart showing the functional context of discovery GWAS SNPs.

Table 1.

Genome-wide main-effect associations in White participants.

	SNP	CHR	POSITION	BIOTYPE	GENE	CONTEXT	A1 WH	MAF WH	A1 NonWH	MAF NonWH	BETA WH	P WH	BETA NonWH	P NonWH	PRIOR STUDIES TRAITS (PUBMEDID)
1	rs8017172	14	55732330	CTCF binding site	-	regulatory region	A	0.45	G	0.30	−131.50	4.75E-15	139.80	2.54E-07	Putamen Vol. (29147026) Parkinson’s (28892059)
2	rs945270	14	55733755	-	-	intergenic	G	0.45	C	0.28	−130.80	6.54E-15	138.00	8.48E-07	Putamen Vol. (31636452)
3	rs8014725	14	55720235	lncRNA	LOC124903320	intron	G	0.47	A	0.26	−129.30	1.49E-14	152.10	7.16E-08
4	rs148470213	14	55726982	-	-	intergenic	C	0.49	T	0.23	−128.90	1.72E-14	144.90	2.24E-06	Putamen Vol. (31636452)
5	rs10143262	14	55718343	lncRNA	LOC124903320	intron	T	0.47	C	0.27	−128.90	1.73E-14	151.80	5.96E-08
6	chr14:55722612	14	55722612	lncRNA	LOC124903320	downstream	G	0.47	T	0.26	−128.80	1.88E-14	153.60	6.20E-08
7	rs868202	14	55729044	-	-	intergenic	C	0.47	T	0.26	−128.60	2.01E-14	151.00	1.10E-07	Putamen Vol. (31636452)
8	rs10129414	14	55726554	lncRNA	LOC124903320	downstream	A	0.47	G	0.28	−128.50	2.19E-14	153.20	2.41E-08	Putamen Vol. (31636452)
9	rs1953351	14	55730507	enhancer	-	regulatory region	G	0.47	A	0.28	−128.10	2.55E-14	151.50	3.72E-08
10	rs1953350	14	55730556	enhancer	-	regulatory region	A	0.47	C	0.28	−128.10	2.55E-14	151.60	3.57E-08
11	rs1959089	14	55731119	enhancer	-	regulatory region	C	0.47	T	0.28	−127.80	2.91E-14	151.40	3.51E-08
12	rs2342589	14	55730350	-	-	intergenic	A	0.47	G	0.32	−127.20	3.91E-14	147.70	2.02E-08
13	rs2342588	14	55730321	-	-	intergenic	C	0.47	T	0.32	−127.20	3.91E-14	147.70	2.02E-08
14	rs1953352	14	55730441	enhancer	-	regulatory region	C	0.47	T	0.32	−127.20	4.00E-14	147.70	2.02E-08
15	rs8012377	14	55724695	lncRNA	LOC124903320	downstream	T	0.47	C	0.28	−127.10	4.22E-14	151.00	3.51E-08	Parkinson’s (38155330)
16	rs9671291	14	55714977	lncRNA	LOC124903320	upstream	G	0.46			−120.20	6.09E-13
17	rs8021018	14	55710790	lncRNA	LOC124903320	upstream	T	0.45	C	0.31	−119.40	9.47E-13	91.43	5.32E-04
18	rs7226995	18	53369385	protein coding	DCC	intron	A	0.49	G	0.42	−108.50	2.06E-10	29.47	0.22
19	rs10502971	18	53361591	protein coding	DCC	intron	A	0.41	A	0.28	110.40	2.44E-10	28.88	0.27
20	rs62097995	18	53316863	protein coding	DCC	intron	A	0.41	A	0.16	110.10	3.19E-10	−0.20	1.00
21	rs62097996	18	53316899	protein coding	DCC	intron	A	0.41	A	0.17	110.10	3.19E-10	−2.95	0.93
22	rs62097994	18	53314592	protein coding	DCC	intron	A	0.41	A	0.17	109.30	4.23E-10	-2.46	0.94
23	rs62098036	18	53367709	protein coding	DCC	intron	A	0.49	G	0.49	−106.20	4.66E-10	20.28	0.40
24	chr18:53300962	18	53300962	protein coding	DCC	intron	A	0.49	T	0.47	105.90	5.17E-10	−23.29	0.34
25	rs9954838	18	53312306	protein coding	DCC	intron	A	0.49	C	0.41	−106.20	5.27E-10	−7.53	0.75
26	rs11662416	18	53321414	protein coding	DCC	intron	G	0.40	G	0.16	108.70	5.73E-10	−1.63	0.96
27	rs7227402	18	53369396	protein coding	DCC	intron	T	0.40	T	0.26	107.40	8.01E-10	37.60	0.16
28	rs6508231	18	53369745	protein coding	DCC	intron	C	0.50	T	0.41	−104.70	8.08E-10	33.52	0.16
29	rs11877400	18	53368227	protein coding	DCC	intron	G	0.50	T	0.41	−104.70	8.08E-10	32.46	0.18
30	rs75048819	18	53318768	protein coding	DCC	intron	G	0.40	G	0.16	107.50	8.34E-10	−0.63	0.99
31	rs8130627	21	40230912	protein coding	DSCAM	intron	G	0.46			−101.90	8.46E-10
32	rs2837518	21	40226239	protein coding	DSCAM	intron	T	0.49			−101.60	8.63E-10
33	rs60218578	18	53369246	protein coding	DCC	intron	T	0.50	G	0.41	−104.50	8.85E-10	33.50	0.16
34	rs6508230	18	53358554	protein coding	DCC	intron	T	0.40	T	0.26	106.60	9.76E-10	33.73	0.21
35	rs77162730	18	53318770	protein coding	DCC	intron	G	0.40	G	0.16	107.00	1.00E-09	−0.63	0.99
36	rs9954990	18	53312275	protein coding	DCC	intron	A	0.49	G	0.40	−104.10	1.04E-09	−9.50	0.70
37	rs1346914	18	53319460	protein coding	DCC	intron	T	0.40	T	0.16	106.90	1.05E-09	−4.14	0.90
38	rs2837517	21	40225992	protein coding	DSCAM	intron	G	0.49			−101.00	1.07E-09
39	rs960697	21	40232312	protein coding	DSCAM	intron	G	0.47			−100.90	1.12E-09
40	rs17503448	18	53331293	protein coding	DCC	intron	C	0.40	C	0.16	106.80	1.13E-09	−4.42	0.89
41	rs77916462	18	53318766	protein coding	DCC	intron	G	0.40	G	0.16	106.70	1.14E-09	−1.42	0.97
42	rs11702542	21	40231751	protein_coding	DSCAM	intron	C	0.46			−100.80	1.29E-09
43	rs17504871	18	53382033	protein coding	DCC	intron	G	0.47	A	0.28	−104.10	1.30E-09	−21.68	0.42
44	rs62098023	18	53367632	protein coding	DCC	intron	C	0.40	C	0.16	106.30	1.33E-09	7.46	0.82
45	rs11664320	18	53346253	protein coding	DCC	intron	C	0.40	C	0.26	106.00	1.40E-09	36.71	0.17
46	rs17412782	18	53361981	protein coding	DCC	intron	C	0.40	C	0.16	105.70	1.47E-09	2.16	0.95
47	rs1504746	18	53331649	protein coding	DCC	intron	C	0.40	C	0.16	106.00	1.49E-09	0.02	1.00
48	rs2837515	21	40225246	protein_coding	DSCAM	intron	C	0.49			−100.20	1.50E-09
49	rs62098022	18	53362521	protein coding	DCC	intron	C	0.40	C	0.17	105.70	1.51E-09	0.71	0.98
50	rs62098014	18	53342715	protein coding	DCC	intron	C	0.40	C	0.16	105.90	1.53E-09	−1.92	0.95

Chromosomal base position corresponds to the human reference genome assembly GRCh38.

Evaluation of the generalizability of GWAS associations

Table 1 also presents the results of aforementioned GWAS discovery SNPs’ associations in distinct and genetically diverse non-White participants. Sixteen of the seventeen most significant discovery SNPs from White participants were also significant in the non-White participants, confirming the generalizability of discovery associations. These SNPs included all five regulatory region variants. However, we observed that the minor allele (A1) was flipped in non-White participants compared to White participants for these sixteen SNPs, indicating genetic heterogeneity between the two population groups. Nonetheless, the direction of the association remained consistent, with the same risk allele being linked with a decrease in putamen volume (see Beta sign in Table 1). In addition, MAF of significant SNPs in White participants (Table 1) varied from 0.40 – 0.50 and varied from 0.16 – 0.49 for the same SNPs in non-White participants. Wilcoxon rank-sum test suggested that the MAFs of the SNPs for the White and non-White participants were not similar and that there was a significant difference between the two paired groups of MAFs (Exact prob < 0.0001).

Gene pathway scoring analysis

Table 2 presents the outcomes of pathway enrichment scoring analysis for the White participants, where all the most significantly associated pathways are listed. Of these, only one pathway, i.e., REACTOME_DSCAM_INTERACTIONS (https://www.reactome.org/content/detail/R-HSA-376172) tentatively reached Bonferroni corrected significance (i.e., ~4.64e-05). The pathway appears to be linked with nervous system development when the DSCAM gene binds netrin-1 protein and works alongside or independently of DCC gene to guide commissural axons across the midline, cooperatively depending on the cellular and developmental context [46, 48].

Table 2.

Gene pathways enrichment scoring analysis from the White participants GWAS analysis.

	Pathway Name	chi2Pvalue	empPvalue
1	REACTOME_DSCAM_INTERACTIONS	9.12E-05	1.81E-04
2	BIOCARTA_BAD_PATHWAY	2.63E-04	5.20E-04
3	REACTOME_RECYCLING_PATHWAY_OF_L1	8.48E-04	9.30E-04
4	REACTOME_L1CAM_INTERACTIONS	2.14E-03	2.51E-03
5	REACTOME_NA_CL_DEPENDENT_NEUROTRANSMITTER_TRANSPORTERS	3.24E-03	5.42E-03
6	REACTOME_MRNA_PROCESSING	3.61E-03	3.67E-03
7	REACTOME_PROCESSING_OF_CAPPED_INTRON_CONTAINING_PRE_MRNA	3.75E-03	4.01E-03
8	REACTOME_MRNA_SPLICING	3.88E-03	4.11E-03
9	BIOCARTA_MYOSIN_PATHWAY	4.11E-03	5.94E-03
10	BIOCARTA_MPR_PATHWAY	4.37E-03	5.04E-03
11	BIOCARTA_TPO_PATHWAY	5.36E-03	7.71E-03
12	BIOCARTA_RANMS_PATHWAY	6.54E-03	4.91E-03
13	REACTOME_AMINE_COMPOUND_SLC_TRANSPORTERS	7.23E-03	9.88E-03
14	KEGG_MTOR_SIGNALING_PATHWAY	7.46E-03	9.40E-03
15	REACTOME_IL_2_SIGNALING	7.61E-03	9.17E-03
16	KEGG_PANCREATIC_CANCER	8.07E-03	7.85E-03
17	REACTOME_SIGNAL_TRANSDUCTION_BY_L1	9.17E-03	1.19E-02
18	REACTOME_CELL_CELL_COMMUNICATION	9.44E-03	7.91E-03

Evaluation of the overlap and replication of known associations from prior studies

The NHGRI GWAS catalog [38] search revealed that prior GWASs have identified a total of 63 SNPs associated with putamen volume [17, 18, 20, 49–51]. Of these, 5 SNPs overlapped with 8 of the most significant discovery SNPs (Table 1), and 17 SNPs showed significant association at Bonferroni corrected threshold (i.e. 7.94 e-4) with putamen volume in ABCD White and/or non-White participants, including 12 SNPs that mapped to six genes (i.e., DCC, DSCAML1, BCL2L1, MYLK2, DLG2, and SHTN1), as presented in Table 3. These outcomes demonstrated the replication of known SNP associations with putamen volume in ABCD participants. Notably, in the most significant five SNPs (i.e., rs8017172, rs945270, rs148470213, rs868202, and rs10129414) that displayed a significant association in both White and non-White participants, there was a flip of minor allele, indicating genetic heterogeneity among the putamen-linked SNPs in Whites and non-Whites, consistent with our previous observations; however, there were consistent directions of association with the same risk allele. Finally, the 17 replicated known SNPs from prior studies (Table 3) included the five most significant discovery SNPs (i.e., rs8017172, rs945270, rs148470213, rs868202, and rs10129414) from our GWAS (Table 1), highlighting the robustness of the discovered associations with putamen volume in our GWAS analysis.

Table 3.

Evaluation of known putamen-SNP associations from prior studies, as identified in NHGRI GWAS catalog, in ABCD dataset.

	SNP	CHR	POSITION	BIOTYPE	MAPPED GENE	CONTEXT	NHGRI P PUTAMEN	PUBMED ID	A1 WH	MAF WH	A1 NonWH	MAF NonWH	BETA WH	P WH	BETA NonWH	P NonWH
1	rs8017172	14	55732330	-	KTN1 - RPL13AP3	regulatory	2.00E-24	29147026	A	0.45	G	0.30	−131.50	4.75E-15	139.80	2.54E-07
2	rs945270	14	55733755	-	KTN1 - RPL13AP3	intergenic	7.00E-53	31636452	G	0.45	C	0.28	−130.80	6.54E-15	138.00	8.48E-07
3	rs148470213	14	55726982	-	KTN1 - RPL13AP3	intergenic	6.00E-40	31636452	C	0.49	T	0.23	−128.90	1.72E-14	144.90	2.24E-06
4	rs868202	14	55729044	-	KTN1 - RPL13AP3	intergenic	1.00E-49	31636452	C	0.47	T	0.26	−128.60	2.01E-14	151.00	1.10E-07
5	rs10129414	14	55726554	lncRNA	KTN1 - RPL13AP3	intergenic	2.00E-50	31636452	A	0.47	G	0.28	−128.50	2.19E-14	153.20	2.41E-08
6	rs62097986	18	53292457	protein_coding	DCC	intron	1.00E-13	25607358	A	0.39	A	0.16	95.94	4.20E-08	−3.90	0.908
7	rs11660938	18	53286366	protein_coding	DCC	intron	4.00E-12	29147026	T	0.40	T	0.16	94.82	5.94E-08	−3.37	0.921
8	rs62098013	18	53337491	protein_coding	DCC	intron	8.00E-20	31636452	A	0.35	A	0.13	92.28	2.72E-07	−13.80	0.705
9	rs7902527	10	116955888	protein_coding	SHTN1	intron	1.00E-09	31636452	A	0.24	A	0.24	96.78	5.11E-07	−9.63	0.728
10	rs10439607	20	31670738	protein_coding	BCL2L1	intron	4.00E-16	31636452	A	0.30			−84.08	4.41E-06
11	rs666845	11	83566501	protein_coding	DLG2	intron	2.00E-08	29147026	T	0.40	T	0.50	78.33	5.24E-06	37.32	0.115
12	rs683250	11	83565125	protein_coding	DLG2	intron	4.00E-11	25607358	G	0.40	G	0.43	77.86	6.10E-06	61.23	0.011
13	rs6087771	20	31718921	protein_coding	BCL2L1	intron	2.00E-09	29147026	C	0.29			−81.98	9.05E-06
14	rs1432054	11	83549182	protein_coding	DLG2	intron	2.00E-14	31636452	G	0.38	G	0.34	74.89	1.61E-05	74.73	3.43E-03
15	rs6060983	20	31833121	protein_coding	MYLK2	intron	2.00E-15	31636452	C	0.30			−72.76	6.67E-05
16	rs35200015	11	117512500	protein_coding	DSCAML1	intron	4.00E-17	31636452	A	0.19	A	0.16	−78.40	2.29E-04	−26.51	0.412
17	rs597583	11	117551084	protein_coding	DSCAML1	intron	4.00E-07	29147026	G	0.22	G	0.24	−67.77	7.38E-04	−62.42	0.026

Shared genetic architecture of putamen volume and related psychiatric disorders

Our NHGRI GWAS catalog [38] search identified the following number of SNPs associated with neuropsychiatric disorders in over 300 prior GWASs and other genetic studies: 2364 SNPs for depression, 4150 SNPs for schizophrenia, 602 SNPs for Parkinson’s disease, 1929 SNPs for ADHD, 1600 SNPs for bipolar disorder, and 275 SNPs for OCD. A subset of these SNPs were available within the quality-controlled ABCD GWAS dataset: 2030 SNPs for depression, 3473 SNPs for schizophrenia, 498 SNPs for Parkinson’s disease, 1744 SNPs for ADHD, 1344 SNPs for bipolar disorder, and 256 SNPs for OCD.

Table 4 presents the known SNPs from prior GWASs of neuropsychiatric disorders that were also associated with putamen volume in the ABCD sample at a stringent Bonferroni-corrected significance threshold. These SNPs included 10 SNPs for depression [52–56], 2 SNPs for schizophrenia [57, 58], 5 SNPs for Parkinson’s disease [59–63], 11 SNPs for ADHD [64–67], 2 SNPs for bipolar disorder [55, 67], and 1 SNP for OCD [68]. Notably, SNPs rs4632195 and rs11663393 from the DCC, a gene identified in our discovery GWAS analysis, reached Bonferroni-corrected significance for association with putamen volume in White participants in the ABCD study. In addition to being associated with putamen volume, SNP rs4632195 shared an association with three neuropsychiatric disorders (depression, ADHD, and bipolar disorder) and rs11663393 shared an association with two neuropsychiatric disorders (depression and bipolar disorder). Furthermore, a DCC gene SNP rs8084351 from a prior OCD GWAS showed Bonferroni-corrected significant association with putamen volume in White participants. This SNP was also shared in GWASs for depression, schizophrenia, ADHD, and bipolar disorder, as listed in the NHGRI GWAS catalog. Supplementary Tables 3–8 present the neuropsychiatric disorder-wise SNP associations with putamen volume that were assessed at 0.05 significance threshold, which include 261 SNPs for depression (Supplementary Table 3), 353 SNPs for schizophrenia (Supplementary Table 4), 66 SNPs for Parkinson’s disease (Supplementary Table 5), 232 SNPs for ADHD (Supplementary Table 6), 150 SNPs for bipolar disorder (Supplementary Table 7), and 32 SNPs for OCD (Supplementary Table 8).

Table 4.

Associations of putamen volume and the SNPs that were implicated with neuropsychiatric disorders in NHGRI GWAS catalog.

	MAPPED TRAIT	SNP	CHR	POSITION	BIOTYPE	MAPPED GENE	CONTEXT	P NHGRI GWAS	PUBMED ID	A1 WH	MAF WH	A1 NonWH	MAF NonWH	BETA WH	P WH	BETA NonWH	P NonWH
1	Depression	rs1431181	18	53253042	protein coding	DCC	intron	3.E-12	29942085	A	0.40	A	0.16	95.13	5.21E-08	−2.51	0.941
2	Depression	rs4277413	18	53172556	protein coding	DCC	intron	1.E-11	29942085	G	0.47	G	0.29	91.13	9.68E-08	−17.73	0.506
3	Depression	rs17410557	18	53250021	protein coding	DCC	intron	9.E-10	34586374	C	0.37	C	0.15	93.26	1.19E-07	−7.16	0.836
4	Depression	rs12968428	18	53223855	protein coding	DCC	intron	2.E-12	29942085	G	0.39			−84.85	1.03E-06
5	Depression	rs8097041	18	53371847	protein coding	DCC	intron	2.E-11	29942085	A	0.39	T	0.47	−79.65	5.41E-06	42.66	0.076
6	Depression	rs7505145	18	53185406	protein coding	DCC	intron	2.E-08	29942085	A	0.47			−74.44	1.10E-05
7	Depression	rs7231178	18	53227371	protein coding	DCC	intron	2.E-09	35017462	G	0.48	G	0.48	−73.54	1.54E-05	−6.26	0.805
8	Depression	rs8099160	18	53226240	protein coding	DCC	intron	2.E-14	29942085	A	0.44	A	0.27	73.91	1.54E-05	−0.59	0.983
9	Depression	rs11663393	18	53088362	protein coding	DCC	intron	2.E-10	31926635	A	0.45	A	0.32	72.22	2.02E-05	5.74	0.825
10	Depression	rs4632195	18	53220378	protein coding	DCC	intron	6.E-18	34045744	T	0.50	T	0.48	71.51	2.34E-05	26.20	0.294
1	Schizophrenia	rs11665242	18	53380757	protein coding	DCC	intron	9.E-10	33169155	G	0.40	G	0.16	104.60	2.57E-09	−2.08	0.951
2	Schizophrenia	rs1431196	18	53305732	protein coding	DCC	intron	6.E-11	31374203	G	0.41	G	0.16	104.00	2.76E-09	0.18	0.996
1	Parkinson disease	rs8017172	14	55732330	-	-	intergenic	2.E-07	28892059	A	0.45	G	0.30	−131.50	4.75E-15	139.80	2.54E-07
2	Parkinson disease	rs8012377	14	55724695	lncRNA	LOC124903320	downstream	2.E-10	38155330	T	0.47	C	0.28	−127.10	4.22E-14	151.00	3.51E-08
3	Parkinson disease	rs947211	1	205783537	lncRNA	-	nc transcript exon	2.E-12	19915576	A	0.28	G	0.49	76.35	6.50E-05	−32.08	0.194
4	Parkinson disease	rs6060983	20	31833121	protein coding	MYLK2	intron	6.E-07	32201043	C	0.30			−72.76	6.67E-05
5	Parkinson disease	rs708726	1	205793011	protein coding	SLC41A1	intron	2.E-08	37805635	T	0.28	G	0.45	75.65	7.22E-05	−22.72	0.364
1	ADHD	rs62098012	18	53335941	protein coding	DCC	intron	7.E-12	34446935	T	0.40	T	0.16	105.50	1.78E-09	−0.73	0.983
2	ADHD	rs17411548	18	53296573	protein coding	DCC	intron	4.E-14	35764056	T	0.40	T	0.17	103.60	3.07E-09	−4.11	0.902
3	ADHD	rs6508220	18	53304806	protein coding	DCC	intron	7.E-17	35764056	A	0.49	A	0.30	95.40	2.29E-08	3.62	0.890
4	ADHD	rs9807529	18	53418531	protein coding	DCC	intron	7.E-11	35764056	G	0.42	G	0.18	96.09	2.90E-08	−11.00	0.728
5	ADHD	rs8083850	18	53180922	protein coding	DCC	intron	7.E-15	35764056	G	0.42			−93.87	4.76E-08
6	ADHD	rs11664788	18	53446816	protein coding	DCC	intron	3.E-13	35764056	T	0.46	G	0.36	-89.77	1.46E-07	−43.13	0.080
7	ADHD	rs11663156	18	53385388	protein coding	DCC	intron	2.E-10	35764056	C	0.32	C	0.23	−87.45	8.68E-07	15.24	0.602
8	ADHD	rs4513167	18	53162159	protein coding	DCC	intron	2.E-08	38565336	T	0.39			−77.34	8.48E-06
9	ADHD	rs10787738	10	117017860	protein coding	SHTN1	intron	3.E-19	34446935	T	0.28	T	0.36	83.09	9.55E-06	18.69	0.453
10	ADHD	rs965943	18	53331060	protein coding	DCC	intron	1.E-10	35764056	G	0.39			−76.72	9.57E-06
11	ADHD	rs4632195	18	53220378	protein coding	DCC	intron	4.E-10	33479212	T	0.50	T	0.48	71.51	2.34E-05	26.20	0.294
1	Bipolar	rs11663393	18	53088362	protein coding	DCC	intron	2.E-10	31926635	A	0.45	A	0.32	72.22	2.02E-05	5.74	0.825
2	Bipolar	rs4632195	18	53220378	protein coding	DCC	intron	4.E-10	33479212	T	0.50	T	0.48	71.51	2.34E-05	26.20	0.294
1	OCD	rs8084351*	18	53200189	protein coding	DCC	intron	4.E-12	31835028	A	0.49	A	0.42	65.89	9.97E-05	−18.00	0.475

These known SNP associations were assessed at a stringent Bonferroni-corrected significance threshold.

*rs8084351 NHGRI GWAS P-VALUES [TRAIT = P(PUBMED ID)]: Depression=4E-12(31835028), Schizophrenia=2E-09 (35396580), ADHD = 4E-09(32606422), Bipolar=4E-12(31835028), and OCD = 4E-12(31835028).

Table 5 presents 41 known SNPs that were associated with at least three neuropsychiatric disorders – depression [55, 56, 68–72], schizophrenia [58, 67, 68, 71–79], Parkinson’s disease [62], ADHD [64, 65, 67, 68, 71, 78], bipolar disorder [55, 67, 68, 71, 72, 75, 80–84], and OCD [68] – as identified in the NHGRI GWAS catalog, and also with putamen volume at a threshold level of 0.05 in ABCD White and/or non-White participants. Pair-wise LD patterns showed that these SNPs were mostly independent with only exception of three SNPs from chromosome 4 and two SNPs from chromosome 18 (Supplementary Table 9). The full list of SNPs that shared an association between putamen volume at this threshold and two or more neuropsychiatric disorders is presented in Supplementary Table 10. A visualization of data from Supplementary Table 10 is presented in Supplementary Figure 6, showing the sharing or overlap of all significantly associated SNPs with putamen volume, depression, schizophrenia, Parkinson’s, ADHD, bipolar disorder, and OCD. There were 15, 10, 16, and 51 SNPs that shared with 5, 4, 3, and 2 neuropsychiatric disorders, respectively, and were also associated with putamen volume in ABCD White and/or non-White participants (Table 5, Supplementary Table 10, and Supplementary Figure 6). These outcomes represent the shared genetic architecture of putamen volume and neuropsychiatric disorders.

Table 5.

Shared genetic architecture of putamen volume and neuropsychiatric disorders, including depression (DEP), schizophrenia (SCH), Parkinson’s disease (PD), ADHD, bipolar disorder (BD), and OCD.

	SNP	CHR	POSITION	BIOTYPE	MAPPED GENE	CONTEXT	A1 WH	MAF WH	A1 NonWH	MAF NonWH	BETA WH	P WH	BETA NonWH	P NonWH	SHARED ASSOCIATIONS TRAITS (PUBMED ID)
1	rs8084351	18	53200189	protein coding	DCC	intron	A	0.49	A	0.42	65.89	9.97E-05	−18.00	0.475	DEP (31835028), SCHZ (35396580), ADHD (32606422), BD (31835028), OCD (31835028)
2	rs35225200	4	102225731	lncRNA	-	intergenic	C	0.08	C	0.03	93.71	2.06E-03	−14.61	0.848	DEP (31835028), SCHZ (31374203), ADHD (31835028), BD (31835028), OCD (31835028)
3	rs174592	11	61851136	protein coding	FADS2	intron	G	0.40	A	0.46	−48.76	4.80E-03	1.92	0.936	DEP (31835028), SCHZ (31835028), ADHD (31835028), BD (34002096), OCD (31835028)
4	rs631399	15	37393066	lncRNA	LOC105370772	intron	C	0.37	C	0.38	46.68	6.29E-03	−30.64	0.210	DEP (31835028), SCHZ (31835028), ADHD (31835028), BD (31835028), OCD (31835028)
5	rs5758265	22	41221893	protein coding	L3MBTL2	intron	A	0.30	A	0.36	43.77	0.018	6.31	0.801	DEP (31835028), SCHZ (32606422), ADHD (31835028), BD (31835028), OCD (31835028)
6	rs5757717	22	39546229	-	RPS19BP1 - CACNA1I	regulatory region	T	0.44	T	0.33	−39.78	0.020	−4.14	0.872	DEP (31835028), SCHZ (26198764), ADHD (31835028), BD (31835028), OCD (31835028)
7	rs77135925	17	29168251	protein coding	MYO18A	intron	C	0.08	C	0.03	70.22	0.023	63.32	0.384	DEP (31835028), SCHZ (31835028), ADHD (31835028), BD (31835028), OCD (31835028)
8	rs55648125	6	50849005	protein coding	TFAP2B	regulatory region	G	0.10	G	0.04	−60.80	0.028	−42.86	0.464	DEP (31835028), SCHZ (35396580), ADHD (31835028), BD (34159505), OCD (31835028)
9	rs9834970	3	36814539	lncRNA	LINC02033	regulatory region	T	0.49	C	0.41	−33.31	0.048	16.30	0.509	DEP (31835028), SCHZ (34159505), ADHD (33686288), BD (34159505), OCD (31835028)
10	rs7200826	16	12972976	protein coding	SHISA9	intron	T	0.24	T	0.10	−35.41	0.067	83.73	0.035	DEP (30718901), SCHZ (31835028), ADHD (31835028), BD (31835028), OCD (31835028)
11	rs2514218	11	113522272	-	DRD2 - TMPRSS5	intergenic	T	0.34	T	0.20	−26.19	0.141	66.91	0.027	DEP (31835028), SCHZ (32606422), ADHD (33479212), BD (32606422), OCD (31835028)
12	rs880090	19	19629920	protein coding	GMIP	3 prime UTR	C	0.28	C	0.36	−23.76	0.203	−84.53	6.25E-04	DEP (31835028), SCHZ (31835028), ADHD (31835028), BD (31835028), OCD (31835028)
13	rs2057884	7	105289803	protein coding	SRPK2	intron	T	0.35	T	0.48	−22.21	0.204	56.00	0.018	DEP (31835028), SCHZ (30285260), ADHD (31835028), BD (31835028), OCD (31835028)
14	rs140505938	1	150059494	lncRNA	-	intron	T	0.16	T	0.06	7.54	0.740	−124.50	0.017	DEP (31835028), SCHZ (32606422), ADHD (31835028), BD (31835028), OCD (31835028)
15	rs1730054	3	158237129	protein coding	RSRC1	intron	A	0.42	A	0.35	0.26	0.988	−49.71	0.047	DEP (31835028), SCHZ (31835028), ADHD (31835028), BD (31835028), OCD (31835028)
16	rs4632195	18	53220378	protein coding	DCC	intron	T	0.50	T	0.48	71.51	2.34E-05	26.20	0.294	DEP (34045744), SCHZ (32606422), ADHD (33479212), BD (33479212)
17	rs13107325	4	102267552	protein coding	SLC39A8	missense	T	0.07	T	0.02	113.90	5.00E-04	−44.95	0.580	DEP (34734193), SCHZ (32606422), Parkinson’s (32201043), ADHD (35764056)
18	rs12443954	16	89675088	protein coding	SPATA33	intron	G	0.26	G	0.38	−59.47	2.01E-03	−19.26	0.441	DEP (23453885), SCHZ (23453885), ADHD (23453885), BD (23453885)
19	rs4650608	1	78772330	-	IFI44 - ADGRL4	intergenic	C	0.32	C	0.26	−51.59	4.18E-03	8.68	0.748	DEP (23453885), SCHZ (24280982), ADHD (23453885), BD (22182935)
20	rs3791556	2	239191276	protein coding	HDAC4	intron	A	0.18	A	0.17	−48.59	0.027	27.90	0.387	DEP (23453885), SCHZ (23453885), ADHD (23453885), BD (23453885)
21	rs6765687	3	52699090	protein coding	GLT8D1	intron	T	0.24	T	0.11	−42.01	0.033	80.42	0.040	DEP (23453885), SCHZ (23453885), ADHD (23453885), BD (23453885)
22	rs6867265	5	131078288		ARL2BPP4 - HINT1	intergenic	G	0.09			−59.18	0.050			DEP (23453885), SCHZ (23453885), ADHD (23453885), BD (23453885)
23	rs30266	5	104636656	lncRNA	-	intron	A	0.30	A	0.13	−29.78	0.101	−91.91	0.014	DEP (30718901), SCHZ (33479212), ADHD (33479212), BD (33479212)
24	rs10250997	7	136461788	-	TRPC6P8 - ZP3P2	intergenic	A	0.10	A	0.18	40.87	0.132	−86.83	6.19E-03	DEP (23453885), SCHZ (23453885), ADHD (23453885), BD (23453885)
25	rs11587682	1	150351808	protein coding	PRPF3	intron	T	0.12	T	0.04	−4.31	0.868	−196.80	1.03E-03	DEP (23453885), SCHZ (23453885), ADHD (23453885), BD (23453885)
26	rs13135092	4	102276925	protein coding	SLC39A8	intron	G	0.08	G	0.03	89.59	3.75E-03	−10.79	0.888	SCHZ (33686288), ADHD (33686288), BD (32606422)
27	rs1054442	12	48995537	protein coding	DDN	3 prime UTR	C	0.37	A	0.49	−43.62	0.012	−17.93	0.467	SCHZ (31374203), ADHD (35764056), BD (27329760)
28	rs7325477	13	109409474	-	LINC00399 - LINC00676	regulatory region	C	0.34	A	0.48	−44.40	0.013	8.99	0.710	DEP (34099189), SCHZ (34099189), BD (34099189)
29	rs2339515	2	22341421	lncRNA	LINC01830	intron	T	0.37			−42.42	0.014			SCHZ (33479212), ADHD (33479212), BD (33479212)
30	rs73033411	7	2841916	protein coding	GNA12	intron	C	0.15	C	0.14	−51.96	0.026	24.47	0.479	DEP (34099189), SCHZ (34099189), BD (34099189)
31	rs7914279	10	120402378	-	RPL21P16 - PLPP4	intergenic	G	0.09			62.87	0.030			DEP (34099189), SCHZ (34099189), BD (34099189)
32	rs59724122	8	27567179	pseudogene	GULOP	intron	T	0.15	T	0.10	−49.65	0.036	29.83	0.453	SCHZ (31268507), ADHD (34446935), BD (29121268)
33	rs4858204	3	20446208	lncRNA	SGO1-AS1	intron	A	0.41	A	0.26	34.75	0.041	3.10	0.912	DEP (34099189), SCHZ (34099189), BD (34099189)
34	rs12658032	5	104568525	lncRNA	-	intron	A	0.34	A	0.24	−26.20	0.135	−83.52	3.19E-03	DEP (31926635), ADHD (32606422), BD (31926635)
35	rs12898878	15	48946233	protein coding	SHC4	intron	C	0.50	T	0.38	20.50	0.224	−49.34	0.048	DEP (34099189), SCHZ (34099189), BD (34099189)
36	rs35841922	5	104736717	lncRNA	-	intron	C	0.15	C	0.06	−22.99	0.321	−102.60	0.046	SCHZ (33479212), ADHD (33479212), BD (33479212)
37	rs12341335	9	25649147	-	RN7SKP120 - TUSC1	intergenic	C	0.11	C	0.21	11.70	0.660	94.89	1.52E-03	DEP (34099189), SCHZ (34099189), BD (34099189)
38	rs7897310	10	36748480	-	Y_RNA - MKNK2P1	intergenic	A	0.34	A	0.22	−3.75	0.831	−72.77	0.015	DEP (34099189), SCHZ (34099189), BD (34099189)
39	rs6484581	11	32494908	lncRNA	WT1-AS	intron	C	0.50	C	0.39	−3.36	0.842	54.46	0.025	DEP (34099189), SCHZ (34099189), BD (34099189)
40	rs3796696	4	184283978	lncRNA	-	intergenic	A	0.06	A	0.08	1.70	0.960	90.22	0.039	DEP (34099189), SCHZ (34099189), BD (34099189)
41	rs4648996	1	24573039	protein coding	NCMAP	intron	C	0.33	C	0.27	0.44	0.980	−56.42	0.042	DEP (34099189), SCHZ (34099189), BD (34099189)

The candidate SNP associations with putamen volume were assessed at the 0.05 significance threshold in ABCD White and/or non-White participants. The listed shared SNP associations with at least three neuropsychiatric disorders, as identified from NHGRI GWAS catalog, in addition to their associations with putamen volume in ABCD dataset.

Discussion

In this study, we successfully evaluated the genetic main effect on putamen volume using GWAS in ABCD White participants and showed the generalizability of their associations in distinct and genetically diverse non-White participants. We performed pathway enrichment scoring analysis using the GWAS results and successfully demonstrated that the known SNPs from prior genetic studies on putamen volumes overlapped with our discovery SNPs and also associated with putamen volume in the ABCD White and/or non-White participants. In addition, we assessed the shared genetic associations between putamen volume and multiple neuropsychiatric disorders, i.e., depression, schizophrenia, Parkinson’s disease, ADHD, bipolar disorder, and OCD. The most notable contributions were the following: 1) We identified 142 SNPs with genome-wide significant SNP-putamen volume associations in ABCD White participants. 2) Of these, more than a dozen of the most significant discovery SNPs were associated with putamen volume in non-White participants, including five regulatory region SNPs (i.e., rs8017172, rs1953350, rs1953351, rs1959089, and rs1953352). 3) Five known SNPs from prior putamen volume GWASs [17, 18, 20, 49–51] overlapped with our most significant discovery SNPs, and the associations of 17 of these known SNPs were replicated in our analysis of the ABCD study dataset. 4) Despite a successful generalization of the discovery SNP associations, we discovered that there is significant genetic heterogeneity between White and non-White participants in context of putamen-linked SNPs. 5) We identified a key pathway REACTOME_DSCAM_INTERACTIONS (https://www.reactome.org/content/detail/R-HSA-376172) linked with putamen volume that involves the DSCAM gene, netrin-1 protein, and DCC gene, depending on the cellular and developmental context [46, 48]. 6) We identified shared genetic associations with putamen volume and a half-dozen neuropsychiatric disorders (depression, schizophrenia, Parkinson’s disease, ADHD, bipolar disorder, and OCD).

The most notable genes identified in our study were the DCC and DSCAM genes. Published literature suggests that DCC is a transmembrane protein with immunoglobulin and fibronectin domains [41], and the DCC axon guidance receptor and its ligands, the netrins, are crucial for guiding axonal projections toward the ventral midline in the developing nervous system [42, 43]. Netrins act as chemoattractants or chemorepellents for axons [85]. Netrin-1 binding to DCC causes chemoattraction, while interaction with UNC5 leads to chemorepulsion [86]. DCC and netrin-1 are essential for commissural axon guidance and neuronal migration [41, 44]. DSCAMs, i.e., Down syndrome cell adhesion molecules, immunoglobulin-like transmembrane proteins with fibronectin III domains, were identified on chromosome 21 and regulate neuronal wiring in humans and other vertebrates [45]. During nervous system development, spinal commissural axons are guided by netrin-1, which activates the DCC receptor, while DSCAM binding to netrin-1 is essential for axon guidance, acting independently or with DCC [46]. Local translation of synaptic mRNAs, crucial for CNS development, is impaired in intellectual disabilities like Down syndrome, where DSCAM may be locally translated in dendrites and axonal growth cones of hippocampal neurons [47].

Observation of the most significant discovery SNP–putamen volume associations in non-White participants demonstrated the generalizability of these associations across a distinct and genetically diverse group of participants. The known SNPs that overlapped with our discovery SNPs (Table 1) and were associated with putamen volume in the ABCD dataset (Table 3) were reported in three prior studies. The datasets analyzed in these studies also included adolescent participants—two of the studies (PubMed IDs: 25607358 and 29147026 [17, 20]), were based on participants aged 9–97 years, and the third study (PubMed ID: 31636452 [18]) included participants aged 9–98 years. However, we observed reduced p-value significance in these replication efforts (Table 3). This could be partly due to differences in participants size, population structure, genetic heterogeneity, or context-specific effects across populations. Overall, our study underscores the importance of replication in validating genetic associations and highlights the complexity of genetic influences across different groups.

Our findings of SNP-putamen volume associations are consistent with the high heritability (i.e., 71–79% [18]) of putamen volume. However, epigenetic influences may overshadow the genetic influences [87] particularly in the presence of stressors [88–90]. Observing relatively fewer associations in non-White participants, which also include Black participants among others, may be partly due to the fact that a substantial body of research indicates that non-White individuals, particularly Black individuals, are exposed to higher levels of stress compared to other racial groups [91]. For example, factors such as prenatal stress [92], adverse childhood experiences (ACEs) [93], and parental depression [94] may give rise to epigenetic changes like DNA methylation [95], which in turn may affect putamen volume. Particularly, ACEs have a profound impact on mental and physical health, disrupting brain development in regions related to stress regulation, including the prefrontal cortex, hippocampus, amygdala, and putamen [96], contributing to mental health disorders [97]. Parental depression, particularly maternal, is also linked with reduced putamen volume in children [94, 98]. Further studies are needed to evaluate the influence of early-life stressors and epigenetics on putamen volume and associated neuropsychiatric disorders.

In addition, we also recognize some limitations of this study. Pooling genetically diverse individuals in a single non-White group for the generalization of discoveries in the White group may introduce confounding due to ancestry-related allele frequency differences. To partially mitigate this concern, we included principal components of genetic ancestry as covariates in all analyses to adjust for population structure. We emphasize that the aggregated generalization results may be viewed as preliminary evidence that requires confirmation in larger, ancestry-specific samples. Our pathway analysis suggested that the ABCD adolescent GWAS SNP associations were enriched for pathways related to neuronal growth, specifically DSCAM’s binding to netrin-1 and its cooperation with, or independence from, DCC in guiding commissural axons. The observed overlap of known SNPs from prior studies with our most significant discovery SNPs could very likely be due to the presence of adolescent participants in those studies [17, 18, 20]. However, it is plausible that putamen–SNP associations in adult populations may be enriched in biological pathways related to the putamen’s roles in maintenance [99], degeneration [100], or compensation [101]—such as pathways involved in neurotransmitter transmembrane transport [102]. Developmental stage across the lifespan may influence the biological interpretation of GWAS findings. Therefore, more work evaluating pathways using genome-wide associations and directly comparing participants in adolescence and adulthood within a longitudinal cohort would help clarify whether age-specific genetic architectures or biological pathways contribute to putamen volume variation.

In conclusion, we identified a comprehensive list of genome-wide significant SNPs associated with putamen volume, discovering (i) robust and generalizable regulatory SNP associations, (ii) genetic heterogeneity in putamen-linked SNPs, (iii) biological pathways involving genes associated to putamen volume, and (iv) genetic associations that are shared between putamen volume and several neuropsychiatric disorders, including depression, schizophrenia, Parkinson’s disease, ADHD, bipolar disorder, and OCD. Our findings provide a deeper understanding of genetic architecture and cross-population differences associated with putamen volume, which may help with the advancement of precision medicine.

Author contributions

AS contributed to the study design, performed data preparation, conducted data analysis, contributed to the interpretation of the data and results, and drafted the manuscript. JP contributed to the study design and interpretation of the data and results, and reviewed the draft. Both authors approved the final manuscript.

Data availability

The data for the Adolescent Brain and Cognitive Development study, which was used for the secondary analysis in this study, is publicly available. The dataset can be accessed through authorized data access from the NIH Brain Development Cohorts (NBDC) Data Hub (https://www.nbdc-datahub.org/abcd-study).

Competing interests

Authors have no conflicts of interest in relation to the work presented in this manuscript.

Ethics approval and consent to participate

Our study does not report any individual participant’s data or health related outcome. It involves a secondary analysis on data of human samples that was previously collected and anonymized by ABCD Study (see DATA AVAILABILITY subsection) as per the informed consent of participants. The use of ABCD Study dataset for this secondary analysis study was approved by Duke Health Institutional Review Board (IRB) protocol number Pro00108729. All methods were performed in accordance with the relevant guidelines and regulations.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41398-026-04299-6.